Flexible CNT-array double helices Strain Sensor with high stretchability for Motion Capture

Motion capture is attracting more and more attention due to its potential wide applications in various fields. However, traditional methods for motion capture still have weakness such as high cost and space consuming. Based on these considerations, a flexible, highly stretchable strain sensor with high gauge factor for motion capture is fabricated with carbon nanotube (CNT) array double helices as the main building block. Ascribed to the unique flexible double helical CNT-array matrix, the strain sensor is able to measure strain up to 410%, with low hysteresis. Moreover, a demonstration of using this strain sensor for capture hand motion and to control a mechanical hand in real time is also achieved. A model based on finite difference method is also made to help understand the mechanism of the strain sensors. Our work demonstrates that strain sensors can measure very large strain while maintaining high sensitivity, and the motion capture based on this strain sensor is expected to be less expensive, more convenient and accessible.


Measurement range test of the strain sensor
. Measurement range test of the CNT array double helices strain sensor. Expansion process is shown with red curve. The strain sensor start to lose conductivity at around 330% strain, as resistance increases sharply. Blue dots indicates the shrinkage process. Even at this condition the strain sensor is able to reverse to its initial state, which shows decent stability.   To understand the mechanism of our sensors more clearly, a model based on finite difference method has been made for the sensors. The CNTADH film can be treated as randomly distributed clusters of CNTADH bundle, simulated by randomly placed conductive circular flake. To simulate the film crack, a random net is created, and the flakes move along the nearest net. Using a constant potential, resistance is calculated by R=V/I, where R is resistance, V is potential, I is current. Based on Ohm's Law:

Different samples and their test results
Where J is current density, σ is electrical conductivity, E is electrical field, resistance can be calculated by finite difference method. The conductivity map at different strain is generated, shown in Figure S5a. Conductivity is proportional to the number of layers of CNT cluster. When strain is applied, the overlap area decreases and crack forms, hence the resistance increases. Figure S5b shows the potential map of the sensor, as strain increases, the electrical field decreases, which lead to smaller current and larger resistance. Simulated results and the experiment results are shown in Figure S5c.
Other side effects may cause the inflection point in the experiment results. One possible mechanism is the reposition of the CNTADH clusters. This is caused by the nonuniform deformation of the sensor caused by the set up. In an ideal situation, the strain sensor is stretched uniformly. However in the test set up, a clap structure is stick to the side of the strain sensor. When stretching, tension is applied to the side of the strain sensor, causing a necking effect ( Figure S6a). This effect may lead to the result that the resistance of real device is larger than the ideal situation. A simple example is used to show this effect ( Figure S6b). The vertical cross section of effective part of the strain sensor is shown, although real condition is much more complex, this example can be used to explain why the necking of the sensor can cause the reposition of CNTADH and hence lead to resistance change from ideal condition. Assume the initial state is shown in the left in Figure S6b. When stretching, a simple model is shown in the right in Figure S6b. It is worth mentioning cross section of real device has a more complex shape. Since the thickness of the device is smaller than the width, necking in horizontal can be neglected. The area of the two cross section is the same: The CNTADH film is on top of the device, assume its initial thickness is t, width is w and length is l. The initial resistance of the device is: ρ ρ is the resistivity of thin film.
Resistance of the stretched state can also be calculated: This is 1.162 times of ideal resistance. This factor is changing while strain increases, which infers that there is a correction function need to be multiplied to the original simulation result. To analyze why this effect can increase resistance, the sensor can be divided into 3 parts: transition part, stretched part, transition part. The ratio of the length of transition part and the total length will decide how much the resistance increases. As the ratio become smaller, resistance increases less, this is because the sensor will have less difference with the original state. To show the basic trends of this mechanism, here we only analyze it qualitatively. When no strain is applied, ideal situation is the same as real situation. When strain is applied ( Figure S6c middle), transition part, which is connecting the edge of the sensor and stretched part, is relatively large, comparing to the length of the whole device. When strain is very large (Figure S6c right), tension of tpe increase sharply, which will cause the stretched part less likely to be stretched.
According to this, transition part become smaller. Thus the overall resistance tend to be similar to the simulation result. Relative length of the transition part determines the difference of real condition and simulation result. So when strain increase, correction function initially increases from 1, and then decreases.
Although the actual correction function cannot be obtained based simple analyze, it is useful to assume some basic function which is in agreement with the previous assumption as the correction function to show how this mechanism might influence the final result ( Figure S7). These analyze still have limitations since they can only qualitatively explain the experiment results, further simulation may be done to improve the results.
ΔR/R versus strain of sensors with different thickness of CNTADH film are also simulated, shown in Figure S5d. It indicates that with thinner film, slope of the curve increases and the gauge factor is larger, but the measurement range will decrease.
Results show if the thickness of the sensors is 0.42 times of current sensor, the electrical connection completely breaks with strain less than 300%. Simulation also shows thicker CNTADH film has a smaller slope and smaller gauge factor. Since the thickness of the CNTADH film can be modulated by controlling the amount of CNTADH used when forming film, this shows using this method strain sensors with different gauge factor and different measurement range can also be made.  Table S1. Comparison of our device and some other CNT based strain sensor.

Comparison with recent CNT based strain sensors
Properties of our device is compared with other CNT based devices, marked as 1 1 , 2 2 , 3 3 , 4 4 and 5 5 .