Multi-zone heater tuning for substrate heaters

Abstract

The system includes a multi-zone heater having heating elements in the substrate support corresponding to multiple zones of the substrate support, a processing device coupled to the heating elements, accessing a temperature matrix having multiple vectors corresponding to the zones, determining a temperature map of the substrate support by multiplying the temperature matrix by a weight vector, the weight vector including estimated weight values ​​for each vector, determining a target thickness map of a film on the substrate based on the initial thickness map and the temperature map, the initial thickness map including data characterizing an initial thickness across an initial film location at a uniform temperature, iteratively updating the estimated weight values ​​such that the temperature map results in a minimization of the standard deviation of the thickness values ​​in the target thickness map, and using the estimated weight values ​​as control values ​​for the heating elements.

Description

[Technical Field]

[0001] SUMMARY OF THE INVENTION Embodiments of the present disclosure relate to multi-zone heater tuning in a substrate heater. [Background technology]

[0002] Substrate heaters generally include a single heating element located within the substrate support and are simply controlled by a single temperature setpoint. However, a film or process layer being processed on a substrate lying on the substrate support can experience uneven temperature distribution across the substrate support. Even fairly small temperature variations can cause a few percent variation in the thickness of the process film and / or other film-related properties, potentially resulting in inconsistent semiconductor fabrication across the surface of the substrate. Sufficient variation can cause fabricated devices with strict tolerance requirements to have to be scrapped. Summary of the Invention

[0003] Some of the embodiments described herein include a system including a multi-zone heater including multiple heating elements in a substrate support configured to heat a substrate undergoing processing. The multiple heating elements correspond to multiple zones of the substrate support. The system may further include a processing device coupled to the multiple heating elements. The processing device accesses a temperature matrix including multiple vectors corresponding to the multiple zones, where a value in each vector represents a temperature rise across a location on the substrate support as a result of a corresponding heating element of the multiple heating elements being activated at a temperature setpoint. The processing device further determines a temperature map of the substrate support by multiplying the temperature matrix with a weight vector. The weight vector may include estimated weight values ​​for each vector of the temperature matrix. The processing device further determines a target thickness map for a film on a substrate to be processed on the substrate support based on the initial thickness map and the temperature map, the initial thickness map including data characterizing an initial thickness across a location of a first film on a first substrate heated at a uniform temperature. The processing device further iteratively updates the estimated weight values ​​of the weight vector so that the temperature map results in a minimization of the standard deviation of the thickness values ​​in the target thickness map. The processing device will further use the estimated weight values ​​as control values ​​for each of the plurality of heating elements while processing subsequent substrates.

[0004] In a related embodiment, a method for tuning a multi-zone heater is described herein. The method can include correlating each heating element of a plurality of heating elements in a multi-zone heater of a substrate support with one of a plurality of zones of the substrate support configured to heat the substrate during processing. The method can further include accessing, by the processing device, a temperature matrix including a plurality of vectors corresponding to the plurality of zones, where a value in each vector represents a temperature rise across a location on the substrate support as a result of a corresponding heating element of the plurality of heating elements being activated at a temperature setpoint. The method can further include determining, by the processing device, a temperature map of the substrate support by multiplying the temperature matrix by a weight vector, where the weight vector includes estimated weight values ​​for each vector of the temperature matrix. The method can further include determining, by the processing device, a target parameter map of a film on a substrate to be processed on the substrate support based on the initial parameter map and the temperature map, where the initial parameter map includes data characterizing process parameters across a location of a first film on a first substrate heated at a uniform temperature. The method may further include iteratively updating, by the processing device, the estimated weight values ​​of the weight vector such that the temperature map results in a temperature map that minimizes a standard deviation of the parameter values ​​in the target parameter map, and using, by the processing device, the estimated weight values ​​as control values ​​for each of the plurality of heating elements while processing subsequent substrates.

[0005] In a further embodiment, a method for training a machine learning model to generate a temperature matrix usable for tuning a multi-zone heater is described herein. The method can include receiving, by a computing device, a training dataset including a plurality of data items, each of the plurality of data items including a set of temperature values ​​across a location on a substrate support including the multi-zone heater and a set of thickness values ​​for a film on a substrate processed while on the substrate support. The method can further include inputting, by the computing device, the training dataset into a machine learning model. The method can further include training, by the computing device, the machine learning model based on the training dataset to generate a trained machine learning model that updates values ​​for a temperature matrix including a plurality of vectors corresponding to a plurality of zones of the multi-zone heater. The values ​​in each vector represent a temperature increase across a location on the substrate support as a result of a corresponding one of a plurality of heating elements of the multi-zone heater being set to one or more temperature setpoints, where the combination of the plurality of vectors includes a target set of temperature values ​​across the substrate support usable to ensure a threshold level of film uniformity.

[0006] In a further embodiment, a method for training a machine learning model for tuning a multi-zone heater is described herein. The method can include correlating each heating element of a plurality of heating elements in a multi-zone heater of the substrate support with one of a plurality of zones of the substrate support configured to heat the substrate during processing. The method can further include receiving a training dataset including a plurality of data items, each data item of the plurality of data items including a plurality of temperature setpoints and a thickness map associated with the substrate. The plurality of temperature setpoints can control the plurality of heating elements, and the thickness map includes data characterizing film thickness across film locations on the substrate. The method can further include inputting the training dataset into the machine learning model by a processing device. The method can further include training the machine learning model based on the training dataset to generate a trained machine learning model that will receive a target thickness as input and output a set of temperature setpoints for the plurality of heating elements that result in a film having the target thickness.

[0007] In a further embodiment, a method for using a machine learning model to infer input data to tune a multi-zone heater is described herein. The method can include receiving a substrate on a substrate support of a processing chamber, the substrate support configured to heat the substrate during processing using multiple heating elements corresponding to multiple zones of a multi-zone heater in the substrate support. The method can further include receiving, by a processing device of the processing chamber, information indicating a process to be performed on the substrate and a target thickness of a film on the substrate. The method can further include inputting at least one of the process to be performed or the target thickness into a trained machine learning model. The trained machine learning model can output a set of temperature setpoints for multiple heating elements associated with the process, where the set of temperature setpoints results in a film on the substrate having a target thickness with a standard deviation of thickness values ​​across locations on the substrate that is less than a threshold. The method can further include setting, by the processing device, the multiple heating elements to operate at the set of temperature setpoints. The method can further include processing the substrate in the processing chamber using the multiple heating elements at the set of temperature setpoints.

[0008] Numerous other features are provided in accordance with these and other embodiments of the present disclosure. Other features and embodiments of the present disclosure will become more fully apparent from the following detailed description, the claims, and the accompanying drawings.

[0009] The present disclosure is illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings in which like references indicate like elements. It should be noted that various references to "one" or "an" 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 is a block diagram of a control system including a multi-zone heater in a substrate support according to various embodiments. [Figure 2A] 2A-2C are top views of various embodiments of substrate supports that may be used as substrate supports in the control system of FIG. 1. [Figure 2B] 2A-2C are top views of various embodiments of substrate supports that may be used as substrate supports in the control system of FIG. 1. [Figure 2C] 2A-2C are top views of various embodiments of substrate supports that may be used as substrate supports in the control system of FIG. 1. [Figure 2D] 2A-2C are top views of various embodiments of substrate supports that may be used as substrate supports in the control system of FIG. 1. [Figure 3A] 1 is a schematic flow diagram of a method for finding a weight vector that minimizes the standard deviation of film thickness across a substrate support according to some embodiments. [Figure 3B] 1 is a flow diagram of a method for finding a weight vector that minimizes the standard deviation of film thickness and uses weight values ​​as control values ​​for multiple heating elements, according to some embodiments. [Figure 4] 1A-1C illustrate a set of images (e.g., maps) showing the temperature contours of an overlay of seven heating elements according to various embodiments. [Figure 5] FIG. 10 illustrates a set of images showing an initial thickness map, an optimized temperature map, and a new thickness map after tuning, according to one embodiment. [Figure 6A] 1 is a schematic flow diagram of a method for finding a weight vector that minimizes the standard deviation of a processing parameter of a film across a substrate support, according to some embodiments. [Figure 6B] 1 is a flow diagram of a method for finding a weight vector that minimizes a standard deviation of a processing parameter of a film across a substrate support and uses the weight vector values ​​as control values ​​for a plurality of heating elements, according to some embodiments. [Figure 7]1 is a schematic flow diagram of a method for finding a weight vector that minimizes the standard deviation of a normalized, weighted combination of thickness and process parameter values ​​across a substrate support, according to some embodiments. [Figure 8A] 1 is a flow diagram of a method for training a machine learning model to generate a temperature matrix that can be used to tune a multi-zone heater, according to various embodiments. [Figure 8B] 1 is a flow diagram of a method for training a machine learning model for tuning a multi-zone heater, according to various embodiments. [Figure 9] 1 is a flow diagram of a method of using a machine learning model to infer input data for tuning a multi-zone heater, according to various embodiments. DETAILED DESCRIPTION OF THE INVENTION

[0011] The embodiments described herein relate to systems and methods for tuning a multi-zone heater located within a substrate support of a processing chamber. The multiple zones, e.g., geographical zones, can be associated with multiple separate heating elements integrated within the substrate support. These geographical zones can be, for example, concentric and / or sectorized zones, as will be described in detail. One approach for tuning the temperature set points of multiple heating elements is to iteratively fix some temperature set points while varying others to find an optimal combination of set points that generally achieves uniformity of film thickness or other film-related parameters. However, with this approach, achieving uniformity can be difficult, and it can take a significant amount of time to finally tune the temperature set points of even two or three heating elements, let alone four or more heating elements (e.g., those of a seven-zone heater). For example, the level of complexity of the tuning process grows exponentially with the number of heating elements in a multi-zone heater.

[0012] To address these complexities in a computer-driven manner that significantly reduces the time required for tuning and significantly increases uniformity, the disclosed system and method uses one or both of a temperature matrix and a machine learning model to generate temperature setpoints for control of multiple heating elements of a multi-zone heater. In one embodiment, the temperature matrix includes multiple vectors corresponding to multiple zones, where the value in each vector represents the temperature rise across a location on the substrate support as a result of a corresponding one of the multiple heating elements being activated at the temperature setpoint. Combining the vectors in this matrix thus allows these temperature rises to be superimposed in a manner that can result in a more uniform temperature across the substrate support. A weight vector composed of estimated weight values ​​can be multiplied with the multiple vectors to generate a temperature map. The target thickness map can be based on the difference between an initial thickness map generated at a uniform temperature and the temperature map. The estimated weight values ​​can be iteratively updated so that the temperature map results in a minimum standard deviation of the thickness values ​​(or other film parameter values) in the target thickness map. The estimated weight values ​​that result after the standard deviation falls below the threshold can be used as (or in the calculation of) control values ​​for the heating elements of the multi-zone heater when processing subsequent substrates.

[0013] In another embodiment, a computing device can train a machine learning model to generate values ​​for the temperature matrix (and optionally also for the weight vector) using a training data set of several data items. Each data item can include a set of temperature values ​​across a location on a substrate support and a set of thickness values ​​for a film on a substrate being processed while located on the substrate support. The computing device can input the training data set into the machine learning model and train the machine learning model based on the training data set to generate a trained machine learning model that updates the values ​​of the temperature matrix. The temperature matrix can then be used in such a process.

[0014] In yet another embodiment, a computing device can train a machine learning model using a training dataset of data items from a previous substrate processing run. For example, each data item can include a set of temperature set points and a thickness map associated with the substrate. The set of temperature set points control multiple heating elements, and the thickness map includes data characterizing film thickness across film locations on the substrate. The computing device can then input the training dataset into the machine learning model and train the machine learning model based on the training dataset to generate a trained machine learning model that will receive a target thickness as input and will output a set of temperature set points for multiple heating elements that will result in a film having the target thickness.

[0015] Thus, advantages of systems and methods implemented according to some embodiments of the present disclosure include, but are not limited to, the ability to effectively tune multiple heating elements of a multi-zone heater, e.g., to determine the temperature setpoint of each heating element within a reasonable period of time and to strict criteria for uniformity of film thickness and / or some other film-related parameter. The disclosed approach can tune the temperature map (which affects the uniformity of film thickness and / or other film-related parameters) in ways not possible without multi-zone tuning knobs associated with multiple heating elements. For example, heat can be concentrated in one or more zones while leaving other zones relatively unheated, thereby providing a high level of granularity in temperature modification across geographic regions of the substrate. Other advantages will be discussed, and still other advantages will be apparent to those skilled in the art having the benefit of this disclosure.

[0016] FIG. 1 is a block diagram of a control system 100 including a multi-zone heater 150 in a substrate support 101 according to various embodiments. The control system 100 includes a pedestal 80 in a processing chamber 10 coupled to the substrate support 101. The processing chamber 10 is adapted to perform a process on a substrate positioned on the substrate support 101. The processing chamber 10 can be an etch chamber, a deposition chamber (e.g., an atomic layer deposition chamber, a chemical vapor deposition chamber, a physical vapor deposition chamber, etc.), or other types of process chambers. The processing chamber 10 can be configured to perform one or more plasma-based processes, such as a plasma etch process or a plasma-enhanced deposition process. In various embodiments, the process can include a film covering the substrate, for example, to include deposition, etching, growth, and anodization processes, where the thickness of the film layer varies with temperature. For example, the film can be thicker at lower temperatures and thinner at higher temperatures. Thus, to produce a more uniform thickness in the film, the computing device 102 can adjust the temperature across the substrate support 101 to be more uniform by varying the temperature set points of the multiple heating elements 154 of the multi-zone heater 150. The temperature set points (or temperature set point control values) can be temperature values ​​and / or power output values ​​or the like.

[0017] In these embodiments, the control system 100 further includes a computing device 102, a display device 115, an input / output (I / O) device 117, and a storage device 120. The computing device 102 can be a computing or computer system. In some embodiments, the computing device 102 includes a processing device 104, a memory 108, and one or more communication interfaces 112 adapted to interface with the multi-zone heater 150, for example, via the pedestal 80, the display device 115, the I / O device 117, and / or the storage device 120. In these embodiments, the storage device 120 stores instructions 124 for including algorithms 126 and other computer code, machine learning (ML) models 128, temperature matrices 130, process parameters 135, data maps 138, zone mapping data 140, a graphical user interface (GUI) 110, and / or training data sets 145. The instructions 124 can also be associated with trained ML models of the ML models 128 and can be trained for execution of such ML models. Other data may be stored on storage device 120, as will be discussed in further detail.

[0018] In various embodiments, the multi-zone heater 150 includes, but is not limited to, multiple heating elements 154, each with its own temperature setpoint that controls the temperature level of that heating element, and one or more sensors 158. Each heating element 154 is associated with one of multiple geographic zones of the substrate support 101 (see, for example, FIG. 1B ) and causes a temperature increase in that zone as well as other locations in other geographic zones on the substrate support 101. The temperature increase caused by a zone decreases the farther the location is from the zone in which the heating element is activated. Thus, the temperature at any given location on the substrate support 101 is a combination of the temperature contributions from all activated heating elements 154.

[0019] In some embodiments, the one or more sensors 158 can be thickness imaging sensors capable of imaging a film located on a substrate located on the substrate support 101. The thickness imaging sensor can detect the thickness across one or more locations of the substrate and provide these thickness values ​​to the computing device 102. In another embodiment, metrology equipment external to the processing chamber 10 is used following processing to determine the thickness of the film. As will be discussed, these thickness values ​​(e.g., a thickness map) can be used in determining a target temperature map or set of target temperature values ​​across the substrate support 101 for subsequent process runs that will minimize the standard deviation of the thickness (or other film-related process parameter) values.

[0020] In various embodiments, processing device 104 is a processor, e.g., a central processing unit (CPU) and / or a graphics processing unit (GPU), and / or other processing device including a processor. Processing device 104 can include one or more general processors, digital signal processors, application-specific integrated circuits, field-programmable gate arrays, digital circuits, optical circuits, analog circuits, combinations thereof, or other now-known or later-developed devices for analyzing and processing data. Processing device 104 can implement instructions 124 or other software programs, e.g., manually programmed or computer-generated code for implementing logical functions. The described logical functions or system elements can process and / or convert analog data sources, such as analog electrical, audio, or video signals, or combinations thereof, into digital data sources for audiovisual purposes, such as for computer processing compatibility, or other digital processing purposes, among other functions.

[0021] The storage device 120 may include a non-transitory computer-readable medium on which one or more sets of instructions 124, e.g., software, may be embedded. In one embodiment, the storage device 120 is a disk or optical drive unit. Furthermore, the instructions 124 may perform one or more of the operations as described herein. The instructions 124 may reside completely or at least partially within the memory 108 and / or within the processing device 104 during execution by the computing device 102.

[0022] Memory 108 and processing device 104 may also include non-transitory computer-readable media as described above. "Computer-readable medium," "computer-readable storage medium," "machine-readable medium," "propagation signal medium," and / or "signal-bearing medium" may include any device that contains, stores, communicates, propagates, or transports software for use by or in connection with an instruction-executable system, apparatus, or device. A machine-readable medium may selectively be, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, device, or propagation medium.

[0023] The I / O devices 117 may include peripheral devices, such as a keyboard or a mouse, configured for user interaction with any of the components of the control system 100. Additionally, the display device 115 may be a liquid crystal display (LCD), a cathode ray tube (CRT), or other display suitable for conveying information via a touchscreen or computer screen. The display device 115 may serve as an interface for a user to view the functionality of the processing device 104 or, specifically, as an interface with software stored in the memory 108 or the storage device 120.

[0024] In some embodiments, the GUI 110 may be presented on the display device 115 to provide an interface to a user via the I / O device 117. The GUI 110 may facilitate submission of a measurement file, including temperature and process parameter constraints. The measurement file may include, for example, an initial thickness map generated (e.g., by a thickness imaging sensor) when the film is exposed to a uniform temperature. For example, within the GUI 110, a user may select a value k for a film-related property, a temperature sensitivity value for a specific percent thickness variation per degree Celsius, a baseline temperature, and the like. The GUI 110 may also provide a "Type" in a selectable drop-down menu to select a type of multi-zone heater, some of which are discussed by way of example with reference to FIGS. 2A-2D .

[0025] In various embodiments, the algorithms 126 may include formulas or other sets of mathematical operations that can operate on stored data related to tuning the multi-zone heater 150. For example, one of the algorithms may correlate multiple zones of the substrate support 101 (see FIG. 1B ) with multiple heating elements 154 configured to heat the substrate during processing to generate zone mapping data 140. The zone mapping data 140 may be used in correlation with finding temperature set points for the multiple heating elements 154 and to determine how combinations of temperature set points affect the data map 138 and / or training data set 145 corresponding to the various zones.

[0026] Other algorithms of algorithm 126 can update and / or operate on temperature matrix 130 along with the corresponding weight vector to determine at least some of data maps 138. Data map 138 can then be used to iteratively update the estimated weight values ​​of the weight vector, as will be discussed in more detail with reference to FIGS. 3A-8A . Data map 138 can include, for example, a temperature map of temperature values ​​across locations of the substrate support 101. Data map 138 can further include a thickness map of film thickness values ​​across locations of a substrate located on the substrate support 101. Data map 138 can further include a parameter map of film-related parameter values ​​across locations of a substrate located on the substrate support 101. In one embodiment, the parameter map is a map of k, which represents the extinction coefficient of a film on a substrate. Other film-related process parameters are also contemplated.

[0027] In another embodiment, the computing device 102 can train the ML model 128 to generate values ​​for the temperature matrix 130 (and optionally also for the weight vector) using a training dataset (e.g., from the training dataset 145) that includes several data items. Each data item can include a set of temperature values ​​across locations on the substrate support 101 and a set of thickness values ​​for a film on a substrate being processed while located on the substrate support 101. The computing device 102 can input the training dataset into the ML model 128 and train the ML model 128 based on the training dataset to generate a trained ML model that updates the values ​​of the temperature matrix 130. The temperature matrix 130 can then be used in tuning the multi-zone heater 150, as will be discussed with reference to FIGS. 3A-7 .

[0028] In an embodiment, the computing device 102 can train the ML model 128 using a training dataset (e.g., from the training dataset 145) that includes data items from a previous substrate processing run. For example, each data item can include a set of temperature set points and a thickness map associated with the substrate. The set of temperature set points control a plurality of heating elements 154, and the thickness map includes data characterizing film thickness across film locations on the substrate. The computing device 102 can then input the training dataset into the ML model 128 and train the ML model 128 based on the training dataset 145 to generate a trained ML model that will receive as input a target thickness and / or a process to be performed and output a set of temperature set points for the plurality of heating elements 154 that will result in a film having the target thickness of the process.

[0029] The machine learning model 128 can be a neural network, a deep learning network, a convolutional neural network, a recurrent neural network, a clustering model, a random forest model, a dimensionality reduction model, a decision tree, a support vector machine, a regression analysis model, a Bayesian network, and / or other types of machine learning models. The machine learning model 128 can be trained using supervised, semi-supervised, or unsupervised learning.

[0030] One type of machine learning model that can be used is an artificial neural network, e.g., 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. Pooling is performed, and nonlinearities can be addressed in lower layers, with multi-layer perceptrons typically added on top to map the top-layer features extracted by the convolutional layers to decisions (e.g., classification outputs).

[0031] Deep learning is a class of machine learning algorithms that uses a series 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. Deep neural networks include a hierarchy of layers, where different layers learn different levels of representation corresponding to different levels of abstraction. In deep learning, each level learns to transform its input data into a slightly more abstract composite representation. In particular, the deep learning process can learn by itself which features to optimally place at which level. The "deep" in "deep learning" refers to the number of layers through which data is transformed. More precisely, deep learning systems have significant credit assignment path (CAP) depth. A CAP is a chain of transformations from input to output. A CAP potentially describes a causal relationship between the input and the output. For feedforward neural networks, the cap depth can be that of the network, or the number of hidden layers plus 1. For recurrent neural networks, where signals can propagate through layers multiple times, the cap depth is potentially unlimited.

[0032] Training a neural network can be achieved with a supervised learning scheme that 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 tuning the network's weights across all its layers and nodes so that the error is minimized using techniques such as deep gradient descent and backpropagation. In many applications, repeating this process across a large number of labeled inputs in the training dataset results in a network that can produce accurate outputs when presented with inputs that differ from those present in the training dataset. In high-dimensional settings, e.g., large images, this generalization is achieved when a sufficiently large and diverse training dataset is made available.

[0033] In some embodiments, the machine learning model is trained to generate a temperature matrix that allows a temperature to be input for a particular zone and the impact on physical outcomes based on that temperature to be determined across multiple zones. Once the temperature matrix is ​​complete, a complete understanding can be achieved of the predicted physical outcomes (e.g., layer thickness) across multiple zones on a wafer based on the temperature setpoints in one or more zones of the multi-zone heater 150. In some embodiments, the machine learning model is trained to output a set of recommended temperature setpoints to achieve a film with layers having a target thickness and / or target uniformity.

[0034] 2A-2D are top views of various embodiments of substrate supports that may be used as the substrate support 101 in the control system 100 of FIG. 1. Each of the substrate supports shown in FIGS. 2A-2D includes multiple zones that may be individually heated, for example, using individual heating elements 154. Each zone may also have its temperature metrics individually monitored and adjusted as needed based on a desired temperature profile. A substrate or wafer (not shown) is to be positioned on the top surface of each substrate support, and is generally centered thereon. Thus, the thickness of a film positioned on the substrate and other processing parameter values ​​associated with such a film are subject to the varying temperature profile.

[0035] 2A shows a top surface of a substrate support 201A having three zones, e.g., an inner zone 205, a middle zone 210, and an outer zone 215. In one embodiment, each of the zones 205, 210, and 215 are concentric. As an example, the inner zone 205 may include a radius of 0 to about 85 millimeters (mm) from the center of the substrate support. As another example, the middle zone 210 may include an inner radius that is substantially the same as the radius of the outer periphery of the inner zone 205 for a radius of about 123 mm. As another example, the outer zone 215 may include an inner radius that is substantially the same as the radius of the outer periphery of the middle zone 210, e.g., about 165 mm, for a radius of about 150 mm or greater, e.g., about 170 mm.

[0036] 2B shows a substrate support 201B in which multiple zones 220A-220C are spread across sections around the substrate support in a pie shape. Each of the zones 220A, 220B, and 220C can be heated similarly or differently depending on the desired processing conditions. The substrate support 201B includes three zones, although the number of zones can be more or less than three.

[0037] In some embodiments, one or more of zones 205, 210, and 215 (or additional concentric zones) are combined with one or more of zones 220A, 220B, and 220C (or additional pie-shaped zones). In these embodiments, the heating elements of each zone can overlap, thereby contributing additively to the heating of different locations in each zone. More specifically, if all of zones 220A, 220B, and 220C are used with all of zones 205, 210, and 215, the heating element of zone 220A would overlap, for example, the upper right sections of all of zones 205, 210, and 215. In this manner, the contributions of multiple heating elements in different zones can overlap to the temperature at different locations across the surface of the substrate support.

[0038] Figure 2C shows a substrate support 201C in which a plurality of zones 220A-220C are provided in a pie shape, similar to the substrate support 201B shown in Figure 2B. However, the substrate support 201C also includes an inner zone 205 similar to the substrate support 201A shown in Figure 2A. The inner zone 205 may extend to a radius of about 50 mm or more, for example, about 80 mm to about 90 mm, from the center of the substrate support 201C.

[0039] 2D shows a substrate support 201D in which a plurality of outer zones 225A and 225B surround an inner zone 205. In one embodiment, each of the plurality of outer zones 225A and 225B is arcuate. In some embodiments, each of the plurality of outer zones 225A and 225B is substantially shaped as a semicircle. The inner zone 205 may extend to a radius of about 50 mm or more from the center of the substrate support 201C, for example, from about 80 mm to about 90 mm in some embodiments.

[0040] Adjustments to the temperature setpoints of the heating elements associated with the zones shown in Figures 2A-2D also affect the substrate through the temperature values ​​at each of the multiple locations across the surface of the substrate support, and the resulting temperature movement. As discussed, temperature values ​​at locations farther away from a zone are less affected by the heating elements in that zone. Conversely, temperature values ​​at locations closer to a zone are more directly affected by the heating elements in that zone. These temperature values ​​can comprise a temperature map of data map 138. The temperature map can be a map representing a wafer divided into multiple zones, where a temperature value can be assigned to each zone.

[0041] In some embodiments, the temperature map is not limited to the zones of a multi-zone heater and can include different temperature values ​​for each point on the wafer. Furthermore, in various embodiments, the computing device 102 is configured to determine a target temperature map for a particular film on the substrate via execution of the algorithm 126 and / or the ML model 128, as will be discussed in the following figures. In these embodiments, the target temperature map can be determined to minimize the standard deviation of thickness values ​​within the target thickness map, for example. The thickness map can be a map representing the wafer (or substrate), where thickness values ​​are assigned to multiple different regions of the thickness map. The regions of the thickness map can be the same as or different from the zones of the temperature map. In embodiments, the thickness map includes thickness values ​​for every point on the wafer.

[0042] 3A is a schematic flow diagram of a method 300A for finding a weight vector 303 that minimizes the standard deviation of film thickness across the substrate support 101 according to some embodiments. As discussed, the temperature matrix 130 includes multiple vectors (here, column vectors), each corresponding to one of multiple zones of the multi-zone heater 150C. Seven zones are shown, but this is merely an example, as the zones of the temperature matrix 130 can correspond to more than one zone of the multi-zone heater 150C.

[0043] A first vector of the temperature matrix 130 can hold a reference temperature value, and each of the other vectors can include a value representing the temperature rise across a location on the substrate support as a result of the corresponding heating element being activated at a temperature setpoint. The values ​​of the multiple vectors are shown by way of example only and can vary significantly depending on the film and different processes being used. As shown, some of the values ​​of the temperature matrix 130 are less than the value of one of the temperature matrices 130, while other values ​​of the temperature matrix 130 are significantly higher, illustrating the different temperature effects of zones at different locations on the substrate support. For example, in some embodiments, the values ​​of one of the multiple vectors of the temperature matrix 130 include a first value for locations within one of the multiple zones corresponding to the vector and a second value for locations not within the zone corresponding to the vector. The first value is greater than the second value, and the second value is smaller the further away from the zone.

[0044] In various embodiments, in operation 304, the computing device 102 calculates the temperature map 311 by matrix multiplication of the weight vector 303 times the temperature matrix 130. The weight vector 303 may include, for example, estimated weight values ​​that will produce a particular linear combination of multiple vectors. As an example in the following equation, a mn The vector is a column vector of the temperature matrix 130, x n The value is the value of the weight vector 303, and b m The value is the value of the temperature map 311. TIFF0007772795000001.tif29170

[0045] In operation 312, the computing device 102 calculates a target thickness map 323 for a film on a substrate to be processed on the substrate support 101 based on the initial thickness map 319 and the temperature map 311. The initial thickness map 319 includes data characterizing the initial thickness across the location of the initial film on the initial substrate heated at a uniform temperature. The initial thickness map 319 can be determined, for example, by the thickness imaging sensor described above. In another embodiment, a metrology instrument external to the processing chamber 10 is used after processing to determine the thickness of the film.

[0046] Referring more specifically to operation 312, the computing device 102 can access the temperature sensitivity value 315, e.g., in the storage device 120 or memory 108, and can determine an updated temperature map by subtracting the uniform temperature from the temperature map 130. The computing device 102 can then multiply the temperature sensitivity value by the updated temperature map to determine a modified thickness map. The temperature sensitivity value referenced throughout this disclosure can be understood as the amount of thickness change per one degree Celsius change in temperature. The temperature sensitivity value 315 is typically negative, such that an increase in temperature causes a decrease in film thickness. However, depending on the process being characterized, the temperature sensitivity value 315 can be positive. The computing device 102 can then determine a target thickness map 323 by adding the modified thickness map to the initial thickness map 319.

[0047] In operation 320, the computing device 102 may update the estimated weight values ​​of the weight vector 303 so that the temperature map 311 results in minimizing the standard deviation of the thickness values ​​in the target thickness map 323. The operation 320 may, in some embodiments, be repeated at least one or more times (e.g., after subsequent iterations of operations 304 and 312) to minimize the standard deviation of the thickness values, e.g., below a threshold value. Once below that threshold, the computing device 102 may adjust the temperature setpoints of the plurality of heating elements 154 using the estimated weight values ​​of the weight vector 303 as control values.

[0048] FIG. 3B is a flow diagram of a method 300B for finding a weight vector that minimizes the standard deviation of film thickness and uses weight values ​​as control values ​​for multiple heating elements, according to some embodiments. Method 300B may be performed by processing logic, which may include hardware (e.g., a processing device, an electrical circuit, dedicated logic, programmable logic, microcode, device hardware, an integrated circuit, etc.), software (e.g., instructions run on or executed by a processing device), or a combination thereof. In some embodiments, method 300B is performed by computing device 102 ( FIG. 1 ) or other computing system having a processing device. While shown in a particular sequence or order, unless otherwise specified, the order of processes can be modified. Therefore, the illustrated embodiments should be understood as merely examples, and the illustrated processes may be performed in a different order, and some processes may be performed in parallel. Additionally, one or more processes may be omitted in various embodiments. Therefore, not all processes are required in every embodiment. Other process flows are possible.

[0049] In operation 350, processing logic accesses a temperature matrix including a plurality of vectors corresponding to a plurality of zones, with the values ​​in each vector representing the temperature rise across a location on the substrate support as a result of a corresponding one of a plurality of heating elements being activated at a temperature setpoint.

[0050] In operation 355, processing logic determines a temperature map of the substrate support by multiplying the temperature matrix with a weight vector, where the weight vector includes estimated weight values ​​for each vector in the temperature matrix.

[0051] In operation 360, processing logic determines a target thickness map for a film on a substrate to be processed on the substrate support based on the initial thickness map and the temperature map. The initial thickness map includes data characterizing an initial thickness across a location of a first film on a first substrate heated at a uniform temperature.

[0052] In operation 365, processing logic iteratively updates the estimated weight values ​​of the weight vector so that the temperature map provides a minimum for the standard deviation of the thickness values ​​in the target thickness map.

[0053] In operation 370, processing logic uses the estimated weight values ​​as control values ​​for each of a plurality of heating elements while processing subsequent substrates.

[0054] FIG. 4 is a set of images (e.g., maps) illustrating the temperature contours of the superposition of multiple heating elements according to various embodiments. Some of the temperature maps may be generated via one or more of zones 205, 210, and 215, one or more of zones 220A, 220B, and 220C, or a combination thereof, as discussed with reference to FIGS. 2A-2D . Temperature map 410 illustrates multiple concentric zones, where the central concentric zone is selectively set to a high temperature. Temperature map 415 illustrates a combination of concentric and pie-shaped zones, where the zone toward the bottom left of the substrate support is selectively set to a high temperature. Temperature map 420 illustrates multiple striped zones progressing from one side of the substrate support to the other, where the left-most zone is selectively set to a high temperature. Temperature map 425 illustrates a combination of concentric zones and striped or pie-shaped zones, where the left-most and central zone is selectively set to a high temperature. Many other combinations of zones to produce different and unique temperature maps are envisioned to optimize film and property thicknesses, as will be detailed.

[0055] FIG. 5 is a set of schematic images showing an initial thickness map 519, an optimized temperature map 511, and a new thickness map 523 after tuning, according to one embodiment. The optimized temperature map 511 can be generated as discussed with reference to the temperature map 311 of FIG. 3A and the operation 355 of FIG. 3B. For example, the optimized temperature map 511 results from selecting estimated weight values ​​for a weight vector such that the temperature map results in a minimum for the standard deviation of the thickness values ​​in the target thickness map. The new thickness map 523 results after a process is performed on a film on a substrate using the optimized temperature map (e.g., the updated weight vector) to control the temperature setpoints of multiple heating elements of the multi-zone heater 150C. The variation in the initial thickness map 519 improved from 4.8% to 2.4%, which is roughly a 50% improvement in this example.

[0056] 6A is a schematic flow diagram of a method 600A for finding a weight vector that minimizes the standard deviation of a processing parameter of a film across a substrate support, in accordance with some embodiments. In this context, the term "processing" may be understood to refer to a deposition, etching, growth, anodization process, or other process in which the thickness of a film layer varies with temperature.

[0057] In various embodiments, in operation 604, the computing device 102 calculates a temperature map 611 by matrix multiplication of a weight vector 603 times a temperature matrix 130. The temperature matrix 130 was previously discussed with reference to FIGS. 1 and 3A-3B. The weight vector 603 may include, for example, estimated weight values ​​that will result in a particular linear combination of multiple vectors. As an example in the following equation: mn The vector is a column vector of the temperature matrix 130, x n The value is the value of the weight vector 603, and b m The value is the value of the temperature map 611. TIFF0007772795000002.tif29170

[0058] In operation 612, the computing device 102 calculates a target parameter map 623 of a film on a substrate to be processed on the substrate support 101 based on the initial parameter map 619 and the temperature map 611. The initial parameter map 619 includes data characterizing values ​​of initial parameters across locations of a first film on a first substrate heated at a uniform temperature. The initial parameter map 619 may be determined, for example, by an imaging sensor or other means of detecting parameter values ​​of the film.

[0059] Referring more specifically to operation 612, the computing device 102 may access the parameter sensitivity values ​​615, for example, in the storage device 120 or memory 108, and determine the updated temperature map by subtracting the uniform temperature from the temperature map 130. The parameter sensitivity values ​​referenced throughout this disclosure may be understood as the amount of parameter value change per one degree Celsius change in temperature. The parameter sensitivity values ​​615 are typically positive, such that an increase in parameter value (e.g., k) causes an increase in film thickness. The computing device 102 may then multiply the parameter sensitivity values ​​by the updated temperature map to determine the modified parameter map. The computing device 102 may then determine the target parameter map 623 by adding the modified parameter map to the initial parameter map 619.

[0060] In operation 620, the computing device 102 may update the estimated weight values ​​of the weight vector 603 so that the temperature map 611 results in minimizing the standard deviation of the parameter values ​​in the target parameter map 623. Operation 620 may, in some embodiments, be repeated at least one or more times (e.g., after subsequent iterations of operations 604 and 612) to minimize the standard deviation of the parameter values, e.g., below a threshold value, once below that threshold. Once below that threshold, the computing device 102 may adjust the temperature setpoints of the plurality of heating elements 154 using the estimated weight values ​​of the weight vector 603 as control values.

[0061] FIG. 6B is a flow diagram of a method 600B for finding a weight vector that minimizes the standard deviation of a processing parameter of a film across a substrate support and uses the weight vector values ​​as control values ​​for multiple heating elements, according to some embodiments. Method 600B may be performed by processing logic, which may include hardware (e.g., a processing device, an electrical circuit, dedicated logic, programmable logic, microcode, device hardware, an integrated circuit, etc.), software (e.g., instructions run on or executed by a processing device), or a combination thereof. In some embodiments, method 600B is performed by computing device 102 ( FIG. 1 ) or other computing system having a processing device. While shown in a particular sequence or order, unless otherwise specified, the order of processes may be modified. Therefore, the illustrated embodiments should be understood as merely examples, and the illustrated processes may be performed in a different order, and some processes may be performed in parallel. Additionally, one or more processes may be omitted in various embodiments. Therefore, not all processes are required in every embodiment. Other process flows are possible.

[0062] In operation 650, processing logic correlates each heating element of a plurality of heating elements in a multi-zone heater of the substrate support with one of a plurality of zones of the substrate support configured to heat the substrate during processing.

[0063] In operation 655, processing logic accesses a temperature matrix including a plurality of vectors corresponding to a plurality of zones, the values ​​in each vector representing the temperature rise across a location on the substrate support as a result of a corresponding one of a plurality of heating elements being activated at a temperature setpoint.

[0064] In operation 660, processing logic determines a temperature map of the substrate support by multiplying the temperature matrix with a weight vector, the weight vector including estimated weight values ​​for each vector in the temperature matrix.

[0065] In operation 665, processing logic determines a target parameter map for a film on a substrate to be processed on the substrate support based on the initial parameter map and the temperature map. The initial parameter map includes data characterizing process parameters across a location of a first film on a first substrate heated at a uniform temperature.

[0066] At operation 670, processing logic iteratively updates the estimated weight values ​​of the weight vector so that the temperature map results in a minimum standard deviation of the parameter values ​​in the target parameter map.

[0067] In operation 675, processing logic uses the estimated weight values ​​as control values ​​for each of a plurality of heating elements while processing subsequent substrates.

[0068] 7 is a schematic flow diagram of a method 700 for finding a weight vector that minimizes the standard deviation of a normalized, weighted combination of thickness and process parameter values ​​across a substrate support, according to some embodiments. Method 700 can be viewed as a combination to optimize for multiple criteria, namely thickness and process parameter, in this example, the extinction coefficient (k).

[0069] In various embodiments, in operation 704, the computing device 102 calculates a temperature map 711 by matrix multiplication of a weight vector 703 times a temperature matrix 130. The weight vector 703 may include, for example, estimated weight values ​​that will produce a particular linear combination of multiple vectors. As an example in the following equation, a mn The vector is a column vector of the temperature matrix 130, x n The value is the value of the weight vector 703, and b m The value is the value of the temperature map 711. TIFF0007772795000003.tif29170

[0070] In operation 712A, the computing device 102 calculates a target thickness map 723A for a film on a substrate to be processed on the substrate support 101 based on the initial thickness map 719A and the temperature map 711. The initial thickness map 719A includes data characterizing an initial thickness across a location of an initial film on an initial substrate heated at a uniform temperature. The initial thickness map 719A may be determined, for example, by a thickness imaging sensor as described above.

[0071] More specifically, with reference to operation 712A, the computing device 102 may determine an updated temperature map by accessing temperature sensitivity values ​​715A, for example, in storage device 120 or memory 108, and subtracting the uniform temperature from temperature map 130. The computing device 102 may then multiply the temperature sensitivity values ​​by the updated temperature map to determine a modified thickness map. The computing device 102 may then determine a target thickness map 723A by adding the modified thickness map to the initial thickness map 719A.

[0072] In operation 712B, the computing device 102 calculates a target parameter map 723B for a film on a substrate to be processed on the substrate support 101 based on the initial parameter map 719B and the temperature map 711. The initial parameter map 719B includes data characterizing values ​​of initial parameters across locations of a first film on a first substrate heated at a uniform temperature. The initial parameter map 719B may be determined, for example, by an imaging sensor or other means of detecting parameter values ​​of the film.

[0073] More specifically, with reference to operation 712B, the computing device 102 may access the parameter sensitivity values ​​715B and determine an updated temperature map by subtracting the uniform temperature from the temperature map 130. The computing device 102 may then multiply the parameter sensitivity values ​​by the updated temperature map to determine a modified parameter map. The computing device may then determine a target parameter map 723B by adding the modified parameter map to the initial parameter map 719B.

[0074] In operation 718, the computing device 102 can combine the target thickness map 723A and the target parameter map 723B using a set of weighting values ​​725, which normalize the combined values ​​into a target-normalized, weighted combination 727 of thickness and process parameter values ​​across the substrate support 101. The weighting values ​​725 are used to calculate the weighting factor of the parameter values ​​(w k ) different thickness value (w thk ) can be weighted in such a way that they sum to 1. For example, if the thickness values ​​are to be given a heavier weight than the parameter values, then w thk can be 0.7, and w k may be 0.3. Normalization may be performed to combine the thickness and k (processing parameter) terms, which may be calculated jointly in operation 720.

[0075] In operation 720, the computing device 102 can update the estimated weight values ​​of the weight vector 703 so that the target normalized, weighted combination 727 of thickness and process parameter values ​​results in a minimized standard deviation of the target normalized, weighted combination 727 of thickness and process parameter values. Operation 720 can be repeated at least one or more times in some embodiments (e.g., after subsequent iterations of operations 704, 712A, 712B, and 718) to minimize the standard deviation of the target normalized, weighted combination 727 of thickness and process parameter values, e.g., to fall below a threshold. Once below that threshold, the computing device 102 can adjust the temperature setpoints of the plurality of heating elements 154 using the estimated weight values ​​203 of the weight vector as control values.

[0076] FIG. 8A is a flow diagram of a method 800A for training a machine learning model to generate a temperature matrix usable for tuning a multi-zone heater, according to various embodiments. Method 800A may be performed by processing logic, which may include hardware (e.g., a processing device, an electrical circuit, dedicated logic, programmable logic, microcode, device hardware, an integrated circuit, etc.), software (e.g., instructions running on or executed by a processing device), or a combination thereof. In some embodiments, method 800A is performed by computing device 102 ( FIG. 1 ) or other computing system having a processing device. While shown in a particular sequence or order, unless otherwise specified, the order of processes may be modified. Therefore, the illustrated embodiments should be understood as merely examples, and the illustrated processes may be performed in a different order, and some processes may be performed in parallel. Additionally, one or more processes may be omitted in various embodiments. Therefore, not all processes are required in every embodiment. Other process flows are possible.

[0077] In operation 810, the processing logic receives a training data set 145 including a plurality of data items, each data item including a set of temperature values ​​across a location of a substrate support including a multi-zone heater and a set of thickness values ​​of a film on a substrate processed while on the substrate support 101.

[0078] At operation 820, processing logic inputs the training dataset 145 into a machine learning model, such as the ML model 128 discussed with reference to Figure 1. In one embodiment, the machine learning model is an artificial neural network, a deep neural network, or other type of trainable artificial intelligence model.

[0079] In operation 830, processing logic trains a machine learning model based on the training data set to generate a trained machine learning model that updates values ​​in a temperature matrix including a plurality of vectors corresponding to a plurality of zones of the multi-zone heater. The values ​​in each vector represent a temperature rise across a location on the substrate support as a result of a corresponding one of a plurality of heating elements of the multi-zone heater being set to one or more temperature setpoints. The combination of the plurality of vectors comprises a target set of temperature values ​​across the substrate support that can be used to ensure a threshold level of film uniformity. In some embodiments, training the machine learning model further includes determining each set of temperature values ​​that can be used in subsequent training iterations by multiplying a weight vector by the temperature matrix before being updated.

[0080] In various embodiments, for subsequent substrates to be processed, method 800A can further include determining a target set of temperature values ​​for the substrate support by multiplying a temperature matrix by a weight vector, where the weight vector includes estimated weight values ​​for each vector of the temperature matrix. Method 800A can further include determining a target set of thickness values ​​for a film on the subsequent substrate based on the target set of temperature values ​​and the immediately preceding set of thickness values ​​for a previously processed substrate. Method 800A can further include iteratively updating the estimated weight values ​​of the weight vector so that the target set of temperature values ​​results in minimization of the standard deviation of the target set of thickness values. Method 800A can further include determining a temperature set point for each of the plurality of heating elements using the estimated weight values ​​while processing the subsequent substrate.

[0081] In a related embodiment, method 800A may further include accessing a temperature sensitivity value, determining an updated set of temperature values ​​by subtracting the set of temperature values ​​from the target set of temperature values, multiplying the temperature sensitivity value and the updated set of temperature values ​​to determine a modified set of thickness values, and determining a target set of thickness values ​​by adding the modified set of thickness values ​​to a previous set of thickness values.

[0082] FIG. 8B is a flow diagram of a method 800B for training a machine learning model for tuning a multi-zone heater according to various embodiments. Method 800B may be performed by processing logic, which may include hardware (e.g., a processing device, an electrical circuit, dedicated logic, programmable logic, microcode, device hardware, an integrated circuit, etc.), software (e.g., instructions running on or executed by a processing device), or a combination thereof. In some embodiments, method 800B is performed by computing device 102 ( FIG. 1 ) or other computing system having a processing device. Although shown in a particular sequence or order, unless otherwise specified, the order of processes can be modified. Therefore, the illustrated embodiments should be understood as merely examples, and the illustrated processes can be performed in a different order, and some processes can be performed in parallel. Additionally, one or more processes may be omitted in various embodiments. Therefore, not all processes are required in every embodiment. Other process flows are possible.

[0083] In operation 850, the processing logic correlates each heating element of the plurality of heating elements 154 in the multi-zone heater 150 of the substrate support 101 with one of the plurality of zones of the substrate support 101 configured to heat the substrate during processing.

[0084] At operation 860, processing logic receives a training data set 145 including a plurality of data items for a plurality of substrates. Each data item of the plurality of data items includes a temperature set point and a thickness map associated with the substrate. The temperature set points can control a plurality of heating elements 154, and the thickness map includes data characterizing film thickness across film locations on the substrate. In some embodiments, processing logic generates each thickness map including thickness values ​​for the film 101 on the substrate using a thickness imaging sensor.

[0085] At operation 870, processing logic inputs the training dataset 145 into a machine learning model, such as the aforementioned ML model 128. In one embodiment, the machine learning model is an artificial neural network, a deep neural network, or other type of trainable artificial intelligence model.

[0086] In operation 880, processing logic trains a machine learning model based on the training dataset 145 to generate a trained machine learning model that will receive as input a target thickness and that will output a set of temperature setpoints for the plurality of heating elements 154 that result in a film having the target thickness. In one embodiment, the target thickness is a substantially uniform thickness with a standard deviation of thickness values ​​across the film locations that is less than a threshold. In one embodiment, training the machine learning model includes correlating, for each substrate, a combination of thickness values ​​of the thickness map across those locations and temperature values ​​as a result of the temperature setpoints across the film locations.

[0087] FIG. 9 is a flow diagram of a method 900 of using a machine learning model to infer input data for tuning a multi-zone heater according to various embodiments. Method 900 may be performed by processing logic, which may include hardware (e.g., a processing device, an electrical circuit, dedicated logic, programmable logic, microcode, device hardware, an integrated circuit, etc.), software (e.g., instructions running on or executed by a processing device), or a combination thereof. In some embodiments, method 900 is performed by computing device 102 ( FIG. 1 ) or other computing system having a processing device. While shown in a particular sequence or order, unless otherwise specified, the order of processes can be modified. Therefore, the illustrated embodiments should be understood as merely examples, and the illustrated processes can be performed in a different order, and some processes can be performed in parallel. Additionally, one or more processes may be omitted in various embodiments. Therefore, not all processes are required in every embodiment. Other process flows are possible.

[0088] At operation 910, processing logic causes the substrate support 101 to receive the substrate onto a substrate support of the processing chamber. The substrate support 101 is configured to heat the substrate during processing using a plurality of heating elements 154 corresponding to a plurality of zones of a multi-zone heater 150 in the substrate support 101.

[0089] At operation 920, processing logic receives information associated with processing chamber 10 and indicative of the process to be performed on the substrate and the target thickness of the film on the substrate.

[0090] At operation 930, processing logic inputs at least one of a process to be performed or a target thickness into a trained machine learning model. In some embodiments, the trained machine learning model outputs a set of temperature set points for the plurality of heating elements 154 associated with the process. In various embodiments, the set of temperature set points results in a film on the substrate having a target thickness with a standard deviation of thickness values ​​across the substrate location that is less than a threshold. The trained machine learning model may have been trained for a particular process chamber or for a particular class of process chambers.

[0091] At operation 940, processing logic sets the plurality of heating elements 154 to operate at the set of temperature setpoints.

[0092] In operation 950, processing logic causes the processing chamber to process the substrate using a process having multiple heating elements 154 set to temperature setpoints.

[0093] The systems and methods disclosed herein can be realized in hardware, software, or a combination of hardware and software. The methods and systems can be realized in a centralized manner in at least one computer system, or in a distributed manner in which different elements are spread across several interconnected computer systems. Any computer system or other device suitable for carrying out the methods described herein is suitable for this disclosure. A typical combination of hardware and software can be a computer system having a computer program that, when loaded and executed, controls the computer system to carry out the methods described herein. Such a programmed computer can be considered a special-purpose computer.

[0094] The methods and systems may also be embodied in a computer program product that includes all features that enable the implementation of the operations described herein and that, when loaded in a computer system, is capable of performing these operations. A computer program in this context means any expression, in any language, code or notation, of a set of instructions intended to cause a system having information processing capabilities to perform a particular function, either directly or after one or both of the following: a) translation into another language, code or notation, b) reproduction in a different material form.

[0095] The present disclosure also relates to apparatus for performing the operations herein. This apparatus may be specially constructed for the intended purposes, or it may comprise a general-purpose computer selectively activated or reconfigured by a computer program stored in the computer. Such a computer program may be stored on a computer-readable storage medium, such as, but not limited to, any type of disk, including floppy disks, optical disks, CD-ROMs, and magneto-optical disks, read-only memory (ROM), random access memory (RAM), EPROM, EEPROM, magnetic or optical cards, or any type of medium suitable for storing electronic instructions, each coupled to a computer system bus.

[0096] The algorithms, operations, 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 carry out the methods. The structure of a variety of these systems will appear as set forth below. Additionally, this disclosure is not described with reference to any particular programming language. It will be understood that a variety of programming languages ​​can be used to implement the teachings of the disclosure as described herein.

[0097] The present disclosure may be provided as a computer program product, or software, which may include a machine-readable medium having stored thereon instructions that can be used to program a computer system (or other electronic device) to perform a process according to the present disclosure. A machine-readable medium includes any mechanism for storing information in a form readable by a machine (e.g., a computer). In some embodiments, the machine-readable (e.g., computer-readable) medium includes a machine (e.g., computer) readable storage medium such as read-only memory ("ROM"), random access memory ("RAM"), magnetic disk storage media, optical storage media, flash memory components, and the like.

[0098] The foregoing 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 but have been presented in simple block diagram form to avoid unnecessarily obscuring the present disclosure. Thus, the specific details set forth are merely exemplary. It is contemplated that particular implementations may vary from these example details and still be within the scope of the present disclosure.

[0099] References throughout this specification to "one embodiment" or "an embodiment" mean that a particular feature, structure, or property described with respect to an 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 do not necessarily all refer to the same embodiment. In addition, the term "or" is intended to mean an inclusive "or" rather than an exclusive "or." When the term "about" or "approximately" is used herein, it is intended to mean that the stated nominal value is accurate to within ±10%.

[0100] 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 certain operations may be performed, at least in part, concurrently with other operations, or certain operations may be performed in reverse order. In alternative embodiments, the instructions or sub-operations of separate operations may be intermittent and / or interleaved.

[0101] It is understood that the foregoing is intended to be illustrative, and not limiting. Numerous other embodiments will become apparent to those skilled in the art upon reading and understanding the foregoing 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. a multi-zone heater comprising a plurality of heating elements in a substrate support configured to heat a substrate undergoing processing, the plurality of heating elements corresponding to a plurality of zones of the substrate support; a processing device coupled to the plurality of heating elements, accessing a temperature matrix including a plurality of vectors corresponding to the plurality of zones, a value in each vector representing a temperature rise across a location on the substrate support as a result of a corresponding one of the plurality of heating elements being activated at a temperature setpoint; determining a temperature map of the substrate support by multiplying the temperature matrix and a weight vector, the weight vector including estimated weight values ​​for each vector of the temperature matrix; determining a target thickness map for a film on a substrate to be processed on the substrate support based on an initial thickness map and the temperature map, the initial thickness map including data characterizing an initial thickness across a location of a first film on a first substrate heated at a uniform temperature; iteratively updating the estimated weight values ​​of the weight vector such that the temperature map provides a minimum for a standard deviation of thickness values ​​in the target thickness map; and Using the estimated weight values ​​as control values ​​for each of the plurality of heating elements during processing of a subsequent substrate. a processing device for performing A system comprising:

2. The system of claim 1 , wherein the minimization of the standard deviation is performed by ensuring that the standard deviation of the thickness values ​​is less than a threshold value.

3. A value in one vector among the plurality of vectors is a first value of a location within one of the plurality of zones corresponding to the vector; and a second value for locations not within the zone corresponding to the vector, the first value being greater than the second value, and the second value decreasing the further away from the zone; and The system of claim 1 , comprising:

4. the processing device: accessing a temperature sensitivity value; determining an updated temperature map by subtracting the uniform temperature from the temperature map; multiplying the temperature sensitivity value by the updated temperature map to determine a modified thickness map; determining the target thickness map by adding the modified thickness map to the initial thickness map; The system of claim 1 , further comprising:

5. Correlating each heating element of a plurality of heating elements in a multi-zone heater of a substrate support with one of a plurality of zones of the substrate support configured to heat the substrate during processing; accessing, by a processing device, a temperature matrix including a plurality of vectors corresponding to the plurality of zones, a value in each vector representing a temperature rise across a location on the substrate support as a result of a corresponding one of the plurality of heating elements being activated at a temperature setpoint; determining, by the processing device, a temperature map of the substrate support by multiplying the temperature matrix and a weight vector, the weight vector including estimated weight values ​​of each vector of the temperature matrix; determining, by the processing device, a target parameter map for a film on a substrate to be processed on the substrate support based on an initial parameter map and the temperature map, the initial parameter map including data characterizing process parameters across a location of a first film on a first substrate heated at a uniform temperature; iteratively updating, by the processing device, the estimated weight values ​​of the weight vector so that the temperature map results in a minimum standard deviation of parameter values ​​in the target parameter map; using, by the processing device, the estimated weight values ​​as control values ​​for each of the plurality of heating elements during processing of a subsequent substrate; A method comprising:

6. The method of claim 5 , wherein minimizing the standard deviation is performed by ensuring that the standard deviation of the parameter values ​​is less than a threshold value.

7. The method of claim 5 , wherein the process parameter is the extinction coefficient (k) of the film.

8. A value in one vector among the plurality of vectors is a first value of a location within one of the plurality of zones corresponding to the vector; and a second value for locations not within the zone corresponding to the vector, the first value being greater than the second value, and the second value decreasing the further away from the zone; and The method of claim 5 , comprising:

9. accessing parameter sensitivity values; determining an updated temperature map by subtracting the uniform temperature from the temperature map; multiplying the parameter sensitivity values ​​by the updated temperature map to determine a modified parameter map; determining the target parameter map by adding the modified parameter map to the initial parameter map; The method of claim 5 further comprising:

10. receiving, by a computing device, a training data set including a plurality of data items, each data item of the plurality of data items including a set of temperature values ​​across a location of a substrate support including a multi-zone heater and a set of thickness values ​​of a film on a substrate processed while on the substrate support; inputting the training dataset into a machine learning model by the computing device; and training, by the computing device, the machine learning model based on the training dataset to generate a trained machine learning model that updates values ​​of a temperature matrix including a plurality of vectors corresponding to a plurality of zones of the multi-zone heater, the value in each vector representing a temperature rise across a location on the substrate support as a result of a corresponding one of a plurality of heating elements of the multi-zone heater being set to one or more temperature setpoints, the combination of the plurality of vectors comprising a target set of temperature values ​​across the substrate support usable to ensure a threshold level of film uniformity; A method comprising:

11. The method of claim 10 , wherein the machine learning model is an artificial neural network.

12. The method of claim 10 , further comprising generating the set of thickness values ​​for the film on the substrate using a thickness imaging sensor.

13. 11. The method of claim 10, wherein training the machine learning model further comprises determining each set of temperature values ​​usable in a subsequent training iteration by multiplying a weight vector with the temperature matrix before updating, the weight vector including estimated weight values ​​for each vector of the temperature matrix.

14. For subsequent substrates to be processed: determining the target set of temperature values ​​for the substrate support by multiplying the temperature matrix and a weight vector, the weight vector including estimated weight values ​​for each vector in the temperature matrix; determining a target set of thickness values ​​for the film on the subsequent substrate based on a previous set of thickness values ​​for a previously processed substrate and the target set of temperature values; The method of claim 10 further comprising:

15. iteratively updating the estimated weight values ​​of the weight vector such that the target set of temperature values ​​results in a minimization of the standard deviation of the target set of thickness values; determining temperature set points for each of the plurality of heating elements using the estimated weight values ​​while processing the subsequent substrate; and 15. The method of claim 14, further comprising:

16. accessing a temperature sensitivity value; determining an updated set of temperature values ​​by subtracting the set of temperature values ​​from the target set of temperature values; multiplying the temperature sensitivity value by the updated set of temperature values ​​to determine a set of modified thickness values; determining the target set of thickness values ​​by adding the modified set of thickness values ​​to the previous set of thickness values; 15. The method of claim 14, further comprising:

17. Correlating each heating element of a plurality of heating elements in a multi-zone heater of a substrate support with a zone of a plurality of zones of the substrate support configured to heat the substrate during processing; receiving a training data set including a plurality of data items, each data item of the plurality of data items including a plurality of temperature set points and a thickness map associated with a substrate, the plurality of temperature set points controlling the plurality of heating elements, and the thickness map including data characterizing film thickness across locations of a film on the substrate; inputting, by a processing device, the training dataset into a machine learning model; training, by the processing device, the machine learning model based on the training dataset to generate a trained machine learning model that receives a target thickness as input and outputs a set of temperature set points for the plurality of heating elements that results in the film having the target thickness; A method comprising:

18. The method of claim 17 , wherein the machine learning model is an artificial neural network.

19. 20. The method of claim 17, wherein the target thickness is a substantially uniform thickness with a standard deviation of thickness values ​​across the film locations that is less than a threshold value.

20. 20. The method of claim 17, further comprising generating, using a thickness imaging sensor, each thickness map comprising the thickness values ​​of the film on the substrate.

21. 18. The method of claim 17, wherein training the machine learning model comprises correlating, for each substrate, combinations of temperature values ​​as a result of the plurality of temperature set points across the locations of the film with thickness values ​​of the thickness map across those locations.

22. receiving a substrate on a substrate support of a processing chamber, the substrate support configured to heat the substrate during processing using a plurality of heating elements corresponding to a plurality of zones of a multi-zone heater within the substrate support; receiving, by a processing device of the processing chamber, information indicative of a process to be performed on the substrate and information indicative of a target thickness of a film on the substrate; inputting at least one of the process to be performed or the target thickness into a trained machine learning model, wherein the trained machine learning model outputs a set of temperature set points for the plurality of heating elements associated with the process, the set of temperature set points resulting in the film on the substrate having the target thickness with a standard deviation of thickness values ​​across locations on the substrate that is less than a threshold; configuring, by the processing device, the plurality of heating elements to operate at the set of temperature set points; processing the substrate in the processing chamber using the plurality of heating elements at the set of temperature setpoints; A method comprising:

Images