Large-scale printed strain sensors based on carbon ink incorporated into an intermittent conductive silver pattern

The preparation of both types of sensors, the short and long, begins with the printing of the intermittent conductive patterns. Silver ink (TAIYO INK ELEPASTE NP1) was screen printed onto a 20 μm thick polyethylene terephthalate (PET) substrate, with dimensions of 30 cm × 30 cm. A stainless steel mesh (Asada Mesh HS-D 650/14) was used for high-resolution printing. The short sensors had 20.3 mm long and 0.3 mm wide construction, divided into five segments, each 3.7 mm long. The gap between the two segments was 0.3 mm [Fig. 1(a)]. After printing, such patterns were cured in a conventional convection oven at 100 °C for 30 min. In the next step, the free spaces of the intermittent structure were filled with a functional material that provided strain sensitivity.

Zoom In Zoom Out Reset image size

In this study, a short strain sensor made of CNT ink (VC101, Chasm Advanced Materials) was compared with a sensor with the same construction but made of a carbon black ink (TOYOBO DY-200L-2). The functional inks were used in a very small quantity through the deposition of single droplets [Fig. 1(b)], using a dispenser (Musashi ML-5000XII). The dispenser had adjustable pressure and discharge times. A pressure of 0.35 MPa and discharge time of 70 ms and 50 ms were set for the carbon black ink and CNT, respectively. Such settings were selected to form droplets that were sufficiently large to fill the 0.3 mm × 0.3 mm free spaces in the intermittent pattern and to provide electrical contact between the consecutive silver segments [Fig. 1(c)]. After the deposition of inks, the entire sensor structures were dried in a convection oven at 130° C for 30 min. The electrical resistance of the dried single CNT droplets [Fig. 1(d)] was 219.5 ± 13.4 Ω, while the entire hybrid (silver–CNT) structure had a resistance of 1303 ± 41 Ω. For comparison, the electrical resistance of dried single droplets made of carbon black was 32.6 ± 2.3 Ω, while the entire hybrid (silver–carbon black) structure had a resistance of 199.3 ± 8.4 Ω.

The large-scale sensors had a 225 mm long and 1 mm wide sensing grid. The intermittent structure was fabricated using the same silver ink. Two types of long sensors that differed by the number of sensing elements distributed along the same sensor length were fabricated. Their construction was divided into 41 segments and 81 segments, respectively, with an end loop in the middle. Thus, the total length of the sensing grid was approximately 450 mm long. The first sensor had 40 sensing elements (N40), while the second had 80 sensing elements (N80). The free spaces in the intermittent structure were 1 mm wide. In contrast to the short sensors, the sensing elements were screen printed onto a large intermittent pattern. Nonetheless, any other instrument, such as an inkjet printer, dispenser, or microplotter, can be implemented. The electrical resistance of the screen-printed sensing elements was 389.8 ± 18.9 Ω. The entire resistance of the large-scale sensors with the hybrid structure was 15.13 ± 0.79 kΩ and 31.78 ± 1.74 kΩ for N40 and N80, respectively.

The PET substrates, with the fabricated sensors, were cut into smaller pieces, corresponding to the sensors' dimensions, that is, 6 mm × 30 mm and 10 mm × 235 mm, for the short and long sensors, respectively. Note that in this study, sensors were not covered at the top. However, the implementation of a protective cover layer is necessary for further practical applications. For this purpose, epoxy-based adhesive sheets ^26,27) or a cold lamination process ⁵⁾ can be used.

The fabricated sensors (short and long) were calibrated on the basis of reference strain measurements that were obtained using commercially available strain gauges. The investigated sensors and the reference conventional sensors were bonded to a metal plate using an epoxy-based adhesive (Araldite 2012). During the analysis, the sensors were connected to a quarter-Wheatstone-bridge circuit with an excitation voltage of 2.4 V delivered from a power supply (Keysight E36311A). Using a tensile test machine (Fig. 2), the plate with the sensors was subjected to cyclic uniaxial bending deformations, corresponding to tensile or compressive strains, depending on the bending direction. The output signals from both types of sensors (voltage) were simultaneously recorded using a 24 bit data acquisition system (National Instruments NI-9238). The measurements were registered and analyzed using a specially prepared computer program (National Instruments LabView). Based on the recorded data, the relative change in resistance (ΔR/R₀) was calculated and compared with the corresponding strain (ε) measured using conventional strain gauges, with a known sensitivity [gauge factor (GF)] of 2.12. The GFs of the analyzed sensors were determined using the following equation:

$$\begin{eqnarray}&&{\rm{GF}}=\displaystyle \frac{{\rm{\Delta }}R/{R}_{0}}{\varepsilon },\end{eqnarray} \tag{ 1 }$$

Zoom In Zoom Out Reset image size

In this study, the conventional strain gauges Kyowa KFGS-20-120-C1-11 and Kyowa KFGS-30-120-C1-11 were used to provide reference strain measurements during the calibration of the short and long sensors, respectively. In the case of the short sensors, reference strain gauges were installed on the plate aside from the investigated sensors. The long sensors had a length of over 20 cm. Such long metal foil strain gauges are not commercially available. Thus, in this study, the long sensors were calibrated based on the three conventional gauges installed on the backside of the plate, along the location corresponding to the investigated sensor installed on the top of the plate. The three conventional gauges were installed in the positions corresponding to the beginning, middle, and end of the developed sensor. During the analysis, the output signal from these three strain gauges was measured, and the average strain value was calculated. The calculated average strain was used to calibrate the developed large-scale sensor.

This study started with the fabrication of CNT-based short sensors (2 cm), according to the procedure described in Sect. 2.1. Although in our previous work, we demonstrated the good performance of carbon black-based sensors with the hybrid (carbon black-silver) construction, ³¹⁾ the CNT-based sensors were of interest because of their good mechanical properties, as reported by other researchers. ¹⁵⁾ Thus, prior to the fabrication of large-scale sensors, the preliminary analysis with the short sensors was intended to select a functional material with regard to the strain sensitivity and linearity of the measured output signal.

To provide more information concerning the basic properties of the CNT-based sensor, the sensor with the hybrid construction was compared with a sensor that was entirely made of CNT ink. Thus, an additional sensor with no intermittent silver structure was fabricated using screen printing (Fig. 3). It had the same length and width as the intermittent pattern. The electrical resistance of such a sensor in the form of a continuous CNT line was 124.8 ± 1.6 kΩ.

Zoom In Zoom Out Reset image size

The strain sensitivity of CNT-based sensors was analyzed by measuring the variations in the electrical resistance of the entire sensor structure when the steel plate was subjected to mechanical deformations (Fig. 2). During the analysis, 10 bending cycles up to approximately 800 microstrain (μ ε) were implemented. When the relative change in resistance was measured by the developed sensors, the reference strain measurement was simultaneously recorded by the conventional strain gauge, attached aside to the analyzed sensors. The results show that within the investigated strain range, both CNT sensors (hybrid and one entirely made of CNT) exhibit a linear output signal, no hysteresis, and almost the same strain sensitivity (Fig. 4). The observed no significant difference in the strain sensitivities shows that the sensing properties depend mainly on the functional ink used to fabricate the resistive sensing elements. This is a key feature of the sensor with an intermittent structure. Nonetheless, the calculated sensitivity (GF ≈ 0.82) was relatively low, especially when compared with the conventional strain gauges whose GF is typically approximately 2 (2.12 in this work). Next, the CNT-based sensor with the intermittent structure was compared with the previously reported sensor with the same construction, but with carbon black ink used as the functional material. ³¹⁾ Figure 5 shows that the GF for the carbon black-based sensor was significantly higher (GF = 7.71) than that of the CNT-based sensor. Thus, based on the collected data, carbon black ink was selected to fabricate the resistive elements in large-scale strain sensors with an intermittent structure.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The concept for the construction of the large-scale sensors shown in Fig. 6, is the same as that for the short sensors discussed in Sect. 3.1. Its construction is based on the hybrid structure of a conductive intermittent pattern with embedded resistive elements of a functional ink (carbon black in this study) that provides strain sensitivity. Thus, the resistive functional ink defines the sensor properties while the intermittent conductive patterns specify a path along which strain is measured at the points corresponding to the positions specified by the resistive sensing elements. Such a construction makes it possible to design a sensor with a desirable electrical resistance, which is especially promising for designing large-scale sensors.

Zoom In Zoom Out Reset image size

Two types of long sensors that differed by the number of sensing elements distributed along the same sensor length were investigated (N40 and N80). For the analysis, both types of long sensors were attached to the metal plate, as shown in Fig. 2. The measurements were performed using the same data acquisition system as for the short (2 cm) sensors. First, the sensors were subjected to dynamic deformations, and the measured output signal was compared with those measured by the three reference conventional strain gauges. Figure 7 shows the data recorded for the sensor with 40 sensing elements. Both sensors (N40 and N80) exhibit similar characteristics. It can be observed that the measured output signal was very consistent with the applied external bending deformation measured by the conventional sensors. Note that the analysis was performed under bending deformations in both directions (positive and negative, alternately). It was noticed that in addition to the good response time, the measured output signal was higher than that of the conventional sensors, which indicates good strain sensitivity of the developed sensor. To determine the strain sensitivity, the collected data for both sensors (N40 and N80) were compared with the strain recorded by the conventional sensors (Fig. 8). Based on Eq. (1), the GFs of 13.26 and 13.69 for the sensors N40 and N80 were calculated, respectively. The calculated values were almost the same, which indicates that the effect of a larger number of sensing elements (and thus more contact resistance points between carbon and silver) has a very low impact on the strain sensitivity. The observed effect of more sensing elements on the increase in GF is noticeable but not significant. This finding is consistent with those of our previous studies. ³¹⁾ This demonstrates that the changes in the electrical resistance of the carbon layer are the dominant factors that define the strain sensitivity of the demonstrated sensors.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Note that the achieved strain sensitivity of the large sensors was higher than that of the short sensor (2 cm) made of the same materials. However, the strain sensitivity, besides the materials, depends on the geometry and thickness of the sensing elements. In the case of the short sensors, the sensing elements were printed using a dispenser, in the case of the long sensors, the resistive elements were screen printed. Using the dispenser, the thickness of the resistive elements was approximately 20 μm, while the screen-printed layer had a thickness of approximately 4 μm. Nonetheless, either the screen-printed sensing elements and those printed using the dispenser provide relatively high strain sensitivity, higher than those of the conventional sensors, which are sufficient for most common applications. Moreover, the sensors exhibit good linearity and no hysteresis within the analyzed strain range.

In addition to the dynamic strain analysis, the sensors were tested under static strains, as demonstrated in Fig. 9. Under static bending deformation, at the high state, the output signal remains constant and it goes back to the low state without a visible delay. In this analysis, the developed sensor was calibrated [Fig. 8(a)], that is, the measured output signal, instead of the change in resistance, shows the measured strains. For both types of sensors (N40 and N80), the collected results show good response time to the applied external deformation and strain analysis, which is consistent with those measured by the reference strain sensor. The above results demonstrate the potential suitability of the developed large-scale sensor for static and dynamic strain analyses.

Zoom In Zoom Out Reset image size