Uniformity control for plasma processing

Abstract

To provide a different approach to uniformity tuning while an approach to solve a problem of process uniformity attempts to make all of parameters uniform across a wafer and the uniformity tuning is unique for equipment.SOLUTION: A processing device receives data including one or more plasma exposure durations of a plasma process. The plasma exposure duration is associated with a set of controlled elements. The processing device causes each set of controlled elements to switch between a first mode of operation and a second mode of operation. Each set of controlled elements exposes an appropriate portion of a substrate to plasma related fluxes. The first set of controlled elements processes the substrate at an increased rate while operating in the first mode of operation relative to the second mode of operation. The processing device causes each set of controlled elements to operate in the first mode of operation for the appropriate time duration based on the received plasma exposure duration data.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] Some embodiments of the present disclosure generally relate to digital control of plasma processing. Embodiments further relate to systems, devices, and methods for plasma delivery and / or plasma processing. [Background technology]

[0002] Plasma processing is widely used in the semiconductor industry. Plasmas can modify the chemistry of process gases (e.g., generate ions, radicals, etc.), create new species, and generate a flux of ions toward the wafer with energies ranging from a fraction of an electron volt (eV) to thousands of eV, without limitations related to the process temperature. There are many types of plasma sources (e.g., capacitively coupled plasma (CCP), inductively coupled plasma (ICP), microwave-generated plasma, electron cyclotron resonance (ECR), etc.) that cover a wide operating process range from a few mTorr to several Torr.

[0003] Today, a common plasma processing specification is high uniformity of process results (e.g., uniformity across the wafer, right to the wafer's edge). This standard is often extremely difficult to achieve because it involves many factors, many of which interfere with other factors. Plasma uniformity, chamber design, wafer temperature distribution, bias electrode design, etc. are just some of those factors. To meet these criteria, both the RF antenna and the processing chamber are designed to achieve the highest level of process uniformity. This often leads to large dimensions for the chamber and power generator (e.g., antenna, coil, electrode, etc.), a large overall plasma volume, and other costly measures such as complex temperature control, coil-split magnetic field screens, etc. Basic process uniformity within a few percent can be fixed by general tool design, but as uniformity standards become more stringent, even these measures are often insufficient. In that case, the chamber must be equipped with elements that can enable individual tuning of the chamber for a specific process. Furthermore, a large plasma volume, by itself, can be problematic for processes requiring rapid changes in chemistry.

[0004] Similarities can be observed between plasma processing and TVs. Originally, TVs were based on cathode ray tube (CRT) technology, in which an electron beam scanned horizontally and vertically within a vacuum tube, exciting phosphor dots on the tube's front panel, producing approximately 25 frames per second. Each phosphor dot flashes light for a short (fixed) time, and the brightness of this flash is controlled by the electron beam current. A receptor (e.g., the eye) then integrates the brightness of the light coming from every phosphor dot over a short duration and averages the brightness. The color of a dot is determined by the ratio of the average brightness of neighboring color dots, and brightness is determined by the overall intensity of light from these dots. The analog nature of CRT image control lies in controlling the intensity of the electron beam, or the peak brightness of every light flash. CRTs and other analog systems present challenges in maintaining specific processing specifications (e.g., power requirements, chamber size, component specification limitations, etc.) due to the stringent specification requirements for performing the process. Like CRTs, analog systems generally lack the flexibility often required by modern innovations in various areas of process control. For this reason, analog systems are often manufactured to specialized specifications for performing a particular narrow range of processes.

[0005] The growing problems with CRT technology were solved by switching to digital technology, which of course required changes to both the hardware, the signals, and the signal controls. Obviously, switching from analog to digital technology in plasma processing would require changes to both the hardware and the controls. Summary of the Invention

[0006] In one exemplary embodiment, a method includes receiving, by a processing device, data indicating one or more plasma exposure durations associated with a plasma process. Each of the one or more plasma exposure durations may be associated with a set of controlled elements. The processing device may cause the first set of controlled elements to switch between a first mode of operation and a second mode of operation. The first set of controlled elements expose a first portion of the substrate to a plasma-related flux. The first plurality of controlled elements process the substrate at an increased rate while operating in the first mode of operation relative to the second mode of operation. The processing device may cause the first plurality of controlled elements to operate in the first mode of operation for a first duration based on the data.

[0007] In an exemplary embodiment, a system includes a processing chamber and a set of controlled elements disposed within the processing chamber. The set of controlled elements exposes a surface of a substrate disposed within the processing chamber to plasma-related fluxes. The system may include a processing device communicatively coupled to the set of controlled elements and configured to control the plurality of controlled elements. The processing device may be configured to receive data indicating one or more plasma exposure durations of a plasma processing. The one or more plasma exposure durations may each be associated with a set of controlled elements. A first selection of the set of controlled elements may cause the processing device to switch between a first mode of operation and a second mode of operation. The first selection of the set of controlled elements exposes a first portion of the surface of the substrate to the plasma-related fluxes. The first selection of the set of controlled elements may process the substrate at an increased rate while operating in the first mode of operation relative to the second mode of operation. The processing device may cause a first of the plurality of controlled elements to operate in the first mode of operation for a first duration based on the data.

[0008] In an exemplary embodiment, a method includes receiving, by a processing device, data indicating one or more plasma exposure durations associated with a plurality of controlled elements configured to expose a substrate to plasma-related fluxes associated with plasma processing. The processing device may receive a first thickness profile, the first thickness profile including a first set of substrate process result thickness values measured after exposing the substrate to plasma-related fluxes, the plasma-related fluxes associated with the first plurality of controlled elements operating in a first mode for the first plasma exposure duration. The processing device may determine a first plurality of substrate processing rates associated with one or more locations across a surface of the substrate corresponding to the set of controlled elements operating in the first mode of operation. In response to determining the first plurality of substrate processing rates and the second plurality of substrate processing rates, the processing device may modify the data by changing one of the one or more plasma exposure durations.

[0009] The present disclosure is illustrated by way of example, and not limitation, in the figures of the accompanying drawings, in which like reference numerals indicate like elements. It should be noted that different references to "an" or "one" embodiment in this disclosure are not necessarily to the same embodiment, and such references mean at least one. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 illustrates a plasma processing system according to an aspect of the present disclosure. [Figure 2] FIG. 1 illustrates an arrangement of plasma elements with time-dependent zone control, according to aspects of the present disclosure. [Figure 3] FIG. 1 illustrates a plasma processing system according to an aspect of the present disclosure. [Figure 4A] FIG. 1 illustrates a plasma delivery device using linear electrodes, according to an aspect of the present disclosure. [Figure 4B] FIG. 1 illustrates a plasma delivery device using linear electrodes, according to an aspect of the present disclosure. [Figure 4C] FIG. 1 illustrates a plasma delivery device using linear electrodes, according to an aspect of the present disclosure. [Figure 5] FIG. 1 illustrates a plasma processing system using multiple gas injection zones, according to aspects of the present disclosure. [Figure 6] 1 is a flowchart of a method for substrate processing according to an aspect of the present disclosure. [Figure 7] 1 is a flowchart of a method for adjusting a plasma process according to an aspect of the present disclosure. [Figure 8] FIG. 10 illustrates a model training workflow and a model application workflow for plasma uniformity process rate determination and plasma uniformity adjustment according to aspects of the present disclosure. [Figure 9] 1 is a flow chart illustrating an embodiment of a method for training a machine learning model for determining a substrate processing rate and / or plasma processing adjustment modification of plasma processing data, in accordance with an embodiment of the present disclosure. [Figure 10] 1 is a flowchart of a method for using a machine learning model to correct plasma exposure data, according to an aspect of the present disclosure. [Figure 11] FIG. 1 is a block diagram of an exemplary computing device capable of plasma delivery and / or processing, operating in accordance with one or more aspects of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0011] Process uniformity control is a challenging problem in semiconductor processing. Semiconductor processing often involves multiple parameters that affect results across a substrate. Process uniformity is often extremely difficult to achieve. Process uniformity can involve many factors, many of which interfere with other factors. For example, plasma uniformity, chamber design, wafer temperature distribution, bias electrode design, etc. are just some of those factors. Radio frequency (RF) antennas and process chambers are often manufactured and assembled to achieve the highest level of process uniformity. This often leads to large dimensions for the chamber and power generators (e.g., antennas, coils, electrodes, etc.), large overall plasma volumes, and other costly measures such as complex temperature control, coil splitting, magnetic field screens, etc. The control and relationships between each of these factors are difficult and unclear.

[0012] Traditionally, the approach to overcoming process uniformity challenges is to attempt to make all of these parameters uniform across the wafer (e.g., uniform flux across the wafer, uniform reaction rate). The actual flux of every species during processing is difficult to track and measure, and as a result, process uniformity is often confusing. A uniform result may not be the result of having uniform parameters across all species in the flux, but rather the result of one non-uniformity action (e.g., the flux of one species) being compensated for by the action of another non-uniformity. For example, by modifying the substrate film, the same plasma / gas that produces a uniform result on a first film may produce a non-uniform result on a second film. This can result because the relative reactivity of the second film to the incoming species may differ from that of the first film. The above example shows that uniformity adjustments are unique for a given process and / or equipment setup.

[0013] However, the devices, systems, and / or methodologies disclosed herein offer a different approach to plasma processing and uniformity adjustment. The problem of uniformity adjustment can be mitigated, and in some cases eliminated, when used in a time-based operating paradigm. The results of a process (e.g., uniformity or non-uniformity) on a substrate increase with time, regardless of whether the process is removing, depositing, and / or growing material. For example, over time, the process etches deeper, deposits thicker, and so on. Controlling any particular region of the substrate using time exposure rather than the local density of incoming fluxes allows non-uniformities to be easily compensated for by applying these fluxes for different durations. A method for controlling local exposure to process fluxes can include using multiple local sources (e.g., controlled elements) associated with these fluxes.

[0014] To control local exposure to process fluxes, multiple local sources of these fluxes can be used. For example, a plasma source can include an array of tens or hundreds of small plasma sources placed on a substrate that meets this requirement. By grouping the sources into zones and operating these sources (or zones) for different durations (e.g., on / off, on1 / on2), local wafer exposure to the plasma can be controlled. The flux generated by a zone of sources operating in a first mode of operation (e.g., on state) can be independent of the mode of operation of the remaining zones with all other sources. For example, the flux generated by a first zone of a source in the on state can be independent of the state of all other zones of the source.

[0015] In some embodiments, to manage time t(x,y) with target or threshold accuracy, the proposed source can use a digital methodology that involves dividing the processing time into several subfields, each having an exposure duration during which the source is activated or switched between a higher processing speed mode of operation and a lower processing speed mode of operation (e.g., ON1 / ON2). A source in a lower processing speed mode of operation (e.g., ON2 or OFF) may not generate plasma and plasma-related flux. This binary approach can be used for local flux Φ(x,y,t) (e.g., on / off independent of other sources) and digital time control. The binary approach to flux can enable greater precision of control compared to traditional multiple parameter balancing techniques.

[0016] In some embodiments (e.g., equipment limitations, accuracy thresholds below a target threshold, processing complexity below a threshold complexity, etc.), neither binary control of the flux source (on / off or on1 / on2) nor a digital version of the time control is required to achieve time control of wafer exposure. While a binary approach is advantageous, achieving a fixed flux in the off state may not be compatible with all plasma processing conditions. Implementing a binary approach may require complex and expensive measures that may only be practically feasible for extremely demanding applications.

[0017] Embodiments of the present disclosure provide devices, methods, and systems employing independently controlled plasma sources that can be pulled into a high or low operational mode. For example, on / off is a specific type of this mode (e.g., first source on and the rest off, or first source off and the rest on). The embodiments disclosed herein are directed to devices, systems, and processes for controlling plasma processing through time-dependent exposure of plasma-related fluxes by controlled elements using localized two-level control (e.g., zone control) of grouped sources.

[0018] In an exemplary embodiment, a method includes receiving, by a processing device, data indicating one or more plasma exposure durations of a plasma process. Each of the one or more plasma exposure durations may be associated with a set of controlled elements (e.g., a plasma source). The processing device may cause a first set of controlled elements to switch between a first mode of operation (e.g., on / on 1, high power mode, increased flux generation mode) and a second mode of operation (e.g., off / on 2, lower power mode, reduced flux generation mode). The first set of controlled elements expose a first portion of the substrate to a plasma-related flux. The first plurality of controlled elements process the substrate at an increased rate (e.g., increased rate of deposition growth, etch depth, annealing thickness, etc.) while operating in the first mode of operation relative to the second mode of operation. The processing device may cause the first plurality of controlled elements to operate in the first mode of operation for a first duration based on the data.

[0019] In an exemplary embodiment, a system includes a processing chamber and a set of controlled elements disposed within the processing chamber. The set of controlled elements exposes a surface of a substrate disposed within the processing chamber to a plasma-related flux. The system may include a processing device communicatively coupled to the set of controlled elements (e.g., a plasma source, an actuator, a heat source, etc.) and configured to control the plurality of controlled elements. The processing device may be configured to receive data indicating one or more plasma exposure durations of a plasma process (e.g., annealing, deposition, etching, etc.). The one or more plasma exposure durations may each be associated with a set of controlled elements. The processing device may cause a first selection of the set of controlled elements to switch between a first mode of operation and a second mode of operation. The first selection of the set of controlled elements exposes a first portion of the surface of the substrate to the plasma-related flux. The first selection of the set of controlled elements processes the substrate at an increased rate while operating in the first mode of operation relative to the second mode of operation. The processing device may cause the first of the plurality of controlled elements to operate in a first mode of operation for a first duration based on the data.

[0020] In an exemplary embodiment, a method includes receiving, by a processing device, data indicating one or more plasma exposure durations associated with a plurality of controlled elements configured to expose a substrate to plasma-related fluxes related to plasma processing. The processing device can receive a first thickness profile, the first thickness profile including a first set of process result thickness values for the substrate measured after exposing the substrate to the plasma-related fluxes, the plasma-related fluxes associated with the first plurality of controlled elements operating in a first mode for the first plasma exposure duration. The processing device can receive a second thickness profile of the substrate profiles, the second thickness profile including a second set of process result thickness values for the substrate measured after exposing the substrate to the plasma-related fluxes, the plasma-related fluxes associated with the first plurality of controlled elements operating in the first mode for the second plasma exposure duration. The processing device can determine a first plurality of substrate processing rates associated with one or more locations across a surface of the substrate associated with the set of controlled elements operating in the first mode of operation. The processing device may, in response to determining the first plurality of substrate processing rates and the second plurality of substrate processing rates, modify the data by changing one of the one or more plasma exposure durations by the processing device.

[0021] FIG. 1 illustrates a plasma processing system 100 according to an aspect of the present disclosure. The plasma processing system 100 may include a processing chamber 122 and a plasma source 114. The plasma source includes a wall 124 (e.g., for maintaining atmospheric pressure), a gas inlet 102, a gas distribution volume defined by the wall, and a plasma generation panel 120 including multiple controlled elements (e.g., discharge cells 110) that generate plasma when the same voltage V(t) (e.g., a constant voltage) is applied to all cells of the panel 120. The processing chamber 122 may be one or more of an etch chamber, a deposition chamber (including chambers for atomic layer etching / deposition, chemical vapor deposition, physical vapor deposition, or plasma-enhanced versions thereof), an anneal chamber, a photoresist strip chamber, etc. The processing chamber 122 includes a wall 126 that maintains an internal vacuum and provides support for the plasma source 114, a substrate support 118, and a gas outlet 112, and may include features described in connection with the processing chambers of other embodiments. The gas inlet 102 and gas outlet 112 may provide a flow of a feed gas through the processing system under process gas pressure. The feed gas may include any of air, O, N, Ar, NH, He, and / or other suitable process gases. The plasma source 114 may include a gas expansion volume of a gas injector (e.g., without plasma) that provides a uniform gas flow through the plasma generation panel 120. The uniform gas flow across the surface of the control plate may result in a common gas composition flowing through each of the gas feed lines associated with the discharge cells 110.

[0022] As shown in FIG. 1 , the plasma source 114 may include a radio frequency (RF) (e.g., above 250 kHz) or low frequency (LF) generator 108 (e.g., below 200 kHz) designed to activate (e.g., power) the discharge cells 110. The plasma source 114 may include a controlled switch 104 that connects the discharge cells to the matcher 106 only when closed and disconnects the discharge cells 110 from the matcher when open. Applying an RF voltage to a collection (e.g., zone) of the discharge cells 110 in closed loop with the matcher 106 causes that collection of the discharge cells 110 to emit plasma-related flux. In some embodiments, the generator 108 (e.g., using RF) may be designed to increase or decrease power (e.g., via a signal received by the matcher 106) to maintain a predetermined voltage regardless of the number of zones of discharge cells configured in closed loop with the matcher (e.g., in an on state). For example, in low frequency (LR) mode, no matching may be required and switching between on and off states may result in less crosstalk than when using RF.

[0023] In some embodiments, the discharge cells 110 may not be turned on or off instantly (e.g., by matching between an RF generator and a source). For example, if each cell were turned on and off individually, there would be thousands of instantiated RF on individual circuit loops, which the matcher 106 may not be able to implement. The range of the matcher 106 may not be sufficient for all possible loads (e.g., independent discharge cell voltage matching). The discharge cells may be combined into one or more zones 130A-C. For example, as shown in FIG. 1, the panel 120 may include discharge cells from a center zone 130A, a middle zone 130B, and an edge zone 130C. The use of zones can limit the closed / open loop switching of the switch 104, thereby reducing the load requirements of the matcher 106 to match the instantaneous load switching between individual discharge cells 110. A small number of zones may also not require addressing of individual discharge cells 110.

[0024] In some embodiments, the RF power level can be synchronized with the number of sources turned on or maintained at a fixed level. In the former case, the flux from each operating source can remain fixed regardless of the number of zones turned on, similar to systems with individually addressed discharge cells 110. Time corrections can be performed iteratively and refined on an individual discharge cell 110 basis. For example, methods 700-800 of FIGS. 7-8 describe exemplary methodologies for processing substrates and determining modifications to a process recipe (e.g., exposure duration).

[0025] In the latter case, the flux in the turned-on zones depends on the number of such zones (turned on), resulting in a non-binary process. To create a process recipe, the process rate for each zone configuration is measured. This information may be uploaded to memory (stored locally in matcher 106, RF generator 108, or a separate processing device). In some embodiments, the process recipe may be stored as a process rate matrix, and plasma processing system 100 may calculate a process rate matrix according to a substrate processing algorithm (e.g., using methods 600 and / or 700) to create uniform wafer results. TIFF2025118636000002.tif8170 can be created. Note that multiple recipes and / or processing algorithms can be used to achieve processing results within the target threshold window.

[0026] 2 shows an arrangement of plasma elements 200 with time-dependent zone control according to an embodiment of the present disclosure. The arrangement of plasma elements 200 may be used in conjunction with or as part of the plasma processing system 100 of FIG. 1. For example, the control may be applied to local plasma generation above a substrate. In another example, the arrangement of plasma elements 200 may be used in combination with a common plasma source (e.g., inductively coupled plasma), and control of plasma elements 202A-C is used to control the local fluence (integral of flux over time) of energetic ions relative to the substrate.

[0027] As shown in FIG. 2, plasma elements 202A-C may be associated with (e.g., connected to) several zones as described in connection with FIG. 1. For example, plasma element 202A may be associated with a first zone (e.g., a center zone), plasma element 202B may be associated with a second zone (e.g., a middle zone), and plasma element 202C may be associated with a third zone (e.g., an edge zone). The time for which an RF voltage (e.g., an RF bias voltage) is applied to each zone may be controlled. Note that the applied voltage to each zone may be independent of the number of active (e.g., on) plasma elements. In some embodiments, to achieve independent applied voltages, the RF generator may operate in a fixed output voltage mode. When the generator operates in a power control mode, the RF generator may provide a signal from a matching circuit (e.g., matcher 106 in FIG. 1) to adjust the power depending on the number of active plasma elements 202A-C.

[0028] The time required to drive each of the multiple zones of the plasma element can be stored as a processing recipe. For example, if only three zones are shown in Figure 2, the processing recipe may include (t(a), t(b), t(c)) representing the exposure duration for each zone. The following is an exemplary method for implementing an exposure file for a system with three zones introduced previously. Assume t(a) < t(b) < t(c). The entire panel (all three zones) operates for a duration of t(a), then zone A is turned off, and only zones B and C operate for a duration of t(b) - t(a), then zone B is turned off, and only zone C operates for the remaining time t(c) - t(b). The processing control of the three zones enables achieving a processing profile with substantially radial uniformity. The plasma elements 202A, 202B, 202C or the three sections of the zone can relax a center-low, center-high, or M-shaped and W-shaped radial profile on the substrate. In some embodiments, all elements can operate in the same way (e.g., at the same emission rate of plasma-related flux), but can operate for an appropriate time (e.g., the exposure duration of plasma-related flux) according to the relevant zones of the plasma element.

[0029] In some embodiments, while the active (on) zones of the plasma element are connected to the RF source, the non-active (off) zones of the plasma element can be grounded or floating (i.e., not connected to anything). In some embodiments, the plasma-related flux for the wafer of all nuclides (e.g., ions, electrons, radicals, etc.) except high-energy ions is independent of the state of the discharge cell. In some embodiments, high-energy ions may only exist above the electrode associated with the active (on) zone.

[0030] Alternatively, plasma elements 202A-C may represent multiple plasma shutters. A plasma source generating a plasma that emits a plasma-related flux may be disposed proximate to the plasma shutter. Each plasma shutter may be designed to switch between an open position and a closed position. While in the open position, the plasma shutter may allow ion and electron flux of the plasma to pass through the control plate, and while in the closed position, the plasma shutter blocks the plasma-related flux from passing through the control plate. For example, the plasma shutters may be associated with zones and controlled to selectively open and close to selectively expose a substrate to plasma-related flux flowing through the associated zone of the plasma shutter.

[0031] FIG. 3 illustrates a plasma processing system 300 according to aspects of the present disclosure. The plasma processing system 300 may include a processing chamber 322 and a plasma source 314. The plasma source may include a gas inlet 302 for directing a feed gas. The processing chamber 322 may include a substrate support structure 318 and a substrate 316. The plasma processing system 300 may include a control panel 320 (e.g., a two-dimensional (2D) control plate) including elements 310 that can be opened and closed. In some embodiments, the elements 310 may be opened and / or closed independently, while in other embodiments, the elements 310 may be associated with a collection of elements 310 or a zone of elements. The zone of elements may be actuated (e.g., opened and / or closed) using a power signal similar to that described with respect to FIG. 1 . In some embodiments, a control device may control the zones of elements 310 and selectively actuate different zones of the control panel 320. For example, a plasma processing recipe may include a duration of plasma exposure and may be processed by the control panel 320 by selectively opening zones of the elements 310 for durations related to the exposure duration of the plasma processing recipe.

[0032] In some embodiments, the plasma shutter includes multiple electrode layers that open and close to allow or block ions and electrons to flow through the plasma shutter (e.g., through a control plate). The plasma shutter may include a first layer of electrodes (e.g., a grid facing the plasma) disposed proximate to and / or to the side of the plasma source. The first layer of electrodes may be maintained at a floating potential. The plasma shutter may include a second layer of electrodes including a first ring of electrodes that is maintained at a floating potential when an associated plasma shutter of the multiple plasma shutters is open and at a negative potential (reflects electrons) when an associated plasma shutter of the multiple plasma shutters is closed. The shutter may include a third layer of electrodes including a ring electrode that is maintained at a floating potential when an associated plasma shutter is open and at a positive potential (reflects ions) when the associated plasma shutter is closed.

[0033] In some embodiments, neutral species flow freely through the plasma shutter, so closing the shutter only reduces the process rate rather than stopping it completely. Control of the shutters in any zone can be provided by connecting or disconnecting the shutter electrodes to a set of voltages (e.g., direct current (DC), alternating current (AC)) using a controlled switch.

[0034] 4A-4C illustrate plasma supply devices 400A-C using linear electrodes according to an embodiment of the present disclosure. The plasma supply devices 400A-C may be used in conjunction with the plasma supply systems 100, 300, or another plasma supply device not described herein. The plasma supply devices may include electrodes B1-B4, G1-G4 positioned such that discharge occurs between the electrodes (e.g., only between the electrodes connected to opposing terminals B and G in the plasma generation zone 418). The electrodes may be offset above the substrate 406 (shown behind the electrodes in FIGS. 4A-4B and on the pedestal 414 below the electrodes in FIG. 4C) and buried within a dielectric surface to isolate the electrodes from the plasma (e.g., as shown in FIG. 4C). The electrodes may be connected to different terminals 402, 404 of a power source (e.g., alternating current (AC) or radio frequency (RF)). A first set of electrodes B1-B4 is connected to a first terminal 402 (e.g., the "B" terminal), and a second set of electrodes G1-G4 is connected to a second terminal 404 (e.g., the "G" terminal). A discharge can be initiated between the B and G electrodes. For example, a discharge can occur between (B1, G1), (B2, G2), (B3, G3), and / or (B4, G4). Connecting and disconnecting the electrodes from the power supply can initiate a discharge between any pair of electrodes. The discharge can be initiated to align with an axis 410 (e.g., an axis parallel to the surface of the substrate). Controlled discharge of the electrodes can control the exposure of the substrate 406 to the plasma along this axis 410 (uniformity control). The electrodes can be controlled by varying the discharge time for any pair of electrodes.

[0035] In some embodiments, linear uniformity control along axis 410 is translated to radial control using wafer rotation. As shown in FIG. 4B, the substrate 406 can be rotated along a rotational path 408 relative to a linear electrode. An electrode across the center of the substrate 406 (Y=0) can create maximum exposure at the center, while an electrode touching the edge (Y≈±R max) can create maximum exposure at the edge. The intermediately disposed electrode pair generates maximum plasma exposure halfway between the edge and center of the substrate 406. The plasma supply device 400B can control the discharge time for every pair of electrodes and can compensate for radial non-uniformities that arise from running the discharge for equal times between all pairs.

[0036] FIG. 5 illustrates a plasma processing system 500 using multiple gas injection zones 502A-C according to an embodiment of the present disclosure. The plasma processing system 500 may include a plasma source 508 and a plasma processing chamber 510. The plasma source may include a gas inlet 504 for directing a feed gas. The plasma source may include multiple gas injection sites with associated switches 520A-C. The multiple gas injection sites may be associated with one or more operating zones 502A-C. The plasma processing system 500 may include a control panel 522 with one or more plasma elements 524. The plasma elements may include RF plasma sources operating at fixed power. The plasma source may include a wall 512 for maintaining pressure within the plasma source and individual walls separating one or more gas injection sites associated with one or more operating zones 502A-C. In some embodiments, as shown in FIG. 5, the gas injection is divided into three substantially circular zones 502A-C, and gas flow through the plasma elements 524 on the control panel 522 is controlled. In other embodiments, various geometric configurations may be used instead of or in addition to the substantially circular zones 502A-C.

[0037] Controlling the process rate of the plasma processing system 500 may include switching gas flow through each operating zone 502A-C on or off (e.g., compared to switching RF power as described in other embodiments). Rotating the gas flow (e.g., closing valve switches 520A-C) may not necessarily reduce the process rate to zero (e.g., the plasma elements may still generate plasma, and diffusive components to the ion and radical flux may still reach the substrate). In some embodiments, binary control may not be realized as on / off, but simply as ON1 / ON2, with two levels of process control based on activation of the operating zones 502A-C resulting from gas flow through the valve switches 520A-C. In some embodiments, the overall flow may be synchronized with switching off zones 502A-C, and the rate (ON2) will decrease to a reduced value related to the pressure of the plasma processing system 500. In some embodiments, the pressure and gas flow through the valve switches 520A-C can be synchronized to have a binary system where the off-zones provide negligible throughput (eg, zero throughput).

[0038] As previously mentioned, in some embodiments, a recipe may be used based on the relationship between the control devices (e.g., valve switches 520A-C) and the controlled devices (e.g., the process rate resulting from activating the associated plasma element 524). Using the on and off rates, the recipe associated with the process result is related to the gas flow through the multiple operating zones 502A-C. For example, plasma processing system 500 may include three zones, and the velocity when all gas lines are open may be expressed in the following manner: r 10 >r 20 >r 30can be associated, where the zones are numbered in descending order of processing speed. When the gas flow through any of these zones is off, the speeds are, respectively, r 11 , r 21 , r 31 . To achieve uniformity, the processing time will need to reverse the order of the processing speeds (e.g., a zone with a faster processing speed should have a shorter activation duration compared to the activation duration of a zone with a slower processing speed). For example, the activation durations can be as follows: t1 < t2 < t3. In this example, the relationship between the activation time, processing speed, and processing result in each zone can be expressed as the following linear equations. r 10 t1 + r 11 (t3 - t1) = h1 = h Equation 1 r 20 t2 + r 21 (t3 - t2) = h2 = h Equation 2 r 30 t3 = h3 = h Equation 3

[0039] Equations 1, 2, and 3 can be represented in the following matrix form. TIFF2025118636000004.tif20170

[0040] Or alternatively, R·T = H Equation 5 is represented as.

[0041] Here, as shown above, R is a matrix, and T and H are vectors. Solving this equation can yield the processing recipe. Conventional matrix equation-solving methods can be utilized to identify the processing recipe vector. T = R -1 ·H Equation 6 For example, the inverse of the relationship matrix R can be determined and operated on along with the process results to identify a process recipe that produces uniform process results within a threshold tolerance.

[0042] In some embodiments, on / off control is not optimal when processing is sensitive to smaller changes in activating and deactivating operating zones 502A-C (e.g., on / off is not instantaneous, or switching an operating zone affects the processing speed of switching operating zones, which affects the processing speed of other static and dynamic operating zones).

[0043] In some embodiments, the plasma processing system may include lamps heated wafers (e.g., instead of substrate 518) combined into several zones (e.g., similar to operating zones 502A-C), using multiple lamps (e.g., instead of plasma element 524). The temperature of each zone is monitored (e.g., similar to the gas flow in FIG. 5), and appropriate power is applied to each zone. Temperature uniformity within each zone may be provided by consistent output performance of each of the lamps. In this embodiment, the cost of lamp control may be minimized, but the cost of replacement and associated corrective actions may be high.

[0044] In some embodiments, temperature uniformity within each zone or across the substrate may involve multiple lamp power levels and digital control of the duty cycle for each lamp (e.g., using addressing). Calibration can be performed using infrared (IR) images of the substrate temperature and, for example, using two power levels (or two voltage levels differing by a threshold percentage (e.g., 10-20%)) to determine the duty cycle for each lamp for either of these levels to provide substrate temperature uniformity within the zone. Switching between two levels, rather than maximum and zero, may extend lamp life. Switching between two levels may also reduce precision requirements for lamp manufacturing (e.g., reducing initial lamp cost and operating costs). For example, instead of periodically replacing lamps, a calibration procedure may be run periodically.

[0045] 6-10 illustrate flow diagrams illustrating exemplary methods 600-1000 relating to uniformity control for plasma processing according to some implementations of the present disclosure. For ease of explanation, the methods 600-1000 are illustrated and described as a series of acts. However, acts according to the present disclosure may occur in various orders and / or simultaneously, as well as with other acts not presented and described herein. Furthermore, not all illustrated acts may be performed to implement the methods 600-1000 according to the disclosed subject matter. Furthermore, those skilled in the art will understand and appreciate that the methods 600-1000 may alternatively be represented as a series of interrelated states via state diagrams or events. The methods 600-1000 may be performed, for example, by the plasma processing systems and devices 100-500 described in connection with FIGS. 1-5. At least some operations of the methods 600-1000 are controlled and / or implemented by a processing chamber controller.

[0046] FIG. 6 is a flowchart of a method 600 for substrate processing according to an embodiment of the present disclosure. Referring to FIG. 6 , at block 601, processing logic receives data indicating one or more plasma exposure durations. The data may include one or more plasma exposure durations associated with selections of controlled elements (e.g., plasma source, plasma shutter, bias electrode, heat source, etc.). For example, a process recipe (e.g., a matrix of exposure durations associated with sets or zones of controlled elements) with control instructions for a process controller (e.g., a power supply, RF or alternating current (AC) generator for controlling one or more controlled elements (e.g., plasma source, plasma shutter, bias electrode, heat source, etc.)). The data may be stored within the plasma processing system (e.g., plasma processing system 100 of FIG. 1 ); alternatively or additionally, the data may be received from a source external to the plasma processing system (e.g., a remote storage medium such as a universal serial bus (USB)).

[0047] In block 602, processing logic causes a selection of controlled elements to switch between a first mode of operation and a second mode of operation. The first mode of operation may be associated with a first level of plasma process control, and the second mode of operation may be associated with a second level of plasma process control. For example, the selection of controlled elements may process the substrate at an increased speed while operating in the first mode of operation relative to the second mode of operation (e.g., one or more locations across the surface of the substrate may be processed at a higher speed while the controlled elements operate in the first mode of operation relative to the second mode of operation).

[0048] In some embodiments, the processing logic may be implemented by a processing device coupled to the power source. The controlled elements may include one or more plasma sources configured to receive power from the power source and generate a plasma-related flux. In some embodiments, the processing device is coupled to a matching circuit. The matching circuit may maintain a first voltage level across the one or more plasma sources when a selection of the controlled elements (e.g., multiple controlled elements) operate in a first mode of operation. The matching circuit may maintain a second voltage level across the one or more plasma sources when a selection of the controlled elements operate in a second mode of operation. In some embodiments, the second voltage level is greater than the first voltage level, and in other embodiments, the second voltage level is less than the first voltage level.

[0049] In some embodiments, switching between the first mode of operation and the second mode of operation may include connecting and / or disconnecting an associated selection of controlled elements from a power source. In some embodiments, as described above, switching between the first mode of operation and the second mode of operation includes increasing / decreasing a power level (e.g., a voltage level) supplied to the selection of controlled elements. In some embodiments, switching between the first mode and the second mode of operation includes opening / closing one or more gas injection valves associated with the plasma processing system.

[0050] In block 603, processing logic causes a selection of the controlled element to operate in a second mode of operation for a first duration based on the data. As mentioned above, the selection of the controlled element may be associated with an operating zone of a plasma processing system (e.g., plasma processing system 100 of FIG. 1).

[0051] At block 604, processing logic determines whether all operating zones have been processed. In response to determining that all operating zones have been processed, processing logic proceeds along the yes path and ends. In response to determining that not all zones have been processed, processing logic proceeds along the no path to block 602 and proceeds with one or more of the remaining zones of the processing element. In some embodiments, processing continues until an end condition for the plasma processing is met (e.g., the processing results meet a threshold condition).

[0052] In some embodiments, a plasma processing system may include a control panel (e.g., control panel 120 of FIG. 1 ) with a first selection of controlled elements disposed in a first region of the control panel and a second selection of controlled elements disposed in a second region of the control panel. The first region may be surrounded by the second region. For example, various zones associated with the controlled elements may provide radial control associated with processing a substrate.

[0053] In some embodiments, one or more zones are processed iteratively. For example, a first zone may be processed, followed by a second zone, followed by a third zone, and so on. In some embodiments, one or more zones are processed at least partially concurrently with one another. For example, a first selection of controlled elements may operate in a first mode of operation at least partially concurrently with a second selection of controlled elements operating in the first mode of operation (e.g., at an increased processing speed, a high power mode, a high gas flow mode, etc.).

[0054] In some embodiments, the power supply may maintain a substantially consistent power level when the controlled element is operating in the first mode and the second mode. For example, a mechanical switch such as a gas inlet valve may be used to open and / or close to vary the gas flow rate, which would switch the controlled element between the first and second modes of operation, but the power supply maintains a consistent voltage across the controlled element while switching between the modes of operation.

[0055] In some embodiments, the selection of one or more of the controlled elements may include one or more sets of linear electrodes. For example, a first set of linear electrodes may be coupled to a first terminal of a power supply (e.g., connected and capable of electrically communicating with each other), and a second set of linear electrodes may be coupled to a second terminal of the power supply. In some embodiments, the processing logic may further cause the first set of linear electrodes and the second set of linear electrodes to discharge a plasma-related flux. The one or more linear electrodes may process the substrate along a first axis parallel to a surface of the substrate.

[0056] In some embodiments, the plasma processing system may include a rotating plate for supporting the substrate during plasma processing. The rotating plate may rotate the substrate about an axis perpendicular to the surface of the substrate. For example, using a linear electrode and a rotating plate, the plasma processing system may be capable of processing the substrate around a 360-degree range of motion by rotating an active processing axis about an axis perpendicular to the surface of the substrate. For example, the processing logic may cause the substrate to rotate about a second axis perpendicular to the surface of the substrate.

[0057] 7 is a flowchart of a method 700 for adjusting plasma processing according to an embodiment of the present disclosure. 700 generally includes processing one or more wafers (e.g., two wafers at a time) using various exposure durations across multiple zones of a controlled element and making a comparison between the resulting processing results (e.g., associated thickness profiles) to determine processing rates associated with the various zones of the controlled element. Modifications to the processing instructions (e.g., changes to one or more exposure durations) can be determined based on the one or more determined processing rates across the surface of the substrate. For example, modifying one of the first data can be in response to determining that a rate of change of thickness values at one or more processing locations meets a threshold value.

[0058] 7, in block 701, processing logic receives data including a set of plasma exposure durations associated with a set of controlled elements. The data may include a process recipe in the form of a vector. For example, each element of the recipe (e.g., a time exposure vector) may be associated with a duration for a given zone of the controlled elements for processing a substrate. In some embodiments, the initial recipe may include uniform values (e.g., each zone of the controlled elements may be activated (e.g., operate in a high processing speed mode or a "first operating mode")) that will be adjusted throughout the processing steps of method 700.

[0059] In some embodiments (e.g., non-binary processing modes of operation), the control elements may not allow independent control of individual plasma sources and / or zones of a plasma source. For example, when involving switching gas injection zones on and off (e.g., the embodiment described in connection with FIG. 5), gas may flow through each plasma source depending on the combination of operating zones that are activated. Using an exemplary three-zone operating system, when all zones are on, gas flows through all zones at a first rate. When the second and / or third operating zones are deactivated (e.g., off), flow to the first operating zone will shift and flow to the second and third sources will be closed (e.g., by valves). In some embodiments (e.g., binary control), the overall flow may be adjusted to maintain independent control of individual plasma sources and / or operating zones, while in other embodiments (e.g., non-binary control), the overall flow is not adjusted depending on the combination of controlled elements and / or operating zones that are activated, and thus substrates may be processed at different rates depending on the combination of controlled elements and / or operating zones that are active at a given moment during plasma processing.

[0060] At block 702, processing logic performs a process on a substrate using a zone of controlled elements. Performing a process on a substrate may include causing the zone of controlled elements to operate in a high process speed mode for an associated exposure duration. For example, processing a substrate may include one or more steps and / or processes associated with one or more blocks of method 600 of FIG. 6.

[0061] At block 703, processing logic receives an associated thickness profile of the substrate generated using zones of controlled elements with associated sets of plasma exposure durations. The thickness profile may be associated with processing individual zones of controlled elements or a combination of one or more zones of controlled elements. For example, one or more zones may be processed for associated plasma exposure durations (t1, t2, t3, ...), and the thickness profile may include resulting processing thickness values (h1, h2, h3, ...).

[0062] At block 704, a set of process rates across the surface of the substrate associated with processing the first zone of the controlled element is determined. In some embodiments, the process rates are determined by solving a matrix equation or a composite matrix equation (e.g., Equation 5, previously described). In some embodiments, the process rates are linear, while in other embodiments, the process rates are not linear and may potentially require complex algorithmic methodologies to determine them (e.g., using a matrix solver, such as decomposition, an iterative matrix solver, and / or a machine learning model), as further described in later embodiments.

[0063] At block 705, processing logic determines whether all operating zones have been processed. In response to determining that all operating zones have been processed, processing logic proceeds along the Yes path to block 706. In response to determining that not all zones have been processed, processing logic proceeds along the No path to block 702 to proceed with one or more of the remaining zones of the processing element. In some embodiments, this processing continues until an end condition for the plasma processing is met (e.g., the processing results meet a threshold condition).

[0064] In some embodiments, one or more zones are processed iteratively. For example, a first may be processed, followed by a second zone, followed by a third zone, and so on. In some embodiments, one or more zones are processed at least partially concurrently with one another. For example, a first selection of controlled elements may operate in a first mode of operation at least partially concurrently with a second selection of controlled elements operating in the first mode of operation (e.g., at an increased processing speed, a high power mode, a high gas flow mode, etc.).

[0065] At block 706, processing logic determines a modification to the data based on one or more sets of process rates across the surface of the substrate. Method 700 may be repeated to obtain multiple modifications to the data. A collection of modifications may be applied to the data to generate updated plasma exposure data that, when processed, produces process results that meet a threshold condition (e.g., meet a target process uniformity).

[0066] 8 illustrates a model training workflow 805 and a model application workflow 817 for plasma uniformity process rate determination and plasma uniformity adjustment according to aspects of the present disclosure. In an embodiment, the model training workflow 805 may be implemented on a server, which may or may not include process rate determination and / or plasma uniformity adjustment application, and the trained model is provided to the uniformity process and / or adjustment application, which may implement the model application workflow 817. The model training workflow 805 and the model application workflow 817 may be implemented by processing logic executed by a processor of a computing device. One or more of these workflows 805, 817 may be implemented, for example, by one or more machine learning modules implementing the processing device and / or other software and / or firmware executing on the processing device.

[0067] The model training workflow 805 will train one or more machine learning models (e.g., deep learning models) to perform one or more determining, predicting, correcting, etc. tasks related to process rate determination and / or plasma process adjustment (e.g., adjusting a process recipe to process results to meet target process result threshold conditions). The model application workflow 817 will apply the one or more trained machine learning models to perform determining and / or adjusting, etc. tasks on plasma process data (e.g., one or more plasma exposure durations, one or more controlled element zone associations). One or more of the machine learning models may receive process result data (e.g., one or more thickness profiles processed with the plasma process data).

[0068] Various machine learning outputs are described herein. Specific numbers and arrangements of machine learning models are described and shown. However, it should be understood that the number and types of machine learning models used and the arrangements of such machine learning models can be modified to achieve the same or similar end results. Therefore, the arrangements of machine learning models described and shown are merely examples and should not be construed as limiting.

[0069] In an embodiment, one or more machine learning models are trained to perform one or more of the following tasks: Each task may be performed by a separate machine learning model; Alternatively, a single machine learning model may perform each of the tasks or a subset of the tasks; Additionally or alternatively, different machine learning models may be trained to perform different combinations of the tasks; In one example, one or several machine learning models may be trained, where the trained machine learning (ML) model is a single shared neural network with multiple shared layers and multiple higher-level separate output layers, each output layer outputting a different prediction, classification, identification, etc. Tasks that the one or more trained machine learning models may be trained to perform are as follows: a. Processing Rate Determination—As previously described, various zones of controlled elements may operate in multiple operating modes (e.g., a high substrate processing rate mode and a low substrate processing rate mode) for various exposure durations (e.g., according to a process recipe). Various zones of controlled elements operating in various operating modes result in various substrate processing rates across multiple locations on the substrate. The model receives the exposure durations and process result profiles (e.g., thickness profiles) and will determine the processing rates that various controlled elements will utilize at various locations across the substrate. b. Plasma Processing Adjustments—As previously described, a plasma process may include a process recipe including a set of plasma exposure durations associated with one or more selections of controlled elements designed to expose a substrate to plasma-related fluxes. The machine learning model may receive the process recipe and the resulting thickness profile associated with processing a substrate using one or more selections of controlled elements (e.g., zones). The model may output data indicative of modifications to the plasma processing recipe (e.g., one or more plasma exposure durations). For example, the modifications, when applied to the process recipe, may result in processing a substrate whose process result profile meets a target threshold condition (e.g., a process uniformity threshold).

[0070] One type of machine learning model that can be used to perform some or all of the above tasks is an artificial neural network, such as a deep neural network. An artificial neural network generally includes a feature representation component with a classifier or recurrent layer that maps features to a desired output space. A convolutional neural network (CNN), for example, hosts multiple layers of convolutional filters. In lower layers, pooling may be performed to address nonlinearities, and a multilayer perceptron is typically added above the lower layers to map the top-layer features extracted by the convolutional layers to a decision (e.g., classification output). Deep learning is a class of machine learning algorithms that uses a cascade of multiple layers of nonlinear processing units for feature extraction and transformation. Each successive layer uses the output from the previous layer as input. Deep neural networks can learn in a supervised (e.g., classification) and / or unsupervised (e.g., pattern analysis) manner. A deep neural network includes a hierarchy of layers, with different layers learning different representation levels corresponding to different levels of abstraction. In deep learning, each level learns to transform its input data into a slightly more abstract and complex representation. In plasma processing tuning, for example, the raw input may be a process result profile (e.g., a thickness profile showing one or more thickness values across the surface of a substrate), a second layer may constitute feature data related to the status of one or more zones of a controlled element of a plasma processing system (e.g., zone orientation, plasma exposure duration, etc.), and a third layer may include a starting recipe (e.g., a recipe used as a starting point to determine an updated process recipe for processing a substrate to produce a process result that meets a threshold criterion). In particular, a deep learning process can independently learn which features should optimally be placed at which level. The "deep" in "deep learning" refers to the number of layers through which data is transformed. More precisely, a deep learning system has substantial credit assignment path (CAP) depth. A CAP is a chain of transformations from input to output. A CAP potentially represents a causal connection between the input and output.For feedforward neural networks, the depth of the CAP may be the depth of the network, which may be the number of hidden layers plus 1. For recurrent neural networks, where a signal may propagate through a layer more than once, the CAP depth is potentially unlimited.

[0071] In one embodiment, the one or more machine learning models are recurrent neural networks (RNNs). An RNN is a type of neural network that includes memory to enable the neural network to capture time dependencies. The RNN is capable of learning an input-output mapping that depends on both current and past inputs. The RNN accounts for past and future flow measurements and will make predictions based on this continuous metrology information. The RNN can be trained using a training data set to generate a fixed number of outputs (e.g., to determine a set of substrate processing rates, to determine modifications to a substrate processing recipe). One type of RNN that can be used is a long short-term memory (LSTM) neural network.

[0072] Training a neural network can be accomplished in a supervised learning fashion, which involves feeding a training dataset consisting of labeled inputs through the network, observing its outputs, defining an error (by measuring the difference between the output and the label value), and using techniques such as deep gradient descent and backpropagation to adjust the network's weights across all its layers and nodes so that the error is minimized. In many applications, repeating this process across many labeled inputs in the training dataset results in a network that can produce correct outputs when presented with inputs that differ from those present in the training dataset.

[0073] For the model training workflow 805, a training data set including hundreds, thousands, tens of thousands, hundreds of thousands, or more plasma processing data 810 (e.g., plasma exposure durations associated with a zone of an element to be controlled) and / or process result data (e.g., one or more thickness profiles associated with the plasma processing data) 812 should be used to form the training data set. In an embodiment, the training data set may also include associated substrate processing rate data 814 (e.g., processing rates across the surface of a surface associated with the plasma processing data 810 and the process result data 812) and process adjustment data (e.g., modifications to the plasma processing data responsive to processing a substrate to meet a threshold criterion such as process uniformity) to form the training data set, where each data point may include various labels or classifications of one or more types of useful information. Each case may include, for example, data indicative of one or more plasma exposure durations, associated thickness profiles of associated substrates processed in association with the one or more plasma exposure durations, substrate processing rates associated with the plasma processing data 810 and the process result data 812, and process adjustment data 816. This data may be processed to generate one or more training data sets 836 for training one or more machine learning models. The machine learning models may be trained, for example, to automate adjustments to a plasma processing process recipe (e.g., determine a modification to the plasma exposure duration, update the recipe, and process another substrate to obtain more data and obtain another modification until a substrate processed with the adjusted process recipe achieves process results that meet a threshold condition, such as process uniformity).

[0074] In one embodiment, generating the one or more training data sets 836 includes collecting one or more plasma process data and process result data associated with the plasma process data. The labels used may depend on what the particular machine learning model is trained to do. For example, to train a machine learning model to perform process rate determinations, the training data set 836 may include data labels indicating locations across the substrate associated with process result measurements and process rate determinations. For example, as described in other embodiments, the process recipe and / or process results may be represented as vectors, and the process rates may be represented as one or more matrices.

[0075] To achieve training, processing logic inputs training dataset(s) 836 to one or more untrained machine learning models. Prior to inputting the first input to the machine learning models, the machine learning models may be initialized. Processing logic trains the untrained machine learning model(s) based on the training dataset(s) to generate one or more trained machine learning models that perform various operations as described above.

[0076] Training may be performed by inputting one or more of the plasma process data 810, the process result data 812, the process rate data 814, and the process adjustment data 816 into the machine learning model one by one.

[0077] A machine learning model processes inputs and generates outputs. An artificial neural network includes an input layer consisting of values at data points. The next layer is called a hidden layer, and nodes in the hidden layer each receive one or more of the input values. Each node includes parameters (e.g., weights) to apply to the input values. Thus, each node essentially inputs the input values into a multivariate function (e.g., a nonlinear mathematical transformation) to produce an output value. The next layer can be another hidden layer or an output layer. In either case, nodes in the next layer receive output values from nodes in the previous layer, and each node applies weights to those values and then generates its own output value. This can be performed in each layer. The final layer is the output layer, with one node for each class, prediction, and / or output that the machine learning model can produce.

[0078] Thus, the output may include one or more predictions or inferences. For example, the output predictions or inferences may include one or more modifications to the plasma processing data (e.g., modifications to one or more plasma exposure durations). Processing logic may cause the substrate to be processed using the updated recipe (e.g., with the identified modifications) and receive an updated thickness profile. Processing logic may compare the updated thickness profile to a target thickness profile and determine whether a threshold criterion is met (e.g., thickness values measured across the surface of the wafer fall within a target threshold window). Processing logic determines an error (i.e., classification error) based on the difference between the updated thickness profile and the target thickness profile. Processing logic adjusts the weights of one or more nodes in the machine learning model based on the error. An error term or delta may be determined for each node in the artificial neural network. Based on this error, the artificial neural network adjusts one or more of its parameters (weights for one or more inputs of the node) for one or more of its nodes. The parameters may be updated in a backpropagation manner, so that nodes in the top layer are updated first, followed by nodes in the next layer, and so on. An artificial neural network includes multiple layers of "neurons," each layer receiving values as inputs from neurons in the previous layer. The parameters for each neuron include weights associated with the values received from each of the neurons in the previous layer. Adjusting the parameters may therefore include adjusting the weights assigned to each of the inputs for one or more neurons in one or more layers in the artificial neural network.

[0079] Once the model parameters are optimized, model validation may be performed to determine whether the model has improved and to determine the current accuracy of the deep learning model. After one or more rounds of training, the processing logic may determine whether a stopping criterion has been met. The stopping criterion may be a target level of accuracy, a target number of processed images from the training dataset, a target amount of change to the parameters over one or more previous data points, a combination thereof, and / or other criteria. In one embodiment, the stopping criterion is met when at least a minimum number of data points have been processed and at least a threshold accuracy has been achieved. The threshold accuracy may be, for example, 70%, 80%, or 90% accuracy. In one embodiment, the stopping criterion is met when the accuracy of the machine learning model stops improving. If the stopping criterion is not met, further training is performed. If the stopping criterion is met, training may be completed. Once the machine learning model is trained, a reserved portion of the training dataset may be used to test the model.

[0080] As an example, in one embodiment, a machine learning model (e.g., process rate determiner 867) is trained to determine substrate processing elements. Similar processes can be performed to train machine learning models to perform other tasks, such as those described above. A large set (e.g., thousands to millions) of process result profiles (e.g., thickness profiles) can be collected, and process rate data 869 related to predicted or inferred concentrations associated with input data 862 can be determined.

[0081] Once the one or more trained machine learning models 838 are generated, they may be stored in model storage 845 and added to the substrate processing rate determination and / or process adjustment application. The substrate processing rate determination and / or process adjustment application may then use the one or more trained ML models 838, as well as additional processing logic, to implement an automatic mode in which manual user input of information is minimized, and in some cases, even eliminated.

[0082] For the model application workflow 817, according to one embodiment, input data 862 may be input to a process rate determiner 867, which may include a trained neural network. Based on the input data 862, the process rate determiner 867 outputs information (e.g., process rate data 869) indicative of a set of process rate values across the surface of the substrate.

[0083] According to one embodiment, the input data 862 may be input to a plasma process adjuster 864, which may include a trained neural network. Based on the input data 862, the plasma process adjuster 864 outputs corrections to and / or updated plasma exposure data (e.g., process adjustment data 866).

[0084] 9 is a flowchart illustrating an embodiment for a method 900 for training a machine learning model for determining substrate processing rates and / or plasma processing adjustment modifications for plasma processing data, in accordance with an embodiment of the present disclosure. At block 902 of method 900, processing logic assembles a training data set, where the training data set may include plasma exposure data and a thickness profile associated with a substrate process using the plasma exposure data along with an associated set of controlled elements. Each data item in the training data set (e.g., a substrate processing recipe and an associated thickness profile) may include one or more known processing rates and / or process recipe modifications.

[0085] At block 904, data items from the training dataset are input into an untrained machine learning model. At block 906, the machine learning model is trained based on the training dataset to generate a trained machine learning model that determines a substrate processing rate and / or modifications to substrate processing data (e.g., process recipe, plasma exposure data). The machine learning model may also be trained to output one or more other types of predictions, classifications, decisions, etc. For example, the machine learning model may also be trained to determine a substrate processing rate and / or modifications to a plasma processing recipe, following an updated substrate processing with an updated plasma processing recipe (e.g., updated plasma exposure duration).

[0086] In one embodiment, at block 910, training data item inputs are input to the machine learning model. The inputs may include plasma processing data (e.g., plasma exposure duration) and process result data (e.g., a thickness profile of a processed substrate associated with the plasma processing data). The output may include predictions and / or inferences of plasma processing rates associated with controlled elements processing the substrate and / or plasma processing data modifications (e.g., modifications to a process recipe, such as modifications to a plasma exposure duration associated with a set of controlled elements).

[0087] At block 914, processing logic may perform an updated plasma process with updated plasma processing data incorporating one or more plasma processing data corrections indicated in one or more outputs from the machine learning model. Processing the substrate using the updated plasma processing data produces updated substrate processing results. The processing logic compares the updated substrate processing results to target substrate processing results. At block 916, processing logic determines an error based on a difference between the updated processing results and the target processing results. At block 918, processing logic adjusts weights of one or more nodes in the machine learning model based on the error.

[0088] At block 920, processing logic determines whether a stopping criterion is met. If the stopping criterion is not met, the method returns to block 910, where another training data item is input to the machine learning model. If the stopping criterion is met, the method proceeds to block 925, where training of the machine learning model is complete.

[0089] In one embodiment, multiple different ML models are trained for substrate processing decisions and / or plasma processing adjustments. Each of the ML models may be trained for decisions and / or classifications on different types of input data. For example, a first ML model may be trained to perform process rate decisions using process result data from multiple substrates processed using the same ML model, a second ML model may be trained to perform process rate decisions from multiple processing steps associated with various operations all performed on the same substrate, and a third ML model may be trained to perform concentration decisions using a combination of data used in the first and second ML models. In one embodiment, a single ML model is trained to perform the operations of the first, second, and third ML models described above.

[0090] 10 is a flowchart of a method 1000 of using a machine learning model to modify plasma exposure data (e.g., plasma exposure durations associated with one or more zones of a controlled element for exposing a substrate to a plasma-related flux) according to an embodiment of the present disclosure. Referring to FIG. 10, at block 1001, processing logic performs a plasma process using the plasma exposure data to generate a substrate having a first thickness profile.

[0091] In one embodiment, performing substrate processing involves activating each zone of the controlled element only once, but for a total processing time t p is divided into a suitable number of subfields, each of which is effectively M times longer than the suitable subfield in the previous embodiment. In this embodiment, some regions may reach the target processing result k earlier. ij, and then stops, while other areas are still processed until the entire substrate reaches the target process result (eg, thickness).

[0092] In another embodiment, every step of the plasma process can be characterized by a fixed time. That time can be replaced by a link to plasma exposure data that controls the processing time for each zone of the controlled element, each of which can have thousands of elements. The exposure data can be stored in a file, such as a uniform vector (all elements equal), which can be easily created manually when no other files exist, or it can be quite complex and utilize complex algorithms to obtain specific processing results.

[0093] At block 1002, processing logic provides the thickness profile as input to a trained machine learning model associated with the target thickness profile. The first thickness associated with processing the process exposure data. The machine learning model can be configured to arrive at a desired target thickness profile. The target thickness profile can be associated with a substrate specification or characteristic.

[0094] At block 1003, processing logic obtains output(s) from the machine learning model that includes modifications to the process exposure data. The machine learning model may receive the first exposure map in various formats. For example, the process exposure data may be received by the machine learning model as a map, array, matrix, series of values, etc., that indicate plasma process exposure instructions.

[0095] At block 1004, processing logic applies one or more of the modifications to the process exposure data to generate modified process exposure data. In some embodiments, the modifications to the process exposure data include changing one or more exposure duration values of the exposed process exposure data.

[0096] At block 1005, processing a substrate with the modified process exposure data to produce a substrate with the target thickness profile. In some embodiments, processing a first substrate with the modified process exposure data produces a substrate with the target thickness profile. In other embodiments, processing a second substrate prior to processing with the previous process exposure data results in the second substrate having the target thickness profile.

[0097] 11 shows a block diagram of an exemplary computing device capable of plasma delivery and / or processing, operating in accordance with one or more aspects of the present disclosure. In various illustrative examples, various components of computing device 1100 may represent various components of a computing device, a controller, and / or a control panel (e.g., similar elements described in connection with FIGS. 1-5).

[0098] The exemplary computing device 1100 may be connected to other computer devices in a local area network (LAN), an intranet, an extranet, and / or the Internet. The computing device 1100 may operate in the capacity of a server in a client-server network environment. The computing device 1100 may be a personal computer (PC), a set-top box (STB), a server, a network router, switch or bridge, or any device capable of executing a set of instructions (sequential or otherwise) that specify actions to be taken by that device. Furthermore, while only a single exemplary computing device is shown, the term "computer" shall also be taken to include any collection of computers that individually or together execute a set (or sets) of instructions to perform any one or more of the methodologies described herein.

[0099] The exemplary computing device 1100 may include a processing device 1102 (also referred to as a processor or CPU), a main memory 1104 (e.g., read-only memory (ROM), flash memory, dynamic random access memory (DRAM) such as synchronous DRAM (SDRAM), etc.), a static memory 1106 (e.g., flash memory, static random access memory (SRAM), etc.), and a secondary memory (e.g., a data storage device 1118), which may communicate with each other via a bus 1130.

[0100] The processing device 1102 represents one or more general-purpose processing devices, such as a microprocessor, a central processing unit, etc. More specifically, the processing device 1102 may be a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, a processor implementing other instruction sets, or a processor implementing a combination of instruction sets. The processing device 1102 may also be one or more special-purpose processing devices, such as an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a digital signal processor (DSP), a network processor, etc. According to one or more aspects of the present disclosure, the processing device 1102 may be configured to execute instructions implementing methods 600-1000 shown in FIGS. 6-10.

[0101] The exemplary computing device 1100 may further include a network interface device 1108, which may be communicatively coupled to a network 1120. The exemplary computing device 1100 may further include a video display 1110 (e.g., a liquid crystal display (LCD), a touch screen, or a cathode ray tube (CRT)), an alphanumeric input device 1112 (e.g., a keyboard), a cursor control device 1114 (e.g., a mouse), and an acoustic signal generating device 1116 (e.g., a speaker).

[0102] The data storage device 1118 may include a machine-readable storage medium (or more specifically, a non-transitory machine-readable storage medium) 1128 having stored thereon one or more sets of executable instructions 1122. According to one or more aspects of the present disclosure, the executable instructions 1122 may comprise executable instructions related to performing the methods 600-1000 illustrated in FIGS.

[0103] The executable instructions 1122 may also reside, completely or at least partially, within the main memory 1104 and / or within the processing device 1102 during execution of the executable instructions 1122 by the exemplary computing device 1100, with the main memory 1104 and the processing device 1102 also constituting computer-readable storage media. The executable instructions 1122 may further be transmitted or received over a network via the network interface device 1108.

[0104] While computer-readable storage medium 1128 is illustrated in FIG. 13 as a single medium, the term "computer-readable storage medium" should be interpreted to include a single medium or multiple media (e.g., centralized or distributed databases, and / or associated caches and servers) that store one or more sets of operating instructions. The term "computer-readable storage medium" should also be interpreted to include any medium that is capable of storing or encoding a set of instructions for execution by a machine that cause the machine to perform any one or more of the methodologies described herein. The term "computer-readable storage medium" should therefore be interpreted to include, but is not limited to, solid-state memory and optical and magnetic media.

[0105] Some portions of the above detailed descriptions have been presented in terms of algorithms and symbolic representations of operations on data bits within a computer memory. These algorithmic descriptions and representations are the means used by those skilled in the data processing arts to most effectively convey the substance of their work to others skilled in the art. An algorithm is here, and generally, conceived to be a self-consistent sequence of steps leading to a desired result. These steps require physical manipulations of physical quantities. Usually, though not necessarily, these quantities take the form of electrical or magnetic signals capable of being stored, transferred, combined, compared, and otherwise manipulated. It has proven convenient at times, principally for reasons of widespread usage, to refer to these signals as bits, values, elements, symbols, characters, terms, numbers, or the like.

[0106] It should be borne in mind, however, that these and similar terms all relate to the appropriate physical quantities and are merely convenient labels applied to these quantities. Unless otherwise expressly stated, as will be apparent from the following description, it will be appreciated that throughout the description, descriptions utilizing terms such as "identifying," "determining," "storing," "adjusting," "causing," "returning," "comparing," "creating," "stopping," "loading," "copying," "throwing," "exchanging, replacing," "performing," and the like refer to the actions and processing of a computer system or similar electronic computing device, which manipulate data represented as physical (electronic) quantities in the computer system's registers and memory, and transform that data into other data that are similarly represented as physical quantities in the computer system's memory or registers, or other such information storage, transmission, or display device.

[0107] Examples of the present disclosure also relate to apparatus for performing the methods described herein. This apparatus may be specially constructed for the required purposes, or it may be a general-purpose computer system selectively programmed by a computer program stored in the computer system. Such computer program may be stored on a computer-readable storage medium, such as any type of disk, including, but not limited to, optical disks, compact disk read-only memories (CD-ROMs), and magneto-optical disks, read-only memory (ROM), random access memory (RAM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), magnetic disk storage media, optical storage media, flash memory devices, other types of machine-accessible storage media, or any type of medium suitable for storing electronic instructions, each coupled to a computer system bus.

[0108] The methods and displays presented herein are not inherently related to any particular computer or other apparatus. Various general-purpose systems may be used with programs in accordance with the teachings herein, or it may prove convenient to construct more specialized apparatus to perform the required method steps. The required structure for a variety of these systems will be described below. Moreover, the scope of the present disclosure is not limited to any particular programming language. It will be appreciated that a variety of programming languages may be used to implement the teachings of the present disclosure.

[0109] The preceding description sets forth numerous specific details, such as examples of particular systems, components, and methods, to provide a thorough understanding of some embodiments of the present disclosure. However, it will be apparent to those skilled in the art that at least some embodiments of the present disclosure may be practiced without these specific details. In other instances, well-known components or methods have not been described in detail or have been presented in a simple block diagram format to avoid unnecessarily obscuring the present disclosure. Thus, the specific details described are by way of example only. It is contemplated that particular implementations may vary from these illustrative details and still fall within the scope of the present disclosure.

[0110] References throughout this specification to "one embodiment" or "an embodiment" mean that a particular feature, structure, or property described with respect to that embodiment is included in at least one embodiment. Thus, the appearances of the phrases "in one embodiment" or "in an embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the term "or" is intended to mean an inclusive "or" rather than an exclusive "or." When the terms "about" or "approximately" are used herein, this is intended to mean that the stated nominal value is accurate to within ±10%.

[0111] Although the operations of the methods herein are shown and described in a particular order, the order of the operations of each method may be changed such that some operations may be performed in reverse order or such that some operations may be performed, at least in part, simultaneously with other operations. In alternative embodiments, the instructions of separate operations or sub-operations may be in an intermittent and / or alternating manner.

[0112] It should be understood that the above description is illustrative, and not restrictive. Many other embodiments will be apparent to those skilled in the art upon reading and understanding the above description. The scope of the present disclosure should, therefore, be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled.

Claims

1. receiving, by a processing device, data indicative of one or more plasma exposure durations of a plasma treatment, each of the one or more plasma exposure durations associated with a plurality of controlled elements; causing a first plurality of controlled elements to switch, by the processing device, between a first mode of operation and a second mode of operation, the first plurality of controlled elements exposing a first portion of a substrate to a plasma-related flux, and causing the first plurality of controlled elements to switch, processing the substrate at an increased rate while operating in the first mode of operation relative to the second mode of operation; causing, by the processing device, the first plurality of controlled elements to operate in the first mode of operation for a first duration based on the data; A method comprising:

2. causing a second plurality of controlled elements to switch, by the processing device, between the first mode of operation and the second mode of operation, the second plurality of controlled elements exposing a second portion of the substrate to a plasma-related flux, the second plurality of controlled elements processing the substrate at an increased rate while operating in the first mode of operation relative to the second mode of operation; causing, by the processing device, the second plurality of controlled elements to operate in the first mode of operation for a second duration based on the data; and The method of claim 1 further comprising:

3. 3. The method of claim 2, wherein the first plurality of controlled elements are arranged in a first region of a control panel, the second plurality of controlled elements are arranged in a second region of the control panel, and the first region is surrounded by the second region.

4. 3. The method of claim 2, wherein the first plurality of controlled elements operate in the first mode of operation at least partially simultaneously with the second plurality of controlled elements operating in the first mode of operation.

5. 10. The method of claim 1, wherein the processing device is coupled to a power source, and the controlled element comprises one or more plasma sources configured to receive power from the power source and generate the plasma-related flux.

6. 6. The method of claim 5, wherein the processing device is coupled to a matching circuit that will maintain a first voltage level across the one or more plasma sources when the first plurality of controlled elements operate in the first mode of operation, and that will maintain a second voltage level, lower than the first voltage level, across the one or more plasma sources when the first plurality of controlled elements operate in the second mode of operation.

7. 6. The method of claim 5, wherein the processing device is coupled to one or more gas injection valves, and causing a second plurality of controlled elements to switch between the first mode of operation and the second mode of operation is associated with opening or closing the one or more gas injection valves.

8. the plurality of controlled elements comprising a first set of linear electrodes coupled to a first terminal of the power source and a second set of linear electrodes coupled to a second terminal of the power source, the method comprising: causing one of the first set of linear electrodes and one of the second set of linear electrodes to discharge the plasma-related flux by the processing device, wherein the one or more linear electrodes discharge the plasma-related flux that will process the substrate along a first axis parallel to a surface of the substrate. The method of claim 5 further comprising:

9. The method of claim 8 , further comprising rotating the substrate about a second axis perpendicular to the surface of the substrate.

10. The method of claim 1 , wherein the first plurality of controlled elements comprises one or more of a plasma shutter, a bias electrode, or a heat source.

11. a processing chamber; a plurality of controlled elements disposed within the processing chamber, the plurality of controlled elements causing a surface of a substrate disposed within the processing chamber to be exposed to a plasma-related flux; a processing device communicatively coupled to the plurality of controlled elements; receiving data indicative of one or more plasma exposure durations of a plasma treatment, each of the one or more plasma exposure durations associated with a plurality of controlled elements; a first selection of the plurality of controlled elements causing a switch between a first mode of operation and a second mode of operation, the first selection of the plurality of controlled elements resulting in exposing a first portion of the surface of the substrate to the plasma-related flux, the first selection of the plurality of controlled elements causing a switch to process the substrate at an increased rate while operating in the first mode of operation relative to the second mode of operation; causing the first selection of the plurality of controlled elements to operate in the first mode of operation for a first duration based on the data; and a processing device configured to control the plurality of controlled elements to perform A system comprising:

12. a second selection of the plurality of controlled elements causing a switch between the first mode of operation and the second mode of operation, the second selection of the plurality of controlled elements causing a switch to expose a second portion of the surface of the substrate to the plasma-related flux; causing the second selection of the plurality of controlled elements to operate in the first mode of operation for a second duration based on the data; and The system of claim 11 further comprising:

13. 12. The system of claim 11, wherein the processing device is coupled to a power source, and the controlled element receives power from the power source and comprises one or more plasma sources for generating the plasma-related flux.

14. 14. The system of claim 13, wherein the power supply is coupled to a matching circuit that will maintain a first voltage level across the one or more plasma sources when the first plurality of controlled elements operate in the first mode, and that will maintain a second voltage level, less than the first voltage level, across the one or more plasma sources when the first selection of the plurality of controlled elements operate in the second mode.

15. 14. The system of claim 13, wherein the processing device is coupled to one or more gas injection valves, and causing a second plurality of controlled elements to switch between the first mode of operation and the second mode of operation comprises at least one of opening or closing the one or more gas injection valves.

16. 14. The system of claim 13, wherein the first plurality of controlled elements comprises a first set of linear electrodes coupled to a first terminal of the power supply and a second set of linear electrodes coupled to a second terminal of the power supply, and wherein the processing device is further configured to cause one of the first set of linear electrodes and one of the second set of linear electrodes to discharge the plasma-related flux, and wherein the one or more linear electrodes process the substrate along a first axis parallel to a surface of the substrate.

17. receiving, by the processing device, data indicative of one or more plasma exposure durations associated with a plurality of controlled elements configured to expose the substrate to a plasma-related flux associated with the plasma processing; receiving, by the processing device, a first thickness profile of the substrate, the first thickness profile including a first plurality of process result thickness values of the substrate measured after exposing the substrate to the plasma-related fluxes, the plasma-related fluxes associated with a first plurality of controlled elements operating in a first mode of operation for a first plasma exposure duration; determining, by the processing device, a first plurality of substrate processing rates associated with one or more locations across a surface of the substrate corresponding to the plurality of controlled elements operating in the first mode of operation; modifying, by the processing device, the data by changing one of the one or more plasma exposure durations in response to determining the first plurality of substrate processing rates; A method comprising:

18. receiving, by the processing device, a second thickness profile of the substrate, the second thickness profile including a second plurality of process result thickness values of the substrate measured after exposing the substrate to the plasma-related fluxes, the plasma-related fluxes associated with a first plurality of controlled elements operating in a second mode of operation for a second plasma exposure duration, the first plurality of controlled elements processing the substrate at an increased rate while operating in the first mode of operation relative to the second mode of operation; determining a second plurality of substrate processing rates associated with the one or more locations across the surface of the substrate corresponding to the plurality of controlled elements operating in the second mode of operation, the data being modified further in response to determining the second plurality of substrate processing rates; 20. The method of claim 17, further comprising:

19. receiving, by the processing device, a third thickness profile of the substrate, the third thickness profile including a third plurality of process result thickness values of the substrate measured after exposing the substrate to the plasma-related fluxes, the plasma-related fluxes associated with a second plurality of controlled elements operating in the first mode for a third plasma exposure duration; determining, by the processing device, a third plurality of substrate processing rates associated with the one or more locations across the surface of the substrate corresponding to the second plurality of controlled elements operating in the first mode, the data being modified further in response to determining the third plurality of substrate processing rates; 20. The method of claim 17, further comprising:

20. using the data and the first thickness profile as inputs to a machine learning model; obtaining one or more outputs of the machine learning model, the one or more outputs indicative of a modification to the one or more plasma exposure durations; 20. The method of claim 17, further comprising:

Images