FSTI device based on two-dimensional layered heterojunction MXene / Bi and application of FSTI device in man-machine intelligent interaction

Abstract

The invention discloses a two-dimensional layered heterojunction MXene / Bi-based FSTI device and application thereof in man-machine intelligent interaction, the FSTI device comprises a drain electrode, a source electrode, an electrolyte and an isolation layer, the drain electrode material is an MXene / Bi material with a two-dimensional layered heterojunction structure, the source electrode material is an MXene (at) Zn material, the electrolyte is a PVDF-HFP-GO solid electrolyte, and the isolation layer is a CNF isolation layer. The invention provides a mechanically gated and self-powered FSTI device for the first time, and the FSTI device can be used for man-machine intelligent interaction. The FSTI device obtains linearity (99.5%), portability, self-energization, long service life (as high as 50,000 cycles) and millisecond-level response time (66.59 ms) in the aspect of static / dynamic response. By introducing the tactile feedback technology into the man-machine intelligent interaction application facing the manipulator, the manipulator can execute flexible and meticulous operation tasks, so that the method is particularly suitable for man-machine intelligent interaction scenes, and a new solution is provided for popularization and application of artificial intelligence.

Description

Technical Field

[0001] The present invention belongs to the field of electronic devices and artificial intelligence technology, and specifically relates to a FSTI-like transistor-like ion electronic device (FSTI), and in particular to a FSTI device based on a two-dimensional layered heterojunction MXene / Bi and its application in human-machine intelligent interaction. Background Art

[0002] Human-computer interaction has achieved remarkable results in research fields such as healthcare and virtual worlds. Nevertheless, these fields usually rely on overly complex multi-sensor integration and face challenges in achieving precise control of robot operations. With the rapid development of the Internet of Things and robotics, more intelligent human-computer interaction technologies have put forward an urgent need for remote collaboration to perform flexible and delicate operation tasks. The realization of intelligent human-computer interaction scenarios is highly dependent on advanced tactile sensor networks as interface devices for human-computer intelligent interaction. In recent years, researchers have proposed various sensor integrations and high-resolution arrays. However, these multi-sensor network integrations often expose many shortcomings, including complex structures, signal crosstalk, and stacking of a large number of signals for serial transmission. In addition, electronic information transmission requires a large amount of data transmission and signal processing protocols, which may cause serious delays and complex wiring in machine control systems. All of the above situations will lead to a significant aggravation of the von Neumann bottleneck.

[0003] In order to extract useful information from the increasingly complex data collected by a single sensor or a sensor network, tactile perception is being integrated with advanced data analysis methods. Therefore, it is particularly urgent to combine a single interactive device with a neural network to solve the drawbacks of multi-sensor network integration. More importantly, the skilled hand movements of the robot are not only regulated by signals from the control system, but also preferably by tactile feedback. In order to facilitate artificial intelligence computing while achieving long-term portable, stable and low-power operation, the above-mentioned human-machine intelligent interaction scenarios require interactive devices with fast response time, excellent cycle performance, static / dynamic response, portability, self-power supply and linearity. At present, tactile sensors based on traditional mechanisms can no longer meet these demanding requirements. Therefore, it is a major challenge to design interactive devices suitable for human-machine intelligent interaction systems with excellent sensing performance and self-powered capabilities. Summary of the invention

[0004] The purpose of the present invention is to provide a FSTI device based on a two-dimensional layered heterojunction MXene / Bi as a drain material, and further apply it in human-machine intelligent interaction scenarios. Inspired by the regulation of carrier transport by gate voltage in electronic transistors, the present invention develops a FSTI device (transistor-like ion electronics device). The device emulates the basic design of traditional electronic transistors in both structure and function ( Figure 1), using mechanical stimulation to effectively regulate ion and electron migration, and realizing mechanically sensitive ion channel valve characteristics. This transistor-like, flexible, all-solid-state, self-powered ion electronics device has the characteristics of fast response time (66.59 ms), long life (up to 50,000 cycles), linear and sensitive static / dynamic response, etc. Importantly, combined with deep learning algorithms, a single FSTI device can decode complex gestures in real time (with a recognition accuracy of up to 95.83%). And by introducing tactile feedback technology into human-machine intelligent interaction applications for manipulators, the manipulator can perform flexible and delicate operation tasks. Specifically, the present invention adopts the following technical solutions to achieve it:

[0005] First, the present invention provides a FSTI device based on a two-dimensional layered heterojunction MXene / Bi, which includes a drain, a source, an electrolyte, and an isolation layer, wherein the drain material is a MXene / Bi material with a two-dimensional layered heterojunction structure, the source material is a MXene@Zn material (Zn coated MXene structure), the electrolyte is a PVDF-HFP-GO solid electrolyte, and the isolation layer is a CNF isolation layer; the drain and the source are made into interdigitated electrodes that cross each other but do not contact each other to form a (drain source) electrode layer, the electrolyte is located on the (drain source) electrode layer, the CNF isolation layer is located between the electrolyte and the drain source electrode layer, and the drain and source are respectively connected with conductive sheets such as copper sheets. The entire device can be coated with an insulating material, such as a polyurethane material.

[0006] Among them, the two-dimensional layered heterojunction MXene / Bi material is a two-dimensional layered structure ( Figure 2 ), which is obtained by uniformly mixing Bi nanosheets with a MXene dispersion, wherein the mass ratio of Bi nanosheets to MXene is 1:1~2. The material inhibits the self-stacking of two-dimensional materials between the MXene and Bi phases, thereby increasing active sites and accelerating the diffusion of zinc ions. Therefore, this heterojunction structure is extremely suitable for FSTI human-machine intelligent interaction that requires high mechanical flexibility and rapid ion transport. The preparation method of the two-dimensional layered heterojunction MXene / Bi material can adopt the following steps: first, the Bi nanosheets are uniformly mixed with the MXene dispersion, and then the mixed dispersion is centrifuged at high speed to obtain agglomerates, and then the obtained agglomerates are dried.

[0007] Secondly, the present invention also provides a method for preparing the FSTI device based on the two-dimensional layered heterojunction MXene / Bi, which includes the preparation of the drain electrode, the preparation of the source electrode, the preparation of the electrolyte, the preparation of the isolation layer, and the assembly of the FSTI device.

[0008] Preferably, the preparation of the drain electrode comprises the following steps:

[0009] Step (1): uniformly mixing Bi nanosheets and MXene dispersion, with a mass ratio of Bi:MXene being 1:1-2;

[0010] Step (2): centrifuging the mixed dispersion obtained in step (1) at high speed to obtain agglomerates;

[0011] Step (3): evenly apply the agglomerate obtained in step (2) on a paper base and dry it for 24 to 36 hours;

[0012] Step (4): The paper base obtained in step (3) is laser engraved to obtain an interdigitated electrode, which is referred to as a MXene / Bi interdigitated electrode and is used as a drain electrode of a transistor-like ionic electronic device.

[0013] Preferably, the preparation of the source electrode comprises the following steps:

[0014] Step (a): centrifuging the MXene dispersion at high speed to obtain agglomerates;

[0015] Step (b): evenly apply the agglomerate obtained in step (a) on a paper base and dry it for 24 to 36 hours;

[0016] Step (c): laser engraving the paper base obtained in step (b) to obtain interdigitated electrodes;

[0017] Step (d): electroplating zinc on the interdigital electrodes obtained in step (c). The interdigital electrodes after electroplating zinc are referred to as MXene@Zn interdigital electrodes, which are used as the source of transistor-like ionic electronic devices.

[0018] Preferably, the electrolyte is prepared by dissolving poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP), graphene oxide (GO) and ZnSO4 in dimethylformamide (DMF) in a mass ratio of 10:3~4:5, stirring the suspension, ultrasonicating, evaporating and drying, and the obtained film is recorded as PVDF-HFP-GO solid electrolyte.

[0019] Preferably, the isolation layer is prepared by dissolving a cellulose nanofiber (CNF) stock solution in an alcohol solution, stirring and drying the mixed solution, and the resulting film is recorded as a CNF isolation layer. The CNF isolation layer is prepared by a common method, and no special requirements are made on the concentration of the CNF stock solution and the alcohol solution. For example, 20-30 vol% of the CNF stock solution can be dissolved in a 60-70 vol% alcohol solution. Of course, it can also be adjusted appropriately as needed. The unit area mass of CNF in the CNF isolation layer is preferably 0.2~0.3 mgcm -2 .

[0020] Preferably, the FSTI device is assembled by attaching the drain and source to an insulating material, then attaching an isolation layer to the surface of the drain and source, then attaching an electrolyte to the isolation layer, and finally packaging it with an insulating material. The drain and source are respectively connected to conductive sheets.

[0021] As a more preferred technical solution, the method for preparing the FSTI device based on the two-dimensional layered heterojunction MXene / Bi can adopt the following steps:

[0022] Step (1): uniformly mixing Bi nanosheets and MXene dispersion, with a mass ratio of Bi:MXene being 1:1-2;

[0023] Step (2): centrifuging the mixed dispersion obtained in step (1) at high speed to obtain agglomerates;

[0024] Step (3): evenly apply the agglomerate obtained in step (2) on a paper base and dry it for 24 to 36 hours;

[0025] Step (4): laser engraving the paper substrate obtained in step (3) to obtain an interdigitated electrode, which is referred to as a MXene / Bi interdigitated electrode and is used as a drain electrode of a transistor-like ionic electronic device;

[0026] Step (5): centrifuging the MXene dispersion at high speed to obtain agglomerates;

[0027] Step (6): evenly apply the agglomerate obtained in step (5) on a paper base and dry it for 24 to 36 hours;

[0028] Step (7): laser engraving the paper base obtained in step (6) to obtain interdigitated electrodes;

[0029] Step (8): electroplating zinc on the interdigital electrode obtained in step (7); the interdigital electrode after electroplating zinc is referred to as MXene@Zn interdigital electrode, and used as the source of a transistor-like ionic electronic device;

[0030] Step (9): poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP), graphene oxide (GO) and ZnSO4 are dissolved in dimethylformamide (DMF) at a mass ratio of 10:3 to 4:5. The suspension is stirred, ultrasonicated, evaporated and dried. The obtained membrane is recorded as PVDF-HFP-GO solid electrolyte;

[0031] Step (10): dissolving the cellulose nanofiber (CNF) stock solution in the alcohol mixed solution, stirring and drying the mixed solution. The obtained film is recorded as the CNF isolation layer.

[0032] Preferably, the MXene dispersion in steps (1) and (5) can be prepared by in-situ etching Ti3AlC2 with lithium fluoride / hydrochloric acid according to the literature method, or can be purchased from a professional manufacturer. There is no special requirement for its concentration. Of course, when it is used for the preparation of drain and source materials, the MXene loading (mass) is preferably 1:1. The paper base used in steps (3) and (6) is a medium-speed qualitative filter paper with a diameter of 10-20 cm; the laser power used for laser engraving in steps (4) and (7) is 10.0~20%; the electroplating solution used for electroplating zinc in step (8) is 1-5M ZnSO4, preferably 2-3M; in step (9), when the masses of PVDF-HFP, GO and ZnSO4 are 200mg, 60~80mg and 100mg respectively, the volume of DMF is 5~10 ml. The suspension is stirred at 40.0-50.0°C until it is fully uniform, ultrasonically treated for 3-5 hours, then cast on a glass plate, evaporated at room temperature to form a film, and the film is dried in a vacuum drying oven at 70.0-90.0°C for 72-120 hours, and finally vacuum dried at 120.0-130.0°C for 4-8 hours. In step (10), when the volume ratio of alcohol to water is 2:1, the mixed solution is stirred until it is fully uniform, then cast on a glass plate, and dried on a heating table at 40.0-60.0°C.

[0033] The FSTI device based on the two-dimensional layered heterojunction MXene / Bi is used for human-machine intelligent interaction applications, characterized by adopting the following method steps: attaching the MXene@Zn interdigital electrode and the MXene / Bi interdigital electrode to polyurethane, sandwiching a CNF isolation layer between the electrode surface and the PVDF-HFP-GO solid electrolyte, and finally packaging with an insulating material such as polyurethane.

[0034] Thirdly, the present invention provides a tactile sensor, including a FSTI device based on a two-dimensional layered heterojunction MXene / Bi.

[0035] Fourthly, the present invention also provides the application of the above-mentioned FSTI device based on the two-dimensional layered heterojunction MXene / Bi in a human-computer intelligent interaction system.

[0036] Fifth, the present invention provides a human-machine intelligent interaction system, including a FSTI device based on a two-dimensional layered heterojunction MXene / Bi, a microprocessor, a manipulator, a signal filtering and amplifying module, and a power module, wherein:

[0037] FSTI device, used to collect human physiological signals and sensory feedback signals of the manipulator;

[0038] A microprocessor, used to process the signal data so as to analyze and evaluate the data through a preset algorithm;

[0039] A robotic arm is used to simulate the gestures of a human hand and perform operations in an anthropomorphic manner;

[0040] A signal filtering and amplifying module, used for amplifying and filtering the signal;

[0041] The power module is used to provide electrical energy to the signal filtering and amplifying module, the microprocessor and the manipulator.

[0042] The transistor-like ion electronic device described in the present invention draws on the working principle of PMOSFET (P-type metal oxide semiconductor field effect transistor). Figure 3 a, showing the working principle of the traditional PMOSFET is that due to the formation of two reverse PN junctions between the P-type source and drain and the N-type substrate, there is no carrier transport between the source and drain, resulting in the transistor being in the off state. In contrast, when the external power supply continuously provides power to the source and drain, if the gate also provides a certain bias voltage, due to the charge induction effect, the free electrons of the N-type substrate are repelled, and a large number of holes gather at the gate, making the hole concentration near the gate relatively high, forming a conductive channel connecting the source and drain. As the gate voltage gradually increases, the conductive channel gradually widens until saturation, and the migration process of holes from the source to the drain is enhanced. At this time, the transistor is in a fully on state. In short, the gate voltage plays a role similar to the valve in a faucet, regulating the transport of holes by forming a conductive channel.

[0043] Similarly, Figure 3 b shows the dynamic gate mechanical sensitivity process of the FSTI device of the present invention. In the absence of external stimulation, the FSTI device does not form ion channels due to the insulating effect of the CNF isolation layer, so it has no energy loss and is in a silent state. On the contrary, during the process of loading external pressure, the MXene@Zn source and MXene / Bi drain overcome the blocking effect of the CNF isolation layer, resulting in an increase in the contact area with the PVDF-HFP-GO solid electrolyte, and ion channels begin to form and the number of channels increases with the increase of pressure. At this time, the source undergoes an oxidation reaction, and the released Zn 2+ Ions are transported through the electrolyte across the source / electrolyte interface and then intercalated into the drain. The entire ion transport process requires the transfer of charge in the external circuit to balance it, thereby generating an electrical signal in the external circuit that increases with pressure. In summary, during device operation, pressure also plays a role similar to the valve in a faucet, regulating the ion transport between the source and drain by forming ion channels to achieve the coded output of electrical signals in the external circuit. Importantly, unlike PMOSFET, which requires an external power supply to achieve switching characteristics, the ion transport process in this device is continuous and spontaneous. Therefore, the device can be self-powered to respond to static and dynamic signals.

[0044] On the other hand, the comprehensive performance of the FSTI device was evaluated using an electrical signal test platform. The FSTI device showed a wide output range (up to 1.1 V and 2.3 μA), which can provide stable load capacity for human-machine intelligent interaction scenarios ( Figure 4 ). Thermodynamically, once the ion concentration of the electrolyte is fixed, the inherent intrinsic potential difference between the two electrodes is also fixed. Therefore, the present invention modulates the ion transport process to change the output capacity of the device intrinsic potential difference. In fact, the number of ion channels formed by external pressure modulation can be reflected in the macro-control of the internal impedance of the device, so that the FSTI device has the ability to distinguish different pressure signals. As depicted in the equivalent circuit diagram of the FSTI device, the measured voltage signal U OC With the internal resistance R d changes with the changes in Figure 5 ). The specific formula is as follows:

[0045]

[0046] Where Uε is the intrinsic potential difference of the device (estimated to be 1.1 V), R V is the internal resistance of the voltmeter (10.0 MΩ). d It will decrease with the increase of pressure. When there is no external pressure, due to the insulation effect of the CNF isolation layer, the FSTI device prevents the formation of ion channels. Therefore, the internal resistance of the device is close to infinity, and the measured voltage U OC Almost 0. On the contrary, when pressure is gradually applied to the surface of the device, the electrode and electrolyte begin to contact and ion channels begin to form. As the pressure increases, the electrode / electrolyte interface area increases and the number of ion channels increases. At this time, R d Gradually decreases, the measured voltage U OC Gradually increases. d Reduced to much less than R V (10.0 MΩ) high resistance value, the measured voltage V OC Approaching U ε Achieving low power consumption of devices has always been a concern. By using an ammeter and a voltmeter to measure the current and voltage respectively, the power consumption of FSTI devices can be quantitatively analyzed. It is not difficult to understand that when pressure is applied to make R d When it approaches 0, the power P of the device reaches its maximum value. For voltage meter detection, the maximum power consumption can be expressed by the formula Calculation shows For ammeter detection, due to R G Very small, the ammeter will not divide the voltage from the circuit. Figure 4 It can be seen that the average current I measured when no excessive pressure is applied is 2,300 nA. Therefore, the maximum power consumption can be calculated by the formula Calculation shows In summary, the power consumption of ammeter detection is an order of magnitude greater than that of voltmeter detection, so voltmeter detection is low power consumption. Based on this, the electrical signal that the device can encode and output should be collected using the voltmeter detection method, thereby greatly reducing the power consumption of the device.

[0047] Thirdly, in order to further prove that U can be adjusted by adjusting the internal impedance of the FSTI device OC The present invention evaluates the current-voltage (IV) characteristics and electrochemical impedance of the FSTI device. From the slope of the IV characteristics, the R d The value decreases with increasing pressure ( Figure 6 a). In addition, when the input voltage (E1) value of the device is 0, the output current signal also increases with the increase of pressure. The above results show that the ion channel can be effectively regulated by external pressure and the device is self-powered. It is worth noting that all IV curves intersect at (-1.1, 0). By analyzing the equation and IV characteristics test circuit diagram ( Figure 6 b), it can be seen that the device’s U ε To clearly characterize the effect of mechanical stimulation on ion transport, we performed electrochemical impedance spectroscopy (EIS) on the device without an isolation layer. Figure 6 c, 6d). On the one hand, the R represented by the semicircular area ct It decreases with increasing pressure, indicating that the charge transfer resistance at the electrode / electrolyte interface decreases with increasing pressure regulation. This process can be represented by a parallel circuit of resistance and capacitance. The intercept value of the high-frequency curve on the real axis (R Ω ) decreases with the application of external pressure. On the other hand, the slope in the low-frequency region is related to the ion diffusion process and can be expressed as the Warburg impedance Z W Under pressure regulation, the slope of the line increases, indicating enhanced ion diffusion in the ion channel.

[0048] Fourthly, thanks to the low-power voltmeter detection method, the device system optimization and the excellent isolation effect of the isolation layer, the device has reached an amazing 50,000 times under the cyclic power test with almost zero loss ( Figure 7 a), ensuring long-term stable signal monitoring in artificial intelligence applications. The device's electromechanical transmission function determines the sensitivity and detection range of pressure detection, and its high adjustability is crucial for expanding the application range of the device. By adjusting the unit area mass of the CNF isolation layer, the electromechanical transmission function of the FSTI device can be effectively regulated ( Figure 7b). Compared with discrete signals, linear signals are more helpful for feature extraction and computational analysis of artificial intelligence systems. Figure 7 c shows that as the external pressure increases, the corresponding output signal of the FSTI device maintains a stable linear mode (99.5%). In addition, the FSTI device exhibits excellent switching characteristics under the isolation of CNF nanofibers ( Figure 7 d). This millisecond-scale ion transport capability reduces power consumption and accommodates the speed necessary for parallel acceleration of deep learning. This is attributed to the drain with excellent fast diffusion kinetics ( Figure 8 ) and the design of the interdigital electrode structure that reduces the ion transport path. Usually, human physiological activities are high-frequency / low-frequency, or even high-frequency and low-frequency simultaneously. Since the ion transport process in the FSTI device is continuous and spontaneous, the high / low frequency response range of the FSTI can be as high as 2.22Hz ( Fig. 9 ).

[0049] Fifth, with the help of deep learning algorithms, a single FSTI device can effectively decode complex gestures in real time by monitoring the subtle skin deformation caused by the median and ulnar nerves, with a recognition accuracy of up to 95.83% ( Fig.10 The system also incorporates tactile perception feedback technology into the human-machine intelligent interaction application of the robot, enabling the robot to perform flexible and precise operation tasks ( Fig.11 ). The device, combined with deep learning, can solve the redundancy problem of using multiple sensing devices to decode multiple joint movements.

[0050] The beneficial effects of the present invention are:

[0051] (1) This invention proposes for the first time a mechanically gated and self-powered FSTI device that can be used for intelligent human-machine interaction. The FSTI device has achieved linearity (99.5%), portability, self-power, long life (up to 50,000 cycles) and millisecond response time (66.59ms) in static / dynamic response, making it particularly suitable for intelligent human-machine interaction scenarios.

[0052] (2) The human-machine intelligent interaction system designed by the present invention, equipped with tactile perception feedback technology, gives the robot accurate tactile perception capabilities. It achieves effective decoding of complex human gestures (up to 95.83% recognition accuracy) through deep learning-assisted single FSTI devices. This method solves the redundancy problem of using multiple sensor devices to decode multiple joint movements. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 This is a structural design diagram of the FSTI device of the present invention; in the figure, 1. insulating material (polyurethane); 2. electrolyte; 3. CNF isolation layer; 4. electrode layer; 5. conductive sheet (copper sheet).

[0054] Figure 2 a is the HAADF-STEM image of the two-dimensional layered heterojunction MXene / Bi of the present invention, Figure 2 b is the cross-sectional SEM image and its corresponding mapping;

[0055] Figure 3 a is the working principle of PMOSFET, Figure 3 b is the dynamic gate mechanical gating process of the FSTI device of the present invention;

[0056] Figure 4 is the signal output range of the FSTI device of the present invention;

[0057] Figure 5 is an equivalent circuit diagram of the FSTI device of the present invention;

[0058] Figure 6 a is the IV curve of the FSTI device of the present invention at different pressures, Figure 6 b is the test circuit diagram of IV characteristics, Figure 6 c is the Nyquist curve measured by the FSTI device of the present invention without adding an isolation layer, Figure 6 d is the corresponding Warburg factor;

[0059] Figure 7 a is the cyclic force and electricity test of the FSTI device of the present invention, Figure 7 b is the electromechanical transmission function of the FSTI device of the present invention under different unit area masses of CNF isolation layers, Figure 7 c is the linearity of the FSTI device of the present invention, Figure 7 d is the response / recovery time of the FSTI device of the present invention;

[0060] Figure 8 a is the electrostatic intermittent titration technology curve of the two-dimensional layered MXene / Bi heterojunction of the present invention, Figure 8 b is the corresponding Zn 2+ Diffusion coefficient;

[0061] Fig. 9 is the frequency response range of the FSTI device of the present invention;

[0062] Fig.10 a is the accuracy confusion matrix of the human-machine intelligent interaction system of the present invention, Fig.10 b. Human-machine interface, Fig.10 c is the gesture action decoded by the system;

[0063] Fig.11 a is a manipulator equipped with a tactile perception negative feedback system. Fig.11b is a manipulator without tactile perception negative feedback system.

[0064] Fig.12 The figure is a schematic diagram of the control principle of a human-machine intelligent interaction system of the present invention. DETAILED DESCRIPTION

[0065] The following embodiments are further descriptions of the content of the present invention as an explanation of the technical content of the present invention, but the essential content of the present invention is not limited to the following embodiments. Ordinary technicians in this field can and should know that any simple changes or replacements based on the essential spirit of the present invention should fall within the scope of protection required by the present invention.

[0066] Example 1

[0067] A method for preparing a FSTI device based on a two-dimensional layered van der Waals heterostructure MXene / Bi comprises the following steps:

[0068] Step (1): uniformly mix Bi nanosheets and MXene dispersion, with a mass ratio of Bi:MXene of 1:1 and a MXene loading of 50 mg;

[0069] Step (2): centrifuging the mixed dispersion obtained in step (1) at high speed to obtain agglomerates;

[0070] Step (3): evenly apply the agglomerate obtained in step (2) on a paper base and dry it for 24 to 36 hours;

[0071] Step (4): laser engraving the paper substrate obtained in step (3) to obtain an interdigitated electrode, which is referred to as a MXene / Bi interdigitated electrode and is used as a drain electrode of a transistor-like ionic electronic device;

[0072] Step (5): centrifuge the MXene dispersion (MXene loading 50 mg) at high speed to obtain aggregates;

[0073] Step (6): evenly apply the agglomerate obtained in step (5) on a paper base and dry it for 24 to 36 hours;

[0074] Step (7): laser engraving the paper base obtained in step (6) to obtain interdigitated electrodes;

[0075] Step (8): electroplating zinc on the interdigital electrode obtained in step (7); the interdigital electrode after electroplating zinc is referred to as MXene@Zn interdigital electrode, and used as the source of a transistor-like ionic electronic device;

[0076] Step (9): PVDF-HFP, graphene oxide (GO) and ZnSO4 were dissolved in DMF at a mass ratio of 10:4:5, wherein the loading amount of PVDF-HFP was 200 mg. The suspension was stirred at 40.0-50.0°C, ultrasonicated, evaporated and dried. The obtained membrane was recorded as PVDF-HFP-GO solid electrolyte;

[0077] Step (10): Dissolve 25 vol% CNF stock solution in 67 vol% alcohol solution (the volume ratio of CNF stock solution to alcohol solution is 1:3), stir and dry the mixed solution. The obtained membrane is recorded as CNF isolation layer, and the unit area mass of CNF in the CNF isolation layer is 0.26 mg cm -2 .

[0078] In Example 1, the MXene / Bi drain, MXene@Zn source, PVDF-HFP-GO solid electrolyte and CNF isolation layer are assembled into a FSTI device. The user first attaches the FSTI device to the wrist artery, connects the signal filtering and amplification module and the microprocessor Arduino Nano 33 BLE Sense to the terminal of the FSTI device, and then performs each gesture 1248 times to generate a large number of time domain signals as a labeled data set. These labeled data sets are cleaned and transmitted to the neural network model. The Keras sequence model based on TensorFlow is used. The cleaned data is parsed, randomly split and assigned input and output pairs in the early stage to facilitate the later neural network model construction and training. After feature extraction and classification in the hidden layer of the neural network, the gradient descent optimization algorithm is used for adaptive moment estimation. This is because the algorithm can reduce the memory requirements of the microprocessor and minimize the gradient noise, thereby solving the limitation problems often seen in artificial intelligence applications constrained by the learning rate hyperparameter. Finally, the fully connected layer uses the SoftMax activation function to synthesize all feature information and associate each output with the input. After the neural network model is built, the model will be converted into a format and embedded in Arduino for real-time analysis. The tactile perception feedback system designed for the robot, specifically, the designed FSTI is attached to the surface of the robot finger, which is used to enrich the tactile perception of the robot when it contacts the target object during operation. With the help of the tactile perception system, the robot makes feedback actions according to the changes in the electrical signals of the FSTI. The corresponding feedback action is to stop the grasping action of the robot finger to prevent the original shape of the tofu from being destroyed. On the contrary, the robot without a connected tactile perception system will destroy the original shape of the tofu without hesitation, such as Fig.11 .

[0079] Example 2

[0080] A main preparation method of a FSTI device based on a two-dimensional layered van der Waals heterostructure MXene / Bi is similar to Example 1, except that in step (10), the unit area mass of CNF in the CNF isolation layer is 0.00 mg cm -2 That is, it does not contain a CNF isolation layer.

[0081] By comparison, it is found that a suitable CNF mass per unit area is required to obtain a linear FSTI device. The mass per unit area of ​​CNF in the CNF isolation layer is between 0.26 and 0.3 mg cm -2 Too much mass per unit area will cause the device to fail to conduct.

[0082] In the technical scheme of the present invention, although some numerical values ​​with better effects are given in the above embodiments, such as the mass per unit area of ​​CNF, the present invention is not limited to the mass per unit area of ​​CNF given in the above embodiments, and the specific mass per unit area is determined according to actual needs; for example, the ratio of Bi and MXene with better effects is given in the embodiments, but the present invention is not limited to the ratio of Bi and MXene given in the above embodiments, but the proportion of MXene should not be too small, and the specific proportion of MXene is determined according to actual needs; that is, the content claimed for protection of the present invention is based on the scope recorded and explained in the claims.

[0083] It should be noted that the above-described embodiments of the present invention are only explanations and clarifications for those skilled in the art to understand the technical essence of the present invention, so the technical contents described are not intended to limit the substantial protection scope of the present invention. The substantial protection scope of the present invention shall be subject to the claims. Those skilled in the art should know that any modifications, equivalent substitutions and improvements based on the substantial spirit of the present invention shall be within the substantial protection scope of the present invention.

Claims

1. A FSTI device based on a two-dimensional layered heterojunction MXene / Bi, characterized in that: It includes a drain electrode, a source electrode, an electrolyte, and an isolation layer, wherein the drain electrode material is a MXene / Bi material with a two-dimensional layered heterojunction structure, the source electrode material is a MXene@Zn material, the electrolyte is a PVDF-HFP-GO solid electrolyte, and the isolation layer is a CNF isolation layer; the MXene / Bi material is obtained by uniformly mixing Bi nanosheets and a MXene dispersion, wherein the mass ratio of Bi nanosheets to MXene is 1:1~2.

2. The method for preparing the FSTI device based on the two-dimensional layered heterojunction MXene / Bi according to claim 1, characterized in that: It includes the preparation of drain electrode, source electrode, electrolyte, isolation layer and FSTI device assembly.

3. The preparation method according to claim 2, characterized in that: The preparation of the drain electrode comprises the following steps: Step (1): uniformly mixing Bi nanosheets and MXene dispersion, with the mass ratio of Bi to MXene being 1:1-2; Step (2): centrifuging the mixed dispersion obtained in step (1) at high speed to obtain agglomerates; Step (3): evenly apply the agglomerate obtained in step (2) on a paper base and dry it for 24 to 36 hours; Step (4): The paper base obtained in step (3) is laser engraved to obtain an interdigitated electrode, which is referred to as a MXene / Bi interdigitated electrode and is used as a drain electrode of a transistor-like ionic electronic device.

4. The preparation method according to claim 2, characterized in that: The preparation of the source electrode comprises the following steps: Step (a): centrifuging the MXene dispersion at high speed to obtain agglomerates; Step (b): evenly apply the agglomerate obtained in step (a) on a paper base and dry it for 24 to 36 hours; Step (c): laser engraving the paper base obtained in step (b) to obtain interdigitated electrodes; Step (d): electroplating zinc on the interdigital electrodes obtained in step (c). The interdigital electrodes after electroplating zinc are referred to as MXene@Zn interdigital electrodes, which are used as the source of transistor-like ionic electronic devices.

5. The preparation method according to claim 2, characterized in that: The electrolyte is prepared by dissolving poly(vinylidene fluoride-co-hexafluoropropylene), graphene oxide and ZnSO4 in dimethylformamide at a mass ratio of 10:3~4:5, stirring the suspension, ultrasonicating, evaporating and drying, and the obtained film is recorded as PVDF-HFP-GO solid electrolyte.

6. The preparation method according to claim 2, characterized in that: The FSTI device is assembled by attaching the drain and source to an insulating material, attaching an isolation layer to the surface of the drain and source, attaching an electrolyte to the isolation layer, and finally packaging with an insulating material. The drain and source are respectively connected with conductive sheets.

7. A tactile sensor, comprising the FSTI device based on a two-dimensional layered heterojunction MXene / Bi as described in claim 1 or obtained by the preparation method according to any one of claims 2 to 6.

8. Application of the FSTI device based on two-dimensional layered heterojunction MXene / Bi as described in claim 1 or obtained by the preparation method according to any one of claims 2 to 6 in a human-machine intelligent interaction system.

9. A human-machine intelligent interaction system, comprising a FSTI device based on a two-dimensional layered heterojunction MXene / Bi, a microprocessor, a manipulator, a signal filtering and amplifying module, and a power module, wherein: FSTI device, used to collect human physiological signals and sensory feedback signals of the manipulator; A microprocessor, used to process the signal data so as to analyze and evaluate the data through a preset algorithm; A robotic arm is used to simulate the gestures of a human hand and perform operations in an anthropomorphic manner; A signal filtering and amplifying module, used for amplifying and filtering the signal; The power module is used to provide electrical energy to the signal filtering and amplifying module, the microprocessor and the manipulator.

Images