1 Copyright © 2006 by ASME Proceedings of IMECE2006 2006 ASME International Mechanical Engineering Congress and Exposition November 5-10, 2006… [605104]

1 Copyright © 2006 by ASME Proceedings of IMECE2006
2006 ASME International Mechanical Engineering Congress and Exposition
November 5-10, 2006 Chicago Illinois
IMECE2006-13211
FABRICATION OF INTEGRATED PRESSU RE-FLOW-TEMPERATURE SENSOR FOR
HYDRAULIC SYSTEMS

Charles W. Groepper
University of Minnesota
1100 Mechanical Engineering
111 Church St SE
Minneapolis Minnesota 55455
[anonimizat]
(612) 845-8690 Dr. Tianhong Cui
University of Minnesota
1100 Mechanical Engineering
111 Church St SE
Minneapolis Minnesota 55455
[anonimizat]
(612) 626-1636

Dr. Perry Y. Li
University of Minnesota
1100 Mechanical Engineering
111 Church St SE
Minneapolis Minnesota 55455
[anonimizat]
(612) 626-7815 Dr. Kim A. Stelson
University of Minnesota
1100 Mechanical Engineering
111 Church St SE
Minneapolis Minnesota 55455
[anonimizat]
(612) 625-6528

ABSTRACT

This work develops a low cost multi-functional micro-
electro mechanical systems (MEM S) sensor for use in fluid
power systems. The device is small to facilitate easy integration
into fluid power components, and has the capability to sense
system pressure, fluid temperature, and small pressure
differences that can be correlated to flow rate. The design of
each of the sensing aspects of th e device is outlined, as well as
their layout on the sensor die. Pressure sensing with the device
is accomplished through the use of polysilicon piezoresistors,
while temperature sensing is accomplished using polysilicon thermisters. The procedure nece ssary to fabricate prototype
units is illustrated in detail, and special processes noted. Performance characteristics of pr ototype sensors compare well
to design model predictions. The polysilicon thermister demonstrated a linearity
of 2.32%, a repeatability of 0.6%, and an accuracy of 1.5
oC.
The differential pressure sensor demonstrated a linearity of
0.4%, a repeatability of 0.13%, and an accuracy of 3.6%. The system pressure sensor demonstrated a linearity of 0.7%, a
repeatability of 0.3% and an accuracy of 4.2%. These
performance characteristics prove the functionality of the
device.
KEYWORDS: MEMS, pressure, sensing, transducer,
temperature, fluid power

INTRODUCTION
Many machines rely on fluid power technology as a
means of energy transmission. Some examples include
assembly line robots, construction equipment, and cranes.
These examples all share one similarity; they rely on the
measurement of pressure, flow , and temperature for process
control and monitoring. Traditionally, the measurement of these process variables is accomplished through the addition of
stand alone sensing components to the fluid power system. In a previous paper, an alternative sensing method for fluid power
systems has been proposed. This method is to integrate small,
low cost multifunction MEMS sensors into existing fluid power
components. The flow rate is to be measured indirectly with the
method, in that small pressure drops that exist due to existing
geometry are correlated to the flow rate. The development of
the suitable methods to calibrate the measured pressure
differential across standard elbow geometry to the flow rate is
detailed in complimentary papers [1, 2].
In this paper, the development of the MEMS sensing
device will be detailed. This device is to have the capability to
measure the system pressure, a differential pressure for
inferring the flow rate, and the fluid temperature. A conceptual
overview of the device can be seen in Figure 1. This device Proceedings of IMECE2006
2006 ASME International Mechanical Engineering Congress and Exposition
November 5-10, 2006, Chicago, Illinois, USA

2 Copyright © 2006 by ASME differs from a commercialized product offered by the Foxborro
company, as flow measurement is to be done without the
introduction of energy losses into the system [3].
First, the design of the device will be briefly
discussed, including the temperature and pressure sensing
aspects. The fabrication procedure of the device will then be
outlined, with important process considerations noted.
Following the design and fabrication of the device, the performance characteristics of each sensor aspect will be
shown. The characteristics incl ude the performance of the
system pressure, differential pressure, and temperature sensing
components. Following the evaluation of each of component
separately, integration of the device into elbow geometry will
be shown.

Figure 1: Overview of the Integrated PQT Sensing Principle
NOMENCLATURE

SENSOR OVERVIEW

Figure 1 shows the pressure, flow and temperature
(PQT) sensing principle. A MEMS chip containing a differential pressure sensor, a gauge pressure sensor and a
thermister is integrated into elbow geometry between the two
pressure taps. When fluid flows through the elbow, a small
differential pressure is produced between the two taps which is
detected by the differential pressu re sensor. System pressure is
detected by the gauge pressure sensor, and the system
temperature is measured by the thermister. SENSOR DESIGN
1.1 Thermister Design

The primary consideration in the design of a thermister
is self-heating error, which appe ars directly as uncertainty in
the temperature measurement. Therefore, design consideration
needs to be given to both the power dissipated by the device,
and heat transfer. Measurement of thermister resistance in the
PQT application uses a Wheatstone bridge, with the thermister
functioning as a single resist ance element. The amount of
energy dissipated by the thermister is a function of its nominal
resistance and the bridge excitation voltage. Figure 2 details the
layout of a typical Wheatstone bridge [4].

Figure 2: Typical Wheatstone Bridge

If R 4 is the thermister, and the bridge is initially balanced,
the power dissipated is a functi on of thermister resistance and
excitation voltage, as given in Equation.
) (21
4 22
R RVsP+=
(1)

A bridge excitation voltage of 5V, will be used for the
MEMS thermister. If the thermist er has a nominal resistance of
10kΩ, the power is 0.625 mW. For comparative purposes
thermisters used to measure blood perfusion dissipate 4mW of
power, and can self heat to 3 or 4 degrees greater than ambient.
This type of device is typically spherical, and not in direct
contact with the fluid [5]. In this application the thermister is
fabricated on the surface of the device in a serpentine fashion,
has a large surface area in direct contact with the fluid, and dissipates 70% less power. Therefore, self heating errors are
expected to be less than 1
0C and can be assumed negligible in
comparison to the expected ch anges in hydraulic fluid
temperature.
To simplify device fabrication, the thermister is built
from doped polysilicon, the same piezoeresistive strain material
used for pressure sensing. Polysilicon has a positive temperature coefficient of resist ance (TCR), that an increase in
temperature results in an incr ease in resistance. For typical
doped polysilicon, the TCR is 0.08%/
0C. Other materials
demonstrate a TCR orders of magnitude higher, but their use
greatly complicates the fabrication process.

3 Copyright © 2006 by ASME 1.2 Pressure Sensor Design

The functional principle for pressure sensing in the
MEMS device can be seen in Figu re 3. A diaphragm is subject
to a pressure, which causes a small deflection and a strain on
the surface. This strain can be converted to an electrical signal
through the use of piezoresistors, or resistors whose resistance
changes with applied strain. The change in resistance can then
be converted to an output signal using a Wheatstone bridge.
Through calibration, the bridge output signal is then related to
the applied pressure.
Differential pressure signals are small. The differential
pressure sensor must be much more sensitive than the system
pressure sensor. For the desired sensitivity to be achieved with
reasonable diaphragm dimensions, the diaphragm must have a thickness on the order of microns. Thickness control of silicon
diaphragms of this size is di fficult; therefore an alternative
material is needed. Because co st is of importance, silicon
dioxide has been selected for use as the diaphragm material.
The required sensitivity of the system pressure sensor is much
lower, as it is subject to a much higher pressure difference
(~4.0MPa). This requires a thicker diaphragm, which can easily be fabricated from the base silicon material. Adjusting the
thickness and size of the diaphragms would allow for design
pressures other than those selected for this demonstration.
The use of potassium hydroxide (KOH) to etch the
bulk of the silicon required the diaphragms to be made square.
Although circular diaphragms have better stress properties due
to more uniform strain around the edges than square, the stress advantages do not outweigh the increased fabrication costs to
etch a circular diaphragm.

Figure 3: Basic piezoresistive pressure sensor theory of
operation. Resistors R1,R2…are defined in Figure 2.

The pressure sensing theory of operation suggests a
logical design path. First, the mechanical properties of the
diaphragm need to be selected to maximize the strain at the
surface, while providing a sufficie nt factor of safety against
breakage. After the diaphragms have been properly sized, the
location and size of the piezoresistors needs to be selected to
maximize the electrical output of the device. Once the size and
location of the piezoresistors ar e established, the physical
location on the die of each func tional component of the device
can be selected. Finally, the layout of the bonding pads and
wire traces can be done. Sizing of the diaphragms was done by modeling them
as plates of fixed periphery. Since the governing differential
equation for a plate of this type has no closed form solution, the
Ritz energy method was used to approximate the solution. The
use of this method to solve the differential equation for plates
of fixed periphery is shown in detail in our previous work [2].
The displacement of the differen tial sensor diaphragm (made of
2μm SiO 2) can be seen in Figure 4.

Figure 4: Displacement of differential pressure sensor diagram
given by Ritz Energy Method approximation to a plate of fixed
boundary conditions subject to maximum design pressure

For simplicity, only one term of the series expansion is
used. This simplification is adequate, as can been seen with a
comparison to the quarter plate solution from ANSYS, a
commercial finite element software package given in Figure 5
[6].

Figure 5: ANSYS displacement solution for differential
pressure sensor diaphragm

The Ritz energy approximation for the plate of fixed
periphery allowed the strain at the surface of the plate to be
known at any location on the su rface. The strain at the top
surface of the differential pressure diaphragm be seen in Figure
6. Note the strain is maximum near the edges of the plate,
where the vertical displacement is zero because of the clamped

4 Copyright © 2006 by ASME boundary condition. Equation 2 gives the expected change in
resistance of a piezoresistor as a function of the applied strain.
Maximizing the applied strain maximizes the percentage
change in resistance.
εGRR=Δ
(2)

Figure 6: Strain in a single direction as a function of position
on the top surface

The piezoresistors were designed in a serpentine
fashion and allowed to vary in both size and location on the
surface of diaphragm. An exampl e of such a resistor layout can
be seen in Figure 7.

Figure 7: Serpentine Resistor Layout

To maximize the output of the se nsor, each of the variables for
each of the piezoresistors was systematically varied. The
change in resistance was estimated by area averaging. To
reduce the number of variables in the solution domain, the
length the resistor extended off of the diaphragm, l s, was held
constant at 100 μm. Similarly, the width of the spacing, w s, and
the width of the resistor, w r, was held constant at 50 μm, a size that could be easily and cheaply fabricated. The output of a
Wheatstone bridge composed of piezoresistors arranged in a
full bridge pattern on the surf ace of the system pressure
diaphragm can be seen in
Figure 8. From the work of Bae it is evident that a single turn
resistor which gives the maximum resistor output would result
in substantially degraded signal to noise ratios [7]. Therefore, a
longer resistor with 2 turns and length, lp, of 250 μm was
selected.
The pressure sensor model under predicts the output
voltage. This is because the model considers only strain
information on the diaphgram, and assumes the strain is zero at
all locations off the diaphragm. The actual physical system
does have strain available off of the plate, resulting in a higher
output signal.

Figure 8: Bridge Output as a Function of Resistor Geometry

A similar design procedure was followed for the
design of the system pressure sensor. For this sensor, a
diaphragm thickness of 90 μm was selected, which required a
diaphragm size of 1.6mm. B ecause the diaphragm size was
larger, 3 resistor tu rns were selected.
After the piezoresistors were placed on the
diaphragms, the die layout was determined. For simplicity, the
pressure sensors were placed in the vertical center of the die,
equidistant from the horizontal centerline. This location was
selected to maximize the seali ng surface around the system
pressure sensor. Large 1.1 x 1.1 mm contact pads were
selected to allow hand soldering a connection to the sensor
without wire bonding equipment. The thermister was placed off
center between two contact pads to simplify wire trace layout.
Large 80μm wide wire traces were used for the wiring of sensor
so they could be seen with the naked eye. The layout of the
device can be seen in Figure 1.

5 Copyright © 2006 by ASME
SENSOR FABRICATION

Fabrication of the MEMS device is done using
standard deposition, lithographic and etching techniques to
reduce cost. A basic overview of the MEMS device fabrication
sequence can be seen in
Figure 9. The fabrication process uses clean 525 μm
thick <1 0 0> oriented double polished wafers. The resistivity
or doping of the base silicon does not affect the device
performance as all electrical as pects of the device are created
with surface micromachining on an insulating oxide layer.
Fabrication starts by wet oxidizing the base wafers to
form a 2 μm thick layer of SiO 2 on both sides. Wet oxidation is
used because it is a faster process, and also because the resulting films have much less residual stress than films
produced by a dry oxidation process. However, dry oxidation
processes result in higher film qualities, and the factor of safety
in the differential diaphragm design was selected based on the
use of the wet process [8, 9]. A cross sectional view of the
wafer with the oxide is shown in Figure 9a.
Following the oxidation of the wafers, a 5000 Ǻ thick
layer of in situ doped polysilicon was deposited using a
standard low pressure chemical vapor deposition process
(LPCVD). The in situ process wa s used in lieu of a separate
doping process to reduce cost and simplify fabrication.
Moreover, in situ doping using the LPCVD process can be
tailored to deposit films with no residual stresses [10]. A cross
sectional view of the wafer with the polysilicon is shown in Figure 9b. A recrystallization post deposition anneal step was
done to the polysilicon to reduce the sheet resistance. This
process was performed in an inert N
2 environment to reduce the
oxidation of the film. Following annealing, the polysilicon was
masked using standard lithographic processes, and the bulk of
the material etched away using reactive ion etching (RIE). The
etched material included that of the backside, where doped
polysilicon was not needed. After etching, the wafer appeared
as in step c of the fabrication sequence shown in Figure 9.
The doped polysilicon piezoresistors were connected
using gold contact traces. Past re searchers have noted adhesion
difficulty in the application of gold directly to silicon dioxide.
Therefore, a standard adhesion promoting 2000 Ǻ thick
chromium layer was deposited using electron beam evaporation prior to evaporation of a 3500 Ǻ thick gold layer. After gold
evaporation, the wafer appear ed as shown in Figure 9d.
Following deposition, a lithographic step masked the wire
traces and bonding pads, and the bulk of the gold, and
chromium was wet etched from the wafer surface to form both
the wire traces and the contact pads. Following gold deposition
and patterning, the wafer now appear s as in step e of Figure 9,
with all topside fabrication is completed.
For ease of prototype fabric ation, a timed etch stop
technique was used to control the thickness of each of the
sensor diaphragms. Therefore, in order to correctly size the
thickness of the system pressure diaphragm, and etch through
the wafer for the correct sizing of the differential pressure
diaphragm, the etching of the differential cavity needs to be
started first. Prior to etching the base silicon material, a hole needs to be created in the silic on dioxide on the backside to
expose it. This is done using buffered oxide etch (BOE)
solution and standard lithographic means. After oxide etching
the wafer appeared as shown in Figure 9f. Upon creation of a
window in the oxide, potassium hydroxide (KOH) was used for
the bulk etching of the Si. Since the silicon dioxide is being
used as the masking layer for th is etch, care needs to be given
to the KOH bath temperature and concentration to ensure sufficient selectivity [11-13]. Once KOH etching was
completed, the wafer appeared as shown in step g of Figure 9.
Care also needed to be given to protecting the
previously fabricated topside features during the KOH etching
of the backside. Through experimentation, it was determined a
combination of mechanical a nd chemical means provided
sufficient protection. Mechanical protection was provided with
a plastic fixture with an o-ring which sealed the topside features
in a liquid tight chamber. Since a pinhole or diaphragm
breakage would still result in KOH solution reaching the
sensitive topside components, a chemical barrier was also
applied [14].
After bulk micromachining to form both of the
pressure sensor diaphragms, the finished Si sensor component
(Figure 9j) was to be anondically bonded to Pyrex 7740 glass.
This process has been successfu lly applied in the commercial
production of many sensors [15]. However, for the purpose of
prototype testing, an alternativ e, faster method was used. The
method was to bond the finished Si sensor to a blank Si backing
plate with allowance for the diffe rential pressure signal using
Loctite® epoxy based adhesive. The epoxy was determined to
be more than adequate for s ealing against the pressures the
sensor would be subject. The completed sensor is shown in
Figure 9l.
Several challenges were encountered during the
fabrication of the MEMS PQT device. These challenges were
substantial enough to slightly alter the planned fabrication sequence and deserve mention.
The purpose behind fabricating prototype sensors was
to demonstrate the feasibility of the integrated PQT concept.
Therefore, fabrication issues such as exacting performance
similarities among devices on a wafer were of low concern.
This allowed cheaper masks fabricated from high resolution
transparencies to be used. The fa brication of these masks limits
the minimum useable feature size to 50 μm, and the feature size
variation from device to device is not negligible. The advantage
to using such masks is a very fast turnaround time, and ~66%
saving compared to CNC produc ed masks. All masks used in
the fabrication of the prototype MEMS sensors were of the
printed transparency type.

6 Copyright © 2006 by ASME

Figure 9: Fabrication Sequence

Difficulty was found in the fabrication of the
differential pressure sensor diaphragm. Silicon dioxide has
considerable compressive residual stress after deposition [8, 9].
During prototype fabrication, it was determined this residual
stress was high enough to cause breakage of the thin diaphragm
once all of the silicon backing material had been etched away.
Several attempts were made to anneal and relax this residual stress, however they were unsuccessful. Success in fabricating
this diaphragm was found through a combination of deep
reactive ion etching (DRIE) and silicon support. The technique
was to follow the standard etching steps as per
Figure 9 until there was approximately 40 µm of
material remaining. Then, DR IE was used to etch away
approximately 25 µm of material, leaving 15 µm of material to
support the SiO
2 diaphragms. The use of the DRIE to finish
etch allowed radii to be formed on the corners in the pressure sensor cavities, which eliminated the sharp edge stress
concentrations that would have ordinarily been present if the
KOH etch was finished. The reduction in the stress
concentration along with the support gained by the remaining
silicon material was sufficient to produce operational and
sensitive differential pressure sensors.
During the initial phases of fabrication, connection
difficulties between the Cr/Au wires and the piezoresistors
were encountered. This difficulty resulted in very low batch
yields, and operational sensors that produced very erratic output
signals. The difficulty was traced to poor cleaning of the
polysilicon layer before the e-beam evaporation of the Cr/Au
layers. Since the polysilicon was both annealed, and patterned
using standard lithographic techniques, there was a thin oxide and residual photoresist present. An intermediate RIE etch prior
to the evaporation of the Cr/Au layers provided the necessary
cleaning.

SENSOR TESTING AND EVALUATION

3.1 Thermister Evaluation

The performance of the thermist er was characterized through
comparison to a K calibration thermocouple using a Analog Devices AD595 thermocouple signal conditioning chip. For the
characterization tests, the MEMS PQT device and the
thermocouple were placed in an oil bath. The bath was heated
and allowed to cool, while both the resistance of the MEMS
PQT device and the temperature of the bath monitored.
Figure 10 details the experimental apparatus, while an example
of the raw data obtained is shown in Figure 11.

Figure 10: Oil Bath for Thermister Evaluation

7 Copyright © 2006 by ASME

Oil B ath Temperature and Thermister R esistance as a F unction of Time
25303540455055606570
0 500 1000 1500 2000 2500 3000 3500
Time (s)Temperature (C)
87408760878088008820884088608880890089208940
Thermister Resistance
(ohm)
Temperature (C) Resistance (ohm )

Figure 11: Thermister Evaluation Data

The small thermal mass of th ermocouple compared to
the PQT sensor can be seen in Figure 11. Since the bath is
subject to convection currents, the large thermal mass of the
PQT sensor works to smooth out any small temperature
fluxuations, while the small thermal mass of the thermocouple
is more sensitive to fluxuations.
An example calibration curve for several thermal
cycles on a single sensor is shown as Figure 12. Note the temperature-resistance characteri stic is linear over the small
temperature. Therefore, a Steinhart-Hart polynomial log fit is
unnecessary. The temperature co efficient of resistance can be
obtained from the trend line slope indicated in Figure 12. From
this slope, a temperature coeffi cient of resistance for the doped
polysilicon used in the MEMS PQT device is calculated to be
0.08%/
oC , which matches the value given by Madou [16].

Resistance as a Function of Temperature
89208930894089508960897089808990900090109020
35 37 39 41 43 45 47 49 51
Temperature (C)Resistance (ohm)
Test 1 Test2 Test3

Figure 12: Thermister Temperature Resistance Curve

When the calibration curve shown in Figure 12 is
used, the performance of the ther mister can be evaluated. The
result is shown graphically in Figure 13. Figure 13 shows the
thermister exhibits a linearity of 2.32%, a repeatability of 0.6%
with 3 trials conducted over 3 days, and an overall accuracy of
3.6%. The percentage accuracy of the device translates to a 1.5
oC uncertainty in temperature measurement with the device. These results are summ arized in Table 1 :MEMS Device
Performance Summary .
Thermister Indicated Temperature as a function of Actual
Temperature
2025303540455055606570
20 25 30 35 40 45 50 55 60 65 70
Actual Temperature (C)Thermister Temperature (C)

Figure 13: Thermister Calibration Curve

3.2 Differential Pressure Evaluation

Testing of the differential pressure sensor required a
holding fixture to be fabricated. This fixture can be seen in
Figure 14, and Figure 15. There are two pressure taps in the fixture allowing
for a differential signal to be applied to the MEMS device.

Figure 14: Fixture for Evaluating Pressure Sensor
Performance

Figure 15: Inside View of Evaluation Fixture

A differential pressure signal was obtained by
subjecting one side of the MEMS differential pressure sensor to

8 Copyright © 2006 by ASME atmospheric pressure, while the other side was subjected to the
pressure generated by a variable head of hydraulic oil. The
differential pressure measurement was measured with a high
accuracy Sensotec LVT differentia l pressure transducer. The
results of the offset corrected pe rformance test can be seen in
Figure 16.

Differential Pressure Sensor Output as a function of
Applied Pressure
00.010.020.030.040.050.06
0 5000 10000 15000 20000 25000 30000 35000
Applied Pressure (Pa)Sensor Output (Vout/Vin)
Cycle 1 Cycle 2 Cycle 3
Figure 16: Pressure Voltage Characteristic of Differential
Pressure Sensor

The linearity of the device was measured to be 0.38%,
the repeatability 0.13% and the accuracy 3.58%. Further
increases in the sens itivity through a reduction in the amount of
silicon support under the differential diaphragm resulted in
undesirable non-linear output characteristics.
The experimental results shown in Figure 16 can be
directly compared to the pred icted performance shown in
Figure 8. As expected, the resistor placement model under
predicted the output of the device. However, the results of the model were comparable to the sensor output, which validates
the use of the model for optimizing the output of the sensor.

3.3 System Pressure Evaluation

Performance testing of the system pressure component
of the prototype MEMS device was very similar to that of the
differential pressure test s. For this test, the output signal of the
MEMS device was compared to a commercially available
Barksdale pressure transducer.
To avoid breakage of the differential pressure
diaphragms during the tests, a high pressure hydraulic signal
was applied to both sides of the test fixture and MEMS device.
Figure 17 shows the performance of the system
pressure measurement. System pressure measurement exhibited
a linearity of 0.75%, a repeatability of 0.33%, and an accuracy
of 4.2% when the turndown (ratio of maximum flow range to
minimum) was restricted to 10:1. The performance is
comparable to commer cially available pressure transducers.
Slightly degraded performance was noted when measuring small pressure signals outside of the turndown range. System Pressure Sensor Output as a Function of Applied
Pressure
-0.0100.010.020.030.040.05
0.E+00 5.E+05 1.E+06 2.E+06 2.E+06 3.E+06 3.E+06 4.E+06
Pre ssure (Pa )Sensor Output (Vout/Vin)
Cycle1 Cycle2 Cycle3

Figure 17: Pressure Voltage Characteristic of System Pressure
Sensor
The results of the tests of the three sensors; thermister,
differential pressure sensor, and system pressure sensor are
summarized in Table 1.

Table 1 :MEMS Device Performance Summary
3.4 Sensor Integration into Elbow Geometry

Integration of the MEMS device into a typical fluid
power component is necessary to prove the validity of the
integrated PQT concept. The geometry is shown in Figure 18.
This geometry incorporates a standard flow bend with a
curvature to diameter ratio of unity, and also has the ability to house the MEMS device. The overall dimensions are 25mm X
25mm X 30mm. The cavity housing the MEMS sensor in the
example is 18mm X 18mm X 6mm.

Figure 18 : Integrated PQT Geometry

Integrating the MEMS device into the fixture shown in
Figure 18 allows all sensing aspects of the device to be tested
simultaneously. Since the temp erature and system pressure
sensor have already been evaluated in the previous sections, the
remaining task is to evaluate flow sensing of the sensor. Flow
sensing with the device requires the use of a calibration
procedure, which is developed in a previous work [1].
Application of the calibration procedure developed in this work

9 Copyright © 2006 by ASME to the geometry shown in Figure 18 results in the flow
performance curve shown in Figure 19.

PQT Indicated Flow Rate as a Function of Actual Flow Rate
02468101214
02468 1 0 1 2 1 4 1 6
Actual Flow Rate (LPM)PQT Indicated Flow Rate (LPM)

Figure 19 :PQT Flow Rate Performance Curve

The flow sensing aspect of the device demonstrates a
linearity of 2.83%, a repeatability of 0.92%, and an average
accuracy of 2.80%.
CONCLUSION
The design of a low cost, multifunction MEMS sensor
for use in fluid power systems was presented. The sensor has
provisions to measure system te mperature, system pressure, and
a differential pressure for inferring flow rate.
Temperature sensing with the device was
accomplished by incorporating a th ermister into the design. For
fabrication simplicity, polysilicon was used as the thermister
material. Although this material does not exhibit large
resistance temperature characteristic of other materials, the
fabrication simplicity gained justified its use.
Pressure sensing with the device was accomplished by
using piezoresistive strain elements arranged in a full bridge fashion. A simple fixed plate diaphragm model was used to
optimize both the mechanical diaphragm properties and the
placement of the piezoresistors on the diaphragm surface.
Independent testing of each of component on the
device showed each has good functionality and performance
characteristics. The uncertainty in the temperature
measurement was determined to be 1.5
0C, while the linearity of
the system pressure and differ ential pressure aspects of the
device was determined to be 0.13% and 0.75% respectively.
Proving the validity of the integrated PQT concept
required the MEMS device to be incorporated into a typical
fluid power component. The linearity of the flow portion of the
integrated PQT concept was determined to be 2.83%.

ACKNOWLEDGMENTS
The work outlined in this paper was sponsored by the
cooperative network for research in motion control through
fluid power.
The authors would like to thank PhD candidates Wei Xue
and Yi Liu for their numerous helpful discussions and instruction during the fabrication of the device.
REFERENCES

1. Groepper, C., et al. Design of Integrated Pressure-
Flow-Temperature Sensor for Hydraulic Systems . in
Bath Symposium on Power Transmission & Motion
Control 2006. University of Bath BATH BA2 7AY
UNITED KINGDOM.
2. Groepper, C. W., et al. Integrated Pressure Flow and
Temperature Sensor for Hydraulic Systems . in ASME
International Mechanical Engineering Congress and
Exposition . 2005. Orlando, Florida USA: ASME.
3. Fung, C., Harris, R., and Zhu, T., Multifunction
Polysilicon Pressure Sensors for Process Control.
Sensors, 1999. October.
4. Hambley, A., Electrical Engineering – Principles and
Applications . 1997, Upper Saddle River, New Jersey:
Prentice Hall. 103-105.
5. Yang, K. and Liu, W., A Novel Model of the Pulse
Method for Measurement of Local Tissue Blood
Perfusion. Medical Engineering and Physics, 2004.
26: p. 215-223.
6. ANSYS Inc . 2006: Southpointe
275 Technology Drive
Canonsburg, PA 15317.
7. Bae, B., et al., Design Optimization of a Piezoresistive
Pressure Sensor Considering the Output Signal-to-
Noise Ratio. Journal of Micromechanics and
Microengineering, 2004. 14: p. 1597-1607.
8. Alexandrova, S., Szekeres, A., and Christova, K.,
Stress in Silicon Dioxide Films. Philosophical
Magazine Letters, 1988. 58(1): p. 33-36.
9. Yu, C., Flinn, P., and Bravman, J., In-Situ
Measurement of Viscous Flow of Thermal Silicon
Dioxide Thin Films at High Temperature. Material
Research Society Symposium, 1997. 446: p. 261-266.
10. Oei, D. and Mccarthy, S., The Effect of Temperature
and Pressure on Residual Stress in LPCVD
Polysilicon Films. Material Research Society
Symposium, 1992. 276: p. 85-90.
11. Kovacs, G., Maluf, N., and Peterson, K., Bulk
Micromachining of Silicon. Proceedings of the IEEE,
1998. 86(8): p. 1536-1551.
12. Merlos, A., et al., Optimized Technology for the
Fabrication of Piezoresistive Pressure Sensors.
Journal of Micromechanics and Microengineering,
2000. 10: p. 505-510.

10 Copyright © 2006 by ASME 13. Seidel, H., et al., Anisotropic Etching of Crystalline
Silicon in Alkaline Solutions. Journal Electrochemical
Society, 1990. 137(11): p. 3612-2635.
14. Ruben, K., Flaim, T., and Li, C., Polymeric Protective
Coatings for MEMS Wet-Etch Process. Proceedings of
SPIE: Micromachining and Microfabrication Process
Technology, 2004. 5342.
15. Mirza, A. and Ayon, A., Silicon Wafer Bonding: Key
to MEMS High Volume Manufacturing. Sensors, 1998.
December.
16. Madou, M., Fundamentals of Microfabrication . 2002,
Boca Raton, Florida: CRC Press LLC.
298,205,295,294-297.