Research Article www.acsami.org
Cite This: ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
High-Performance and Multifunctional Skinlike Strain Sensors Based on Graphene/Springlike Mesh Network Wenjing Yuan,*,†,‡ Jinzheng Yang,† Kai Yang,† Huifen Peng,†,‡ and Fuxing Yin*,†,‡ †
School of Materials Science & Engineering and ‡Tianjin Key Laboratory of Materials Laminating Fabrication and Interface Control Technology, Hebei University of Technology, Tianjin 300130, China S Supporting Information *
ABSTRACT: The development of skinlike strain sensors that are integrated with multiple sensing functions has attracted tremendous attention in recent years. To mimic human skin, strain sensors should have the abilities to detect various deformations such as pressing, stretching, bending, and even subtle vibrations. Here, we developed a facile, cost-effective, and scalable method for fabrication of high-performance strain sensors based on a graphene-coated springlike mesh network. This composite-based sensor exhibits an incorporation of low detection limit (LOD) for minute deformation (LOD of 1.38 Pa for pressure, 0.1% for tensile strain, and 10 μm for vibration), multiple sensing functions, long-term stability, and wide maximal sensing range (up to 80 kPa for pressure and 110% for tensile strain). On the basis of its superior performance, it can be applied for in situ monitoring of human motions ranging from subtle physiological signals (e.g., pulse, respiration, and phonation) to substantial movements (e.g., finger bending). KEYWORDS: strain sensor, springlike mesh network, graphene, multifunction, human motion detection
■
INTRODUCTION Skinlike strain sensors that mimic human skin to interact with the surroundings and sensitive to environmental stimuli have attracted considerable attention because of their applications in human motion detection, health monitoring, and human− machine interfaces.1−9 For practical applications, strain sensors integrated with high sensing performance and multiple sensing functions, for example, pressure, stretch, and bending detection, are highly desired.1,10−15 Recently, various conductive materials have been adopted as sensing materials for the design of novel strain sensors, including metal nanoparticles,16 metal nanowires,17−19 conducting polymers,20−24 and carbon nanomaterials.25−30 Among these, graphene has become a promising candidate because of its one-atom-thick two-dimensional (2D) structure, excellent conductivity, large surface area, and mechanical flexible, yet robust properties.31−38 In fact, graphene has been widely investigated for sensing applications, for example, gas sensing,39,40 chemical sensing,41,42 biological sensing,43,44 and strain sensing.45 The large surface area and abundant surface functional groups endow graphene with various sensing sties that can interact with gas molecules, chemicals, or biomolecules through chemical bonding or noncovalent bonding.42 The unique one-atom-thick structure makes graphene sheets easily deformed in the direction normal to its surface and provides good flexibility, making it appropriate for strain sensors. Conventional resistive-type strain sensors comprise two main parts: conductive sensing layer and flexible substrates.46,47 Compared to zero-dimensional or one-dimensional nanomaterials, 2D-graphene can offer large © XXXX American Chemical Society
contact area with the substrates, resulting in strong interface bonding between the substrates and conductive materials, endowing the sensor with long-term stability.47 However, most of the conventional graphene-based planer strain sensors exhibited narrow sensing ranges because of the intrinsic small deformations of graphene sheets or low sensitivity for small deformations due to the fixed connections of sensing materials.48,49 To obtain high sensitivity, low detection limit (LOD) together with large sensing range, microstructured flexible substrates or conductive layers were adopted. Up to now, flexible substrates with micropyramid arrays,50 microdome arrays,47 and interlocking microstructures51,52 have been used for fabrication of highly sensitive strain sensors. It is worth noting that the aforementioned microstructures were mainly fabricated by the lithography technique, which is costly, complicated, and time-consuming. To solve this problem, other methods have been adopted to fabricate flexible substrates with microstructures. For instance, Zhang et al. reported highly sensitive pressure sensors based on hierarchical structures molded from natural leaves.47 Wang and co-workers reported a pressure sensor with microstructured polydimethylsiloxane (PDMS) films molded from silk textiles.53 Liu et al. reported a strain sensor with fish scalelike graphene for sensitive detection of human stretch.11 However, there are still some challenges to be addressed before achieving an ideal strain Received: April 21, 2018 Accepted: May 23, 2018
A
DOI: 10.1021/acsami.8b06496 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces
GSMN film to connect copper foil as electrodes. The relative orientation of the GSMN films could be parallel to each other with the direction of PU fibers parallel to the copper foil. Finally, the sensor was sandwiched by two pieces of PDMS films for protection. The resulting device is shown in Figure S1a. For tensile strain sensing, only one piece of GSMN was required. After silver paste coating and copper foil connecting, the sensor was sandwiched by two pieces of PDMS films for protection. The conductance variations of the strain sensor were recorded on an Electrochemical Analyzer (CHI 760E potentiostat− galvanostat, CH Instruments Inc.) by applying a constant bias voltage of 1 V. Repetitive and consecutive pressures, tensile strains, or vibrations were provided by dynamic mechanical analysis (DMA, Q800, TA Instruments). The bending forces were applied by hands and paper rings with fixed radius were used to ensure the same angle when the bending forces were exerted. Characterization. The morphologies of the samples were recorded by using a field emission scanning electron microscope (JEOL JSM7100F). Raman spectra were recorded using a Renishaw Raman microscope (RM2000) with a 514 nm laser in the range of 200−3200 cm−1 at a power density of 4.7 mW.
sensor. For example, most of the reported sensors can monitor only one or at most two types of physical forces.30,54,55 Particularly, simultaneous detection of pressure, stretch, bending, and vibration with low LOD in the broad sensing range can be hardly achieved. Here, we developed a multifunctional skinlike strain sensor based on springlike microstructured graphene/polymer mesh network that can monitor pressure and stretch together with vibration and bending deformations. The polymer mesh network adopted here has springlike inner architectures (springlike mesh network, SMN) together with the properties of soft, breathable, highly flexible, and comfortable to human skin. Chemically converted graphene (CCG) sheets can be conformally attached onto the surface of SMN, resulting in a springlike conductive mesh network. Two pieces of the conductive mesh network were assembled face to face for a graphene-coated springlike mesh network (GSMN)-based sensor. The resulting strain sensor exhibited a high sensitivity of 72 kPa−1, low LOD of 1.38 Pa, and large sensing range up to 80 kPa for pressure. The smallest tensile strain detectable for this sensor is as minute as 0.1% with a sensing range up to 110%. Meanwhile, a vibration amplitude as low as 10 μm can be identified by this GSMN-based sensor successfully. Because of its superior sensitivity for minute deformation, this strain sensor can detect full-range human motions from tiny physiological signals, for example, phonation, respiration, and pulse beat to large-scale physical activities, for example, finger bending, making it practical for electronic skin applications.
■
■
RESULTS AND DISCUSSION The morphology of springlike mesh network was characterized by scanning electron microscopy (SEM) (Figure 1a), which shows rich surface textures in microscale. Highly elastic PU core fibers with diameters around 40 μm were helically winded by lots of PAI fibers with average diameters of 15 μm, forming springlike fiber architectures. These spring-structured fibers were interconnected by PAI fibers to form a hierarchical mesh network in large scale, serving as the stretchable scaffold. After assembly of CCG, the mesh network shows a morphology of wrinkled CCG coating (Figure 1b and inset), and the overall thickness of the resulting GSMN film was measured to be around 250 μm. CCG sheets adopted here were single-layer structures with lateral dimensions ranging from hundreds of nanometers to 2 μm (Figure S1b). The conformal coating of the CCG sheets on the spring-structured fibers enables the possibility to fabricate skinlike strain sensors with high sensitivity and stability. The Raman spectrum of this GSMN film displays a D-band at 1337 cm−1 and a G-band at 1593 cm−1, confirming the successful coating of CCG sheets (Figure S2). The G-band is associated with the first-order scattering of the E2g mode. The D-band is attributed to the structural defects related to the partially disordered structures or the attachment of functional groups on the graphitic domains.40 Pressure Sensing. To measure the responses of our GSMN-based strain sensor dynamically, a home-made system containing a DMA system and an Electrochemical Analyzer were designed. Such a system can provide a dynamic pressure up to 80 kPa with electrical signals simultaneously recorded. Figure 2a shows relative conductance profiles (ΔG/G0) of the GSMN-based sensor at different pressures for multiple loading−unloading cycles. The sensitivity of the strain sensor is defined as S = (ΔG/G0 %)/ΔP, where G0 is the conductance of sensor under no load, ΔG is the conductance change upon pressure loading, and ΔP is the change in applied pressure. As seen in Figure 2a, the conductance of the sensor undergoes an intensive increase after pressure loading. Once the pressure is unloaded, the conductance quickly returns to its pristine value, indicating fast sensing ability. The smallest pressure that can be provided by our pressure-supply instrument is 10 Pa, and a significant response of 0.54% can be output by our sensor upon this pressure loading. At a higher pressure region, the stable, continuous, and noise-free responses could be observed up to 80 kPa (Figure S3). The relative conductance variation−
EXPERIMENTAL SECTION
Preparation of CCG. Graphene oxide (GO) was prepared through a modified Hummers’ method. Details are reported in the literature studies.56,57 CCG was prepared following the protocol reported before.58 In typical, 25 mg of GO was dispersed in 100 mL of water by sonication. Successively, 375 μL of ammonium hydroxide (30%) and 28 μL of hydrazine were added to the GO dispersion. Mild sonication was applied for uniform dispersion. Then, the mixture was heated at 92 °C for 1 h for reduction. Finally, a small amount of aggregation was removed by filtration to yield a stable CCG suspension. Assembly of CCG Sheets. SMN adopted here is commercially available core-spun yarn textile [20% polyurethane (PU) fibers and 80% polyamide (PAI) fibers, and detailed structures are shown in Figure 1a]. A piece of purchased SMN was alternately rinsed with
Figure 1. SEM images of SMN (a) before and (b) after assembly of CCG sheets. Inset is the enlarged image of the GSMN. deionized water and ethanol under sonication. After being dried in air and further treated by using a UV−ozone cleaner, the mesh network was hydrophilic. Then, it was immersed in CCG solution (0.2 mg mL−1) for 10 min and dried at room temperature, forming the CCG/ springlike mesh network, denoted as GSMN. Construction and Testing of the GSMN-Based Strain Sensor. Two pieces of GSMN films were placed face to face to construct the strain sensor for pressure, bending, vibration sensing, and wearable electronics applications. Silver paste was coated on one edge of each B
DOI: 10.1021/acsami.8b06496 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces
Figure 2. (a) Plot of conductance variation (ΔG/G0) as a function of time at different pressures ranging from 10 Pa to 2 kPa applied on the GSMNbased strain sensor. (b) Plot of conductance variation as a function of pressure ranging from 10 Pa to 80 kPa. (c) Comparison of five different construction types of strain sensors (GSMN−GSMN sensor: black, GPU−GSMN sensor: red, GPU−GPU sensor: blue, GPAI−GSMN sensor: green, and GPAI−GPAI sensor: grey) in regard to conductance variation at 500 Pa. (d) Relative responses of the GSMN sensor upon water droplets. (e) Durability test upon a pressure of 200 Pa for 2500 cycles, 4 s for each cycle.
Figure 3. (a) Typical current−strain curve of a GSMN-based sensor recorded at a stretching rate of 10% min−1. (b) Plot of conductance variation as a function of time at different tensile strain ranging from 0.1 to 30%. (c) Comparison of GSMN-sensor, GPU-sensor, and GPAI-sensor in regard to conductance variation at a tensile strain of 5%. (d) Stress−strain curves of the GSMN, GPU, and GPAI films. (e) Performance of the GSMN-based strain sensor at 1% strain for 500 loading−unloading cycles, 12 s for each cycle.
and bottom mesh network (GSMN−GSMN sensor) exhibited the highest sensitivity. At a pressure of 500 Pa, the GSMN− GSMN sensor showed a response of 24%, while GPU−GSMN, GPU−GPU, GPAI−GSMN, and GPAI−GPAI sensors showed much less conductance change (13.67, 6.14, 1.76, and 0.16% respectively). It is obvious that pressure stimuli can lead to larger deformation for the GSMN films, thus resulting in more significant response than that of planar GPU or GPAI films. In fact, the microstructures of GSMN can endow the resulting sensor with a high conductive pathway increase ratio. In the original state, the top and bottom GSMN films partially contact with each other and form conductive pathways. Upon pressure loading, the contact area increased. Different from bulk planar films, GSMN has hierarchical microstructures with springlike internal arrangement, exhibiting uneven surfaces. A small
pressure curve (Figure 2b) demonstrates that the GSMN strain sensor has a broad sensing range (up to 80 kPa), much wider than most of the pressure sensors reported before.47,59,60 In the higher pressure region, the conductance variation and applied pressure accord with the exponential function (ΔG/G = −2.372 e(−P/27.5) + 2.5). An approximately linear relationship exhibits between ΔG/G0 and P in a pressure region lower than 0.3 kPa. The corresponding sensitivity in this range is calculated to be 72 kPa−1. As a contrast, we adopted graphene-coated planar PAI films (GPAI) or planar PU films (GPU) as device component for comparison. The following five types of pressure sensors were fabricated: GSMN−GSMN sensor, GPAI−GSMN sensor, GPAI−GPAI sensor, GPU−GSMN sensor, and GPU−GPU sensor. As shown in Figure 2c, the strain sensor with both top C
DOI: 10.1021/acsami.8b06496 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX
Research Article
ACS Applied Materials & Interfaces
within 150% strain (480 times and 3.9 × 104 times lower than that of planar PU or PAI films), suggesting a high elasticity. It is generally accepted that both low modulus and large elongation are extremely important for flexible strain sensors. The conductance change in GSMN could be explained by a “compression spring” theory.14 The springlike inner structures can prestore deformations. Upon stretching, the prestored deformation inside the structure released, like a compression spring. To further understand the mechanism underlying the conductance variation in strain sensing, the microstructures of the GSMN layer before and after stretched for 50% strain were studied (Figure 4a,b). The SEM results suggested that the
compressive deformation can enable more contact area increase between or inside the GSMN films, leading to more conductive pathways. To further demonstrate the LOD of the sensor, water droplets were dropped onto the sensor to induce minute pressure loading. As seen in Figure 2d, the sensor exhibits a significant conductance change of 0.057% upon a drop of water. It is calculated that the pressure induced by a drop of water is only 1.38 Pa. This value is much lower or at least comparable to most of the reported pressure sensors. With increasing the number of droplets, the conductance increase corresponding to each droplet is almost constant, demonstrating good reliability and linear behavior of the sensor. The durability of the GSMNbased strain sensor was demonstrated in Figure 2e. Under a pressure of 200 Pa, the conductance response exhibited negligible changes after 2500 loading−unloading cycles, implying a high signal stability for long-term pressure input. Stretch Sensing. The stretch sensing performance of the GSMN-based sensor was monitored by recording the relative conductance variations (ΔG/G0) upon stretching to different strains (ε). Upon applying a constant electrical voltage, the current flowing through the GSMN sensor decreases gradually with the increase of tensile strain (Figure 3a) because of the increase in its resistance. As shown in Figure 3a, similar to most of the reported strain sensors, the plot of current versus strain is composed of more than one region. When the applied tensile strain is higher than 10%, the sensitivity [S = (ΔG/G0)/ε] of the device is 1.63. Whereas when the tensile strain is in the range of 10−100%, the sensitivity is calculated to be 0.37. In the higher deformation region, the sensitivity is lower, indicating a saturation trend. This phenomenon is related to the microstructures of the GSMN. At small strain, the parallel winding loops gradually separated with each other, which impact the conductance of the film intensively. However, when the applying strain becomes larger, the conductance changes mainly because of the winding angel change or even the stretching of core fibers and winding fibers, which impact the conductance change less significantly. The relative conductance−strain curve also demonstrates that the sensor has a broad sensing range (up to 110%). We performed cycling stretching and releasing for different strains ranging from 0 to 30%, and the resultant curves are plotted in Figure 3b, demonstrating the sensor with excellent cycling sensing performance. Benefiting from the high sensitivity at a small tensile strain range, the LOD could be as minute as 0.1%. The output signal was highly reproducible at minute strains (Figure S4). This extremely low LOD can ensure our GSMN-based sensor for monitoring subtle tensile strain deformations. In contrast, the previously reported strain sensors, including those based on graphene materials, usually have narrow sensing ranges (
