PDMS multimodal deformation sensors

Accepted Manuscript Title: Effect of contact material and ambient humidity on the performance of MWCNT/PDMS multimodal deformation sensors Authors: Kaur Leemets, Uno M¨aeorg, Alvo Aabloo, Tarmo Tamm PII: DOI: Reference:

S0924-4247(18)30722-2 https://doi.org/10.1016/j.sna.2018.09.042 SNA 11017

To appear in:

Sensors and Actuators A

Received date: Revised date: Accepted date:

27-4-2018 14-9-2018 17-9-2018

Please cite this article as: Leemets K, M¨aeorg U, Aabloo A, Tamm T, Effect of contact material and ambient humidity on the performance of MWCNT/PDMS multimodal deformation sensors, Sensors and amp; Actuators: A. Physical (2018), https://doi.org/10.1016/j.sna.2018.09.042 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Effect of contact material and ambient humidity on the performance of MWCNT/PDMS multimodal deformation sensors

IP T

Kaur Leemets,*a, Uno Mäeorg b, Alvo Aabloo a Tarmo Tamm a,

IMS Lab, Institute of Technology, Tartu University, Nooruse 1, 50411 Tartu, Estonia

b

Institute of Chemistry, Tartu University, Ravila 14a, 50411, Tartu, Estonia

SC R

a

N

U

* Corresponding author e-mail: [email protected]

CC E

PT

ED

M

A

Graphical abstract

Highlights

A



Humidity effects the signal of resistive and capacitive PDMS/MWCNT deformation sensors.



The effect of humidity strongly depends on the contact materials used



Carbon mesh contacts outperform copper in terms of both electrical and mechanical stability



13-layer mechanical sensor distinguishing between elongation and bending simultaneously.

Abstract The soft sensors and electronic skin concept often relies on composites based on an elastomer

IP T

(like PDMS) with some conducting fillers (like carbon nanotubes). In this paper, the ambient moisture sensitivity of MWCNT/PDMS composite multimodal mechanical sensors is

SC R

demonstrated and tested in both resistance and capacitance measurements. In addition to the sensitivity of the sensor material itself, the contact material used in the construction was

found to play an important role. For our sensor design, the effects on the resistive part of the sensor are balanced out by using the Wheatstone bridge setup. While small, the effect of

U

ambient humidity to the capacitive part of the sensor does indeed exist. For obtaining

N

accurate deformation readings, the effect of humidity needs to be taken into account

M

A

whenever dealing with PDMS based sensors.

ED

Keywords

Introduction

CC E

1

PT

Deformation sensor; PDMS; MWCNT; humidity; moisture; contact material

Electronic skin or e-skin has been a popular research subject for a few decades whether it is for human machine interfacing [1–3], motion tracking [4–7] or prosthetic skin [5]. The most popular materials used in these e-skins are polydimethylsiloxane (PDMS) [1,8–11] and

A

different kinds of textiles [4,7], whether they are just incorporated in the sensor skins as a structural material or as part of the sensor [1,4]. In the course of operation, these kinds of sensors are exposed to changing ambient conditions, as they move in different environments along with the wearer. Therefore, the effect of ambient conditions to the output signal of the sensors must be understood in order to avoid faulty reading caused by change in, for example humidity, as such sensors are seldom ideal.

An ideal sensor has high sensitivity and selectivity, being sensitive to everything and differentiating between all of the stimuli. In reality, sensors are never fully selective and certain measures have to be taken to limit the effects of other stimuli that could affect the sensor output. For instance, carbon based fillers are popular in resistive sensor elements for mechanical sensing but they are also sensitive to temperature. Furthermore, different carbon materials can have different sign for the thermal coefficient. This can be beneficial, as a practically zero temperature coefficient of resistance has been achieved by combining

IP T

multiwall carbon nanotubes (MWCNT) and carbon back based conductive PDMS

composites[12]. It has been shown that the resistivity of pure CNT films increases with the

SC R

presence of solvent vapor, including water vapor[13,14], in the environment. We will demonstrate that the resistivity of MWCNT/PDMS composites is also sensitive to

environment humidity. Capacitance is another sensing signal that is used for both mechanical sensors and environmental sensors. Polyimide, for example, is popular for capacitive

U

moisture sensors as it has a large water uptake from air [15], [16]. PDMS has high

N

permeability to water vapor as well [17–20], thus, capacitive sensors fabricated using PDMS

A

matrix can be sensitive towards water vapor content in the surrounding air.

M

While our previously reported multimodal sensors have shown good stability and reproducibility in ambient conditions [21], the possible effects caused by water vapor

ED

permeability should also be addressed. Both moisture from the ambient humidity or emitted by the wearer may potentially affect the accuracy of the multimodal sensors.

PT

The aim of this study was to quantify the effect the environment humidity has on PDMS composite sensors using our previously reported sensor[21] as an example system. It was found that indeed due to the vapor permeability[20] of PDMS, environmental humidity

CC E

influences the sensor output which must be dealt with, for example using a Wheatstone bridge setup or adding an calibration sensor. The sensors under study were elastic multimodal shape sensors able to measure elongation of the sensor and its curvature simultaneously,

A

while distinguishing between the signals. The sensor achieves that by incorporating a Wheatstone bridge resistive sensor and a capacitive sensor in its layered design. A full Wheatstone bridge setup is used to measure the bending of the sensor while the capacitive part is used to obtain the elongation response. As the materials used in the construction of this sensor, namely PDMS and CNTs, are currently very popular in soft sensors, this research can potentially be beneficial to a wider soft sensor community.

2

Material and methods

Composite sensors tested in this work where fabricated from PDMS (Quantum Silicone QM240T) and MWCNT (Bayer Baytubes C 150 P). The PDMS was chosen as it is rated for 350 % elongation and 83 N/cm tear strength, which was found suitable for the application A detailed description of how the sensors were made can be also found elsewhere [21]. A planetary ball mill (Retsch PM 100) was used for mixing the MWCNT and PDMS together

IP T

with the help of a small amount of chloroform to form the conductive silicone which was used for the conductive elements in the sensor. Pristine PDMS was used for the isolating

parts of the sensor. Contacts for the sensor were made from non-woven carbon fiber matt (60-

SC R

70 µm thickness) or copper film (20 µm in thickness) by cutting them into shape with a Silhouette Portrait® electronic cutting machine.

Sensors were cast into a custom compression mold one layer at a time alternating between

U

conductive and isolating layers and curing at 120 °C for 15-20 minutes in between layers.

N

When the stacked layers are connected as shown on Figure 1, they form a Wheatstone bridge

A

and a capacitive sensor combination that can measure the elongation of the sensor and the bending of the sensor independently and simultaneously. For bending detection, the full

M

Wheatstone bridge is used consisting of the 4 layers: 2 outer layers on each side of the sensor. When the sensor is bent the convex side layers are stretched and the concave side layers are

ED

slightly compressed (or stay unstrained), resulting in a voltage change between the sensing lines marked on the figure by S23 and S14. For elongation detection, the capacitance between

PT

the 2 innermost layers is used. Capacitance increases with elongation as the layers get thinner and their surface area increases.

CC E

The conductive and isolating layers that make up the sensor had thicknesses of either 0.3 mm or 0.1 mm, the MWCNT loadings in PDMS of 2% and 5% by mass were used for the conductive layers. The most complex version consisted of 13 layers: six conductive and

A

seven isolating layers. Most of the measurements reported here were performed with sensors having fewer layers. Figure 2 shows a photograph of a 13-layer sensor and a SEM image of the samples cross-section, taken from one end of the sensor, this way the contact material fibers can be used to distinguish between conductive and isolating layers. Qualitative tests on the effect of environment moisture were conducted by introducing the sensors to different environments. For dry environment, glovebox (Vigor SciLab Single

Station Glovebox) with dry nitrogen atmosphere was used, where the water content stayed below 1 ppm for the extent of the measurements. For the measurements made under ambient conditions, the relative humidity and temperature where logged (using PCE Instruments, PCE-THB 40). For high humidity (“Humid” on the figures), an equilibrium was maintained between water vapor and liquid water in a sealed glass aquarium. Schematic representation of the setup can be seen on Figure 3. To evaluate the effect of moisture on the sensors, impedance measurements were conducted on the resistive layers of the sensors using

IP T

IviumStat Compactstat.e. The measurements were conducted at two different intervals, once a day and once an hour. Due to instrumental limitations, once per hour measurements were

SC R

conducted on one sample at a time, while once per day measurement series were conducted on different sensors in parallel. To mitigate the effect of long transmission lines in the glovebox, impedance was measured at a relatively high AC amplitude of 1 V. To get as

comparable results as possible, all samples were measured at the same AC potential and the

U

current measurement region of 1 mA.

N

9 samples were taken for the daily measurements. 3 for each of the 3 environments (high

A

humidity, low humidity and control in ambient humidity). For each, one sample had copper

M

contacts and 2 resistive layers, one had carbon contacts and 2 resistive layers, and one with 1 resistive layer and carbon contacts (only single layers resistance was measured at a time). Once per hour measurement were made with the samples previously measured in the

ED

glovebox’s dry atmosphere. This assured the samples were fully dry before the experiment.

PT

To compare different contact materials, a test sample was made, which had carbon mat and copper contacts in pairs for one piece of the conductive PDMS MWCNT composite layers. This was done to eliminate differences other than those of the contacts used. For evaluation,

CC E

impedance spectroscopy was used. Quantitative measurements were carried out in an environment chamber, where the moisture content of the nitrogen atmosphere was controlled from 10% to 90% in 20% increments.

A

Humidity was controlled with a humidity generator (ProUmid MHG32) with at least 30 min equilibration time at each humidity level. The schematic representation of the setup used is presented in Figure 4. For this test, the sensors’ capacity was chosen as the signal to evaluate. While the resistance of the sensor layers is also effected by humidity, that effect will be negated by the applied Wheatstone bridge setup , as all the sensor layers experience the environment effects

simultaneously[22,23]. The sensors capacitance on the other hand is not in a bridge setup, thus, it will be susceptible to environmental effects. A freshly made sample was used for these experiments.. The capacitance was logged with a bench LCR meter (ROHDE and SCHWARZ HAMEG HMS8118) at 120 Hz. The sensor was also stretched during the measurements to determine whether the air humidity has an effect on the sensors responsiveness. For every humidity level, the sensor was stretched and released for two cycles in increments of 0.5 mm up to 5 mm elongation. For stretching the sensor, a custom

IP T

3D printed vice-like device was used (depicted in Figure 5). Both ends of the sensor were clamped down, leaving 1 cm between the clamps free to stretch. The clamps were moved

SC R

apart in 0.5 mm steps. This corresponds to stretching the middle part of the sensor in steps of 5%. The data was averaged over the two cycles. Later the data was fitted both elongation versus capacitance at different humidity levels, and humidity versus capacitance at different elongations. This fitting also gave the standard deviation. The R2 was calculated by

U

correlating the measured data to theoretical data calculated using the humidity effect formula

Results and discussion

M

3

A

N

derived from the measurement data.

Dry samples that were placed in a high relative humidity atmosphere showed an increase in resistance, especially those with copper contacts. With increasing humidity, the sensors with

ED

copper contacts had significant increase in low frequency impedance, the part most likely due to the contact. Figure 6 compares the Bode plots of samples of 2% MWCNT loading with

PT

carbon contacts and copper contacts. The overall resistance rose for both types of sensors under high relative humidity, but the sensor with copper electrodes also showed a change in

CC E

the overall shape of the frequency response. The most significant (over 30%) increase of impedance was found in the low frequency part of the Bode plot), the change in the shape indicates a change in the charge transfer mechanism.

A

The resistivity of samples placed under the dry atmosphere of the glovebox decreased slightly. The decrease was smaller than the increase of the samples in high humidity, and in the same order of magnitude to the signal level differences due to the reconnecting of the samples. This is most likely caused by the fact that the samples were cured in a high temperature oven which already decreased their moisture content prior to moving the samples to the glovebox. They were further dried in the vacuum cycling of the antechamber of the glovebox. Overall, from the long-term tests under different moisture levels, it can be

concluded that both the resistivity of the material and that of the contact interface increases with increased moisture content. To confirm the result of the long-term measurements, a shorter timescale experiment with hourly measurements was made with similar samples that were previously kept under the dry atmosphere of the glovebox. From these measurements (Figure 7 and Figure 8), a similar conclusion can be drawn that the sensors with carbon contacts were more stable than those

while the resistance of the samples with copper contacts changed in time.

IP T

with copper contacts as their resistance remained virtually constant after the initial jump,

SC R

While the samples were enclosed in an aquarium isolating them from outside air, the changes of outside temperature could still affect the samples. To exclude the possibility of

inconsistent room conditions affecting the result, the air temperature of the lab was logged

U

and compared to the resistance change, no correlation between the two was found. Figure 7 shows the response of the samples with carbon contacts to high humidity (closed

N

glass aquarium with water/water vapor equilibrium). The “dry” measurement on the Bode

A

and Nyquist plots (A and C) was measured with the sample still in the glovebox. After being

M

transferred from dry atmosphere to the glass aquarium with high humidity atmosphere, there is an initial jump in the impedance that takes place in the first few minutes, following which

ED

the impedance remains relatively stable with a small increase during the first few hours. It can be said that the initial jump is very fast as it had occurred by the time the first measurement outside the glovebox was started. The peak at the 28 kHz is an artefact due to

PT

the measurement setup used[24]. In reality the shape of the “Humid” state curves should be similar to the “Dry” state curve only with a higher absolute value. We can see the values

CC E

change similarly for all the chosen regions and the main change in the impedance occurred during the first hour.

The Bode and Nyquist impedance plots of the short-term experiments on a sample with

A

copper electrodes are presented in Figure 8A and C, respectively. From the Nyquist plot we can see that the two different charge transfer processes are present even in the dry environment, but they become fully distinguishable in the high humidity environment. In Figure 8B, the real part of the impedance at selected frequencies is presented. It can be seen that after the initial jump in the resistance, the high frequency impedance exhibits very little change. At low frequency, where the proposed sensor would be measured in application scenarios, the real part of the impedance is not very stable with gradual increases and

decreases in the course of days. The peaks in the lab temperature readings (Figure 8, D) do not correlate with the features of the impedance. The measurements of the samples with different contact material indicate that two different conduction mechanisms (stages) are present when using copper as the contact material. In the PDMS-CNT mixture, the conduction can be simulated as small resistances and capacitances in parallel. At the contact region, there is a larger capacitance and larger resistance in parallel.

IP T

These two systems connected in series simulate the whole system. The equivalent circuit is shown in the inset of Figure 9. In case of samples with carbon contacts, one of the two

conduction stages disappeared and the overall resistance was lower. Therefore, the current

SC R

flow from carbon contacts to MWCNT is similar enough to the current flow between

individual nanotubes, the impedance measurement does not differentiate between them. On the other hand, the current flow from the copper contact to the MWCNT differs enough to

U

make a noticeable change in the conduction mechanism and the respective impedance.

N

Increasing the MWCNT loading in the composite to 5% for improved conductivity further

A

emphasized the contact material effect. While the conductivity of the sensor layer material increased, that of the contact interface remained the same. This is illustrated in Figure 9,

M

showing the Bode plot of sensors with different contact materials and 5 % MWCNT loading. These tests were conducted under ambient conditions, ~25% RH. While there are still two

ED

separate charge transfer mechanisms detectable, unlike the samples with 2% loading of MWCNT, here the higher frequency one has lower impedance, pointing towards higher

PT

material conductivity. Again, just one charge transfer mechanism could be detected when only carbon contacts were used. We believe that the additional low frequency charge transfer mechanism in case of copper contacts is related to the large interface capacitance and

CC E

resistance between the copper contacts and the conductive silicone composite. The high frequency charge transfer corresponds to that between individual MWCNTs in the PDMS matrix, which has a very small capacitive part and low resistance compared to the copper -

A

silicone interface. It can be seen in the graph that using one carbon and one copper contact in the same sample yields a roughly median signal between carbon-carbon and copper-copper versions. Table 1 lists values for the individual components of the proposed equivalent circuit (inset of Figure 9) in the case of a sample with carbon and copper contact interfaces. The values were

obtained by fitting the data using IviumSoft built-in analysis tools. The relatively large error in the material (Bulk) part is due to noise in the measurement data in the high frequency part. Previous results show that carbon contacts have multiple advantages compared to copper contacts. Most notable of them is the improved integration with the sensor material offering a better connection both electrically and mechanically. Carbon contacts, however, have one major drawback: obtaining a reliable connection from the carbon mat to a wire or PCB pad is

IP T

very difficult compared to copper contacts that can be easily soldered to any metallic parts. To increase the contact reliability of the carbon mesh contact material, the contact ends were

SC R

electroplated with copper to allow soldering wires or PCB pads. It makes the connection

between the wire or pad and contact stable as well as simplifies the fabrication process of the sensors.

U

For the electroplating, a lab power supply was used at 2 V without current limiting. For the anode, a pure copper slab was used, the cathode was the carbon mesh contacts to be coated,

N

and a 1 M solution of CuSO4 was used as the electrolyte.

A

From the results above, it can be concluded that ambient humidity has an effect on the

M

resistivity of the material of the individual sensor layers. However, as the resistance difference is the signal in a Wheatstone bridge setup, the effects of humidity, temperature,

ED

etc. are cancelled out by design. Therefore, there is no need to include a correction mechanism for the bridge output that is used to measure the bending of the sensor.

PT

On the other hand, the capacitance signal that is used to measure the elongation of the sensor does not have a self-correcting mechanism. Thus, the humidity effect on the sensors

CC E

capacitance must be studied.

The amount of change in the capacitance and the absolute value of sensor capacitance were in

A

good agreement with ballpark calculations. Applying the parallel plate capacitor formula: (1) C=εrε0S/d

Where C is the electric capacitance, εr is the relative permittivity, ε0 is the electric constant, S is the area and d is the thickness of the isolating layer, and using relative permittivity 2.7 of common PDMS, we get a resting state capacitance of 48 pF. We are stretching only the middle 1 cm of the sensor which would have a capacitance of 23.9 pF. If that part of the sensor would be stretched by 20%, both the width and thickness will be reduced by 10%

resulting in a capacitance of 28.7 pF. The difference of around 5 pF agrees well with the results shown above in Figure 10, with 20% elongation corresponding to 5 pF at 10% humidity. While our system is far from an ideal parallel plate capacitor, the equation compares well with measurement data from both thicker and thinner sensors that we have tested. We believe that the change in capacitance that is induced by the environment humidity is due

IP T

to the water vapor diffusing into PDMS, water has higher dielectric constant (≈ 80) than PDMS itself (≈ 3), therefore, the capacitance increases with increasing water content in the

SC R

isolating layers of our sensor.

Figure 10 illustrates the change in the capacitance signal that is induced by the changing ambient humidity at different strain values. The change that is induced by moisture appears virtually independent of the level of elongation up to 35%, therefore, the effect could be

U

easily calibrated out with an added humidity sensor. For the range of 0-35% elongation, the

N

effect of moisture gave an average slope of 0.0082 pF/% of humidity with a standard

A

deviation of 0.00099 pF/% (this means the capacitance changes by 8.2 fF with every percent of relative humidity) In the extended range from 0 to 50% elongation, the average humidity

M

effect was 0.0099 pF/% with a standard deviation of 0.0031 pF/%. This can be further described as an equation:

ED

(2) Chum1 = Chum0 ∙ (1 + 0.0001 ∙ ΔRH) Where Chum1 is the capacitance at given humidity, Chum0 is a known capacitance at a known

PT

humidity level, and ΔRH is the difference between relative humidity of given level hum1 and known level hum0. The dataset correlates well with the proposed equation (2), resulting in a

CC E

squared correlation coefficient of 0.9990. While this difference arising from the humidity effect appears small, for example comparing to polyimide based sensors where the slope of a similar equation was 0.0049 and the relative

A

change in the capacitance was even larger as the base capacitance was smaller[16]. From the results, we can see that between 11% to 90% relative humidity, the capacity change induced by humidity change corresponds to less than 2.5%, or up to 3% of elongation under extreme strain. Therefore, the effect of moisture is indeed small, but it exists, and it can be a problem if highest accuracy is required. A simple way to solve the problem would be to add to the net of sensors at least one reference sensor which is made from the same materials but not

affected by mechanical stimuli, providing a correction term for the signals of the actual mechanical sensors. Using the same materials for the calibration sensor has the added benefit that if there are other factors affecting the sensors, those effects would be accounted for inherently as well.

4

Conclusions

IP T

In this study, the emphasis was on evaluating the effect of ambient humidity on the multimodal deformation sensors, focusing on the contact materials. While the resistance of sensor layers was found to be noticeably sensitive to humidity, this is offset by using a

SC R

Wheatstone bridge setup that eliminates environment effects that act on all the sensor layers

simultaneously. The role of the contact material was surprisingly strong, especially at higher MWCNT concentrations, where the contact resistance makes up a considerable part of the

U

total resistance. A 10-fold difference between samples using copper electrodes vs carbon

several advantages compared to the copper foil:

porous unlike metal contacts)

A

better integration with the soft silicon material (less brittle than carbon paper and

M



N

electrodes was observed. It was established that carbon mesh is the preferred material due to

no delamination



higher conductivity between contact and sensor material



simpler conduction mechanism



compliance with the structure (less likely to pierce sensor layers than copper mesh)

CC E

PT

ED



There is, however, one major disadvantage: 

almost impossible to directly make soldered connections

A

The main disadvantage – carbon not being suitable for soldering – can be alleviated by electroplating the ends of the carbon contacts with copper, yielding a metal-coated end bit for soldering and a compliant carbon contact inside the sensor. The effect of humidity on the capacitance signal was studied more thoroughly as the sensor design had no inherent way to compensate it. The effect of changing humidity remained systematic at different strain levels, increasing the capacitance by a constant factor. This

offset caused by relative humidity rising from 11% to 90% corresponds to approximately 2.5% elongation. Consequently, the sensors can be used with no humidity compensation when the moisture stays stable or somewhat lower resolution/accuracy is acceptable. In case of higher precision requirements, the influence of humidity has to be considered as a factor. For example, a mechanically isolated capacitive humidity sensor made from the same materials could be used as a reference for eliminating any undesired effects.

IP T

As the materials used here are common in many soft sensor designs, the results and the proposed approaches can be applied for many similar sensor systems such as pressure and

SC R

elongation sensors using PDMS in its sensing element construction.

Acknowledgements

U

This work was supported by Institutional Research Funding project IUT20-24 from the

A

CC E

PT

ED

M

A

N

Estonian Research Council.

5

References

[1]

S. El-Molla, A. Albrecht, E. Cagatay, P. Mittendorfer, G. Cheng, P. Lugli, et al., Integration of a Thin Film PDMS-Based Capacitive Sensor for Tactile Sensing in an Electronic Skin, J. Sensors. 2016 (2016) 1–7. doi:10.1155/2016/1736169.

[2]

L. Pan, A. Chortos, G. Yu, Y. Wang, S. Isaacson, R. Allen, et al., An ultra-sensitive resistive pressure sensor based on hollow-sphere microstructure induced elasticity in

[3]

IP T

conducting polymer film, Nat. Commun. 5 (2014). doi:10.1038/ncomms4002.

S. Jung, J.H. Kim, J. Kim, S. Choi, J. Lee, I. Park, et al., Reverse-Micelle-Induced

SC R

Porous Pressure-Sensitive Rubber for Wearable Human-Machine Interfaces, Adv. Mater. 26 (2014) 4825–4830. doi:10.1002/adma.201401364. [4]

F. Lorussi, N. Carbonaro, D. De Rossi, A. Tognetti, Strain- and Angular-Sensing

U

Fabrics for Human Motion Analysis in Daily Life, in: Smart Text., Springer

J. Kim, M. Lee, H.J. Shim, R. Ghaffari, H.R. Cho, D. Son, et al., Stretchable silicon

A

[5]

N

International Publishing, 2017: pp. 49–70. doi:10.1007/978-3-319-50124-6_4.

doi:10.1038/ncomms6747.

S.G. Yoon, H.-J. Koo, S.T. Chang, Highly Stretchable and Transparent Microfluidic

ED

[6]

M

nanoribbon electronics for skin prosthesis, Nat. Commun. 5 (2014) 5747.

Strain Sensors for Monitoring Human Body Motions, ACS Appl. Mater. Interfaces. 7

[7]

PT

(2015) 27562–27570. doi:10.1021/acsami.5b08404. A. Tognetti, F. Lorussi, N. Carbonaro, D. de Rossi, Wearable Goniometer and

CC E

Accelerometer Sensory Fusion for Knee Joint Angle Measurement in Daily Life, Sensors. 15 (2015) 28435–28455. doi:10.3390/s151128435.

[8]

Y.R. Jeong, H. Park, S.W. Jin, S.Y. Hong, S.-S. Lee, J.S. Ha, Highly Stretchable and

A

Sensitive Strain Sensors Using Fragmentized Graphene Foam, Adv. Funct. Mater.

[9]

(2015) n/a-n/a. doi:10.1002/adfm.201501000. C. Lee, L. Jug, E. Meng, High strain biocompatible polydimethylsiloxane-based conductive graphene and multiwalled carbon nanotube nanocomposite strain sensors, Appl. Phys. Lett. 102 (2013) 183511. doi:10.1063/1.4804580.

[10] D. Zhou, H. Wang, Design and evaluation of a skin-like sensor with high stretchability

for contact pressure measurement, Sensors Actuators A Phys. 204 (2013) 114–121. doi:10.1016/j.sna.2013.10.002. [11] C.X. Liu, J.W. Choi, Strain-dependent resistance of PDMS and carbon nanotubes composite microstructures, IEEE Trans. Nanotechnol. 9 (2010) 590–595. doi:10.1109/TNANO.2010.2060350. [12] K. Chu, S.-C. Lee, S. Lee, D. Kim, C. Moon, S.-H. Park, Smart conducting polymer

IP T

composites having zero temperature coefficient of resistance, Nanoscale. 7 (2015) 471–478. doi:10.1039/C4NR04489D.

SC R

[13] K. Parikh, K. Cattanach, R. Rao, D.-S. Suh, A. Wu, S.K. Manohar, Flexible vapour

sensors using single walled carbon nanotubes, Sensors Actuators B Chem. 113 (2006) 55–63. doi:10.1016/j.snb.2005.02.021.

U

[14] O.K. Varghese, P.D. Kichambre, D. Gong, K.G. Ong, E.C. Dickey, C.A. Grimes, Gas

N

sensing characteristics of multi-wall carbon nanotubes, Sensors Actuators B Chem. 81

A

(2001) 32–41. doi:10.1016/S0925-4005(01)00923-6.

M

[15] C.-Y. Lee, G.-W. Wu, W.-J. Hsieh, Fabrication of micro sensors on a flexible substrate, Sensors Actuators A Phys. 147 (2008) 173–176.

ED

doi:10.1016/j.sna.2008.05.004.

[16] A. Rivadeneyra, J. Fernández-Salmerón, M. Agudo-Acemel, J.A. López-Villanueva,

PT

A.J. Palma, L.F. Capitan-Vallvey, A printed capacitive–resistive double sensor for toluene and moisture sensing, Sensors Actuators B Chem. 210 (2015) 542–549.

CC E

doi:10.1016/j.snb.2015.01.036. [17] P. Jaakson, A. Aabloo, T. Tamm, Encapsulation of ionic electroactive polymers: reducing the interaction with environment, in: Y. Bar-Cohen, F. Vidal (Eds.),

A

Electroact. Polym. Actuators Devices 2016, International Society for Optics and Photonics, 2016: p. 979825. doi:10.1117/12.2219053.

[18] E. Lee, H. Zhang, J.K. Jackson, C.J. Lim, M. Chiao, Janus films with stretchable and waterproof properties for wound care and drug delivery applications, RSC Adv. 6 (2016) 79900–79909. doi:10.1039/C6RA16232K. [19] Y. Zhang, M. Ishida, Y. Kazoe, Y. Sato, N. Miki, Water-vapor permeability control of

PDMS by the dispersion of collagen powder, IEEJ Trans. Electr. Electron. Eng. 4 (2009) 442–449. doi:10.1002/tee.20429. [20] S.M. Allen, M. Fujii, V. Stannett, H.B. Hopfenberg, J.L. Williams, The barrier properties of polyacrylonitrile, J. Memb. Sci. 2 (1977) 153–163. doi:10.1016/S03767388(00)83241-X.

deformation sensors, Sensors Actuators A Phys. 252 (2016) 42–47. doi:10.1016/j.sna.2016.11.015.

IP T

[21] K. Leemets, U. Mäeorg, T. Tamm, Development of soft and compliant multimodal

SC R

[22] K. Hoffmann, Applying the wheatstone bridge circuit., HBM, Darmstadt, 1974.

http://paginas.fisica.uson.mx/horacio.munguia/aula_virtual/Cursos/Instrumentacion II/Documentos/Teoria Ctos Puentes.PDF (accessed July 27, 2017).

U

[23] N. Lu, C. Lu, S. Yang, J. Rogers, Highly Sensitive Skin-Mountable Strain Gauges

N

Based Entirely on Elastomers, Adv. Funct. Mater. 22 (2012) 4044–4050.

A

doi:10.1002/adfm.201200498.

M

[24] B.W. Veal, P.M. Baldo, A.P. Paulikas, J.A. Eastman, Understanding Artifacts in Impedance Spectroscopy, J. Electrochem. Soc. 162 (2014) H47–H57.

A

CC E

PT

ED

doi:10.1149/2.0791501jes.

Vitae Kaur Leemets received the Bachelor and Masters degree in Material science from the University of Tartu in 2012 and 2014, respectively. He is now pursuing the PhD degree in the Institute of Technology, University of Tartu in the field of soft wearable technology. Uno Mäeorg graduated in 1974 from the University of Tartu (Estonia) in Chemistry. He was

IP T

awarded his PhD in 1985, having worked under the guidance of prof. Ülo Lille, and has continued working at his alma mater where he was made Associate Professor of Organic Chemistry in 1992. His research interests are mainly in the area of organic synthesis,

SC R

materials science and methodology development. He is interested in the development of

analytical methods in organic synthesis as well as the applications of synthetized compounds in material science.

U

A. Aabloo was born in Estonia. He received his Ph.D. in solid state physics in 1994 from University of Tartu. In 1995–1996 he was postdoc in Ångström Laboratory, Uppsala

N

University. Since 2007 he is a professor of polymer materials technology in the Institute of

A

Technology, University of Tartu. His current research interests include atomistic and

M

multiscale modeling of polymer electrolytes and their inter faces, radiation damage studies, electromechanically active polymer transducers and their applications in robotics and micro

ED

devices

Tarmo Tamm is a senior research fellow of material sciences at the Institute of Technology,

PT

University of Tartu, Estonia. He received his PhD in chemistry in 2003, from the University of Tartu. Among his main research interests are both the experimental and theoretical studies

A

CC E

of electroactive materials, functional polymers and composites.

Figure 1 Schematic illustration of the stacked layers connected to form the working sensor including a Wheatstone bridge for measuring bending of the sensor and a capacitive sensor part for registration of the elongation. Figure 2 Cross-section of a 13-layer sensor, the approximate cut line is represented by a green line on the image above. On the SEM image the individual layers are marked together with their thickness. In the conductive layers, the contact material fibers are visible as the cut

IP T

was made in the contact region. Figure 3 Setup for high humidity measurements consisting of: (1) sample under study, ()

SC R

Petri dish to avoid direct contact with water, (3) wires leading to the potentiostat, (4) glass

lid, (5) vacuum grease used to seal the lid to the glass aquarium (6), (7) water in the bottom of the aquarium.

U

Figure 4 Measurement setup for controlled humidity and elongation test. Setup consists of the sample (1) that is clamped in the vice like device (2) used to elongate the sample. A

N

picture of the device is on Figure 5. The sample is connected to measuring equipment with

A

wires running through the Styrofoam lid (5) together with the Alan key (4) used to actuate the

M

vice-like device, and the output of the humidity generator (6) all of this was enclosed in a glass aquarium (7).

ED

Figure 5 Vice-like device used to stretch the sensors: (1) Alan key used to crank the vice mechanism, (2) body of the device, (3) sample under strain, (4) moving jaw of the vice

PT

mechanism, (5) clamps to fix the sensor to the device. Figure 6 Bode plots for a sample with carbon contacts (A) and copper contacts (B) under

CC E

high relative humidity atmosphere. Plots shown are from the start of the experiment, after 20 h and after 9 days.

Figure 7 Results of Impedance measurement for a sample of 2% MWCNT loading with

A

carbon mesh contacts transferred from dry to high relative humidity atmosphere. A) the Bode plot with 4 frequencies marked with vertical lines, the real part of the impedance for those 4 frequencies is shown in time space in (B) (dry atmosphere reading was omitted from the graph to show the small changes better); (C) the Nyquist plot; and (D) the ambient temperature during the measurement.

Figure 8 Results of Impedance measurement for a sample of 2% MWCNT loading with copper contacts transferred from dry to high relative humidity atmosphere. (A) the Bode plot with some of the frequencies marked with vertical lines, those frequencies are shown in time scale in (B) (dry atmosphere reading was omitted from the graph); (C) the Nyquist plot; and (D) the ambient temperature during the measurement. Figure 9 Bode plot comparison of impedances of sensors with different contact materials.

IP T

The inset shows the proposed equivalent circuit where “interface” is the interface between the contact and the conductive silicone, and “bulk” is the conductive silicone itself. The sample used for the Cu-C measurements had one of each electrode as the opposite terminals of the

SC R

conductive PDMS/MWCNT composite with 5% loading of filler by mass. It can be seen that the copper electrodes have two charge transfer modes and the carbon ones have only one,

while the combinations of the two different contacts is essentially the intermediate of the two

U

other curves.

A

CC E

PT

ED

M

A

N

Figure 10 Capacitance signal as a function of humidity at different elongation values.

A ED

PT

CC E

IP T

SC R

U

N

A

M

A ED

PT

CC E

IP T

SC R

U

N

A

M

A ED

PT

CC E

IP T

SC R

U

N

A

M

A ED

PT

CC E

IP T

SC R

U

N

A

M

Table 1 Fitted results of impedance measurement between copper and carbon contacts

Value 5.093E+03 3.991E+02 8.480E-09 3.696E-10

Error % 0.68 5.58 1.59 17.76

Ω Ω F F

A

CC E

PT

ED

M

A

N

U

SC R

IP T

Parameter RInterface RBulk CInterface CBulk