Multiscale and hierarchical wrinkle enhanced graphene/Ecoflex sensors integrated with human-machine interfaces and cloud-platform

Current state-of-the-art stretchable/flexible sensors have received stringent demands on sensitivity, flexibility, linearity, and wide-range measurement capability. Herein, we report a methodology of strain sensors based on graphene/Ecoflex composites by modulating multiscale/hierarchical wrinkles on flexible substrates. The sensor shows an ultra-high sensitivity with a gauge factor of 1078.1, a stretchability of 650%, a response time of ~140 ms, and a superior cycling durability. It can detect wide-range physiological signals including vigorous body motions, pulse monitoring and speech recognition, and be used for monitoring of human respirations in real-time using a cloud platform, showing a great potential for the healthcare internet of things. Complex gestures/sign languages can be precisely detected. Human-machine interface is demonstrated by using a sensor-integrated glove to remotely control an external manipulator to remotely defuse a bomb. This study provides strategies for real-time/long-range medical diagnosis and remote assistance to perform dangerous tasks in industry and military fields.


Supplementary Note 2. A comparative study of graphene/Ecoflex Sensors and CNT/Ecoflex Sensors
The conductive properties of materials and a good matching between the conductive materials and substrate is very important. In order to investigate the effects of different conductive materials on the performance of sensor, we have used CNT and graphene (both are high performance nano-carbon based materials) as the conductive materials for strain sensors for comparisons. We kept the same experimental parameters for the other preparation processes.

Supplementary Note 3. Hansen solubility parameters of various materials in this study and their effects
Generally, for two substances to be mixed/interacted effectively in a molecular scale, their Hansen solubility coefficient parameters must be close to each other. 1 The Hansen parameters of graphene (δD= 18 MPa 1/2 , δP= 9.3 MPa 1/2 , δH= 7.7 MPa 1/2 ) are much closer to those of NMP (δD=18 MPa 1/2 , δP=12.3 MPa 1/2 , δH=7.2 MPa 1/2 ), if compared with those of Ecoflex (δD≈13 MPa 1/2 , δP=0 MPa 1/2 , δH=0 MPa 1/2 ). Therefore, without the water involved, it is more energetically favorable for the graphene to be attached with NMP rather than the Ecoflex. In that case, it is unlikely that the graphene is effectively attached onto the surface of the Ecoflex. 2 To address this problem, we made it less energetically favorable for the graphene to remain dispersed uniformly in the NMP by adding water into the graphene-NMP dispersion. The  are 1 mg/ml, 2 mg/ml, 3 mg/ml, 4 mg/ml and 5 mg/ml, and when the ratio of NMP/water is greater than 1:4, the initial resistance of the composites decreases with the increase of the proportion of water in the graphene dispersion, which shows that adding water can effectively transfer graphene nanoflakes from NMP onto Ecoflex. When the ratio of water in the graphene dispersion is further increased to ¼ of NMP/water, the resistance of the composites shows the smallest value. With the further increase of the water contents, the resistance of the composite becomes increasing. The possible reason is that large content of water causes the decrease of the solubility of graphene in the dispersion, and the graphene becomes agglomerated, thus affecting the adsorption of graphene by the Ecoflex. Therefore, the optimum proportion of NMP/water to achieve the best conductivity of the sensor is ¼ in this study. Moreover, when the concentration of graphene is 4 mg/ml and the ratio of NMP/water is ¼, the resistance of the sensor shows the smallest value. In addition, the resistance of the composites is quite high, but its strain detection range is low. The reason is that there are not many conductive paths.
When the composites are loaded with a large strain, the conductive paths are easily broken, so the detection range is small. From Supplementary Figure 3b~f, we can see that when the ratios of NMP/water in the graphene dispersion are 1:3, 1:4 and 1:5, and the strain detection range of the prepared composite can reach up to 650%, indicating that water in graphene dispersion plays an important role in improving the sensing performance of the prepared composite. In addition, the composites with the graphene concentration of 1-4 mg/ml and the ratio of NMP/water of 1:4 has shown much higher sensitivities than those of the other ratios of NMP/water but the same graphene concentration value ranges. When the graphene concentration is 4 mg/ml and the NMP/water ratio is 1:4, the sensitivity of the composite reaches its highest value. In summary, the optimal ratio of NMP/water in graphene dispersion is 1:4 and the optimal concentration of graphene is 4 mg/ml in this study. The intensity ratio of D peak to G peak is usually used as an important parameter to characterize the defect density in the graphene. The ID/IG in Raman spectra follows the following equation: where rS and rA are the radii of the 'structurally disordered' area and the 'activated' area surrounding the defect, respectively. 6 The factor CA is defined by the electron-phonon matrix elements. Supplementary Reference. [7] reported that, if the wavelength of the excitation laser is in the visible light range, the mean distance between two adjacent defects (LD, nm) in the graphene can be calculated using the following equation: For the excitation laser with a wavelength of 532 nm (e.g., an energy of 2.33 eV), the mean distance between two adjacent defects (LD, nm) in the graphene can be calculated from: The defect density (nD) can be calculated using D (cm −2 ) = 10 14 /(π 2 ), therefore, composite thin film. 10 The sensor with these microsize cracks showed an ultrahigh sensitivity (with a maximum value of 11344) but a limited stretchability (ε≤50%). Chen et al. also developed a strain sensor based a microcrack mechanism using an acid-interface engineering method. 11 The sensor showed a high sensitivity, but its detection range was very limited.
Kim et al. reported a strain sensor made of composite film with a densely packed microprism-array architecture. This sensor simultaneously achieved a good sensitivity (gauge factor≈81 at >130% strain) and a large stretchability (150%), as well as a long-term reliability (10000 cycles at a strain of 150%). 12 This densely packed microprism-array architecture leads to significantly morphological changes in the metal nanowires percolation network upon stretching. However, the equipment for the preparation of micropatterned silicon master design is relatively expensive, and the preparation process is complicated.
Different from all those previously reported methods of flexible strain sensors, we proposed a novel methodology by modulating multiscale and hierarchical wrinkles on the surface of a flexible substrate to be integrated into graphene/Ecoflex composite strain sensors.
The process mainly involves solution treatment method, which does not involve any expensive equipment and/or complicated preparation processes. Besides, by precisely controlling the solution concentration and treatment durations, high performance samples can be obtained (e.g., a large stretchability of up to 650% strain and a GF of up to 1078.1), with good process repeatability. Although the sensitivity of our sensor is slightly less than those based on microcrack mechanism, our sensor's detection range is much wider, which indicates that they can be used for a wide-range applications of both large and small strain scenario.
Of course, our sensor and its process still have some limitations. Firstly, compared with those based on a single step treatment, our two-step treatment method could increase the time of preparation process. In addition, our method is not IC compatible. Finally, the strain sensor shows obvious overshoots during its usage, which is due to the tensile stress-relaxation behavior under the applied strains.

Supplementary Note 13. Graphene/Ecoflex composite strain sensor applied for human motion detection
The wearable sensors were conformally attached to different positions of a person's body