**1. Introduction**

Micro‐electro‐mechanical systems (MEMS) technology has opened the doorway to miniatur‐ izing and integrating various mechanical and electrical components on the same chip. The principal fabrication technologies for producing MEMS devices, such as anisotropic wet etching for single crystal silicon, sacrificial etching for producing mechanical elements, and anodic bonding for the packaging, were invented in the 1960s. These technologies have been used for producing various miniaturized sensors from the 1970s onwards. Silicon diaphragm structures produced by anisotropic wet etching are used as sensing elements for pressure sensors. An ion sensitive field‐effective‐transistor was developed as a chemical sensor. MEMS sensors started to be integrated on silicon wafers in the middle of the 1970s. A pressure sensor and its electrical circuits are integrated on the same silicon waferin orderto reduce the parasitic capacitance and footprint. During this time, a miniaturized gas chromatograph for use in aerospace was fabricated on a 2‐inch silicon wafer; it integrated a column, valve seat, and sensing heater. Moreover, a mass‐flow controller driven by piezo‐electric actuators for precisely regulating gas flow in a pipe was developed in the late 1980s. Since the 1990s, MEMS technologies have spread to many different fields, and these days there are miniaturized sensors, for example, for acceleration, gyroscopic, flow, infrared, and tactile sensors.

## **1.1. Attractive features of MEMS sensors**

MEMS technologies can produce miniaturized sensors a few mm in size that offerthe following very useful features for industry.


© 2013 Shikida; licensee InTech. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2013 Shikida; licensee InTech. This is a paper distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

thermal sensor applications. This is because the heat capacity decreases as the sensor volume decreases.

**e.** Compatibility of specialized fabrication processes, such those described in (a) to (c), with semiconductor electrical circuit fabrication processes has to be carefully

MEMS Sensors and Their Applications 333

**a.** Mechanics for designing an adequate deformation vs. force relation and ensuring

**c.** Electrical circuit analysis for designing the sensing circuits and electrical properties

The multi‐physics approach is taken because the above points are related to each

**4.** Scaling effects: The sizes of structural elements such as cantilevers and diaphragms are on the order of 10 micrometers. Such a level of miniaturization means that forces that are proportional to the area of the structural element dominate the motion of the sensor. Other forces, such as magnetic and inertial forces which are proportional to the volume of elements, are greatly reduced as a result of miniaturization. The effects of the electrostatic force and surface tension have to be carefully considered when designing MEMS sensors.

MEMS technologies have a wide variety of applications, and there are many types (examples are shown in Table 1). In the following sections, tactile sensors for human interfaces and flow

The single crystal silicon is often used as a MEMS material because it is a high‐quality semiconductor with excellent mechanical properties [1]. It is said that the yield strength of silicon is comparable to that of stainless steel. Silicon sensors for gyroscopes and for measuring pressure, acceleration, flow, and inclination have already been commercialized and are being used in the automotive and information industries [2‐5]. Silicon has also been used to make miniaturized tactile sensors as interface devices in robotic automation systems since in the middle of the 1980s. Various tactile sensors based on silicon MEMS technologies have been developed [6‐9]. In particular, Takao et al. integrated a tactile sensor with a CMOS device on the same chip [10]. Moreover, Shikida et al. proposed to integrate an actuator mechanism onto

sensors for air‐conditioning systems and medical applications will be focused.

a tactile sensor as means of measuring the hardness of objects [11‐13].

considered when MEMS sensors are to be integrated with electrical circuits.

**3.** Multi‐disciplinary know‐how: MEMS sensors' development requires multi‐physics knowledge. For example, developers of acceleration sensors need a thorough knowledge

**b.** Vibration analysis for designing the resonant frequency of the device,

(capacitance and resistance) of the sensing structures, and

**d.** Fabrication technology for producing the sensors.

of the following areas:

reliability,

other.

**1.3. Applications of MEMS sensors**

**2. Tactile sensors for human interfaces**


#### **1.2. Difficulties in developing MEMS sensors**

Despite the above‐mentioned advantages, MEMS sensors face a number of difficulties in regard to their continued development.

	- **a.** Sacrificial etching is used to make movable elements on a silicon wafer. Parts of the deposited layers are selectively etched away. This etching is normally done by isotropic wet etching, and it is especially used to produce comb‐drive actuators. Sticking between the released movable element and the substrate underneath it is sometimes a problem, but it can be solved by using a specialized drying method and thin film deposition.
	- **b.** Anisotropic wet etching is used to produce diaphragm and cantilever structures. The micro‐structures are formed by orientation‐dependent etching of single crystal silicon.
	- **c.** Dry etching is normally used to make comb‐drive actuators. To produce the designed structure precisely, one has to overcome loading and notch effects.
	- **d.** Variation in packaging is an issue since each MEMS sensor requires an order‐made packaging process. For instance, tactile force and flow velocity sensors have to be exposed to their environment in order to detect physical and chemical quantities of interest. On the other hand, acceleration and gyro sensors are completely encapsu‐ lated in a shell structure. The pressure inside the shell is kept at a vacuum level in order to increase the Q‐factor at the resonant frequency.
	- **a.** Mechanics for designing an adequate deformation vs. force relation and ensuring reliability,
	- **b.** Vibration analysis for designing the resonant frequency of the device,
	- **c.** Electrical circuit analysis for designing the sensing circuits and electrical properties (capacitance and resistance) of the sensing structures, and
	- **d.** Fabrication technology for producing the sensors.

The multi‐physics approach is taken because the above points are related to each other.

**4.** Scaling effects: The sizes of structural elements such as cantilevers and diaphragms are on the order of 10 micrometers. Such a level of miniaturization means that forces that are proportional to the area of the structural element dominate the motion of the sensor. Other forces, such as magnetic and inertial forces which are proportional to the volume of elements, are greatly reduced as a result of miniaturization. The effects of the electrostatic force and surface tension have to be carefully considered when designing MEMS sensors.

#### **1.3. Applications of MEMS sensors**

thermal sensor applications. This is because the heat capacity decreases as the sensor

**3.** Small parasitic capacitance and dead space: MEMS sensors can be integrated with electrical circuits on the same silicon wafer, and thus, one can reduce the parasitic capacitance in the sensing circuit. MEMS technologies can integrate different components, for example, sensors and actuators, on the same wafer to build up the system, and thus,

Micro-Nano Mechatronics — New Trends in Material, Measurement, Control, Manufacturing and Their Applications in

**4.** 2D analysis: MEMS sensors can be integrated on silicon wafer in the form of an array, and thus, one can analyze the distribution of the physical quantities, for example, pressure

Despite the above‐mentioned advantages, MEMS sensors face a number of difficulties in

**1.** Large capital investment: The MEMS technologies, especially in those used in the fabrication process, are based on semiconductor technologies that have been developed for producing 3D microstructures from silicon wafers. A large amount of capital invest‐

**2.** Variations in the fabrication process: MEMS sensors typically have cantilever, diaphragm, or comb structures. They are suspended and actuated to detect physical quantities. Many

**a.** Sacrificial etching is used to make movable elements on a silicon wafer. Parts of the deposited layers are selectively etched away. This etching is normally done by isotropic wet etching, and it is especially used to produce comb‐drive actuators. Sticking between the released movable element and the substrate underneath it is sometimes a problem, but it can be solved by using a specialized drying method and

**b.** Anisotropic wet etching is used to produce diaphragm and cantilever structures. The micro‐structures are formed by orientation‐dependent etching of single crystal

**c.** Dry etching is normally used to make comb‐drive actuators. To produce the designed

**d.** Variation in packaging is an issue since each MEMS sensor requires an order‐made packaging process. For instance, tactile force and flow velocity sensors have to be exposed to their environment in order to detect physical and chemical quantities of interest. On the other hand, acceleration and gyro sensors are completely encapsu‐ lated in a shell structure. The pressure inside the shell is kept at a vacuum level in

structure precisely, one has to overcome loading and notch effects.

order to increase the Q‐factor at the resonant frequency.

they make it possible to reduce the dead space in the system.

ment is generally required in the fabrication of MEMS sensors.

technologies are required for producing such structures.

and force, acting on a plane with high spatial resolution.

**1.2. Difficulties in developing MEMS sensors**

regard to their continued development.

thin film deposition.

silicon.

volume decreases.

Biomedical Engineering

332

MEMS technologies have a wide variety of applications, and there are many types (examples are shown in Table 1). In the following sections, tactile sensors for human interfaces and flow sensors for air‐conditioning systems and medical applications will be focused.

## **2. Tactile sensors for human interfaces**

The single crystal silicon is often used as a MEMS material because it is a high‐quality semiconductor with excellent mechanical properties [1]. It is said that the yield strength of silicon is comparable to that of stainless steel. Silicon sensors for gyroscopes and for measuring pressure, acceleration, flow, and inclination have already been commercialized and are being used in the automotive and information industries [2‐5]. Silicon has also been used to make miniaturized tactile sensors as interface devices in robotic automation systems since in the middle of the 1980s. Various tactile sensors based on silicon MEMS technologies have been developed [6‐9]. In particular, Takao et al. integrated a tactile sensor with a CMOS device on the same chip [10]. Moreover, Shikida et al. proposed to integrate an actuator mechanism onto a tactile sensor as means of measuring the hardness of objects [11‐13].


**Figure 1.** Magnetically driven Si-based tactile sensor. Republished with permission of IOP Published Ltd from Ref. [13].

MEMS Sensors and Their Applications 335

**Figure 2.** Detection principle of Si-based active tactile sensor. Republished with permission of IOP Published Ltd from

Ref. [13].

#### **Table 1.** MEMS sensor applications

MEMS sensors are being used to study human activity by embedding them in clothes and everyday consumer products. Silicon tactile sensors have high space and time resolutions. However, it is difficult to make them wearable because silicon is so brittle. To overcome this problem, MEMS researchers have started using resin materials to fabricate flexible sensor structures [14‐17]. Notably, Konishi et al. and N. Chen et al. devised knitted structures for electrical wiring [18, 19]. These structures have advantages whereby they can fit on arbitrary bendable surfaces. Shikida et al. proposed an artificial hollow fiber as a MEMS material to implement a tactile sensing function on a fabric and developed a tactile sensor by weaving together fibers [20‐23]. They studied the unit's ability to detect force, cross‐talk, etc., by applying concentrated loads to it.

The following sections focus on silicon‐based and fabric tactile sensors.

#### **2.1. Si‐based tactile sensors**

A schematic view of a magnetically driven tactile sensor is shown in Figure 1. It consists of a diaphragm with a mesa at the center, piezo‐resistive strain sensors for detecting the displace‐ ment of the diaphragm at the periphery, a permanent magnet on the backside of the dia‐ phragm, and a flat coil. The magnetic actuation system is formed by pairing the permanent magnet and the flat coil, and the sensing diaphragm is driven up and down by the magnetic force generated by this actuation system. The sensor is used in two different modes, as shown in Figure 2.

**Figure 1.** Magnetically driven Si-based tactile sensor. Republished with permission of IOP Published Ltd from Ref. [13].

MEMS sensors are being used to study human activity by embedding them in clothes and everyday consumer products. Silicon tactile sensors have high space and time resolutions. However, it is difficult to make them wearable because silicon is so brittle. To overcome this problem, MEMS researchers have started using resin materials to fabricate flexible sensor structures [14‐17]. Notably, Konishi et al. and N. Chen et al. devised knitted structures for electrical wiring [18, 19]. These structures have advantages whereby they can fit on arbitrary bendable surfaces. Shikida et al. proposed an artificial hollow fiber as a MEMS material to implement a tactile sensing function on a fabric and developed a tactile sensor by weaving together fibers [20‐23]. They studied the unit's ability to detect force, cross‐talk, etc., by

Micro-Nano Mechatronics — New Trends in Material, Measurement, Control, Manufacturing and Their Applications in

A schematic view of a magnetically driven tactile sensor is shown in Figure 1. It consists of a diaphragm with a mesa at the center, piezo‐resistive strain sensors for detecting the displace‐ ment of the diaphragm at the periphery, a permanent magnet on the backside of the dia‐ phragm, and a flat coil. The magnetic actuation system is formed by pairing the permanent magnet and the flat coil, and the sensing diaphragm is driven up and down by the magnetic force generated by this actuation system. The sensor is used in two different modes, as shown

The following sections focus on silicon‐based and fabric tactile sensors.

applying concentrated loads to it.

**2.1. Si‐based tactile sensors**

**Table 1.** MEMS sensor applications

Biomedical Engineering

334

in Figure 2.

**Figure 2.** Detection principle of Si-based active tactile sensor. Republished with permission of IOP Published Ltd from Ref. [13].

**2.** *ms* is not negligible compared to *md*

**a.** Elastic coefficient *Ks*

**b.** Damping coefficient *Cs*

**2.2. Fabric tactile sensors**

the following advantages.

as if they were cloth.

**2.** Wearable sensors

**1.** Tactile sensing at arbitrary surface

gloves, and shoes, or by patching them to these items.

**3.** Detecting 2D contact‐force distributions

are described in the references [11‐13].

following procedure is used to estimate physical quantities.

and resonant frequency obtained in vibration mode.

*Cs* of the contacted object is estimated from *Ks*, *ms*, and the Q‐factor.

In this case, the sample's equivalent mass is not negligible compared to that of the mesa with diaphragm. Thus, the resonance characteristics, such as the resonant frequency and the Q‐ factor, depend on the equivalent mass and the elastic coefficient of the contacted object. The

MEMS Sensors and Their Applications 337

Both operation modes are required to estimate *Ks* and the equivalent mass *ms* of the object. First, the sensor precisely detects the elastic coefficient of the object using quasi‐static mode. Then, the equivalent mass of the object is estimated from the elastic coefficient of the object

The fabrication process and the sensor characteristics driven by the two means of actuation

Figure 3(a) shows the applications of fabric tactile sensors. These sensors can be used for monitoring the force distribution at human interfaces. A schematic view and operation principle are shown in Figures 3(b)‐(d). An artificial hollow fiber is made from a single elastic hollow fiber covered with metal and insulation layers. The sensor is fabricated by weaving the artificial hollow fibers in a reticular pattern. The applied normal load is detected by measuring the change in capacitance between the warp and weft fibers at their intersection point. When a normal load is applied to an intersection point of the fabric tactile sensor, the resulting deformation increases the contact area between the warp and weft fibers. Thus, the capacitance between them increases as the applied load increases. The fabric tactile sensor is therefore able to detect the normal load by measuring this capacitance change. This fabric tactile sensor has

The fabric sensor can fit on any surface, and it can perform tactile sensing by weaving the fibers

Wearable tactile sensors can be made by directly weaving artificial hollow fibers into clothes,

The sensor can detect the 2D normal‐load distribution by sequentially measuring the capaci‐ tance changes at all intersection points. The resolution of the 2D normal‐force distribution is

#### **Quasi‐static mode**

In this mode of operation, the tactile sensor detects two physical quantities, the contact force and the elasticity of the contacted object. The values are measured when the forces working on the object and the actuating magnetic force are in equilibrium. The detection principle is as follows. The mesa structure on the diaphragm moves only one direction in this mode. When the mesas on the diaphragm of the tactile sensor arrays come into contact with the object, some of the mesa structures are pushed inward by the bumpy surface of the object (Figure 2(a)). The contact force is detected by piezo‐resistive strain sensors measuring the deflection of the diaphragm. The force distribution and a 2D surface image of the object can be generated by using an array of sensors. After the contact force detection, the contacting mesas are driven against the object by the magnetic actuation system to measure the hardness of the object. As a result, the contacting regions are deformed in a way that depends on the elasticity of the object. Thus, one can calculate the elasticity of the object by using the relationship between the displacement of the diaphragm and the applied force, which is measured quasi‐statically.

## **Vibration mode**

In this mode, the mesa structure is alternately driven up and down at the resonance frequency of the diaphragm. The sensor can estimate the elastic and damping coefficients of the object in contact. Each sensing diaphragm is vibrated at the resonant frequency by magnetic force, as shown in Figure 2(b). Sensor outputs, such as the amplitude of the diaphragm, resonant frequency, and phase shift, change in accordance with the mechanical properties of the contacted object. The lumped parameter circuit model of this sensor system and the formulas for the resonant frequency and Q‐factor are shown in Figure 2(b‐3). The resonant frequency and Q‐factor depend on the elastic coefficient and equivalent mass of the object (see the formulas). The equivalent mass, *ms*, depends on the vibration amplitude, which is controlled with the measurements. The detection procedure depends on the magnitude of difference between the device mass *md*, and object mass *ms*, as follows:

#### **1.** *ms* << *md*

In this case, the vibrating mass of the sample is much less than that of the mesa with the diaphragm. The mass of the mesa and diaphragm dominates the total mass in vibration; thus, one can estimate elastic‐ and damping‐ coefficients of the contacted object with the following procedure.

**a.** Elastic coefficient *Ks*

The resonant frequency depends only on the elastic coefficient of the object, and it increases as the elastic coefficient *Ks* increases, as shown in Formula 1. Thus, one can calculate *Ks* of the contacted object simply by measuring the change in the resonant frequency.

**b.** Damping coefficient *Cs*

The Q‐factor of the vibrating diaphragm structure depends on the elastic and damping coefficients (*Ks* and *Cs*), as shown in Formula 2. Thus, one can estimate *Cs* of the contacted object from the Q‐factor and *Ks*.

**2.** *ms* is not negligible compared to *md*

In this case, the sample's equivalent mass is not negligible compared to that of the mesa with diaphragm. Thus, the resonance characteristics, such as the resonant frequency and the Q‐ factor, depend on the equivalent mass and the elastic coefficient of the contacted object. The following procedure is used to estimate physical quantities.

**a.** Elastic coefficient *Ks*

**Quasi‐static mode**

Biomedical Engineering

336

**Vibration mode**

**1.** *ms* << *md*

procedure.

**a.** Elastic coefficient *Ks*

**b.** Damping coefficient *Cs*

from the Q‐factor and *Ks*.

In this mode of operation, the tactile sensor detects two physical quantities, the contact force and the elasticity of the contacted object. The values are measured when the forces working on the object and the actuating magnetic force are in equilibrium. The detection principle is as follows. The mesa structure on the diaphragm moves only one direction in this mode. When the mesas on the diaphragm of the tactile sensor arrays come into contact with the object, some of the mesa structures are pushed inward by the bumpy surface of the object (Figure 2(a)). The contact force is detected by piezo‐resistive strain sensors measuring the deflection of the diaphragm. The force distribution and a 2D surface image of the object can be generated by using an array of sensors. After the contact force detection, the contacting mesas are driven against the object by the magnetic actuation system to measure the hardness of the object. As a result, the contacting regions are deformed in a way that depends on the elasticity of the object. Thus, one can calculate the elasticity of the object by using the relationship between the displacement of the diaphragm and the applied force, which is measured quasi‐statically.

Micro-Nano Mechatronics — New Trends in Material, Measurement, Control, Manufacturing and Their Applications in

In this mode, the mesa structure is alternately driven up and down at the resonance frequency of the diaphragm. The sensor can estimate the elastic and damping coefficients of the object in contact. Each sensing diaphragm is vibrated at the resonant frequency by magnetic force, as shown in Figure 2(b). Sensor outputs, such as the amplitude of the diaphragm, resonant frequency, and phase shift, change in accordance with the mechanical properties of the contacted object. The lumped parameter circuit model of this sensor system and the formulas for the resonant frequency and Q‐factor are shown in Figure 2(b‐3). The resonant frequency and Q‐factor depend on the elastic coefficient and equivalent mass of the object (see the formulas). The equivalent mass, *ms*, depends on the vibration amplitude, which is controlled with the measurements. The detection procedure depends on the magnitude of difference

In this case, the vibrating mass of the sample is much less than that of the mesa with the diaphragm. The mass of the mesa and diaphragm dominates the total mass in vibration; thus, one can estimate elastic‐ and damping‐ coefficients of the contacted object with the following

The resonant frequency depends only on the elastic coefficient of the object, and it increases as the elastic coefficient *Ks* increases, as shown in Formula 1. Thus, one can calculate *Ks* of the

The Q‐factor of the vibrating diaphragm structure depends on the elastic and damping coefficients (*Ks* and *Cs*), as shown in Formula 2. Thus, one can estimate *Cs* of the contacted object

contacted object simply by measuring the change in the resonant frequency.

between the device mass *md*, and object mass *ms*, as follows:

Both operation modes are required to estimate *Ks* and the equivalent mass *ms* of the object. First, the sensor precisely detects the elastic coefficient of the object using quasi‐static mode. Then, the equivalent mass of the object is estimated from the elastic coefficient of the object and resonant frequency obtained in vibration mode.

**b.** Damping coefficient *Cs*

*Cs* of the contacted object is estimated from *Ks*, *ms*, and the Q‐factor.

The fabrication process and the sensor characteristics driven by the two means of actuation are described in the references [11‐13].

#### **2.2. Fabric tactile sensors**

Figure 3(a) shows the applications of fabric tactile sensors. These sensors can be used for monitoring the force distribution at human interfaces. A schematic view and operation principle are shown in Figures 3(b)‐(d). An artificial hollow fiber is made from a single elastic hollow fiber covered with metal and insulation layers. The sensor is fabricated by weaving the artificial hollow fibers in a reticular pattern. The applied normal load is detected by measuring the change in capacitance between the warp and weft fibers at their intersection point. When a normal load is applied to an intersection point of the fabric tactile sensor, the resulting deformation increases the contact area between the warp and weft fibers. Thus, the capacitance between them increases as the applied load increases. The fabric tactile sensor is therefore able to detect the normal load by measuring this capacitance change. This fabric tactile sensor has the following advantages.

**1.** Tactile sensing at arbitrary surface

The fabric sensor can fit on any surface, and it can perform tactile sensing by weaving the fibers as if they were cloth.

**2.** Wearable sensors

Wearable tactile sensors can be made by directly weaving artificial hollow fibers into clothes, gloves, and shoes, or by patching them to these items.

**3.** Detecting 2D contact‐force distributions

The sensor can detect the 2D normal‐load distribution by sequentially measuring the capaci‐ tance changes at all intersection points. The resolution of the 2D normal‐force distribution is

**Figure 3.** Fabric tactile sensor and its applications. Republished with permission of IOP Published Ltd from Ref. [21].

determined by the grid spacing between the warp and weft fibers, and it is easily adjusted for any purpose by combining the artificial hollow fibers with cotton yarn.

#### **4.** Tactile sensing over a large surface area

Silicon‐based MEMS sensors have limited area coverage, because they use a small silicon wafer as a substrate. However, fabric tactile sensors consist of fibers; therefore, they can be made as big as needed for the application.

A large rectangular fabric tactile sensor was fabricated by weaving together hollow fibers, brass wires, and conventional cotton yarn (0.3 mm in diameter). The fabric sensor during the weaving process is shown in the right of Fig. 4(a). Figure 4(b) shows one of the produced fabric tactile sensors. The sensor measures 87.5 mm x 52.0 mm, and its sensing points are arranged in a 4 x 4 array. The pitch of detection points is 7.0 mm horizontally and vertically.

**Figure 5.** Applied force and shape detection using a fabric tactile sensor. Republished with permission of The Institute

MEMS Sensors and Their Applications 339

A fabric tactile sensor measuring 180.0 mm x 56.0 mm was evaluated as to its shape detection capability. The unit had a 4 x 10 array of sensing points (total number: 40). The number of the warp wires and weft fibers varied from A to D and 1 to 10, respectively. Thus, the intersection points were each labeled with a letter and number combination (see Fig. 5). The pitch of the detection points was 18.0 mm horizontally and 14.0 mm vertically. Figure 5 shows how the applied pressure and the object's shape are detected. A square plate measuring 20 mm x 20 mm was placed on the sensor (the area from B3 to B4 and from C3 to C4), and a force of 2925 mN was applied to it (Figure 5(a)). Two different metal weights were placed on the sensor (Figure 5(b)). A heavy weight of 269 g and a light weight of 187 g were directly placed on the area from B3 to B4 and from C3 to C4 and the area from B7 to B8 and from C7 to C8. Both weights were the same size, i.e., 24.2 mm x 24.2 mm. The fabric sensor successfully detected

MEMS flow sensors fabricated on Si wafers have a long history, and various types have been developed. C. Liu et al. developed a thermal type of flow sensor to detect shear stress [24]. J. Zhe et al. used a micro‐machined cantilever structure for measuring shear stress from the flow [25]. Si‐MEMS flow sensors have excellent space and time resolutions, and they are used in semiconductor equipment and in automobiles for controlling gas flows precisely. Addition‐ ally, Unnikrishan et al. developed a MEMS‐on‐tube assembly to simplify the packaging process [26]. They integrated Si‐MEMS devices directly on a glass tube, which is compatible

of Engineering and Technology from Ref. [23].

the two‐dimensional shape of the contacted objects.

**3. Flow sensors and their applications**

**Figure 4.** Fabrication of fabric tactile sensor. Republished with permission of The Institute of Engineering and Technol‐ ogy from Ref. [23].

**Figure 5.** Applied force and shape detection using a fabric tactile sensor. Republished with permission of The Institute of Engineering and Technology from Ref. [23].

A fabric tactile sensor measuring 180.0 mm x 56.0 mm was evaluated as to its shape detection capability. The unit had a 4 x 10 array of sensing points (total number: 40). The number of the warp wires and weft fibers varied from A to D and 1 to 10, respectively. Thus, the intersection points were each labeled with a letter and number combination (see Fig. 5). The pitch of the detection points was 18.0 mm horizontally and 14.0 mm vertically. Figure 5 shows how the applied pressure and the object's shape are detected. A square plate measuring 20 mm x 20 mm was placed on the sensor (the area from B3 to B4 and from C3 to C4), and a force of 2925 mN was applied to it (Figure 5(a)). Two different metal weights were placed on the sensor (Figure 5(b)). A heavy weight of 269 g and a light weight of 187 g were directly placed on the area from B3 to B4 and from C3 to C4 and the area from B7 to B8 and from C7 to C8. Both weights were the same size, i.e., 24.2 mm x 24.2 mm. The fabric sensor successfully detected the two‐dimensional shape of the contacted objects.

## **3. Flow sensors and their applications**

determined by the grid spacing between the warp and weft fibers, and it is easily adjusted for

**Figure 3.** Fabric tactile sensor and its applications. Republished with permission of IOP Published Ltd from Ref. [21].

Micro-Nano Mechatronics — New Trends in Material, Measurement, Control, Manufacturing and Their Applications in

Silicon‐based MEMS sensors have limited area coverage, because they use a small silicon wafer as a substrate. However, fabric tactile sensors consist of fibers; therefore, they can be made as

A large rectangular fabric tactile sensor was fabricated by weaving together hollow fibers, brass wires, and conventional cotton yarn (0.3 mm in diameter). The fabric sensor during the weaving process is shown in the right of Fig. 4(a). Figure 4(b) shows one of the produced fabric tactile sensors. The sensor measures 87.5 mm x 52.0 mm, and its sensing points are arranged

**Figure 4.** Fabrication of fabric tactile sensor. Republished with permission of The Institute of Engineering and Technol‐

in a 4 x 4 array. The pitch of detection points is 7.0 mm horizontally and vertically.

any purpose by combining the artificial hollow fibers with cotton yarn.

**4.** Tactile sensing over a large surface area

big as needed for the application.

Biomedical Engineering

338

ogy from Ref. [23].

MEMS flow sensors fabricated on Si wafers have a long history, and various types have been developed. C. Liu et al. developed a thermal type of flow sensor to detect shear stress [24]. J. Zhe et al. used a micro‐machined cantilever structure for measuring shear stress from the flow [25]. Si‐MEMS flow sensors have excellent space and time resolutions, and they are used in semiconductor equipment and in automobiles for controlling gas flows precisely. Addition‐ ally, Unnikrishan et al. developed a MEMS‐on‐tube assembly to simplify the packaging process [26]. They integrated Si‐MEMS devices directly on a glass tube, which is compatible with a Swagelok® connector, by fusion bonding. These and other Si‐based flow sensors are summarized in the references [27, 28]. The sensors are assembled onto a rigid flat board in the packaging process, to compensate for the brittleness of the Si material. This means these sensors are difficult to put on curved surfaces.

As stated above, the hot‐wire method detects a flow by measuring the heat it dissipates to the fluid by forced convection. The relationship between the dissipated power *Q* and flow velocity

<sup>0</sup>

where *A*, *B*, and *n* are constants depending on the geometry of the wire element. *Th* and *T0* are the temperatures of the wire and in the fluid, respectively. If the wire were infinitely long, *n*

Two different operations are commonly used for heating the wire electrically. One is constant voltage, or current mode. The flow rate is calculated from the change in electrical resistance, and this method has the advantage of needing only relatively simple electrical circuits. However, it has a drawback in that it is difficult to shorten the response time. To improve the response of the sensing wire, another mode, i.e., constant temperature (CT), is frequently used in flow rate measurements. In this mode, the temperature of the wire is kept constant by using a feedback circuitto shorten the response time, and the flow rate is calculated from the feedback voltage. CT mode is normally used in practice. The electrical energy supplied to the hot wire in CT mode is equal to the power dissipated from the wire to the fluid. As a result, equation

where *Vh* and *Rh* are the voltage difference and electrical resistance at the wire, respectively. Here, the square of the voltage at the wire is proportional to the n‐th power of the flow velocity.

The most common hot‐wire anemometers are placed in the center of flow channel structures. Thus, the devices detect the flow rate through the center of a tube. This measurement method

The flow velocity at the center of tube depends on the flow conditions. For example, the flow distribution in the tube depends on the distance from the entrance. Thus, the sensor has to be placed in the hydraulically fully developed flow region to obtain a constant value. This means

The sensor output depends on the radial position in the tube, because of the flow rate distri‐

*h*

0

(3)

*<sup>n</sup> Q A BU T T <sup>h</sup>* (2)

MEMS Sensors and Their Applications 341

*U* in a low Reynolds number flow is given by King's law [35]:

(2) is modified as follows:

has the following drawbacks.

**1.** Requirement of large inlet length

**2.** Requirement of position accuracy

would be 0.5. However, *n* differs form 0.5 in actual wires of finite length.

2

*h*

*R*

that it needs to be a certain distance from the tube entrance.

bution. This is true even in a fully developed flow.

*3.1.2. On‐wall in‐tube flexible thermal flow sensors*

*h n*

*<sup>V</sup> A BU T T*

Improved flexibility is the key to making MEMS sensors suitable for a larger range of appli‐ cations, and this can be accomplished by using resin materials as a substrate. Zhu et al. fabricated a flow senor directly on a flexible printed circuit board with electrical circuits, and mounted it onto the curved surface of a wing to control an aircraft [29]. Ma et al., produced a flexible flow sensor made of polyimide film for detecting the dynamic wave flows in a water channel [30]. For medical applications, Li et al., used Kapton film as a substrate, and they integrated pressure, temperature, glucose, and oxygen sensors on it for blood analysis [31]. Naito et al. fabricated a flow sensor on a flexible 7.5‐μm‐thick polyimide film and produced a miniaturized on‐wall in‐tube flow sensor [32]. They used it to fabricate a catheter flow sensor for measuring aspirated‐ and inspired‐air characteristics [33, 34]. Recently, flow sensors have been used to control air supplies precisely in large scale air‐conditioning network systems for the purpose of reducing wasted energy.

### **3.1. Thermal flow sensors**

Three principles, thermal anemometry, calorimetric flow sensing, and time‐of‐flight sensing, are used in thermal flow sensing. Thermal anemometry, i.e., hot‐wire anemometry, detects the flow rate by its cooling effect on a heated element. A calorimetric flow sensor, on the other hand, uses a heater and two sensing elements; the heater is placed between the two sensing elements, and the flow is detected from the difference in heating effects at the two sensing elements. The time‐of‐flight method uses the travelling time of a thermal pulse from the heater to the sensing element for detecting flow. The methods are explained in detail in the references [27, 28]. Hot‐wire anemometry flow sensors are widely used because they can measure high flow rates and have the simplest structure.

### *3.1.1. Hot‐wire anemometry*

Hot‐wire anemometry uses the cooling effect of forced convection to detect changes in flow. When a heated wire is placed in a flow, the equilibrium temperature at the wire depends on the amount of convection in the fluid. On the other hand, the electrical resistance of the wire generally depends on its temperature, and it can be expressed as,

$$R\_2 = R\_1 \cdot \left[1 + \alpha \cdot \left(T\_2 - T\_1\right)\right] \tag{1}$$

where *R1* and *R2*, are the electrical resistances of the wire at temperature *T1* and *T2*, and *α* is the temperature coefficient of resistance (TCR). Flow changes are detected from changes in the electrical resistance of the wire. Therefore, in practice, this method requires a large and stable TCR value.

As stated above, the hot‐wire method detects a flow by measuring the heat it dissipates to the fluid by forced convection. The relationship between the dissipated power *Q* and flow velocity *U* in a low Reynolds number flow is given by King's law [35]:

$$Q = \left(A + BLI^n\right) \cdot \left(T\_h - T\_0\right) \tag{2}$$

where *A*, *B*, and *n* are constants depending on the geometry of the wire element. *Th* and *T0* are the temperatures of the wire and in the fluid, respectively. If the wire were infinitely long, *n* would be 0.5. However, *n* differs form 0.5 in actual wires of finite length.

Two different operations are commonly used for heating the wire electrically. One is constant voltage, or current mode. The flow rate is calculated from the change in electrical resistance, and this method has the advantage of needing only relatively simple electrical circuits. However, it has a drawback in that it is difficult to shorten the response time. To improve the response of the sensing wire, another mode, i.e., constant temperature (CT), is frequently used in flow rate measurements. In this mode, the temperature of the wire is kept constant by using a feedback circuitto shorten the response time, and the flow rate is calculated from the feedback voltage. CT mode is normally used in practice. The electrical energy supplied to the hot wire in CT mode is equal to the power dissipated from the wire to the fluid. As a result, equation (2) is modified as follows:

$$\frac{\left(V\_{h}^{2}\right)}{R\_{h}} = \left(A + B\mathcal{U}^{\prime}\right) \cdot \left(T\_{h} - T\_{0}\right) \tag{3}$$

where *Vh* and *Rh* are the voltage difference and electrical resistance at the wire, respectively. Here, the square of the voltage at the wire is proportional to the n‐th power of the flow velocity.

#### *3.1.2. On‐wall in‐tube flexible thermal flow sensors*

The most common hot‐wire anemometers are placed in the center of flow channel structures. Thus, the devices detect the flow rate through the center of a tube. This measurement method has the following drawbacks.

**1.** Requirement of large inlet length

with a Swagelok® connector, by fusion bonding. These and other Si‐based flow sensors are summarized in the references [27, 28]. The sensors are assembled onto a rigid flat board in the packaging process, to compensate for the brittleness of the Si material. This means these

Micro-Nano Mechatronics — New Trends in Material, Measurement, Control, Manufacturing and Their Applications in

Improved flexibility is the key to making MEMS sensors suitable for a larger range of appli‐ cations, and this can be accomplished by using resin materials as a substrate. Zhu et al. fabricated a flow senor directly on a flexible printed circuit board with electrical circuits, and mounted it onto the curved surface of a wing to control an aircraft [29]. Ma et al., produced a flexible flow sensor made of polyimide film for detecting the dynamic wave flows in a water channel [30]. For medical applications, Li et al., used Kapton film as a substrate, and they integrated pressure, temperature, glucose, and oxygen sensors on it for blood analysis [31]. Naito et al. fabricated a flow sensor on a flexible 7.5‐μm‐thick polyimide film and produced a miniaturized on‐wall in‐tube flow sensor [32]. They used it to fabricate a catheter flow sensor for measuring aspirated‐ and inspired‐air characteristics [33, 34]. Recently, flow sensors have been used to control air supplies precisely in large scale air‐conditioning network systems for

Three principles, thermal anemometry, calorimetric flow sensing, and time‐of‐flight sensing, are used in thermal flow sensing. Thermal anemometry, i.e., hot‐wire anemometry, detects the flow rate by its cooling effect on a heated element. A calorimetric flow sensor, on the other hand, uses a heater and two sensing elements; the heater is placed between the two sensing elements, and the flow is detected from the difference in heating effects at the two sensing elements. The time‐of‐flight method uses the travelling time of a thermal pulse from the heater to the sensing element for detecting flow. The methods are explained in detail in the references [27, 28]. Hot‐wire anemometry flow sensors are widely used because they can measure high

Hot‐wire anemometry uses the cooling effect of forced convection to detect changes in flow. When a heated wire is placed in a flow, the equilibrium temperature at the wire depends on the amount of convection in the fluid. On the other hand, the electrical resistance of the wire

where *R1* and *R2*, are the electrical resistances of the wire at temperature *T1* and *T2*, and *α* is the temperature coefficient of resistance (TCR). Flow changes are detected from changes in the electrical resistance of the wire. Therefore, in practice, this method requires a large and stable

(1)

 2 1 21 *R R TT* 1 

sensors are difficult to put on curved surfaces.

the purpose of reducing wasted energy.

flow rates and have the simplest structure.

generally depends on its temperature, and it can be expressed as,

**3.1. Thermal flow sensors**

Biomedical Engineering

340

*3.1.1. Hot‐wire anemometry*

TCR value.

The flow velocity at the center of tube depends on the flow conditions. For example, the flow distribution in the tube depends on the distance from the entrance. Thus, the sensor has to be placed in the hydraulically fully developed flow region to obtain a constant value. This means that it needs to be a certain distance from the tube entrance.

**2.** Requirement of position accuracy

The sensor output depends on the radial position in the tube, because of the flow rate distri‐ bution. This is true even in a fully developed flow.

### **3.** Difficulty of measuring flow rates in curved tubes

The flow velocity distribution changes depending on the curvature radius and angle of the tube and on the distance from the end of the curved region. Thus, the sensor cannot output a constant value in a curved tube because the velocity distribution is asymmetric in the radial direction.

**3.2. Flow sensors for air‐conditioning networks**

to be precisely measured downstream of the bent duct.

in the sensor's development.

much as possible.

of the secondary flow.

of the duct.

Air conditioning systems are located outside of buildings and they delivertreated airto rooms through a network of ducts. These ducts usually have complicated configurations and a number of bends because they have to go inside a limited amount empty space between the beams of the building's ceiling. To reduce wasted energy, the air supplied to each room has to be controlled at an ideal sensing point near the outlet port. This means the flow rate must

MEMS Sensors and Their Applications 343

To measure the flow rate downstream of a bent duct,the following points have to be considered

**1.** The flow velocity distribution in the tube is expressed as a quadratic function. This means that the flow velocity reaches a maximum at the duct's center and a minimum at the inside surface of the duct. Thus, to reduce the air flow resistance of the sensor itself, the sensor has to be put on the inside surface of the duct, and its thickness has to be minimized as

**2.** The flow sensor structure has to be flexible in order to fit on the rounded inside surface

**3.** The air flow downstream of a bent duct is complicated because of the secondary flow caused by the bend. This means that it is difficult to measure the flow rate by using a single‐point measurement. To overcome this problem, a number of sensors have to be attached to the inside surface, and their outputs should be averaged to reduce the effect

A patch‐type flexible flow sensor was developed as a way to control the air supply precisely in large‐scale air‐conditioning network systems. The sensor is based on polyimide film, and it was fabricated by photolithography and thin‐film deposition. First, the photoresist was patterned by photolithography, and Au/Cr film was deposited by sputtering. The metal was then patterned by selectively removing the photoresist (lift‐off process). The Au and Cr thicknesses were 250 nm and 10 nm, respectively. Figure 7(a) shows flexible film sensors fabricated on a 3‐inch wafer. Each sensor measures 10 mm x 10 mm. A flexible printed circuit was used for the electrical connection. The sensor was placed on the printed circuit, and it was bonded to the board by adhesive. The electrical connection was manually made with silver paste. The electrical contact area was covered with adhesive to increase its mechanical strength. To form a cavity for thermal isolation, the sensor was placed on a silicone rubber sheet with a 5.0 mm x 5.0 mm hole in it, and the sensor was fixed to the sheet with adhesive polyimide film. The rubber sheet was 0.5‐mm thick. Figure 7(b) shows an assembled flexible flow sensor.

In an experimental evaluation, four flexible flow sensors were attached at an angle of 90° to the inside surface of an 8‐inch duct, and the outputs from them were averaged. The relationship between the output and flow rate obeyed King's equation for flows of from 0 to 3000 m3

The averaged sensor outputs depended on the distance of the sensorfrom the bend of the duct, andtheir values were slightly higherthan those obtainedin a straightduct.Aconversion factor,

h‐<sup>1</sup> .

To overcome these problems, a wall‐mounted in‐tube thermal flexible sensor able to measure the flow rate under both hydraulically developing and fully developed conditions was developed [36, 37]. The sensor consists of a ring‐shaped heater element on a flexible polyimide film (Figure 6), and it is fabricated on a polyimide film substrate and mounted on the inner wall surface of the tube to form a ring‐shaped sensor structure.

**Figure 6.** On-wall in-tube thermal flow sensor. Republished with permission of IOP Published Ltd from Ref. [36].

The advantages of the sensor are as follows.

**1.** It can measure flows in hydraulically developing regions

The flow develops in the direction from the inner wall surface to the center of the tube. The flow rate near the inner wall surface becomes steady first. Thus, this sensor can measure the flow even if the flow is not fully developed in all regions in the tube, because it measures the flow rate only near the inner wall. This means that the sensor can measure even in the hydraulically developing region of the tube, which shortens the distance from the entrance required by the sensor.

**2.** It can measure axially asymmetric flows.

The output signal from the sensoris insensitive to the velocity distribution in the tube, because ring‐shaped sensing structure averages out the flow distribution. This means the sensor outputs a steady value even if the velocity distribution is not axial symmetric.

The on‐wall in‐tube flow sensor is described in the references [36, 37]. The following sections describe two of its applications.

#### **3.2. Flow sensors for air‐conditioning networks**

**3.** Difficulty of measuring flow rates in curved tubes

wall surface of the tube to form a ring‐shaped sensor structure.

The advantages of the sensor are as follows.

**2.** It can measure axially asymmetric flows.

required by the sensor.

describe two of its applications.

**1.** It can measure flows in hydraulically developing regions

direction.

342

Biomedical Engineering

The flow velocity distribution changes depending on the curvature radius and angle of the tube and on the distance from the end of the curved region. Thus, the sensor cannot output a constant value in a curved tube because the velocity distribution is asymmetric in the radial

Micro-Nano Mechatronics — New Trends in Material, Measurement, Control, Manufacturing and Their Applications in

To overcome these problems, a wall‐mounted in‐tube thermal flexible sensor able to measure the flow rate under both hydraulically developing and fully developed conditions was developed [36, 37]. The sensor consists of a ring‐shaped heater element on a flexible polyimide film (Figure 6), and it is fabricated on a polyimide film substrate and mounted on the inner

**Figure 6.** On-wall in-tube thermal flow sensor. Republished with permission of IOP Published Ltd from Ref. [36].

The flow develops in the direction from the inner wall surface to the center of the tube. The flow rate near the inner wall surface becomes steady first. Thus, this sensor can measure the flow even if the flow is not fully developed in all regions in the tube, because it measures the flow rate only near the inner wall. This means that the sensor can measure even in the hydraulically developing region of the tube, which shortens the distance from the entrance

The output signal from the sensoris insensitive to the velocity distribution in the tube, because ring‐shaped sensing structure averages out the flow distribution. This means the sensor

The on‐wall in‐tube flow sensor is described in the references [36, 37]. The following sections

outputs a steady value even if the velocity distribution is not axial symmetric.

Air conditioning systems are located outside of buildings and they delivertreated airto rooms through a network of ducts. These ducts usually have complicated configurations and a number of bends because they have to go inside a limited amount empty space between the beams of the building's ceiling. To reduce wasted energy, the air supplied to each room has to be controlled at an ideal sensing point near the outlet port. This means the flow rate must to be precisely measured downstream of the bent duct.

To measure the flow rate downstream of a bent duct,the following points have to be considered in the sensor's development.


A patch‐type flexible flow sensor was developed as a way to control the air supply precisely in large‐scale air‐conditioning network systems. The sensor is based on polyimide film, and it was fabricated by photolithography and thin‐film deposition. First, the photoresist was patterned by photolithography, and Au/Cr film was deposited by sputtering. The metal was then patterned by selectively removing the photoresist (lift‐off process). The Au and Cr thicknesses were 250 nm and 10 nm, respectively. Figure 7(a) shows flexible film sensors fabricated on a 3‐inch wafer. Each sensor measures 10 mm x 10 mm. A flexible printed circuit was used for the electrical connection. The sensor was placed on the printed circuit, and it was bonded to the board by adhesive. The electrical connection was manually made with silver paste. The electrical contact area was covered with adhesive to increase its mechanical strength. To form a cavity for thermal isolation, the sensor was placed on a silicone rubber sheet with a 5.0 mm x 5.0 mm hole in it, and the sensor was fixed to the sheet with adhesive polyimide film. The rubber sheet was 0.5‐mm thick. Figure 7(b) shows an assembled flexible flow sensor.

In an experimental evaluation, four flexible flow sensors were attached at an angle of 90° to the inside surface of an 8‐inch duct, and the outputs from them were averaged. The relationship between the output and flow rate obeyed King's equation for flows of from 0 to 3000 m3 h‐<sup>1</sup> . The averaged sensor outputs depended on the distance of the sensorfrom the bend of the duct, andtheir values were slightly higherthan those obtainedin a straightduct.Aconversion factor,

**Figure 7.** Patch-type flexible flow sensor for large scale air-conditioning network systems.

which enables us to calculate the flow rate values from the obtained sensor outputs in bent ducts, was derived [38].

**Figure 8.** Concept of catheter flow sensor for trans-bronchial measurement. Republished with permission of IOP Pub‐

MEMS Sensors and Their Applications 345

Polymer‐MEMS technologies and heat shrinkable tubes are used to produce the sensor. The

Polyimide film is used as a substrate. The thickness of the film affects the thermal response performance (it has to be relatively thick; 7.5 μm). The metal film heater structures are formed on the film by using photolithography and sputtering, and they are patterned with a lift‐off process. The typical size of the sensor is 2.8 mm x 5.5 mm. Two heaters are formed on the film in orderto detect the flow direction. Each film sensoris mechanically cut before being mounted

**b.** The heat shrinkable tube is baked at 110°C. The Teflon tube shrinking to almost half its original size as a result of heating, and the film sensor is automatically mounted on the inner wall surface during the shrinking process and becomes fixed on the tube surface.

**c.** A cavity structure is formed under the heating element to improve thermal isolation. To produce the active structure, a slit is formed on the tube, and it is covered with a one‐ mode Teflon heat‐shrinkable tube to seal it. The outer diameter of the inner tube is only a little bit larger than the inner diameter of the outer one, after the heat shrinking process.

Figures 9(a) and 9(b) show a schematic diagram and a photograph of the fabricated catheter‐ type flow sensor. The inner and outer diameters of the tube are 1.0mm and 1.8 mm. The

lished Ltd from Ref. [34].

inside the tube.

fabrication process consists of two steps.

**2.** Mounting film sensor inside a tube

The sensor is mounted on the inside surface of the tube as follows.

**a.** The sensor is inserted into a heat shrinkable tube made of Teflon.

Thus, these two Teflon tubes are tightly fixed to one other.

**1.** Flexible film sensor fabrication

## **3.3. Flow sensors for medical applications**

The numbers of cardiac disease and cerebral stroke patients are gradually decreasing, thanks to developments in medical devices and improved health guidance. However, chronic obstructive pulmonary disease (COPD) is still on the increase. Spirometry is normally used to evaluate the progress of COPD. It measures the flow rate in the human mouth. The respiratory system consists of numerous levels of bronchi just like a tree, and the lung alveoli are located at the end of diverging bronchi. In the case of COPD, the alveoli structures gradually collapse as the patient ages and because of absorbed cigarette smoke or other air‐polluting substances. Lesions develop at the ends of the diverging bronchi. The current method of measurement, which evaluates lesions in the mouth, cannot be used to evaluate such small lesions in the bronchi.

To overcome the above problem, Shikida et al. [34, 39] devised a catheter‐type flow sensorthat can measure aspirated‐ and inspired‐airflow characteristics trans‐bronchially. The flow sensor (Figure 8) can be inserted into a small bronchus for measuring aspirated and inspired air characteristics. An on‐wall in‐tube thermal flow sensor is mounted on the inside of the tube, and it is used in the bronchoscope. The external diameter of the tube is only a few mm, and therefore, it can reach into the small bronchus. Two heaters were formed on the film in order to detect the flow direction.

**Figure 8.** Concept of catheter flow sensor for trans-bronchial measurement. Republished with permission of IOP Pub‐ lished Ltd from Ref. [34].

Polymer‐MEMS technologies and heat shrinkable tubes are used to produce the sensor. The fabrication process consists of two steps.

**1.** Flexible film sensor fabrication

which enables us to calculate the flow rate values from the obtained sensor outputs in bent

Micro-Nano Mechatronics — New Trends in Material, Measurement, Control, Manufacturing and Their Applications in

**Figure 7.** Patch-type flexible flow sensor for large scale air-conditioning network systems.

The numbers of cardiac disease and cerebral stroke patients are gradually decreasing, thanks to developments in medical devices and improved health guidance. However, chronic obstructive pulmonary disease (COPD) is still on the increase. Spirometry is normally used to evaluate the progress of COPD. It measures the flow rate in the human mouth. The respiratory system consists of numerous levels of bronchi just like a tree, and the lung alveoli are located at the end of diverging bronchi. In the case of COPD, the alveoli structures gradually collapse as the patient ages and because of absorbed cigarette smoke or other air‐polluting substances. Lesions develop at the ends of the diverging bronchi. The current method of measurement, which evaluates lesions in the mouth, cannot be used to evaluate such small lesions in the

To overcome the above problem, Shikida et al. [34, 39] devised a catheter‐type flow sensorthat can measure aspirated‐ and inspired‐airflow characteristics trans‐bronchially. The flow sensor (Figure 8) can be inserted into a small bronchus for measuring aspirated and inspired air characteristics. An on‐wall in‐tube thermal flow sensor is mounted on the inside of the tube, and it is used in the bronchoscope. The external diameter of the tube is only a few mm, and therefore, it can reach into the small bronchus. Two heaters were formed on the film in order

ducts, was derived [38].

Biomedical Engineering

344

bronchi.

to detect the flow direction.

**3.3. Flow sensors for medical applications**

Polyimide film is used as a substrate. The thickness of the film affects the thermal response performance (it has to be relatively thick; 7.5 μm). The metal film heater structures are formed on the film by using photolithography and sputtering, and they are patterned with a lift‐off process. The typical size of the sensor is 2.8 mm x 5.5 mm. Two heaters are formed on the film in orderto detect the flow direction. Each film sensoris mechanically cut before being mounted inside the tube.

**2.** Mounting film sensor inside a tube

The sensor is mounted on the inside surface of the tube as follows.


Figures 9(a) and 9(b) show a schematic diagram and a photograph of the fabricated catheter‐ type flow sensor. The inner and outer diameters of the tube are 1.0mm and 1.8 mm. The temperature coefficient of resistance of the sensor is 0.0025K‐<sup>1</sup> . The package method of the on‐ wall in‐tube film mounting is suitable for miniaturization because the sensor structure itself does not disturb the flow stream.

was inspired in a short time, and a large amount of air was aspirated at the beginning of the aspiration mode, and the amount gradually decreased after that. The sensor signal quantita‐ tively corresponded to this natural respiratory mechanism. From these results, the catheter flow sensor will be useful for evaluating the flow characteristics in the small bronchus region

MEMS Sensors and Their Applications 347

**Figure 10.** Breathing waveform of rat measured with intubated catheter flow sensor. Republished with permission of

**Figure 11.** Stent flow sensor for evaluation of nasal respiration. Republished with permission of IOP Published Ltd

in the future.

IOP Published Ltd from Ref. [39].

from Ref. [40].

**Figure 9.** Fabricated catheter type flow sensor. Republished with permission of IOP Published Ltd from Ref. [39].

Experiments were done under the regulations set out in the Nagoya University Animal Experiments Guidelines, and they were approved by the animal ethics committee. The experiments were done under anesthesia. Thus, the following data are aspirated‐ and inspired‐ air characteristics under anesthesia. Infiltration of bodily fluids into the tube was prevented by using Teflon as the tube material. The catheter flow sensor is intended to be incorporated into a bronchoscope and to be inserted in the small airways from the mouth. Thus, the sensor was first tested on rats. An optical fiberscope with an outer diameter of 0.8 mm, instead of a bronchoscope, was used in the tests. Inserting the flow sensor into the airway from the mouth of the rat with the fiberscope involved three steps.


The breathing waveform of the rat is shown in Figure 10. A period of 820 ms for inspiration and aspiration was obtained. This means that the respiration frequency was 1.1 Hz. The ventilated air volume was calculated from this breathing waveform, and a value ranging from 1.01–1.09 cc was obtained. The known respiration frequency and ventilated air values of rats range from 1.1–1.9 for the former and from 0.60–1.25 for the latter. The measured values coincided with the physiological values in the literature.

The air was inspired for a short time period, suddenly becoming aspirated for a time period. In aspirated mode, a large amount of air was aspirated at the beginning, and the aspirated air gradually decreased afterwards. Inspiration and aspiration were done by moving the dia‐ phragm. The air was inspired by expanding the thoracis. This was done by contracting the diaphragm. The air was simply aspirated by the restorative force of the thoracis. Thus, the air was inspired in a short time, and a large amount of air was aspirated at the beginning of the aspiration mode, and the amount gradually decreased after that. The sensor signal quantita‐ tively corresponded to this natural respiratory mechanism. From these results, the catheter flow sensor will be useful for evaluating the flow characteristics in the small bronchus region in the future.

temperature coefficient of resistance of the sensor is 0.0025K‐<sup>1</sup>

of the rat with the fiberscope involved three steps.

coincided with the physiological values in the literature.

Teflon tube guide.

location.

does not disturb the flow stream.

Biomedical Engineering

346

wall in‐tube film mounting is suitable for miniaturization because the sensor structure itself

Micro-Nano Mechatronics — New Trends in Material, Measurement, Control, Manufacturing and Their Applications in

**Figure 9.** Fabricated catheter type flow sensor. Republished with permission of IOP Published Ltd from Ref. [39].

Experiments were done under the regulations set out in the Nagoya University Animal Experiments Guidelines, and they were approved by the animal ethics committee. The experiments were done under anesthesia. Thus, the following data are aspirated‐ and inspired‐ air characteristics under anesthesia. Infiltration of bodily fluids into the tube was prevented by using Teflon as the tube material. The catheter flow sensor is intended to be incorporated into a bronchoscope and to be inserted in the small airways from the mouth. Thus, the sensor was first tested on rats. An optical fiberscope with an outer diameter of 0.8 mm, instead of a bronchoscope, was used in the tests. Inserting the flow sensor into the airway from the mouth

**1.** Only the fiberscope was inserted, from the mouth to the targeted location, by observing the inside of the airway. Then, the sensor was inserted to the targeted location with a

**3.** The Teflon tube guide was carefully extracted and only the flow sensor remained at the

The breathing waveform of the rat is shown in Figure 10. A period of 820 ms for inspiration and aspiration was obtained. This means that the respiration frequency was 1.1 Hz. The ventilated air volume was calculated from this breathing waveform, and a value ranging from 1.01–1.09 cc was obtained. The known respiration frequency and ventilated air values of rats range from 1.1–1.9 for the former and from 0.60–1.25 for the latter. The measured values

The air was inspired for a short time period, suddenly becoming aspirated for a time period. In aspirated mode, a large amount of air was aspirated at the beginning, and the aspirated air gradually decreased afterwards. Inspiration and aspiration were done by moving the dia‐ phragm. The air was inspired by expanding the thoracis. This was done by contracting the diaphragm. The air was simply aspirated by the restorative force of the thoracis. Thus, the air

**2.** The fiberscope was withdrawn when the flow sensor reached the target location.

. The package method of the on‐

**Figure 10.** Breathing waveform of rat measured with intubated catheter flow sensor. Republished with permission of IOP Published Ltd from Ref. [39].

**Figure 11.** Stent flow sensor for evaluation of nasal respiration. Republished with permission of IOP Published Ltd from Ref. [40].

A human being breathes through the mouth and the nose. Bacteria and viruses are trapped in the nose cavity, and only clean inspired air is supplied to the hung. The inspired air is also humidified in the cavity in order for lungs to adsorb oxygen effectively. The inspired air is warmed inside the body is then expired outside; this is a method of heat exchange. The flow characteristics of the nose are deeply related to various health concerns. Thus, a stent‐type of thermal flow sensor was also developed for measuring nasalrespiration. The MEMS stent flow sensor is shown in Figure 11. It is a thermal flow sensor fabricated on polymer film and monolithically integrated on the stent structure [40].

[3] Yamasaki H., editor. Handbook of sensors and actuators – Intelligent sensors (vol. 3).

MEMS Sensors and Their Applications 349

[5] Maluf N. An introduction to microelectromechanical systems engineering. USA: Ar‐

[6] Chu Z, Sarro P. M, Middlehoek S. Silicon three‐axial tactile sensor, In: Technical Di‐ gest of The 8th International Conference on Solid‐State Sensors and Actuators, June

[7] Kane B. J, Cutkosky M. R, Kovacs G. T. A. A traction stress sensor array for use in high‐resolution robotic tactile imaging. Journal of microelectromechanical systems

[8] Kasten K, Amelung J, Mokwa W. CMOS‐compatible capacitive high temperature

[9] Kim K, Lee K. R, Kim Y. K, Lee D. S, Cho N. K, Kim W. H, Park K. B, Park H. D, Park Y. K, Kim J. H, Pak J. J. 3‐axes flexible tactile sensor fabricated by Si micromachining and packaging technology. In: Proceedings of IEEE MEMSʹ06. January 2006, Istanbul,

[10] Takao H, Yawata M, Sawada K, Ishida M. A robust and sensitive silicon‐MEMS tac‐ tile‐imager with scratch resistant surface and over‐range protection. Technical Digest of The 14th International Conference on Solid‐State Sensors, Actuators and Microsys‐

[11] Shikida M, Shimizu T, Sato K, Itoigawa K. Active tactile sensor for detecting contact force and hardness of an object. Sensors and Actuators A: Physical 2003;103(1‐2)

[12] Hasegawa Y, Shimizu T, Miyaji T, Shikida M, Sasaki H, Sato K, Itoigawa K. A micro‐ machined active tactile sensor for hardness detection. Sensors and Actuators A :

[13] Hasegawa Y, Shikida M, Sasaki H, Sato K, Itoigawa K. An active tactile sensor for de‐ tecting mechanical characteristics of contacted objects. Journal of micromechanics

[14] Engel J, Chen J, Liu C. Development of polyimide flexible tactile sensor skin. Journal

[15] Kim S‐H, Engel J, Liu C, Jones D. L. Texture classification using a polymer‐based MEMS tactile sensor. Journal of micromechanics and microengineering 2005;15(5)

pressure sensors. Sensors and Actuators A: Physical 2000;85(1‐3) 147‐152.

[4] Ristic L. Sensor technology and devices. USA: Artech House, Inc.; 1994.

The Netherlands: Elsevier Science B. V.; 1996. 3.

tech House, Inc.; 2000.

2000;9(4) 425‐434.

Turkey; 2006.

213‐218.

912‐920.

1995, Stockholm, Sweden; 1995.

tems. June 2007, Lyon, France; 2007.

Physical 2004;114(2‐3) 141‐146.

and microengineering 2006;16(8) 1625‐1632.

of micromechanics and microengineering 2003;13(3) 359‐366.
