[go: up one dir, main page]

0% found this document useful (0 votes)
71 views9 pages

Shikida2017 Article FabricationAndFlow-sensorAppli

Uploaded by

yohannes
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
71 views9 pages

Shikida2017 Article FabricationAndFlow-sensorAppli

Uploaded by

yohannes
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 9

Microsyst Technol (2017) 23:677–685

DOI 10.1007/s00542-015-2704-3

TECHNICAL PAPER

Fabrication and flow‑sensor application of flexible thermal


MEMS device based on Cu on polyimide substrate
Mitsuhiro Shikida1 · Yosuke Niimi2 · Shunji Shibata3 

Received: 30 September 2015 / Accepted: 6 October 2015 / Published online: 17 October 2015
© Springer-Verlag Berlin Heidelberg 2015

Abstract  A Cu on polyimide (COP) substrate was pro- 1 Introduction


posed as a MEMS material, and the fabrication process for
a flexible thermal MEMS sensor was developed. The COP Micro-electromechanical systems (MEMS) technologies
substrate application to MEMS devices has the advantage can miniaturize the sensor devices to less than millimeter
that typical MEMS structures fabricated in a SOI wafer in in size for producing high-value-added systems, and vari-
the past—such as a diaphragm, a beam, a heater formed on ous types of physical MEMS sensors have therefore been
a diaphragm—can also be easily produced in the COP sub- commercialized. Two kinds of structures, diaphragms and
strate in the flexible fashion. These structures can be used beams, are normally used as mechanical sensing elements
as the sensing element in various physical sensors, such as in MEMS sensors. The former structure is mostly used in
flow, acceleration, and shear stress sensors. A flexible ther- pressure and force measurements, and these physical val-
mal MEMS sensor was produced by using a lift-off pro- ues are reflected by deformation of the diaphragm. The
cess and sacrificial etching of a copper layer on the COP deformation values, in turn, are generally converted into
substrate. A metal film working as a flow sensing element electrical signals reflecting deformation-induced changes in
was formed on a thin polyimide membrane produced by the capacitance or resistance. When a beam structure is used
sacrificial etching. The fabricated flexible thermal MEMS in force, mass, or acceleration measurements, the cantilever
sensor was used as a flow sensor, and its characteristics is frequently operated at its resonance frequency. MEMS
were evaluated. The obtained sensor output versus the flow technologies can increase the resonance frequency by min-
rate curve closely matched the approximate curve derived iaturizing the resonators and thus can be used to make pro-
using King’s law. The rising and falling response times duces for atomic force microscopes and chemical substance
obtained were 0.50 and 0.67 s, respectively. sensors. A double-supported beam is also used as a main
part of a comb drive electrostatic actuator for the MEMS
acceleration sensor.
A heater formed on a thermally isolated diaphragm also
has been used in MEMS sensors, and thermal convective
phenomena have been widely used in MEMS systems as
flow-sensing mechanisms (Meijer and Herwaarden 1994;
Trautweiler 1997), shear stress sensors, and accelera-
* Mitsuhiro Shikida tion sensors. Three different thermal principles—thermal
shikida@hiroshima‑cu.ac.jp
anemometry, calorimetric sensing, and time-of-flight sens-
1
Department of Frontier Science, Hiroshima City University, ing—are used in MEMS sensors (Elwnspoek and Wieg-
Hiroshima, Japan erink 2001; Gianchandani et al. 2008). Thermal anemom-
2
Department of Micro‑Nano Systems Engineering, Nagoya etry is used to detect shear stress acting on the surface of
University, Nagoya, Japan an object moving in fluid (Ho and Tai 1998; Lin et al. 2005;
3
Department of Mechanical Science Engineering, Nagoya Liu et al. 1999; Sawada et al. 2014). The amount of shear
University, Nagoya, Japan

13

678 Microsyst Technol (2017) 23:677–685

stress caused by the friction between the fluid and the mov-
ing object—for example, between air and an airplane, auto-
mobile, or bullet train—strongly affects mechanical energy
loss. On the other hand, calorimetric sensing is applied to
detect the inertia of the air (Courteaud et al. 2008; Daud-
erstät et al. 1995; Mally et al. 2003). A strong advantage
of the acceleration sensor based on the above thermal
sensing mechanism is that it does not require any moving
solid proof mass in its sensing. Thus, it can achieve high
mechanical robustness. Because of low performance at
response speed, however, its frequency band is not broad.
Normally, silicon is used as a material for fabricating
MEMS sensor devices because of its compatibility with
MEMS manufacturing processes. Thus, MEMS sensor
structures such as diaphragms, beams, and heaters on dia-
phragms are normally fabricated on silicon or silicon-on-
insulator (SOI) wafers. Generally, Si-MEMS sensors have
excellent space and time resolutions because of silicon’s
high Young’ modulus, and thus they have been used in vari-
ous industrial applications. However, it is difficult for them
to be used on a curved surface in a flexible configuration
because of their brittleness. Consequently, it is difficult to
incorporate Si- MEMS sensors in human interfaces and
wearable sensor devices.
To overcome this problem, polymer materials, such as
polyimide, parylene, and silicone resin, have been used
as substrate materials in the fabrication of flexible MEMS Fig. 1  Schematic and exploded views of flexible thermal MEMS
sensors (Shikida et al. 2012, 2013). MEMS technologies, sensor based on COP substrate
for example, photolithography, metal film deposition, and
wet- and dry-etching, are also used to produce miniatur-
ized mechanical structures in these polymer materials. An that have in the past been fabricated on a SOI wafer, can
artificial hollow fiber was proposed as a MEMS material also be fabricated easily on a COP substrate.
for producing a fabric tactile sensor for part of a wearable Schematic and exploded views of a flexible thermal
sensing system (Hasegawa et al. 2008; Kita et al. 2010). MEMS sensor based on the COP substrate are shown in
Moreover, a copper-on-polyimide (COP) substrate was Fig.  1a, b. A metal heater pattern working as the MEMS
introduced as a starting material for fabricating flexible sensor is formed on the thin polyimide film membrane, and
MEMS sensors (Niimi et al. 2014), and it was used to make it is thermally isolated by the cavity to shorten the response
a thermal-flow sensor (Shibata et al. 2014). In this paper time. A cavity and an electrical feedthrough are produced
the fabrication process and flow sensing performance of the by selectively etching part of the copper layer (sacrificial
flexible thermal MEMS sensor based on the COP substrate etching). The thermal MEMS sensor based on COP sub-
are described in detail. strate has the flowing advantages.

1. It is flexible and thus can be easily mounted on a


2 Flexible thermal MEMS sensor curved surface.
2. The electrical feedthrough connecting the sensing pads
2.1 Sensor design to electrical pads for the external connection can be
monolithically fabricated with a low resistance because
Like a SOI wafer, a COP substrate with polyimide layer part of the substrate’s copper layer works as the electri-
consists of three layers: a base polyimide film, a copper cal feedthrough formed by the sacrificial etching.
layer, and a thin polyimide layer. Thus, the usage of the 3. It can be used as the sensing element in flow, accelera-
COP substrate with a polyimide layer has the advantage tion, and shear stress sensors. And if by thin film depo-
that the typical MEMS-structures, such as the diaphragm, sition after the cavity formation we seal the holes in the
beam, and heater formed on thermally isolated diaphragm

13
Microsyst Technol (2017) 23:677–685 679

thin polyimide layer that were formed for the sacrifi-


cial etching, it can also be used for pressure sensing.

2.2 Operation principle

Three different operation principles—thermal anemometry,


calorimetric, and time-of-flight—have been used in MEMS
thermal sensors (Elwnspoek and Wiegerink 2001; Gian-
chandani et al. 2008). To use the flexible thermal MEMS
sensor as an airflow sensor, the thermal anemometry mech-
anism was chosen in this study because it has a wide range
of flow rate sensing and the simplest structure. The sensor
measured the flow velocity by using the cooling effect of
forced convection heat transfer to the airflow at the heater
formed on the diaphragm. When the heater working as the
flow velocity sensor is operated by a constant temperature
circuit having a feedback control to shorten the response
time, the relationship between the flow velocity and electri-
cal energy supplied to the heater under the forced convec-
tion heat transfer condition can be expressed as
P
= G0 + Kun (1)
∆T
where P is the electrical energy supplied to the heater, ΔT
is the temperature difference between the heater and the air
(Shibata et al. 2014). Here, G0, K, and n are constants, and
u is the average flow velocity. The equation is translated, by
using the applied voltage to the heater, as

V 2 = a + bun (2)
where V is the applied voltage to the heater (Shibata et al.
2014). Here, a, b, and n are constants depending on the
heater structures. This equation was originally derived by
Fig. 2  Fabrication sequence for thermal MEMS sensors based on
King (King 1914), and detailed information on the con- COP substrates
stants is described in the same reference. The square of
the voltage at the heater is proportional to the nth power
of flow velocity under the forced convection heat transfer
condition.

3 Fabrication

The fabrication process for the flexible thermal MEMS


sensor based on the COP substrate consists of the following
steps:

(a) Preparation of the COP substrate (Fig. 2a)


(b) Deposition and patterning of a photosensitive polyim-
ide film (Fig. 2b)
(c) Metal film deposition and patterning by the lift-off pro-
cess (Fig. 2c, d)
Fig. 3  Surface roughness of copper in COP substrate measured by
(d) Cavity and feedthrough formation by sacrificial etch-
AFM
ing of the copper layer (Fig. 2e, f)

13

680 Microsyst Technol (2017) 23:677–685

Fig. 4  Surface roughness
change of polyimide film by
oxygen plasma etching

3.1 Roughness of copper layer in COP substrate in order to determine how thick the photosensitive poly-
imide should be. The obtained AFM image is shown in
A thin photosensitive polyimide film serves as the mem- Fig. 3. The arithmetic mean roughness and the maximum
brane supporting the heater in thermal MEMS sensors. height roughness calculated from the AFM image were
A liquid of polyimide is coated on the copper layer in 0.12 and 1.02 μm, respectively. Thus, any photosensitive
a COP substrate by a spinner, and then it is solidified polyimide thickness >2.0 μm was thought to be sufficient
by heat. The surface roughness of the copper layer was for completely covering the tiny roughness of the copper
therefore evaluated by atomic force microscopy (AFM) surface.

13
Microsyst Technol (2017) 23:677–685 681

Fig. 7  Etching characteristics in case of single-line mask pattern


Fig. 5  Relationship between surface roughness of polyimide film and
plasma etching time

3.2 Polyimide surface treatment by plasma

The metal film working as the heater was obtained by


depositing a Cr/Au film <1.0 μm thick on the chemi-
cally inactive polyimide film surface. Thus, the polyimide
surface after its solidification was roughened by oxygen
plasma etching to increase the physical adhesion between
the polyimide film and the metal film. The roughness
dependency on the etching time was evaluated experimen-
tally. The polyimide film was etched by the oxygen plasma
at 100 W with gas flow of 100 cm3/min. The etched sur-
face was measured by AFM, and surface roughness images
obtained (Fig. 4) show that the surface was uniformly
roughen at each etching time. As shown in Fig. 5, both
the arithmetic mean roughness and the maximum height
roughness calculated from AFM images increased linearly
with etching time. The arithmetic mean roughness of the
polyimide film before etching was 0.5 nm, and it increased
to 24 nm with 5-min etching. The maximum height rough-
ness was 150 nm when the etching time was 5.0-min. Thus,
the etching time was chosen as 3.0-min by considering the
metal film thickness of 250 nm to cover the roughened pol-
Fig. 6  Mask patterns for copper wet etching, and side etching of line yimide surface.
pattern

13

682 Microsyst Technol (2017) 23:677–685

Fig. 8  Etching characteristics in case of double-line and hole mask


patterns

3.3 Sacrificial etching of copper layer for cavity


and feedthrough formation
Fig. 9  Designed flexible thermal MEMS sensor
Both the cavity for thermal isolation and the electrical
feedthrough structures were formed by selectively etching
part of the copper layer based on the photosensitive pol- pattern are shown in Fig. 7. The etching amounts in the
yimide pattern working as the etching mask in the same depth and side directions increased linearly with etching
process (sacrificial etching). An iron(II) chloride solu- time. Thus, the etching rates in the depth and side direc-
tion was used as the copper etching solution. The etch- tions were estimated to 13 and 11 μm/min, respectively.
ing is dominated by the diffusion-limited reaction, and The etching properties in the case of the double-line and
as a result, copper layer is isotropically etched according hole patterns are shown in Fig. 8. Two different gaps in
to the mask pattern. In the case of the isotropic wet etch- between lines were used, of 10 and 50 μm, and the diam-
ing, the etching reaction strongly depends on the mask eter of the hole was 10 μm. The following results are
pattern because the etching depends on the solution dif- obtained.
fusion. Thus, the etching characteristics obtained with
different mask patterns were investigated for determin- (a) The etching amounts in the depth and side directions
ing the shapes of the cavity and feedthrough into the cop- strongly depended on the mask pattern and etching
per layer. The three different mask patterns used in the time.
experiments—a single line, a double line, and a hole—are (b) The etching rate depended on the width of the gap
shown in Fig. 6a. The amounts of the etching depth in ver- between the lines and was higher for the larger gap.
tical direction and the side etching under the mask, shown (c) The etching rates in the depth and side directions
in Fig. 6b. were measured by an optical microscope. Both were lower with the hole pattern than they were with
of the etching characteristics in the case of the single-line the line patterns.

13
Microsyst Technol (2017) 23:677–685 683

Fig. 11  Fabricated flexible thermal MEMS sensor

(Photoneece; PW-1500, 1500 cp, Toray Industries, Inc.)


was coated and patterned to define a structure in the cop-
per layer (Fig. 10b). The patterned polyimide film 5.0 μm
thick worked as an etching mask for forming the cavity and
the feedthrough in the copper layer in the final sacrificial
Cu etching, and it also became a membrane structure. Part
of the patterned polyimide film was used as a connection
area between the deposited metal film and the electrical
feedthrough in the copper layer. The metal film became
the heater for sensor and the electrical wiring connecting
Fig. 10  Detail fabrication process for flexible thermal MEMS sensor between the heater and the electrical feedthrough structures
in the copper layer. Thanks to the forward tapered shape at
3.4 Overview of fabrication the cross-section in the patterned polyimide film, the metal
film covered the sidewall surface of the tapered shape area
A flexible thermal MEMS sensor based on the above three in the polyimide film completely. The patterned polyim-
considerations was designed (Fig. 9) and was fabricated on ide film surface was roughened by 3.0-min oxygen plasma
a COP substrate (Fig. 10). As shown in Fig. 9, in this study etching (input power 100 W, flow rate 100 cm3/min) to
the hole patterns were used in the fabrication of the cavities increase the adhesion between the polyimide and metal
and the electrical feedthroughs. The overview of its fabrica- film with the anchor effect.
tion process is as follows. A negative photoresist (ZPN1150-90, Zeon Corporation)
A commercially available COP substrate (Ube Exsymo developed for the lift-off process, was coated on the poly-
Co., Ltd.) was used as the starting material. The thick- imide film surface and patterned with UV light to define the
nesses of the copper and polyimide were 18 and 50 μm, shapes of the metal films (Fig. 10c). After its patterning, all
respectively (Fig. 10a). The photosensitive polyimide the surfaces were exposed for 1.0 min to oxygen plasma

13

684 Microsyst Technol (2017) 23:677–685

Fig. 12  Temperature coefficient of resistance in fabricated flexible


thermal MEMS sensor

Fig. 14  Response waveform of fabricated flexible thermal MEMS


sensor

Fig. 13  Relationship between sensor output and flow rate at fabri-


cated flexible thermal MEMS sensor A fabricated flexible thermal MEMS sensor is shown in
Fig. 11. The sensing metal on the polyimide thin membrane
was successfully fabricated on the COP substrate. The
(input power 50 W, flow rate 50 cm3/min) to remove the cavity and electrical feedthoughs were also formed by the
residual organic substance on the polyimide and the copper selective etching of the Cu layer. The heater located in the
surface. The Au/Cu film was deposited by sputtering and center of the three metal lines was designed as the flow sen-
patterned by selectively removing the photoresist (lift-off sor. Both sides of the metal patterns were designed as the
process) (Fig. 10d). The thicknesses of the Au and Cr were temperature sensors in the accelerometer application.
250 and 10 nm, respectively.
A positive photoresist (OFPR800-100, Tokyo Ohka
Kogyo, Ltd.) was applied and patterned on the metal film 4 Flow‑sensor application
to serve as a protection layer during the following Cu
etching (Fig. 10e). Finally, both the cavities and electrical To use the fabricated flexible thermal MEMS sensor for
feedthroughs were formed by selectively etching part of the gas flow sensing, the temperature coefficient of resistance
copper layer based on the polyimide mask pattern (sacri- (TCR) was evaluated experimentally. As shown in Fig. 12,
ficial etching). An iron(II) chloride solution (Sunhayato the electrical resistance increased linearly with tempera-
Corp.) was used as an etchant. The etching temperature was ture. A TCR of 0.0019/K was obtained from the gradient of
set to 40 °C. After the 8.0-min etching, the metal patterns the obtained graph.
working as the sensor formed on polyimide membrane and The flow sensing characteristics of the flexible thermal
the electrical feedthrough were obtained on the COP sub- MEMS sensor were evaluated experimentally. Compressed
strate (Fig. 10f). air was used as a gas source, and its flow rate was regulated
by using a mass flow controller (KOFLOC, Model 3200,

13
Microsyst Technol (2017) 23:677–685 685

Kojima Instruments, Inc.). A heater for detecting the flow References


rate was driven by a constant-temperature circuit to shorten
the response time. First the relationship between the sensor Courteaud J, Crespy N, Combette P, Sorli B, Giani A (2008) Studies
output and the flow rate was evaluated because it works as a and optimization of the frequency response of a micromachined
thermal accelerometer. Sens Actuator A 147:75–85
calibration curve when a flow rate is measured. As shown in Dauderstät UA, de Vries PHS, Hiratsuka R, Sarro PM (1995) Sili-
Fig. 13, we confirmed that the obtained graph closely matched con accelerometer based on thermopiles. Sens Actuator A
the approximate curve derived using the King’s law theoreti- 46:201–204
cal model (King 1914). Then a response time was evaluated Elwnspoek M, Wiegerink R (2001) Mechanical microsensors.
Springer, Berlin
as a dynamic property. A three-port solenoid valve was placed Gianchandani Y, Tabata O, Zappe H (2008) Comprehensive MEMS
between the sensor and mass flow controller to change the gas (flow sensor). Elsevier, Amsterdam, pp 209–272
flow direction. Outlet port 1 was used for measuring the sen- Hasegawa Y, Shikida M, Ogura D, Suzuki Y, Sato K (2008) Fabrica-
tion of a wearable fabric tactile sensor produced by artificial hol-
sor response time, and outlet port 2 as the gas exhaust port.
low fiber. J. Micromech Microeng 18:085014
The response time was measured by switching the connec- Ho CM, Tai YC (1998) Micro-electro-mechanical-systems (MEMS)
tion tube from port 2 to port 1. An air flow of 300 cm3/min and fluid flow. Annu Rev Fluid Mech 30:579–612
was input to the sensor. The response waveforms of the sen- King LV (1914) On the convection of heat from small cylinders in
a stream of fluid: determination of the convection constants of
sor are shown in Fig. 14. The rising and falling response times
small platinum wires with applications to hot-wire anemometry.
obtained were 0.50 and 0.67 s, respectively. The difference Philos Trans R Soc Lond Ser A Contain Pap Math Phys Char
between these times was due to the different heating and cool- 214:373–432
ing mechanisms. In the case of the flow increase, the heater Kita G, Shikida M, Suzuki Y, Tsuji Y, Sato K (2010) Large-sized fab-
ric tactile sensors for detecting contacted objects. Micro Nano
temperature was actively controlled to be a constant by the
Lett 5(6):389–392. doi:10.1049/mnl.2010.0130
electrical circuit. However, the heater was passively cooled by Lin Q, Xu Y, Jiang F, Tai TC (2005) A parametrized three-dimen-
heat dissipation. Thus, the response time for the cooling mode sional model for MEMS thermal shear-stress sensors. J Micro-
was a little longer than that for the heating mode. electromech Syst 14(3):625–633
Liu C, Huang JB, Zhu Z, Jiang F, Tung S, Tai YC, Ho CM (1999) A
Micromachined flow shear-stress sensor based on thermal trans-
fer principle. J Microelectromech Syst 8(1):90–99
5 Conclusion Mally F, Giani A, Martinez A, Bonnot R, Temple-Boyer P, Boyer A
(2003) Micromachined thermal accelerometer. Sens Actuator A
103:359–363
We proposed a Cu on polyimide (COP) substrate as a start-
Meijer GCM, Herwaarden AW (1994) Thermal sensor. IOP publish-
ing material for producing flexible thermal MEMS sensors ing Ltd, Bristol
and, using it, obtained the following results. Niimi Y, Shibata S, Shikida M (2014) Polymer micromachining
based on Cu on polyimide substrate and its application to flex-
ible MEMS sensor. In: Technical digests of IEEE micro electro
1. The fabrication process for a flexible thermal MEMS
mechanical systems conference, pp 528–531
sensor was developed. The sensor was made by using Sawada T, Terashima O, Sakai Y, Nagata K, Shikida M, Ito Y, Mura-
the lift-off process and sacrificial etching of the copper matsu S (2014) On the measurement of wall shear stress with a
layer on a COP substrate. An advantage of the process patch-type micro-fabricated hot-film sensor. J Fluid Sci Technol.
doi:10.1299/jfst.2014jfst0029
has that both the cavity and feedthrough structures in
Shibata S, Niimi Y, Shikida M (2014) Flexible thermal MEMS flow
the copper layer can be made in the same fabrication sensor based on Cu on polyimide substrate. Proc IEEE Sens
step. 2014:424–427
2. The fabricated flexible thermal MEMS sensor was used Shikida M, Yoshikawa K, Iwai S, Sato K (2012) A MEMS flow sensor
applied in a variable-air-volume unit in a building air-condition-
as a flow sensor, and its characteristics were evaluated
ing system. Sens Actuator A 188:2–8
experimentally. The obtained sensor output versus Shikida M, Yamazaki Y, Yoshikawa K, Sato K (2013) Flexible flow
the flow rate curve closely matched the approximate sensor for large scale air-conditioning network systems. Sens
curve derived using King’s law. The rising and falling Actuator A 189:212–217
Trautweiler SF (1997) Silicon hot film flow sensor. Physical electron-
response times were 0.50 and 0.67 s, respectively.
ics Laboratory, Swiss Federal Institute of Technology, Zurich.
ISBN 3-907574-18-4
Acknowledgments  This research was supported by The Canon
Foundation and Grant-in-Aid for Scientific Research (B) No.
26286034 from the Ministry of Education, Culture, Sports, Science
and Technology (MEXT), Japan.

13

You might also like