A transparent and stretchable all-graphene multifunctional electronic-skin sensor matrix is developed. Three different functional sensors are included in this matrix: humidity, thermal, and pressure sensors. These are judiciously integrated into a layer-by-layer geometry through a simple lamination process. Electronic skins (E-skins) are flexible circuitry matrices in which each sensor cell can transduce external stimuli on epidermis to electronic signals. E-skins are potentially useful in the fabrication of wearable human-machine interfacing devices, remote real-time health monitoring, implantable prosthetics, and multifunctional smart skins.1-6 For example, pressure-sensitive rubbers integrated with organic transistor matrices have been used to map pressure distributions.7-11 A variety of skin-like pressure or strain sensors fabricated by sandwiching elastomeric materials between soft conductors have also been reported.12-17 A piezoelectric material or interdigitated electrode design has been introduced into a graphene transistor matrix to prepare a stretchable tactile E-skin.18-20 Previous studies of E-skins have mainly focused on tactile sensors that transduce physical variables (pressure, shear, or strain) into electronic signals. Plausible mimics of multifunctional human skin will require multimodal detection, including temperature, humidity, and pressure, integrated into a single pixel. Practical E-skin matrices require two additional qualifications: (i) simultaneous multiple stimuli sensing and (ii) low-cost and facile fabrication processes that minimize materials. The specifications may be met by developing rational device architecture designs using simple sensing materials. Graphene, a 2D hybridized carbon layer with a hexagonal honeycomb lattice, has attracted attention for its utility in a variety of electronic device applications. Its unique charge transport allows a high carrier concentration (1013 cm–2) with a mobility exceeding 104 cm2 V–1 s–1 under ambient condition.21 Excellent mechanical properties extend the applicability of graphene to stretchable devices, and a good thermal conductivity enhances heat dissipation in highly integrated circuits.22, 23 High-quality large-area graphene may be used in transparent conducting electrodes fabricated through chemical vapor deposition (CVD) and subsequent roll-to-roll transfer processes. These graphene electrodes exhibit a sheet resistance as low as 125 Ω sq−1 and a 97% optical transmittance.24 Graphene oxide (GO) or its reduced form, reduced graphene oxide (rGO), two graphene derivatives, may also be mass-produced using a solution process called Hummers method by exfoliating the materials from graphite.25-27 The surface functional groups on GO or rGO, including hydroxyl, carboxyl, and epoxy groups, are very sensitive to environmental conditions, including humidity, chemicals, and temperature, and they have been widely used as sensing materials.28-33 The adoption of versatile graphene derivatives in E-skin applications paves the way to achieve transparent and multifunctional sensors with facile fabrication process. Here, we developed a transparent and stretchable all-graphene multifunctional E-skin sensor matrix. Three different functional sensors were included in this matrix: humidity, thermal, and pressure sensors, and were judiciously integrated into a layer-by-layer geometry through a simple lamination process. CVD-grown graphene was used to form the electrodes and interconnects for these three sensors, whereas GO and rGO were used as the active sensing materials for the humidity and temperature sensors, respectively. The top polydimethylsiloxane (PDMS) substrate, which bore the GO humidity sensor array, was laminated in a crisscross fashion onto the top of the bottom PDMS substrate, which bore the rGO temperature sensor array. The arrays were prepared to have the same geometry. The top PDMS substrate sandwiched between two CVD-graphene electrodes acted as an active layer for the capacitive pressure and strain sensors. Together, the sensors monitored a variety of daily life sensations (e.g., a hot wind blowing, breathing, and finger touching) with excellent sensitivity. Each sensor in the matrix exhibited simplex sensing performance: it was only sensitive to its specific stimulation and gave no response to other stimulations. The three sensors in the matrix detected external stimuli simultaneously and relayed independent electrical signals. 2D color mappings of the simultaneous multifunctional sensing were collected. The device architecture developed here for use as a multifunctional E-skin sensor matrix not only avoided the preparation of several materials separately; it enabled sensor integration using a simple lamination method. Figure 1a shows a schematic diagram of the method used to fabricate the multifunctional sensor matrix (a 6 × 6 sensors array in this work, demonstrated using a simple 2 × 2 array). The top GO-based humidity sensor and the bottom rGO-based thermal sensor were prepared using the same geometry: four CVD-graphene (Gr) subelectrodes on each substrate shared two CVD-Gr main electrodes. The sheet resistance of the CVD-graphene electrode was measured to be around 500 Ω sq−1. Prior to transferring the CVD-Gr electrodes onto the PDMS substrate, the substrate was treated with O2 plasma to form a hydrophilic surface. The GO or rGO dispersion (characterized by AFM, XPS, and FT-IR spectroscopy, as shown in Figure S1 in the Supporting Information) was spray-coated onto the region between the patterned CVD-Gr electrodes on the two PDMS substrates to form the sensing channel. The top PDMS substrate prepared with a GO-based humidity sensor array was then laminated onto the top of the bottom PDMS substrate prepared with the rGO thermal sensor array, after rotation through a 90° angle, followed by a subsequent degassing process in a vacuum chamber over 30 min to remove bubbles between the two layers. Prior to lamination, the backside of the top PDMS substrate was treated with O2 plasma to promote interfacial adhesion between the two PDMS layers via the dehydration reaction between -Si-OH groups present on the PDMS surface.34 The top PDMS substrate sandwiched between the top and bottom CVD-Gr electrodes formed capacitive pressure sensors. Note that only graphene derivatives such as CVD-Gr, GO, and rGO were utilized to fabricate the multimodal sensing matrix (Figure 1b), thereby avoiding the use of several distinct materials. The simple fabrication process, which included lamination and degassing steps, permitted the facile integration of a variety of sensors at high yield and low cost. An equivalent circuit diagram of the multifunctional sensors matrix is displayed in Figure 1c. The matrix consisted of three different sensors: a GO-based impedance humidity sensor (red) on the top layer, an rGO-based resistive thermal sensor (blue) on the bottom layer, and a PDMS-based capacitive pressure sensor (green) sandwiched between the top (black line) and bottom (gray line) CVD-Gr electrodes. The resistive sensors were tested using a DC input, whereas the impedance and capacitive sensors were tested using an AC input. In the multimodal E-skin sensor matrix, it was crucial to distinguish the specific output signals from each sensor simultaneously (discussed further below). The good matrix transparency and excellent mechanical properties of the graphene derivatives and PDMS-enabled conformal contact between the matrix and human skin, with a transmittance exceeding 90% over the range 400–1000 nm, as shown in Figure 1d. The sensing performance of each sensor in the matrix was characterized. Figure 2a plots the sensing properties of the GO-based humidity sensor. The variation in the GO capacitance was collected at a specific input frequency of 10 MHz. As the relative humidity (RH) was increased from 20% to 90%, the capacitance increased from 0.15 to 4.27 pF because the adsorbed water molecules increased the capacitance of GO. Two distinct regimes were present in the curves. Under a low RH environment (RH less than 60%), water molecules were adsorbed onto the GO surface through double hydrogen bonding. In this regime, the hopping of protons between adjacent hydroxyl groups increased the leak conductivity in the film, which enhanced the capacitance of the GO. As the RH was increased above 60%, a larger number of water molecules adsorbed onto the GO surface and penetrated the GO films. These water molecules facilitated the hydrolysis of various functional groups, including carboxyl, epoxy, and hydroxyl groups, on the GO. These ionic species dramatically enhanced the ionic conductivity and sharply increased the capacitance.35, 36 The impedance spectrum obtained from the GO-based humidity sensor could be modeled by an electrical equivalent circuit that included a single charge transfer resistance (Rct) value and two constant phase elements, CPE1 and CPE2, as shown in Figure S2a in the Supporting Information. The impedance of a constant phase element (ZCPE) is defined as ZCPE = Q–1(iω)–n (0 ≤ n ≤ 1), where Q is a real parameter, i is the imaginary unit, ω is the frequency, and n is a real parameter, the value of which can vary from 0 (pure resistor) to 1 (pure capacitor).29 The spectrum fit well to the proposed equivalent circuit across the entire frequency range (Figures S2b and S2c, Supporting Information), and the extracted fitting parameters Rct, n1, and n2 were consistent with those obtained from previous GO humidity sensors. Figure 2b plots the real-time humidity sensing properties of the device at different RHs of 30%, 40%, 50%, 60%, and 90%. The capacitance of GO increased gradually to higher values under each RH condition and then returned to the initial value after the RH was recovered to ambient humidity. The stable signal saturation behavior was monitored at three different RHs (40%, 60%, and 75%) as shown in Figure S3 in the Supporting Information, which is consistent with previous reports related with the GO humidity sensors.37, 38 In addition, the delamination issue of the GO layers from the channel (Figure S4, Supporting Information) should be addressed because the GO was exposed to the outside environment. Figure 2c plots the sensing properties of the rGO-based thermal sensor. A DC voltage of 1 V was applied between the two CVD-Gr electrodes. The resistance of rGO decreased linearly from 0.62 to 0.28 MΩ as the temperature increased from 0 to 100 °C. The temperature dependence of the rGO conductance (G) was examined, and a plot of G versus T–1/3 was fit numerically (Figure S5, Supporting Information). The experimental data fit well over the entire temperature range, suggesting that charge transport in the rGO multilayer films was governed by a 3D variable range hopping (VRH) model supplemented with parallel quantum tunneling.39, 40 In this model, the temperature-dependent conductance of the rGO film could be described as G = Gh·exp(–H/T1/3)+Gt, where H is a hopping parameter, Gh·exp(–H/T1/3) represents the hopping contribution, and Gt represents the quantum tunneling contribution. High temperatures facilitated thermally activated charge hopping among the localized states, which enhanced the electrical conductance of the film.39, 40 Figure 2d shows the real-time measurements of the resistance of the rGO sensor as the temperature was increased and then held at four different temperatures (30, 60, 80, and 100 °C) for several minutes. The output resistance of rGO decreased gradually and was maintained at each value. As the temperature decreased to 0 °C, the rGO resistance returned to its initial value. The sensing properties of the GO and rGO devices depended on the amount of GO or rGO deposited onto the channel (Figure S6, Supporting Information) and should be addressed through further systematic studies. These results suggested that both GO and rGO could be used as excellent sensing materials for humidity and thermal sensors, respectively. The sensing properties of the PDMS-based pressure and strain sensors prepared with the CVD-Gr/PDMS/CVD-Gr structure were characterized. A custom pressure gauge comprising a plastic pole terminated with a square-shaped glass unit (contact area = 1 mm2) was used to apply pressures ranging from 0 to 450 kPa onto the device. PDMS layers with three different thicknesses (3, 5, and 10 μm) were tested, as shown in Figure S7 in the Supporting Information. Although the 10 μm thick PDMS layer exhibited the highest sensitivity, it did not form conformal contact with human skin. The trade-off between the contact properties with human skin and the sensing performance was optimized, and a 5 μm thick PDMS layer was selected. Figure 2e plots the capacitance as a function of the pressures applied to the capacitive PDMS pressure sensor. The curve exhibited two distinct regimes. As the applied pressure dipped below 10 kPa, the capacitance increased from 7.32 to 7.52 pF in a steep slope because the initially high pressure easily deformed the PDMS layer. Under high pressures exceeding 10 kPa, however, the slope decreased due to the reduced deformation space available to the compressed PDMS layer. The sensitivity of the PDMS pressure sensor, defined as (ΔC/C0)/P, was 0.002 kPa−1 (Figure S8, Supporting Information). Figure 2f plots the real-time measured capacitance values as a function of pressure. Application of five different pressures: 2, 4, 20, 100, and 200 kPa, resulted in immediate capacitance responses over a short response time of <0.2 s. The minimum detectable pressure was actually lower than 0.5 kPa, but could not be displayed due to the detection limitation of our pressure gauge. Apart from its pressure sensing capabilities, the PDMS pressure sensor also detected bending strain, as shown in Figure S9 in the Supporting Information. The durability bending test applied over 2000 cycles confirmed that our lamination and degassing processes were useful for preparing matrix-bound capacitive PDMS pressure sensors (Figure S10, Supporting Information). A piezoelectric poly(vinylidene fluoride trifluoroethylene) [P(VDF-TrFE)] layer could be inserted between the two CVD-Gr electrodes in place of the PDMS layer for both pressure and strain sensing (Figure S11, Supporting Information).41, 42 Notably, all sensing materials in the matrix, including GO, rGO, and the sandwiched PDMS, were utilized as obtained. A higher sensitivity may potentially be achieved from the E-skin by further optimizing these materials. Stretchability test was also conducted on the E-skin matrix (Figure S12, Supporting Information). The sensing properties of each sensor were monitored during stretching cycles. The stretching strain of 3% was applied and released repeatedly to the matrix in the direction parallel with the sensing channel. Note that the sensing performances of each sensor were invariant even after 500 cycles. The good stretchability of the E-skin was attributed to the mechanical properties of the graphene derivatives used in the sensor matrix. The integration of all three sensors into a matrix required that each sensor provide output responses to a specific stimulus without displaying sensitivity to other stimuli. The simplex sensing performance of the GO-based humidity sensor was tested under three stimuli: temperature, humidity, and pressure, as shown in Figure 3a. The GO capacitance responded only to the humidity and not to the temperature and pressure. A slight increase in the GO capacitance above 80 °C was observed, but this change was much smaller than the response arising from a humidity change. The resistance of the rGO-based thermal sensor only changed with the temperature (Figure 3b). The rGO resistance decreased linearly with temperature, but no obvious variations were observed under varying pressures. The top PDMS substrate in the sensors matrix was laminated onto the bottom PDMS substrate bearing the rGO-based thermal sensor array; thus, the thermal sensors were passivated by the top 5 μm thick PDMS layer. This lamination process effectively protected the rGO that interacted with external water molecules (humidity), thereby reducing the humidity sensitivity of rGO.43 The pressure detection properties of the PDMS-based pressure sensor under different environmental conditions were tested at different temperatures and RHs (Figure 3c,d). The left panel in Figure 3c shows the real-time pressure sensing measurements conducted at different temperatures. As the temperature increased from 22 to 90 °C, the capacitance decreased from 7.3 to 5.7 pF because the thickness of the PDMS layer increased due to the thermal expansion of the PDMS;44 however, the corresponding pressure sensing properties at each temperature were not significantly affected. As shown in the right panel of Figure 3c, the absolute values of the capacitance (the difference between the capacitance values measured in the presence or absence of the applied pressures) were similar at different temperatures (indicated by the blue arrows). Similar pressure tests were conducted under different humidity conditions (Figure 3d). As RH was increased from 30% to 80%, the capacitance of the PDMS layer increased from 7.3 to 8.4 pF. This increase was attributed to the fact that higher numbers of water molecules were adsorbed onto the PDMS surface at higher RH to increase the capacitance of the PDMS layer.45, 46 The pressure sensing properties were less affected by the RH (blue arrows in the right panel of Figure 3d). Consequently, the pressures applied to the pressure sensor under different environments could be detected by measuring the signal variations. The ability of the sensors to mimic the multifunctionality of human skin was assessed by measuring the output signals of each sensor in our multifunctional sensor matrix induced by external stimuli. The single signal detection capabilities of our impedance analyzer permitted the simultaneous monitoring of two different output signals from the pressure and thermal sensors. The two output signals from the humidity and thermal sensors could also be monitored simultaneously under similar external stimuli. These two groups of measured output signals were analyzed in a single graph. Figure 4a–c plot the output signals collected from the three sensors under three different stimuli (hot wind blowing, finger pressing, and breathing). The hot wind blowing on the matrix dramatically decreased the rGO resistance (blue curve in Figure 4a) but did not induce any variations in the capacitance values of both the GO and PDMS. The finger pressing stimulus resulted in a change in the output signals of all three sensors (Figure 4b). Under certain pressures, the PDMS capacitance decreased (green curve in Figure 4b). This capacitance change opposed the capacitance increase observed with the pressure applied using the insulating plastic pole. The fringing electric field was intercepted and shunted to ground by the finger, which decreased the capacitance. The GO capacitance increased upon introduction of moisture from the finger (red curve in Figure 4b). At the same time, the rGO resistance increased as the channel temperature increased to the human body temperature (blue curve in Figure 4b). The temperature sensing response was relatively slower to the other sensor responses due to the slow heat transfer from the finger. Finally, the responses to human breathing detected by our sensor matrix are plotted in Figure 4c. The GO capacitance values were immediately sensitive to the humidity changes induced by breathing (red curve in Figure 4c). The surrounding temperature also increased slightly, as indicated by the resistance decrease of the rGO sensor (blue curve in Figure 4c). No obvious changes in the pressure were observed. These results indicated that different output signals extracted from the corresponding sensors under external stimuli could be clearly distinguished, which is important for mimicking multifunctional human skin. The proof-of-concept E-skin sensor matrix (6 × 6 array) fabricated using the procedure illustrated in Figure 1a was tested for its ability to capture human action stimulation. The pressure, temperature, and humidity changes were monitored as a finger was pressed at two corners of the matrix (Figure 4d). As shown in Figure 4e, the output sensing signals from different sensors were collected and plotted to obtain the corresponding 2D color maps. The upper black-and-white mapping shown in Figure 4e plotted the calculated sensitivities of the three sensors under finger pressing. The capacitance and resistance sensitivity were defined as ΔC/C0 and ΔR/R0, respectively. The bottom color mapping represented the distributions of the corresponding temperature (blue), humidity (red), and pressure (green) during the finger pressing event, as estimated from the relevant sensitivities plotted in the upper black-and-white mappings. In conclusion, we demonstrated the fabrication of an all-graphene transparent conformable multifunctional E-skin matrix. CVD-Gr was used as the electrodes in the matrix, and GO and rGO were adopted as the sensing materials. A simple lamination process was used to dexterously integrate the humidity, temperature, and pressure sensors into a single unit. Each sensor was sensitive to its relevant external stimulus but was not affected by the other two stimuli. Moreover, all sensors worked simultaneously and reported on different stimulations individually. The distributions of the temperature, humidity, and pressure during finger pressing were represented in 2D color maps. This work suggests a facile fabrication process using a combination of graphene derivatives to produce a transparent conformable E-skin device without using conventional complex E-skin fabrication processes. Additional sensors and wireless communication units may be integrated into this simple lamination process to assist in realizing interactive remote healthcare applications in the future. The use of several graphene derivatives as the main components in the E-skin may also speed up the industrial-scale use of graphene. Materials Preparation: Graphene oxide was prepared from natural graphite (Alfa Aesar, 99.999% purity, –200 mesh) using a modified Hummers method.47 Briefly, 10 g graphite and 230 mL H2SO4 were mixed in a flask. Next, 30 g KMnO4 were added slowly over 1 h. Stirring was continued for 2 h in an ice water bath. After the mixture had been stirred vigorously for 18 h at room temperature, 460 mL deionized water were added, and the solution was stirred for 10 min in an ice water bath. 50 mL of H2O2 (30 wt% aqueous solution) were then added, and the mixture was stirred for 2 h at room temperature. The resulting mixture was precipitated and filtered to obtain the graphite oxide powder. The graphite oxide was then exfoliated into GO nanosheets (100 mg L−1) in deionized water by bath sonication for 1 h. The GO was reduced in solution by exchanging the dispersion medium of the GO nanosheets with N-methyl-2-pyrrolidone to prevent aggregation. Hydrazine monohydrate (N2H4) was then added dropwise to the GO solution to a final concentration of 4 mM, followed by heating at 100 °C for 24 h. Large-area graphene electrodes were grown on a Cu foil (10 × 10 cm2) through CVD. The Cu foil was loaded into a quartz tube and annealed at 1000 °C under an H2 atmosphere at low pressures for 1 h. Then, 5 sccm CH4 was introduced to trigger graphene growth under a continuous H2 (10 sccm) flow. After 30 min, the CH4 flow was ceased and the tube was cooled to room temperature under the H2 flow. The graphene obtained on the Cu foil was transferred onto a PDMS substrate using standard photolithography and etching processes. Device Fabrication: The PDMS solution comprising a base prepolymer and a crosslinking agent (10:1 weight ratio) was poured onto the HMDS-treated glass, and then the sample was put into a chamber to remove bubbles in the PDMS layer by degassing. After degassing, the PDMS layer on the glass was spin-coated at 1500 rpm and then cured at 120 °C for 1 h. The PDMS surface was treated with O2 plasma to form a hydrophilic surface. The CVD-grown graphene was transferred onto the PDMS using a previously reported method19 and then patterned using photolithography and subsequent reactive ion etching (RIE). The sensing channels were formed by spray-coating rGO (100 mg L−1) or GO (1 g L−1) onto the channel through a shadow mask. Both channels were then laminated vertically with the guidance of an aligner mark. The degassing process was conducted in the vacuum chamber over 30 min, and the fabricated matrix was pressed to achieve conformal contact. Device Measurements: The current–voltage (I–V) properties of the thermal sensing devices were measured using Keithley 238 and 6517A semiconductor parameter analyzers with a temperature controller. All impedance properties were measured using a HP4284A high-precision LCR meter with a customized humidity control chamber. The pressure sensing properties were measured using an Agilent 4980A precision LCR meter with a mark-10 force gauge M5-05. The Au pads were deposited onto the terminals of each CVD-graphene electrode for electrical wiring with the conductive wires. All the sensing signals were collected through the conductive wires connected to the source measurement units. D.H.H. and Q.S. contributed equally to this work. This work was supported by a grant from the Center for Advanced Soft Electronics (CASE) under the Global Frontier Research Program (2013M3A6A5073177 and 2014M3A6A5060932) and Basic Science Research Program (2009-0083540) of the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology, Korea. As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
