Residual Thickness Compensation

Abstract

The method includes identifying materials associated with a substrate processing step of a recipe. The method further includes determining an expected total residual thickness value after the substrate processing step. The method further includes determining an expected material thickness value for the materials associated with the substrate processing step based on the expected total residual thickness value. The method further includes updating the recipe based on the materials and the expected material thickness value for the materials to generate an updated recipe. The method further includes processing the substrate based on the updated recipe.

Description

[Technical Field]

[0001] FIELD OF THE DISCLOSURE The present disclosure relates to compensation, and more particularly to residual thickness compensation during substrate manufacturing. [Background technology]

[0002] The products may be manufactured by performing one or more manufacturing processes using the manufacturing equipment, for example, a substrate processing device may be used to manufacture substrates through a substrate processing step. Summary of the Invention

[0003] The following is a simplified summary of the disclosure to provide a basic understanding of some aspects of the disclosure. This summary is not an exhaustive summary of the disclosure. It is not intended to identify key or critical elements of the disclosure, nor is it intended to delineate the scope of particular embodiments of the disclosure or the scope of the claims. The sole purpose of this summary is to present some concepts of the disclosure in a simplified form as a prelude to the more detailed description that is presented later.

[0004] One aspect of the present disclosure includes a method for identifying materials associated with a substrate processing step of a recipe. The method further includes determining an expected total residual thickness value after the substrate processing step. The method further includes determining an expected material thickness value for the materials associated with the substrate processing step based on the expected total residual thickness value. The method further includes updating the recipe based on the materials and the expected material thickness value for the materials to generate an updated recipe. The method further includes processing a substrate based on the updated recipe.

[0005] A further aspect of the present disclosure includes a non-transitory computer-readable storage medium containing instructions that, when executed by a processing device operatively connected to a memory, perform processes including identifying a material associated with a substrate processing operation of a recipe. The process further includes determining an expected total residual thickness value after the substrate processing operation. The process further includes determining an expected material thickness value for the material associated with the substrate processing operation based on the expected total residual thickness value. The process further includes updating the recipe based on the material and the expected material thickness value for the material to generate an updated recipe. The process further includes processing a substrate based on the updated recipe.

[0006] Another aspect of the present disclosure includes a system including a memory and a processing device coupled to the memory. The processing device identifies materials associated with a substrate processing step of a recipe. The processing device further determines an expected total residual thickness value after the substrate processing step. The processing device further determines an expected material thickness value for the materials associated with the substrate processing step based on the expected total residual thickness value. The processing device further updates the recipe based on the materials and the expected material thickness value for the materials to generate an updated recipe. The processing device further processes a substrate based on the updated recipe.

[0007] The present disclosure is illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings. [Brief explanation of the drawings]

[0008] [Figure 1] FIG. 1 is a block diagram illustrating an exemplary system architecture according to certain embodiments. [Figure 2] 1 illustrates a data set generator associated with residual thickness compensation, according to certain embodiments. [Figure 3] FIG. 10 is a block diagram illustrating the determination of predictive data associated with residual thickness compensation, according to certain embodiments. [Figure 4A]FIG. 1 is a flow diagram of a method associated with residual thickness compensation, according to certain embodiments. [Figure 4B] FIG. 1 is a flow diagram of a method associated with residual thickness compensation, according to certain embodiments. [Figure 4C] FIG. 1 is a flow diagram of a method associated with residual thickness compensation, according to certain embodiments. [Figure 5] FIG. 1 is a block diagram illustrating a computer system according to certain embodiments. DETAILED DESCRIPTION OF THE INVENTION

[0009] Described herein are techniques directed to residual thickness compensation (e.g., adjusting film deposition parameters based on residual thickness during substrate manufacturing, scalable run-time software techniques to compensate for deposition thickness drift based on chamber residual thickness, chamber residual-based loop time adjustment).

[0010] The manufacturing equipment can deposit a film on the surface of a substrate according to a process recipe. The manufacturing equipment can deposit multiple layers of a film on the surface of the substrate and perform an etching process to form a pattern in the deposited film. For example, the manufacturing equipment can perform a chemical vapor deposition (CVD) process to deposit alternating layers on the substrate. The film can include one or more layers of material formed during the deposition process, and each layer can include a particular thickness gradient (e.g., a change in thickness along a layer of the deposited film). For example, a first layer (referred to as the proximal layer or proximal end of the film) can be formed directly on the surface of the substrate and can have a first thickness. After the first layer is formed on the substrate surface, a second layer having a second thickness can be formed on the first layer. This process continues until the deposition process is completed and the final layer of the film (referred to as the distal layer or distal end of the film) is formed.

[0011] The film may be subjected to, for example, an etching process to form a pattern on the surface of the substrate, a chemical-mechanical polishing (CMP) process to smooth the surface of the film, or other processes to produce a finished substrate. An etching process may involve applying a high-energy process gas (e.g., plasma) to the sample surface to break down material on the surface, which may then be removed by a vacuum system.

[0012] A processing chamber can perform each substrate manufacturing process (e.g., a deposition process, an etching process, a polishing process, etc.) according to a process recipe. A process recipe defines a particular set of steps to be performed on a substrate during that process and can include one or more settings associated with each step. For example, a deposition process recipe can include processing chamber temperature settings, processing chamber pressure settings, precursor flow settings for materials included in a film to be deposited on a substrate surface, etc. Thus, the thickness of each film layer can be correlated to these processing chamber settings.

[0013] A film may include alternating layers of various materials. For example, a film may include alternating layers of oxide and nitride (oxide-nitride-oxide-nitride stack, i.e., ONON (oxide-nitride-oxide-nitride) stack), alternating layers of oxide and polysilicon (oxide-polysilicon-oxide-polysilicon stack, i.e., OPOP (oxide-polysilicon-oxide-polysilicon) stack), etc. Each set of alternating layers may be referred to as a loop. For example, a film may include 40 loops (e.g., 40 pairs of oxide-nitride layers), where the layer thicknesses in some loops may be different from the layer thicknesses in other loops. For example, a film stack can include 40 oxide-nitride loops (e.g., 80 layers of alternating 40 oxide layers and 40 nitride layers), where the first loop of the film stack has oxide layers of a first thickness and nitride layers of a second thickness, then the next 9 loops of the film stack have oxide layers of a third thickness and nitride layers of a fourth thickness, and the final 30 loops of the film stack have oxide layers of a fifth thickness and nitride layers of a sixth thickness.

[0014] The film may also include layers of different materials with variable (non-alternating) patterns (e.g., oxide-nitride-nitride-oxide stack (ONNO), oxide-nitride-oxide-oxide stack (ONOO), etc.). During this substrate fabrication process, the thickness of each loop may change (drift) due to continually changing deposition parameters and fluctuations in processing chamber conditions (e.g., contaminant buildup, residual thickness on the processing chamber walls, erosion of certain components, etc.). Variations in layer thickness may move the gas distribution plate closer to or farther from the substrate surface, thus affecting plasma flow and / or temperature and causing further deformation of the film.

[0015] To maintain the desired overall thickness of the film stack, in some manufacturing systems, these variations are compensated for by manually increasing or decreasing the deposition time of subsequent loops. For example, if the thickness of a first loop is larger than required by the process recipe, a technician can manually decrease the deposition time of a second loop in the process recipe to produce a thinner loop than required by the process recipe. A processing chamber may have multiple substrate processing regions or slots. After a deposition step, deposition drift (e.g., caused in part by different residual thicknesses on the walls of each substrate processing region) may vary between individual substrate processing regions (e.g., slots) within a single processing chamber. Compensating for these variations between substrate processing regions can also generally be performed by manually increasing or decreasing the deposition time of subsequent layers to maintain the desired overall thickness of the film stack. However, such a process is not scalable, prone to error, and time-consuming. It can also result in manufacturing delays, lost throughput, film defects, inconsistent and abnormal products, unscheduled user time or downtime, and rejected parts. It also significantly increases the time required to perform process recipe parameter optimization.

[0016] Aspects and embodiments of the present disclosure address the above and other shortcomings of existing techniques by performing residual thickness compensation, for example, by adjusting deposition parameters based on the residual thickness during substrate manufacturing (e.g., by compensating for drift in deposition thickness based on chamber residual thickness).

[0017] The processing device identifies a material associated with a substrate processing step of the recipe. In some examples, the processing device determines that the substrate processing step is to deposit a nitride on the substrate. In some examples, the processing device determines that the substrate processing step is to deposit an oxide on the substrate. In some examples, the processing device determines that the substrate processing step is to deposit a conductor (e.g., copper, tungsten, etc.) on the substrate.

[0018] The processing device determines an expected total residual thickness value after the substrate processing step, which may be an expected total residual thickness value on the substrate and / or on the walls of the processing chamber depending on the substrate processing steps of the recipe up to the end of the current substrate processing step.

[0019] The processing device determines a predicted material thickness value (e.g., for a current substrate processing operation) for a material associated with the substrate processing operation based on the predicted total residual thickness value. In some embodiments, the predicted material thickness value is the difference between the desired material thickness value and the actual material thickness value. In some embodiments, to determine the predicted material thickness value, the processing device provides the predicted total residual thickness value as an input to a trained machine learning model. In some embodiments, to determine the predicted material thickness value, the processing device receives output from the trained machine learning model associated with prediction data, the material thickness value being associated with the prediction data. In some embodiments, the trained machine learning model is trained with data inputs including historical total residual thickness values ​​and target outputs of historical material thickness values.

[0020] The processing device updates the recipe based on the material and the expected material thickness value for the material to generate an updated recipe. In some embodiments, to update the recipe, the processing device determines an updated time value associated with the substrate processing operation. In some embodiments, the processing device determines the updated time value associated with the substrate processing operation based on the time value associated with the substrate processing operation, the expected total remaining thickness value, and the expected material thickness value. In some embodiments, updating the recipe includes determining at least one of an updated radio frequency (RF) power for the substrate processing operation, an updated interval value for the substrate processing operation, an updated gas flow rate value for the substrate processing operation, or an updated chamber pressure value for the substrate processing operation.

[0021] The processing device processes the substrate based on the updated recipe.

[0022] Aspects of the present disclosure provide technical advantages. Aspects of the present disclosure avoid the time-consuming and error-prone process of manually calculating an "operation time offset" for each loop and entering the offset into a table. Aspects of the present disclosure are scalable. Aspects of the present disclosure prevent resulting delays in production, throughput losses, and / or film defects. Aspects of the present disclosure result in a significant reduction in the time required to perform parameter optimization of a process recipe. The present disclosure also generates diagnostic data from which corrective actions can be taken to avoid inconsistent and abnormal products and unscheduled user time or downtime. Aspects of the present disclosure enable deposition drift compensation based on material type.

[0023] FIG. 1 is a block diagram illustrating an example system 100 (an example system architecture) according to certain embodiments. System 100 (e.g., via corrective action component 122 and / or prediction component 114) can perform methods described herein (e.g., methods 400A-400C of FIGS. 4A-4C). System 100 includes client device 120, manufacturing equipment 124, sensors 126, metrology equipment 128, prediction server 112, and data store 140. In some embodiments, prediction server 112 is part of prediction system 110. In some embodiments, prediction system 110 further includes server machines 170 and 180.

[0024] In some embodiments, one or more of client device 120, manufacturing equipment 124, sensors 126, metrology equipment 128, prediction server 112, data store 140, server machine 170, and / or server machine 180 are connected to one another via network 130 to generate prediction data 160 for performing residual-based adjustments of film deposition parameters during substrate fabrication. In some embodiments, network 130 is a public network that provides client device 120 with access to prediction server 112, data store 140, and other publicly available computing devices. In some embodiments, network 130 is a private network that provides client device 120 with access to manufacturing equipment 124, sensors 126, metrology equipment 128, data store 140, and other privately available computing devices. In some embodiments, network 130 includes one or more wide area networks (WANs), local area networks (LANs), wired networks (e.g., Ethernet networks), wireless networks (e.g., 802.11 networks or Wi-Fi networks), cellular networks (e.g., Long Term Evolution (LTE) networks), routers, hubs, switches, server computers, cloud computing networks, and / or combinations thereof.

[0025] In some embodiments, client device 120 includes a computing device such as a personal computer (PC), a laptop, a mobile phone, a smartphone, a tablet computer, a netbook computer, etc. In some embodiments, client device 120 includes a corrective action component 122. In some embodiments, corrective action component 122 may also be included in prediction system 110 (e.g., a machine learning processing system). In some embodiments, corrective action component 122 is alternatively included in prediction system 110 (e.g., instead of being included in client device 120). Client device 120 includes an operating system that enables a user to one or more of: integrate, generate, view, edit data, provide instructions to prediction system 110 (e.g., a machine learning processing system), etc.

[0026] In some embodiments, the corrective action component 122 receives one or more of user input (e.g., via a graphical user interface (GUI) displayed by the client device 120), characteristic data 142, performance data 152, etc. In some embodiments, the characteristic data 142 may be a predicted total residual thickness value, a predicted material thickness value, etc. In some embodiments, the corrective action component 122 transmits data (e.g., user input, characteristic data 142, performance data 152, etc.) to the prediction system 110, receives prediction data 160 from the prediction system 110, determines corrective actions based on the prediction data 160, and causes the corrective actions to be implemented. In some embodiments, the corrective action component 122 stores the data (e.g., user input, characteristic data 142, performance data 152, etc.) in the data store 140, and the prediction server 112 retrieves the data from the data store 140. In some embodiments, prediction server 112 stores the output of trained machine learning model 190 (e.g., predicted data 160) in data store 140, and client device 120 retrieves the output from data store 140. In some embodiments, corrective action component 122 receives corrective action instructions (e.g., based on predicted data 160) from prediction system 110 and causes the corrective action to be performed.

[0027] The manufacturing tools 124 may perform processes according to a recipe or over a period of time to produce products such as substrates, wafers, semiconductors, or electronic devices. The manufacturing tools 124 may include processing chambers. The processing chambers may be adapted to perform any number of processes on substrates. The same or different substrate processes may be performed in each processing chamber or substrate processing region. Substrate processing processes may include atomic layer deposition (ALD), physical vapor deposition (PVD), or chemical vapor deposition (CVD), etching, annealing, hardening, pre-cleaning, or metal or metal oxide removal. Other processes may be performed on the substrates therein. The processing chambers may include one or more sensors configured to capture data on the substrate before, after, or during a substrate processing process. For example, the one or more sensors may be configured to acquire spectral and / or non-spectral data about a portion of the substrate during the substrate processing process. In other or similar embodiments, one or more sensors may be configured to acquire data associated with the environment within the processing chamber before, after, or during a substrate processing operation. For example, the one or more sensors may be configured to acquire data associated with the temperature, pressure, gas concentrations, etc. of the environment within the processing chamber during a substrate processing operation.

[0028] The processing chamber may be used for a process in which a material is deposited on a substrate. For example, the processing chamber may be a chamber for a deposition process, as described above. In some embodiments, the processing chamber includes a chamber body and a showerhead enclosing the interior space. The showerhead may include a showerhead base and a showerhead gas distribution plate. Alternatively, the showerhead may be replaced with a lid and a nozzle in some embodiments, or with multiple pie-shaped showerhead sections and a plasma generation unit in other embodiments. The chamber body may be fabricated from aluminum, stainless steel, or other suitable materials, such as titanium (Ti). The chamber body generally includes a sidewall and a bottom. An exhaust port may be defined in the chamber body and may connect the interior space to a pumping system. The pumping system may include one or more pumps and a throttle valve, which are utilized to evacuate and adjust the pressure of the interior space of the processing chamber.

[0029] The showerhead may be supported on a sidewall of the chamber body. The showerhead (or lid) may be open to allow access to the interior space of the processing chamber and may provide a seal for the processing chamber when closed. A gas panel may be connected to the processing chamber to provide process gases and / or cleaning gases to the interior space via the showerhead or lid and nozzle (e.g., through apertures in the showerhead or lid and nozzle). For example, the gas panel may provide precursors for the film material to be deposited on the surface of the substrate. In some embodiments, the precursors may include silicon-based precursors or boron-based precursors. The showerhead may include a gas distribution plate (GDP) having multiple gas supply holes (also referred to as channels) throughout the GDP. A substrate support assembly is positioned within the interior space of the processing chamber below the showerhead. The substrate support assembly holds the substrate during processing (e.g., during a deposition process), for example, using an electrostatic chuck.

[0030] In some embodiments, a processing chamber may include a metrology device (e.g., metrology device 128) and / or a sensor (e.g., sensor 126) configured to generate in-situ measurements (e.g., metrology data) and / or sensor measurements (e.g., sensor data) during a process performed in the processing chamber. In some embodiments, the measurements and / or sensor measurements may be a subset of the characteristic data 142 and / or performance data 152. The metrology device and / or sensor may be operably connected to a system controller. In some embodiments, the metrology device may be configured to generate measurements (e.g., thickness) of a film during a particular instance of a deposition process. In some embodiments, the sensor may be configured to generate sensor measurements (e.g., thickness) of a film during a particular instance of a deposition process. The system controller may generate a thickness profile for the film based on the measurements received from the metrology device. The system controller may generate a thickness profile for the film based on the sensor measurements received from the sensor. In other or similar embodiments, the processing chamber does not include a metrology device. In such an embodiment, the system controller can receive one or more measurements for the film after completion of a deposition process in the processing chamber, can determine a deposition rate based on the one or more measurements, and can generate a thickness profile for the film based on the determined concentration gradient of the deposition process and the determined deposition rate.

[0031] The manufacturing tool 124 can perform processes on substrates (e.g., wafers) in processing chambers. Examples of substrate processes include deposition processes that deposit one or more layers of film on the surface of the substrate, etching processes that form patterns on the surface of the substrate, etc. The manufacturing tool 124 can perform each process according to a process recipe. The process recipe defines a particular set of steps to be performed on the substrate during that process and can include one or more settings associated with each step. For example, a recipe for a deposition process can include temperature settings for the processing chamber, pressure settings for the processing chamber, flow rates of precursors for materials included in the film to be deposited on the substrate surface, etc.

[0032] In some embodiments, the manufacturing tool 124 includes a sensor 126 configured to generate data associated with a substrate processed in the manufacturing system 100. For example, a processing chamber may include one or more sensors configured to generate a residual thickness profile (e.g., a thickness of a material on a processing chamber wall) associated with the processing chamber before, during, and / or after a process (e.g., a deposition process). For example, a processing chamber may include one or more sensors configured to generate spectral or non-spectral data associated with a substrate before, during, and / or after a process (e.g., a deposition process) is performed on the substrate. In some embodiments, the spectral data generated by the sensor 126 may indicate the concentration of one or more materials deposited on the surface of the substrate. The sensor 126 configured to generate spectral data associated with the substrate may include a reflectance sensor, an ellipsometry sensor, a thermal spectral sensor, a capacitance sensor, etc. The sensor 126 configured to generate non-spectral data associated with the substrate may include a residual thickness sensor, a temperature sensor, a pressure sensor, a flow rate sensor, a voltage sensor, etc.

[0033] The metrology tool 128 can provide metrology data associated with substrates processed by the fabrication tool 124. In some embodiments, the metrology data can be a subset of the characteristic data 142 and / or the performance data 152. The metrology data can include values ​​for film property data (e.g., wafer-space film properties), dimensions (e.g., thickness, height, etc.), dielectric constant, dopant concentration, density, defects, etc. In some embodiments, the metrology data can further include values ​​for one or more surface profile property data (e.g., etch rate, etch rate uniformity, critical dimensions of one or more features included in the surface of the substrate, critical dimension uniformity across the surface of the substrate, edge placement error, etc.). The metrology data can be for finished or unfinished products. The metrology data can vary from substrate to substrate. The metrology data can be generated using, for example, reflectometry techniques, ellipsometry techniques, transmission electron microscopy (TEM) techniques, etc.

[0034] In some embodiments, prediction server 112, server machine 170, and server machine 180 each include one or more computing devices, such as a rack-mounted server, a router computer, a server computer, a personal computer, a mainframe computer, a laptop computer, a tablet computer, a desktop computer, a graphics processing unit (GPU), an accelerator application-specific integrated circuit (ASIC, e.g., a tensor processing unit (TPU)), etc.

[0035] The prediction server 112 includes a prediction component 114. In some embodiments, the prediction component 114 identifies (e.g., received from the client device 120 and retrieved from the data store 140) characteristic data 142 (e.g., expected total residual thickness values, expected material thickness values, etc.) and generates prediction data 160 associated with performing corrective actions (e.g., recipe updates, deposition process parameter updates, process parameter updates, etc.). In some embodiments, the prediction component 114 uses one or more trained machine learning models 190 to determine the prediction data 160. In some embodiments, the trained machine learning models 190 are trained using historical characteristic data 144 and historical performance data 154.

[0036] For example, deposition process parameters may include deposition time for each layer and / or each loop of a process recipe, a temperature setting for a process chamber, a pressure setting for a process chamber, a flow rate setting for a precursor for a material included in a film to be deposited on a substrate surface, a showerhead height, etc. In some embodiments, updating a recipe includes updating deposition process parameters. For example, the deposition process parameters may include a time value, an updated time value, a recipe, an updated recipe, an RF power for a substrate processing operation, an interval value for a substrate processing operation, a gas flow rate value for a substrate processing operation, or a chamber pressure value for a substrate processing operation, an updated RF power for a substrate processing operation, an updated interval value for a substrate processing operation, an updated gas flow rate value for a substrate processing operation, an updated chamber pressure value for a substrate processing operation, etc.

[0037] In some embodiments, the prediction system 110 (e.g., prediction server 112, prediction component 114) uses supervised machine learning (e.g., supervised dataset, historical characteristic data 144 labeled with historical performance data 154, etc.) to generate the predicted data 160. In some embodiments, the prediction system 110 uses semi-supervised learning (e.g., semi-supervised dataset, where performance data 152 is a predicted percentage, etc.) to generate the predicted data 160. In some embodiments, the prediction system 110 uses unsupervised machine learning (e.g., unsupervised dataset, clustering, clustering based on historical characteristic data 144, etc.) to generate the predicted data 160.

[0038] In some embodiments, the manufacturing equipment 124 (e.g., a deposition chamber, a cluster tool, a wafer backgrinding system, a wafer saw, a die attach machine, a wire bonder, a die overcoat system, a molding machine, a hermetic sealing machine, a metal can welding machine, a deflash / trim / form / singulation (DTFS) machine, a branding machine, and / or an exterior processing machine, etc.) is part of a substrate processing system (e.g., an integrated processing system). The manufacturing equipment 124 includes one or more of a controller, an enclosure system (e.g., a substrate carrier, a front opening unified pod (FOUP), an auto-teach FOUP, a process kit enclosure system, a substrate enclosure system, a cassette, etc.), a side storage pod (SSP), an aligner device (e.g., an aligner chamber), a factory interface (e.g., an equipment front end module (EFEM)), a load lock, a transfer chamber, one or more processing chambers, and / or a robotic arm (e.g., disposed in a transfer chamber, disposed in a front interface, etc.), etc. The enclosure system, SSP, and load lock are attached to a factory interface, and a robot arm disposed within the factory interface transfers content (e.g., substrates, process kit rings, carriers, validation wafers, etc.) between the enclosure system, SSP, load lock, and factory interface. An aligner device is disposed within the factory interface to align the content. The load lock and processing chamber are attached to a transfer chamber, and a robot arm disposed within the transfer chamber transfers content (e.g., substrates, process kit rings, carriers, validation wafers, etc.) between the load lock, processing chamber, and transfer chamber. In some embodiments, the manufacturing tool 124 includes components of a substrate processing system.In some embodiments, the processing chamber or substrate characteristic data 142 is obtained from the processing chamber or substrate being subjected to one or more processes (e.g., deposition, etching, heating, cooling, transport, treatment, flow, etc.) performed by components of the manufacturing tool 124.

[0039] In some embodiments, the sensor 126 provides characteristic data 142 (e.g., sensor values ​​such as past and current sensor values), i.e., characteristic data 142 of a processing chamber (e.g., expected total residual thickness) or characteristic data 142 of a substrate processed by the manufacturing tool 124 (e.g., expected material thickness).

[0040] In some embodiments, the sensor 126 includes one or more metrology tools, such as an ellipsometer (used to characterize the surface of a thin film by measuring material properties such as layer thickness, optical constants, surface roughness, composition, optical anisotropy, etc.), an ion mill (used to prepare non-uniform bulk materials when large areas of material are uniformly thinned), a capacitance versus voltage (CV) system (used to measure the CV and capacitance versus time (Ct) characteristics of semiconductor devices), an interferometer (used to measure distance in terms of wavelength and to determine the wavelength of a particular light source), a source measurement unit (SME) magnetometer, optical and imaging systems, a surface profilometer, a wafer prober (used to test semiconductor wafers before separating them into individual dies or chips), an imaging station, a critical-dimension scanning electron microscope (CD-SEM), a These include microscopes (used to ensure the stability of manufacturing processes by measuring critical dimensions of substrates), reflectometers (used to measure the reflectivity and radiance from surfaces), resistance probes (used to measure the resistivity of thin films), resistance high-energy electron diffraction (RHEED) systems (used to measure or monitor the crystal structure or crystal orientation of epitaxial thin films of silicon or other materials), and / or X-ray diffractometers (used to unambiguously determine the crystal structure, crystal orientation, film thickness, and residual stress of silicon wafers, epitaxial films, or other substrates).

[0041] In some embodiments, the characteristic data 142 is used for equipment health and / or product health (e.g., product quality). In some embodiments, the characteristic data 142 is received over a period of time.

[0042] In some embodiments, the sensors 126 and / or metrology devices 128 provide characteristic data 142, which may include one or more of morphology data, size attribute data, dimensional attribute data, image data, scanning electron microscope (SEM) images, energy dispersive x-ray (EDX) images, defect distribution data, spatial location data, elemental analysis data, wafer signature data, chip layers, chip layout data, edge data, gray level data, signal-to-noise data, temperature data, spacing data, current data, power data, and / or voltage data, etc.

[0043] In some embodiments, the characteristic data includes morphology data (e.g., data related to the morphology of the substrate, such as the thickness or residual thickness of a deposited layer). In some embodiments, the characteristic data 142 includes size attribute data (e.g., data representing the size of an attribute of the substrate). In some embodiments, the characteristic data 142 includes dimensional attribute data (e.g., data representing the dimension of an attribute of the substrate). In some embodiments, the characteristic data 142 includes an SEM image (e.g., an image taken by a scanning electron microscope using a focused beam of electrons to scan the surface of the substrate and generate a high-resolution image). In some embodiments, the characteristic data 142 includes an EDX image (e.g., an image generated from data collected using X-ray techniques to identify the elemental composition of a material). In some embodiments, the characteristic data 142 includes defect distribution data (e.g., data representing the distribution (e.g., spatial, temporal distribution, etc.) of defects on the substrate). In some embodiments, the characteristic data 142 includes spatial location data (e.g., data representing the spatial location of an attribute, defect, element, etc. of the substrate). In some embodiments, characteristic data 142 includes elemental analysis data (e.g., data representing the elemental composition of the substrate). In some embodiments, characteristic data 142 includes wafer signature data (e.g., data representing the distribution of wafer defects across the substrate resulting from a single manufacturing issue). In some embodiments, characteristic data 142 includes chip layer data (e.g., associated with a layer or step in a substrate manufacturing process). In some embodiments, characteristic data 142 includes chip layout data (e.g., data representing the layout of the substrate in terms of its planar geometric shape). In some embodiments, characteristic data 142 includes edge data (e.g., data representing the edge of the wafer, such as chipped edges, wafer edge thickness, wafer bow and / or warpage, etc.). In some embodiments, characteristic data 142 includes gray level data (e.g., data representing the brightness of pixels in an image of the substrate) and signal-to-noise data (e.g., data representing the signal-to-noise ratio of measurements of the substrate, e.g., by a spectroscopic measurement device).

[0044] In some embodiments, characteristic data 142 (e.g., historical characteristic data 144, current characteristic data 146, etc.) is processed (e.g., by client device 120 and / or by prediction server 112). In some embodiments, processing characteristic data 142 includes generating features. In some embodiments, a feature is a pattern in characteristic data 142 (e.g., slope, width, height, peak, etc.) or a combination of values ​​from characteristic data 142 (e.g., power derived from voltage and current, etc.). In some embodiments, characteristic data 142 includes features used by prediction component 114 to derive prediction data 160.

[0045] In some embodiments, the metrology tool 128 may be included as part of the fabrication tool 124. For example, the metrology tool 128 may be included within or connected to a processing chamber and configured to generate metrology data (e.g., characteristic data 142, performance data 152, etc.) of the interior of the processing chamber or the substrate while the substrate remains in the processing chamber, before, during, and / or after a process (e.g., a deposition process, an etch process, etc.). In some examples, the metrology tool 128 may be referred to as an in-situ metrology tool. In other examples, the metrology tool 128 may be connected to other stations of the fabrication tool 124. For example, the metrology tool may be connected to a transfer chamber, a load lock, or a factory interface.

[0046] In some embodiments, the sensor 126 may be included as part of the manufacturing tool 124. For example, the sensor 126 may be included within or connected to a processing chamber and configured to generate sensor data of the processing chamber or the substrate while the substrate remains in the processing chamber before, during, and / or after a process (e.g., a deposition process, an etch process, etc.). In some cases, the sensor 126 may be referred to as an in-situ sensor. In other examples, the sensor 126 may be connected to other stations of the manufacturing tool 124. For example, the sensor may be connected to a transfer chamber, a load lock, or a factory interface.

[0047] In some embodiments, metrology tool 128 (e.g., ellipsometry tool, imaging tool, spectroscopy tool, etc.) is used to determine metrology data (e.g., inspection data, image data, spectroscopy data, ellipsometry data, material composition data, optical data, structural data, etc.) corresponding to the interior (e.g., surface) of a processing chamber or a substrate fabricated by fabrication tool 124 (e.g., substrate processing equipment). In some examples, metrology tool 128 is used to inspect portions (e.g., layers) of a substrate and / or the interior of a processing chamber after fabrication tool 124 processes the substrate. In some embodiments, metrology tool 128 performs scanning acoustic microscopy (SAM), ultrasonic inspection, X-ray inspection, and / or computed tomography (CT) inspection. In some examples, metrology tool 128 is used to determine the quality of the processed substrate (e.g., layer thickness, layer uniformity, and / or interlayer spacing of layers, etc.) after fabrication tool 124 deposits one or more layers on a substrate. In some embodiments, the metrology devices 128 include imaging devices (e.g., SAM devices, ultrasound devices, X-ray devices, and / or CT devices). In some embodiments, the characteristic data 142 includes sensor data from the sensors 126 and / or metrology data from the metrology devices 128. In some embodiments, the characteristic data 142 includes sensor data from the sensors 126 and / or metrology data from the metrology devices 128 that are located in situ (inside the process chamber). In some embodiments, the performance data 152 includes user input via the client device 120 and / or metrology data from the metrology devices 128. The characteristic data 142 can include metrology data from a first subset of the metrology devices 128, and the performance data 152 can include metrology data from a second subset of the metrology devices 128.

[0048] In some embodiments, the performance data 152 may be associated with the performance of a recipe (e.g., a deposition recipe, an updated recipe, etc.) For example, the performance data 152 may be of a recipe and / or a substrate or processing chamber subjected to a processing step of the recipe.

[0049] In some embodiments, the characteristic data 142 may be derived from metrology data and / or sensor data. Metrology data may be data describing metrology of the substrate. Sensor data may be data representing internal conditions and characteristics of the process chamber. In some embodiments, the characteristic data may include a deposition thickness value (e.g., the actual or expected amount of material deposited on the substrate). In some embodiments, the characteristic data may include a total remaining thickness value (e.g., the actual or expected amount of material remaining to deposit on the chamber walls). In some embodiments, the deposition thickness value is derived from a deposition process recipe (e.g., the deposition thickness is equal to the expected deposition thickness for the deposition process).

[0050] In some embodiments, the actual deposition thickness value or actual remaining thickness value may refer to the amount of material actually deposited on the substrate or the amount of deposited material remaining on the process chamber walls after a deposition process. The actual deposition thickness value or actual total remaining thickness value may be measured using a metrology device or sensor.

[0051] In some embodiments, the expected deposition thickness value or expected residual thickness value may refer to the amount of material expected to deposit on a substrate or remain deposited on a process chamber wall after a deposition process. In some embodiments, the expected deposition thickness value or actual total residual thickness value may be derived from a process recipe (e.g., deposition process parameters). In some embodiments, the expected deposition thickness value or actual residual thickness value may be derived from previous actual thickness values ​​(e.g., on a substrate and / or on a process chamber wall) corresponding to the same deposition process.

[0052] In some embodiments, data store 140 is memory (e.g., random access memory), a drive (e.g., a hard drive, a flash drive), a database system, or another type of component or device capable of storing data. In some embodiments, data store 140 includes multiple storage components (e.g., multiple drives or multiple databases) spread across multiple computing devices (e.g., multiple server computers). In some embodiments, data store 140 stores one or more of characteristic data 142, performance data 152, and / or prediction data 160.

[0053] The characteristic data 142 may include predicted total residual thickness values, predicted material thickness values, actual total residual thickness values, actual material thickness values, desired total residual thickness values, desired material thickness values, etc. The performance data 152 may include predicted total residual thickness values, predicted material thickness values, actual total residual thickness values, actual material thickness values, desired total residual thickness values, desired material thickness values, etc.

[0054] In some embodiments, the data store 140 may store a predicted total residual thickness value and a predicted material thickness value. The predicted total residual thickness value may include one or more data points associated with a predicted residual film profile expected to be produced by a particular process recipe. In some embodiments, the predicted total residual thickness value may include a desired thickness of the film, a desired thickness of one or more layers of the film, and / or a desired thickness of one or more loops of the film, etc. The predicted material thickness value may include one or more data points associated with a current film thickness generated by the fabrication tool 124. For example, the predicted material thickness value may include a predicted thickness of the film, a predicted thickness of one or more layers of the film, and / or a predicted thickness of one or more loops of the film, etc. For example, the predicted material thickness value may include a measured thickness of the film, a measured thickness of one or more layers of the film, and / or a measured thickness of one or more loops of the film, etc. The predicted material thickness value may be measured using the metrology tool 128. The updated recipe may include one or more adjustments or offsets applied to the process chamber parameters or the process recipe. For example, an updated recipe may include adjustments to deposition times for film layers and / or loops, temperature settings for processing chambers, pressure settings for processing chambers, flow rates of precursors for materials included in the film being deposited on the substrate surface, power supplied to processing chambers, ratios of two or more settings, etc. An updated recipe may be generated by comparing an expected total residual thickness (e.g., the residual thickness expected to be produced by the process recipe) and determining adjustments to apply to parameters of the process recipe to achieve the expected material thickness using algorithms, libraries of known defect patterns, etc. An updated recipe may be applied to steps associated with deposition processes, etch processes, etc.

[0055] In some embodiments, the data store 140 may be configured to store data associated with known failure patterns. A failure pattern may be one or more values ​​(e.g., a vector, a scalar, etc.) associated with one or more problems or faults associated with a processing chamber subsystem. In some embodiments, a failure pattern may be associated with a corrective action. For example, a failure pattern may include parameter adjustment steps to correct the problem or fault indicated by the failure pattern. For example, a predictive system may compare a determined failure pattern with a library of known failure patterns to determine the type of failure that occurred in the subsystem, the cause of the failure, recommended corrective actions to correct the failure, etc.

[0056] In some embodiments, the corrective action may be, for example, updating a recipe (e.g., process recipe, deposition recipe), determining an expected material thickness value, processing a substrate based on the updated recipe, training a machine learning model using data inputs including historical residual thickness values ​​and target outputs including historical performance data, and / or receiving output associated with predicted data using the trained machine learning model, where the expected material thickness value is associated with the predicted data, etc. In some embodiments, updating the process recipe may include updating process parameters (e.g., deposition time, flow rate, temperature, etc.). In some embodiments, the corrective action may include providing machine learning (e.g., causing a recipe update based on the predicted data 160, an update of process / step parameters, etc.).

[0057] In some embodiments, the prediction data 160 is associated with a corrective action. In some embodiments, the corrective action is associated with one or more of updating a recipe (e.g., process recipe, deposition recipe), determining an expected material thickness value, processing a substrate based on the updated recipe, training a machine learning model using data inputs including historical residual thickness values ​​and target outputs including historical performance data, using the trained machine learning model to receive an output associated with the predicted data, repairing one or more pieces of the manufacturing equipment 124, replacing one or more pieces of the manufacturing equipment 124, computational process control (CPC), statistical process control (SPC) (e.g., SPC for comparison with a three-sigma graph), advanced process control (APC), model-based process control, preventative operational maintenance, design optimization, manufacturing parameter updates, wafer recipe modifications, feedback control, and / or machine learning modifications.

[0058] In some embodiments, data store 140 may be configured to store data that is inaccessible to users of the manufacturing system. For example, process data, spectral data, contextual data, etc. obtained for substrates being processed in the manufacturing system may be inaccessible to users (e.g., operators) of the manufacturing system. In some embodiments, all data stored in data store 140 may be inaccessible to users of the manufacturing system. In some embodiments, one portion of the data stored in data store 140 may be inaccessible to a user, while other portions of the data stored in data store 140 may be accessible to the user. In some embodiments, one or more portions of the data stored in data store 140 may be encrypted using an encryption mechanism unknown to the user (e.g., the data is encrypted using a private encryption key). In some embodiments, data store 140 may include multiple data stores, where data inaccessible to a user is stored in one or more first data stores and data accessible to a user is stored in one or more second data stores.

[0059] The performance data 142 includes historical characteristic data 144 and current characteristic data 146. In some embodiments, the characteristic data 142 (e.g., sensor data) may include an expected total residual thickness value, an expected material thickness value, an updated RF power for a substrate processing operation, an updated interval value for a substrate processing operation, an updated gas flow rate value for a substrate processing operation, an updated chamber pressure value for a substrate processing operation, actual deposition residual thickness data, pressure data, temperature data, a temperature range, power data, a comparison parameter for comparing inspection data with threshold data, threshold data, cooling rate data, and / or a cooling rate range, etc. In some embodiments, at least a portion of the characteristic data 142 is from the sensors 126 and / or metrology devices 128.

[0060] Performance data 152 includes historical performance data 154 and current performance data 156. Performance data 152 may indicate whether a substrate is properly designed, properly manufactured, uniform with other substrates, and / or properly functioning. Performance data 152 may indicate whether a substrate processing step was performed accurately. For example, performance data 152 may indicate the actual thickness deposited on a wafer or in a processing chamber during a deposition step (e.g., material thickness value, total residual thickness value, historical material thickness value, deposition thickness on the walls of the processing chamber, etc.). Performance data 152 may indicate whether a substrate processing step (e.g., a deposition step) was performed effectively. For example, performance data 152 may indicate deposition drift before, during, or after a deposition step, and may indicate a deposition residual thickness value, a material thickness value, etc.

[0061] In some embodiments, at least a portion of the performance data 152 is associated with the quality of products manufactured by the manufacturing tool 124. In some embodiments, at least a portion of the performance data 152 is based on metrology data from the metrology tool 128 (e.g., the historical performance data 154 includes metrology data indicative of successfully processed substrates, substrate characteristic data, yield, material thickness values, etc.) or metrology data from the sensor 126 (e.g., the historical performance data 154 includes sensor data indicative of successfully processed substrates, substrate characteristic data, yield, material thickness values, etc.). In some embodiments, at least a portion of the performance data 152 is based on inspection of substrates or the interior of the processing chamber (e.g., current performance data 156 based on actual inspection). In some embodiments, the performance data 152 includes user input (e.g., via the client device 120) indicative of substrate quality or deposition drift within the processing chamber / substrate processing region or on the substrate. In some embodiments, the performance data 152 includes an indication of an absolute value (e.g., the inspection data for a substrate indicates that the threshold data is below the threshold data by a calculated value, and the drift value is below the threshold drift value by a calculated value), or a relative value (e.g., the inspection data for a film deposition is below the threshold data by 5%, and the drift value is below the threshold drift value by 5%). In some embodiments, the performance data 152 indicates satisfaction of a threshold error amount (e.g., at least a 5% error in deposition drift after a deposition step, at least a 5% error in fabrication, at least a 5% error in flow, at least a 5% error in deformation, a specification limit).

[0062] In some embodiments, the historical data includes one or more of historical characteristic data 144 and / or historical performance data 154 (e.g., at least a portion for training machine learning model 190). The current data includes one or more of current characteristic data 146 and / or current performance data 156 (e.g., at least a portion input to trained machine learning model 190 after training of the model 190 using historical data). In some embodiments, the current data is used to retrain the trained machine learning model 190.

[0063] In some embodiments, the predictive data 160 is used to trigger the implementation of corrective actions on a process recipe, a deposition process parameter of a recipe / process, a manufacturing tool, a substrate processing system, or a component of a substrate processing apparatus.

[0064] Performing multiple types of measurements on multiple layers of a product or processing chambers to determine whether to perform corrective action is costly in terms of time used, metrology equipment 128 used, energy consumed, bandwidth used to transmit the metrology data, processor overhead to process the metrology data, etc. By providing characteristic data 142 to model 190 and receiving predicted data 160 from model 190, system 100 has the technical advantage of avoiding the costly process of using metrology equipment 128 on multiple layers of a product and / or sensors on processing chambers, avoiding wasted time and scrapped substrates.

[0065] Performing a manufacturing process (e.g., deposition) using manufacturing equipment 124 and / or manufacturing parameters (e.g., deposition process parameters) that result in defective products or damage to the manufacturing equipment is costly in terms of time, energy, product, manufacturing equipment 124, cost, etc. to identify corrective actions to prevent the occurrence of defective products. By providing characteristic data 142 to model 190, receiving prediction data 160 from model 190, and triggering corrective actions (e.g., recipe updates) based on prediction data 160, system 100 has the technical advantage of avoiding the costs of manufacturing, identifying, and discarding defective substrates.

[0066] In some embodiments, prediction system 110 further includes server machine 170 and server machine 180. Server machine 170 includes dataset generator 172 that can generate datasets (e.g., a set of data inputs and a set of target outputs) for training, validating, and / or testing machine learning model 190. Dataset generator 172 has a function called data collection, compilation, reduction, and / or partitioning to prepare data for machine learning. In some embodiments (e.g., for small datasets), partitioning for validation after training (e.g., explicit partitioning) is not used. During training, repeated cross-validation (e.g., 5-fold cross-validation, leave-one-out cross-validation) can be used, where a given dataset is effectively split repeatedly into various training and validation sets during training. A model (e.g., the best model, the most accurate model, etc.) is selected from the vector of models for the automatically separated combination subsets. In some embodiments, dataset generator 172 explicitly splits historical data (e.g., historical characteristic data 144 and corresponding historical performance data 154) into a training set (e.g., 60 percent of the historical data), a validation set (e.g., 20 percent of the historical data), and a test set (e.g., 20 percent of the historical data). Some steps of dataset generator 172, according to some embodiments, are described in detail below with respect to FIG. 2. In some embodiments, prediction system 110 (e.g., via prediction component 114) generates multiple sets of features (e.g., training features).In some examples, the first set of features corresponds to a first set of types of characteristic data (e.g., from the first set of sensors) (a first combination of values ​​from the first set of sensors, a first pattern in values ​​from the first set of sensors), which corresponds to each of the above datasets (e.g., a training set, a validation set, and a test set), and the second set of features corresponds to a second set of types of characteristic data (e.g., from a second set of sensors different from the first set of sensors) (a second combination of values ​​different from the first combination, a second pattern different from the first pattern), which corresponds to each of the above datasets.

[0067] Server machine 180 includes a training engine 182, a validation engine 184, a selection engine 185, and / or a test engine 186. In some embodiments, engines (e.g., training engine 182, validation engine 184, selection engine 185, and test engine 186) refer to hardware (e.g., circuitry, dedicated logic, programmable logic, microcode, processing device, etc.), software (e.g., instructions executed on a processing unit, general-purpose computer system, or dedicated machine), firmware, microcode, or a combination thereof. Training engine 182 can train machine learning model 190 using one or more sets of features associated with a training set from dataset generator 172. In some embodiments, training engine 182 generates multiple trained machine learning models 190, where each trained machine learning model 190 corresponds to a distinct set of parameters (e.g., characteristic data 142) and corresponding responses (e.g., performance data 152) of the training set. In some embodiments, multiple models are trained with the same parameters but with different objectives to model multiple effects. In some examples, a first trained machine learning model is trained using characteristic data 142 from all sensors 126 (e.g., sensors 1-5), a second trained machine learning model is trained using a first subset of characteristic data (e.g., from sensors 1, 2, and 4), and a third trained machine learning model is trained using a second subset of characteristic data that overlaps with the first subset of features (e.g., from sensors 1, 3, 4, and 5).

[0068] The validation engine 184 may validate the trained machine learning models 190 using a corresponding set of features in the validation set from the dataset generator 172. For example, a first trained machine learning model 190 trained using a first set of features in the training set is validated using a first set of features in the validation set. The validation engine 184 determines the accuracy of each of the trained machine learning models 190 based on the corresponding set of features in the validation set. The validation engine 184 evaluates and flags trained machine learning models 190 whose accuracy does not meet a threshold accuracy (e.g., should be discarded). In some embodiments, the selection engine 185 may select one or more trained machine learning models 190 whose accuracy meets a threshold accuracy. In some embodiments, the selection engine 185 may select the trained machine learning model 190 with the highest accuracy among the trained machine learning models 190.

[0069] The testing engine 186 can test the trained machine learning models 190 using a corresponding set of test set features from the dataset generator 172. For example, a first trained machine learning model 190 trained using a first set of training set features is tested using a first set of test set features. The testing engine 186 determines the trained machine learning model 190 with the highest accuracy among all the trained machine learning models based on the test set.

[0070] In some embodiments, machine learning model 190 (e.g., used for classification) refers to a model artifact (e.g., correctly classifying a condition or ordinal level for each training input) generated by training engine 182 using a training set that includes data inputs and corresponding target outputs. A pattern in the dataset that maps the data inputs to the target outputs (correct classifications or levels) can be found, and the mapping is provided to machine learning model 190 that captures this pattern. In some embodiments, machine learning model 190 uses one or more of Gaussian Process Regression (GPR), Gaussian Process Classification (GPC), Bayesian Neural Networks, Neural Network Gaussian Processes, Deep Belief Networks, Gaussian Mixture Models, or other probabilistic learning methods. Non-probabilistic techniques may also be used, including one or more of a support vector machine (SVM), a radial basis function (RBF), clustering, a k-nearest neighbor algorithm (k-NN), linear regression, random forests, neural networks (e.g., artificial neural networks), etc. In some embodiments, the machine learning model 190 is a multi-variate analysis (MVA) regression model.

[0071] The prediction component 114 provides the current characteristic data 146 (e.g., as input) to the trained machine learning model 190 and executes the trained machine learning model 190 (e.g., on the input to obtain one or more outputs). The prediction component 114 can determine (e.g., extract) predicted data 160 from the trained machine learning model 190 and determine (e.g., extract) uncertainty data that indicates a level of confidence that the predicted data 160 corresponds to the current performance data 156. In some embodiments, the prediction component 114 or the corrective action component 122 uses the uncertainty data (e.g., an uncertainty function or a retrieval function derived from the uncertainty function) to decide whether to use the predicted data 160 to perform corrective action or whether to further train the model 190.

[0072] For purposes of illustration and not limitation, aspects of the present disclosure describe training one or more machine learning models 190 using historical data (i.e., prior data, historical characteristic data 144, and historical performance data 154) and providing current characteristic data 146 to the one or more trained probabilistic machine learning models 190 to determine predicted data 160. In other embodiments, a heuristic or rule-based model is used to determine predicted data 160 (e.g., without using a trained machine learning model). In other embodiments, a non-probabilistic machine learning model may be used. The prediction component 114 monitors the historical characteristic data 144 and the historical performance data 154. In some embodiments, any of the information described with respect to data input 210 in FIG. 2 is monitored or otherwise used in the heuristic or rule-based model.

[0073] In some embodiments, the functionality of client device 120, prediction server 112, server machine 170, and server machine 180 is provided by fewer machines. For example, in some embodiments, server machines 170 and 180 are combined into a single machine, while in some other embodiments, server machine 170, server machine 180, and prediction server 112 are combined into a single machine. In some embodiments, client device 120 and prediction server 112 are combined into a single machine.

[0074] In general, functionality described in one embodiment as being performed by client device 120, prediction server 112, server machine 170, and server machine 180 may also be performed on prediction server 112 in other embodiments, where appropriate. Furthermore, functionality attributed to a particular component may be performed by various or multiple components operating together. For example, in some embodiments, prediction server 112 determines corrective actions based on prediction data 160. In other examples, client device 120 determines prediction data 160 based on data received from a trained machine learning model.

[0075] Additionally, the functionality of a particular element may be performed by various or multiple components working together. In some embodiments, one or more of prediction server 112, server machine 170, or server machine 180 are accessed as services offered to other systems or devices via a suitable application programming interface (API).

[0076] In some embodiments, a "user" is presented as a single individual. However, other embodiments of the present disclosure encompass a "user" being an entity controlled by multiple users and / or automated sources. In some instances, a collection of individual users linked as a group of administrators is considered a "user."

[0077] Although embodiments of the present disclosure are described in terms of determining predictive data 160 for residual thickness compensation (e.g., updating a recipe based on thickness values) during substrate manufacturing during substrate processing in a fabrication facility (e.g., a substrate processing facility), in some embodiments the present disclosure is also applicable generally to corrective actions within a fabrication facility. Embodiments may be applied generally to determining part quality based on various types of data.

[0078] 2 illustrates a dataset generator 272 (e.g., dataset generator 172 of FIG. 1 ) for generating a dataset for a machine learning model (e.g., model 190 of FIG. 1 ) (e.g., associated with residual thickness compensation, methods 400A-400C, etc.), according to certain embodiments. In some embodiments, dataset generator 272 is part of server machine 170 of FIG. 1 . The dataset generated by dataset generator 272 of FIG. 2 can be used to train a machine learning model (e.g., see FIG. 4B ) that causes corrective action to be taken (e.g., see FIG. 4C ).

[0079] A dataset generator 272 (e.g., dataset generator 172 in FIG. 1 ) generates a dataset for a machine learning model (e.g., model 190 in FIG. 1 ). The dataset generator 272 generates the dataset using historical characteristic data 244 (e.g., historical characteristic data 144 in FIG. 1 ) and historical performance data 254 (e.g., historical performance data 154 in FIG. 1 ). System 200 in FIG. 2 shows dataset generator 272, data input 210, and target output 220.

[0080] In some embodiments, the dataset generator 272 generates a dataset (e.g., a training set, a validation set, a test set) that includes one or more data inputs 210 (e.g., training inputs, validation inputs, test inputs). In some embodiments, the dataset generator 272 does not generate a target output (e.g., for unsupervised learning). In some embodiments, the dataset generator generates one or more target outputs 220 corresponding to the data inputs 210 (e.g., for supervised learning). The dataset also includes mapping data that maps the data inputs 210 to the target outputs 220. The data inputs 210 are also referred to as “features,” “attributes,” or “information.” In some embodiments, the dataset generator 272 provides a dataset to the training engine 182, the validation engine 184, or the test engine 186, where the dataset is used to train, validate, or test a machine learning model 190 (e.g., associated with adjusting film deposition parameters based on residual thickness during substrate manufacturing, methods 400A-C, etc.).

[0081] In some embodiments, the data set generator 272 generates the data input 210 and the target output 220. In some embodiments, the data input 210 includes one or more sets of historical characteristic data 244 (e.g., total residual thickness values, material thickness values, etc.) (e.g., associated with adjusting film deposition parameters based on residual thickness during substrate fabrication, methods 400A-400C, etc.). In some embodiments, the historical characteristic data 244 includes one or more of: characteristic data from one or more sensors and / or metrology devices; a combination of characteristic data from one or more sensors and / or metrology devices; and / or a pattern from characteristic data from one or more sensors and / or metrology devices, etc.

[0082] In some embodiments, the dataset generator 272 generates a first data input corresponding to a first set of historical characteristic data 244A for training, validating, or testing a first machine learning model (e.g., associated with adjusting film deposition parameters based on residual thickness during substrate manufacturing, methods 400A-400C, etc.), and the dataset generator 272 generates a second data input corresponding to a second set of historical characteristic data 244B for training, validating, or testing a second machine learning model (e.g., associated with adjusting film deposition parameters based on residual thickness during substrate manufacturing, methods 400A-400C, etc.).

[0083] In some embodiments, the dataset generator 272 discretizes (e.g., segments) one or more of the data inputs 210 or the target outputs 220 (e.g., for use in a classification algorithm for a regression problem). Discretizing the data inputs 210 or the target outputs 220 (e.g., segmenting via a sliding window) converts continuous values ​​of variables into discrete values. In some embodiments, the discrete values ​​of the data inputs 210 represent discrete historical characteristic data 144 to obtain the target outputs 220 (e.g., discrete historical performance data 154).

[0084] The data input 210 and target output 220 for training, validating, or testing the machine learning model include information about a particular facility (e.g., a particular substrate manufacturing facility). In some examples, the historical characteristic data 244 and the historical performance data 254 are for the same manufacturing facility (e.g., associated with adjusting film deposition parameters based on residual thickness during substrate manufacturing, methods 400A-400C, etc.).

[0085] In some embodiments, the information used to train the machine learning model is from manufacturing tools 124 of a particular type at a manufacturing facility having particular characteristics, allowing the trained machine learning model (e.g., associated with adjusting film deposition parameters based on residual thickness during substrate manufacturing, methods 400A-400C, etc.) to determine an outcome for a particular group of manufacturing tools 124 based on input of current parameters (e.g., current characteristic data 146) associated with one or more components that share the characteristics of the particular group. In some embodiments, the information used to train the machine learning model is for components from more than one manufacturing facility, allowing the trained machine learning model to determine an outcome for a component based on input from one manufacturing facility.

[0086] In some embodiments, after generating the dataset and using the dataset to train, validate, or test the machine learning model 190, the machine learning model 190 (e.g., associated with adjusting film deposition parameters based on residual thickness during substrate manufacturing, methods 400A-400C, etc.) is further trained, validated, or tested (e.g., current performance data 156 in FIG. 1 ) or adjusted (e.g., by adjusting weights associated with the input data of the machine learning model 190, such as connection weights in a neural network).

[0087] A machine learning model processes inputs and generates outputs (e.g., associated with adjusting film deposition parameters based on residual thickness during substrate manufacturing, methods 400A-400C, etc.). An artificial neural network includes an input layer consisting of values ​​in data points. The next layer is called a hidden layer, and each node in the hidden layer receives one or more input values. Each node contains parameters (e.g., weights) to apply to the input values. Thus, each node essentially takes the input values ​​and turns them into a multivariate function (e.g., a nonlinear mathematical transformation) to generate an output value. The next layer can be another hidden layer or an output layer. In either case, the nodes in the next layer receive output values ​​from the nodes in the previous layer, and each node applies a weight to that value before generating its own output value. This can occur at each layer. The final layer is the output layer, where there is one node for each class, prediction, and / or output that the machine learning model can generate.

[0088] Thus, the input may include one or more predictions or estimates (e.g., associated with adjusting film deposition parameters based on residual thickness during substrate fabrication, methods 400A-400C, etc.). For example, the output predictions or estimates may include one or more predictions of deposition drift, film deposition on chamber components, erosion of chamber components, predicted failure of chamber components, predicted failure of the deposition process, etc. Processing logic determines an error (i.e., classification error) based on the difference between the output (e.g., prediction or estimate) of the machine learning model and a target label associated with the input training data. Processing logic adjusts 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 the error, the artificial neural network adjusts one or more of its own parameters (weights for one or more inputs of a node) of one or more of its own nodes. Parameters may be updated in a back-propagation manner, such 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 neuron in the previous layer. Adjusting the parameters may therefore include adjusting the weights assigned to each input for one or more neurons in one or more layers within the artificial neural network.

[0089] After one or more rounds of training, the processing logic can determine whether a stopping criterion is met. The stopping criterion can be a target level of accuracy, a target number of processed images from the training dataset, a target amount of change to a parameter relative to one or more previous data points, combinations thereof, and / or other criteria. In some embodiments, 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 can be, for example, 70%, 80%, or 90% accuracy. In some embodiments, the stopping criterion is met when the accuracy of the machine learning model no longer improves. If the stopping criterion is not met, further training is performed. If the stopping criterion is met, training can be completed. Once the machine learning model is trained, a reserved portion of the training dataset is used to test the model.

[0090] 3 is a block diagram illustrating a system 300 for generating predicted data 360 (e.g., predicted data 160 of FIG. 1), according to certain embodiments. System 300 is used to determine predicted data 360 via a trained machine learning model (e.g., model 190 of FIG. 1) (e.g., associated with residual thickness compensation, methods 400A-400C, etc.).

[0091] In block 310, the system 300 (e.g., the prediction system 110 of FIG. 1 ) performs data partitioning (e.g., via the data set generator 172 of the server machine 170 of FIG. 1 ) of historical data (e.g., the historical characteristic data 344 and the historical performance data 354 of the model 190 of FIG. 1 ) to generate a training set 302, a validation set 304, and a test set 306 (e.g., associated with adjusting film deposition parameters based on residual thickness during substrate manufacturing, methods 400A-400C, etc.). In some examples, the training set represents 60% of the historical data, the validation set represents 20% of the historical data, and the test set represents 20% of the historical data. The system 300 generates multiple sets of features for each of the training set, validation set, and test set. In some examples, if the historical data includes features from 20 sensors (e.g., sensor 126 of FIG. 1 , sensors on manufacturing equipment, and / or metrology devices) and 100 products (e.g., products each corresponding to characteristic data from the 20 sensors), the first set of features would be sensors 1-10, the second set of features would be sensors 11-20, the training set would be products 1-60, the validation set would be products 61-80, and the test set would be products 81-100. In this example, the first set of features in the training set would be parameters from sensors 1-10 for products 1-60.

[0092] In block 312, the system 300 performs model training (e.g., via the training engine 182 of FIG. 1 associated with adjusting film deposition parameters based on residual thickness during substrate manufacturing, methods 400A-400C, etc.) using the training set 302. In some embodiments, the system 300 trains multiple models using multiple sets of features in the training set 302 (e.g., a first set of features in the training set 302, a second set of features in the training set 302, etc.). For example, the system 300 trains machine learning models to generate a first trained machine learning model using a first set of features in the training set (e.g., characteristic data from sensors 1-10 for products 1-60) and to generate a second trained machine learning model using a second set of features in the training set (e.g., characteristic data from sensors 11-20 for products 1-60). In some embodiments, the first trained machine learning model and the second trained machine learning model are combined to generate a third trained machine learning model (e.g., in some embodiments, the third trained machine learning model is itself a better predictor than either the first trained machine learning model or the second trained machine learning model). In some embodiments, the feature sets used in comparing the models overlap (e.g., a first set of features is characteristic data from sensors 1-15, and a second set of features is characteristic data from sensors 5-20). In some embodiments, hundreds of models are generated, including models with various permutations of features and combinations of models.

[0093] At block 314, system 300 performs model validation (e.g., via validation engine 184 of FIG. 1 ) using validation set 304. System 300 validates each of the trained models (e.g., associated with adjusting film deposition parameters based on residual thickness during substrate manufacturing, methods 400A-400C, etc.) using a corresponding set of features in validation set 304. For example, system 300 validates a first trained machine learning model using a first set of features in the validation set (e.g., parameters from sensors 1-10 for products 61-80) and generates a second trained machine learning model using a second set of features in the validation set (e.g., parameters from sensors 11-20 for products 61-80). In some embodiments, system 300 validates hundreds of models (e.g., various permutations of features, combinations of models, etc.) generated in block 312. In block 314, the system 300 determines the accuracy of each of the one or more trained models (e.g., via model validation) and determines whether one or more of the trained models have an accuracy that meets the threshold accuracy. In response to a determination that none of the trained models have an accuracy that meets the threshold accuracy, the flow returns to block 312, where the system 300 performs model training using a different feature set from the training set. In response to a determination that one or more of the trained models have an accuracy that meets the threshold accuracy, the flow proceeds to block 316. The system 300 discards trained machine learning models whose accuracy is below the threshold accuracy (e.g., based on a validation set).

[0094] In block 316, the system 300 performs model selection (e.g., via selection engine 185 of FIG. 1 ) to determine which of the one or more trained models that meet the threshold accuracy has the highest accuracy (e.g., based on the validation of selected model 308, block 314). In response to a determination that two or more trained models that meet the threshold accuracy have the same accuracy, flow returns to block 312, where the system 300 performs further training of models using further refined training sets corresponding to further refined feature sets to determine the trained model with the highest accuracy.

[0095] At block 318, the system 300 performs model testing (e.g., via the test engine 186 of FIG. 1 ) using the test set 306 to test the selected model 308. The system 300 tests the first trained machine learning model using a first set of features in the test set (e.g., characteristic data from sensors 1-10 for products 81-100) and determines (e.g., based on the first set of features in the test set 306) whether the first trained machine learning model meets a threshold accuracy. In response to the accuracy of the selected model 308 not meeting the threshold accuracy (e.g., the selected model 308 is overfitted to the training set 302 and / or the validation set 304 and cannot be applied to other datasets, such as the test set 306), flow proceeds to block 312, where the system 300 performs model training (e.g., model retraining) using a different training set (e.g., characteristic data from a different sensor) corresponding to a different feature set. In response to a determination based on the test set 306 that the selected model 308 has an accuracy that meets the threshold accuracy, flow proceeds to block 320. At least in block 312, the model learns patterns in the historical data to make predictions, and in block 318, the system 300 applies the model to the remaining data (e.g., the test set 306) to test the predictions (e.g., associated with adjusting film deposition parameters based on residual thickness during substrate manufacturing, methods 400A-400C, etc.).

[0096] In block 320, the system 300 receives current characteristic data 346 (e.g., current characteristic data 146 of FIG. 1 ) using the trained model (e.g., selected model 308), determines (e.g., extracts) predicted data 360 (e.g., predicted data 160 of FIG. 1 ) from the trained model, and performs corrective action (e.g., updating a process recipe, causing a modification of deposition process parameters, processing a substrate based on the updated recipe, etc.) for adjusting film deposition parameters based on residual thickness during substrate manufacturing. In some embodiments, the current characteristic data 346 corresponds to the same type of features in the historical characteristic data 344. In some embodiments, the current characteristic data 346 corresponds to the same type of features as the type of subset of features in the historical characteristic data 344 used to train the selected model 308 (e.g., associated with adjusting film deposition parameters based on residual thickness during substrate manufacturing, methods 400A-400C, etc.).

[0097] In some embodiments, current data is received. In some embodiments, the current data includes current performance data 356 (e.g., current performance data 156 of FIG. 1 ) and / or current characteristic data 346 (e.g., associated with adjusting film deposition parameters based on residual thickness during substrate fabrication, methods 400A-400C, etc.). In some embodiments, at least a portion of the current data is received from a metrology tool (e.g., metrology tool 128 of FIG. 1 ) or via user input. In some embodiments, a model is retrained based on the current data. In some embodiments, a new model is trained based on the current performance data 356 and the current characteristic data 346.

[0098] In some embodiments, one or more of blocks 310-320 are performed in various orders and / or with other steps not presented and described herein. In some embodiments, one or more of blocks 310-320 are not performed. For example, in some embodiments, one or more of data partitioning of block 310, model validation of block 314, model selection of block 316, and / or model testing of block 318 are not performed.

[0099] 4A-4C are flow diagrams of methods 400A-400C associated with residual thickness compensation (e.g., residual-based adjustment of film deposition parameters during substrate fabrication), according to certain embodiments. In some embodiments, methods 400A-400C are performed by processing logic, including hardware (e.g., circuitry, dedicated logic, programmable logic, microcode, processing device, etc.), software (e.g., instructions executed on a processing device, general-purpose computer system, or dedicated machine), firmware, microcode, or a combination thereof. In one embodiment, method 400A may be performed by a computer system, such as computer system architecture 100 of FIG. 1. In other or similar embodiments, one or more steps of method 400A may be performed by one or more other machines not depicted in the figures. In some embodiments, methods 400A-400C are performed, at least in part, by prediction system 110. In some embodiments, method 400A is performed by client device 120 (e.g., corrective action component 122) and / or prediction system 110 (e.g., prediction component). In some embodiments, method 400B is performed by server machine 180 (e.g., training engine 182, etc.). In some embodiments, method 400C is performed by prediction server 112 (e.g., prediction component 114) and / or client device 120 (e.g., corrective action component 122). In some embodiments, a non-transitory storage medium stores instructions that, when executed by a processing device (e.g., prediction system 110, server machine 180, prediction server 112, client device 120, etc.), cause the processing device to perform one or more of methods 400A-400C.

[0100] For ease of explanation, methods 400A-400C are shown and described as a series of steps. However, steps according to the present disclosure may be performed in various orders and / or simultaneously, and with other steps not shown or described herein. Moreover, in some embodiments, not all illustrated steps are performed to implement methods 400A-400C in accordance with the disclosed subject matter. Furthermore, those skilled in the art will understand and appreciate that method 800 may alternatively be represented as a series of interrelated states or events via a state diagram.

[0101] FIG. 4A is a flow diagram of a method associated with adjusting film deposition parameters based on deposition thickness during substrate manufacturing according to an embodiment of the present disclosure.

[0102] Referring to FIG. 4A , in some embodiments, in block 401, processing logic performing method 400A identifies a material associated with a substrate processing step of a recipe. In some embodiments, the identified material can be reused between loops, and a loop can have any number of materials in any order. In some embodiments, deposition drift compensation is based on the identified material and the substrate processing step associated with the material. In some embodiments, processing logic can receive user input identifying a material associated with a substrate processing step of a recipe. In other embodiments, processing logic can automatically select a material associated with a substrate processing step of a recipe. For example, processing logic can identify a material associated with a substrate processing step of a recipe for which at least one process run has completed and a material thickness value has been determined. In yet another embodiment, processing logic can identify a material associated with a substrate processing step of a recipe based on a currently running process recipe. For example, processing logic can perform a deposition process on a substrate according to a process recipe. The deposition process can be performed in one or more processing chambers. The process recipe can include one or more deposition parameters for the deposition process. For example, deposition parameters may include deposition time for each layer and / or each loop of a process recipe, temperature settings for the process chamber, pressure settings for the process chamber, precursor flow settings for materials included in the film being deposited on the substrate surface, showerhead height, etc. In some embodiments, deposition parameters may vary for each substrate processing region. A deposition process may deposit multiple layers on a substrate. For example, a deposition process may deposit alternating layers of oxide and nitride, or alternating layers of oxide and polysilicon.

[0103] In block 402, the processing logic determines an expected total residual thickness value after the substrate processing operation. For example, the expected total residual thickness value may include a measured thickness of the residual film, a measured thickness of one or more layers of the residual film, and / or a measured thickness of one or more loops of the residual film. The expected total residual thickness value may be measured using metrology tool 128. In some embodiments, a thickness profile is obtained from data store 140. In some embodiments, the expected total residual thickness may be derived from a process recipe. For example, the process recipe may be designed to deposit a 200 angstrom film. In such a case, the expected total residual thickness may be 200 angstroms. In some embodiments, the expected total residual thickness value may be measured using sensor 126 that measures deposition on the process chamber walls. In some embodiments, the expected total residual thickness may be determined based on at least one deposition process parameter (e.g., deposition time, deposition rate, etc.).

[0104] In some embodiments, the processing logic can determine the characteristic data (e.g., an expected total residual thickness value) by a sensor collecting the characteristic data (e.g., by measuring a deposition residual thickness on a processing chamber wall). In some embodiments, the processing logic can identify the characteristic data based on an expected thickness to be added to the substrate (e.g., a thickness parameter of a deposition process, a measured thickness from a previous run of the same deposition process, etc.). In some embodiments, the processing logic can identify the characteristic data based on the material deposited and the amount of time it was deposited (e.g., a process recipe, deposition process parameters, etc.). In some embodiments, the processing logic can identify the characteristic data based on sensor data and / or metrology data (e.g., a total residual thickness value, a material thickness value, an RF power of a substrate processing process, an interval value of a substrate processing process, a gas flow rate value of a substrate processing process, or a chamber pressure value of a substrate processing process, etc.) of a substrate subjected to a substrate processing process (e.g., a deposition process).

[0105] In block 403, processing logic determines an expected material thickness value for the material associated with the substrate processing step based on the expected total residual thickness value.

[0106] In some embodiments, processing logic can use one or more mathematical formulas or models to determine an expected material thickness value associated with a substrate processing step. In some embodiments, a relationship can exist between the expected total residual thickness and the expected material thickness (e.g., based on the identified materials associated with the substrate processing step in the recipe). In some embodiments, this relationship can be a mathematical relationship, formula, function, or the like. In some embodiments, this relationship can be derived using at least the expected material thickness (e.g., data points representing the material thickness after the deposition step), the expected total residual thickness (e.g., data points representing the wall residual thickness, contamination, etc.), and the deposition time (data points representing the duration of the deposition step in the process recipe). In some embodiments, the input to the formula or function is the expected total residual thickness, and the output of the formula is the expected material thickness. In some embodiments, processing logic can appropriately scale the deposition step time to compensate for drift.

[0107] In some embodiments, a first relationship associated with a material can correspond to a first substrate processing region (e.g., a slot), and a second relationship associated with the same material can correspond to a second substrate processing region (e.g., a relationship based on the same material can be specific to a particular slot).

[0108] In some embodiments, processing logic may generate a mathematical equation (e.g., polynomial, linear, logarithmic, etc.) associated with the identified material. In some embodiments, the mathematical equation may be reused between loops. In some embodiments, there may be a material-based mathematical equation specific to each substrate processing region (e.g., slot). For example, a third-order polynomial may be used, where y=ax 3 +bx2 +cx+d, where (x, y) are coordinates and a, b, c, and d are constants. While a third-order polynomial is used as an example, polynomials (of any order), linear, logarithmic, exponential, etc., may be used. In some embodiments, the formula may be generated using a set of expected total residual thickness values ​​(e.g., chamber wall residual thickness after loop 1, chamber wall residual thickness after loop 2, chamber wall residual thickness after loop 3, etc.) as x variables and expected material thickness values ​​(e.g., expected deposition thickness at various chamber wall residual thicknesses) as y coordinates. The (x, y) values ​​may be obtained from the expected profile. The set of (x, y) coordinates are used to determine constants for the formula. In some embodiments, the total residual thickness (e.g., chamber wall residual thickness) is the seasoning thickness plus the deposition thickness, where the deposition thickness is equal to the total thickness of all previous loops. The seasoning layer can include a layer (e.g., a silicon oxide layer) that overlies the chamber walls before a substrate is introduced into the chamber for processing. The deposited seasoning layer reduces the likelihood of contaminants interfering with subsequent processing steps.

[0109] In some embodiments, processing logic can use the formula and the total residual thickness to determine the value of (y) for a particular loop. In particular, processing logic can receive input (e.g., user-based input, automated input, etc.) indicating a layer or a loop of a deposition process. Processing logic can then input the actual material thickness of the loop or layer (obtained from the thickness profile) into the formula to calculate the value of (y) for the loop or layer.

[0110] In some embodiments, the expected material thickness value may include a desired thickness of the film, a desired thickness of one or more layers of the film, and / or a desired thickness of one or more loops of the film, etc. In some embodiments, the expected material thickness value is retrieved from data store 140. In some embodiments, the expected material thickness value may be the difference between the desired material thickness value and the actual material thickness value. In some embodiments, determining the expected material thickness value may include providing the expected total residual thickness value as an input to a trained machine learning model. In some embodiments, determining the expected material thickness value may include receiving output associated with prediction data from the trained machine learning model, where the expected material thickness value is associated with the prediction data. In some embodiments, the trained machine learning model may be trained using data input including historical total residual thickness values ​​and a target output of the historical material thickness values.

[0111] At block 404, processing logic updates the recipe based on the material and the expected material thickness value to generate an updated recipe (including updated deposition process parameters such as, for example, a time value (for deposition), an updated time value, an RF power for the substrate processing process, an interval value for the substrate processing process, a gas flow rate value for the substrate processing process, a chamber pressure value for the substrate processing process, a chamber temperature, etc.). In some embodiments, updating the recipe may include determining an updated time value associated with the substrate processing process. In some embodiments, determining an updated time value associated with the substrate processing process may be based on a time value associated with the substrate processing process, an expected total remaining thickness value, and an expected material thickness value. In some embodiments, updating the recipe may include determining at least one of an updated RF power for the substrate processing process, an updated interval value for the substrate processing process, an updated gas flow rate value for the substrate processing process, or an updated chamber pressure value for the substrate processing process.

[0112] In some embodiments, the updated recipe may include one or more corrective actions (e.g., deposition step parameter updates, process recipe updates, etc.) applied to parameters of a process recipe (e.g., associated with a processing chamber during one or more steps of the process recipe). In particular, the updated recipe may include adjustments to the deposition time, temperature settings of a processing chamber, pressure settings of a processing chamber, flow rates of precursors for materials included in a film to be deposited on the substrate surface, power supplied to a processing chamber, ratios of two or more settings, etc. for each of one or more layers and / or loops. For example, the updated recipe may include a deposition time adjustment for a loop of a process recipe. In some embodiments, the updated recipe may include a set of parameter adjustments for each layer and / or loop of a process recipe. For example, the updated recipe may include adjustments to the deposition time for a first loop, a second loop, a third loop, etc., up to the final loop. Each adjustment may be applied to a respective deposition step to adjust the thickness of one or more loops or layers so that the thickness of the film stack is equal to the expected film thickness indicated by the expected total residual thickness. For example, the expected total residual thickness may include the measured thickness of the film, the measured thickness of one or more layers of the film, and / or the measured thickness of one or more loops of the film, etc. The thickness profile may be measured using metrology device 128 and / or one or more sensors 126. In some embodiments, the thickness profile is obtained from data store 140.

[0113] For example, if the expected film thickness after loop 39 (e.g., expected total residual thickness) is a first value (e.g., 30,000 nm), the expected film thickness after loop 40 is a second value (e.g., 30,500 nm), and the actual film thickness (e.g., actual material thickness) during deposition and after loop 39 is a third value (e.g., 30,050), the updated recipe may indicate a modification to the deposition time of loop 40 (e.g., shortening the deposition time of loop 40 by a certain period of time) so that the actual film thickness after loop 40 equals the expected film thickness (e.g., 30,500 nm).

[0114] In some embodiments, processing logic can generate an updated recipe using one or more mathematical formulas or models associated with materials associated with the substrate processing steps of the recipe. For example, processing logic can generate a curve-fitting model using data values ​​from the predicted total remaining thickness and / or predicted material thickness, and then use the curve-fitting model to determine offset time values ​​for particular steps during the current deposition process.

[0115] In some embodiments, for example, a mathematical expression (e.g., a polynomial of degree 3, y=ax 3 +bx 2 +cx+d, where (x, y) are coordinates and a, b, c, d are constants), may be used to update a recipe (e.g., a recipe for a substrate processing region). For example, in some embodiments, processing logic may use a formula and a total residual thickness to determine the value of (y) for a particular loop. In particular, processing logic may receive input (e.g., user-based input, automatic input, etc.) indicating a layer or a loop of a deposition process. Processing logic may then input the actual material thickness of the loop or layer (obtained from the thickness profile) into the formula to calculate the value of (y) for the loop or layer. In some embodiments, processing logic generates an updated recipe based on the calculated value of (y). In some embodiments, the updated recipe may be generated based on the following formula: For example, updated recipe=(y[first loop] / y[current loop number])*t step where t step is the expected time for the selected loop. Although curve fitting methods are described for determining the correction profile (e.g., updated recipe), other models, including but not limited to regression analysis, least squares methods, etc., may be used to generate the correction profile.

[0116] In some embodiments, processing logic may generate the updated recipe using a machine learning model (e.g., machine learning model 190) or using an inference engine.

[0117] In block 405, the processing logic processes the substrate based on the updated recipe. For example, the processing logic may deposit a first set of film layers on the substrate (e.g., execute a first set of substrate processing steps), determine an expected total residual thickness value for the deposited film, generate an updated recipe to correct any defects detected during deposition of the first set of film layers, apply the updated recipe to the process recipe, and deposit a second set of film layers on the substrate (e.g., execute a second set of substrate processing steps). Thus, the deposition process recipe may be adjusted in real time or near real time (e.g., substantially real time). This process may be repeated for each deposition step of the process recipe.

[0118] In some embodiments, the updated recipe of block 405 may still cause drift in the processing chamber and during deposition of layers on the substrate. In some embodiments, performance data associated with the updated recipe is identified after block 405. In some embodiments, certain portions of method 400A may be updated based on the performance data (e.g., updating the expected total residual thickness associated with the substrate processing step of the recipe, updating the expected material thickness value, updating deposition process parameters, etc.). In some embodiments, the updated method 400A may be repeated to more accurately execute the substrate processing step of the recipe. In some embodiments, processing logic may determine whether the drift (e.g., the material thickness value) meets a first threshold (e.g., exceeds a certain value) and repeat method 400A until the drift meets a second threshold (e.g., falls below a certain value). FIG. 4B illustrates a method of training a machine learning model (e.g., model 190 of FIG. 1 ) to determine predicted data associated with residual thickness compensation (e.g., predicted data 160 of FIG. 1 ) according to an aspect of the present disclosure.

[0119] 4B, in block 410 of method 400B, processing logic identifies historical property data for the substrate (e.g., historical total residual thickness values, historical materials associated with historical substrate processing steps of historical recipes, historical material thickness values, historical property data 144, etc.). The historical property data may include data from historical recipes, historical deposition process parameters, historical substrates, and / or historical process chamber property data, etc.

[0120] In some embodiments, at block 412, processing logic identifies historical performance data (e.g., historical material thickness values, the difference between actual and predicted material thickness values, the difference between actual and predicted total residual thickness values, historical performance data 154 of FIG. 1 ) of historical deposition process parameters, process recipes, substrates, and / or process chambers (e.g., substrates meeting a threshold, substrates not meeting a threshold, recipes meeting a threshold, recipes not meeting a threshold, etc.). The historical performance data can include historical material thickness values ​​(e.g., the difference between actual and predicted material thickness values, the difference between actual and predicted total residual thickness values, etc.) and / or historical process chamber data (e.g., one or more values ​​of the accuracy of the deposition process and / or fulfillment of a predicted total residual thickness value (e.g., after the deposition process), etc.). The performance data, including historical performance data, may include sensor data and / or metrology data (e.g., total residual thickness values, material thickness values, RF power values, spacing values, gas flow rates, or chamber pressure values, etc.), or user input indicative of the performance of a substrate when certain parameters are met or when a certain level of performance is achieved (e.g., the ability to pass a probe test that measures voltage). The performance data, including historical performance data, may include sensor data and / or metrology data, or user input indicative of the performance of a processing chamber when certain parameters are met or when a certain level of performance is achieved (e.g., the ability to pass a deposition drift test). At least a portion of the historical characteristic data and historical performance data may be associated with parts of a new substrate processing apparatus (e.g., used for benchmarking). At least a portion of the historical characteristic data and historical performance data may be associated with manufactured substrates. At least a portion of the historical characteristic data and historical performance data may be associated with a processing chamber.

[0121] At block 414, the processing logic trains the machine learning model using data inputs including historical characteristic data 144 and / or target outputs including historical performance data 154 to generate a trained machine learning model.

[0122] In some embodiments, the historical characteristic data is of a previous substrate or processing chamber, and / or the historical performance data corresponds to a previous substrate or processing chamber. In some embodiments, the historical characteristic data corresponds to a deposition process or process recipe, or to a substrate or processing chamber subjected to a deposition process or process recipe. In some embodiments, the historical characteristic data includes previous measurements of a previous substrate or processing chamber, and / or the historical performance data corresponds to a previous substrate or processing chamber. The historical performance data may be associated with substrate quality, such as substrate metrology data, substrate throughput, or substrate defects. The historical performance data may be associated with the quality of a process recipe (e.g., a deposition process), such as the agreement of actual and predicted total residual thickness values. The historical performance data may be associated with the quality of process recipe or deposition process parameters, such as the ability to accurately deposit each layer and / or film in accordance with the predicted total residual thickness of the deposition process. The historical performance data may be associated with the quality of components of a substrate processing equipment, such as test data, substrate metrology data, or substrate failure times.

[0123] At block 414, the processing logic trains a machine learning model using data inputs including historical characteristic data 144 (e.g., historical total residual thickness values) and / or target outputs including historical performance data 154 (e.g., historical material thickness values) to generate a trained machine learning model.

[0124] In some embodiments, the historical characteristic data is of a previous substrate or processing chamber, and / or the historical performance data corresponds to a previous substrate or processing chamber. In some embodiments, the historical characteristic data includes previous measurements of a previous substrate or processing chamber, and / or the historical performance data corresponds to a previous substrate or processing chamber. The historical performance data may be associated with substrate quality, such as substrate metrology data (e.g., total residual thickness values, material thickness values, etc.), substrate throughput, and substrate defects. The historical performance data may be associated with the quality of a process recipe or deposition operation, such as the ability to accurately deposit an expected total residual thickness and / or an expected material thickness. The historical performance data may be associated with the quality of components of a substrate processing equipment, such as test data, substrate metrology data, and substrate failure times.

[0125] In block 414, a machine learning model may be trained using a target output including the historical characteristic data 144 and / or the historical performance data 154 to generate a trained machine learning model configured to update a recipe, generate an updated recipe, and / or trigger corrective action (e.g., process a substrate based on the updated recipe) based on the characteristic data. In some embodiments, the trained machine learning model may be configured to predict performance data 152 (e.g., performance data including an updated process recipe, updated deposition process parameters, performance data for substrates processed with the updated process recipe, performance data for substrates processed with the updated deposition process parameters, etc.) based on the characteristic data 142 (e.g., the predicted total residual thickness value of block 402 of FIG. 4A , the predicted material thickness value of block 403 of FIG. 4A ). In response to the predicted performance data meeting a threshold (e.g., deposition drift exceeding a certain value), processing logic may trigger corrective action (e.g., update the deposition process parameters, update the process recipe, process a substrate based on the updated recipe, etc.). In response to the predicted performance data not meeting a threshold, the processing logic may prevent corrective action from being taken (e.g., ensuring the process recipe remains the same, ensuring the deposition process parameters remain the same, etc.).

[0126] In some embodiments, the deposition process parameters may include a time value, an updated time value, a recipe, an updated recipe, an RF power for the substrate processing process, an interval value for the substrate processing process, a gas flow rate value for the substrate processing process, a chamber pressure value for the substrate processing process, an updated RF power for the substrate processing process, an updated interval value for the substrate processing process, an updated gas flow rate value for the substrate processing process, an updated chamber pressure value for the substrate processing process, etc.

[0127] FIG. 4C is a method 400C of using a trained machine learning model (e.g., model 190 of FIG. 1) associated with adjusting deposition parameters based on residual thickness during substrate manufacturing to trigger the execution of corrective action.

[0128] 4C, in block 420 of method 400C, processing logic identifies characteristic data. In some embodiments, the characteristic data in block 420 includes an expected total residual thickness value, an expected material thickness value, etc. In some embodiments, block 420 is similar to blocks 401 and 402 of FIG. 4A.

[0129] At block 422, processing logic provides the characteristic data as data input to a trained machine learning model (e.g., trained via block 414 of FIG. 4B ). In some embodiments, the trained machine learning model can be associated with a variable relationship between expected total residual thickness and expected material thickness.

[0130] At block 424, processing logic receives output from the trained machine learning model associated with the prediction data, where the updated recipe is based on the prediction data.

[0131] At block 426, processing logic causes the execution of corrective actions based on the predictive data.

[0132] 4A includes training a machine learning model to determine an expected material thickness value for a material associated with a substrate processing operation based on the expected total residual thickness value. In some embodiments, block 403 of FIG. 4A includes using the trained machine learning model to: Determining an expected material thickness value for a material associated with the substrate processing step based on the expected total residual thickness value.

[0133] In some embodiments, block 404 of Figure 4A includes training a machine learning model to update the recipe based on the materials and expected material thickness values ​​of the materials to generate an updated recipe. In some embodiments, block 404 of Figure 4A includes using a machine learning model to update the recipe based on the materials and expected material thickness values ​​of the materials to generate an updated recipe.

[0134] In some embodiments, the characteristic data 142 is an expected total residual thickness value (e.g., an expected deposition thickness associated with a deposition process), and the trained machine learning model of block 422 was trained using data inputs including historical expected total residual thickness values ​​and / or historical actual total residual thickness values, and target outputs including historical performance data 154 (e.g., actual total residual thickness values, actual material thickness values, etc.).

[0135] In some embodiments, the characteristic data 142 is deposition thickness characteristic data (e.g., expected total residual thickness value, expected material thickness value, etc.), and the trained machine learning model of block 422 was trained using a data input including historical deposition thickness characteristic data and a target output including historical performance data 154 including historical deposition thickness characteristic data of past substrates or past deposition processes. The predicted data 160 of block 424 can be associated with predicted performance data (e.g., performance data for a substrate, or performance data for a deposition recipe or process) based on the characteristic data. In response to the predicted performance data meeting a threshold (e.g., deposition drift exceeding a certain level), processing logic can trigger corrective action (e.g., updating the process recipe, updating deposition parameters, etc.). In response to the substrate not meeting the threshold, processing logic can prevent corrective action (e.g., ensuring the process recipe remains the same, ensuring the deposition process parameters remain the same, etc.).

[0136] 5 is a block diagram illustrating a computer system 500, according to certain embodiments. In some embodiments, computer system 500 is one or more of client device 120, prediction system 110, server machine 170, server machine 180, and / or prediction server 112, etc.

[0137] In some embodiments, computer system 500 is connected to other computer systems (e.g., via a network such as a local area network (LAN), an intranet, an extranet, or the Internet). In some embodiments, computer system 500 operates in the capacity of a server or a client computer in a client-server environment, or as a peer computer in a peer-to-peer or distributed network environment. In some embodiments, computer system 500 is provided by a personal computer (PC), a tablet PC, a set-top box (STB), a personal digital assistant (PDA), a mobile phone, a web appliance, a server, a network router, switch, or bridge, or any device capable of executing (sequentially or otherwise) a set of instructions that specify actions to be performed by the device. Furthermore, the term "computer" is intended to include any collection of computers that individually or jointly execute a set (or sets) of instructions to perform any one or more of the methods described herein.

[0138] In a further aspect, the computer system 500 includes a processing device 502, a volatile memory 504 (e.g., random access memory (RAM)), a non-volatile memory 506 (e.g., read-only memory (ROM) or electrically erasable programmable ROM (EEPROM)), and a data storage device 518, which communicate with each other via a bus 508.

[0139] In some embodiments, processing device 502 is provided by one or more processors, such as a general-purpose processor (e.g., a Complex Instruction Set Computing (CISC) microprocessor, a Reduced Instruction Set Computing (RISC) microprocessor, a Very Long Instruction Word (VLIW) microprocessor, a microprocessor implementing another type of instruction set or a combination of instruction set types), or a specialized processor (e.g., an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA), a Digital Signal Processor (DSP), or a network processor).

[0140] In some embodiments, computer system 500 further includes a network interface device 522 (e.g., connected to a network 574). In some embodiments, computer system 500 also includes a video display unit 510 (e.g., a liquid crystal display (LCD)), an alphanumeric input device 512 (e.g., a keyboard), a cursor control device 514 (e.g., a mouse), and a signal generating device 520.

[0141] In some embodiments, data storage device 518 includes a non-transitory computer-readable storage medium 524 storing instructions 526 that encode any one or more of the methods or functions described herein, where the instructions 526 include instructions for encoding the components of FIG. 1 (e.g., corrective action component 122, prediction component 114, etc.) and for performing the methods described herein (e.g., one or more of methods 400A-400C).

[0142] In some embodiments, the instructions 526 also reside, completely or partially, within the volatile memory 504 and / or within the processing device 502 during execution by the computer system 500; thus, in some embodiments, the volatile memory 504 and the processing device 502 also constitute machine-readable storage media.

[0143] Although the computer-readable storage medium 524 is shown as a single medium in the illustrated example, the term "computer-readable storage medium" is intended 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 executable instructions. The term "computer-readable storage medium" is also intended to include any tangible medium that can store or encode a set of instructions for execution by a computer that cause the computer to perform any one or more of the methods described herein. The term "computer-readable storage medium" is intended to include, but is not limited to, solid-state memory, optical media, and magnetic media.

[0144] In some embodiments, the methods, components, and features described herein may be implemented by discrete hardware components or may be incorporated into the functionality of other hardware components, such as an application-specific integrated circuit (ASIC), FPGA, DSP, or similar device. Additionally, the methods, components, and features may be implemented by firmware modules or functional circuits within a hardware device. Furthermore, the methods, components, and features may be implemented in any combination of hardware devices and computer program components, or in a computer program.

[0145] Unless otherwise specifically stated, terms such as "identifying," "determining," "updating," "causing," "providing," "receiving," "performing," "obtaining," "accessing," "adding," "training," and the like refer to actions and processes performed by or implemented by a computer system that manipulates and converts data represented as physical (electronic) quantities in the registers and memory of the computer system into other data similarly represented as physical quantities in the memory or registers of the computer system, or other such information storage, transmission, or display devices. Furthermore, in this specification, terms such as "first," "second," "third," "fourth," and the like are presented as labels to distinguish among various elements and may not have a sequential meaning according to their numerical designations.

[0146] The embodiments described herein also relate to apparatus for performing the methods described herein. The apparatus may be specially configured to perform the methods described herein, or may comprise a general-purpose computer system selectively programmed by a computer program stored on the computer system, such computer program being stored on a computer-readable tangible storage medium.

[0147] The methods and illustrative embodiments described herein are not inherently related to any particular computer or other apparatus. Various general-purpose systems may be used in accordance with the teachings described herein, or it may be more convenient to construct more specialized apparatus to perform the methods described herein, and / or each individual function, routine, subroutine, or step thereof. Examples of configurations for various such systems are set forth in the preceding specification.

[0148] The foregoing specification is intended to be illustrative, not limiting. While the present disclosure has been described with reference to certain exemplary implementations and embodiments, it will be understood that the present disclosure is not limited to the described examples and embodiments. The scope of the present disclosure should be defined with reference to the following claims, along with the full scope of equivalents to which such claims are entitled.

Claims

1. Identifying materials associated with substrate processing steps of the recipe; determining an expected total residual thickness value after said substrate processing step; determining an expected material thickness value for the material associated with the substrate processing operation based on the expected total residual thickness value; updating deposition process parameters based on the material and the predicted material thickness values ​​for the material to compensate for deposition drift to generate an updated recipe; processing the substrate based on the updated recipe; A method comprising:

2. The method of claim 1 , wherein the predicted material thickness value is the difference between a desired material thickness value and an actual material thickness value.

3. The method of claim 1 , wherein updating the recipe comprises determining updated time values ​​associated with the substrate processing steps.

4. 4. The method of claim 3, wherein determining the updated time value associated with the substrate processing operation is based on a time value associated with the substrate processing operation, the predicted total residual thickness value, and the predicted material thickness value.

5. 10. The method of claim 1, wherein updating the recipe includes determining at least one of an updated radio frequency (RF) power for the substrate processing operation, an updated interval value for the substrate processing operation, an updated gas flow rate value for the substrate processing operation, or an updated chamber pressure value for the substrate processing operation.

6. Determining the expected material thickness value includes: providing the predicted total residual thickness value as an input to a trained machine learning model; 10. The method of claim 1, further comprising receiving output from the trained machine learning model associated with prediction data, wherein the predicted material thickness value is associated with the prediction data.

7. 7. The method of claim 6, wherein the trained machine learning model is trained using data inputs including historical total residual thickness values ​​and target outputs of historical material thickness values.

8. A non-transitory computer-readable storage medium storing instructions that, when executed, cause a processing device to: Identifying materials associated with substrate processing steps of the recipe; determining an expected total residual thickness value after said substrate processing step; determining an expected material thickness value for the material associated with the substrate processing operation based on the expected total residual thickness value; updating deposition process parameters based on the material and the predicted material thickness values ​​for the material to compensate for deposition drift to generate an updated recipe; processing the substrate based on the updated recipe; A non-transitory computer-readable storage medium for causing a computer to perform steps including:

9. 9. The non-transitory computer-readable storage medium of claim 8, wherein the predicted material thickness value is a difference between a desired material thickness value and an actual material thickness value.

10. The non-transitory computer-readable storage medium of claim 8 , wherein updating the recipe comprises determining updated time values ​​associated with the substrate processing operations.

11. 11. The non-transitory computer-readable storage medium of claim 10, wherein determining the updated time value associated with the substrate processing operation is based on a time value associated with the substrate processing operation, the predicted total residual thickness value, and the predicted material thickness value.

12. 10. The non-transitory computer-readable storage medium of claim 8, wherein updating the recipe includes determining at least one of an updated radio frequency (RF) power for the substrate processing operation, an updated interval value for the substrate processing operation, an updated gas flow rate value for the substrate processing operation, or an updated chamber pressure value for the substrate processing operation.

13. Determining the expected material thickness value includes: providing the predicted total residual thickness value as an input to a trained machine learning model; 10. The non-transitory computer-readable storage medium of claim 8, further comprising: receiving output from the trained machine learning model associated with prediction data, wherein the expected material thickness value is associated with the prediction data.

14. 14. The non-transitory computer-readable storage medium of claim 13, wherein the trained machine learning model is trained using data inputs including historical total residual thickness values ​​and target outputs of historical material thickness values.

15. 1. A system comprising: Memory and a processing device coupled to the memory; wherein the processing device comprises: Identifying materials associated with substrate processing steps of the recipe; determining an expected total residual thickness value after said substrate processing step; determining an expected material thickness value for the material associated with the substrate processing operation based on the expected total residual thickness value; updating deposition process parameters based on the material and the predicted material thickness values ​​for the material to compensate for deposition drift to generate an updated recipe; processing the substrate based on the updated recipe; The system.

16. The system of claim 15 , wherein the predicted material thickness value is the difference between a desired material thickness value and an actual material thickness value.

17. 16. The system of claim 15, wherein to update the recipe, the processing device determines an updated time value associated with the substrate processing operation.

18. 20. The system of claim 17, wherein the processing device determines the updated time value associated with the substrate processing operation based on a time value associated with the substrate processing operation, the predicted total residual thickness value, and the predicted material thickness value.

19. To determine the expected material thickness value, the processing device providing the predicted total residual thickness value as an input to a trained machine learning model; receiving an output from the trained machine learning model associated with prediction data, the predicted material thickness value being associated with the prediction data; and The system of claim 15 , further comprising:

20. 20. The system of claim 19, wherein the trained machine learning model is trained using data inputs including historical total residual thickness values ​​and target outputs of historical material thickness values.

Images