Laser maintenance apparatus and method

Abstract

Disclosed are devices and methods that use one or more trained classifiers to determine whether to take automated containment measures, and if so, prioritize less invasive automated containment measures, and take more invasive automated containment measures only if the less invasive automated containment measures are ineffective. Also disclosed are devices and methods that train a first classifier using a first data type to obtain a first classification using the first classifier, and use the first classification to train a second classifier that classifies a second data type.

Description

[Technical Field]

[0001] (CROSS-REFERENCE TO RELATED APPLICATIONS)

[0001] This application claims priority to U.S. Application No. 63 / 434,785, entitled "LASER MAINTENANCE APPARATUS AND METHOD," filed December 22, 2022, the entire contents of which are incorporated herein by reference.

[0002]

[0002] The subject matter of this disclosure relates to an apparatus and method for maintenance of a laser system, and more particularly to an apparatus and method for automatically identifying health conditions of a laser system that may benefit from the automatic application of maintenance containment measures. [Background technology]

[0003]

[0003] Laser systems are used, for example, as radiation sources in facilities, or "fabs," that manufacture semiconductor devices. Laser systems typically require periodic maintenance activities to ensure they operate according to specifications and to avoid manufacturing defects. These maintenance activities include maintenance events in which a field service engineer (FSE) visits the fab where the laser system is installed and physically interacts with the laser system. Such maintenance events include, for example, troubleshooting, part replacement, calibration, and alignment.

[0004] A callback is a maintenance event that requires an FSE to return to a previously serviced laser system within a set time (e.g., four weeks) after the initial maintenance event. Currently, if tool performance is insufficient after a maintenance event, the FSE must travel to the fab and manually make the necessary changes, which can add significant downtime to the downtime associated with the initial maintenance event. As a result, callbacks are highly unpopular with customers. Introducing an apparatus and method that could reduce the number of callbacks or reduce the need for an FSE to interact with a laser system would be highly beneficial. This situation creates a need for the subject matter of the present disclosure. Summary of the Invention

[0005] The following presents a simplified summary of one or more embodiments to provide a better understanding of the subject matter of the present disclosure. This summary is not intended to provide a detailed overview of all of the embodiments discussed, nor is it intended to deem any element essential or critical. Nor is it intended to delineate the full scope of any embodiment. Its sole purpose is to present some concepts of one or more embodiments in a simplified form to the reader prior to the more detailed description that is presented later.

[0006] According to one aspect of the embodiment, an automated maintenance system is disclosed that can monitor the performance of a laser system and request automatic intervention when certain conditions are met. These interventions are also referred to as "containment measures" or simply "containments" because they are measures aimed at restoring operation within specifications. According to another aspect of the embodiment, the containments are ranked based on invasiveness, with less invasive containments attempted first and more invasive containments performed only if they fail to restore normal operation.

[0007]

[0007] According to another aspect of an embodiment, a method for maintaining a laser system is disclosed, the method including: using a classifier to determine whether a negative laser condition exists that can be corrected by an automatic containment action; selecting a least invasive automatic containment action from a group of automatic containment actions as a first automatic containment action; performing the first automatic containment action; using a classifier to determine whether the negative laser condition was corrected by performing the first automatic containment action; if the classifier determines that the negative laser condition has not been corrected by the first automatic containment action, selecting a next least invasive containment action from the group of automatic containment actions as a next automatic containment action; and performing the next automatic containment action.

[0008]

[0008] The group of automatic containment measures may include N automatic containment measures ranked according to invasiveness, and the method may further include sequentially performing one of the N automatic containment measures until the classifier determines that either one of the N automatic containment measures corrected the negative laser condition or determines that none of the N automatic containment measures corrected the negative laser condition.

[0009]

[0009] According to another aspect of an embodiment, a method for maintaining a laser system is disclosed, the method including training a first classifier using a first data type, classifying first data of the first data type using the first classifier, training a second classifier that classifies second data of a second data type using the first data, classifying the second data using the second classifier and determining whether to take an automatic containment action, and if the second classifier determines that an automatic containment action should be taken, selecting the automatic containment action to be taken and taking the automatic containment action.

[0010] The first data type may comprise data obtained from a gain scan performed by increasing the magnitude of a voltage applied to an electrode in a discharge chamber of the laser system and determining the laser output energy as a function of the voltage magnitude. The first data type may comprise data obtained from a timing scan performed by sequentially increasing the time gap between applying a main oscillator trigger voltage to a main oscillator chamber of the laser system and applying a power amplifier trigger voltage to a power amplifier chamber of the laser system and determining a variation in an operating parameter as a function of the duration of the time gap. The second data type may comprise burst statistical data. Selecting an automatic containment action may include selecting the least invasive automatic containment action that has not yet been taken from a group of automatic containment actions.

[0011]

[0011] According to another aspect of an embodiment, a method for maintaining a laser system is disclosed, the method including classifying a first data type using a first classifier to obtain the first type of classified data, training a second classifier that classifies a second data type using the first type of classified data, determining using the second classifier whether to take an automatic containment action based on the second classifier's classification of the second data type, selecting the automatic containment action to be taken if the second classifier determines that an automatic containment action should be taken, and taking the automatic containment action.

[0012] The first data type may comprise data obtained by a gain scan that sequentially increases the magnitude of a voltage applied to an electrode in a discharge chamber of the laser system and determines the laser output energy as a function of the voltage magnitude. The first data type may comprise data obtained from a timing scan that sequentially increases the time gap between applying a main oscillator trigger voltage to a main oscillator chamber of the laser system and applying a power amplifier trigger voltage to a power amplifier chamber of the laser system and determining a variation in an operating parameter as a function of the duration of the time gap. The second data type may comprise burst statistical data. Selecting an automatic containment action may include selecting the least invasive automatic containment action that has not yet been taken from a group of automatic containment actions.

[0013]

[0013] According to another aspect of an embodiment, a maintenance apparatus for a laser system is disclosed, the apparatus comprising: a classifier adapted to determine whether a negative laser condition exists that can be corrected by an automatic containment measure; an automatic containment measure selection unit adapted to select a least invasive automatic containment measure from a plurality of automatic containment measures as a first automatic containment measure; and a first automatic containment measure execution unit adapted to cause the laser system to perform the first automatic containment measure, wherein the classifier is further adapted to determine whether the negative laser condition has been corrected by performing the first automatic containment measure, and if it is determined that the negative laser condition has not been corrected by performing the first automatic containment measure, selects a more invasive automatic containment measure from the plurality of automatic containment measures as the next automatic containment measure, and the apparatus causes the laser system to perform a second automatic containment measure if the classifier determines that the negative laser condition has not been corrected by performing the first automatic containment measure.

[0014]

[0014] The multiple automatic containment measures may comprise N automatic containment measures ranked according to invasiveness, and the device may further comprise a unit for respectively performing one of the N automatic containment measures until the classifier determines that either one of the N automatic containment measures has corrected the negative laser condition or determines that none of the N automatic containment measures has corrected the negative laser condition.

[0015]

[0015] According to another aspect of an embodiment, a maintenance device for a laser system is disclosed, the device comprising: a first classifier trained using a first data type; a second classifier trained using data classified by the first classifier and a second data type and adapted to determine whether to take an automatic containment action; and a selector for selecting the automatic containment action to be taken if the second classifier determines that an automatic containment action should be taken.

[0016] The first data type may comprise data obtained from a gain scan performed by increasing the magnitude of a voltage applied to an electrode in a discharge chamber of the laser system and determining the laser output energy as a function of the voltage magnitude. The first data type may comprise data obtained from a timing scan performed by sequentially increasing the time gap between applying a main oscillator trigger voltage to a main oscillator chamber of the laser system and applying a power amplifier trigger voltage to a power amplifier chamber of the laser system and determining the variation of an operating parameter as a function of the duration of the time gap. The second data type may be burst statistical data. The selector may be adapted to select the least invasive automatic containment measure that has not yet been taken.

[0017]

[0017] According to another aspect of an embodiment, a non-transitory computer-readable medium is disclosed having stored thereon instructions that, when executed by a processor, cause the processor to perform operations including: using a classifier to determine whether a negative laser condition exists that can be corrected by an automatic containment measure; selecting a least invasive automatic containment measure from the plurality of automatic containment measures as a first automatic containment measure; performing the first automatic containment measure; using the classifier to determine whether the negative laser condition was corrected by performing the first automatic containment measure; and if the classifier determines that the negative laser condition has not been corrected by the first automatic containment measure, selecting the next least invasive containment measure from the plurality of automatic containment measures as a next automatic containment measure, and performing the next automatic containment measure.

[0018] Further embodiments, features, and advantages of the subject matter of the present disclosure, as well as the structure and operation of the various embodiments, are described in detail below with reference to the accompanying drawings. [Brief explanation of the drawings]

[0019]

[0019] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate the subject matter of the present disclosure and, together with the following description, serve to explain the principles of the subject matter of the present disclosure and to enable one skilled in the art to make and use the subject matter of the present disclosure.

[0020] [Figure 1]

[0020] FIG. 1 is a diagram of a photolithography system that may benefit from examples of certain aspects of the embodiments. [Figure 2]

[0021] 1 is a diagram of a light source for a photolithography system that may benefit from examples of certain aspects of the embodiments. [Figure 3]

[0022] FIG. 1 is a functional block diagram of a configuration for automated maintenance of a laser system according to an aspect of an embodiment. [Figure 4]

[0023] FIG. 1 is a functional block diagram of a configuration for training a classifier for use in an automated maintenance system according to an aspect of an embodiment. [Figure 5]

[0024] 1 is a flowchart of a method for automated maintenance of a laser system in accordance with an aspect of an embodiment. [Figures 6A-6D]

[0025] FIG. 1 illustrates an operational mode of an automated maintenance system according to an aspect of an embodiment. [Figure 7]

[0026] 1 is a flowchart of a method for automated maintenance of a laser system in accordance with an aspect of an embodiment. [Figures 8A-8C]

[0027] FIG. 1 illustrates operational stages of an automated maintenance system according to an aspect of an embodiment. [Figure 9]

[0028] 1 is a flowchart of a method for automated maintenance of a laser system in accordance with an aspect of an embodiment. [Figure 10]

[0029] FIG. 1 is a functional block diagram of a configuration for automated maintenance of a laser system according to an aspect of an embodiment.

[0021]

[0030] Further features and advantages of the presently disclosed subject matter, as well as the structure and operation of various embodiments of the presently disclosed subject matter, are described in detail below with reference to the accompanying drawings. It should be noted that the scope of the present disclosure is not limited to the specific embodiments expressly described herein. Such embodiments are included herein for illustrative purposes only. Additional embodiments will be apparent to those skilled in the art based on the teachings presented herein. DETAILED DESCRIPTION OF THE INVENTION

[0022]

[0031] Various embodiments are described below with reference to the drawings, in which like reference numerals indicate like or corresponding elements. In the following description, numerous specific details are set forth in order to provide a more thorough understanding of one or more embodiments. It will be apparent, upon review of this disclosure, that in some or all cases, any of the embodiments described below can be practiced without adopting the specific design details described below. In some instances, well-known structures and devices are shown in block diagram form to facilitate describing one or more embodiments.

[0023]

[0032] Systems such as those described herein may provide advantages in a wide range of applications and implementations. For ease of illustration, one such application, as a specific, non-limiting example, is semiconductor photolithography. Furthermore, the following examples relate to systems that generate radiation in the deep ultraviolet (DUV) portion of the electromagnetic spectrum, i.e., light having wavelengths ranging from about 100 nanometers (nm) to about 400 nm. However, it will be apparent that the principles disclosed herein are also applicable to systems that generate radiation in other portions of the spectrum, e.g., the extreme ultraviolet (EUV) portion of the spectrum, i.e., light having wavelengths ranging from about 5 nm to 20 nm, as well as to systems that generate particle beams, such as ion beams or electron beams.

[0024]

[0033] 1, optical system 100 includes a system controller 104, an output device controller 102, and an illumination system 110 that generates a pulsed laser light beam 106. Light beam 106 may be directed to an output device, such as a stepper / scanner 105. Stepper / scanner 105 is a photolithography exposure tool that uses light beam 106 to pattern microelectronic features on a wafer 108. In a photolithography system, components within illumination system 110 (see FIG. 2) determine parameters of light beam 106, which in turn determine parameters of the microelectronic features patterned on wafer 108 by stepper / scanner 105.

[0025]

[0034] 2 is a functional block diagram illustrating an example configuration of illumination system 110. As shown in FIG. 2, illumination system 110 includes a gas discharge seed laser system 260. Seed laser system 260 is configured to generate seed laser output pulses from a master oscillator (“MO”) 263. MO 263 may be configured as a chamber having a pair of electrodes (not shown) such that an electrical discharge occurs between the electrodes to generate a laser gas discharge in a lasing gas, e.g., ArF, KrF, F2, and / or XeF, within the chamber, producing relatively broadband radiation.

[0026]

[0035] The resulting broadband radiation may be modified by a line narrowing module (LNM) 262 to select a relatively very narrow bandwidth center wavelength. LNM 262 may include a grating (not shown). A master oscillator output coupler ("MO OC") 264 receives radiation from MO 263. The output of MO OC 264 may be directed to a line center analysis module ("LAM") 266, which generates output 269.

[0027]

[0036] The output 269 propagates to relay optics 265. The relay optics 265 includes an MO wavefront engineering box ("WEB") 268, which may include a multi-prism beam expander (not shown) and an optical delay path (not shown). The WEB 268 can be used to redirect the output 269 of the seed laser system 260 to a power amplifier ("PA") stage 270. The PA may be configured as a powering amplifier, or PRA.

[0028]

[0037] PA stage 270 includes a beam reverser 272, a PA laser oscillation chamber 273, and a PA web 278. PA web 278 is arranged to receive the redirected output 269 of seed laser system 260 from MO web 268. PA web 278 may include a partially reflective input / output coupler (not shown), a maximum reflecting mirror for the nominal operating wavelength, and one or more prisms. PA web 278 may be provided with seed beam injection and output coupling optics (not shown) so that the beam is redirected by beam reverser 272 through a gain medium in PA laser oscillation chamber 273.

[0029]

[0038] The PA laser oscillation chamber 273 includes a chamber (not shown) having a pair of electrodes between which an electrical discharge can be generated to generate broadband radiation.

[0030]

[0039] The laser light beam pulse output from PA stage 270 is directed by PA WEB 278 to output subsystem 275, which measures and modifies parameters of the laser light beam pulse before generating final light beam 106. Output subsystem 275 includes a bandwidth analysis module ("BAM") 274 that receives the output from PA stage 270 and extracts a portion of the laser light beam pulse for metrology purposes, for example, to measure bandwidth or pulse energy. The laser light beam pulse then passes through an optical pulse stretcher ("OPuS") 276 within output subsystem 275, which modifies the light beam pulse.

[0031]

[0040] Components within OPuS 276 can be configured to convert a single output pulse into a pulse train. Secondary pulses created from the original single output pulse are delayed from one another to extend the effective pulse length of the laser and lower the peak pulse intensity. The light beam from OPuS 276 passes through a Complex Auto Shutter and Measurement Module ("CASMM") 277 or pulse energy meter within output subsystem 275 before light beam 106 is emitted from illumination system 110.

[0032]

[0041] As previously mentioned, such laser systems may require maintenance. Such maintenance may require an FSE to travel to the fab where the laser system is operating, take the laser system offline, and manually address the system, potentially resulting in a significant downtime penalty. This downtime penalty can be minimized by a maintenance system that monitors the performance of the laser system and initiates containment measures when one or more predetermined conditions are met. In some embodiments, the maintenance system can initiate containment measures sequentially until a successful containment measure is initiated. The order in which the maintenance system initiates containment measures may be according to predetermined parameters, such as, for example, invasiveness. Thus, the maintenance system may begin by initiating the least invasive containment measure and proceed to the next, most invasive containment measure only if the least invasive containment measure fails to restore acceptable operation.

[0033]

[0042] As used herein, "invasiveness" refers to the burden of performing containment measures, e.g., in terms of monetary cost and / or downtime. Containment measures range from less invasive operations such as conditioning burn-in (firing chamber-conditioning laser pulses) to more invasive operations such as refilling gas in the MO laser chamber and / or PA laser chamber.

[0034]

[0043] In some embodiments, the maintenance system is implemented as one or more machine learning models using wafer-level aggregate data to predict when the laser system is in a state requiring containment measures, e.g., imminent failure, and prevent failure before it occurs without requiring on-site maintenance personnel to call back to the system. If the maintenance system determines that the laser system is trending toward failure or is still in a failed state, the maintenance system is empowered to automatically trigger a series of potential solutions, ranging from firing a conditioning pulse to changing the laser configuration to requesting a gas refill, for example.

[0035]

[0044] The maintenance system continuously tracks performance and, if the first minimally invasive solution fails, selects among untried, more invasive potential responses, adopting a conservative policy that prioritizes tool availability and utilizes the most costly interventions only as a last resort.

[0036]

[0045] The maintenance system can instantly, independently, and predictively initiate containment when the laser system's performance is degrading toward a problem but is not yet sensitive enough to force downtime. Manual intervention is only required if all available automated solutions have failed to restore satisfactory operation. Even in the worst-case scenario where the maintenance system ultimately fails to circumvent the problem, the laser system can proceed with the callback as it currently does, with little or no penalty for attempting to circumvent the callback.

[0037]

[0046] In other words, according to one aspect of an embodiment, the maintenance system analyzes the state of the laser system and uses a machine learning model to direct the initiation of a series of increasingly more invasive potential solutions, whereby machine learning is used to determine whether to attempt a potential solution, and if so, which of the ranked series of potential solutions to try, continuing in order of rank until one of the solutions restores satisfactory operation or until all of the potential solutions have failed.

[0038]

[0047] In one embodiment, the maintenance system includes a trained classifier for identifying conditions requiring attention for the laser system. A classifier is a machine that classifies a given set of data into two or more classes or categories. The classifier can implement any of a number of classification methods, including logistic regression, naive Bayes, stochastic gradient descent, k-nearest neighbors, decision trees, random forests, artificial neural networks, and support vector machines. Classification can be binary, classifying data into one of two classes, or multi-class, classifying data into three or more possible classes. Classifiers must be trained. Training is the process of taking data known to belong to a particular class (true data or ground truth data) and creating a classifier based on that known data. Classification is the process of taking a classifier built on such a training dataset and running it on unknown data to determine the class membership of the unknown data.

[0039]

[0048] For example, one example of classification for a laser system is to classify the laser system as healthy or impaired. One way to determine the health of a laser system, i.e., whether the laser system is on the verge of starting to malfunction, is to examine so-called burst statistics. A system for obtaining burst statistics is disclosed, for example, in U.S. Patent Application Publication No. 2022 / 0365445, entitled "Burst Statistics Data Aggregation Filter," published November 17, 2022.

[0040]

[0049] All patent applications, patents, and publications cited herein are incorporated by reference in their entirety as part of this specification, except to the extent that they contain definitions, disclaimers or disclaimers of subject matter, and which contradict the explicit disclosure herein.

[0041]

[0050] FIG. 3 illustrates a maintenance system according to some embodiments, including a data collection and analysis system 320. The data collection and analysis system 320 is configured to receive data 330 from the laser source 110. Additionally or alternatively, the data collection and analysis system 320 can transmit data and / or control information 340 back to the laser source 110. The data 330 may include, for example, measurements of beam energy, center wavelength, and bandwidth. The data 330 may also include the repetition rate of pulses of pulsed laser radiation from the laser source 110. These pulses may be emitted in bursts, as in the example of FIG. 3 . This data is therefore referred to as burst statistics data, or simply burst statistics.

[0042]

[0051] The raw burst statistics data represent aggregated pulse-by-pulse, wafer-level data (e.g., streaming data) received from the laser source 110 by the data acquisition and analysis system 320 and associated with the laser source 110. These data may be accumulated as minimum, maximum, and average values.

[0043]

[0052] In some examples, data collection and analysis system 320 can be co-located with laser source 110. Additionally or alternatively, data collection and analysis system 320 can be located at a central location that receives data from one or more laser sources. In some examples, data collection and analysis system 320 can include components that are co-located with laser source 110 and / or distributed across different geographic locations and communicate with each other using one or more networks.

[0044]

[0053] 3 also includes a classifier 350. The classifier 350 is configured to receive data 360 from the data acquisition and analysis system 320 and generate an output 370 indicating whether the laser source 110 is healthy. As used herein, "healthy" refers to a state in which the laser system is operating normally and is not failing or on the verge of failing. "Failure" or "negative" refers to a state in which the laser system has failed or is on the verge of failing and containment measures must be taken to restore the laser system to a healthy state.

[0045]

[0054] The output 370 is provided to a containment action selection module 380, which determines the automatic containment action to take. As mentioned above, according to one aspect of an embodiment, the selected automatic containment action can be based on factors such as the degree of invasiveness, with the least invasive containment action being selected first, and if that is unsuccessful, successively more invasive automatic containment actions are taken until one is successful. If none of the containment actions are successful, the maintenance system may request that a non-automatic containment action be taken manually.

[0046]

[0055] The classifier 350 is trained before it is used. One way to train the classifier 350 is to use an ensemble of burst statistics 400, as shown in Figure 4. The ensemble of burst statistics may be made more tractable by known techniques such as dimensionality reduction and feature extraction by module 410. Dimensionality reduction may be performed using any of the known techniques, including Functional Principal Component Analysis (FPCA).

[0047]

[0056] A collection of ground truth data 420 for the burst statistics is also provided. This collection may be provided, for example, by expert analysis of a subset of the burst statistics. Known techniques such as bootstrapping (sampling with replacement) may be used to increase the size of the ground truth data.

[0048]

[0057] FIG. 5 is a flowchart illustrating a method for using a classifier to identify conditions in a laser system that require containment. In step S10, true data is provided. This step may be performed, for example, by manual classification of a subset of the data set performed by an expert. Next, in step S20, a target data set, such as burst statistics, is reduced, for example, by dimensionality reduction or feature selection. Next, in step S30, a classifier is trained using the reduced data set and the manually classified true data. Next, in step S40, the trained classifier is used to identify conditions in the laser system that require containment.

[0049]

[0058] As previously described, according to one aspect of the embodiment, if it is determined that the laser system requires containment, the automatic containment action to be taken can be selected, starting with the least invasive automatic containment action and proceeding in order of increasing invasiveness, until an effective automatic containment action is taken or it is determined that the available automatic containment actions are unable to restore the system to healthy operation. For example, FIG. 6A illustrates an array 390 of N automatic containment actions, where N is a positive integer. A pointer 395 in FIG. 6A is located at automatic containment action 1, which may be, for example, the least invasive automatic containment action. For example, this may be a chamber burn-in action by firing pulses to condition the chamber. In FIG. 6B, it is assumed that automatic containment action 1 was ineffective in restoring healthy operation of the laser system. The maintenance system then attempts automatic containment action 2, the next least invasive automatic containment action among the N automatic containment actions. The shading in the box for Automatic Containment Measure 1 indicates that this measure was attempted but was unsuccessful and will not be attempted again in this example. Automatic containment measures may be repeated if it is known that such measures may be more effective after more invasive containment measures have been implemented.

[0050]

[0059] In Figure 6C, it is assumed that all automatic containment measures have been taken up to the most invasive containment measure on the list, automatic containment measure N. This automatic containment measure N may correspond to, for example, time-consuming and costly replenishment. In Figure 6D, if it is determined that automatic containment measure N was ineffective, it is determined that no effective automatic containment options are available.

[0051]

[0060] FIG. 7 is a flowchart outlining a procedure for identifying automatic containment actions, if any, to be taken, according to one aspect of an embodiment. In step S110, the health of the laser system is monitored. In this context, "laser health" refers to whether the laser is operating beyond or on the verge of exceeding acceptable performance limits. In step S120, it is determined whether some form of containment action is necessary based on the determined laser health. If containment action is not necessary, the maintenance system continues to continuously monitor the laser health in step S110. However, if containment action is necessary, the maintenance system performs a first automatic containment action in step S130. As previously mentioned, this first automatic containment action may be selected from among multiple automatic containment actions ranked according to some parameter, such as invasiveness.

[0052]

[0061] After the first automatic containment action is taken, in step S140, the laser health is rechecked. Next, in step S150, it is determined whether containment action is still required. If containment action is not still required, in step S110, the process returns to monitoring the laser health. If containment action is still required, i.e., if the first automatic containment action was ineffective, in step S160, a second automatic containment action is taken. In this example, this is the next least invasive containment action, such as adjusting the chamber pressure.

[0053]

[0062] The process continues until step S170, where it is determined whether a final automated containment action is required. Next, in step S180, the last available automated containment action is taken, which may be the most invasive containment action, such as refilling the laser chamber with gas. Next, in step S190, the laser health is rechecked. If, in step S200, it is determined that containment is still required, it is determined that the available automated containment action cannot restore the laser health, and a maintenance action, such as a callback, is initiated in step S210. On the other hand, if, in step S200, it is determined that containment is no longer required, the process returns to step S110, where continuous monitoring of the laser health is resumed.

[0054]

[0063] As previously mentioned, an example of an automatic, relatively non-invasive containment measure is firing a null pulse into the chamber to condition or burn in the chamber. Here, a "null pulse" refers to a pulse generated for purposes other than substrate patterning. An example of an automatic, relatively non-invasive containment measure is gas refilling. In typical laser systems, the gas mixture is replaced using a procedure called a refill procedure, which involves removing and replacing the gas mixture in the chamber. Refill procedures often consume (waste) expensive gases, such as neon, that make up the gas mixture. Furthermore, the laser system cannot operate during a refill procedure. For at least these reasons, it is desirable to limit the number of refill procedures. Furthermore, it is desirable to perform a refill procedure only when other, less invasive procedures have failed, i.e., when necessary.

[0055]

[0064] In some applications, it may be desirable to perform a refill procedure only if the condition to be alleviated by the refill procedure is known to be correctable by refilling. For example, a performance condition may include "MO rollover," which refers to the transition of an MO into an operating region where an increase in voltage results in energy loss. This is considered a chronic inefficiency of an MO gas discharge chamber. A MO rollover condition is typically considered to be remedied by refilling. Therefore, in some applications, it may be beneficial to train a maintenance system to not only identify a binary chamber as healthy or failed, but also identify the performance condition that is causing the chamber to fail and select containment measures based on that determination.

[0056]

[0065] Besides a full refill, other automatic containment measures could include adjusting the gas composition, pressure, temperature, or blower speed of the MO or PA chamber. These automatic containment measures would constitute a more invasive form of containment than the delivery of a burn impulse, but less invasive than a full refill.

[0057]

[0066] For example, one hierarchy of automated containment measures may be as follows, in order of increasing invasiveness: (1) Emitting a control pulse (burn-in) (2) Adjusting the gas pressure (3) Adjusting the gas temperature (4) Rich refill (adding fluorine gas to the gas mixture in the MO or PA chamber), and (5) Complete refilling of either or both of the MO and PA chambers.

[0058]

[0067] It will be appreciated that the order and the means employed may vary depending on the characteristics of a given laser system.

[0059]

[0068] The above describes an example of a maintenance system in which burst statistics are used to train a classifier, which is then fed back to the trained classifier to identify faulty laser systems. However, burst statistics have attributes that can practically limit their usefulness in certain situations. One such attribute is the large number of burst statistics available. For example, a fab may send burst statistics to a laser manufacturer for the purpose of monitoring the operation of a laser system for many bursts, each consisting of many pulses. This volume of data can be unmanageable. Furthermore, because burst statistics are obtained from production operations at the fab, they are subject to factors known only to the fab operator. For example, a fab may use process recipes that affect burst statistics, but this may not be known to the laser system provider. For these reasons, in some embodiments, it may be beneficial to use a different type of dataset as a first dataset to combine with a dataset consisting of burst statistics as a second dataset.

[0060]

[0069] For example, another type of data available about the state of a laser system is data obtained from one or more scans performed after an automatic gas optimization (AGO) procedure. In an AGO procedure, the chamber is overfilled and then used to generate pulses. The pressure in the chamber is then reduced until the chamber operates satisfactorily. Scans are then performed for each of one or more operating parameters.

[0061]

[0070] One such operating parameter is laser gain, a measurement of a laser system that indicates the amount of energy produced by the laser as a function of the voltage applied between the electrodes in the laser chamber. This relationship is typically expressed as the derivative of voltage with respect to energy, dv / de, or its reciprocal, de / dv. Once the AGO procedure is complete, the FSE can obtain a characteristic AGO gain curve by applying a sequence of (scanning) electrode voltages and measuring the laser output energy. These AGO gain curves can be used to determine whether the system is exhibiting poor or degraded performance.

[0062]

[0071] Another type of scan data is data obtained from a scan performed after AGO that relates to the time difference (Δt) between (1) when a trigger signal is applied to the MO chamber and (2) when a trigger signal is applied to the PA chamber. This value is known as dtMOPA or ΔtMOPA. By scanning various values ​​of dtMOPA and monitoring the laser system's performance, such as bandwidth and center wavelength, as a function of dtMOPA, engineers can obtain AGO dtMOPA curves. These curves can also be correlated to systems that are in the process of failing or are close to failing, i.e., failing.

[0063]

[0072] The advantage of using the data resulting from these scans (hereafter referred to as AGO data or AGO curve data) is that it is generated under known conditions and can therefore be reliably correlated with the health of the laser system. However, the disadvantage of using these AGO data is that AGO scans are generally performed only after a refill and therefore are not as voluminous as data obtained from burst statistics.

[0064]

[0073] According to one aspect of the embodiment, these AGO data sets and burst statistics data sets are used together to mutually exploit the benefits of the reliability of the AGO data against the benefits of the more easily obtainable burst statistics.

[0065]

[0074] More specifically, in some embodiments, the AGO data is classified according to the state (healthy or unhealthy) of the laser system when the AGO data was acquired. The classified AGO data is then used as true data to train a burst statistics classifier that classifies burst statistics according to the state of the laser system. The training burst statistics are correlated with the AGO data. As an example, the burst statistics used to train the burst statistics classifier may be burst statistics acquired within a certain time interval after acquisition of the correlated AGO data. This results in a burst statistics classifier that can classify the state of the laser system based on the input burst statistics.

[0066]

[0075] The preliminary classification of the AGO data used to train the burst statistics classifier can theoretically be obtained by one or more individuals ("experts") with in-depth knowledge of the laser system's behavior associating various instances of burst statistics with what they determine to be the state of the laser system at the time the burst statistics were obtained. This process is sometimes referred to as "manual classification by experts." Manual classification is time-consuming and labor-intensive, and there is a practical limit to the amount of data that an expert can manually classify. It is generally desirable to have more AGO true data than can actually be manually classified. Therefore, in some embodiments, the AGO data used as true data for training the burst statistics classifier is obtained using a preliminary classifier. The preliminary classifier uses a set of manually classified AGO data as true data and classifies the input AGO data according to the state of the laser system. This results in a larger set of classified AGO data that can be used as true data for the burst statistics classifier.

[0067]

[0076] In some embodiments, the AGO data can be subjected to FPCA analysis to reduce the dimensionality of the AGO data. FPCA takes the AGO data and outputs coefficients for a set of orthonormal basis functions (eigenfunctions) that encompass all variations in the data set. The reduced subset of AGO data can then be manually analyzed by an expert, who correlates the AGO data based on the state of the equipment, i.e., the estimated system state. In one embodiment, the process is binary, with the expert classifying the AGO data based on whether it corresponds to a healthy or unhealthy equipment state. However, it will be appreciated that this AGO data can also be analyzed more precisely based not only on whether the laser system is healthy or unhealthy, but also on one or more specific performance issues that are causing the unhealthy state.

[0068]

[0077] The larger set of mechanically classified AGO data is then used as true data to train a second classifier, such as a support vector classifier (SVC) with a radial basis function (RBF) kernel. SVC represents the training data as points in space and builds a model that separates the points as widely spaced as possible, given that a given training example is labeled as belonging to one of two categories. New points are then added to the space by predicting which category and space they belong to, thereby assigning the new example to one of the categories. In this way, the second classifier classifies burst statistics based on whether or not they require containment. Burst statistics may be reduced, for example, using feature selection using known techniques.

[0069]

[0078] The trained second classifier can then predict the state of the laser system (determined by the correlation AGO) and classify the burst statistics into two cases. The first case is a healthy laser system that is operating within specifications and has no indication of imminent danger of going out of specifications within a certain time period in the future. In this case, no containment action is required. The other case is when the laser system is out of specifications or is likely to go out of specifications within a certain time period in the future, i.e., is faulty. In this case, no containment action is required. Once this classification is obtained, the maintenance system can continuously attempt automatic containment actions as described above.

[0070]

[0079] Such an automatic containment maintenance process is illustrated in the functional block diagrams of Figures 8A to 8C. In Figure 8A, an initial dataset 800, which may be, for example, AGO curve data, is obtained. This initial dataset 800 is provided to a dimensionality reduction module 810, which reduces the dimensionality of the first dataset 800, for example, by performing an FCPA analysis and deriving coefficients of eigenfunctions of the first dataset. A portion of the reduced initial dataset 800 is manually classified to generate manually classified data 820, which is provided as true data to a first classifier 830. Another portion of the initial dataset 800 is used as input to the first classifier 830, which classifies data input from the reduced initial dataset 800, to generate a classified initial dataset 840 (Figure 8B).

[0071]

[0080] In this manner, the arrangement of Figure 8A is used to obtain an initial classified data set 840 as shown in Figure 8B. The initial classified data set 840 is then used to train a second classifier 870 on a second data set 850, e.g., burst statistics data. The size of the second data set 850 may also be reduced by standard techniques, such as feature selection in a feature selection module 860, as shown.

[0072]

[0081] 8C, a trained second classifier 870 is then provided with operational burst statistics 880, such as information regarding one or more burst parameters, such as gain. The second classifier 870 then provides an indication of the status of the laser system based on the provided data. This laser system status information is provided to a containment selection module 900, which determines which automatic containment action to take based on the information from the second classifier 870.

[0073]

[0082] FIG. 9 is a flowchart outlining an example of a process that can be implemented on the above-described system. In step S310, an AGO classifier is trained using, for example, manually classified AGO curve data as true data. Next, in step S320, the trained AGO classifier is operated to obtain a classification of the AGO curve according to the health of the laser system. Next, in step S330, a second classifier, for example, a burst statistics classifier, is trained using burst statistics acquired around (e.g., temporally correlated with) the associated AGO data as input and the state of the laser system (healthy or failing) as output. The ground truth data is a collection of classified AGO data curves generated by the AGO classifier. Next, in step S340, the burst statistics classifier is used in real time to classify the burst statistics to determine whether the laser system is in a failing state, i.e., failed or on the verge of failure.

[0074]

[0083] 10, various embodiments and components thereof may be implemented using one or more known computer systems, such as the exemplary embodiments, systems, and / or devices shown in the figures or otherwise described. Computer system 1200 may be any known computer capable of performing the functions described herein.

[0075]

[0084] Computer system 1200 includes one or more processors (also called central processing units or CPUs), such as processor 1210. Processor 1210 is connected to a communication infrastructure or bus 1220.

[0076]

[0085] One or more of the processors 1210 may each be a graphics processing unit (GPU). In one embodiment, a GPU is a processor that is a specialized electronic circuit designed to handle mathematically intensive applications. A GPU may have a parallel architecture that is efficient for parallel processing of large blocks of data, such as the mathematically intensive large data common in computer graphics applications, images, video, etc.

[0077]

[0086] The computer system 1200 also includes user input / output devices 1230 such as a monitor, keyboard, pointing device, etc., and communicates with the communications infrastructure 1220 via a user input / output interface 1240 .

[0078]

[0087] The computer system 1200 also includes a primary memory 1250, such as random access memory (RAM), which may include one or more levels of cache. The primary memory 1250 stores control logic (i.e., computer software) and / or data.

[0079]

[0088] Computer system 1200 may also include one or more secondary storage devices or memories 1260. Secondary memory 1260 may include, for example, a hard disk drive 1280 and / or a removable storage device or drive 1290. Removable storage drive 1290 may be a floppy disk drive, a magnetic tape drive, a compact disk drive, an optical storage device, a tape backup device, and / or other storage device / drive.

[0080]

[0089] The removable storage drive 1290 may interface with a removable storage unit 1300. The removable storage unit 1300 comprises a computer usable or readable storage device on which computer software (control logic) and / or data is stored. The removable storage unit 1300 may be a floppy disk, magnetic tape, compact disk, DVD, optical storage disk, and / or other computer data storage device. The removable storage drive 1290 reads from and / or writes to the removable storage unit 1300 in known manner.

[0081]

[0090] According to one example embodiment, secondary memory 1260 may include other means, devices, or approaches for making computer programs and / or other instructions and / or data accessible by computer system 1200. Such means, devices, or approaches may include, for example, removable storage unit 1310. Examples of removable storage unit 1310 include program cartridges and cartridge interfaces (such as those found in video game devices), removable memory chips (such as EPROMs or PROMs) and associated sockets, memory sticks and USB ports, memory cards and associated memory card slots, and / or other removable storage units and associated interfaces.

[0082]

[0091] Computer system 1200 may further include a communications interface or network interface 1320. Communications interface 1320 enables computer system 1200 to communicate and interact with any combination of remote devices, remote networks, remote entities, etc. (each singly or collectively referred to as 1330). For example, communications interface 1320 may enable computer system 1200 to communicate with remote devices 1330 via communications paths 1340, which may be wired and / or wireless and may include any combination of a LAN, a WAN, and / or the Internet. Control logic and / or data may be transmitted to and from computer system 1200 via communications paths 1340.

[0083]

[0092] In one embodiment, a non-transitory, tangible apparatus or article of manufacture having control logic (software) stored on a non-transitory, tangible, computer-usable or readable medium is also referred to herein as a computer program product or program storage device, including, but not limited to, a tangible article of manufacture embodying computer system 1200, main memory 1250, secondary memory 1260, removable storage units 1290 and 1300, and any combination thereof. Such control logic, when executed by one or more data processing devices (e.g., computer system 1200), causes such data processing devices to operate as described herein.

[0084]

[0093] Based on the teachings contained herein, it will be apparent to one skilled in the relevant art how to make and use embodiments of the present disclosure with data processing devices, computer systems, and / or computer architectures other than those shown in Figure 10. In particular, embodiments may operate with software, hardware, and / or operating system implementations other than those described herein.

[0085]

[0094] Although specific reference has been made above to the use of embodiments in the context of optical lithography, it will be appreciated that embodiments may also be used in other applications, for example in imprint lithography, and are not limited to optical lithography where the context allows.

[0086]

[0095] It is to be understood that the terms or phrases used herein are for purposes of description and not of limitation, and that they should be interpreted by one skilled in the relevant art based on the teachings contained herein.

[0087]

[0096] It should be understood that the Detailed Description section is used to interpret the claims. The Summary and Abstract sections describe one or more exemplary embodiments, but are not intended to be exhaustive and therefore not to limit the scope of the embodiments and appended claims in any way.

[0088]

[0097] The embodiments have been described by showing functional building blocks for implementing specific functions and their relationships. The boundaries of these functional building blocks have been arbitrarily defined herein for the convenience of explanation. Alternative boundaries may be defined as long as the specific functions are properly performed.

[0089]

[0098] The foregoing description of the specific embodiments sufficiently clarifies the general nature of the present embodiments so that those skilled in the art can, by applying their technical knowledge, readily modify and / or adapt such specific embodiments to various applications without undue experimentation and without departing from the general concept of the present embodiments. Accordingly, such modifications and variations are intended to be within the meaning and range of equivalents of the embodiments provided herein, based on the teaching and guidance set forth herein.

[0090]

[0099] The above description includes examples of multiple embodiments. It is, of course, not possible to describe every conceivable combination of components or methodologies for purposes of describing these embodiments; however, those skilled in the art will recognize that additional combinations and variations of the various embodiments are possible. Accordingly, the described embodiments are intended to embrace all alterations, modifications, and variations that fall within the spirit and scope of the appended claims. Furthermore, even if elements of the described aspects and / or embodiments are described or claimed in the singular, the plural is contemplated unless limitation to the singular is expressly stated. Furthermore, all or a portion of any aspect and / or embodiment can be utilized in combination with all or a portion of any other aspect and / or embodiment, unless otherwise specified.

[0091]

[0100] The embodiments can be further described using the following clauses. 1. Using a classifier to determine whether a negative laser condition exists that can be corrected by automated containment measures; selecting a least invasive automatic containment measure from the group of automatic containment measures as a first automatic containment measure; Implementing the first automatic containment measure; determining, using the classifier, whether the negative laser condition has been corrected by taking a first automatic containment action; selecting a next least invasive containment action from the group of automatic containment actions as the next automatic containment action if the classifier determines that the first automatic containment action did not correct the negative laser condition; and Taking the following automatic containment measures: 1. A method for maintaining a laser system, comprising: 2. The method of clause 1, wherein the group of automatic containment actions comprises N automatic containment actions ranked according to invasiveness, and the method further comprises sequentially performing one of the N automatic containment actions until the classifier determines that either one of the N automatic containment actions corrected the negative laser condition or determines that none of the N automatic containment actions corrected the negative laser condition. 3. Training a first classifier using the first data type; classifying first data of a first data type using a first classifier; training a second classifier using the first data to classify second data of a second data type; classifying the second data using a second classifier to determine whether to take automatic containment action; selecting an automatic containment action to be taken if the second classifier determines that an automatic containment action should be taken; and implementing automatic containment measures; 1. A method for maintaining a laser system, comprising: 4. The method of clause 3, wherein the first type of data comprises data obtained from a gain scan performed by increasing the magnitude of a voltage applied to an electrode in a discharge chamber of the laser system and determining the laser output energy as a function of the magnitude of the voltage. 5. The method of clause 3, wherein the first data type comprises data obtained from a timing scan performed by sequentially increasing a time gap between applying a main oscillator trigger voltage to a main oscillator chamber of the laser system and applying a power amplifier trigger voltage to a power amplifier chamber of the laser system and determining a variation in an operating parameter as a function of the duration of the time gap. 6. The method of clause 3, wherein the second data type comprises burst statistical data. 7. The method of clause 3, wherein selecting an automatic containment measure includes selecting, from a group of automatic containment measures, the least invasive automatic containment measure that has not already been taken. 8. Classifying the first data type using a first classifier to obtain first type classified data; training a second classifier that uses the first type of classification data to classify a second data type; using a second classifier to determine whether to take an automated containment action based on the second classifier classifying the second data type; selecting an automatic containment action to be taken if the second classifier determines that an automatic containment action should be taken; and implementing automatic containment measures; 1. A method for maintaining a laser system, comprising: 9. The method of clause 8, wherein the first type of data comprises data obtained from a gain scan performed by sequentially increasing the magnitude of a voltage applied to electrodes in a discharge chamber of the laser system and determining the laser output energy as a function of the magnitude of the voltage. 10. The method of clause 8, wherein the first data type comprises data obtained from a timing scan performed by sequentially increasing a time gap between applying a main oscillator trigger voltage to a main oscillator chamber of the laser system and applying a power amplifier trigger voltage to a power amplifier chamber of the laser system and determining a variation in an operating parameter as a function of the duration of the time gap. 11. The method of clause 8, wherein the second data type comprises burst statistical data. 12. The method described in clause 8, wherein selecting an automatic containment measure includes selecting, from a group of automatic containment measures, the least invasive automatic containment measure that has not already been taken. 13. A classifier adapted to determine whether a negative laser condition exists that can be corrected by automatic containment measures; an automatic containment measure selection unit adapted to select a least invasive automatic containment measure from the plurality of automatic containment measures as a first automatic containment measure; a first automatic containment action execution unit adapted to cause the laser system to perform a first automatic containment action; The classifier is further adapted to determine whether the negative laser condition has been corrected by performing the first automatic containment action, and if it is determined that the negative laser condition has not been corrected by performing the first automatic containment action, select a more invasive automatic containment action from the plurality of automatic containment actions as a next automatic containment action; and if the classifier determines that the negative laser condition is not corrected by taking the first automatic containment action, causing the laser system to take a second automatic containment action. Laser system maintenance equipment. 14. The apparatus of clause 13, wherein the plurality of automatic containment measures comprises N automatic containment measures ranked according to invasiveness, and further comprising a unit for respectively performing one of the N automatic containment measures until the classifier determines that either one of the N automatic containment measures corrected the negative laser condition or that none of the N automatic containment measures corrected the negative laser condition. 15. A first classifier trained using a first data type; a second classifier trained using the data classified by the first classifier and a second data type, and adapted to determine whether to take automatic containment action; a selector for selecting an automatic containment action to be taken if the second classifier determines that an automatic containment action should be taken; A maintenance device for a laser system comprising: 16. The apparatus described in clause 15, wherein the first data type comprises data obtained from a gain scan performed by increasing the magnitude of a voltage applied to an electrode in a discharge chamber of the laser system and determining the laser output energy as a function of the magnitude of the voltage. 17. The apparatus of clause 15, wherein the first data type comprises data obtained from a timing scan performed by sequentially increasing a time gap between applying a main oscillator trigger voltage to a main oscillator chamber of the laser system and applying a power amplifier trigger voltage to a power amplifier chamber of the laser system and determining a variation in an operating parameter as a function of the duration of the time gap. 18. The apparatus of clause 15, wherein the second data type comprises burst statistical data. 19. The method of clause 15, wherein the selector is adapted to select the least invasive automated containment measure that has not yet been taken. 20. When executed by a processor, causes the processor to: using a classifier to determine whether a negative laser condition exists that can be corrected by automated containment measures; selecting a least invasive automatic containment measure from the plurality of automatic containment measures as a first automatic containment measure; Implementing the first automatic containment measure; determining, using the classifier, whether the negative laser condition has been corrected by taking a first automatic containment action; selecting a next least invasive containment action from the plurality of automatic containment actions as a next automatic containment action if the classifier determines that the first automatic containment action has not corrected the negative laser condition; and Taking the following automatic containment measures: A non-transitory computer-readable medium having stored thereon instructions for performing operations including:

[0092]

[0101] These and other embodiments are within the scope of the following claims.

Claims

1. using a classifier to determine whether a negative laser condition exists that can be corrected by automated containment measures; selecting a least invasive automatic containment measure from the group of automatic containment measures as a first automatic containment measure; performing said first automatic containment measure; using the classifier to determine whether the negative laser condition was corrected by taking the first automatic containment action; selecting a next least invasive containment action from the group of containment actions as a next containment action if the classifier determines that the first containment action has not corrected the negative laser condition; and take the following automatic containment measures:

1. A method for maintaining a laser system, comprising:

2. 2. The method of claim 1, wherein the group of automatic containment actions comprises N automatic containment actions ranked according to invasiveness, the method further comprising sequentially taking one of the N automatic containment actions until the classifier determines that either one of the N automatic containment actions corrected the negative laser condition or that none of the N automatic containment actions corrected the negative laser condition.

3. training a first classifier using a first data type; classifying first data of the first data type using the first classifier; training a second classifier using the first data to classify second data of a second data type; classifying the second data using the second classifier to determine whether to take automatic containment action; selecting an automatic containment action to be taken if the second classifier determines that an automatic containment action should be taken; and performing said automatic containment measures; 1. A method for maintaining a laser system, comprising:

4. 4. The method of claim 3, wherein the first data type comprises data obtained from a gain scan performed by increasing the magnitude of a voltage applied to an electrode in a discharge chamber of the laser system and determining the laser output energy as a function of the magnitude of the voltage.

5. 4. The method of claim 3, wherein the first data type comprises data obtained from a timing scan performed by sequentially increasing a time gap between applying a master oscillator trigger voltage to a master oscillator chamber of the laser system and applying a power amplifier trigger voltage to a power amplifier chamber of the laser system and determining a variation in an operating parameter as a function of the duration of the time gap.

6. The method of claim 3 , wherein the second data type comprises burst statistics data.

7. The method of claim 3 , wherein selecting an automatic containment action includes selecting the least invasive automatic containment action that has not yet been taken from a group of automatic containment actions.

8. classifying the first data type using a first classifier to obtain first type classified data; training a second classifier using the first type of classified data to classify a second data type; using the second classifier to determine whether to take automatic containment action based on the second classifier's classification of the second data type; selecting an automatic containment action to be taken if the second classifier determines that an automatic containment action should be taken; and performing said automatic containment measures; 1. A method for maintaining a laser system, comprising:

9. 9. The method of claim 8, wherein the first type of data comprises data obtained from a gain scan performed by sequentially increasing the magnitude of a voltage applied to an electrode in a discharge chamber of the laser system and determining the laser output energy as a function of the magnitude of the voltage.

10. 10. The method of claim 8, wherein the first data type comprises data obtained from a timing scan performed by sequentially increasing a time gap between applying a master oscillator trigger voltage to a master oscillator chamber of the laser system and applying a power amplifier trigger voltage to a power amplifier chamber of the laser system and determining a variation in an operating parameter as a function of the duration of the time gap.

11. The method of claim 8 , wherein the second data type comprises burst statistical data.

12. The method of claim 8 , wherein selecting an automatic containment action includes selecting the least invasive automatic containment action that has not yet been taken from a group of automatic containment actions.

13. a classifier adapted to determine whether a negative laser condition exists that can be corrected by automatic containment measures; an automatic containment measure selection unit adapted to select a least invasive automatic containment measure from the plurality of automatic containment measures as a first automatic containment measure; a first automatic containment action execution unit adapted to cause the laser system to perform the first automatic containment action; The classifier is further adapted to determine whether the negative laser state has been corrected by performing the first automatic containment action, and if it is determined that the negative laser state has not been corrected by performing the first automatic containment action, select a more invasive automatic containment action from the plurality of automatic containment actions as a next automatic containment action; if the classifier determines that the negative laser condition is not corrected by taking the first automatic containment action, causing the laser system to take a second automatic containment action. Laser system maintenance equipment.

14. 14. The apparatus of claim 13, wherein the plurality of automatic containment actions comprises N automatic containment actions ranked according to invasiveness, and further comprising: a unit for respectively taking one of the N automatic containment actions until the classifier determines that either one of the N automatic containment actions corrected the negative laser condition or that none of the N automatic containment actions corrected the negative laser condition.

15. a first classifier trained using a first data type; a second classifier trained using the data classified by the first classifier and a second data type, and adapted to determine whether to take automatic containment action; and a selector for selecting an automatic containment action to be taken if the second classifier determines that an automatic containment action should be taken; A maintenance device for a laser system comprising:

16. 16. The apparatus of claim 15, wherein the first data type comprises data obtained from a gain scan performed by increasing the magnitude of a voltage applied to an electrode in a discharge chamber of the laser system and determining the laser output energy as a function of the magnitude of the voltage.

17. 16. The apparatus of claim 15, wherein the first data type comprises data obtained from a timing scan performed by sequentially increasing a time gap between applying a master oscillator trigger voltage to a master oscillator chamber of the laser system and applying a power amplifier trigger voltage to a power amplifier chamber of the laser system and determining a variation in an operating parameter as a function of the duration of the time gap.

18. 16. The apparatus of claim 15, wherein the second data type comprises burst statistical data.

19. The method of claim 15 , wherein the selector is adapted to select the least invasive automatic containment measure that has not yet been taken.

20. When executed by a processor, the processor: using a classifier to determine whether a negative laser condition exists that can be corrected by automated containment measures; selecting a least invasive automatic containment measure from the plurality of automatic containment measures as a first automatic containment measure; performing said first automatic containment measure; using the classifier to determine whether the negative laser condition was corrected by taking the first automatic containment action; selecting a next least invasive containment action from the plurality of automatic containment actions as a next automatic containment action if the classifier determines that the first automatic containment action has not corrected the negative laser condition; and take the following automatic containment measures: A non-transitory computer-readable medium having stored thereon instructions for performing operations including:

Images