Perspective Chapter: Dielectric Elastomer Sensor Capable of Measuring Large Deformation and Pressure

Abstract

Most of the conventional sensors used for measuring deformation, pressure, etc., use metal, ceramics, piezo, or the like. Many of them are very rigid, and when the object is deformed or when the pressure on the object changes currently, it is necessary to arrange a large number of sensors with different conditions side by side. However, it is still difficult to measure all changes over time. With the newly developed dielectric elastomer sensor, even a very thin (0.1–0.2 mm) elastomer thickness could be deformed in difficult environments (e.g., places with large temperature changes or large vibrations), and it would be possible to measure any pressure changes due to its deformation. By applying this sensor, it can be used as a position sensor (including a three-dimensional sensor) or an acceleration sensor, so that it could be applied to the control of the arms and legs of a robot, smart shoes, and the like.

1. Introduction

Materials consisting of new metal materials, high-performance polymer materials, fine ceramics, and composite materials are expected to be used in cutting-edge technologies that support various industries and economies in the twenty-first century, along with electronics and biotechnology [1].

Piezoelectric composite materials [2], conductive polymer materials [3], bimetals [4], etc., are known as composite materials that can be applied as sensor materials. The former two are excellent materials that combine the advantages of both organic materials with inorganic materials in a matrix. The latter combines dissimilar metals to enable temperature sensors.

Engineering sensors are extremely simple devices at the present stage. As seen in partially integrated pressure sensors and magnetic sensors, they are becoming more intelligent, but they have not yet reached the level of intelligent sensors. As an example of the development of an integrated pressure sensor, T. Sarutani et al. prototyped an integrated pressure sensor using a shear gauge as a piezo resistance gauge [2]. The temperature of the pressure sensor is corrected on the same chip. As a feature of this sensor, the pressure that can be measured is only the pressure at the time when the pressure is applied, and accurate measurement cannot be performed unless the temperature is corrected.

As a sensor for measuring elongation, a strain sensor can be mentioned. The type using a piezo is well known [5]. However, the measurement of elongation is quite limited. Stretch sensors that support greater elongation using elastomers have recently emerged. Electrodes were attached to the top and bottom of the elastomer, and the elements were stacked in several stages for use [6]. It is an application of the so-called dielectric elastomer (DE). However, since a plurality of these elements are laminated, it is considered that the flexibility is not so high, so it is considered that the element is not sufficiently deformed unless the force required for deformation is large.

In this experiment, carbon grease, carbon black, and single-wall carbon nanotubes (SWCNTs) were applied to thin elastomer films of silicon, acrylic, and hydrogenated nitrile rubber (HNBR) as electrodes, and each film was used as a pressure sensor. Then, how each film behaved as a pressure sensor was observed. We also verified the performance as a stretch sensor by using the same combination of film and electrodes as above. In this way, we discussed whether a film with the same structure could be used as an intelligent sensor capable of two different types of sensing.

2. Background of dielectric elastomers (DEs)

DE was first created in 1990 by S. Chiba, R. Pelrine et al. of the Stanford Research Institute in the United States. After that, various researchers started their own development [7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28]. Currently, that development is spreading to DE actuators (DEA) and DE power generators (DEGs) that generate electricity by reversing the drive [28]. The concept of the DE sensor was introduced in the latter half of the 1990s, and research and development are currently underway in this field as well [8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25].

The structure and principle of the DE sensor (see Figure 1) are the same as those of the DEA and DEG, and the driving principle is that the output of DEA and its elongation are in inverse proportion to each other. Also, the change in capacitance of the DE sensor and DEA and its elongation are in direct proportional relationship [9] (see Figure 2).

Figure 1.

Structure and principle of DE sensor.

Figure 2.

Relationship between the elongation of the DE sensor and capacitance.

The relationship between DEA and capacitance can be expressed by the following equation;

$$C=\epsilon \frac{S}{d}$$E1

Here, C is capacity (F), ε is the dielectric constant of polymer film (F/m), S is area of Electrode (m²), and d is distance between electrodes (m).

So far, a wide range of research and development have been carried out for various applications. This research also includes the examination of material types and DE sensor shapes [10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24]. First, as given an example of the material, H. Sun et al. claimed that SiR-Fe containing 8% FeCl₃ could be effectively used as a medical DE sensor [10]. Perhaps this polymer is hard and unsuitable for precision sensors. B. O’brien et al. made a DE sensor by using two layers of thick silicone elastomer with a size of 20 mm x 70 mm and applying carbon powder to the electrodes [11]. M. Han et al. proposed a pressure sensor in which the medical wearable sensor consisted of a carbon nanotube-polydimethylsiloxane (CNT-PDMS) composite electrode and a porous polymer dielectric layer [12]. The concept is interesting, but this wearable sensor is unlikely to work well because the electrodes were not flexible enough and polymers used were hard.

H. Liebscher et al. tried to increase the dielectric constant of polyurethane by adding barium titanate to the polyurethane elastomer [13]. J. Bae et al. tried to control the robot by feeding back the information [14]. C. Briggs et al. tried to develop a tactile sensor for a robot’s hand by attaching a dome-shaped DE sensor or DE sensor mounted on a cylindrical pad to a robot gripper [15]. L. Agostini et al. also conducted preliminary experimental studies seeking the most influential elements in order to optimize the performance of hemispherical anti-collision sensors [16].

As an example of circuit studies, K. Jung et al. used modulation technology to realize DEA operation and its detection and used a system that mixed a low-frequency signal for operation and a high-frequency signal with small amplitude for detection [17]. H. Bose et al. tried to create a thicker mat-shaped pressure sensor because the sensor signal was too small to detect in the sheet-shaped DE [18]. Finally, as an application example, J. Bae et al. also explained applications using PVDF-based materials, silicones, and acrylic materials [14]. M. Rosenthal et al. discussed diagnostic tools for industrial equipment and applicable sensors for system monitoring [19].

R. Walker et al. conducted a unique study of attaching a DE sensor to a wetsuit for divers as an underwater application for DE. [20]. C. Larson et al. also arranged a large number of DE sensors on participants’ bodies and tried to apply them to a virtual reality game that would enable them to play the game by moving their bodies [21]. As another system research example, a DEA system could be developed that combines a DE sensor and an actuator to assist the movement of a patient’s fingers, hands, feet, etc., and to accurately evaluate their rehabilitation progress [22]. R Venkatraman et al. tried to measure blood pressure using a DEA cuff device [23].

In this way, DEA can be used simultaneously an actuator and a pressure sensor and/or a position sensor [24]. In the near future, it is expected that the arms and legs of intelligent robots, nursing care equipment, and the like will become possible. In addition, a DEA/sensor system that assists the movement of the patient’s fingers, hands, feet, etc., and accurately evaluates the rehabilitation situation will be possible [25]. In developing such devices and devices, it is necessary to improve the performance of DE sensors. The major factors will be to improve the conductivity of the electrode and the flexibility of the electrode [26, 27].

3. Experimental procedure

In this experiment, carbon grease, carbon black, and SWCNT (ZEONANO®-SG101, Zeon Corp., Tokyo, Japan) were used as electrodes for four types of elastomers to create DE pressure sensors, and the performance of each was compared. If necessary, soft resin blocks were laid under the elastomer films to further increase the sensitivity of the DE sensor. Similarly, for the stretch sensor, carbon grease, carbon black, and SWCNT electrodes were used for each, and the performance was compared.

3.1 Test materials

In this study, we used a US-made acrylic material (3 M/4905), a US-made acrylic distortion-corrected film, and a silicon material (ELASTOSIL FILM 2030250). In addition, hydrogenated acrylonitrile butadiene rubber (HNBR) film synthesized by Zeon Corp. was used. The thickness of the acrylic material and silicon of the test piece was 0.5 mm, and the thickness of HNBR was 0.2 mm. HNBR is a material that absorbs the vibrations of automobile engines and is too hard as it is, so we optimized the amount of cross-linking agent added and cut a part of the double bond to improve the elongation [26]. Using the above films and resin blocks, a tensile test and dynamic viscoelasticity test were also performed to correlate sensor performance with material differences.

3.1.1 Tensile test

A tensile test of the dumbbell-shaped test piece as shown in Figure 3 was performed using the Orientech tabletop material tester STA-1150. The displacement velocities were set to 100 mm/min, 200 mm/min, 300 mm/min, and 500 mm/min in order to evaluate the effect of the displacement velocities on the mechanical properties.

Figure 3.

A tensile test of the dumbbell-shaped test piece.

3.1.2 Dynamic viscoelasticity test

The dynamic viscoelasticity tester used was an MCR302 rheometer manufactured by Anton Paar, and the viscoelastic behavior of acrylic DE and silicone DE in frequency dependence was investigated. The measurement conditions were a frequency of 0.1–20 Hz, a shear strain of 1%, and a room temperature of 20° C, and the size of the test piece was 25 mm in diameter.

3.1.3 Elastomer hardness measurements

In this study, the hardness of the acrylic film (3 M/4905), the strain-corrected film of the acrylic film, silicon (ELASTOSIL FILM 2030250), HNBR film, and resin block materials were measured using Askar C method. The details of the resin blocks are shown in Section 3.3.

Askar C is a durometer C (spring type hardness tester) specified in SRIS0101 (Japan Rubber Association standard) and is a measuring instrument for measuring hardness. If Asker C is described as 20 in the physical properties table, it means that the value measured by the Asker C hardness tester is 20, and contrary to the needle insertion degree and consistency test, the larger the number, the harder the material.

3.2 How to make a donut-shaped sheet DE sensor

Each elastomer was cut into a circle and molded to an outer diameter of approximately 8 cm using 100% prestrain. A carbon grease electrode, a carbon black electrode, and an SWCNT electrode were attached to the upper and lower parts of these films to form a donut-shaped sheet DE sensor (see Figure 4). Carbon grease was applied as electrodes with a small brush. As the carbon black electrode, a carbon black spray (carbon black dissolved in a solvent, mixed with a small amount of binder, and packed in a spray can) was used. In the case of SWCNT electrodes, the electrodes were attached using SWCNT spray (ZEONANO®-SG101/SWCNT dissolved in a solvent, mixed with a small amount of binder, and packed in a spray can) [25]. The thickness of the carbon grease electrode was 100 μm, and the thickness was confirmed using a double-scan high-precision laser measuring instrument (LT-9500 & LT-9010 M) manufactured by Keyence. The thickness of carbon black was 70 μm and that of SWCNT was 50 μm, and the thickness was similarly confirmed using a double-scan high-precision laser measuring instrument.

Figure 4.

An example of a donut-shaped sheet DE sensor (using SWCNT electrode).

3.3 Pressure measurement/evaluation methods using donut-shaped sheet DE sensors

As an evaluation method for the various DE pressure sensors described above, pressure was applied using the evaluation system shown in Figure 5 with a resin block as a cushioning material (Blue Forest Trading LLC: KG-01) laid under the DE sensor. It was measured using a small precision vise attached to the Z-axis precision stage, and a load is applied from the top of the donut-shaped DE sensor to deform it. The experimental donut-shaped DE sensor is installed at the top of the load cell. The load was measured using a load cell capable of measuring up to 20 kg. The analog signal output from the load cell was taken into the CPU via a dedicated AD conversion IC (HX711). After that, the calibration process was performed by the CPU, and then the measured value was displayed on the LDC. The experimental equipment was calibrated using weights of 10 g, 500 g, 1 kg, 5 kg, and 15 kg, and it was confirmed that the error at each measurement point was within ±1%. The electrostatic capacity of the donut-shaped DE sensor was measured using LCR METER (ZM2372) manufactured by NF corporation.

Figure 5.

Overview of electrostatic capacity measuring device for donut-shaped DE sensor.

As described above, in the donut-shaped DE sensor, a conical resin block made of an acrylic gel sheet was installed in the central part, and the central part was installed so as to be raised by about 5 mm. Figure 6 shows a donut-shaped DE sensor with a resin block installed in order to rise the central part by about 5 mm. The shape of the resin block is shown in Figure 7. The height of the resin block is 6 mm, but this block is quite soft, and when a donut-shaped DE sensor is placed on it, it sinks 1 mm and becomes 5 mm. There are two reasons for inserting this resin block:

• To deform the elastomer more greatly, and

• When using as a carbon grease as an electrode, deformation or leakage will occur if force is applied to the carbon grease layer, so the resin block as a cushioning materials were adapted. Naturally, it had need to be used to disperse the pressure.

Figure 6.

Donut-shaped DE sensor installed so that the central part rises about 5 mm.

Figure 7.

Shapes of resin blocks used.

For the acrylic and its strain-removed film, the block shown in Figure 6a was used, but since silicon and HNBR are hard, it was necessary to enlarge the block as shown in Figure 6b. The reason is that with a hard film, the block sinks more than necessary, making it difficult to measure the electrostatic capacity. So, it is necessary to increase the area that receives pressure (see Eq. (1) above).

The response speed of the DE sensor was verified by measuring the time from the time when the load was applied to the DE sensor until the actual capacitance of the DE sensor changed. The presence or absence of a load on the DE sensor was determined by monitoring the amount of deformation of the DE sensor using an ultrahigh-precision laser displacement meter (controller: LC-2400, laser sensor: LC-2440) manufactured by Keyence Co., Ltd.

3.4 DE sheet-type sensors without cushioning materials

In addition to the above experiments, experiments without cushioning material in acrylic sheet sensors, HNBR, and silicon sheet sensors using SWCNT electrodes were executed. The loads were increased from 10 kg to 120 kg at 10 kg intervals, and the capacitance at each load was measured. The electrodes of sensors used were the SWCNT spray described above, and the shape was a circle with a diameter of 20 mm. Figure 8 shows a sheet DE sensor as the experimental prototype.

Figure 8.

An experimental sensor with a diameter of 20 mm.

Figure 9 shows an outline of the experimental equipment. By sandwiching each prototype sensor in a precision vise and tightening the precision vise, a load is applied to the experimental sensor to deform it. The experimental sensor was attached to a precision vise by sandwiching it between acrylic blocks for insulation. For the load measurement, a load cell capable of measuring up to 120 kg was used. The analog signal from the load cell was taken into the CPU via a dedicated AD conversion IC (HX711). After that, after performing the calibration process with the CPU, the measured value was displayed on the LDC. The measurement accuracy was confirmed at three points of 10 kg, 30 kg, and 50 kg using a weight whose mass was measured in advance, and it was confirmed that the error was 5% or less at each point. The electrostatic capacity that changes due to the deformation of the experimental sensor was measured using LCR METER (ZM2372) manufactured by NF corporation.

Next, using a DE sheet sensor with a SWCNT electrode attached to a US-made acrylic distortion-corrected film, the pressure was applied up to 120 kgf by sandwiching it between flanges, and its behavior was observed. The meaning of this experiment is that when bolting the flange, it can be confirmed by the pressure sensor whether or not the bolts are tightened evenly. Figure 10 shows the installation of DE sensors on the flange. Figure 11 shows sensor bases with four load cells. Figure 12 shows a pressure change measurement system by tightening bolts using DE sensors.

Figure 9.

Outline of the experimental device for sheet-type sensors.

Figure 10.

Installation of DE sensors on the flange and flange fixed with M6 bolts.

Figure 11.

Sensor bases with four load cells.

Figure 12.

Pressure change measurement system by tightening bolts using DE sensor.

Acrylic DE sensors (strain removal type) using SWCNT electrodes were installed at four locations on the flange. Capacitance was measured while tightening only the lower right bolt (No. 1) (see Figure 12) by 1/2 or 1 turn compared with the others. A calibration was performed in advance by comparing with the value of the DE sensor using a load cell with a maximum measured value of 120 kg (see Figure 13). The flange was fixed with four M6 (φ6 mm) bolts attached at 90° intervals. The DE seat sensor was attached to the top of the aluminum plate so that it was located right next to the flange fixing bolt. A glass epoxy plate having a thickness of 0.3 mm is installed on the surface of the aluminum plate for the purpose of electrically insulating the DE sensor and the aluminum block (see Figure 12).

Figure 13.

Examples of weighing scales using various load cells capable of measuring up to 120 kg.

As a circuit to take the data from the DE sensor into the PC, four sets of circuits in which the DE sensor was connected to the CV conversion and the voltage adjustment circuit were prepared. The output from each amplifier was AD-converted and taken into the PC.

Figure 13 shows examples of weighing scales using various load cells capable of measuring up to 120 kg. The load cell alone cannot be used, and a mounting base, weighted part, etc., are required. For example, the weight of a load cell that can weigh up to 120 kg is 200 g, but when the mounting base and weighted parts are added, it becomes about 600 g (see Figure 14). In addition to the above, the load cell for measuring several kg has a size of 80 x 12.7 x 12.7 mm and weighs about 30 g. Moreover, in order to use the load cell, a structure that supports the fulcrum with strength that does not deform even with the maximum load is required, which makes it even larger and heavier.

Figure 14.

Load cell and load cell mount.

3.5 Manufacturing method of stretch sensor

The elastomer, which had the best performance with the pressure sensor, was cut into a rectangle, molded into a width of 3 cm and a length of 5 cm using 100% prestrain, and an electrode with a width of 2 cm and a length of 3 cm was applied to the center (see Figure 15). In order to confirm the difference in performance due to the difference in electrodes, general carbon grease, carbon black, and SWCNT were used as electrodes. For the method of making carbon grease, carbon black, and SWCNT electrodes and the methods of checking those thicknesses, refer to 3.2: How to make a donut-shaped sheet DE sensor.

Figure 15.

Sample photograph of a stretch sensor (using carbon grease electrode): Measure the change in conductivity while increasing the stretch by 0.5 cm in the right direction of this sample.

3.6 Performance comparison by stretch sensor elongation

The change in conductivity was measured while increasing the elongation by 0.5 cm for the sensor using carbon grease, the sensor using carbon black, and the sensor using SWCNT (see Figure 15).

4. Results

The above various experiments were performed, and the following results were obtained.

4.1 Pressure measurement result by donut-type sheet DE sensor

The pressure measurement results of the donut-shaped sheet DE sensors prepared using three types of electrodes and four types of elastomer materials are shown in Figure 16 (carbon grease electrodes), Figure 17 (carbon black electrodes), Figure 18 (SWCNT electrodes), and those are summarized in Table 1. As shown in Table 1a, the silicon material (ELASTOSIL FILM 2030250) has a larger maximum measurable pressure than others. The acrylics have a smaller minimum measurable pressure value than silicon and HNBR. The acrylic with removed distortion and the acrylic with remaining distortion, the measuring minimum pressure value of the removing distortion was about 2 g smaller.

Figure 16.

Changes in electrostatic capacities due to load when using carbon grease electrodes.

Figure 17.

Changes in electrostatic capacities due to load when using carbon black electrodes.

Figure 18.

Changes in electrostatic capacities due to load when using SWCNT electrodes.

Material	Measurement range (g)
a) Measurement range when carbon grease is used for the electrodes.
Silicone material (ELASTOSIL FILM 2030 250)	30 ∼ 9,000
Modified HNBR	20 ∼ 8,500
Acrylic material made in the United States (3 M / 4905)	10 ∼ 8,000
The film that corrected the distortion of the US-made acrylic	8 ∼ 8,000
b) Measurement range when carbon black is used for the electrodes.
Silicone material (ELASTOSIL FILM 2030 250)	30 ∼ 10,000
Modified HNBR	20 ∼ 9,000
Acrylic material made in the United States (3 M / 4905)	10 ∼ 8,500
The film that corrected the distortion of the US-made acrylic	8 ∼ 8,500
c) Measurement range when SWCNT is used for the electrode.
Silicone material (ELASTOSIL FILM 2030 250)	30 ∼ 11,000
Modified HNBR	20 ∼ 10,500
Acrylic material made in the United States (3 M / 4905)	8 ∼ 10,100
The film that corrected the distortion of the US-made acrylic	5 ∼ 10,100

Material	Maximum measurement speed (msec)
	Carbon grease	Carbon black	SWCNT
d) Electrodes and measurement speeds.
Silicone material (ELASTOSIL FILM 2030 250)	69	66	59
Modified HNBR	80	78	68
Acrylic material made in the United States (3 M / 4905)	93	86	73
The film that corrected the distortion of the US-made acrylic	94	88	74

Table 1.

Pressure measurement results of donut-type sheet DE sensor.

Note: Each of the above measurement data was confirmed in the load cell.

The difference in electrodes did not significantly affect the measurement range. Furthermore, as in the above, there was not much difference in the detection speed due to the difference in the electrodes. Silicon is relatively faster than other membranes (see Table 1d). The DE sensor using the SWCNT electrode gave slightly better results than the others. The DE sensor using the SWCNT electrode gave slightly better results than the others. The circuit used for the experiment was fixed to the circuit shown in Figure 5 for the experiment.

Figure 19 shows the results of SS curves and the results of viscoelasticity tests (storage and loss modulus) of four different elastomers. Table 2 shows the hardness of the elastomer films and the hardness of the resin blocks used in the experiment.

Figure 19.

The results of SS curves and the results of viscoelasticity tests (storage and loss modulus) of 4 different elastomers. (a) Tensile test results for 4 types of elastomers. (b) 4 types of Dynamic viscoelasticity test results (upper figure: Storage, lower figure: Loss modulus).

Material	Hardness (Asker C)
a) Elastomers used for DE sensors
HNBR	50
Acrylic material made in the United States (3 M / 4905)	18.0
The film that corrected the distortion of the US-made acrylic	17.5
Silicone material (ELASTOSIL FILM 2030 400)	59.0
b) Resin blocks.
Material	Hardness (Asker C)
Acrylic gel sheet (Blue Forest Trading LLC:KG-01)	10
Urethane gel (EXSEAL Co., Ltd.:H5–100)	2

Table 2.

The hardness of the elastomer films and the hardness of the resin blocks used in the experiment.

Note: This time, the experiment was conducted using only the acrylic gel sheet. Next time, we plan to perform comparative verification using a softer urethane gel sheet.

4.1.1 Experiment conducted without resin cushion material

Figure 20 shows the results of measuring the capacitance by increasing the load from 10 kg to 120 kg at 10 kg intervals. Furthermore, in Figure 20, “Change in electrostatic capacity due to load” shows the measured values from 10 kgf to 120 kgf, but this is a value limited by the measurement range of the 120 kg load cell used for the measurement. It has been confirmed that the measurement range of the sheet-type DE sensor used in this experiment can be measured from about 4 kgf (using acrylic film with corrected distortion) by changing the load cell used for measurement. However, in order to unify the experimental conditions, only the values measured with a 120 kg load cell are shown in Figure 20.

Figure 20.

Change in capacitance due to load.

It can be seen that the capacitance of acrylic changes almost linearly from 10 kg to the measurement limit of 120 kg. On the other hand, silicon changed linearly from 30 kg, but the amount of change decreased from 100 kg. In the case of silicon, the change in capacitance stabilizes in about 10 seconds, but in the case of acrylic, it took about 20–25 seconds. Since the silicon used in this experiment is harder than acrylic, the region that changes linearly is narrow, but the response speed is considered to be fast. HNBR had intermediate values between silicon and acrylic.

4.1.2 Demonstration experiment to see if bolts are tightened evenly

As explained above, capacitance was measured while tightening only the lower right bolt (No. 1) (see Figure 12) by 1/2 or 1 turn compared with the others. As a result, it was confirmed that the pressure near the bolt at the lower right was higher than the others. Figure 21 is a graph showing the change in capacitance when bolts are tightened. When only the lower right screw (CH1) was rotated 1/2, the capacitance was about 22.42 pF. This is equivalent to 8.46 kg when converted to load. In addition, the capacitance when only the lower right screw is rotated once is about 24.15 pF, which is equivalent to 30.89 kg.

Figure 21.

Change in capacitance due to measured load: * the load was calculated from the capacitance based on Figure 13. * CH1: Lower right, CH2: Upper right, CH3: Upper left, CH4: Lower left.

4.2 Stretch sensor elongation comparison due to differences in the performance of different types of electrodes

Figure 22 shows the change in conductivity when a DE sensor with carbon grease, carbon black, and SWCNT electrodes attached to the acrylic that has been subjected to strain removal treatment is stretched by 5 mm. As mentioned in above, the thickness of the carbon grease electrode was 100 μm.

Figure 22.

The change in conductivity when a DE sensor with carbon grease, carbon black, and SWCNT electrodes attached to the acrylic that has been subjected to strain removal treatment is stretched by 5 mm.

The thickness of the carbon black was 70 μm and that of the SWCNT was 50 μm.

5. Discussion

Section 5.1 discusses the results of the experiment in this time, and Section 5.2 discusses the possibility of applying DE sensors to robotics and human-robot interaction (HRI).

5.1 Discussion of experimental results

First, the values obtained with all the data in Table 1a, b and c and the sensors with the four SWCNT electrodes used in the flanges were compared with the values obtained with the load cell in each case. The values of both types of sensors were almost the same. With the donut-shaped DE sensor, there was no significant difference in the measured pressure range due to the difference in the electrodes. The case where SWCNT was used for the electrode had the widest measurement range. The reason for this will be described in the stretch sensor section below. When the SWCNT electrode was used for the acrylic from which the strain had been removed, a change in capacitance appeared from the time the load reached 5 g, and a change in capacitance was confirmed up to 10,100 g. With this electrode, however, it is considered that the change in capacitance was small because the deformation of the resin block reached the limit at 10,000 g or more. With this electrode, it is considered that the change in capacitance was small because the deformation of the resin block reached the limit at 10,000 g or more. Moreover, it changes almost linearly between 500 g and 10,000 g. In the case of the sensors used the other electrodes, however, the boundary that changes linearly is narrower than that. In this sensor, the capacitance changes by about 5% by changing the test frequency from 120 Hz to 1 kHz. The reason is considered to be the influence of the frequency characteristics of carbon grease. When this electrode is used, the frequency characteristic of AC is worse than expected, and the capacitance changes more than the deformation of the elastomer.

We also conducted a demonstration experiment to see if it was possible to detect abnormal bolt tightening on the flange. After tightening the bolt in the lower right of the flange more than the other three places, we measured the pressures in four places (including where the bolt was tightened more) with the load cells and DE sensors using SWCNT electrodes, and each was compared. As a result, it was confirmed that the pressure of the tightened bolt was higher than the others. In the experiment above, the DE sensor was placed on the left side of the bolt, but for confirmation, the DE sensor was placed again on the right side of the bolt, the pressure was checked, and it was confirmed that the values were the same. When actually using these sensors, it is better to make a hole in the DE sensor and put it though in the bolt. It seems that this can be applied to various joints in the future, for example, by tightening pipes and tires that allow dangerous substances to flow, to support safer operation. The reason why the capacitance was measured by tightening the lower right bolt (No. 1) with two steps (1/2 or one turn) is that if those were suddenly tightened with a large force, they might be damaged. Therefore, two-stage measurement was employed.

In the case of the DE sensor without cushion, in the preliminary study, the film thickness of the silicon was as thin as 0.5 mm, and it was quite hard, so the range of change was estimated to be narrow. However, even with silicon, it was confirmed that the load changed from 10 kg to 100 kg linearly from 30 kg to 100 kg. The sheet-type sensor using acrylic shows a change in capacitance from a load of several kg, and changes almost linearly from 10 kg to the measurement limit of 120 kg. HNBR showed an intermediate value between them. In any case, with a load of 10 kg or less, linear displacement did not occur, and with acrylic, the limit was about 4 kg. The silicon weighed 9 kg and the HNBR weighed around 7.5 kg. When measuring a load lower than this value, it seems possible to measure it sufficiently by adding a cushion material (this discussion will be described in detail below).

When measuring cases with large pressures, the above three types of DE sensors could measure in a relatively wide range and could be effective in various applications. For example, as mentioned above, these sensors have a thin elastomer of 0.5 mm or less and seem to be effective as a sensor for checking the tightened condition of the screws in the flange part, which is often used for pipelines and plant piping. It may also be used as a sensor to detect failures in the flange due to aging or vibration. The silicon might be better than the acrylic or HNBR for applications that require faster measurement speeds.

Next, looking at the four elastomer materials individually, from the SS curve measurement results (Figure 19a) and (Table 1a and b), the silicon material (ELASTOSIL FILM 2030250) is harder than other films. So, the film seems to have a relatively large measurable pressure change range. That is, there is a drag against pressure, and the film cannot be easily crushed by pressure. The measurement results of dynamic viscoelasticity support this (Figure 19b) [26]. In addition, the hardness of the four types of elastomers above was measured using Askar C, and the results were in the order of silicon, HNBR, acrylic with strain, and acrylic with strain removed. Thus, evidences support that the explanation above is correct (See Table 2).

As acrylic deforms greatly, it can be detected from the smallest value and can be measured from 5 g. In other words, acrylic is so soft that it can quickly detect even the slightest pressure.

That is, as described in eq. (1), when the elastomer is crushed by pressure or some forces and the electrodes are closer to each other, a larger capacitance can be obtained. On the other hand, with the hardest silicon, no change was observed at about several g, and a force of about 30 g was required.

The difference in electrodes did not significantly affect the measurement range, but in terms of measurement speed, silicon was slightly faster than other films. The reason is that the silicon film is harder than others, so even if it is pushed by pressure, it returns quickly to its original state [26]. The speed of acrylic is a little slower than that of silicon, but it seems to be sufficient for practical use. As shown in Table 1c, SWCNTs with good conductivity showed slightly better results. The reason for this is the same as above, and the better the conductivity, the larger the Coulomb force that can be generated, and as a result, the film is distorted quickly, and the reaction is quick to return. There was little difference in measurement speed depending on the electrodes, but it seems that even a small difference in speed could be a key factor for sensors used in smaller devices such as mm size.

By using a very soft resin cushioning material installed in the center of the donut-shaped sheet DE sensor used, it was possible to measure even with a slight pressure, which led to the success of this experiment. Considering the hardness and elongation of the abovementioned elastomer, it can be inferred that the measurement range can be changed by changing the material of the cushioning material installed on the elastomer film [29]. As mentioned in the Background of dielectric elastomers (DEs) above, if the shape is semicircular, it could come into contact with the object to be measured faster. In such a form, as the pressure or pushing force increases, the ground plane of the elastomer becomes larger by that amount, and it is considered that a sufficient effect in terms of capacitance is produced. In this experiment, a small cushion was sufficient without having to make such a semicircular sensor shape. In this experiment, we did not conduct an experiment to change the shape of the cushion material. If the shape of the cushion material is such that the convex lens is viewed from the side, it is considered that the pressure range width can be increased while maintaining the small shape.

Furthermore, we will conduct the experiment again using the ultra-soft modeling resin: H5-100/H5-600J] manufactured by EXCEL Co., Ltd., as the new cushioning material. This material is softer than the resin used this time, so it seems that it would be possible to detect even with a smaller pressure (see Table 2b). The minimum measurement depends on the hardness of the elastomer used in the donut-shaped sheet DE sensor, so it is necessary to use a cushion to stretch the elastomer as described above. However, in this respect, acrylic, which has a small minimum measurement value, could be an effective material.

As an interesting result, the pressure measurement curve shape from the sheet type and the donut-shaped DE sensor are opposite of each other. The reason is that in the case of a donut-shaped DE sensor, the elastomer is stretched by the resin block, and then an external force is applied, the resin block shrinks, the elongation of the elastomer is attenuated, and the capacitance is reduced (See Figures 16–18). In the sheet type, an external force is directly applied to the sheet, and the entire sheet is stretched, whereby the amount of electrostatic charge increases (see Figure 19a). The reason for conducting the donut-type and sheet-type experiments was that we wanted to prove that the donut type could handle the pressure of the gram order to the 10 kg order, and the sheet type could handle the pressure of the 10 kg order to the 100 kg order. As shown in the results of this experiment, this is because that it might be able to serve as a guideline for proper development within industries.

As shown in Figure 19a, one more interesting finding is that when comparing the SS curves of a film from which the strain of the film has been removed with 3 M acrylic and the SS curve of the film without removing the strain, the curves are almost the same to some extent, and the breaking stress is also almost the same. However, it turned out that the final elongation was different. This is thought to have caused a difference in the pressure measurement range and measurement speed in this experiment. It is presumed that the film was more uniform and therefore more easily stretched (see Figure 19a). In addition, since 100% pre-stretching was applied, the hardness of the acrylic became moderately hard, and dynamic viscoelasticity was also moderately present (see Figure 19b).

The various DE sensors in this experiment can measure any pressure changes from deformation instantaneously within the measurement range. It might depend more on the circuit design. The circuit used this experiment was the same to the circuit mentioned in Section 3.3. If a circuit having a higher degree of amplification is used for this circuit, even a smaller change can be measured, so that the measurement range can be expanded. As mentioned above, the selection of the elastomer material and cushioning material used for the sensor is important for improving the measurement speed, but the tune-up of the circuit is also important for achieving a faster measurement speed.

In 2011, the authors of this paper, in collaboration with the Japan Agency for Marine-Earth Science and Technology (JAMSTEC), examined a DEA that can be driven even on a 100,00 m deep seabed [30, 31]. Using a pressure chamber that could reproduce the pressure of the seabed of 10,000 m, a roll-type DEA was manufactured by rolling a sheet-type DE using an acrylic material, and when an experiment was conducted, it was able to be sufficiently driven even under a pressure of 10,000 m. We believe that the DE sensor can be sufficiently driven as a sensor even under a pressure of 10,000 m if an elastomer with sufficient thickness is used and a cushion material with an appropriate thickness is selected.

Next, regarding the stretch sensor, the silicon material is harder than other films, and as a sensor, the operation speed is fast. However, since the elongation width is small, it seems that the intended use could be limited. Materials with low dynamic viscoelasticity, such as silicone, might not be well suited for DEA/sensor materials [26]. In comparison, acrylic is softer and more stretchable than other films. It could be suitable for DEA/sensor materials. The reaction speed is also sufficient for practical use, and it seems to be optimal as a sensor that matches human movements. In addition, the pressure change range that can be measured is large. Acrylic’s upward-sloping dynamic viscoelasticity supports this evidence (see Figure 19b).

The difference in conductivity between carbon grease, carbon black, and SWCNT electrodes is clear, and SWCNT is the most suitable for stretch sensors (see Figure 22). The SWCNTs are more conductive, but not only that, SWCNTs are packed in a spray and sprayed to achieve a uniform, thinner electrode [25]. If you look at a commercial basis, however, the cost of carbon black is much cheaper than SWCNTs and multi-walled CNTs (MWCNTs). So, this choice should not be disregarded.

In 2012, we created a system to check how much a finger bends by attaching a DE sensor with a carbon black electrode on a non-distorted acrylic film to a rubber glove. (See Figure 23) [31, 32, 33].

Figure 23.

Rubber glove with DE sheet sensor.

However, due to the carbon black, the stretch sensor that follows the movement of the finger did not stretch sufficiently. So, we would like to verify it again with a system that reflects these data.

5.2 Application to robotics and human robot interaction (HRI)

The best use of the DE pressure sensor is to apply it to finger pressure sensors in robots or magic hands. This type of sensor is thought to be able to measure pressure ranging from 1 gram to about 150 kg. Additionally, it is extremely flexible, thin, and small, so dexterity rivaling that of human fingers could be realized. When working with humans, it goes without saying that a robot with humans-like abilities would be easier to work with. S. Chiba and M. Waki used a DEA to create a model in which a robot finger is driven by tendons similar to those found in a human finger. Next time, we plan to incorporate DE sensors into this model to create a finger that is more like a human finger and has senses [34].

If this stretch sensor is used well, it can be used as a three-dimensional position sensor for the arms and legs of robots and the like. The principle of 3D position sensor will be explained step by step below.

A stretch sensor is attached to the upper and lower sides of the robot’s arm to move the arm upward. The upper sensor is extended and the lower sensor is the contracted. As a result, it is possible to two-dimensionally determine at what angle the arm is bent. Furthermore, by deploying another pair of sensors on the side of the arm, it is possible to sense diagonal movement. That is, the three-dimensional position can be easily determined from the difference between the upper and lower capacitance and the difference in the capacitance on the side surfaces. For this, a calculation table might be helpful. In the next experiment, we plan to attach such a sensor to the robot. As it is now possible to lift an 8 kgf weight by 1 mm or more at 88mms with 0.15 g (0.96 g including reinforcement material) of acrylic [25, 35, 36], we will use it for this system. If you created a finger model that incorporates DEA and DE sensors, it seems that you could make a human-like finger. In addition, it seems that the arm with three pairs of the above sensors could detect the speed of movement of the arm. The expansion and contraction speed of DEA moving in the three-dimensional direction can be calculated from the change in capacitance and the position of the arm (based on the information from the position sensor). Alternatively, each value could be imported as a variable into such a calculated table. Environmental resistance is also an important factor for the actual use of robots. Regarding the environmental resistance of DE sensors, silicon with SWCNT electrodes and HNBR sensors can be used at temperatures 150–200°C [37]. At low temperatures, silicon is said to be usable down to around −100°C [38]. If it could be used in this temperature range, it might be expected to be used as a robot sensor for space environments, deep seas, and extremely cold regions.

6. Conclusion

The difference in the type of elastomers, the thicknesses and shapes of the cushioning materials placed under the elastomer films, the difference in the presence or absence of distortion of the films, the effect of the difference in the type of electrodes were examined, and the following conclusions are obtained:

• The silicon sensor has a small deformation, but the reaction speed as a sensor is fast.

• The acrylic sensor has a large amount of deformation, and the reaction speed as a sensor is practically sufficient. Therefore, it is suitable for stretch sensors. In addition, as a pressure sensor, it can measure pressure in a certain range width steplessly.

• In acrylic, compared with strain-removed films and the untreated films, the treated films had better performance.

• Compared with the difference in measurement speed between each electrode, but there was no significant difference.

• When four acrylic sensors using SWCNT as electrodes were installed and a bolt was tightened in a certain place more than the other three, the pressure was higher than the other three.

• The cushioning material allows you to extend the measurement pressure range. However, even in the case without cushioning materials, the sensors using the SWCNT electrodes were able to measure the pressure accurately in a large pressure range of about 10 kg or more.

Acknowledgments

We would like to thank Mr. M. Uejima, Mr. H. Uchida, and Mr. M. Takeshita of ZEON Corporation for providing SWCNT (ZEONANO® -SG101) and HNBR free of charge for carrying out our experiment.