With the development of miniaturized and portable computing devices and sensors and burgeoning interest in personalized healthcare and new concepts in human–machine interaction, smart wearable devices [1
] have attracted wide interest because of their potential in health monitoring [4
], motion tracking [6
] and assessment of in vitro models of gastric motility [12
]. Among these devices, “electronic skin” [14
] has developed as a prominent class. Beginning in the 1970s, the concept emerged in science fiction, and in the new century, with the improvement of science and technology, electronic skin began to move from fiction to reality [18
Early electronic skin consisted primarily of non-conductive elastomers and flexible conductive sensing elements. Their sensing mechanisms were also relatively simple and were generally divided into two types: Resistive [6
] and capacitive [20
]. In recent years, many researchers have pursued development of capacitive electronic skin, with carbon nanotubes (CNTs) [6
], graphene [26
] and carbon black [29
] often used as conductive elements, while others have used silver nanowires (AgNW) [21
] or metallic nanoparticles [32
]. Silicones [35
], spun elastic fibers [36
] and conventional textiles [37
] have been used as dielectric layers. Among them, the silicone poly(dimethyl siloxane) (PDMS) [35
] is most often used as an inexpensive non-conductive elastomer, with its relative ease of fabrication and chemical modification and its low glass transition temperature contributing to its high flexibility [38
]. All these systems have depended upon attachment of the conductive material to the surface of the flexible substrate, by spraying or other deposition methods [6
]. Sensitive devices have been produced, but challenges remain, such as durability. The interface between the flexible substrate and its conductive coating must maintain its integrity, which can be challenging due to the risk of delamination and shedding [46
]. Some researchers have reported endurance testing for durability [40
], though typically over short durations. Other researchers have employed elastomeric ionic hydrogels or ionomer polymers as their conductive medium [49
], pursuing long-term stability and biocompatibility. Ionic sensors [49
] do not suffer from low-motion artifacts and have high sensitivity to external mechanical stimulation, but their production processes are relatively complex and may not be suitable for mass production. Therefore, it is still challenging to develop a flexible, conductive sensor material that is highly durable, offers simple assembly, has sufficient performance, is suitable for large deformation and flexion angle human motion and yet is low cost.
The conductive mechanism of carbon black [30
] (CB) is explained by seepage theory [51
], where as long as the distances between the carbon particles are small enough, conduction can be achieved. Guo et al. [29
] developed a multilayer (10) textile-attached capacitive pressure sensor that used a PDMS/CB blend as its dielectric layer, with PDMS as shielding layers and organo-silicone adhesive as electrode layers. Tsouti et al. [30
] suggested that it was possible to develop a simpler arrangement for a PDMS/CB based capacitive strain sensor, where a PDMS/CB blend was used for the electrodes and PDMS as the dielectric. However, their experimental work was limited to very low strain (0.05%) measurement of a cantilever. In 2018, during the development of this current work, Shintake et al. [53
] published a characterization of a similar five layer arrangement (Ecoflex™ silicone dielectric layer sandwiched by blended silicone/CB conductive layers and packaged in silicone layers) over large scale deformation (500% strain) and over high numbers of repetitive strain cycles (10,100). This study also included sensor integration with a glove and monitoring of finger bending over ~20% sensor strain.
In this paper we demonstrate how similar materials, in a simple five layer arrangement, can be made effectively at first hand and can be used in a large-strain (50%), durable, low-cost sensor that is suitable for use as a wearable device for monitoring high flexion-angle body motion and large tissue deformation. The device was prepared by sequential casting and curing of the silicone layers—a simple, low-cost, industrially relevant method. Durability during large strain (50%) repetitive motion was established via testing over 10,000 tensile strain cycles, while body motion monitoring was demonstrated by fixing planar sensors to paper elbow and knee bands, with tissue deformation examined via fixing a sensor to the silicone stomach of an in vitro model of human digestion. [54
] The latter, by monitoring the peristaltic contractions that drive gastric motility, opens the possibility of enhancing in-demand in vitro models of gastric digestion [12
]. Additionally, the effects of temperature and humidity upon sensor performance were examined.
3. Results and Discussion
An approximately 1 mm thick, five-layer, flexible, PDMS based capacitive strain sensor was fabricated by sequential addition and curing of the layers. This process is low-cost and simple and is amenable to commercial scale-up via translation to roll-to-roll manufacturing, or other polymer film casting methods. In Figure 2
a a cross sectional view SEM micrograph of the sensor is given, with the insulating layers marked as I and the flexible conductive silicone rubber (25% CB, see in Table A1
) layers marked as II. The central insulator layer does not contain conductive carbon black and is the dielectric material and the outer two layers form insulating encapsulation, or packaging, layers for the sensor. The sensing mechanism is as follows: When the elastomeric sensor is stretched, the thickness of the dielectric layer decreases and the area of electrodes (A) increases, resulting in a significant increase in capacitance [22
]. This is described by Equation (1). From the SEM micrograph, it can be seen that, from top to bottom, the individual layers were 175, 275, 200, 275 and 150 µm thick. Variations in the packaging layer thicknesses were due to variations in the bench-top fabrication method employed here. From Figure 2
b, the overall thickness is 1.08 mm, consistent with the SEM image. The performance, fabrication method and characterization of comparable published flexible capacitive sensors are summarized in Table 1
is capacitance, ε0
are the dielectric constant of a vacuum and the relative permittivity of dielectric media, respectively, A
is area of electrodes and d
is thickness of the dielectric layer.
summarizes the conductive properties of different blends of PDMS and CB. It can be seen that as carbon content of the PDMS blends was increased from 15% to 30% electrical conductivity gradually increased, with useful conductivity achieved with 25% and 30% CB. Though it was the most conductive, the 30% CB silicone was difficult to prepare and exhibited rough surfaces, so 25% CB was used for sensor fabrication.
a gives sensor ΔC/C0
values at fixed temperature [57
] (15, 25, 37 °C) and varying relative humidity (20, 40, 60 %RH), over a range of 0 to 50 % elongation. Capacitance values measured at 25 °C and 40% RH ranged from ~260 to 380 pF and the total range of measurements was ~200 to 500 pF. These relatively large capacitive values are easily measured with miniaturized equipment and require much less sensitive data acquisition systems than some other [32
] published sensors. For example, finger presses on smart phone capacitive touch screens typically register on the order of 1 pF change [58
]. This also gives scope for device miniaturization.
It can be seen from Figure 3
a that the ΔC/C0
varies little with humidity [59
] when temperature is constant, with the most difference apparent at 25 °C. Conversely, Figure 3
b gives Sensor ΔC/C0
values at fixed relative humidity (20, 40, 60 %RH) and varying temperature (15, 25, 37 °C). Again, little variance is displayed, with perhaps the most at 60 %RH. PDMS is water permeable [60
] and will swell with water [61
], but this effect was not apparent. The exterior PDMS packaging layers may have effectively isolated the internal conductive and dielectric layers, preventing significant influence from water vapor. Similarly, temperature has little influence over the measured range. Therefore, it can be concluded that temperature and humidity have at most only a minor effect on the sensor’s capacitive strain detection. These effects are unknown for most published sensor systems.
The sensor’s capacitive Gauge factor (G), defined by
, was also assessed at the different temperature and humidity combinations, with the values given in Table 2
. It was very stable and within the median range for for published systems (~0.5 to 1.0). Table 1
makes this comparison with several comparable publications.
a,b is SEM cross-sectional views of the sensor taken before being subjected to 10,000 strain cycles. It can be seen from the figure that the surface of the sample is relatively flat before stretching. Figure 4
c,d is SEM cross-sectional views of the sensor taken after endurance testing at 25 °C (50% strain). Small rough patches of ~10 μm size are infrequently observed. Figure 4
g,h shows the SEM profile of the sensor taken after endurance testing at 37 °C (50% strain). Again small rough patches of ~10 μm size are sparsely distributed. These do not appear to be distinct cracks [47
], but rather may be carbon particles disturbed by the strain-cycling, though it is not possible to be certain without further chemical imaging. Microcracks may have occurred that were not visible at this scale of magnification. Most importantly, sensor conductive layer morphology did not appear to change appreciably over 10,000 strain cycles of testing, demonstrating a seldom seen level of durability.
Consistency of sensing performance with use was established by examining repeated strains at various elongations and frequencies, after 0 to 10,000 strain cycles. These results are illustrated in Figure 5
, where consistent magnitudes of relative change in capacitance were observed. More variance within these strain cycle test groups was observed at the 0.5 Hz cycling frequency, but this may have been due to the limitations of the measurement, which restricted signal sampling to the same 0.5 Hz frequency.
It can be seen from Figure 5
that the sensor stretched at 0.5 Hz with 50% elongation for 100 times, 500 times, 1000 times, 5000 times and 10,000 times, separately, perform well. The value of the ΔC/C0
is 0.35. Even with different strain frequency, the high value of the ΔC/C0
is also 0.35. This consistency is observed again in Figure A2
, Figure A3
and Figure A4
. In Figure A3
, with 20% elongation, the value of ΔC/C0
is consistently 0.14, which does not vary with strain frequency. This is also true of 10% elongation results of Figure A4
and the 5% elongation values of Figure A5
, with ΔC/C0
is also 0.07 and 0.035 respectively.
Finally, Figure 6
displays side-by-side repeated strain cycles, recorded at different frequencies after 10,000 cycles, for 50% elongation. Again stability of signal and durability, are displayed by the modest variance in capacitive change. This confirms the results of Shintake et al. [53
] with a similar 5-layer PDMS/CB sensor, who earlier demonstrated durability over 10,100 cycles. These are the only known studies that have examined durability to this extent and have demonstrated wearable applications of their sensor materials.
Cohen et al. [20
] also demonstrated signal stability with their CNT based capacitive strain sensor, but over 3000 strain cycles, while He et al. [27
] showed similar signal stability for their graphene based capacitive pressure sensor over 1050 cycles of strain. The greatest duration study of capacitive sensor stability was performed by White et al. [56
], who examined an IGT/silicone blend and silicone based multi-layer sensor over 100,000 strain cycles, observing considerable performance deterioration after 25 k cycles. This indicates that even further durability testing may need to be performed in future.
displays the measured capacitance of a wearable version of the sensor, when it is at rest on a table and when it worn on the stationary elbow of a human subject, where it is in contact with the skin. Wearability was achieved by integrating the sensor with an elbow band. Mean capacitance was 259.4 ± 0.2 pF on the bench, where the standard deviation is equivalent to ΔC/C0
of ±0.0008. Mean capacitance was 266 ± 1.5 pF on the stationary elbow, where the standard deviation is equivalent to ΔC/C0
of ±0.0056. Attaching the sensor to skin had the effect of a one off capacitance increase of ~2.5%, easily compensated for. Increased signal variance from skin contact had little impact [62
] on the measurement result.
a illustrates monitoring of elbow flexion angles of 30°, 45° and 90°, during repetitive motion. The capacitance varies from ~260 to 285 pF for bending from 0–30°, from ~260 to 310 pF for bending from 0–45° and from 260 to 375 pF for bending from 0–90°. As the flexion angle increases, the maximum value of the capacitance increases accordingly. As the bending angle becomes larger, sensor area increases and the spacing between the two flexible electrodes becomes smaller. This relationship is in keeping with the capacitance formula C = ε0
], where ε0
is the space permittivity, εr
is the relative dielectric constant of the dielectric material, A is the area of the capacitor and d is the distance between separated electrodes. The sensor appears to report fairly repeatable and stable signals during testing at different flexion angles, with much of the variance likely due to varying degrees of motion from the human test subject, or the 0.5 Hz sampling rate limit of the data acquisition system. Relative signal changes, of ~40% over the full range of motion, given by ΔC/C0
, give clear indications of this motion and due to the 0.86 gauge factor, are comparable to the highest range observed in this field [64
The results of testing of the sensor at different elbow flexion angles and at two different repetitive movement frequencies for each flexion angle are shown in Figure 8
b. It can be seen that the characteristics of these under these different motions can be identified well, reflecting the movements of the human arm. In addition, the characteristics are generally repeatable for the same maximum bending angle and the magnitude of capacitance change increases with increasing flexion angle. All of the above suggest that the current sensor, despite its simplicity, functions well for large angle bodily motion detection.
Further characterization of the device for large flexion angle motion was achieved via fixing it to knee band and monitoring capacitance during running and squatting and rising. This is illustrated in Figure 9
. The motions are clearly distinguished, with relative signal changes up to ~90%, again comparable with the highest range observed in this field. [64
] Together, these results represent the first application of a CB/silicone based capacitive strain sensor to this class of motion.
A final application of the sensor material was used to gauge its suitability for monitoring large scale soft tissue deformation, of the type that would be seen with mammalian organs. This was achieved by laminating the sensor to an anatomically realistic, silicone stomach model, from an in vitro human digestion simulation system [54
]. Sensor position is illustrated in Figure A2
from Appendix A
and is located on the anterior body wall. The peristaltic motion that drives gastric motility is simulated by applying rollers to the model and a brief video of this process is given in the Supplementary Material
. The stomach is initially compressed in the default position, which increases as the rollers move, eventually moving past the model and allowing it to relax. The rollers then return to a position at the other end of the model, where they change direction and resume the original motion. The compression and relaxation phases of this process are easily distinguished in Figure 10
, a capacitive signal trace of several iterations. A video of the process has been included in the Supplementary Material
This experiment further demonstrates the suitability of the sensor material to monitoring large scale human motions, this time through compression deformation of soft tissue and may be the first such application to this type of motion. Currently this study has been limited to in vitro models, which indicate that the material can be used in these systems, or externally on the body. Further development may include surface modification for use in animal models of disease.