Bulk Acoustic Wave Gyroscopes: Houri Johari

5 Bulk Acoustic Wave
Gyroscopes
Houri Johari
CONTENTS
5.1 Introduction............................................................................................................................. 91
5.2 Principle of Operation.............................................................................................................92
5.3 Design of BAW Gyros............................................................................................................. 93
5.3.1 Angular Gain Estimation............................................................................................. 95
5.3.2 Sensitivity Analysis.....................................................................................................96
5.3.3 Resolution Analysis..................................................................................................... 98
5.3.4 Dynamic Range........................................................................................................... 98
5.3.5 Thermoelastic Damping..............................................................................................99
5.4 Implementation of BAW Gyros............................................................................................. 101
5.4.1 Implementation in (100) SCS..................................................................................... 101
5.4.2 Fabrication Method.................................................................................................... 103
5.5 Measurements of BAW Gyros............................................................................................... 103
5.5.1 Frequency Characterization and Mode Matching..................................................... 103
5.5.2 Performance Characterization................................................................................... 104
5.5.3 Quality Factor Characterization................................................................................ 104
5.6 Summary............................................................................................................................... 107
Acknowledgments........................................................................................................................... 109
References....................................................................................................................................... 109
5.1 INTRODUCTION
Coriolis-based microgyroscopes have increasingly been used from a utomotive safety to consumer
electronics to personal navigation. Examples include antiskid stability, control systems in auto-
mobiles, image stabilization in digital cameras, gaming consoles, and dead reckoning in portable
devices using inertial measurement units. As high-volume consumer applications for microma-
chined gyroscopes continue to grow, the design and manufacturing techniques that improve the
dynamic range, shock survivability, and bias stability of the device while reducing the manufactur-
ing cost and die size become increasingly valuable.
The vast majority of high-performance micromachined vibratory gyroscopes reported in the
literature to date utilize the flexural modes of a microstructure, which typically fall in the frequency
range 3–50 kHz [1–5]. Low-frequency flexural-mode microgyroscopes rely on increases in mass
and vibration amplitude to reduce the noise floor and to improve the bias stability [6,7], at the
cost of large form factors and high operating voltages. In bulk acoustic wave (BAW) gyroscopes,
single-crystal silicon (SCS) disk gyroscopes are designed to operate in their degenerate elliptic bulk
acoustic modes with frequencies in the range 1–20 MHz. This enhances the operational bandwidth
of the gyroscope in the mode-matched condition compared to low-frequency (<100 kHz) flexural-
mode gyroscopes. Operating gyroscopes in the mode-matched condition with a high-quality factor
(Q) enhances the signal-to-noise ratio significantly and improves the performance of the gyroscope.
91
92 MEMS: Fundamental Technology and Applications
Also, migrating to the bulk modes operation reduces the form factor while achieving high shock
tolerances, high Q, and high temperature stability.
This chapter gives an overview of the high-frequency BAW gyroscopes. The concept of the
BAW gyroscopes is introduced and the principle of operation for these devices is discussed in
Section 5.1. The fundamental parameters of the BAW gyroscope design are addressed in Section
5.2. The performance parameters including sensitivity, resolution, and dynamic range are analyzed
and explained in Section 5.3. Section 5.4 describes the design requirements for the implementation
of these devices in silicon. Finally, the preliminary characterization and measurement of these
devices are presented in Section 5.5, followed by the environmental effects on the performance of
the devices.
5.2 PRINCIPLE OF OPERATION

The schematic view of a capacitive BAW disk gyroscope is shown in Figure 5.1. It consists of a
Downloaded by [Universiti Malaysia Perlis] at 02:42 01 February 2016
center-supported silicon disk with capacitive drive, sense, and control electrodes surrounding the
disk. The locations of the electrodes depend on the modes of operation and the crystal orientation of
the silicon. Electrodes are located on antinode lines of the two degenerate bulk modes, to provide a
large transduction area and to facilitate mode-matched operation.
The BAW disk gyroscope is made on a silicon-on-insulator (SOI) substrate. The resonating
disk is supported at its center by a small island of the buried oxide (BOX) layer of the SOI
(Figure 5.1). To fabricate these devices on a standard SOI, circular perforations or release holes
should be added to the solid disk to enable the release of the structure from the front side of the
wafer (though not shown in Figure 5.1). The size and configuration of such perforations need to
be carefully optimized to minimize the frequency split between the two modes [8]. As verified
by finite element analysis software (ANSYS) simulations, which are discussed in Section 5.4,
small holes (with diameter <10 μm) are preferable to minimize the mode shape distortion and
frequency separation. The polysilicon traces attached to the center of the disk, to which they are
self-aligned, provide a DC bias voltage to the disk structure.
To excite the bulk acoustic modes of the disk gyroscope, large electrostatic forces are required,
which in turn necessitate strong capacitive coupling. This is obtained with high-aspect-ratio nar-
row capacitive gaps. Thus, the device is implemented in thick SOI substrates (30–60 μm) with deep
submicron capacitive gaps (of typically 200 nm).
z-axis
Sense electrode
Polysilicon trace
Center 30°
support
Drive electrode
Silicon handle
Tuning
electrodes
Silicon device layer
FIGURE 5.1 Schematic diagram of a capacitive BAW disk gyro in (100) silicon.
Bulk Acoustic Wave Gyroscopes 93
To actuate the capacitive BAW disk gyroscope, an AC signal is applied to the drive electrode
at the selected resonance frequency, while a DC bias is applied to the disk itself. When the disk
is driven in its elliptic resonant mode, Coriolis acceleration causes a transfer of energy between
the two degenerate elliptic modes due to the application of rotation rate around the normal axis
of the device. As energy builds up in the degenerate mode, output current is created at the sense
electrodes. A custom-built interface circuit amplifies, demodulates, and filters the sense current to
yield rotation-induced Coriolis signal [9]. The principal features of the design are the axisymmetry
of the disk along with the cyclic-symmetric arrangement of the release holes. These features allow
the intended modes of vibration to exist in degenerate pairs [10]. To facilitate the mode-matching
process and consequently improve the performance, typically, two degenerate modes with the same
resonance frequency are used as the drive and sense modes [11,12]. When the resonating disk is
implemented in an isotropic material such as (111) SCS or polycrystalline silicon [13,14], the two
primary elliptic degenerate modes that are spatially 45° apart will have identical resonance fre-
quencies (degenerate pair). However, if the disk gyroscope is implemented in (100) silicon, the two
primary degenerative modes have a frequency split of about 1 MHz. This makes them unsuitable for
mode-matched operation of a gyroscope. Instead, the secondary elliptic modes of the same device,
which are spatially 30° apart, have nearly identical frequencies and should be employed to enable
mode-matched operation in (100) silicon.
Fabrication of nonidealities and the existence of perforations inevitably introduce some amount
of frequency separation between the two degenerate modes. Some electrodes around the device are
used to electrostatically tune and null the frequency split between the two modes [15].
5.3 DESIGN OF BAW GYROS

In this section, the equations of motion for the BAW disk gyroscopes are discussed and the response
to an applied rotation rate is calculated. For simplicity, a solid disk structure is assumed in the entire
analysis.
In general, the displacement of a linear elastic body due to vibration can be written as a linear
combination of its normal mode shape functions and its amplitudes [16,17]. For a disk gyroscope,
operating in its degenerate elliptic modes, the vibration displacement of each point on the disk is a
summation of both degenerate mode displacements written as
ur = q1fr1 + q2fr 2
(5.1)
uq = q1fq1 + q2fq 2
where ur and u θ are the radial and circumferential components of the displacement vector in a
vibrating disk, ϕr and ϕθ are the mode shape functions in the radial and circumferential directions,
respectively, and q1 and q2 are the amplitudes of the two degenerate modes. Subscripts 1 and 2,
respectively, represent the first degenerate mode (drive) and the second degenerative mode (sense)
in the entirety of this section. The mode shape functions for a solid disk [18,19] operating in its
bulk modes are
fr1 = Ur cos(mq ), fq1 = Uq sin(mq )

(5.2)
fr 2 = −Ur sin(mq ), fq 2 = Uq cos( mq )

where m is the mode order number; for the primary elliptic mode, m = 2 and for the secondary
elliptic mode, m = 3. Ur and Uθ are the normalized dimensionless displacements in the radial and
circumferential directions, respectively [16], defined as
 d  k r m  h r
Ur =  R J m  m  + Rxm J m  m  
 dr  R  r  R 
(5.3)
 m  k r d  h r
Uq =  − RJ m  m  − Rxm Jm  m  
 r  R  dr  R  

where R is the disk radius at the perimeter of the disk. The normalized displacements are expressed
in terms of Bessel functions of the first kind (Jm) as well as km and hm, which are both dimensionless
frequency parameters and expressed in Ref. [20]. ξm is the next parameter in Equation 5.3, which
was derived and explained in detail in Refs. [16,20] and it depends both on the material’s Poisson
ratio and on the operating mode number (m). With km, hm, and ξm being known, the normalized dis-
placements (Ur and Uθ) can be estimated, and the mode shape functions (ϕr and ϕθ) are fully defined.
Next, the expressions for the amplitude of the two modes (q1, q2) should be determined to fully
quantify the displacements (ur and u θ). To accomplish this, the normal mode differential equations
were solved, using the Lagrange method [20]. The details of these derivations can be found in
Ref. [20]. The coupled second-order differential equations used to model the disk gyroscope are
expressed in (5.4) and can be solved for q1 and q2:
M1q1 + D1q1 − 2g Ω z q2 − g Ω q − M Ω2 q + ( K − K )q = F

z 2 1 z 1 m1 e1 1 1
(5.4)

M 2 q2 + D2 q2 − 2g Ω z q1 − g Ω z q1 − M 2 Ω z q2 + ( K m 2 − K e 2 )q2 = F2
2

where Km1 and Ke1 are mechanical and electrical stiffnesses and are defined in Ref. [20]. Also, D1
and D2 are the damping coefficients for the drive and sense modes, respectively. F1 and F2, the elec-
trostatic forces originating from the electrical potential term, are derived and expressed in Ref. [20].
F2 is identical to F1 except that drive voltages are replaced with sense voltages. Also, M1 and γ are,
respectively, the effective mass and Coriolis coupling coefficient and are defined as
M1 =∫ r (f + f ) dV
2
r1
2
q1
(5.5)

γ =∫ r (f f − f f ) dV
r1 q 2 r 2 q1
The schematic view of a capacitive disk gyroscope operating in its secondary elliptic modes is
shown in Figure 5.2. As shown, Δθn is the angle at which the electrode spans, θn the angle from
the x-axis to the center of the nth electrode, and Δd the capacitive gap change, which is defined as
Δd = ur = qr1ϕr1 + qr2ϕr2.
Δd
θn
Δθn
vd
d0
VP
FIGURE 5.2 Schematic view of a capacitive disk gyroscope operating at its secondary elliptic modes.
It is worth noting that the change in the capacitive gap (Δd) depends on the mode of vibration
and the location of the electrode. Since the capacitive gap (d0) is much smaller than the radius (R)
of the disk gyroscope, the circumferential displacement is very small in comparison with a radial
displacement and can be neglected. Thus, only the radial displacement is considered for the capaci-
tive gap changes. Given that the drive electrodes are simultaneously located on the antinodal lines
of the drive mode and on the nodal lines of the sense mode, it implies that (ϕr1 ≫ ϕr2) at the drive
electrodes and similarly (ϕr2 ≫ ϕr1) at the sense electrodes. However, owing to the fabrication of
nonidealities, the antinodal lines of the mode shapes might be slightly misaligned with respect to
the electrode center. This can be corrected for by electrically canceling the quadrature errors.
5.3.1 Angular Gain Estimation

One of the critical parameters in designing a Coriolis-based gyroscope is the angular gain (Ag).
The angular gain is the amount of the effective mass during resonance that couples into the sense
Coriolis mode. The angular gain has the maximum value of 1 for the frame-based (rectangular-
shaped) gyros. However, the angular gain has a smaller value than one for the shell-typed gyro
structures. The angular gain in shell-typed gyros depends on the sensor structure as well as the
type of the resonant modes in operation. In shell-typed gyros, when a rotation rate is applied
to the sensor, the nodal lines of the vibration pattern lag behind the rotation of the sensor. The
angular gain (Ag) is defined as the ratio of the change in the vibration pattern angle to the applied
angle of rotation (Ωz). To estimate the angular gain in BAW gyros, matched drive and sense
frequencies, along with zero damping, are assumed. Also, the angular acceleration coupling is
neglected due to its small magnitude compared to the Coriolis coupling. With these assumptions,
the normal model Equations 5.4 are simplified and solved for the drive and sense amplitudes
(q1, q2) as expressed as
g 
q1 = A cos  Ω z t  sin(w t )
M 
(5.6)
g 
q2 = − A sin  Ω z t  sin(w t )
 M 

where A is a constant amplitude term and ω is defined as w 0 2 − Ω z 2 (1 − (g /M )2 ) in which ω0, Ωz,
γ, and M are, respectively, the resonant angular frequency, an applied rotation rate around the z-axis,
the Coriolis coupling coefficient, and the effective mass.
The drive mode amplitude (q1) is assumed to have the maximum amplitude (A) at 0°. This is
because the electrode located at 0° is aligned with the antinodal line of the drive mode. With the
application of a constant rotation rate (Ωz), the vibration pattern also rotates at a related constant
rate. After a certain time (t), the drive mode has transferred completely to the sense mode, due to
the mode’s degenerate interrelation. At this time, the term in the first parentheses in Equation 5.6,
(γ Ωzt/M), is equal to 90°, and the amplitude of the sense mode will reach its maximum value. The
sensor has rotated through an angle β = Ωzt = πM/2γ. While the sensor rotates through β, Coriolis
force shifts the vibration pattern through an angle (θ) from the drive mode to its degenerate mode
(sense mode).
It is worth noting that the degenerate primary elliptic modes in (111) SCS are separated by 45°
about the disk’s axis. However, the secondary elliptic modes in (100) SCS are separated by 30°.
Taking the ratio of the vibration pattern angle (θ) to the angle of rotation in the sensor (β), the angu-
lar gain for a disk operating in its elliptic modes is calculated as
q p / 2m g
Ag = = = (5.7)
b pM / 2g mM
TABLE 5.1
Design Parameters in BAW Solid Disk Gyroscopes
(111) SCS Primary (100) SCS Secondary
Material/Mode Elliptic Mode (m = 2) Elliptic Mode (m = 3)
Frequency parameter (km) 1.4243 2.2988
Frequency parameter (hm) 2.3447 3.5836
Elastic wave ratio (ξm) 2.2071 1.1369
Resonant frequency (MHz) ANSYS 5.01 ANSYS 7.51
Theoretical 5.00 Theoretical 7.42
Normalized radial displacement (Ur) at r = R 1.5105 0.9420
Normalized tangential displacement (Uθ) at r = R 0.4621 0.0708
Effective mass (m, kg) 2.0676e − 8 1.4661e − 008
Coriolis coupling coefficient (γ, kg) 1.8954e − 8 1.0175e − 008

Angular gain (Ag) 0.45 0.23
To evaluate the angular gain (Ag) for primary and secondary elliptic modes, the mode shape
functions for the solid disk need to be integrated over the disk volume to determine the Coriolis cou-
pling coefficient (γ) and the effective mass (M). Given the disk radius (R) and the material properties
(E, ν) of the disk as well as the resonance frequency of operation (ω0), the frequency parameters (km)
and (hm) should be estimated first. Next, the ratio of the elastic wave constants (ξm) is estimated by
solving the Bessel function with the frequency parameters. Then, the normalized displacements (Ur
and Uθ) are calculated and the mode shape functions are determined in terms of (r, θ) and integrated
over the disk volume. Finally, the angular gain values are calculated and summarized in Table 5.1.
As can be seen, the angular gain is found to be ~1.9 times larger for primary elliptic modes in (111)
silicon than for secondary elliptic modes in (100) silicon.
5.3.2 Sensitivity Analysis
The sensitivity of a gyroscope is determined by the capacitance change at the sense electrodes, orig-
inating from the sense mode excitation due to the Coriolis force. Here, we consider an ideal device
with matched degenerate mode frequencies and symmetric damping. As the angular acceleration
coupling is insignificant in comparison with Coriolis coupling, this can be neglected. Similarly, the
centrifugal stiffness, being very small compared to mechanical and electrical stiffnesses, is also
assumed to be negligible. Consequently, the second-order differential equations for the drive and
sense modes are simplified and expressed as
w0 F
q1 + q + w 02 q1 = 1
Q 1 M1
(5.8)
w0
q2 + q + 2 mAg Ω z q1 + w 02 q2 = 0
Q 2
These equations are solved for the drive and sense amplitudes, and their ratio is derived [20] as
stated as
A2 q 2 mAg Ω z Q
= 2 =− (5.9)
A1 q1 w0
As can be seen, the sense mode amplitude is linearly proportional to the mode-matched quality
factor (Q), angular gain (Ag), and the drive mode amplitude (q1), and is inversely proportional to
the operating resonance frequency (ω0). As a result, high mode-matched quality factor is required
to enhance the sense mode amplitude and the sensitivity in high-frequency gyroscopes due to their
small drive amplitude (a few tens of nm). To determine the sensitivity of a BAW gyroscope, the
total capacitance change at the sense electrode should be evaluated. The capacitance change (δCs)
due to the vibration motion of the sense mode at the sense electrode is derived in Ref. [20] and is
expressed as
dCs 4 me0 hRq1 Ag QUr = R  ∆q 

= sin(mq )sin  m (5.10)
Ωz w 0 md0 2  2 

It is worth noting that sin(mθ) is equal to 1 for both primary and secondary elliptic modes due
to the multiplication of the mode number (m) by the vibration pattern angle (θ). Also, the sensitiv-
ity is linearly proportional to the angular gain. Thus, higher sensitivity can be achieved for mode-
matched devices implemented in (111) SCS and operated in their primary elliptic modes.
The change of the sense capacitance due to the applied rotation is very small and needs to be
amplified properly to be readable. Therefore, each sense electrode is accompanied by a low-noise
transimpedance amplifier (TIA) and two cascaded voltage amplifiers. The fully amplified signal
is a cross-product of two signals (carrier and Coriolis) and can thus be treated as an amplitude-
modulated (AM) signal. It is then demodulated to extract the rotation rate (Coriolis signal) from the
device by mixing the carrier signal from the drive loop and the AM output. Finally, a low-pass filter
is used to extract the Coriolis signal from the mixer output. Figure 5.3 shows the configuration for
detecting the rotation response from the output signal at the sense electrode.
The sensitivity of the disk gyroscope in terms of output voltage, using readout circuitry in
Figure 5.3, can be expressed as
uout 4e0 hRq1 Ag QUr = R  m∆q 

= VP RF Av sin  (5.11)
Ωz d02  2 

CF
VGAIN
R2
Is (AM-IN)
δC
Vp – R1
–
+ Vout (AM-OUT)
+
Cpad
+
VREF
VREF1 –
VP
R3
R4 Rate
output
LPF
VREF2
Demodulator
Carrier signal
(from drive loop)
FIGURE 5.3 Readout circuitry used for measuring the rotation rate in BAW disk gyroscopes.
where RF and Av are, respectively, the gain of the TIA and the total gain from the cascaded voltage
amplifiers.
5.3.3 Resolution Analysis
Resolution, defined as the minimum detectable rotation rate, is a major performance parameter
in a gyroscope. The total resolution has two principal components: the Brownian noise, originat-
ing from the mechanical motion of the resonating structure, and the electronic noise, originating
from the interface circuitry. By equating the displacement caused by Brownian motion [21,22] to
the displacement induced by Coriolis acceleration, the mechanical resolution of BAW gyroscopes
operating in their elliptic modes is derived in Ref. [20] and is expressed as
1 4 kBT
Ω z (Brownian ) = (5.12)
2 mAg q1 w 0 MQ

where q1 is the drive amplitude, ω0, M, and Q are, respectively, the natural angular frequency, effective
mass, and mode-matched quality factor, kB is the Boltzmann constant, and T is the absolute tempera-
ture. The effective mass for disk gyroscopes operating in their BAW modes can be calculated by the
integral of the kinetic energy [20], normalized to the maximum displacement at the disk edge, Ur=R.
Furthermore, the electronic noise arising from the interface circuitry can be determined by assuming
that the output noise (Inoise) has a white spectrum near the resonant frequency. The electronic noise
of the interface circuitry at mode-matched operation in a disk is derived in Ref. [20] and is stated as
d02 I noise BW (5.13)

Ω z (Electronics) =
4VP q1QAg (e0 hRUr = R sin (m∆q / 2))

where Inoise is the input-referred current noise of the interface circuitry. To fully realize the poten-
tial of BAW gyroscopes, the design and performance parameters for a typical BAW gyroscope are
evaluated and summarized in Table 5.2.
5.3.4 Dynamic Range

Dynamic range, another important performance parameter in gyroscopes, refers to the range of
input rates over which the output is detectable. It is typically computed as the ratio between the
maximum input rotation rate (full-scale rate) that the sensor can tolerate and the system noise
floor. To ensure linear operation, the maximum input rate is considered to be reached when the
disk deforms one-tenth of the capacitive gap. Also, it is assumed that the electronic noise can be
reduced substantially employing advanced low-noise front-end circuitry, leaving Brownian noise
as the dominant noise source. To estimate the dynamic range, the maximum input rotation rate is
divided by the product of the Brownian noise and the square root of the device bandwidth. The
bandwidth of the device is replaced by (fresonance /2Qmode-matched) and the dynamic range for these
devices is stated as
q2 - maxw 01.5 M 0.1d0w 0 4pM

Dynamic Range = = (5.14)
4 kBTQ BW 4kBT

In Equation 5.14, d0, ω0, and M are, respectively, the initial gap, the angular frequency, and the
effective mass, kB is the Boltzmann constant, and T is the absolute temperature. These devices, due
TABLE 5.2
Analytical Estimation of Performance Parameters in BAW Solid Disk Gyros
(111) SCS Primary (100) SCS Secondary
Material/Mode Elliptic Mode (m = 2) Elliptic Mode (m = 3)
Disk diameter (μm) 800 800
Disk thickness (μm) 40 40
Electrode length (Le) (μm) 175 175
Capacitive gap (nm) 200 200
Drive amplitude (nm) 20 20
Effective quality factor (Q) 100,000 100,000
Operating frequency (MHz) 5.00 7.42
Ur=R 1.5105 0.9420
Effective mass coefficient 0.4413 0.3129

Polarization voltage (V) 10 10
Inoise (pA/√Hz) 2.5 2.5
Frequency bandwidth (Hz) 25 38
Sensitivity (aF/°/s) 4.76 1.01
MNEΩ (m°/s/√Hz) 0.79 1
ENEΩ (m°/s/√Hz) 1.7 5.3
Dynamic range (dB) 112 114
to high operating frequency and mode-matched operating, offer a wide dynamic range up to 114 dB,
as listed in Table 5.2, with a large bandwidth.
5.3.5 Thermoelastic Damping
Thermoelastic damping (TED) is a dominant loss mechanism in flexural-based gyroscopes, intro-
ducing highly temperature-dependent performance deterioration that circumscribes the thermal
stability of these sensitive devices [23–25]. Thus, designing micromachined gyroscopes such that
their Qs are not largely affected by TED is always desirable. To do this, the effect of TED is studied
for solid and perforated disk structures operating at their primary and secondary elliptic modes as
well as at their breathing modes.
As a starting point, it is noted that the total Q of a resonating structure [26] generally can be
expressed as
−1
 1 1 1 1  (5.15)
Q= + + + 
 QAir QTED QSupport QOther 
−1 −1 −1 −1
where QAir , QTED , QSupport , and QOther are, respectively, defined as air damping losses, TED losses,
the losses through support, and other losses not included in the preceding terms. The latter might
originate from, for example, surface roughness or intrinsic material energy dissipation. As one can
expect, each of the Qs on the right-hand side of Equation 5.15 has less impact upon the overall Q
when it is larger. This implies that the total Q would be less dominated by QTED if the QTED value
was very large compared to the rest of the Qs.
To determine the QTED for a disk operating in BAW modes, the coupled deformation and tem-
perature partial differential equations (PDEs) were estimated and solved for their eigenvalues [27].
Consequently, for each resonant mode, QTED can be evaluated as the ratio of the imaginary part to
the real part of the corresponding eigenvalue. The scaled thermoelastic equations were derived in
(a) (b)
7.871e–6 lambda(59)=1.519636e–4 to 4.642986e–13i Subdomain: T 5.184e–6
lambda(41)=9.954261e–5 to 2.514262e–12i Subdomain: T ×10–6
×10–6
Deformation: u,v,w Deformation: u,v,w 5
6 4
3
4
2
2 1
0
0
–1
–2 –2
–3
–4
–4
–6 –5
OTED = 37e6 OTED = 170e6
–6.788e–6 –5.169e–6
FIGURE 5.4 Steady-state vibration-induced temperature distribution in an 800-μm diameter solid disk for
(a) primary elliptic and (b) secondary elliptic modes, simulated in COMSOL.
Ref. [20] and were solved in COMSOL multiphysics engineering simulation software. The steady-
state temperature distributions and QTED for primary and secondary elliptic modes in an exemplary
800-µm diameter solid disk structure are shown in Figure 5.4.
As one can observe, QTED in these modes for an exemplary 800-μm diameter disk is very large
and it has the value in the range of low hundreds of millions. This implies that in a solid disk operat-
ing in its primary and secondary elliptic modes, QTEDs can respectively contribute only ~0.027 and
0.006 ppm on the total Q.
Furthermore, thermoelastic losses are modeled for a variety of disk geometries. Figure 5.5 shows
that decreasing the disk diameter lowers QTED in BAW modes. This clearly indicates that the total
Q becomes more reliant on QTED when the disk dimensions are reduced, driving the resonant fre-
quencies higher. Interestingly, the rate of the decrease in QTED for primary elliptic mode is less than
QTED for secondary elliptic modes. This may be simply due to the higher frequency of secondary
elliptic modes.
Next, symmetrical release holes are added to the disk, as dictated by fabrication needs (Section
5.4). Including perforations in the disk introduces localized dynamic strain concentrations, which
(a) (b)
10,000 1000
Primary elliptic mode Primary elliptic mode
Breathing mode
1000 Breathing mode
Secondary elliptic mode
Secondary elliptic mode 100
QTED (×106)
QTED (×106)
100
10
10
1 1
0.1 0.1
0.0001 0.001 0.01 0.1 1 10 0 5 10 15 20 25 30 35 40 45
1/R (μm) Release hole diameter (μm)
FIGURE 5.5 The estimated QTED simulated in COMSOL (a) for solid disk structures in the range of 0.5 MHz
to 2 GHz and (b) the effect of release holes’ sizes in QTED.
induce temperature gradients and result in additional TED losses. To study the effect of release hole
sizes on QTED, an exemplary 800-μm diameter disk with a variety of hole sizes were modeled in
COMSOL. In each model, the perforation sizes are uniform across the disk and equally spaced as
shown in Figure 5.5.
As seen, QTED drops substantially for perforated disks with large release holes. This establishes
that the optimization of QTED for perforated disks favors very small release holes (<8 μm). Also,
according to the simulation results, beyond a certain release hole size, the rate of decrease in
QTED becomes less pronounced, perhaps because the operating frequency also begins to decrease.
Also, it was observed that for disks without any perforations, QTED of secondary elliptic mode is
~4.5 times larger than the primary elliptic mode. However, when perforations are introduced, QTED
of the secondary elliptic mode reduces to that of the primary elliptic modes. Furthermore, due to the
common axisymmetry between the breathing mode and the release holes pattern, QTED in this mode
is much higher than in the elliptic modes with perforations present. Although breathing modes are
not suitable for disk gyroscopes, they could be utilized for disk resonators.
5.4 IMPLEMENTATION OF BAW GYROS

5.4.1 Implementation in (100) SCS
This chapter focuses on implementing BAW gyroscopes in (100) SCS substrates because silicon
resonant structures are significantly less susceptible to the energy losses compared with polysili-
con, resulting in higher quality factors (Q). In addition, there is no limitation on the thickness of
the structure as opposed to polysilicon structures, depending on the number of depositions. The
primary elliptic modes, which are 45° offset, have a large frequency split of (>1 MHz in 4.5 MHz)
between two degenerate modes in (100) SCS, an anisotropic material. This makes them unsuitable
for gyroscope applications. On the other hand, the secondary elliptic modes of a (100) silicon disk,
which are 30° offset, have nearly identical frequencies (Δf < 100 Hz in 6 MHz) as shown in Figure
5.6. This implies that for (100) SCS, the secondary elliptic modes should be employed to enable
matched-mode operation.
Implementation of a solid disk requires a backside etch step, presenting the need for both top-
side and backside packaging. To eliminate the latter, perforations were added to the disk, enabling
its release from the front side of SOI wafer. However, adding perforations to the disk introduces
frequency shift between the modes. Finite element analysis is used to model the introduced fre-
quency split as well as to guide the optimization of the release holes accordingly. To minimize the
effect of perforations on frequency split, some basic design rules have been investigated and are
discussed in the following section. First, the orientations of the perforations need to be compatible
with the antinodes and nodes of the vibration patterns to have the equal effects on both degenerative
modes. This results in minimizing the frequency split between two modes [20]. Second, perfora-
tions tend to distort the mode shapes. This effect can be mitigated by choosing the shape, size, and
configuration of the perforations appropriately. A variety of configurations such as radial grooves,
tangential grooves, and holes with different sizes were modeled in ANSYS. As a result, holes are
preferable over grooves with respect to lessening mode shape distortion and frequency separation.
Further optimizations addressed the holes’ distributions and sizes. To do this, two major categories
of equally spaced rings of release holes were developed. The first category includes nonuniform-
sized holes along the radial direction of the disk. This approach placed the holes exactly at every
30° and 15°, coincident with the nodal lines of mode shapes. The second category is comprised of
uniform small-sized holes, located mostly at the fraction of 30° offset. Both structure types are sym-
metric about the x- and y-axes.
The results for a variety of uniform hole sizes are shown in Figure 5.7. As expected, the disks
with uniform small release holes exhibit very small frequency split (Δf < 150 Hz). This is because
the frequency behavior of a disk with uniform small holes closely mirrors that of the solid disk.
(a)
1 1
STEP = 1 ANSYS STEP = 1 MX ANSYS
SUB = 3 SUB = 9
FREQ = .438E+07 FREQ = .546E+07
UX (AVG) UX (AVG)
RSYS = 1 RSYS = 1 Y
DMX = 6485 Y DMX = 6703 Z X
SMN = –6482 Z X MX SMN = –6702
SMX = 6483 SMX = 6698
MN
MN
–6482 –3601 –720 .176 2161 5042 –6702 –3724 –746.174 2232 5209
–5042 –2161 720.404 3602 6483 –5213 –2235 742.668 3720 6698
(b)
1 1 MX
STEP = 1 ANSYS STEP = 1 ANSYS
SUB = 4 SUB = 5
FREQ = .752E+07 FREQ = .752E+07
UX (AVG) UX (AVG)
RSYS = 1 Y MX RSYS = 1
DMX = 8542 Z X DMX = 8540 Y
SMN = –8443 Z X
SMN = –8442 MN
SMX = 8445 SMX = 8443
MN
–8442 –4690 –93.8956 2816 6569 –8443 –4691 –938.055 2815 6567
–6566 –2813 939.478 4692 8445 –6567 –2814 938.245 4691 8443
FIGURE 5.6 ANSYS simulation results of an 800 μm diameter solid disk gyroscope in (100) silicon.
(a) Primary elliptic modes with Δf = 1.08 MHz and (b) secondary elliptic modes with Δf = 70 Hz.
600 7.60
500 7.50
400 7.40
Frequency (MHz)
Δf (Hz)
300 7.30
Freq. split
Freq.
200 7.20
100 7.10
0 7.00
0 4 6 8 10 18
Release hole diameter (μm)
FIGURE 5.7 The effect of release hole sizes on the frequency split (Δf) of two modes and operating
frequencies.
Release holes VDC through poly trace
FIGURE 5.8 The process flow of a center-supported silicon disk gyro on SOI wafer.
If an isotropic material (such as [111] SCS or polysilicon) is used, the two primary elliptic degen-
erative modes, spatially 45° apart, have identical resonance frequencies (Δf = 70 Hz). The same
perforation design method employed for (100) SCS disks can be applied to (111) SCS disks [20].
However, in (111) SCS substrates, release holes need to be placed and repeated symmetrically every
45°, instead of every 30° as required in (100) SCS substrates.
5.4.2 Fabrication Method

The BAW gyroscopes should be fabricated in a high-aspect-ratio process. Owing to the high fre-
quency of operation, large electrostatic forces should be applied to enable excitation of the device.
As a result, narrow gaps (in submicrometer) are necessary for these devices. The signal-to-noise
ratio can be enhanced largely if thick structures are employed. Thus, thick SOI substrates using the
HARPSS process [21,28,29] are used to implement these devices. The cross-section of fabricated
BAW gyros is shown in Figure 5.8 and is outlined thoroughly in Ref. [12].
It is worth noting that the center post diameter is approximately one-twentieth of the disk diam-
eter. The small center post size minimizes losses through support [20] and enhances the Q of the
device.
5.5 MEASUREMENTS OF BAW GYROS

A number of fabricated silicon disk gyroscopes were fabricated and characterized on a variety of
thick SOI substrates (30–60 µm). The preliminary measurement results of 40-μm-thick disk gyro-
scopes implemented in (100) SCS substrates are presented here. The MEMS gyroscope chip was
mounted on a printed circuit board (PCB) including a drive loop and sense amplifiers [9,20]. All
coupled drive, sense, and tuning electrodes as well as polarization voltage (VP) pads are connected
to the board through wire bonds. The board is placed inside a vacuum chamber, and coaxial cables
are utilized to apply signals to the device. The test setup used here is similar to the one reported in
Refs. [7,20].
The scanning electron microscope (SEM) view of a fabricated and exemplary 800-μm diameter
disk gyroscope with nonuniform perforations is shown in Figure 5.9. The close-up view of the elec-
trode area shows the capacitive gap of 200 nm between the electrode and the disk in 40-μm-thick
SOI, providing a gap aspect ratio of 200.
5.5.1 Frequency Characterization and Mode Matching

The secondary elliptic modes of this prototype were observed at 5.95 MHz with a frequency split
of 450 Hz. This is in good agreement with ANSYS simulation results, as shown in Figure 5.10. The
Qs were measured to be 200,560 and 207,660 for the drive and sense modes in 1 mTorr vacuum,
Release Poly-silicon 200-nm capacitive gap

holes trace
(100) SCS
disk
40-µm thick
poly-silicon
SE 03-Mar-08 WD32.2 mm 5.00 kV x100 500 um
electrode
WD22.5 mm 5.00 kV x1.8 k 25 um
FIGURE 5.9 SEM view of an 800-μm diameter silicon disk gyro with a 200-nm gap and a 40-μm-thick disk.
Ach log MAG 3 dB/ REF –12dB –4.6831 dB Ach log MAG 3 dB/Reg –12 dB –4.7719 dB
Step = 1
Sub = 4
5.949167 MHz Step = 1 5.949624 MHz
Freq. = .612E+07 Sub = 5
UX
RSYS = 1
(AVG) TrkΔ Freq. = .613E+07
UX (AVG) TrkΔ
RSYS = 1
DKX = 8934
SMN = –8913 BW: 29.663 Hz DMX = 8945
SKN = –8895
BW: 28.651 Hz
SMX = 8888
0Δ Cent: 5.9491679 MHz SMX = 8923
0Δ Cent: 5.9496243 MHz
Q: 200.56 k Q: 207.66 k
–8913 –4957 –1001 2955 6910 2 3 Loss: –4.6831 dB 2 3 Loss: –4.7719 dB
–6935 –2979 976.709 4932 8888 –8895 –4935 –975.682 2984 6943
ΔL.F: –16.923 Hz –6915 –2955 1004 4964 8923

ΔL.F: –19.98 Hz
ΔR.F: 12.741 Hz ΔR.F: 8.671 Hz
Freq. = 5.95 MHz

Freq. = 5.95 MHz
QDrive = 200,560
QSense = 207,660
BW = 15 Hz
BW = 15 Hz
In 1 mTorr vacuum
In 1 mTorr vacuum
IF BW 10 Hz Power 2 dBm SWP 47.28 s IF BW 10 Hz Power 2 dBm SWP 47.28 s
Center 5.94917 MHz Span 200 Hz Center 5.94963 MHz Span 200 Hz
FIGURE 5.10 The measured frequency response and ANSYS simulation results for an 800-μm diameter
silicon disk gyroscope operating in its secondary elliptic modes, having the resonance frequency of 5.95 MHz
with Δf = 450 Hz at VP = 6 V in 1 mTorr vacuum.
respectively. The corresponding Q values at 1 Torr vacuum were still very high and recorded to be
185,280 and 192,060, respectively.
The as-fabricated frequency split of 450 Hz (~75 ppm compared to the operating frequency
of 5.95 MHz) was electrostatically nulled to improve the device performance (Figure 5.11). The
quadrature error was also minimized by adjusting the appropriate quadrature voltages. The mode-
matched Qs of 180,000 and 235,810 are measured, respectively, before and after quadrature signal
minimization.
5.5.2 Performance Characterization
The scale factor is measured only from one sense electrode both at split modes and at matched
mode. The rate sensitivity in mode-matched condition after quadrature cancellation (Figure 5.11c)
was also measured. Figure 5.12 shows the measured rate sensitivity of 267 μV/°/s with discrete elec-
tronics at VP of 13.2 V and mode-matched Q of 235,810 in 1 mTorr vacuum.
5.5.3 Quality Factor Characterization

The temperature-induced frequency drift of a similar prototype was measured using a tempera-
ture-controlled chamber. As shown in Figure 5.13a, the measured frequency drifts for both drive
and sense modes have a linear trend with a slope of −26 ppm/°C over the temperature range −5°C
Bulk Acoustic Wave Gyroscopes
(a) (b) (c)

Ach log MAG 5 dB/ REF –25 dB –40.92 dE Ach log MAG 5 dB/ REF –5 dB 8.689 dB A/R log MAG 3.5 dB/ REF –10 dB 2.9769 dB
5.947025 MHz 5.94827675 MHz
1 2 5.94816025 MHz
TrkΔ 0 1 TrkΔ
BW: 86.867 Hz BW: 25.224 Hz
0 Cent: 5.9482471 MHz 2 3
Cent: 5.9481589 MHz
2 1 3 Q: 68.475 k Q: 235.81 k
Loss: 8.689 dB
Loss: 2.9769 dB
ΔL.F: –93.011 Hz
ΔL.F: –8.916 Hz
ΔR.F: –6.144 Hz
ΔR.F: 16.309 Hz
Δf = 450 Hz Δf = 60 Hz Δf = 0 Hz
IF BW 10 Hz Power 2 dBm SWP 47.26 s IF BW 100 Hz Power –10 dBm SWP 5.93 s IF BW 10 Hz Power –3 dBm SWP 94.69 s
Center 5.949525 MHz Span 5 kHz Center 5.94829675 MHz 1 kHz Center 5.94815525 MHz Span 1 kHz
FIGURE 5.11 Collection of plots showing electrostatic frequency tuning and mode-matched operation in BAW disk gyroscope in (100) SCS.
105
25
20
15
10
5
Vout (mV)
0
–100 –75 –50 –25 –5 0 25 50 75 100
–10
–15
–20
–25
Rotation rate (°/s)
FIGURE 5.12 The measured rate sensitivity results from silicon disk gyroscope at mode-matched operation
after quadrature cancellation with VP = 13 V and Q = 236,000, using discrete electronics.
(a) (b)
5.898 175
~ 4% reduction in Q over ΔT = 60°C
5.896 170
5.894 165
Frequency (MHz)
5.892 160
155
5.890
Q (K)
Slope = –26.2 ppm/°C 150

5.888
145
5.886
140
5.884 Sense mode (1 mTorr)
Drive mode 135 Drive mode (1 mTorr)
5.882 Sense mode Sense mode (2 Torr)
130
5.880 Drive mode (2 Torr)
–5 10 25 40 55 125
–5 10 25 40 55
Temperature (°C) Temperature (°C)
FIGURE 5.13 Measured temperature characteristics of the frequency and Q in BAW gyros, showing a linear
frequency drift profile with a slope of −26 ppm/°C. The Q dropped ~4% for both drive and sense modes at
ΔT = 60°C. An optimized design characterization.
to 55°C. The frequency drift is mainly due to the temperature dependence of Young’s modulus of
silicon, resulting in both degenerate modes tracking each other with the same slope. This implies
the stability of the frequency separation between the modes over a temperature range 60°C. Also,
the Qs of both modes were characterized over the same temperature range both in 1 mTorr and in
2 Torr vacuum pressure and are shown in Figure 5.13. The Qs were dropped ~4% over the 60°C
temperature range. This indicates that the Qs in BAW modes are not mainly limited by the TED
mechanism in contrast with low-frequency gyros [3,30], resulting in performance stability over
temperature.
The design of the perforations in the disk gyroscopes was optimized to significantly reduce
the frequency split between the two modes. An SEM view of a fabricated 800-μm diameter (100)
SCS disk gyroscope including uniform 6-μm diameter holes on a 40-μm-thick SOI along with the
ANSYS simulation results is shown in Figure 5.14.
The frequency response of this prototype was measured and shown in Figure 5.15 and is in
very good agreement with the ANSYS simulation results (with a difference of 2.8%). This very
small difference can be readily attributed to slight geometric differences between the fabricated
and designed structures. The secondary elliptic modes were observed to be matched at 7.30 MHz
in 1 mTorr vacuum and air (with very small frequency separation of less than 100 Hz). The
(a)
Symmetrical
uniform 6-μm
release holes
Poly-silicon trace
30 μm X_GAO Scan speed = 6 Width = 1.030 mm 111 X WD = 36.7 mm
16 Aug 2008 18:13:31 Frame Int Busy N = 150 Stage at T = 36 6° 500 kV Stage A = SE2
(b)
1 ANSYS 1 ANSYS
Nodal solution Nodal solution
STEP = 1 STEP = 1
SUB = 5 SUB = 6
FREQ = .750E+07 FREQ = .750E+07
UX (AVG) UX (AVG)
RSYS = 1 z x RSYS = 1 zx
DMX = 8701 DMX = 8703
SMN = –8655 SMN = –8655
SMX = 8655 SMX = 8658
–8655 –4808 –961.444 2885 6732 –8655 –4808 –960 .269 2887 6734
–6731 –2885 961.901 4809 8655 –6731 –2884 963.422 4811 8658
FIGURE 5.14 (a) SEM view of a disk gyro with uniform 6-μm diameter holes in a 40-μm-thick silicon disk
and (b) ANSYS simulation results.
mode-matched Qs of 73,000 and 14,435, respectively, were measured in 1 mTorr vacuum and air
at VP = 7 V.
The rotation rate sensitivity was measured only from one sense electrode in air, as shown in
Figure 5.15c. The measured rate sensitivity of 64.5 μV/°/s is recorded with discrete electronics and
VP of 12.5 V, with Q at a mode-matched Q of 14,435. The measurement results in air confirms that
the large stiffness of BAW gyroscopes makes them less sensitive to environmental pressure, obvi-
ating the need for high-vacuum encapsulation. This simplifies wafer-scale packaging for reduced
manufacturing cost in these devices.
5.6 SUMMARY
An overview of high-frequency BAW gyroscopes is presented in this chapter. The concept of the
BAW gyroscopes is introduced and the principle of operation for these devices is discussed. The
fundamental parameters of the BAW gyroscopes design are addressed. The degenerate in-plane
elliptic modes are employed for the operation of these devices and the normal bulk mode model
is derived using the Lagrange method. The performance parameters including sensitivity, resolu-
tion, and dynamic range are analyzed and explained. The implementation methods for BAW gyros
are presented. Finally, the preliminary characterization and measurement of these devices are pre-
sented, followed by the environmental effects on the performance of the devices.
108
(a) (b) (c)

A/R log MAG 2 dB/REF –16 dB –10.086 dB A/R log MAG 5 dB/REF –50 dB –34.735 dB
7.300165 MHz 7.3001 MHz
TrkΔ TrkΔ 6
0Δ BW: 99.822 Hz 0Δ BW: 505.711 Hz 5
Cent: 7.3001619 MHz Cent: 7.3001578 MHz
2 3 Rate sensitivity 4
Q: 72.881 k Q: 14.435 k
Loss: –10.086 dB Loss: –34.735 dB
64.5 µV/°/s 3
2 3
ΔL.F: –62.975 Hz ΔL.F: –195.015 Hz In air 2
ΔR.F: 36.847 Hz ΔR.F: 310.896 Hz 1
Vout (mV)
0
–90 –70 –50 –30 –10 10 30 50 70 90
MEMS: Fundamental Technology and Applications

–1
–2
–3
Freq. = 7.30 MHz Freq. = 7.30 MHz
–4
Qeffective = 72,881 Qeffective = 14,345 –5
BW = 50 Hz BW = 505 Hz –6
In 1 mTorr vaccum In air Rotation rate (°/s)
IF BW 10 Hz Power –2dBm SWP 94.69 s IF BW 30 Hz Power 0 dBm SWP 23.54 s

Center 7.300175 MHz Span 2 kHz Center 7.3001 MHz Span 10 kHz
FIGURE 5.15 The measured response in BAW gyro with uniform 6-μm diameter holes, showing the mode-matched operation without tuning and Q of (a) 73 kHz in
vacuum, (b) 14.4 k in air, and (c) the rate results in air.
ACKNOWLEDGMENTS
This research was performed at Georgia Institute of Technology in IMEMS group from 2004 to
2008. The author would like to thank several individuals who provided assistance in the course of this
research: Dr. Farrokh Ayazi for his valuable mentoring; Dr. Siavash Pourkamali, Dr. Reza Abdolvand,
Dr. Ajit Sharma, and Dr. Mohammad Faisal Zaman for their valuable discussion; and Mrs Jalpa Shah
and Mr. Milap Dalal for their assistance in characterizations. Also, the author would like to thank the
staff of Microelectronics Research Center (MiRC) at Georgia Tech for their assistance.
REFERENCES
1. W. A. Clark, R. T. Howe, and R. Horowitz, Surface micromachined z-axis vibratory rate gyroscope, In
Proceedings of Solid-State Sensors and Actuators Workshop, Hilton Head Island, SC, June 1996, pp. 283–287.
2. H. Xie and G. Fedder, A DRIE CMOS-MEMS gyroscope, In Proceedings of the IEEE Sensors’04,
Vienna, Austria, October 2004, pp. 1413–1416.

3. M. F. Zaman, A. Sharma, and F. Ayazi, High performance matched-mode tuning fork gyroscope, In
Proceedings of IEEE/ASME Conference on Microelectromechanical System (MEMS’06), Istanbul,
Turkey, January 2006, pp. 66–69.
4. F. Ayazi and K. Najafi, A HARPSS polysilicon vibrating ring gyroscope, IEEE Journal of
Microelectromechanical Systems, 10(2), 161–179, 2001.
5. M. F. Zaman, A. Sharma, B. V. Amini, and F. Ayazi, The resonating star gyroscope, In Proceedings of
the IEEE/ASME Conference on Microelectromechanical System (MEMS’05), Miami, FL, January 2005,
pp. 355–358.
6. A. Sharma, M. F. Zaman, M. Zurcher, and F. Ayazi, A 0.1°/hr bias drift electrically matched tuning fork
microgyroscopes, IEEE/ASME Conference on Microelectromechanical System (MEMS’08), Tucson, AR,
January 2008, pp. 6–9.
7. M. F. Zaman, Degree-per-hour mode-matched micromachined silicon vibratory gyroscopes, PhD
Dissertation, Electrical and Computer Engineering Department, Georgia Institute of Technology, April 2008.
8. H. Johari and F. Ayazi, Capacitive bulk acoustic wave silicon disk gyroscopes, In Technology Digest of
International Electron Device Meeting (IEDM), San Francisco, CA, December 2006, pp. 513–516.
9. J. Shah, H. Johari, A. Sharma, and F. Ayazi, CMOS ASIC for MHz silicon BAW gyroscope, In Proceedings
of International Symposium on Circuits and Systems (ISCAS), Seattle, WA, May 2008, pp. 2458–2461.
10. M. E. McNie et al. Design and fabrication of a micromachined silicon ring gyroscope, In Proceedings
of the 12th European Conference on Solid-State Tranducers and the 9th UK Conference on Sensors and
Their Applications, Southampton, UK, Vol. 1, 1998, pp. 369–372.
11. M. W. Putty and K. Najafi, A micromachined vibrating ring gyroscope, In Technology Digest of Solid-
State Sensor and Actuator Workshop, Hilton Head Island, SC, June 1994, pp. 213–220.
12. H. Johari and F. Ayazi, High frequency capacitive disk gyroscopes in (100) and (111) silicon, Proceedings
of the 20th IEEE International Conference on Micro Electro Mechanical Systems (MEMS 2007), Kobe,
Japan, January 2007, pp. 47–50.
13. J. Kim et al. Why is (111) silicon a better mechanical material for MEMS? In Technology Digest of the
International Conference on Solid-State Sensors and Actuators (Transducers’01), Munich, Germany,
June 2001, pp. 662–665.
14. H. Guohong and K. Najafi, A single-crystal silicon vibrating ring gyroscope, In Proceedings of IEEE
Conference on Microelectromechanical System (MEMS’02), Los Vegas, January 2002, pp. 718–721.
15. M. W. Putty, A micromachined vibrating ring gyroscope, PhD Dissertation, University of Michigan,
December 2000.
16. Z. Hao, S. Pourkamali, and F. Ayazi, VHF single crystal silicon elliptic bulk-mode capacitive disk resonators;
Part I: Design and modeling, IEEE Journal of Microelectromechanical Systems, 13(6), 1043–1053, 2004.
17. W. T. Thompson, Theory of Vibration with Applications, Prentice-Hall, Englewood Cliffs, NJ, 1998.
18. S. S. Rao, Mechanical Vibration, 4th edition, Pearson Prentice-Hall, Upper Saddle River, NJ, 2004.
19. M. Onoe, Contour vibrations of isotropic circular plates, Journal of Acoustic Society American, 28(6),
1158–1162, 1956.
20. H. Johari, Micromachined capacitive silicon bulk acoustic wave gyroscopes, PhD Dissertation,
Mechanical Engineering Department, Georgia Institute of Technology, December 2008.
21. F. Ayazi, A high aspect ratio high-performance polysilicon vibrating gyroscopes, PhD Thesis, University
of Michigan, 2000.
22. R. A. Johnson, Mechanical Filters in Electronics, Wiley, New York, 1983.

23. Z. Hao and F. Ayazi, Thermoelastic damping in flexural-mode ring gyroscopes, In ASME International
Mechanical Engineering Congress and Exposition (IMECE 2005), Orlando, FL, November 2005.
24. A. Duwel, J. Gorman, M. Weinstein, J. Borenstein, and P. Ward, Quality factors of MEMS gyros and the
role of thermoelastic damping, In Proceedings of IEEE/ASME Conference on Microelectromechanical
Systems (MEMS’02), Los Vagas, NV, January 2002, pp. 214–19.
25. A. Duwel, J. Gorman, M. Weinstein, J. Borenstein, and P. Ward, Experimental study of thermoelastic
damping in MEMS gyros, Sensors and Actuators A, A103(1–2), 70–75, 2003.
26. J. Yang, T. Ono, and M. Esashi, Energy dissipation in submicrometer thick single-crystal silicon cantile-
vers, Journal of Microelectromechanical Systems, 11(6), 775–783, 2002.
27. W. Nowacki, Dynamic Problems of Thermoelasticity, Noordhoof International, Leyden, 1975.
28. F. Ayazi and K. Najafi, High aspect-ratio combined poly and single-crystal silicon (HARPSS) MEMS
technology, IEEE Journal of Microelectromechanical Systems, 93, 288–294, 2000.
29. S. Pourkamali, Z. Hao, and F. Ayazi, VHF single crystal silicon elliptic bulk-mode capacitive disk reso-
nators. Part II: Implementation and characterization, IEEE Journal of Microelectromechanical Systems,
13(6), 1054–1062, 2004.

30. M. F. Zaman, A. Sharma, Z. Hao, and F. Ayazi, A mode-matched silicon yaw tuning fork gyroscope
with sub-deg/hr Allan deviation bias instability, IEEE Journal of Microelectromechanical Systems, 17(6),
1526–1536, 2008.

Bulk Acoustic Wave Gyroscopes: Houri Johari

Uploaded by

Copyright:

Available Formats

You might also like

Bulk Acoustic Wave Gyroscopes: Houri Johari

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Bulk Acoustic Wave Gyroscopes: Houri Johari

Uploaded by

Copyright:

Available Formats

5 Bulk Acoustic Wave

5.2 PRINCIPLE OF OPERATION

5.3 DESIGN OF BAW GYROS

fr1 = Ur cos(mq ), fq1 = Uq sin(mq )

M1q1 + D1q1 − 2g Ω z q2 − g Ω  q − M Ω2 q + ( K − K )q = F

5.3.1 Angular Gain Estimation

Coriolis coupling coefficient (γ, kg) 1.8954e − 8 1.0175e − 008

dCs 4 me0 hRq1 Ag QUr = R  ∆q 

uout 4e0 hRq1 Ag QUr = R  m∆q 

d02 I noise BW (5.13)

5.3.4 Dynamic Range

q2 - maxw 01.5 M 0.1d0w 0 4pM

Effective mass coefficient 0.4413 0.3129

5.4 IMPLEMENTATION OF BAW GYROS

Release holes VDC through poly trace

5.4.2 Fabrication Method

5.5 MEASUREMENTS OF BAW GYROS

5.5.1 Frequency Characterization and Mode Matching

Release Poly-silicon 200-nm capacitive gap

ΔL.F: –16.923 Hz –6915 –2955 1004 4964 8923

Freq. = 5.95 MHz

5.5.3 Quality Factor Characterization

(a) (b) (c)

Slope = –26.2 ppm/°C 150

(a) (b) (c)

MEMS: Fundamental Technology and Applications

IF BW 10 Hz Power –2dBm SWP 94.69 s IF BW 30 Hz Power 0 dBm SWP 23.54 s

Vienna, Austria, October 2004, pp. 1413–1416.

22. R. A. Johnson, Mechanical Filters in Electronics, Wiley, New York, 1983.

13(6), 1054–1062, 2004.

You might also like

M1q1 + D1q1 − 2g Ω z q2 − g Ω q − M Ω2 q + ( K − K )q = F