Self-powered 2D nanofluidic graphene pressure sensor with Serosa-Mimetic structure

Abstract

High-performance self-powered pressure sensors have attracted much attention due to their potential applications in bionic limbs, healthy motion detection, medical devices, and other fields. However, existing devices are either imperceptible to the pressure direction or require an external power source as a driving force. Here, a pressure sensor inspired by the structure of serosal membranes is reported, which contains a novel partially reduced graphene oxide (prGO) membrane. In this Serosa-Mimetic structured membrane, a narrow “Trail” structure containing COOH is considered as an ion filter and ion conductor, which facilitates the directional migration of cations under external stimuli, resulting in a directional flow of net charges, resulting in current (or voltage) signal. Therefore, the prGO-Trail membrane can be used as an ion transport layer to facilitate self-powered pressure sensing. This pressure-driven output voltage (and current), produced by ion selectivity, is linear with the applied pressure. The fabricated self-powered bionic pressure sensor has good performance, the optimized response sensitivity is 0.282 nA Pa⁻¹, the response/recovery time is 90/110 ms, and long-term stability (1000 cycles), which provides a meaningful design idea and a larger open field of vision for the next generation of self-driving bionic pressure sensors.

INTRODUCTION

As a common sensor type, pressure sensor is able to convert the recognized information into a specific signal output under a specific pressure. According to the different working mechanisms, most pressure sensors can be divided into piezoresistive sensors,^1,2 piezoelectric sensors^3,4 and capacitive sensors.⁵ However, these devices usually require an additional external power source as a driving force, and the output signal often exhibits a nonlinear relationship with the measured pressure, thus greatly limiting their applications in future bioelectronic devices: such as bionic limbs,⁶ health and motion monitoring,⁷ medical equipment,⁸ and so forth. And with the massive consumption of non-renewable resources and the growing demand for carbon neutrality, the development of a self-powered pressure sensor that uses clean energy to supply energy is the future trend of bioelectronic devices: such as pulse detection, speech recognition and load weighting.^9–11 In addition, the key problem of self-powered pressure sensors is to realize the conversion between mechanical energy and electrical energy, but it is often plagued by the nonlinear transfer function problem between these two energy forms.¹²

Electrokinetic energy conversion (EKEC) technology is a proven solution.¹³ Permselective membranes are the core component of EKEC technology, which produces a continuous direct-current electrical signal instead of a pulsed output when electrolyte flows through the membrane under external pressure. Specifically, owing to the electrostatic repulsion behavior of surface charge on the permselective membranes, only counter-ion (relative to the surface charge) can move through the permselective membranes, resulting in directional motion of net charge that promotes the generation of electrical signal under pressure or salinity difference. Small integrated electronic devices made of EKEC sensors can be self-powered and output controllable sensing signals, which can meet the needs of practical applications and greatly reduce our dependence on non-renewable fossil energy. In general, in nanofluidic channels with nano- and sub-nanometer confinement, the transport of ions tends to occur within the Debye length of finite thickness. Therefore, when reducing the nanofluidic channel radius to the Debye length, the ion transport of confined nanochannels is controlled by the surface charge. From the Helmholtz-Smoluchowski equation,¹⁴ it can be known that the electrical energy generated by the ion-selectivity under pressure can achieve a substantially linear relationship. An ideal permselective membrane should have high permeability, high selectivity and sufficient mechanical strength. Recently, ultrafast permeability and highly selective ion transport have been observed in two-dimensional (2D) nanofluids, which opens up new avenues for EKEC technologies because of their great potential in overcoming the bottlenecks of conventional membranes. Currently, typical 2D materials, such as graphene oxide,¹⁵ MXenes,^16,17 MoS₂,¹⁸ and boron nitride,¹⁴ have been used to prepare permselective membranes.^19,20 The interlayer spacing existing between the stacked 2D nanosheets acts as a nanofluidic channel, reducing the transport resistance and enabling the rapid selective permeation of ions. However, there are some inherent drawbacks in the practical application of 2D material-based films, including longer ion diffusion distance, lower ion selectivity and poor durability in water.²¹ Therefore, developing EKEC pressure sensors based on high-performance permselective membranes remains a challenge.

Over billions of years of evolution, nature has developed countless perfect systems. Learning from nature is the eternal theme of human survival and sustainable development. The structure of the serosa is widespread in organisms, and the selective permeation and transport of ions in the protein ion channels on the serosa are the basis for the action of many living species.²² Through the rapid and direction-selective ion transport of membrane proteins, organisms can respond to a range of stimuli including pressure, heat, odor, light and sound. The serous membrane structure is highly selective, for example Na⁺ and K⁺ can be selectively transported across the cell membrane, resulting in a transmembrane potential.^23,24 Inspired by this high-efficiency ion transport dynamic system, a variety of energy systems have been designed to extract energy from ion transport as new energy sources, such as artificial electric organs,²⁵ nanofluidic devices,^26–28 and hydroelectric generators.²⁹ This provides the basis for a serous membrane-like pressure sensor based on EKEC technology.

Inspired by this unique biological nanostructure, we report for the first time a biomimetic membrane composed of partially reduced graphene oxide and artificially regulated ion channels for self-powered pressure sensing. It has been found that layered membranes based on 2D materials face swelling problems in water, so improving the stability of 2D membranes in aqueous solutions is a current challenge.²⁶ In order to solve this problem, we obtained partially reduced graphene oxide (prGO) by controlling the degree of hydrothermal reaction. Due to the reduction of oxygen-containing functional groups on prGO nanosheets, the hydrophilicity is reduced, so the swelling rate of the membrane will be lower than that of general graphite oxide. At the same time, since the prGO membrane still retains a large number of carboxyl functional groups and a large number of negative charges on the surface, it can selectively recognize cations. Jiang et al. utilized multilayer graphene hydrogel membranes as nanofluidic generators to convert hydraulic motion into streaming ionic currents. And further predict the upsurge of functional 2D nanofluidic devices.³⁰ However, as a typical 2D material, laminated graphene layers tend to aggregate and self-overlap to form compact structures due to the strong van der Waals interactions between planes. This inevitably leads to inefficient ion transport within the interlayer nanofluidic channels. To obtain interlayer networks with open or porous structures capable of reducing ion transport resistance, the sacrificial template method is an effective method. Gogotsi et al. utilized polymethyl methacrylate (PMMA) microspheres with diameters of 2–3 μm as templates to fabricate macroporous MXene anodes for sodium ion storage, enabling high-rate performance.³¹ Therefore, we employed a sacrificial template method to introduce tetragonal needle-like zinc oxide (T-ZnO) between 2D layered prGO membranes, and then acid-etched T-ZnO to fabricate artificially tuned ion channels to increase ion flux. In this Serosa-Mimetic structured membrane, the narrow Trail structure containing COOH is considered as an ion filter and ion conductor, which can facilitate the transport of permeable ions with high-cation selectivity. We use this prGO-Trail membrane with artificially regulated high-flux ion channels as an ion transport layer, mimicking part of the function of the serosa membrane, allowing the directional migration of cations through external stimuli, resulting in the directional flow of net charges, thereby generating current (or voltage) (Figure S1). Based on the aforementioned prGO-Trail membrane as an ion-selective membrane, we assembled it into a self-powered pressure sensor. The pressure sensor is manufactured with a silicone ring as a carrier and a carbon paste film as an electrode. Encapsulating the device with two elastic rubber films allows for tighter contact between the ion transport layer, carrier, and electrodes, resulting in a more accurate signal without noise and scattering. The sensor performs well and can be used to monitor low frequency vibration signals, acoustic signals, micropulse stimulation and non-contact vibration. At the same time, we also study the pressure adaptive mechanism of the sensor, which provides meaningful design ideas and a larger open field of vision for the next generation of self-powered biomimetic pressure sensors.

RESULTS AND DISCUSSION Design of Serosa-Mimetic 2D nanofluidic pressure sensor

The design of Serosa-Mimetic 2D nanofluidic pressure sensor based on EKEC technology is shown in Figure 1. And Figure 1A shows the plasma membrane structure. Phospholipid bilayers form the basic structure of the plasma membrane and provide mosaic sites for integral protein. Under external stimuli, cations can shuttle within ion-selective channels. Therefore, the serosa membrane structure enables high-ion selective permeation and fast ion transport to generate transmembrane potential. This potential difference provided by ion selectivity provides inspiration for our sensor. The cation-selective prGO-Trail membrane is the core component of our sensing device (Figure 1B). The prGO-Trail membrane itself is a large-scale nanochannel integration with negative charges, and due to the electrostatic repulsion on the surface of the nanochannel, it exhibits good cation selectivity in the sub-nanometer confinement. At this point, intermolecular forces, namely short-range steric interactions and hydration forces, as well as long-range van der Waals forces and electrostatic interactions, begin to dominate.^32–35 Ions (or hydrated ions) in an electrolyte solution under pressure stimulation selectively pass through an ion-selective membrane (depending on the surface charge) as a nanofluid, resulting in a directional movement of a net charge that facilitates the generation of flowing voltages and currents (Figure 1C and Figure S1). To prevent the redox reaction of the electrodes themselves from becoming a signal interfering factor, we used carbon paste as the active electrode at both ends. The electrolyte is 1.0 M NaCl solution, in which Na⁺ acts as a transmembrane transport ion to generate transmembrane potential. The prGO nanosheets mimic the plasma membrane structure with a large number of “Trail” channels (vertical and parallel transport pathway) and etched T-ZnO leaving active oxygen functional groups (COOH, OH),³⁶ thus allowing fast ion transport and high cation permeability (Figure 1C).

FIGURE 1. Schematic diagram of plasma membrane structure and Serosa-Mimetic 2D nanofluidic sensor. (A) Schematic illustrations of the plasma membrane structure. (B) The prGO-Trail membrane and its SEM cross-sectional photo. (C) Detailed schematic illustration of 2D nanofluidic sensor response mechanism. (D) Brief schematic illustration of the bionic 2D nanofluidic sensor response mechanism and corresponding response/recovery signal.

It is worth noting that our prepared serosal-like pressure sensor produces a phenomenon similar to biological adaptation.^37,38 When there is no external mechanical stimulation, due to the same concentration of hydrated sodium ions on both sides of the electrolyte solution, no electrical signal is output at this time, corresponding to the blue area of the current curve (Figure 1D). However, when a constant external stimulus was applied to the self-powered sensor, the sodium ions on the pressure side diffused rapidly to the opposite side and produced an electrical signal. After reaching a certain concentration difference, the sodium ion concentrations in the solutions on both sides tended to equilibrium again, and the electrical signal slowly decayed to zero, corresponding to the red area of the current curve (Figure 1D). Through the EKEC mechanism, we can successfully encode external mechanical stimuli into current/voltage signals without relying on an external power source (the enlarged schematic in Figure 1D details this process).

Figure S2, S3 and the experimental section show the fabrication process of the prGO-Trail membrane. By controlling the degree of reduction reaction, the oxygen-containing functional groups on GO nanosheets can be partially reduced, and the obtained prGO nanosheets still have a large number of carboxyl functional groups. The average size of the prepared prGO nanosheets was 3.5 μm (Figure S2A and 2B). X-ray photoelectron spectroscopy (XPS) analysis revealed that abundant surface functional groups, including hydroxyl (OH) and carboxyl (COOH), remained after the partial reduction of GO (Figure S4). By controlling the temperature of the hydrothermal reaction, the C/O atomic ratio in prGO can be adjusted to achieve the purpose of artificially adjusting oxygen-containing functional groups in ion channels (Table S1). We finally chose to partially reduce GO by hydrothermal method at 120°C as the prGO we used. T-ZnO was prepared by chemical vapor deposition (CVD) method using zinc powder as raw material (Figure S2C and 2D). The obtained prGO and T-ZnO were dispersed in water to form a stable mixed suspension and reconstituted by vacuum-assisted filtration to form a black paper-like flexible membrane with metallic color (Figure S3). The removal of T-ZnO using a sacrificial template method resulted in a self-supporting prGO-Trail membrane with a typical layered microstructure (Figure S3B). The cross-sectional scanning electron microscope (SEM) image is shown in Figure S3C. From Figure S5, it can be observed that the needle-like T-ZnO and prGO formed a composite membrane. The SEM images clearly showed that T-ZnO caused obvious wrinkles in the composite membrane. When acid is used to etch away the T-ZnO, these folds form numerous “Trails”, further increasing the ion permeability. As shown in Figure S6, this is because the use of T-ZnO as an in-situ sacrificial template increases the interlayer open structure of the graphene membrane, effectively increasing the accessibility of the electrolyte solution, thereby further increasing the cation flux. The hydrophilicity of the prepared self-supporting membrane is lower than that of the general GO membrane, and the contact angle is about 65.2° (Figure S7).

Charge-governed ionic transport

In order to further verify the working mechanism of prGO-Trail membrane based on ion-selective permeation under EKEC technology, it is necessary to understand the inherent ion transport properties of prGO-Trail membrane. We first studied the transmembrane current–voltage (I-V) response of the layered membrane under different NaCl concentrations, which provides basic information about the surface charge of the prGO-Trail nanofluidic channel. Unless otherwise specified, all ion transport experiments are performed at a membrane thickness of 5.0 μm. The current passing through the prGO-Trail membrane was measured with a pair of Ag/AgCl electrodes, as shown in Figure 2A (see the experimental section for details).

FIGURE 2. Charge-governed ion transport. (A) Ionic transport measurement through prGO-Trail lamellar membrane. (B) I − V characteristics of prGO-Trail membrane for varying NaCl concentration of 0.01–100 mM. (C) Conductance versus salt concentration for the prGO-Trail membrane. (D) I–V curves of composite membrane in 1000-fold NaCl concentration gradient under diffusion. The low concentration side is set to 0.1 mM.

Figure 2B shows a series of I-V curves recorded in NaCl electrolytes with concentrations ranging from 10⁻⁵ M to 10⁻¹ M. They all show linear ohmic behavior, and the ionic current rectification phenomenon is negligible, which can be attributed to the symmetrical structure of the composite system.³⁹ Conductivity measurements show that ion transport through the prGO-Trail membrane is controlled by charge. As shown in Figure 2C, the measured ion conductance has two distinct characteristic behaviors. Specifically, the ion conductance follows the body law in the high-concentration area, showing a linear relationship.

However, when the salt concentration is lower than 10⁻³ M, the ion conductance gradually becomes a plateau, showing typical surface charge-controlled ion transport characteristics. At low concentrations, ion conductance becomes independent of ion concentration, and is mainly controlled by ion transport controlled by surface charge, where the thickness of the electric double layer increases close to the Debye shielding length and is comparable to the size of the nanochannel.⁴⁰ Since the surface of the prGO nanosheets is negatively charged, oppositely charged cations (i.e., Na⁺) will be enriched in the channel, while the same charged anions (i.e., Cl⁻) will be excluded, forming a restricted unipolar environment. In this case, the ion concentration in the channel is determined by the surface charge on the prGO, not by the bulk concentration. The increase in the ion carrier concentration will help improve the conductivity.

Secondly, the use of NaCl concentration gradient system to apply a chemical potential gradient on the prGO-Trail membrane can further understand the transport process of restricted ions. Since the negatively charged composite membrane is cation selective, it can preferentially transfer cations (i.e., Na⁺) from the high-concentration side to the low-concentration side, thereby generating a diffusion current (I_diff) and a diffusion potential (V_diff).⁴¹ By collecting the I-V curve in the presence of a transmembrane concentration gradient, the value of V_diff and I_diff can be directly calculated according to the measured open-circuit voltage (V_OC) and short-circuit current (I_SC). Figure 2D shows a typical I-V curve of prGO-Trail membrane recorded under a concentration gradient of 10⁻⁴ M/10⁻¹ M NaCl. The absolute values of I_SC and V_OC are about 7.98 μA and 150 mV, respectively. According to equation1, the ion migration number of the nanochannel in the NaCl solution is calculated,⁴² and t₊ approaching 0.78 of Na⁺ can be obtained, which indicates the difference in the diffusion flux of anion and cation.[Image Omitted. See PDF]

where t₊ is the cation transference number; E_diff refers to the diffusion potential; R, T, z, F, refer to the gas constant, temperature, valence charge and Faraday constant respectively; λ and c refer to ion activity coefficient and concentration.

Response of self-powered bio-inspired sensor

The use of ion selective membranes can obtain pressure-driven electric power generation as an electrical signal source. Here, we used the prGO-Trail membrane with excellent ion transport ability to achieve pressure-driven osmotic energy conversion. In short, the prGO-Trail membrane is assembled between two neutral compartments, which are filled with the same concentration of NaCl solution. The Poly tetra fluoroethylene (PTFE) membrane coated with carbon paste is used as the electrode, and the cover plate is assembled on the two neutral chambers to assemble the sensor (Figure S8). Driven by pressure, the ions in the solution always diffuse from the direction of the pressure source to the other side, causing uneven charge distribution in the two regions (Figure S1). Because the prGO-Trail membrane exhibits good cation selectivity, under pressure, Na⁺ diffuses from the pressure source to the other side where no pressure is applied in the solution. With the help of the electrode on two sides, the pressure gradient energy can be collected and converted into electrical energy, so as to achieve self-powered function and the purpose of pressure direction sensing (Figure 3A).²³

FIGURE 3. Response of self-powered bio-inspired sensor. (A) Schematic representation of the electrical signal testing setup. (B) In a test circuit (left), the ionic current (right) was repeatedly produced through press two sides of the pressure sensor (applying alternating pressure on both sides of the sensor). (C) The I-T curves under serial pressures (applying pressure on only one side of the sensor). (D) Sensitivity of the sensors. The optimized prGO-Trail membrane has higher sensitivity than the pristine prGO/T-ZnO composite membrane. (E) The I-T curve indicates the response time and recovery time of the sensor under 549 Pa. (F) The frequency response performance of self-powered sensors exhibits excellent frequency uniformity and stability under 549 Pa.

In order to study the current change characteristics of ion channel pressure sensors, a high-precision electrical signal testing system (including electrical signal detection devices, motion propulsion devices and electrical signal acquisition computers) is used to test the pressure-sensitive characteristics of the sensors. Figure 3B shows the I-T curves of applying alternating pressure on both sides of the sensor, and the current signal has typical directional response. This is due to the different directions of Na⁺ diffusion driven by alternating pressure on both sides. Therefore, the digital multimeter collects a set of current signals with completely opposite directions. Figure 3C shows a typical pressure-driven current response in 1.0 M NaCl aqueous solution. At this time, the test method is to apply pressure on only one side of the sensor, and maintain the pressure on one side in the subsequent pressure test to quantify the performance of the 2D nanofluidic pressure sensor. When different levels of external pressure are applied, the current signal is significantly improved compared to the sensor assembled with the original prGO membrane (Figure S10). Membrane thickness is also an important physical factor affecting energy conversion. As shown in Figure S11, the electrokinetic current increases almost linearly as the membrane thickness decreases from 930 nm to 420 nm at a pressure of 549 Pa, and the electrokinetic current increases when the membrane thickness is <420 nm It is slightly reduced. This is because when the membrane thickness is in the range of 420 ~ 930 nm, it only affects the transmembrane resistance rather than the charge selectivity, which is in line with the classical Ohmic dependence. The lower selectivity caused by the short ion path after the membrane thickness is less than 420 nm will impair electrical energy conversion. We selected a prGO-Trail membrane with a thickness of 5 μm for pressure sensing testing after considering the energy conversion properties and durability of the membrane. Observing the results in Figure S9, the XRD results show that the 2theta angle of the prGO-Trail membrane prepared by the sacrificial template method is shifted to the left by 0.5° compared with the pristine prGO membrane (002). It shows that T-ZnO acts as a sacrificial template to increase the interlayer spacing, resulting in more ion paths. Raman results show that partially reduced graphene oxide has similar peak shapes to graphene oxide. It shows that by controlling the degree of hydrothermal reaction, the C/O atomic ratio can be controlled while maintaining the active oxygen functional groups on prGO, so as to achieve the purpose of artificially controlling the ion channels in the prGO-Trail membrane. And as shown in Figure 3D, in the electrolyte solution of a certain concentration (1.0 M NaCl), the current I has a linear relationship with the applied pressure difference Δp. This pressure-driven output voltage (and current) generated by ion selectivity is linear with the applied pressure and can be described by the Helmholtz-Smoluchowski equation,¹⁴ [Image Omitted. See PDF]

Where all the parameters ε, ζ, A, η, and L are constants, and ε is the permittivity of the solution, ζ is the Zeta potential of the channel surface, η is the viscosity of the solution, A is the cross sectional area of the channel and L is the length of the channel in a certain electrolyte solution and ion pore channel. Therefore, by measuring the current, the pressure difference Δp value corresponding to the ion selective transport mechanism can be calculated. And constructive requirements for self-powered pressure sensors with linear transfer functions are expected to be fulfilled.

We define the current response sensitivity of this nanofluidic pressure sensor as S = ΔI/ΔP, where ΔI is the current density change produced under an external pressure change of ΔP. Figure 3D clearly shows that the optimized prGO-Trail membrane has a sensitivity of 0.282 nA Pa⁻¹, which is ~3.2 times of magnitude improvement compared to the pristine prGO/T-ZnO composite membrane (0.087 nA Pa⁻¹). In addition, another important parameter of the sensor is the response time. The fast response without lag can ensure timely response under external pressure. The response/recovery time of the sensor at 549 Pa is 90 ms/110 ms (Figure 3E), which is sufficient to meet the requirements of practical applications. In order to further evaluate the durability and stability of the self-powered sensor based on prGO-Trail, 1000 load-unload tests were performed on one side under 549 Pa, and the results showed that the current fluctuation was very small, as shown in Figure 3F. The enlarged illustration clearly shows that there is no obvious change in the current amplitude at the starting point and ending point, which confirms that the sensor has good stability.

Pressure sensing mechanism

The main working mechanism of the biomimetic pressure sensor based on EKEC technology in this study is the change of the directional flow of the measured local net charge caused by external stimuli.^43,44 This is highly related to the prGO-Trail membrane as an ion transport layer. The prGO-Trail membrane acts as an integration of large-scale nanofluidic channels and selectively provide a transport channel for cations. Its ionic conductivity is closely related to the concentration of the electrolyte solution. By controlling the concentration of the electrolyte solution, the surface charge of the nanofluidic channel can be controlled, thereby further regulating ion transport. From Figure 2C, we know that the ionic conductivity is only related to the diffusion behavior of the bulk solution in the high-concentration electrolyte, while the ion transport behavior is only controlled by the surface charge of the nanoflow channel in the low-concentration electrolyte. Therefore, 1.0 M NaCl solution is used as the electrolyte of the pressure sensor. When the sensor is stimulated by external pressure, the directional movement of ions of a single charge (in this case, cations) will cause a directional flow of net charge, resulting in an electric current (or voltage).

The specific working mechanism test steps for pressure sensors based on EKEC technology are as follows: gently place a weight on the bionic ion channel pressure sensor, wait for a period of time and then remove the weight. As shown in Figure 4A and 4B, when 5 g, 10 g, and 20 g weights are pressed on the sensor, the pressures what generate are calculated as 327 Pa, 432 Pa, and 613 Pa, respectively. The sensors generate adaptive signals for voltage and current respectively. It is observed that the electrical signal of the sensor continues to decay over a period of time, which indicates that the sensor has gradually adapted to the applied pressure. When the weight is removed, the electrical signal is reversed and then attenuated to a minimum.

FIGURE 4. Operating principles and output performance of biomimetic pressure sensor. The image depicts the voltage (A) and current (B) response of the device under the pressure of different weights. (C) A schematic structure of sensor. i: the image shows the sensor is in equilibrium state; ii: after adding pressure, the sensor is under non-equilibrium state; iii: the sensor is in equilibrium state under constant pressure; iv: after withdrawing the pressure, the balance that establish in pressure is broken, then backing to state i. (D) The complete net charge flow process of a biomimetic pressure sensor after being stimulated by pressure.

Figure 4D records the complete net charge flow process of the sensor after being stimulated by pressure. When a constant weight is placed on the self-powered sensor, the current signal undergoes a rapid response to a slow decay to zero. When the constant weight is removed, the current signal reverses and then slowly decays to zero again. This phenomenon can be illustrated by the four processes shown in Figure 4C. Figure 4C i shows the original state of the device. At this time, the device is in an equilibrium state, so the ion concentration of the electrolysis chambers on both sides is in an equilibrium state. Figure 4C ii shows that when the device is under external pressure, the cations in the electrolysis chamber on the pressure side diffuse through the prGO-Trail membrane to the electrolysis chamber on the opposite side. This causes the cation concentration of the electrolysis chamber on the opposite side to be higher than that on the pressure side, so a current signal is generated in the external circuit. When pressure is continuously applied to the device, the cations in the pressure side gradually no longer diffuse to the opposite side (Figure 4C iii), which leads to the phenomenon shown in Figure 4D, where the voltage and current signal continues to decay within a few seconds. As shown in Figure 4C iv, when the pressure applied to the device is release, the cation concentration in the electrolysis chambers on both sides begins to rebalance, and the cations in the opposite side begin to diffuse into the pressure side. Eventually, the cation concentration of the electrolysis chambers on both sides returned to the original state, and the charge gradually stabilized and tended to balance. Therefore, after a fixed pressure is applied to the sensor, a constantly decaying streaming current and streaming potential will be generated, so as to realize the conversion and output of energy. Since the current/voltage signal generation process is dynamic and adaptive, this makes this device is more suitable for sensing dynamic pressure rather than continuous pressure. In addition, the adaptive current signal of the sensor can be maintained for longer than 25 s (Figure 4D), while the normal pulse of a human takes about 0.6–1 s to beat once. Therefore, the sensor adaptive sensing duration based on EKEC technology can completely cover the pulse duration of human pulse. And because of the linear relationship between the sensor's electrical signal and the applied pressure, this sensor has great application potential for human pulse health monitoring.

Practical applications of the self-powered bio-inspired sensor

The sensor based on prGO-Trail membrane has the characteristics of high sensitivity, fast response time, high stability, and no need for external power source. It has great potential in monitoring low-frequency vibration signals, acoustic signals, tiny pulse stimuli, and non-contact vibration. Figure 5A shows the test signal of the water drop. A drop of water (~12 mg) will cause vibration with pressure of about 1.5 Pa to the sensor. Since the electrode of the sensor is thin-membrane, it is easy to sense vibration. In addition, the sensor can sense the non-contact vibration of the human body, thereby generating an electrical signal response. Figure 5B shows the signal generated by a volunteer blowing continuously to the sensor. According to calculations, the pressure generated by the volunteers blowing is less than 1.5 Pa. As shown in Figure 5C, clapping at a position ~3 cm away from the sensor, it can be observed that the sensor generates a considerable electrical signal.

FIGURE 5. Application of ion channel pressure sensor. (A) The response of drop of water (about 1.5 Pa), (B) exhale air pressure ([less than]1.5 Pa), (C) clapping at a position ~3 cm away from the sensor. (D) Pulse pressure test device. (E) Pulse pressure from the wrists of normal conditions (about 86 beats per minute) and postexercise (96 beats per minute). (F) The details of single pulse. (G) Schematic illustration of setup for acoustic vibration test by using ion channel pressure sensor. (H) The response (the above line) of a series of discontinuous audio (the below line) at different strength. (I) The response corresponds coincident to original audio from time and strength.

At the same time, in order to monitor the pulse condition of the human body in real time, ion channel sensors are placed on the wrist of an adult male to detect tiny radial pulse waves (Figure 5D). Pulse wave is an important physiological signal used to detect blood pressure, heart rate, arteriosclerosis, and so forth. As shown in Figure 5E, the pulse of the human body at motion state (96 beats per minute) and calm state (84 beats per minute) can be clearly identified. Figure 5F shows the signal of a single pulse. The sensor can identify the systolic peak (P1) and diastolic peak (P2) of the wrist artery. According to electrical signal parameters, some health indicators of the human body can be obtained, such as reflection index (R.I. = b/a × 100%, b is the systolic peak, a is the diastolic peak) and arterial stiffness index (S.I. = volunteer height/△T, △T is the peak systolic and diastolic delay time of the carpal artery).^45,46 The calculated volunteer pulse reflex index and arterial index were 59% and 5.1. These two indices are indicators for measuring the degree of arteriosclerosis developed by the international medical community. Atherosclerosis will lead to the decline of the two indices, leading to cardiovascular diseases such as hypertension. Therefore, the real-time monitoring of the reflex index and the arterial index is very important. Therefore, our sensor shows excellent potential for medical health detection.

In addition, in order to further explore the practical application of ion channel sensors, we designed a test of the device's response to sound. Figure 5G shows a schematic diagram of the test device, which mainly includes a speaker, an Agilent digital source meter and a sponge with the effect of isolating external vibrations. When the horn plays sound, the sound propagates through the air and causes the electrode in the ion channel pressure sensor to vibrate. The vibration causes the transport of ions inside the device, and finally generates a current signal. The signal shown in Figure 5H is the detection signal and the original digital audio signal of three audio signals with different strengths and a slightly longer interval. The upper curve is the electrical signal measured by the experiment, and the lower curve is the original digital audio signal. The intensity and time interval of the detected electrical signal and the original sound signal can be completely matched, which means that the sensor has a fairly good degree of restoration of the sound signal. Figure 5I shows the detection signal and the original digital audio signal of a short interval audio signal. The lower curve is the original digital audio signal, and the upper is the signal detected by the sensor. The signal strength is different and the frequency is higher. After careful comparison, we found that although the peak shape is different, the peak of the audio waveform corresponds to the peak of the current signal of our sensor. Therefore, it can be said that the ion channel pressure sensor we prepared has good sensing characteristics for low-frequency small vibration signals, and can completely restore low-frequency audio signals and can restore high-frequency sound signals to a certain extent.

CONCLUSIONS

In summary, a self-powered bio-inspired pressure sensor inspired by serosa-mimetic structure ion selectivity is demonstrated. The partially reduced GO membrane is used to make nanochannels to simulate the function of biological ion channels. The sacrificial template method is used to introduce T-ZnO between the two-dimensional layered prGO membranes and then etch them to prepare ion exchange membranes with artificially adjusted ion channel performance. The experiment results systematacially proved the cation selectivity of prGO-Trail membrane. Therefore, the prGO-Trail membrane can be used as an ion transport layer to promote self-powered pressure sensing. The test shows that the sensor has good performance, the optimal response sensitivity is 0.282 nA Pa⁻¹, the response time is 90 ms, the recovery time is 110 ms, and the minimum detection limit is about 1.5 Pa. In addition, the pressure-driven output voltage (and current) generated by ion selectivity is linear with the applied pressure, which satisfies the construction requirements of future self-powered pressure sensors with linear transfer functions. We use the inspiration from nature to prepare a bionic pressure sensor that can be used for human health detection, which is a harmonious embodiment of nature and human life and health. Our sensors have the ability to recognize pressure directions, respond to airflow and hands, monitor pulse, and detect acoustic vibrations in real time. In particular, the sensor has the characteristics of constant pressure self-adaptation. We look forward to the potential applications of the various functions of the ion channel pressure sensor developed in soft robots and wearable electronic devices.

Experimental section Materials

Sodium nitrate (NaNO₃), concentrated sulfuric acid (H₂SO₄), potassium permanganate (KMnO₄), hydrogen peroxide (H₂O₂), zinc powder (Zn) and hydrochloric acid (HCl) were purchased from Sinopharm group. Flake graphite was purchased from Aldrich. All the chemical reagents were used as received without further purification.

Synthesis of graphene oxide

GO was prepared by oxidation of natural graphite powder by using a modified Hummers' method. Briefly, 23 ml sulfuric acid was stirred at ice-bath and the mixture of 1.0 g graphite and 0.5 g sodium nitrate was added. Potassium permanganate (3.0 g) was added slowly under vigorous agitation, and it is very important to keep the temperature of the suspension lower than 20°C. Successively, 35°C of water-bath was needed for the reaction system for 2 h until a thick paste was formed. Under vigorous agitation, deionized water (46 ml) was added continuously and slowly to keep the temperature lower than 98°C, after finish it, the reaction system was transferred to 98°C oil-bath for about 15 min, turning the color of the solution from brown to yellow. 300 ml 3% hydrogen peroxide solution was added to react with excessive potassium permanganate for 30 min. The resulting solution was filtered and washed with 3.0% HCl aqueous solution (300 ml). The resulting solid was dispersed in water by ultrasonication for about 1.5 h to make a GO aqueous dispersion (10 mg/ml).

Preparation of partial reduced graphene oxide

Typically, 5.0 ml graphene oxide dispersion (10 mg/ml) and 100 ml DI water were added in a wide mouth bottle, after being vigorously shaken for a few minutes, then the bottle was put in oven (95, 120 or 150°C) for 3 h. The resulting colloid is a partially reduced graphene oxide colloid.

Synthesis of tetragonal needle-like zinc oxide

Place a quartz boat containing 0.5 g of zinc powder in a single-ended quartz tube. Then place a silicon wafer 10 cm away from the quartz boat. Finally, the entire quartz tube was placed in a CVD furnace with a dual temperature zone at a heating rate of 10°C min⁻¹, heated to 950°C under atmospheric pressure and kept for 5 min, and then the temperature was quickly lowered. The generated tetragonal needle-shaped zinc oxide will be deposited on the silicon wafer.

Fabrication of the prGO-trail membrane

The prepared T-ZnO was added to DI water, stirred and sonicated to form a 1 mg ml⁻¹ suspension. Then prGO colloid was mixed with 10 wt% T-ZnO suspension. The mixture was sonicated for 15 min and stirred for 2 h to form a uniform suspension. The prGO/T-ZnO mixture is vacuum filtered on a nitrocellulose membrane. After being dried in the air for 24 hours, the prGO/T-ZnO composite membrane was easily peeled from the substrate. Then the composite membrane was immersed in 0.1 M HCl for 12 h to etch the T-ZnO, and dried in air to obtain the prGO-Trail membrane.

Characterization and electrical measurements

The structures and morphologies of prGO nanosheets, T-ZnO, prGO-Trail membrane were characterized by a high-resolution scanning transmission electron microscope (TEM, FEI Titan G2 60–300), a scanning electron microscope (SEM, FEI Nova NanoSEM 450, 10 kV). X-ray diffractometer (XRD, PANalytical B.V. Empyrean) was also used to further characterize the chemical compositions and the structure. The X-ray photoelectron spectroscopy (XPS) was performed in AXIS-ULTRA DLD-600 W. And the different hydrophilies of GO membrane, prGO-Trail membrane were characterized by contact angle measuring instrument (CE-100ACA) as well. The sensing performance of Bio-inspired self-powered pressure sensor was tested on the system with a computer-controlled single-axis motor, a force gauge, and a digital multimeter (Keithley DMM7510).

The transmembrane ionic transport tests were performed with source meter (Agilent B2901A). The prGO-Trail membrane was mounted between a custom-made two-compartment electrochemical cell. The testing membrane area was about 0.785 cm². Homemade Ag/AgCl electrodes were used to apply a transmembrane electrical potential and remained stable during the testing process. The testing solutions were all prepared using ultrapure water (18.2 MΩ cm⁻¹).

AUTHOR CONTRIBUTIONS

Ziqi Ren, and Hang Zhang, share the same contribution to the work. Nishuang Liu supervised the whole project. Nishuang Liu, Ziqi Ren, and Hang Zhang con-ceived and designed the experiments. Ziqi Ren, Hang Zhang, Dandan Lei, Qixiang Zhang, Tuoyi Su and Luoxin Wang performed the experiments. Yihua Gao and Jun Su rendered helpful discussions. Ziqi Ren wrote the paper. All authors commented on the manuscript.

ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China (NSFC: 51872106) (Nishuang Liu), NSFC: 11874025 (Yihua Gao) and the Natural Science Foundation of Hubei Province (NSFHB: 2016CFB43) (Nishuang Liu). The authors also thank the Analysis and Testing Center of HUST for the XRD, XPS and Raman characterizations. Thank the Center of Optoelectronic Micro & Nano Fabrication and Characterizing Facility, Wuhan National Laboratory for Optoelectronics of HUST for the support in the SEM tests.

CONFLICT OF INTEREST

The authors declare no conflict of interest.