sensors

Article
Fabrication and Evaluation of a Flexible MEMS-Based
Microthermal Flow Sensor
Myoung-Ock Cho 1 , Woojin Jang 1 and Si-Hyung Lim 2, *

1 Department of Mechanical Systems Engineering, Graduate School, Kookmin University, Seoul 02707, Korea;
myock@kookmin.ac.kr (M.-O.C.); jngwj2061@kookmin.ac.kr (W.J.)
2 School of Mechanical Engineering, Kookmin University, Seoul 02707, Korea
* Correspondence: shlim@kookmin.ac.kr; Tel.: +82-2-910-4672

Abstract: Based on the results of computational fluid dynamics simulations, this study designed and
fabricated a flexible thermal-type micro flow sensor comprising one microheater and two thermistors
using a micro-electromechanical system (MEMS) process on a flexible polyimide film. The thermistors
were connected to a Wheatstone bridge circuit, and the resistance difference between the thermistors
resulting from the generation of a flow was converted into an output voltage signal using LabVIEW
software. A mini tube flow test was conducted to demonstrate the sensor’s detection of fluid velocity
in gas and liquid flows. A good correlation was found between the experimental results and the
simulation data. However, the results for the gas and liquid flows differed in that for gas, the output
voltage increased with the fluid’s velocity and decreased against the liquid’s flow velocity. This
study’s MEMS-based flexible microthermal flow sensor achieved a resolution of 1.1 cm/s in a liquid
flow and 0.64 cm/s in a gas flow, respectively, within a fluid flow velocity range of 0–40 cm/s. The
sensor is suitable for many applications; however, with some adaptations to its electrical packaging, it
will be particularly suitable for detecting biosignals in healthcare applications, including measuring
respiration and body fluids.



Keywords: MEMS; flexible thermal flow sensor; microheater; flow measurement



Citation: Cho, M.-O.; Jang, W.; Lim,

S.-H. Fabrication and Evaluation of a
Flexible MEMS-Based Microthermal 1. Introduction
Flow Sensor. Sensors 2021, 21, 8153.
The use of micro-electromechanical system (MEMS) technology satisfies various de-
https://doi.org/10.3390/s21238153
mands in the field of microsensors for small size, low power consumption, low cost, mass
production, high accuracy, high sensitivity, etc. These advantages have led to multiple
Received: 1 November 2021
Accepted: 4 December 2021
advances in the field of micro flow sensors [1]. Currently, MEMS-based micro flow sen-
Published: 6 December 2021
sors are found in various applications in industrial settings. The combination of MEMS
technology and automatic control/AI technology is suitable for use in smart homes/green
Publisher’s Note: MDPI stays neutral
buildings, safety/environmental sensing, and healthcare fields [2], with the objective of re-
with regard to jurisdictional claims in
alizing Internet of Things’ technology in relation to the construction of smart homes/green
published maps and institutional affil- buildings. In addition, since it is also important to present a mathematical model in the
iations. design process prior to developing these MEMS devices, there was an effort to build a
physical-mathematical model that is a basis for both presenting the operating conditions of
the device and an objective explanation of the results. [3]. However, in the field of micro
flow sensors, the most notable recent developments have been related to the healthcare sec-
Copyright: © 2021 by the authors.
tor and the measurement of biosignals, and many studies and advancements are currently
Licensee MDPI, Basel, Switzerland.
underway in this area [4–6].
This article is an open access article
There are three main types of MEMS-based flow sensors: thermal flow sensors, piezore-
distributed under the terms and sistive flow sensors, and piezoelectric flow sensors. Of these, the simple structure and
conditions of the Creative Commons principle of the thermal flow sensor and its ability to measure a wide range of flow rates
Attribution (CC BY) license (https:// in both liquids and gases makes it popular in many different fields [7]. Three different
creativecommons.org/licenses/by/ thermal flow sensor methods can be used according to the means of measurement, and
4.0/). an appropriate technique can be selected according to the type of fluid and flow rate.

Sensors 2021, 21, 8153. https://doi.org/10.3390/s21238153 https://www.mdpi.com/journal/sensors

Sensors 2021, 21, 8153 2 of 12

The hot-wire method measures heat loss, the calorimetric method measures the thermal
equilibrium between two thermistors, and the time-of-flight method measures the arrival
time of heat pulses. In some studies, advanced sensor performance has been realized using
a combination of these principles to maximize the advantages of each method [8–11].
To date, many different types of thermal flow sensors have been studied and devel-
oped, although few have progressed into commercial products. The majority of thermal
flow sensors are used to measure the low flow rates of gases or liquids up to several
milliliters/minute, and most reported data refer only to continuous flow rates. Preventing
heat loss from a sensor by blocking the conduction of heat to a substrate is challenging for
the thermal flow sensor.
Ashauer et al. realized a response time of under 2 ms and a resolution of 0.1 mm/s in
a thermal flow sensor across a wide range of flow rates by creating a thermally isolated
structure using thermopiles on a silicon nitride membrane [12]. Dijkstra et al. implemented
a calorimetric-based flow sensor with low hydraulic resistance that included a microchannel
that was thermally isolated from the substrate with a channel height of several tens of
micrometers to prevent any step coverage problems from occurring. The developed sensor
detected a signal fluid flow as low as 300 nL/min [13]. Ahmed et al. developed a wireless,
dual-mode, low-powered, calorimetric-based complementary metal–oxide semiconductor
MEMS flow sensor to measure bidirectional gas [14,15].
The studies of micro thermal flow sensors for biomedical application are remark-
able. Mistry and Mahapatra modeled a thermal flow sensor to develop a MEMS-based
implantable sensor to monitor arterial blood flow and simulated the thermal distribution
of the fluid flow [16]. Li et al. developed a smart catheter flow sensor to monitor cere-
bral blood flow; they used the constant temperature mode in the thermal diffusion flow
measurement principle to establish a method for providing reliable data without drift [17].
Additionally, they used a compensation circuit system to solve the problem of continually
increasing temperature in a medium. Jiang et al. developed a non-invasive respiratory
monitoring system using a flexible hot-film smart sensing strip that was attached to a
person’s upper lip to measure real-time respiration [18]. They used a Bluetooth module
for wireless data acquisition and a smart device as a display module, which could be
configured as a wearable device to monitor conditions such as sleep apnea. Okihara et al.
fabricated a detachable flexible flow sensor to evaluate the partial respiration characteristics
of the airway and measured the local flow [19]. The flexible sensors can be applied to the
medical and industrial fields that require micro-flow measurement and analysis since they
have a deformable property, which can access narrow and winding areas that are diffi-
cult to access with conventional sensors, and thus more accurate results can be expected.
Finally, Hedrich et al. developed a MEMS-based thermal membrane anemometric flow
sensor for connection to a breathing mask for long-term respiratory monitoring related to
cardiovascular disease [20].
In this study, we developed a flexible microthermal flow sensor for simultaneously
measuring high flow rates in gases and liquids from tens to hundreds of milliliters/minute
using the MEMS process on a flexible substrate. It is anticipated that the sensor could be
used in many fields for different purposes, including healthcare. Generally, the differential
pressure meter, which is known to be suitable for high flow, has a limited dynamic range,
and is also difficult to apply to a pulsating flow [12]. However, the present study’s flexible
thermal flow sensor can measure high-flow fluids and has obtained meaningful results
with pulsating liquid flows. In addition, the developed sensor can effectively reduce heat
loss compared to metal or silicon-based sensors by using a polyimide film with a low
thermal conductivity. This suggests that the developed sensor has the potential for future
use in medical applications.
This study’s sensor electrode was designed with one central microheater and two
thermistors—one on either side. It was developed by applying a calorimetric principle that
measures the flow rate by detecting the change in the equilibrium state of both thermistors
under a flow. A computational fluid dynamics (CFD) analysis determined the structure
2021, 21, x FOR PEER REVIEW 3 of 12

Sensors 2021, 21, 8153 3 of 12

thermistors under a flow. A computational fluid dynamics (CFD) analysis determined the
structure and size of the sensor, and the output voltage from the electrical circuit was
designed and monitored in real time using LabVIEW software. The sensor signal was
measured within anda size
flowofrange
the sensor, andSCCM
of 25–200 the output
using voltage
N2 gas at from the electrical
25–150 mL/min circuit was designed and
for a con-
monitored in real time using LabVIEW software. The sensor signal
tinuous liquid flow with a syringe pump. The signal was also verified in a pulsating liquid was measured within a
flow range of 25–200 SCCM using
flow using an artificial heartbeat model within a flow N 2 gas at 25–150 mL/min for
range of 25–200 mL/min. A highlya continuous liquid flow
with a syringe pump. The signal was also verified in a pulsating
accurate result was obtained using the simplest type of bridge circuit based on the calori- liquid flow using an
artificial heartbeat model within a flow range of 25–200 mL/min.
metric principle; many studies use a complicated circuit to compensate for the signal drift A highly accurate result
phenomenon. was obtained using the simplest type of bridge circuit based on the calorimetric principle;
This studymany studies
fabricated use a complicated
a MEMS-based circuit
flexible microto flow
compensate
sensor and for the signal drift
conducted flowphenomenon.
velocity measurement tests on gases and liquids using a basic technique. Important ex-conducted flow
This study fabricated a MEMS-based flexible micro flow sensor and
velocity
perimental results weremeasurement
obtained withintests on gases
a range andapplicable
of sizes liquids usingto thea basic
human technique.
body. Important
experimental results were obtained within a range of sizes
The future development of electrical packaging for the developed sensor can extend its applicable to the human body.
The future development of electrical packaging for the developed
application to the measurement of biosignals in the healthcare field, including for the sensor can extend its
application to the measurement
measurement of respiration and body fluids. of biosignals in the healthcare field, including for the
measurement of respiration and body fluids.
2. Materials and Methods
2. Materials and Methods
2.1. Modeling and Simulation
2.1. Modeling and Simulation
To review the To anticipated
review the difficulties
anticipated in designing
difficultiesand manufacturing
in designing the flexible
and manufacturing the flexible
microthermal flow sensor and to predict the results, a CFD analysis was conducted
microthermal flow sensor and to predict the results, a CFD analysis was conducted using using
ANSYS 17.0 FluentANSYS software beforesoftware
17.0 Fluent the development
before theofdevelopment
the sensor [21]. Forsensor
of the simplicity,
[21]. the
For simplicity, the
flow path was flow
regarded as a rectangular pipe, and the physical properties
path was regarded as a rectangular pipe, and the physical properties of the fluids of the fluids
were established wereforestablished
gas (air) and forliquid (blood),
gas (air) respectively.
and liquid (blood),The study predicted
respectively. The studythe predicted the
temperature change of the thermistors
temperature change of the on both sides ofon
thermistors theboth
microsensor
sides of theelectrode resulting
microsensor electrode resulting
from the change in the
from thefluid’s
changevelocity. The microheater
in the fluid’s velocity. The and thermistors
microheater andcomprising
thermistors thecomprising the
sensor were setsensor
in platinum
were set(Pt).
in Considering
platinum (Pt). theConsidering
size of the sensor
the sizetoofbethe
developed,
sensor to the be developed, the
size of the sensor’s
size electrode for theelectrode
of the sensor’s heater and forthe
thethermistors
heater andwas the 0.1 × 1.0 mm,was
thermistors and0.1the× 1.0 mm, and
electrode’s elements were separated
the electrode’s elements from oneseparated
were side of thefrom flowonepathside
using the flow
of the “linespath
fromusing the “lines
sketches” menu. from The ends ofmenu.
sketches” the flow The path
ends were
of thedefined
flow path as were
the inlet
definedandasoutlet,
the inlet and outlet,
respectively, and the fluid velocity
respectively, and thewas fluidprovided
velocity differentially
was providedatdifferentially
10–40 cm/s at atthe
10–40inlet.
cm/s at the inlet.
The ◦ C above that of the surrounding fluid,
The temperature oftemperature
the microheater of thewasmicroheater
set to 4 °C wasabove setthat
to 4of the surrounding fluid,
and the difference
and the temperature temperature difference
between thebetween the two thermistors
two thermistors accordingaccording
to the fluidto the fluid velocity
was simulated. Figure 1a presents a schematic design
velocity was simulated. Figure 1a presents a schematic design of the simulation area of the simulation area setup, the size
and arrangement of the sensor electrode, and the direction
setup, the size and arrangement of the sensor electrode, and the direction of the fluid, of the fluid, whereas Figure 1b
whereas Figureshows
1b shows the the
temperature
temperature distribution
distribution in the setting
in the zone
setting zoneand andthethesensor
sensorelectrode under
electrode underflow
flowconditions.
conditions.TheThestudy
studyoptimized
optimized the the distance between the heater and and the thermistor
the thermistor byby adjusting
adjustingthe thedistance
distancebetween
betweenthe thetwotwo components
components andand analyzing
analyzing thethe linearity of
linearity of the temperature difference according to the flow rate. Table 1 summarizes the the shape and
the temperature difference according to the flow rate. Table 1 summarizes
boundary
shape and boundary conditions
conditions and and the material
the material properties
properties usedused
in thein flow
the flow analysis.
analysis.

Figure
Figure 1. (a) 1. (a) design
Schematic Schematic design
of the of the simulation
simulation area, electrode,
area, the sensor the sensorand
electrode, and the
the direction ofdirection
the fluid.of(b)
the
Result of the
fluid. (b) Result of the temperature distribution in the setting zone
temperature distribution in the setting zone and the sensor electrode under a flow. and the sensor electrode under
a flow.
 Table 1. The boundary conditions and the material properties used in the flow simulation.

Materials
Dimension
Sensors (mm).
2021, 21, 8153 Boundary Condition 4 of 12
Air Blood Platinum
Density
Flow path 1×1×3 3 1.225 1060 21450 Heater Temperature 314 k
(kg/m ) conditions and the material properties used in the flow simulation.
Table 1. The boundary
Thermal
Materials Heat Flux
Dimension (mm) condition
Boundary Condition
Air Blood Platinum
Flow path
Heater 1 ××31
1 ×0.3 Specific Heat3 )(j/kg·k)
Density (kg/m 1.2251006 10603762 21450 134 Heater Temperature
RTD1,2 3141kLayer
Shell con-
Thermal (Thickness
Heat Flux
condition
duction (m): 2 ×
Heater 0.3 × 1 Specific Heat (j/kg·k) 1006 3762 134 RTD1,2
Shell 1 Layer (Thickness
10−5)
conduction (m): 2 × 10−5 )
Velocity
Velocity
Thermal Conductiv-
Thermal 0.1~0.4
0.1~0.4
RTD1
RTD 1 0.2 ××
0.2 1 1 Conductivity 0.0242
0.0242 0.52 0.52 69.1 69.1 Inlet Inlet (m/s) (m/s)
ity (w/m·k)
(w/m·k) Temperature
Temperature 310 k310 k
RTD 2 0.2 × 1 Viscosity (kg/m·s) 1.79 × 10−5 −50.00278 Outlet Temperature 310 k
RTD 2 0.2 × 1 Viscosity (kg/m·s) 1.79 × 10 0.00278 Outlet Temperature 310 k

2.2.
2.2. Design and Fabrication of the MEMS MEMS Sensor
Sensor Electrode
Electrode
This study
study used a flexible
flexible microthermal
microthermal flow sensor in conjunction with a thermo-
transfer method to to measure
measurethe theflow
flowvelocity
velocityininvarious
variousfluids.
fluids.TheThesizesize
andand shape
shape of
of the
the sensor were designed by establishing the correct temperature
sensor were designed by establishing the correct temperature of the microheater and cal- of the microheater and
calculating
culating thethe heat
heat capacity
capacity ofofthe
thesensor.
sensor.The
Thesensor
sensorshape
shapewas wassetset to
to aa structure in which
which
the typical calorimetric
the typical calorimetricmetricmetricprinciple
principlewaswas applied
applied byby arranging
arranging an upstream
an upstream sensor
sensor and
and a downstream
a downstream sensor sensor on both
on both sidessides
of theofcentral
the central heater,
heater, and theandsize
the ofsize of each
each sensor
sensor elec-
electrodes
trodes were were set0.1
set to to by
0.11by mm1 mm in obedience
in obedience to simulation
to the the simulation setting.
setting. EachEach electrode
electrode was
was in the form of a coil, and the central heater was set by calculating
in the form of a coil, and the central heater was set by calculating the width and length of the width and length
of
thethe
coilcoil within
within thefinal
the finalwidth
widthofof100100μmµm inin consideration
consideration of of the
the physical
physical and
and thermal
thermal
properties
properties of the metal to meet the set temperature. The two thermistors were minimized
of the metal to meet the set temperature. The two thermistors were minimized
in
in the
the width
width of of the
the coil
coil to
to increase
increase thethe resistance
resistance soso that
that aa linear
linear result
result could
could bebe expected
expected
by
by maximizing the temperature change, and the total width of the two thermistors diddid
maximizing the temperature change, and the total width of the two thermistors not
not exceed
exceed 100 µm,
100 μm, respectively.
respectively. The final
The final designdesign
of theofsensor’s
the sensor’s electrode
electrode is shownis shown in
in Figure
Figure
2a. 2a.

Figure 2. The designed and fabricated sensor electrode. (a) Final design of the sensor electrode,
bar = 500 µm. (b) Schematic of the MEMS process to form the sensor electrode. (c,d) Sensor electrode
fabricated by the MEMS process, bar = 6 mm and 500 µm, respectively. (e,f) Sensor electrode inserted
in the thermal flow sensor module, bar = 5 mm and 1 mm, respectively.

The dimensions of the sensor were approximately 1.3 mm (width) × 1.6 mm (length).
To maximize the temperature difference between the two thermistors, the distance to the
 fabricated by the MEMS process, bar = 6 mm and 500 μm, respectively. (e,f) Sensor electrode in-
serted in the thermal flow sensor module, bar = 5 mm and 1 mm, respectively.

The dimensions of the sensor were approximately 1.3 mm (width) × 1.6 mm (length).
Sensors 2021, 21, To
8153maximize the temperature difference between the two thermistors, the distance to the 5 of 12
heater was minimized by placing the heater and the thermistors 20 μm apart. The sensor
electrode was fabricated using a MEMS process based on a previously designed sensor
structure. To ensure a thin, flexible sensor, the substrate was formed from a polyimide
heater was minimized by placing the heater and the thermistors 20 µm apart. The sensor
film rather thanelectrode
a general wassilicon wafer.
fabricated First,
using a 75 μm
a MEMS thick based
process polyimide film was laid
on a previously on a sensor
designed
silicon wafer, and O2 activation
structure. To ensure was performed
a thin, to enhance
flexible sensor, the bonding
the substrate with from
was formed the photore-
a polyimide film
sistor. Chrome rather
(Cr) and
thanPt were sputtered
a general at aFirst,
silicon wafer. thickness
a 75 µmofthick
20 and 200 nm,
polyimide respectively,
film was laid on a silicon
and the sensor wafer,
electrode
and was completed
O2 activation wasvia a lift-offtoprocess.
performed enhanceFigure 2b presents
the bonding with thea photoresistor.
sche-
Chrome
matic of the MEMS (Cr) and
process forPt were sputtered
forming at a thickness
the sensor electrode,ofand
20 and 200 nm,
Figure 2c,drespectively,
shows the and the
sensorcompleted
finished item. The electrode was completed
sensor via a lift-off
electrode process. Figure
demonstrated 2b presents
a resistance of aapproxi-
schematic of the
MEMS process for forming the sensor electrode, and Figure 2c,d shows the finished item.
mately 100 Ω in the microheater and approximately 700 Ω in both thermistors.
The completed sensor electrode demonstrated a resistance of approximately 100 Ω in the
microheater and approximately 700 Ω in both thermistors.
2.3. Fabrication of the Sensor Module
2.3. Fabrication
A polyimide-based of the
flexible Sensor circuit
printed Module board (PCB) was fabricated to directly con-
A polyimide-based
nect the MEMS Pt electrode flexible circuit.
to an electrical printed A circuit board
section of (PCB) was was
electrode fabricated to directly
cut and
connect PCB,
attached to the flexible the MEMS
and thePt electrode
connection to pads
an electrical circuit.
on the PCB andAthe section
sensorof electrode
electrodewas cut
were connectedand attached
using silvertopaste.
the flexible PCB, and
The surface the sensor
of the connection padswas
module on the PCB and
double the sensor
coated
electrode were connected using silver paste. The surface of the sensor module was double
by immersion in a diluted PDMS solution to provide insulation while preserving high
coated by immersion in a diluted PDMS solution to provide insulation while preserving
sensitivity. Thehigh
thickness of the PDMS layer was determined via a scanning electron mi-
sensitivity. The thickness of the PDMS layer was determined via a scanning electron
croscope (SEM)microscope
to be approximately 5–6 μm. A SEM
(SEM) to be approximately image
5–6 µm. of the
A SEM PDMS
image layer
of the PDMSon layer
the on the
sensor module sensor
is presented in Figure 3. However, for the liquid flow test, the insulation
module is presented in Figure 3. However, for the liquid flow test, the insulation
layer was replaced
layerwith a coatingwith
was replaced of parylene-C since the former
a coating of parylene-C since theis former
more vulnerable to
is more vulnerable to
electrical short electrical short circuits,
circuits, unlike unlike
a gas flow a gas
[22], andflow [22],
the and the of
thickness thickness of thelayer
the coating coating
waslayer was
3 μm. 3 µm.

Figure 3.Figure 3. Aelectron

A scanning scanning electron microscope
microscope (SEM) image (SEM) image of the polydimethylsiloxane
of the polydimethylsiloxane (PDMS) layer on (PDMS) layer
the flexible sensor
on the flexible sensor substrate (cross-sectional view). The red circle shows a 10×
substrate (cross-sectional view). The red circle shows a 10× magnified image of the PDMS layer. magnified image
of the PDMS layer.
The sensor module was completed by attaching the pitch connector on the flexible
The sensorPCB to thewas
module electrode sensorby
completed andattaching
connecting
thethe jumper
pitch cables into
connector it (Figure
on the 2e,f). The
flexible
PCB to the electrode sensor and connecting the jumper cables into it (Figure 2e,f). The sensor
microheater of the sensing unit was subjected to a heating test using the completed
module.
microheater of the A thermal
sensing unit wasimaging camera
subjected to amonitored the using
heating test heat generated when asensor
the completed 1.25 V current
was applied to the microheater via the sensor’s custom-made PCB.
module. A thermal imaging camera monitored the heat generated when a 1.25 V current
was applied to the
2.4. microheater viaSystem
Electronic Circuit the sensor’s custom-made PCB.
This study designed an electrical circuit system, which included a Wheatstone bridge
circuit for the sensor, and fabricated a 5 × 5 cm PCB. Figure 4 shows the circuit diagram and
the signal measurement workflow. The two thermistors on either side of the microheater
were connected to the Wheatstone bridge circuit, respectively. A voltage of 5 V was applied
to the circuit from the DAQ, and a reduced voltage of 1.25 V was applied to the microheater
via a regulator on the PCB. Under a flow, the temperature difference changes the resistance
of the thermistors, which generates a voltage in the Wheatstone bridge circuit. The output
 2.4. Electronic Circuit System
This study designed an electrical circuit system, which included a Wheatstone bridge
circuit for the sensor, and fabricated a 5 × 5 cm PCB. Figure 4 shows the circuit diagram
and the signal measurement workflow. The two thermistors on either side of the micro-
heater were connected to the Wheatstone bridge circuit, respectively. A voltage of 5 V was
Sensors 2021, 21, 8153 6 of 12
applied to the circuit from the DAQ, and a reduced voltage of 1.25 V was applied to the
microheater via a regulator on the PCB. Under a flow, the temperature difference changes
the resistance of the thermistors, which generates a voltage in the Wheatstone bridge cir-
voltage
cuit. Thewas amplified
output voltage 50 was
timesamplified
via a differential
50 timesamplifier (LM358; National
via a differential amplifiersemiconductor,
(LM358; Na-
SantaClara,
tional CA, USA),
semiconductor, and the high-frequency
SantaClara, CA, USA), andnoise was cut off by noise
the high-frequency a resistor–capacitor
was cut off by
a(RC) filter. The output
resistor–capacitor (RC)voltage
filter. signal that passed
The output voltagethrough
signal the
thatcircuit
passed system
throughwasthe
digitized
circuit
by DAQ before being analyzed and visualized in dedicated software. To
system was digitized by DAQ before being analyzed and visualized in dedicated soft- analyze the signals
of the To
ware. flow sensor,the
analyze thissignals
study of
constructed an analysis
the flow sensor, algorithm
this study using LabVIEW
constructed software
an analysis algo-
(LabVIEW,
rithm using NI, Austin,software
LabVIEW TX, USA), and the voltage
(LabVIEW, signal
NI, Austin, was
TX, acquired
USA), at avoltage
and the frequency
signalof
20 Hz.
was acquired at a frequency of 20 Hz.

Figure 4.
Figure Circuit diagram
4. Circuit diagram for sensor driving
driving and
and signal
signalmeasurement
measurementworkflow.
workflow.The
Thecircuit system
circuit systemis
is powered
powered byby
5 V5from
V from the data
the data acquisition
acquisition (DAQ)
(DAQ) devicedevice (myDAQ:
(myDAQ: NI, USA),
NI, USA), and theand the voltage
output output
voltage is displayed
is displayed via LabVIEW
via LabVIEW S/W (viaS/W (via DAQ).
DAQ).

3. Results
3. ResultsandandDiscussion
Discussion
3.1. Visualization of the
3.1. Visualization of the Temperature
Temperature Changes
Changes by
by Simulation
Simulation
This section
This section uses
uses computational
computational simulations
simulations to to predict
predict the
the general
general performance
performance and and
results of the detection of fluid flow using the fabricated sensor. Several factors
results of the detection of fluid flow using the fabricated sensor. Several factors to be con- to be
considered when fabricating the sensor were checked in advance and were reflected in
sidered when fabricating the sensor were checked in advance and were reflected in the
the design of the device. The simulation visualized the temperature difference between
design of the device. The simulation visualized the temperature difference between the
the two thermistors, and the variations provided the magnitude of the output voltage,
two thermistors, and the variations provided the magnitude of the output voltage, which
which is directly or inversely proportional to the flow velocity depending on the properties
is directly or inversely proportional to the flow velocity depending on the properties of
of the fluid. According to the CFD analysis for the temperature difference between the
the fluid. According to the CFD analysis for the temperature difference between the ther-
thermistors of the sensor electrodes with respect to the flow rate of the fluid, in the case of
mistors of the sensor electrodes with respect to the flow rate of the fluid, in the case of gas,
gas, ∆T (ThermistorDownstream − ThermistorUpstream ) showed a tendency to increase as the
ΔT (ThermistorDownstream − ThermistorUpstream) showed a tendency to increase as the velocity
velocity also increased. However, when the physical properties of the fluid were changed
also increased. However, when the physical properties of the fluid were changed to those
to those of blood (which is a liquid), there was a tendency for ∆T to decrease at speeds
of blood (which is a liquid), there was a tendency for ΔT to decrease at speeds exceeding
exceeding 1 cm/s.
1 cm/s.Generally, in a calorimetric thermal flow sensor, the temperature of a downstream
sensor is higherinthan
Generally, a calorimetric thermal flow
that of an upstream sensor, thetotemperature
one according of a downstream
the heat provided by a heater.
sensor is higher
Accordingly, thethan that of an
temperature upstream∆T,
difference, oneis according
defined as tothethe heatobtained
value provided byby a heater.
subtracting
Accordingly, the temperature difference, ΔT, is defined as the value obtained
the upstream temperature from the downstream temperature. For gas, the temperature by subtract-
ing the of
decrease upstream temperature
the upstream from the
sensor according downstream
to the temperature.
flow rate is greater than thatFor gas,down-
of the the
stream sensor; therefore, the ∆T increases according to the increase in flow rate. However,
for liquid, there is a rapid temperature decrease for the upstream sensor under a flow;
consequently, the additional temperature reduction according to the flow rate is much
lower than the size of the temperature decrease in the downstream sensor. Therefore, the
temperature difference between the two sensors is inversely proportional to the increase
in flow rate. It is believed that this phenomenon is attributed to the difference in ther-
mal conductivity between gas (air) and liquid (blood) of 0.025–0.03 and 0.52–0.6 W/m·k,
respectively, representing a difference of approximately 17–20 times.
 that of the downstream sensor; therefore, the ΔT increases according to the increase in
flow rate. However, for liquid, there is a rapid temperature decrease for the upstream
sensor under a flow; consequently, the additional temperature reduction according to the
flow rate is much lower than the size of the temperature decrease in the downstream sen-
sor. Therefore, the temperature difference between the two sensors is inversely propor-
Sensors 2021, 21, 8153 7 of 12
tional to the increase in flow rate. It is believed that this phenomenon is attributed to the
difference in thermal conductivity between gas (air) and liquid (blood) of 0.025–0.03 and
0.52–0.6 W/m·k, respectively, representing a difference of approximately 17–20 times.
Additionally, when
Additionally, when the width of both thermistors
thermistors andand thethe microheater
microheaterwas was0.1
0.1mmmm
and the
and the distance
distancebetween
betweenthe theheater and
heater and thermistors
thermistors waswas
symmetrically
symmetrically 0.01–0.02 mm, mm,
0.01–0.02 the
temperature
the temperaturedifference between
difference the two
between the twothermistors increased
thermistors or decreased
increased almostalmost
or decreased line-
arly with
linearly thethe
with change
changein in
thethe
flow rate
flow of of
rate the fluid.
the fluid.Figures
Figures5 5andand6 6present
presentthe
thesimulation
simulation
results in
results in graph form
form for
forgas
gasand
andliquid,
liquid,respectively.
respectively. When
When thethe
velocity of the
velocity of fluid in-
the fluid
creased by 10 cm/s, the temperature difference between the two thermistors
increased by 10 cm/s, the temperature difference between the two thermistors increased increased by
approximately 0.1 °C ◦
for gas, whereas it decreased by 0.2 °C
by approximately 0.1 C for gas, whereas it decreased by 0.2 C for liquid. ◦
for liquid.

Figure5.5. Simulation
Figure Simulation results
results for gas flow with respect
respect to
to the
the distance
distance between
betweenthethemicroheater
microheaterandand
thermistors. (a,b)
thermistors. (a,b) Distance
Distance between
between thethe microheater
microheater and
and the
the electrode
electrodeisissymmetrically
symmetricallylocated
locatedbyby
0.2and
0.2 and0.1
0.1 mm
mm increments,
increments, respectively. (c) Electrodes
Electrodes are
are located
located asymmetrically
asymmetrically0.08
0.08mm
mmforforthe
the
sensor upstream
sensor upstream to to the
the heater
heater and
and 0.1
0.1 mm
mm for
for the
the sensor
sensor downstream
downstreamto tothe
theheater.
heater.(d)
(d)Electrodes
Electrodes
are located asymmetrically 0.1 mm for the sensor upstream to the heater and 0.09 mm for the heater
are located asymmetrically 0.1 mm for the sensor upstream to the heater and 0.09 mm for the heater
downstream to the sensor.
downstream to the sensor.

3.2. Experimental Results for Gas Flow

To evaluate the performance of the developed flexible thermal flow sensor, this study
configured an experimental setup to detect the change in voltage according to the change
in flow rate. The MEMS flow sensor module was inserted into 3 mm diameter tubing to
locate the sensor electrode in the center of the tube, and the cables from the sensor were
connected to the electrical circuit system outside the tubing. The two thermistors on either
side of the sensor electrode were connected to the Wheatstone bridge circuit, respectively,
and 1.2 V of voltage was applied to the microheater. This experiment used nitrogen gas
and a gas generator to control the flow rate, and the gas pipe was connected to the tubing
containing the micro flow sensor. The flow rates of the N2 gas (in SCCM (cm/s)) in the tube
were controlled to 25.65 (5), 51.3 (10), 77 (15), 102.6 (20), 128 (25), 154 (30), 179.5 (35), and
205.2 SCCM (40 cm/s) (the numerical value in parentheses indicates the velocity of the fluid
inside the tubing). A signal monitoring system performed real-time checks on the output
voltage, whereas nitrogen (N2 ) gas flowed at a specific flow rate from the gas generator to
the micro flow sensor.
 Sensors2021,
Sensors 2021,21,
21,8153
x FOR PEER REVIEW 8 of
8 of1212

Figure 6. Simulation results for the liquid flow with respect to the distance between the microheater
Figure 6. Simulation results for the liquid flow with respect to the distance between the microheater
and thermistors. (a,b) Distance between the microheater and the electrode is located symmetrically
and thermistors. (a,b) Distance between the microheater and the electrode is located symmetrically
by
by 0.05
0.05 mm
mm andand 0.02
0.02 mm
mm increments,
increments, respectively. (c) Electrodes
respectively. (c) Electrodes are
are located
locatedasymmetrically
asymmetrically0.020.02
mm
mm for
for the
the sensor
sensor upstream
upstream to to the
the heater
heater and
and 0.05
0.05mmmmfor forthe
thesensor
sensordownstream
downstreamtotothetheheater.
heater.(d)
(d)
Electrodes
Electrodesarearelocated
locatedasymmetrically
asymmetrically0.050.05mm
mmforforthe
thesensor
sensorupstream
upstreamtotothe heater
the and
heater 0.02
and mm
0.02 mm for
for downstream
the the downstream sensor
sensor to the
to the heater.
heater.

As the flow Results

3.2. Experimental rate variedfor Gas between
Flow 5 and 40 cm/s, the change in resistance resulting
from Totheevaluate
temperature the performancebetween
difference the two thermistors
of the developed (located
flexible thermal flowonsensor,
eitherthis
sidestudy
of the
microheater)
configured an experimental setup to detect the change in voltage according to the change7.
was converted into a voltage change. The results are presented in Figure
Figure
in flow7a contains
rate. The MEMSa stair-like graph of
flow sensor increasing
module flow rate
was inserted over
into time.diameter
3 mm A graphtubingshowing
to
the voltage value against the flow rate was derived using the average
locate the sensor electrode in the center of the tube, and the cables from the sensor were voltage value from
each flow rate
connected to the range, and acircuit
electrical simulation
systemresult forthe
outside thetubing.
gas flow was
The two plotted in Figure
thermistors 7b for
on either
Sensors 2021, 21, x FOR PEER REVIEW 9 of 12
reference.
side of the sensor electrode were connected to the Wheatstone bridge circuit, respectively,of
The result of the gas flow measurements for the coefficient of determination
the
andgraph
1.2 Vfor the change
of voltage wasinapplied
the voltage
to the against a flow rate
microheater. Thisbetween
experiment5 andused40 cm/s was gas
nitrogen 0.99,
which demonstrates an excellent linearity. For each flow rate range,
and a gas generator to control the flow rate, and the gas pipe was connected to the tubing the average standard
deviation
deviation
containing ofofthe
the
thevalues
values
micro was
flow was 0.38
0.38
sensor. mV;
mV;
The asflow
asthethe
flow
flow
rates rate
of theincreased
rate Nincreased by by
1 cm/s,
1 cm/s,the voltage
the voltage
2 gas (in SCCM (cm/s)) in the tube
increased
increased by
by 1.2
1.2 mV. The maximum
maximum noise signal signal was
was 0.77
0.77 mV,
mV,andandaaresolution
resolutionofof0.64 0.64cm/s
were controlled to 25.65 (5), 51.3 (10), 77 (15), 102.6 (20), 128 (25), 154 (30), 179.5 (35), and
cm/s
was was achieved
achieved in thein flow
the flowrate. rate.
205.2 SCCM (40 cm/s) (the numerical value in parentheses indicates the velocity of the fluid
inside the tubing). A signal monitoring system performed real-time checks on the output
voltage, whereas nitrogen (N2) gas flowed at a specific flow rate from the gas generator to
the micro flow sensor.
As the flow rate varied between 5 and 40 cm/s, the change in resistance resulting from
the temperature difference between the two thermistors (located on either side of the mi-
croheater) was converted into a voltage change. The results are presented in Figure 7. Fig-
ure 7a contains a stair-like graph of increasing flow rate over time. A graph showing the
voltage value against the flow rate was derived using the average voltage value from each
flow rate range, and a simulation result for the gas flow was plotted in Figure 7b for ref-
erence. The result of the gas flow measurements for the coefficient of determination of the
graph for the change in the voltage against a flow rate between 5 and 40 cm/s was 0.99,
Figure
Figure7.7.Results
Resultsofof the
thegas
gasflow
flowtesttest
using thethe
using flexible micro
flexible thermal
micro flowflow
thermal sensor. (a) Graph
sensor. showing
(a) Graph showing
which demonstrates an excellent linearity. For each flow rate range, the average standard
voltage value for gas flow rate over time. (b) Graph comparing the temperature difference (the sim-
voltage value for gas flow rate over time. (b) Graph comparing the temperature difference (the
ulation result) with the voltage value against the flow rate.
simulation result) with the voltage value against the flow rate.

3.3. MEMS Flow Sensor Test in Continuous Liquid Flow

The sensor’s performance was analyzed in a continuous liquid flow system using a
syringe pump (KDS 200; KD Scientific, Holliston, MA, USA). A 40 mm diameter syringe
with a capacity of 200 mL was mounted on the syringe pump, and a 3 mm diameter tube
Sensors 2021, 21, 8153 9 of 12

3.3. MEMS Flow Sensor Test in Continuous Liquid Flow

The sensor’s performance was analyzed in a continuous liquid flow system using a
syringe pump (KDS 200; KD Scientific, Holliston, MA, USA). A 40 mm diameter syringe
with a capacity of 200 mL was mounted on the syringe pump, and a 3 mm diameter tube
was connected to the syringe. The micro flow sensor was inserted in the middle of the
tube in the perpendicular direction of the flow—the sensor electrodes meet the fluid in
the order of an upstream sensor, the heater, and a downstream sensor—and was sealed
tightly to prevent leakage. A water temperature of 30–40 ◦ C was maintained throughout
the experiment. Figure 8a,b illustrates the experimental setups for analyzing the flow
of a liquid using a syringe pump and an artificial heartbeat model, respectively. The
correlation between the output voltage generated from the sensor within the specific flow
rate range and the fluid flow rate generated by the system was determined by adjusting
the flow rate to achieve the desired speed within the tube containing the sensor. Within
the tube, the flow rate was controlled to between 5 and 30 cm/s, and the sensor’s output
voltage was measured within the controlled flow. The two thermistors and the microheater
were connected to the PCB, which included the Wheatstone bridge circuit, and the output
value was adjusted to zero by controlling the variable resistance inserted in the PCB under
the initial flow. Then, the output voltage generated according to the fluid flow rate was
monitored. It was revealed that as the flow rate increased, the output voltage decreased
accordingly. Figure 9a shows the change in the output voltage of the sensor with varying
flow rates over time. Figure 9b presents a graph showing the output voltage value of
the sensor according to the flow rate of the fluid with the liquid flow simulation rate for
comparison. At over 0.98, the coefficient of determination for the voltage value against
the flow rate revealed a good linearity. Despite maintaining the same flow rate for over
10 s, the voltage value remained constant. The results of this experiment reveal that an
increase in flow rate of 1 cm/s caused an average decrease of 3.6 mV in the output voltage;
furthermore, the average standard deviation of each flow rate section was ±2 mV,
Sensors 2021, 21, x FOR PEER REVIEW 10 ofwhich
12
indicates a sufficient resolution of 2 cm/s or less.

Figure
Figure 8. Schematics
8. Schematics ofof theexperimental
the experimentalsetups
setups for
for (a)
(a) the
the continuous
continuous liquid
liquidflow
flowexperiment
experimentusing a syringe
using pump
a syringe andand
pump
(b) (b)
thethe pulsating
pulsating liquid
liquid flow
flow experimentusing
experiment usingan
anartificial
artificial heartbeat
heartbeat model.
model.

3.4. Results of the Liquid Flow Experiment Using an Artificial Heartbeat Model
To evaluate the performance of the flow sensor in a liquid flow under constant pulsa-
tion (rather than under only a constant flow), this experiment used a cardiovascular model
to simulate the heartbeat movement of a living body. The cardiovascular model comprised
two water reservoirs, a pump, 3 mm diameter tubing (to replicate blood vessels), and a
signal control system. A commercial flow sensor (FD-XS1; Keyence, USA) was inserted
in the center of the tube to measure the real-time flow rate, and the MEMS flow sensor
was equipped with the same line as the commercial sensor. The control system of the
cardiovascular model allowed the beat rate, relative pressure, and stroke to be adjusted to
provide a linear
Figure 9. Results of the liquid flow test using increase
the flexibleinmicrothermal
flow rate with
flow increases in relative
sensor. (a) Graph pressure
showing andvalue
the voltage stroke. The
for liquid flow rate over time. (b) Graph showing the voltage value and the temperature difference (i.e., the simulation
result) against the liquid flow velocity. (c) Graph showing output voltage value of the MEMS flow sensor in accordance
with the flow velocity in the artificial heartbeat model.

3.4. Results of the Liquid Flow Experiment Using an Artificial Heartbeat Model
Sensors 2021, 21, x FOR PEER REVIEW 10 of 12

Sensors 2021, 21, 8153 10 of 12

parameters for this experiment included a beat rate of 40 bpm and an internal flow rate
velocity range of 5–40 cm/s; the data from the MEMS flow sensor were obtained within
this range. Figure 9c shows the output voltage value of the MEMS flow sensor according
to the flow rate produced by the artificial heartbeat. This experiment found that when
the fluid velocity increased by 1 cm/s, the voltage decreased by an average of 0.36 mV.
The average standard deviation in each flow rate range was 0.09 mV. The flexible MEMS
flow sensor fabricated in this study was confirmed to realize a resolution of 0.5 cm/s for a
pulsating liquid flow. The resolution was different between continuous flows and pulsatile
flows in this study, and it is thought that the high level of noise signal in the continuous
Figure 8. Schematics of the flow reducedsetups
experimental the resolution. Table 2 summarizes
for (a) the continuous the performance
liquid flow experiment of the flexible
using a syringe pump andMEMS
flow sensors in the gas and liquid flow tests.
(b) the pulsating liquid flow experiment using an artificial heartbeat model.

Figure 9. Results of the liquid flow test using the flexible microthermal flow sensor.
sensor. (a) Graph showing the voltage value
rate over
for liquid flow rate over time.
time. (b) Graph showing the voltage value and the temperature
temperature difference (i.e.,
(i.e., the
the simulation
simulation
result) against
result) against the
the liquid
liquid flow
flow velocity.
velocity. (c)
(c) Graph
Graph showing
showing output
output voltage
voltage value
value of
of the
the MEMS
MEMS flow
flow sensor
sensor in in accordance
accordance
with the
with the flow
flow velocity
velocity in
in the
the artificial
artificial heartbeat
heartbeat model.
model.

3.4. Results
Table 2. Performance of the Liquid
indicators for gasFlow Experiment
and liquid Usingflexible
flows using an Artificial
MEMSHeartbeat Model
flow sensors.
To evaluate the performance of the flow sensor in a liquid flow under constant pul-
Liquid
sation (ratherGas than under only a constant flow), this experiment used a cardiovascular
model to simulate the heartbeat movement Continuous of aFlow
living body. The Pulsating Flow model
cardiovascular
Temperature (∆T, comprised◦ two water reservoirs, a pump, 3 mm diameter tubing (to replicate blood ves-
0.1 C ↑/10 cm/s 0.2 ◦ C ↓/10 cm/s
simulation) sels), and a signal control system. A commercial flow sensor (FD-XS1; Keyence, USA) was
Voltage inserted in the center
1.2 mV/cm of the tube to3.6
± 0.4 mV measure
mV/cm the real-time
± 3.7 mV flow 0.36
rate,mV/cm
and the± MEMS
0.12 mVflow
Resolution sensor was0.64equipped
cm/s with the same line 1.1 as cm/s
the commercial sensor. The0.5controlcm/s system of
the cardiovascular model allowed the beat rate, relative pressure, and stroke to be ad-
Noise level 0.77 mV 4 mV 0.18 mV
justed to provide a linear increase in flow rate with increases in relative pressure and
Range 0~40 cm/s 0~30 cm/s
stroke. The parameters for this experiment included a beat rate of 40 bpm and 0~40 cm/san internal
Power consumption flow rate velocity range of 5–40 cm/s; the data 40 mW from the MEMS flow sensor were obtained
within this range. Figure 9c shows the output voltage value of the MEMS flow sensor
according to the flow rate produced by the artificial heartbeat. This experiment found that
4. Conclusions
when the fluid velocity increased by 1 cm/s, the voltage decreased by an average of 0.36
mV. This study designed
The average standard and fabricated
deviation in aeach
flexible
flowmicrothermal
rate range was flow0.09
sensor
mV.using MEMS
The flexible
technology. A CFD analysis provided the basis for the design of the
MEMS flow sensor fabricated in this study was confirmed to realize a resolution of 0.5 sensor electrode,
and a sensor module capable of measuring fluid flow signals was fabricated along with
cm/s for a pulsating liquid flow. The resolution was different between continuous flows
an electrical circuit system. LabVIEW software was used for the real-time monitoring
and pulsatile flows in this study, and it is thought that the high level of noise signal in the
of the output voltage signals from the sensor. The fabricated flow sensor module was
continuous flow reduced the resolution. Table 2 summarizes the performance of the flex-
confirmed to measure both liquid and gas flows; furthermore, its successful operation
ible MEMS flow sensors in the gas and liquid flow tests.
under a continuous liquid flow (with a syringe pump) and a pulsating flow (with an
artificial heartbeat model) was verified. This study’s fabricated sensor can operate with
a variety of fluids and can measure a wide range of flow rates. It is anticipated that the
flexible micro flow sensor technology developed in this study can be applied in various
industrial fields. Although some adaptations to its electrical packaging are required, the
Sensors 2021, 21, 8153 11 of 12

sensor has significant potential for application in healthcare fields, particularly for the
measurement of biosignals, including respiration and the flows of body fluids.

Author Contributions: Conceptualization, S.-H.L.; methodology, M.-O.C.; software, M.-O.C. and

W.J.; validation, M.-O.C. and W.J.; formal analysis, M.-O.C.; investigation, M.-O.C. and W.J.; re-
sources, S.-H.L.; writing—original draft preparation, M.-O.C.; writing—review and editing, S.-H.L.;
visualization, M.-O.C. and W.J.; supervision, S.-H.L.; project administration, S.-H.L. All authors have
read and agreed to the published version of the manuscript.
Funding: This material is based upon work supported by the Ministry of Trade, Industry and Energy
(MOTIE, Korea), the Ministry of Science and ICT (MSIT, Korea), and the Ministry of Health and
Welfare (MOHW, Korea) under the Technology Development Program for Ai-Bio-Robot-Medicine
Convergence (20001704).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Please contact the corresponding author for data requests.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Ciuti, G.; Ricotti, L.; Menciassi, A.; Dario, P. MEMS sensor technologies for human centred applications in healthcare, physical
activities, safety and environmental sensing: A review on research activities in Italy. Sensors 2015, 15, 6441–6468. [CrossRef]
[PubMed]
2. Idjeri, B.; Laghrouche, M.; Boussey, J. Wind Measurement Based on MEMS Micro-Anemometer with High Accuracy Using ANN
Technique. IEEE Sens. J. 2017, 17, 4181–4188. [CrossRef]
3. Versasi, M.; Jannelli, A.; Angiulli, G. Electrostatic Micro-Electro-Mechanical-Systems(MEMS) Devices: A Comparison Among
Numerical Techniques for Recovering the Membrane Profile. IEEE Access 2020, 8, 125874–125886. [CrossRef]
4. Tsuchiya, T.; Lee, G.-B.; Institute of Electrical and Electronics Engineers. Integration of catheter flow sensor onto tracheal
intubation tube system. In Proceedings of the IEEE 26th International Conference on Micro Electro Mechanical Systems (MEMS
2013), Taipei, Taiwan, 20–24 January 2013.
5. Kagie, A.; Bishop, D.K.; Burdick, J.; La Belle, J.T.; Dymond, R.; Felder, R.; Wang, J. Flexible rolled thick-film miniaturized flow-cell
for minimally invasive amperometric sensing. Electroanalysis 2008, 20, 1610–1614. [CrossRef]
6. Li, C.; Wu, P.; Wu, Z.; Mehan, N.; Bhattacharjee, N.; Hartings, J.A.; Ahn, C.H.; Narayan, R.K. Highly accurate thin-film flexible
microsensor for continuous and quantitative measurement of cerebral blood flow. Biomed. Microdevices 2015, 17, 87. [CrossRef]
[PubMed]
7. Ejeian, F.; Azadi, S.; Razmjou, A.; Orooji, Y.; Kottapalli, A.; Warkiani, M.E.; Asadnia, M. Design and applications of MEMS flow
sensors: A review. Sens. Actuators A Phys. 2019, 295, 483–502. [CrossRef]
8. Kuo, J.T.W.; Yu, L.; Meng, E. Micromachined thermal flow sensors—A review. Micromachines 2012, 3, 550–573. [CrossRef]
9. Meng, E.; Li, P.Y.; Tai, Y.C. A biocompatible Parylene thermal flow sensing array. Sens. Actuators A Phys. 2008, 144, 18–28.
[CrossRef]
10. De Luca, A.; Udrea, F. CMOS MEMS Hot-Film Thermoelectronic Flow Sensor. IEEE Sens. Lett. 2017, 1, 1–4. [CrossRef]
11. Yoon, H.J.; Kim, S.Y.; Lee, S.W.; Yang, S.S. Fabrication of a Micro Electromagnetic Flow Sensor for Micro Flow Rate Measurement.
In Proceedings of the SPIE’s 7th Annual International Symposium on Smart Structures and Materials, Newport Beach, CA, USA,
6–9 March 2000. [CrossRef]
12. Ashauer, M.; Glosch, H.; Hedrich, F.; Hey, N.; Sandmaier, H.; Lang, W. Thermal flow sensor for liquids and gases based on
combinations of two principles. Sens. Actuators A Phys. 1999, 73, 7–13. [CrossRef]
13. Dijkstra, M.; de Boer, M.J.; Berenschot, J.W.; Lammerink, T.S.J.; Wiegerink, R.J.; Elwenspoek, M. Miniaturized thermal flow sensor
with planar-integrated sensor structures on semicircular surface channels. Sens. Actuators A Phys. 2008, 143, 1–6. [CrossRef]
14. Ahmed, M.; Xu, W.; Mohamad, S.; Boussaid, F.; Lee, Y.K.; Bermak, A. Fully Integrated Bidirectional CMOS-MEMS Flow Sensor
with Low Power Pulse Operation. IEEE Sens. J. 2019, 19, 3415–3424. [CrossRef]
15. Xu, W.; Lijin, B.; Duan, M.; Wang, X.; Wicaksana, J.; Min, A.; Ahmed, M.; Wang, R.; Fang, N.X.; Bermak, A.; et al. A wireless
dual-mode micro thermal flow sensor system with extended flow range by using 0.18 µm Cmos-MEMS process. In Proceedings
of the 2018 IEEE Micro Electro Mechanical Systems (MEMS), Belfast, UK, 21–25 January 2018. [CrossRef]
16. Mistry, K.K.; Mahapatra, A. Design and simulation of a thermo transfer type MEMS based micro flow sensor for arterial blood
flow measurement. Microsyst. Technol. 2012, 18, 683–692. [CrossRef]
17. Li, C.; Wu, P.M.; Hartings, J.A.; Wu, Z.; Ahn, C.H.; LeDoux, D.; Shutter, L.A.; Narayan, R.K. Smart catheter flow sensor for
real-time continuous regional cerebral blood flow monitoring. Appl. Phys. Lett. 2011, 99, 233705. [CrossRef]
Sensors 2021, 21, 8153 12 of 12

18. Jiang, P.; Zhao, S.; Zhu, R. Smart sensing strip using monolithically integrated flexible flow sensor for noninvasively monitoring
respiratory flow. Sensors 2015, 15, 31738–31750. [CrossRef] [PubMed]
19. Okihara, C.; Hasegawa, Y.; Shikida, M.; Matsushima, M.; Kawabe, T. Development of cylinder hollow structure with flow
sensor by film transfer technology. In Proceedings of the 2016 IEEE SENSORS, Orlando, FL, USA, 30 October–2 November 2016.
[CrossRef]
20. Hedrich, F.; Kliche, K.; Storz, M.; Billat, S.; Ashauer, M.; Zengerle, R. Thermal flow sensors for MEMS spirometric devices. in Sens.
Actuators A Phys. 2010, 162, 373–378. [CrossRef]
21. Lee, J.; Kim, T.G.; Kim, D.; Kim, J.; Lim, S.H. Improvements in the performance of a microthermal flow sensor using asymmetrically
located temperature sensors. Int. J. Precis. Eng. Manuf. 2017, 18, 227–231. [CrossRef]
22. di Pancrazio, A.; Bruschi, P.; Piotto, M. Low power, MEMS liquid flow sensor with silicone coating electrical insulation. In
Proceedings of the 2014 Symposium on Design, Test, Integration and Packaging of MEMS/MOEMS (DTIP), Cannes, France,
1–4 April 2014. [CrossRef]