Event-based detection of rapidly changing phenomena

Abstract

Methods and apparatus for monitoring a processing apparatus and event-based detection of detecting rapidly changing phenomena are provided. In some embodiments, techniques for detecting rapidly changing phenomena may include: responsive to an occurrence of a phenomenon, determining one or more changes in an intensity of optical signals, the optical signals obtained by an optical sensor; detecting one or more changes in voltage correlating to the one or more changes in intensity of optical signals; and based on the one or more changes in voltage staying above a threshold, generating images representative of the phenomenon and comprising indications of the one or more changes in intensity of the optical signals occurring at corresponding pixels of the images.

Description

Docket No. LAM1P014WOEVENT-BASED DETECTION OF RAPIDLY CHANGING PHENOMENACROSS-REFERENCE TO RELATED APPLICATIONS

[0001] A PCT Request Form is filed concurrently with this specification as part of the present application. Each application that the present application claims benefit of or priority to as identified in the concurrently filed PCT Request Form is incorporated by reference herein in its entirety and for all purposes.BACKGROUND

[0002] Tools and systems configured to deposit, etch and / or manipulate materials and films on semiconductor substrates using gaseous chemistries often use one or more sensors to monitor and detect phenomena and changes thereof. A variety of sensed information can be used to understand the state of the system. However, there are types of phenomena that suddenly occur in these systems at rapid, instantaneous, or near-instantaneous speeds that many sensors cannot resolve temporally, spatially, or both. These rapidly occurring and rapidly changing phenomena reveal the limitations of currently used sensors.

[0003] For example, cameras and visual sensors can capture plasma and mechanical motions spatially with a fixed frame rate but have a limitation in temporal resolution (e.g., insufficient frames per second to capture information visually). Photodiode sensors, while they can detect fast-acting phenomena such as arcs, are limited by spatial resolution as they cannot tell the location of a phenomenon. Other plasma phenomena such as plasmoids are not captured by spectral sensors (e.g., those that use optical emission spectroscopy (OES) to observe elemental compositions of a material), or photodetectors that generally provide a single data point per sample or timestamp, but can be resolved by a spatial sensor (e.g., a camera). Vibrations at different frequencies may not be captured by most sensors except accelerometers used with select products and tools. Mechanical motions are not typically visible under low light conditions. Some examples of mechanical motions include lift pin motion and motion of other robotic components. Relatedly, clamping failures may not be detected until much later where the chances of process impact is high. Valve-related failures may also not always be captured quickly and easily, especially under low light or low resolution. Other types of sensors can be used as well to acquire data such as VIX, voltage and current (VI), power (based, e.g., on phase, impedance, frequency), pressure, temperature, etc.Docket No. LAM1P014WO

[0004] Each of these sensors can provide one or few aspects but not provide global insight. For instance, a temporal sensor configured to measure temporal characteristics of pulsed light or other bursts of light can detect fast arcs, but not provide the location of such arcs. Examples of temporal sensors can include photodiode optical sensors, micro-photodiodes, phototransistors, photocells, pulse characterization sensors, photomultiplier tubes, solid-state photomultipliers, other fast, one-dimension radiation (typically but not necessarily limited to visible light) sensors, or arrays thereof. In many situations, these sensors acquire data at a fixed rate (e.g., continuous recording by a camera at a set frame per second) where transient phenomena may not be caught or misinterpreted. This requires parsing of information to determine phenomena of interest, which can be computationally expensive or slow, or require manual interpretation. These methods of data collection produce large volumes of data and are either analyzed manually or in a batch process, resulting in large file sizes, high storage requirements, and heavy computational solutions that use numerous or large computers, graphics processing units (GPUs), and / or similar equipment. In general, temporal sensors only reach total observed intensity above a defined threshold or set of thresholds, and accordingly, not only cannot provide locality but also cannot provide the number of events and the relative intensity of each event.

[0005] To these ends, it would be desirable to use a more efficient approach with a single type of sensor to capture suddenly occurring, rapidly changing, fast-acting, or transient phenomena without high computational and data storage requirements.

[0006] The background description provided herein is for the purposes of generally presenting the context of the disclosure. Work of the presently named inventors, to the extent it is described in this background section, as well as aspects of the description that may not otherwise qualify as prior art at the time of filing, are neither expressly nor implicitly admitted as prior art against the present disclosure.SUMMARY

[0007] In one aspect of the present disclosure, a method for detecting rapidly changing phenomena is disclosed. In some embodiments, the method may include: responsive to an occurrence of a phenomenon, determining one or more changes in an intensity of optical signals, the optical signals obtained by an optical sensor; detecting one or more changes in voltage correlating to the one or more changes in intensity of optical signals; and based on the one or more changes in voltage staying above a threshold, generating images representative ofDocket No. LAM1P014WO the phenomenon and comprising indications of the one or more changes in intensity of the optical signals occurring at corresponding pixels of the images.

[0008] In some implementations thereof, the determining of the one or more changes in voltage may include determining a positive change in voltage associated with the optical signals with respect to time. In some implementations thereof, the determining of the one or more changes in voltage may include determining a negative change in voltage associated with the optical signals with respect to time.

[0009] In some scenarios, the phenomenon may include a plasma ignition, plasma motion, plasmoid formation or movement, a plasma discharge, an electrical arc, motion of a component of a processing apparatus, or a hollow cathode discharge (HCD), or a ceasing thereof.

[0010] In some cases, the phenomenon may include the electrical arc within the processing apparatus; and the method may further include, based on a computerized visual evaluation of the pixels of the images, identifying one or more images of the generated images associated with the electrical arc.

[0011] In some cases, the phenomenon may include the HCD within the processing apparatus; and the method may further include, based on a computerized visual evaluation of the pixels of the images, identifying one or more images of the generated images associated with the HCD.

[0012] In some cases, the phenomenon may include the motion of the component of the processing apparatus, the motion including an actuation of a valve; and the method may further include determining a degradation of the valve based on a frequency of a vibration occurring during the actuation being under a frequency threshold, or based on a change in the frequency of the vibration occurring during the actuation, the frequency of the vibration determined based on the generated images associated with the motion of the component.

[0013] In another aspect of the present disclosure, a method for mitigating a failure occurring at a processing apparatus is disclosed. In some embodiments, the method may include: obtaining a plurality of images representative of a phenomenon occurring with respect to the processing apparatus, the plurality of images including indications of one or more changes in an intensity of optical signals obtained by an optical sensor; and based on a difference between at least a first image and a second image of the plurality of images exceeding a threshold, causing the processing apparatus to adjust an ongoing process associated with the phenomenon.

[0014] In some implementations thereof, the adjustment of the ongoing process may include an issuance of a warning or an error, or a shutdown of the processing apparatus or a component thereof.Docket No. LAM1P014WO

[0015] In some embodiments, the method may include: obtaining a plurality of images representative of a phenomenon occurring with respect to the processing apparatus, each of the plurality of images including a plurality of pixels each indicative of a change in an intensity of optical signals obtained by an optical sensor; and based at least on a difference in the intensity of optical signals indicated by at least a first pixel of a first image of the plurality of images and a corresponding first pixel of a second image of the plurality of images exceeding a threshold, causing the processing apparatus to adjust an ongoing process associated with the phenomenon.

[0016] In some implementations thereof, the adjustment of the ongoing process may include an issuance of a warning or an error, or a shutdown of the processing apparatus or a component thereof.

[0017] These and other features of the disclosed embodiments will be described in detail below with reference to the associated drawings.BRIEF DESCRIPTION OF DRAWINGS

[0018] FIG. 1A shows an example processing apparatus for depositing or etching a film on or over a substrate utilizing a plasma process, in which the example processing apparatus may include a camera sensor.

[0019] FIG. IB presents a schematic view of an implementation of a multi-station processing tool; the tool includes four camera sensors.

[0020] FIG. 1C presents a top view of an example electronic device fabrication system having four multi-station fabrication tools, one of which includes camera sensors.

[0021] FIG. ID presents an example of spectral sensitivity ranges for “standard” (visible- sensitive), IR-sensitive, and UV-sensitive sensor elements in a camera sensor or a combination of camera sensors.

[0022] FIG. IE schematically depicts a process chamber with a camera oriented to capture images along a horizontal line of sign into the chamber interior.

[0023] FIG. IF schematically depicts a process chamber with cameras oriented to capture images along vertical lines of sight into the chamber interior.

[0024] FIG. 1G illustrates an embodiment in which a camera sensor is mounted to access a viewport or optical access aperture of a chamber wall in a fabrication tool.

[0025] FIG. 2 is a simplified diagram of a traditional imaging scheme with a fixed frame rate.

[0026] FIG. 3 is a simplified diagram of an example imaging scheme with a dynamic, eventbased frame rate.Docket No. LAM1P014WO

[0027] FIGS. 4A and 4B are flow diagrams of example processes for an event-based detection of phenomena using comparison of frames, according to some embodiments.

[0028] FIG. 5 is a flow diagram of an example process for event-based arc detection, according to some embodiments.

[0029] FIG. 6 is a flow diagram of an example process for event-based transients detection, according to some embodiments.

[0030] FIGS. 7 A and 7B illustrate a sequence of example frames representative of a rapidly occurring phenomenon such as activation of a light source, captured using event-based detection of the phenomenon, according to some embodiments.

[0031] FIGS. 8 A and 8B illustrate another sequence of example frames representative of a rapidly occurring phenomenon such as deactivation of a light source, captured using eventbased detection of the phenomenon, according to some embodiments.

[0032] FIGS. 9 A and 9B illustrate another sequence of example frames representative of a rapidly occurring phenomenon such as an electrical arc, captured using event-based detection of the phenomenon, according to some embodiments.

[0033] FIG. 10 is a flow diagram of an example process for detection of an electrical arc and adjustment of a process based thereon, according to some embodiments.

[0034] FIGS. 11A and 11B illustrate another sequence of example frames representative of a rapidly occurring phenomenon such as hollow cathode discharge (HCD), captured using eventbased detection of the phenomenon, according to some embodiments.

[0035] FIG. 12 is a flow diagram of an example method for detecting rapidly changing phenomena, according to some embodiments.

[0036] FIG. 13 is a flow diagram of an example method for mitigating a failure occurring at a processing apparatus, according to some embodiments.

[0037] FIG. 14 is a flow diagram of another example method for mitigating a failure occurring at a processing apparatus, according to some embodiments.

[0038] FIG. 15 shows a schematic of an example process system that may be used to perform the methods described herein.DETAILED DESCRIPTION

[0039] The following terms are used throughout the instant specification:

[0040] The terms “semiconductor wafer,” “wafer,” “semiconductor substrate,” “substrate,” “wafer substrate” and “partially fabricated integrated circuit” may be used interchangeably. Those of ordinary skill in the art understand that the term “partially fabricated integratedDocket No. LAM1P014WO circuit” can refer to a semiconductor wafer during any of many stages of integrated circuit fabrication thereon. A wafer or substrate used in the semiconductor device industry typically has a diameter of 200 mm, 300 mm, or 450 mm. Examples of wafer materials include silicon (Si), gallium arsenide (GaAs), and silicon germanium (SiGe). Besides semiconductor wafers, other workpieces that may take advantage of the disclosed embodiments include various articles such as printed circuit boards, magnetic recording media, magnetic recording sensors, mirrors, optical elements, display devices or components such as backplanes for pixelated display devices, flat panel displays, micro-mechanical devices and the like. The workpiece may be of various shapes, sizes, and materials.

[0041] “Manufacturing equipment” or “fabrication tool” refers to equipment in which a manufacturing process takes place. Manufacturing equipment may include a processing chamber in which the workpiece resides during processing. Typically, when in use, manufacturing equipment performs one or more electronic device fabrication operations. Examples of manufacturing equipment for semiconductor device fabrication include subtractive process reactors and additive process reactors. Examples of subtractive process reactors include dry etch reactors (e.g., chemical and / or physical etch reactors), wet etch reactors, and ashers. Examples of additive process reactors include chemical vapor deposition reactors, and atomic layer deposition reactors, physical vapor deposition reactors, wet chemical deposition reactors, electroless metal deposition cells, and electroplating cells.

[0042] In various embodiments, a process reactor or other manufacturing equipment includes a tool for holding a substrate during processing. Such tool is often a pedestal or chuck, and these terms are sometimes used herein as a shorthand for referring to all types of substrate holding or supporting tools that are included in manufacturing equipment.

[0043] As used herein and unless otherwise qualified, the term “camera,” “camera sensor,” or “sensor” is not limited to sensors designed or configured to work with a camera. These terms include other multi-pixel radiation sensors, with or without color or multispectral filters, that can provide sensed information that can provide an image of a radiation distribution within a fabrication tool. According to different implementations, optical (including visual) sensors or cameras, or infrared (IR) sensors or cameras, or both may be part of a camera or a sensor.

[0044] The term “image” may refer to a spatial representation of a physical domain including one or more features. An image may be provided in the form of data or signals arranged to represent the physical domain. An image may be produced by a pixelated sensor such as a camera sensor. An image may contain the spatial representation of the physical domain in oneDocket No. LAM1P014WO dimension, two dimensions, or three dimensions. Multiple images obtained consecutively over time may form a video representation of the physical domain.

[0045] Information about conditions within a process chamber, as captured with a camera sensor or pixelated sensor, may include radiation intensity values as a function of position within the process chamber. In some embodiments, the radiation intensity values are provided as an image. In some embodiments, the radiation intensity values are provided as two- dimensional or three-dimensional pixelated values. In other embodiments, the radiation intensity values are provided in only one dimension such as along a slit or interface between components. In some implementations a one-dimensional sensor or array is configured to scan the interior of a process chamber to generate, e.g., a two-dimensional image of a portion of the process chamber interior.

[0046] Optionally, the intensity values are also provided as a function of wavelength. In some embodiments, a camera sensor or other pixelated sensor comprises separate detection elements, each configured to capture radiation values at a given location, but with different spectral sensitivity profiles; e.g., in the red, green, and blue regions. Some camera sensors are configured to capture radiation intensity values in discrete wavelength ranges that are sometimes referred to as bins. Such sensors include hyperspectral imagers that capture intensity values in narrow wavelength bins and multispectral imagers that capture intensity values over broader wavelength bins. Optionally, the intensity values are provided as function or time; for example, images may be captured as video frames. In some embodiments, multiple camera sensors provide information from different overlapping or contiguous regions within a fabrication tool.

[0047] Figure ID presents an example of spectral sensitivity ranges for “standard” (visible- sensitive), IR-sensitive, and UV-sensitive sensor elements in a camera sensor or a combination of camera sensors.

[0048] A “semiconductor device fabrication operation” or “fabrication operation” as used herein may refer to an operation performed during fabrication of semiconductor devices. Typically, the overall fabrication process includes multiple semiconductor device fabrication operations, each performed in its own semiconductor fabrication tool such as a plasma reactor, an electroplating cell, a chemical mechanical planarization tool, a wet etch tool, and the like. Categories of semiconductor device fabrication operations include subtractive processes, such as etch processes and planarization processes, and material additive processes, such as deposition processes (e.g., physical vapor deposition, chemical vapor deposition, atomic layer deposition, electrochemical deposition, electroless deposition). In the context of etch processes,Docket No. LAM1P014WO a substrate etch process includes processes that etch a mask layer or, more generally, processes that etch any layer of material previously deposited on and / or otherwise residing on a substrate surface. Such an etch process may etch a stack of layers in the substrate.

[0049] In various approaches, materials (such as the above) in stacks of layers (e.g., thin films) may be deposited through chemical vapor deposition (CVD) techniques such as plasma enhanced chemical vapor deposition (PECVD), low pressure chemical vapor deposition (LPCVD), atmospheric pressure chemical vapor deposition (APCVD), metal organic chemical vapor deposition (MOCVD), atomic layer deposition (ALD), plasma enhanced atomic layer deposition (PEALD), or through direct metal deposition (DMD), etc. These examples are not intended to be limiting. Different deposition techniques may be used for different ends. For example, deposition may be useful whenever wafer stress and / or bowing are induced due to material present on the frontside of the wafer. Notably, different deposition techniques may be associated with or subject to different temperatures, pressures, precursors, or other deposition processes, conditions, or recipes. For instance, certain deposition techniques may overlap in temperatures ranges that can be used, which may advantageously be used to strategically deposit multiple layers at once or certain layers only.Location and Integration of a Sensor in a Processing Apparatus

[0050] Figure 1A shows an example fabrication tool denoted as an example substrate processing apparatus 100. Processing apparatus 100 may be configured for depositing films on or over a semiconductor substrate utilizing any number of processes. For example, processing apparatus 100 may be adapted for performing, in particular, plasma-enhanced chemical vapor deposition (PECVD) or plasma-enhanced atomic layer deposition (PEAED). Processing apparatus 100 may include a sensor 117 on a chamber wall. Sensor 117 may be configured to capture data (e.g., images from a visual or optical sensor, spectral data from a spectral sensor, temporal data from a photodiode) from the interior of processing apparatus 100. Note that while sensor 117 is shown as a single block, it is intended to depict implementations in which one, two, or more sensors are located proximate to one another, optionally sharing a single viewport or other window into a process chamber or a process station 102. In some cases, the individual sensors within block 117 may be trained on different components or fields of vision within a chamber interior. In some cases, the individual sensors within block 117 may be configured to capture different spectral ranges (far infrared (IR), near IR, visible, ultraviolet (UV), etc.). In some embodiments, block 117 and / or sensor(s) represented thereby may be configured to capture information based on events, which will be described in more detail below.Docket No. LAM1P014WO

[0051] In some example configurations, processing apparatus 100 of Figure 1A may employ a single process station 102 of a process chamber with a single substrate holder 108 (e.g., a pedestal) in an interior volume, which may be maintained under vacuum by a vacuum pump 118. A showerhead 106 and a gas delivery system 101, which are fluidically coupled to the process chamber, may permit the delivery of film precursors, for example, as well as carrier and / or purge and / or process gases, secondary reactants, etc.

[0052] In Figure 1A, gas delivery system 101 may include a mixing vessel 104 for blending and / or conditioning process gases for delivery to showerhead 106. One or more mixing vessel inlet valves 120 may control introduction of process gases to mixing vessel 104. Particular reactants may be stored in liquid form (e.g., in an ampoule source) prior to vaporization and subsequent delivery to process station 102 of a process chamber. The implementation of Figure 1 A may include a vaporization point 103 for vaporizing liquid reactant to be supplied to mixing vessel 104. In some implementations, vaporization point 103 may include a heated liquid injection module. In some other implementations, vaporization point 103 may include a heated vaporizer. In yet other implementations, vaporization point 103 may be eliminated from the process station. In some implementations, a liquid flow controller upstream of vaporization point 103 may be provided for controlling a mass flow of liquid for vaporization and delivery to process station 102.

[0053] Showerhead 106 may operate to distribute process gases and / or reactants (e.g., film precursors) toward a substrate 112 at the process station, the flow of which may be controlled by one or more mechanical valves upstream from the showerhead (e.g., valves 120, 120A, 105). In the implementation depicted in Figure 1 A, substrate 112 is depicted as located beneath showerhead 106, and is shown resting on a pedestal 108. Showerhead 106 may include any suitable shape and may include any suitable number and arrangement of ports for distributing process gases to substrate 112. In some implementations involving two or more process stations, gas delivery system 101 may include valves or other flow control structures upstream from the showerhead, which can independently control the flow of process gases and / or reactants to each station so as to, e.g., permit gas flow to one station while prohibiting gas flow to a second station. Furthermore, gas delivery system 101 may be configured to independently control process gases and / or reactants delivered to each station in a multi-station apparatus such that the gas composition provided to different stations is different; e.g., the partial pressure of a gas component may vary between stations at the same time.

[0054] In the implementation of Figure 1A, gas volume 107 is depicted as being located beneath showerhead 106. In some implementations, pedestal 108 may be raised or lowered toDocket No. LAM1P014WO expose substrate 112 to gas volume 107 and / or to vary the size of gas volume 107. The separation between pedestal 108 and showerhead 106 is sometimes referred to as a “gap.” Optionally, pedestal 108 may be lowered and / or raised during portions of the deposition process to modulate process pressure, reactant concentration, etc., within gas volume 107. Showerhead 106 and pedestal 108 are depicted as being electrically coupled to a radio frequency (RF) signal generator 114 and matching network 116 for coupling power to a plasma generator. Thus, showerhead 106 may function as an electrode for coupling radio frequency power into process station 102. RF signal generator 114 and matching network 116 may be operated at any suitable RF power level, which may operate to form plasma having a desired composition of radical species, ions, and electrons. In addition, RF signal generator 114 may provide RF power having more than one frequency component, such as a low-frequency component (e.g., less than about 2 MHz) as well as a high frequency component (e.g., greater than about 2 MHz). In some implementations, plasma ignition and maintenance conditions are controlled with appropriate hardware and / or appropriate machine- or computer-readable instructions in a system controller which may provide control instructions via a sequence of input / output control instructions.

[0055] In general, any plasma-assisted fabrication tool may be used to implement the disclosed embodiments, including integration of a sensor configured to capture information (e.g., images of plasmas and / or plasma-related phenomena, or images of or relating to components of the fabrication tool). Example fabrication tools may include deposition apparatuses, which may include, but are not limited to, apparatus from the ALTUS® product family, the VECTOR® product family, and / or the SPEED® product family, the KIYO® product family, the STRIKER® product family, and the VERSYS® product family, each available from Lam Research Corp., of Fremont, California, or any of a variety of other fabrication tools employing plasma.

[0056] For simplicity, processing apparatus 100 is depicted in Figure 1 A as a standalone station (102) of a process chamber for maintaining a low-pressure environment. However, some fabrication tools employ a plurality of process stations such as shown in Figure IB, which schematically depicts an implementation of a multi-station fabrication tool 150, an example of a process apparatus. Fabrication tool 150 may employ a process chamber 165 that includes multiple fabrication process stations, each of which may be used to perform processing operations on a substrate held in a wafer holder, such as pedestal 108 of Figure 1A, at a particular process station. In the implementation of Figure IB, the process chamber 165 is shown as having four process stations 151, 152, 153, and 154. However, in certain otherDocket No. LAM1P014WO implementations, multi-station processing apparatuses may have more or fewer process stations depending on the implementation and, for instance, the desired level of parallel wafer processing, size or space constraints, cost constraints, etc. Figure IB additionally shows a substrate handler robot 175, which may operate under the control of a system controller 140, and configured to move substrates from a wafer cassette (not shown in Figure IB) from loading port 180 and into multi-station process chamber 165, and onto one of process stations 151, 152, 153, and 154.

[0057] As depicted, process station 153 has an associated sensor 121 located and configured to obtain data from within process station 153 and, in some embodiments, from within process chamber 154. Process station 151 has two associated sensors 123 and 125. Sensor 123 is located and configured to obtain data from within process station 151 and, in some embodiments, from within process chamber 152. Sensor 125 is located and configured to obtain data from within process station 151 and, in some embodiments, from within process chamber 153. Process station 152 has an associated sensor 127 located and configured to obtain data from within process station 152 and, in some embodiments, from within process chamber 154. Any one or more of sensors 121, 123, 125, and 127 may be optically coupled to the interior of process chamber 165 via a viewport or other window disposed in or on the chamber wall. Additionally, while not shown in Figure IB, some embodiments may have one or more sensors adjacent to process station 154.

[0058] Fabrication tool 150 may include a system controller 140 configured to control process conditions and hardware states of the fabrication tool 150. System controller 140 may interact with one or more sensors, gas flow subsystems, temperature subsystems, and / or plasma subsystems — collectively represented as block 141 — to control process gas flow, thermal conditions, and plasma conditions as appropriate for controlling a fabrication process. System controller 190 and subsystems (represented as block 141) may act to implement a recipe or other process conditions in the stations of process chamber 165.

[0059] In multi-station fabrication tools, an RF signal generator may be coupled to an RF signal distribution unit, which may be configured to divide the power of the input signal into, for example, four output signals, which may correspond to four respective process stations. Output signals from an RF signal distribution unit may possess similar levels of RF voltage and RF current, which may be conveyed to individual stations of a multi-station fabrication tool.

[0060] Figure 1C provides a top view of an electronic device fabrication system 182 having four quad-station fabrication tools 188, 189, 193, and 195. Each quad-station tool contains four process stations, each configured to hold and process a substrate. At the front end of systemDocket No. LAM1P014WO182 are three Front Opening Unified Pods (FOUPs) 183a, 183b, and 183c accessible by a frontend wafer handling robot 185, which may be configured to transfer wafers between the FOUPs and a load lock 187. A first wafer handler 190 may be located and configured to transfer wafers between load lock 187 and quad-station fabrication tools 188 and 189. Wafer handler 190 may also be configured to transfer wafers to second load lock 191 that makes wafer available to quad-station fabrication tools 193 and 195 via a second wafer handler 192.

[0061] In an example, quad-station tool 195 may include three sensors 196, 197, and 198 disposed around its outer wall. The sensors are shown vertically affixed to three sides of the four-sided chamber of tool 195. The only side without a sensor in this example is the side next to the wafer handler 192. While not shown in Figure 1C, similar sensor arrangements can be provided on any one or more of each of the three other quad station chambers in the system. It should be understood that, in some cases, a system controller may be configured to modify a position or orientation of a given sensor (e.g., move up, down, left, or right, or rotate with respect to an axis including a central axis).

[0062] Figures IE and IF schematically illustrate fabrication tools with sensors oriented to capture information from a horizontally directed and a vertically directed line of sight, respectively. Figure IE shows a process chamber 170 having a chamber wall 171, a showerhead 172, and a pedestal 173, all designed and constructed in any manner that is known in the art. Figure IE also includes a sensor 174 arranged to capture information relating to the interior of process chamber 170 via a viewport or a window 176 designed for camera access. Traditionally sensed data may include, e.g., images from a visual or optical sensor, spectral data from a spectral sensor, temporal data from a photodiode. However, in some embodiments described herein, captured information may include event-based data, which will be described in more detail below. Sensor 174 may have or be set to capture a field of view defined by edges 177 and 178. As illustrated, sensor 174 and window 176 are arranged to allow the sensor 174 to capture information relating to plasma and / or images including the vertical edge of pedestal 173. Sensor 174 may be configured to adopt various fields of view, angles, zoom levels, and other parameters to capture the desired location or component within the process chamber 170. Other arrangements or configurations (including sensor 174 or window 176) may permit the sensor 174 to capture information relating to other vertical edges in the process chamber, including, e.g., thermal information or thermal images.

[0063] Figure IF shows a similar process chamber 181 but with a camera having a vertical line of sight to allow capture of information relating to other features. Specifically, chamber 181 may include a chamber wall 179, a showerhead 184, and a pedestal 186, all designed andDocket No. LAM1P014WO constructed in any manner that is known in the art. Figure IF also includes vertically oriented sensors 194 and 194’ arranged to capture information relating to the interior of process chamber 181 via viewports or windows 199 and 199’ designed for camera access. Each of sensors 194 and 194’ may be an example of sensor 174 of Figure IE and configured similarly. Sensor 194 may have or be set to capture a field of view defined by edges 161 and 162, while sensor 194’ may have or be set to capture a field of view defined by edges 163 and 164. As illustrated, sensor 194 and window 199 may be arranged to allow the sensor 194 to capture plasma information and / or images including an edge of a stem and / or lower side of pedestal 186. As illustrated, sensor 194’ and window 199’ may be arranged to allow the camera to capture plasma information and / or images including an edge of a stem and / or backside of showerhead 184. Other arrangements may permit a sensor to capture information relating to other edges in the process chamber.

[0064] Figure 1G illustrates an arrangement for attaching a sensor to the wall of a fabrication tool. The view is from the inside of a chamber, which is shown in cross-section. As shown, a sensor 166 may be attached to a wall of a process chamber 167. Sensor 166 may be configured and oriented to capture images of the interior of process chamber 167 via a viewport 168. Sensor 166 may be disposed in a protective enclosure 169 also attached to process chamber wall 167.

[0065] Alternatives to the arrangements shown in Figures 1A, IB, 1C, IE, IF, and / or 1G, may include (a) sensors on corners of chambers (not necessarily on flat walls) and / or (b) light pipes (rigid or flexible) or optical fibers disposed within the reactor (e.g., between one or more viewports or windows in a chamber wall and one or more locations within a chamber interior). Further, it should be understood that systems employing one, two, or more sensors are not limited to quad-station chambers or even multi-station chambers. As illustrated in the figures, in certain embodiments, a fabrication tool or a station in a fabrication tool may be outfitted with more than one sensor. In some cases, a fabrication tool or station has 3 or more sensors, or 5 or more sensors, or 8 or more sensors, or 10 or more sensors. In some embodiments, a station of a multi-station tool may have 1 to 3 sensors. In some embodiments, 2, 3, or more sensors may share a single window or viewport. In some chamber designs, there may not be a viewport, or there may be insufficient viewports to accommodate all sensors. In such cases, chamber designs may include a wall region to accommodate one or more sensors. In general, there can be any of various combination of a single sensor and / or multiple sensors and / or lighting systems that can be placed at different locations, including any of various viewports or other windows.Docket No. LAM1P014WO

[0066] The individual sensors of a multi-sensor tool or station may be positioned and configured to capture different fields of view within the tool or station interior. In some implementations, different sensors may be located and oriented to capture images of the tool interior at different angles. In some implementations, different sensors are located and oriented to capture information relating to the tool interior at different translational offsets. In some cases, multiple sensors oriented in such manner may, for example, be arranged to share a single window or viewport. In some embodiments, sensor analysis logic is configured to stitch or otherwise combine images from two or more individual sensors located and oriented to capture different regions and / or angles within a tool interior.

[0067] As indicated, in embodiments employing multi-chamber fabrication tools, one or more sensors may be located and oriented to capture information about two or more chambers. This may be convenient when two or more stations are along a line of sight from a viewport or other window outfitted with a sensor. In some implementations, sensor analysis logic is configured to use information about structural features of adjacent stations such as station walls, or showerheads, or pedestals, to provide context or frame of reference for plasma radiation data collected from a different station.

[0068] In some cases, a fabrication tool may include a lighting system configured to illuminate all or one or more portions of the tool interior. In some implementations, a lighting system may be configured to allow a sensor to take an illuminated image when the plasma is off (e.g., outside operation or in between pulses). Note that in some cases no lighting system may be employed and lighting may be used from the plasma itself. In some implementations, a lighting system may employ one or more light-emitting diodes (LEDs) or other light sources. The light sources may be monochromatic, polychromatic with discrete emission wavelengths, or broad spectrum. The light source may be active continuously, pulsed synchronously with one or more camera shutters, pulsed asynchronously with one or more camera shutters, or pulsed synchronously with other process parameters such as RF generators or gas delivery valves. In other implementations, multiple light sources may be employed in different locations within or outside the chamber. These multiple light sources can be energized continuously or sequentially with timing managed to enable structured lighting to be utilized to construct superresolution images of features within the chamber. In some implementations, one or more notch or bandpass filters may be provided in front of a light source to produce effects that can support analysis (e.g., identification of particular chemical species by their emission spectra).

[0069] A sensor is typically disposed outside of a fabrication tool, although in some embodiments, it may be integrated with a chamber wall or other component or assembly withinDocket No. LAM1P014WO the chamber. In certain embodiments, a window specially constructed for one or more sensors may be integrated into a chamber wall. In some cases, a window may be constructed to permit a lighting system to shine light on the chamber interior to thereby allow the sensor to capture images of illuminated chamber components. In certain embodiments, a sensor may be optically coupled to an interior of a fabrication tool using an existing viewport that is provided in a chamber wall to allow visual inspection of the tool interior.

[0070] In some embodiments, a sensor may be directly attached to a wall or window of a fabrication tool. A sensor may be affixed to a fabrication tool by various mechanisms such as an adhesive, a bolt or other mechanical fixture, a magnet, etc. In some embodiments, a sensor is disposed at a remote location from a fabrication tool. For example, a sensor may be optically coupled to a viewport via a fiber or other light conduit. Some embodiments allow for a sensor to be mounted in a protective enclosure within a fabrication tool interior. In some embodiments, a sensor has an associated cooling system or thermal management device or component. Examples of thermal management elements include an insulating material (e.g., rubber gaskets), one or more heat dissipative structures, a flowing liquid heat exchanger, etc.

[0071] A viewport or other window for allowing a sensor to “view” a chamber interior may be made from any of a variety of materials. Examples include UV fused silica, UV fused quartz, sapphire, borosilicate glass, and calcium fluoride. In other embodiments, laminates or composites of multiple materials may be utilized in fabricating the windows. In certain embodiments, the window may be substantially transmissive over a spectral range of about 100- 6000 nm or about 100-1000 nm. To make such a wide spectral range usefully available to a sensor, a commercial sensor may need to be modified by removing one or more wavelengthspecific or wavelength range-limiting filters on the sensor as manufactured or sold. In some embodiments, a window may include anti-reflective coating to avoid glare from accompanied illumination in the system described elsewhere herein.

[0072] A viewport or other window for allowing a sensor to view a chamber interior may have any of various sizes and shapes. In certain embodiments, a window has a circular, elliptical, rectangular, or polygonal shape. In some embodiments, a window in a chamber wall is constructed as (or including) an optical element such as a mirror, lens, filter, polarizer, or grating. In some embodiments, a window is integrated with a mirror. Some embodiments further may include optical mirrors or other optical components not integrated with the window but rather located within the chamber to, e.g., enable optical access to regions of the chamber that do not have line-of-sight to a viewport or window. In certain embodiments, the window may be a cylindrical piece of sapphire. In certain embodiments, the window may be coatedDocket No. LAM1P014WO with one or more antireflective films. In certain embodiments, a window may have a maximum cross-sectional dimension (e.g., a diameter or diagonal) of about 5 cm or less, or about 5mm or less.

[0073] In some embodiments, optical elements permit indirect optical information to be captured by a sensor. “Indirect” optical information may include image information outside a line of sight of a sensor. Indirect optical information may be reflected, refracted, scattered or otherwise directed from its source, which is outside a field of view of the sensor, to a position inside the field of view of the sensor. To this end, a fabrication tool may include a mirror or other optical element configured to direct light or other optical information into a sensor’ s field of view. In some cases, the optical element is part of a process chamber that has a non-optical function. For example, an aluminum component such as an aluminum chamber wall may reflect light on the IR region of the electromagnetic spectrum.

[0074] Viewports have been observed to produce thermal and electrical anomalies. Therefore, eliminating viewports and replacing them with small dimension windows may provide benefits to the fabrication tool processing environment in certain embodiments.

[0075] One or more sensors (e.g., visual or optical sensor such as a camera) may be arranged to provide a multiplexed processing of images. In some embodiments, a single remote sensor may process optical information (and generate images) from multiple locations (e.g., multiple viewports). For example, a single camera sensor may support multiple viewports. In some embodiments, a fabrication tool may employ one camera sensor to capture image data from two more stations of a multi-station chamber. For example, a tool may have a first window on a chamber wall adjacent a first station and a second window on the chamber wall adjacent a second station. The tool may additionally include a first optical fiber or light pipe optically coupling the first window to a camera sensor and a second optical fiber or light pipe optically coupling the second window to the camera sensor. The camera sensor may be configured to multiplex signals from the first and second optical fibers or light pipes. In some embodiments, the tool includes an array of light pipes and / or an array of optical fibers for conveying optical signals between a source in the tool and a camera sensor.

[0076] Images or video clips from one or more camera sensors may be processed in a multiplexed fashion by image analysis logic running on hardware at any of various locations. In some embodiments, this approach is applied to study states of a fabrication tool or conduct other evaluation outside of real time.

[0077] In some embodiments, one or more fabrication tools and associated sensors may have local edge computers. An edge computer may be configured to execute programs forDocket No. LAM1P014WO processing and / or managing sensor data. Examples of such programs include image analysis programs and image / video multiplexing programs. In certain embodiments, an edge computer may contain a program for multiplexing image / video data from one or more sensors at a fixed rate. In some cases, an edge computer may be configured to execute one thread for multiplexing video / images from multiple sensors and execute a different thread for analyzing images and / or video. An edge computer may run separate virtual machines for its various responsibilities associated with sensors.

[0078] In certain embodiments, an edge computer may be provided for a single station, but the computer is configured to multiplex and / or otherwise process data from multiple sensors, some of which may not be located at the single station.Characteristics of Cameras for Sensing Plasma Conditions

[0079] Camera sensors and other visual / optical sensors are characterized by various parameters including the number of pixels, range of wavelengths captured, and the like. In some embodiments, a camera sensor for capturing information about a plasma may be capable of sensing intensity values of electromagnetic radiation at wavelengths including at least portion of the UV spectrum, at least a portion of the visible spectrum, at least a portion of the IR spectrum, or any combination thereof. As an example, a camera sensor may be configured to sense intensity values over range including about 100 nm to about 1000 nm.

[0080] As examples for any embodiments herein, camera sensors may be constructed as charge-coupled devices (CCDs) or complementary metal-oxide- semiconductor (CMOS) arrays. In certain embodiments, a camera sensor as used herein may have at least about 5 megapixels or at least about 12 megapixels. In some embodiments, a camera sensor used herein may have as few as about 2 megapixels. In some embodiments, a camera sensor used herein may have as many as about 20 megapixels, 200 megapixels, or more.

[0081] In some implementations, an image capture device may be a line or one-dimensional array of sensors or pixels. Such device may be configured to scan across a two-dimensional field of view. The scan direction may be substantially perpendicular to the axis of the line of sensors. In some embodiments, a one-dimensional image capture device may be oriented perpendicular to a wafer or chamber component and optionally configured to scan from one side of the chamber to the other (or within other portion or field of view within the chamber).

[0082] In certain embodiments, a camera as used in any of the embodiments herein is configured with a shutter. In some implementations, a camera may be configured to capture video data of a plasma (or other phenomenon or event) in or at a fabrication tool. In certain embodiments, a camera may be configured to capture video information of a plasma (or otherDocket No. LAM1P014WO phenomenon or event) in or at a fabrication tool at a frame rate of about 30 to about 120 frames per second (FPS). In certain embodiments, some types of sensors and cameras may be configured to capture video information of a plasma (or other phenomenon or event) in or at a fabrication tool at a frame rate of at least about 500 FPS, at least about 10,000 FPS, or at least about 1 million FPS. In certain embodiments, a sensor may be configured to capture video information at such frame rates of a plasma (or other phenomenon or event) in or at a fabrication tool dynamically, in which frames or images are captured based on certain events or conditions, rather than at fixed shutter or capture speeds. Resulting images and videos may have an effective frame rate of at least about 500 FPS, 1,000 FPS, 10,000 FPS, 1 million FPS, or more. A sensor or camera may thus be capable of capturing frames over a short period of time, e.g., an accumulation time of optical signals of 10 nanoseconds (ns). Various accumulation times may be possible, e.g., up to 10 ns, as low as 100 ns, or over 1 second. The term “effective frame rate” or “effective FPS” or “effective sampling rate” may, in the context of the present disclosure, refer to capturing of frames where a change occurs in a phenomenon (e.g., a rapidly changing phenomenon) as opposed to continuously capturing images at a consistent frame rate. Compare timeline 200 representing a fixed rate with timeline 300 representing an event-based frame rate, as described with respect to FIGS. 2 and 3. In specific cases, the aforementioned terms may refer to a sampling rate or a quantity of frames captured over a given time period based on occurrence of events, parameters (e.g., light intensity, voltage) being above a threshold (e.g., intensity threshold, voltage threshold, dV / dt threshold), or changes in said parameters, rather than a consistent rate with a consistent shutter speed. In some approaches, changes in pixels may be accumulated over an accumulation time, which can be extrapolated to a per-second basis.

[0083] Some fabrication tools may include a still image or video display. Such display may be employed to allow process engineers or other staff to view the too interior when a camera sensor or light conduit blocks access to a viewport from outside the tool. In some embodiments, a view to a chamber interior is provided via images or video streamed electronically (e.g., using the real-time streaming protocol (RTSP), the real-time messaging protocol (RTMP), low- latency HTTP live streaming (HLS), secure reliable transport (SRT), WebRTC, or the like), optionally to a remote location via a web application. Examples of remote sites include a fabrication monitoring room or facility, a smart phone, a tablet, and / or a desktop computer system. In some embodiments, communication of the image(s) or video may be made via network that includes, as a node, a camera on a process chamber. Such network may be wiredDocket No. LAM1P014WO or wireless, e.g., a mesh network. In certain embodiments, a network employs a protocol employing Wi-Fi, Bluetooth, cellular, etc.Other Sensor Types that May Be Used, in Conjunction with Cameras

[0084] In some embodiments, a fabrication tool may include one or more sensors in addition to the sensors discussed above. Such additional sensor(s) may be configured to sense a plasma or other conditions in situ. Such sensors may include, but are not limited to, mechanical limit sensors, inertial sensors (e.g., accelerometers or gyroscopes), infrared (IR) sensors, acoustic sensors, mass flow sensors, pressure sensors such as the pressure manometers, and temperature sensors such as thermocouples, which may be located in a process gas delivery system, a pedestal, a chuck, etc. Specific examples of additional sensors include current sensors (e.g., VI probes), which may be affixed to one or more structural components such as a showerhead or pedestal, an in situ spectroscopic sensor configured to capture emitted radiation from a wafer or reactor component in the UV, visible, and / or IR spectrum (e.g., an optical emission spectroscopy sensor (OES)), an in situ sensor configured to detect optical absorption characteristics of gases in the process chamber, an in situ optical metrology tool such as a reflectometer.

[0085] Another example of an additional sensor may be a capacitive voltage sensor having a relatively high input impedance. Another example of an additional sensor may be an inductive current transformer having a relatively low input impedance that occasionally or periodically samples a current conducted from an RF signal generator without bringing about any significant voltage drop. In some embodiments, a current or voltage sensor may be coupled in series between RF signal generator and a multi-station fabrication tool.Camera Image Analysis

[0086] Image analysis logic may be configured to receive sensed values from one or more camera sensors or other visual or spatial sensors (or in some embodiments, optical sensors, such as the event-driven sensor discussed herein) on a fabrication tool. In certain embodiments, inputs to the image analysis logic may include pixel-by-pixel intensity values as a function an observable parameter such as wavelength, time, location, polarization, or any combination thereof. In certain embodiments, input data from a camera sensor may be provided in the form of image data, video data, spectral values, time series data, wafer metrology data, or the like. In some embodiments, the input data may be filtered by wavelength, polarization, etc. In some embodiments, analysis logic may be configured to receive and act on additional input information beyond camera sensor intensity data. Such additional input information mayDocket No. LAM1P014WO include metadata about the camera sensor and / or associated camera components, substrate metrology information, historical information about the fabrication tool, etc.

[0087] The analysis logic may be configured to output one or more properties of plasma in a fabrication tool and / or a classification of a state of the fabrication tool or a component thereof. Some examples of plasma properties were presented above. Some examples of a state of the fabrication tool or a component thereof may include temperature of a heat jacket, gas delivery line, or a heating element. In some embodiments, the analysis logic may be configured as a classifier for diagnostic purposes, or for predictive purposes, or for control purposes. Some examples of diagnostic classifications may include fault detection and anomalous conditions. Some examples of predictive classifications may include process or mechanical drift (e.g., a varying shape of a showerhead or other component) and associated predictive maintenance (generated by, e.g., regression analysis). Further examples are provided in International Application No. PCT / US2021 / 058550, entitled “PREDICTIVE MAINTENANCE FOR SEMICONDUCTOR MANUFACTURING EQUIPMENT” and filed November 9, 2021, which is assigned to the assignee hereof and incorporated herein by reference in its entirety. Examples of control classifications may include recommended modifications to apparatus or processes.

[0088] Camera sensor analysis logic may comprise any of various types of classifiers or models such as deep neural networks (e.g., convolutional neural networks, autoencoders, UNet, etc.), traditional or classical computer vision methods such as edge detection, image modification (such as blurring, changing contrast), intensity thresholding, color channel thresholding, etc.

[0089] Analysis logic may be configured to perform an image processing routine such as a segmentation or other edge finding routine. In certain embodiments, analysis logic may be configured to use segmentation or other edge detection method to determine plasma property information relative to a system component. Segmentation can isolate the component. For example, identification of showerhead and detection of plasma properties some distance away from it may employ segmentation. In the field, when mechanical mounting may not always be consistent, segmentation can help minimize errors (as opposed to fixing the (x, y) location for all images / video generated by a fleet of tools).

[0090] The logic may employ any of various techniques for edge detection or segmentation. For example, the logic may employ a threshold-based method, a deep learning model, etc. In some embodiments, the edge of a plasma or the boundary of a subregion within a plasma having defined plasma characteristic may be determined using a processing sequence such as theDocket No. LAM1P014WO following: (a) data reduction, (b) denoising (e.g., gaussian blur), and (c) edge finding / thresholding (e.g., a Canny sequence of filter).

[0091] In some embodiments, image analysis logic may be configured to determine a location of a plasma based on, at least partly, a centroid of a region occupied by the plasma. A plasma’s centroid may be a geometric centroid determined from a region deemed by analysis logic to be within boundaries of the plasma. In some cases, the centroid may be calculated by considering the intensity of radiation within an image of the plasma. This may be implemented by weighting pixels or regions of an image based on, at least, the intensity values of the pixels or regions. A plasma’s centroid may also be determined by simply applying an intensity threshold to the pixel or region values, and only considering those pixels or regions having intensity values above the threshold when calculating the centroid.

[0092] Analysis and / or control logic may employ a plasma centroid to determine the alignment or tilt of the process chamber, the showerhead, the pedestal, and / or another component. For example, the logic may compare the centroid position of a process chamber under consideration to an expected or baseline centroid position for a properly aligned process chamber or process chamber component. If the centroids do not agree to within a defined tolerance, the logic may flag the current system as being out of alignment.

[0093] In some embodiments, image analysis logic may be configured to determine a location of a plasma by integrating or summing optical intensity values over a bounded region of interest within a field of view of a camera sensor. Examples of regions of interest include a gap between a pedestal and showerhead or a subregion of the gap such as shown in the figures.

[0094] In some cases, analysis logic may be configured to determine a plasma’s location based on, at least, a point or boundary of a plasma having a defined spectral characteristic. The defined spectral characteristic may be a region of the electromagnetic (EM) spectrum associated with a gas or component in the process chamber. The spectral region may be associated with an emission spectrum and / or one or more emission lines of a gas or component in the process chamber. A spectrally limited location of the plasma may be used to identify the composition and / or location of a gas within a process chamber or a station thereof. For example, if gas of a particular composition is to be located only above a showerhead and it is found to be located in the gap between a pedestal and showerhead, the analysis logic may flag the process chamber or the current process as needing inspection or modification. Moreover, in some cases, the analysis logic may be able to capture an imbalance of gas at the showerhead at the point of plasma ignition, as well as the transient period before a steady state in plasma is achieved. The analysis may be able to indicate whether any detected imbalance is time bound,Docket No. LAM1P014WO which may allow transient behaviors to be better understood and optimized more quickly. For example, the analysis logic may be used to determine a uniformity of plasma during the initial plasma ignition phase before steady state is reached, for example, whether the showerhead lights up uniformly or whether there is a transient effect where a first region of the showerhead lights up prior to a second region of the showerhead. Moreover, the analysis logic may be used to determine whether any detected transient effects are reproducible over several cycles or process runs. For example, analysis of the transient behavior may allow for process parameter changes, compensation, and / or a diagnosis of hardware that is malfunctioning (e.g., to replace the hardware, etc.).

[0095] In certain embodiments, analysis logic may be configured to perform streaming data analysis. As viewports are replaced or integrated with sensors, engineers will require a “window” to the fabrication tool interior. In certain embodiments, an image or video processing system is configured to provide access to live stream of data which may be accessed through a fabrication tool’s computer, local wireless streams to cellphone apps, and other intranet channels accessible via, e.g., a browser. In some embodiments, analysis logic is configured to perform analytics such as component segmentation and classification in real time. As an example, analysis logic may be configured to flag faults such as a stuck pin in a wafer support. The logic may ensure that such issues are addressed before, e.g., processing another wafer.

[0096] In certain embodiments, analysis logic may be configured to perform fixed frame (image) and / or video analysis. As explained elsewhere herein, the analysis logic may be configured to analyze and interpret where static image or multiple temporal frames depending on the use case.

[0097] In certain embodiments, analysis logic may be configured to with edge computing capabilities. To minimize network traffic, at least some computation may be performed on an edge node (or multiple edge nodes) and only limited data is transferred to remote computational or memory resources (e.g., remote storage databases). Decisions on control, feedback, warnings, and the like may be based on computing capabilities on the edge nodes.

[0098] In certain embodiments, analysis logic may be configured to perform feed forward and / or feedback for multiple applications (process control, auto calibrations, hardware adjustments, etc.). In certain embodiments, computational resources may be configured to feed results of a camera analysis to a controller of process conditions at a subsequent (downstream) tool. For example, analysis may be conducted on plasma images captured while a wafer is being processed in a single station plasma tool. The results of analysis may be used to control conditions in an adjacent module for additional deposition. As another examples, informationDocket No. LAM1P014WO on non-uniformity on station 1 can be used to compensate process conditions at station 2, 3 or 4 (or another), or in the current or subsequent process steps.

[0099] In certain embodiments, analysis logic may be configured to perform multiplexing and / or stitching of images. Multiplexing (e.g., multi-threaded processing) may allow a single processor to process images from more than one camera. Multiplexing may allow a single processor to handle images from multiple wavelength ranges. In certain embodiments, multiple cameras may be used to generate not only a rich stream of temporal data but also spatial locations from multiple perspectives. While there may be some delays due to switching from camera to camera, for some applications, analysis logic can generate a combined image capturing more than just the field of view of one camera. In some embodiments, analysis logic may be configured to reconstruct three-dimensional information by combining additional sensors and cameras (and optionally modeling data).Traditional Imaging & Event-based Imaging

[0100] In traditional imaging approaches, an imaging sensor (such as a spatial sensor, a visual sensor, or an optical sensor) is configured to collect light or other types of electromagnetic (EM) signals under certain exposure conditions. In some configurations, the imaging sensor may include one or more photosensors or photodiodes, such as a charge-coupled device (CCD) capable of converting optical signals into electrical signals and read out as voltage signal on a per-pixel basis over accumulation times, providing spatial resolution and temporal resolution.

[0101] In traditional imaging approaches, capturing EM signals may be based on preprogrammed or preset shutter speeds and / or exposure times. The light may also interact with the light sensor and generate pixels based on preconfigured lens, aperture, and / or filter arrangements. For a video, multiple frames are captured at a fixed frame rate. That is, a shutter may open and close at a fixed rate, as illustrated in FIG. 2, which is a simplified diagram of a traditional imaging scheme with a fixed frame rate. Example frames 1, 2 ... N are depicted. A timeline 200 represents an example shutter opened and closed (or opening and closing) at times tl and t2, respectively, to capture visual information for a first frame, the example shutter opened and closed at times t3 and t4 to capture visual information for a second frame, etc. The length of time between tl and t2 (x) may be equal to the length of time between t3 and t4 (x) and to subsequent shutter open exposure times. Similarly, the length of time between t2 and t3 (y) may be equal to the length of time between t4 and tn (y) and to subsequent shutter closed times.

[0102] However, there are limitations to traditional imaging. For example, frames are captured at fixed intervals with full resolution at the prescribed setting regardless of any meaningfulDocket No. LAM1P014WO change in the image. That is, all information is captured without discriminating saliency or conditions of the object or view being captured. This results in high data requirements, high network requirements (e.g., if streaming or transferring captured frames or other image data), and high computational requirements for downstream applications (e.g., deep learning, machine learning, computer vision, big data analysis, image manipulation). Another example of a limitation is that overexposure and occur because of a limited dynamic range. Overexposure can increase saturation and “wash out” objects in the image, making it difficult or impossible to identify or observe certain portions or objects or phenomena in the image. Although overexposure can be overcome by algorithms with high dynamic range capabilities, this approach still has limitations and can limit frame rates. As another example, autoexposure can impact image quality, which can in turn lead to errors in downstream applications. As another example, a moving object could suffer from motion blur because of finite imaging capabilities or frame rates. These limitations make it difficult or impossible to capture shortlived or rapidly changing phenomena. Typical cameras lack the ability to detect valuable information, including transient phenomena and rapidly changing phenomena such as electrical arcs and other anomalies.

[0103] In the context of the present disclosure, a phenomenon or event that is “rapidly changing,” “rapidly occurring,” “transient,” “instantaneous,” or “nearly instantaneous” may refer to a phenomenon or event that change state in a very short period of time, e.g., within 5 nanoseconds (ns), 10 ns, 100 ns, 1 microsecond (ps). Some such phenomena may last in the millisecond (ms) range, for example, 1 to 20 ms in the case of a hollow cathode discharge (HCD), and hence may also be considered a transient phenomenon. Some such phenomena can involve changing light emission and intensity of light. Examples can include an electrical arc, a plasma discharge, a plasmoid formation or movement (e.g., moving toward an edge of a showerhead), a HCD, illumination of a light source, or the ending thereof. Some other examples can include flow of wet chemistries and liquids (including droplets), vibration different from expected motion (e.g., expected movement from point A to B), unexpected wafer movement in the system (e.g., due to pressure change). Not all rapidly changing phenomena are transient and lasting for a very short period of time; however, all transient phenomena are rapidly changing. It is typically infeasible to capture such phenomena using traditional imaging based on fixed frame rates, even at high rates such as 1 million FPS, and would be prohibitively resource intensive for monitoring applications (e.g., to detect and visualize an electrical arc when it happens), as it would require constant recording and storage of images, especially at very high frame rates. Other examples of the phenomena or events discussed herein can includeDocket No. LAM1P014WO rapid (or non-rapid) motion of an object (or even a person), for example, of a physical hardware component, e.g., vibrations of a component (e.g., a valve such as an ALD valve), or opening and closing of a valve. In such cases, the phenomenon or event can have sufficiently visually detectable changes in position (vibrations may be up to 500 Hz, for example) to be captured by traditional imaging. However, it can still be costly to monitor and inexact when using visual imaging or traditional computer vision to analyze and study minute changes in position (e.g., during vibrations), and in some cases, determine an action to take with respect to the component or an apparatus or system using the component.

[0104] For example, electrical plasma arcs and other transient anomalies can last a very short time (e.g., on the scale of microseconds (ps) or even on the scale of nanoseconds (ns), e.g., under 100 ns, under 50 ns, under 10 ns, under 5 ns) and often have higher signal intensities than other objects (e.g., plasma) or events that can be present within a processing apparatus. It may result in waste of resources to try to capture these phenomena that are not reliably detectable. Moreover, traditional cameras would not be able to detect them with accuracy, or at all. However, equipment and wafer damage and waste could be prevented if such elusive phenomena were detected early.

[0105] To overcome these limitations and obtain meaningful information, including transient phenomena and rapidly changing phenomena, event-driven cameras and sensors may be used in the disclosed embodiments, implementations, and examples. Such event-driven cameras may detect changes with respect to a previous frame or a set of frames. As such, event-driven cameras can overcome the aforementioned saturation problem that cameras with set exposure times are susceptible to. Event-driven cameras can also detect changes according to set process condition or parameters needed. One example application would be to gain a deeper understanding of plasma ignitions by capturing transient regions during plasma ignition, and obtain better cycle-to-cycle and station-to- station control. In addition, other types of changes can be observed. For example, physical movements or health of a component (e.g., valve, chuck, vacuum clamp), plasma motion, formation of plasmoids, fluctuations in plasma, and changes occurring at a surface of a process chamber could be observed. Applications are not limited to short-lived phenomena and anomalies but also minute variations in process conditions that would otherwise be undetected by a typical camera or the human eye.

[0106] FIG. 3 is a simplified diagram of an imaging scheme with a dynamic, event-based frame rate. A timeline 300 represents various example frames 1 through N captured over time. Frames 1, 2 and 3 may be captured consecutively at a consistent rate with a time period of y between captures. Frames 1, 2 and 3 may be captured based on some event occurring at a region orDocket No. LAM1P014WO space (e.g., at or within a processing apparatus) that is observed by a spatial sensor or an optical sensor (e.g., an event-driven sensor of the type discussed herein).

[0107] In some embodiments, a specialized, event-driven sensor configured to perform ultrafast sensing via frames captured over a short period of time (e.g., up to 10 ns, 100 ns, 10 ps, 100 ps, 1 ms, 10 ms of accumulation time) may be used to obtain optical signals over a short accumulation time per frame and produce electrical signals having voltage. Examples of an event may include an electrical arc, motion of a component (e.g., vibration, opening and closing of a valve), a plasma ignition, plasmoid formation or movement, a plasma discharge, a HCD, or other transients or anomalies. Other examples of an event may include a ceasing of the foregoing transient phenomena. That is, the end of a phenomenon may also be considered an event and detectable using the frames, as will be further described below. In fact, starting and stopping of an event may both be treated as being equally important as they can be used to control a processing apparatus. As an example scenario, detection of an anomaly may cause the start of control (e.g., a controller may signal the system with instructions to start or end a process), or cause a change to the system state (e.g., via instructions to modify an ongoing process or hardware settings) to adjust to a different system state and / or reach a steady state. Resolution of the anomaly may cause the end of the control, e.g., by stopping aforementioned instructions or sending new instructions to stop, slow, or reverse the modifications. As alluded to above, the end of a phenomenon may be detected with at least equal importance.

[0108] More specifically, in some embodiments, frames 1, 2 and 3 in this example may be captured based on optical signals or a parameter thereof meeting certain conditions. In some implementations, intensity of the optical signals obtained via the sensor may be above an intensity threshold. In some implementations, a change in voltage (dV / dt) associated with the optical signals may be above a dV / dt threshold. Other types of parameter and thresholds may be implemented according to the desired application, e.g., voltage level, sensor output offset or bias, slew rate (maximum rate of change of voltage). In some cases, a parameter and threshold may be determined statistically, e.g., a mean value, a weighted mean, a median value, a maximum value, a minimum value, a standard deviation, or other relevant derivations.

[0109] In some approaches to thresholding, debounce, hysteresis, and / or the time horizon (event width) may be considered. Debounce refers to a technique used to ensure that a signal or a parameter has consistently crossed a threshold before it is considered valid. This can eliminate false positives caused by noise or rapid fluctuations around the threshold. Debouncing may involve waiting for the signal or parameter to remain in a stable state above or below the threshold for a certain period of time (event width) before confirming that theDocket No. LAM1P014WO signal or parameter has met the relevant threshold. Hysteresis refers to introducing two different thresholds for switching states: one for when a signal is rising and another for when a signal is falling. This creates stability with a buffer zone that prevents rapid switching back and forth caused by noise or small fluctuations. In some implementations of hysteresis, a higher threshold for determining that a signal has met or exceeded the threshold and a lower threshold for determining that a signal no longer meets the threshold it off (or vice versa) may be defined.

[0110] In some embodiments, frames 1, 2 and 3 in this example may be captured based on the existence of optical signals or a parameter. For example, if certain pixels of an image have a non-zero value of optical intensity, it may be indicative of some level of light being present at that location of the image. In such embodiments, this may be sufficient to trigger a frame capture.

[0111] In some embodiments, a parameter associated with a frame may be compared against a corresponding parameter of another frame. For instance, in the example of FIG. 3, light intensity associated with frame 1 may be compared against light intensity associated with frame 2. Based on the change in light intensity being above a threshold, at least frame 2 may be captured. In some cases, frame 1 may also be captured as an initial point of reference or based on comparison with a previous frame. Association with a frame in this case may refer to, e.g., at a pixel or at least portion(s) of the frame, or based on statistical determination.

[0112] Returning to timeline 300, it can be seen that frame 4 is captured a time period of y’ after frame 3, where y’ is different (longer) from the time period of y. This may be due to similar reasons as the capture of frames 1, 2 and / or 3. For example, no appreciable (e.g., above a threshold) change in parameter may have been determined between optical signal(s) at time tl and time t2. Frame 4 may have been captured based on a condition being met with respect to parameters of frames 3 and 4, similar to parameters and conditions described above. Similarly, frame N is captured a time period of y ” after frame 4, where y’ ’ is different (shorter) from the time period of y’ (but longer than the time period y), on similar basis.

[0113] Frames 1, 2, 3, 4 and N can thereby be captured based on events, rather than at a constant capture rate as in traditional imaging. Incidentally, the length of time periods y, y’ and y” are depicted as varying merely to illustrate that the frames are not captured at a consistent frame rate, but rather, more dynamically based on events meeting criteria, thereby achieving outcomes that overcome traditional limitations and offer advantages as described elsewhere herein.

[0114] FIG. 4A is a flow diagram of an example process 400 for an event-based detection of phenomena using comparison of frames. Examples of phenomena may include short-lived,Docket No. LAM1P014WO transient, rapidly changing, instantaneous (or near- instantaneous) phenomena such as the types described above and elsewhere herein.

[0115] In some embodiments, at 402, a first frame may be obtained. For example, an event- driven sensor (e.g., camera) configured to perform ultra-fast sensing may be used to obtain the first frame.

[0116] At 404, a second frame subsequent to the first frame may be obtained. In addition, the second frame may be evaluated to determine any change with respect to the first frame. In some approaches, a change in a parameter of the second frame relative to the second frame may be determined, e.g., at a pixel or at least portion(s) of the frame, or based on statistical determination. In some implementations, the parameter may be an intensity, voltage, or a change thereof. This change may be evaluated against a threshold.

[0117] As an example, a positive change in voltage associated with a portion of frame 2 (which may be a single pixel) relative to a corresponding portion of frame 1 (which may be a corresponding pixel at the same position of the frame) may be determined. Positive change in voltage (+dV / dt) may correspond to an increase in optical signals received over time, e.g., light emitted from a phenomenon (e.g., electrical arc, HCD, light turning on), or a change in position of an edge of an object which is now detected at a slightly different location (e.g., an adjacent pixel or region of the frame). Even if the phenomenon becomes brighter to the point the light intensity causes saturation and will not allow perception with typical sensors (e.g., CMOS), pixels based on dV / dt will be visible in the frames, with spatial information and temporal information, as discussed herein. In some scenarios, +dV / dt may be indicative of the phenomenon starting. In some cases, light from a phenomenon or other light sources may be redirected by a mirror or optical element. In some scenarios, +dV / dt may be indicative of motion of an object. This +dV / dt (where dt may be the difference in capture times of frames 1 and 2) may be compared against a threshold, such as a threshold for a change in voltage. That is, if +dV / dt is above a threshold dV / dt, then the change in voltage would meet the threshold, and an event would be determined based on the occurrence of frame 2. Put another way, a relatively constant voltage where dV / dt is below a threshold or close to zero would mean that there are no changes to the system nor any appreciable phenomena occurring.

[0118] As a contrasting example, a negative change in voltage associated with a portion of frame 2 (which may be a single pixel) relative to a corresponding portion of frame 1 (which may be a corresponding pixel at the same position of the frame) may be determined. Negative change in voltage (-dV / dt) may correspond to a decrease in optical signals received over time, e.g., diminishing or disappearance of light from a phenomenon, or a change in position of anDocket No. LAM1P014WO edge of an object which is no longer detected at a location. In some scenarios, -dV / dt may be indicative of the phenomenon ending (e.g., electrical discharge disappearing, HCD disappearing, light turning off). In some scenarios, -dV / dt may be indicative of motion of an object, e.g., in a different or opposite direction compared to a motion corresponding to +dV / dt. This -dV / dt (where dt may be the difference in capture times of frames 1 and 2) may be compared against a threshold, such as a threshold for a change in voltage). That is, if -dV / dt is below a threshold -dV / dt (or the absolute value of -dV / dt is above the absolute value of the - dV / dt threshold, farther from zero), then the change in voltage would meet the threshold, and an event (e.g., a ceasing of a phenomenon such as the end of an electrical arc) would be determined based on the occurrence of frame 2. Put another way, a relatively constant voltage where dV / dt (including -dV / dt) is below a threshold or close to zero would mean that there are no changes to the system nor any appreciable phenomena occurring.

[0119] In some implementations, responsive to the determination of the event, frame 2 may be stored, e.g., as raw image data or encoded data (e.g., intensity value per pixel). Other frames not meeting the threshold may be discarded. In other approaches, other thresholds may be used, such as a voltage threshold for a voltage value associated with a portion of frame 2, an intensity threshold for an intensity value associated with a portion of frame 2.

[0120] At 406, whether the determined event triggers a failure condition or a warning condition (or other error conditions) may be evaluated. In some embodiments, if the parameter or change in parameter exceeds a second threshold (and / or third, fourth, etc.) that is indicative of an error, failure, warning, or another reason to flag the event, then the example process 400 may proceed to example process 450 (described further with respect to FIG. 4B). For example, if the change in light intensity of the acquired optical signals is too fast (exceeding a second threshold), then it may indicate the presence of an electrical arc (or other anomaly, e.g., HCD) in a process chamber. Another condition for flagging the event may be a number of times that the second threshold is exceeded, e.g., over a period of time or over a consecutive number of previous frames, which may reduce any false positives. If, for instance, the change in light intensity exceeds the second threshold N number of times, then it may indicate a higher confidence that the event is an anomaly.

[0121] On the other hand, if the parameter or change in parameter does not exceed the second threshold, then the example process 400 may proceed to 410. At 410, whether more frames are available may be determine. If so, the example process 400 may return to 404 to obtain a third frame, which may be evaluated to determine any change with respect to the previous frameDocket No. LAM1P014WO(e.g., the second frame). In some cases, the third frame may be evaluated with respect to any previous frame (e.g., the first frame).

[0122] If there are no more frames available, the example process 400 may end.

[0123] If a failure condition or a warning condition or other error is triggered at 406, the example process 400 may proceed to example process 450 shown in FIG. 4B. In some embodiments, a control change to be made may be determined, e.g., stop the ignition of plasma, stop or slow inflow of process gases, stop or reduce provision of RF power, etc.

[0124] At 452, it may be determined whether such a control change is within the physical limits of compensators and components of the processing apparatus or system. It may also be determined whether the control change is able to correct the failure, warning, or error condition. If so, at 454, the control change may be made. In some implementations, the control change may be made on a cycle-to-cycle basis. A controller may send instructions to make the control change for a particular cycle, e.g., after the current one is complete. In some implementations, the control change may be made immediately or during a current cycle, e.g., if the process involves a long plasma generation or exposure time.

[0125] If not, at 456, any ongoing process may be stopped. The detected issues may not be able to be resolved by the system or its components. For example, a physical actuation such as opening or closing a valve may be insufficient to mitigate the error condition. In some scenarios, an electrical arc or a large electrical discharge may be damaging the wafer(s) and / or hardware and cannot be prevented without stopping the process, as opposed to controlling the process.

[0126] Hence, example process 450 may be considered a controlling process, while example process 400 may be considered an observation and reporting process. Example process 450 may be triggered when there is a need to control a process where needed, e.g., depending on whether an error or anomalous event occurs. Rapidly occurring or transient phenomena can be captured at a highly granular frame rate, providing opportunities to monitor and mitigate any failures on a cycle-to-cycle basis, along with station-to-station control by independently controlling selected station(s). Opportunities to provide users and engineers with a deeper understanding of the process (e.g., during plasma ignition) can be provided as well.

[0127] The above general approach can be applied to myriad use cases. In anomaly detection, for example, it can be used for detecting electrical arcs and discharges, which may last merely up to hundreds of nanoseconds or hundreds of microseconds. It can also be used for detecting other system transients such as plasma motion (e.g., expansion), plasmoids, HCDs. The general approach can also be applied to valve operations, where slight motions or vibrations can beDocket No. LAM1P014WO determined, and used to detect, e.g., mechanical degradations based on changes in dampening, open-close cycles or latency time, or vibration frequency. The general approach can also be applied to perform object motion tracking generally. The general approach can also be applied to low-light wafer motion tracking. The general approach can also be applied to perform steady-state plasma analysis, in which there may be small fluctuations in a plasma that would otherwise be undetected by a typical camera or the human eye. This could be used to understand the stability of the plasma by capturing pixel-level event changes, e.g., at least 10,000 events / second. The general approach can also be applied to determining quality of surfaces, e.g., process chamber walls, which can be used to cause performance of surface cleaning or change of surface. The general approach can also be applied to monitor the health of components such as clamps and chucks. The general approach can also be applied to wet chemistry flow sensing, e.g., based on motion of reactants.

[0128] Specific examples of use cases are described below.Arc Detection

[0129] Electrical arcs may occur unpredictably in an instant with or without other plasma in the system. Where arcs occur with no other plasma in the system, an approach that does not require monitoring at a constant frame rate (such as with traditional imaging systems) would advantageously allow detection at a fast temporal resolution with spatial resolution based on when the event (e.g., arc) occurs. While a photodiode may be faster, it does not provide spatial resolution such as the approaches described above and below.

[0130] Where arcs are present with other plasma in the system, fast temporal resolution with spatial resolution may be enabled using the approaches described above and below, as well as an ability to avoid additional thresholding required to filter out background plasma. While a photodiode may be faster, it does not provide spatial resolution and may require additional tuning to differentiate between the arc and background plasma light.

[0131] Accumulation of optical signals and a dynamic frame rate can allow understanding dynamics of the arcing process. An example method for arc detection is provided below.

[0132] FIG. 5 is a flow diagram of an example process 500 for event-based arc detection, according to some embodiments. In some embodiments, at 502, one or more frames may be generated based on one or more accumulation values. Accumulation values may refer to a period of time during which optical signals are received and accumulated by an event-driven sensor configured to perform ultra-fast sensing to generate a frame. For example, optical signals may be accumulated for 10 ns, 100 ns, 10 ps, 100 ps, 500 ps, 1 ms, 10 ms, 50 ms, and / or another length of time. The longer the accumulation time, the more the optical signalsDocket No. LAM1P014WO can be obtained. Different accumulation settings may be appropriate depending on the phenomena being monitored. A lower accumulation setting may be used for very short lived phenomena, such as electrical arcs. However, longer accumulation can be used to determine motion.

[0133] Subsequently, at 504, arc detection may be performed on the generated one or more frames. Various methods may be used to detect an electrical arc in the one or more frames. In some embodiments, an arc may be identified based on machine vision techniques. For example, such algorithm may involve traditional computer vision algorithms such as edge detection, contour generation, or any morphological operations or transforms needed to confirm. In some approaches, arc detection (or determination of the type of other phenomena) may be based on deep learning algorithms or models trained using machine learning techniques such as supervised or unsupervised training, which may involve selection of a training dataset (train set) that includes various training data including frames or portions thereof (or pixels) as well as ground truth frames (ideal expected results), minimization of a loss function, selection of a learning rate and a regularization rate, selection of a portion (e.g., 20%) of the training data as a validation set, and / or adjustments of weights, biases, and other parameters. Such a training dataset may be stored on a non-transitory computer-readable storage medium. The training dataset may then be used by a computerized apparatus or system (e.g., a controller) to train a deep learning or machine learning model. Pre-training and fine tuning may also be performed in some cases. Models used for such deep learning may include neural networks, convolutional neural networks, and / or Generative Adversarial Networks (GANs). Other approaches such as diffusion models or generative approaches may be used as well. In some scenarios, transfer learning may be used for model training between system to system.

[0134] At 506, in some embodiments, the resulting trained model may be configured to determine an event, namely, the presence of an arc, at least within a portion (e.g., region) of a frame. The trained model may also output a likelihood and / or confidence of the likelihood of the presence of the arc. In some implementations, the event may be identified as an arc if it has a likelihood or confidence above a certain level.

[0135] In some embodiments, in addition to machine vision algorithms, comparator logic may also be employed. For example, fast arcs will have a significantly faster dV / dt (both positive and negative) compared to, e.g., turning on a light-emitting diode (LED) in the system or another slower anomaly. Comparator logic may be used to compare dV / dt of a first set of frames with dV / dt of a second set frames to determine, e.g., which is larger (and therefore more likely to be an arc) and whether the first and second set of frames can be identified as an arc,Docket No. LAM1P014WO for example. Comparator logic may be used to compare dV / dt to a threshold, such as a slew rate threshold; or it may be used to compare light intensity to an intensity threshold. If dV / dt is positive and maintained above a threshold, a phenomenon may be beginning to occur or becoming more intense, and thus of salient interest. In addition, the end of these phenomena may also be detected, e.g., via negative dV / dt staying above a threshold (in absolute value). If dV / dt is negative and maintained below a threshold (or above, in absolute value), a phenomena may be ceasing or becoming less intense, and also of salient interest. In some cases, negative dV / dt may indicate a phenomenon as well. In some cases, dV / dt (either positive or negative) may indicate motion of an object.

[0136] If an arc is not detected, the example process 500 may return back to 502. If an arc is detected, the event may be analyzed and recorded or stored at 508. For example, events (e.g., occurrence of an arc), formation of the arc, location of the arc, and / or other information relating to the arc (e.g., time relative to a start of a process) can be written to a data file as they are identified. Possible causes of the arc can be better understood with more events are identified and collected over time. Generated frames and / or image data may also be stored, e.g., on a non- transitory computer-readable storage medium.Transients Detection

[0137] FIG. 6 is a flow diagram of an example process 600 for event-based transients detection, according to some embodiments. At 602, one or more frames may be generated based on one or more accumulation values, which may be similar to 502 and omitted herein for brevity.

[0138] At 604, presence of transients may be determined. In different embodiments, transients may be determined in a way similar to 504, including via machine / computer vision, machine learning, deep learning, etc. Transients referred to in the present disclosure may be of the type that may occur within or at a processing apparatus configured to perform fabrication operations (e.g., deposition processes involving plasma). Examples of such transients may include HCD, a plasma discharge, plasma ignition, and plasmoid formation and groupings or movement. Plasma instabilities can lead to the formation of multiple plasmoid structures that may interact with one another. In specific scenarios, electrical plasma arcs may be a transient phenomenon. Hence, example process 600 may be a generalized process of the example process 500 of FIG. 5. Each of these phenomena may generate light and thus be detectable by an event-driven sensor configured to perform ultra-fast sensing based on optical signals and dV / dt associated with the optical signals, as described above. In addition, the end of these phenomena may also be detected, e.g., via negative dV / dt staying above a threshold (in absolute value).Docket No. LAM1P014WO

[0139] Furthermore, comparator logic may be used in some embodiments. dV / dt associated with occurrence or disappearance of transients may be compared with dV / dt associated with occurrence or ending of other phenomena. For example, in the case of transients, such as HCD, fast arcs have a dV / dt (or a range thereof) that is significantly different than dV / dt (or range thereof) of HCD. Types of phenomena or events can be differentiated based on where the dV / dt falls within a range or scale compared to other dV / dt that has been measured or been known. In some situations, it may be possible that an arc is slow enough to match or overlap with the dV / dt associated with HCDs. In either scenario, both are anomalies that are damaging to the tool, and hence, detection of either can lead to correction or mitigation of an ongoing process, including modifying the operation of a processing apparatus or a component, or slowing, stopping, or shutting down of the processing apparatus or the component. Hence, a combination of both detection and comparator approaches can corroborate with determination and classification of phenomena.

[0140] At 606, if the phenomenon is not detected, the example process 600 may return to 602. If the phenomenon is detected, the example process 600 may proceed to 608, where the event may be analyzed and recorded or stored, similar to 508.Component Detection

[0141] Another use case of event-based imaging and detection is monitoring operation of components of a processing apparatus or tool. An illustrative example is a valve used in a deposition tool, such as an atomic layer deposition (ALD) valve. A valve may be used to control the flow of precursor gases and reactants into a process chamber where reactions occur with a semiconductor substrate, plasma, and / or other chemistries, thereby causing deposition of material on the substrate. A valve ensures the precise and controlled delivery of precursor gases and reactants. This precision is crucial because the deposition process, particularly in ALD, relies on sequential surface reactions to build up a thin film layer by layer. Moreover, sequences are repeated in a cyclic fashion to build a thin film with atomic layer precision, with the valve precisely controlling the introduction and removal of precursors and reactants in each cycle. Therefore, accurate cyclic timing control of the valve over time or many cycles is required, and valves may be designed to open and close very quickly or controlled to actuate with accuracy and precision to the desired timing, allowing for precise timing of precursor influx and pulses thereof. In the case of ALD, this timing control contributes to the characteristic layer-by-layer growth of the thin film.

[0142] ALD valve operations are rated for millions of cycles. There is an expected frequency for which these valves should open and close. However, as they age, the response to theDocket No. LAM1P014WO actuation mechanism (e.g., diaphragm change caused by pneumatic pressure) may mechanically degrade or become less accurate or reliable. There is an expected mechanical degradation that could result in changes in dampening and vibration patterns. Valves in the same module or tool or across different stations or chambers may also be compared against one another. Detecting these changes from the onset could help with understanding the health of each individual valve (or even a related component, such as a hinge) over time and performance as the valve approaches the end of its lifetime. For example, using event-based detection, information such as position of a valve (e.g., open or closed, distance from pipe), vibration frequency of a valve, latency time of opening and closing, dampening frequency, and number of open-close cycles (cycle frequency) can be obtained. As a specific example, a valve that typically performs at a time required to open of 10 ms may drift to 11, 12, 13, and so on. The vibrations associated with a valve that takes 10 ms to open may be different from vibrations associated with a valve that now takes 13 ms. Predictive maintenance can involve monitoring of vibrations and / or their frequencies throughout the life of a valve, and replacing or adjusting processes where valves are not at peak performance.

[0143] Other examples of components of a fabrication tool that may be monitored using eventbased imaging include an edge ring (e.g., on which a wafer rests during processing) and / or one or more lift pins (e.g., which are used to support and / or move a wafer to position and / or transfer the wafer). For example, an edge ring and / or a lift pin may be monitored for breakage, to ascertain whether it has moved in positioning, to determine whether all lift pins of a tool are moving (e.g., raising or lowering) at the same time, etc. In some embodiments, event-based imaging may be used to detect wafer breaks. In some cases, breakage of an edge ring, lift pin, or a wafer may occur at a relatively slow speed that is generally not detected by event-based imaging systems. However, in some embodiments, an LED may be used in tandem with the event-based system to produce artificial events which are detected by the event-based system. For example, the LED may be flashed in a strobe-like manner, and the strobe-like flashing of the LED may correspond to events that trigger the event-based imaging system. Components of the fabrication tool and / or a wafer undergoing processing may be captured by the eventbased imaging system which is triggered to capture by the flashing LED, thereby allowing visualization of relatively slower events by the event-based imaging system. Note that, in some embodiments, the techniques described herein for flashing an LED in a strobe-like manner may be used to diagnose malfunctioning of the LED itself.Docket No. LAM1P014WOVisualization of Phenomena

[0144] As noted above, change in voltage over time (dV / dt) may be indicative of an event occurring or ceasing. More specifically, the voltage detected by an event-driven sensor configured to perform ultra-fast sensing of optical signals, such as light emitted by a rapidly changing phenomenon or event.

[0145] FIGS. 7A and 7B illustrate a sequence 700 of example frames 702, 704, 706, 708 representative of a rapidly changing phenomenon such as activation of a light source, captured using event-based detection of the phenomenon, according to some embodiments. In this example, the rapidly occurring phenomenon may be an LED turning on. When observed using the human eye, the LED light may appear to turn on instantaneously. However, it can be seen that that is not the case when captured with the event-driven sensor described in the present disclosure. Example frames 702, 704, 706, 708 are sequential frames at times tl, t2, t3 and t4, which may not be temporally spaced equally apart but shown purely for illustrative purposes.

[0146] Example frame 702 shows a region 703 within which the phenomenon is originating (e.g., LED light source) at tl, indicated by a cluster of pixels 701. In some implementations, the region 703 may include pixels 701 that are shown based on a parameter staying above a threshold. For example, dV / dt (also referred to as +dV / dt for clarity that it is a positive change in voltage) may be above a dV / dt threshold (also referred to as +dV / dt threshold) in each pixel that is shown in example frame 702. More directly, region 703 is where more optical signals are suddenly being detected, per pixel, over time (and in example frame 702, at tl). Since, in this case, dV / dt must stay above the dV / dt threshold, it means that there is a continuous change (e.g., increase) in light intensity to meet the dV / dt threshold, as opposed to staying at the relatively same light intensity (which would mean dV / dt is closer to zero or possibly below the dV / dt threshold but not enough to meet a -dV / dt threshold as discussed with respect to FIGS. 8A and 8B). In some implementations, the parameter may be intensity or a voltage, which may be above an intensity threshold or a voltage threshold.

[0147] Example frame 704 shows additional pixels 705 at t2 which have expanded outward from the region 703. As in example frame 702, the pixels 705 shown in example frame 704 may be based on the parameter staying above the threshold. For example, dV / dt may be above a dV / dt threshold in each pixel that is shown in example frame 704. As in example frame 702, there is a continuous change (e.g., increase) in light intensity such that dV / dt stays above the dV / dt threshold. In other words, each of pixels 705 may indicate a pixel of the example frame 704 where light intensity of detected optical signals is increasing significantly (or above the threshold amount). Region 703 no longer has the pixels 701 (or most of them) that were shownDocket No. LAM1P014WO in example frame 702. This may signify that dV / dt is no longer above the dV / dt threshold. For example, light intensity in region 703 may not be increasing sufficiently to meet the dV / dt threshold. Light intensity in region 703 may even be decreasing to some extent, although, in some representations, if the decrease meets a -dV / dt threshold, it may be indicated in a different way within a given frame, as will be discussed further with respect to FIGS. 8 A and 8B. In some implementations, the parameter may be intensity or a voltage, which may be above an intensity threshold or a voltage threshold.

[0148] Example frame 706 shows pixels 707 at t3. Compared to example frame 704, there are fewer pixels whose parameter (e.g., dV / dt) is staying above the threshold. For example, the activation of the LED may be nearing a steady state (e.g., where the light is completely on), so fewer pixels may be capturing changes in optical signal intensities and experiencing a dV / dt above a dV / dt threshold in example frame 706.

[0149] Example frame 708 shows even fewer pixels 709 at t4. There are almost no pixels remaining by time t4. The activation of the LED may be almost complete. Upon completion of the phenomenon (e.g., LED turning on), a frame representing changes in detected optical signals would have no pixels.

[0150] Note that tl, t2, t3 and t4 may be in the order of nanoseconds, e.g., spaced 5 to 50 ns apart. Using frames such as example frames 702, 704, 706, 708 (and / or other frames prior to or subsequent to those example frames) generated based on optical signals obtained via an event-driven optical sensor, a fixed-FPS sequence of images or a video having an equivalent or effective FPS (e.g., a sampling rate of 500 FPS) may be obtained. In other implementations, the sampling rate can exceed 10,000 FPS. The event-driven optical sensor may be configured to rapidly respond to the change in lighting and zero out the signals when no longer changed to the extent of interest (e.g., not above a threshold such as a dV / dt threshold). The videos having such a high equivalent FPS can be slowed for human observation, e.g., FPS of 15, 30, 60, 120, etc., where, e.g., example frames 702, 704, 706, 708 are spaced 1 / 60 of a second apart. Such a video (or sequence of images) may be a video representative of am LED turning on, which may possibly be a rapidly occurring phenomenon of interest to detect. Such a video may not visually show an LED light but rather show pixels representing the change in the phenomena occurring over tl through t4. In some approaches, background visual noise can be reduced. Any background illumination (as long as relatively constant) does not impact resulting frames since there are no changes to light intensity occurring. That is, the relevant parameter (e.g., dV / dt) would not reach the threshold (e.g., dV / dt threshold).Docket No. LAM1P014WO

[0151] Obtaining frames or images representative of a rapidly occurring and / or transient phenomenon enables adjustment to an ongoing process or operation, such as at a processing apparatus where the phenomenon is occurring. In some cases, such an adjustment may include up to a shutdown of the system and the processing apparatus. Example approaches for detecting the phenomenon include an image-based comparison between frames or a pixel-based comparison between frames.

[0152] In an image-based comparison, a plurality of images representative of an event occurring with respect to the processing apparatus may be obtained. The plurality of images may include indications of one or more changes in an intensity of optical signals obtained by an optical sensor such as the aforementioned event-driven optical sensor. Two or more of the example frames 702, 704, 706, 708 may be examples of the plurality of images representative of the event. The event may be, for example, an LED light turning on. However, myriad events such as electrical arcs, HCDs, motion or vibration of a component, and / or other transients may be detected in similar fashion to above using the event-driven optical sensor.

[0153] Based on a difference between at least a first image and a second image of the plurality of images exceeding a threshold, the processing apparatus may adjust an ongoing process associated with the event, where the adjustment may be caused by, e.g., a controller or instructions from or to the controller. For instance, comparing example frames 702 and 704 representative of changes captured with respect to a rapidly occurring phenomenon as shown in FIG. 7A, it can be seen that significant changes have occurred between tl and t2. More specifically, in light of pixels 701 shown in example frame 702, it can be seen that some phenomenon has progressed by t2 in example frame 704 relative to what has begun to occur at tl. In some approaches, the first frame may be a reference frame in which there are no pixels. As such, any at least one pixel that appears on the second frame may be an indication that there has been some change in optical signals (e.g., from a phenomenon) in a region or area of interest being monitored. In some cases, the first frame may have been obtained at any past point in time when no phenomenon was occurring at the region or area of interest.

[0154] Various approaches can be used to determine the changes (if any) between frames. For example, as mentioned above, computer vision techniques or algorithms, machine learning or deep learning models, neural networks, or other similar known techniques or methods may be used. In some approaches, changes can be statistically determined and compared. For example, one frame (e.g., example frame 702) may have a first average intensity value or a first pixel count, and another frame (e.g., example frame 704) may have a second average intensity value or a second pixel count. In the case of example frames 702 and 704, example frame 704 hasDocket No. LAM1P014WO many more pixels 705 than pixels 701 shown in example frame 702 (e.g., exceeding a quantity threshold or a difference threshold). Hence, it may be determined that a phenomenon is continuing or progressing based on the statistically determined parameter. In another example, changes can be determined based on distance of the change in where pixels are occurring. Example frame 704 has pixels present in a newly formed region 703a, while example frame 702 does not. The distance between region 703 and region 703a may indicate a significant progression of the phenomenon, e.g., if the distance exceeds a distance threshold. Other parameters may be evaluated similarly, e.g., pixel cluster size, pixel value in implementations in which a pixel may not only be present but indicated with one of multiple values depending on corresponding light intensity, rate of change in distance, and / or others that will be apparent to those having ordinary skill in the relevant arts. In some implementations, multiple of the foregoing approaches or parameters can be considered or used to corroborate a determination that a rapidly occurring event has occurred, and an ongoing operation or process may need adjustment. For example, between example frames 704 and 706, there are fewer and smaller distance differences between pixel locations, but there is a difference in the number of pixels 705 and 707. There are fewer pixels 707 at t3 compared to pixels 705 at t2, which may indicate that a phenomenon has occurred and is reaching a steady state.

[0155] In a pixel-based comparison, even one pixel occurrence or change (or more in some approaches) between frames may be sufficient to modify an ongoing operation or modify an operation of the system (e.g., cause a shutdown of a processing apparatus, a component (e.g., an RF generator), or the entire system). In some approaches, the first frame may be a reference frame in which there are no pixels. As such, any at least one pixel that appears on the second frame may be an indication that there has been some change in optical signals (e.g., from a phenomenon) in a region or area of interest being monitored. In some cases, the first frame may have been obtained at any past point in time when no phenomenon was occurring at the region or area of interest. Visual noise may be managed using, e.g., a slew rate or intensity threshold. A pixel-based comparison may be based on one pixel change, as opposed to multiple pixels or an image as a whole (e.g., using a statistical determination of a parameter, or differences based on different regions of a frame) in an image-based comparison. In some configurations, comparator logic may be employed for image-based or pixel-based comparison.

[0156] Referring now to FIGS. 8A and 8B, another sequence 800 of example frames 802, 804, 806, 808 representative of a rapidly changing phenomenon such as deactivation of a light source, captured using event-based detection of the phenomenon, according to someDocket No. LAM1P014WO embodiments, is illustrated. In this example, the rapidly occurring phenomenon may be an LED turning off, e.g., from an on state. As noted elsewhere herein, the ending of a phenomenon may be of salient interest. Hence, the ending or ceasing of a phenomenon may also be considered a phenomenon.

[0157] When observed using the human eye, the LED light may appear to turn off instantaneously. However, it can be seen that that is not the case when captured with the event- driven sensor described in the present disclosure. Example frames 802, 804, 806, 808 are sequential frames at times tl, t2, t3 and t4, which may not be temporally spaced equally apart but shown purely for illustrative purposes.

[0158] Example frame 802 shows a region 803 within which the phenomenon is ending at an origin of the phenomenon (e.g., LED light source) at tl, indicated by a cluster of pixels 801. In some implementations, the region 803 may include pixels 801 that are shown based on a parameter staying below a negative threshold or above a threshold in absolute value. For example, -dV / dt may be below a -dV / dt threshold in each pixel that is shown in example frame 802. In another example, the absolute value of -dV / dt may be above a dV / dt threshold. Put these examples another way, voltage corresponding to region 803 is dropping more than the threshold. More directly, region 803 is where, suddenly, less optical signals are being detected, per pixel, over time (and in example frame 802, at tl). Since, in this case, -dV / dt must stay below the -dV / dt threshold, it means that there is a continuous change (e.g., decrease) in light intensity to meet the -dV / dt threshold, as opposed to staying at the relatively same light intensity (which would mean dV / dt is closer to zero or possibly above the -dV / dt threshold but not enough to meet a +dV / dt threshold of the type discussed with respect to FIGS. 7A and 7B). In some implementations, the parameter may be intensity or a voltage, which may be below an intensity threshold or a voltage threshold.

[0159] Example frame 804 shows additional pixels 805 at t2 which have expanded outward from the region 803. There are also fewer or less concentrated pixels in region 803. As in example frame 802, the pixels 805 shown in example frame 804 may be based on the parameter staying below the threshold. For example, -dV / dt may be below a -dV / dt threshold in each pixel that is shown in example frame 804. As in example frame 802, there is a continuous change (e.g., decrease) in light intensity such that -dV / dt stays below the -dV / dt threshold. In other words, each of pixels 705 may indicate a pixel of the example frame 704 where light intensity of detected optical signals is decreasing significantly (or below the threshold amount).

[0160] Example frame 806 shows pixels 807 at t3. Compared to example frame 804, there are fewer pixels whose parameter (e.g., -dV / dt) is staying below the threshold. For example, theDocket No. LAM1P014WO deactivation of the LED may be nearing a steady state (e.g., where the light is completely off), so fewer pixels may be capturing changes in optical signal intensities and experiencing a -dV / dt below a -dV / dt threshold in example frame 706.

[0161] Example frame 808 shows even fewer pixels 809 at t4. There are almost no pixels remaining by time t4. The deactivation of the LED may be almost complete. Upon completion of the phenomenon (e.g., LED turning off), a frame representing changes in detected optical signals would have no pixels.

[0162] Similar to the example frames 702, 704, 706, 708, tl, t2, t3 and t4 may be in the order of nanoseconds, e.g., spaced 5 to 50 ns apart. Although example frames 802, 804, 806, 808 are also at tl, t2, t3 and t4, they may not be spaced apart or correspond to the timing of example frames 702, 704, 706, 708. Nonetheless, a sequence of images or video having an effective FPS of, e.g., 500 or 10,000 or more may be obtained using example frames 802, 804, 806, 808 (and / or other frames prior to or subsequent to those example frames), as noted above with respect to example frames 702, 704, 706, 708. Such a video (or sequence of images) may be a video representative of an LED turning off, which may possibly be a rapidly occurring phenomenon of interest to detect.

[0163] In some embodiments, pixels shown in the frames can represent different types of parameters. For example, in example frames 702, 704, 706, 708, pixels of one color (e.g., blue) may be used to represent increasing light intensities (+dV / dt), while in example frames 802, 804, 806, 808, pixels of another color (e.g., black) may be used to represent a decreasing light intensities (-dV / dt). In some implementations, degrees or values of the parameter can be represented using other visual effects, such as brightness of the pixels. For instance, a higher dV / dt (defined, e.g., by a higher threshold or range) may be indicated by a brighter pixel than lower dV / dt that still exceeds the dV / dt threshold. Hence, pixels may provide multi-class indications and also not be a binary indication of optical signals meeting or not meeting a criterion or threshold within a frame.

[0164] Obtaining frames or images representative of the ending of a rapidly occurring and / or transient phenomenon enables adjustment to an ongoing process or operation, such as at a processing apparatus where the phenomenon is occurring. In some cases, such an adjustment may include up to a shutdown of the system and the processing apparatus. Example approaches for detecting the ending of the phenomenon include an image-based comparison between frames or a pixel-based comparison between frames, as discussed above.

[0165] Another example of a possible rapidly occurring phenomenon of interest to detect is an electrical arc, which is damaging to the processing apparatus or tool and wafers being processedDocket No. LAM1P014WO therein but difficult and cumbersome (e.g., high memory requirements) to detect using traditional fixed-rate cameras.

[0166] FIGS. 9A and 9B illustrate another sequence 900 of example frames 902, 904, 906, 908 representative of a rapidly occurring phenomenon such as an electrical arc, captured using event-based detection of the phenomenon, according to some embodiments. Formation of an electrical arc may be a highly transient event that is undetectable by the human eye or traditional imaging. While the presence of an electrical arc can be detectable using a temporal sensor, as noted above, such temporal sensor provides limited information. Spatial information (e.g., location indicated by pixels) captured with the event-driven sensor or camera described in the present disclosure could provide insight that can be evaluated to confirm or determine that an arc has occurred and where. Example frames 902, 904, 906, 908 are sequential frames at times tl, t2, t3 and t4, which may not be temporally spaced equally apart but shown purely for illustrative purposes.

[0167] Example frame 902 is representative of a monitored space at tl, before an arc (or other phenomena) has formed. A change in light intensity, measured by a parameter such as dV / dt meeting or exceeding a threshold such as a dV / dt threshold (or other parameters and thresholds relating to intensity and voltage), would be indicated by pixels being shown in the frame.

[0168] Example frame 904 shows a region 910 within which the phenomenon (e.g., an electrical arc) is forming or beginning to form at t2, indicated by a cluster of pixels 905. In some implementations, the region 910 may include pixels 905 that are shown based on a parameter staying above a threshold. For example, dV / dt may be above a dV / dt threshold in each pixel that is shown in example frame 904. More directly, region 910 is where more optical signals are suddenly being detected, per pixel, over time (and in example frame 904, at t2). Since, in this case, dV / dt must stay above the dV / dt threshold, it means that there is a continuous change (e.g., increase) in light intensity to meet the dV / dt threshold, as opposed to staying at the relatively same light intensity (which would mean dV / dt is closer to zero or possibly below the dV / dt threshold but not enough to meet a -dV / dt threshold).

[0169] In addition, the general shape of the cluster of pixels 905 resembles a linear segment or a line. In some implementations, the type of phenomenon can be identified using visual analysis of a cluster of pixels (e.g., using a computer vision technique, machine learning, neural networks, or another general visual approach) and / or comparing against known shapes or patterns of clusters. In the example frame 904, the linear shape could corroborate a determination (with or without some certainty level) that the phenomenon being detected is an electrical arc. In some cases, such as example frames 702, 704, 802 and 804, the tight circularDocket No. LAM1P014WO shape of the cluster of pixels and / or the bursting star-like shape (and / or the progression from the circular shape to the star-like shape) can indicate or corroborate a determination that the phenomenon being detected is an activation or a deactivation of a light source, e.g., a light source of a certain size, or specifically a LED light. In yet other examples, such as that discussed with respect to FIGS. 11A and 1 IB, other shapes can lead to a determination of other phenomena.

[0170] Example frame 906 shows pixels 905 within the region 910 at t3. Compared to example frame 904, there are fewer pixels whose parameter (e.g., dV / dt) is staying above the threshold. For example, an electrical arc may be finishing its formation or reaching its peak intensity, and thus, dV / dt may be decreasing (and falling below the dV / dt threshold) at more of the pixels within the region 910.

[0171] Example frame 908 shows even fewer pixels 905 within the region 910 at t4. There are almost no pixels remaining by time t4. The event — the formation and presence of the electrical arc — may be nearly over. The electrical arc may have its peak intensity at t4, or it may be disappearing or have disappeared by t4. In visualizations using -dV / dt, disappearance of the electrical arc may also be detected separately. Either way, once the event has completed, all pixels would disappear after t4.

[0172] tl, t2, t3 and t4 may be in the order of nanoseconds, e.g., spaced 5 to 50 ns apart, which can be generated based on optical signals obtained via the event-driven optical sensor configured to rapidly respond to the change in the optical signals (e.g., light, lighting) and provide high temporal and spatial resolutions. The sequence of frames (and / or other frames prior to or subsequent to example frames 902, 904, 906, 908) can result in a video having an equivalent or effective FPS of 500, 10,000, or higher. Similar to example frames 702, 704, 706, 708, 802, 804, 806, 808, the sequence of frames or video may be slowed down for manual review and / or computer image analysis (e.g., using visual or machine techniques described elsewhere herein). Such a video (or sequence of images) may be a video representative of an electrical arc forming, which may possibly be a rapidly occurring phenomenon of interest to detect, e.g., at a processing apparatus, tool, or system, and use as a basis for adjusting a process or operation (including shutting down the system). However, note that the resultant video would not visually show the arc itself, as a traditional camera would. Example approaches for detecting the electrical arc include an image-based comparison between frames or a pixel-based comparison between frames, as discussed above. Another example approach will be described now.Docket No. LAM1P014WO

[0173] FIG. 10 is a flow diagram of an example process 1000 for detection of an electrical arc and adjustment of a process based thereon, according to some embodiments.

[0174] At 1002, raw data may be received over a prescribed time period. In some embodiments, raw data may include optical signals received at an event-driven optical sensor configured to perform ultra-fast sensing of optical signals and rapidly respond to the change in the optical signals, such as the type discussed so far. Depending on application, the event- driven optical sensor may be positioned (e.g., at a viewport, front end, next to a valve) so as to obtain the optical signals from a region of interest or a space of interest, e.g., at a particular portion of a process chamber or processing apparatus, tool, or system, where rapidly changing phenomena or transient phenomena might occur. Such phenomena may occur sporadically or unexpectedly over a very short period of time, e.g., within nanoseconds or microseconds. Thus, constant monitoring may be needed, depending on the application. In some applications, deviations from known behavior may be monitored, such as vibrations of a component such as a valve. In the context of example process 1000, however, the system may be monitored for electrical arcs (or other transient phenomena). Optical signals may be received at the event- driven optical sensor as a stream of data. However, the optical signals may not necessarily result in storage of frames, image data, video data, or other generated data. Advantageously, in some implementations, frames and video may be generated responsive to an electrical arc being detected (e.g., when a relevant parameter or change in parameter meets a condition or threshold), which would significantly reduce the need to constantly capture and store image / video data as with traditional imaging. In other implementations, however, frames and video may be optionally generated outside of detecting a phenomenon, at 1004. For example, frames may be periodically generated based on a prescribed accumulation rate of acquiring optical signals, e.g., an accumulation rate of 1 ms, 100 ms, 1 second, 1 minute, or shorter or longer. That is, one frame representative of a parameter or change in parameter meeting a condition can be generated at intervals corresponding to the accumulation rate. This can still reduce storage requirements over traditional imaging. For instance, generating frames with an accumulation rate of 1 second would result in an effective frame rate of 1 FPS, as opposed to typical frame rates such as 15 or 30 FPS.

[0175] At 1006, an arc detection routine may be performed. In some embodiments, frames may be generated based on a parameter, and an image-based comparison of the frames may be performed.

[0176] In some implementations, a plurality of frames may be generated in which pixels are formed based on a parameter such as dV / dt associated with the raw data (e.g., optical signals)Docket No. LAM1P014WO meeting a condition such as a dV / dt threshold, on a per-pixel basis. For example, a frame may be generated responsive to dV / dt associated with a given pixel meeting or exceeding a dV / dt threshold. In some implementations, other parameters such as voltage or light intensity may be evaluated against a voltage threshold or an intensity threshold. The plurality of frames may thus include indications of one or more changes in an intensity of the optical signals. Example frames 902 - 908 may be examples of a frame that can be generated. At least two generated frames may then be compared (e.g., using comparator logic or controller logic) to determine a difference, e.g., between a first frame and a second frame. Various approaches as described above with respect to FIGS. 7 A and 7B may be used to determine the changes (if any) between the frames.

[0177] In some embodiments, a pixel-based comparison may be performed. More specifically, a change of one pixel (or more) between frames may be determined. For example, a first frame and a second frame may be compared to determine whether there is a pixel in one frame that is not present in another.

[0178] In some implementations, the first frame may be a reference frame in which there are no pixels. As such, any at least one pixel that appears on the second frame may be an indication that there has been some change in optical signals (e.g., from an electrical arc) in the region or space of interest being monitored. In some cases, the reference frame may have been obtained at any past point in time when arc was occurring at the region or space of interest.

[0179] In some implementations, a comparator or comparator logic may be used to compare frames or pixels. In some implementations, a controller, controller apparatus or system (e.g., a group of controllers), or controller logic may be used to compare frames or pixels.

[0180] In some embodiments, a shape or pattern of the pixels may be used to identify the electrical arc. For example, the general shape of a cluster of pixels present in a frame could resemble or match a known shape or pattern of pixel clusters. The cluster of pixels 905 shown in example frame 904 or example frame 906 may be an example of the appearance of a cluster of pixels corresponding to an electrical arc.

[0181] At 1008, whether an electrical arc is detected may be determined, e.g., based on the arc detection routine of 1006. If an arc is detected, at 1010, a process may be adjusted based at least on the presence of the arc. The process may be, e.g., an ongoing operation of a processing apparatus or system. The adjustment of the process may involve, e.g., changing an operational parameter or a process parameter, such as flow rate of a gas, temperature, pressure, timing of plasma ignition or plasma exposure, valve opening or closing rate, rotation speed of a chuck,Docket No. LAM1P014WORF power, etc. The adjustment of the process may involve, e.g., issuance of a warning or error, or a shutdown of the processing apparatus or system or a component thereof.

[0182] In some implementations, the process may be adjusted further based on a certainty associated with the arc. That is, as an illustrative example, the adjustment may be performed only if the arc is detected with a certainty level of the arc meeting a particular certainty threshold of, say, 70%. Such certainty threshold may be selected based on the desired sensitivity. In some implementations, different actions or adjustments can be performed depending on the certainty level. For example, a certainty level of at least 0.7 (70%) may result in a shutdown of the processing apparatus or system or a component thereof, while a certainty level of 0.2-0.6 (20-60%) may result in a warning or a soft shutdown. The certainty level of the arc may be determined based on various characteristics of the pixels, e.g., intensity (especially where more than binary (multi-class) values are used), size, location, pixel density, shape of pixel cluster, a machine learning model (which may have been trained using frames known to represent an arc). In some implementations, debounce and / or hysteresis as described above may be used to confirm or increase the certainty level.

[0183] If an arc is not detected, the example process 1000 may revert to 1002, where the arc may continue to be monitored based on frames generated, e.g., based on optical signals, and the arc detection routine.

[0184] FIGS. 11A and 11B illustrate another sequence 1100 of example frames 1102, 1104, 1106 representative of a rapidly occurring phenomenon such as hollow cathode discharge (HCD), captured using event-based detection of the phenomenon, according to some embodiments. HCDs are transient anomalies that could last several milliseconds (e.g., 1 - 20 ms), or the length of a RF cycle, which may occur in a gap between a showerhead and a pedestal, and may negatively affect wafer processing, but may in some cases be undetectable by the human eye or traditional imaging. HCDs can be benign in some cases but potentially point toward the liability of an electrical arc happening. HCDs can also interfere with sensor detection and measurements as well as potentially affect deposition or etching processes. Thus, detection of HCDs can be beneficial as at least a warning sign for degradation of the process or the tool. Example frames 1102, 1104, 1106 are sequential frames at times tl, t2 and t3, which may not be temporally spaced equally apart but shown purely for illustrative purposes.

[0185] Example frame 1102 shows a region 1110a within which the phenomenon (e.g., an HCD) is beginning to form at tl, indicated by a cluster of pixels 1103a. In some implementations, the region 1110a may include pixels 1103a that are shown based on a parameter staying above a threshold. For example, dV / dt may be above a dV / dt threshold inDocket No. LAM1P014WO each pixel that is shown in example frame 1102. More directly, region 1110a is where more optical signals are suddenly being detected, per pixel, over time (and in example frame 1102, at tl). Since, in this case, dV / dt must stay above the dV / dt threshold, it means that there is a continuous change (e.g., increase) in light intensity to meet the dV / dt threshold, as opposed to staying at the relatively same light intensity. In some implementations, the parameter may be intensity or a voltage, which may be above an intensity threshold or a voltage threshold.

[0186] Example frame 1104 shows additional pixels 1103a and 1103b at t2. Cluster of pixels 1103b is within another region 1110b of shown in example frame 704. As in example frame 1102, the pixels 1103b shown in example frame 704 may be based on the parameter (e.g., dV / dt) staying above the threshold (e.g., dV / dt threshold).

[0187] Example frame 1106 shows pixels 1103a and 1103b at t3. Compared to example frame 1104, there are fewer pixels, especially in the center portion of the clusters of pixels 1103a and 1103b. Fewer pixels may be capturing changes in optical signal intensities and experiencing a dV / dt above a dV / dt threshold. The formation of HCDs may be nearing a steady state or beginning to disappear.

[0188] tl, t2 and t3 may be in the order of nanoseconds, e.g., spaced 5 to 50 ns apart, which can be generated based on optical signals obtained via the event-driven optical sensor configured to rapidly respond to the change in the optical signals (e.g., light, lighting) and provide high temporal and spatial resolutions. The sequence of frames (and / or other frames prior to or subsequent to example frames 1102, 1104, 1106) can result in a video having an equivalent or effective FPS of 500, 10,000, or higher. The sequence of frames or video may be slowed down for manual review and / or computer image analysis (e.g., using visual or machine techniques described elsewhere herein). Such a video (or sequence of images) may be a video representative of HCD(s) forming, which may possibly be a rapidly occurring phenomenon of interest to detect, e.g., at a processing apparatus, tool, or system, and use as a basis for adjusting a process or operation (including shutting down the system). However, note that the resultant video would not visually show the HCDs themselves, as a traditional camera would. Example approaches for detecting the HCDs include an image-based comparison between frames or a pixel-based comparison between frames, as discussed above.

[0189] In addition, the ring-like shape of the clusters of pixels 1103a and 1103b may be used as a basis for identifying the type of phenomenon occurring. For example, visual analysis of a cluster of pixels (e.g., using a computer vision technique, machine learning, neural networks, or another general visual approach) and / or comparing against known shapes or patterns of clusters may enable a determination (with or without some certainty level) that the phenomenonDocket No. LAM1P014WO being detected is an HCD(s). In some cases, the progression from a circular cluster of pixels to the ring-like shape from example frame 1104 to example frame 1106 may indicate or provide corroboration for the determination that the phenomenon being detected is an HCD(s).Methods

[0190] FIG. 12 is a flow diagram of an example of a method 1200 for detecting rapidly changing phenomena, according to some embodiments. One or more of the functions of the method 1200 may be performed by or caused by a computerized apparatus or system or a component thereof (e.g., a sensor apparatus or a hardware and / or software component thereof). In some embodiments, the computerized apparatus or system such as a deposition apparatus or deposition tool configured to perform deposition of material (e.g., film) on a substrate. In some embodiments, the sensor apparatus may be a sensor, such as the event-driven sensor (e.g., camera or optical sensor) discussed herein. Structure for performing the functionality illustrated in one or more of the blocks shown in FIG. 12 may be performed by a hardware and / or software component of such computerized apparatus or system, such as, for example, a controller apparatus, a computerized system, or a computer-readable apparatus including a storage medium storing computer-readable and / or computer-executable instructions that are configured to, when executed by a processor apparatus, cause the at least one processor apparatus or a computerized apparatus to perform the operations. A controller may be one example of the computerized apparatus or system. A process chamber may be another example or component of the computerized apparatus or system. Example components of above apparatus or system are illustrated in, e.g., FIGS. 1 A - 1C and 15, and described in more detail elsewhere herein.

[0191] It should also be noted that the operations of FIG. 12 may be performed in any suitable order, not necessarily the order depicted in FIG. 12. Further, the process shown in FIG. 12 may include additional or fewer operations than those depicted in FIG. 12.

[0192] At 1210, the method 1200 may include, responsive to an occurrence of a phenomenon, determining one or more changes in an intensity of optical signals, the optical signals obtained by an optical sensor.

[0193] In some scenarios, the determining of the one or more changes may be responsive to an occurrence of a phenomenon. In some cases, the phenomenon may include a plasma ignition, plasma motion, plasmoid formation (or movement), a plasma discharge, an electrical arc, motion of a component of a processing apparatus (e.g., vibration, motion, opening / closing / actuation of a valve, back-and-forth motion based on dampening forces andDocket No. LAM1P014WO external forces), or a hollow cathode discharge (HCD). In some cases, the phenomenon may include a ceasing or ending of the aforementioned types of phenomena.

[0194] In some cases, the phenomenon may include the electrical arc within the processing apparatus; and the method 1200 may further include, based on a computerized visual evaluation of the pixels of the images, identifying one or more images of the generated images associated with the electrical arc.

[0195] In some cases, the phenomenon may include the HCD within the processing apparatus; and the method 1200 may further include, based on a computerized visual evaluation of the pixels of the images, identifying one or more images of the generated images associated with the HCD.

[0196] In some cases, the phenomenon may include the motion of the component of the processing apparatus, the motion comprising an actuation of a valve; and the method 1200 may further include determining a degradation of the valve based on a frequency of a vibration occurring during the actuation being under a frequency threshold, or based on a change in the frequency of the vibration occurring during the actuation, the frequency of the vibration determined based on the generated images associated with the motion of the component.

[0197] As discussed above, in different implementations, the computerized visual evaluation of the pixels of the images may include computer vision algorithms (e.g., edge detection, contour generation), machine learning techniques (e.g., training a machine learning model to determine the type of phenomenon), deep learning, neural networks, and / or other techniques.

[0198] At 1220, the method 1200 may include detecting one or more changes in voltage correlating to the one or more changes in intensity of optical signals.

[0199] In some embodiments, the determining of the one or more changes in voltage may include determining a positive change in voltage associated with the optical signals with respect to time. In some approaches, this positive change in voltage may be referred to as +dV / dt as discussed above. In some embodiments, the determining of the one or more changes in voltage may include determining a negative change in voltage associated with the optical signals with respect to time. In some approaches, this negative change in voltage may be expressed as - dV / dt as discussed above.

[0200] At 1230, the method 1200 may include based on the one or more changes in voltage (or the intensity of the optical signals) staying above a threshold, generating images representative of the event and comprising indications of one or more changes in the intensity of the optical signals occurring at corresponding pixels of the images. In some embodiments, the generating of the images may be based on an accumulation time of the optical signals obtained by theDocket No. LAM1P014WO optical sensor. In some implementations, the accumulation time may be a time period of up to 50 ms, 500 ps, 100 ps, 10 ps, 100 ns, or 10 ns. In some implementations, the accumulation time may be a time period of at least 1 second (e.g., multiple seconds) or at least 100 ns. In some embodiments, the threshold may include a threshold in a change in voltage associated with the intensity of the optical signals (such as a dV / dt threshold).

[0201] In some embodiments, the method 1200 may further include determining a difference between at least a first image and a second image of the generated images. Optionally, at 1240, the method 1200 may include, based on the difference between at least the first image and the second image exceeding a threshold, causing the processing apparatus to adjust an ongoing process at the processing apparatus. In some implementations, the method 1200 may further include determining a difference between at least a third image and a fourth image of the generated images different from the first and second images, based on the difference between at least the first image and the second image not exceeding a threshold.

[0202] In some embodiments, the method 1200 may further include detecting the occurrence of the phenomenon based on a certainty level associated with the occurrence of the phenomenon exceeding a threshold. In some implementations, the method 1200 may further include adjusting an ongoing process at the processing apparatus based on the certainty level associated with the occurrence of the phenomenon exceeding the certainty threshold. In some implementations, the adjusting of the ongoing process may include stopping the ongoing process. This may include a shutdown of the processing apparatus or system or a component thereof, per 1246. In some implementations, the adjusting of the ongoing process may include changing a process parameter, per 1242. In some implementations, the adjusting of the ongoing process may include issuing a warning or error, per 1244.

[0203] In some embodiments, the method 1200 may further include detecting the occurrence of the phenomenon based on a certainty level associated with the occurrence of the phenomenon exceeding a threshold. In some implementations, debounce and / or hysteresis may be used to increase the reliability of the detecting of the occurrence of the phenomenon.

[0204] In some embodiments, an effective sampling rate of the generated images may be at least 500 frames per second, the at least 500 frames being representative of changes occurring within an accumulation time of the optical signals during the occurrence of the phenomenon. In some embodiments, the effective sampling rate of the generated images may be at least 10,000 frames per second. In some embodiments, the effective sampling rate of the generated images may be at least 100,000 images per second, the at least 100,000 frames being representative of changes occurring within an accumulation time of the optical signals duringDocket No. LAM1P014WO the occurrence of the phenomenon. In some embodiments, the effective sampling rate of the generated images may be at least 10 million frames per second. In some embodiments, the effective sampling rate of the generated images may be below 500 frames per second. The frames may be representative of changes occurring within an accumulation time of the optical signals during the occurrence of the phenomenon, or in some specific cases, within one second of the phenomenon.

[0205] FIG. 13 is a flow diagram of an example of a method 1300 for mitigating a failure occurring at a processing apparatus, according to some embodiments. One or more of the functions of the method 1300 may be performed by or caused by a computerized apparatus or system, such as a deposition apparatus or deposition tool configured to perform deposition of material (e.g., film) on a substrate. Structure for performing the functionality illustrated in one or more of the blocks shown in FIG. 13 may be performed by hardware and / or software components of such computerized apparatus or system, such as, for example, a controller apparatus, a computerized system, or a computer-readable apparatus including a storage medium storing computer-readable and / or computer-executable instructions that are configured to, when executed by a processor apparatus, cause the at least one processor apparatus or a computerized apparatus to perform the operations. A controller may be one example of the computerized apparatus or system. A process chamber may be another example or component of the computerized apparatus or system. Example components of above apparatus are illustrated in, e.g., FIGS. 1A - 1C and 15, and described in more detail elsewhere herein.

[0206] It should also be noted that the operations of FIG. 13 may be performed in any suitable order, not necessarily the order depicted in FIG. 13. Further, the process shown in FIG. 13 may include additional or fewer operations than those depicted in FIG. 13.

[0207] At 1310, the method 1300 may include obtaining a plurality of images representative of a phenomenon occurring with respect to the processing apparatus, the plurality of images comprising indications of one or more changes in an intensity of optical signals obtained by an optical sensor. The one or more changes in the intensity of optical signals may be represented as dV / dt, or more specifically, +dV / dt or -dV / dt.

[0208] At 1320, the method 1300 may include, based on a difference between at least a first image and a second image of the plurality of images exceeding a threshold, causing the processing apparatus to adjust an ongoing process associated with the phenomenon. In some embodiments, the difference between the first image and the second image may be determinedDocket No. LAM1P014WO using one or more of various methods for an image -based comparison as discussed above, e.g., with respect to FIGS. 7A, 7B, 8A, 8B, 9A, 9B, 11A and 11B.

[0209] In some embodiments, the adjustment of the ongoing process may include an issuance of a warning or an error, or a shutdown of the processing apparatus or a component thereof.

[0210] FIG. 14 is a flow diagram of another example of a method 1400 for mitigating a failure occurring at a processing apparatus, according to some embodiments. One or more of the functions of the method 1400 may be performed by or caused by a computerized apparatus or system, such as a deposition apparatus or deposition tool configured to perform deposition of material (e.g., film) on a substrate. Structure for performing the functionality illustrated in one or more of the blocks shown in FIG. 14 may be performed by hardware and / or software components of such computerized apparatus or system, such as, for example, a controller apparatus, a computerized system, or a computer-readable apparatus including a storage medium storing computer-readable and / or computer-executable instructions that are configured to, when executed by a processor apparatus, cause the at least one processor apparatus or a computerized apparatus to perform the operations. A controller may be one example of the computerized apparatus or system. A process chamber may be another example or component of the computerized apparatus or system. Example components of above apparatus are illustrated in, e.g., FIGS. 1A - 1C and 15, and described in more detail elsewhere herein.

[0211] It should also be noted that the operations of FIG. 14 may be performed in any suitable order, not necessarily the order depicted in FIG. 14. Further, the process shown in FIG. 14 may include additional or fewer operations than those depicted in FIG. 14.

[0212] At 1410, the method 1400 may include obtaining a plurality of images representative of a phenomenon occurring with respect to the processing apparatus, each of the plurality of images comprising a plurality of pixels each indicative of a change in an intensity of optical signals obtained by an optical sensor. The change in the intensity of optical signals may be represented as dV / dt, or more specifically, +dV / dt or -dV / dt.

[0213] At 1420, the method 1400 may include, based at least on a difference in the intensity of optical signals indicated by at least a first pixel of a first image of the plurality of images and a corresponding first pixel of a second image of the plurality of images exceeding a threshold, causing the processing apparatus to adjust an ongoing process associated with the phenomenon. In some embodiments, the difference between the first image and the second image may be determined using one or more of various methods for an image -basedDocket No. LAM1P014WO comparison as discussed above, e.g., with respect to FIGS. 7A, 7B, 8A, 8B, 9A, 9B, 11A and 11B.

[0214] It should be noted that, in some embodiments, captured event images may be processed to recreate a time-based set of frames or images (e.g., a video). In some embodiments, the video may be monochromatic (e.g., in grayscale or black and white). This may be useful to allow an end user to interpret the captured event data, e.g., by visualizing the captured event data as a video or sequence of frames. Processing of captured event data into a sequence of frames / images may involve utilizing one or more models trained to convert event data to image frames. In some embodiments, a trained event data to image data model may be supplemented with a second model trained to denoise the images generated by the first model. Example architectures of models include convolutional neural networks (CNNs), U-Nets, or the like.

[0215] In some embodiments, the adjustment of the ongoing process may include an issuance of a warning or an error, or a shutdown of the processing apparatus or a component thereof.Apparatus

[0001] FIG. 15 is a schematic of a process system suitable for conducting deposition processes, such as frontside deposition processes, in accordance with embodiments. The system 1500 includes a transfer module 1503. The transfer module 1503 provides a clean, pressurized environment to minimize risk of contamination of substrates being processed as they are moved between various reactor modules. Mounted on the transfer module 1503 is a multi-station reactor 1509 capable of performing ALD, treatment, and CVD according to various embodiments. Multi-station reactor 1509 may include multiple stations 1511, 1513, 1515, and 1517 that may sequentially perform operations in accordance with disclosed embodiments. Stations may include a heated pedestal or substrate support, one or more gas inlets or showerhead or dispersion plate.

[0002] Mounted on the transfer module 1503 may be one or more single or multi-station modules 1507 capable of performing plasma or chemical (non-plasma) pre-cleans, other deposition operations, or etch operations. The module may also be used for various treatments to, for example, prepare a substrate for a deposition process. The system 1500 also includes one or more wafer source modules 1501, where wafers are stored before and after processing. An atmospheric robot (not shown) in the atmospheric transfer chamber 1519 may first remove wafers from the wafer source modules 1501 to loadlocks 1521. A wafer transfer device (generally a robot arm unit) in the transfer module 1503 moves the wafers from loadlocks 1521 to and among the modules mounted on the transfer module 1503.Docket No. LAM1P014WO

[0003] In various embodiments, a system controller 1542 is employed to control process conditions during deposition. The system controller 1542 will typically include one or more memory devices and one or more processors. A processor may include a central processing unit (CPU), graphics processing unit (GPU), logic, or other computing unit (computer) or processing unit, analog and / or digital input / output connections, stepper motor controller boards, etc.

[0004] The system controller 1542 may control all the activities of the deposition apparatus. The system controller 1542 executes system control software, including sets of instructions for controlling the timing, mixture of gases, chamber pressure, chamber temperature, wafer temperature, radio frequency (RF) power levels, wafer chuck or pedestal position, and other parameters of a particular process. Other computer programs stored on memory devices associated with the system controller 1542 may be employed in some embodiments.

[0005] The depicted embodiment includes a user interface associated with the system controller 1542. The user interface may include a display screen, graphical software displays of the apparatus and / or process conditions, and user input devices such as pointing devices, keyboards, touch screens, microphones, etc.

[0006] System control logic may be configured in any suitable way. In general, the logic can be designed or configured in hardware and / or software. The instructions for controlling the drive circuitry may be hard coded or provided as software. The instructions may be provided by “programming.” Such programming is understood to include logic of any form, including hard coded logic in digital signal processors, application-specific integrated circuits, and other devices which have specific algorithms implemented as hardware. Programming is also understood to include software or firmware instructions that may be executed on a general- purpose processor. System control software may be coded in any suitable computer readable programming language.

[0007] The computer program code for controlling the processes in a process sequence can be written in any conventional computer readable programming language: for example, assembly language, C, C++, Pascal, Fortran, or others. Compiled object code or script is executed by the processor to perform the tasks identified in the program. Also as indicated, the program code may be hard coded.

[0008] The controller parameters relate to process conditions, such as, for example, process gas composition and flow rates, temperature, pressure, cooling gas pressure, substrate temperature, and chamber wall temperature. These parameters are provided to the user in the form of a recipe and may be entered utilizing the user interface.Docket No. LAM1P014WO

[0009] Signals for monitoring the process may be provided by analog and / or digital input connections of the system controller 1542. The signals for controlling the process are output on the analog and digital output connections of the system 1500.

[0010] The system software may be designed or configured in different ways. For example, various chamber component subroutines or control objects may be written to control operation of the chamber components necessary to carry out the deposition processes in accordance with the disclosed embodiments. Examples of programs or sections of programs for this purpose include substrate positioning code, process gas control code, pressure control code, and heater control code.

[0011] In some implementations, a system controller 1542 is part of a system, which may be part of the above-described examples. Such systems can include semiconductor processing equipment, including a processing tool or tools, chamber or chambers, a platform or platforms for processing, and / or specific processing components (a wafer pedestal, a gas flow system, etc.). These systems may be integrated with electronics for controlling their operation before, during, and after processing of a semiconductor wafer or substrate. The electronics may be referred to as the “controller,” which may control various components or subparts of the system or systems. The system controller 1542, depending on the processing requirements and / or the type of system, may be programmed to control any of the processes disclosed herein, including the delivery of processing gases, temperature settings (e.g., heating and / or cooling), pressure settings, vacuum settings, power settings, radio frequency (RF) generator settings in some systems, RF matching circuit settings, frequency settings, flow rate settings, fluid delivery settings, positional and operation settings, wafer transfers into and out of a tool and other transfer tools and / or load locks connected to or interfaced with a specific system.System Controllers

[0012] Broadly speaking, a controller may be defined as electronics having various integrated circuits, logic, memory, and / or software that receive instructions, issue instructions, control operation, enable cleaning operations, enable endpoint measurements, and the like. Such controller may be used in or with any of the apparatus described herein. The integrated circuits may include chips in the form of firmware that store program instructions, digital signal processors (DSPs), chips defined as application specific integrated circuits (ASICs), and / or one or more microprocessors, or microcontrollers that execute program instructions (e.g., software). Program instructions may be instructions communicated to the controller in the form of various individual settings (or program files), defining operational parameters for carrying out a particular process on or for a semiconductor wafer or to a system. The operationalDocket No. LAM1P014WO parameters may, in some embodiments, be part of a recipe defined by process engineers to accomplish one or more processing steps during the fabrication of one or more layers, materials, metals, oxides, silicon, silicon dioxide, surfaces, circuits, and / or dies of a wafer.

[0013] A system controller may be a part of or coupled to a computer that is integrated with, coupled to the system, otherwise networked to the system, or a combination thereof. For example, a system controller may be in the “cloud” or all or a part of a fab host computer system, which can allow for remote access of the wafer processing. The computer may enable remote access to the system to monitor current progress of fabrication operations, examine a history of past fabrication operations, examine trends or performance metrics from a plurality of fabrication operations, to change parameters of current processing, to set processing steps to follow a current processing, or to start a new process. In some examples, a remote computer (e.g., a server) can provide process recipes to a system over a network, which may include a local network or the Internet. The remote computer may include a user interface that enables entry or programming of parameters and / or settings, which are then communicated to the system from the remote computer. In some examples, the controller receives instructions in the form of data, which specify parameters for each of the processing steps to be performed during one or more operations. It should be understood that the parameters may be specific to the type of process to be performed and the type of tool that the controller is configured to interface with or control. Thus, as described above, the controller may be distributed, such as by including one or more discrete controllers that are networked together and working towards a common purpose, such as the processes and controls described herein. An example of a distributed controller for such purposes would be one or more integrated circuits on a chamber in communication with one or more integrated circuits located remotely (such as at the platform level or as part of a remote computer) that combine to control a process on the chamber.

[0014] Without limitation, example systems may include a plasma etch chamber or module, a deposition chamber or module, a spin-rinse chamber or module, a metal plating chamber or module, a clean chamber or module, a bevel edge etch chamber or module, a physical vapor deposition (PVD) chamber or module, a CVD chamber or module, an ALD chamber or module, an atomic layer etch (ALE) chamber or module, an ion implantation chamber or module, a track chamber or module, and any other semiconductor processing systems that may be associated or used in the fabrication and / or manufacturing of semiconductor wafers.

[0015] As noted above, depending on the process step or steps to be performed by the tool, the controller might communicate with one or more of other tool circuits or modules, other tool components, cluster tools, other tool interfaces, adjacent tools, neighboring tools, tools locatedDocket No. LAM1P014WO throughout a factory, a main computer, another controller, or tools used in material transport that bring containers of wafers to and from tool locations and / or load ports in a semiconductor manufacturing factory.

[0016] A system controller may include various programs. A substrate positioning program may include program code for controlling chamber components that are used to load the substrate onto a pedestal or chuck and to control the spacing between the substrate and other parts of the chamber such as a gas inlet and / or target. A process gas control program may include code for controlling gas composition, flow rates, pulse times, and optionally for flowing gas into the chamber prior to deposition in order to stabilize the pressure in the chamber. A pressure control program may include code for controlling the pressure in the chamber by regulating, e.g., a throttle valve in the exhaust system of the chamber. A heater control program may include code for controlling the current to a heating unit that is used to heat the substrate. Alternatively, the heater control program may control delivery of a heat transfer gas such as helium to the wafer chuck.

[0017] Examples of chamber sensors that may be monitored during deposition include mass flow controllers, pressure sensors such as manometers, and thermocouples located in the pedestal or chuck. Appropriately programmed feedback and control algorithms may be used with data from these sensors to maintain desired process conditions.

[0018] The foregoing describes implementation of disclosed embodiments in a single or multichamber semiconductor processing tool. The apparatus and process described herein may be used in conjunction with lithographic patterning tools or processes, for example, for the fabrication or manufacture of semiconductor devices, displays, LEDs, photovoltaic panels, and the like. Typically, though not necessarily, such tools / processes will be used or conducted together in a common fabrication facility. Lithographic patterning of a film typically includes some or all of the following steps, each step provided with a number of possible tools: (1) application of photoresist on a workpiece, i.e., substrate, using a spin-on or spray-on tool; (2) curing of photoresist using a hot plate or furnace or UV curing tool; (3) exposing the photoresist to visible or UV or x-ray light with a tool such as a wafer stepper; (4) developing the resist so as to selectively remove resist and thereby pattern it using a tool such as a wet bench; (5) transferring the resist pattern into an underlying film or workpiece by using a dry or plasma- assisted etching tool; and (6) removing the resist using a tool such as an RF or microwave plasma resist stripper.

[0019] Although the foregoing embodiments have been described in some detail for purposes of clarity of understanding, it will be apparent that certain changes and modifications may beDocket No. LAM1P014WO practiced within the scope of the appended claims. Embodiments disclosed herein may be practiced without some or all of these specific details. In other instances, well-known process operations have not been described in detail to not unnecessarily obscure the disclosed embodiments. Further, while the disclosed embodiments will be described in conjunction with specific embodiments, it will be understood that the specific embodiments are not intended to limit the disclosed embodiments. It should be noted that there are many alternative ways of implementing the processes, systems, and apparatus of the present embodiments. Accordingly, the present embodiments are to be considered as illustrative and not restrictive, and the embodiments are not to be limited to the details given herein.

Claims

Docket No. LAM1P014WOCLAIMSWhat is claimed is:

1. A method for detecting rapidly changing phenomena, the method comprising: responsive to an occurrence of a phenomenon, determining one or more changes in an intensity of optical signals, the optical signals obtained by an optical sensor; detecting one or more changes in voltage correlating to the one or more changes in intensity of optical signals; and based on the one or more changes in voltage staying above a threshold, generating images representative of the phenomenon and comprising indications of the one or more changes in intensity of the optical signals occurring at corresponding pixels of the images.

2. The method of claim 1, wherein the determining of the one or more changes in voltage comprises determining a positive change in voltage associated with the optical signals with respect to time.

3. The method of claim 1, wherein the determining of the one or more changes in voltage comprises determining a negative change in voltage associated with the optical signals with respect to time.

4. The method of claim 1, further comprising detecting the occurrence of the phenomenon based on a certainty level associated with the occurrence of the phenomenon exceeding a threshold.

5. The method of claim 1, wherein the phenomenon comprises a plasma ignition, plasma motion, plasmoid formation or movement, a plasma discharge, an electrical arc, motion of a component of a processing apparatus, breakage of a component of the processing apparatus, breakage of wafer undergoing processing in the processing apparatus, or a hollow cathode discharge (HCD), or a ceasing thereof.

6. The method of claim 5, wherein: the phenomenon comprises the electrical arc within the processing apparatus; and the method further comprises, based on a computerized visual evaluation of the pixels of the images, identifying one or more images of the generated images associated with the electrical arc.Docket No. LAM1P014WO7. The method of claim 5, wherein: the phenomenon comprises the HCD within the processing apparatus; and the method further comprises, based on a computerized visual evaluation of the pixels of the images, identifying one or more images of the generated images associated with the HCD.

8. The method of claim 5, wherein: the phenomenon comprises the motion of the component of the processing apparatus, the motion comprising an actuation of a valve; and the method further comprises determining a degradation of the valve based on a frequency of a vibration occurring during the actuation being under a frequency threshold, or based on a change in the frequency of the vibration occurring during the actuation, the frequency of the vibration determined based on the generated images associated with the motion of the component.

9. The method of claim 5, wherein the phenomenon comprises breakage of a component, and wherein the component is one of an edge ring or a lift pin.

10. The method of claim 1, wherein an effective sampling rate of the generated images is at least 500 frames per second, the at least 500 frames being representative of changes occurring within an accumulation time of the optical signals during the occurrence of the phenomenon.

11. The method of claim 10, wherein the effective sampling rate of the generated images is at least 100,000 frames per second, the at least 100,000 frames being representative of changes occurring within an accumulation time of the optical signals during the occurrence of the phenomenon.

12. The method of claim 1, wherein the generating of the images is based on an accumulation time of the optical signals obtained by the optical sensor.

13. The method of claim 11 , wherein the accumulation time comprises a time period of at least 100 nanoseconds (ns).Docket No. LAM1P014WO14. The method of claim 1, further comprising determining a difference between at least a first image and a second image of the generated images.

15. The method of claim 14, further comprising, based on the difference between at least the first image and the second image exceeding a threshold, causing the processing apparatus to adjust an ongoing process at a processing apparatus.

16. The method of claim 14, further comprising determining a difference between at least a third image and a fourth image of the generated images different from the first and second images, based on the difference between at least the first image and the second image not exceeding a threshold.

17. The method of claim 1, further comprising detecting the occurrence of the phenomenon based on a certainty level associated with the occurrence of the phenomenon exceeding a certainty threshold.

18. The method of claim 17, further comprising adjusting an ongoing process at a processing apparatus based on the certainty level associated with the occurrence of the phenomenon exceeding the certainty threshold.

19. The method of claim 18, wherein the adjusting of the ongoing process comprises stopping the ongoing process.

20. A method for mitigating a failure occurring at a processing apparatus, the method comprising: obtaining a plurality of images representative of a phenomenon occurring with respect to the processing apparatus, the plurality of images comprising indications of one or more changes in an intensity of optical signals obtained by an optical sensor; and based on a difference between at least a first image and a second image of the plurality of images exceeding a threshold, causing the processing apparatus to adjust an ongoing process associated with the phenomenon.Docket No. LAM1P014WO21. The method of claim 20, wherein the adjustment of the ongoing process comprises an issuance of a warning or an error, or a shutdown of the processing apparatus or a component thereof.

22. A method for mitigating a failure occurring in a processing apparatus, the method comprising: obtaining a plurality of images representative of a phenomenon occurring with respect to the processing apparatus, each of the plurality of images comprising a plurality of pixels each indicative of a change in an intensity of optical signals obtained by an optical sensor; and based at least on a difference in the intensity of optical signals indicated by at least a first pixel of a first image of the plurality of images and a corresponding first pixel of a second image of the plurality of images exceeding a threshold, causing the processing apparatus to adjust an ongoing process associated with the phenomenon.

23. The method of claim 22, wherein the adjustment of the ongoing process comprises an issuance of a warning or an error, or a shutdown of the processing apparatus or a component thereof.

Images