Enhancing Accuracy of Flexible Piezoresistive Pressure Sensors by Suppressing Seebeck Effect

SUI JingliangDU MinzhiJING YuanyuanYANG XiaonaWAN Xuefen34YANG YiZHANG Kun

1 Key Laboratory of Textile Science & Technology of Ministry of Education, Donghua University, Shanghai 201620, China2 College of Textiles, Donghua University, Shanghai 201620, China3 Hebei Engineering Technology Research Center for IoT Data Acquisition & Processing, North China Institute of Science and Technology, Langfang 065000, China4 College of Computer, North China Institute of Science and Technology, Langfang 065000, China5 College of Information Science and Technology, Donghua University, Shanghai 201620, China

Abstract: Flexible piezoresistive pressure sensors can offer convenient detection of mechanical deformations for wearable electronics. Previous studies of flexible piezoresistive pressure sensors focus on the sensitivity but the low-cost and self-powered sensors remain a challenge due to the deviation of resistance signal acquisition caused by thermoelectric voltage. Here, piezoresistive pressure sensors with ultralow Seebeck coefficient of -0.72 μV/K based on carbon nanotubes(CNTs)/polyethyleneimine(PEI)/melamine(CPM) sponge are reported. Due to the diminished Seebeck effect, the CPM sponge pressure sensors successfully reduce the deviation to 18.75% and can keep stable sensitivity and resistance change under a very low working voltage and change temperature environment. The stable properties of the sensors make them successful to work for real-time sensing in self-powered wearable electronics.

Key words: piezoresistive pressure sensor; carbon nanotube(CNT); Seebeck effect; deviation in resistance measurement; thermoelectric voltage

Introduction

With the booming development of artificial intelligence and Internet of Things(IoT), flexible pressure sensors have attracted a great deal attention of researchers due to the broad application in e-skin[1-2]human-machine interaction[3-4], wearable health monitoring systems[5]and soft robotics[6]. So far, rapid progress of pressure sensors has been made in various types of flexible sensors based on piezoresistive[7], capacitive[8], piezoelectric[9], and triboelectric[10]sensing mechanisms. In particular, piezoresistive pressure sensors are widely used owing to their high sensitivity, simple device structure, low-cost and easy signal acquisition. Under external pressure stimuli, the piezoresistive pressure sensors will deform and transduce external force into a resistance signal. A typical flexible piezoresistive pressure sensor is the sandwich structure, consisting of two electrodes and a conducting elastomer. The sensing ability of a piezoresistive pressure sensor depends on the conducting elastomer. In order to obtain a high-sensitive piezoresistive pressure sensor, high conductivity materials such as carbon nanotubes(CNTs)[11-12], graphene[13], MXene[14], and poly(3, 4-ethylenedioxythiophene): poly(styrenesulfonate)(PEDOT: PSS)[15]are widely adopted.

However, as a result of the Seebeck effect, these semiconductor materials will produce thermoelectric potential under temperature difference, which can cause inaccuracy in the signal acquisition of piezoresistive pressure sensors in low working voltage. In the previous studies, most of researchers focus on the high sensitivity of piezoresistive pressure sensors. But the errors caused by thermoelectric potential is ignored because the working voltage of pressure sensors is relatively high to volt magnitude in their researches. For the next generation wearable electronics, flexible, low-cost, and battery-free are essential to improve their practical value, so the piezoresistive pressure sensor needs other power supply. Flexible thermoelectric generators based on organic thermoelectric materials[16-18]are ideal energy sources for the piezoresistive pressure sensors owing to their wearable properties[19-20]and continuous voltage output. But the output voltage of the flexible thermoelectric devices is usually a few hundred microvolts. For example, Zhangetal.[21]reported a self-powered pressure sensor based on piezoresistive and thermoelectric mechanism, the generated voltages are 173 μV and 345 μV under the temperature differences of 5 K and 10 K, respectively. In this working condition, the thermoelectric potential produced by the pressure sensors is close to the voltage from the thermoelectric devices that cannot be ignored. What is more, the piezoresistive pressure sensors are usually applied on the skin or nearby the skin. The temperature difference between human body and external environment can easily reach 2 K but cannot remain stable because of the changed body temperature and air flow of environment, which results in an erratic thermoelectric potential. The deformation under external force stimuli of the piezoresistive pressure sensors will exacerbate the instability of the temperature difference because of the extra air convection. In this complex temperature condition and low working voltage, the thermoelectric potential produced by the semiconductor materials will cause huge deviations in the resistance signal collection.

Here, an ultralow Seebeck coefficient piezoresistive pressure senor was fabricated. The piezoresistive pressure sensor is a composite of supporting component and conductive component. We chose melamine sponge as the supporting material owing to its highly porous and interconnected structure with good compressibility and resilience. For the conductive component, CNTs with easily adjustable Seebeck coefficient are used. Specifically, we choose single-walled carbon nanotube(SWCNT), and SWCNT is more sensitive to polyethyleneimine(PEI) doping, which has been demonstrated in the previous study[22]. The composite of CNTs and melamine sponge was fabricated by the dip coating method, and then the adjustment of Seebeck coefficient of the CNTs/melamine(CM) sponge was achieved by doping PEI. The Seebeck coefficient of the CNTs/PEI/melamine(CPM) pressure sensor can be decreased to as low as -0.72 μV/K, and it successfully avoids the deviations of the signal acquisition. The main aim of our work is to eliminate the errors of piezoresistive pressure sensors signal collection caused by the thermoelectric effect, which is essential for the low working voltage for piezoresistive pressure sensors used in the next generation wearable electronics.

1 Experiments

1.1 Materials

Melamine sponge was purchased from Sichuan Chaoju Sinoyqx New Material Technology Co., Ltd., Chengdu, China. CNTs were purchased from China Science Times Nanotechnology Co., Ltd., Chengdu, China. Sodium deoxycholate(SDOC) was purchased from Shanghai Yien Chemical Technology Co., Ltd., Shanghai, China. Branched PEI and ethanol were all purchased from Aldrich. And silver paste was purchased from Jingte Technology Company, Dongguan, China. All the chemicals were used without further purification.

1.2 CNTs dispersion preparation

CNTs were dispersed in the deionized(DI) water using SDOC as the surfactant. For each experiment, the CNTs were used with the concentration of 2.5 mg·mL-1and the weight ratio of CNTs to SDOC is 3:10. The dispersions were sonicated for 60 min using a high-power probe ultrasonication at 30 W. Afterward, the mixture was centrifuged at 3 000 g for 30 min.

1.3 CM sponge preparation

The melamine sponge was cut into cuboid-shaped pieces with 10 mm in length, 10 mm in width, and 5 mm in thickness. And the melamine sponge was washed in DI water and dried. Then the cuboid-shaped melamine sponge was dipped in the CNTs dispersion for 20 min. After the excessive solution was squeezed out of the sponge, the sponge was dried in a vacuum oven at 60 ℃ for 1 h. In order to get continuous coating inside the sponge, the dip coating process needs to be repeated for 10 times.

1.4 PEI doping

Firstly, PEI was dissolved in ethanol at different concentrations(0.05%, 0.10%, 1%, and 10% by weight). CM sponge was immersed in the PEI solution for 30, 60, 90, 120 s, and then the sponge was squeezed to remove excessive solution. Finally, the CPM sponge was dried at 70 ℃ for 30 min in a vacuum oven.

1.5 Fabrication of pressure sensors

The pressure sensors were assembled by attaching two copper sheets to both upper and lower surfaces of the CPM sponge, and silver paste was brushed onto the interfaces for good electrical contact. Then copper wires were welded to copper sheets for electrical connection between the sandwich structure sensors and testing devices.

1.6 Measurement of pressure sensors

The Seebeck voltage in this study was measured by the Keithley 2182A nanovoltmeter, Shanghai, China. The temperature was detected by an infrared thermoscope. The average Seebeck coefficient was extracted from a straight line fit of the measured slope.

Pressure was applied by an electromechanical testing machine(ZQ-990A, Dongguan, China) controlled by a computer. The current and resistance were recorded by a Keithley 2400 SourceMeter(Shanghai, China) controlled by a computer through the LabVIEW program.

To mimic the limited output voltage of the flexible thermoelectric devices, we set the output voltage of Keithley 2400 SourceMeter in the range of 200-300 μV.

2 Results and Discussion

2.1 Seebeck coefficient influenced by PEI doping

The CNTs showp-type properties due to the oxygen doping when they are exposed in air. The positive Seebeck coefficient of the CM sponge is at 13.12 μV/K. The Seebeck coefficient of CNTs is very sensitive to doping and both the concentration of the PEI and immersion time remarkably influence the doping effect. As shown in Fig. 1, the positive Seebeck coefficient of the CM sponge rapidly decreases to a low level after being treated for 30 s. In the doping process, the amine-rich PEI molecules coated on the surface of CNTs, which act as electron donors. The major carriers convert from holes to electrons and the Seebeck coefficient switches from positive to negative during the PEI doping process. With the immersion time increasing, the negative Seebeck value increases and reaches the steady state after 90 s.

Fig. 1 Seebeck coefficient of the CM sponge after PEI doping for 30, 60, 90 and 120 s, respectively

The doping effect of the PEI concentration shows similar trend to the immersion time. From Fig. 2, we can see that the Seebeck coefficient switches from positive to negative(-0.72 μV/K) when the concentration of PEI increases to 0.05% by weight. As the concentration of PEI increases further, the negative Seebeck coefficient increases slowly and approaches saturation when the PEI concentration increased to 0.10% by weight. It seems that the thermoelectric properties of the CM sponge are more sensitive in the range of lower PEI concentrations. The network structure of sponge contributes to the PEI doping with lower concentrations because the 3D network structure coated with CNTs can provide physical absorption positions for PEI molecules compared to 1D or 2D materials.

2.2 Reduction of resistance deviations caused by Seebeck effect

The resistance change caused by the temperature variation is another source of error in the signal acquisition process for piezoresistive pressure sensors. As shown in Fig. 3, the temperature coefficient of resistance(TCR) of the CM sponge is negative(0.003 4 K-1). The slight decrease of resistance with the increase of temperature can be ignored because the temperature variation cannot reach such high level in real-time sensing.

The temperature condition of sensors is complex, and it may rapidly change due to the air convection induced by external pressure stimuli and deformation of sensors. Besides, the air flow of environment and the unstable contact between sensors and skin may influence the temperature difference as well. To simulate the actual working conditions, we set a Peltier tile under the pressure sensor to provide continuously changed temperature difference with 4.5 K/min in heating or cooling. In the CM sponge pressure sensor withp-type thermoelectric properties, the holes move under temperature difference and assemble on the cold side of the sponge. Therefore, the CM sponge sensor generates a thermoelectric voltage shown in Fig. 4, and the sensor was connected in series with the Keithley 2400 SourceMeter. Decreasing of the detected resistance would be observed when the external current was applied in the circuit.

This is the reason that the blue dot line shifts downward apparently in Fig. 5(a), and it is independent of the TCR because they have opposite effect to the resistance change. When the bottom of CM sponge pressure sensor is set on the hot side, the direction of the generated voltage is opposite when it is placed on the cold side, so the red dot line shifts up as shown in Fig. 5(a). For the thermoelectric voltage ΔV=S× ΔT, where, ΔVrepresents thermoelectric voltage,Srepresents the Seebeck coefficient, and ΔTrepresents the temperature difference. With the larger temperature difference, this shift should be more obvious while we observe the contrary phenomenon. The intervals of the three lines are equal at the beginning of the curve, and the three lines are gradually getting closer at the end of the lines. This is attributed to the compression of the sponge. The highly porous structure of sponge provides a very low thermal conductivity that helps to maintain a constant temperature gradient, so the sponge generates stable thermoelectric voltage. With the increase of the pressure, the sponge is compressed resulting in the decrease of the temperature difference, which weakens the Seebeck effect. For the CPM sponge with ultralow Seebeck value, the resistance shifts vanish and the three lines almost coincide in Fig. 5(b). This is because the Seebeck of the CPM sponge pressure sensor is reduced and the deviation of the resistance signal acquisition is reduced. It also indicates that the TCR is negative in the pressure sensing process. In Fig. 5, the initial resistances of CM sponge sensors and CPM sponge sensors are different. This is because the electrical conductivity of CPM sponge sensors decreases from 1.41 × 10-3S/cm to 2.08 ×10-4S/cm after being doped by PEI.

Another important indicator of piezoresistive pressure sensor is the sensitivityS= δ(ΔR/R0)/δP,where,Srepresents the sensitivity, ΔRrepresents the resistance change,R0represents the initial resistance andPrepresents the applied pressure. The thermoelectric voltage produced by sensors also causes errors to the sensitivity. In Fig. 6 the resistance decreases rapidly with the increase of the pressure in low pressure region. This negative piezoresistive effect is attributed to the increase of the contact area inside the sponge and forms more conductive paths. With the further increase of pressure, this effect becomes weaker due to the reduction of inner space of the sponge. In Fig. 6(a), the sensitivityS1of CM sponge sensors without temperature difference is 0.32 kPa-1in low pressure region. When the sensors are set on the cold side of the Peltier tile, the sensitivityS2shifts upward from the origin line and reach to 0.36 kPa-1. When the sensors are set on the hot side of the Peltier tile, the sensitivityS3is changed to 0.26 kPa-1, which shows a deviation of 18.75%. The deviations are resulted from the errors in measured resistance induced by Seebeck effect. The inaccuracy of the sensitivity will cause unreliability of the data in real-time sensing. For the CPM sponge sensor shown in Fig. 6(b), the sensitivityS4keeps stable in changeable working environment.

3 Conclusions

In conclusion, piezoresistive pressure sensors with ultralow Seebeck coefficient are fabricated by a facial dip coating approach and subsequent PEI doping treatment. The CPM sponge pressure sensors reduce the deviation of the resistance signal acquisition and enhance the accuracy of the sensing in low voltage and complicated temperature environment. For real-time sensing, the CPM sponge sensors can retain high precision and sensitivity in self-powered systems with very low voltage. The stability and low cost of sensing properties suggest that the CPM sponge sensors have good potential in the next generation wearable electronics.