Process and system for maintenance of a light source

Abstract

A system and a maintenance process of a light source includes (a) collecting a first set of parameters of a first apparatus of the light source; (b) analyzing the first set of parameters to calculate a first score of a likelihood of a failure mode of the first apparatus; and (c) in response to the first score exceeding a first pre-determined value, (1) generating one or more first action plans for maintenance and / or replacement of the first apparatus, (2) providing a notification of the one or more first action plans, and (3) storing the first set of parameters, the first score, and the one or more first action plans in a storage medium.

Description

PROCESS AND SYSTEM FOR MAINTENANCE OF A LIGHT SOURCECROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to US Application No. 63 / 738,056, filed December 23, 2024, titled PROCESS AND SYSTEM FOR MAINTENANCE OF DUV LIGHT SOURCE, which is incorporated herein by reference in its entirety.FIELD

[0002] The disclosed subject matter relates to systems and processes for maintenance of light source systems, particularly to systems and processes for predictive maintenance and maintenance planning for DUV light source systems.BACKGROUND

[0003] Photolithography is a process by which semiconductor circuitry is patterned on a semiconductor substrate such as a silicon or other semiconductor wafer. During a semiconductor manufacturing process, an inspection or a metrology apparatus can scan a surface of a reticle and / or the surface of a semiconductor substrate to detect pattern defects or other anomalies that may impact yield. A deep ultraviolet (DUV) light source can provide light used to expose a photoresist on the wafer, measure a dimension of targets, or characterize physical properties of components. Often, a DUV light source is an excimer laser system, and the light is in the form of light pulses generated by the laser system. The light pulses are passed through a beam delivery unit and a reticle or a mask, and then projected onto a photoresist on a wafer. Multiple exposures are made to cover a desired area of the photoresist. In this way, a portion of a chip design is patterned into the photoresist, which is then developed, etched, and cleaned, and then used as a mask for implantation, deposition, etching, and / or other processes on or in the structure of the wafer. This process then repeats.

[0004] Excimer laser based DUV light sources generally include multiple apparatuses or modules. The performance of each module is important to the performance of the light source. When one or more modules require maintenance or replacement, efficiently conducting the maintenance and / or replacement is desirable to keep downtime to a minimum.SUMMARY

[0005] In some general aspects an automated predictive maintenance process of a deep ultraviolet (DUV) light source is provided, the process including: (a) automatically collecting a first set of parameters of a first apparatus of the DUV light source; (b) automatically analyzing the first set of parameters to calculate a first score of a likelihood of a failure mode of the first apparatus; and (c) automatically (1) generating one or more first action plans for maintenance and / or replacement of the first apparatus, (2) providing a notification of the one or more first action plans, and (3) storing thefirst set of parameters, the first score, and the one or more first action plans in a storage medium, when the first score exceeds a first pre-determined value.

[0006] Implementations can include one or more of the following.

[0007] The process can further include automatically repeating actions (a), (b), and (c) when the first score does not exceed the first pre -determined value. The process can further include automatically repeating actions (a), (b), and (c) at a predetermined frequency. The process can further include servicing the first apparatus according to one of more of the one or more first action plans or a revision or derivative thereof.

[0008] The process can further including: (d) automatically collecting a second set of parameters from a second apparatus of the DUV light source; (e) automatically analyzing the second set of parameters to calculate a second score of a likelihood of a failure mode of the second apparatus; and (f) automatically (4) generating one or more second action plans for maintenance and / or replacement of the second apparatus, and (5) triggering a notification of the one or more second action plans, and (6) storing the second set of parameters, the second score, and the one or more second action plans in a storage medium, when the second score exceeds a second pre-determined value. The process can further include repeating actions (d), (e), and (f) when the second score does not exceed the second pre -determined value. The process can further include automatically repeating actions (d), (e), and (f) at a predetermined frequency. The process can further include servicing the second apparatus according to one of more of the one or more second action plans or a revision or derivative thereof.

[0009] Automatically analyzing the first set of parameters to calculate a first score can include performing an evaluation to determine the probability of the first apparatus experiencing an operating fault within a predetermined second number of (upcoming) light pulses when the first apparatus has participated in producing at least a first number of light pulses. Automatically analyzing the first set of parameters to calculate a first score of a likelihood of a failure mode of the first apparatus can include performing an evaluation to determine the probability of the first apparatus experiencing an operating fault within a predetermined fourth number of (upcoming) light pulses when the first apparatus has remained in service and participated in producing at least a third number of light pulses since last being evaluated to determine the probably of an operating fault.

[0010] Automatically generating the one or more first action plans for maintenance and / or replacement of the first apparatus can include automatically generating the one or more first action plans using a large language model (LLM) with inputs, training inputs, and / or references including the first score of the likelihood of a failure mode and one or more of (1) stored light source and / or apparatus metadata, (2) stored maintenance and / or replacement cycle logic, and (3) a library of past scores of likelihoods of failure modes and associated action plans. Automatically generating the one or more first action plans for maintenance and / or replacement of the first apparatus can include automatically generating the one or more first action plans using an LLM trained on a library of pastlikelihood of failure mode predictions and associated action plans, with an input (prompt) including the first score of the likelihood of a failure mode.

[0011] The process can further include providing access for authorized users to the first set of parameters, the first score, and the one or more first action plans in the storage medium. The process can further include providing, for authorized users, change access to the one or more first action plans in the storage medium. The process can further include providing, for authorized users, review and approval access to the one or more first action plans in the storage medium.

[0012] In another aspect, an automated predictive maintenance process of a deep ultraviolet (DUV) light source is provided, the process including: (a) automatically collecting respective sets of parameters of two or more apparatuses of the DUV light source; (b) automatically analyzing the respective sets of parameters to calculate respective scores of a likelihood of a failure mode of the two or more apparatuses; (c) automatically (1) generating an action plan for maintenance and / or replacement of each respective apparatus of the two or more apparatuses whose respective score of a likelihood of a failure mode exceeds a respective predetermined value, (2) triggering a notification of the action plan, and (3) storing the action plan in a storage medium; and (d) automatically periodically repeating actions (a) through (c) at least until the action plan is generated.

[0013] Implementations can include one or more of the following.

[0014] The process can further include servicing the respective apparatus according to the action plan or a revision or derivative thereof. Automatically analyzing the respective sets of parameters to calculate respective scores can include performing respective automated evaluations to determine respective probabilities the two or more apparatuses experiencing a respective operating fault within a respective predetermined number of (upcoming) light pulses.

[0015] Automatically generating the respective action plan for maintenance and / or replacement of each respective apparatus can include using a large language model (LLM) based on data including each respective set of parameters and each respective score of the likelihood of a failure mode, and based on data and / or training including (1) stored light source and / or apparatus metadata, (2) stored maintenance and / or replacement cycle logic, and (3) a library of past scores of likelihoods of failure modes and associated action plans. Automatically generating the action plan for maintenance and / or replacement of each respective apparatus can include using an LLM trained on a library of past scores of likelihoods of failure modes and associated action plans and based on data (prompt(s)) including each respective set of parameters and each respective score of the likelihood of a failure mode.

[0016] According to an additional aspect, a computer-implemented process includes: (a) analyzing by a computing device respective sets of parameters of two or more apparatuses of a DUV light source to calculate respective scores of a likelihood of a failure mode of the two or more apparatuses; (b) by the computing device, generating or triggering the automatic generation an action plan for maintenance and / or replacement of each respective apparatus of the two or more apparatuses whose respective score of a likelihood of a failure mode exceeds a respective predetermined value, triggeringa notification of the action plan, and storing the action plan in a storage medium; and (c) repeating (a) and (b) at least until the action plan is generated by the computing device.

[0017] Implementations can include one or more of the following.

[0018] The automatic generation of the action plan for maintenance and / or replacement of each respective apparatus can include using a large language model (LLM) based on data including each respective set of parameters and each respective score of the likelihood of a failure mode, and based on data and / or training including (1) stored light source and / or apparatus metadata, (2) stored maintenance and / or replacement cycle logic, and (3) a library of past scores of likelihoods of failure modes and associated action plans.

[0019] In a further aspect, a non-transitory computer-readable storage medium includes executable instructions for causing a processor to perform operations, the instructions including instructions for performing the following operations: (a) analyzing respective sets of parameters of two or more apparatuses of a DUV light source to calculate respective scores of a likelihood of a failure mode of the two or more apparatuses; (b) generating an action plan for maintenance and / or replacement of each respective apparatus of the two or more apparatuses whose respective score of a likelihood of a failure mode exceeds a respective predetermined value, triggering a notification of the action plan, and storing the action plan in a storage medium; and (c) periodically repeating actions (a) and (b) at least until the action plan is generated.

[0020] Implementations can include one or more of the following.

[0021] Generating the action plan can include using a large language model (LLM) based on (using and prompt or input) data including each respective set of parameters and each respective score of the likelihood of a failure mode, and based on data and / or training including (1) stored light source and / or apparatus metadata, (2) stored maintenance and / or replacement cycle logic, and (3) a library of past scores of likelihoods of failure modes and associated action plans.

[0022] In yet another aspect, an automated maintenance process of a deep ultraviolet (DUV) light source is provided, the process including: (a) automatically collecting a set of parameters of the DUV light source; (b) automatically analyzing the set of parameters to calculate a score of a likelihood of a failure mode of a laser chamber and to detect one or more significant changes and / or out-of-range conditions in the performance of the light source or of apparatuses or components thereof; and (c) (1) automatically generating an action plan for maintenance and / or replacement of the laser chamber when the score exceeds a first pre-determined value, and for maintenance to correct the one or more significant changes and / or out-of-range conditions when present, (2) automatically providing a notification of the action plan, and (3) automatically storing the set of parameters, the score, the one or more significant changes and / or out-of-range conditions, and the action plan in a storage medium.

[0023] Implementations can include one or more of the following.

[0024] Automatically generating the action plan can include using a large language model (LLM) based on data (based on inputs or prompts) including the set of parameters and the score of thelikelihood of a failure mode and the one or more significant changes and / or out-of-range conditions, and based on data and / or training including (1) stored light source and / or apparatus metadata, (2) stored maintenance and / or replacement cycle logic, and (3) a library of past scores of the likelihood of a failure mode and past one or more significant changes and / or out-of-range conditions and associated action plans.

[0025] In still another aspect, a system for automated maintenance of a deep ultraviolet (DUV) light source includes: (a) a data collection system configured to automatically receive and collect a set of parameters of the DUV light source; (b) a failure prediction analyzer configured to automatically receive from the data collection system and analyze the set or sets of parameters to calculate a score of a likelihood of a failure mode of a laser chamber; and (c) an action plan generator configured to (1) automatically generate an action plan for maintenance and / or replacement of the laser chamber when the calculated score exceeds a first pre-determined value, (2) automatically provide a notification of the action plan, and (3) automatically store the set of parameters, the score, and the action plan in a storage medium.

[0026] Implementations can include one or more of the following.

[0027] The action plan generator can include a large language model (LLM) configured to generate an action plan based on data including the set of parameters and the score of the likelihood of the failure mode, and based on data and / or training including (1) stored light source and / or apparatus metadata, (2) stored maintenance and / or replacement cycle logic, and (3) a library of past scores of the likelihood of a failure mode and associated sets of parameters and associated action plans.

[0028] The system can further further include a performance analyzer configured to automatically receive the set of parameters from the data collection system and to detect one or more significant changes and / or out-of-range conditions in the performance of the light source or of apparatuses or components thereof, and the action plan generator can be further configured to generate the action plan including actions to correct the one or more significant changes and / or out-of-range conditions when present.

[0029] The details of one or more implementations are set forth in the accompanying drawings and the description below. Other features will be apparent from the description and drawings, and from the claims.DRAWING DESCRIPTION

[0030] FIG. 1 is a schematic diagram, not to scale, of an overall broad conception of an exposure system.

[0031] FIG. 2 is a schematic diagram, not to scale, of an overall broad conception of an excimer laser based DUV light source.

[0032] FIG. 3 is a flow diagram of aspects of a process for predictive maintenance and maintenance planning for a DUV light source.

[0033] FIG. 3 A is a flow diagram of a variation of the process of the flow diagram of FIG. 3.

[0034] FIG. 4 is a schematic diagram of elements of a system implementing a process such as the process of FIG. 3 or FIG. 3A.DETAILED DESCRIPTION

[0035] Various aspects and implementations are now described with reference to the drawings, wherein like reference numerals are used to refer to like elements throughout. When a particular feature, structure, or characteristic is described in connection with an implementation, it is understood that it is within the knowledge of one skilled in the art to effect such feature, structure, or characteristic in connection with other implementations whether or not explicitly described.

[0036] Some features or aspects of the present disclosure may be implemented in hardware, firmware, software, or any combination thereof. For example, implementations of the present disclosure may also be implemented as instructions stored on a machine -readable medium, which may be read and executed by one or more processors. A machine -readable medium may include any mechanism for storing or transmitting information in a form readable by a machine (e.g., a computing device). For example, a machine-readable medium may include read only memory (ROM); random access memory (RAM); magnetic disk storage media; optical storage media; flash memory devices; cloud-implemented storage, electrical, optical, acoustic, or other forms of propagated signals (e.g., carrier waves, infrared signals, digital signals, etc.), and others. Further, firmware, software, routines, instructions may be described herein as performing certain actions. However, it should be appreciated that such descriptions are merely for convenience and that such actions in fact result from computing devices, processors, controllers, or other devices executing the firmware, software, routines, instructions, and so forth.

[0037] FIG. 1 shows an exposure system 100 that includes a light source 102. The light source 102 produces a pulsed light beam 104 and directs it to an exposure apparatus 106, such as a scanner that patterns microelectronic and other features on a wafer 110, and a metrology tool that measures and analyzes critical dimensions on the wafer with high precision. The wafer 110 is placed on a wafer table 112 constructed to hold the wafer 110 and connected to a positioner 114 configured to accurately position the wafer 110 in accordance with certain parameters.

[0038] The pulsed light beam 104 has a wavelength in the DUV range, with a wavelength of 248 nanometers (nm) or 193 nm, for example. The exposure apparatus 106 includes an optical arrangement 108 having, for example, one or more condenser lenses, a mask, and an objective arrangement. The mask is movable along one or more directions, such as along an optical axis of the pulsed light beam 104 or in a plane that is perpendicular to the optical axis. The objective arrangement includes a projection lens and enables an image transfer to occur from the mask to photoresist on the wafer 110. The optical arrangement 108 helps adjust the range of angles for the pulsed light beam 104 impinging on the mask 110. The optical arrangement 108 also helps tohomogenize (make uniform) the intensity distribution of the pulsed light beam 104 across the mask 110.

[0039] The exposure apparatus 106 can include, among other features, a controller 116 that controls how layers are printed on the wafer 110. The controller 116 may include a memory that stores information such as process recipes that determine the parameters including a length of the exposure on the wafer 110 based on, for example, the mask used, as well as other factors that affect exposure. During lithography or metrology processes, a burst of pulses of the pulsed light beam 104 illuminates the same area of the wafer 110 to constitute an illumination dose.

[0040] The exposure system 100 also includes a control system 118. In general, the control system 118 includes one or more of digital electronic circuitry, computer hardware, firmware, and software. The control system 118 can be centralized or be partially or wholly distributed throughout the exposure system 100, and / or may be at least in part distributed even beyond any of the physical structure of the exposure system 100.

[0041] FIG. 2 shows a DUV light source 202 in the form of a DUV gas-discharge pulsed laser system. In some embodiments, the DUV light source 202 produces a pulsed beam 204 that can serve as the light beam 104 of FIG. 1. FIG. 2 shows a two-chamber laser system including a seed laser stage 220, such as a solid state or gas discharge master oscillator, a power amplification stage 240 such as a single-pass power amplifier (“PA”), a power ring amplifier (“PRA”), or a power oscillator (“PO”). Single chamber systems and systems with three or more chambers can also be used. The DUV light source 202 further includes relay optics 230, and an output subsystem 250.

[0042] The seed laser stage 220 may include, e.g., a master oscillator (“MO”) chamber 224 which includes a pair of electrodes 223 and 225. The seed laser stage 220 may also include a master oscillator output coupler (“OC”) 228, which may comprise a partially reflective mirror (not shown), that forms, together with a reflective grating (not shown) in a line narrowing module (“LNM”) 222, an oscillator (optical cavity) in which a beam oscillates to form a seed laser output pulse. The seed laser stage 220 may also include a first spectrum analysis module 226. The relay optics 230 may include an MO wavefront engineering box (“WEB”) 232 that may serve to redirect the output of the seed laser stage 220 toward the power amplification stage 240, and may include, a multi prism beam expander (not shown) and an optical delay path (not shown).

[0043] The power amplification stage 240 may include, for example, a power amplifier discharge chamber 244. The power amplifier discharge chamber 244 may include a pair of electrodes 243 and 245. The power amplifier discharge chamber 244 may be part of an oscillator. The oscillator may be formed or defined by (1) seed-beam injection and output coupling optics (not shown) that may be incorporated into a PRA WEB 248 and (2) a beam reverser (“BR”) 242. The PRA WEB 248 may incorporate a partially reflective input / output coupler (not shown) and a maximally reflective mirror for the nominal operating wavelength (e.g., at around 193 nm for an ArF system) and one or more prisms.

[0044] A second spectrum analysis module 246 may receive the light beam oscillating in the power amplification stage 240 and pick off a portion of the light beam for metrology purposes, e.g., to measure the output bandwidth and pulse energy. The laser light beam 249 of pulses then passes through the PRA WEB 248 to an optical pulse stretcher (“OPuS”) 252, and then to an autoshutter, both within the output subsystem 250. In the implementation of the output subsystem 250 shown in FIG. 2, the autoshutter is in the form of, or included within, a combined autoshutter metrology module (“CASMM”) 254, which may also include a pulse energy meter.

[0045] The power amplifier discharge chamber 244 and the MO discharge chamber 224 are configured as chambers in which electrical discharges between the electrodes create an inverted population of high energy molecules, including, e.g., Ar, Kr, F2, and / or Xe to produce a relatively broad-band light amplification potential. The wavelength(s) that are permitted to oscillate, and accordingly receive significant amplification, can be line-narrowed to a relatively very narrow bandwidth around a center wavelength selected by adjustments made in the LNM 222.

[0046] In the course of semiconductor device production using a light source such as DUV light source 202 of FIG. 2, it can be necessary to perform maintenance on or to replace modules or apparatuses within the light source, including but not necessarily limited to the MO discharge chamber 224 and the power amplifier discharge chamber 244. For example, gases used as the lasing media in the chambers 224 and 244 can age and require refreshment or replacement. The chambers 224 and 244 and / or other components of the light source 202 can themselves age and require replacement. When the need arises for maintenance or replacement of apparatuses or components of the DUV light source 202 such as the chambers 224 and / or 244, it is desirable that maintenance and / or replacement be performed with as efficiently as possible and with as little disturbance as possible to the continued or scheduled use of an associated exposure system, such as exposure system 100 of FIG. 1.

[0047] In an aspect of the present disclosure, a predictive maintenance process of a DUV light source includes (a) collecting a first set of parameters of a first apparatus of the DUV light source, such as chamber 224 or 244; (b) analyzing the first set of parameters to calculate a first score of a likelihood of a failure mode of the first apparatus; and (c) when the first score exceeds a first pre-determined value: (1) generating one or more first action plans for maintenance and / or replacement of the first apparatus, (2) providing a notification of the one or more first action plans, and (3) storing the first set of parameters, the first score, and the one or more first action plans in a storage medium.

[0048] In another aspect, a predictive maintenance process of a deep ultraviolet (DUV) light source includes: (a) collecting respective sets of parameters of two or more apparatuses of the DUV light source, such as the chambers 224 and 224; (b) analyzing the respective sets of parameters to calculate respective scores of a likelihood of a failure mode of the two or more apparatuses; (c) generating an action plan for maintenance and / or replacement of each respective apparatus of the two or more apparatuses whose respective score of a likelihood of a failure mode exceeds a respectivepredetermined value, triggering a notification of the action plan, and storing the action plan in a storage medium; and (d) periodically repeating actions (a), (b), and (c) at least until the action plan is generated.

[0049] FIG. 3 is a flow diagram illustrating certain features of processes such as these. A process P300 includes first checking whether a failure prediction is needed or has been requested (S302). A failure prediction can be triggered (or considered “needed”) at a specified time interval, for example, daily, or may be requested manually by a user. A failure prediction can be triggered additionally, or alternatively, by a count of the number of light pulses produced by a light source or a count of the respective number of light pulses produced with the participation of a respective one of the apparatuses of the light source. If a failure prediction is not needed or requested (S302, “N” branch), the check can be performed again (S302), such as at a specified time interval, such as daily. If a failure prediction is needed or requested (S302, “Y” branch), a failure prediction algorithm is executed for one or more apparatuses or modules of the light source (S304). The failure prediction algorithm uses, among other information, data collected previously up to the time of performing the failure prediction, such as respective sets of performance parameters relating to the one or more apparatuses of the light source for which a prediction is to be made.

[0050] Failure prediction can take the form of performing an evaluation to determine a score representing a probability of an apparatus of the light source experiencing an operating fault within a predetermined second number of (upcoming) light pulses after the apparatus has participated in producing at least a predetermined first (or initial) number of light pulses. Additionally or alternatively, failure prediction can also take the form of performing an evaluation to determine a score representing the probability of the apparatus experiencing an operating fault within a predetermined fourth number of (upcoming) light pulses after the first apparatus has remained in service and participated in producing at least a third (or additional) number of light pulses since last being evaluated to determine the probably of experiencing an operating fault. Evaluating the probability can be performed using a model, including an analytical model and / or a trained machine learning model or other machine learning model.

[0051] Next, a check is performed to determine whether an action (such as a maintenance / repair or replacement action) is needed (S306). In implementations, a maintenance / repair or replacement action will generally be needed when a score exceeds (or exceeds or equals) a predetermined value. (Note that “exceeds” in this usage can mean going lower than a limit if lower numbers indicate higher probability of an operating fault or going higher than a limit if higher numbers indicate a higher probability of an operating fault.) Determination that an action is needed can be made in various ways. In implementations of the process, an action can typically be determined to be needed after only one failure or fault prediction score exceeding a given limit. Alternatively, for one or more of the apparatuses or modules of the light source, an action can be determined to be needed only after more than one failure or fault prediction score exceeding a given limit, such as after two or moreconsecutive failure or fault predictions (of the same apparatus), such as on consecutive days, for example. Other variations are of course possible. Also in some implementations, if a failure prediction is made that is identical to the previous day’s failure prediction for which an action plan has already been generated, an action plan can be determined as not needed. If an action plan is not needed (S306, “N” branch), the process returns to check for the need for further failure predictions (S302).

[0052] If an action plan is needed (S306, “Y” branch), an action plan is generated (S308). The action plan is generated by a large language model (LLM) or an equivalent or similar process, with data input or “prompt” including the failure prediction score(s) and the associated respective sets of parameters, relating to the one or more apparatuses of the light source predicted to fail. Light source metadata (R312) can also be included in the data input or “prompt” and / or can be used as a present resource or a prior training resource of the LLM in the generation of the action plan as shown. Data in the form of repair and / or replacement cycle logic R14 and in the form of a library R310 of past failure predictions and associated data (and in some implementations metadata) sets and action plans (R310) can also be used as a current resource and / or prior training resource of the LLM generating the action plan.

[0053] After the action plan has been generated, logic is executed to notify a user such as a member of a “local team” or the like of the newly created action plan (S316). The action plan itself is stored in a location such as a physical or virtual memory location accessible to the users for further study, modification, or use and the like (S318). The action plan or a revision or derivative thereof is then used (followed or executed) to maintain / repair and / or replace the respective apparatus or apparatuses of the light source.

[0054] In the failure prediction algorithm, the one or more apparatuses or modules of the light source, such as the chambers 224 and 244 of the light source 202 of FIG. 2, for example, can be analyzed for failure probabilities sequentially or in parallel. The analysis can include a first pre-specified number of light pulses below which no failure mode is predicted, and after which the analysis predicts the likelihood of a failure mode within a second number of (upcoming) pulses. The analysis can further include the probability of an apparatus experiencing an operating fault within a predetermined fourth number of (upcoming) light pulses when the apparatus has remained in service and participated in producing at least a third number of light pulses since last being evaluated to determine the probably of an operating fault.

[0055] In embodiments, the generated plan can include multiple sub-parts. In embodiments, the generated plan can include alternatives from which a user such as a user can select. The generated plan can be in the form of multiple plans, each for different aspects of the associated repair and / or replacement activities. The generated plan can also be in the form of a unified whole.

[0056] A process such as process P302 of FIG. 3 can further include providing access, such as read access or other access, for authorized users, to the stored generated action plan(s). The process canalso include, for authorized users, granting change access and / or review and approval access to the stored one or more action plans.

[0057] In implementations, the process, in forms described above and below herein and in other variations, can be performed by a computing device. In implementations, the process, in various forms described above and below herein and in other forms, can be instantiated in a non-transitory computer-readable storage medium comprising executable instructions for causing a processor to perform the operations of the process.

[0058] FIG. 3 A is a flow diagram of a process P300A having additional actions relative to the process P300 of FIG. 3. In the process P300A of FIG. 3A, performance measures such as wavelength, bandwidth, beam power, beam power stability, dose control (or energy control of bursts of pulses), fraction or raw number of dropped pulses over time, efficiency, operating voltage and the like can be monitored essentially continuously for any significant change or departure from targets or targeted ranges, that is, for out-of-range conditions (S303). Performance monitoring (S303) and failure prediction (S304) can be executed asynchronously in parallel, or synchronously and / or sequentially if desired.

[0059] If a significant change in performance and / or an out-of-range condition is detected (S303, “Y” branch), a check is performed (S306A) to determine whether an action that requires an action plan (such as a maintenance / repair or replacement action) is needed due to the performance change and / or out-of-range condition, and / or due to a failure prediction, if present. If a sufficiently significant failure prediction (a probability score of failure within a prespecified interval exceeding a threshold value) or if a sufficiently significant performance change or out-of-range condition is present, such that one or more significant and / or manual maintenance or replacement actions are needed (S306A, “Y” branch), an action plan is generated including all actions needed to address all current significant failure predictions and sufficiently significant performance changes and / or out-of-range conditions. Data in the form a library R310A of past failure predictions and performance changes and out-or range conditions — and associated data (and in some implementations metadata) sets — and associated action plans (R310A) can be used as a current resource and / or prior training resource of the LLM in generating the action plan. This process allows automating drafting of action plans addressing a wide range of performance issues in addition to refills and / or replacements of discharge chambers.

[0060] Other aspects of the process P300A are similar to or the same as in process P300 of FIG. 3 described above. Additional variations, examples, and details of processes for performance monitoring and failure prediction can be found in US Patent Publication 2024152063 Al, “Maintenance of Modules for Light Sources Used in Semiconductor Photolithography.” (Note that any patent applications, patents, and printed publications cited herein are incorporated herein by reference in their entireties, except for any definitions, subject matter disclaimers or disavowals, and except to the extent that the incorporated material is inconsistent with the express disclosure herein, in which case the language in this disclosure controls.)

[0061] FIG. 4 a schematic diagram of some elements of an implementation of an apparatus or system 480 for performing a process such as the process of FIG. 3 and / or of 3A or one or more variations thereof. As shown in FIG. 4, a light source LS supplies current and / or recent performance data 450, such as respective sets of performance parameters relating to one or more apparatuses of the light source LS to a data collection system LSD. These can include current and / or recent performance parameters relating to the performance of the light source LS as a whole. A source of light source metadata source “LMD”, such as a curated and stored listing of the metadata for the light source LS, supplies light source metadata 452c. The light source metadata can include product types or versions, model numbers of the light source or of modules thereof, dates of manufacture, part codes of parts or specific combinations of parts present in the light source or assemblies or modules thereof, software types and versions utilized in the light source or its control system, and the like. The performance data 450 and the metadata 452c can be brought together by the data collection system LSD and supplied (454) to a failure prediction analyzer (“FPA”) and, in implementations, to a performance analyzer PA. The PA when present may be part of or in parallel to the FPA, as suggested by the dotted line separation in the figure.

[0062] The FPA generates score, for a respective one or more apparatuses or modules of the light source, representing the probability of an operating fault or “failure” or “failure mode” occurring within a specified time (from present) and / or within a specified number of (upcoming) light pulses produced by the light source. For each apparatus or module having a score over a prespecified limit, the FPA transmits the respective score and associated data based on which the score was generated (represented as arrows 456-1, 456-2, through 456-n) to an action plan generator (“APG”). The FPA can also directly alert users of any score over the prespecified limit (457), such as through a preexisting standard communication system CS, such as an email or text message system, for example.

[0063] The PA when present analyzes the performance of the light source or of modules thereof on multiple measures of performance, for example, measures such as but not limited to wavelength of light produced, bandwidth of light produced, energy per pulse and / or energy per burst of pulses, energy stability over pulses and / or bursts, pointing stability, positioning stability, lost pulse counts or percentages, energy efficiency, and statistics based on these or combinations of these. For each performance measure found by the PA to be changing significantly or to be in an out-of-range condition, the PA transmits to the APG data representing the significant change and / or out-of-range condition together with the data based on which the significant change and / or out-of-range condition was detected (represented as arrows 455-1, 455-2, through 455 -n). The PA when present can also directly alert users of any significant change in performance and / or any out-of-range condition through the same or similar channels (457) as used by the FPA

[0064] The APG includes an action plan processor (“APP”) and an action plan storage (“APS”). The APP is a large language model (LLM) or equivalent or similar-functioning processor, and uses therespective data sets from the FPA (456-1, 456-2, through 456-n) and from the PA when present (455-1, 455-2, through 455 -n). In some implementations, light source metadata can be accessed or received (452ap) by the APP as a resource or reference, such as in implementation in which the metadata is not included in the data sets from the FPA. The APP also uses, as reference and / or as prior training material, stored maintenance and / or replacement cycle logic RCL which is transmitted to and / or interrogated by the APP (452cl), and a library LIB of past likelihood of failure mode predictions and associated data (and optionally of significant changes and / or out-of-range conditions) and associated resulting action plans, also transmitted to and / or interrogated by the APP (4521b). Numeric and other data is ordered in a specified standard order for presentation to the APP, and the same order is used for training and / or reference from the library LIB such that sequencer(s) of the APP can readily interpret the data it receives. The action plan generated by the APP is transmitted to (460) and stored in the action plan storage APS, along with additional separate supporting documents or data 460s- 1, 460s-2, through 460s-n, if any.

[0065] At or about the same time the action plan is stored, the APP (or the APG) transmits a notification alert 461 to one or more users or similar personnel or to a system accessible to them, notifying of the existence of the generated and stored action plan(s). The notification alert 461 can come through a preexisting standard communication system CS, such as an email or text message system. The notification alert can include instructions or a link for accessing the stored action plan(s). Access (462) to the stored action plan(s) in the storage medium of the APS can be through the preexisting standard communication system CS, or as shown in this implementation, through a user interface UI (which may take the same from as the preexisting standard communication system CS) or through a landing page reached by a link such as a transmitted link, or a graphical user interface or standard computer terminal interface at or remote from the light source LS itself, through similar systems. The access 462 can be limited to qualified users and can include change access and / or review and approval access for specifically qualified users. A request for approval of the generated plan (or of an edited version or a derivative of the generated plan) and a subsequent approval (464) can be communicated between the APG and the users or others, as desired. The APG can also forward the generated plan (or of an edited version or a derivative of the generated plan) (466) to a customer or facility owner or operator (“CUST”) for notice and / or review and approval, users or others can of course communicate directly with customers or facility operators regarding the generated plan or its derivative(s) (467).

[0066] Using the processes and apparatuses of the present disclosure can improve the reliability and efficiency of maintenance and / or replacement steps performed in maintaining a DUV light source. Automatic notification of users and / or others through existing communication systems such as email and / or text and the like helps prevent users from missing notifications and being inadequately prepared when repair or replacement is called for. Automatic generation of an action plan or action plans saves time in the planning of actions and in the preparation of documentation for repair and / orreplacement activities. This can in some cases allow relatively earlier repair and / or replacement when needed. Use of stored and maintained metadata by an action plan processor that generates the action plan or plans ensures that up-to-date metadata is used and avoids potential difficulties that may otherwise arise in expeditiously identifying correct metadata and applying or entering it without error. Use of an action plan storage location in the form of a central storage can help ensure version control of an action plan or plans in preparation and enable clear identification and communication of an action plan finalized and / or approved for execution.

[0067] Aspects and implementations of the present disclosure can be further described using the following numbered clauses:1. A predictive maintenance process of a light source, the process including: (a) collecting a first set of parameters of a first apparatus of the light source; (b) analyzing the first set of parameters to calculate a first score of a likelihood of a failure mode of the first apparatus; and (c) in response to the first score exceeding a first pre-determined value, (1) generating one or more first action plans for maintenance and / or replacement of the first apparatus, (2) providing a notification of the one or more first action plans, and (3) storing the first set of parameters, the first score, and the one or more first action plans in a storage medium.2. The predictive maintenance process of clause 1, further including , in response to the first score being below the first pre -determined value, repeating actions (a) and (b).3. The predictive maintenance process of clause 1, further including repeating actions (a), (b), and (c) at a predetermined frequency.4. The predictive maintenance process of clause 1, further including servicing the first apparatus according to one of more of the one or more first action plans or a revision or derivative thereof. 5. The predictive maintenance process of clause 1, further including: (d) collecting a second set of parameters from a second apparatus of the light source; (e) analyzing the second set of parameters to calculate a second score of a likelihood of a failure mode of the second apparatus; and (f) in response to the second score exceeding a second pre-determined value, (4) generating one or more second action plans for maintenance and / or replacement of the second apparatus, and (5) triggering a notification of the one or more second action plans, and (6) storing the second set of parameters, the second score, and the one or more second action plans in a storage medium.6. The predictive maintenance process of clause 5, further including, in response to the second score being below the second pre-determined value, repeating actions (d) and (e).7. The predictive maintenance process of clause 5, further including repeating actions (d), (e), and (f) at a predetermined frequency.8. The predictive maintenance process of clause 5, further including servicing the second apparatus according to one of more of the one or more second action plans or a revision or derivative thereof.9. The predictive maintenance process of clause 1, wherein analyzing the first set of parameters to calculate the first score includes performing an evaluation to determine the probability of the firstapparatus, in response to participating in producing at least a first number of light pulses, experiencing an operating fault within a predetermined second number of (upcoming) light pulses.10. The predictive maintenance process of clause 9, wherein analyzing the first set of parameters to calculate the first score of the likelihood of the failure mode of the first apparatus includes performing an evaluation to determine the probability of the first apparatus, in response to being remained in service after participating in producing at least a third number of light pulses since being evaluated to determine the probability of operating fault, experiencing an operating fault within a predetermined fourth number of (upcoming) light pulses.11. The predictive maintenance process of clause 1, wherein generating the one or more first action plans for maintenance and / or replacement of the first apparatus includes generating the one or more first action plans using a large language model (LLM) with inputs, training inputs, and / or references including the first score of the likelihood of a failure mode and one or more of (1) stored light source and / or apparatus metadata, (2) stored maintenance and / or replacement cycle logic, and (3) a library of past scores of likelihoods of failure modes and associated action plans.12. The predictive maintenance process of clause 11, wherein generating the one or more first action plans for maintenance and / or replacement of the first apparatus includes automatically generating the one or more first action plans using an LLM trained on a library of past likelihood of failure mode predictions and associated action plans, with an input (prompt) including the first score of the likelihood of a failure mode.13. The predictive maintenance process of clause 1, wherein generating the one or more first action plans for maintenance and / or replacement of the first apparatus includes generating the one or more first action plans using an LLM trained on a library of past scores of likelihood of failure mode predictions and associated action plans, with an input (prompt) including the first score of the likelihood of the failure mode.14. The predictive maintenance process of clause 1, further including providing access for authorized users to the first set of parameters, the first score, and the one or more first action plans in the storage medium.15. The predictive maintenance process of clause 1, further including providing, for authorized users, change access to the one or more first action plans in the storage medium.16. The predictive maintenance process of clause 1, further including providing, for authorized users, review and approval access to the one or more first action plans in the storage medium.17. A predictive maintenance process of a deep ultraviolet (DUV) light source, the process including: (a) collecting respective sets of parameters of two or more apparatuses of the DUV light source; (b) analyzing the respective sets of parameters to calculate respective scores of a likelihood of a failure mode of the two or more apparatuses; (c) (1) generating an action plan for maintenance and / or replacement of each respective apparatus of the two or more apparatuses whose respective score of a likelihood of the failure mode exceeds a respective predetermined value, (2) triggering a notificationof the action plan, and (3) storing the action plan in a storage medium; and (d) periodically repeating actions (a) through (c) at least until the action plan is generated.18. The predictive maintenance process of clause 17, further including servicing the respective apparatus according to the action plan or a revision or derivative thereof.19. The predictive maintenance process of clause 17, wherein analyzing the respective sets of parameters to calculate respective scores includes performing respective evaluations to determine respective probabilities the two or more apparatuses experiencing a respective operating fault within a respective predetermined number of (upcoming) light pulses.20. The predictive maintenance process of clause 17, wherein generating the respective action plan for maintenance and / or replacement of each respective apparatus includes using a large language model (LLM) based on data including each respective set of parameters and each respective score of the likelihood of the failure mode, and based on data and / or training including (1) stored light source and / or apparatus metadata, (2) stored maintenance and / or replacement cycle logic, and (3) a library of past scores of likelihoods of failure modes and associated action plans.21. The predictive maintenance process of clause 17, wherein generating the action plan for maintenance and / or replacement of each respective apparatus includes using an LLM trained on a library of past scores of likelihoods of failure modes and associated action plans and based on data (prompt(s)) including each respective set of parameters and each respective score of the likelihood of the failure mode.22. A computer-implemented process including: (a) analyzing by a computing device respective sets of parameters of two or more apparatuses of a DUV light source to calculate respective scores of a likelihood of a failure mode of the two or more apparatuses; (b) by the computing device, generating or triggering the generation an action plan for maintenance and / or replacement of each respective apparatus of the two or more apparatuses whose respective score of a likelihood of a failure mode exceeds a respective predetermined value, triggering a notification of the action plan, and storing the action plan in a storage medium; and (c) repeating (a) and (b) at least until the action plan is generated by the computing device.23. The computer-implemented process of clause 22, wherein the generation of the action plan for maintenance and / or replacement of each respective apparatus includes using a large language model (LLM) based on data including each respective set of parameters and each respective score of the likelihood of a failure mode, and based on data and / or training including (1) stored light source and / or apparatus metadata, (2) stored maintenance and / or replacement cycle logic, and (3) a library of past scores of likelihoods of failure modes and associated action plans.24. A non-transitory computer-readable storage medium including executable instructions for causing a processor to perform operations, the instructions including instructions for performing the following operations: (a) analyzing respective sets of parameters of two or more apparatuses of a DUV light source to calculate respective scores of a likelihood of a failure mode of the two or more apparatuses;(b) generating an action plan for maintenance and / or replacement of each respective apparatus of the two or more apparatuses whose respective score of a likelihood of a failure mode exceeds a respective predetermined value, triggering a notification of the action plan, and storing the action plan in a storage medium; and (c) periodically repeating actions (a) and (b) at least until the action plan is generated.25. The non-transitory computer-readable storage medium of clause 22, wherein generating the action plan includes using a large language model (LLM) based on data including each respective set of parameters and each respective score of the likelihood of a failure mode, and based on data and / or training including (1) stored light source and / or apparatus metadata, (2) stored maintenance and / or replacement cycle logic, and (3) a library of past scores of likelihoods of failure modes and associated action plans.26. A maintenance process of a deep ultraviolet (DUV) light source, the process including: (a) collecting a set of parameters of the DUV light source; (b) analyzing the set of parameters to calculate a score of a likelihood of a failure mode of a laser chamber and to detect one or more significant changes and / or out-of-range conditions in the performance of the light source or of apparatuses or components thereof; and (c) (1) generating an action plan for maintenance and / or replacement of the laser chamber when the score exceeds a first pre -determined value, and for maintenance to correct the one or more significant changes and / or out-of-range conditions when present, (2) providing a notification of the action plan, and (3) storing the set of parameters, the score, the one or more significant changes and / or out-of-range conditions, and the action plan in a storage medium.27. The maintenance process of clause 26, wherein generating the action plan includes using a large language model (LLM) based on data including the set of parameters and the score of the likelihood of a failure mode and the one or more significant changes and / or out-of-range conditions, and based on data and / or training including (1) stored light source and / or apparatus metadata, (2) stored maintenance and / or replacement cycle logic, and (3) a library of past scores of the likelihood of a failure mode and past one or more significant changes and / or out-of-range conditions and associated action plans.28. A system for automated maintenance of a deep ultraviolet (DUV) light source, the system including: (a) a data collection system configured to automatically receive and collect a set of parameters of the DUV light source; (b) a failure prediction analyzer configured to automatically receive from the data collection system and analyze the set or sets of parameters to calculate a score of a likelihood of a failure mode of a laser chamber; and (c) an action plan generator configured to (1) automatically generate an action plan for maintenance and / or replacement of the laser chamber when the calculated score exceeds a first pre-determined value, (2) automatically provide a notification of the action plan, and (3) automatically store the set of parameters, the score, and the action plan in a storage medium.29. The system of clause 28, wherein the action plan generator includes a large language model (LLM) configured to generate an action plan based on data including the set of parameters and the score of the likelihood of the failure mode, and based on data and / or training including (1) stored light source and / or apparatus metadata, (2) stored maintenance and / or replacement cycle logic, and (3) a library of past scores of the likelihood of a failure mode and associated sets of parameters and associated action plans.30. The system of clause 28, further including a performance analyzer configured to automatically receive the set of parameters from the data collection system and to detect one or more significant changes and / or out-of-range conditions in the performance of the light source or of apparatuses or components thereof, and wherein the action plan generator is further configured to generate the action plan including actions to correct the one or more significant changes and / or out-of-range conditions when present.

[0068] The above-described aspects and implementations and other implementations are within the scope of the following claims.

Claims

CLAIMS1. A predictive maintenance process of a light source, the process comprising:(a) collecting a first set of parameters of a first apparatus of the light source;(b) analyzing the first set of parameters to calculate a first score of a likelihood of a failure mode of the first apparatus; and(c) in response to the first score exceeding a first pre -determined value, (1) generating one or more first action plans for maintenance and / or replacement of the first apparatus, (2) providing a notification of the one or more first action plans, and (3) storing the first set of parameters, the first score, and the one or more first action plans in a storage medium.

2. The predictive maintenance process of claim 1, further comprising , in response to the first score being below the first pre-determined value, repeating actions (a) and (b).

3. The predictive maintenance process of claim 1, further comprising repeating actions (a), (b), and (c) at a predetermined frequency.

4. The predictive maintenance process of claim 1, further comprising servicing the first apparatus according to one of more of the one or more first action plans or a revision or derivative thereof.

5. The predictive maintenance process of claim 1, further comprising:(d) collecting a second set of parameters from a second apparatus of the light source;(e) analyzing the second set of parameters to calculate a second score of a likelihood of a failure mode of the second apparatus; and(f) in response to the second score exceeding a second pre-determined value, (4) generating one or more second action plans for maintenance and / or replacement of the second apparatus, and (5) triggering a notification of the one or more second action plans, and (6) storing the second set of parameters, the second score, and the one or more second action plans in a storage medium.

6. The predictive maintenance process of claim 5, , in response to the second score being below the second pre-determined value, further comprising repeating actions (d) and (e).

7. The predictive maintenance process of claim 5, further comprising repeating actions (d), (e), and (f) at a predetermined frequency.

8. The predictive maintenance process of claim 5, further comprising servicing the second apparatus according to one of more of the one or more second action plans or a revision or derivative thereof.

9. The predictive maintenance process of claim 1, wherein analyzing the first set of parameters to calculate the first score comprises performing an evaluation to determine the probability of the first apparatus, in response to participating in producing at least a first number of light pulses, experiencing an operating fault within a predetermined second number of upcoming light pulses.

10. The predictive maintenance process of claim 9, wherein analyzing the first set of parameters to calculate the first score of the likelihood of the failure mode of the first apparatus comprises performing an evaluation to determine the probability of the first apparatus, in response to being remained in service after participating in producing at least a third number of light pulses since being evaluated to determine the probability of operating fault, experiencing an operating fault within a predetermined fourth number of upcoming light pulses.

11. The predictive maintenance process of claim 1, wherein generating the one or more first action plans for maintenance and / or replacement of the first apparatus comprises generating the one or more first action plans using a large language model (LLM) with inputs, training inputs, and / or references including the first score of the likelihood of the failure mode and one or more of (1) stored light source and / or apparatus metadata, (2) stored maintenance and / or replacement cycle logic, and (3) a library of past scores of likelihoods of failure modes and associated action plans.

12. The predictive maintenance process of claim 1, wherein generating the one or more first action plans for maintenance and / or replacement of the first apparatus comprises generating the one or more first action plans using an LLM trained on a library of past scores of likelihood of failure mode predictions and associated action plans, with and input (prompt) including the first score of the likelihood of the failure mode.

13. A predictive maintenance process of a deep ultraviolet (DUV) light source, the process comprising:(a) collecting respective sets of parameters of two or more apparatuses of the DUV light source;(b) analyzing the respective sets of parameters to calculate respective scores of a likelihood of a failure mode of the two or more apparatuses;(c) (1) generating an action plan for maintenance and / or replacement of each respective apparatus of the two or more apparatuses whose respective score of a likelihood of the failure mode exceeds a respective predetermined value, (2) triggering a notification of the action plan, and (3) storing the action plan in a storage medium; and(d) periodically repeating actions (a) through (c) at least until the action plan is generated.

14. The predictive maintenance process of claim 13, further comprising servicing the respective apparatus according to the action plan or a revision or derivative thereof.

15. The predictive maintenance process of claim 13, wherein analyzing the respective sets of parameters to calculate respective scores comprises performing respective evaluations to determine respective probabilities the two or more apparatuses experiencing a respective operating fault within a respective predetermined number of upcoming light pulses.

16. The predictive maintenance process of claim 13, wherein generating the respective action plan for maintenance and / or replacement of each respective apparatus comprises using a large language model (LLM) based on data including each respective set of parameters and each respective score of the likelihood of the failure mode, and based on data and / or training including (1) stored light source and / or apparatus metadata, (2) stored maintenance and / or replacement cycle logic, and (3) a library of past scores of likelihoods of failure modes and associated action plans.

17. The predictive maintenance process of claim 13, wherein generating the action plan for maintenance and / or replacement of each respective apparatus comprises using an LLM trained on a library of past scores of likelihoods of failure modes and associated action plans and based on data (prompt(s)) including each respective set of parameters and each respective score of the likelihood of the failure mode.

18. A system for automated maintenance of a deep ultraviolet (DUV) light source, the system comprising:(a) a data collection system configured to automatically receive and collect a set of parameters of the DUV light source;(b) a failure prediction analyzer configured to automatically receive from the data collection system and analyze the set or sets of parameters to calculate a score of a likelihood of a failure mode of a laser chamber; and(c) an action plan generator configured to (1) automatically generate an action plan for maintenance and / or replacement of the laser chamber when the calculated score exceeds a first predetermined value, (2) automatically provide a notification of the action plan, and (3) automatically store the set of parameters, the score, and the action plan in a storage medium.

19. The system of claim 18, wherein the action plan generator comprises a large language model (LLM) configured to generate an action plan based on data including the set of parameters and the score of the likelihood of the failure mode, and based on data and / or training including (1) stored light source and / or apparatus metadata, (2) stored maintenance and / or replacement cycle logic, and (3) alibrary of past scores of the likelihood of a failure mode and associated sets of parameters and associated action plans.

20. The system of claim 18, further comprising a performance analyzer configured to automatically receive the set of parameters from the data collection system and to detect one or more significant changes and / or out-of-range conditions in the performance of the light source or of apparatuses or components thereof, and wherein the action plan generator is further configured to generate the action plan including actions to correct the one or more significant changes and / or out-of-range conditions when present.

Images