Supersensitive all-fabric pressure sensors using printed textile electrode arrays for human motion monitoring and human–machine interaction

Abstract

Integrating the advantages of ultrahigh sensitivity, good breathability, low-cost and large-area fabrication processes, and facile integration with other functional devices is full of challenges for wearable pressure sensors. Here, a novel all-fabric piezoresistive pressure sensor is designed with a bottom interdigitated textile electrode screen-printed using silver paste and a top bridge of AgNW-coated cotton fabric. The entire fabrication process is facile, economical and suitable for large-scale integrated production. Benefiting from the highly porous microstructure, large surface roughness and ultra-low resistance of the conductive fabric, our piezoresistive pressure sensors show excellent detection performance, including an extra-high sensitivity of 2.46 × 10⁴ kPa⁻¹ to 5.65 × 10⁵ kPa⁻¹ over a wide pressure regime (0–30 kPa), a giant on/off ratio of ≈10⁶, a fast response time (6 ms), and a low detection limit (0.76 Pa). Thanks to these merits, the devices not only have the ability to detect various tiny signals of the human body, but also can be widely applied in a human–computer interactive system as a real wearable sensor platform, which was demonstrated by playing the piano and computer games.

---

Introduction

Electronic textiles (E-textiles) have attracted a great deal of attention as next-generation wearable electronics due to their facile integration into clothes.^1,2 Various electronic elements such as physical or biological sensors,^3,4 energy harvesting devices,⁵ field-effect transistors⁶ and antennas⁷ have been integrated into E-textiles. In particular, wearable pressure sensors with force-sensitive properties, which can mimic and surpass the function of human skin by transducing tiny external stimuli such as touching, shear and mechanical vibration into electrical signals, have been widely explored for diagnostics, monitoring patients, human motion detection, etc.^8–10 To meet the needs of wearable electronics, pressure sensors are required to be ultra-sensitive, comfortable, breathable, washable, flexible and robust.^11,12 However, most of the traditional wearable sensors are fabricated using airproof and non-renewable elastic substrates, such as polydimethylsiloxane (PDMS),^13,14 polyethylene terephthalate (PET)¹⁵ and polyurethane (PU),¹⁶ which are unfavourable for large-area applications due to skin's breathing. These disadvantages will greatly limit their prospects in the application of E-skin.¹⁷ In contrast, textile substrates are skin-friendly, renewable and offer enormous opportunities to deploy sensors and other devices, which are promising for E-textiles.

So far, various operation types of pressure sensors such as piezoresistive,^9,17 capacitive,¹⁸ piezoelectric,¹⁹ and triboelectric²⁰ have been developed. Particularly, piezoresistive sensors are widely used owing to their facile fabrication process, superior sensitivity, read-out mechanism and potentially high pixel density.²¹ In order to obtain a highly sensitive textile-based piezoresistive pressure sensor, fabric with high electrical conductivity and large surface roughness is essential. Various conductive materials such as carbon nanomaterials (CNTs and rGO),^8,11 metallic nanoparticles/nanowires (AgNPs and AgNWs)^12,24 and organic conductive polymers (PEDOT)²⁵ combined with fibers or fabrics have been reported to construct piezoresistive pressure sensors. AgNWs have superior conductivity as well as excellent stability, which are ideal candidates for wearable pressure sensors. Recently, Wei et al.¹² demonstrated a useful wearable pressure sensor with the structure of two conductive cotton sheets coated with AgNWs. Despite this progress, large-area whole-textile pressure sensor arrays able to map pressure stimuli are seldom reported.¹¹ Besides, the inferior sensitivity of textile pressure sensors developed before makes the computing system hard to identify the loading/unloading signals, which limits the implementation of pressure sensors used in human–computer interaction systems.²²

Here, we demonstrate a novel structure of all-textile-based pressure sensors and large-area sensor arrays. The resistive textile sensor unit is composed of a bottom interdigitated textile electrode screen-printed using silver paste and a top bridge of AgNW-coated cotton fabric. The pressure sensors maintained an ultra-sensitivity of 2.46 × 10⁴ to 5.65 × 10⁵ kPa⁻¹ in a wide pressure range (<30 kPa), which is among the best results for wearable pressure sensors. Moreover, the textile sensor could operate under a low voltage of 0.1 V while achieving a fast response/relaxation time (6/16 ms), a low detection limit (0.76 Pa), and a high stability (>1000 loading/unloading cycles). These remarkable detective performances show that our sensors could detect varied tiny stimulations and facilely integrate with other functional devices. The textile electrodes are screen-printed and the AgNW-coated cotton could be fabricated using the wet chemistry process. Both of the two manufacturing processes are facile, low-cost, and suitable for large-area and high-volume production in the future.

Results and discussion

The fabrication process of the textile-based pressure sensor is illustrated in Fig. 1a. We synthesized high-quality AgNWs by a polyol process according to the method reported recently.²⁶ It is noted that the AgNWs we prepared were about 25 nm in width and 30 μm in length, and have a high aspect ratio >1000 (Fig. S1, ESI†). In order to obtain a uniform conductivity of the AgNW-decorated fabric, AgNWs were deposited into the pre-cleaned cotton by dip coating followed by a drying process. The sheet resistance of the AgNW-decorated cotton reaches 797.9 Ω sq.⁻¹ after only one dip-coating cycle and 8.18 Ω sq.⁻¹ after 5 dip-coating cycles (Fig. S2, ESI†). This sheet resistance is lower than those of most previously reported conductive materials used in piezoresistive pressure sensors, such as CNT-coated fabric (77.4 kΩ sq.⁻¹),¹¹ SWNT/PDMS composites (3.5 × 10⁴ Ω sq.⁻¹),¹³ and AuNW-coated paper (2.5 ± 0.4 MΩ sq.⁻¹).²¹

Fig. 1 The fabrication of wearable pressure sensors. (a) Schematic illustration of fabrication of the textile-based pressure sensors. (b) Schematic illustration of the working principles. (c) A photograph of the pressure sensor. (d) Digital photograph shows that the device is flexible. (e) Large-area circuits on the cotton substrate. (f–h) The SEM images of AgNW-coated cotton fibers at different magnification. (i) SEM image of a screen-printed silver electrode on cotton.

Flexible silver circuits on textiles are fabricated by a screen printing method (Fig. 1a). The Ag paste flows through a designed screen mesh after a squeegee stroke and then wets the cotton substrate. After drying at 80 °C for 25 min, two silver electrodes with an interdigitated configuration are coated on the cotton substrate with an ultrahigh conductivity (sheet resistance is about 0.37 Ω sq.⁻¹). Subsequently, we covered a piece of AgNW-decorated cotton on the top of the silver-coated electrodes, followed by placing a thin VHB adhesive tape for encapsulation while maintaining mechanical flexibility. Such fabricated devices are soft and wearable due to the flexible nature of both cotton and AgNWs (Fig. 1c and d). Fig. 1e shows that a complex circuit of about 50 square centimetres is printed on the fabric with a minimum gap size of only 430 μm (Fig. S3, ESI†), suggesting that it is easy to screen-print large-area and high-density flexible circuits. In addition, screen-printing is also compatible with industrial roll-to-roll processes, which is suitable for low-cost mass fabrication.^27,28

The detector performance of the textile-based pressure is influenced by its microstructure. SEM images of AgNW-decorated cotton (Fig. 1f and g) show the cotton fibers with natural layered and porous microstructures. At a higher magnification, AgNW networks tightly and evenly wrapped around the surface of the cotton fibers (Fig. 1h). Besides, unlike traditional flexible electrodes based on flat substrates, the silver circuits on textiles also have a rough microstructure (Fig. 1i and Fig. S5, ESI†). 3D confocal imaging results show a nonuniform height distribution within a range of 100–800 μm for the AgNW-coated cotton substrate (Fig. S4, ESI†), while the microstructure height of paper and patterned elastomers is normally lower than 100 μm, indicating that the textile has a larger surface roughness.^29,30Fig. 1b shows the sensing mechanism of our pressure sensor. Without pressure, only a small number of contact points exist between those conductive fibers and the silver electrode. When an external pressure is exerted on the textile-based pressure sensor, the micron-scale cotton fibers would deform with the result that the contact area and current transmission pathways between the AgNW-coated fibers and the silver textile electrodes would increase sharply. Upon unloading, both the cotton fibers and the silver electrode recovered to their original shapes, resulting in a reproducible sensing performance.

A SourceMeter (Keithley 2400) was used to explore the electrical performance of textile-based pressure sensors. I–V curves at different pressures (Fig. 2a) are observed to linearly increase with the applied voltage range indicating that the device is in line with Ohm's law.⁹ The device response is steady under a wide range of static pressure (0–50 kPa). Fig. 2b shows static results of current response under various pressures at 0.1 V for the pressure sensor constructed by different dip-coating cycles of AgNWs. When no pressure is applied, the current cannot be detected no matter how many dip-coating cycles of AgNWs because there are no enough contact points between AgNW-decorated cotton and silver circuits as mentioned before. Then as the pressure increases the current shows a sharp increase by four to six orders of magnitude, respectively, demonstrating that our sensors with different dip-coating cycles all have an outstanding current-switching behaviour and achieve a giant high I_on/I_off switching ratio of 10⁶ (I_off is the initial current; I_on is the current under applied pressure). This switching feature is much higher than in most previous reports (Table S1, ESI†),^8–23 making it easier for computers to recognize different pressures through significant changes in current, which is quite suitable for human–computer interactive applications.^22,31 The extreme current switching behavior can be attributed to the fact that the OFF-state is initially at a break insulating condition; when applying pressure, a high ON-state current flow can be attained by highly conductive AgNW-decorated cotton that bridges the two interdigitated silver electrodes. Like many other piezoresistive pressure sensors, electrical hysteresis was present in our devices (Fig. S6a, ESI†) due to local inelastic deformation processes that always exist in the textile (Fig. S6b, ESI†).

Fig. 2 Evaluation of the sensing performance. (a) I–V curves of the device with applied different pressures. (b) Static results of the current response under various pressures at 0.1 V for the pressure sensor constructed using different dip-coating cycles of cotton. (c) The performance comparison of the AgNW-coated cotton with different dip-coating cycles. (d) The cycling test of the flexible pressure sensor under a pressure of 2500 Pa. (e) Response/release time of the device. (f) Current response to the application and removal of a leaf and a small flower on the textile-based pressure sensor, corresponding to pressures of only 1.82 Pa and 0.76 Pa, respectively.

In addition, it is worth mentioning that the sensitivity (S = δ(ΔI/I₀)/δP, where ΔI is the relative change in current, I₀ is the initial current, and P is the applied pressure) of the textile-based pressure sensor at a pressure below 0.6 kPa is strongly dependent on dip-coating cycles of AgNWs, which is greatly improved as indicated in Fig. 2c (Region I), from about 8.82 × 10³ kPa⁻¹ to 5.65 × 10⁶ kPa⁻¹ for the conductive fibers with 1 and 5 dip-coating cycles, respectively. The SEM images of AgNW-decorated fibers demonstrate that the AgNW networks with 5 dip-coating cycles are more massively interlaced than those with 1 dip-coating cycle (Fig. S7, ESI†). The improvement in sensitivity is mainly due to an increase of the electrical contact area caused by the increase of the AgNW density.

To our knowledge, the sensitivity (≈5.65 × 10⁵ kPa⁻¹) in the low-pressure range (0–0.6 kPa) is one of the highest values reported so far (Table S1, ESI†), which is about three to six orders of magnitude higher than those based on polypyrrole (PPy) hydrogels on the PET substrate (<133.1 kPa⁻¹),⁹ CNT-coated fabrics on the Ni electrode (14.4 kPa⁻¹),¹¹ AuNW-decorated paper on the interdigitated electrode (3.5 kPa⁻¹),²¹ and complex micro-pyramid arrays patterned on PDMS (5.53 kPa⁻¹).³² Furthermore, our textile-based pressure sensor preserves an extremely high sensitivity (≈2.0 × 10⁴ kPa⁻¹) over such a wide pressure regime (10–30 kPa), which has never been reported in the previous literature (Table S1, ESI†).^8–23 The ultra-high-sensitivity might be due to the excellent mechanical and structural properties of our textile-based pressure sensor. The degree of increase in contact area with applied pressure depends on the elastic modulus of sensing elements. The large deformation of AgNW-decorated cotton with low elastic moduli (Fig. S6a, ESI†) produces a large increase in contact area.³³ A surface structure with a wide size distribution (Fig. S4, ESI†) is also proposed to improve sensitivity.³⁰ The large surface roughness of the fabrics further increases the contact area between the AgNW-coated cotton and the silver textile electrodes under deformation,³⁴ which effectively promotes the sensitivity of our piezoresistive pressure sensors. In addition, most piezoresistive fibre-based pressure sensors are based on semiconductor nanomaterials (CNTs and graphene) or organic polymers (PEDOT and PPy). Compared with these materials, both AgNW networks and silver electrodes used in our device have outstanding electronic conductivity. Therefore, a tiny pressure-induced change in contact area can contribute to huge current changes, which further improves the detection sensitivity.

We explored the durability of our flexible device under a pressure of 3.4 kPa at a frequency of 1 Hz. The results show that current amplitude almost remains unchanged after about 1000 cycles of repeated loading/unloading (Fig. 2d). The high stability is also indicated in a bending test and the sensing performance of our flexible device shows negligible variation after repeated bending for over 1000 cycles at 25 mm radius (Fig. S8, ESI†). Furthermore, our device shows a rapid response and relaxation properties (6 ms and 13 ms, respectively) to an instantaneous pressure of 500 Pa (Fig. 2e). Fig. 2f illustrates our pressure sensors are still extremely highly sensitive when a very light object such as a leaf and a piece of petal is placed and removed, corresponding to the pressure of 1.82 Pa and 0.76 Pa, respectively. In contrast to previous flexible pressure sensors (Table S1, ESI†), our device not only shows a comparable low response time but also offers remarkable sensitivity to tiny pressures.

Owing to the natural flexibility and the high sensitivity, our textile-based pressure sensor is wearable and has the ability to detect various mechanical signals such as finger motions, hand motions, wind pressure and acoustic vibrations. As shown in Fig. 3a, when the bending angles of our flexible pressure sensors gradually decrease, the contact area between the AgNW-coated fibers and the silver textile electrodes will decrease at the same time. As a result, the current response shows a corresponding decline. Moreover, as the pressure sensor is stretched (Fig. S9a, ESI†) and touched (Fig. S9b, ESI†), the corresponding current signals show significant changes and show excellent repeatability about these multimodal deformations. The results above suggest that our pressure sensor has great potential in monitoring the movement of other body parts. In order to investigate the ability of the textile-based sensor to recognize the sound signal, the device is attached on a speaker that is directly connected to the computer (Fig. 3b). When the speaker plays a section of music (music name: Burning), fluctuations in the output current curves are observed in Fig. 3c, in which the amplitude and the frequency of the curves represent sound features. It is worth noting that our devices exhibited excellent reproducibility as the music was played again, demonstrating its potential uses in various applications such as dynamic sound detections and voice recognition.¹³ Due to the ultrahigh sensitivity, our device also could distinguish gentle wind blowing from the mouth. Once the wind reaches the surface of our device, the tiny mechanical vibration caused by wind pressure could be quickly recorded and then the response current gradually returned to the initial state as the wind disappears (see Fig. 3d). To further verify its ability to detect wind signals, the device was directly attached on the air outlet of the mask. It can accurately monitor the changes in respiratory frequency before (14.1 times per min) and after exercise (30.3 times per min) as shown in Fig. 3f, providing a facile and effective way for future respiratory signal testing.

Fig. 3 Real-time current response to different mechanical forces using our textile-based sensor. (a) Optical images and the current signal of our devices in response to different bending angles. (b and c) Photograph images of the device attached on a speaker and the real-time current signal responding to playing the same music twice (music name: Burning). (d) Digital photographs and the current curve for detecting the wind blowing from the mouth. (e) The device was directly attached on the air outlet of the mask. (f) The current signal responds to respiration under before and after exercise. (g) Photographs showing that the flexible detector is fixed on the wrist by an elastic spandex band, and the human motion of bending the wrist (Motion I), waving (Motion II) and holding fist (Motion III). (h) Real-time current response corresponding to different human motions. (i) The current signal in response to the whole process of lifting and lowering the cup.

As our pressure sensor is mainly based on the cotton substrate, it is convenient to stitch our all-textile device onto fabrics. Fig. 3g shows that the flexible detector secured to the wrist can be used for monitoring different hand motions, such as bending wrist (Motion I), waving hand (Motion II) and making a fist (Motion III) (Fig. 3h). What's more, the device exhibited excellent detection performance and we could recognize the entire process of lifting and lowering the cup through the current waveforms, as shown in Fig. 3i. Interestingly, the current amplitude during holding the cup was significantly greater than the current without the cup, which was mainly due to more muscle deformation when holding the object. These results suggest that our device is wearable, flexible, and highly sensitive in the low-pressure range, and has potential applications in wearable gesture controllers and various healthcare for assessing walking steps or sleeping quality.^35,36

LED arrays are integrated over a large area on a textile substrate. Fig. 4a shows that the 4 × 4 pixel sensing arrays (5 × 5 mm² each) are respectively connected to the 4 × 4 LED display dot matrix according to the mirror sequence, which makes external forces become visible and human-readable signals. We sandwiched AgNW-coated cotton as the top electrode and the VHB film for encapsulation. The sensing arrays and the LED display dot matrix can be folded as well as the cloth (see Fig. 4b). The current response of our pressure sensor touched by two fingers can be visually observed in Fig. 4c, and the position of the bright LED represents the touching area, while the related brightness is in response to applied pressures. The pressure response of a single LED with the connected pressure sensor is depicted in Fig. 4d. When the applied pressure is larger than 100 Pa, an obvious increase of current flow can be observed. Fig. 4e illustrates the current and related brightness of the LEDs in response to different applied pressures at a voltage of 4 V. The results indicate that both the current and brightness increase drastically with the applied pressure increased from 200 Pa to 3.2 kPa. From the slope of the response curve, our user-interactive system exhibited high sensitivity (80.1–998.7 Cd m⁻² kPa⁻¹ for the pressure range of 0.2–1.2 kPa, ≈1400 Cd m⁻² kPa⁻¹ for the pressure range of 1.2–3.2 kPa). Our user-interactive system is sensitive enough that we can rapidly identify the external pressure based on the brightness of the LED with the naked eye when an empty glass bottle (14.4 g) is placed on the sensor arrays (Fig. 4f and g). In contrast with the previous reports that sophisticated data acquisition circuits and electronic boards are needed, our textile-based pressure sensors can be seamlessly integrated with functional devices on the textile, which can simplify the circuit design and reduce large-scale integration costs.^22,23

Fig. 4 Large-scale integration for the user-interactive E-textile. (a) Photograph of the pressure sensor arrays and LED arrays were integrated on a textile substrate. (b) The user-interactive E-textile was folded. (c) Digital photograph shows that the sensor arrays were pressed by two fingers and the brightness of the LEDs monitors the pressed area. (d) I–V characteristics of the LED unit and the pressure sensor integration under different applied pressures. (e) The current and related brightness of the LEDs in response to different applied pressures at a voltage of 4 V. (f and g) Photograph of an empty glass bottle (14.4 g) was placed on the sensor arrays and the LED arrays display and map the pressure.

Furthermore, our textile-based pressure sensors can be used for a human–computer interactive system as a real wearable electronic platform. The sensing arrays were integrated into a microcontroller unit (MCU), which was designed to help communication between the sensors and the personal computer (Fig. 5a and Fig. S10, ESI†). A block diagram of the designed MCU is shown in Fig. 5b. The current response of each sensor was amplified and converted to a digital form in order for the computer to recognize it. Sensing arrays with nine independent detection units were sewed onto a sleeve (Fig. 5c). When numbers “2”, “4”, “6”, and “8” are pressed separately (Fig. 5d), each pressure sensor exhibited more than four orders of magnitude change in current without interference between each sensor. Such large current changes can be acquired and identified. Based on this mechanism, various applications can be realized by integrating the textile sensing arrays. As shown in Fig. 5e and f, playing the piano and computer games could be accomplished via our smart wearable system (Movies s1 and s2, ESI†).

Fig. 5 A real-wearable sensor platform for a human–computer interactive system. (a) Schematic of a human–computer interactive system. (b) A block diagram of the designed MCU. (c) The sensing arrays in which nine independent detection units were touched by a finger. (d) Current response of all nine sensors against a human finger. (e and f) The sensing arrays can be used to play piano and play computer games.

In summary, supersensitive and large-area integrated all-fabric pressure sensors have been fabricated using a low-cost and scalable fabrication process. Our pressure sensor is composed of a bottom interdigitated textile electrode screen-printed using silver paste and a top bridge of AgNW-coated cotton fabric. Benefiting from the layered and porous microstructure of the cotton fibers, the resulting all-textile pressure sensors achieve a high sensitivity of 2.46 × 10⁴ kPa⁻¹ to 5.65 × 10⁶ kPa⁻¹ over a wide pressure regime (0–30 kPa), a giant high on/off ratio of ≈10⁶, a fast response time (6 ms), and a low detection limit (0.76 Pa). The devices could detect various mechanical signals such as acoustic vibrations, hand motions and even respiration with excellent repeatability. Large-area sensor arrays could be facilely printed on clothes and applied in human–computer interactive systems, which was demonstrated by playing the piano and computer games. Such pressure sensor arrays and a fabrication process should be promising in the field of interactive wearable electronics, smart fabrics and robotic systems.

Experimental section

The preparation of AgNWs

AgNWs were synthesized through a polyol method. First, 50 mM NaBr and 100 mM AgNO₃ solution in ethylene glycol (EG) were prepared, and 30 mg of polyvinyl pyrrolidone (PVP) was dissolved in 6 mL of EG and heated in an oil bath at 160 °C for 60 min under magnetic stirring (500 rpm). Then 50 μL of NaBr and AgNO₃ solution was transferred into the vial. After 5 min, 2 mL of the AgNO₃ solution was added to the vial at an injection rate of 5 μL s⁻¹. When the mixture solution became shiny, the reaction was quenched by an ice-water bath for 30 min. Subsequently, AgNWs were washed with acetone and EG and collected by centrifugation at 3000 rpm.

Fabrication of the all-textile pressure sensors

First, the cotton sheets were cleaned with anhydrous ethanol and deionized water (DI) several times, and then the cotton sheets were dipped into an aqueous solution of AgNWs followed by drying at 150 °C for 10 min. The dip-coating cycles were repeated for 2 to 5 cycles, leading to AgNW-coated cotton with different sheet resistances. Subsequently, the commercially conductive silver paste (ENSON CD-03, China) was printed on the pre-cleaned cotton through screen-printing. After drying at 80 °C for 25 min, patterned silver electrodes on the cotton substrate with ultrahigh conductivity were obtained. At last, the bottom silver electrodes and top AgNW-coated cotton were encapsulated using a VHB adhesive film.

Large-area patterning and fabrication

The patterned silver electrodes arrays (5 × 5 mm² each) and the conductive silver circuit were fabricated on the cotton substrates using screen printing. The AgNW-coated cotton was cut using a knife into 5 × 5 mm² pieces and then each piece was placed on the electrode pixels and sandwiched between a 0.2 mm-thick VHB film, leading to large-area, patterned pressure sensors.

Device characterization

The SEM images were obtained using a QUANTA 250 (FEI, America). To observe the responses of piezoresistive sensors against multiple stimuli, the current and sheet resistance were recorded using a digital source meter (Keithley 2400, America) and an electrochemical workstation (CHI 660E, China). The sensitivity of the textile-based pressure sensor was measured using a computer controlled force gauge (HP-10, China Handpi Instruments), which was applied as the external force. The electroluminescence intensity of the green LED was characterized using a luminance meter (ST-86LA, China).

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors gratefully acknowledge the National Natural Science Foundation of China (51503022, U1663229, and 61705026); the Scientific and Technological Research Program of Chongqing Municipal Education Commission (KJ1601114); the Talent Introduction Project of Chongqing University of Arts and Sciences (R2016XC14); the Overseas Returnees Support Program for Innovation and Entrepreneurship of Chongqing (cx2018136) and the Basic and Frontier Research Program of Chongqing Municipality (cstc2016jcyjA0577) for providing support to this work.

Notes and references

1. Y. Wei, S. Chen, X. Yuan, P. Wang and L. Liu, Adv. Funct. Mater., 2016, 26, 5078–5085.
2. J. Lee, H. Kwon, J. Seo, S. Shin, J. H. Koo, C. Pang, S. Son, J. H. Kim, Y. H. Jang, D. E. Kim and T. Lee, Adv. Mater., 2015, 27, 2433–2439.
3. N. Luo, W. Dai, C. Li, Z. Zhou, L. Lu, C. C. Y. Poon, S. Chen, Y. Zhang and N. Zhao, Adv. Funct. Mater., 2016, 26, 1178–1187.
4. C. Pang, G. Lee, T. Kim, S. M. Kim, H. N. Kim, S. Ahn and K. Suh, Nat. Mater., 2012, 11, 795–801.
5. J. Chen, Y. Huang, N. Zhang, H. Zou, R. Liu, C. Tao, X. Fan and Z. L. Wang, Nat. Energy, 2016, 1, 16138.
6. M. Hamedi, R. Forchheimer and O. Inganäs, Nat. Mater., 2007, 6, 357–362.
7. R. Salvado, C. Loss, R. Gonçalves and P. Pinho, Sensors, 2012, 12, 15841–15857.
8. Z. Ma, A. Wei, J. Z. Ma, L. Shao, H. Jiang, D. Dong, Z. Ji, Q. Wang and S. Kang, Nanoscale, 2018, 10, 7116–7126.
9. L. Pan, A. Chortos, G. Yu, Y. Wang, S. Isaacson, R. Allen, Y. Shi, R. Dauskardt and Z. Bao, Nat. Commun., 2014, 5, 3002.
10. H. Chou, A. Nguyen, A. Chortos, J. W. F. To, C. Lu, J. Mei, T. Kurosawa, W. Bae, J. B. H. Tok and Z. Bao, Nat. Commun., 2015, 6, 8011.
11. M. Liu, X. Pu, C. Jiang, T. Liu, X. Huang, L. Chen, C. Du, J. Sun, W. Hu and Z. L. Wang, Adv. Mater., 2017, 29, 1703700.
12. Y. Wei, S. Chen, Y. Lin, X. Yuan and L. Liu, J. Mater. Chem. C, 2016, 4, 935–943.
13. X. Wang, Y. Gu, Z. Xiong, Z. Cui and T. Zhang, Adv. Mater., 2014, 26, 1336–1342.
14. S. Y. Kim, S. Park, H. W. Park, D. H. Park, Y. Jeong and D. H. Kim, Adv. Mater., 2015, 27, 4178–4185.
15. X. Wang, H. Zhang, L. Dong, X. Han, W. Du, J. Zhai, C. Pan and Z. L. Wang, Adv. Mater., 2016, 28, 2896–2903.
16. B. Hwang, J. Lee, T. Q. Trung, E. Roh, D. Kim, S. Kim and N. Lee, ACS Nano, 2015, 9, 8801–8810.
17. Y. Wei, S. Chen, X. Dong, Y. Lin and L. Liu, Carbon, 2017, 113, 395–403.
18. S. Yao and Y. Zhu, Nanoscale, 2014, 6, 2345–2352.
19. J. Chun, K. Y. Lee, C. Kang, M. W. Kim, S. Kim and J. M. Baik, Adv. Funct. Mater., 2014, 24, 2038–2043.
20. X. Wang, H. Zhang, L. Dong, X. Han, W. Du, J. Zhai, C. Pan and Z. L. Wang, Adv. Mater., 2016, 28, 2896–2903.
21. S. Gong, W. Schwalb, Y. Wang, Y. Chen, Y. Tang, J. Si, B. Shirinzadeh and W. Cheng, Nat. Commun., 2014, 5, 3132.
22. Y. Lai, B. Ye, C. Lu, C. Chen, M. Jao, W. Su, W. Hung, T. Lin and Y. Chen, Adv. Funct. Mater., 2016, 26, 1286–1295.
23. C. Wang, D. Hwang, Z. Yu, K. Takei, J. Park, T. Chen, B. Ma and A. Javey, Nat. Mater., 2013, 12, 899–904.
24. N. Matsuhisa, D. Inoue, P. Zalar, H. Jin, Y. Matsuba, A. Itoh, T. Yokota, D. Hashizume and T. Someya, Nat. Mater., 2017, 16, 834–840.
25. S. Takamatsu, T. Lonjaret, E. Ismailova, A. Masuda, T. Itoh and G. G. Malliaras, Adv. Mater., 2016, 28, 4485–4488.
26. R. R. Da Silva, M. Yang, S. Choi, M. Chi, M. Luo, C. Zhang, Z. Li, P. H. C. Camargo, S. J. L. Ribeiro and Y. Xia, ACS Nano, 2016, 10, 7892–7900.
27. Y. S. Rim, S. Bae, H. Chen, N. De Marco and Y. Yang, Adv. Mater., 2016, 28, 4415–4440.
28. R. Faddoul, N. Reverdy-Bruas and A. Blayo, Mater. Sci. Eng., B, 2012, 177, 1053–1066.
29. J. Shi, L. Wang, Z. Dai, L. Zhao, M. Du, H. Li and Y. Fang, Small, 2018, 14, 1800819.
30. P. Yu, K. Zhang, Y. Zhen, J. Song, Z. Ju, Y. Li, X. Wang, D. Wang, M. Jian and Y. Zhang, ACS Nano, 2018, 12, 2346–2354.
31. L. Mottola and G. P. Picco, ACM Comput. Surv., 2011, 43, 1–51.
32. B. Zhu, Z. Niu, H. Wang, W. R. Leow, H. Wang, Y. Li, L. Zheng, J. Wei, F. Huo and X. Chen, Small, 2014, 10, 3625–3631.
33. J. Park, Y. Lee, J. Hong, M. Ha, Y. D. Jung, H. Lim, S. Y. Kim and H. Ko, ACS Nano, 2014, 8, 4689–4697.
34. W. Zeng, L. Shu, Q. Li, S. Chen, F. Wang and X. Tao, Adv. Mater., 2014, 26, 5310–5336.
35. 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. Kim, Nat. Nanotechnol., 2014, 9, 397–404.
36. J. Duan, X. Liang, J. Guo, K. Zhu and L. Zhang, Adv. Mater., 2016, 28, 8037–8044.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c8tc02716a

---