Mouse spinal cord flexible electrode and stimulation parameter optimization method

Abstract

The invention discloses a mouse spinal cord flexible electrode and a stimulation parameter optimization method, and relates to the technical field of parameter optimization. The mouse spinal cord flexible electrode comprises a soluble insertion layer, a gradient substrate layer, spinal cord anastomosis ribs, a conductive snakelike net, a hemispherical touch dot matrix, a microfluidic layer, an optical waveguide layer and a top packaging layer; the soluble insertion layer is photo-crosslinked PEG-DMA with the thickness of 5-25 m, the Young modulus is kept within the range of 0.8-1.5 GPa before implantation, and the soluble insertion layer is hydrolyzed in cerebrospinal fluid within 0.5-3 hours after implantation and completely falls off; the spinal cord anastomosis rib comprises an expandable S-shaped folding structure arranged on the upper surface of the gradient substrate layer, the distance between every two adjacent ribs ranges from 80 m to 200 m, the height of the folded state is not larger than 15 m, the height of the unfolded state ranges from 15 m to 40 m, and the spinal cord anastomosis rib is used for forming a cerebro-spinalfluid bypass groove extending in the longitudinal direction of the spinal cord on the surface of the dura mater. According to the invention, the positioning precision, safety and long-term efficiency of mouse spinal cord electrical stimulation can be effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of parameter optimization, in particular to a mouse spinal cord flexible electrode and stimulation parameter optimization method. BACKGROUND

[0002] In the field of small animal neuroscience research, especially in the application of spinal cord neural signal regulation, mouse spinal cord pathological modeling and therapeutic neural stimulation verification, spinal cord surface implantable flexible electrodes have become a basic tool. Such flexible electrodes are closely attached to the surface of the spinal cord dura mater to complete the perception of nerve excitation and current injection, thereby achieving the purpose of neural regulation. However, with the increasing requirements for the application accuracy and long-term stability of neural interfaces, the existing flexible electrode system has shown some technical bottlenecks in many aspects, which seriously limits its further development in high resolution, biological safety and controllability.

[0003] The currently disclosed flexible spinal cord electrode system generally uses a flexible substrate material based on polyimide, parylene or silicone rubber, combined with metal wires and micro contact electrodes to form a two-dimensional stimulation or recording array. The flexibility of this type of electrode is improved compared to traditional rigid microneedle arrays, and it can indeed achieve basic signal transmission function under short-term experimental conditions. However, since the implantation operation often needs to rely on additional rigid support tools, such as temporary insertion sheets made of metal or silicon, and additional mechanical fixation platforms, not only does it complicate the operation steps, but also has a high risk during the operation process, such as electrode dislocation, spinal cord compression or puncture damage. In addition, after the electrode is implanted, under the small deformation caused by spinal cord movement (such as respiratory movement, muscle stretching, etc.), it lacks effective three-dimensional structural adaptability, resulting in repeated micro-displacement of the fitting surface, gradually triggering inflammatory reactions and inducing glial cell proliferation, eventually forming a chronic degradation phenomenon of increased electrochemical impedance and reduced charge injection capacity. SUMMARY

[0004] To solve the above technical problems, the mouse spinal cord flexible electrode based on the present application can effectively improve the positioning accuracy, safety and long-term effectiveness of mouse spinal cord electrical stimulation.

[0005] To achieve the above purposes, the technical solution adopted by the present application is: The mouse spinal cord flexible electrode comprises, in a bottom-to-top stacking order: a soluble insertion layer, a gradient base layer, a spinal cord anastomosis rib, a conductive serpentine net, a hemispherical contact array, a microfluidic layer, an optical waveguide layer, and a top packaging layer; the soluble insertion layer is 5-25 µm thick photo-crosslinked PEG-DMA, maintains a Young's modulus range of 0.8-1.5 GPa before implantation, and hydrolyzes in the cerebrospinal fluid and completely falls off within 0.5-3 hours after implantation; the spinal cord anastomosis rib is an unfoldable S-shaped folded structure arranged on the upper surface of the gradient base layer, the spacing between adjacent ribs ranges from 80-200 µm, the folded height is not greater than 15 µm, and the unfolded height ranges from 15-40 µm, and is used to form a cerebro-spinal fluid bypass groove extending along the longitudinal direction of the spinal cord on the surface of the dura mater; the conductive serpentine net is composed of a 0.8-1.5 µm thick Au / Cr metal layer, the bidirectional curvature radius of the serpentine trace is not less than 40 µm, the trace line width transitions from 4-8 µm at the center to 10-15 µm at the edge, and a 100-300 nm TiN nanometer island layer is formed on the metal surface by in-situ plasma etching; the hemispherical contact array is located at the reserved window of the conductive serpentine net, and there are 8-32 convex hemispherical contacts, each contact has a diameter ranging from 120-160 µm and a convex height ranging from 10-25 µm, and the contact surface is a PEDOT:PSS-IrOx co-coating layer, and the impedance of a single contact at 1 kHz is ≤2 kΩ.

[0006] Further, the gradient base layer has a thickness ranging from 2-8 µm and is formed by continuous transition of a lower liquid crystal polymer and an upper difluoro-p-xylene substituted poly-p-xylene, wherein the thickness of the lower layer is greater than that of the upper layer and the Young's modulus decreases from 3-6 GPa to 1-3 GPa.

[0007] Further, the microfluidic layer is a 4-12 µm thick Parylene-C and SU-8 stack layer, at least two columns of Y-shaped microfluidic channels are formed inside, the channel width ranges from 10-30 µm, and the channels are respectively connected to external independent drug injection ports and are windowed around the contacts with a 3-10 µm tapered orifice; the optical waveguide layer comprises a 30-70 µm wide and 3-7 µm high trapezoidal cross-section silicon nitride waveguide core and a 1-3 µm thick Parylene-HT waveguide cladding layer covering the outside of the waveguide core, supports 470 nm single-mode transmission, and the minimum bending radius is ≤0.6 mm.

[0008] Further, the top packaging layer is a 1-4 µm thick Parylene-CS moisture barrier layer that covers and seals the soluble insertion layer, the gradient base layer, the spinal cord anastomosis rib, the conductive serpentine net, the hemispherical contact array, the microfluidic layer, and the optical waveguide layer; the conductive serpentine net, the current contact, the microfluidic layer, and the optical waveguide layer are collectively extended into an integrated flexible tail wire at the distal end, realizing multiple physical interfaces.

[0009] Further, at least two through micro-holes are arranged on each folding unit of the spinal cord anastomosis rib along the longitudinal direction, the diameter of the through micro-holes ranges from 5 microns to 15 microns, and the center distance between adjacent through micro-holes ranges from 40 microns to 80 microns, so as to form a pressure relief discharge channel after the rib is unfolded, reduce the instantaneous liquid pressure between the dura mater and the flexible electrode, and improve the attachment stability.

[0010] Further, the integrated flexible tail line is sequentially provided with a metal pad group, a micro-fluid interface seat and an optical fiber sleeve at the distal end, and the three are integrally packaged by the same poly-p-xylylene coating layer, wherein the metal pad group is used for electrical signal transmission, the micro-fluid interface seat is used for drug injection connection, and the optical fiber sleeve is used for external light source coupling, and the three physical interfaces are electrically insulated, liquid isolated and optically independent from each other.

[0011] The method for optimizing the stimulation parameters of the mouse spinal cord flexible electrode, the method comprises: before the mouse spinal cord flexible electrode is implanted, the soluble insertion layer is sampled by partition, the scanning electron microscope is combined with the laser confocal microscope, the three-dimensional topography imaging and the optical path interference thickness meter are used to obtain the thickness of the soluble insertion layer before implantation, and the curvature radius of the mouse spinal cord is recorded; the bending stiffness difference of the liquid crystal polymer and the difluorinated p-xylene substituted poly-p-xylylene in the gradient base layer is measured respectively; according to the curvature radius of the mouse spinal cord, the thickness of the soluble insertion layer and the bending stiffness difference, the mechanical-electric coupling coefficient of the mouse spinal cord flexible electrode is calculated; the reference pulse charge is set, and the target pulse charge is calculated by combining the mechanical-electric coupling coefficient; the comprehensive target function is set, and the comprehensive target function value is minimized under the premise that the maximum allowed current, the interface voltage safety threshold, the single-phase safety charge and the upper limit of the mechanical stiffness are not exceeded, so that the mouse spinal cord flexible electrode reaches the maximum stimulation efficiency.

[0012] Further, the mechanical-electric coupling coefficient is: ; wherein, is the thickness of the soluble insertion layer; is the curvature radius of the mouse spinal cord; is the bending stiffness of the liquid crystal polymer in the gradient base layer; is the bending stiffness of the difluorinated p-xylene substituted poly-p-xylylene in the gradient base layer; is the reference stiffness; is the common Poisson's ratio, which is 0.34; is the thickness of the liquid crystal polymer in the gradient base layer; is the thickness of the difluorinated p-xylene substituted poly-p-xylylene in the gradient base layer.

[0013] Further, the comprehensive objective function is: ; Wherein, is the comprehensive objective function value; is the target pulse charge; is the maximum allowed current; is the flexible tail line DC resistance; is the interface voltage safety threshold; is the single contact 1kHz impedance; is the spinal cord tissue conductivity; is the contact-spinal cord average gap; is the reference pulse charge; is the upper limit of mechanical stiffness; is the single-phase safe charge; is the total contact area; is the total number of hemispherical contacts; is the single contact diameter.

[0014] Compared with the prior art, the beneficial effects of the present application are: first, the present application realizes the self-consistent transition of the required rigidity in the implantation stage and the long-term flexibility after implantation by introducing a soluble insertion layer, avoids the compression or damage of the mouse spinal cord caused by additional rigid guide equipment, and improves the operation safety and implantation precision. Secondly, through the modulus gradient design of the gradient base layer, the shear stress concentration between the functional layers of the electrode is effectively relieved, the structural fatigue and contact drift under the condition of spinal cord micro-motion are significantly reduced, and the mechanical stability of long-term adhesion is ensured. At the same time, the spinal cord anastomosis rib arranged on the base can be self-deployed after implantation, forming a liquid bypass channel extending along the longitudinal direction of the spinal cord, optimizing the local cerebrospinal fluid circulation environment, and reducing the risk of interface impedance fluctuation and neural tissue inflammatory reaction caused by tissue fluid retention. Further, the conductive serpentine net and the hemispherical contact array cooperatively construct a multi-channel stimulation interface that takes into account the deformation compensation capability and low impedance characteristics, ensuring that the stimulation signal still has stable transmission capability in the structure dynamic change. On this basis, the present application also embeds microfluidic channels and optical waveguide structures in the flexible electrode body, so that neural stimulation is not limited to electrical signal control, but also realizes the joint regulation of neurochemical environment and light-sensitive pathway, expanding the flexibility of experimental design and the strategy space of neural regulation. Finally, by including all structural parameters and physical properties into the stimulation parameter optimization method, a complete structure-electrical-biological response coupling model is constructed, so that each stimulation parameter setting can achieve an optimal balance between safety, electrical efficiency and biological adaptability, greatly improving the experimental stability, repeatability and neural response efficiency of the electrode system in mouse spinal cord research. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1Structure diagram of the flexible mouse spinal cord electrode according to the present application; Figure 2 Top view of the spinal cord anastomosis rib according to the embodiment of the present application; Figure 3 Side view of the spinal cord anastomosis rib according to the embodiment of the present application; Figure 4 Cross-sectional view of the spinal cord anastomosis rib according to the embodiment of the present application; Figure 5 Structure diagram of the conductive serpentine mesh according to the embodiment of the present application; Figure 6 Diagram of the Y-shaped convergence structure according to the embodiment of the present application. DETAILED DESCRIPTION

[0016] The following description is provided to enable any person skilled in the art to practice the present application. The preferred embodiments in the following description are only examples of the present application and other obvious modifications are possible.

[0017] Reference Signs Figure 1 As shown in the figure, the flexible mouse spinal cord electrode comprises, in the order from bottom to top, a soluble insertion layer, a gradient base layer, a spinal cord anastomosis rib, a conductive serpentine mesh, a hemispherical contact array, a microfluidic layer, an optical waveguide layer, and a top packaging layer; the soluble insertion layer is 5-25 µm thick photo-crosslinked PEG-DMA, maintains a Young's modulus in the range of 0.8-1.5 GPa before implantation, and hydrolyzes in the cerebrospinal fluid and falls off completely within 0.5-3 hours after implantation; the spinal cord anastomosis rib is an unfoldable S-shaped folded structure arranged on the upper surface of the gradient base layer, the distance between adjacent ribs is in the range of 80-200 µm, the folded height is not greater than 15 µm, and the unfolded height is in the range of 15-40 µm, for forming a cerebro-spinal fluid bypass groove extending longitudinally along the spinal cord on the surface of the dura mater; the conductive serpentine mesh is composed of a 0.8-1.5 µm thick Au / Cr metal layer, the bidirectional curvature radius of the serpentine trace is not less than 40 µm, the trace line width transitions from 4-8 µm at the center to 10-15 µm at the edge, and a 100-300 nm TiN nanometer island layer is formed on the metal surface by in-situ plasma etching; the hemispherical contact array is located at the reserved window of the conductive serpentine mesh, and there are 8-32 convex hemispherical contacts, each contact has a diameter in the range of 120-160 µm and a convex height in the range of 10-25 µm, the contact surface is a PEDOT:PSS-IrOx co-coating layer, and the impedance of a single contact at 1 kHz is ≤2 kΩ.

[0018] In the implantation stage, the spinal cord tissue is covered by the dura mater, which has a much higher mechanical strength than the electrode body. In order to avoid pushing deformation and obtain repeatable positioning accuracy without the help of additional hard guide needles, the present scheme introduces a soluble insertion layer at the bottom layer. The light-crosslinked PEG-DMA has a Young's modulus in the range of 0.8GPa to 1.5GPa after UV curing, which is sufficient to maintain shape stability under naked eye visible micro-operation; when the device is in place and cerebrospinal fluid penetrates into the soluble insertion layer, the crosslinked network breaks down as the ester bond hydrolysis reaction gradually splits, causing the modulus to rapidly drop to tens of megapascal and then completely dissolve, converting the high rigidity required during implantation into the ultra-soft required for smooth operation, greatly reducing the inflammatory response caused by long-term mechanical stimulation. The sudden drop in modulus also affects the change in the electrode-tissue gap, and the stimulation parameter optimization method uses this change as the initial mechanical boundary condition to calculate the coupling coefficient of the subsequent current density distribution.

[0019] The gradient substrate layer that follows uses the continuous deposition of liquid crystal polymers and difluorinated p-xylene substituted poly-p-xylene to produce a longitudinal bending stiffness that continuously decreases through the thickness difference and intrinsic modulus difference. When the electrode periodically bends with the small-scale expansion and contraction movements of the spinal cord breathing and gait, traditional single-layer homogeneous thin films often accumulate shear stress at the junction between the soft end and the hard periphery. The present scheme gradually releases strain layer by layer with the inherent gradient, ensuring that the conductive serpentine mesh and the hemispherical contact array are maintained at a micron-level flatness under dynamic load, macroscopically ensuring stable stimulation field distribution and microscopically reducing the probability of fatigue crack initiation in the metal film. The presence of the gradient also allows the spinal cord anastomosis ribs to maintain a spatial slot structure with minimal external force after unfolding, forming a cerebro-spinal fluid bypass slot along the longitudinal direction of the spinal cord. This slot not only provides an additional heat dissipation path, but more importantly, it dilutes the gas and acidic byproducts produced by the electrode-tissue interface electrochemical reaction through continuous low-speed fluid shear, slowing down the interface polarization accumulation rate. The stimulation parameter optimization method considers this heat dissipation and dilution effect as a dynamic heat-mass transfer boundary condition, establishing a time-iterative thermal-electro-chemical coupling model to dynamically adjust the pulse width and frequency to avoid transient local pH deviation or temperature rise above the threshold.

[0020] The conductive serpentine mesh assumes the dual responsibilities of long-distance low-loss conduction and near-field reversible deformation. The bidirectional serpentine geometry with a curvature radius ≥ 40 µm and a line width increasing from the center to the edge releases strain concentration on the tension side and prevents wrinkle accumulation on the compression side. TiN nano-islands formed by in-situ plasma etching provide capacitive active sites at the electrochemical level, reducing interface impedance and inhibiting hydrogen evolution side reactions; at the mechanical level, the micro-nano concave-convex structure breaks the critical crack size of the film surface, delaying fatigue cracking. The hemispherical contact array is generated by double electrochemical deposition within a set window, and the hemispherical shape naturally enlarges the effective area and homogenizes the contact pressure with a circular arc profile, achieving a single contact impedance ≤ 2 kΩ at 1 kHz while avoiding sharp edges that can tear the soft film. The PEDOT:PSS-IrOx dual-phase coating on the surface of the contact points assumes the charge injection in an ion-electron cooperative manner when the charge flows in both directions, allowing the stimulation parameter optimization method to search for the global optimum within a larger safe charge window when optimizing the single-phase pulse width.

[0021] The microfluidic layer deeply embeds Y-shaped channels with a width of 10 µm to 30 µm in a Parylene-C / SU-8 stack, and releases anti-inflammatory agents, tracer dyes, or ion buffers as needed by an external micropump driven tail line. Its outlet is distributed around the periphery of each hemispherical contact, and the drug solution acts on the tissue coaxially with the stimulation current in the dispersion flow field. An algorithm can determine whether to trigger micro-liquid injection based on real-time impedance drift rate. The optical waveguide layer forms a soft total internal reflection channel through a silicon nitride high-refractive-index core and a Parylene-HT low-refractive-index cladding, providing 470 nm single-mode transmission while maintaining a minimum bending radius of 0.6 mm. Electrical stimulation and optical stimulation are two complementary methods of neural intervention, and the stimulation parameter optimization method maps the optical power to the equivalent membrane potential modulation gain in the background, so when searching for the pulse frequency, it can calculate the solution that reduces the current while maintaining the same effect according to the light-electricity cooperative excitation curve, thereby reducing heat deposition. The top encapsulation layer is entirely vapor-deposited with Parylene-CS, forming a continuous and ultra-thin water vapor barrier. All lower functional modules are connected to the tail line area through trench-bridge via holes and tail line area, and the tail line area is covered with the same parylene package layer, achieving electrical, liquid, and optical separation of the metal pad group, microfluidic interface seat, and optical fiber sleeve. In this way, on the one hand, the external measurement system can collect real-time changes in the flexible tail line DC resistance, hemispherical contact array impedance, and spinal cord tissue conductivity; on the other hand, the micropump flow rate and light source power can be written into the same communication bus, allowing the stimulation parameter optimization method to have a truly multi-physical closed loop.

[0022] Furthermore, the gradient base layer has a thickness ranging from 2 µm to 8 µm, and is formed by a continuous transition from a lower layer of liquid crystal polymer to an upper layer of difluoro-substituted parylene-substituted parylene, wherein the thickness of the lower layer is greater than that of the upper layer and the Young's modulus decreases from 3 GPa to 6 GPa to 1 GPa to 3 GPa.

[0023] The gradient substrate layer is located at the core carrier position of the flexible electrode of the mouse spinal cord, and its thickness is controlled between 2 µm and 8 µm. In the continuous process of photolithography and chemical vapor deposition, the lower liquid crystal polymer is formed by spin coating or blade coating first, and then the vapor deposition of difluorinated p-xylene substituted p-xylene is carried out in the same vacuum chamber, so that the two polymers interpenetrate at the nanoscale, and then a continuous transition structure from bottom to top is formed. The lower liquid crystal polymer is oriented in the nematic phase, which uses a weak magnetic field to guide the molecular axis arrangement during the room temperature annealing stage, so that the in-plane isotropic and thick anisotropic stiffness distribution is obtained, so that the Young's modulus is stabilized in the high value interval of 3 GPa to 6 GPa; the difluorinated p-xylene substituted p-xylene deposited subsequently introduces difluorinated substituents into the molecular skeleton, so that the van der Waals force between the chain segments decreases and a controllable free volume is generated, providing soft buffering in the modulus interval of 1 GPa to 3 GPa.

[0024] Because the lower layer thickness is always greater than the upper layer thickness, the neutral plane of the overall structure is intentionally sunk to the position close to the liquid crystal polymer. When attached to the mouse spinal cord, the outer curved side mainly bears the skeleton support provided by the liquid crystal polymer, and the inner curved side absorbs deformation by the difluorinated p-xylene substituted p-xylene; this modulus gradient disperses the interlayer shear to multiple thin interfaces, inhibiting the stress concentration and peeling risk commonly seen in traditional double-layer structures. During the process, the researchers monitor the deposition rate and chemical composition in real time through in-situ ellipsometry thickness and Raman shift, ensuring that the modulus decreases continuously and without abrupt changes; after deposition, the upper surface is slightly roughened by low-power oxygen plasma to provide adhesion for subsequent deposition of the conductive serpentine mesh. In the stimulation parameter optimization link, the bending stiffness difference of the gradient substrate layer is input into the mechanical-electric coupling model in real time to calculate the actual strain distribution of the electrode-spinal cord interface after the soluble insertion layer falls off. This strain distribution determines the effective contact area of the contact-tissue and the uniformity of the current density, thereby affecting the interface voltage and single-phase safety charge items in the comprehensive objective function; by dynamically adjusting the pulse width and amplitude during the iterative optimization process, the soft accommodation provided by the gradient substrate layer and the high modulus support of the liquid crystal polymer are balanced, ensuring mechanical compliance while optimizing stimulation efficiency.

[0025] Further, the microfluidic layer is a 4 µm to 12 µm thick Parylene-C and SU-8 stack, which forms at least two columns of Y-shaped microflow channels inside, with a channel width ranging from 10 to 30 µm, respectively connecting external independent drug injection ports and opening windows around the contact with 3-10 µm tapered orifices; the optical waveguide layer includes a trapezoidal cross-section silicon nitride waveguide core with a width of 30 µm to 70 µm and a height of 3 µm to 7 µm, and a 1 µm to 3 µm thick Parylene-HT waveguide cladding layer covering it, supporting 470 nm single-mode transmission with a minimum bending radius ≤0.6 mm.

[0026] The microfluidic layer introduces more than two rows of Y-shaped microfluidic channels inside the Parylene-C and SU-8 stack with a thickness of 4 µm to 12 µm, the channel width is controlled at 10 µm to 30 µm, and the branch angle is matched with the interlayer stress field through a one-time double-sided exposure and twice high-temperature reflow shaping process, then a 3 µm to 10 µm tapered orifice is processed around the contact to transition the lumen radially to an in-plane outlet, and a stable laminar sheath is formed before the liquid jet enters the tissue using the local high surface energy difference, thereby achieving localized drug delivery on the surface of the spinal cord without damaging the adhesion state of the electrode and tissue interface; the external independent drug injection port is connected through an integrated flexible tail line, which can not only inject inflammatory inhibitors in real time to reduce long-term contact impedance, but also can drop a small amount of ion balance solution during the stimulation parameter optimization process to correct the interface voltage drift. The optical waveguide layer uses silicon nitride material to form a trapezoidal cross-section waveguide core with a width of 30 µm to 70 µm and a height of 3 µm to 7 µm on the microfluidic layer through plasma-enhanced chemical vapor deposition, and then a continuous slope is introduced at the edge of the waveguide through reactive ion etching to reduce off-mode leakage; the outer 1 µm to 3 µm thick Parylene-HT waveguide cladding is formed by low-temperature vapor phase polymerization, providing the same flexibility as the microfluidic layer and blocking water vapor, so that the waveguide still maintains 470 nm single-mode transmission under a bending radius of not more than 0.6 mm; the distal end of the waveguide is collimated and coupled with a fiber sleeve, and the proximal end is opened in the central ring-shaped light window of the hemispherical contact array, and the light field distribution and current density are concentrically diffused in the tissue, thereby synchronously writing the light-electric dual-channel effect into the comprehensive objective function in the stimulation parameter optimization algorithm, and adjusting the pulse width and light power ratio in real time, so that the current-light power composite stimulation releases more neural regulation freedom while ensuring safe charging density. The entire manufacturing process uses Parylene-C, SU-8 and Parylene-HT alternately to realize the interlayer chemical homogeneity and mechanical modulus gradual transition of microfluidic and optical structures, ensuring that the flexible tail line maintains channel integrity during repeated bending; during the chronic experiment stage, the drug delivered by the Y-shaped microfluidic channel can dynamically adjust the local ion environment, and the precise light field distribution feedback of the optical waveguide layer provides real-time adjustable biochemical dimensions for stimulation parameter optimization, thereby allowing the mouse spinal cord flexible electrode to continuously output low-impedance and high-selectivity composite stimulation during long-term implantation.

[0027] Further, the top encapsulation layer is a 1 µm to 4 µm thick Parylene-CS moisture barrier layer that covers and seals the soluble insertion layer, the gradient base layer, the spinal cord anastomosis rib, the conductive serpentine mesh, the hemispherical contact array, the microfluidic layer, and the optical waveguide layer; the conductive serpentine mesh, the current contact, the microfluidic layer, and the optical waveguide layer are collectively extended into an integrated flexible tail line at the distal end, realizing multiple physical interfaces.

[0028] The top encapsulation layer is formed by vapor deposition of parylene co-crystals at room temperature at the end of the overall manufacturing process, forming a 1-4 pm thick Parylene-CS moisture barrier layer, whose chlorine and sulfur substituents form in-situ radical crosslinks with the surface of the underlying soluble insertion layer, gradient base layer, spinal cord anastomosis rib, conductive serpentine mesh, hemispherical contact array, microfluidic layer, and optical waveguide layer during the deposition process, building seamless molecular-level interfaces that lengthen the water vapor diffusion path and maintain a very low water content for a long time; this layer has a dual effect on the soluble insertion layer, on the one hand providing overall rigidity reinforcement before implantation to prevent interlayer warping, and on the other hand maintaining external integrity after the soluble insertion layer is hydrolyzed and falls off, leaving no cavity on the surface of the spinal cord; when the spinal cord anastomosis rib is deployed and forms a cerebro-spinal fluid bypass groove on the surface of the spinal dura mater, the compliance of the top encapsulation layer allows it to bend with the rib without appearing indentation cracks, thereby ensuring the effective drainage function of the bypass groove in the chronic stage; the conductive serpentine mesh, current contact, microfluidic layer, and optical waveguide layer extend synchronously at the distal end under the top encapsulation layer, using the same photolithography and plasma activation steps to in-plane bond with the top encapsulation layer, and finally presenting an integrated flexible tail line after laser cutting, the outer edge of the tail line is still covered with a continuous Parylene-CS moisture barrier layer, so that the multi-physical interface remains airtight, electrically isolated, and optical path stable when bending, stretching, and twisting; in the stimulation parameter optimization method, the water-blocking effect of the top encapsulation layer on the spinal cord tissue stabilizes the interfacial electrochemical environment of the hemispherical contact array, slows down the impedance drift, and the impedance value returned by the flexible tail line in real time is used as a dynamic input together with the ion balance liquid flow rate delivered by the microfluidic layer to iteratively adjust the pulse amplitude and drug solution concentration, achieving the synergistic optimization of mechanical compliance, chemical homeostasis, and electrical stimulation.

[0029] Further, at least two through micro-holes are arranged on each folding unit of the spinal cord anastomosis rib in the longitudinal direction, the diameter of the through micro-holes ranges from 5 microns to 15 microns, and the center-to-center distance between adjacent through micro-holes ranges from 40 microns to 80 microns, so as to form a reduced pressure drainage channel after the rib is deployed, reduce the instantaneous liquid pressure between the spinal dura mater and the flexible electrode, and improve the attachment stability.

[0030] The spinal cord anastomosis rib is endowed with decompression and drainage functions in the design stage. The diameter of the through micro-pores arranged longitudinally in each folding unit is maintained between 5 microns and 15 microns, and the center-to-center distance between adjacent through micro-pores is maintained between 40 microns and 80 microns. This combination of size and spacing is derived from the minimum critical drainage cross-section requirement obtained by reverse calculation based on the measured data of the mouse dura mater permeability coefficient and the cerebrospinal fluid instantaneous flow rate, while taking into account the statistical safety factor of the rib structure integrity and folding elasticity. In the manufacturing process, a short etching delay is added at the end of deep silicon etching through a double-layer negative photoresist mask to make the wall surface of each through micro-pore present a micro-cone structure, further reducing the inlet hydraulic impedance; then low-temperature poly-p-xylene vapor deposition is used to form a dense coating on the pore wall to block the lateral infiltration of subsequent metal and conductive polymer deposition, ensuring electrical insulation. When the rib is not unfolded, the micro-pores are half-shielded by the adjacent folding layers, which hardly affects the implant stiffness; when the soluble insertion layer is hydrolyzed in the cerebrospinal fluid and releases the rib elasticity, the folding structure is unfolded, the longitudinal through micro-pores are stretched to completely overlap, forming a continuous decompression and drainage channel. The instantaneous pressure difference between the outer surface of the rib and the back of the flexible electrode is quickly diverted, and the liquid enters the rib lumen through the through micro-pores and then flows to the cerebro-spinal fluid bypass groove on both sides along the folding valley direction, thereby effectively suppressing the lifting force and shear stress generated by local liquid accumulation during the initial stage of electrode adhesion. Real-time optical coherence tomography verification shows that under the action of micro-pore conduction, the liquid layer thickness between the dura mater and the flexible electrode can be reduced to less than 2 microns within 10 seconds after unfolding, and remains stable in the subsequent pressure fluctuations caused by respiratory movements, which directly reduces the chronic inflammation caused by micro-slip at the contact-tissue interface. The stimulation parameter optimization algorithm incorporates the aperture and spacing information into the mechanical-electrical coupling model, and infers the interface fluid dynamic state by real-time measured contact impedance changes. Once an upward trend in impedance is found, the algorithm can drive the microfluidic layer to inject a small amount of ion balance liquid to improve lubrication, while fine-tuning the pulse interval to reduce heat accumulation; this shows that the through micro-pores not only improve the adhesion stability at the physical level, but also intervene in the electrophysiological regulation process through closed-loop feedback, enabling the mouse spinal cord flexible electrode to maintain low-power and high-reliability stimulation performance in long-term experiments.

[0031] Further, the integrated flexible tail wire is provided with a metal pad group, a microfluidic interface seat and an optical fiber sleeve at the distal end in sequence, which are integrally packaged by the same poly-p-xylene coating layer, wherein the metal pad group is used for electrical signal transmission, the microfluidic interface seat is used for drug injection connection, and the optical fiber sleeve is used for external light source coupling. The three physical interfaces are electrically insulated, liquid isolated and optically independent from each other.

[0032] The distal region of the integrated flexible tail wire is continuously stretched into a gradient transition zone after the combined process of micro-lithography and laser cutting. In this zone, a metal layer is first deposited and photo-lithographically patterned to form a metal pad group, then a hollow step structure of the micro-fluidic interface seat is vertically stacked using stereolithography on the same substrate, and finally a fiber sleeve is embedded into the residual window using a micro-grinding alignment jig and activated by low-temperature plasma. The entire region then enters a eutectic vapor deposition chamber, where a continuous parylene coating is grown in one step, embedding the metal pad group, micro-fluidic interface seat, and fiber sleeve in a homologous layer in the material sense. The parylene coating shrinks slightly when solidified at room temperature, firmly fixing the three physical interfaces through the tensile self-locking effect, and establishing a long-term air-tight barrier for the micro-fluidic channel using its extremely low water vapor transmission rate. The surface of the metal pad group is electrochemically annealed at sub-millivolt to obtain a smooth crystal plane, enabling high-integrity soldering at low temperature with an external micro-soldering machine, and achieving electrical isolation with the adjacent micro-fluidic interface seat through the built-in insulating ribs in the coating. The micro-fluidic interface seat reserves a tapered self-sealing valve structure inside the coating, which automatically closes under the shrinkage stress of the coating when the external injection needle is withdrawn, ensuring that the liquid only enters the tapered orifice around the contact point along the pre-designed micro-fluidic channel, and does not penetrate into the metal pad group or the fiber sleeve. The inner wall of the fiber sleeve is sputtered with silicon oxide and polished to achieve sub-micron core coupling with the end face of the external optical fiber, and the coating forms a symmetric annular reinforcing rib at this point, which not only shares the shear force of plugging and unplugging, but also maintains optical independence with the metal pad group and the micro-fluidic interface seat. The three types of interfaces are integrally packaged by the same parylene coating, making the tail wire exhibit a single neutral plane when bending and twisting, greatly suppressing the stress mismatch between multiple material interfaces. During the stimulation parameter optimization process, the low-resistance signal path provided by the metal pad group can real-time feedback the contact impedance and spinal evoked potential, the micro-fluidic interface seat can output micro-droplet injection instructions to the target area according to the algorithm, and the fiber sleeve can output calibration light pulses simultaneously to verify the spatio-temporal accuracy of the optical-electric combined stimulation. The algorithm uses these three real-time data to cooperatively update the comprehensive objective function, adaptively adjusting the pulse width, drug flow rate, and light power ratio while ensuring the current safety threshold and temperature upper limit, so that the mouse spinal cord flexible electrode can maintain a low-power and high-efficiency neural regulation state during the long-term implantation period. The air-tight, electrically-insulating, and optically-independent framework constructed by the parylene coating also provides repeatable interface robustness for multiple batches of experiments.

[0033] The method for optimizing stimulation parameters of a mouse spinal cord flexible electrode comprises: before implanting the mouse spinal cord flexible electrode, sampling a soluble insertion layer by partition, using a scanning electron microscope combined with a laser confocal microscope thickness gauge to perform three-dimensional topography imaging and optical path interference thickness measurement on each sampling point, obtaining the thickness of the soluble insertion layer before implantation, and recording the radius of curvature of the mouse spinal cord; respectively measuring the bending stiffness difference of liquid crystal polymer and difluorinated p-xylene substituted p-xylene in the gradient base layer; according to the radius of curvature of the mouse spinal cord, the thickness of the soluble insertion layer and the bending stiffness difference, the mechanical-electric coupling coefficient of the mouse spinal cord flexible electrode is calculated; setting a reference pulse charge, and combining the mechanical-electric coupling coefficient, the target pulse charge is calculated; setting a comprehensive target function, minimizing the value of the comprehensive target function under the premise of not exceeding the maximum allowed current, the interface voltage safety threshold, the single-phase safety charge and the upper limit of the mechanical stiffness, so that the mouse spinal cord flexible electrode reaches the maximum stimulation efficiency.

[0034] The basic logic of the stimulation parameter optimization method is to unify the structural information that can be measured before implantation and the physiological boundary conditions that need to be met in real time after implantation into the same calculable framework. First, researchers partition the soluble insertion layer on the micro-operation platform for sampling, select several representative coordinate points in a two-dimensional matrix, and then use a scanning electron microscope to complete three-dimensional topographic imaging. Subsequently, the laser confocal micro-thickness meter is switched to implement optical path interference thickness measurement on the same coordinate system. The two types of data are perfectly aligned in space, so they can not only give the overall thickness distribution of the soluble insertion layer on a macro scale, but also reveal the thickness fluctuations caused by local microscopic pits and adjacent bosses. These fluctuations will cause differences in the evolution rate of Young's modulus when actually implanted. The operator simultaneously calls the three-dimensional reconstruction module and the curvature calibration module on the terminal software to bind the thickness density map and the curvature radius of the mouse spinal cord to the unified coordinate system before attachment to obtain the initial strain field. Next, the system applies the micro-beam bending test to the liquid crystal polymer and the difluoro-substituted p-xylene substituted p-xylene in the gradient substrate layer, respectively, and outputs the load-displacement curve of both in real time. The bending stiffness difference is extracted by curve fitting, and this difference is input into the physical simulation engine together with the soluble insertion layer thickness matrix and the curvature radius of the spinal cord to calculate the mechanical-electric coupling coefficient. This coefficient reflects the amplification or attenuation effect of the change of the contact interface resistance of the electrode in the process of external bending compression and internal bending stretching on the strain distribution, and is the core basis for subsequent pulse charge setting. Subsequently, the algorithm sets the reference pulse charge as the origin of the electric stimulation space, and then multiplies or divides the mechanical-electric coupling coefficient to obtain the target pulse charge, thereby ensuring that the charge density and tissue compliance are adjusted synchronously in the same dimension. In order to further consider biological safety and energy efficiency, the system introduces a comprehensive objective function, sets the maximum allowed current, interface voltage safety threshold, single-phase safe charge and upper limit of mechanical stiffness as four hard constraints, and sets the contribution of the target pulse charge to the induced motor potential as the efficiency item, and the influence of energy consumption on tissue temperature rise as the penalty item.

[0035] The optimizer adopts a hierarchical hybrid strategy: the outer layer uses the particle swarm algorithm to explore the minimum region of the comprehensive objective function globally, and the inner layer uses the quasi-Newton method to refine the parameters locally. The two are executed alternately until the comprehensive objective function converges to the preset threshold. After the implantation surgery is completed, the system collects the impedance spectrum and evoked potential amplitude of the hemispherical contact array in real time through the flexible tail line. Whenever the interface impedance shows a consistent upward trend due to tissue reaction, the optimizer triggers the microfluidic layer to add low-concentration ion balancing solution to reduce the interface voltage, while slightly adjusting the target pulse charge to maintain a safety margin; if the impedance recovers, it automatically withdraws the drug solution and slightly adjusts the pulse width to save energy and reduce ion load. The algorithm also listens to the micro-displacement caused by the respiratory and cardiac cycles, dynamically recalibrates the mechanical-electric coupling coefficient, and updates the target pulse charge in real time to maintain a stable stimulation threshold and repeatability in long-term experiments. During the entire process, all data are written to the cloud database in real time, facilitating horizontal comparison between batches of experiments and providing training samples for subsequent model accuracy iteration. The final results show that this stimulation parameter optimization method can achieve predictable, reproducible, and efficient motor-evoked potential output for mouse spinal cord flexible electrodes with the lowest energy consumption under the strict premise of not exceeding the maximum allowed current, not touching the interface voltage safety threshold, not exceeding the single-phase safety charge, and not crossing the upper limit of mechanical stiffness. It provides a set of replicable closed-loop optimization paradigm for spinal cord injury mechanism research and minimally invasive neural regulation treatment.

[0036] Further, the mechanical-electric coupling coefficient is: ; wherein, is the thickness of the soluble insertion layer; is the radius of curvature of the mouse spinal cord; is the bending stiffness of the liquid crystal polymer in the gradient base layer; is the bending stiffness of the difluorinated p-xylene substituted parylene in the gradient base layer; is the reference stiffness; is the common Poisson's ratio, which is 0.34; is the thickness of the liquid crystal polymer in the gradient base layer; is the thickness of the difluorinated p-xylene substituted parylene in the gradient base layer.

[0037] The mechanical-electric coupling coefficient The coefficient is designed to compress the matching degree of geometric compliance and interlayer stiffness into a single dimensionless scalar, thus providing a shortcut for the "mechanical→electrical" mapping of the stimulation parameter optimization algorithm. The numerator of the coefficient only contains a difference term, which compares the ratio of the thickness of the soluble insertion layer to the radius of curvature of the spinal cord. The soluble insertion layer is the outermost source of rigidity of the electrode before implantation and the first barrier to directly contact the spinal cord surface during the attachment process. If the thickness of this layer is close to or exceeds the order of magnitude of the radius of curvature, significant passive compression will occur when the film spreads on the curved surface, and the true contact area of the contact will be weakened in the early stage of attachment; on the contrary, when the thickness is much smaller than the radius of curvature, the film hardly produces additional compression, the geometric compliance is high, and the mechanical coupling efficiency is naturally improved. The linear ratio of the two is compared with "1", which essentially quantifies the geometric fault tolerance space with the residual compression: the larger the ratio, the higher the residual compression, and the coefficient is proportionally weakened; the ratio close to zero indicates that the soluble insertion layer hardly burdens the compliance in geometry.

[0038] The core of the denominator is the bending stiffness difference, which takes the square of the difference between the bending stiffness of the liquid crystal polymer layer and the difluoro-p-xylene substituted poly-p-xylene layer, and then normalizes it with the average stiffness. When the multilayer composite structure of the film is bent on the curved surface, if there is a significant difference in the bending stiffness of the upper and lower layers, the entire system will exhibit migration of the neutral surface and interlayer shear concentration; these mechanical distortions will eventually be reflected in the interface impedance and local current density of the contact through the deformation path of the conductive serpentine net and the channel compression of the microfluidic layer. Taking the square of the stiffness difference can screen the sign information of "which layer is harder" and only retain the absolute size of the difference; normalizing it with the average value eliminates the influence of the thickness scale and the modulus scale of the material itself, making the index comparable for different material systems. The denominator adds a constant "1" to the item, which has two purposes: one is to ensure that when the stiffness of the two layers is completely matched, the denominator is stable at "1" and will not amplify the value; the second is to avoid numerical singularity when the stiffness difference approaches zero due to experimental measurement errors.

[0039] The bending stiffness itself appears as "modulus x thickness cubed", which is a direct product of classical plate theory: for the same material, a small increase in thickness can bring a three-fold increase in bending resistance, which shows that in such an ultra-thin system as a flexible electrode, even if the Young's modulus of the two materials is not greatly different, the thickness deviation is enough to create a significant bending difference. In the formula, the Poisson's ratio is set to zero point three four, because within the acceptable manufacturing tolerance of the polymer film, the change of Poisson's ratio of the two layers has a much weaker influence on the bending performance than the modulus and thickness, so using a single constant not only simplifies the measurement but also avoids introducing unnecessary random noise into the coefficient calculation.

[0040] In summary, The construction of the formula follows two design ideas: one is to define the "thickness-curvature" ratio as the geometric compliance index, which explicitly shows the potential impact of unavoidable passive compression on the electronic interface before attachment; the second is to define the "bending stiffness difference / average stiffness" as the material matching index, which highlights the difference by squaring operation, removes the dimensional dependence of absolute value by normalization, and ensures that the index will not be amplified in the ideal matching case by "plus one" operation. In this way, the coefficient falls between zero and one, and the closer the value is to one, the better the geometric compliance and material matching, and the algorithm allows a larger pulse charge within the same safety boundary; a lower value means that any point design mismatch will be punished as a contraction of the electric stimulation capacity.

[0041] In the actual implementation process, the system first obtains the thickness matrix and spinal curvature, calculates the numerator; then obtains the modulus and thickness of each layer through nanomechanical experiments, calculates the bending stiffness, and completes the denominator. All processes are strictly time-stamped synchronized to avoid time drift of multi-source data. Subsequently, As a weight factor sent into the closed-loop optimizer, it directly determines the initial pulse setting and the subsequent adaptive adjustment direction. When the soluble insertion layer is effectively thinned due to hydrolysis in the body fluid, or the curvature field changes slowly due to tissue repair, the system updates the parameters in real time through the flexible tail line and recalculates the numerator; similarly, if chronic tissue reactions cause additional silicon oxidation or hydration at the material interface, causing changes in the effective modulus between layers, the system dynamically refreshes the denominator by inverting the bending stiffness difference through multi-frequency impedance spectroscopy. Thus, Throughout the experimental period, the latest mechanical-electric coupling state is always depicted, providing a real-time reliable physical benchmark for stimulation parameter optimization.

[0042] Further, the comprehensive objective function is: ; Wherein, is the value of the comprehensive objective function; is the target pulse charge; is the maximum allowed current; is the DC resistance of the flexible tail line; is the interface voltage safety threshold; is the single-touch 1kHz impedance; is the spinal cord tissue conductivity; is the average gap between the touch and the spinal cord; is the reference pulse charge; is the upper limit of mechanical stiffness; is the single-phase safe charge; is the total area of the touch; is the total number of hemispherical touches; is the diameter of a single touch.

[0043] The comprehensive objective function compresses four inherently incompatible indicators—electrical boundary constraints, organizational interface safety, charge budget, and mechanical tolerance—into a single dimension, and through square normalization, ensures that each indicator is compatible with the overall objective function. The contributions are dimensionally equivalent, thus providing an unbiased trade-off platform for the closed-loop optimizer. First, the target pulse charge... With maximum allowable current The ratio characterizes whether the instantaneous discharge amplitude reaches the neuron's instantaneous sodium channel inactivation threshold. The squared term means that the penalty increases exponentially as the amplitude approaches the upper limit, forcing the algorithm to prioritize finding margin in the width or duty cycle dimension rather than continuing to increase the peak value. Secondly, Multiply by resistance - impedance - tissue conductivity composite transfer impedance Then, with the interface voltage safety threshold The comparison reflects the peak voltage generated on both sides of the interface when current passes through the equivalent circuit formed by the flexible tail wire, hemispherical contact array, and spinal cord tissue. Excessive voltage will induce a Faraday reaction or tissue ablation; therefore, the square penalty numerically transforms any local high resistance in the transmission channel into macroscopic energy consumption and safety disadvantages, causing the optimizer to tend to reduce... Or dynamically add ion balance solution to improve To mitigate the risk. The third item is the reference pulse charge. Same single-phase safe charge The comparison reflects the distance between the minimum energy required for reversible depolarization of a neuron by a single hemispherical contact within a specified time window and the long-term cumulative damage threshold of the tissue; appearing in a squared form, any exceeding this threshold... The settings will be magnified quickly. The forced algorithm seeks strategies to reduce the repetition frequency or time-division multi-touch stimulation to balance charge diffusion.

[0044] The last item replaces half of the sum of the bending stiffness and the upper limit of the mechanical stiffness with liquid crystal polymer and difluoro-p-xylene. In comparison, the overall bending resistance of the material stack is directly included in the same safety domain: when the bending moment of the flexible structure approaches the upper limit under breathing or gait tension, the squared penalty ensures that the optimizer reduces... This reduces the additional thermal stress introduced by electrical stimulation, and also triggers the algorithm to re-evaluate. And the attachment strategy was revised. The sum of the four items resulted in... It possesses three main characteristics: First, each term is composed of directly measurable or real-time estimable physical quantities, without relying on empirical coefficients; second, all denominators are hard safety thresholds or design limits, ensuring... That means a certain safety bound has been touched, the optimizer can be used as a stopping criterion; third, square normalization makes the gradient descent or quasi-Newton local optimizer get a smooth convex domain in four-dimensional space, avoiding the gradient vanishing or saddle point stagnation. In the specific implementation, the system synchronously records Actual waveforms, updated by four-probe measurements , fitted with multi-frequency impedance spectroscopy , and estimated by optical coherence tomography , and calculated in real time Effective area decay, updated in real time If is less than a preset threshold and remains stable for thirty cycles, the algorithm will gently increase the upper limit of the test energy efficiency; if approaches one or instantaneously surges, the pulse width is immediately reduced and the rest interval is increased, while triggering the microfluidic layer to inject liquid or reducing the external light power, simultaneously reducing pressure from multiple dimensions. The entire mechanism ensures that the mouse spinal cord flexible electrode always operates at a safe, low-power, and efficient dynamic equilibrium point throughout the entire process of chronic experiments, and through continuous retrieval of historical data, it can provide intuitive and comparable performance indicators for animal experiments of different batches or different disease models.

[0045] Reference pulse charge is: ; where is the spinal cord nerve threshold current density; is the single-phase pulse width (adjustable); is the equivalent thickness of the tissue hydration layer.

[0046] First, estimate the total charge that needs to be injected into the tissue per unit time using the most basic surface charge conservation relationship, then introduce microtransport corrections at the contact-tissue interface, and simultaneously incorporate the effects of electrode geometry on nerve threshold and the effects of tissue chemical environment on charge diffusion into the same dimension. representing the lower limit current density required to trigger an action potential in the spinal cord white matter, is the most direct quantitative indicator of whether electrical stimulation can cause nerve excitation; multiplying it by the total contact projection area is equivalent to expanding the nerve threshold from surface current density to bulk charge rate, and then multiplying it by to convert the rate to the total amount of single pulses. This triple product naturally gives the minimum charge required in an ideal ohmic medium, when the contact is in close contact with the tissue, and without considering interface polarization. However, there is always a hydration layer on the surface of the real spinal cord, with a thickness of nanometers to microns, which is composed of proteins, glycosaminoglycans, and ionic hydration sheaths, forming an additional series resistance to the cross-interface electric field; The equivalent thickness of this kind of diffusion layer is described. Since the hydration layer thickness is proportional to the average gap between the contact and the tissue The equivalent thickness of this kind of diffusion layer is described. Since the hydration layer thickness is proportional to the average gap between the contact and the tissue The ratio is established, which quantifies the additional impedance caused by the non-ideal adhesion of the interface. When the contact is in close contact with the tissue The additional impedance is small, and the gain term rises, forcing The adjustment is higher, thus compensating for the loss of charge at the algorithm level; if the adhesion is gradually expanded by chronic tissue remodeling, the gap gradually expands, and the gain term drops, providing the algorithm with the basis to actively reduce the reference pulse charge, which can also reduce energy consumption and heat load. This coupling of the expanded charge and the interface gap compensation factor makes the reference pulse charge a composite index that pays attention to both the electrical threshold and respects the interface geometry and chemical state.

[0047] This index plays a dual role of "starting point" and "warning line" in the optimization process. As a starting point, it provides a linear scale for the initial estimation of the target pulse charge The algorithm first adjusts the after the geometric-material coupling to obtain , and then performs the first convergence search against the four safety boundaries in the comprehensive objective function. As a warning line, It appears in the third term of the comprehensive objective function, and the ratio with the single-phase safe charge monitors in real time whether the single stimulation falls in the gray area between reversible depolarization and cumulative damage of the tissue. When long-term implantation leads to changes or changes in tissue response , the impedance spectrum returned by the encapsulated tail line will trigger the system to re-measure both, update the gain term and refresh ; then the optimizer adjusts according to the new , ensuring that the maximum stimulation efficiency is maintained within the safety threshold. Through this continuous cycle of updating, the reference pulse charge is always synchronized in real time with the contact geometry, interface chemistry, and neural excitation threshold throughout the entire chronic experimental period, ensuring that the electrode does not over-inject charge due to interface changes, and preventing energy allocation from remaining at conservative settings when impedance decreases, thus allowing the mouse spinal cord flexible electrode to always operate at an optimized point of high efficiency, low consumption, and safety.

[0048] The target pulse current is: ; Where is the cerebrospinal fluid resistivity; is the resistivity of the conductive serpentine mesh film; is the direct current resistance of the flexible tail line.

[0049] First, The leftmost one translates the comprehensive evaluation of the match between the attachment geometry and the interfacial stiffness directly into a proportional factor for the amplification or attenuation of the current: if the flexible electrode fits the curvature of the spinal cord well and the material mismatch is minimal, Close to the next term, the amplification effect of the formula is not limited; once there is a mismatch in geometry or material, The entire product will be depressed, prompting the optimizer to reduce the peak current to prevent the formation of local high stress areas in the contact during microslip, which can exacerbate interface inflammation. The next The baseline pulse charge, which has already taken into account the nerve threshold, electrode area, and hydration layer effect, is divided by the single-phase pulse width in the previous step, converting the static charge requirement into a time-resolved current amplitude; the benefit of this is that the algorithm can adjust The flexible trade-off between current peak and pulse duration meets the differentiated needs of different experiments in terms of spatial and temporal selectivity.

[0050] The first bracket in the formula represents the relative size between the cerebrospinal fluid resistivity and the conductive serpentine net film resistivity when the interfacial electric field is driven. Cerebrospinal fluid is a semiconductive salt solution, and its resistivity can vary by up to two percent when the body temperature fluctuates, ion concentration changes, or trace proteins are deposited; the Au / Cr double-layer thin film resistivity of the conductive serpentine net changes slightly with small deviations in film thickness and grain boundary scattering, and after strengthening the TiN nano-island, it will undergo a long-term stabilization process. Taking the ratio of the two and weighting it in the form of "1+", it is equivalent to telling the optimizer that if the conductive ability of the cerebrospinal fluid is weaker than that of the metal film, the interfacial ohmic voltage drop is larger, and to ensure that the given charge reaches the neuron membrane within the same pulse width, the peak current must be increased by the same proportion; if the ion strength of the cerebrospinal fluid increases, causing its resistivity to decrease, the current path loss is reduced, the bracket value falls, and the amplification effect of is automatically weakened. Since can be measured in real time in vivo by a four-probe microelectrode, and is written into the on-chip storage after manufacturing calibration, the real-time ratio of the two can be updated at a millisecond level in the operating room environment, providing fast enough feedback for closed-loop control.

[0051] The second bracket couples the single-touch 1 kHz impedance with the flexible tail wire DC resistance to evaluate the contribution of the electrode internal conductor to the overall loop voltage drop. After being coated with poly-p-xylylene, the copper or gold traces in the flexible tail wire may develop micro-cracks during repeated bending, resulting in an equivalent DC resistance The degradation rate increases slowly over time; simultaneously, the PEDOT:PSS-IrOx co-coating on the hemispherical contact surface also undergoes ion exchange and pseudocapacitance aging in the physiological fluid environment, leading to... Both the real and imaginary parts of the value can increase. If the increase in single-contact impedance is faster than the change in the DC resistance of the tail wire, the value of the brackets increases, and the system amplifies the drive current to maintain the effective voltage across the pulse; conversely, when material aging is mainly concentrated in the tail wire metal... The current rises faster, the increase in current within brackets is suppressed, and the algorithm is more inclined to replace the tail wire module rather than simply increase the current. The reason for comparing the two rather than simply summing them is that, when faced with the same order of magnitude of impedance increase, increasing the current to overcome the polarization voltage drop at the contact interface is often more efficient than overcoming the ohmic loss of the tail wire; therefore, through a relative measure, the optimizer can automatically allocate resources to the most profitable adjustment path.

[0052] The experimental procedure first involves performing four-probe calibration of the Au / Cr serpentine mesh and tail wire in a clean room to obtain the initial... and During the implantation surgery, the system uses confocal fiber electrodes to measure the cerebrospinal fluid conductivity on the spinal cord surface and converts it in real time. And write it into the formula; within the same time window, the multi-frequency impedance spectroscopy scan results provide All measured values ​​are pushed to the central processing core via the on-chip bus, a process that can be completed in just a few hundred microseconds. The refresh then directly drives the pulse generator to output a new peak value. Since the pulse sequence typically operates at a repetition rate in the hundreds of hertz range, the system can revise the current of the next frame based on the measurement results of the previous frame without introducing any perceptible delay. This allows the electrodes to stably deliver the equivalent charge to the target neuron group even when faced with fluctuations in cerebrospinal fluid ion composition, contact interface polarization, and conductor aging.

[0053] When working in conjunction with the integrated objective function, It acts as the iterable variable: the outer particle swarm optimization algorithm generates a set of candidates. , The combination, with the inner layer based on actual measurements. , , , And the latest Calculate the corresponding Then substitute the obtained values ​​back into the four penalty functions for calculation. .if If the particle swarm fails to reach the target range, the outer layer will update the particle swarm velocity and position and resample the parameters; if If both safety and efficiency thresholds are met, then The target pulse current is locked and issued to the hardware controller. In this way, the target pulse current not only fully embodies the mechanical coupling of the flexible electrode, the interface electrochemical state and the circuit ohmic loss, but also ensures that it does not exceed the biological safety boundary at any time through a multi-layer optimization strategy, and finally realizes a sustainable dynamic balance between extremely low thermal load, extremely high stimulation selectivity and long-term material stability for the mouse spinal cord flexible electrode.

[0054] Reference Figures 2 to 4 From the top view, the significant difference between the folded state and the unfolded state of the rib can be observed. In the folded state before implantation, the S-shaped structure is compressed into a compact form, with the overall height strictly controlled within 15 microns, ensuring minimally invasive during implantation. The rib surface presents a continuous wavy profile, with wave crests and troughs arranged alternately, forming a unique serpentine geometry. When the electrode is implanted on the surface of the spinal cord dura mater, the rib naturally unfolds in the physiological environment, with the height expanding to the range of 15-40 microns, and the unfolding process utilizes the elastic recovery characteristics of the material.

[0055] An accurate spacing of 80-200 microns is maintained between adjacent ribs, and this spacing is designed based on the anatomical features of the spinal cord surface and the hydrodynamic characteristics of cerebrospinal fluid. Too small spacing will result in insufficient bypass slot volume, affecting the decompression effect; too large spacing will reduce the stability of the electrode attachment. Through optimization design, it is ensured that effective decompression is maintained while good mechanical contact performance is maintained. Each folded unit of the rib is provided with multiple through micro-holes along the longitudinal direction, and the micro-hole diameter is accurately controlled within the range of 5-15 microns. The design of these micro-holes follows the principles of fluid mechanics, and the diameter size not only ensures sufficient fluid passing capacity, but also avoids the risk of blockage caused by tissue embedding. The center-to-center spacing between adjacent through micro-holes is set to 40-80 microns, forming a uniformly distributed drainage network. The presence of micro-holes significantly enhances the drainage capacity of the rib, and when transient liquid pressure occurs between the dura mater and the electrode, cerebrospinal fluid can quickly dissipate through these micro-holes.

[0056] The side view shows the extension of the ribs along the longitudinal direction of the spinal cord. Multiple rib units are arranged in sequence on the surface of the gradient base layer, forming a continuous support structure. This longitudinal arrangement design fully considers the anatomical features of the spinal cord, ensuring that the electrodes can fully conform to the natural curvature of the spinal cord surface. Each rib unit can independently exert support and flow guiding effects after unfolding, forming a synergistic effect between each other. The cross-sectional view details the formation mechanism and working principle of the cerebrospinal fluid bypass groove. When the rib is fully unfolded, its three-dimensional structure forms a series of regular groove channels on the surface of the spinal dura mater. These channels constitute a bypass system for cerebrospinal fluid, providing an additional flow path for physiological cerebrospinal fluid circulation. Under normal physiological conditions, cerebrospinal fluid flows freely along these bypass grooves, maintaining the normal pressure balance around the spinal cord. When local pressure rises during electrode implantation or after implantation, the bypass groove system plays a key role in pressure reduction. Excess cerebrospinal fluid can quickly dissipate to the surrounding area through the gaps between the ribs and the through micropores, effectively reducing the instantaneous pressure between the spinal dura mater and the electrode. This pressure reduction mechanism is of great significance for preventing electrode displacement and maintaining stable electrophysiological contact.

[0057] The material selection and surface treatment of the ribs are also carefully designed. Biocompatible materials compatible with the gradient base layer are used to ensure good biocompatibility and long-term stability. The surface is specially treated to reduce the coefficient of friction and reduce mechanical stimulation to the surrounding tissue. At the same time, the elastic modulus of the ribs is optimally adjusted to ensure sufficient support strength while avoiding damage to soft neural tissue. The entire bypass groove system design embodies the concept of bionics, simulating efficient fluid transport structures in nature. Through precise geometric parameter control and reasonable spatial layout, the perfect unity of pressure reduction and flow discharge, stable adhesion, and biocompatibility is achieved, providing reliable technical support for long-term implantation of spinal cord electrodes.

[0058] Figure 5 The precise structural design of the conductive serpentine net and the electrode contact mechanism of the hemispherical contact array are demonstrated. The conductive serpentine net is composed of a 0.8-1.5 micron thick Au / Cr bi-metallic layer, where the gold layer provides excellent electrical conductivity and corrosion resistance, and the chromium layer serves as a transition layer to enhance the bonding strength between the metal and the substrate. The geometric design of the serpentine network fully considers the mechanical stress distribution characteristics of flexible electrodes. The traces adopt a continuous S-shaped curve layout, with a bidirectional curvature radius strictly controlled to be no less than 40 microns, effectively avoiding stress concentration at sharp corners. This smooth curve design significantly improves the fatigue life of the metal wire under repeated bending and deformation, ensuring the electrical connection reliability during long-term implantation.

[0059] The wire width is designed to vary gradually from 4-8 microns in the center region to 10-15 microns in the edge region. This variable width design is based on the principle of current density distribution optimization. The center region carries the main signal transmission function, and a narrower wire width is used to increase the integration density. The edge region mainly bears mechanical stress, and a wider wire width is used to enhance the structural strength. The gradual transition of the wire width ensures the continuity of the resistance and the uniformity of the current distribution. A hemispherical contact array is set at the reserved window, containing 8-32 independent convex hemispherical contacts. Each contact is precisely controlled within the diameter range of 120-160 microns, with a convex height of 10-25 microns. The hemispherical geometry design can maximize the contact area with the spinal cord tissue while reducing the contact impedance. The PEDOT:PSS-IrOx co-coating on the surface of the contact provides excellent electrochemical performance, with the PEDOT:PSS layer having good biocompatibility and lower impedance characteristics, and the IrOx layer providing stable charge transport capacity and anti-electrochemical corrosion performance. The metal surface is formed with a 100-300 nanometer thick TiN nanometer island layer through in-situ plasma etching process. These nanometer island structures significantly increase the effective surface area of the electrode, reducing the electrode-tissue interface impedance. TiN material has excellent biocompatibility and electrochemical stability, and the random distribution of nanometer islands forms a large number of micro-electrode units, achieving uniform dispersion and transmission of current. The impedance of a single contact at a frequency of 1 kilohertz is strictly controlled within 2 kilohertz, ensuring high-quality electrophysiological signal transmission and effective electrical stimulation output.

[0060] Reference Figure 6The Y-shaped converging structure is the key to realize precise drug delivery in the microfluidic layer, and the double-branch converging design ensures the controllable mixing of multiple drug solutions. The structure is composed of two independent drug solution input branches and one mixed output channel. The convergence angle of the branch and the main channel is optimized by fluid mechanics to ensure that the drug solution forms a stable laminar flow interface at the convergence point. The two input branches are connected to the injection ports of drug solution A and drug solution B, respectively, allowing independent input control of different drugs. By adjusting the flow rate ratio of the two branches, the concentration ratio of the mixed drug solution can be accurately controlled. The branch channel wall is precisely shaped by photolithography, with an inner surface smoothness of nanometer level, effectively reducing flow resistance and drug residue. The convergence area is the key area where the two drugs begin to mix. The geometry of this area is designed with a gradual transition to avoid vortex and dead volume caused by sharp corners. The mixing process in this area is driven by diffusion, and the mixing efficiency depends on the matching of diffusion coefficient, flow rate and channel geometry. The pressure gradient distribution along the channel direction shows a continuous decreasing trend, and the pressure reaches a balance state at the convergence point, ensuring that the two drug solutions can stably converge without backflow phenomenon. The mixed drug solution is transported through the single output channel to the downstream tapered window opening, and finally realizes precise drug release around the hemispherical contact point. The entire Y-shaped structure design fully considers the special properties of micro-scale fluid, achieving precise control of drug ratio and ensuring mixing uniformity.

[0061] The following examples demonstrate the process from pre-implantation characterization to the first safe stimulation parameters given by the closed-loop optimizer. First, after grid sampling of the soluble insertion layer on the micro-operation platform, the scanning electron microscope and the laser confocal micro-thickness meter jointly give the thickness distribution. The average thickness is denoted as . The curvature radius of the same adult C5 segment mouse spinal cord is measured by a three-dimensional optical profiler .

[0062] The upper and lower layers of the gradient substrate layer have thicknesses of , . The Young's modulus of the liquid crystal polymer is taken as , and the Young's modulus of the difluoro-p-xylene substituted p-xylene is taken as . The common Poisson's ratio is . The bending stiffness is calculated using the plate and shell small deflection formula :

[0063] .

[0064] Substitute the mechanical-electric coupling coefficient formula .

[0065] Next, determine the reference pulse charge. Literature reports that the white matter action potential threshold current density is about . The total number of hemispherical contacts designed is Single contact diameter taken Single contact cross-sectional area taken .

[0066] Single phase pulse width taken .

[0067] Interface hydration layer equivalent thickness taken , Contact-spinal cord average gap taken .

[0068] Thus .

[0069] Target pulse current calculated. In experiments, cerebrospinal fluid path equivalent resistivity Au / Cr thin film equivalent resistivity Contact one kilohertz impedance Flexible tail wire DC resistance .

[0070] .

[0071] Safety check evaluates comprehensive objective function Maximum allowed current taken Spinal cord tissue conductivity taken Total contact effective area Interface voltage safety threshold Single phase safety charge Mechanical stiffness upper limit Thus:

[0072] .

[0073] Since All four hard constraints are satisfied; the closed loop optimizer latches , as the first round of stimulation configuration and issues to the pulse generator. Subsequently the system enters real-time monitoring mode: if contact impedance or cerebrospinal fluid resistivity drifts, the measured values will update each term of the formula and recalculate and to ensure that the current amplitude is always within the biological safety domain while maintaining close to the maximum stimulation efficiency.

[0074] The above shows and describes the basic principles, main features and advantages of the present application. Those skilled in the art should understand that the present application is not limited to the above-mentioned embodiments, and the above-mentioned embodiments and descriptions in the specification are only the principles of the present application. Various changes and improvements can be made without departing from the spirit and scope of the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection of the present application is defined by the appended claims and their equivalents.

Claims

1. Flexible spinal cord electrodes for mice, stacked in order from bottom to top, include: The invention comprises a soluble insert layer, a gradient basal layer, spinal cord anastomosis ribs, a conductive serpentine mesh, a hemispherical contact array, a microfluidic layer, an optical waveguide layer, and a top encapsulation layer. The soluble insert layer is characterized by being a 5µm to 25µm thick photocrosslinked PEG-DMA layer, maintaining a Young's modulus of 0.8GPa to 1.5GPa before implantation, and hydrolyzing and completely detaching from the cerebrospinal fluid within 0.5 to 3 hours after implantation. The spinal cord anastomosis ribs are unfoldable S-shaped folded structures disposed on the upper surface of the gradient basal layer, with adjacent rib spacing ranging from 80µm to 200µm, a folded height not exceeding 15µm, and an unfolded height ranging from 15µm to 40µm, used to form a cerebrospinal fluid duct extending longitudinally along the spinal cord on the dura mater surface. The rebro-spinalfluid bypass trench; the conductive serpentine mesh is composed of a 0.8µm to 1.5µm thick Au / Cr metal layer, with a bidirectional curvature radius of not less than 40µm for the serpentine traces, and the trace width transitions from 4 to 8µm at the center to 10 to 15µm at the edge. The metal surface is in-situ plasma etched to form a 100nm to 300nm TiN nano-island layer; the hemispherical contact array is located at the reserved window of the conductive serpentine mesh, with a total of 8 to 32 convex hemispherical contacts, each with a diameter ranging from 120µm to 160µm and a convex height ranging from 10µm to 25µm. The contact surface is a PEDOT:PSS-IrOx co-coating, and the impedance of a single contact at 1kHz is ≤2kΩ.

2. The flexible mouse spinal cord electrode as described in claim 1, characterized in that, The gradient substrate layer, with a thickness ranging from 2µm to 8µm, is formed by a continuous transition between a lower liquid crystal polymer and an upper difluoro-substituted p-xylene. The thickness of the lower layer is greater than that of the upper layer, and the Young's modulus gradually decreases from 3GPa to 6GPa to 1GPa to 3GPa.

3. The flexible mouse spinal cord electrode as described in claim 2, characterized in that, The microfluidic layer is a 4µm to 12µm thick Parylene-C and SU-8 stack, forming at least two rows of Y-shaped microfluidic channels with a channel width ranging from 10 to 30µm. These channels are connected to external independent drug injection ports and have 3-10µm tapered openings around the contacts. The optical waveguide layer includes a silicon nitride waveguide core with a trapezoidal cross-section of 30µm to 70µm wide and 3µm to 7µm high, and a 1µm to 3µm thick Parylene-HT waveguide cladding covering it. It supports 470nm single-mode transmission and has a minimum bending radius ≤0.6mm.

4. The flexible mouse spinal cord electrode as described in claim 3, characterized in that, The top encapsulation layer is a Parylene-CS moisture barrier layer with a thickness of 1µm to 4µm, which covers and seals the soluble insertion layer, gradient base layer, spinal cord anastomosis rib, conductive serpentine mesh, hemispherical contact array, microfluidic layer and optical waveguide layer; the conductive serpentine mesh, current contacts, microfluidic layer and optical waveguide layer extend together at the distal end to form an integrated flexible tail wire to realize multiple physical interfaces.

5. The flexible mouse spinal cord electrode as described in claim 4, characterized in that, Each fold unit of the spinal cord anastomosis rib has at least two through-holes along the longitudinal direction. The diameter of the through-holes ranges from 5 micrometers to 15 micrometers, and the center-to-center distance between adjacent through-holes ranges from 40 micrometers to 80 micrometers. This is to form a decompression and drainage channel after the rib is unfolded, thereby reducing the instantaneous liquid pressure between the dura mater and the flexible electrode and improving the adhesion stability.

6. The flexible mouse spinal cord electrode as described in claim 5, characterized in that, The integrated flexible tail cable has a metal pad group, a microfluidic interface seat and an optical fiber sleeve arranged sequentially at the far end. The three are encapsulated by the same parylene coating layer. The metal pad group is used for electrical signal transmission, the microfluidic interface seat is used for drug injection connection, and the optical fiber sleeve is used for external light source coupling. The three physical interfaces are electrically insulated, liquid isolated and optically independent of each other.

7. A method for optimizing stimulation parameters of a flexible mouse spinal cord electrode according to any one of claims 1 to 6, characterized in that, The method includes: before implantation of a flexible spinal cord electrode in mice, sampling the soluble insert layer in sections, using a scanning electron microscope combined with a laser confocal microscopy thickness gauge to perform three-dimensional morphological imaging and optical path interference thickness measurement at each sampling point to obtain the thickness of the soluble insert layer before implantation, and recording the radius of curvature of the mouse spinal cord; measuring the bending stiffness difference between the liquid crystal polymer and difluoro-p-xylene-substituted parylene in the gradient basal layer; calculating the mechano-electric coupling coefficient of the flexible spinal cord electrode based on the radius of curvature of the mouse spinal cord, the thickness of the soluble insert layer, and the bending stiffness difference; setting a reference pulse charge and calculating the target pulse charge based on the mechano-electric coupling coefficient; setting a comprehensive objective function to minimize the comprehensive objective function value under the premise of not exceeding the maximum allowable current, the interface voltage safety threshold, the single-phase safe charge, and the upper limit of mechanical stiffness, so that the flexible spinal cord electrode in mice achieves maximum stimulation efficacy.

8. The method for optimizing stimulation parameters of a flexible mouse spinal cord electrode as described in claim 7, characterized in that, Mechanical-electric coupling coefficient for: ； in, The thickness of the soluble insertion layer; The radius of curvature of the mouse spinal cord; The bending stiffness of the liquid crystal polymer in the gradient substrate layer; The bending stiffness of difluoro-p-xylene-substituted poly(p-xylene) in the gradient substrate layer; For reference stiffness; To use a common Poisson's ratio, a value of 0.34 is adopted; This refers to the thickness of the liquid crystal polymer in the gradient substrate layer; The thickness of difluoro-p-xylene-substituted poly(p-xylene) in the gradient substrate layer.

9. The method for optimizing stimulation parameters of a flexible mouse spinal cord electrode as described in claim 8, characterized in that, The overall objective function is: ； in, The value of the comprehensive objective function; The target pulse charge; Maximum allowable current; For the flexible tail wire DC resistance; The interface voltage safety threshold; The impedance is 1kHz for a single contact. Electrical conductivity of spinal cord tissue; The average intervertebral space between the contact point and the spinal cord; The reference pulse charge; This represents the upper limit of mechanical stiffness. It is a single-phase safe charge; Total contact area; This represents the total number of contact points on the hemisphere. This refers to the diameter of a single contact point.

Images