Highly selective flexible tactile strain and temperature sensors against substrate bending for an artificial skin

Abstract

Flexible devices can conformally cover any surfaces and interact with different stimuli such as human touch. Although a flexible tactile sensor has been reported as an artificial skin application, distinguishing between a tactile force and strain due to substrate bending remains challenging. Here we report a highly selective tactile force sensor against bending on the basis of strain engineering by fabricating a cantilever structure. The proposed device achieves a 4–23 times improvement in selectivity compared to conventional pressure sensitive rubber. As a proof-of-concept for e-skin, an array composed of highly selective tactile force sensors and temperature sensors is successfully demonstrated to imitate human skin.

---

1. Introduction

Flexible and wearable devices have received much attention due to their potential in electronic applications such as robotics/prostheses,^1–7 health monitoring,^8–12 and human interactive devices.^13–15 An interesting flexible device is an artificial electronic skin (e-skin) that can detect tactile pressure and temperature distribution similar to human skin. In fact, a lot of approaches have been reported to realize e-skin (i.e., tactile pressure and temperature sensors) and other tactile pressure sensor-based devices.^{1–7,16–21} Typically tactile pressure sensors are resistive^{1,5–7,16,18–21} or capacitive^2,17 sensitive sensors that detect a structural deformation upon applying pressure or force.

One of the challenges for e-skin applications is that conventional flexible tactile pressure sensors for e-skin applications also detect substrate bending because bending also causes a strain distribution that corresponds to the structural deformation in a flexible substrate where the tactile pressure sensor is integrated, which inhibits a precise tactile pressure reading.

Another key technique that needs to be developed is a practical fabrication method. Most previous reports fabricated flexible devices using semiconductor infrastructure for metallization and/or active material patterning.^{1–5,7,8,10,13–15,17,19,21} (There are a few exceptions.^{6,9,18,20,22,23}) Although the device cost depends on the application, typical applications of flexible devices must be economically fabricated on macroscale substrates. Because the semiconductor infrastructure limits devices to tens of centimeters in size, an alternate fabrication method is necessary. One potential solution is a printing method using a screen printer,^6,9,24 gravure printer,^23,25 etc.

To achieve economically fabricated strain and temperature sensors using a screen printer by developing sensor inks, we previously reported a macroscale e-skin, which can monitor tactile force, friction force, and temperature.⁶ However, this e-skin also detects substrate bending, which prevents precisely measuring the tactile pressure.

In this study, we propose a strain-engineered highly selective tactile pressure sensor using a printed strain sensor on a cantilever structure of a flexible substrate. Strain due to bending of the substrate is not applied on a cantilever, which should result in a high selectivity between the strain sensor and substrate bending. Since this report targets e-skin applications, an array on a flexible substrate, which is comprised of strain sensors on a cantilever structure and temperature sensors, is demonstrated to mimic human skin using fully printed methods.

2. Experimental

2.1 Fabrication process

Fig. 1a depicts the fabrication process of a strain sensor and a temperature sensor with a cantilever structure. Sensors in this unique structure are printed on both surfaces of a polyethylene terephthalate (PET) film, allowing the sensors to be highly integrated. Briefly, silver (Ag) electrodes were screen-printed on the top surface of a PET film as interconnections for the strain sensor using a commercially available Ag ink (Asahi Chem. Res. Lab., Japan). Then Ag electrodes for the temperature sensor were printed on the bottom PET surface so that they are aligned with the Ag electrodes on the top surface. For electrodes on both surfaces, the Ag film after printing was cured at 130 °C for 30 min. Next, an array of strain sensor based on Ag nanoparticle (AgNP) and carbon nanotube (CNT) was screen-printed. It is noteworthy that this strain sensor reads the resistance change as a function of the applied force. As reported previously, the resistance change is caused by the change in distance between AgNPs in the mixed film upon applying a tensile and a compressive stress.^6,24,26 Next, temperature sensors based on CNT ink and poly(3,4-ethylenedioxythiophene)-poly(styrenesulfonate) (PEDOT:PSS) solution were printed on the bottom PET surface. This temperature sensor also reads the resistance change as a function of the temperature change. The sensing mechanism is on the basis of electron hopping at the interface between CNT and PEDOT:PSS.^6,9

To create a cantilever structure and via holes to connect the top and bottom surfaces electrically, a laser cutter cut and patterned the PET substrate. Ag ink was then printed and cured over the via holes for the electrical connections. Finally, the gap between the cantilever and the surface where the device is placed was fabricated by laminating the top surface with a one-sided adhesive polyester film (more than 100 μm thick) (i.e., strain sensor side). This polyester allows a cantilever bending even the sample is placed on a solid surface, so that the thicker film makes higher dynamic range of a tactile pressure sensing. Fig. 1b–d shows the schematics and a picture of the fabricated e-skin device. The device including contact pads measures 8 cm × 8 cm and contains 4 × 4 strain sensor and 3 × 4 temperature sensor arrays. Enlarged pictures and scanning electron microscopy images of the resistive strain and temperature sensors show that the alignment accuracy is relatively good (within 150 μm) for the manual screen printer (Fig. 1e and g) and the surface morphology (Fig. 1f and h), respectively.

Fig. 1 (a) Schematic of the e-skin fabrication process on a PET substrate using a printing method. Schematics of (b) the final structure and (c) a cross-section of one pixel. (d) Photo of the final fabricated e-skin with a 4 × 4 tactile force sensor array and a 3 × 4 temperature sensor array. Enlarged photos and scanning electron microscopy images of (e), (f) a strain sensor and (g), (h) a temperature sensor.

2.2 Strain sensor

For the strain sensor, AgNP ink (Paru, Korea) and CNT ink (SWeNT, USA) were mixed with a 3 : 5 weight ratio. The inks were manually mixed together by stirring the inks. Using a commercially available screen mask (Sonocom, Japan), the mixed ink was screen-printed. After printing, the ink was cured at 70 °C for at least 60 min in air ambient. By changing the composition ratio of Ag ink and CNT ink, the sensitivity can be tuned as reported previously.^6,24 We here chose the ratio of a 3 : 5 weight ratio to realize relative high sensitivity.

2.3 Temperature sensor

For temperature sensor, CNT ink (SWeNT, USA) and PEDOT:PSS solution (Sigma-Aldrich, USA) were mixed with different composition ratios of the inks to optimize the condition for the high sensitivity. The inks were manually mixed and stirred. Using a shadow mask fabricated by cutting a polyester film using a laser cutter tool, the mixed ink was printed on the bottom PET surface by putting the ink over the mask. The curing temperature of the temperature sensor film was 70 °C for at least 60 min in air ambient.

3. Result and discussion

First, the sensitivity was measured as a function of the applied force by bending the cantilever structure using a force sensor (KYOWA, Japan). Fig. 2a shows the normalized resistance change, ΔR (=R − R₀)/R₀, at different applied forces, where R and R₀ are the resistance with and without the applied force, respectively (R₀ = 48 Ω). The normalized resistance change is proportional to the square of the applied tactile force. Because the experimental results and fitting curve agree well, the resistance change can be readily converted into the applied force. This cantilever-type tactile force sensor requires a slightly stronger touch than a gentle human touch (∼50 mN) for some applications such as human interactive communications. If a higher sensitivity is required to behave exactly like human skin, the substrate must be a thicker but softer material (i.e., a lower Young modulus material) to create a higher strain in the strain sensor at a lower applied force than the PET substrate used in this report.

Fig. 2 (a) Normalized resistance change (ΔR/R₀) of the CNT–AgNP strain sensor as a function of tactile force. (b) Normalized resistance change (ΔR/R₀) of repeatable cycle measurements of the strain sensor. (c) Normalized resistance change of the PEDOT:PSS–CNT temperature sensor as a function of temperature for various composition ratios of PEDOT:PSS and CNT inks. (d) Sensitivity of the temperature sensor as a function of the ink composition ratio.

Repeatability and uniformity of the strain sensor are another factors to characterize the strain sensor due to the printed mechanical sensor. Fig. 2b shows the normalized resistance change of cycle test of the strain sensor to characterize the repeatability when the force of 20 mN is applied. On the basis of the results, R₀ (i.e. applied force 0 mN) is gradually increased by applying the force continuously while the sensitivity is almost constant. The increase of the base resistance, R₀, is most likely due to mechanical damage to the strain sensor caused by mechanical stress during the measurement. Next, the uniformity of sensitivity of the strain sensor is discussed. The average sensitivity is ∼0.4% at the applied force of 200 mN with the standard deviation of ∼0.4%. This large standard deviation is due to nonuniformity of strain sensor printing because we here use a manual-operated screen printer. This manual printer creates pressure difference of a squeegee easily when the CNT–AgNP ink is printed, resulting in that the thickness of the strain sensor (i.e. sensitivity) is nonuniform. These problems need to address by considering the passivation layer, developing a new material system, and/or using a pressure-controllable screen printer in the future.

Next, the printing condition of the temperature sensor was optimized to achieve maximum sensitivity by changing the composition ratio of the mixture inks of PEDOT:PSS and CNTs. After printing and curing the temperature sensors, the resistance changes were measured by placing them on a hotplate. Fig. 2c shows representative normalized resistance changes (ΔR (=R − R₀)/R₀) of the temperature sensors as a function of hotplate temperature, where R₀ = 30 kΩ and R are the resistance at room temperature (22 °C) and at the measured temperature, respectively. For precise measurements, the temperature of the hotplate was also monitored using an infrared temperature sensor. Since the main electron hopping mechanism at the interface between PEDOT:PSS and CNT was used to detect the temperature difference with a high sensitivity,⁹ an ideal composition ratio should exist to create highly efficient electron hopping. As expected, the slopes, which correspond to the sensitivity, clearly change for the different composition ratios (Fig. 2c). Fig. 2d compiles the sensitivity extracted from the linear fitting of the resistance change at various temperatures. The maximum sensitivity of PEDOT:PSS and the CNT temperature sensor of ∼0.78% per °C at a 3 : 1 mixture weight percent ratio is achieved compared to our previous study (0.61% per °C).

Next a finite element method (FEM) simulation was conducted to observe the strain distribution in a cantilever structure when the substrate was bent to a 32 mm curvature radius. To find the directional dependence, the cantilever structures with vertical (bending direction “A”) and parallel (bending direction “B”) directions along the bending direction were simulated (Fig. 3a). The curvature radius is defined as the radius of the substrate bent as described in Fig. 3b. Fig. 3c and d indicate that the strain in the cantilever structure is ∼0.015% (∼0.02%) for bending direction “A” (“B”), while the area outside of the cantilever structure shows a strain distribution that is more than ten times higher, suggesting that the strain sensor is independent of substrate bending in both directions. To prevent a strain effect due to substrate bending, the strain sensor was patterned on the cantilever structure while the Ag interconnection was not. It should be noted that the resistance of the Ag interconnection does not change under a small strain caused by substrate bending.

Fig. 3 Schematics of (a) a one-pixel device with two bending directions “A” and “B” and (b) cross-section of the experimental setup to measure sensors while bending with a specific curvature radius. FEM simulation results for a substrate bent with a 32 mm curvature radius in (c) direction “A” and (d) direction “B”. (e) Normalized resistance change of the cantilever type strain sensor and a conventional pressure sensitive rubber (PSR) as a function of the bending radius. The results with bending direction “A” and “B” are overlapped in the figure due to small resistance change up to 7 mm curvature radius. (f) Normalized resistance change of an integrated temperature sensor as a function of the bending curvature radius of the substrate.

To confirm experimentally that the strain sensor is independent of bending, the normalized resistance change of the strain sensor on a cantilever structure as a function of the bending curvature radius of substrate was measured. For comparison, conventional pressure sensitive rubber (PSR) (PCR Technical, Japan), which is often used for e-skin applications,^5,7,13 was also measured. For PSR, a pair of Ag electrodes were printed on one side of PSR, and the resistance changes as a function of the bending radius were measured. The normalized resistance change of a strain sensor on a cantilever structure is only ∼0.1% (∼0.6%) for the strain sensor bent toward direction “A” (“B”) up to a bending radius of 7 mm (Fig. 3e). When the selectivity is considered, the ratio of the real resistance change due to tactile pressure to the error resistance change due to bending is important. The selectivity of the resistance change by a tactile force and substrate bending (=S_force/S_bend) is about 4–23 depending on the bending direction, whereas that for PSR is about 0.5 at the same applied force (∼400 mN) and bending radius (∼34 mm). Here S_force and S_bend are the normalized resistance changes when the tactile force and bending of the substrate are applied, respectively. These observations strongly suggest that the selectivity of the strain-engineered tactile pressure sensor using a cantilever structure can be dramatically improved compared to conventional PSR.

Although we used only one pattern design to fabricate a cantilever-type strain sensor, further optimizing the structure can reduce the resistance change upon bending. It is noteworthy that the sensitivity under substrate bending is slightly lower for bending direction “B” due to the slight bending of the cantilever structure (i.e., the spring constant of the cantilever structure increases), which results in a stronger required force to bend the cantilever with the same bending displacement as the flat state or bending direction “A”. Furthermore, the bending tolerance with high selectivity depends on the size of a cantilever structure, so that the design of a cantilever is also important factor. In addition, the temperature sensor also has a high selectivity of ∼13 for a 1 °C temperature change (∼0.78%) and substrate bending with a curvature radius of 33 mm (∼0.06%) (Fig. 2c and 3f). In addition, we have already reported that the strain sensor has low sensitivity of temperature change (∼0.03% per °C).²⁴ On the basis of these, the strain and temperature sensor also have high selectively to distinguish each stimulus.

Finally, as a proof-of-concept, e-skin monolithically integrated with a 4 × 4 tactile force sensor array and a 3 × 4 temperature sensor array was demonstrated. Each pixel was connected to a multi-channel data logger to measure all pixels simultaneously. Due to the range limitation of the resistance (maximum resistance is 100 Ω) of the data logger, the temperature sensors were connected to a Wheatstone bridge circuit to convert the output resistance change into voltage. Fig. 4 depicts a tactile force measurement and temperature mappings when two fingers are placed on the e-skin device. The device can successfully map the two-dimensional tactile force and temperature distribution similar to human skin. However, the pixel resolution of the tactile force and the temperature distribution is large (∼2 cm pitch) compared to human skin. Because this report focuses on demonstrating highly selective tactile force and temperature sensors against substrate bending as a proof-of-concept, a high-density integration was not conducted. If a high resolution is needed, further study is required to fabricate the sensors with high resolution patterning of the printing methods.

Fig. 4 (a) Photo of the measurement setup. (b) Two-dimensional tactile force and temperature mappings when the e-skin is touched manually by two fingers.

4. Conclusions

In conclusion, we present highly selective tactile force and temperature sensors against bending of a flexible substrate for e-skin and flexible device applications. A strain-engineered cantilever structure is used to achieve the high selectivity; compared to a conventional PSR-based tactile force sensor, which has a selectivity <1, the improvement is about 4–23 for the tactile force sensor depending on the bending direction and about 13 for the temperature sensor. This is a significant advance in the field of flexible tactile force sensors. In addition, modifying the design may further improve the selectivity. As a proof-of-concept of the cantilever-type e-skin, the two-dimensional tactile force and temperature distributions similar to human skin are successfully demonstrated. More importantly, the e-skin is fabricated using solely a printing method, which should eventually realize a platform to fabricate macroscale and economical electronics for various applications, including a wearable and flexible device.

Acknowledgements

This work was partially supported by JSPS KAKENHI Grant (#26630164 & 26709026), the Mazda Foundation, Tateishi Science and Technology Foundation, and Japan Prize Foundation.

Notes and references

1. J. Park, Y. Lee, J. Hong, M. Ha, Y.-D. Jung, H. Lim, S. Y. Kim and H. Ko, ACS Nano, 2014, 8, 4689.
2. G. Schwartz, B. C.-K. Tee, J. Mei, A. L. Appleton, D. H. Kim, H. Wang and Z. Bao, Nat. Commun., 2013, 4, 1859.
3. D.-H. Kim, N. Lu, R. Ma, Y.-S. Kim, R.-H. Kim, S. Wang, J. Wu, S. M. Won, H. Tao, A. Islam, K. J. Yu, T.-i. Kim, R. Chowdhury, M. Ying, L. Xu, M. Li, H.-J. Chung, H. Keum, M. McCormick, P. Liu, Y.-W. Zhang, F. G. Omenetto, Y. Huang, T. Coleman and J. A. Rogers, Science, 2011, 333, 838.
4. W. Wu, X. Wen and Z. L. Wang, Science, 2013, 340, 952.
5. K. Takei, T. Takahashi, J. C. Ho, H. Ko, A. G. Gillies, P. W. Leu, R. S. Fearing and A. Javey, Nat. Mater., 2010, 9, 821.
6. S. Harada, K. Kanao, Y. Yamamoto, T. Arie, S. Akita and K. Takei, ACS Nano, 2014, 8, 12851.
7. T. Someya, Y. Kato, T. Sekitani, S. Iba, Y. Noguchi, Y. Murase, H. Kawaguchi and T. Sakurai, Proc. Natl. Acad. Sci. U. S. A., 2005, 102, 12321.
8. D. Son, J. Lee, S. Qiao, R. Ghaffari, J. Kim, J. E. Lee, C. Song, S. J. Kim, D. J. Lee, S. W. Jun, S. Yang, M. Park, J. Shin, K. Do, M. Lee, K. Kang, C. S. Hwang, N. Lu, T. Hyeon and D.-H. Kim, Nat. Nanotechnol., 2014, 9, 397.
9. W. Honda, S. Harada, T. Arie, S. Akita and K. Takei, Adv. Funct. Mater., 2014, 24, 3299.
10. K.-I. Jang, S. Y. Han, S. Xu, K. E. Mathewson, Y. Zhang, J.-W. Jeong, G.-T. Kim, R. C. Webb, J. W. Lee, T. J. Dawidczyk, R. H. Kim, Y. M. Song, W.-H. Yeo, S. Kim, H. Cheng, S. I. Rhee, J. Chung, B. Kim, H. U. Chung, D. Lee, Y. Yang, M. Cho, J. G. Gaspar, R. Carbonari, M. Fabiani, G. Gratton, Y. Huang and J. A. Rogers, Nat. Commun., 2014, 5, 4779.
11. K. Takei, W. Honda, S. Harada, T. Arie and S. Akita, Adv. Healthcare Mater., 2015, 4, 487.
12. X. Wang, Y. Gu, Z. Xiong, Z. Cui and T. Zhang, Adv. Mater., 2014, 26, 1336.
13. C. Wang, D. Hwang, Z. Yu, K. Takei, J. Park, T. Chen, B. Ma and A. Javey, Nat. Mater., 2013, 12, 899.
14. S. Jung, J. H. Kim, J. Kim, S. Choi, J. Lee, I. Park, T. Hyeon and D.-H. Kim, Adv. Mater., 2014, 26, 4825.
15. J.-W. Jeong, W.-H. Yeo, A. Akhtar, J. J. S. Norton, Y.-J. Kwack, S. Li, S.-Y. Jung, Y. Su, W. Lee, J. Xia, H. Cheng, Y. Huang, W.-S. Choi, T. Bretl and J. A. Rogers, Adv. Mater., 2013, 25, 6839.
16. J. T. Muth, D. M. Vogt, R. L. Truby, Y. Megnuc, D. B. Kolesky, R. J. Wood and J. A. Lewis, Adv. Mater., 2014, 26, 6307.
17. H. Vandeparre, D. Watson and S. P. Lacour, Appl. Phys. Lett., 2013, 103, 204103.
18. A. Morteza, A. Pichitpajongkit, S. Lee, S. Ryu and I. Park, ACS Nano, 2014, 8, 5154.
19. Y. Zhou, S.-T. Han, Z.-X. Xu and V. A. L. Roy, Nanoscale, 2013, 5, 1972.
20. T. Yamada, Y. Hayamizu, Y. Yamamoto, Y. Yomogida, A. Izada-Najafabadi, D. N. Futaba and K. Hata, Nat. Nanotechnol., 2011, 6, 296.
21. N. T. Tien, S. Jeon, D.-I. Kim, T. Q. Trung, M. Jang, B.-U. Hwang, K.-E. Byun, J. Bae, E. Lee, J. B.-H. Tok, Z. Bao, N.-E. Lee and J.-J. Park, Adv. Mater., 2014, 26, 796.
22. X. Liu, L. Gu, Q. Zhang, J. Wu, Y. Long and Z. Fan, Nat. Commun., 2014, 5, 4007.
23. P. H. Lau, K. Takei, C. Wang, Y. Ju, J. Kim, Z. Yu, T. Takahashi, G. Cho and A. Javey, Nano Lett., 2013, 13, 3864.
24. S. Harada, W. Honda, T. Arie, S. Akita and K. Takei, ACS Nano, 2014, 8, 3921.
25. M. Jung, J. Kim, J. Noh, N. Lim, C. Lim, G. Lee, J. Kim, H. Kang, K. Jung, A. D. Leonard, J. M. Tour and G. Cho, IEEE Electron Device Lett., 2010, 57, 571.
26. K. Takei, Z. Yu, M. Zheng, H. Ota, T. Takahashi and A. Javey, Proc. Natl. Acad. Sci. U. S. A., 2014, 111, 1703.

---