Inertia Sensors

•Displacement x is a measure of sensitivity. However mass results in

higher sensitivity
•Higher spring constant makes it stiff and lower sensitivity
 Parallel-Plate Capacitive
Accelerometer
One of the earliest fully integrated capacitive acceleration sensors is surface
micromachined on a wafer with integrated MOS detection circuitry.
The sensor consists of a metal coated oxide cantilever with a 0.35μg-thick
electroplated gold patch at its distal end serving as a proof mass. The length, width,
and the thickness of the cantilever are 108 μm,25 μm and 0.46 μm respectively.
The counter electrode is made of heavily doped p type silicon.
The capacitor gap is defined by an epitaxy silicon layer grown on a silicon surface.
• A surface micromachining process was developed using thermal oxide as the
cantilever structural material and epitaxially grown silicon as the sacrificial layer.
• Structural layer used to form the microstructure (cantilever)
• Sacrificial layers deposited between structural layers removed leaving the
microstructure.
 Fabrication Steps of Cantilever
Capacitive Accelerometer
• Step a) The process starts with an n-type, (100) silicon wafer.

20 3
• Step b) A heavily boron doped region (with concentration of 10 / cm ) is
made using an oxide layer as the doping barrier. Between steps a and b,
certain detailed procedures such as the oxide growth, deposition and
patterning of photoresist, and the subsequent oxide etch and photoresist
removal were performed
• Step c) An expitaxial silicon layer with a resistivity of is grown to a
thickness of over the entire wafer

• Step d) Another layer of oxide is deposited and patterned

• Step e) Oxide layer serves as a mask for etching via hole

• Step f) It also serves as a barrier for doping (to form drain, source, and
electrical conduction paths on the slopes of the via hole). The doping is
conducted using ion implantation at 100 keV energy and 5 10 / cm dose.
14 2

During the via hole etch, the heavily doped region will not be attacked
because the etchant reduces its etch rate on heavily doped silicon.
• Step g) The oxide barrier is removed

• Step h) Another layer of thick oxide is grown, which serves as the

dielectric insulator, the cantilever, and etching barrier in regions other than
the gate
• Step i) A layer of metal is deposited and patterned. It provides electrical

interconnects to the bottom p electrode, electrode on top of the oxide
cantilever, and gate of the field effect transistor. The metal layer consists of
20-nm thick Cr followed by 40-nm-thick gold, with the Cr used to satisfy a
critical requirement of enhancing the adhesion between the gold and the
substrate.

• Step j) A wet silicon etch is performed to undercut epitaxial silicon

beneath the oxide cantilever
• The released cantilever is naturally bent due to intrinsic stress present in
the metal and oxide thin films. The upward bending is approximately
1.5˚at the end of the cantilever. Under the influence of applied
acceleration, the beam will further deform from the stationary profile. For
these reasons, the cantilever surface is not perfectly parallel with the
substrate.
• The capacitance value between a curved cantilever and the doped counter
electrode can be estimated in a piece-wise fashion, by summing incremental
capacitances contributed by longitudinal segments of the cantilever. By
ignoring fringe capacitance, the total capacitance is estimated as

where L and b are the length and width of the cantilever, respectively, and
the permitivity of the air medium.
• 
The
0 capacitance change is read using a relatively simple impedance
converter. The sensor is capable of 2.2 mV/g of acceleration sensitivity,
corresponding to a beam displacement of 68 nm/g. The mechanical resonant
frequency of the cantilever is 22 kHz.
 Parallel-Plate Pressure Sensors
• A membrane pressure sensor can detect pressure differential across the
membrane. Two pressure ports are typically required. To simplify the
pressure sensor design and use, absolute pressure sensors are often
desirable.
• In such sensors, the reference pressure at one side of the membrane is
integrated. One popular choice is to provide a zero pressure
reference(vacuum) by hermetic sealing.
• The use of vacuum avoids expansion of trapped air and increases the
bandwidth by eliminating air damping inside the cavity.
• The device must retain resolution over a temperature range from -25˚C to
85 ˚C.
• A membrane made of doped silicon serves as the pressure sensing element
and one electrode. The counter electrode consists of patterned metal thin
film on the bottom substrate made of glass.
Fabrication Process of Pressure Sensor
with Sealed Cavity
• Step a) The process begins with a (100) silicon wafer

• Step b) An oxide mask is deposited and patterned

• Step c) Oxide layer serves as a chemical barrier during a wet anisotropic
etching of silicon (using KOH solutions)

• Step d) A 9μm -deep recessed region is created with the slopes being
surfaces

• Step e) Another layer of oxide is grown, this time as a conformal coating.

• Step f) The oxide is then photolithographically patterned

• Step g) A boron diffusion step at 1175˚C is conducted to form doped

regions as thick as 15μm.

• Step h) Stripping of the oxide, and growth and patterning of yet another
oxide layer
• Step i) The second oxide layer is used in a subsequent boron diffusion step
to define a doped region (depth=3μm) which becomes the thickness of the
membrane diaphragm

• Step j) A layer of silicon oxide is deposited and patterned, to form a

dielectric insulation

• Step k) Patterned holes in the oxide allow a subsequently deposited

polysilicon to contact the boron doped region and provide electrical contact
with the membrane later
• Step l) A layer of metal (consisting of Cr and Au) is deposited and
patterned, with the gold facing the front side of the wafer. An oxide layer
is deposited and patterned to reside on the bottom of the cavity to provide
electrical isolation in case the top membrane touches the bottom electrode.

• Step m) The researchers flip bonded the wafer onto a glass wafer, which is
coated with a composite Ti-Pt-Au layer. Wafer-level anodic bonding to the
glass is performed in vacuum ( torr) at 400˚C for 30 min
1 106
• Step n) The backside of the silicon wafer is etched in an anisotropic silicon
etchant to dissolve the silicon other than the heavily doped, raised
membrane

• Due to the large gap, the sensor has a wide dynamic range (500–800 torr)
along with a very high resolution (25 mtorr, equivalent of altitude
difference of one foot at sea level) after readout and digital compensation.
The device yielded a pressure sensitivity of 25 fF/torr (or 3000 ppm/torr).
 Flow Sensors
• Fluid flowing past a solid surface introduces a boundary layer, inside
which the flow velocity is reduced. Inside the boundary layer, the velocity
varies with the distance to the wall surface (y).
• The shear stress is defined as the velocity gradient at the boundary
multiplied by the viscosity of the fluid:
du
w  
dy
The term μ is the dynamic viscosity with the unit being kg/(m.s)
• Shear stress sensors reveal critical fluid flow conditions at the bottom of
the boundary flow, which are difficult to measure conventionally. The area
integral of shear stress produces drag force. The shear stress information
can be used for active control of turbulent flow field, for actively
monitoring fluid drag, and for achieving drag reduction.
• Techniques for measuring fluid shear stress fall into two categories: the
hot-wire/hot-film anemometer (indirect measurement) and the floating-
element technique (direct measurement).
• A floating-element shear stress sensor was the first MEMS shear stress
sensor developed. The floating element shear stress sensor determines the
magnitude of local shear stress directly by measuring the drag force it
experiences.
• A suspended floating element is flush mounted on the surface of a wall.
The displacement of the floating element due to the shear force (drag
force) acting on the plate is transduced into plate displacement, which can
be measured by a variety of techniques, including electrostatics,
piezoresistivity, piezoelectricity and optical sensing.
• The capacitive floating element consists of a plate (with area of W0  L0 ),
suspended by four fixed-guided cantilevers, each with a length, width, and
thickness of L1 , w1 and t1 respectively. The plate is assumed to be a rigid
body. A distributed drag force is applied on the element as well as on the
cantilevers.
• Under a given flow shear stress  w ,a distributed force P acting on the
floating plate and a distributed force q (N/m) on the four fixed-guided
beams are given as
1
P   wW0 Lo
and
q   wW1
2

• To estimate the total displacement, we assume a force of P/4 applied as a

point load at the guided end of one fixed-guided beam, along with a
distributed load.
• The total displacement, under the combined forces, is a linear summation
of displacements under each applied force.
• The electronics detection uses a differential capacitance readout scheme.
• Three passivated electrodes are located on the surface of the wafer
underneath the element and a thin conductor is embedded in the polyimide.
• The coupled capacitances between the drive electrode in the center and the
two symmetrically placed sense electrodes are modified by motion of the
floating element. This change in capacitance is transduced by connecting
the sense electrodes to a pair of matched depletion-mode MOSFET’s on
chip. (Fringe capacitances are ignored in the analysis.) The drive voltage
Vd V p Cdp
is coupled to the plate through
• The sense capacitance C ps1 and C ps 2 vary linearly with deflection of the
plate δ, according to
Fabrication Process of Flow Sensor
• The fabrication process for the device begins from a silicon wafer with
MOS circuits already defined.
• The entire wafer is passivated with 750-nm-thick atmospheric chemical
vapor deposition of silicon dioxide and a 1μm polyimide layer (Dupont
2545).
• The polyimide is added between the passivated electrodes and the
sacrificial layer to eliminate stress cracking in the silicon dioxide layer.
• A 3 μm -thick aluminum layer is evaporated as the sacrificial layer. It is
patterned photolithographically. A 1μm -thick polyimide layer is coated
again and cured. This is followed by the evaporation of a 30-nm thick
chromium layer, which serves as the floating electrode.
• A 30μm -thick polyimide layer is applied in seven coating steps. A layer of
aluminum is deposited on the top layer and serves as a mask for etching the
polyimide to define the plate and the cantilever. The undercut is 6μm
corresponding to a 30μm deep etching.
• The aluminum is removed using a mixture of phosphoric acid, acetic acid,
nitric acid, and water. It takes approximately 2 h to completely release a
500μm by 500μm floating element.
• Each support beam is 1 mm long, 10μm wide, and 30μm thick.The
Young’s modulus of the polyimide is 4 GPa. The cantilevers are loaded in
axial tensile stress due to the residual stress in the polyimide plate.
 Capacitive Tactile sensors
• A parallel-plate capacitor is formed by bonding a silicon wafer with glass.
• One piece consists of a cone-shaped silicon mesa suspended by a circular
silicon membrane with a thickness of t and a radius of a.
• The glass piece consists of a recessed region in which electrodes are
patterned. Four electrodes, each with an area ofL2 are arranged in a quad
configuration. Four capacitors are formed, between the four electrodes and
the suspended plate. These are denoted C1 through C4
• If a normal force is applied perpendicular to the substrate, the distance
between the movable mass and the bottom electrodes is reduced uniformly
for all four capacitors. The capacitance change is related to the
displacement by
 r  0 L2
C  2
d
d
• If a shear force is applied to induce rotational movement of the silicon
mass, the changes of capacitance for the four capacitors will be different.
Two capacitances increase while the other two decrease, with almost the
same degree of change.
• Under a tilting angle θ, the total capacitance of the tilted plate capacitor is
estimated as

 r  0 L2
C 
d  0.5 L
 Fabrication Process of Tactile
Sensor
• Step a) The processing of the silicon part began with a standard p-type
<100> silicon wafer, which is polished on both sides

• Step b) A buried n-type layer (3.5 μm deep) was formed by doping. Then a
6 μm–thick n-type epitaxial silicon layer was grown.The buried n type
layer and the epitaxial layer constitute the thickness of a flexible
membrane.
• Step c) A deep p-type diffusion doping is performed to electrically isolate
each capacitor electrodes. A composite layer of silicon oxide followed by
silicon nitride is grown on both sides.

• Step d) The silicon nitride and oxide on the backside is patterned to serve
as an etch mask for wet anisotropic etching
• Step e) After the etching, a contact pad on top of a membrane is formed by
anisotropic silicon wet etch.

• Step f) The silicon nitride and oxide layers are then removed using wet
chemical etchants. The silicon wafer is then bonded to a glass wafer, which
consists of a recessed region (3μm deep) with patterned electrodes on the
bottom. Anodic bonding is achieved at 400˚C with a voltage bias of 1000–
1200 V.
Characteristics of sensor output w.r.t applied calibration forces:

• In the range of 0 to 1 gram, normal forces causes a capacitance change of

0.13 pF, whereas shear forces causes a differential capacitance of 0.32 pF.
The capacitance change is linearly proportional to the calibration force
within this range.
 Parallel Plate Actuators
• Parallel-plate capacitors can be used for micro actuation.
• Most common applications involve linear displacement vertical to the
plane of electrodes or rotational displacement.
• The amount of displacement that can be achieved with a parallel-plate
capacitive actuator is limited by the initial gap spacing. Increasing the
initial spacing allows longer range movement but limits the magnitude of
forces.
• Parallel-plate actuators have been used to achieve long-range, in plane
movement, by employing a device called the scratch drive actuator (SDA).
• The principle of the SDA is shown in Figure.

• Each SDA consists of a parallel plate with a bushing along one edge.
• Under no applied voltage bias, the parallel plate is parallel to the substrate.
When a bias voltage is applied, one edge of the parallel plate will contact
the substrate first. As the bias voltage gradually increases, the contact area
between the top plate and the bottom substrate increases. This is often
referred as a “zipping” motion
• As the zipping motion progresses towards the edge with the bushing, the
bushing is forced to rotate and “skids” when the lateral force caused by the
zipping motion exceeds the friction force.
• Upon the removal of the bias voltage, the parallel-plate capacitor returns to
the horizontal plane, but the plate travels by a small in-plane increment,
anchored by the landed bushing.
• Rapid succession of periodic actuation causes the scratch drive to achieve
high-speed linear displacement.
• Velocities of SDA drives can reach 80 μm/s at 1000 kHz activation
frequency, with the speed linearly proportional to the frequency at lower
range. The linear output force of SDA is known to increase significantly,
form 10 μN to 60 μN when the voltage peak increases from 68 to 112 V.
The maximum output force of SDA may reach 100 μN.
 Inertia Sensors
• In parallel plate capacitive accelerometer, a parallel capacitor is formed
between the moving mass and the substrate.
• In inertia sensors by IDT configuration, two sets of comb fingers that move
out of plane with respect to each other are used.
• A change in capacitance is caused by the variation in the interelectrode
overlap area. C 

• For small angle displacement, the change in capacitance under an angular

displacement is given by

• Where lm , l f , d and n are the length of the inertia mass, the length of the
sensing finger, the gap distance, and the number of sense fingers.
• The rotational angle is related to the torque M by the expression

where a and are the acceleration and a correctional factor (0.281)

accounting for the rectangular cross section of the torsion beam. The
length, width, and thickness of the torsional beams are lm , l f , and t,
respectively.
 Fabrication Process of Inertia
Sensor
• Step a) The silicon wafer is heavily doped by boron to a depth of 12μm,
which corresponds to the thickness of the top electrode plates.

• Step b) Electroplated nickel is used (not shown) to mask against a deep

reactive ion etching through the deeply doped region to define the fingers
and mass
• Step c, d) A piece of glass is etched to make a cavity in a selective region

• Step e) The silicon wafer and the glass are bonded

• Step f) The undoped silicon layer is dissolved in silicon wet etchants. The
etching solution does not attack the deeply doped silicon and the glass
significantly. A recessed cavity below the bank of moving comb fingers allows
larger range of finger displacement and reduced aerodynamic damping.

• Compared with parallel-plate capacitors, the capacitance between two

neighboring set of fingers are relatively small. However, one can achieve large
capacitance and force by increasing the number of comb pairs.The sensor is
tested using a switched capacitance integrated circuit with a gain of 15 mV/fF.
The sensitivity of the device is 300 mV/g.