Example: membrane pressure sensor

18. Piezoresistive Pressure Sensors : Case Study

Introduction
Piezoresistance
Motorola MAP sensor

Piezoresistive pressure sensor design

Capacitive pressure sensor design

The piezoresistive straing gauge are usually made of doped poly-Si and are
designed in pairs with a readout cicuitry (usually a Wheatstone bridge).
While strain-pressure reponses of the membrane have been modelled in
previous chapter, practical devices are usually rather calibrated, and their
response stored on-chip in a look-up table.
The response of the device to applied pressure is related through the
mechanical response of membrane, piezoresistive response of transducer:

Vout R ( P P0 )

In this design, a capacitive bridge can be formed wit two reference

capacitors, and the output voltage is related to the deflection of the
membrane x and hence the differential pressure (P - P0) through:

Vout C x ( P P0 )

By controlling the background pressure P0 it is possible to fabricate the

following types of pressure sensors:

An absolute pressure sensor that is referenced to vacuum (P0 = 0)

A gauge-type pressure sensor that is referenced to atmoshperic
pressure (P0 = 1 atm)
A differential sensor where P0 is maintained at a known value

Piezoresistive vs capacitive approaches

Piezoresistive pressure sensors

This chapter introduces piezoresistive devices through the

specific case study of membrane pressure sensors
While other approaches such as capacitive effects can be
used for such applications, silicons also possess the
property of piezoresistance whose implementations as
transduction mechanism in membrane is somewhat more
straightforward

Piezoresistive designs are the most employed because of its low cost,
robustness, and ease of circuit integration

Piezoresistivity

9. Piezoresistive Pressure Sensors : Case Study

Introduction
Piezoresistance
Motorola MAP sensor

Piezoresistivity is the dependence of electrical resistivity on strain

Such an effect is related to the rearrangement of energy bands of a
solid under applied strain (above)

Piezoresistivity

Analytic formulation in cubic materials

Assuming effect is linear, the relationships between electric field

and current density is given by:

= [e + ] J
where e is the resistivity tensor, is the piezoresistive tensor, is
the stress tensor, and J is the current density

Note: while and J are vectors, e and are second rank tensors,
while is a fourth rank tensor
However, in a cubic crystal the resistivity tensor is diagonal and
characterized by a unique diagonal value e
In addition, as previously described, the stress tensor can be
reduced to six independent elements and re-annoted as such:

Such strain will modify both the bandgap as well as the effective
masses (and thus mobilities) ascribed to the bands
The effect is isotropic in as much as a given strain may increase
resistivity along one direction while decreasing it along others

Analytic formulation in cubic materials (ctnd.)

Thus, the above equation can be written along the three principal
directions of the cubic lattice

3
= [1 + 113 + 12 (1 + 2 )]J 3 + 44 (13J1 + 23J 2 )
e
where the three independent piezoresistive coefficients are:

e 11 = 1111
e 12 = 1122
e 44 = 2 2323

23 = 32 = 23

22 = 2

31 = 13 = 31

33 = 3

12 = 21 = 12

Longitudinal and transverse piezoresistance

t
l
J

1
= [1 + 111 + 12 ( 2 + 3 )]J1 + 44 (12 J 2 + 13J 3 )
e
2
= [1 + 11 2 + 12 (1 + 3 )]J 2 + 44 (12 J1 + 23J 3 )
e

11 = 1

l
t

If a relatively long and narrow resistor is defined in a planar

structure, then the primary current density and electric field are
both along the long axis of the resistor.
Structures are usually designed so that one of the axes of the
principle in-plane stress is also along the resistor axis.
This simplifies the set of equations to following simplified
formulation:

R
=
= l l + t t
R

where R is the resistance of the resistor and the subscripts l & t

refer to transverse and longitudinal stresses along the resistor
axis.

Longitudinal and transverse piezo... (ctnd.)

The orientations of resistor is not necessarily aligned with

crystalline orientations of device

Longitudinal and transverse piezo... (ctnd.)

Piezoresistors are often oriented in the [110] directions.

The directional cosines are:

The general expressions l and t are related to the original tensor

through:

l = 11 2(11 12 44 )(l12 m12 + l12 n12 + m12 n12 )

and

t = 12 + (11 12 44 )(l12l 22 + m12 m 22 + n12 n 22 )

1 1
(l1 , m1, n1 ) =
,
,0
2 2
Thus:

1
(11 + 12 + 44 )
2
1
= (11 + 12 44 )
2

l,110 =

where (l1, m1, n1) are the directional cosines between the
longitudinal resistor direction and the crystal axis and (l2, m2, n2)
is the direction cosines between the transverse direction and the
crystal axes

1 1
(l 2 , m 2 , n 2 ) =
,
,0
2 2

t ,110

Note: by this definition l1l2 + m1m2 + n1n2 = 0 given that these

direction cosines are orthogonal to each other.

Piezoresistive coefficients in Si

Design example

The resistors are fabrictated along the [110] directions

Design example (ctnd.)

Design example (ctnd.)

For n-type resistors:

For p-type resistors:

l = 71.8 1011 Pa -1

1
1
2
2
1
1
t = ( 11 + 12 44 ) = (102.2 + 53.4 13.6) = 17.6 1011 Pa -1
2
2

l = ( 11 + 12 + 44 ) = (102.2 + 53.4 13.6) = 31.2 1011 Pa -1

t = 66.3 1011 Pa -1
thus, p-type is better suited to perform piezoresistive
readout in this direction

Design example (ctnd.)

Numerical example

Which is better, a longitudinal or a transverse resistor?

The transverse resistor is fully plunged in region of maximum
strain, but will also be greatly affected by placement error
The longitudinal resistor spans over a wider region of stresses. It
will be less sensitive, but will be less prone to alignment errors
from device to device

The n-type cantilever is 200 m long 20 m wide and 5 m thick. It is bent

by a point load on its end. The p=type piezoresistors are 20 m long and
2m wide. A force of 10 N is applied at extremity of device. Calculate
change of resistance
From Senturia, section 9.3:

4L3

4(200 106 )3
F =
10 10 6 = 0.8m
w max =
EWH3
(160 109 )(20 10 6 )(5 10 6 )3

The cantilever deflection at any point x is given by:

w (x) =

FL 2
x
x 1

2EI 3L

Numerical example (ctnd.)

Alternate design

The radius of curvature is then given by:

1
2w
F
=
= (L x )
( x ) x 2
EI

Since = zE , the stress at surface (z =-H/2) is given by:

Given that I = WH 12 :

( x )
3

( x ) =

6F
H 2w

EH FH
=
(L x )
2
2I

( L x ) = 1.2 1011 ( L x )

It therefore spans from = 24 MPa at x = 0 to 21.6 MPa at x = 20 m

An average stress of = 22.8 MPa is therefore used to calculate change of

resistance. We finally get:

R
= (67.6 10 11 )( 22.8 106 ) = 1.54 %
R0

Alternate design (cntd)

Resistors R1 and R3 experience stresses in their longitudinal

direction and a stress in their transverse direction:

R 1 R 3
=
= l + t = (67.6 10 11 )
R1
R3

(note: the Poisson ratio in the [110] direction is = 0.064)

All four resistors are aligned along one of the [110] directions, and
aligned with the principle axes of stresses

Alternate design (ctnd.)

Resistors R2 and R4 experience stresses in their transverse

direction and a stress in their longitudinal direction:

R 2 R 4
=
= l + t = 61.7 10 11 l
R2
R4

Alternate design (ctnd.)

Averaging over doping variations

0(z)

surface diffusion

Connecting the four resistors in a Wheatstone bridge configuration,

we get:

V0
R1 R3 R2 R4
1 + 2
=

Vs ( R1 + R2 )( R3 + R4 ) 2(1 + 1 2 )

where:

Real life piezoresistors will present some degree of non-uniformity

with respect to the doping levels and the stress distribution they are
subjected to.
For instance, creation of piezoresistor through surface diffusion
doping will create a depth-dependent profile of the unstrained
resistivity 0, as seen above.

1 = ( l + t )
2 = ( l + t )

Averaging over doping variations (cntd)

Averaging over stress variations

substrate

doped
resistor
L

z=0
Ec

z=H

Ef

R0

Ev
z=0

z=zj

To calculate the nominally unstrained resistance R0 , we evaluate the

zj
following integral:

( z ) =

1
W
=
z
R 0 0 Le,0 (z)

where zj is the junction depth equal to the edge of the space-charge

junction.

The neutral axis is at z = H/2 thus:

E ( H 2 z)

Transverse stresses are neglected, thus, the stress-induced

resistivity change is:
Thus:

e (z) = e,0 (z)[1 + l l (z)]

zj

1
W
=
[1 l l (z)]z
R 0 Le,0 (z)
where we assumed that: [1 + l ]

[1 + l ]

Averaging over stress variations (cntd)

z=0

z=H

9. Piezoresistive Pressure Sensors : Case Study

9.1 Introduction
9.2 Piezoresistance
9.3 Motorola MAP sensor

Rearranging:
zj

W
R = R 0 1 + R 0
ll (z)z
Le,0 (z)

Motorola MAP sensor

The Motorola MAP sensor has been developped to measure the

absolute pressure in the intake of automobile engines

Motorola MAP sensor

The Motorola MAP sensor has been developped to measure the

absolute pressure in the intake of automobile engines

Motorola MAP sensor: transistor fabrication

Motorola MAP sensor: transistor fabrication

Motorola MAP sensor: diaphragm fabrication

Motorola MAP sensor: diaphragm fabrication (cntd.)

Design of piezoresistor

Design of piezoresistor (ctnd.)

I
1

WR

+
V2

LR

Vo
-

The resistor is probed in a four-point configuration as follows:

piezoresistor in the [100] direction

Design of piezoresistor (ctnd.)

2
3

The current density J in resistor is J = J1 along resistor axis but is J = 0 along the
other direction.

Design of piezoresistor (ctnd.)

However the field in the transverse direction is not zero due to the piezoresistive
properties of the materials

V1 =

Using the piezoresistive equations:

2
= [1 + 11 2 + 12 (1 + 3 )]J 2 + 44 (12 J1 + 23J 3 )
e

LR

1x1 = e (1 + 111 + 122 )J1

LR

(1111 + 1212 )x1

The transverse voltage is given by:

V2 =

since J2 = J3 = 0 and 3 = 13 = 23 = 0, we obtain:

WR

WR

2x 2 = e 4412J1x 2 = e 4412 J1WR

W
V2 = 44 12 R V1
LR

1 = e (1 + 111 + 12 2 )J1
2 = e 44 12 J1
3 = 0

LR

1
V1 = e L R J1 1 +
L R

1
= [1 + 111 + 12 ( 2 + 3 )]J1 + 44 (12 J 2 + 13J 3 )
e

3
= [1 + 113 + 12 (1 + 2 )]J 3 + 44 (13J1 + 23J 2 )
e

The voltage across the piezoresistor is given by:

Thus the transverse voltage V2 only depends on the shear stress present in
regions between the two taps.

10

Stress analysis

Stress analysis

A uniform plate under uniform pressure possess a displacement function

of:
C
2y
2x
w ( x, y) = 1 1 + cos

1 + cos
4
L
L
with C1 is the displacement in center of plate, and is given by:
P=

4 EH 3
6(1 2 )L4

C1

The radius of curvature in the middle of the edge is given by

1 2w
=
x x 2

x = L 2, y = 0

2 C
= 1
L 2

1
2

A numerical analysis of that structure would actually yield:

2

which is a factor of two off from our analytical solution. This numerical
result is used for the rest of the analysis.

EH
x =
2 x

x =

L
x = 0.294 P
H

The related surface stress at that location is:

Thus:

Using = 0.06 for the [110] direction:

L
x = 0.606 P
H

where P is the applied pressure.

Since we are dealing with a plate, the y direction stress at center of

the edge is:
y = x

L
(1 2 ) P
H

Stress analysis (ctnd.)

Now, lets calculate the shear stress on the resistors arranged as above.
The axial stresses in the x and y directions add up to a shear stress
2
given by:
x y
L
12 =
= 0.141 P
2
H

2
Thus:
W
V2
L W
= 44 12 R = 0.141 44 R P
V1
H LR
LR
With 44 = 138 10 11 Pa 1 , L H = 50, and WR L R = 1 / 5

we get:

V2
mV
= 0.096
V1
V kPa

11