Bionic dynamic nano-fluidic device for brain-like computing and regulation method

Abstract

The application discloses a kind of bionic dynamic nanofluidion devices and regulation methods for brain-like computing, the bionic dynamic nanofluidion device includes nanometer scale nanofluid channel and flexible support layer;The nanofluid channel can be dynamically bent and the inner surface of channel has distributed dynamically adjustable charge.The bionic dynamic nanofluidion device can be regulated by applying stress, pulse voltage intensity, pulse voltage quantity, positive and negative pulse voltage to bionic nanofluidion device current, capacitance and conductance, to realize the regulation, memory and storage of information.The bionic nanofluidion device has positive voltage conductance reading and writing and negative voltage conductance erasing ability under the action of voltage pulse.The bionic nanofluidion device has high plasticity, highly parallel processing, adaptive learning, ultra-low energy consumption and other advantages in brain-like computing process.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of bionic nanofluidic ionics, and particularly relates to the fields of nanotechnology, ion devices and brain-like computing. BACKGROUND

[0002] Brain-like computing aims to simulate and understand the information processing principles and mechanisms of the human brain to achieve more efficient, intelligent and adaptive computing systems, which need high-performance electronic devices to support to achieve the goal of simulating the brain to achieve efficient operation to solve the current problems faced by artificial intelligence. Currently, ion devices as a new type of electronic device have broad application prospects in this field, but the following problems faced by current ion devices limit their development and application, such as stability problems: the ion channels and ion transport materials in ion devices need to have good reliability and long-term stability. However, some ion materials may be deactivated, drifted or degraded under long-term use or high temperature conditions, limiting the service life and stability of ion devices. Responsiveness problems: the speed of ion transport is usually slow, and the response time of ion devices is longer compared with traditional electronic devices, which may limit their performance and efficiency in some application scenarios. Manufacturing technology problems: the manufacturing and integration of ion devices face some challenges. On the one hand, it is still difficult to manufacture nanoscale ion channels and ion transport materials with consistency and controllability; on the other hand, the integration problems between ion devices and traditional electronic devices, such as interface matching, signal conversion and interface optimization, also need to be solved. Energy consumption problems: although ion devices have potential energy efficiency advantages in some aspects, some ion transport processes still need to consume high energy. In addition, the power consumption and energy consumption of ion devices are closely related to factors such as size, structure and operating conditions. Scaling problems: large-scale production and commercial application of ion devices still face some challenges. From the laboratory stage to commercial production, the process needs to overcome problems such as technical difficulties, cost reduction, reliability and sustainability, so it is urgent to develop new ion devices with low cost, stability, strong controllability and low energy consumption. SUMMARY

[0003] The main purpose of the present application is to provide a bionic dynamic nano ionic device for brain-like computing. The bionic nano ionic device, as a new type of electronic device developed based on bionic nano ionic science, provides a new direction for solving the above problems. Specifically, the nano ionic channel in the living organism realizes the organic combination of regulation of information transmission process, memory and storage based on dynamic ion transmission regulation, and establishes a bridge between the biological brain and the artificial brain for developing signal transmission and translation media for brain-like computing, so as to solve the bottleneck problem of current artificial intelligence in the field of human-computer interaction and the problem of high efficiency and energy consumption of current electronic devices in the process of brain-like computing. Therefore, the bionic nano ionic device regulates ion transmission by imitating the structure and function of the nano-scale ion channel in the living organism to realize information processing, memory and storage conversion, which helps to solve the problems of poor controllability, low operation efficiency and high energy consumption of traditional ion devices in the field of brain-like computing, and promotes the development of bionic nano ionic devices in the field of brain-like computing.

[0004] The present application provides a bionic dynamic nano ionic device, which comprises a nano flow channel and a flexible support material, the inner diameter of the nano flow channel is 1-500 nm, the inner surface of the channel has electric charge, the nano flow channel will dynamically bend under the condition of external pressure, and the nano flow channel contains an electrolyte solution.

[0005] The nano flow channel in the bionic nano ionic device dynamically bends under external pressure, and protons change the dynamic confinement space to realize dynamic regulation of channel current, conductance and capacitance.

[0006] The bionic nano ionic device produces current memory hysteresis curve and conductance memory hysteresis curve under the action of alternating voltage.

[0007] The bionic nano ionic device has the ability of reading and writing of positive voltage conductance and erasing ability of negative voltage conductance under the action of voltage pulse.

[0008] Preferably, the inner diameter of the nano flow channel in the bionic nano ionic device is 1-10 nm.

[0009] Preferably, the flexible support material in the bionic nano ionic device is one or more of polydimethylsiloxane, epoxy resin, polyvinyl alcohol, polyester or polynaphthalene dimethyl glycol ester. The preferred curing temperature of the flexible support material is 50-100 DEG C, and the curing time is 2-24 h.

[0010] Preferably, the nanofluidic channel base structure in the biomimetic nanofluidic device is one or more of single-walled carbon nanotubes, double-walled carbon nanotubes, multi-walled carbon nanotubes, titanium dioxide nanotubes, titanate nanotubes, and titania nanotubes. The inner diameter of the nanotube channel is preferably 10-500 nm. In another preferred embodiment of the present application, the nanofluidic channel material comprises one or more of single-walled carbon nanotubes, double-walled carbon nanotubes, multi-walled carbon nanotubes, titanium dioxide nanotubes, titanate nanotubes, and titania nanotubes. The inner diameter of the nanofluidic channel is 1-10 nm.

[0011] Preferably, the ion activation power of the biomimetic nanofluidic ion nanofluidic channel is 50-300 W, and the activation time is 5-30 min.

[0012] Preferably, the biomimetic nanofluidic ion nanofluidic channel is internally modified with an organic polyelectrolyte polymer.

[0013] Preferably, the organic polyelectrolyte polymer comprises one or more of polydopamine, polyacrylamide, polyacrylic acid, and polystyrene sulfonic acid, with a concentration of 0.1 g / L-5 g / L and a reaction time of 4-24 h.

[0014] Optionally, the nanofluidic channel modification method comprises, but is not limited to, free radical polymerization, chemical grafting, interfacial binding, layer-by-layer polymerization, and chemical deposition.

[0015] In a preferred embodiment of the present application, the nanofluidic channel in the biomimetic nanofluidic ion device is one or more.

[0016] In a preferred embodiment of the present application, the concentration of the electrolyte solution is 0.1 mM-1000 mM.

[0017] In a preferred embodiment of the present application, the applied voltage intensity is +0.1-+10.0 V.

[0018] In a preferred embodiment of the present application, the applied pressure is 1000-100000 Pa.

[0019] The initiator of the organic polyelectrolyte polymer comprises inorganic initiators such as ammonium persulfate, potassium persulfate, and sodium persulfate, and organic initiators such as benzoyl peroxide, di-tert-butyl peroxide, and methyl ethyl ketone peroxide, with a concentration of 0.01 g / L-1 g / L.

[0020] Another object of the present application is to provide a regulation method based on the biomimetic dynamic nanofluidic device, wherein the inner surface of the nanofluidic channel is in an electrolyte solution and has positive or negative charge, the current, conductance, capacitance, etc. of the biomimetic dynamic nanofluidic device are regulated by applying stress, adjusting proton concentration, pulse voltage, and alternating voltage, the regulation mechanism such as presynaptic potential, postsynaptic potential, and synaptic weight of brain neurons is simulated, and efficient brain-like computing is realized.

[0021] In the preferred embodiment of the present application, the conductance of the biomimetic nanofluidic device is regulated by applying stress, proton concentration, pulse voltage intensity, pulse voltage quantity, and positive and negative pulse voltage, and adaptive learning, highly parallel processing, ultra-low energy consumption, plasticity, and multi-modal perception of brain-like computing are realized.

[0022] Optionally, the accuracy of the biomimetic nanofluidic device in brain-like computing process is above 90%.

[0023] The advantages of the biomimetic nanofluidic device of the present application compared with current ion devices are as follows:

[0024] 1. The present application is based on the biomimetic dynamic nanochannel in nanoscale confined space, which can simulate the ion transmission behavior between neuron synapses and realize the transmission, memory, and storage of information.

[0025] 2. The biomimetic dynamic nanochannel of the present application can simulate the dynamic bending function of biological channels and realize the dynamic regulation of ion transmission.

[0026] 3. The biomimetic nanofluidic device of the present application has the ability of positive voltage conductance reading and writing and negative voltage conductance erasing under the action of voltage pulse, and realizes memory and forgetting in the process of brain-like computing.

[0027] 4. The biomimetic nanofluidic device of the present application has high recognition efficiency for handwritten digit recognition in the process of brain-like computing based on neural network, which is above 90%.

[0028] 5. The biomimetic nanofluidic device of the present application can regulate the conductance of the biomimetic nanofluidic device by applying stress, pulse voltage intensity, pulse voltage quantity, and positive and negative pulse voltage, and realize adaptive learning, anti-forgetting characteristics, ultra-low energy consumption, plasticity, and multi-modal perception of brain-like computing. BRIEF DESCRIPTION OF DRAWINGS

[0029] The present application will be further described below in combination with the drawings and examples.

[0030] Figure 1 It is a schematic diagram of the bending and recovery process of the biomimetic dynamic nanofluidic channel under the action of external pressure.

[0031] Figure 2 Current-voltage memory hysteresis curve of the biomimetic dynamic nanofluidic device for Example 1.

[0032] Figure 3 Conductance-voltage memory hysteresis curve of the biomimetic dynamic nanofluidic device for Example 1.

[0033] Figure 4 Resistance writing curve of the biomimetic dynamic nanofluidic device for Example 2 under the action of pulse voltage.

[0034] Figure 5 Resistance erasing curve of the biomimetic dynamic nanofluidic device for Example 2 under the action of pulse voltage.

[0035] Figure 6 Brain-computer computing result of the biomimetic dynamic nanofluidic device for Example 2 in handwritten digit recognition.

[0036] Figure 7 The dynamic neural network constructed based on the biomimetic dynamic nanofluidic device for Example 2 has dynamic anti-forgetting property compared with the conventional neural network. DETAILED DESCRIPTION

[0037] The present application is further explained with reference to the accompanying drawings and specific examples. The drawings of the present application are only schematic and the specific proportions of the various elements in the drawings can be altered in accordance with design requirements.

[0038] (1) Nanotube channel device preparation

[0039] A square silicon wafer with a width of 1 cm is used as a growth substrate, and then acetone and alcohol are used for backwashing and cleaning twice and dried by blowing. Then, the carbon nanotube array growth substrate is modified, the development conditions are screened, and the improvement and optimization of the condition parameters such as metal particle evaporation are carried out. Through chemical vapor deposition method, nanotube nanofluidic channel array with different materials and different inner diameters is obtained. Then, the plasma activation equipment is used for activation treatment. Subsequently, the flexible polymer is poured in the gap outside the nanotube wall and solidified by high temperature. Finally, the cold ultrathin sectioning method is used to remove the excess polymer and nanotube at both ends of the channel, and a bidirectional conducting nanofluidic channel assembly is obtained.

[0040] (2) Performance characterization and results of nanotube nanofluidic channel device

[0041] ① Device memory performance characterization and results: The ion controllable transmission test characterization is carried out by electrochemical method. The designed control system is divided into two modules, which are symmetrical structure, and the biomimetic nanofluidic device is placed in the middle. A silver / silver chloride electrode is placed on the left and right sides of the upper part, and a pressure sensor is connected to the right side of the assembly. Then different concentrations of electrolyte solution are injected into the interior. Under different pH proton concentration conditions, the current and conductance changes of the device are measured by applying 100Hz, 0.1V AC voltage on the electrode. Then a +1V pulse voltage is applied on the electrode for 10s, and after 10s, a +0.1V reading voltage is applied for 10s as a pulse writing process. A-1V pulse voltage is applied for 10s, and after 10s, a-0.1V reading voltage is applied for 10s as a pulse erasing process. Finally, the device electrode is pulsed with +0.1V for 5 times, and then a +1V pulse voltage is applied once, and then the pulse voltage +0.1V is applied.

[0042] ② Device brain-like computing characterization and results: Finally, the device is embedded as a neural network node into the neural network brain-like computing process to perform handwritten digital recognition operation.

[0043] Example 1

[0044] (1) Preparation of biomimetic nanofluidic device nanofluidic channel

[0045] A square silicon wafer with a width of 1cm is used as the growth substrate, and then acetone and alcohol are used for reverse washing twice and blow-drying. Then the carbon nanotube array growth substrate is modified, the development conditions are selected, and the metal particle evaporation conditions are improved and optimized. Through chemical vapor deposition method, a multi-walled carbon nanotube nanofluidic channel array with an inner diameter of 100nm is obtained. Then the plasma activation equipment is activated at a power of 200W for 5min. Then the polyepoxy resin is poured into the gap outside the carbon nanotube, and then it is solidified at a high temperature of 60℃ for 3h. Finally, the cold section method is used to remove the excess epoxy resin and carbon nanotubes at both ends of the channel, and a bidirectional conducting nanofluidic channel assembly is obtained.

[0046] (2) Polyelectrolyte modification of the inner surface of the biomimetic nanofluidic device nanofluidic channel

[0047] The inner surface of the nanofluidic channel is modified by free radical polymerization. Specifically, the flexible carbon nanotube channel is cleaned with ethanol and soaked in 0.1M dilute hydrochloric acid for 24h to remove the residual product on the surface. Then, it is placed in a plasma cleaning machine for activation at a power of 150W for 5min. Subsequently, it is immediately placed in a Tris buffer mixed solution containing 1.5g / L dopamine and 0.5g / L polyethyleneimine for ultrasonic treatment for 4min and then left for 12h. Then, the unreacted dopamine and polyethyleneimine are removed with pure water. The device is added to a 4g / L polyacrylic acid and 1g / L ammonium persulfate solution for 12h to obtain the prepared biomimetic nanofluidic ion channel device.

[0048] (3) Performance characterization and results of the biomimetic nanofluidic ion device

[0049] ① Device memory performance characterization and results: The ion controllable transmission test is characterized by electrochemical method. The designed control system is divided into two modules, which are symmetrical structures. The biomimetic nanofluidic ion device is placed in the middle. One silver / silver chloride electrode is placed on each side of the upper part. A pressure sensor is connected to the right side of the assembly. Then, 0.1mM electrolyte solution is injected into the interior. Under the condition of pH=4, the current and conductance change of the device are measured by applying 100Hz, +0.1V alternating voltage on the electrode, generating a memory hysteresis curve, which shows the memory function. Then, a +1V pulse voltage is applied on the electrode for 10s, and a +0.1V reading voltage is applied for 10s after 10s interval, which is a pulse writing process. The results show that the conductance of the device increases from 156nS to 276nS, which shows that the conductance increases with the increase of the number of pulses. A-1V pulse voltage is applied for 10s, and a-0.1V reading voltage is applied for 10s after 10s interval, which is a pulse erasing process. The results show that the conductance of the device decreases from 376nS to 156nS, which shows that the conductance of the device gradually decreases with the increase of the number of pulse erasing processes. The device shows storage and erasing functions. Finally, the electrode of the device is applied with a pulse voltage of +0.1V for 5 times, and the conductance is 123nS. Then, a pulse voltage of +1V is applied once, and then a pulse voltage of +0.1V is applied, and the conductance is changed to 136nS. The test shows that the conductance of the device on the surface of the structure is enhanced after the application of 1V pulse voltage, and the device shows memory enhancement effect.

[0050] ② Device brain-like computing characterization and results: Finally, the device is embedded as a neural network node into the neural network brain-like computing process to perform handwritten digital recognition operation. The operation results show that the operation accuracy of the device reaches 81%.

[0051] Example 2

[0052] (1) Preparation of nanofluidic channel of biomimetic nanofluidic ion device

[0053] The square silicon wafer with a width of 1 cm is used as a growth substrate, and then cleaned twice by reverse washing with acetone and alcohol and dried, and then the carbon nanotube array growth substrate is modified, the development conditions are selected, and the improvement and optimization of the condition parameters such as metal particle evaporation are performed, so that the multi-walled carbon nanotube nanoflow channel array with an inner diameter of 200 nm is obtained. The plasma activation equipment is used to activate for 5 min under the condition of 150 functions, then the polyepoxy resin is poured into the gap outside the carbon nanotube, and is cured at a high temperature of 60°C for 18 h, finally the excess polydimethylsiloxane and carbon nanotube at both ends of the channel are removed by freezing section method, and the bidirectional nanoflow channel assembly is obtained.

[0054] (2) Modification of the inner surface of the biomimetic nanoflow ion device nanoflow channel with polyelectrolyte

[0055] The inner surface of the nanoflow channel is modified by chemical grafting method, specifically, the flexible carbon nanotube channel is first cleaned with ethanol and immersed in 0.1M dilute hydrochloric acid for 24 h to remove the residual product on the surface, then it is placed in the plasma cleaning machine for 15 min of plasma activation treatment at a power of 75 W, then it is immediately placed in a 2g / L dopamine Tris buffer mixed solution for ultrasonic treatment for 10 min and then static treatment for 24 h, then the unreacted dopamine is removed with pure water, and the device is added into a 2g / L sodium styrene sulfonate polyacrylic acid and 0.01g / L potassium persulfate solution for 6 h to obtain the prepared biomimetic nanoflow ion channel device.

[0056] (3) Performance characterization and results of the biomimetic nanoflow ion device

[0057] ①Device memory performance characterization and results: The ion controllable transport test characterization is carried out by electrochemical method. The designed control system is divided into two modules, which are symmetrical structure, and the biomimetic nanofluidic device is placed in the middle. A silver / silver chloride electrode is placed on each side of the upper part, a pressure sensor is connected to the right side of the assembly, and a pressure of 10000 Pa is applied. Then 1 mM electrolyte solution is injected into the interior. Under the condition of pH = 4, the current and conductance changes of the device are measured by applying 100 Hz, +0.1 V alternating voltage on the electrode, and the memory hysteresis curve is generated, which shows the memory function. Then a +1 V pulse voltage is applied on the electrode for 10 s, and a +0.1 V reading voltage is applied for 10 s after an interval of 10 s, which is a pulse writing process. The results show that the conductance of the device increases from 95 nS to 176 nS, which shows that the conductance increases with the increase of the number of pulses. A -1 V pulse voltage is applied for 10 s, and a -0.1 V reading voltage is applied for 10 s after an interval of 10 s, which is a pulse erasing process. The results show that the conductance of the device decreases from 176 nS to 96 nS, which shows that the conductance gradually decreases with the increase of the number of pulse erasing processes. The device shows storage and erasing functions. Finally, the electrode time of the device is +0.1 V pulse voltage for 5 times, and the conductance is 83 nS. Then a +1 V pulse voltage is applied once, and then a pulse voltage of +0.1 V is applied, and the conductance is changed to 119 nS. The test shows that the conductance of the device on the surface structure is enhanced after the application of +1 V pulse voltage, and the device shows memory enhancement effect.

[0058] ②Device brain-like computing characterization and results: Finally, the device is embedded as a neural network node into the neural network brain-like computing process to perform handwritten digital recognition operation. The operation results show that the operation accuracy of the device reaches 89%.

[0059] Example 3

[0060] (1) Preparation of nanofluidic channel of biomimetic nanofluidic device

[0061] A square silicon wafer with a width of 1 cm is used as a growth substrate, and then acetone and alcohol are used for reverse washing twice and blow-drying. Then, by modifying the carbon nanotube array growth substrate, screening the developing conditions, and improving and optimizing the condition parameters such as metal particle evaporation, a multi-walled carbon nanotube nanofluidic channel array with an inner diameter of 8 nm is obtained. Then, the plasma activation equipment is activated at 50 W for 30 min, and then the polyepoxy resin is poured into the gap outside the carbon nanotube wall and solidified at 50°C for 24 h. Finally, the excess polydimethylsiloxane and carbon nanotubes at both ends of the channel are removed by freeze sectioning method, and a bidirectional conducting nanofluidic channel assembly is obtained.

[0062] (2) Polyelectrolyte modification of the inner surface of the nanofluidic channel of the biomimetic nanofluidic device

[0063] Firstly, the flexible carbon nanotube channel was cleaned with ethanol and immersed in 0.1M dilute hydrochloric acid for 24h to remove the residual product on the surface, and then was placed in a plasma cleaning machine with an activation power of 50W for 25min. Subsequently, it was immediately placed in a Tris buffer mixed solution containing 2g / L polyethyleneimine and ultrasonicated for 4min, and then rested for 12h. The unreacted dopamine and polyethyleneimine were removed with pure water. The device was added to a solution of 0.1g / L acrylic acid and 0.2g / L di-tert-butyl peroxide for 24h to obtain the prepared biomimetic nanofluidic channel device.

[0064] (3) Performance characterization and results of the biomimetic nanofluidic device

[0065] ① Device memory performance characterization and results: The ion controllable transmission test characterization was carried out by electrochemical method. The designed control system was divided into two modules, which were symmetrical structure, and the biomimetic nanofluidic device was placed in the middle. One silver / silver chloride electrode was placed on each side of the upper part, a pressure sensor was connected to the right side of the assembly, a pressure of 1000Pa was applied, and then 10mM electrolyte solution was injected into the inside. Under the condition of pH=4, the current and conductance change of the device were measured by applying 100Hz, +0.1V alternating voltage on the electrode, a memory hysteresis curve was generated, and the memory function was exhibited. Subsequently, a +1V pulse voltage was applied on the electrode for 10s, and then a +0.1V reading voltage was applied for 10s with an interval of 10s as a pulse writing process. The results showed that the conductance of the device increased from 76nS to 141nS, which exhibited that the conductance increased with the increase of the number of pulses. A-1V pulse voltage was applied for 10s, and then a-0.1V reading voltage was applied for 10s with an interval of 10s as a pulse erasing process. The results showed that the conductance of the device decreased from 140nS to 74nS, which exhibited that the conductance of the device gradually decreased with the increase of the number of pulse erasing processes. The device exhibited storage and erasing functions. Finally, the electrode time of the device was 0.1V pulse voltage for 5 times, and the conductance was 63nS. A +1V pulse voltage was applied, and then the pulse voltage was changed to 0.1V, and the conductance was changed to 78nS. The test showed that the conductance of the device on the surface was enhanced after the application of +1V pulse voltage, and the device exhibited memory enhancement effect.

[0066] ② Device brain-like computing characterization and results: Finally, the device was embedded as a neural network node into the neural network brain-like computing process to perform handwritten digital recognition operation. The operation results showed that the operation accuracy of the device reached 96%.

[0067] Example 4

[0068] (1) Preparation of the biomimetic nanofluidic device nanofluidic channel

[0069] The square silicon wafer with width of 1 cm was used as the growth substrate, and then cleaned twice by acetone and alcohol backwashing and dried, and then the carbon nanotube array growth substrate was modified, the development conditions were selected, and the improvement and optimization of the condition parameters such as metal particle evaporation were carried out, and the titanium oxide nanotube nanoflow channel array with an inner diameter of 500 nm was obtained, and then the polyepoxy resin was poured into the gap outside the carbon nanotube under the condition of 300 W plasma activation equipment for 12 min, and then cured at 100°C for 12 h, and finally the excess polydimethylsiloxane and carbon nanotube at both ends of the channel were removed by freezing section method, and the bidirectional conducting nanoflow channel assembly was obtained.

[0070] (2) Modification of the inner surface of the biomimetic nanoflow ion device nanoflow channel with polyelectrolyte

[0071] First, the flexible carbon nanotube channel was cleaned with ethanol and immersed in 0.1M dilute hydrochloric acid for 24h to remove the residual product on the surface, and then placed in a plasma cleaning machine with an activation power of 300W for 5min, and then immediately placed in a Tris buffer mixed solution containing 0.5g / L dopamine and 1.5g / L polyethyleneimine for ultrasonic treatment for 10min and then static for 24h, and then the unreacted dopamine and polyethyleneimine were removed with pure water, and the device was added into a solution of 0.75g / L styrene sulfonic acid and 0.9g / L benzoyl peroxide for 24h to obtain the prepared biomimetic nanoflow ion channel device.

[0072] (3) Performance characterization and results of the biomimetic nanoflow ion device

[0073] ①Device memory performance characterization and results: The ion controllable transport test characterization is carried out by electrochemical method. The designed control system is divided into two modules, which are symmetrical structure, and the biomimetic nanofluidic device is placed in the middle. A silver / silver chloride electrode is placed on the left and right sides of the upper part, and a pressure sensor is connected to the right side of the assembly. A pressure of 1000 Pa is applied, and then 1 mM electrolyte solution is injected into the interior. Under the condition of pH = 4, the current and conductance changes of the device are measured by applying 100 Hz, +0.1 V alternating voltage on the electrode, and the memory hysteresis curve is generated, which shows the memory function. Then a +1 V pulse voltage is applied on the electrode for 10 s, and then a +0.1 V reading voltage is applied for 10 s after an interval of 10 s, which is a pulse writing process. The results show that the conductance of the device increases from 156 nS to 193 nS, which shows that the conductance increases with the increase of the number of pulses. A -1 V pulse voltage is applied for 10 s, and then a -0.1 V reading voltage is applied for 10 s after an interval of 10 s, which is a pulse erasing process. The results show that the conductance of the device decreases from 190 nS to 154 nS, which shows that the conductance gradually decreases with the increase of the number of pulse erasing processes. The device shows storage and erasing functions. Finally, the electrode time of the device is +0.1 V pulse voltage for 5 times, and the conductance is 151 nS. Then a +1 V pulse voltage is applied once, and then a pulse voltage of +0.1 V is applied, and the conductance changes to 169 nS. The test shows that the conductance of the device on the surface structure is enhanced after the application of +1 V pulse voltage, and the device shows memory enhancement effect.

[0074] ②Device brain-like computing characterization and results: Finally, the device is embedded as a neural network node into the neural network brain-like computing process to perform handwritten digital recognition operation. The operation results show that the operation accuracy of the device reaches 91%.

[0075] Example 5

[0076] (1) Preparation of nanofluidic channel of biomimetic nanofluidic device

[0077] A square silicon wafer with a width of 1 cm is used as a growth substrate, and then acetone and alcohol are used for reverse washing twice and blow-drying. Then the carbon nanotube array growth substrate is modified, the development conditions are selected, and the metal particle evaporation conditions are improved and optimized. A titanium dioxide nanotube nanofluidic channel array with an inner diameter of 100 nm is obtained. Then the plasma activation equipment is used for activation at a power of 120 W for 8 min. Then polyethylene glycol is poured into the gap outside the titanium dioxide nanotube, and high-temperature solidification is carried out at 70°C for 12 h. Finally, the excess polydimethylsiloxane and titanium dioxide nanotube at both ends of the channel are removed by freeze sectioning method, and a bidirectional conducting nanofluidic channel assembly is obtained.

[0078] (2) Polyelectrolyte modification of the inner surface of the nanofluidic channel of the biomimetic nanofluidic device

[0079] Firstly, the flexible carbon nanotube channel was cleaned with ethanol and immersed in 0.1M dilute hydrochloric acid for 24h to remove the residual product on the surface, and then was placed in a plasma cleaning machine with an activation power of 60W for 18min, and then immediately placed in a mixed solution of 1.9g / L dopamine and 0.1g / L polyethyleneimine in Tris buffer for 2min of ultrasonic treatment and 4h of standing, and then the unreacted dopamine and polyethyleneimine were removed with pure water. The device was added to a solution of 2.5g / L acrylic acid and 0.05g / L methyl ethyl ketone peroxide for 18h to obtain the prepared biomimetic nanofluidic channel.

[0080] (3) Performance characterization and results of the biomimetic nanofluidic device

[0081] ① Device memory performance characterization and results: The ion controllable transport test was characterized by electrochemical method. The designed control system was divided into two modules, which were symmetrical structure, and the biomimetic nanofluidic device was placed in the middle. One silver / silver chloride electrode was placed on each side of the upper part, a pressure sensor was connected to the right side of the assembly, a pressure of 50000Pa was applied, and then 10mM electrolyte solution was injected into the inside. Under the condition of pH=4, the current and conductance change of the device were measured by applying 100Hz, +0.1V alternating voltage on the electrode, a memory hysteresis curve was generated, and the memory function was exhibited. Then a pulse voltage of +1V was applied on the electrode for 10s, and a reading voltage of 0.1V was applied for 10s after 10s interval, which was a pulse writing process. The results showed that the conductance of the device increased from 91nS to 115nS, which exhibited that the conductance increased with the increase of pulse number. A pulse voltage of-1V was applied for 10s, and a reading voltage of-0.1V was applied for 10s after 10s interval, which was a pulse erasing process. The results showed that the conductance of the device decreased from 116nS to 90nS, which exhibited that the conductance gradually decreased with the increase of pulse erasing process number. The device exhibited storage and erasing functions. Finally, a pulse voltage of +0.1V was applied on the electrode for 5 times, the conductance was 78nS, and then a pulse voltage of +1V was applied once, and then the pulse voltage was changed to +0.1V, and the conductance was changed to 96nS. The test showed that the conductance of the device on the surface of the structure increased after the application of a pulse voltage of +1V, and the device exhibited memory enhancement effect.

[0082] ② Device brain-like computing characterization and results: Finally, the device was embedded as a neural network node into the neural network brain-like computing process to perform handwritten digital recognition operation. The operation results showed that the operation accuracy of the device reached 94%.

[0083] Comparative Example 1

[0084] (1) Preparation of nanofluidic channel of the biomimetic nanofluidic device

[0085] The square silicon wafer with a width of 1 cm was used as a growth substrate, and then cleaned twice by reverse washing with acetone and alcohol and dried by blowing, and then the carbon nanotube array growth substrate was modified, the development conditions were screened, and the condition parameters such as metal particle evaporation were improved and optimized, and then a multi-walled carbon nanotube nanoflow channel array with an inner diameter of 8 nm was obtained by chemical vapor deposition, and then the polydimethylsiloxane was poured into the gap outside the carbon nanotube wall by using a plasma activation device under the condition of 100 W power for 10 min, and then the polydimethylsiloxane and the carbon nanotube were removed by the freeze ultra-thin sectioning method, and finally a bidirectional nanoflow channel assembly was obtained.

[0086] (2) Performance characterization and results of the carbon nanotube channel device

[0087] ① Device memory performance characterization and results: The ion controllable transmission test characterization was carried out by an electrochemical method. The designed control system is symmetrical structure, and the biomimetic nanoflow ion device is placed in the middle. Each of the upper left and right sides is placed into a silver / silver chloride electrode, a pressure sensor is connected to the right side of the assembly, different pressures are applied, and then 100 mM electrolyte solution is injected into the inside. Under the condition of pH = 4, the current and conductance change of the device were measured by applying 100 Hz, 0.1 V alternating voltage on the electrode, the results showed that the current-voltage was linear, no memory hysteresis curve was generated, then a pulse writing process was carried out by applying +1 V pulse voltage on the electrode for 10 s, and then applying +0.1 V reading voltage for 10 s with an interval of 10 s. The research results showed that the current remained at 12 μA, the conductance was 1.2×10 -5 S, and did not increase with the increase of the number of pulse processes, the results showed that the conductance did not change; a pulse erasing process was carried out by applying -1 V pulse voltage for 10 s, and then applying -0.1 V reading voltage for 10 s with an interval of 10 s, the current still remained at 12 μA, the conductance was 1.2×10 -5 S, and did not increase with the increase of the number of pulse processes, the device showed no information storage and erasing function. Finally, the device electrode was applied with pulse voltage +0.1 V for 5 times, and then applied with pulse voltage +1 V once, and then the pulse voltage was changed to +0.1 V, the test results showed that the conductance of the device did not change after applying pulse voltage +1 V, and the device showed no memory enhancement effect.

[0088] ② Device brain-like computing characterization and results: Finally, the device was embedded as a neural network node into the neural network brain-like computing process, and the handwritten digital recognition operation was carried out, the operation results showed that the operation accuracy rate of the device reached 0%, which was mainly because the device had no memory, information storage and control function.

[0089] Comparative Example 2

[0090] (4) Preparation of nanofluidic channel of biomimetic nanofluidic device

[0091] The square silicon wafer with a width of 1 cm was used as a growth substrate, and then cleaned by acetone and alcohol backwashing twice and dried. Then, the carbon nanotube array growth substrate was modified, the development conditions were screened, and the improvement and optimization of the condition parameters such as metal particle evaporation were performed, so that the multi-walled carbon nanotube nanofluidic channel array with an inner diameter of 8 nm was obtained. Then, the plasma activation equipment was used to activate for 30 min under the condition of 50 W. Subsequently, the polyepoxy resin was poured into the gap outside the carbon nanotube, and then cured at a high temperature of 60°C for 24 h. Finally, the excess polydimethylsiloxane and carbon nanotube at both ends of the channel were removed by freezing sectioning method, and the bidirectional conducting nanofluidic channel assembly was obtained.

[0092] (5) Inner surface polyelectrolyte modification of nanofluidic channel of biomimetic nanofluidic device

[0093] First, the flexible carbon nanotube channel was cleaned with ethanol and immersed in 0.1M dilute hydrochloric acid for 24 h to remove the residual product on the surface. Then, it was placed in the plasma cleaning machine and activated at a power of 50 W for 25 min. Subsequently, it was immediately placed in a Tris buffer mixed solution containing 2 g / L polyethyleneimine and ultrasonicated for 4 min, and then left for 12 h. Then, the unreacted dopamine and polyethyleneimine were removed with pure water. The device was added into a solution containing 0.1 g / L acrylic acid and 0.2 g / L di-t-butyl peroxide for 24 h, and the prepared biomimetic nanofluidic channel device was obtained.

[0094] (6) Performance characterization and results of biomimetic nanofluidic device

[0095] Device memory performance characterization and results: The ion controllable transmission test characterization was performed by electrochemical method. The designed control system

[0096] The device is divided into two modules, left and right symmetrical structure, the middle is placed in bionic nanofluidic device, the upper left and right sides each put into a silver / silver chloride electrode, the right side of the assembly is connected with pressure sensor, no pressure is applied, then 0.1 mM electrolyte solution is injected into the inside. Under the condition of pH = 7, the current and conductance change of the device are measured by applying 100 Hz, +0.1 V alternating voltage on the electrode, no memory hysteresis curve appears, indicating that there is no memory function; then +1 V pulse voltage is applied on the electrode for 10 s, and then +0.1 V reading voltage is applied for 10 s with an interval of 10 s as a pulse writing process, the results show that the conductance of the device is 76 nS and remains unchanged; -1 V pulse voltage is applied for 10 s, and then -0.1 V reading voltage is applied for 10 s with an interval of 10 s as a pulse erasing process, the results show that the conductance of the device is 74 nS and remains stable, the results show that the device has no storage and erasing function. Finally, the electrode of the device is applied with 0.1 V pulse voltage for 5 times, the conductance is 63 nS, and then +1 V pulse voltage is applied once, then the pulse voltage is changed to 0.1 V, and the conductance is changed to 64 nS, the test structure shows that the conductance of the device almost does not change after the application of +1 V pulse voltage, and the device has no memory enhancement effect.

[0097] Device brain-like computing representation and results: finally, the device is embedded into the neural network brain-like computing process as a neural network node

[0098] In the middle, the handwritten digital recognition operation is carried out, and the operation result shows that the operation accuracy of the device reaches 90%, which is mainly because the device has no memory, information storage and regulation function.

[0099] The above is only the preferred embodiment of the present application, therefore cannot limit the scope of the present application, that is, equivalent changes and modifications made according to the scope and content of the present application should still be within the scope of the present application.

Claims

1. A biomimetic dynamic nano-fluidic device for brain-like computing, characterized in that: The application relates to a biomimetic nanofluidic ion device. The biomimetic nanofluidic ion device comprises a nanofluidic channel and a flexible support material, the nanofluidic channel has an inner diameter of 1-500 nm, the nanofluidic channel of the biomimetic nanofluidic ion device is internally decorated with an organic polyelectrolyte, and the inner surface of the channel has electric charges; the nanofluidic channel can be dynamically bent under the condition of external pressure, and the nanofluidic channel contains an electrolyte solution.

2. The bio-inspired dynamic nanofluidic device for brain-like computing of claim 1, wherein: The organic polyelectrolyte comprises one or more of polydopamine, polyacrylamide, polyacrylic acid and polystyrene sulfonic acid.

3. The bio-inspired dynamic nanofluidic device for brain-like computing of claim 1, wherein: The nanofluidic channel is dynamically bent under external pressure, and protons change the dynamic confinement space to realize dynamic regulation of channel current, conductance and capacitance.

4. The bio-inspired dynamic nanofluidic device for brain-like computing of claim 1, wherein: The biomimetic nanofluidic ion device generates current memory hysteresis curves and conductance memory hysteresis curves under the action of an alternating voltage.

5. The bio-inspired dynamic nanofluidic device for brain-like computing of claim 1, wherein: The biomimetic nanofluidic ion device has positive voltage conductance reading and writing and negative voltage conductance erasing ability under the action of a voltage pulse.

6. The bio-inspired dynamic nanofluidic device for brain-like computing of claim 1, wherein: The flexible support material comprises at least one or more of polydimethylsiloxane, epoxy resin, polyvinyl alcohol, polyester or polynaphthalene dimethyl glycol ester.

7. The bio-inspired dynamic nanofluidic device for brain-like computing of claim 6, wherein: The curing temperature of the flexible support material is 50-100 DEG C, and the curing time is 2-24 hours.

8. The bio-inspired dynamic nanofluidic device for brain-like computing of claim 1, wherein: The nanofluidic channel base structure is one or more of a single-walled carbon nanotube, a double-walled carbon nanotube, a multi-walled carbon nanotube, a titanium dioxide nanotube, a titanate nanotube and a titania nanotube. 9.A regulation method based on a biomimetic dynamic nanofluidic ion device, wherein the nanofluidic channel of the biomimetic dynamic nanofluidic ion device has an inner diameter of 1-10 nm, the inner surface of the nanofluidic channel is positively or negatively charged in an electrolyte solution, and the current, conductance and capacitance of the biomimetic dynamic nanofluidic ion device are regulated by applying stress, adjusting the proton concentration, pulsing voltage and alternating voltage to simulate the presynaptic potential, postsynaptic potential and synaptic weight regulation mechanism of brain neurons.

Images