Halogen injection decision method and system for excimer lasers

By collecting pulse energy and charging voltage, the rate of change of light power is calculated, the energy voltage slope and threshold voltage are identified, the total pressure component is removed, and the sources of halogen consumption and impurity accumulation are determined using the response vector. This solves the problems of halogen injection decision lag and ineffective injection, extends the whole-cavity ventilation interval, and improves halogen utilization efficiency.

CN123121231APending Publication Date: 2026-11-17HANGZHOU HUASHENGMAI TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611406085.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-10
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies cannot effectively distinguish and identify processes such as halogen consumption, impurity accumulation, electrode ablation, and decreased optical transmittance in excimer lasers, leading to a decrease in light output capability. Furthermore, halogen injection decisions are delayed or ineffective, making it impossible to accurately determine the timing of cavity ventilation.

Method used

By collecting pulse energy and charging voltage, the rate of change of the sliding estimate of light capability is calculated. Zero-mean charging voltage perturbation is applied to identify the energy voltage slope and threshold voltage. The total pressure component in the cavity is stripped off. The direction angle of the response vector is used to determine the source of attenuation. The timing of cavity ventilation is determined by normalized recovery rate and efficiency sequence.

Benefits of technology

It enables accurate identification of sources of halogen consumption and impurity accumulation without the need for additional diagnostic devices, avoiding ineffective injection, extending the ventilation interval of the entire chamber, improving halogen utilization efficiency, and reducing maintenance costs and downtime.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123121231A_ABST
    Figure CN123121231A_ABST
Patent Text Reader

Abstract

This invention discloses a halogen injection decision method and system for excimer lasers, belonging to the field of laser technology. The method obtains the rate of change of the sliding estimate of the light output capability from the pulse energy and charging voltage. When the rate of change reaches a threshold, a halogen injection is performed, and the increment of halogen partial voltage and the increment of total voltage are measured. Zero-mean charging voltage perturbations are applied before and after injection to identify the energy-voltage slope and threshold voltage. The total voltage component is extracted from the changes in these two components, yielding two improvement quantities constituting the response vector. The normalized recovery rate is obtained by weighting the two components and dividing by the relative halogen partial voltage increment. The dominant attenuation source is determined by the azimuth angle range, and the injection amount is adjusted accordingly, or partial emission and replacement are performed. The normalized recovery rates from multiple injections constitute a halogen utilization efficiency sequence, and the timing of cavity ventilation is determined by the cumulative deviation detection point. This method can separate reversible and irreversible attenuation without adding in-cavity sensors, avoiding ineffective injection and extending the cavity ventilation interval.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of excimer laser technology, and more specifically to a halogen injection decision method for excimer lasers and a decision system for implementing this method. Background Technology

[0002] Excimer lasers use a mixture of rare gases, halogen gases, and buffer gases as their working medium and are important light sources in photolithography and precision machining. During operation, halogens are continuously consumed, impurities such as fluorine compounds accumulate, electrodes are sputtered and ablated, and the ultraviolet transmittance of cavity windows and internal optical surfaces decreases. All of these factors contribute to a continuous increase in the charging voltage required to maintain the target energy under constant pulse energy operation, a phenomenon known as light extraction capability attenuation.

[0003] Existing technologies employ a three-stage gas management system—halogen micro-injection, partial gas emission replacement, and full-cavity ventilation—to delay halogen decay. Triggering factors are all open-loop quantities such as cumulative pulse count, runtime, and intracavity pressure. However, halogen consumption involves both discharge dissociation channels related to the cumulative pulse count and chemical reaction channels related to exposure time and surface condition. The weights of these two channels drift with operating conditions, and injecting a fixed amount at fixed intervals will result in alternating instances of insufficient and excessive injection.

[0004] Furthermore, existing technologies measure the deviation of charging voltage or pulse energy as a single scalar. Four processes—halogen consumption, impurity accumulation, electrode ablation, and decreased optical transmittance—are superimposed in the same direction on this scalar and indistinguishable. Based on this, decisions are made to continue applying halogen injection to address irreversible degradation. Such injection does not restore light emission capability; instead, it leads to halogen excess and accelerates the formation of fluorine-containing impurities. These impurities then consume halogen and absorb deep ultraviolet light, further reducing light emission capability. For irreversible components, existing technologies either rely on the prior assumption that their timescale is much longer than that of gas processes to rule them out, or they attempt post-hoc confirmation when the halogen partial pressure can no longer be reduced. The former fails at the end of the cavity's lifespan and with optical surface contamination, while the latter lags behind the actual occurrence of degradation. Neither approach can definitively determine whether the problem lies with the electrode or the optical surface. Replacing the cavity and replacing the window differs by an order of magnitude in terms of downtime and spare parts lead time.

[0005] To identify the source, it is usually necessary to add a sampling optical path, a discharge current probe, or an online gas composition detection device. However, the excimer laser cavity is a high-cleanliness sealed component subjected to deep ultraviolet radiation, and adding feedthroughs and intracavity devices can easily become new sources of contamination and failure points. In addition, halogen injection increases both the halogen partial pressure and the total pressure inside the cavity. The ratio obtained by directly dividing the difference in light output capability by the increase in halogen partial pressure contains a total pressure component unrelated to halogen. Using a moving average to find the inflection point flattens the inflection point and introduces phase lag, thus delaying the gas exchange of the entire cavity into the impurity self-acceleration region. Summary of the Invention

[0006] To address the above problems, this invention provides a halogen injection decision method and system for excimer lasers, which solves the problems of existing technologies where the sources of optical power attenuation are indistinguishable on a single scalar observation, ineffective injection is applied to irreversible attenuation, the identification of irreversible components is delayed and does not distinguish between discharge channels and optical channels, and the timing of whole-cavity gas exchange is passively triggered.

[0007] This invention proposes a halogen injection decision method for excimer lasers, comprising: The pulse energy and charging voltage are collected to obtain the rate of change of the sliding estimate of the light emission capability relative to the cumulative number of pulses. When the rate of change reaches the trigger threshold, a halogen injection is performed according to the injection amount, and the halogen partial pressure increment and the total pressure increment in the cavity are measured. Before the injection and at the end of the stabilization waiting period after the injection, a zero-mean charging voltage perturbation is applied, and the energy voltage slope and threshold voltage are identified by the pulse energy and charging voltage under the perturbation. The total pressure component is extracted from the changes in the energy voltage slope and the threshold voltage based on the total pressure increment in the cavity, resulting in the relative improvement in slope and the relative improvement in threshold voltage that constitute the response vector. The normalized recovery rate is obtained by weighting the two components of the response vector and dividing it by the ratio of the halogen partial pressure increment to the halogen partial pressure before injection. The dominant source of this attenuation is determined by the angle range in which the direction angle of the response vector is located. When the primary source is halogen consumption, the next injection amount is determined by the ratio of the preset change rate target value to the change rate and taking into account the normalized recovery rate; otherwise, halogen injection is suppressed and partial discharge and replacement are performed according to the upper limit of the total pressure in the cavity. The normalized recovery rate from multiple injections constitutes a halogen utilization efficiency sequence. The deviation between the extrapolated value and the measured value of the first segment is accumulated to detect the change point, and the timing of whole-cavity ventilation is determined.

[0008] The above scheme uses the existing pulse energy and charging voltage measurements as the only inputs, and obtains the response in two degrees of freedom before and after a halogen injection. This allows the four types of attenuation sources, which were originally superimposed in the same direction on a single scalar, to be separated according to the response direction. The injection decision is changed from open-loop timing and quantification to closed-loop tuning based on the measured response. Thus, without adding in-cavity diagnostic devices, the scheme achieves the purpose of avoiding invalid injection and extending the ventilation interval of the entire cavity.

[0009] Furthermore, the sliding estimate of the light emission capability is recursively calculated pulse by pulse using the following formula: in, and The first Pulse and the first Sliding estimate of light output capability after pulse generation; and The first The pulse energy and charging voltage of the generated pulse; The sliding coefficient; The rate of change is calculated using the following formula: in, For the first Rate of change; and The first Next and first The cumulative number of pulses; and Take the cumulative pulse count respectively and The sliding estimate of the light output capability at that time.

[0010] The light capability is characterized by the ratio of pulse energy to charging voltage and is recursively calculated pulse by pulse, allowing continuous tracking of attenuation without interrupting the production cycle. The triggering basis is the rate of change of the relative cumulative number of pulses rather than the absolute level, so that the triggering time is not affected by the target energy setting value and the model difference.

[0011] Furthermore, the charging voltage perturbation is a sequence of pulse-by-pulse sign taking, the sequence being in each... The sum of the signs within the sub-block formed by the emitted pulses is zero, and the charging voltage varies with the sequence according to the following formula: in, and The first The charging voltage of the pulse and the first The charging voltage is obtained by energy control through pulse generation. The relative amplitude of the perturbation; For the first The sign of the pulse, with a value of or ; The number of pulses contained in the sub-block; The energy voltage slope and the threshold voltage are covered by the perturbation. The pulse energy and charging voltage are obtained by weighted least squares fitting using the following formula, and the estimated standard deviations of both are also obtained: in, and These are pulse energy and charging voltage, respectively. The energy voltage slope; Threshold voltage; The number of pulses covered by the perturbation.

[0012] The perturbation sequence has zero mean and self-cancellation of signs within sub-blocks. The cumulative dose deviation is zero within any window not shorter than the sub-block length, thus achieving the required voltage span for identification without affecting dose stability. Weighted least squares simultaneously provide the estimated standard deviations of two quantities, providing a scale that can be directly compared with the fluctuation level for subsequent criteria.

[0013] Further, the total stripping pressure component is calculated using the following formula: in, and These represent the change in energy voltage slope and the change in threshold voltage after stripping the total voltage component, respectively. and These are the energy voltage slopes before and after halogen implantation, respectively. and These are the threshold voltages before and after halogen implantation, respectively. This represents the total pressure increment within the cavity; and These are the sensitivity coefficients of the energy voltage slope and the threshold voltage to the total pressure inside the cavity, respectively. The sensitivity coefficient is calibrated as follows: after the entire cavity is ventilated, a halogen-free buffer gas is injected into the cavity. A period of the charging voltage perturbation is applied before and after the buffer gas injection. The change in the energy voltage slope and the threshold voltage are respectively divided by the total pressure increment in the cavity caused by the buffer gas injection.

[0014] The halogen premixed gas increases the halogen partial pressure and the total pressure in the cavity at the same time. The change obtained after removing this component only reflects the improvement brought by the halogen itself. The discrimination threshold no longer drifts with the current total pressure level in the cavity. The sensitivity coefficient is calibrated by injecting pure buffer gas. The calibration process does not change the halogen partial pressure. All the measured changes can be attributed to the total pressure in the cavity.

[0015] Further, the relative improvement in slope and the relative improvement in threshold voltage are obtained by the following formula: in, and These represent the relative improvement in slope and the relative improvement in threshold voltage, respectively. The normalized recovery rate is obtained by the following formula: in, Normalized recovery rate; and These are the halogen partial pressure increment and the halogen partial pressure before injection, respectively; and These are the weights for the relative improvement in slope and the relative improvement in threshold voltage, respectively. and These are the standard deviations of the relative improvement in slope and the relative improvement in threshold voltage, respectively.

[0016] The two improvement measures are relative rather than absolute, so that the responses of different models and cavity ages can be compared on the same scale. The components with smaller standard deviations are given greater weight by weighting by the inverse of variance, and the signal-to-noise ratio of the composite is not lower than that of any of the components. Then, the recovery rates obtained under different injection volumes can be directly compared by normalizing with the relative halogen partial pressure increment, providing a unified benchmark for successive comparisons.

[0017] Furthermore, the composite magnitude of the response vector, the standard deviation of the composite magnitude, and the direction angle are obtained by the following formula: in, This is the synthesized amplitude; The standard deviation of the composite amplitude; It is the direction angle; The primary sources for determining this attenuation include: Not less than , Greater than zero and When the angle is in the first range, the dominant source is determined to be halogen consumption; Not less than , Greater than zero and When the angle is less than the lower bound of the first angle range, the dominant source is determined to be impurity accumulation; Less than or If the value is not greater than zero, it is determined that the halogen injection did not produce a valid response. The first angle range is obtained by expanding the median value of the directional angle obtained after several halogen injections following a complete cavity ventilation by three times the standard deviation of the directional angle to both sides.

[0018] The ratio of the synthetic amplitude to its standard deviation provides the criterion for whether the current response exceeds the fluctuation zone, while the directional angle undertakes the factorization function, thus decoupling the magnitude of the recovery quantity from its source attribution; the first angle interval is reconstructed according to the new cavity reference segment after each cavity air exchange, and can be automatically adapted to random type and cavity age without the need to preset discrimination parameters for different models.

[0019] Furthermore, when the dominant source is determined to be impurity accumulation, the weighted sum of the current data is included in the impurity attribution amount; when the current halogen implantation does not produce an effective response, the ratio of the monotonically increasing relative amount of the threshold voltage before implantation to the monotonically decreasing relative amount of the energy voltage slope before implantation is taken between several adjacent halogen implantations. If the ratio is greater than a set multiple, the dominant source is determined to be discharge channel degradation, and the monotonically increasing relative amount is included in the discharge channel attribution amount; if the ratio is less than the reciprocal of the set multiple, the dominant source is determined to be optical channel degradation, and the monotonically decreasing relative amount is included in the optical channel attribution amount. The discharge channel attribution and the optical channel attribution are respectively extrapolated to their respective tolerances using the following formulas to obtain the remaining number of pulses for cavity switching and the remaining number of pulses for window switching: in, and These are the remaining pulse counts for cavity switching and window switching, respectively. and These are the cumulative values ​​of the discharge channel attribution and the optical channel attribution, respectively. and These are the tolerances for the two, respectively; and These represent the growth rates of the relative cumulative pulse counts for each.

[0020] The irreversible components are further divided into two categories: discharge channels and optical channels. Each category is accumulated and extrapolated separately, and two independent predictions of the remaining pulse number are output. This allows two types of maintenance parts with a difference of one order of magnitude in downtime, spare parts cost, and spare parts lead time to be prepared and scheduled separately. Furthermore, maintenance actions that fall within the same time window can be merged into the same downtime for execution.

[0021] Furthermore, the setting of the next injection amount is performed according to the following formula: in, and The first Next and first The injection volume per time; The preset target value for the rate of change; For the first Rate of change; For the first Normalized recovery rate of the second time; The median of the normalized recovery rate obtained after several halogen injections following a complete cavity ventilation; The relative amplitude and the number of pulses covered by the next charging voltage perturbation are then adjusted according to the standard deviation of the normalized recovery rate. in, and The first Next and first The relative amplitude of the perturbation; and The first Next and first The number of pulses covered by the perturbation; For the first The standard deviation of the normalized recovery rate; This is the target value for the standard deviation of the normalized recovery rate.

[0022] The injection volume tuning formula includes a recovery rate ratio term. When the improvement amount obtained by unit halogen partial pressure decreases, the amount is increased proportionally to avoid chasing the target value of change rate with a fixed amount when the halogen utilization efficiency has already decreased. The perturbation and waiting period are only lengthened and increased when the standard deviation increases, so that the confidence interval of the direction angle is maintained within the width of the angle interval throughout the entire life cycle, while the energy fluctuation and production cycle occupancy are kept at a low level throughout the entire life cycle.

[0023] Furthermore, the halogen utilization efficiency sequence, the extrapolated value, the deviation, and the cumulative deviation are obtained by the following formulas: in, For the first Secondary halogen utilization efficiency; For the first The cumulative number of pulses; For the first Normalized recovery rate of the second time; The median of the normalized recovery rate obtained after several halogen injections following a complete cavity ventilation; To extrapolate the linear relationship fitted from the first segment of the sequence to the th... The extrapolated value obtained next; and These are the intercept and slope of the linear relationship, respectively; and The first The deviation amount and cumulative deviation value; The relaxation coefficient; The standard deviation of the deviation of the first segment of the sequence; Greater than The change point is determined to have occurred at that time, and is then... The cumulative number of pulses corresponding to the initial sequence of changes from zero to a continuously positive value is taken as the position of the change point. This is the cumulative threshold coefficient; The timing for determining the cavity ventilation is as follows: the change point has occurred, the proportion of the impurity attribution amount in the sum of the impurity attribution amount, the discharge channel attribution amount, and the optical channel attribution amount is not less than the proportion threshold, and the difference between the upper limit of the cavity total pressure and the current cavity total pressure is less than the cavity total pressure margin required for a preset number of partial discharges and replacements. When all three conditions are met, cavity ventilation is performed.

[0024] The cumulative structure first subtracts the relaxation amount and then takes the maximum of zero. The deviation caused by normal fluctuations is absorbed by the relaxation amount. The cumulative amount is repeatedly truncated to zero in the early stage without drifting, so there is no need to smooth the sequence and no smoothing phase lag is introduced. After entering the accelerated descent stage, the small deviations are superimposed in time, and the detection sensitivity is not limited by the single-point signal-to-noise ratio. After detection, the change point is located by tracing back along the cumulative amount and checked by two conditions: impurity ratio and total pressure margin. This ensures that the entire chamber gas exchange does not fall within the impurity self-acceleration zone, nor is it falsely triggered in cases where the electrode or optical surface is the main cause.

[0025] Another aspect of the present invention proposes a halogen injection decision system for an excimer laser, comprising: The observation module is used to collect pulse energy and charging voltage, obtain the rate of change of the sliding estimate of light output capability relative to the cumulative number of pulses, and initiate a halogen injection according to the injection amount when the rate of change reaches the trigger threshold, and measure the halogen partial pressure increment and the total pressure increment in the cavity. The identification module is used to apply a zero-mean charging voltage perturbation before the injection and at the end of the stabilization waiting period after the injection, and to identify the energy voltage slope and threshold voltage from the pulse energy and charging voltage under the perturbation; the attribution module is used to strip the total pressure component according to the total pressure increment in the cavity to obtain two relative improvement quantities that constitute the response vector, and to obtain the normalized recovery rate and the dominant source from the response vector. The execution module is used to perform injection volume tuning or injection suppression and to perform partial emission and replacement according to the dominant source; The ventilation decision module is used to construct a halogen utilization efficiency sequence from the normalized recovery rates of multiple injections and to determine the timing of whole-cavity ventilation.

[0026] The advantages of this invention compared to the prior art are: 1. This invention applies a zero-mean charging voltage perturbation before and after halogen implantation, identifies the energy voltage slope and threshold voltage based on the correspondence between pulse energy and charging voltage, and strips the total pressure component according to the increase in total intracavity pressure caused by the implantation, normalizing it with the relative halogen partial pressure increment. This expands the light emission capability attenuation, which was originally only observable on a single scalar, into a two-dimensional response. Halogen consumption causes the energy voltage slope to decrease and the threshold voltage to increase; both recover synchronously after implantation. After impurity accumulation implantation, only the slope partially increases while the threshold voltage does not decrease. Electrode ablation mainly raises the threshold voltage. Decreased optical surface transmittance mainly lowers the slope but hardly changes the threshold voltage. The response directions of the four types of processes are different from each other. The response direction of a single injection can determine the source. Therefore, it is not necessary to add a sampling optical path, discharge current probe or cavity gas composition detection device to complete the factorization. This avoids the pollution sources and failure points brought about by the addition of vacuum feedthrough and cavity devices. Furthermore, since the total pressure component has been stripped and the recovery rate has been normalized, the discrimination threshold does not drift with the cavity total pressure level. When the cavity total pressure is close to the upper limit, the recovery rate will not be underestimated due to the weakening of the marginal gain. When the cavity total pressure is low, the irreversible condition will not be judged as reversible.

[0027] 2. This invention, by dividing the injection volume into a single path based on the dominant source and adjusting or suppressing the injection, and converting it into partial discharge and replacement, blocks the ineffective injection of irreversible decay. The halogen in the cavity is no longer excessive, and the generation rate of fluorine-containing impurities decreases accordingly. The decline of the halogen utilization efficiency curve is slowed down, and the interval between two cavity gas replacements is thus extended. At the same time, the irreversible component is further divided into discharge channel attribution and optical channel attribution, which are accumulated and extrapolated to their respective tolerances. The remaining pulse count for cavity replacement and the remaining pulse count for window replacement are predicted independently. This allows two types of maintenance parts with a difference of one order of magnitude in downtime, spare parts cost, and spare parts lead time to be prepared and produced separately. Maintenance actions falling within the same time window can be merged into the same downtime for execution, thereby transforming the general conclusion of cavity replacement into a production-schedulable maintenance plan.

[0028] 3. This invention detects change points by performing a preliminary linear extrapolation and deviation accumulation on the halogen utilization efficiency sequence. It replaces the moving average with a cumulative structure that first reduces the relaxation amount and then rounds to zero, allowing deviations caused by normal fluctuations to be absorbed by the relaxation amount. The cumulative amount is repeatedly truncated to zero in the preliminary stage without drifting, thus avoiding the phase lag introduced by smoothing. Furthermore, it allows successive small deviations to accumulate over time, resulting in a significant increase in the cumulative amount before a single deviation exceeds the fluctuation level. Therefore, the detection sensitivity is not limited by the single-point signal-to-noise ratio. After detection, the change point can be located by tracing back along the cumulative amount, ensuring that the entire cavity ventilation no longer falls within the impurity self-acceleration zone. Simultaneously, the relative amplitude of the next perturbation, the number of pulses covered, the stabilization waiting period, and the hysteresis bandwidth of the trigger threshold are adjusted using the standard deviation of the recovery rate. This ensures that the confidence interval of the direction angle remains within the width of the angle interval throughout the entire lifespan, allowing factorization to be completed even at the end of the cavity's lifespan. The perturbation and waiting period only increase when the standard deviation rises, keeping energy fluctuations and production cycle occupancy at a low level throughout the entire lifespan. All the input quantities required by this invention are existing measurement quantities necessary for this type of laser to complete energy stability control and gas filling, and can be implemented on existing models as control software upgrades. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 This is a block diagram of a halogen injection decision system for an excimer laser according to an embodiment of the present invention; Figure 2 This is a flowchart of a halogen injection decision method for an excimer laser according to an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the direction angle interval discrimination of the micro-injection response vector according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the change point detection and cavity ventilation timing of the halogen utilization efficiency sequence in an embodiment of the present invention. Detailed Implementation

[0031] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, so as to provide a thorough understanding of the embodiments of the present invention. However, those skilled in the art will understand that the present invention may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail. The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0032] This invention provides a halogen injection decision method and system for excimer lasers, applicable to gas management in pulsed gas discharge lasers where halogen gas is one of the components. Typical applications include excimer lasers used in photolithography, annealing, and precision machining. The required inputs include pulse energy per pulse, charging voltage per pulse, total cavity pressure, and the injection volume measurement value from the injection valve. These are all existing measurements necessary for energy stability control and gas filling in this type of laser, eliminating the need for additional intracavity sensors, sampling optical paths, or online gas composition detection devices.

[0033] The laser emits light continuously or intermittently in a constant pulse energy mode. As halogen is consumed, impurities accumulate, electrodes are ablated, and optical transmittance decreases, the charging voltage required to maintain the target pulse energy gradually increases. This invention treats a single halogen injection as an actively applied excitation and the laser's response to this excitation as an observation. The four types of degradation processes affect the energy-voltage slope and threshold voltage in different ways. The direction of the response in the two-dimensional plane constitutes the criterion for the source, and the weighted sum of the responses constitutes a quantitative measure of recovery. The response of a single injection is used to determine whether the action is appropriate and how to adjust the injection volume for the next injection. The response sequence of multiple injections is used to determine the timing of cavity ventilation and the remaining available pulses for the two types of maintenance components.

[0034] Example 1 This embodiment provides a halogen injection decision system for an excimer laser, such as Figure 1 As shown, the system includes an observation module, an identification module, an attribution module, an execution module, and a ventilation decision module.

[0035] The observation module acquires pulse energy and charging voltage synchronously pulse by pulse, and also collects the total pressure inside the cavity and the injection volume measurement values ​​of the gas injection valve, marking each sample with a cumulative pulse count. A sliding estimate of the light emission capability is generated from the pulse-by-pulse data, and the rate of change of this estimate relative to the cumulative pulse count is calculated. This rate of change is compared with a trigger threshold to determine whether to initiate a halogen injection. The trigger threshold has a hysteresis bandwidth to prevent repeated injections when the rate of change fluctuates near the threshold. The rate of change obtained by the observation module is simultaneously sent to the execution module to participate in the injection volume tuning.

[0036] The identification module applies a first segment of charging voltage perturbation after the observation module issues a trigger but before halogen injection is executed, and applies a second segment of charging voltage perturbation at the end of the stabilization waiting period after halogen injection is executed. The perturbation superimposes a pulse-by-pulse sign sequence onto the target energy of the energy control loop, ensuring the charging voltage has a sufficient range within the controlled amplitude. The identification module fits the energy voltage slope and threshold voltage to the pulse energy and charging voltage data covered by each perturbation segment, and provides the estimated standard deviation and fitting condition number for both. If the condition number or residual exceeds the limit, the sequence is lengthened or the amplitude is increased before re-identification. The relative amplitude of the perturbation and the number of pulses covered by the identification module are given by the execution module and are limited by the amplitude upper limit set by the cavity constraints.

[0037] The attribution module receives a set of energy voltage slopes and threshold voltages before and after injection from the identification module, as well as the current halogen partial pressure increment and total pressure increment measured by the injection valve. It then separates the total pressure component from the changes in these two quantities using a pre-calibrated total pressure sensitivity coefficient, obtaining two relative improvement quantities that constitute the response vector. The weighted sum of the two components of the response vector is calculated as the ratio to the relative halogen partial pressure increment, serving as the normalized recovery rate. The module also calculates the composite amplitude, standard deviation, and orientation angle of the response vector, and assigns the source based on the angular interval of the orientation angle. The attribution module sends the normalized recovery rate and attribution conclusions to the execution module, and successively sends the normalized recovery rate and attribution flags to the gas exchange decision module.

[0038] The execution module operates according to the attribution conclusions. When the dominant source is halogen consumption, it adjusts the next injection amount and maintains the injection cycle based on the ratio of the target change rate to the measured change rate, and also includes the ratio of the normalized recovery rate to the new cavity baseline. When the dominant source is impurity accumulation, it suppresses halogen injection and performs partial evacuation and replacement based on the upper limit of the total cavity pressure. When the dominant source is discharge channel degradation or optical channel degradation, it suppresses halogen injection and includes this response in the corresponding attribution amount. Simultaneously, the execution module adjusts the relative amplitude of the next perturbation, the number of pulses covered, the length of the stabilization waiting period, and the hysteresis bandwidth of the trigger threshold based on the standard deviation of the normalized recovery rate.

[0039] The ventilation decision module divides the successive normalized recovery rate by the new cavity benchmark to obtain the halogen utilization efficiency sequence. It then uses recursive least squares to fit a linear relationship to the first part of this sequence and extrapolates it. Deviations where the measured value is lower than the extrapolated value are accumulated with relaxation. When the accumulated amount exceeds a threshold, a change point is determined to have occurred, and the module traces back along the accumulated amount to locate the corresponding accumulated pulse count. The ventilation decision module accumulates three types of attribution values—impurities, discharge channels, and optical channels—accumulates them separately according to attribution markers. When a change point has occurred, the proportion of impurity attribution values ​​reaches a threshold, and the total cavity pressure margin is insufficient to support a preset number of partial emissions and replacements simultaneously, it determines to execute full-cavity ventilation. The discharge channel attribution value and the optical channel attribution value are extrapolated to their respective tolerances, and the remaining pulse counts for cavity ventilation and window ventilation are output. The conclusion that a change point has occurred is sent back to the execution module to tighten the trigger threshold and extend the stabilization waiting period.

[0040] The actions and parameters applied by each of the above modules are all limited by the physical constraints within the cavity, including the upper and lower limits of the total pressure, halogen partial pressure, single injection volume, relative amplitude of perturbation, and stabilization waiting period. The lower limit of the stabilization waiting period is the time required for three to five gas circulation cycles within the cavity, determined by the fan flow rate and cavity volume; the lower limit of the single injection volume is determined by the minimum controllable opening of the injection valve.

[0041] Example 2 This embodiment provides a halogen injection decision method for excimer lasers, such as... Figure 2 As shown, the method includes the following steps.

[0042] Step S1: Collect pulse energy and charging voltage to obtain the rate of change of the sliding estimate of light emission capability relative to the cumulative number of pulses. When the rate of change reaches the trigger threshold, perform a halogen injection according to the injection amount and measure the halogen partial pressure increment and the total pressure increment in the cavity.

[0043] The sliding estimate of light emission capability is calculated pulse-by-pulse using the following formula: in, and The first Pulse and the first Sliding estimate of light output capability after pulse generation; and The first The pulse energy and charging voltage of the generated pulse; This is the sliding coefficient.

[0044] This formula uses the output energy obtained per unit charging voltage as a measure of light output capability. In constant pulse energy mode, the pulse energy is constant while the charging voltage increases with decay; in constant charging voltage mode, the charging voltage is constant while the pulse energy decreases with decay. In both modes, this estimated value decreases monotonically with decay, thus it does not depend on the specific operating mode. The sliding coefficient is taken from 0.001 to 0.02, corresponding to an equivalent sliding window of approximately 50 to 1000 pulses. The smaller the value, the stronger the fluctuation suppression but the slower the follow-up.

[0045] The rate of change is calculated relative to the cumulative number of pulses using the following formula: in, For the first Rate of change; and The first Next and first The cumulative number of pulses; and Take the cumulative pulse count respectively and The sliding estimate of the light output capability at that time.

[0046] In engineering, the rate of change is expressed as a relative decrease percentage per million pulses. The trigger condition is that this relative decrease percentage is not less than the trigger threshold, which is set at 0.05% to 0.5% per million pulses. After one trigger, the next trigger is allowed only after the relative decrease percentage falls back to the trigger threshold minus the hysteresis bandwidth, which is set at two to five times the standard deviation of the normalized recovery rate.

[0047] Step S2: Apply a zero-mean charging voltage perturbation before the injection and at the end of the stabilization waiting period after the injection. Identify the energy voltage slope and threshold voltage from the pulse energy and charging voltage under the perturbation.

[0048] Within the operating point neighborhood, the pulse energy and charging voltage satisfy the following linear relationship: in, and These are pulse energy and charging voltage, respectively. The energy voltage slope; This is the threshold voltage.

[0049] The energy voltage slope characterizes the output energy obtained per unit charging voltage, and is jointly determined by the discharge energy deposition efficiency, laser medium gain, and optical channel transmittance; the threshold voltage characterizes the minimum charging voltage required for excitation, and is jointly determined by the discharge initiation condition and the laser threshold gain condition. These are two independent degrees of freedom on the same operating characteristic, respectively emphasizing the optical channel and the discharge channel, and thus can be used to distinguish the sources of attenuation.

[0050] Under constant pulse energy control, the charging voltage is clamped within a very narrow band by the servo loop. The voltage span of the pulse energy and charging voltage data pair is insufficient to simultaneously and stably identify both parameters, resulting in an ill-conditioned fitting problem. Therefore, a perturbation is introduced before halogen injection and at the end of the stabilization period. A pulse-by-pulse sign sequence is superimposed on the target energy of the energy control loop, and the charging voltage varies with this sequence according to the following formula: in, and The first The charging voltage of the pulse and the first The charging voltage is obtained by energy control through pulse generation. The relative amplitude of the perturbation; For the first The sign of the pulse, with a value of or .

[0051] This sequence is in each The sum of the signs within a sub-block formed by the emitted pulses is zero. Dose control is concerned with the cumulative dose within the moving window, not the energy of a single shot. The signs cancel each other out within the sub-block, ensuring that any length is not less than [a certain value]. Within the window, the cumulative dose deviation is zero, achieving sufficient voltage span without degrading the stability of the moving average dose; when using signed and non-zero sequences or unidirectional steps, the cumulative dose may exhibit random walks or systematic bias. The relative amplitude of the perturbation ranges from 0.3% to 1.5%, with its upper limit given by the dose stability specification, not exceeding 2%; the number of pulses covered by the perturbation ranges from 200 to 2000; the number of pulses contained in the sub-block ranges from 8 to 64.

[0052] For each perturbation segment, the pulse energy and charging voltage data pairs are fitted using weighted least squares to obtain the energy-voltage slope and threshold voltage. The estimated standard deviation and condition number of the two are then given by the residual variance and the design matrix. If the condition number or residual exceeds the limit, the number of pulses covered or the relative amplitude of the perturbation is increased before re-identification.

[0053] Step S3: Extract the total pressure component from the change in the energy voltage slope and the threshold voltage according to the total pressure increment in the cavity, and obtain the relative improvement in slope and relative improvement in threshold voltage that constitute the response vector.

[0054] Perform one halogen premixed gas injection according to the currently set injection volume, and measure the increase in halogen partial pressure and the increase in total pressure in the cavity. The halogen volume fraction in the premixed gas is taken as 0.5% to 5%, the halogen partial pressure increase is taken as 1 Pa to 30 Pa, and the corresponding increase in total pressure in the cavity does not exceed 2% of the total pressure in the cavity. After the injection is completed, a stabilization waiting period is performed, which is taken as 3 s to 30 s, with the lower limit being the time required for three to five cycles of gas circulation in the cavity.

[0055] Halogen injection increases both the halogen partial pressure and the total pressure within the cavity. This total pressure, through discharge impedance, breakdown conditions, and exciter density, affects the energy voltage slope and threshold voltage. Without stripping, this improvement is mistakenly attributed to the halogen. Stripping is performed using the following formula: in, and These represent the change in energy voltage slope and the change in threshold voltage after stripping the total voltage component, respectively. and These are the energy voltage slopes before and after halogen implantation, respectively. and These are the threshold voltages before and after halogen implantation, respectively. This represents the increase in total intracavitary pressure caused by this injection; and These are the sensitivity coefficients of the energy voltage slope and the threshold voltage to the total pressure inside the cavity, respectively.

[0056] The calibration method for the two sensitivity coefficients is as follows: After the entire cavity is vented, a halogen-free buffer gas is injected into the cavity. A period of the charging voltage perturbation is applied before and after the buffer gas injection. The changes in the energy voltage slope and threshold voltage are divided by the increase in the total cavity pressure caused by the buffer gas injection, respectively. The buffer gas injection does not change the halogen partial pressure; all measured changes are attributed to the total cavity pressure, and the calibration is unaffected by halogens. During cavity aging, the two sensitivity coefficients are corrected to the first order according to the halogen utilization efficiency sequence.

[0057] The two relative improvements that constitute the response vector are obtained by the following formula: in, and These represent the relative improvement in slope and the relative improvement in threshold voltage, respectively. Taking the negative sign, the direction of improvement, which lowers the threshold voltage, is positive, and... The positive direction is consistent.

[0058] Step S4: The normalized recovery rate is obtained by dividing the weighted sum of the two components of the response vector by the ratio of the halogen partial pressure increment to the halogen partial pressure before injection, and the dominant source of this attenuation is determined by the angle range in which the direction angle of the response vector is located.

[0059] The normalized rate of recovery is obtained by the following formula: in, Normalized recovery rate; and These represent the increase in halogen partial pressure caused by this injection and the halogen partial pressure before injection, respectively. and These are the weights for the relative improvement in slope and the relative improvement in threshold voltage, respectively. and These are the standard deviations of the relative improvement in slope and the relative improvement in threshold voltage, respectively, which are obtained by propagating the estimated standard deviations of energy voltage slope and threshold voltage according to the above definition.

[0060] The slope improvement and threshold voltage improvement are two observed projections of the physical process of halogen concentration recovery on two degrees of freedom. They compensate for each other rather than enabling each other. Therefore, the numerator is the weighted sum of the two relative improvements. The weight is the inverse normalized of the square of the standard deviation. The projection with higher observation accuracy has a larger proportion in the synthesized quantity, minimizing the variance of the synthesized quantity when the accuracy of the two projections changes with the operating conditions. The denominator is the relative increment of halogen partial pressure rather than the absolute increment. The normalized recovery rate is a dimensionless quantity, and its value does not drift with the current halogen partial pressure level. Therefore, the discrimination threshold does not drift with the total pressure in the cavity.

[0061] The composite magnitude, standard deviation, and orientation angle of the response vector are obtained by the following formula: in, This is the synthesized amplitude; The standard deviation of the composite amplitude; This is the direction angle.

[0062] The four types of degradation processes affect the energy voltage slope and threshold voltage in different ways, and their corresponding relationships are shown in Table 1.

[0063] Table 1. The influence of four types of attenuation sources on the energy voltage slope and threshold voltage. The causes of each action direction in Table 1 are as follows: Halogen consumption causes the gain to decrease as the halogen concentration decreases, while simultaneously increasing the pump power density required to reach the threshold gain. Therefore, the slope decreases and the threshold voltage increases, both recovering synchronously after injection. Impurity accumulation enhances deep ultraviolet absorption within the cavity and quenches the upper energy levels. Simultaneously, electronegative impurities alter the electron energy distribution and degrade discharge uniformity. The injected halogen can only compensate for the gain but not remove the impurities, thus only the slope partially recovers. Electrode ablation increases the discharge gap and distorts the field pattern, raising the breakdown and threshold conditions, thus significantly increasing the threshold voltage. The slope decreases slightly due to the mismatch between the discharge volume and the optical cavity mode. Decreased optical surface transmittance causes multiplicative attenuation of the output, thus significantly decreasing the slope. The excimer dielectric gain is extremely high, compressing the effect of cavity loss changes on the threshold conditions, and the threshold voltage remains essentially unchanged.

[0064] The direction of the response vector is the criterion for determining the source, and the discrimination rule is as follows: Figure 3 As shown. When the synthesized amplitude is not less than three times the standard deviation of the synthesized amplitude, the relative improvement in slope is greater than zero, and the direction angle is within the first angle interval, the two relative improvements improve in the same direction and are of similar magnitude, indicating that the dominant source is halogen consumption; when the synthesized amplitude is not less than three times the standard deviation of the synthesized amplitude, the relative improvement in slope is greater than zero, and the direction angle is less than the lower limit of the first angle interval, the slope rises significantly while the threshold voltage does not fall, indicating that the injected halogen participated in gain recovery but the discharge conditions did not improve accordingly, indicating that the dominant source is impurity accumulation; when the synthesized amplitude is less than three times the standard deviation of the synthesized amplitude, or the relative improvement in slope is not greater than zero, no distinguishable response is produced by this injection, indicating that this halogen injection did not produce an effective response; when the synthesized amplitude is not less than three times the standard deviation of the synthesized amplitude, the relative improvement in slope is greater than zero, and the direction angle is greater than the upper limit of the first angle interval, the threshold voltage improves but the slope does not improve, which is inconsistent with the action mode of the four types of processes, indicating that the total voltage sensitivity coefficient needs to be recalibrated and the perturbation segment needs to be redone.

[0065] The first angular interval is obtained by expanding the median value of the directional angles obtained after several halogen injections following a complete cavitation gas exchange by three times the standard deviation of those directional angles to both sides. The number of injections is between 5 and 20. Since the irreversible components after cavitation gas exchange are negligible, the response direction at this point serves as the directional reference for pure halogen consumption. Using its own dispersion to define the boundary allows the false positive probability to be directly linked to the confidence level. The lower bound of the first angular interval is 10° to 30°, and the upper bound is 60° to 80°.

[0066] In step S5, when the dominant source is halogen consumption, the next injection amount is determined by the ratio of the preset change rate target value to the change rate and taking into account the normalized recovery rate; otherwise, halogen injection is suppressed and partial discharge and replacement are performed according to the upper limit of the total pressure in the cavity.

[0067] When the primary source of energy consumption is halogen, the next injection volume is determined by the following formula, while the injection cycle remains unchanged: in, and The first Next and first The injection volume per time; The preset target value for the rate of change; For the first Rate of change; For the first Normalized recovery rate of the second time; This represents the median of the normalized recovery rate obtained after several halogen injections following a complete cavity ventilation.

[0068] The median value is insensitive to outliers and can reduce the impact of individual distorted samples caused by valve malfunctions or production interruptions in the perturbation segment on the baseline. Therefore, the median value is used instead of the average value. The recovery rate ratio term in the above formula increases the amount of improvement gained per unit halogen partial pressure proportionally when the improvement decreases. The tuning result is limited by the upper and lower limits of the single injection volume to avoid over-injection to catch up with the target value of the rate of change in a stage where the utilization efficiency has already decreased significantly.

[0069] When the primary source is impurity accumulation, halogen injection is suppressed, and partial evacuation and replacement are performed based on the upper limit of the total pressure in the cavity. The upper limit of the total pressure in the cavity is 300 kPa to 500 kPa, and the halogen partial pressure is maintained between 70% and 120% of the new filling value.

[0070] When a halogen implantation fails to produce a valid response, the ratio of the monotonically increasing relative amount of the threshold voltage before implantation to the monotonically decreasing relative amount of the energy voltage slope before implantation is taken between several adjacent halogen implantations. The number of implantations is between 5 and 20. If this ratio is greater than a set multiple, the dominant source is determined to be discharge channel degradation, and the monotonically increasing relative amount is accumulated into the discharge channel attribution. If this ratio is less than the reciprocal of the set multiple, the dominant source is determined to be optical channel degradation, and the monotonically decreasing relative amount is accumulated into the optical channel attribution. The set multiple is between 2 and 5. When the dominant source is determined to be impurity accumulation, the weighted sum of the current halogen implantation is accumulated into the impurity attribution. The two types of irreversible attribution are extrapolated to their respective tolerances to obtain two independent predictions of the remaining pulse count: in, and These are the remaining pulse counts for cavity switching and window switching, respectively. and These are the cumulative values ​​of the discharge channel attribution and the optical channel attribution, respectively. and These are the tolerances for the two devices, with the former ranging from 0.15 to 0.25 and the latter from 0.20 to 0.30, determined by reverse calculation from the upper limit of the overall charging voltage and the energy specification. and These represent the growth rates of the relative cumulative pulse counts for the two, given by the least squares slope of the last 5 to 20 samples.

[0071] Replacing the cavity and replacing the window differ by an order of magnitude in terms of downtime, spare parts cost, and spare parts lead time. By comparing the remaining pulse counts of the two items with the expected time of the entire cavity air exchange on the same cumulative pulse count axis, actions falling within the same time window can be combined and scheduled for the same maintenance shutdown.

[0072] Simultaneously, the relative amplitude of the next perturbation and the number of pulses it covers are tuned according to the standard deviation of the normalized recovery rate: in, and The first Next and first The relative amplitude of the perturbation; and The first Next and first The number of pulses covered by the perturbation; For the first The standard deviation of the normalized recovery rate; The target value for the normalized recovery rate standard deviation is taken as 5% to 15% of the new cavity baseline recovery rate.

[0073] The variance of the least squares estimate is inversely proportional to the sample size and the square of the relative amplitude of the perturbation. To reduce the standard deviation by the same factor, the number of pulses covered needs to be increased according to the square relationship; therefore, the above formula is squared. The stabilization waiting period is adjusted according to the following formula. The waiting period setting factor ranges from 0.2 to 1.0. in, and The first Next and first A period of stabilization is expected.

[0074] At the end of the cavity age, energy fluctuations increase and the standard deviation rises. According to this tuning relationship, stronger and longer perturbations and longer waiting periods can be automatically obtained, while the confidence interval for the orientation angle remains within the width of the angle interval, allowing factorization to be completed even at the most critical stage. When the standard deviation falls, the perturbations and waiting periods fall synchronously. When uniformly using smaller amplitudes and shorter sequences, the confidence interval for the orientation angle at the end of the cavity age will be wider than the angle interval; when uniformly using larger amplitudes and longer sequences, the energy fluctuations and cycle time occupancy under normal operating conditions also increase.

[0075] Step S6: The normalized recovery rate from multiple injections forms a halogen utilization efficiency sequence. The deviation between the extrapolated value and the measured value of the first segment is used to accumulate the detection change point to determine the timing of the whole chamber ventilation.

[0076] The halogen utilization efficiency sequence, the extrapolated value, the deviation, and the cumulative deviation are obtained by the following formulas: in, For the first Secondary halogen utilization efficiency; To extrapolate the linear relationship fitted by recursive least squares from the first segment of the sequence to the th... The extrapolated value obtained next; and These are the intercept and slope of the linear relationship, respectively. Not greater than zero to indicate a slow descent; and The first The deviation amount and cumulative deviation value; The relaxation factor is taken as 0.3 to 0.8; This represents the standard deviation of the first segment of the sequence.

[0077] The accumulator employs a structure of first subtracting the relaxation factor and then rounding to zero, rather than directly summing the deviations or taking the second difference of the smoothed sequence. After subtracting the relaxation factor, the deviations caused by normal fluctuations are absorbed by the relaxation factor, and the accumulator is repeatedly truncated to zero in the early stage without drifting, eliminating the need for sequence smoothing and thus avoiding the introduction of smoothing phase lag. In the accelerated descent phase, the deviation remains positive and increases progressively. The accumulator structure superimposes these small deviations over time, resulting in a significant increase in the accumulator before a single deviation exceeds the fluctuation level, thus the detection sensitivity is not limited by the single-point signal-to-noise ratio. Rounding the accumulator to zero records the starting point of the deviation's persistence, which can be directly used for backtracking and localization.

[0078] Greater than It was determined that a change point had occurred. The cumulative threshold coefficient is set to 4 to 6. Then, tracing back along the cumulative amount, we find the starting sequence in which the cumulative amount changes from zero to a continuously positive value; the corresponding number of cumulative pulses is the location of the turning point. Figure 4 As shown. If the detection occurs when the threshold is crossed, but the change point is actually located before that, the backtracking step separates the identification time from the change point location; the detection lag is no longer equal to the lag in the change point location estimate. After the change point is confirmed, the extrapolation of the preceding linear relationship is stopped, and the final relationship is used instead.

[0079] The entire cavity ventilation is performed when all three of the following conditions are met simultaneously: a change point has occurred; the proportion of the impurity attribution amount in the sum of the impurity attribution amount, the discharge channel attribution amount, and the optical channel attribution amount is not less than the proportion threshold, which is between 0.4 and 0.7; and the difference between the upper limit of the total cavity pressure and the current total cavity pressure is less than the total cavity pressure margin required for partial discharge and replacement after a preset number of times, which is between 2 and 5 times.

[0080] The three conditions are presented in parallel for the following reasons: When only the first condition is met, although the utilization efficiency has already entered a period of accelerated decline, if the main cause of the accelerated decline is the electrode or optical surface, gas exchange cannot reverse the decline. In this case, cavity replacement or window replacement should be performed instead of gas exchange. When only the second condition is met, although the impurity ratio is high, if it is still in the slow decline phase, partial emission and replacement are still sufficient to maintain efficiency, and gas exchange is premature. The third condition is the feasibility condition for implementation. Partial emission and replacement need to leave a margin below the upper limit of the total pressure in the cavity. If the margin is insufficient, this path has reached its end.

[0081] When the conditions are met, the entire cavity is evacuated and refilled, and then the new cavity reference section is re-executed: the two total pressure sensitivity coefficients are recalibrated by injecting pure buffer gas, the new cavity reference recovery rate and the upper and lower limits of the first angle interval are recalculated, and the three types of attribution accounts are cleared to zero.

[0082] After the change point is confirmed, the trigger threshold is tightened by multiplying it by a tightening coefficient, and the stabilization waiting period is extended by multiplying it by an extension coefficient, both coefficients being between 0.2 and 0.8. Utilizing the fact that efficiency enters the accelerated decline phase, the same rate of change no longer easily triggers injection; each judgment has a longer stabilization time to lower the standard deviation, thus suppressing ineffective injections at the end of the cavity age. The impurities generated by excess halogen at the end of the process will superimpose with the already high impurity levels, making this stage the most effective in suppressing ineffective injections.

[0083] The key technical parameters and their value ranges in this embodiment are shown in Table 2.

[0084] Table 2 Key technical parameters and their value ranges Example 3 Based on the halogen injection decision method for excimer lasers described in Embodiment 2, this embodiment is a specific application of the method when the dominant source is determined to be halogen consumption. Specifically, it involves executing steps S1 to S5 sequentially and entering a set of specific values ​​in the halogen consumption branch. The purpose is to illustrate how, when the direction of the response vector enters the first angle interval, the next injection amount is determined by a preset target value of the rate of change and the normalized recovery rate, and to include the difference brought about by the recovery rate ratio term compared to setting it only by the rate of change.

[0085] In this embodiment, the device used is an argon-fluorine excimer laser with a repetition frequency of 4 kHz, constant pulse energy operation, a target pulse energy of 10.0 mJ, an operating charging voltage of approximately 11.9 kV, a total cavity pressure of 320 kPa, an upper limit of 400 kPa, a halogen partial pressure of 250 Pa, a halogen volume fraction of 1% in the premixed gas, and a halogen injection rhythm of approximately once every 2 million pulses, with a cumulative pulse count of [missing information]. hair.

[0086] With a sliding coefficient of 0.005, the estimated value of light emission capability decreased from 0.8483 mJ / kV to 0.8403 mJ / kV in the last 5 million shots, with a measured change rate of 0.19% per million shots. This exceeds the trigger threshold of 0.15% per million shots, initiating a halogen injection.

[0087] The perturbation segment before injection had a relative perturbation amplitude of 1.0%, covered 1500 pulses, and included 30 sub-block pulses. Weighted least-squares fitting of this segment yielded an energy voltage slope of 3.05000 mJ / kV before injection, with an estimated standard deviation of 0.0087 mJ / kV; the threshold voltage before injection was 8.62000 kV, with an estimated standard deviation of 0.0093 kV. The cumulative dose deviation for this segment of 1500 pulses was zero, and the moving average dose remained unchanged.

[0088] After one halogen injection, the measured halogen partial pressure increment was 8.5 Pa, corresponding to a total pressure increment of 0.85 kPa. The total pressure in the cavity rose to 320.85 kPa, which did not reach the upper limit. After an 8-second stabilization period, the post-injection perturbation phase was executed, yielding an energy voltage slope of 3.08976 mJ / kV and a threshold voltage of 8.54823 kV.

[0089] The new cavity reference section, calibrated with pure buffer gas injection, yielded a slope total pressure sensitivity of 0.0012 mJ / (kV·kPa) and a threshold voltage total pressure sensitivity of 0.0190 kV / kPa. After stripping the total pressure component, the change in energy voltage slope was 0.03874 mJ / kV, and the change in threshold voltage was -0.08792 kV. Without stripping, the change in threshold voltage was -0.07177 kV, indicating that the total pressure component masked approximately 18% of the halogen attribution improvement.

[0090] The resulting relative improvement in slope is 0.0127, and the relative improvement in threshold voltage is 0.0102. The standard deviations of the two relative improvements are 0.00285 and 0.00108, respectively, with corresponding weights of 0.126 and 0.874. The combined amplitude is 0.0163, and the standard deviation of the combined amplitude is 0.00101. The ratio of the two is 16.1, which is much greater than 3. The relative improvement in slope is greater than zero. The orientation angle is 38.8°, falling within the first angle range of 21° to 69° determined by the new cavity reference segment, indicating that the dominant source is halogen consumption.

[0091] The relative halogen partial pressure increment is 0.034, and the normalized recovery rate is 0.309; the new cavity reference recovery rate is 0.385, therefore the halogen utilization efficiency is 0.803. The standard deviation of the normalized recovery rate is 0.030, which is 7.8% of the new cavity reference recovery rate, lower than the standard deviation target value of 10%. The relative amplitude of the perturbation and the number of pulses covered remain unchanged, the stabilization waiting period remains unchanged, and the trigger threshold hysteresis bandwidth is taken as 0.090.

[0092] The target change rate is 0.12% per million rounds, and the actual change rate is 0.19% per million rounds. According to the injection volume tuning formula, the next injection volume is 0.787 times that of the current one, which does not reach the limit, and the injection cycle remains unchanged. Without considering the recovery rate ratio, the tuning result is 0.632 times that of the current one, which is about 20% lower than the above formula. Under the condition that the halogen utilization efficiency has dropped to 0.803, this injection volume is not enough to bring the change rate back to the target value, and an additional injection cycle is required for convergence.

[0093] Example 4 Based on the halogen injection decision method for excimer lasers described in Example 2, this example is a specific application of the method in the case where a single injection at the end of the cavity age does not produce an effective response. Specifically, after determining in step S4 that no effective response has been produced, the process proceeds to step S5, which involves a set of specific values ​​for the long-term drift factor. The purpose is to illustrate how to distinguish between discharge channel degradation and optical channel degradation by the ratio of the monotonically increasing relative amount of the threshold voltage before injection to the monotonically decreasing relative amount of the energy voltage slope before injection, thereby suppressing ineffective injection and providing the remaining number of pulses for cavity switching and window switching, respectively.

[0094] In this embodiment, the devices and operating conditions are the same as in Embodiment 3, and the cumulative pulse count is... The device is in the late stage of its cavity life, with the working charging voltage having risen to approximately 14.0 kV, the total cavity pressure at 371 kPa, and the halogen partial pressure at 268 Pa.

[0095] The sliding estimate of light emission capability decreased by 0.41% per million shots in the last 2 million shots, far exceeding the trigger threshold, thus initiating a halogen injection.

[0096] The relative standard deviation of the pulse energy at the end of the cavity age increased from 0.4% to 0.5%. The standard deviation of the normalized recovery rate obtained in the previous round was 0.061, which is 15.8% of the new cavity reference recovery rate of 0.385, and the ratio to the standard deviation target value of 0.0385 is 1.58. According to the tuning formula, the relative amplitude of the perturbation was tuned from 0.9% to 1.4%, which did not reach the upper limit of 2%; the number of pulses covered was tuned from 1200 pulses to 2996 pulses according to the square relationship, which was truncated by the upper limit of 2000 pulses; the stabilization waiting period was tuned from 8s to 10s. The energy voltage slope before injection was identified as 2.41000mJ / kV, with an estimated standard deviation of 0.0057mJ / kV; the threshold voltage before injection was 9.86000kV, with an estimated standard deviation of 0.0098kV; the corresponding normalized recovery rate standard deviation fell back to 0.027, which is 7.1% of the new cavity reference recovery rate.

[0097] Halogen injection was performed, with a halogen partial pressure increase of 9.0 Pa and a total cavity pressure increase of 0.90 kPa, bringing the total cavity pressure to 371.9 kPa, still below the upper limit of 400 kPa. After waiting 10 seconds, the energy voltage slope after injection was re-identified as 2.41333 mJ / kV, and the threshold voltage after injection was 9.87151 kV.

[0098] After first-order correction according to the halogen utilization efficiency sequence, the two total pressure sensitivity coefficients are 0.0011 mJ / (kV·kPa) and 0.0175 kV / kPa, respectively. After stripping the total pressure component, the change in energy voltage slope is 0.00234 mJ / kV, and the change in threshold voltage is -0.00424 kV. The unstripped threshold voltage change is positive 0.01151 kV, meaning that the threshold voltage increases instead of decreasing after injection. This increase comes from the increase in intracavity total pressure brought in by the injection, which, if not stripped, would be misinterpreted as a deterioration of discharge conditions. The stripped threshold voltage change of -0.00424 kV is a very small but correct improvement.

[0099] Therefore, the relative improvement in slope is 0.00097, and the relative improvement in threshold voltage is 0.00043. The standard deviations of the two relative improvements are 0.00237 and 0.00099, respectively, with corresponding weights of 0.149 and 0.851. The composite amplitude is 0.00106, and the standard deviation of the composite amplitude is 0.00091. The ratio of the two is 1.16, which is less than 3. Therefore, it is determined that the halogen injection did not produce an effective response, and the corresponding normalized recovery rate is only 0.015, which is 4% of the new cavity reference recovery rate.

[0100] Instead, a long-term comparison was made: spanning approximately 24 million halogen implantations, the energy voltage slope before implantation monotonically decreased from 2.42800 mJ / kV to 2.41000 mJ / kV, a relative decrease of 0.74%; the threshold voltage before implantation monotonically increased from 9.84500 kV to 9.86000 kV, a relative increase of 0.15%. The ratio of the relative amount of the monotonically increasing threshold voltage to the relative amount of the monotonically decreasing slope was 0.20, which is less than the reciprocal of the set factor of 3, indicating that the dominant source was optical channel degradation.

[0101] This suppresses subsequent halogen injections, and the relative decrease is accumulated in the optical channel attribution. The cumulative values, growth rates, tolerances, and extrapolated residual pulse numbers of the two types of attributions are shown in Table 3.

[0102] Table 3 Extrapolation results of the remaining pulse count of the two types of maintenance components in Example 4 As shown in Table 3, the number of remaining pulses after window replacement differs from that after cavity replacement by approximately 14 times. Therefore, material preparation and production scheduling should be based on window replacement at this stage, as there is still ample capacity in the cavity. According to the single scalar criterion of existing technologies, the 0.41% decrease per million rounds would be interpreted as insufficient halogen and requiring additional injection. This invention, however, identifies optical channel degradation immediately after one halogen injection and inhibits subsequent injections. Based on an injection rate of 9 Pa per 2 million rounds, approximately 75 ineffective injections can be blocked within the remaining approximately 150 million rounds before window replacement, equivalent to avoiding approximately 680 Pa of cumulative halogen overload.

[0103] Example 5 Based on the halogen injection decision method for excimer lasers described in Example 2, this example is a specific application of the method when the dominant source is determined to be impurity accumulation, which subsequently triggers whole-cavity venting. Specifically, it is a set of specific values ​​that advance from the partial emission and displacement branch in step S5 to the venting determination in step S6. The purpose is to illustrate how to suppress halogen injection and switch to partial emission and displacement when the directional angle is below the lower boundary of the first angle interval, and how the three conditions of change point, impurity attribution ratio, and total pressure margin in the cavity jointly determine the timing of whole-cavity venting.

[0104] In this embodiment, the device used is a krypton-fluorine excimer laser with a repetition rate of 2 kHz, intermittent emission, a duty cycle of approximately 45%, a target pulse energy of 10.0 mJ, an operating charging voltage of approximately 14.0 kV, a total cavity pressure of 386 kPa, an upper limit of 420 kPa, a halogen partial pressure of 195 Pa, and a cumulative pulse count of [missing information]. hair.

[0105] After initiating halogen injection, a pre-injection perturbation segment was executed with a perturbation relative amplitude of 1.0% and a coverage of 1500 pulses. The pre-injection energy voltage slope was identified as 2.76000 mJ / kV, with an estimated standard deviation of 0.0083 mJ / kV; the pre-injection threshold voltage was 10.42000 kV, with an estimated standard deviation of 0.0109 kV. Injection was performed, with a halogen partial voltage increment of 6.2 Pa and a total intracavity pressure increment of 0.62 kPa. After waiting 10 seconds, the post-injection energy voltage slope was identified as 2.80607 mJ / kV, and the post-injection threshold voltage was 10.42451 kV.

[0106] By separating the total pressure component from the corrected slope total pressure sensitivity of 0.0013 mJ / (kV·kPa) and the threshold voltage total pressure sensitivity of 0.0182 kV / kPa at this cavity age, the energy voltage slope change is 0.04526 mJ / kV, and the threshold voltage change is -0.00677 kV; corresponding to a relative improvement in slope of 0.0164 and a relative improvement in threshold voltage of 0.00065. The standard deviations of the two relative improvements are 0.0030 and 0.00104, respectively, with corresponding weights of 0.107 and 0.893; the composite amplitude is 0.0164, and the standard deviation of the composite amplitude is 0.00098, with a ratio of 16.8, which is greater than 3; the relative improvement in slope is greater than zero; the direction angle is 2.3°, which is less than the lower limit of the first angle interval of 19°, indicating that the dominant source is impurity accumulation.

[0107] This assessment is consistent with the physical process: the significant increase in the slope indicates that the injected halogen participated in gain recovery, while the near-absence of the threshold voltage indicates that the discharge conditions did not improve accordingly, consistent with the impurity characteristics of coexisting deep ultraviolet absorption and deterioration of discharge uniformity within the cavity. The normalized recovery rate in this study was 0.073, far lower than the baseline recovery rate of 0.341 for the new cavity in this machine.

[0108] Based on this, halogen injection is suppressed, and partial evacuation and replacement are performed according to the upper limit of the total pressure in the cavity. The total pressure margin required for this partial evacuation and replacement is 8 kPa, and the current remaining margin is 34 kPa, which can support 4 times, which is greater than the preset replacement number of 3 times. Therefore, no full cavity gas exchange is performed this time. After the replacement is completed, the weighted synthesis amount of this response is accumulated into the impurity attribution amount, and its cumulative value reaches 0.412.

[0109] The cumulative pulse count has been increased to At the time of the event, the cumulative deviation of the halogen utilization efficiency sequence reached 5.3 times the standard deviation of the previous deviation, exceeding the cumulative threshold coefficient of 5, indicating that a change point had occurred; tracing back along the cumulative value, the starting point where the cumulative value changed from zero to a continuously positive value corresponds to the cumulative pulse number. The change point actually occurs approximately before the detection point. At this point, the cumulative impurity attribution value is 0.518, the cumulative discharge channel attribution value is 0.062, the cumulative optical channel attribution value is 0.104, and the impurity attribution percentage is 0.757, exceeding the percentage threshold of 0.55; the total pressure inside the cavity is 405 kPa, with a remaining margin of 15 kPa, which is less than the 24 kPa required for the preset number of replacements (3). All three conditions are met simultaneously, and a complete cavity ventilation is performed.

[0110] After the entire cavity is ventilated, the new cavity reference is reconstructed: the slope total pressure sensitivity of 0.0014 mJ / (kV·kPa) and the threshold voltage total pressure sensitivity of 0.0206 kV / kPa are recalibrated by injecting pure buffer gas; the new cavity reference recovery rate of 0.341 is reconstructed by the median value of the normalized recovery rate obtained from the subsequent 12 halogen injections; the first angle range of 20° to 71° is reconstructed by the directional angle distribution of these 12 injections; the three types of attribution accounts are cleared.

[0111] In this example, the change point location is given by backtracking, and its difference from the detected point is... Following the method of calculating the second difference of the smooth sequence, the identification point will be further shifted back at the same fluctuation level, and the whole-cavity ventilation action will fall within the impurity self-acceleration zone. In this example, whole-cavity ventilation is performed when the impurity attribution ratio reaches 0.757, and ventilation can recover utilization efficiency; when ventilation is performed based solely on the change point without checking this ratio, in the case of optical channel degradation as in Example 4, ventilation cannot recover utilization efficiency.

[0112] The identification and discrimination results of Examples 3 to 5 are summarized in Table 4.

[0113] Table 4 Identification and discrimination results of Examples 3 to 5 In Table 4, the slope before and after injection both refer to the energy voltage slope, with the unit being mJ / kV; the threshold voltage is in kV; the total pressure in the cavity and the increment of the total pressure in the cavity are in kPa; the halogen partial pressure and the increment of the halogen partial pressure are in Pa; and the relative improvement of the slope, the relative improvement of the threshold voltage, the composite amplitude, the standard deviation of the composite amplitude, and the normalized recovery rate are all dimensionless quantities.

[0114] As shown in Table 4, the ratios of the synthesized amplitude to its standard deviation for the three embodiments are 16.1, 1.16, and 16.8, respectively. Embodiment 4 falls within the fluctuation band, while Embodiments 3 and 5 fall outside it, indicating that the criterion has sufficient discriminative power. The synthesized amplitudes of Embodiments 3 and 5 are both on the order of 0.0164, but their orientation angles are 38.8° and 2.3°, respectively, and they are judged to be different dominant sources. It is evident that the magnitude of the recovery is insufficient to distinguish the source, and the orientation angle serves as a factor, which cannot be provided by a single scalar observation.

[0115] Example 6 Based on the halogen injection decision method for excimer lasers described in Example 2, this example is a specific application of this method under the condition of intermittent light output at low duty cycle. Specifically, it involves executing a set of specific values ​​for the perturbation and stabilization waiting period self-tuning in step S5 for three consecutive rounds. The purpose is to illustrate how, when the standard deviation of a single identification is too high, the relative amplitude of the perturbation, the number of pulses covered by the perturbation, the stabilization waiting period, and the trigger threshold hysteresis bandwidth can be adjusted in reverse from the standard deviation of the normalized recovery rate, so that the confidence of the direction angle discrimination can meet the standard under different duty cycle conditions without the need to preset parameter tables separately.

[0116] In this embodiment, the device used is an argon-fluorine excimer laser with a repetition rate of 6 kHz, a duty cycle of approximately 12%, a total cavity pressure of 305 kPa, a halogen partial pressure of 242 Pa, a single-shot halogen partial pressure increment of 5.5 Pa, a local new cavity reference refractive index of 0.371, and a target value for the normalized refractive index standard deviation of 0.0371, which is 10% of the new cavity reference refractive index. The cumulative number of pulses is... hair.

[0117] At low duty cycles, halogen consumption is primarily driven by chemical reaction pathways related to exposure time, leading to an underestimation of consumption by the open-loop model based on cumulative pulse count. Simultaneously, long intervals limit the number of continuous pulses available for each perturbation segment, resulting in a higher standard deviation for single-shot identification. The results of the three rounds of parameter tuning are shown in Table 5.

[0118] Table 5. Results of Three-Round Perturbation and Waiting Period Tuning in Example 6 The standard deviation of the normalized recovery rate obtained in round j is 0.052, which is 14.0% of the new cavity reference recovery rate, and the ratio to the target standard deviation is 1.40. The relative amplitude of the perturbation is tuned to... The perturbation pulse number is less than the upper limit of 2%; the perturbation pulse number is tuned according to the square relationship. The firing rate is 1200 rounds, which is less than the maximum of 2000 rounds; the stable waiting period is set at a waiting period setting factor of 0.5. s, take 10.0s; trigger threshold hysteresis bandwidth is taken as 0.156.

[0119] In the j+1 round of identification using the tuned parameters, the normalized recovery rate standard deviation was 0.034, which is 9.2% of the new cavity baseline recovery rate, already below the target value. The three-standard-deviation half-width of the azimuth angle narrowed from 12.4° to 8.1°, less than one-third of the first angle interval half-width of 24°, indicating that the confidence level for azimuth angle discrimination met the standard. The dominant source of this round of identification was determined to be halogen consumption; the injection volume was tuned according to the injection volume tuning method, and the injection cycle was maintained.

[0120] The standard deviation of the normalized recovery rate obtained in the j+2 round is 0.031, and the ratio of the standard deviation target value is 0.84, which is less than 1. Therefore, reverse tuning is performed: the relative amplitude of the perturbation drops to 0.70%, the number of perturbation pulses drops to 850, the stabilization waiting period drops to 9.2s, and the perturbation amplitude and production cycle time decrease synchronously.

[0121] As shown in Table 5, under the same tuning relationship, the high standard deviation condition automatically obtains stronger and longer perturbations and longer waiting periods to maintain the confidence level of the azimuth angle determination, while the low standard deviation condition automatically decreases to reduce the perturbation amplitude, eliminating the need to preset parameter tables for different duty cycle conditions. When a uniform perturbation relative amplitude of 0.60% and a perturbation pulse count of 600 are taken, the half-width of the azimuth angle (three times the standard deviation) in this example will consistently remain above 12°. The first angle interval is reconstructed according to the new cavity reference segment after each cavity air change, and its angle gradually narrows with random aging. At the end of the model's lifespan, the width of the first angle interval narrows from 48° to approximately 30°, and this half-width will exceed one-third of the interval's half-width. When a uniform perturbation relative amplitude of 1.4% and a perturbation pulse count of 2000 are taken at the end of Example 4, the energy jitter and the approximately 2000 pulses per cycle occupancy during the entire lifespan of the conventional condition like in Example 3 also increase accordingly.

[0122] The numerical values ​​in the above embodiments are implementation examples and are not intended to limit the scope of protection of the present invention.

[0123] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A halogen injection decision method for an excimer laser, characterized in that, include: The pulse energy and charging voltage are collected to obtain the rate of change of the sliding estimate of the light emission capability relative to the cumulative number of pulses. When the rate of change reaches the trigger threshold, a halogen injection is performed according to the injection amount, and the halogen partial pressure increment and the total pressure increment in the cavity are measured. Before the injection and at the end of the stabilization waiting period after the injection, a zero-mean charging voltage perturbation is applied, and the energy voltage slope and threshold voltage are identified by the pulse energy and charging voltage under the perturbation. The total pressure component is extracted from the changes in the energy voltage slope and the threshold voltage based on the total pressure increment in the cavity, resulting in the relative improvement in slope and the relative improvement in threshold voltage that constitute the response vector. The normalized recovery rate is obtained by weighting the two components of the response vector and dividing it by the ratio of the halogen partial pressure increment to the halogen partial pressure before injection. The dominant source of this attenuation is determined by the angle range in which the direction angle of the response vector is located. When the primary source is halogen consumption, the next injection amount is adjusted according to the ratio of the preset change rate target value to the change rate and taking into account the normalized recovery rate; otherwise, halogen injection is suppressed and partial discharge and replacement are performed according to the upper limit of the total pressure in the cavity. The normalized recovery rate from multiple injections constitutes a halogen utilization efficiency sequence. The deviation between the extrapolated value and the measured value of the first segment is accumulated to detect the change point, and the timing of whole-cavity ventilation is determined.

2. The halogen injection decision method for excimer lasers according to claim 1, characterized in that, The sliding estimate of the light extraction capability is calculated pulse-by-pulse using the following formula: in, and The first Pulse and the first Sliding estimate of light output capability after pulse generation; and The first The pulse energy and charging voltage of the generated pulse; The sliding coefficient; The rate of change is calculated using the following formula: in, For the first Rate of change; and The first Next and first The cumulative number of pulses; and Take the cumulative pulse count respectively and The sliding estimate of the light output capability at that time.

3. The halogen injection decision method for excimer lasers according to claim 1, characterized in that, The charging voltage perturbation is a sequence of positive and negative signs taken pulse by pulse, and the sequence is in each The sum of the signs within the sub-block formed by the emitted pulses is zero, and the charging voltage varies with the sequence according to the following formula: in, and The first The charging voltage of the pulse and the first The charging voltage is obtained by energy control through pulse generation. The relative amplitude of the perturbation; For the first The symbol for the pulse; The number of pulses contained in the sub-block; The energy voltage slope and the threshold voltage are covered by the perturbation. The pulse energy and charging voltage are obtained by weighted least squares fitting using the following formula, and the estimated standard deviations of both are also obtained: in, and These are pulse energy and charging voltage, respectively. The energy voltage slope; Threshold voltage; The number of pulses covered by the perturbation.

4. The halogen injection decision method for excimer lasers according to claim 3, characterized in that, The total stripping pressure component is calculated using the following formula: in, and These represent the change in energy voltage slope and the change in threshold voltage after stripping the total voltage component, respectively. and These are the energy voltage slopes before and after halogen implantation, respectively. and These are the threshold voltages before and after halogen implantation, respectively. This represents the total pressure increment within the cavity; and These are the sensitivity coefficients of the energy voltage slope and the threshold voltage to the total pressure inside the cavity, respectively. The sensitivity coefficient is calibrated as follows: after the entire cavity is ventilated, a halogen-free buffer gas is injected into the cavity. A period of the charging voltage perturbation is applied before and after the buffer gas injection. The change in the energy voltage slope and the threshold voltage are respectively divided by the total pressure increment in the cavity caused by the buffer gas injection.

5. The halogen injection decision method for excimer lasers according to claim 4, characterized in that, The relative improvement in slope and the relative improvement in threshold voltage are obtained by the following formula: in, and These represent the relative improvement in slope and the relative improvement in threshold voltage, respectively. The normalized recovery rate is obtained by the following formula: in, Normalized recovery rate; and These are the halogen partial pressure increment and the halogen partial pressure before injection, respectively; and These are the weights for the relative improvement in slope and the relative improvement in threshold voltage, respectively. and These are the standard deviations of the relative improvement in slope and the relative improvement in threshold voltage, respectively.

6. The halogen injection decision method for excimer lasers according to claim 5, characterized in that, The composite magnitude of the response vector, the standard deviation of the composite magnitude, and the direction angle are obtained by the following formula: in, This is the synthesized amplitude; The standard deviation of the composite amplitude; It is the direction angle; The primary sources for determining this attenuation include: Not less than , Greater than zero and When the angle is in the first range, the dominant source is determined to be halogen consumption; Not less than , Greater than zero and When the angle is less than the lower bound of the first angle range, the dominant source is determined to be impurity accumulation; Less than or If the value is not greater than zero, it is determined that the halogen injection did not produce a valid response. The first angle range is obtained by expanding the median value of the directional angle obtained after several halogen injections following a complete cavity ventilation by three times the standard deviation of the directional angle to both sides.

7. The halogen injection decision method for excimer lasers according to claim 6, characterized in that, When the primary source is determined to be impurity accumulation, the weighted sum of the current data is included in the impurity attribution amount. When the current halogen implantation does not produce an effective response, the ratio of the monotonically increasing relative amount of the threshold voltage before implantation to the monotonically decreasing relative amount of the energy voltage slope before implantation is taken between several adjacent halogen implantations. If the ratio is greater than a set multiple, the primary source is determined to be discharge channel degradation, and the monotonically increasing relative amount is included in the discharge channel attribution amount. If the ratio is less than the reciprocal of the set multiple, the primary source is determined to be optical channel degradation, and the monotonically decreasing relative amount is included in the optical channel attribution amount. The discharge channel attribution and the optical channel attribution are respectively extrapolated to their respective tolerances using the following formulas to obtain the remaining number of pulses for cavity switching and the remaining number of pulses for window switching: in, and These are the remaining pulse counts for cavity switching and window switching, respectively. and These are the cumulative values ​​of the discharge channel attribution and the optical channel attribution, respectively. and These are the tolerances for the two, respectively; and These represent the growth rates of the relative cumulative pulse counts for each.

8. The halogen injection decision method for excimer lasers according to claim 7, characterized in that, The next injection volume is determined using the following formula: in, and The first Next and first The injection volume per time; The preset target value for the rate of change; For the first Rate of change; For the first Normalized recovery rate of the second time; The median of the normalized recovery rate obtained after several halogen injections following a complete cavity ventilation; The relative amplitude and the number of pulses covered by the next charging voltage perturbation are then adjusted according to the standard deviation of the normalized recovery rate. in, and The first Next and first The relative amplitude of the perturbation; and The first Next and first The number of pulses covered by the perturbation; For the first The standard deviation of the normalized recovery rate; This is the target value for the standard deviation of the normalized recovery rate.

9. The halogen injection decision method for excimer lasers according to claim 7, characterized in that, The halogen utilization efficiency sequence, the extrapolated value, the deviation, and the cumulative deviation are obtained by the following formulas: in, For the first Secondary halogen utilization efficiency; For the first The cumulative number of pulses; For the first Normalized recovery rate of the second time; The median of the normalized recovery rate obtained after several halogen injections following a complete cavity ventilation; To extrapolate the linear relationship fitted from the first segment of the sequence to the th... The extrapolated value obtained next; and These are the intercept and slope of the linear relationship, respectively; and The first The deviation amount and cumulative deviation value; The relaxation coefficient; The standard deviation of the deviation of the first segment of the sequence; Greater than The change point is determined to have occurred at that time, and is then... The cumulative number of pulses corresponding to the initial sequence of changes from zero to a continuously positive value is taken as the position of the change point. This is the cumulative threshold coefficient; The timing for determining the cavity ventilation is as follows: the change point has occurred, the proportion of the impurity attribution amount in the sum of the impurity attribution amount, the discharge channel attribution amount, and the optical channel attribution amount is not less than the proportion threshold, and the difference between the upper limit of the cavity total pressure and the current cavity total pressure is less than the cavity total pressure margin required for a preset number of partial discharges and replacements. When all three conditions are met, cavity ventilation is performed.

10. A halogen injection decision system for an excimer laser, characterized in that, include: The observation module is used to collect pulse energy and charging voltage, obtain the rate of change of the sliding estimate of light output capability relative to the cumulative number of pulses, and initiate a halogen injection according to the injection amount when the rate of change reaches the trigger threshold, and measure the halogen partial pressure increment and the total pressure increment in the cavity. The identification module is used to apply a zero-mean charging voltage perturbation before the injection and at the end of the stabilization waiting period after the injection, and to identify the energy voltage slope and threshold voltage from the pulse energy and charging voltage under the perturbation. The attribution module is used to separate the total pressure component from the change in the energy voltage slope and the threshold voltage according to the total pressure increment in the cavity, to obtain the relative improvement of the slope and the relative improvement of the threshold voltage that constitute the response vector, to obtain the normalized recovery rate by dividing the weighted sum of the two components of the response vector by the ratio of the halogen partial pressure increment to the halogen partial pressure before injection, and to determine the dominant source of this attenuation by the angle range in which the direction angle of the response vector is located. The execution module is used to, when the dominant source is halogen consumption, adjust the next injection amount according to the ratio of a preset change rate target value to the change rate and take into account the normalized recovery rate; otherwise, it suppresses halogen injection and performs partial discharge and replacement according to the upper limit of the total pressure in the cavity. The ventilation decision module is used to construct a halogen utilization efficiency sequence from the normalized recovery rates of multiple injections, and to detect change points by accumulating the deviation between the extrapolated value and the measured value of the first segment, so as to determine the timing of whole-cavity ventilation.