A multi-workpiece table reverse motion anti-vibration decoupling lithography workpiece table system

Abstract

This invention discloses a multi-stage reverse motion anti-vibration decoupling lithography stage system, belonging to the field of precision motion platforms for semiconductor lithography equipment. The system includes a fixed partitioned planar coil table, a metrology gantry, multiple magnetically levitated stage pieces, a wireless optical communication subsystem, and docking charging and heat-conducting composite contacts. A Halbach permanent magnet array is installed on the bottom surface of the stage, while the coil table remains stationary. A grating ruler is fixed to the upper metrology gantry, and the read head moves with the stage. The stage levitates without power or signal cables, powered by an onboard solid-state battery module and transmitting data via wireless optical communication. The docking station receives power and heat through the composite contacts. The reverse motion of the stage pieces at the process station and the counter-punching station provides anti-vibration decoupling. The metrology gantry and coil table are mechanically and thermally decoupled. The dynamic composite positioning error of the stage is ≤0.3 nm RMS, suitable for 1nm optical resolution lithography equipment.

Description

Technical Field

[0001] This invention belongs to the technical field of precision motion platform for high-end semiconductor lithography equipment. Specifically, it relates to a multi-stage reverse motion anti-vibration decoupling lithography stage system, which is particularly suitable for ultra-high precision lithography equipment with 1nm optical resolution, such as EUV extreme ultraviolet lithography and DUV deep ultraviolet lithography. It is mainly used in 300mm wafer nanometer-level scanning exposure, high-precision overlay, multi-station parallel preprocessing, vacuum cableless precision motion, and full-domain sub-nanometer stable metrology scenarios.

[0002] This invention addresses core technical scenarios such as inertial disturbance suppression of multi-station magnetic levitation workpiece stages, elimination of crosstalk between multiple magnetic fields, multi-dimensional decoupling of mechanical and thermal measurement, wireless vacuum operation, cableless power supply and heat exchange, and long-term thermal steady-state control. It overturns the traditional moving coil, moving grating, and drag chain cable architecture, and achieves high-precision, high-stability, high-throughput, and low-disturbance precision motion control under a multi-workpiece stage scalable architecture. It belongs to the core precision motion platform technology of next-generation advanced process lithography equipment. Background Technology

[0003] The overlay accuracy, critical dimension uniformity, and imaging stability of high-end semiconductor lithography equipment are highly dependent on the workpiece stage motion positioning accuracy and the overall machine reference stability. During the high-speed, high-acceleration XY scanning motion of the workpiece stage, significant inertial reaction forces, overturning moments, and dynamic impact vibrations are generated. These disturbances directly couple to the overall machine metrology reference, causing grating metrology drift and optical reference micro-jitter, ultimately degrading the lithography imaging quality and overlay accuracy.

[0004] Currently, the mainstream momentum cancellation solution for lithography stages in the industry is represented by the existing dual-stage architecture, whose core structure is a moving coil + moving grating architecture: the linear motor coil moves synchronously with the stage, and the metrology grating reference plate is directly fixed on the moving stage body. This traditional architecture has a series of inherent technical defects that cannot be avoided, which seriously limit the performance iteration of ultra-high precision, multi-station, high-throughput lithography equipment. The specific defects are as follows: First, traditional moving coil architecture requires a large number of power cables, signal drag cables, and water-cooled drag chains to move with the stage. Repeated bending and dragging of the cables will introduce nonlinear frictional disturbances, additional inertial disturbances, and hysteresis disturbances, which will directly destroy the dynamic balance accuracy of momentum cancellation. At the same time, cables and flexible water-cooled pipelines are prone to producing a small amount of gas and particle shedding under ultra-high vacuum baking conditions, which will contaminate the vacuum chamber environment, affect the cleanliness of the photolithography process, and the movement of flexible pipelines will easily cause vacuum leakage risks.

[0005] Second, in traditional architecture, the metrological reference moves synchronously with the workpiece stage. Coil heating, workpiece stage thermal drift, and motion deformation are directly coupled into the metrological reference, making it impossible to achieve thermal and mechanical decoupling. In the case of multiple workpiece stages, multiple independent moving grating references have problems such as scale splicing errors and inconsistent coordinate references. The synchronization accuracy of multi-machine momentum cancellation trajectory is poor, making it impossible to achieve stable collaborative operation of three or more machines in multiple workstations.

[0006] Third, the traditional dual-workpiece stage architecture has fixed functions and extremely poor scalability. The hardware role of the workpiece stage is fixed, and it can only support the alternating operation of two workpiece stages. It cannot realize multi-station parallel preprocessing and multiple sets of momentum cancellation pairing. The timing of processes such as loading pre-alignment, baking and resting, process scanning, and error calibration compete with each other, and there is a hardware architecture bottleneck in improving the throughput of the whole machine.

[0007] Fourth, traditional wired power supply, wired communication, and wired water cooling architecture cannot achieve completely cable-free levitation operation of the workpiece stage. The drag chain constraint restricts the workpiece stage's degree of freedom of movement and results in poor dynamic response consistency, failing to meet the precision motion requirements of ultra-high precision and ultra-low disturbance. At the same time, frequent plugging and unplugging and bending of wired interfaces can easily lead to unstable contact, signal jitter, and pipeline leakage, further reducing system stability.

[0008] Fifth, existing multi-workpiece stage magnetic levitation solutions do not establish a global calibration and dynamic compensation mechanism for the magnetic field crosstalk problem of multi-machine synchronous motion. The magnetic fields of multiple Halbach arrays and motion coils are superimposed and coupled to each other, generating periodic magnetic field disturbances, which degrade the stability of magnetic levitation height and thrust output accuracy, and cannot meet the sub-nanometer level dynamic positioning requirements.

[0009] Sixth, in existing technologies, the metrology framework, motion platform, and drive coil thermal control system are coupled together, without an independent isolating constant temperature system and a workpiece stage autonomous temperature equalization structure. The Joule heat generated by the continuous operation of the coil, the heat generated by the onboard read head and other devices on the workpiece stage, and the pulse heat generated by EUV exposure cannot be effectively controlled, which easily leads to irreversible thermal drift errors. Furthermore, the lack of a redundant measurement scheme with multiple read heads at four corners makes it difficult to suppress Abbe errors and false angle errors caused by thermal bending of the substrate, and it is impossible to maintain long-term stable metrology at the sub-nanometer level.

[0010] Regarding reaction force compensation, the industry typically uses a balance mass to withstand the reaction force generated by the stage movement. Existing balance mass solutions have evolved through two generations: the first is a single-layer balance mass solution, where a stationary magnet plate capable of slight drift is placed below the stage travel as the balance mass, along with a low-bandwidth drift control loop to pull the slowly drifting balance mass back to the center of the travel; the second is a layered solution of reactive mass and reactive actuator, using two layers of counterweights that can move in opposite directions to share the reaction force, expanding the reaction force compensation travel through the combined movement of the two layers of counterweights. The inherent limitations of these solutions are: the balance mass itself is an inert mass block, not involved in the lithography process, and its movement relies entirely on an additional drift control loop; the accuracy and bandwidth of the reaction force cancellation are constrained by the loop bandwidth and the balance mass travel; the inert balance mass also occupies space and mass budget below the travel, making it difficult to linearly replicate as the number of stages increases; and existing architectures with balance masses still use moving coils as moving parts, requiring power cables, signal cables, and water-cooling pipes that accompany the stage movement.

[0011] In summary, existing traditional lithography stage architectures suffer from numerous technical shortcomings, including reference coupling disturbances, cable drag chain disturbances, severe magnetic field crosstalk, poor scalability, insufficient thermal stability, low multi-machine collaborative accuracy, outdated heat exchange modes, and insufficient six-degree-of-freedom metrology redundancy. These limitations prevent them from meeting the stringent technical requirements of next-generation 1nm optical resolution lithography equipment, which demand sub-nanometer dynamic positioning accuracy, extremely low reference disturbances, high throughput across multiple stations, and stable vacuum cableless operation. Therefore, there is an urgent need to develop a novel multi-stage reverse motion anti-vibration decoupling system to systematically address the deficiencies of existing technologies at the levels of hardware architecture, electromagnetic design, metrology design, thermal control design, power supply and heat exchange, communication architecture, and control algorithms. Summary of the Invention

[0012] Purpose of the invention To address the aforementioned deficiencies in existing technologies, the present invention aims to provide a multi-stage reverse motion anti-vibration decoupled lithography stage system. This system overturns the traditional moving coil, moving grating, and drag chain cable architecture, constructing a novel decoupled architecture consisting of a fixed partitioned coil + bottom Halbach magnetic array + upper static independent metrology frame + four sets of Littrow heterodyne grating readheads at the four corners of the floating platform + moving readhead + airborne energy storage cableless power supply + multi-access parallel free space laser communication (FSOC) + sealed self-equalizing periodic heat exchange. This invention features a cavity with symmetrically arranged process stations and counter-current stations. The workpiece stage hardware itself does not have a fixed process or counter-current function; its role is determined by its location within the station. The workpiece stage can queue and move between loading stations, waiting stations, counter-current stations, and process stations. All workpiece stages are of the same specification and are precisely counterweighted, with consistent mass and inertia tensors. Any workpiece stage scheduled to the counter-current station automatically becomes a momentum counter-current station. The invention allows for scalable configuration of three or more workpiece stages. Momentum counter-current is achieved through reverse symmetrical motion of paired stations, addressing the fundamental mechanical issues of large disturbances, low accuracy, poor scalability, insufficient stability, thermal lag, communication interruptions, limited power supply, and insufficient attitude measurement accuracy in traditional architectures. This improves the dynamic synthetic positioning accuracy of the workpiece stage. It meets the precision operation requirements of 1nm optical resolution lithography equipment.

[0013] Core Invention Principles and Overall Technical Concept The overall technical concept of this invention is as follows: a bottom-fixed partitioned planar coil desktop is used as the basis for full-domain electromagnetic drive, eliminating the moving coil structure that moves with the stage and all moving drag chains, thus eliminating the parasitic disturbances and pollution risks caused by cables and pipelines; four sets of one-dimensional Halbach permanent magnet arrays (XY diagonally alternating, totaling 64 permanent magnet units) are standardized and arranged at the four corners of the bottom of each independent magnetic levitation workpiece stage to form a "fixed coil moving magnet" drive mode, eliminating the inherent defects such as moving coil heating and moving with the stage, and cable dragging disturbances.

[0014] An independent, suspended ultra-low expansion microcrystalline glass (Zerodur or ULE) metrology truss is erected on the upper part of the cavity as a static global metrology reference. The entire seamless grating master scale is fixed to the lower surface of the truss without any light-transmitting openings to ensure the integrity of the metrology reference. A set of Littrow heterodyne grating read heads is arranged at each of the four physical corners of the workpiece stage floating platform. The four sets of read heads work together to calculate the complete 6DOF position and attitude, and align with the static grating master scale to obtain measurement data. This forms a three-layer split decoupled physical architecture: bottom fixed partitioned coil desktop → middle moving workpiece stage → upper static metrology truss.

[0015] During the workpiece stage levitation operation, all cable chains, including power cables, signal fibers, and water-cooled hoses, are eliminated. Power is supplied by a parallel module of onboard solid-state batteries and supercapacitors within the workpiece stage. The supercapacitors handle instantaneous peak power consumption, while the solid-state batteries provide steady-state continuous power. High-speed parallel data exchange between multiple workpiece stages is achieved through dual-band FSOC free-space laser communication. Only when the workpiece stage enters the loading docking station do the TiN nanocomposite contacts on both side walls vertically press against the cold plate contacts of the frame, simultaneously completing battery recharging and heat dissipation from the workpiece stage. During levitation, the contacts separate, eliminating any mechanical contact or sliding friction, thus avoiding disturbance and vacuum particle contamination at the source. TiN can also be replaced with TaN, both being vacuum-compatible hard, low-emission coatings suitable for ultra-high vacuum baking conditions.

[0016] The thermal control system employs a dual-layer system of "rack constant temperature isolation + workpiece stage sealed self-equilibrium temperature + periodic solid-state heat exchange". Four Littrow heterodyne grating read heads serve as the main steady-state onboard heat source for the workpiece stage. Together with the PZT micro-nano compensation platform and micro circulation pump, they generate approximately 15 W of steady-state heat. The heat is smoothed by the sealed perfluoropolyether (PFPE) liquid storage chamber inside the workpiece stage, strictly controlling the internal temperature gradient to less than 50 mK to avoid false Rx / Ry angle errors caused by thermal bending of the silicon carbide substrate. The workpiece stage integrates a permanently sealed PFPE liquid storage temperature equalization chamber with no external fluid pipelines. It autonomously circulates and smooths the temperature gradient throughout the process. Only the docking station completes periodic heat dissipation through TiN contacts, achieving zero heat loss and return for single-wafer operation cycles. The metering truss and coil table are equipped with two independent constant temperature circuits, combined with heat-insulating air gaps and high-reflectivity radiation shielding layers, to achieve dual decoupling of mechanics and thermality.

[0017] The cavity is equipped with symmetrically arranged process stations and counter-counter stations. The hardware of the workpiece stage does not pre-define the identity of the process stage or the counter-counter station, but the station role is dynamically assigned by the scheduling system. The two equally weighted workpiece stages located at the process stage and the counter-counter station read the same set of global 6DOF coordinates and perform high-precision synchronous reverse motion, relying on momentum conservation to counteract inertial disturbances. The remaining workpiece stages can perform wafer loading, pre-alignment, vacuum baking, and metrology calibration in parallel, realizing parallel process efficiency improvement. The number of workpiece stages can be expanded indefinitely, supporting multiple sets of momentum counter-counter pairs for synchronous operation.

[0018] At the electromagnetic level, the partitioned planar coil desktop is composed of arrayed planar coil tiles. The tiles are grouped into multiple independent drive partitions by row or column. Hall sensor groups are embedded in each tile. Through factory full-domain thrust matrix calibration, online self-calibration and real-time feedforward compensation, the crosstalk of magnetic fields between multiple machines is suppressed. A whole piece of permalloy shielding layer is arranged above the Halbach array of the workpiece stage to suppress the interference of stray magnetic fields on the metering and optical communication devices. At the same time, eddy current thrust disturbances are eliminated through eddy current feedforward compensation.

[0019] At the communication level, the FSOC multiple access architecture, which combines sector floodlight coverage, cat's eye backlighting, and wavelength division multiple access, is adopted. A single base station deployment can cover the entire route and serve multiple workstations, solving the defects of traditional point-to-point link disconnection and inability to run multiple machines in parallel. Combined with the Smith predictor to compensate for link delay jitter, it ensures stable communication.

[0020] The control layer employs a three-layer hierarchical cooperative motion controller, coupled with Robust control, DOB disturbance observer, notch filter, RSS end-link error budgeting, and multi-level fault protection enable high-precision and high-stability control across all operating conditions, and the system dynamically synthesizes positioning errors. .

[0021] 1. Detailed composition of each subsystem of the system 1.1 Vacuum Chamber The vacuum chamber serves as the mounting platform for the entire unit, providing an ultra-high vacuum (UHV) working environment to meet the stringent vacuum requirements of EUV and DUV lithography processes. A rigid flange at the bottom of the chamber securely mounts the partitioned planar coil tabletop. An independent vibration-isolated mounting interface is located at the top of the chamber, supported by a designed granite truss vibration-isolated support, achieving mechanical decoupling between the metering system and the chamber and drive system. The side walls of the chamber are reserved for FSOC debugging windows, temperature measurement, and auxiliary power supply interfaces, ensuring the entire unit meets… Under high-temperature baking conditions, all internal components are made of low-emission, non-volatile, and vacuum-compatible materials. The cavity is spatially symmetrically designed with process stations and counter-processing stations, along with loading and waiting stations, allowing multiple magnetic levitation workpiece stages to dynamically queue and rotate.

[0022] 1.2 Fixed-zone planar coil desktop The coil tabletop is fixedly mounted at the bottom of the vacuum chamber and does not move with any workpiece stage throughout the process, abandoning the traditional moving coil architecture. The substrate is made of high-rigidity, low-thermal-expansion, and high-resistivity silicon carbide material, which can effectively suppress eddy current heating and magnetic field distortion; an independent silicon carbide cold plate constant temperature circuit is installed on the underside of the substrate, ensuring precise temperature control. It continuously removes the Joule heat from the coil, enabling independent thermal control of the drive system.

[0023] The surface of the coil desktop substrate is composed of arrayed planar coil tiles. Each coil tile is a physically independent block, arranged in an array with adjacent seams, covering the entire travel range of the workpiece stage. Within each coil tile, X- and Y-axis racetrack-shaped coil windings are stacked in layers along the Z-axis. The windings within adjacent tiles in the same coordinate direction are electrically connected at the seams, ensuring electrical continuity of the windings in that direction and uninterrupted thrust output. Each tile is physically independent and can be individually disassembled and replaced, with the Joule heat of the coils dispersed to the corresponding cold plate area of ​​each tile. Several adjacent coil tiles are grouped by row or column into a drive zone, with the physical zone being consistent with the drive zone. Each drive zone is equipped with an independent power drive channel, and the number of drive zones is no less than the maximum number of workpiece stages operating simultaneously in the system. Each drive zone can independently output current vectors and calculate thrust torque. The controller activates only the coil tile windings within the corresponding drive zone to output drive force based on the real-time spatial position of the workpiece stage, adapting to the parallel levitation motion of multiple workpiece stages.

[0024] Before leaving the factory, the equipment undergoes full-field thrust matrix calibration, traversing different workpiece stage positions and different coil current conditions, accurately collecting magnetic field crosstalk data under multi-machine synchronous motion, and ensuring calibration accuracy. During equipment operation, the calibration matrix is ​​embedded in the controller in real time to complete feedforward compensation, and the residual is compensated by superimposing the magnetic fields of multiple moving machines. This suppresses magnetic field cross-coupling disturbances between multiple magnetically levitated workpiece stages. The coil tabletop has no rigid mechanical connection to the upper metering truss, achieving mechanical decoupling and suppressing the transmission of vibration and deformation across the system.

[0025] A set of Hall sensors is embedded between the upper surface of the substrate and the lower edge of the winding of each coil tile. Each set consists of no less than six Hall sensing elements arranged orthogonally and differentially, outputting three orthogonal components of the spatial magnetic field directly above the tile and its gradient components along three coordinate directions, for a total of six magnetic field observation components. Based on the deterministic distribution of the magnetic field of the Halbach permanent magnet array along the spatial period, the Hall data of a single tile can be used to inversely solve the local three-dimensional displacement and normal tilt of the non-contact moving permanent magnet directly above the tile. The Hall data of multiple adjacent tiles are jointly solved, and the known poses of each workpiece stage are used as priors for successive approximation to separate the magnetic field contribution generated by each permanent magnet array at this tile.

[0026] The measurement data from the Hall sensor array is collected by the stator-side acquisition circuit and used for: firstly, determining the commutation phase of the drive zone where the tile is located online, and performing redundant verification with the commutation phase calculated based on the overall metering data, so that the independent power drive channels of each drive zone maintain commutation correctness even when the overall metering data is briefly abnormal; secondly, self-calibrating the thrust coefficient of the drive zone online, so that the zone thrust calibration matrix is ​​continuously updated during operation, and is mutually verified with the full-field thrust matrix full-domain calibration data completed when the equipment leaves the factory.

[0027] Third, the magnetic field generated by the Harbach permanent magnet array of other magnetically levitated workpiece stages at the position of this tile is directly measured, the crosstalk of the multi-machine magnetic field is calculated online and the magnetic field crosstalk feedforward compensation matrix is ​​updated in real time, and crosstalk data obtained from factory calibration is used for mutual verification; the Hall sensor group and the grating read heads at the four corners of the workpiece stage floating platform form a heterogeneous redundant observation, so that some position observation functions are undertaken by the stator side, reducing the heat generation and counterweight pressure of the airborne observation device of the workpiece stage.

[0028] The Hall sensor has a spatial resolution on the micrometer scale and is used only for zone commutation, online thrust self-calibration, and online observation of magnetic field crosstalk. The remanence of the Harbach permanent magnet array drifts with temperature, and its magnetic field amplitude is compensated in real time according to the remanence temperature coefficient of the permanent magnet and the temperature measurement data built into the workpiece stage; single-point magnetic field measurement uses the factory-calibrated strength data of the permanent magnet as the prior to the calculation. The Hall sensor does not participate in the nanometer-level position and attitude measurement chain of the workpiece stage. The nanometer-level six-degree-of-freedom position and attitude of the workpiece stage is still obtained by the four sets of follow-up grating read heads at the four corners of the workpiece stage floating platform reading the stationary grating master scale and then jointly calculating.

[0029] The Hall sensor is fixed by ceramic encapsulation and inorganic bonding, and its temperature resistance is no less than that of the cavity. High-temperature baking temperature, while meeting the requirements for low vacuum venting.

[0030] 1.3 Independent Suspended Ultra-Low Expansion Metering Truss The metering truss is made of Class 0 Zerodur / ULE ultra-low expansion microcrystalline glass. It is formed into a space truss structure by welding and inorganic vacuum low-gas bonding process. The whole structure is erected on multiple independent granite vibration isolation supports. The vibration isolation supports are independently arranged on the cavity base and have no rigid connection with the coil table and the vacuum cavity shell, which blocks the transmission of mechanical vibration at the bottom to the metering reference and achieves mechanical decoupling.

[0031] A single, seamless two-dimensional phase grating master scale is rigidly mounted on the lower surface of the truss, without any light-transmitting openings, fully preserving the integrity of the metrological reference. The grating master scale has a grating pitch of 400 nm and is compatible with a 785 nm metrological light source. Four Littrow heterodyne grating readouts are mounted at the four physical corners of the workpiece stage's floating platform, reading the signal from the seamless grating master scale upwards. The multi-channel measurement data from the four readouts is used to jointly calculate the 6DOF position and attitude of the workpiece stage. The readouts serve as the main onboard heat source for the workpiece stage. The truss incorporates a distributed PT1000 high-precision temperature measurement array, ensuring high temperature measurement accuracy. It can monitor the reference temperature field distribution in real time; the truss has a pre-embedded closed flow channel, which is introduced with low-exhaust fluorinated liquid to form an independent constant temperature fluid circuit, ensuring the temperature control accuracy of the truss body. Temperature difference throughout the venue This achieves ultra-high thermal stability of the metrological standard.

[0032] The truss is laid with multiple layers on the lower surface of the coil tabletop. A passivated high-reflectivity radiation shielding foil is used, with a 5 mm heat-insulating air gap reserved between it and the coil tabletop. Through the dual structure of air insulation and radiation shielding, the Joule heat of the coil is blocked from being transferred to the metrological reference through radiation and convection, thus achieving thermal decoupling. An infrared radiation shielding cover is added to the outside of the truss to isolate the radiation disturbance caused by the periodic temperature fluctuations of the workpiece table and avoid thermal deformation of the reference grating.

[0033] An integrated laser interferometric reference target is rigidly connected to the truss body at the corners of the truss, serving as the self-calibration reference for the entire machine. After the equipment has finished baking, after a long period of shutdown, or after the temperature drift exceeds the threshold, the system automatically initiates the self-calibration process and generates... High-resolution LUT compensation table, correcting for grating periodic harmonic nonlinearity error and truss minute thermal deformation, static total uncertainty of metrology subsystem. .

[0034] The FSOC base station optical transceiver is fixedly mounted on the upper part of the truss, sharing the same static metrological reference with the grating master scale. Joint calibration is completed at the factory, and the optical axis offset compensation matrix is ​​fixed to the controller. Base station power and fiber optic backhaul use fixed cabling; only the link between the workpiece stage and the truss is wireless, balancing reference uniformity with the advantages of cableless movement. The grating master scale is equivalent to a parallel flat window for the FSOC communication band, allowing for zero-order direct beam transmission without the need for openings and preventing diffraction crosstalk.

[0035] 1.4 Halbach magnetic levitation workpiece stage All workpiece stages are configured to the same specifications, with standardized hardware structures and magnetic array parameters. Each workpiece stage is individually weighed and precisely counterweighted to ensure that its mass and inertia tensor (including rotational inertia about three axes) are consistent with each other. The mass deviation between any two stages is less than 0.1 g, and the rotational inertia deviation about three axes is less than 0.1%, ensuring the consistency of mass and inertia in the synchronous motion of multiple machines. Each workpiece stage adopts a two-level positioning architecture: a lower Halbach permanent magnet coarse motion suspension + an upper 7-unit PZT six-degree-of-freedom micro-nano compensation platform. The entire process involves non-contact suspension motion without any mechanical support or dragging structure.

[0036] 1.4.1 Halbach Permanent Magnet Coarse Array A recessed groove is formed on the lower surface of the silicon carbide main base plate of the workpiece stage, housing four sets of one-dimensional Halbach permanent magnet arrays arranged at the four corners, achieving an alternating XY diagonal arrangement, totaling 64 permanent magnet units; each array contains 16 permanent magnet units, employing... Flat tile structure, single magnetic block size Single array outer contour The overall outer envelope of the four arrays The four arrays are centrally located within the base plate recess, with a 10 mm safety margin reserved around the base plate. The rated suspension air gap of the workpiece stage is 1.0 mm. Under this air gap, the above four arrays provide a rated suspension force of 600 N and a thrust density of 0.8 N / cm², maintaining a suspension force margin of more than 1.5 times for a 40 kg workpiece stage (gravity of approximately 392 N).

[0037] This invention relates to a one-dimensional Halbach array specifically referring to a magnetic field configuration where the magnetization modulation period is along a single coordinate axis. Two diagonal arrays have magnetization periods with no modulation only along the X and Y directions, while another pair of diagonal arrays have magnetization periods with no modulation only along the Y and X directions. The magnetization directions of the magnetic blocks within each group are as follows: The sequential rotation enhances the driving magnetic field towards the coil tabletop, improving the utilization rate and stability of the levitation thrust. The four one-dimensional arrays are arranged diagonally alternately, and their thrust synthesis generates a parasitic torque around the Z-axis. This parasitic torque is calculated and canceled in real time by the bottom current inner loop according to the differential current of the four arrays, so that the positioning accuracy in the Rz direction is not affected by the arrangement of the four arrays.

[0038] A 0.3-0.5 mm seamless 1J85 permalloy shielding layer is attached between the array's upper surface and the silicon carbide substrate. This single-point grounding effectively absorbs stray magnetic fields escaping upwards from the array, suppressing the stray magnetic field at a height of 100 mm above the workpiece stage to below 0.5 mT. This suppresses magnetic field interference with the signal modulation of the grating read head and FSOC optoelectronic devices. The shielding layer is a thin sheet with grooves etched along a direction perpendicular to the eddy current loop, cutting off the eddy current loop and controlling the eddy current loss during workpiece stage scanning to the order of 1 W. This eddy current loss is included in the workpiece stage thermal management budget. The controller incorporates an eddy current feedforward compensation algorithm to eliminate the induced thrust generated by residual eddy currents. The gaps between the magnetic blocks are filled with a vacuum inorganic adhesive, eliminating organic adhesive volatilization and making it suitable for ultra-high vacuum conditions.

[0039] All workpiece stage Halbach arrays undergo precise magnetic block sorting and pairing, with a total mass deviation of less than 0.1 g for each workpiece stage array and a magnetic flux distribution deviation of less than 0.1 g. This ensures the maximum consistency of quality and inertia in the synchronous motion of multiple machines.

[0040] 1.4.2 7-Unit PZT Micro / Nano Compensation Platform A 99.7% high-purity alumina ceramic reference base is erected above the main base plate via low-heat insulation support columns. Seven PZT drive units are arranged above the base, forming a six-degree-of-freedom micro / nano compensation mechanism: three Z-axis vertical PZTs are positioned... The distribution enables Z-axis translation and Rx / Ry attitude adjustment; two X-axis and two Y-axis horizontal PZTs are symmetrically positioned to achieve X / Y translation and Rz attitude correction, accurately compensating for the residual error of coarse motion suspension.

[0041] Each PZT is equipped with micro-ceramic universal flexible hinges at both ends, allowing Small-angle oscillation eliminates assembly stress and lateral compression, preventing PZT cracking and damage; multiple 0.3 mm thick high-purity alumina ceramic thin springs are arranged on the outer ring of the floating load-bearing platform to form a ring-shaped weakly elastic pre-tightening and reset mechanism, ensuring pre-tightening stiffness. A restoring pre-tension force is continuously applied to the floating platform to ensure that all PZT devices are subjected to only compressive stress throughout the process, avoiding damage from tensile stress and improving the service life and stability of the devices.

[0042] The PZT has a rated working stroke of 0~10 μm and is only used to compensate for residual position errors, micro-vibrations, and attitude tilts during the scanning process; it does not undertake large-stroke XY scanning motion. A 12-inch porous ceramic vacuum chuck is installed on the upper surface of the floating platform for wafer adsorption. A Littrow heterodyne grating reader is installed at each of the four physical corners of the floating platform, for a total of four grating readers. The airborne FSOC optical transceiver is located in the four empty corner positions of the floating platform. The grating reader bracket is made of low-expansion silicon carbide material. A high thermal conductivity coupling block is installed between the heat dissipation surface of the reader housing and the wall of the sealed liquid-cooled isothermal chamber inside the workpiece stage. A coupling block is installed between the reader and the upper surface of the floating platform. A heat-insulating layer directs the heat generated by the read head into the liquid storage chamber instead of conducting it to the floating platform and suction cup. Each read head housing integrates a temperature sensor, whose temperature signal is connected to the controller for real-time correction of the Abbe error compensation matrix. Each read head integrates a 785nm laser, photoelectric receiving components, a PZT tilt correction mirror, and a pre-amplifier circuit, all housed within a μ-metal shielded shell. The four read heads output multiple X / Y / Z measurement raw data, and the controller calculates the complete 6DOF (X / Y / Z / Rx / Ry / Rz) position and attitude of the workpiece stage based on the multiple sets of measurements from the four read heads. The read head light source is 785 nm with 2048x electronic subdivision. The system has a built-in 10 kHz update rate Abbe error compensation matrix to correct the Abbe offset introduced by the Rx / Ry / Rz attitude in real time. The four grating read heads are the main steady-state onboard heat source for the workpiece stage. Together with the PZT micro-nano compensation platform and the micro-circulation pump, they generate approximately 15 W of steady-state heat. This heat generation is included in the workpiece stage's thermal management budget. The heat is uniformly smoothed by the PFPE sealed liquid storage chamber inside the workpiece stage, controlling the internal temperature gradient. To avoid false Rx / Ry angle errors introduced by the bending of the silicon carbide substrate, the accuracy responsibility of the two-stage positioning architecture is divided as follows: the lower Halbach magnetic levitation coarse motion stage is only responsible for large-stroke scanning motion and disturbance suppression in the mid-frequency and above frequency bands. Its tracking residual (including position-related thrust ripple and low-frequency drift caused by the flexible deformation of the magnetic plate) is absorbed by the upper PZT six-degree-of-freedom micro-nano compensation platform within a closed loop within a stroke of 0~10 μm. The closed-loop bandwidth of the PZT micro-nano compensation platform is not less than 1 kHz (not less than twice the upper limit of the coarse motion stage position loop bandwidth). The spectral energy of the tracking residual of the coarse motion stage is mainly concentrated in the low-frequency band below 100 Hz, and the residual in this low-frequency band is absorbed by the PZT stage closed loop. The high-frequency residual above 500 Hz is suppressed by the coarse motion stage itself. The dynamic synthetic positioning error described later... This refers to the synthetic positioning error after closed-loop correction by the PZT micro-nano compensation platform, with the stationary grating master scale as the measurement benchmark, rather than the tracking error of the magnetic levitation coarse motion stage. The Rx / Ry / Rz attitude measured by the four sets of grating read heads is corrected by the aforementioned 10 kHz Abbe error compensation matrix in a closed-loop manner, so that the attitude angle control accuracy reaches the nrad level, and the Abbe residual is compressed to the value listed in the subsequent error budget.

[0043] 1.4.3 Airborne energy storage subsystem and sidewall thermoelectric composite charging and conducting mechanism The workpiece stage has an internal cavity that integrates an airborne energy storage subsystem, namely a parallel energy storage module of solid-state lithium battery and supercapacitor. It is a vacuum-compatible cableless power supply architecture. The entire levitation scanning process is powered by the airborne module without any external power supply cables. The supercapacitor handles the instantaneous peak power consumption of PZT attitude adjustment and grating read head operation, while the solid-state battery provides steady-state continuous power supply, which is suitable for power consumption requirements under all working conditions.

[0044] The workpiece stage has two sets of adjacent sidewalls with TiN nano-polished thermoelectric composite contact surfaces. Sidewall A is the positive electrode conductive and heat-conducting surface, and sidewall B is the negative electrode recirculation heat-conducting surface. The TiN coating thickness is 50~200 nm, with a Ti transition layer to improve adhesion, and the surface is ultra-precision polished with Ra. It features ultra-high vacuum with low gas escaping, resistance to vacuum cold welding, high wear resistance, and high electrical and thermal conductivity, and can be replaced with a TaN coating of the same performance. The frame is equipped with low-temperature cold plate contacts with corrugated elastic buffers at the corresponding workstations, which can compensate for minor deviations in the workpiece table's positioning and ensure synchronous and uniform pressing of both contact surfaces.

[0045] The workpiece stage only completes mechanical contact at the loading and stopping station. The clamping force is precisely controlled by the magnetic levitation XY lateral closed-loop force. The line of action of the clamping force passes through the center plane of gravity of the workpiece stage, and the clamping torques on both sides cancel each other out, without introducing Rx / Ry attitude disturbances. The bonding action adopts a pure axial approach, without relative slippage, avoiding scratches on the coating and the generation of vacuum particles. During the levitation motion stage, the TiN contact surface maintains a safe gap of 0.8~1.2 mm, maintaining non-contact and without any dragging or friction disturbances.

[0046] A single stop and bonding process simultaneously completes high-current battery charging and workpiece stage heat dissipation. The Joule heat from the charging contact is incorporated into the overall machine thermal budget. The system can calibrate contact resistance and contact thermal resistance based on real-time clamping force, dynamically correct the thermal balance model, and compensate for parameter drift caused by long-term wear of the coating. The SiC substrate is insulated, with only the conductive TiN exposed in the sidewall contact area, while the remaining areas are insulated to avoid the risk of leakage.

[0047] 1.5 Sealed Liquid Storage Self-Equalizing Temperature Thermal Management Subsystem This invention abandons the traditional follower-type water-cooled drag chain architecture and adopts an integrated permanent sealed liquid storage and temperature equalization cavity built into the workpiece stage. The cavity is equipped with honeycomb partitions to effectively suppress liquid sloshing, reduce dynamic center of gravity disturbance of magnetic levitation motion, and improve suspension stability. The cavity integrates a low-outgas micro circulation pump to drive the coolant to circulate autonomously.

[0048] The coolant uses high-purity degassed perfluoropolyether (PFPE) medium, which has extremely low saturated vapor pressure, ultra-high vacuum compatibility, electrical insulation, and low heat volatility. The liquid is permanently sealed inside the workpiece stage, with no external fluid pipelines or plug-in connectors, eliminating the risk of vacuum leakage and fluid drag chain disturbance.

[0049] The system's heat sources mainly include onboard steady-state heat sources and pulse heat sources: the grating read head, PZT micro-nano compensation platform, and micro circulation pump together generate approximately 15 W of steady-state heat. During EUV exposure, the wafer absorbs photons to generate pulse heat, which is preferentially intercepted by the upper electrostatic chuck (ESC) microchannel cooling system, significantly reducing the pulse heat load conducted downwards to the workpiece stage and lowering the workpiece stage's thermal management pressure.

[0050] Thermal management adopts a steady-state control mode of "cycled heat storage + cycled heat dissipation", achieving zero thermal balance in single-wafer operation cycles: During the exposure and scanning stage, the TiN contact surface separates, and there is no thermal or electrical contact between the workpiece stage and the frame. An internal micro-pump drives the PFPE liquid to circulate at high speed, quickly smoothing out hot spots and local temperature differences in the four corners, and strictly controlling the internal temperature gradient of the workpiece stage. Only a small amount of heat storage overall ( (Scale); during the docking stage, the TiN contact surface is pressed tightly against the cold plate of the frame, and all the accumulated heat stored in this cycle is discharged through solid contact heat conduction, and thermal balance adjustment is completed in combination with charging heat dissipation.

[0051] The system establishes a precise set of thermal balance equations, calibrates the relationship between clamping force and contact thermal resistance and contact resistance in real time, and dynamically adjusts the heat exchange temperature difference and heat exchange time to ensure long-term thermal stability and suppress the accumulation of thermal drift.

[0052] 1.6 Dual-band FSOC Multiple Access Parallel Wireless Optical Communication Subsystem This invention employs a wireless communication subsystem. In this embodiment, the wireless communication subsystem is a drag chain-free dual-band FSOC free-space laser communication subsystem, eliminating the need for moving optical fibers, enabling cableless bidirectional data interaction between the workpiece stage and the main controller, avoiding radio frequency electromagnetic interference, not affecting grating measurement and magnetic levitation field stability, and its low latency characteristics are adapted to the nanometer-level servo closed-loop bandwidth requirements.

[0053] The FSOC system consists of a fixed base station optical transceiver on the upper truss and an optical transceiver mounted on the workpiece table. The base station and the grating master share the same static constant-temperature metrological reference. After joint calibration, their relative pose remains stable over a long period. The base station coordinates are incorporated into the grating metrological coordinate system, with no additional reference drift. The power supply and fiber optic backhaul on the base station side are all fixed cabling. The wireless link only covers the gaps in the workpiece table's movement, balancing stability and the advantages of cableless movement. The grating master has a grating pitch of 400 nm. The FSOC's operating wavelengths of 635 nm and 1550 nm are both greater than the grating pitch, placing it in the subwavelength region. The beam only has zero-order direct transmission, with no diffraction or beam splitting interference, eliminating the need for a light-passing aperture.

[0054] The system adopts a core architecture of "downlink sector floodlighting + uplink cat's-eye retroreflection + wavelength division multiple parallel operation" to solve the defects of traditional point-to-point link breakage and inability to coordinate multiple machines: the downlink uses the base station sector floodlight transmitting antenna to expand the 1550 nm high-speed communication light beam into a full-coverage travel cone, which can be stably received at any position of the workpiece stage without the need for precise acquisition; the uplink relies on the airborne cat's-eye retroreflector to reflect back along the original incident path, and the uplink data is loaded by the airborne modulator, without the need for continuous tracking and alignment, and the 2 kHz bandwidth fast reflector (FSM) is only used for uplink beam pointing compensation; the multi-machine parallel operation adopts wavelength division multiple access technology, which allocates an independent 635 / 655 / 675 nm beacon wavelength to each workpiece stage, and the base station receives angle measurement in parallel through dichroic devices and a four-quadrant detector (QD) array. A single base station can serve three or more workpiece stages at the same time, and the angle measurement data is synchronously used as an attitude redundancy measurement signal.

[0055] The spectrum employs a three-color isolation design, with the 635 nm beacon band, 1550 nm communication band, and 785 nm grating metrology light source band spaced apart. Combined with a narrowband interference filter, the optical path crosstalk isolation is improved. No optical interference; communication data update rate of 10 kHz, link stability bit error rate All optomechanical components meet Baking, ultra-high vacuum, and low gas output performance.

[0056] The FSOC link reports optical power, bit error rate, and latency jitter parameters in real time, which are then sent to the Smith predictor module of the controller to accurately compensate for fixed latency and random jitter. When link quality deteriorates, the closed-loop bandwidth is adaptively reduced to maintain system stability margin. In the event of a severe link interruption, the workpiece stage is triggered for smooth docking protection to avoid failure risks. All airborne optical transceivers are strictly ballasted, and their weight deviations are included in the overall mass consistency ballast of the workpiece stage to ensure that the mass and inertia consistency of the synchronous motion of multiple machines are not disrupted.

[0057] 1.7 Hierarchical Cooperative Motion Controller The controller adopts a three-layer bandwidth hierarchical isolation architecture to avoid loop coupling resonance, achieving high-precision and high-robust control. The hierarchical division of labor is clear and the bandwidth gradient is reasonable. 1) Upper Trajectory Momentum Coordination Layer (0.1~10 Hz): Receives process exposure instructions, completes workpiece stage station queuing and scheduling, Jerk amplitude-limited S-curve trajectory generation, and real-time monitoring of total system momentum; adaptively adjusts scanning speed according to momentum threshold, automatically reduces speed or smoothly stops when the threshold is exceeded, ensuring momentum cancellation accuracy and system safety.

[0058] 2) Middle layer 6DOF MIMO robust control layer (100~500 Hz): Employing Hybrid Sensitivity Controller norm Equipped with a DOB disturbance observer, it monitors and compensates for lumped disturbances such as magnetic field crosstalk, thermal deformation, and communication jitter in real time; it features a multi-order notch filter to accurately avoid resonant modes of truss and worktable structures; and it incorporates a Smith predictor to compensate for FSOC link delay jitter. The controller can tolerate dynamic parameters. Perturbation, phase margin Gain margin It is extremely robust.

[0059] 3) Inner Loop Current of Bottom Coil (25 kHz): Each coil is equipped with an independent PI high-speed current closed loop, with current hard limiting to prevent coil overheating and force saturation. The electromagnetic actuator is equivalent to an ideal force source, reducing the difficulty of outer loop control and increasing the bandwidth separation ratio between inner and outer loops. To avoid coupled oscillations.

[0060] The system establishes an RSS square root full-link error budget model, comprehensively considering all dimensions of random errors, including grating measurement error, magnetic levitation tracking error, thermal drift error, momentum cancellation residual disturbance, magnetic field crosstalk error, communication jitter error, and PZT compensation residual. The static and dynamic standard deviations of each error source are shown in the table below. After synthesis according to the square root, the static synthesis positioning error is 0.169 nm RMS, and the dynamic synthesis positioning error is 0.265 nm RMS, which meets the requirements. The design specifications clearly distinguish the independent technical definitions of 1nm optical imaging resolution and 0.3nm workpiece stage motion measurement error.

[0061] Error source Static σi (nm RMS) Dynamic σi (nm RMS) Grating measurement error 0.099 0.120 Magnetic levitation tracking error 0.100 0.160 Thermal drift error 0.055 0.070 Momentum cancels residual disturbance 0 0.105 Magnetic field crosstalk error 0.045 0.060 Communication jitter error 0.035 0.060 PZT compensated residual 0.050 0.085 RSS (Rectangular and Root) Synthesis 0.169 0.265 The machine is equipped with multi-level fault protection logic, which monitors the workpiece stage posture, system momentum, FSOC link status, coil current, and metering signal validity in real time. It triggers alarms and speed reductions and provides smooth stopping protection in stages to avoid the risk of high-speed collisions.

[0062] 2. Complete workflow of this system S1 Dock Pre-processing Stage: The standby workpiece stage travels at low speed to the loading station. Magnetic levitation lateral force controls the dual TiN contacts to vertically press against the cold plate contacts of the frame, simultaneously completing onboard battery charging and periodic heat storage removal; wafer loading, vacuum pre-alignment, and cavity baking pre-processing are completed. After the operation is completed, the lateral clamping force is released, the contacts separate, and the workpiece stage enters a cable-free levitation state. Data exchange is completed via onboard energy storage and FSOC wireless optical communication, awaiting station scheduling. Other workpiece stages can wait in line at the waiting station, enabling parallel pre-processing procedures.

[0063] S2 Scanning Exposure and Momentum Cancellation Stage: Two identical, equally weighted workpiece stages are scheduled to the spatially symmetrical process station and counter-processing station, respectively. The workpiece stage scheduled to the counter-processing station automatically serves as the momentum cancellation workpiece stage. The process station workpiece stage carries the wafer and performs high-speed S-curve scanning exposure, while the counter-processing station workpiece stage simultaneously reads global 6DOF metrological data and performs precise reverse symmetrical motion. Relying on momentum conservation, it cancels the inertial force and overturning moment generated by the process scanning, suppressing the overall reference disturbance. Throughout the scanning process, the workpiece stage has no contact with any drag chains, cables, or pipelines. The internal PFPE liquid self-circulates to equalize the temperature and maintain a stable temperature field. The layered controller completes trajectory control, disturbance suppression, error compensation, and link monitoring throughout the process.

[0064] S3 Multi-Station Parallel Operation: The system supports configurations of 3, 4 or more workpiece stages, and can simultaneously set up multiple independent process counter-counter stations for pairing. After a single scan is completed, the workpiece stages are unpaired and re-queued for rotation. Workpiece stages that have already been processed return to the loading / waiting station for preprocessing, and newly scheduled workpiece stages are put into operation, realizing parallel flow of exposure process and wafer preprocessing, which greatly improves the overall throughput.

[0065] S4 Periodic Self-Calibration of the Whole Machine: When the cavity baking is completed, the machine is shut down for a long time, or the temperature drift of the metering gantry exceeds the threshold, the system automatically starts the self-calibration process, scans the reference target on the workpiece stage, updates the LUT error compensation table and Abbe error compensation matrix, and refreshes the metering, optical communication and magnetic field compensation parameters to maintain long-term operational accuracy.

[0066] S5 Status Monitoring and Fault Handling: The controller monitors the operating parameters of the entire system in real time. When the momentum exceeds the soft threshold, it automatically reduces speed and suppresses vibration. When hard faults such as momentum over-limit, communication link failure, metering signal failure, and current over-limit are triggered, the workpiece stage is smoothly reduced and stopped through the interpolation algorithm to avoid collision risks.

[0067] 3. Beneficial effects 1. Revolutionizing the traditional architecture and eliminating drag chain disturbances: Adopting a fixed coil moving magnet and a static grating reference architecture, all drag chain structures such as power supply cables, signal optical fibers, and water-cooled hoses that follow the workpiece stage are eliminated. This eliminates inherent defects such as cable friction, inertial disturbances, vacuum particle contamination, and pipeline leakage from the source, significantly improving motion stability and cavity cleanliness.

[0068] 2. Achieve true cable-free levitation operation under all working conditions: During the levitation scanning stage, there is no physical connection between the workpiece stage and the frame. It is powered autonomously by the onboard battery and supercapacitor parallel module. FSOC multi-access parallel optical communication enables cable-free high-speed data interaction. Only the docking station completes intermittent power replenishment and heat dissipation through TiN composite contacts, taking into account both the advantages of cable-free movement and the requirements of long-term steady-state operation.

[0069] 3. Highly scalable multi-machine collaborative architecture: The workpiece stage has no fixed functional identity; the workstations are dynamically rotated and roles are automatically switched; all workpiece stages are precisely counterweighted to achieve equal weight, with a mass deviation of less than 0.1 g and a rotational inertia deviation of less than 0.1% around the three axes. Any workpiece stage scheduled to the counterweight station automatically undertakes the reverse motion without the need for mass screening and pairing; compared with the existing balancing mass scheme that uses an inert iron block as the balancing mass and requires a low-bandwidth drift control loop, this invention uses another in-service workpiece stage of equal weight as the balancing mass. Both stages are actively controlled and kinematically symmetrical and opposite, with reaction force cancellation without additional drift loops and no residual reaction force injected into the foundation; it supports expansion to 3 or more workpiece stages, and can build multiple momentum cancellation units, allowing preprocessing and exposure processes to run in parallel, significantly improving the overall throughput and breaking through the hardware bottleneck of the traditional dual-workpiece stage architecture.

[0070] 4. Precise suppression and online calibration of crosstalk across the entire magnetic field: The coil tabletop adopts an array of planar coil tiles, with physical partitions consistent with the drive partitions; Hall sensor groups are embedded in each tile to achieve online determination of the commutation phase of each partition, online self-calibration of the thrust of each partition, and online direct measurement of crosstalk between multiple machines. The factory calibration of the entire thrust matrix and the online calibration during operation are mutually verified, controlling the residual crosstalk between multiple machines to within 0.08 mT; combined with the back-side permalloy magnetic shielding and eddy current compensation of the workpiece stage, the coupling disturbance of the magnetic field between multiple machines is suppressed, ensuring the accuracy of magnetic levitation thrust and the stability of levitation.

[0071] 5. Dual Decoupling of Mechanics and Thermodynamics: The independent suspended metering truss has no rigid connection with the drive system, achieving mechanical decoupling; the metering truss and the coil tabletop have dual independent constant-temperature circuits, coupled with thermal insulation gaps, radiation shielding, and a sealed self-equalizing structure for the workpiece stage, blocking the coupling paths of heat conduction and radiation, and suppressing thermal drift errors; the seamless grating master scale ensures uniform coordinate reference across the entire domain, eliminating splicing errors; simultaneously, four sets of Littrow heterodyne readers at the four corners of the workpiece stage jointly calculate the complete six-degree-of-freedom attitude, with multi-point redundancy suppressing Abbe errors, and the heat generated by the readers is incorporated into the thermal budget of the sealed liquid storage chamber of the workpiece stage, strictly constraining the temperature gradient within the stage. This avoids false angular errors introduced by thermal bending of the substrate.

[0072] 6. Novel TiN thermoelectric integrated contact design: Symmetrical pressing structure on both sides, torque self-cancellation without posture disturbance; pure axial bonding without slippage, no detectable vacuum particles are generated; charging and solid-state heat conduction and dissipation are completed simultaneously in a single bonding, accurately controlling the thermal balance of a single cycle, and adapting to long-term continuous operation conditions.

[0073] 7. High-stability multiple access FSOC optical communication: Optimized sector floodlight + cat's eye backlight + wavelength division architecture, solving the core defects of traditional point-to-point link breakage and inability of multiple machines to run in parallel. A single base station covers the entire route and serves multiple workstations. Spectral isolation without crosstalk, accurate delay compensation, and significantly improved link stability and redundancy.

[0074] 8. High-precision robust control and comprehensive error management: A three-layer hierarchical control architecture avoids loop coupling. Robust control combined with DOB disturbance observer adapts to parameter perturbations, while notch filtering and Smith delay compensation precisely suppress various disturbances; RSS end-to-end error budgeting achieves sub-nanometer precision control and dynamic error positioning. It precisely matches the requirements of 1nm optical resolution lithography equipment; the coarse magnetic levitation stage only undertakes the large stroke motion and mid-to-high frequency disturbance suppression, and its tracking residual is absorbed by the upper PZT micro-nano compensation platform in a closed loop within the micron-level stroke. The sub-nanometer level synthesis accuracy is guaranteed by the fine adjustment stage rather than the coarse motor, which is isomorphic to the division of labor logic of long-stroke and short-stroke in lithography equipment. Attached Figure Description

[0075] Figure 1 This is a schematic diagram of the overall cross-sectional layout of the system of the present invention.

[0076] Figure 2 This diagram illustrates the topological division of labor and workstation layout for three workpiece stations, showing the process station, the counter-clamping station, and the loading / waiting station.

[0077] Figure 3This is a schematic diagram of the topology and workstation queuing scheduling of 4 workstations, showing the spatially symmetrical arrangement of process stations, counter-counter stations, and the workstation flow queuing relationship.

[0078] Figure 4 A schematic diagram of four sets of one-dimensional Halbach permanent magnet arrays arranged at the four corners.

[0079] Figure 5 This diagram illustrates the tile splicing and drive partitioning of a fixed partitioned planar coil tabletop. The coil tabletop is composed of arrayed planar coil tiles spliced ​​together with seams. Each tile contains X and Y track-shaped coil windings stacked in layers along the Z direction. The drive partition in the diagram is composed of one or more coil tiles grouped together, and Hall sensor groups embedded in each coil tile are shown.

[0080] Figure 6 This is a cross-sectional schematic diagram of a single magnetic levitation workpiece stage, showing from top to bottom the grating read head, PZT micro-nano compensation platform, airborne energy storage module, silicon carbide base plate, Halbach array, and docking contacts.

[0081] Figure 7 This is a schematic diagram of a hierarchical collaborative motion controller architecture.

[0082] Figure 8 This is a schematic diagram of the mechanical and thermal decoupling of the Zerodur metering truss, featuring dual independent isothermal circuits and radiation shielding.

[0083] Figure 9 This is a schematic diagram of the optical path principle for dual-band FSOC sector floodlight coverage and retroreflection multiple access. The airborne cat's eye retroreflector reflects back along the original path, and multiple units are located in parallel according to the beacon wavelength.

[0084] Figure 10 This is a schematic diagram of the components of the RSS multi-source error budget.

[0085] Marker Lookup Table mark name Maps appear 1 vacuum chamber Figure 1 2 Fixed partition planar coil desktop (including silicon carbide substrate) Figure 1 、 Figure 4 、 Figure 5 、 Figure 6 、 Figure 8 3 Zerodur Measurement Truss Figure 1 、 Figure 8 、 Figure 9 4 Grating master scale (785 nm two-dimensional phase grating, grating pitch 400 nm) Figure 1 、 Figure 6 、 Figure 8 、 Figure 9 5 Magnetic levitation workpiece stage (5a~5c / 5-1~5-4 are specific stages) Figure 1 、 Figure 2 、 Figure 3 、 Figure 9 6 Halbach permanent magnet array (6a-6d: four sets of one-dimensional arrays) Figure 1 、 Figure 4 、 Figure 6 7 Grating reader (4 sets per unit, arranged at the four corners of the floating platform, Littrow heterodyne type) Figure 1 、 Figure 2 、 Figure 3 、 Figure 6 、 Figure 7 、 Figure 9 8 FSOC base station optical transceiver (fixed to the upper part of the truss, stationary) Figure 1 、 Figure 9 9 FSOC airborne optical transceiver Figure 1 、 Figure 6 、 Figure 9 10 Dockable thermally conductive composite contact (workpiece stage side, TiN / TaN coating) Figure 1 、 Figure 2 、 Figure 6 11 Rack contact base (including bellows elastic buffer) Figure 6 12 Granite vibration isolation pier Figure 1 、 Figure 8 13 PZT Six-DOF Micro / Nano Compensation Platform (7 Units) Figure 1 、 Figure 6 14 silicon carbide substrate Figure 1 、 Figure 4 、 Figure 6 15 Permalloy (1J85) shielding layer Figure 1 、 Figure 4 、 Figure 6 16 High-reflectivity radiation shielding foil Figure 8 17 Silicon carbide cold plate constant temperature circuit ( ) Figure 1 、 Figure 4 、 Figure 5 、 Figure 6 、 Figure 8 18 Fluorinated liquid constant temperature circuit ( ) Figure 8 19 wafer Figure 1 、 Figure 6 20 Porous ceramic vacuum suction cup (12 inches) Figure 1 、 Figure 6 21 Drive partitions (211-213) Figure 5 22 Arrayed planar coil tiles (internal layered stacking of X-axis / Y-axis raceway windings) Figure 1 、 Figure 4 、 Figure 5 、 Figure 6 、 Figure 8 23 Airborne energy storage subsystem (parallel module of solid-state battery and supercapacitor) Figure 6 24 Layered Cooperative Motion Controller (24-1 Upper Layer / 24-2 Middle Layer / 24-3 Bottom Layer) Figure 7 25 Process station Figure 1 、 Figure 2 、 Figure 3 26 Hedging station Figure 1 、 Figure 2 、 Figure 3 27 Material loading station Figure 2 、 Figure 3 28 Waiting workstation Figure 2 、 Figure 3 29 Laser Interferometric Reference Target Figure 8 30 Fast Mirror (FSM, bandwidth) ) Figure 9 31 Four-quadrant detector (QD) Figure 9 32 Read head μ metal shielding shell Figure 6 、 Figure 9 33 Alumina ceramic reference base Figure 1 、 Figure 6 34 Ceramic universal flexible hinge ( ) Figure 6 35 Alumina ceramic thin spring preload reset mechanism (preload stiffness) ) Figure 6 36 Infrared radiation shield Figure 8 37 Insulating air gap (5 mm) Figure 8 38 Airborne cat's eye retroreflector (reflects back along the original path, no tracking or switching required) Figure 9 39 Base station sector floodlight transmitting antenna (covering the entire range) Figure 9 40 Hall sensor array (embedded within each coil tile) Figure 5 41 Coil tile splicing Figure 5 Detailed Implementation

[0086] The following four embodiments are used to illustrate the technical solution of the present invention.

[0087] Example 1: Minimum feasible configuration of 3 workpiece stages This embodiment represents the minimum hardware configuration of the system. The cavity is equipped with symmetrical process stations and counter-clamping stations, as well as loading and waiting stations. There are a total of 3 magnetically levitated workpiece stages with identical hardware. The workpiece stages do not have fixed functional identities; their roles are determined by the station they are in. The workpiece stages can queue and move between the loading, waiting, process, and counter-clamping stations. All workpiece stages are precisely balanced to ensure equal weight, and the mass and inertia tensor of each stage are consistent with each other. Any stage scheduled to the counter-clamping station automatically becomes a workpiece stage for momentum cancellation.

[0088] A fixed, partitioned planar coil table is installed at the bottom of the vacuum chamber. The surface of the coil table substrate is composed of arrayed planar coil tiles spliced ​​together with seams. Within each tile, X- and Y-axis racetrack-shaped coil windings are stacked in layers along the Z-axis. The windings within adjacent tiles in the same direction are electrically connected at the seams, while the tiles themselves are physically independent and replaceable in sections. Hall sensor groups are embedded within each tile. The system is divided into three drive zones by row grouping, each drive zone equipped with an independent power drive channel to complete the full-field thrust matrix calibration and magnetic field crosstalk feedforward compensation. An independent suspended Zerodur metering truss is erected above the chamber, supported by granite vibration isolation supports, mechanically decoupled from the chamber and coil table. The truss is equipped with an independent fluorinated liquid constant-temperature circuit; the coil table is equipped with an independent silicon carbide cold plate constant-temperature circuit; a 5 mm thermal insulation air gap is set in the middle. A high-reflectivity radiation shielding layer blocks thermal radiation coupling.

[0089] Each workpiece stage base plate is fitted with four sets of one-dimensional Halbach permanent magnet arrays arranged at the four corners; each set contains 16 permanent magnet units. Planar layout, single piece Single outer contour Two diagonally opposite sets of one-dimensional Halbach arrays have magnetization periods along the X-direction and no magnetization periods in the Y-direction; another two diagonally opposite sets of one-dimensional Halbach arrays have magnetization periods along the Y-direction and no magnetization periods in the X-direction; the magnetic blocks within each set are arranged in a one-dimensional Halbach configuration, and the magnetization directions of adjacent magnetic blocks within the set are... The four arrays are rotated sequentially. The upper side of each array is covered with a 0.4 mm thick 1J85 permalloy shielding layer and grounded at a single point to suppress upward stray magnetic fields. The upper part of the workpiece stage has a 7-unit PZT micro-nano compensation platform with an alumina ceramic spring pre-tightening mechanism to ensure that all PZTs only bear compressive stress. A set of Littrow heterodyne grating read heads is arranged at each of the four corners of the floating platform. The four sets of read heads jointly calculate the 6DOF attitude. The read heads, PZT micro-nano compensation platform and micro circulation pump together generate about 15 W of steady-state heat. The heat is uniformly heated by the sealed PFPE liquid storage chamber inside the workpiece stage, and the internal temperature gradient is controlled to be less than 50 mK.

[0090] The workpiece stage is equipped with an onboard solid-state battery and supercapacitor parallel energy storage module, an onboard FSOC optical transceiver, and a μ-metal shielded encapsulated grating read head; the workpiece stage base plate is configured with TiN thermoelectric composite contacts on the sidewalls. A single, seamless 785nm two-dimensional phase grating master scale is fixed to the lower side of the metrology truss; laser interferometric reference targets are integrated at the corners of the truss, supporting automatic self-calibration of the entire unit. The FSOC base station optical transceiver is fixed to the upper part of the metrology truss, establishing a bidirectional optical communication link with the three workpiece stage onboard optical transceivers.

[0091] The hierarchical collaborative motion controller executes the workstation queuing and scheduling logic: one workpiece stage is assigned to the loading station, where magnetic levitation lateral force presses the TiN contacts, completing charging and heat conduction, and performing wafer loading pre-alignment; one workpiece stage is scheduled to the process station for high-speed scanning exposure; one workpiece stage is scheduled to the counter-clamping station, automatically acting as a reverse motion workpiece stage to perform synchronous reverse motion to achieve momentum cancellation; the workpiece stages at the loading station can complete the pre-processing of the next wafer in parallel. In this embodiment, the RSS dynamic synthesis positioning error is 0.265 nm RMS, which meets the requirements. The specifications are compatible with lithography equipment with 1 nm optical resolution.

[0092] Example 2: Queueing and scheduling configuration for 4 workpiece workstations Based on Example 1, an additional magnetically levitated workpiece stage is added, resulting in a total of four identical workpiece stages. The cavity is spatially symmetrically configured with process stations and counterbalancing stations, as well as loading and waiting stations. All workpiece stages have no fixed process / counterbalancing role and can queue and move between loading, waiting, process, and counterbalancing stations. The four workpiece stages are precisely counterweighted to ensure consistent mass and inertia tensor; any stage scheduled to the counterbalancing station automatically becomes a momentum-counterbalancing workpiece stage. All four workpiece stages utilize four sets of one-dimensional Halbach permanent magnet arrays at the four corners; each set contains 16 permanent magnet units. Planar arrangement; all workpiece arrays are sorted and paired by magnetic blocks, with the total mass deviation of each array less than 0.1 g and the magnetic flux distribution deviation... Each workpiece stage floating platform has four sets of Littrow heterodyne grating read heads arranged at its four corners. The heat generated by the read heads is uniformly controlled by a sealed PFPE liquid storage chamber.

[0093] Operating Condition 1: The controller schedules one workpiece stage to the process station for loaded scanning exposure, and schedules another workpiece stage to the counter-current station, which automatically acts as the reverse motion workpiece stage to perform reverse synchronous momentum cancellation; the remaining two workpiece stages are respectively at the loading station for wafer replacement, pre-alignment baking, and waiting in line at the waiting station. After scanning is completed, all workpiece stages re-enter the queuing sequence.

[0094] Working condition 2: After completing one round of exposure, the workpiece stage is rescheduled and reassigned, and the workpiece stage is moved to the process station and the counter-exposure station to carry out exposure operations alternately, which effectively improves the throughput of the whole machine.

[0095] The remaining mechanical, electromagnetic, metrological, thermal control, communication, and control parameters are completely consistent with those in Example 1. Four units are one configuration of this invention, which can be expanded to five or six workpiece stages; two sets of momentum cancellation pairing groups can be set simultaneously, allowing multiple standby preprocessing stages to process multiple wafers in parallel.

[0096] Example 3: Fault Protection Operation Example Throughout the entire normal operation of the system, the hierarchical collaborative motion controller monitors in real time: FSOC link optical power, communication bit error rate, coil output current, workpiece stage 6DOF position and attitude, system real-time total momentum, and the validity of grating metering signals.

[0097] When the total momentum of the system exceeds the set soft threshold, the upper trajectory momentum coordination layer automatically reduces the workpiece stage scanning acceleration and scanning speed to reduce inertial disturbances. When any of the following hard fault conditions occur: total momentum exceeds the hard threshold, FSOC communication link is lost, grating metering signal fails, or coil current exceeds the limit, the controller does not execute an emergency stop. Instead, it uses an interpolation algorithm to output a smooth transition control command, driving all suspended workpiece stages to smoothly decelerate and slowly fall back to the mechanical support docking position, avoiding high-speed collisions of the workpiece stages and achieving safe fault handling.

[0098] Example 4: Example of a vibration-damping decoupling control method for reverse motion of multiple workpiece stages The following S1 to S5 are an overview of the system workflow, which corresponds to S01 to S05 of the method claims.

[0099] This embodiment implements a multi-workpiece stage reverse motion anti-vibration decoupling control method. This method is based on the aforementioned system of three or more magnetically levitated workpiece stages. The vacuum chamber is equipped with spatially symmetrical process stations and counter-processing stations, as well as loading and waiting stations. The workpiece stage hardware has no fixed functional identity; its role is determined by its location within the specified station. The workpiece stage can queue and move between loading, waiting, process, and counter-processing stations. The method execution steps are as follows: S01. Pre-processing docking steps: The magnetic levitation workpiece stage to be operated is dispatched to the loading station, and the docking and heat-conducting composite contact is vertically pressed. The built-in power storage module of the workpiece stage is charged and the heat of the workpiece stage is discharged. After the wafer loading and pre-alignment are completed, the workpiece stage is dispatched to the process station or the counter-clamping station according to the system queuing and dispatching logic. After the dispatching is completed, the contact is separated, and the workpiece stage enters the levitation cableless working state. It is powered by the power storage module and completes bidirectional measurement data and control command interaction through the dual-band FSOC free space laser communication subsystem.

[0100] S02, Momentum cancellation station automatic allocation steps: The two workpiece stages are respectively scheduled to the spatially symmetrical process station and the counterweight station; each workpiece stage is a workpiece stage of the same specification and is precisely counterweighted, with the mass and inertia tensor being consistent with each other. The workpiece stage scheduled to the counterweight station automatically becomes the reverse motion workpiece stage, and the workpiece stage in the process station becomes the process scanning workpiece stage and carries the wafer to be exposed.

[0101] S03, Reverse Momentum Cancellation Scanning Step: The process scanning stage at the control station performs high-speed scanning exposure motion according to the Jerk limiting S-curve trajectory; the reverse motion stage at the synchronous control station reads the global 6DOF position and velocity information output from the Zerodur metering gantry and performs synchronous reverse motion with the process scanning stage, relying on momentum conservation to cancel the inertial force and overturning moment generated by the process scanning stage; the stage remains suspended throughout the scanning process, without the constraints of power cables or signal cables; this step uses a layered collaborative motion controller for control; the upper-layer trajectory momentum coordination layer completes station queuing scheduling, trajectory generation, and total momentum monitoring; the middle-layer MIMO robust control layer combines a DOB disturbance observer to compensate for magnetic field crosstalk and thermal deformation lumped disturbance; the bottom-layer current inner loop outputs the coil drive current; the Smith predictor is used to compensate for the delay and jitter of the wireless optical communication link, and a notch filter is set to avoid modal resonance of the gantry and stage structure.

[0102] S04, Workstation Rotation Queuing Steps: After a single round of scanning exposure is completed, the current workpiece stage pairing relationship is released; the controller reallocates workpiece stage positions according to the queuing scheduling logic, scheduling workpiece stages that have completed their work to the loading station or waiting station, and reallocating workpiece stages that are in the queue to be processed to the process station or counter-clocking station, and entering the next round of scanning operation; the remaining workpiece stages perform wafer preprocessing, vacuum baking, or standby queuing in parallel at the loading station / waiting station; the system can simultaneously build multiple sets of independent "process station counter-clocking station" pairings, and multiple pairs perform reverse momentum cancellation scanning in parallel, with multiple workpiece stages participating in the queuing flow at the same time.

[0103] S05. Status Monitoring and Fault Handling Procedures: Throughout the entire system operation process, the total momentum of the system, the optical power and bit error rate of the FSOC wireless optical communication link, the 6DOF position and attitude of the workpiece stage, the coil drive current, and the validity of the grating metering signal are monitored in real time. When the total momentum of the system exceeds the soft threshold alarm, the scanning acceleration and scanning speed are automatically reduced. When any of the following hard fault conditions are triggered: total momentum exceeding the limit, communication link loss, metering signal failure, or drive current exceeding the limit, the controller outputs an interpolation smooth transition command to control all suspended workpiece stages to perform smooth deceleration and smooth docking, avoiding the risk of collision.

Claims

1. A multi-stage reverse motion anti-vibration decoupling photolithography stage system, characterized in that, The system includes a vacuum chamber, a fixed-section planar coil table, a metering truss, three or more magnetically levitated workpiece stages, a wireless communication subsystem, an onboard energy storage subsystem, a sidewall thermoelectric composite charging and conducting heat mechanism, and a thermal management subsystem. The vacuum chamber contains symmetrically arranged process stations and counter-processing stations, as well as loading and waiting stations. The fixed-section planar coil table remains stationary. Each magnetically levitated workpiece stage has four sets of one-dimensional Halbach permanent magnet arrays on its lower base plate. The fixed-section planar coil table and the Halbach... A permanent magnet array is used to drive the workpiece stage to levitate without contact. A metering truss is stationary above all the magnetically levitated workpiece stages. A single, seamless optical grating master scale is fixed to the lower surface of the metering truss. Each magnetically levitated workpiece stage has a set of follower optical grating readers at each of its four corners. These four sets of readers read the optical grating master scale upwards and jointly calculate the six degrees of freedom position and attitude of the workpiece stage. Each magnetically levitated workpiece stage has an onboard energy storage module consisting of a solid-state battery and a supercapacitor connected in parallel. In the levitated operation state, all following power cables, signal fibers, and water-cooled drag chains are eliminated, and the onboard energy storage module provides independent power. The wireless communication subsystem enables seamless communication between the workpiece stage and the system's main control unit. The cable supports bidirectional data transmission; the sidewall thermoelectric composite charging and conducting mechanism only works at the docking station, realizing intermittent power replenishment and periodic heat dissipation; the thermal management subsystem adopts a self-equilibrating temperature structure with a built-in sealed liquid storage in the workpiece stage, realizing autonomous temperature control without pipelines; the workpiece stage hardware does not have a fixed functional identity, but its role is determined by the station it is in; the three or more magnetically levitated workpiece stages are of the same specification and are precisely counterweighted, with the mass and inertia tensor of each stage being consistent with each other; the magnetically levitated workpiece stage scheduled to the counter-current station acts as a reverse-moving workpiece stage, performing synchronous reverse motion with the process scanning workpiece stage at the process station, canceling the inertial disturbance generated by the motion of the process scanning workpiece stage in a momentum cancellation manner.

2. The multi-stage reverse motion anti-vibration decoupling lithography stage system according to claim 1, characterized in that, The fixed-zone planar coil tabletop is composed of arrayed planar coil tiles spliced ​​together with seams. Each coil tile is a physically independent block, and the windings within adjacent coil tiles in the same direction are electrically connected in series at the seams. Several adjacent coil tiles are grouped to form a driving zone, and each driving zone is configured with an independent power driving channel. A Hall sensor group is embedded between the upper surface of the substrate of each coil tile and the lower edge of the winding. The Hall sensor group consists of no less than six Hall sensing elements arranged orthogonally and differentially, used to output the three orthogonal components of the magnetic field directly above the tile and its gradient components along the three coordinate directions. The Hall sensor group is used to determine the commutation phase of the driving zone online, update the thrust calibration matrix of the driving zone online, and solve the magnetic field crosstalk generated by the Halbach permanent magnet array of other magnetic levitation workpiece stages at this tile online and update the magnetic field crosstalk feedforward compensation matrix.

3. The multi-stage reverse motion anti-vibration decoupling lithography stage system according to claim 1, characterized in that, It also includes a hierarchical collaborative motion controller, which has a three-layer bandwidth isolation architecture consisting of an upper trajectory momentum coordination layer, a middle multi-input multi-output robust control layer, and a bottom current inner loop; the upper trajectory momentum coordination layer is used for workstation queuing scheduling, trajectory generation, and momentum monitoring; the middle layer is equipped with... The controller and disturbance observer are configured with a notch filter to avoid structural modes; the bottom layer is used for high-speed current closed loop; the hierarchical cooperative motion controller establishes a root sum of squares (RSS) error budget model, the input of which includes at least the uncertainty of the grating metrology subsystem, the tracking error of the magnetic levitation closed loop, the thermal deformation drift error, the momentum cancellation residual disturbance, the truss micro-vibration transmission disturbance, the magnetic field crosstalk error, the free space laser communication time delay jitter error, and the PZT micro-motion compensation residual. The dynamic synthetic positioning error of the workpiece stage is controlled as follows: .

4. The multi-stage reverse motion anti-vibration decoupling lithography stage system according to claim 1, characterized in that, The system is equipped with fault protection logic, which monitors the workpiece stage position and attitude, total system momentum, free space laser communication link status, coil current and metering signal validity in real time. When the total system momentum exceeds the soft threshold, the scanning speed and acceleration are automatically reduced. When a hard fault condition such as total momentum exceeding the limit, communication link failure, metering failure or current exceeding the limit is triggered, an interpolation smoothing command is output to make the workpiece stage decelerate and stop smoothly.

5. The multi-stage reverse motion anti-vibration decoupling lithography stage system according to claim 1, characterized in that, The thermal management subsystem autonomously stores and equalizes heat within the workpiece stage during the exposure phase, and dissipates heat via solid-state conduction through the sidewall thermoelectric composite heat conduction mechanism during the docking phase, ensuring that the heat balance of the workpiece stage reaches zero within a single operating cycle; the temperature gradient inside the workpiece stage is controlled as follows: .

6. The multi-stage reverse motion anti-vibration decoupling lithography stage system according to claim 1, characterized in that, The Halbach permanent magnet array consists of four independent one-dimensional Halbach arrays arranged at the four corners of the workpiece stage base plate. The magnetization period of the two diagonally opposite one-dimensional Halbach arrays is along the X direction and has no modulation in the Y direction. The magnetization period of the other two diagonally opposite one-dimensional Halbach arrays is along the Y direction and has no modulation in the X direction. The magnetic blocks in the four arrays are arranged according to the one-dimensional Halbach rule. The magnetization is alternating; the total mass deviation of the Halbach permanent magnet array of each magnetically levitated workpiece stage is less than 0.1 g, and the magnetic flux deviation is... The magnetic levitation workpiece platforms were weighed and precisely counterweighted one by one, and the mass deviation of each platform was less than 0.1 g.

7. The multi-stage reverse motion anti-vibration decoupling lithography stage system according to claim 1, characterized in that, The metering truss is mechanically decoupled from the vacuum chamber and the fixed partitioned planar coil tabletop via independent granite vibration isolation supports. The metering truss and the fixed partitioned planar coil tabletop are each equipped with an independent constant-temperature circuit, with a 5 mm thermal insulation air gap between them. High-reflectivity radiation shielding foil enables thermal decoupling between the metering truss and the coil tabletop.

8. A method for vibration-resistant decoupling control of reverse motion of multiple workpiece stages, characterized in that, The method is based on three or more magnetically levitated workpiece stages, which are arranged in a vacuum chamber equipped with process stations, punching stations, loading stations, and waiting stations. The hardware of the workpiece stages does not have a fixed functional identity; its role is determined by the station it is located at. The magnetically levitated workpiece stages are of the same specifications and are precisely counterweighted, with each stage having the same mass and inertia tensor. The method includes the following steps: S01. Pre-processing and docking steps: The magnetic levitation workpiece stage to be operated is dispatched to the loading station. The TiN thermoelectric composite contacts on the side wall are vertically pressed against the cold plate contacts of the frame by the lateral force of magnetic levitation, and the onboard energy storage module is powered. At the same time, the heat stored in the workpiece stage is discharged periodically. After the wafer loading, pre-alignment and cavity baking pre-processing are completed, the lateral pressing force is removed, the contacts are separated, and the workpiece stage enters the cableless levitation operation state. It is powered autonomously by the onboard energy storage and completes two-way data interaction through the FSOC wireless optical communication subsystem. S02, Momentum cancellation station automatic allocation step: The two workpiece stages are respectively allocated to the spatially symmetrical process station and the counter-clamping station; the mass and inertia tensor of the magnetically levitated workpiece stages are consistent with each other, the workpiece stage allocated to the counter-clamping station automatically becomes the reverse motion workpiece stage, and the workpiece stage in the process station becomes the process scanning workpiece stage. S03, Reverse Momentum Cancellation Scanning Step: The process scanning workpiece stage is controlled to complete high-speed scanning exposure according to the Jerk limiting S-curve trajectory. Simultaneously, the reverse motion workpiece stage is controlled to read the global 6DOF position and velocity data output by the metering gantry and execute synchronous reverse motion. The inertial force and overturning moment generated by the process scanning are canceled by the conservation of momentum. The workpiece stage has no drag chain or external contact throughout the scanning process. The perfluoropolyether liquid inside the workpiece stage is self-circulating and temperature equalized. The layer controller completes trajectory adjustment, disturbance suppression and error compensation in real time. S04, Workstation rotation queuing steps: After a single round of scanning exposure is completed, the workpiece table pairing relationship is released, the controller dynamically schedules workstation rotation, the workpiece table that has completed the operation returns to the loading station or waits for the station to complete the pre-processing, and the workpiece table newly scheduled to the process station and the punching station is connected to carry out the next round of operation; multiple workpiece tables are queued and flowed in parallel to realize the synchronous progress of exposure and pre-processing processes. S05. Status monitoring and fault handling steps: Monitor the total momentum of the system, the optical power and bit error rate of the FSOC link, the workpiece stage posture, the coil current and the metering signal status in real time throughout the process; when the total momentum of the system exceeds the soft threshold, automatically reduce the scanning speed and acceleration; when a hard fault condition is triggered, output the interpolation smoothing control command to control the workpiece stage to decelerate and stop smoothly.

9. The multi-workpiece stage reverse motion anti-vibration decoupling control method according to claim 8, characterized in that, In step S01, the clamping force line of the magnetic levitation lateral force passes through the center plane of the workpiece stage, driving the contact points on both sides of the workpiece stage to synchronously and vertically fit with the cold plate contact of the frame; in step S04, the system simultaneously constructs multiple sets of independent process stations and counter-clamping stations, and each set of pairs performs reverse momentum cancellation scanning in parallel.

10. The multi-workpiece stage reverse motion anti-vibration decoupling control method according to claim 8, characterized in that, In step S03, each magnetically levitated workpiece stage is weighed and precisely counterweighted, with the mass deviation between any two stages being less than 0.1 g. In step S05, the hard fault conditions include total momentum exceeding the limit, communication link failure, metering failure, and current exceeding the limit. The interpolation smoothing control command causes all levitated workpiece stages to synchronously and smoothly decelerate and stop.

Images