Micro-control system for intelligent feedback alignment and adjustment of MLA lens of photoetching machine

Abstract

The present application relates to the technical field of optical regulation of photoetching machine, discloses a MLA lens intelligent feedback alignment adjustment micro-control system of photoetching machine.The system comprises a light intensity distribution acquisition module, a lens state mapping module, an offset optimization module, an exposure compensation decision module and a dynamic calibration feedback module.The light intensity distribution acquisition module acquires light intensity distribution data in the exposure field in real time, analyzes the spatial characteristics and time change sequence to generate extraction results;the lens state mapping module associates the physical coordinates of the lens unit, calculates the optical parameter offset to generate a mapping set;the offset optimization module analyzes the offset cumulative distribution and fluctuation range to generate a multi-dimensional optimization parameter set;the exposure compensation decision module dynamically allocates compensation priorities, calculates compensation coefficients and fine adjustment vectors to generate an execution instruction set;the dynamic calibration feedback module executes the instructions and monitors the feedback calibration deviation.The system realizes accurate adjustment of lens alignment and guarantees exposure quality.

Description

Technical Field

[0001] This invention relates to the field of optical control technology for lithography machines, specifically to a micro-control system for intelligent feedback alignment adjustment of MLA lenses in lithography machines. Background Technology

[0002] In the field of integrated circuit manufacturing, lithography machines are core equipment, and their exposure accuracy directly affects the performance and yield of chips. As chip manufacturing processes continue to break through to the nanometer scale, increasingly stringent requirements are placed on the alignment accuracy and stability of optical components in lithography systems. MLA (microlens array), as a key component of the lithography machine's exposure optical system, is composed of a large number of microlens units, and its positional accuracy and the consistency of its optical parameters are crucial to exposure quality.

[0003] Traditional MLA lens adjustment methods rely heavily on static adjustments using preset calibration parameters, lacking the ability to respond to real-time changes during exposure. In actual exposure, factors such as changes in ambient temperature, mechanical vibration, and fluctuations in light source stability can cause slight shifts in the position and optical parameters of the MLA lens unit. If these shifts are not detected and compensated in time, they will gradually accumulate and lead to uneven light intensity distribution within the exposure field, thereby affecting the transfer accuracy of the chip pattern.

[0004] Existing adjustment systems often limit their analysis of light intensity distribution data to spatial characteristics at a single time point, making it difficult to capture the dynamic changes in offset over time, resulting in lag in calibration decisions. Furthermore, in the coordinated adjustment of multiple lens units, the lack of correlation analysis and priority allocation mechanisms for the offsets of each unit easily leads to wasted compensation resources or insufficient adjustment of key units, further restricting the exposure accuracy and stability of the lithography machine. Summary of the Invention

[0005] The purpose of this invention is to provide a smart feedback alignment adjustment microcontroller system for MLA lenses in lithography machines, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides a micro-control system for intelligent feedback alignment adjustment of MLA lenses in lithography machines, the system comprising:

[0007] The light intensity distribution acquisition module is used to acquire light intensity distribution data in the exposure field in real time, analyze the spatial characteristics and temporal change sequence of the light intensity distribution data, and generate light intensity distribution feature extraction results.

[0008] The lens state mapping module is used to receive the light intensity distribution feature extraction results, associate the physical position coordinates of the MLA lens unit, calculate the optical parameter offset of each lens unit, and generate a lens offset feature mapping set.

[0009] The offset optimization module is used to match the preset optical calibration reference value based on the lens offset feature mapping set, analyze the cumulative distribution and fluctuation range of the offset, and generate a multi-dimensional offset optimization parameter set.

[0010] The exposure compensation decision module is used to dynamically allocate the compensation priority of the lens unit according to the multi-dimensional offset optimization parameter set, calculate the exposure energy compensation coefficient and the position fine-tuning vector, and generate the lens compensation execution instruction set.

[0011] The dynamic calibration feedback module is used to execute the lens compensation execution instruction set, monitor the changes in light intensity distribution after compensation in real time, compare the compensation target value with the actual calibration result, and generate a calibration deviation feedback report.

[0012] Preferably, the light intensity distribution feature extraction result includes light intensity spatial gradient value, time decay coefficient and abnormal fluctuation region identifier;

[0013] The lens offset feature mapping set includes the refractive error, axial offset angle, and speckle distribution anomaly of each lens unit;

[0014] The multi-dimensional offset optimization parameter set includes compensation weight coefficients, tolerable offset thresholds, and compensation response time windows;

[0015] The lens compensation execution instruction set includes the energy compensation amplitude, the lens array displacement vector, and the compensation timing schedule table;

[0016] The calibration deviation feedback report includes the residual offset distribution, compensation efficiency index, and secondary calibration requirement identifier.

[0017] Preferably, the operation of the light intensity distribution acquisition module includes:

[0018] A high-resolution light intensity sensor array is used to collect continuous time-series light intensity sampling point data within the exposure field, and the spatial coordinates and time-series labels of the sampling points are marked.

[0019] The light intensity sampling point data is divided into gridded analysis units. The mean and standard deviation of light intensity in each unit are calculated to identify abnormal light intensity regions that exceed the preset fluctuation range.

[0020] The light intensity gradient changes of adjacent analysis units are correlated to construct a light intensity attenuation trend curve, and the light intensity distribution feature extraction result is generated and transmitted to the lens state mapping module.

[0021] Preferably, the operation of the lens state mapping module includes:

[0022] Receive the light intensity gradient value and abnormal fluctuation region identifier from the light intensity distribution feature extraction result, and match the physical coordinates of the corresponding lens unit in the MLA lens array;

[0023] Calculate the light focusing deviation angle and diopter change of each lens unit, and calibrate the lens optical performance attenuation index by combining speckle distribution data;

[0024] The lens offset feature map set is generated by integrating the light focusing deviation angle, diopter change, and optical performance attenuation index, and then input into the offset optimization module.

[0025] Preferably, the operation of the offset optimization module includes:

[0026] The refractive error and axial offset angle are extracted from the lens offset feature map set and compared with the preset optical calibration reference value range;

[0027] Analyze the spatial distribution density of offsets exceeding calibration reference values ​​in the lens array and calculate compensation priority weights;

[0028] Based on the compensation priority weight and offset fluctuation range, the multi-dimensional offset optimization parameter set is generated and sent to the exposure compensation decision module.

[0029] Preferably, the operation of the exposure compensation decision module includes:

[0030] Based on the compensation weight coefficients and compensation response time window in the multi-dimensional offset optimization parameter set, the energy compensation sequence of the lens unit is planned.

[0031] Calculate the required exposure energy gain and micro-displacement vector for each lens unit, and generate a compensation action sequence by combining the lithography machine exposure timing.

[0032] The compensation action sequence is encapsulated into the lens compensation execution instruction set and sent to the dynamic calibration feedback module.

[0033] Preferably, the operation of the dynamic calibration feedback module includes:

[0034] The energy compensation amplitude and lens array displacement vector in the lens compensation execution instruction set are executed, and the compensated light intensity distribution data are collected in real time.

[0035] By comparing the compensation target value with the actual light intensity distribution characteristics, the distribution ratio of the residual offset in the lens array is calculated.

[0036] When the proportion of residual offset distribution exceeds the tolerable offset threshold, a calibration deviation feedback report containing a secondary calibration requirement identifier is generated.

[0037] Preferably, the system further includes a calibration performance evaluation module:

[0038] Receive the residual offset distribution and compensation efficiency index from the calibration deviation feedback report, and associate them with historical calibration record data;

[0039] The residual offset decay rate after multiple calibration operations is statistically analyzed to generate a lens calibration performance degradation curve.

[0040] When the slope of the lens calibration performance degradation curve exceeds the preset degradation threshold, a calibration strategy correction command is output to the offset optimization module.

[0041] Preferably, the offset optimization module further includes the following operations:

[0042] Receive the calibration strategy correction instruction and adjust the tolerable offset threshold and the compensation response time window;

[0043] Recalculate the compensation priority weights of the lens units to generate an updated set of multi-dimensional offset optimization parameters;

[0044] The updated multi-dimensional offset optimization parameter set is fed back to the exposure compensation decision module.

[0045] Preferably, the system further includes an exposure field zoning compensation module for performing:

[0046] The optical parameter offset of the lens offset feature mapping set is received, and the compensation region grid is divided in combination with the exposure field topology.

[0047] Calculate the dispersion and spatial correlation of lens offset within the grid of each compensation region to generate regional compensation synergy coefficients;

[0048] The compensation action sequence of adjacent grids is dynamically adjusted according to the regional compensation coordination coefficient, generating a partitioned collaborative compensation instruction and merging it into the lens compensation execution instruction set.

[0049] Compared with the prior art, the beneficial effects of the present invention are:

[0050] By acquiring real-time light intensity distribution data within the exposure field through a light intensity distribution acquisition module and analyzing its spatial characteristics and temporal variation sequence, the system can comprehensively capture the dynamic offset patterns of the MLA lens unit during the exposure process, providing accurate feature basis for subsequent calibration. This synchronous analysis of spatiotemporal characteristics breaks through the limitations of traditional static detection, enabling the system to promptly perceive the cumulative trend of minute offsets.

[0051] The lens state mapping module correlates light intensity distribution characteristics with the physical position coordinates of MLA lens units, calculates the optical parameter offsets of each unit, and generates a mapping set, achieving a precise mapping from changes in light intensity distribution to the state of a specific lens unit. This correlation mechanism allows the offset of each lens unit to be clearly located, avoiding the problem of generalizing the overall offset and ignoring individual differences in traditional adjustment, thus laying the foundation for targeted compensation.

[0052] The offset optimization module matches preset optical calibration benchmarks to a mapping set, analyzes the cumulative distribution and fluctuation range of offsets, and generates a multi-dimensional set of optimization parameters. Through in-depth analysis of the distribution characteristics and fluctuation patterns of offsets, it can identify key areas and potential risk points within the offset concentration, ensuring that the optimization parameters consider not only the offset of individual units but also the overall offset distribution, thus improving the scientific rigor and comprehensiveness of the compensation strategy.

[0053] The exposure compensation decision module dynamically allocates the compensation priority of lens units based on a multi-dimensional optimized parameter set, calculates the exposure energy compensation coefficient and position fine-tuning vector, and generates an execution instruction set. This priority allocation mechanism ensures that, with limited compensation resources, lens units that have a greater impact on exposure quality are processed first, avoiding resource misallocation. At the same time, the synergistic effect of energy compensation and position fine-tuning enables multi-dimensional and precise adjustment of lens units, correcting positional offsets and compensating for changes in optical parameters, thus improving the completeness of the compensation effect.

[0054] After executing the compensation command, the dynamic calibration feedback module monitors the changes in light intensity distribution after compensation in real time, compares the target value with the actual result, and generates a deviation feedback report. This closed-loop feedback mechanism enables the system to verify the compensation effect in a timely manner. If a deviation exists, a new round of adjustment can be quickly initiated, forming a continuously optimized dynamic adjustment cycle to ensure that the state of the MLA lens unit is always kept within the ideal range, thus maintaining the stability of exposure quality. Attached Figure Description

[0055] Figure 1 This is a schematic diagram illustrating the working principle of the intelligent feedback alignment adjustment microcontroller system for the MLA lens of the lithography machine described in this invention.

[0056] Figure 2 A flowchart illustrating the operation of the light intensity distribution acquisition module;

[0057] Figure 3 A flowchart illustrating the operation of the lens state mapping module;

[0058] Figure 4 A flowchart for the operation of the exposure compensation decision module;

[0059] Figure 5 This is a flowchart of the calibration performance evaluation module operation. Detailed Implementation

[0060] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0061] Please see Figure 1 The present invention provides a micro-control system for intelligent feedback alignment adjustment of MLA lens in lithography machine. The system includes: light intensity distribution acquisition module, lens state mapping module, offset optimization module, exposure compensation decision module and dynamic calibration feedback module.

[0062] The light intensity distribution acquisition module acquires real-time light intensity distribution data within the exposure field, analyzes the spatial characteristics and temporal variation sequence of this data, generates light intensity distribution feature extraction results, and transmits them to the lens state mapping module. The lens state mapping module receives the light intensity distribution feature extraction results, associates them with the physical position coordinates of the MLA lens units, calculates the optical parameter offsets of each lens unit, generates a lens offset feature mapping set, and sends it to the offset optimization module. Based on the lens offset feature mapping set, the offset optimization module matches a preset optical calibration benchmark value, analyzes the cumulative distribution and fluctuation range of the offset, generates a multi-dimensional offset optimization parameter set, and transmits it to the exposure compensation decision module. The exposure compensation decision module dynamically allocates the compensation priority of the lens units according to the multi-dimensional offset optimization parameter set, calculates the exposure energy compensation coefficient and position fine-tuning vector, generates a lens compensation execution instruction set, and outputs it to the dynamic calibration feedback module. The dynamic calibration feedback module executes the lens compensation execution instruction set, monitors the changes in light intensity distribution after compensation in real time, compares the compensation target value with the actual calibration result, generates a calibration deviation feedback report, and feeds it back to other parts of the system or the operating interface. All modules interact through a data bus, and a real-time embedded system is used to implement data transmission and control logic.

[0063] Example 1: See Figure 2 The operation of the light intensity distribution acquisition module begins with the deployment of a high-resolution light intensity sensor array, which covers the entire exposure field surface of the lithography machine in a matrix form. Each sensor unit has an independent photoelectric conversion function, capturing the instantaneous light intensity value within the exposure field at microsecond-level sampling intervals. Spatial coordinate positioning uses Cartesian coordinate system mapping, with the XY plane coordinate values ​​strictly aligned with the physical position of the lens array. Time-series labels record timestamps accurate to the nanosecond level, forming a continuous time-series dataset of light intensity sampling points. The data acquisition process continues within a single exposure cycle, with the sampling frequency consistently maintained at over 10,000 times per second, ensuring the capture of transient changes in light intensity.

[0064] The light intensity sampling data is meshed into an analysis unit grid covering the entire exposure field. Each analysis unit covers a fixed area, defined as a 2μm × 2μm square block. The grid division algorithm uses the principle of equal area segmentation, and interpolation technology is used for boundary processing to ensure no data omissions. The mean light intensity within a unit is calculated by arithmetic mean of all sampling points, and the standard deviation is calculated based on the principle of statistical variance analysis. The preset fluctuation range is set to ±10% of the historical mean of the unit according to the lithography process requirements. By comparing the current fluctuation value with the preset range in real time, units exceeding the range are automatically marked as light intensity anomaly areas. The logic for judging anomaly areas includes spatial continuity verification, and adjacent anomaly units are automatically merged into continuous area identifiers.

[0065] The calculation of light intensity gradients between adjacent analysis units involves spatial differentiation operations. The gradient value calculation employs the central difference algorithm, using the current unit as the core and extracting the average light intensity of the eight surrounding adjacent units as input parameters. Gradient direction calculation uses vector decomposition techniques, outputting horizontal and vertical gradient components, ultimately synthesizing spatial gradient vectors and orientation angle data. The analysis of attenuation characteristics in the time dimension relies on time series modeling. The light intensity attenuation trend curve fitting utilizes a sliding window regression technique, constructing an exponential attenuation model using sampling points within each time window. The time attenuation coefficient is extracted from the exponential term parameter in the model, recorded as a percentage attenuation per millisecond. Abnormal fluctuation area identification generates a binary mask matrix based on the spatial distribution of exceeding units, and the identification data is integrated into the feature extraction results.

[0066] In the data integration phase, the spatial gradient values ​​of light intensity, the time decay coefficient, and the identifiers of abnormal fluctuation areas are encapsulated into a multi-dimensional data structure. The spatial gradient values ​​are stored as a two-dimensional floating-point array, with the matrix size consistent with the grid cell layout. The time decay coefficient uses a time-series queue storage structure, recording data sequentially according to the exposure process. The abnormal area identifiers are mapped to a binary matrix, with each bit corresponding to a single cell state. The data transmission interface uses a high-speed serial bus protocol, and the light intensity distribution feature extraction results are pushed to the lens state mapping module in real time. The data packet verification mechanism ensures transmission integrity through cyclic redundancy check codes. The entire process is executed in parallel within the hardware acceleration unit, with data processing latency controlled within milliseconds.

[0067] The module incorporates a real-time monitoring mechanism to continuously diagnose the sensor array's operational status. An anomaly detection algorithm scans the sensor response curves, identifies drifting or failed units, and triggers an automatic calibration procedure. The sensor calibration process employs reference light source calibration technology, with standard light intensity values ​​covering the dynamic range within the operating range. Calibration data is stored in non-volatile memory, and calibration coefficients are loaded into the processing unit upon system startup. The data acquisition logic includes anti-interference design; environmental noise filtering utilizes digital signal processing technology, including a low-pass filter algorithm to remove high-frequency interference components.

[0068] The size parameters of the meshed analysis unit can be dynamically adjusted via a configuration interface, covering a range from 1 micrometer to 5 micrometers. This parameter adjustment mechanism adapts to the requirements of different lithography processes, and adjustment commands are issued by the upper-level control system. The data processing thread adopts a multi-stage pipeline architecture, with the acquisition thread, mesh calculation thread, and feature extraction thread executing asynchronously in parallel. The cache management unit allocates independent storage blocks for each thread, and data exchange achieves zero-latency transmission through dual-port memory. The hardware accelerator undertakes the core computing tasks, including a dedicated arithmetic logic unit to handle repetitive operations such as mean calculation and variance calculation.

[0069] The differential operation for spatial gradient calculation is implemented using a programmable logic device, and the processing step size can be configured to single-order or second-order difference mode. The gradient vector output format includes both polar and rectangular coordinates, supporting on-demand parsing by subsequent modules. The time decay model parameter fitting adopts the iterative least squares method, and the initial value estimation is based on historical data trend prediction. An absolute error threshold is set for the convergence condition, and the maximum number of iterations in the fitting process is limited to within 100. Morphological processing operations are added to the anomaly region detection process, using dilatational erosion algorithms to smooth region edges and filter out misjudgments caused by isolated noise points.

[0070] Example 2: See Figure 3 The lens state mapping module receives the light intensity distribution feature extraction results from the light intensity distribution acquisition module. These results include the light intensity spatial gradient value, time decay coefficient, and abnormal fluctuation region identifiers. Internally, the module first parses the input data packet to extract the light intensity gradient value and abnormal fluctuation region identifiers. The light intensity gradient value is stored in the form of a two-dimensional matrix, with matrix elements... Indicates the first Line number The light intensity gradient value of the gridded analysis cell is included. This parameter contains a horizontal gradient component (X-axis) and a vertical gradient component (Y-axis), which together reflect the spatial amplitude and direction of the light intensity variation in that cell. and These are the row and column indices of the grid, with a value range consistent with the total number of grid cells in the lithography machine's exposure field. Abnormal fluctuation regions are identified using a binary matrix, with matrix elements... A value of 1 indicates the corresponding coordinates. There is an anomaly in light intensity at this location. and This is the coordinate index of the abnormal region.

[0071] The module aligns the light intensity feature data with the physical location of the MLA lens array using a coordinate mapping algorithm. The coordinate mapping process is based on a pre-stored lens position database, in which the coordinates of the center point of each lens unit are recorded. ,in This serves as a unique identifier for the lens unit. The mapping algorithm calculates the coverage relationship between the intensity grid units and the lens units, and uses the nearest neighbor matching principle to determine each intensity gradient value. Corresponding lens unit After matching is complete, the module assigns a set of light intensity gradient data to each lens unit, forming a lens-gradient association dataset.

[0072] Calculating the optical parameter offsets of each lens unit involves the derivation of several physical quantities. The light focusing deviation angle is one such quantity. The relationship between light intensity gradient and lens optical properties was calculated using the following model:

[0073] ,

[0074] in: Indicates lens unit The corresponding light intensity gradient vector, This is the nominal focal length of the lens. In this formula... Lens unit The angle of light focusing deviation, expressed in radians; This is the arctangent function, used to calculate the angular offset caused by the gradient; It represents the magnitude of the gradient vector, reflecting the degree of drastic change in light intensity.

[0075] Change in diopter The derivation is based on a correlation model between the light intensity attenuation coefficient and the lens material properties. The module extracts the time attenuation coefficient from the input data. This coefficient represents the light intensity in the lens unit. The rate of time decay within the covered area. Combined with the refractive index of the lens material. With thickness The change in refractive power is calculated using the material optical response formula, and the output is in refractive powers. Speckle distribution anomaly. The analysis relies on abnormal fluctuation area identification data, and the module statistical lens unit Anomaly indicators within the coverage area The distribution density is combined with historical speckle pattern characteristics to calculate the current speckle anomaly score.

[0076] Optical performance degradation index The module integrates and calculates the above three parameters. , and Weighting coefficients are assigned separately, and the mixture is generated by linear weighted summation. The weighting coefficients are dynamically adjusted based on the lens's position in the array, with edge lens units assigned higher sensitivity weights. After calculation, the module sorts the offset parameters of all lens units by their physical location coordinates, generating a lens offset feature map set. This dataset is stored in a structured table format, with each row recording the identifier of a lens unit. ,coordinate Light focusing deviation angle Change in refractive power anomaly in speckle distribution and optical performance degradation index .

[0077] The offset optimization module receives the lens offset feature map set and begins performing calibration reference value comparison. The module loads preset optical calibration reference values ​​from non-volatile memory, including refractive error thresholds. and axial offset angle threshold For each lens unit Module comparison and The absolute value, and and The absolute value of the value is used to mark units exceeding the threshold as objects to be calibrated. The comparison results generate a binary state vector. ,in Indicates lens unit Calibration is required.

[0078] Spatial distribution density analysis is for all The module divides the exposure field into several analysis regions and calculates the proportion of lens units to be calibrated in each region. Density calculation uses a kernel density estimation algorithm, taking the lens unit locations as the input point set and outputting a continuous density distribution function. Compensation priority weights are also considered. The calculation combines density distribution and offset magnitude, assigning higher weights to severely offset lens units in high-density regions. After weight calculation, the module integrates all parameters to generate a multi-dimensional offset optimization parameter set. This parameter set includes lens unit identifiers. Compensation priority weight Tolerable offset threshold and compensation response time window ,in Dynamically adjust according to lens type It was calculated based on the exposure timing constraints of the lithography machine.

[0079] The module operates a real-time diagnostic mechanism to monitor data consistency during the calculation process. An anomaly detection algorithm scans the statistical distribution of optical parameter offsets to identify outliers exceeding reasonable ranges. When an anomaly is detected, the module automatically triggers a data re-acquisition process, obtaining the latest input data from the light intensity distribution acquisition module. A historical data recording function continuously saves the offset feature mapping set from each calculation for long-term performance trend analysis. Data storage employs a time-series database structure, supporting rapid retrieval of historical records by time range.

[0080] Example 3: See Figure 4 The exposure compensation decision module receives a multi-dimensional offset optimization parameter set from the offset optimization module. This dataset includes information such as the compensation weight coefficients of the lens units, the tolerable offset threshold, and the compensation response time window. Internally, the module first parses the input data and establishes a parameter lookup table indexed by the lens unit identifier. The compensation weight coefficients reflect the calibration priority of each lens unit, with values ​​ranging from 0 to 1, and are calculated by the offset optimization module based on the lens offset feature mapping set. The tolerable offset threshold defines the maximum allowable residual offset for each lens unit; this threshold parameter is directly related to the lithography process requirements. The compensation response time window specifies that the compensation action must be completed within a specific time period; the start and end times of this window are synchronized with the lithography machine's exposure sequence.

[0081] The module plans the energy compensation order of lens units based on compensation weight coefficients. The sorting algorithm comprehensively considers the relationship between weight values ​​and the spatial position of the lenses, prioritizing compensation for lens units with higher weights and grouping spatially adjacent units together as much as possible. After the compensation order list is generated, the module calculates the required exposure energy gain value for each lens unit. The energy compensation amount is calculated using an optical transfer function model, with the input being the refractive error and axial offset angle data from the lens offset feature map set. Exposure energy gain value. The calculation formula is as follows:

[0082] ,

[0083] in: Indicates lens unit The amount of refractive error deviation, This is the reference value for refractive power; Lens unit axial offset angle, This is the angle reference value. The exposure energy gain value in the above formula... This is a dimensionless scaling factor, whose value characterizes the intensity of exposure energy compensation for lens unit k relative to the reference reference. The larger the value, the higher the exposure energy that the lens unit needs to compensate for; The calculation results are directly used as the core reference for the exposure energy compensation range, and are converted by the exposure compensation decision module into specific light source power adjustment parameters to adapt to the energy output requirements of the lithography machine exposure sequence.

[0084] The calculation of the position fine-tuning vector involves the mechanical adjustment parameters of the lens array. Based on the spatial coordinate data in the lens offset feature map set, the module derives the required displacement direction and distance for each lens unit. The vector calculation considers the mutual influence of adjacent lens units to avoid local adjustments leading to deterioration of the optical performance of surrounding areas. The fine-tuning vector output is a displacement in a three-dimensional coordinate system, including translational components along the X and Y axes and a focusing adjustment along the Z axis.

[0085] The integration process of the lithography machine's exposure timing incorporates compensation actions into the equipment's work cycle. The module analyzes the lithography machine's control signal timing diagram and inserts compensation instructions at predetermined blank intervals. Constraints on the compensation response time window ensure that all actions are completed within the exposure gap, without affecting the normal exposure process. An action sequence arrangement algorithm optimizes the instruction order, reducing the idle travel time of the actuators. For spatially adjacent lens units, their compensation actions are merged into consecutive instruction batches for execution.

[0086] The lens compensation execution instruction set is encapsulated using a hierarchical data structure. The top-level instruction header contains global parameters, such as the total number of instructions, estimated execution time, and checksum. The middle-level structure is grouped by compensation region, recording the region identifier, the list of included lens units, and common parameters. The bottom-level entries describe the compensation details of each lens unit, including the unit identifier, energy gain value, displacement vector, and precise timestamp. Data is encoded in binary format, compressed, and transmitted to the dynamic calibration feedback module via a high-speed interface.

[0087] The instruction transmission process employs a redundancy check mechanism to ensure data integrity. The sending end calculates a cyclic redundancy check (CRC) code and appends it to the end of the data packet; the receiving end verifies the correctness of the check code. If a transmission error is detected, the receiving module requests a retransmission of the affected data block. The communication protocol defines an acknowledgment mechanism; if the sending end does not receive an acknowledgment within a specified time, it automatically retransmits. The transmission link has bandwidth monitoring capabilities, dynamically adjusting the data packet size to adapt to network load conditions.

[0088] Upon receiving the lens compensation execution command set, the dynamic calibration feedback module immediately initiates the compensation action execution process. The energy compensation module adjusts the power output of the exposure light source, adjusting the irradiation intensity of specific lens units according to the gain value specified in the command. The mechanical fine-tuning mechanism drives the positioning device of the lens array to achieve sub-micron level position correction. The execution process strictly follows the time sequence requirements in the command, with a high-precision timer controlling the triggering time of each action.

[0089] The compensation effect monitoring system acquires real-time data on the corrected light intensity distribution. A sensor array scans the exposure field at millisecond intervals, measuring the actual light intensity value in each region. The data acquisition module converts the raw signal into a digital quantity and transmits it to the analysis unit. An algorithm for extracting light intensity distribution characteristics calculates the difference between the current light field and the target distribution, quantifying the residual offset. The difference analysis results are statistically analyzed by lens unit region, calculating the percentage of calibration deviation for each unit.

[0090] The calibration deviation feedback report integrates monitoring data and preset standards. The report data structure includes fields such as lens unit identifier, coordinate position, target compensation value, actual achieved value, and deviation amount. A residual offset distribution map visually displays the calibration status of the entire exposure field, using color gradients to represent the degree of deviation in different areas. The compensation efficiency index calculates the degree of agreement between the actual compensation effect and the expected target; the closer the value is to 1, the higher the compensation accuracy. Secondary calibration requirement indicators are automatically set based on deviation thresholds; areas exceeding the standard are marked as requiring further processing.

[0091] The module's internal operational status monitoring mechanism tracks key parameters during the compensation process. Real-time monitoring indicators include the actual output power of energy compensation, the accuracy of mechanical displacement, and the execution time of each action. An anomaly detection algorithm compares the expected values ​​with the measured values, triggering an alarm signal when the difference exceeds the allowable range. Execution timeout monitoring ensures that the compensation action is completed within a specified time window, and timeout events are recorded in the system log for subsequent analysis.

[0092] The hardware architecture employs a distributed processing unit design, with the main controller coordinating multiple dedicated processors. The energy compensation calculation unit is responsible for the power adjustment algorithm of the exposure light source and is equipped with a high-precision digital-to-analog converter interface. The mechanical control processor manages the positioning system of the lens array, achieving nanometer-level motion control. The sensor data processing unit processes multiple light intensity signals in parallel and extracts feature parameters in real time. The communication bus adopts a star topology, establishing dedicated data channels between the master node and each submodule.

[0093] Fault-tolerant mechanisms handle abnormal situations during execution. The energy compensation over-limit protection circuit monitors the light source output power and immediately cuts off the drive signal when it exceeds the safety threshold. Mechanical motion limit sensors prevent the lens array displacement from exceeding its physical range, triggering an emergency stop mechanism. The data verification process verifies the rationality of each compensation command and filters parameter combinations that do not conform to physical constraints. A backup control path takes over critical functions and maintains basic compensation capabilities in the event of a main system failure.

[0094] Performance optimization techniques improve the module's response speed and processing efficiency. An instruction prefetching mechanism analyzes the timing patterns of the compensation sequence and preloads subsequent instructions into the cache. A parallel pipeline architecture allows energy compensation calculations, mechanical motion planning, and sensor data processing to be executed concurrently. The memory management unit employs a block allocation strategy, allocating dedicated storage areas for different types of data. A real-time task scheduling algorithm prioritizes processor resources for time-critical operations.

[0095] The configuration management interface supports dynamic adjustment of compensation parameters. External systems can update key parameters such as reference baseline values, deviation thresholds, and time windows. Once a parameter modification request is validated, it takes effect immediately in subsequent compensation calculations. The version control system records all parameter change history, supporting quick rollback to previous configurations. Parameter import / export functions facilitate the migration of settings between different processes, maintaining consistency in compensation strategies.

[0096] The real-time performance monitoring interface displays key metrics of the compensation process. The dashboard view visually presents the current compensation progress, residual offset distribution, and system load status. The detailed parameter panel lists the calibration data and execution status for each lens unit. The historical trend graph shows the compensation efficiency index changing over time. The alarm information area scrolls to display recently occurring anomalies, categorized by severity. The interface update frequency is synchronized with the lithography machine's work cycle, ensuring operators have access to the latest system status.

[0097] The modular software architecture adopts a layered design, isolating hardware-related code from business logic. The device driver layer encapsulates the underlying hardware operation interface, providing a unified set of control commands. The service layer implements the core compensation algorithm and data processing flow. The interface layer defines the communication protocol and data format for external systems. The configuration management layer maintains parameter settings and system status information. Well-defined APIs facilitate interaction between layers, reducing internal coupling within the modules.

[0098] The maintenance mode supports hardware calibration and performance testing. The light source power calibration program connects to a standard optical power meter to establish output characteristic curves. Position sensor calibration uses a laser interferometer to verify displacement measurement accuracy. The mechanical transmission system undergoes backlash testing to compensate for motion return errors. The signal acquisition channel connects to a precision signal generator to calibrate the range and linearity. Calibration results are stored in non-volatile memory for real-time compensation during daily operation.

[0099] Example 4: See Figure 5The calibration performance evaluation module receives a calibration deviation feedback report from the dynamic calibration feedback module. This report includes data such as residual offset distribution, compensation efficiency index, and secondary calibration requirement identifier. The module first parses the input data packet, extracts key fields, and stores them in an internal buffer. The residual offset distribution data is recorded in a structured format, with each entry corresponding to the post-calibration status of a lens unit, including the unit identifier, coordinate position, residual offset value, and region number. The compensation efficiency index is a scalar value reflecting the overall performance of a single calibration operation, defined between 0 and 1, with higher values ​​indicating more significant compensation effects. The secondary calibration requirement identifier is a Boolean variable, indicating whether an additional calibration process is needed.

[0100] The module accesses the historical calibration record database to retrieve the complete dataset of the 10 most recent calibration operations. Historical data is stored in a time-series structure, with each record containing a timestamp, calibration parameter set, execution result, and performance indicators. The data association process is based on a time window matching algorithm, comparing the current calibration result with similar historical operations. Comparison fields include the residual offset distribution ratio, the trend of the compensation efficiency index, and the frequency of secondary calibration triggers. The historical data retrieval range is configurable, defaulting to operation records within the last 24 hours, but the time span can be adjusted according to process requirements.

[0101] Statistical analysis methods are applied to the residual offset data from multiple calibration operations. The module calculates the average residual offset after each calibration and constructs a time-series curve. Attenuation rate analysis employs linear regression, with the slope of the fitted curve characterizing the long-term trend of the system's calibration performance. The lens calibration performance degradation curve generation process includes data smoothing, using a moving average algorithm to eliminate short-term fluctuations. The degradation curve output is a discrete point sequence, with each point corresponding to the statistical value of the residual offset after one calibration cycle. The curve slope is calculated using the difference method, taking the average of three consecutive cycles as the current slope estimate.

[0102] A preset degradation threshold is stored in the module configuration parameters, defining the critical rate at which calibration performance deteriorates. The threshold comparison logic monitors the absolute value of the curve slope in real time, and when a condition exceeding the threshold is detected, a calibration strategy correction instruction is generated. The instruction contains parameter adjustment suggestions, mainly involving the resetting of the tolerable offset threshold and the compensation response time window. The instruction generation process references historical optimization records, selecting the parameter combination that performed best in similar past scenarios as the benchmark. The correction instruction is transmitted to the offset optimization module via the data bus, and the transmission protocol includes fields such as instruction type, parameter list, and effective time.

[0103] After receiving the calibration strategy correction instruction, the offset optimization module initiates the parameter update process. The tolerable offset threshold is dynamically adjusted based on the instruction's suggested value, with the adjustment magnitude proportional to the degree of degradation. The compensation response time window is recalculated considering the current operating state of the lithography machine and combining it with exposure timing constraints to determine the maximum allowable adjustment range. The module internally maintains a parameter version control system, recording the numerical differences before and after each update, as well as the effective time. Before the parameters take effect, simulation verification is performed, using historical data to test the predicted compensation effect under the new parameters.

[0104] The recalculation of compensation priority weights is based on the updated parameter set. The weight allocation algorithm integrates optical parameter offsets from the lens offset feature map set, the new tolerable offset threshold, and regional density distribution data. The calculation process uses a weighted summation model, with the proportion of different factors dynamically configured according to process requirements. The updated multi-dimensional offset optimization parameter set includes lens unit identifiers, newly calculated weight values, adjusted tolerable thresholds, and optimized response time windows. This parameter set is fed back to the exposure compensation decision module via a high-speed interface to generate the next round of compensation instructions.

[0105] The system monitoring mechanism continuously tracks changes in calibration performance after parameter updates. The real-time data acquisition cycle has been shortened to 50% of the original interval to quickly capture feedback on the effects of parameter adjustments. Monitoring indicators include the real-time distribution of residual offsets, the efficiency of compensation actions, and the frequency of secondary calibration requests. A data visualization interface dynamically displays the relative position of the calibration performance degradation curve to the threshold line, allowing operators to intuitively understand the system status. An anomaly detection algorithm scans for sudden changes in key indicators after parameter updates; when abnormal fluctuations are detected, an automatic rollback mechanism is triggered to restore the system to the previous stable parameter configuration. Table 1 shows typical historical data analysis results output by the calibration performance evaluation module, including key indicators such as calibration cycle, mean residual offset, and slope changes.

[0106] Table 1. Typical historical data analysis results of the calibration performance evaluation module output.

[0107]

[0108] The module's hardware implementation is based on a high-performance embedded processor, equipped with a dedicated memory area for storing historical datasets. Real-time processing threads run on an independent kernel, ensuring that analysis tasks do not interfere with the system's main control flow. Data persistence uses high-speed solid-state storage media, supporting millisecond-level read and write access. The communication interface incorporates a dual-channel redundancy design, automatically switching to the backup link in case of primary channel failure. The module's software architecture adopts a microservice design, with functional units interacting asynchronously through message queues, enhancing the overall system's fault tolerance.

[0109] The dynamic configuration interface allows external systems to adjust analysis parameters, including the historical record retrieval range, degradation threshold definition, and monitoring frequency. The interface protocol defines standardized parameter read / write commands, supporting remote configuration updates. The diagnostic log recording module records key operational events; log entries include timestamps accurate to microseconds, operation type, and descriptions of the impact range. Log analysis tools can trace parameter change history, assisting in troubleshooting abnormal system behavior.

[0110] The fault tolerance mechanism handles data anomalies, including missing historical records, out-of-bounds values, and time series breaks. The data repair algorithm automatically fills in missing values ​​using linear interpolation of adjacent data. The input data validation process detects illegal characters or format errors and filters out non-compliant data packets. Timeout monitoring forcibly interrupts long-running analysis tasks to prevent system resource exhaustion. Resource isolation technology ensures the basic functional availability of modules under extreme loads, with core services receiving priority computing resources.

[0111] Example 5: The operation of the exposure field zoning compensation module begins by receiving optical parameter offset data from the lens offset feature map set. This dataset contains optical performance parameters for each lens unit, such as refractive error, axial offset angle, and speckle distribution anomaly. Internally, the module first parses the input data, establishes a spatial index structure based on the physical coordinates of the lens units, and converts the discrete lens unit data into a continuous two-dimensional parameter field. The parameter field construction uses an interpolation algorithm to handle missing data points, ensuring the complete parameter coverage of the entire exposure field.

[0112] The exposure field topology is partitioned based on the physical layout characteristics of the lithography machine. The module decomposes the entire exposure field into multiple compensation region grids, each grid covering a fixed number of lens units. The grid partitioning algorithm considers the arrangement pattern of the lens array and uses the hexagonal tiling principle to optimize boundary processing. The size of a single compensation region grid can be dynamically configured, with the default setting being a square block containing 50×50 lens units. The grid boundaries adopt an overlapping design, and there is an adjustable-width transition band between adjacent grids to avoid boundary effects during compensation.

[0113] The lens offset analysis within each compensation region grid includes discreteness calculation and spatial correlation assessment. Discreteness quantification reflects the magnitude of lens parameter variation within the grid, and the calculation process statistically analyzes the standard deviation of the offsets of all lens elements within the grid. Spatial correlation assessment uses the covariance matrix method to analyze the synchronicity of parameter changes in adjacent lens elements. The module generates two key indicators for each compensation region grid: offset dispersion score and spatial correlation coefficient. The offset dispersion score indicates the consistency of lens parameters within the region; a lower value indicates a more concentrated compensation need. The spatial correlation coefficient reflects the correlation of parameter changes between this region and surrounding regions; a higher value indicates a greater likelihood of needing coordinated compensation.

[0114] The generation of the regional compensation synergy coefficient integrates the calculation results of the two indicators mentioned above. The module internally defines the calculation rules for the synergy coefficient, fusing the offset dispersion score and spatial correlation coefficient according to a weighted ratio. The weight allocation is dynamically adjusted according to the process requirements of the exposure field; the weight of the edge region is biased towards spatial correlation, while the weight of the central region focuses more on parameter dispersion. The output range of the synergy coefficient is 0 to 1; a higher value indicates that the compensation action of that region needs stronger synchronization with the surrounding regions. After the coefficient calculation is completed, the module establishes a topological graph of the compensation region grid, where nodes represent compensation regions, and edge weights are determined by the synergy coefficient.

[0115] The process of dynamically adjusting the compensation action sequence is based on a coordination coefficient topology graph. A module operation graph analysis algorithm identifies clusters of regions with high coordination coefficients. For compensation regions within a cluster, the module generates uniform compensation parameters, including the same exposure energy adjustment amplitude and position fine-tuning direction. Independent regions with low coordination coefficients retain personalized compensation schemes. The timing arrangement of compensation actions considers the movement constraints of the physical actuators. The position fine-tuning of the MLA lens array is achieved by a micro-displacement platform driven by a high-precision robotic arm, which is the core execution component for adjusting the lens unit position. During the partitioned coordinated compensation process, the compensation actions for high coordination coefficient regions are continuously time-sequenced, allowing the robotic arm to perform fine-tuning actions directly according to the spatial topology sequence without returning to the initial position after completing compensation for adjacent regions. This reduces the robotic arm's ineffective movement distance, decreases idle movement time, and improves the execution efficiency of compensation actions. Actions in high coordination regions are arranged to be executed within consecutive time periods, further reducing the robotic arm's idle movement time.

[0116] The generation of zoned collaborative compensation instructions ultimately integrates all adjustment results. The instruction data structure includes fields such as compensation area identifier, collaborative coefficient value, energy compensation amount, displacement vector, and execution time window. The instruction set is merged into the lens compensation execution instruction set through a high-speed data interface, and data compression technology is used during transmission to reduce communication latency. When the actuator drive system parses the instructions, it prioritizes processing cluster instructions for areas with high collaborative coefficients to ensure the compensation synchronization of critical areas.

[0117] The module operates a real-time monitoring mechanism to track the execution effect of compensation commands. A feedback data acquisition system records the actual adjustment parameters for each compensation region and compares them with the target values. The difference analysis results are used to dynamically update the synergy coefficient calculation model, forming a closed-loop optimization. Historical synergy coefficient data is stored in a circular buffer, supporting rapid backtracking of the regional synergy patterns from the last five compensation operations. An anomaly detection algorithm scans for sudden changes in the synergy coefficients; when a significant deviation from the historical pattern is detected, a re-partitioning calculation process is triggered.

[0118] The hardware acceleration unit handles core computational tasks, including computationally intensive operations such as spatial correlation analysis, cooperability coefficient calculation, and graph topology processing. A dedicated processor features a multi-threaded architecture, processing data from different compensation regions in parallel. Memory management employs block caching technology, allocating independent storage space for data in each compensation region. The communication interface supports bidirectional data transmission, capable of receiving lens offset parameters and sending compensation commands to the execution system.

[0119] The configuration management interface allows external adjustment of partition compensation parameters, including grid size, coordination coefficient weights, and execution priorities. The parameter update mechanism employs version control, recording the change time, operator, and reason for each modification. The module status monitoring interface displays the grid division of the compensation area, the distribution of coordination coefficients, and the queue of pending instructions in real time. Visualization tools use different colors to indicate the coordination level of each area, facilitating intuitive understanding of the system status by operators.

[0120] Fault tolerance mechanisms handle exceptional situations during computation. Data validation algorithms check the rationality of input parameters and filter out outliers exceeding the physically possible range. Computation timeout monitoring forcibly interrupts long-running analysis tasks to prevent system resource exhaustion. Backup computing paths take over core algorithms in the event of a main processor failure, ensuring the continuous generation of compensation instructions. Resource allocation strategies prioritize the computational needs of highly collaborative regions, ensuring the timely execution of critical compensation actions.

[0121] The module's performance optimization employs incremental computation technology. For areas where the changes between two consecutive input data are minimal, most of the intermediate data from the previous calculation is reused. The change detection algorithm compares the differences between the current input and historical data, triggering a complete recalculation process only for areas with significant changes. This design significantly reduces processing latency, enabling the module to keep up with the real-time demands of high-speed exposure systems.

[0122] The logging system meticulously records the key steps of the partitioning compensation decision-making process, including grid partitioning parameters, the calculation process of the coordination coefficient, and the final instruction generation logic. Log entries contain microsecond-level timestamps and operational context information for post-event analysis and system debugging. Log files are stored cyclically, retaining operation records from the most recent 48 hours. Diagnostic tools can retrieve log data for specific time periods to reconstruct the compensation decision-making process.

[0123] The module collaborates with other parts of the system through standardized data interfaces. Input data formats have clearly defined field types and unit specifications to avoid parsing ambiguity. Output instruction sets use binary encoding to improve transmission efficiency. The interface protocol includes a data verification mechanism to detect any errors or data loss during transmission. The communication link has a redundant design, automatically switching to a backup path when the primary channel is interrupted.

[0124] Dynamic adjustment capabilities enable the module to adapt to different lithography process requirements. Process switching commands trigger the reset of internal parameters and load compensation strategies that match the new process. Gradual adjustments are used during the transition phase to avoid compensation oscillations caused by sudden parameter changes. A process feature recognition algorithm automatically analyzes the statistical characteristics of the input data and recommends suitable mesh partitioning schemes and collaborative calculation parameters.

[0125] The module's maintenance interface supports remote diagnostics and software updates. The firmware upgrade process employs a dual-storage design to ensure a rollback to a stable version even if the update is interrupted. Debug mode can freeze real-time data processing and inject test datasets to verify specific functionalities. Performance analysis tools statistically analyze the execution time distribution of each computational task to identify potential optimization points. Maintenance operation records are stored independently and managed separately from regular operation logs.

[0126] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0127] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A micro-control system for intelligent feedback alignment adjustment of MLA lenses in a lithography machine, characterized in that, include: The light intensity distribution acquisition module is used to acquire light intensity distribution data in the exposure field in real time, analyze the spatial characteristics and temporal change sequence of the light intensity distribution data, and generate light intensity distribution feature extraction results. The lens state mapping module is used to receive the light intensity distribution feature extraction results, associate the physical position coordinates of the MLA lens unit, calculate the optical parameter offset of each lens unit, and generate a lens offset feature mapping set. The offset optimization module is used to match the preset optical calibration reference value based on the lens offset feature mapping set, analyze the cumulative distribution and fluctuation range of the offset, and generate a multi-dimensional offset optimization parameter set. The exposure compensation decision module is used to dynamically allocate the compensation priority of the lens unit according to the multi-dimensional offset optimization parameter set, calculate the exposure energy compensation coefficient and the position fine-tuning vector, and generate the lens compensation execution instruction set. The dynamic calibration feedback module is used to execute the lens compensation execution instruction set, monitor the changes in light intensity distribution after compensation in real time, compare the compensation target value with the actual calibration result, and generate a calibration deviation feedback report.

2. The intelligent feedback alignment adjustment micro-control system for the MLA lens of a lithography machine according to claim 1, characterized in that: The light intensity distribution feature extraction results include light intensity spatial gradient values, time decay coefficients, and abnormal fluctuation region identifiers. The lens offset feature mapping set includes the refractive error, axial offset angle, and speckle distribution anomaly of each lens unit; The multi-dimensional offset optimization parameter set includes compensation weight coefficients, tolerable offset thresholds, and compensation response time windows; The lens compensation execution instruction set includes the energy compensation amplitude, the lens array displacement vector, and the compensation timing schedule table; The calibration deviation feedback report includes the residual offset distribution, compensation efficiency index, and secondary calibration requirement identifier.

3. The intelligent feedback alignment adjustment micro-control system for the MLA lens of a lithography machine according to claim 2, characterized in that, The operation of the light intensity distribution acquisition module includes: A high-resolution light intensity sensor array is used to collect continuous time-series light intensity sampling point data within the exposure field, and the spatial coordinates and time-series labels of the sampling points are marked. The light intensity sampling point data is divided into gridded analysis units. The mean and standard deviation of light intensity in each unit are calculated to identify abnormal light intensity regions that exceed the preset fluctuation range. The light intensity gradient changes of adjacent analysis units are correlated to construct a light intensity attenuation trend curve, and the light intensity distribution feature extraction result is generated and transmitted to the lens state mapping module.

4. The intelligent feedback alignment adjustment micro-control system for the MLA lens of a lithography machine according to claim 3, characterized in that, The operation of the lens state mapping module includes: Receive the light intensity gradient value and abnormal fluctuation region identifier from the light intensity distribution feature extraction result, and match the physical coordinates of the corresponding lens unit in the MLA lens array; Calculate the light focusing deviation angle and diopter change of each lens unit, and calibrate the lens optical performance attenuation index by combining speckle distribution data; The lens offset feature map set is generated by integrating the light focusing deviation angle, diopter change, and optical performance attenuation index, and then input into the offset optimization module.

5. The intelligent feedback alignment adjustment micro-control system for the MLA lens of a lithography machine according to claim 4, characterized in that, The operation of the offset optimization module includes: The refractive error and axial offset angle are extracted from the lens offset feature map set and compared with the preset optical calibration reference value range; Analyze the spatial distribution density of offsets exceeding calibration reference values ​​in the lens array and calculate compensation priority weights; Based on the compensation priority weight and offset fluctuation range, the multi-dimensional offset optimization parameter set is generated and sent to the exposure compensation decision module.

6. The intelligent feedback alignment adjustment micro-control system for the MLA lens of a lithography machine according to claim 5, characterized in that, The operation of the exposure compensation decision module includes: Based on the compensation weight coefficients and compensation response time window in the multi-dimensional offset optimization parameter set, the energy compensation sequence of the lens unit is planned. Calculate the required exposure energy gain and micro-displacement vector for each lens unit, and generate a compensation action sequence by combining the lithography machine exposure timing. The compensation action sequence is encapsulated into the lens compensation execution instruction set and sent to the dynamic calibration feedback module.

7. The intelligent feedback alignment adjustment micro-control system for the MLA lens of a lithography machine according to claim 6, characterized in that, The operation of the dynamic calibration feedback module includes: The energy compensation amplitude and lens array displacement vector in the lens compensation execution instruction set are executed, and the compensated light intensity distribution data are collected in real time. By comparing the compensation target value with the actual light intensity distribution characteristics, the distribution ratio of the residual offset in the lens array is calculated. When the proportion of residual offset distribution exceeds the tolerable offset threshold, a calibration deviation feedback report containing a secondary calibration requirement identifier is generated.

8. The intelligent feedback alignment adjustment micro-control system for the MLA lens of a lithography machine according to claim 7, characterized in that, It also includes a calibration performance evaluation module: Receive the residual offset distribution and compensation efficiency index from the calibration deviation feedback report, and associate them with historical calibration record data; The residual offset decay rate after multiple calibration operations is statistically analyzed to generate a lens calibration performance degradation curve. When the slope of the lens calibration performance degradation curve exceeds the preset degradation threshold, a calibration strategy correction command is output to the offset optimization module.

9. The intelligent feedback alignment adjustment micro-control system for the MLA lens of a lithography machine according to claim 8, characterized in that, The offset optimization module further includes the following operations: Receive the calibration strategy correction instruction and adjust the tolerable offset threshold and the compensation response time window; Recalculate the compensation priority weights of the lens units to generate an updated set of multi-dimensional offset optimization parameters; The updated multi-dimensional offset optimization parameter set is fed back to the exposure compensation decision module.

10. The intelligent feedback alignment adjustment micro-control system for the MLA lens of a lithography machine according to claim 9, characterized in that, It also includes an exposure field zoning compensation module, used to perform: The optical parameter offset of the lens offset feature mapping set is received, and the compensation region grid is divided in combination with the exposure field topology. Calculate the dispersion and spatial correlation of lens offset within the grid of each compensation region to generate regional compensation synergy coefficients; The compensation action sequence of adjacent grids is dynamically adjusted according to the regional compensation coordination coefficient, generating a partitioned collaborative compensation instruction and merging it into the lens compensation execution instruction set.

Images