Structural design and analysis of micromachined ring-type
vibrating sensor of both yaw rate and linear acceleration
Jui-Hong Weng
a,1, Wei-Hua Chieng
a,∗, Jenn-Min Lai
b,2aDepartment of Mechanical Engineering, National Chiao Tung University, 1001 Ta Hsueh Rd, Hsinchu Taiwan, ROC bR&D Center, NeoStones MicroFabrication Corporation, 10F, No. 1, Li-Hsin 5th Road, SBIP, Hsin-chu, Taiwan, ROC
Received 10 March 2004; received in revised form 1 June 2004; accepted 17 June 2004 Available online 27 July 2004
Abstract
This work presents the design and analysis of a micromachined ring-type motion sensor which can sense both the yaw rate and linear acceleration of a device in that it is installed. It integrates the functions of a gyroscope and an accelerometer. The design of the microstructure is based on the structure of a vibrating ring gyroscope. The analysis, considering the symmetry of the structure, implies that the rotational symmetry and mirror symmetry of the microstructure markedly affect the performance of the sensor. The modes of vibration determined using finite element analysis and the theoretical calculation of the resonant frequency of the structure are presented. The sensitivities of the characteristic of the sensor to the design specification, the fabrication errors and the theoretical errors are discussed. The structure is fabricated based on deep silicon etching and wafer bonding techniques. The process is developed for eliminating the aspect ratio dependent effect (ARDE) in dry etching process.
© 2004 Elsevier B.V. All rights reserved.
Keywords: Gyroscope; Accelerometer; Symmetry; MEMS
1. Introduction
Sensors for detecting motion have received considerable interest in recent decades. The gyroscope and the accelerom-eter represent two important categories of inertial technol-ogy for detecting motion. Sensor technoltechnol-ogy has evolved form discrete, expensive and inflexible units to smart, self-calibrating, silicon-based devices, which are integrated, low-cost and small, and exhibits other advantageous properties
[1]. The rapid development of semiconductor technology has enabled micro electromechanical systems (MEMS), includ-ing micromachined gyroscopes and accelerometers, to keep up with the requirements of modern sensors. The gyroscope is
∗Corresponding author. Tel.: +886 3 5712121x55152;
fax: +886 3 5720634.
E-mail address: whc@cc.nctu.edu.tw (W.-H. Chieng). 1 Tel: +886 3 5712121x55170, fax: +886 3 5731754. E-mail address:juihongweng.me87g@nctu.edu.tw.
2 Tel: +886 3 5712121x55170, fax: +886 3 5731754. E-mail address:a1005@neostones.com.tw.
used in many applications, including guidance systems, nav-igation systems and control systems. Specific applications of the gyroscope include tactical weapon guidance, ship and airplane navigation, craft attitude control, motor sensors and commercial electronics.
Many important ideas related to gyroscope have been successfully developed. The conventional gimbaled spinning wheel or rotor is the most well known gyroscope. The laser ring gyro is also extensively applied and has been proven to be successful in inertial navigation applications[2]. Al-though highly accurate, these devices are too expensive for low-cost applications and too large to meet requirements of minimal size. The vibrating gyroscope is another successful class of gyro designs. It uses vibrating mechanical elements to sense rotation, and so requires no spinning or lubricated parts. In the 1960s, engineers tried to use vibrating elements, rather than rotating bodies, to sense the Coriolis force. Most gyroscope engineers developed vibrating gyroscopes based on vibrating strings and tuning forks. The use of vibrating gyroscopes not only keeps down the cost of devices but also
0924-4247/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.sna.2004.06.016
Fig. 1. Top and detailed views of structure that includes electrodes.
maintains performance because they can be manufactured precisely. The vibrating string gyro, the tuning fork gyro and the vibrating shell gyro are known as vibratory rate sensors or rotation sensors[2]. All can be implemented with micro-machining technology, and many designs have been carried out[3–18]. The hemispherical resonator gyroscope (HRG)
[19], a vibrating shell gyro, has achieved inertial navigation performance levels that rival those of the laser ring gyro. The vibrating ring-type gyro, which is a simplified case of the vibrating shell gyro, is the subject of this work. The specific structural design of such gyro makes it particularly candidates for integration into accelerometers.
The accelerometer has been used in many engineering ap-plications over several decades and many successful designs exist. Since Roylance developed the piezo-resistive silicon
accelerometer [20] in 1977, the silicon-based
accelerom-eter has become a mainstream detector of acceleration, replacing the convectional pendulous accelerometer. Such silicon-based devices can be classified into piezo-resistive, capacitive, piezo-electric, junction-effected, thermal,
elec-Fig. 2. Schematic overview of control loops and structure.
tromagnetic and electron tunneling accelerometers[21–26]. Its structural design and simplicity of fabrication make the capacitive accelerometer a popular type of accelerometer. Furthermore, the basic scheme of the capacitive accelerom-eter exploits electrostatic force, which can drive a vibrating microstructure.
Notably, the vibration of a circular ring can be theoret-ically decoupled into several modes. The first mode stands for the linear motion, and the second involves rotation. Two independent vibration modes exist concurrently. Therefore, the capacitive accelerometer and the vibrating ring gyroscope can be integrated in a motion sensor that detects both accel-eration and rotation.
2. Sensor structure and control loops
The sensor design comprises the structure and the control loops.Fig. 1presents the top and detailed views of the struc-ture of the sensor. The main ring in the center is the main
body designed to vibrate in the flexural mode.Fig. 2 indi-cates how the main ring is connected to eight support rings that surround it. The similarity between the shape of the main ring and that of the support rings simplifies the analysis. The main ring and the support rings are designed to vibrate in the plane, as depicted inFig. 7. Electrodes are located at the centers of the support rings, as shown inFig. 1, forming ca-pacitors and driving the structure. The potential difference between the support ring and the electrode generates an elec-trostatic force that attracts them toward each other, and drives the structure to vibrate. The electrodes are shaped to maxi-mize effective displacement when the structure vibrates in the desired modes of vibration.
Fig. 2schematically depicts the control loops that are used to manipulate the sensor. The main driving loop causes the structure to vibrate as desired. The vibration is enhanced by phase-locked loops, and the structure vibrates at the resonant frequency to yield large amplitudes[27]. A rebalancing loop is designed to rebalance the vibration induced by the Corio-lis force. Two rebalancing loops are designed to balance the vibration induced by linear motion. Three rebalancing loops cooperate to balance the planar motion.
3. Governing equation of the vibrating ring
Fig. 3depicts the coordinate system of the motion sensor, including the inertial coordinate, the translating and rotating coordinate and the sensor coordinate. The sensor translates, without rotation, relative to a translating and rotating frame, which moves relative to a fixed inertial coordinate, where
→
r1specifies the position vector from inertial coordinate to
the translating and rotating coordinate and→r2specifies the
position vector from the frame to the sensor. The sensor is taken as a vibrating ring with a radius of r3. The vibration of
the main ring is considered to include radial and tangential displacements, denoted as u andv, respectively. Assume that the elongation of the central line of the ring remains constant in the flexural mode. The radial and tangential displacements
Fig. 3. Coordinate system of the sensor.
are,
u =(an cosnθ + bn sin nθ)
v = −an
n sinnθ + bnn cosnθ
, (1)
where a and b are functions of time, and n denotes the number of modes[28]. These terms, cos nθ and sin nθ, are orthogo-nal on the interval 0≤ θ ≤ 2π. The designed structure is a complete ring, such that one time function is independent of the others.
To simplify the derivation, r1is omitted and the position
of the ring is expressed as,
X =→r2+→r3+→u +→v , (2)
and the velocity V of the sensor is expressed as,
V = d
dt(X) + ΩX, (3)
whereΩ is the rate of rotation of the frame. The kinetic energy and potential energy of the ring are,
T = 1 2ρtwr3 2π 0 V · V dθ U = EI 2r3 2π 0 ∂2u ∂θ2 + u 2 dθ (4)
whereρ is ring’s density; t the ring’s thickness; w the ring’s width; E the Young’s modulus of the material, and I its mo-ment of inertia. Lagrange’s equation, taking a2as a
general-ized coordinates yields the differential equation of motion, 5 8m¨a2− mΩz ˙ b2−1 2m ˙Ωzb2 +9πEI r3 3 a2+ 5 16ma2Ω 2 x− 5 16ma2Ω 2 y− 5 8ma2Ω 2 z + 3 8mr3Ω 2 x− 3 8mr3Ω 2 y = fa2. (5)
This equation governs the second mode of flexural vibra-tion of the ring, as depicted inFig. 4. The detection of rota-tion depends on the second mode. This expression involves the centrifugal force associated with a2and r3. In this work,
these terms are made relatively small by designing the struc-ture with a high effective stiffness, which affects the flexural vibration. Besides, those terms associated with r3will not
ap-pear in the higher order vibrations, such as the third mode. A higher order vibration may be a solution when a low stiffness and a large radius are required.
Fig. 5(a) indicates the second mode of vibrations of the ring, including the primary mode and secondary mode. The modes are mutually independent, and spatially orthogonal.
Fig. 5(b) shows the ring that vibrates along the 0◦axis. The ring is simply divided into four sectors, and there is an equiv-alent velocity representing the velocity of each sector. When the ring is rotated, the Coriolis force is induced and the re-sultant force lies along the 45◦axis, as indicated inFig. 5(c). Hence, the Coriolis force generates the secondary mode of vibration.
Fig. 4. Motion of the ring, comprising first and second modes of vibration.
ConsideringEq. (4), using Lagrange’s equation, and tak-ing a1 as the generalized coordinate, yield the differential
equation of motion is expressed as,
m¨a1− (2mΩzb1˙ + mb1Ω˙z)+ (mΩxΩyb1− ma1Ω2z −ma1Ω2
y)+ m¨x − (2mΩzy + my ˙Ω˙ z− 2mΩy˙z − mz ˙Ωy)
+(mzΩxΩz+ myΩxΩy− mxΩy2− mxΩy2)= fa1, (6)
where x, y and z are relative position vectors of the sensor in the translating and rotating frame, and a1and b1represent the
rigid body motion of the ring. This equation governs the first mode of flexural vibration of the ring, as depicted inFig. 4. The detection of linear acceleration depends on the first mode. In the first row ofEq. (6), the centrifugal terms are smaller than other terms on micro scale. The last two rows inEq. (6)
represent the rigid body motion of the sensor. They can be shifted to the right of the equation and considered as part of the applied force, such that the applied force is a function of position. The force is comparable to other terms in the governing equation and may pull the structure far away from its linear region. Hence, the rebalancing loop must be used.
Fig. 5. (a) Second mode of vibration and its generalized coordinates; (b) primary mode on 0◦axis; and (c) the secondary mode and resultant Coriolis force on 45◦axis.
4. Symmetry of structure
The structural design of the motion sensor is determined with reference to both rotation and acceleration. The structure consists of the main ring and the supports. The influence of the symmetry on the support is discussed.
The first consideration in designing the support is the function of a gyroscope. The main ring surrounded by four asymmetric supports, as shown inFig. 6(a), is the simplest arrangement for the gyroscopic function. Those supports are placed in directions a2and b2, such that the structure is
equally stiff in these directions. However, two of the second modes of vibration of the structure are not exactly at 45◦to each other and their natural frequencies are different. The structure exhibits planar motion when it vibrates, causing extra energy to be lost. The deformation of the support is not uniform, generating noise that affects detection. Thus, the supports should be placed as presented inFig. 6(b), with
rotational symmetry denoted by C8[29]. This arrangement
yields a balanced device with two modes that have the same natural frequency[30]at 45◦to each other.
The centrifugal force and Coriolis force are generally ap-plied simultaneously. The performance of a gyroscope under
Fig. 6. Arrangements of supports: (a) four supports; (b) eight asymmetric supports; and (c) eight symmetric supports.
acceleration must be considered. When those structures il-lustrated in Fig. 6(a) and (b) are exposed to accelerations, main rings become deformed. The acceleration induces non-collinear displacement, which generates detected noise. The asymmetry of the supports is responsible for these defects. A feedback control loop may be used to rebalance those phe-nomena, but the design of control loops is complicated. For-tunately, those effects are not very serious on the micro scale, yet they cause undesired noise. Besides, asymmetric supports induce much more stress on the connection between the main ring and the supports.
A higher degree of symmetry of the structure presents bet-ter performance.Fig. 6(c) depicts the structure that contains symmetric supports. The structure is both mirrors symmet-ric and rotationally symmetsymmet-ric, and the degree of symmetry is denoted as D4. The shape of the support is the same as
that of the main ring, simplifying the estimation of the
dy-namic response. As expected, two modes are 45◦ to each
other as shown inFig. 7(a) and (b). This structure exhibits a collinear force–displacement relationship, and is insensi-tive to those defects described above. The mirror symmetry of the structure facilitates the design of the rebalance loops that compensate for accelerations. The compensating volt-age is proportional to the acceleration, so the structure can be used as an accelerometer. The symmetry of the structure, on which the use of the sensor as a gyroscope and accelerometer depends, is both rotational and mirrors symmetry.
5. Scheme for detecting rotation and acceleration
When both the main ring and support rings of the sen-sor are considered, the governing equation of the nth mode vibration is represented as follows.
A¨an− BΩ˙bn− B2Ωb˙ n+ c˙an+ Can = Fan A¨bn+ BΩ˙an+ B2Ωa˙ n+ c˙bn+ Cbn = Fbn
, (7)
where A is the effective mass; B and B2 are the effective
masses associated with the Coriolis force; c is the effective damping; C is the effective stiffness;Fan andFbn are the generalized forces, and ¨anand ¨bnare generalized coordinates of the nth mode of vibration.
This work addresses the first and second modes. The first mode involves linear acceleration, and the second involves rotation.Fig. 4indicates the motion of the ring consists of the first and second modes of vibration.Fig. 2implies that the second mode of vibration is driven by the main driving loop. The rebalancing loop for detection is added to sense the rotation without which the signal process would be much more complex. Based onEq. (7), with reference to the second mode but ignoring the angular acceleration, the driving forces are as follows.
Fa2 = sin ωt Fb2 = −Kb2
, (8)
whereω is the natural frequency of the second mode and
equals√C/A and K is gain of the rebalancing force which is exerted by the rebalancing loop. The output signal determined
Fig. 7. (a and b) Scheme for detecting rotation of gyroscope; (c and d) scheme for detecting linear acceleration of accelerometer.
Natural frequency ω1= 3 5Eρw 2 r4 ω2= 1.24ω1 ω3= 0.4ω1, ω3= 2.58ω1
by the rebalancing loop is,
Vrot= Kb2= −ΩB
c sinωt. (9)
The output voltage is proportional to the rate of rotation. This finding is obtained by setting a large gain of rebalancing force, making b2much smaller than a2inEq. (7), such that
the terms in b2can be neglected.Fig. 7(a) and (b) depict the
detection scheme used to measure the rate of rotation. The fixed amplitude of the primary mode is driven in a2and the
Coriolis force induces a response in the form of the secondary mode in b2.
As shown inFig. 2, the first mode of vibration is rebal-anced by two rebalancing loops to sense linear accelerations. The amplitude of induced displacement is constrained by the applied force, the driving frequency, effective stiffness, and gain of the rebalancing force. When the driving frequency is much smaller than the resonant frequency of the first mode, the amplitude is approximated to,
f
C + K. (10)
The effective stiffness is designed to be high to reduce the amplitude as much as possible. Also, the rebalance loop is added to yield the readout circuit.Fig. 7(c) and (d) present the detection scheme for sensing acceleration. The opposing support rings are displaced equally when the structure is ac-celerated. The difference between the induced signals of the displacements of the opposing support rings represents the acceleration.
Table 2
Comparison of calculated and simulated results
Analytical FEM (Ansys5.3)
Resonant frequency of center fixed ring (Hz) 20926 20990
Resonant frequency of side fixed extension ring (Hz) 25955 25457
Resonant frequency of side fixed rotation ring (Hz) 8456 9047
Resonant frequency of total structure (Hz) 21935 22823
Displacement caused by electrostatic force (mm) 2.56E−5 2.46E−5
Displacement caused by acceleration (mm) 1.07E−5 1.19E−5
Ring radius: 1 mm; ring width: 20 um; ring thickness: 250 um; Young’s modulus: 150 GPa; material density: 2300 kg/m3; applied electrostatic force: 5.43N; applied acceleration: 9.8 m/s2.
6. Natural frequency
Many microstructures vibrate at natural frequencies with large amplitude, as does the sensor proposed herein. The cal-culation of the natural frequency is important, because the frequency relates to the working point of the circuit of the sensor. The readout circuit is designed to suit this frequency, which also affects the performance of the sensor. Designing the circuit with an excessively high frequency raises many problems, such as the band limitation of an operational am-plifier.
The process of estimating the natural frequency is sepa-rating the structure into three parts; determining the natural frequency of each part individually, and then calculating the composite natural frequency of the structure. The structure is separated into the main ring, extending support rings and ro-tating support rings. The similarity between the main ring and the supports simplifies the calculation.Table 1lists the mode and natural frequency of separated parts of the structure.
Problematically, the natural frequency of the rotating sup-port ring is not close toω1. Accordingly, the frequencies of
the two modes close to ω1, with the same boundary
condi-tions, are estimated. The exact frequency is determined using two concepts. The first is to weight the effective mass. The mode of vibration of a composite structure should be similar to that of a substructure with a greater effective mass. In the design proposed here, the effective mass of the main ring is weighted more heavily and the mode of the structure is close to that of the main ring. The extending support rings are also heavily weighted. Hence, the natural frequency of the struc-ture should be betweenω1andω2. The second concept is the impedance for each mode. The impedance of a mass-spring
system is expressed as follows;
Z = XF = |K − Mω2|. (11)
The natural frequency of the structure is assumed to be
ω1. The impedances of the two modes are,
Z3= 0.105Mω2 1 Z
3= 0.53Mω21
. (12)
Clearly,Z3is about five times Z3. The authors believe that
the natural vibration of the structure must be efficient. Thus,
the mode associated with Z3should be the desired mode of
the rotating support rings.
The natural frequency of the structure is determined by adding the total kinetic energy and the total potential energy. The natural frequency of the structure,ω is expressed as,
ω ∼= 1.048ω1. (13)
This frequency is close toω1and the mode is very similar to the second mode, and so the requirements are satisfied.
Table 2compares the numerical and theoretical results, and the errors between these results are less than 5%. The part of the structure is modeled by quadrilateral shell ele-ments, and the type of the element is Shell63.Fig. 8indicates the convergence of the natural frequency with mesh refine-ment. The natural frequency of the second mode of the silicon structure is 22,823 Hz.
7. Sensitivity of sensor characteristics to the design specification, fabrication errors and theoretical errors
The sensitivity of the sensor characteristic to the design specification is obtained by calculating the natural frequency of the structure. Thereafter, many important system param-eters can be derived and sensor performance can be esti-mated, for example, structural displacement and S/N ratio. The design parameters of this structure are the ring radius r,
Fig. 8. Finite element analysis: the part of the structure modeled by quadri-lateral shell elements and convergence of the natural frequency with mesh refinement. Table 3 Dimensions of parameters Symbol Formula Resonant frequency ωn E ρwr2 Electrostatic force Fe εVcc2drt2 Coriolis force Fc ΩQεEρ× Vcc2tr3 wd2
Rotation signal Vrot ΩQε √ρ
E√E× Vcc3 r 6 w4d3 Acceleration signal Vacc aEρ× Vcc r4
dw2 Signal to noise ratio for SNR(rot) ΩQε2 1
E× Vcc3 r 3t w5d3 rotation
Signal to noise ratio for SNR(acc) aεEρ× Vccr3t d2w acceleration
E: Young’s modulus of material;ρ: material density; w: ring width; r: ring radius;ε: dielectric parameter; Vcc: static applied voltage; t: ring thickness; d: gap between ring and electrode;Ω: rotation rate; Q: quality factor; a: acceleration.
the ring widthw and the ring thickness t.Table 3 lists the results.
The radius, width and thickness of the ring and the width of the gap between the ring and the electrode may all man-ifest manufacturing errors. The sensitivities of sensor char-acteristics to the fabrication errors are estimated from the
dimensional parameters, which are given inTable 3.
Sin-gle crystal silicon is a suitable material, because its material properties, including density and Young’s modulus, are sta-ble. In fact, lithography seldom produces an erroneous ra-dius of the ring. Hence, only an error in the width of the ring that arises during fabrication influences the resonant
fre-quency. The ring width designed here is 10m/20 m, and
the resultant errors are between 1 and 2m, or 10%. The
error in the width of the ring leads to an error in the width
of the gap. The gap width used here is 2m, so the error
may strongly affect the performance of the sensor. The width of the gap increases as the width of the ring declines. An error in the width of the gap is the most common manufac-turing error.Fig. 9plots the normalized sensitivities, listed inTable 3, versus the change in the width of the gap. The width of the ring is designed to be ten times the width of the gap. The maximum error in the width of the gap is typically equal to the width of the designed gap, because of the etch-ing process used in this work. The normalized sensitivities of the sensor’s characteristics of interest to the inaccuracy in the width of the gap, as plotted inFig. 9, imply that ev-ery such characteristic deteriorates as the width of the gap increases. In the worst case, parameters such as the S/N ra-tio decline to approximately one-fifth of their designed val-ues. The anisotropy of the characteristics of the structure can cause coupling of the vibration modes, which is associated with the crystallographic orientation of the silicon wafer[31]
and rotational misalignment during fabrication. Aligning the first mask with the crystals in the wafer can eliminate the anisotropy[32].
Fig. 9. Normalized sensitivities of sensor characteristics of interest to man-ufacturing errors in the gap width.
The sources of theoretical errors discussed herein include shifting of the natural frequency and angular acceleration. The structure is designed to vibrate at its resonant frequency, is close to the natural frequency if damping is weak. The factors that influence the resonant frequency must be consid-ered. The first is damping. Providing a stable environment in which the sensor can operate is a solution. Acceleration is the second factor that influences the natural frequency. Theoret-ically, a force or acceleration causes the natural frequency to drift, governed by,
∆ω = ωn×
r2
EIn2π2× F. (14)
According to the simulation results, this drift is neglect able. This effect depends on the ratio of the applied force to the relative stiffness of the structure. In this work, this ratio on the micro scale is too small to have any effect.
Eq. (7)specifies the errors due to the angular acceleration. As mentioned, the sensor incorporates designed rebalancing loops. A steady state error actually occurs in the rebalancing loops. This error is simply treated as a constant force applying to the structure.Eq. (7)for the second mode of vibration is solved as follows; q1= 1 CF1+ B2 ˙ Ω C2+KC+(B 2Ω)˙ 2F2 q2= −B2Ω˙ C2+KC+(B 2Ω)˙ 2 F1+ C C2+KC+(B 2Ω)˙ 2 F2 (15)
When a force is applied, an angular acceleration causes the force in a single direction to have an unwanted influence in another direction, generating noise. Applying a sinusoidal force also creates this problem. Only increasing the effective stiffness and the gain of the rebalancing force can solve it.
A process for fabricating the motion sensor and reducing
the ARDE is developed [33], and described below.Fig. 10
displays a schematic view of the process described in the following sections, observed along AA¯ inFig. 1. The pro-cess starts with a {1, 0, 0} n-type silicon wafer, which is patterned using photo resist, as depicted inFig. 10(a). The patterned wafer is etched by silicon dry etching, as depicted inFig. 10(b). A protective 500 ˚A thick layer of oxide is de-posited following dry etching, as depicted inFig. 10(c). This layer protects the structure during the subsequent etching of the backside, as revealed inFig. 10(e). The 5000 ˚A thick in-terlayer of nickel is deposited on the glass wafer, as indicated inFig. 10(d). This layer prevents the bonding of the glass to those areas that which should be removed during wet etching. It also prevents the structure from bonding to the glass, ensur-ing that the release of the structure is satisfactory. The silicon wafer and the glass wafer are bonded using an anodic bon-der. The upper part of the resulting SOI structure is removed by back-sided wet etching, forming the structure depicted in
Fig. 10(e). The areas designed to be removed are removed during this step. These areas are added to the design of the mask. The width of the line on the wafer is thus uniform, effectively reducing the aspect ratio-dependent effect during dry etching. The aforementioned protective layer guarantees the bonding from the silicon to the glass and resists the wet etching process. The structure is released by the wet etching of the glass wafer, as indicated inFig. 10(f).Fig. 11shows the resulting structure, with a thickness of 180m.
Fig. 12(a) presents the gap between the support ring and the electrode in their neutral positions.Fig. 12(b) indicates that the gap width is reduced when the electrode attracts the
Fig. 10. Schematic view of fabrication process: (a) silicon patterning with photo resist; (b) deep silicon etching; (c) deposition of protective layer; (d) anodic bonding; (e) back-sided wet etching; and (f) release of structure.
Fig. 11. Photographs of the ring-type motion sensor: top view of the structure with a height of 180m, backlight through the glass and detailed view of structure of the gap between the ring and the electrode (SEM).
Fig. 12. (a) Photograph of the gap between the support and the electrode in their neutral positions; (b) photograph of the ring-type motion sensor when the electrode is attracting the support ting; (c) deformed structure; and (d) electrical connections.
support ring.Fig. 12(c) depicts the support ring is deformed into an oval shape.Fig. 12(d) depicts the electrical connection of the driving electrode. This deformed shape is very close to the designed mode of vibration. Due to the over-etching in dry etching process, the width of gap increases to 4m and the natural frequency of the structure decreases to 19 kHz. This result follows the estimation of sensitivities.
9. Conclusion
This work presents a motion sensor that performs the func-tions of a gyroscope and an accelerometer. The capacitive
accelerometer and the vibrating ring gyroscope can be in-tegrated in a motion sensor that detects both acceleration and rotation. The vibration of a circular ring can be theo-retically decoupled into many modes, all of which exist si-multaneously. The first mode stands for linear motion, while the second involves rotation. The detected signal can be out-put simply and clearly by adding rebalancing control loops and designing a structure with relatively high stiffness. The rotational symmetry of the structure enables the sensor to detect rotation. The mirror symmetry maintains the perfor-mance of the sensor under acceleration. Rebalancing loops enhance the sensor as both a gyroscope and an accelerome-ter. A lack of both rotational symmetry and mirror symmetry
sults are less than 5%, and the modes of vibration are exactly as required. The error in the gap width is the major concern regarding the fabrication of the structure presented herein. Every considered characteristic deteriorates as the width of the gap increases. In the worst case, parameters such as the S/N ratio fall to approximately one-fifth of the designed val-ues. Structural testing demonstrates that the deformed shape of the structure and its driving frequency are very close to those of the designed mode of vibration. Based on those de-sign considerations, the vibrating ring-type motion sensor can be used to integrate the functions of a gyroscope and an accelerometer.
Acknowledgement
The authors would like to thank the National Science Council of the Republic of China, Taiwan, for financially supporting this research under Contract No. NSC89-2212-E-009-062.
References
[1] L.K. Baxter, Capacitive Sensors Design and Applications, IEEE Press, New York, 1997.
[2] A. Lawrence, Modern Inertial Technology Navigation Guidance and Control, New York, Springer-Verlag, 1993.
[3] W.H. Quick, Theory of the vibrating string as an angular motion sensor, Trans. ASME, J. Appl. Mech. 31 (1964) 523–534. [4] W.D. Gates, Vibrating angular rate sensors may threaten the
gyro-scope, Electronics 10 (1968) 130–134.
[5] D. Boocock, L. Maunder, Vibration of a symmetry tuning fork, J. Mech. Eng. Sci. 11 (4.) (1969).
[6] L. Parameswaran, C. Hsu, M.A. Schmidt, A merged MEMS-CMOS process using silicon wafer bonding, in: Proceedings of the IEEE International Electron Devices Meeting, Washington, DC, December 1995.
[7] P. Greiff, B. Boxenhorn, T. King, L. Niles, Silicon monolithic mi-cromechanical gyroscope, in: Proceedings of the 6th International Conference on Solid-State Sensors and Actuators (Transducers’91), San Francisco, CA, June 1991, pp. 966–968.
[8] W.A. Clark, R.T. Howe, R. Horowitz, Surface micromachined-axis vibratory rate gyroscope, in: Proceedings of the Tech. Dig. Solid-State Sensor and Actuator Workshop, Hilton Head Island, SC, June 1996, pp. 283–287.
[9] T. Juneau, A.P. Pisano, Micromachined dual input axis angular rate sensor, in: Proceedings of the Tech. Dig. Solid-State Sensor and Ac-tuator Workshop, Hilton Head Island, SC, June 1996, pp. 299–302. [10] W. Geiger, B. Folkmer, J. Merz, H. Sandmaier, W. Lang, A new silicon rate gyroscope, in: Proceedings of the IEEE Micro Electro Mechanical Systems Workshop (MEMS’98), Heidelberg, Germany, February 1998, pp. 615–620.
silicon micromachined gyroscope with piezoresistive detection and electromagnetic excitation, IEEE J. Microelectromech. Syst. 8 (3) (1999) 243–250.
[14] D.R. Sparks, S.R. Zarabadi, J.D. Johnson, Q. Jiang, M. Chia, O. Larsen, W. Higdon, P. Castillo-Borelley, A CMOS integrated sur-face micromachined angular rate sensor: its automotive applications, in: Proceedings of the Tech. Dig. 9th Int. Conf. Solid-State Sen-sors and Actuators (Transducers’97), Chicago, IL, June 1997, pp. 851–854.
[15] F. Ayazi, K. Najafi, A HARPSS polysilicon vibrating ring gyroscope, IEEE J. Microelectromech. Syst. 10 (2) (2001) 169–179.
[16] A. Selvakumar, K. Najafi, High density vertical comb array mi-croactuators fabricated using a novel bulk/poly-silicon trench refill technology, in: Proceedings of the Tech. Dig. Solid-State Sensor and Actuator Workshop, Hilton Head Island, SC, USA, 1994, pp. 138–141.
[17] J.S. Burdess, T. Wren, The theory of a piezoelectric disc gyroscope, IEEE Trans. Aerospace Electron. Syst. 22 (4) (1986) 411–418. [18] B.J. Gallacher, J.S. Burdess, A.J. Harris, Principles of a three-axis
vibrating gyroscope, IEEE Trans. Aerospace Electron. Syst. 37 (4) (2001) 1333–1343.
[19] E.J. Loper, D.D. Lynch, Projected system performance based on recent HRG test results, in: Proceedings of the 5th Digital Avionics System Conference, Seattle, Washington, November 1983. [20] L.M. Roylance, J.B. Angeel, A batch-fabricated silicon
accelerome-ter, IEEE Trans. Electron. Devices 26 (12) (1979) 1911–1917. [21] H. Seidel, H. Riedel, R. Jolbeck, G. Muck, W. Kuoke, M. Koniger,
Capacitive silicon accelerometer with highly symmetrical design, Sens. Actuators A21–A23 (1990) 312–315.
[22] K. Okada, Tri-axial piezo-electric accelerometer, in: Proceedings of Transducer’95, 25–29 June 1995, pp. 566–569.
[23] B. Puers, L. Reynaret, W. Soneyw, W.M.C. Sansen, A new unaxial accelerometer in silicon based on the piezo-junction effect, IEEE Trans. Electron. Devices 35 (8) (1988) 764–770.
[24] R. Hiratsuka, D.C.V. Duyn, T. Otaredain, P. de Vries, Design con-sideration for the thermal accelerometer, Sens. Actuators A32 (1992) 380–385.
[25] E. Abbaspour-Sani, R.S. Huang, C.Y. Kwok, A linear electro-magnetic accelerometer, Sens. Actuators A44 (1994) 103–109. [26] C.-H. Liu, T.W. Kenny, A high-precision, wide-bandwidth
microma-chined tunneling accelerometer, IEEE J. Microelectromech. Syst. 10 (3) (2001) 425–433.
[27] H. Bruschi, G. Torzo, Method for accurate resonant frequency mea-surement with a phase modulated feedback loop, Rev. Sci. Instrum. 58 (1978) 2181–2184.
[28] S. Timoshenko, D.H. Young, W. Weaver J.R., Vibration Problems in Engineering, fourth ed., John Wiley & Sons, 1974.
[29] G.H. Duffey, Applied Group Theory for Physicists and Chemists, Prentice-Hall, Englewood Cliffs, 1992.
[30] W. Soedel, Vibrations of Shells and Plates, Marcel Dekker, New York, Basel, 1981.
[31] R. Eley, C.H.J. Fox, S. McWilliam, Anisotropy effects on the vibra-tion of circular rings made from crystalline silicon, J. Sound Vib. 228 (1) (1999) 11–35.
[32] J.M. Lai, W.H. Chieng, Y-C. Huang, Precision alignment of mask etching with respect to crystal orientation, J. Micromech. Microeng. 8 (4) (1998) 327–329.
[33] J.H. Weng, W.H. Chieng, J.M. Lai, Structural fabrication of a ring-type motion sensor, J. Micromech. Microeng. 14 (5) (2004) 710–716.
Biographies
Professor Wei-Hua Chieng received a PhD in mechanical engineering at Columbia University in 1989. He has been a full professor in mechanical engineering at National Chiao-Tung University since 1996 and was appointed associated professor in 1989. His current research is in MEMS and electrical circuit design.
Mr. Jui-Hong Weng received BS and MS degrees in mechanical engineering at National Chiao-Tung University, Taiwan, in 1996 and
1998, respectively. He has been a PhD candidate in mechanical engineering at National Chiao-Tung University, Taiwan, since 2000. From 1998, he is researching on MEMS. His current research is micromachined gyroscope and accelerometer.
Dr. Jenn-Min Lai received a PhD in mechanical engineering department of National Chiao-Tung University, Taiwan, in 1999. Since 2000, he has been a senior engineer of R&D Center in NeoStones Micro-Fabrication Corporation. He is in charge of process development of MEMS.
