Surface roughness and emissivity determination

Abstract

A system includes a radiation source configured to emit a radiation beam. The system further includes a first optical sensor configured to detect a first intensity of a first portion of the radiation beam reflected from a surface of an object. The system further includes a second optical sensor configured to detect a second intensity of a second portion of the radiation beam scattered by the surface of the object. The system further includes a processing device communicatively coupled with the first optical sensor and the second optical sensor. The processing device is configured to determine at least one of roughness or emissivity of the surface of the object based on a comparison of the first intensity and the second intensity.

Description

Technical Field

[0001] Embodiments of the present disclosure relate generally to determination of surface roughness and emissivity of an object and, more particularly, to systems, methods, and apparatus for optically determining surface roughness and emissivity of an object. Background Art

[0002] Emissivity is a fundamental property of materials. In particular, in semiconductor processing, accurate characterization of the emissivity and / or surface roughness of chamber component surfaces can have a direct impact on the quality of processed substrates. Emissivity can be affected by a variety of material parameters including topography (e.g., surface roughness), reflectivity, etc. 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 extensive overview of the disclosure. It is not intended to identify the key or important elements of the disclosure, nor is it intended to delimit any scope of a particular embodiment of the disclosure or any scope of the claims. Its sole purpose is to present some concepts of the disclosure in a simplified form as a prelude to a more detailed description presented later.

[0004] Some embodiments described herein encompass a system comprising: a radiation source configured to emit a radiation beam. The system further comprises: a first optical sensor configured to detect a first intensity of a first portion of the radiation beam reflected from a surface of an object. The system further comprises: a second optical sensor configured to detect a second intensity of a second portion of the radiation beam scattered by the surface of the object. The system further comprises: a processing device communicatively coupled to the first optical sensor and the second optical sensor. The processing device is configured to determine at least one of a roughness of the surface of the object or an emissivity of the surface of the object based on a comparison of the first intensity and the second intensity.

[0005] Additional or related embodiments described herein encompass a method comprising: emitting a radiation beam from a radiation source. The method further comprises: detecting, by a first optical sensor, a first intensity of a first portion of the radiation beam reflected from a surface of a chamber component of a processing chamber. The method further comprises: detecting, by a second optical sensor, a second intensity of a second portion of the radiation beam scattered by the surface of the chamber component. The method further comprises: determining, via a processing device communicatively coupled to the first optical sensor and the second optical sensor, at least one of a roughness of the surface of the chamber component or an emissivity of the surface of the chamber component based on a comparison of the first intensity and the second intensity.

[0006] In a further embodiment, a non-transitory machine-readable storage medium comprising instructions, which when executed by a processing device, cause the processing device to perform operations including: receiving data associated with at least one of an emissivity or a roughness of a surface of a chamber component of a processing chamber. The operations further include: inputting the data associated with at least one of the emissivity or the roughness of the surface of the chamber component into a trained machine learning model. The operations further include: receiving an output from the trained machine learning model, the output comprising a predicted substrate process result. The predicted substrate process result corresponds to a future substrate to be processed using the chamber component in the processing chamber.

[0007] According to these and other aspects of the present disclosure, numerous other features are provided. Other features and aspects of the present disclosure will become more apparent from the following detailed description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] The present disclosure is illustrated by way of example and not limitation in the accompanying drawings in which like reference numerals indicate like elements. It should be noted that different references to "a" or "one" embodiment in the present disclosure do not necessarily refer to the same embodiment, and such references refer to at least one.

[0009] Figure 1A A simplified side view of a system for optically determining the emissivity and / or surface roughness of an object is shown in accordance with aspects of the present disclosure.

[0010] Figure 1B A simplified side view of a system for optically determining the emissivity and / or surface roughness of an object is shown in accordance with aspects of the present disclosure.

[0011] Figure 2 A cross-sectional view of one embodiment of a processing chamber is depicted.

[0012] Figure 3 An illustrative computer system architecture in accordance with aspects of the present disclosure is depicted.

[0013] Figure 4 A model training workflow and a model application workflow for determining predicted processed substrate outcomes according to aspects of the present disclosure are shown.

[0014] Figure 5A is a flowchart of a method for generating a training data set for training a machine learning model according to aspects of the present disclosure.

[0015] Figure 5Bis a flow chart of a method for generating predicted processed substrate outcomes using a trained machine learning model according to aspects of the present disclosure.

[0016] Figure 6 is a flow chart of a method for optically determining the emissivity and / or surface roughness of an object according to aspects of the present disclosure.

[0017] Figure 7 A diagrammatic representation of a machine in the example form of a computing device is described within which a set of instructions, for causing the machine to perform any one or more of the methodologies discussed herein, may be executed. DETAILED DESCRIPTION

[0018] Embodiments of the present disclosure relate to systems and methods for determining surface roughness and emissivity. The process results of a manufacturing process depend on many factors, including process recipes and chamber component conditions. For example, based on the emissivity and / or surface roughness of components of a processing chamber used to perform a process (such as a deposition process, an etching process, etc.) on a substrate, the process results across the substrate surface may vary. For example, based on the conditions of a showerhead, the conditions of a cover, the conditions of a nozzle, the conditions of a substrate holder supporting a substrate, the conditions of a chamber liner, the conditions of a pump and / or a valve, etc., the process results across the substrate surface may vary. The emissivity and / or surface roughness of one or more of these components may directly affect the quality of a thin film deposited on a substrate. The emissivity of an object (such as a chamber component) may also be affected by various factors, including topography (such as surface roughness). Therefore, it may be useful to classify the surface roughness of an object together with the emissivity.

[0019] Typically, the wavelength range in which the emissivity of chamber components has the greatest impact on the quality of processed substrates is in the mid-IR, particularly in the 3-5 micron range. Conventional emissometers (i.e., tools used to measure emissivity) typically measure and report emissivity in this wavelength range. These conventional emissometers operate on the principle that, for a given sample, there is a direct relationship between emissivity and reflected radiation (i.e., radiation reflected by the surface of the object being measured). Therefore, conventional emissometers operate by illuminating the object with light from a source and collecting the reflected light from the surface of the object. The reflected light is detected and then reported.

[0020] Conventional systems and methods for detecting the emissivity of an object have many drawbacks. First, conventional systems have little control over the size of the area illuminated by the light source (i.e., the "spot size"). Therefore, conventional systems and methods cannot effectively analyze small surface areas or geometric shapes.

[0021] Second, and relatedly, conventional systems utilize relatively weak radiation from an omnidirectional radiation source (e.g., via an aperture) to illuminate the surface of an object. The omnidirectional radiation of conventional systems results in limited collection of reflected light (e.g., reflected radiation). As a result, conventional systems are inherently sensitive to noise and cannot provide the accuracy required to characterize small object geometries (e.g., less than 1,000 microns). To improve accuracy, conventional systems can slow down the measurement process and utilize certain techniques to improve the signal-to-noise ratio. To improve the signal-to-noise ratio, some conventional systems use larger apertures to deliver radiation to the object, but as discussed above, this can result in system inaccuracies and larger spot sizes.

[0022] Aspects and embodiments of the present disclosure address the above-mentioned and other disadvantages of conventional systems by providing a system (e.g., an optical measurement tool) for detecting the emissivity and / or surface roughness of an object. In some embodiments, the system includes a radiation source, such as a supercontinuum laser operating in the mid-infrared range, which emits a radiation beam (e.g., a laser beam). The radiation beam can be directed to the surface of the object through one or more mirrors and / or lenses. In some embodiments, the lens focuses the radiation beam to a "point" on the surface of the object. The surface of the object reflects and / or scatters a portion of the radiation beam. In some embodiments, the reflected portion having a first intensity is reflected back to the system and detected by an optical detector of the system. In some embodiments, the scattered portion having a second intensity is collected by the system (e.g., collected by a reflective objective lens (e.g., a Schwarzschild objective lens)) and detected by another optical detector of the system. A processing device (e.g., a computing device, etc.) determines the surface roughness and / or emissivity of the object based on comparing the intensity of the reflected radiation (e.g., the first intensity) with the intensity of the scattered radiation (e.g., the second intensity).

[0023] Compared with the conventional systems described above, embodiments of the present disclosure are more advantageous. In particular, some embodiments described herein detect emissivity with higher accuracy by providing a radiation source that emits a radiation beam, rather than the omnidirectional radiation source of the conventional system. The radiation beam is stronger (e.g., has a greater intensity) and is more focused, so the intensity of reflected radiation and / or scattered radiation from the surface of the object is greater. This greater intensity reduces the sensitivity of the system to signal noise, thereby achieving higher accuracy. In addition, some embodiments described herein can also simultaneously detect and characterize both the emissivity and surface roughness of the object under test. By using two optical detectors, both reflected radiation (e.g., "bright field") and scattered radiation (e.g., "dark field") can be measured to provide data for characterizing the emissivity and surface roughness of the object. This data can be used (e.g., via the machine learning techniques described below) to predict substrate process results for substrates to be processed in a processing chamber using a chamber component under test (e.g., an object under test). In addition, the radiation beam used in the embodiments described herein allows for faster measurement of emissivity compared to conventional systems.

[0024] Figure 1A A simplified side view of a system 100A for optically determining the emissivity and / or surface roughness of an object is shown, in accordance with aspects of the present disclosure. In some embodiments, the system 100A is an optical measurement tool (eg, an emissivity meter).

[0025] System 100A includes a radiation source 102 configured to emit a radiation beam 103, which may be a focused radiation beam. In an embodiment, radiation source 102 is a laser, such as a semiconductor laser (e.g., a laser using a laser diode). Other types of layers that may be used include gas lasers, solid-state lasers, fiber lasers, and liquid lasers. In some embodiments, radiation source 102 is a supercontinuum laser. In optics, a supercontinuum is formed when a series of nonlinear processes act together on a pump beam, causing the spectrum of the original pump beam to be severely broadened. The result is a continuous spectrum. In some embodiments, radiation source 102 is a supercontinuum laser configured to operate in the mid-infrared range (e.g., radiation source 102 is a mid-infrared supercontinuum laser). In some embodiments, the wavelength of electromagnetic radiation emitted by radiation source 102 is in the range of 1-6 microns. In further embodiments, the wavelength of radiation emitted by radiation source 102 is in the range of 3-5 microns. In some embodiments, radiation beam 103 is a collimated beam (e.g., radiation source 102 is configured to emit a collimated beam). In some embodiments, the diameter of radiation beam 103 is between about 1 mm and about 10 mm. In some embodiments, the diameter of radiation beam 103 is about 5 mm.

[0026] In some embodiments, the radiation beam 103 is directed through a polarizing filter 104 (also referred to as a polarizer). The polarizing filter 104 can be disposed between the radiation source 102 and the beam splitter 106 along the optical axis of the radiation beam 103. In some embodiments, the polarizing filter 104 is configured to polarize the radiation beam 103 emitted from the radiation source 102. In some embodiments, the polarizing filter 104 linearly polarizes the radiation beam 103. In some embodiments, the polarizing filter 104 is omitted.

[0027] Typically, beam splitters such as beam splitter 106 are polarization dependent, meaning that the ratio of reflected radiation to transmitted radiation is a function of the polarization and wavelength of the incident radiation. Although the radiation beam 103 emitted by the radiation source 102 may be substantially unpolarized, there may be some residual and varying polarization preference between the horizontal and vertical directions. Under such conditions, the amount of radiation transmitted by the beam splitter 106 and / or the polarization of the radiation may appear slightly modulated. Such modulation may introduce errors in the normalization process associated with the optical sensor 108 described herein. Therefore, by including a polarization filter 104 in some embodiments, any shift in the instantaneous polarization of the radiation beam 103 will be converted into an amplitude fluctuation that will affect the radiation transmitted by the beam splitter 106 and the radiation reflected by the beam splitter 106 in the same manner (e.g., during the period when the polarization of the radiation beam 103 changes, the amplitude of both the transmitted radiation and the reflected radiation will increase or decrease). More functionality of the beam splitter will be described below.

[0028] In some embodiments, radiation beam 103 passes through beam splitter 106 (optionally after passing through polarizing filter 103). In some embodiments, a one-way mirror is used instead of a beam splitter.

[0029] In some embodiments, all or substantially all of radiation beam 103 passes through beam splitter 106. Alternatively, a portion of the radiation beam may be reflected by beam splitter 106 and directed toward optical sensor 108, while another portion of the radiation beam is transmitted by the beam splitter (e.g., toward lens 110). In some embodiments, a majority of the intensity of radiation beam 103 is transmitted through beam splitter 106, while a small portion (e.g., 2-10%) of the intensity of radiation beam 103 is reflected toward optical sensor 108. In some embodiments, radiation beams 103 of substantially equal intensity are transmitted through beam splitter 106 and reflected (e.g., by beam splitter 106) toward optical sensor 108.

[0030] The optical sensor 108 can be configured to detect the intensity of the portion of the radiation beam reflected by the beam splitter 106. The optical sensor 108 (as well as the optical sensors 116, 130) can be or include a sensor having one or more (e.g., a matrix) sensing components. In some embodiments, the sensing component is a charge coupled device (CCD) sensor. In some embodiments, the sensing component is a complementary metal oxide semiconductor (CMOS) type image sensor. In some embodiments, the sensing component is a mercury cadmium telluride (HgCdTe) photoconductive detector. Other types of image sensors known to those skilled in the art can also be used for the optical sensors 108, 116, 130. In some embodiments, the optical sensors 108, 116 and / or 130 include a current meter or are each coupled to a current meter to measure the current induced by receiving radiation.

[0031] As described below, the intensity detected by optical sensor 108 can be used to normalize the radiation intensity detected by optical sensor 116 and / or optical sensor 130. For example, changes in the radiation intensity detected by optical sensor 108 can be used to attenuate changes in the radiation intensity output by radiation source 102 and detected by optical sensor 116 and / or optical sensor 130. Specifically, the radiation intensity detected by optical sensor 108 can be used as a relative reference for optical sensor 116 and optical sensor 130 because, in some embodiments, the radiation intensity detected by optical sensor 108 is directly related to the intensity of radiation beam 103. In some examples, power fluctuations of radiation beam 103 output by source 102 can be detected by optical sensor 108. In some embodiments, the signal output by optical sensor 108 is used to stabilize system 100. In embodiments, variations in the measurement system (e.g., variations between measurements of the same radiation intensity) can be reduced based on the signal output by optical sensor 108. In some embodiments, the signal output by optical sensor 108 can reduce variations in measured values ​​(such as emissivity and / or surface roughness) to less than 0.1%. Therefore, the use of the beam splitter 106 and the optical sensor 108 can improve the stability of the system 100, so that the variation in the embodiment is less than 0.1%. In other embodiments, the variation can be less than 0.2%, less than 0.3%, less than 0.4%, less than 0.5%, less than 0.6%, less than 0.6%, less than 0.7%, less than 0.8%, less than 0.9% or less than 1.0%.

[0032] The use of the polarizing filter 104 described above further improves the stability of the system 100. In particular, the polarization of the radiation output by the radiation source 102 may fluctuate slightly. The amount of the radiation beam 103 that passes through the beam splitter 106 and the amount of the radiation beam 103 that is reflected by the beam splitter 106 may have a certain dependence on the polarization. Therefore, slight fluctuations in polarization may be detected as changes in intensity detected by one or more of the optical sensors 108, 116, 130, thereby causing system instability. However, by introducing the polarizing filter 104, any fluctuations in the polarization of the radiation beam 103 are removed, thereby improving the stability of the system measurement (reducing measurement variation).

[0033] In some embodiments, beam splitter 106 transmits the radiation beam (e.g., a portion of the radiation beam, a majority of the radiation beam, all of the radiation beam except for the portion reflected toward optical sensor 108, etc.) toward one or more lenses 110, which can be located on the optical axis of the system. In some embodiments, lens 110 is a doublet lens. In some embodiments, lens 110 is an objective lens. Lens 110 can be configured to focus the radiation beam to enhance the radiation beam and / or reduce the diameter of the radiation beam. The focal length of lens 110 can be about 50 mm to about 100 mm. In some embodiments, the focal length of lens 110 can be about 75 mm. In some embodiments, lens 110 can focus the radiation beam on the surface of object 114 to a spot size of less than about 200 microns in diameter. In some embodiments, lens 110 focuses radiation beam 103 to a spot size of less than 300 microns. In some embodiments, lens 110 can focus radiation beam 103 to a spot size of less than 500 microns. In some embodiments, the spot size is about 50 microns to 90 microns. The spot size can be a function of the focal length of the lens 110, the wavelength of the radiation beam 103, and the initial width of the radiation beam 103. In some embodiments, the spot size can be variable. For example, in some embodiments, the lens 110 is attached to an actuator or other translation mechanism that can move the position of the lens 110 along the optical axis of the system 100. This movement of the position of the lens 110 can change the focus setting of the optical system. In some embodiments, an actuator coupled to the lens 110 can move the lens 110 along the optical axis to change the spot size on the surface of the object under test 114.

[0034] In some embodiments, the object 114 is located on a support 135. The support 135 can be a movable platform. In some embodiments, the support 135 can move around one or more axes (e.g., one axis, two axes, three axes, etc.). For example, the support 135 can move in an XY plane that is orthogonal (e.g., substantially orthogonal) to the direction of the incident radiation beam. In some embodiments, the support 135 can rotate around one or more axes. In some embodiments, the support 135 can have six or fewer degrees of freedom.

[0035] In some embodiments, the radiation beam is transmitted through lens 110 toward angled mirror 112, which reflects the focused radiation beam onto a surface of object 114. In some embodiments, the surface of object 114 is at least partially emissive and may have a surface roughness. In some examples, the surface of object 114 may reflect radiation and / or scatter radiation. The amount of reflected radiation and / or scattered radiation may depend on one or more properties of the surface of the object, such as roughness, reflectivity, absorptivity, refractive index, etc. The reflected radiation and / or scattered radiation may be measured (e.g., via system 100). In some embodiments, object 114 is a chamber component of a substrate processing chamber, such as Figure 2 A component of the processing chamber 200 of the apparatus. In some embodiments, the surface of the object 114 to be measured can be substantially perpendicular to the incident radiation beam. Therefore, if the object has an uneven surface, the orientation of the object relative to the system 100 can be changed when measuring different parts of the object so that the normal of a point on the surface of the object to be measured is aligned with the ray of the radiation beam 103.

[0036] In some embodiments, the mirror 112 is coupled to the bottom surface of the convex mirror 124 of the reflective objective 120 (e.g., as shown). The position and / or size of the mirror 112 can be such that scattered radiation (e.g., scattered radiation from the surface of the object 114) is not blocked by the mirror 112. A first portion of the radiation beam can be reflected back toward the mirror 112 by the surface of the object 114. The first portion of the radiation beam can be referred to as a reflected radiation beam. The reflected radiation beam can then be reflected from the mirror 112, reflected back through the lens 110, and reflected from the beam splitter 106 toward the optical sensor 116.

[0037] In some embodiments, beam splitter 106 reflects the reflected portion of the radiation beam to optical sensor 116. Optical sensor 116 may be configured to detect the intensity of the portion of the radiation beam reflected by the surface of object 114 (i.e., the reflected radiation beam). In some embodiments, the radiation intensity detected by optical sensor 116 (e.g., the intensity of the reflected radiation beam) is related to the emissivity and / or surface roughness of the surface of object 114.

[0038] In some embodiments, radiation scattered from the surface of the object 114 (e.g., represented by the dashed arrows in FIG. 1 and referred to as a scattered radiation beam) is collected by a reflective objective 120, which is referred to as a light collector. The reflective objective 120 may include a concave mirrored inner surface 122 and a convex mirror 124 disposed below or near the inner surface 122. The scattered radiation beam may be collected by the concave mirrored inner surface 122 and reflected toward the convex mirror 124. In some embodiments, the convex mirror 124 forms a central shielded area of ​​the reflective objective 120. In some embodiments, the convex mirror 124 reflects the radiation collected from the scattered radiation beam through a hole 126 in the concave mirrored inner surface 122 toward an optical sensor 130. In some embodiments, the reflective objective 120 is a Schwarzschild objective. However, one skilled in the art will recognize that other reflective objective lenses may also be used. In some embodiments, the optical sensor 130 is configured to detect the intensity of the radiation scattered from the surface of the object 114 (e.g., the intensity of the scattered radiation beam). In some embodiments, the intensity of radiation detected by optical sensor 130 (eg, the intensity of radiation scattered by the surface of object 114 ) is related to the emissivity and / or surface roughness of the surface of object 114 .

[0039] In some embodiments, the system controller 160 (e.g., a computing device, a processing device, etc.) can be communicatively coupled to the optical sensor 108, the optical sensor 116, and / or the optical sensor 130. The system controller 160 can be and / or can include a computing device, such as a personal computer, a server computer, a programmable logic controller (PLC), a microcontroller, a system-on-chip (SoC), etc. The system controller 132 can include one or more processing devices, which can be general-purpose processing devices, such as a microprocessor, a central processing unit, or the like. In more detail, the processing device can be a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, or a processor that implements other instruction sets or a processor that implements a combination of instruction sets. The processing device can also be one or more special-purpose processing devices, such as an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), a network processor, or the like. The system controller 132 can include a data storage device (e.g., one or more disk drives and / or solid-state drives), a main memory, a static memory, a network interface, and / or other components. The system controller 132 may execute instructions to perform any one or more of the methods and / or embodiments described herein. The instructions may be stored on a computer-readable storage medium, which may include a main memory, a static memory, an auxiliary memory, and / or a processing device (during execution of the instructions). The system controller 132 may also be configured to allow a human operator to input and display data, operating commands, and the like.

[0040] The system controller 160 may receive output signals from each optical sensor. In some embodiments, the system controller 160 may determine (e.g., via processing logic) the roughness and / or emissivity of the surface of the object 114 based on a comparison of the radiation intensity detected by the optical sensor 116 and the radiation intensity detected by the optical sensor 130. In some embodiments, the emissivity is equivalent to 1 minus the reflectivity of the surface of the object 114 (e.g., 1-reflectivity). In some embodiments, the intensity of the reflected radiation (e.g., the intensity detected by the optical sensor 116) is related to the emissivity. For example, the emissivity may be considered complementary to the reflectivity. The reflectivity may be calculated by the ratio of the intensity of the reflected radiation beam (the intensity being represented, for example, by the radiation intensity detected by the optical sensor 116) to the radiation intensity detected by the optical sensor 108. By the relationship of emissivity=1-reflectivity, the reflectivity may indicate the emissivity of the surface of the object 114. In some embodiments, the ratio of the scattered radiation intensity to the reflected radiation intensity indicates the surface roughness.

[0041] In some examples, a higher reflected radiation intensity (e.g., reflected radiation intensity detected by optical sensor 116) compared to a scattered radiation intensity (e.g., scattered radiation intensity detected by optical sensor 130) may indicate a lower emissivity and / or a lower surface roughness of the surface of object 114. In some examples, a lower reflected radiation intensity may indicate a higher emissivity and / or a higher surface roughness compared to the scattered radiation intensity. In some examples, a higher scattered radiation intensity (e.g., scattered radiation intensity detected by optical sensor 130) may indicate a higher surface roughness compared to the reflected radiation intensity, while a lower scattered radiation intensity may indicate a lower surface roughness compared to the reflected radiation intensity.

[0042] In some embodiments, system controller 160 may determine that the surface roughness of object 114 is related to the ratio of scattered radiation intensity to reflected radiation intensity. Thus, processing device may determine the surface roughness of object 114 based on the ratio of radiation intensity detected by optical sensor 130 to radiation intensity detected by optical sensor 116.

[0043] In some embodiments, as described above, the system controller 160 may further determine the surface roughness and / or emissivity based on the sensor data from the optical sensor 108. Specifically, the system controller 160 may determine a normalization coefficient based on the sensor data from the optical sensor 108. The normalization coefficient may be used to normalize the sensor data from the optical sensor 116 and / or the optical sensor 130. For example, a change in the amplitude of the radiation beam 103 may cause a corresponding change in the reflected radiation detected by the optical sensor 116 and / or the scattered radiation detected by the optical sensor 130. These changes may cause changes in the surface roughness and / or emissivity calculated by the system controller 160. However, the optical sensor 108 may also detect changes in the amplitude of the radiation beam 103. By determining the normalization coefficient based on the sensor data from the optical sensor 108 (e.g., where the sensor data corresponds to a change in the amplitude of the radiation beam 103), changes in the sensor data from the optical sensors 116 and 130 may be normalized (e.g., based on the normalization coefficient). The system controller 160 may use the normalized sensor data to determine the surface roughness and / or emissivity. In some embodiments, the normalization factor is proportional to the product of the intensity detected by optical sensor 108 and the target intensity. The output signal of optical sensor 116 and / or optical sensor 130 can be multiplied by the normalization factor to determine a corrected signal. In some embodiments, the normalization factor can account for frequency and / or phase mismatches between sensors, nonlinearity of sensor measurements, and / or other non-idealities of system 100A.

[0044] In some embodiments, the system 100A can generate a surface roughness map and / or emissivity map of the object 114. The surface roughness map and / or emissivity map can be generated by moving the object 114 relative to the incident radiation beam (e.g., via the movable support 135) and determining the surface roughness and / or emissivity at various discrete points on the surface of the object. In some embodiments, the generated surface roughness map and / or emissivity map can be based on the measurement results of the surface roughness and / or emissivity at various known points on the surface of the entire object 114. This map can be used to determine various predicted elements described below.

[0045] In some embodiments, system 100 includes a camera instead of or in addition to one or more components of system 100A. In some examples, a camera operating in the mid-infrared range can image the surface of object 114 to determine emissivity and / or roughness information of the surface of object 114.

[0046] Figure 1BA simplified side view of a system 100B for optically determining emissivity and / or surface roughness according to aspects of the present disclosure is shown. Features of system 200 that are similarly numbered to features of system 100 may have similar structures and / or functions as described above. In some embodiments, system 100B includes a rotatable mirror 152 configured to direct a radiation beam toward lens 110. In some examples, the optical axis of radiation source 102 may be set to be approximately at a right angle (e.g., 90 degrees) to the optical axis of lens 110. In some embodiments, rotatable mirror 152 is substantially disposed at the focal plane of lens 110. In some embodiments, rotatable mirror 152 may reflect a radiation beam from the radiation source toward lens 110 at an angle. In response to rotation of rotatable mirror 152 about an axis perpendicular to the optical axis of lens 110 (e.g., an axis in the up-down direction of a drawing page or an axis in and out of a drawing page), rotatable mirror 152 may move the radiation beam across the surface of object 114 in a periodic motion. For example, lens 110 can transform the angular motion of rotatable mirror 152 into lateral motion (e.g., periodic lateral motion) of the radiation beam. In some examples, rotation of rotatable mirror 152 can cause the radiation beam to periodically move back and forth (e.g., in a periodic motion) across the surface of object 114. The back and forth motion of the radiation beam can allow the surface of object 114 to be scanned.

[0047] In some embodiments, while the object 114 is slowly moved in the Y direction of the XY plane by the support 135, the radiation beam is quickly moved back and forth in the X direction (e.g., the X direction of the XY plane) to scan the surface of the object 114. The data collected during the scan (e.g., reflected radiation intensity and / or scattered radiation intensity) can be used to determine (e.g., by the system controller 160) an emissivity surface profile map and / or a surface roughness profile map (e.g., one or more profile maps) of the object 114.

[0048] Figure 2 is a cross-sectional view of a processing chamber 200 (eg, a semiconductor processing chamber, a display processing chamber, etc.) having a plurality of processing chambers that have been used according to an embodiment of the present disclosure. Figure 1A System 100A or Figure 1BOne or more chamber components of the system 100B can be characterized. For example, the processing chamber 200 can be used for a process that provides a corrosive plasma environment having plasma processing conditions. For example, the processing chamber 200 can be a chamber for a plasma etcher or plasma etching reactor, a plasma cleaner, and the like. Other types of chambers can include deposition chambers, cleaning chambers, oxidation chambers, and the like. Examples of chamber components whose surface roughness and / or emissivity can be characterized include a substrate support assembly 248, an electrostatic chuck (ESC), a ring (e.g., a process kit ring or a single ring), a chamber wall, a base, a gas distribution plate, a showerhead 230, a gas line, a nozzle, a lid, a gasket, a gasket kit, a shield, a plasma screen, a flow equalizer, a cooling base, a chamber viewport, a chamber lid, and the like. The chamber components can be composed of metals, metal alloys, ceramics, and any combination thereof. The chamber components can include coatings, such as plasma resistant coatings or corrosion resistant coatings, and their surfaces can be characterized using Figure 1A-1B The coating can be deposited or grown by atomic layer deposition, plasma spraying, chemical vapor deposition, ion-assisted deposition, sputtering, physical vapor deposition, electroplating, anodization, etc.

[0049] In one embodiment, the processing chamber 200 includes a chamber body 202 and a showerhead 230, which enclose an internal volume 206. The showerhead 230 can include a showerhead base and a showerhead gas distribution plate. Alternatively, in some embodiments, the showerhead 230 can be replaced by a cover and a nozzle. The chamber body 202 can be made of aluminum, stainless steel, or other suitable materials. The chamber body 202 generally includes a sidewall 208 and a bottom 210. Any of the showerhead 230 (or cover and / or nozzle), the sidewall 208, and / or the bottom 210 can include the characterized coating.

[0050] An outer liner 216 can be disposed adjacent the sidewall 208 to protect the chamber body 202. The outer liner 216 can be characterized as follows. In one embodiment, the outer liner 216 is made of aluminum oxide.

[0051] An exhaust port 226 may be defined in the chamber body 202 and may couple the interior volume 206 to a pump system 228. The pump system 228 may include one or more pumps and throttle valves used to evacuate and regulate the pressure of the interior volume 206 of the processing chamber 200.

[0052] The showerhead 230 can be supported on the sidewalls 208 and / or the top of the chamber body 202. In some embodiments, the showerhead 230 (or the lid) can be opened to allow access to the interior volume 206 of the processing chamber 200, and can provide a seal for the processing chamber 200 when closed. A gas panel 258 can be coupled to the processing chamber 200 to provide process gases and / or cleaning gases to the interior volume 206 through the showerhead 230 or the lid and nozzle. The showerhead 230 is used to process the chamber for dielectric etching (etching dielectric materials). The showerhead 230 can include a gas distribution plate (GDP) with a plurality of gas delivery holes 232 throughout the GDP. The showerhead 230 can include a GDP bonded to an aluminum showerhead base or an anodized aluminum showerhead base. The GDP 233 can be made of Si or SiC, or a ceramic such as Y2O3, Al2O3, YAG, etc. In embodiments, the showerhead 230 and the delivery holes 232 can be characterized using the system 100 or 150. For a processing chamber used for conductor etching (etching conductive materials), a lid can be used instead of a showerhead. The lid can include a central nozzle that fits into a central hole in the lid. The lid can be a ceramic such as Al2O3, Y2O3, YAG, or a ceramic compound including a solid solution of Y2O3-ZrO2 and Y4Al2O9. The nozzle can also be a ceramic such as Y2O3, YAG, or a ceramic compound including a solid solution of Y2O3-ZrO2 and Y4Al2O9. According to one embodiment, the lid, showerhead 230 (which includes, for example, a showerhead base, a GDP, and / or a gas delivery conduit / hole), and / or the nozzle can be characterized using system 100 or 150.

[0053] A substrate support assembly 248 is disposed in the interior volume 206 of the processing chamber 200 beneath the showerhead 230 or lid. The substrate support assembly 248 holds the substrate 244 during processing and may include an electrostatic chuck bonded to a cooling plate.

[0054] An inner liner may be located at the periphery of the substrate support assembly 248. The inner liner may be a halogen-containing gas resistant material, such as discussed with reference to the outer liner 216. In one embodiment, the inner liner 218 may be made of the same material as the outer liner 216. Additionally, in an embodiment, the inner liner 218 may also be characterized using the system 100 or 150.

[0055] Figure 3An illustrative computer system architecture 300 according to aspects of the present disclosure is depicted. The computer system architecture 300 includes a client device 320, manufacturing equipment 322, an optical metrology tool 326, a prediction server 312 (which is used, for example, to generate prediction data, provide model adaptation, use a knowledge base, etc.), and a data store 350. The prediction server 312 can be part of a prediction system 310. The prediction system 310 can further include server machines 370 and 380. In some embodiments, the computer system architecture 300 can include a manufacturing system for processing substrates, or an optical metrology tool 326, or can be part of the manufacturing system or the optical metrology tool. Additional details about the optical metrology tool 326 will be described in detail below. Figure 1A-1B supply.

[0056] Components of client device 320, manufacturing equipment 322, optical measurement tool 326, prediction system 310, and / or data storage 350 may be coupled to each other via network 340. In some embodiments, network 340 is a public network that provides client device 320 with access to prediction server 312, data storage 350, and other publicly available computing devices. In some embodiments, network 340 is a private network that provides client device 320 with access to manufacturing equipment 322, optical measurement tool 326, data storage 350, and / or other privately available computing devices. Network 340 may include 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.

[0057] Client devices 320 may include computing devices such as personal computers (PCs), laptops, mobile phones, smartphones, tablet computers, pocket computers, network-connected televisions (“smart TVs”), network-connected media players (e.g., Blu-ray players), set-top boxes, over-the-top (OTT) streaming devices, operator boxes, etc.

[0058] The manufacturing equipment 322 may produce products according to the recipe. In some embodiments, the manufacturing equipment 322 may include or may be part of a manufacturing system including one or more stations (e.g., process chambers, transfer chambers, load locks, factory interfaces, etc.) configured to perform different operations on substrates.

[0059] The optical measurement tool 326 may be a tool (e.g., a system) for determining the emissivity and / or roughness of the surface of the object under test. The optical measurement tool 326 may be configured to generate data associated with the emissivity and / or surface roughness of the object measured by the optical measurement tool 326. In some embodiments, the optical measurement tool corresponds to the system 100A or the system 100B. In some embodiments, such data (e.g., emissivity data, surface roughness data, etc.) may be stored in a data storage 350, where the data may be accessed (e.g., via a network 340). The optical measurement tool 326 may include one or more sensors (e.g., multiple optical sensors) configured to detect radiation and generate data associated with the object under test. In some embodiments, the optical measurement tool 326 includes a radiation source to provide a radiation beam, which is used to irradiate the surface of the object under test (e.g., a chamber component of a substrate processing chamber of the manufacturing equipment 322, etc.). The optical sensor of the optical measurement tool 326 may detect radiation reflected and / or scattered by the surface of the object. In some embodiments, the optical measurement tool 326 can generate emissivity data and / or surface roughness data based on the intensity of the reflected radiation and / or scattered radiation detected by the optical sensor. In some embodiments, the optical measurement tool 326 can generate a surface roughness profile and / or an emissivity profile of the surface of the object under test by measuring the emissivity and / or surface roughness at multiple locations on the surface of the object under test. In some embodiments, the optical measurement tool 326 can be included in a system for manufacturing components (such as processing chamber components) of the manufacturing equipment 322.

[0060] The data storage 350 may be a 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. The data storage 350 may include multiple storage components (e.g., multiple drives or multiple databases) that may span multiple computing devices (e.g., multiple server computers). The data storage 350 may store emissivity data and surface roughness data (e.g., generated by the optical measurement tool 326).

[0061] One or more portions of the data storage 350 may be configured to store data that is inaccessible to users of the manufacturing system. In some embodiments, all data stored at the data storage 350 may be inaccessible to users of the manufacturing system. In other or similar embodiments, a portion of the data stored at the data storage 350 is inaccessible to the user, while another portion of the data stored at the data storage 350 is accessible to the user. In some embodiments, the inaccessible data stored at the data storage 350 is encrypted using an encryption mechanism unknown to the user (e.g., the data is encrypted using a private encryption key). In other or similar embodiments, the data storage 350 may include multiple data storages, wherein data that is inaccessible to the user is stored in a first data storage, and data that is accessible to the user is stored in a second data storage.

[0062] In some embodiments, prediction system 310 includes server machine 370 and server machine 380. Server machine 370 includes training set generator 372, which is capable of generating a training data set (e.g., a set of data inputs and a set of target outputs) to train, validate, and / or test machine learning model 390 or a set of machine learning models 390. Figure 4 and 5A Some operations of the training set generator 372 are described in detail. In some embodiments, the training set generator 372 can divide the training data into a training set, a validation set, and a test set.

[0063] The server machine 380 may include a training engine 382. An engine may refer to hardware (e.g., circuitry, dedicated logic, programmable logic, microcode, processing device, etc.), software (e.g., instructions running on a processing device, a general-purpose computer system, or a dedicated machine), firmware, microcode, or a combination thereof. The training engine 382 may be capable of training a machine learning model 390 or a set of machine learning models 390. The machine learning model 390 may refer to a model artifact created by the training engine 382 using training data. The training data may include training inputs and corresponding target outputs (correct answers for the corresponding training inputs). The training engine 382 may find patterns in the training data that map the training inputs to the target outputs (the answers to be predicted). The training engine 382 may then ultimately provide a machine learning model 390 that captures these patterns. The machine learning model 390 may include a linear regression model, a partial least squares regression model, a Gaussian regression model, a random forest model, a support vector machine model, a neural network, a ridge regression model, etc. In some embodiments, instead of or in addition to the machine learning model, the machine learning model 390 is a physics-based model.

[0064] The training engine 382 can also validate the trained machine learning model 390 using a corresponding set of features from the validation set of the training set generator 372. In some embodiments, the training engine 382 can assign a performance rating to each trained machine learning model in a set of trained machine learning models 390. The performance rating can correspond to the accuracy of the corresponding trained model, the speed of the corresponding model, and / or the efficiency of the corresponding model. According to some embodiments described herein, the training engine 382 can select a trained machine learning model 390 whose performance rating meets the performance criteria to be used by the prediction engine 314. Figure 5A Additional details regarding training engine 382 are provided.

[0065] The prediction server 312 includes a prediction engine 314 that can provide data (e.g., emissivity data and / or surface roughness data) from the optical metrology tool 326 as input to a trained machine learning model 390. The prediction engine can execute the trained model 390 on the input to obtain one or more outputs. In an embodiment, the trained model 390 is trained based on training data that includes surface profiles of roughness and / or emissivity of chamber components and one or more quality metrics of one or more processed substrates. Figure 5B As further described, in some embodiments, the prediction engine 314 processes input data (e.g., surface profiles of roughness and / or emissivity of chamber components) using the model 390 to predict substrate process results (e.g., one or more substrate quality metrics) for future substrates to be processed in the processing chamber using the chamber components measured by the optical metrology tool 326.

[0066] It should be noted that in some other embodiments, the functionality of server machines 370 and 380 and prediction server 312 may be provided by a greater or lesser number of machines. For example, in some embodiments, server machines 370 and 380 may be integrated into a single machine. In other embodiments, server machines 370 and 380 and / or prediction server 312 may be integrated into a single machine. In general, the functionality described as being performed by server machine 370, server machine 380, and / or prediction server 312 in one embodiment may also be performed on client device 320. In addition, functionality attributed to a particular component may also be performed by a different component or multiple components operating together.

[0067] Figure 4A model training workflow 405 and a model application workflow 417 for determining predicted processed substrate outcomes based on a surface profile map of one or more chamber components are shown according to one embodiment. The model training workflow 405 and the model application workflow 417 may be performed by processing logic executed by a processor of a computing device. One or more of these workflows 405, 417 may be implemented by, for example, one or more machine learning models implemented on a processing device and / or other software and / or firmware executed on the processing device.

[0068] The model training workflow 405 is used to train one or more machine learning models (e.g., deep learning models) to determine predicted substrate results for a substrate processed in a process chamber, the process chamber including one or more chamber components having a measured emissivity surface profile and / or roughness surface profile. The model application workflow 417 is used to apply the one or more trained machine learning models to perform substrate result evaluation. Each component emissivity / roughness data 412 may include surface emissivity and / or roughness at multiple locations of a chamber component of the processing chamber. For example, each component emissivity / roughness data 412 may include an array of surface emissivity measurements and / or surface roughness measurements of the corresponding chamber component. In some embodiments, the component emissivity / roughness data 412 includes one or more emissivity maps and / or roughness maps (e.g., profile maps) of an object surface (e.g., a chamber component surface). In some embodiments, the emissivity map and / or roughness map may be generated via the system 100A or 100B described above.

[0069] Various machine learning outputs are described herein. A specific number and arrangement of machine learning models are described and illustrated. However, it should be understood that the number and type of machine learning models used and the arrangement of such machine learning models can be modified to achieve the same or similar end results. Therefore, the arrangement of machine learning models described and illustrated is only an example and should not be construed as limiting.

[0070] In some embodiments, one or more machine learning models are trained to perform one or more substrate result estimation tasks. Each task can be performed by a separate machine learning model. Alternatively, a single machine learning model can perform each task or a subset of tasks. For example, a first machine learning model can be trained to determine a substrate process result, and a second machine learning model can be trained to determine a corresponding corrective action. Additionally or alternatively, different machine learning models can be trained to perform different combinations of tasks. In one example, one or several machine learning models can be trained. The trained machine learning (ML) model can be a single shared neural network having multiple shared layers and multiple higher-level distinct output layers, where each output layer outputs a different prediction, classification, recognition, etc. For example, a first higher-level output layer can determine a substrate process result based on input data corresponding to a first chamber component, and a second higher-level output layer can determine a substrate process result based on input data corresponding to a second chamber component.

[0071] One type of machine learning model that can be used to perform some or all of the tasks listed above is an artificial neural network, such as a deep neural network. An artificial neural network typically includes a feature representation component with a classifier or regression layer that maps features to a target output space. For example, a convolutional neural network (CNN) contains multiple layers of convolutional filters. Deep learning is a class of machine learning algorithms that uses a cascade of multiple layers of nonlinear processing units for feature extraction and transformation. Each successive layer uses the output from the previous layer as input. Deep neural networks can learn in a supervised (e.g., classification) and / or unsupervised (e.g., pattern analysis) manner. A deep neural network includes a hierarchy of layers, with different layers learning different levels of representation corresponding to different levels of abstraction. In deep learning, each layer learns to transform its input data into a slightly more abstract and comprehensive representation. Notably, the deep learning process can learn on its own which features are best placed at which layer. The "depth" in "deep learning" refers to the number of layers through which the data is transformed. More specifically, deep learning systems have a considerable credit assignment path (CAP) depth. A CAP is a chain of transformations from an input to an output. A CAP describes the underlying cause-effect relationship between an input and an output. For feedforward neural networks, the depth of CAP can be the depth of the network and can be the number of hidden layers plus 1. For recurrent neural networks where signals can propagate through a layer more than once, the CAP depth can be infinite.

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

[0073] For the model training workflow 405, a training dataset containing hundreds, thousands, tens of thousands, hundreds of thousands, or more instances of component emissivity / roughness data 412 (e.g., surface emissivity / roughness maps) should be used to form a training dataset. For example, the data may include chamber component emissivity measurements determined using a given number of measurements. In some embodiments, multiple measurements are performed to generate a surface emissivity map of the chamber component surface. This data may be processed to generate one or more training datasets 436, which are used to train one or more machine learning models. The training data items in the training dataset 436 may include component emissivity / roughness data 412, substrate results of substrates processed in a processing chamber using the measured chamber components, and / or one or more images of the processed substrate.

[0074] To implement training, processing logic inputs the training data set 436 into one or more untrained machine learning models. Before the first input is input into the machine learning model, the machine learning model can be initialized. Processing logic trains the untrained machine learning model based on the training data set to generate one or more trained machine learning models that perform the various operations described above. Training can be performed by inputting input data such as component emissivity / roughness data 412, images and / or results of processed substrates into the machine learning model one by one.

[0075] Machine learning models process inputs to generate outputs. An artificial neural network includes an input layer, which consists of the values ​​in the data points. The next layer is called a hidden layer, and the nodes at the hidden layer each receive one or more input values. Each node contains parameters (such as weights) that are applied to the input values. Therefore, each node essentially inputs the input values ​​into a multivariate function (for example, a nonlinear mathematical transformation) to produce an output value. The next layer may be another hidden layer, or an output layer. In both cases, the nodes at the next layer receive output values ​​from the nodes at the previous layer, each node applies weights to those values, and then generates its own output value. This can be performed at each layer. The last layer is the output layer, where there is a node for each category, prediction, and / or output that the machine learning model can produce.

[0076] Thus, the output may include one or more predictions or inferences (e.g., an estimate of the outcome of a processed substrate for a substrate processed in a process chamber in which the measured substrate was processed using a particular chamber component). The processing logic may compare the output estimated substrate outcome with the historical substrate outcome. The processing logic determines an error (i.e., a classification error) based on the difference between the estimated substrate outcome and the target substrate outcome. The processing logic adjusts the weights of one or more nodes in the machine learning model based on the error. An error term or delta may be determined for each node in the artificial neural network. Based on this error, the artificial neural network adjusts one or more parameters of one or more of its nodes (the weights of one or more inputs of the node). The parameters may be updated in a back-propagation manner such that the nodes at the highest layer are updated first, then the nodes at the next layer, and so on. The artificial neural network includes multiple layers of "neurons", where each layer receives as input values ​​from neurons at the previous layer. The parameters of each neuron include weights associated with the values ​​received from each neuron at the previous layer. Thus, adjusting the parameters may include adjusting the weights assigned to each input of one or more neurons of one or more layers in the artificial neural network.

[0077] Once the model parameters are optimized, model validation can be performed to determine whether the model has improved and to determine the current accuracy of the deep learning model. After one or more rounds of training, the processing logic can determine whether the stopping criteria have been met. The stopping criteria can be a target level of accuracy, a target number of processed images from a training data set, a target amount of change in a parameter relative to one or more previous data points, a combination thereof, and / or other criteria. In one embodiment, the stopping criteria are met when at least a minimum number of data points have been processed and at least a threshold accuracy has been reached. The threshold accuracy can be, for example, 70%, 80%, or 90% accuracy. In one embodiment, if the accuracy of the machine learning model has stopped improving, then the stopping criteria are met. If the stopping criteria are not met, then further training is performed. If the stopping criteria are met, then the training may be complete. Once the machine learning model is trained, a retained portion of the training data set can be used to test the model. Once one or more trained machine learning models 438 are generated, they can be stored in a model storage device 445 and can be added to the processed substrate result engine 430.

[0078] For the model application workflow 417, according to one embodiment, input data 462 may be input into one or more processed substrate result determiners 467, each of which may include a trained neural network or other model. Additionally or alternatively, one or more processed substrate result determiners 467 may apply an image processing algorithm to determine the results of the processed substrate. The input data may include a chamber component surface emissivity profile and / or a roughness profile (such as measured / generated using the optical measurement tools described herein). In addition, the input data may optionally include one or more images of the measured chamber components. Based on the input data 462, the processed substrate result determiner 467 may output one or more estimated processed substrate results 469. The processed substrate results 469 may include predicted qualities (e.g., thickness, uniformity, etc.) of one or more films deposited or etched on a substrate to be processed in a process chamber using the measured chamber components.

[0079] The action determiner 472 may determine one or more actions 470 to be performed based on the results 469 for the processed substrates. In one embodiment, the action determiner 472 compares the result estimates for the processed substrates to one or more result thresholds for the processed substrates. If the result estimates for one or more processed substrates meet or exceed the result thresholds for the processed substrates, the action determiner 472 may determine that it is recommended to replace chamber components and / or update process parameters for future substrate processing. In this case, the action determiner 472 may output a recommendation or notification to replace chamber components and / or update process parameters. In some embodiments, the action determiner 472 automatically updates the process parameters based on the results 469 for the processed substrates that meet one or more criteria. In some examples, the results 469 for the processed substrates may include an estimated condition of the substrate after one or more processing operations. In some embodiments, the estimated conditions may be used to determine one or more updates to the process parameters for future substrate processing using the chamber components in the processing chamber.

[0080] Figure 5A 1 is a flow chart of a method 500A for generating a training data set for training a machine learning model to perform substrate result evaluation according to aspects of the present disclosure. The method 500A is performed by processing logic, which may include hardware (circuitry, dedicated logic, etc.), software (such as running on a general-purpose computer system or a dedicated machine), firmware, or some combination thereof. In one embodiment, the method 500A may be performed by a computer system (such as Figure 3 In other or similar embodiments, one or more operations of method 500A may be performed by one or more other machines not depicted in the drawings.

[0081] At block 510 , processing logic initializes a training set T to an empty set (eg, {}).

[0082] At block 512, processing logic obtains substrate process result data associated with a substrate processed at a processing chamber of a manufacturing system (e.g., data associated with a surface of a thin film on a substrate, such as film thickness, uniformity, etc.). In some embodiments, processing logic obtains historical substrate process result data corresponding to substrates processed in a processing chamber using one or more historian chamber components.

[0083] At block 514, the processing logic obtains surface emissivity information and / or surface roughness information of a component included in the processing chamber, the component having processed the substrate. As previously described, the surface emissivity information and / or surface roughness information may be obtained by an optical measurement tool (e.g., Figure 3 The surface emissivity information and / or surface roughness information may be obtained by an optical measurement tool 326 of the embodiment of the present invention or a system for optically determining emissivity and / or surface roughness, such as the system 200 or the system 200. In some embodiments, the surface emissivity information and / or surface roughness information may include a profile map of the surface of the chamber component under test. In some embodiments, the processing logic obtains historical chamber component surface roughness data and / or historical chamber component emissivity data corresponding to historical measurement results of historical chamber components.

[0084] At block 516, processing logic generates training inputs based on the data obtained for chamber component surface emissivity and / or roughness at block 514. In some embodiments, the training inputs may include a normalized set of sensor data (e.g., normalized intensity of reflected and / or scattered radiation, normalized emissivity and / or surface roughness measurements, etc.).

[0085] At block 518, processing logic may generate a target output based on the substrate process result data obtained at block 512. The target output may correspond to a substrate result metric (data indicative of the quality of the processed substrate) for a substrate processed in the processing chamber.

[0086] At block 520, processing logic generates an input / output map. The input / output map refers to a training input that includes or is based on the data of the chamber components and a target output of the training input, wherein the target output identifies a substrate process result, and wherein the training input is associated with (or mapped to) the target output. At block 522, processing logic adds the input / output map to the training set T.

[0087] At box 524, processing logic determines whether the training set T includes an amount of training data sufficient to train the machine learning model. It should be noted that in some embodiments, the sufficiency of the training set T can be determined based solely on the number of input / output mappings in the training set, while in some other embodiments, the sufficiency of the training set T can be determined in addition to or instead of the number of input / output mappings based on one or more other criteria (e.g., a measure of the diversity of training examples, etc.). In response to determining that the training set T includes an amount of training data sufficient to train the machine learning model, processing logic provides the training set T to train the machine learning model. In response to determining that the training set does not include an amount of training data sufficient to train the machine learning model, method 500 returns to box 512.

[0088] At block 526, processing logic provides a training set T to train the machine learning model. In some embodiments, the training set T is provided to (e.g., Figure 3 The training engine 382 of the server machine 380 is used to perform the training. In the case of a neural network, for example, the input values ​​of a given input / output mapping (e.g., spectral data and / or chamber data for a previous substrate) are input to the neural network, and the output values ​​of the input / output mapping are stored in the output nodes of the neural network. The connection weights in the neural network are then adjusted according to a learning algorithm (e.g., back propagation, etc.), and the procedure is repeated for other input / output mappings in the training set T. After block 526, the machine learning model (e.g., Figure 3 The machine learning model 390) can be used to provide predicted substrate process outcomes for substrates processed in a processing chamber using the measured chamber components.

[0089] Figure 5B is a flow chart of a method 500B for generating predicted processed substrate results using a trained machine learning model according to aspects of the present disclosure. The method 500B is performed by processing logic, which may include hardware (circuitry, dedicated logic, etc.), software (such as running on a general-purpose computer system or a dedicated machine), firmware, or some combination thereof. In one embodiment, the method 500B may be performed by a computer system (such as Figure 3 In other or similar embodiments, one or more operations of method 500B may be performed by one or more other machines not depicted in the drawings.

[0090] At block 552, processing logic receives data associated with the emissivity and / or roughness of a surface of a chamber component of a processing chamber. In some embodiments, the data is obtained from an optical measurement tool (e.g., Figure 3The data may be received by an optical measurement tool 326 (e.g., an optical measurement tool 326) or a system for optically determining emissivity and / or surface roughness (e.g., system 100A or system 100B). The data may be raw sensor data or data that has been processed (e.g., by a processing device, a computing device, etc.) to determine surface emissivity and / or roughness. In some embodiments, the data is in the form of one or more surface profiles that indicate surface emissivity and / or roughness.

[0091] At block 554, processing logic inputs the data received at block 552 into a trained machine learning model. In some embodiments, the trained machine learning model is trained using the Figure 3 , Figure 4 and / or Figure 5A The trained machine learning model can be trained using the techniques described herein. The trained machine learning model can be trained using data inputs including historical surface roughness data and / or historical surface emissivity data labeled with corresponding target output data including historical substrate process result data. The trained machine learning model can be trained to output one or more predicted substrate process results based on the data inputs associated with the surface emissivity and / or roughness of chamber components.

[0092] At block 556, processing logic receives output from the trained machine learning model, the output comprising predicted substrate process results corresponding to future substrates to be processed using chamber components in the processing chamber. In some embodiments, the surface emissivity and / or roughness of the chamber components may affect the results of substrates processed in the processing chamber. The predicted substrate process results may reflect these effects.

[0093] Figure 6 6 is a flow chart of a method 600 for optically determining emissivity and / or surface roughness according to aspects of the present disclosure. The method 600 is performed by a system that may include hardware (circuitry, dedicated logic, optical measurement tools described herein, etc.), software (such as running on a general purpose computer system or a dedicated machine), firmware, or some combination thereof. In one embodiment, the method 600 may be performed by a computer system (such as Figure 3 In other or similar embodiments, one or more operations of method 600 may be performed by one or more other machines not depicted in the drawings.

[0094] At block 602, a radiation source of the system emits a radiation beam. In some embodiments, the radiation beam is an infrared radiation beam in the mid-infrared range (e.g., an infrared radiation beam output by a laser). For example, the wavelength range of the radiation beam may be 1-6 microns. In another example, the wavelength range of the radiation beam may be 3-5 microns. In some embodiments, the radiation source is a mid-infrared supercontinuum laser emitter configured to operate in the mid-infrared range. Thus, in some embodiments, the radiation beam is a mid-infrared infrared laser beam. In some embodiments, the radiation beam is directed to the surface of an object via one or more mirrors, filters (e.g., polarizing filters), lenses, and / or beam splitters. Due at least in part to the emissivity and / or roughness of the surface of the object, the surface of the object may reflect a portion of the radiation beam and / or may scatter a portion of the radiation beam (e.g., another portion). In some embodiments, the object is a chamber component of a substrate processing chamber.

[0095] At block 604, a first optical sensor of the system detects an intensity of a portion of a radiation beam reflected from a surface of an object (e.g., a chamber component). The intensity of the reflected radiation may indicate at least the emissivity and / or roughness of the surface of the object. In some embodiments, the reflected portion of the beam is directed to the first optical sensor via one or more mirrors, lenses, and / or beam splitters. In some examples, the reflected portion of the radiation beam returns to the radiation source along at least a portion of the path. The beam splitter may direct the reflected radiation on the path to the optical sensor.

[0096] At block 606, a second optical sensor of the system detects an intensity of a portion of the radiation beam that is scattered by a surface of the object (e.g., a chamber component). The intensity of the scattered radiation may be indicative of at least the emissivity and / or roughness of the surface of the object. In some embodiments, the scattered radiation is collected by a reflective lens (e.g., a Schwarzschild lens) and directed (e.g., reflected and / or focused) toward the second optical sensor. In some embodiments, the reflective lens is disposed substantially above the object (e.g., such as a Figure 1A and 1B In some embodiments, one or more mirrors, filters, lenses, etc. may direct, process, manipulate, reflect, etc. the scattered radiation.

[0097] At block 608, a processing device communicatively coupled to the first optical sensor and the second optical sensor may determine (e.g., via processing logic) at least one of a roughness of a surface of an object (e.g., a chamber component) or an emissivity of a surface of the object. In some embodiments, the processing device makes this determination based on a comparison of the intensity of the reflected radiation and the intensity of the scattered radiation described above. In some embodiments, a manufacturing process parameter (e.g., a manufacturing recipe, a manufacturing operation, etc.) corresponding to the object is updated (e.g., adjusted) based on the measured roughness and / or the measured emissivity. For example, a manufacturing process for a chamber component may be updated based on a measured value of a surface roughness and / or emissivity of a surface of a sample chamber component. In such an example, the measured value may indicate that the sample chamber component does not meet a target threshold (e.g., a target surface roughness threshold and / or a target emissivity threshold). The update to the manufacturing process may be to manufacture future chamber components within the target threshold according to the updated manufacturing process parameters.

[0098] In some embodiments, the processing device is communicatively coupled to a third optical sensor (e.g., a normalized sensor, optical sensor 108 of FIG. 1 , etc.). The processing device may use the signal received from the third optical sensor to normalize the signals received from the first optical sensor and the second optical sensor (e.g., normalize the intensity of reflected radiation and / or scattered radiation). In some embodiments, a surface map of the object is generated by capturing multiple measurements at different locations on the surface of the object. The surface map may indicate surface roughness and / or emissivity across the surface of the object. In some embodiments, the surface map may be generated by taking multiple separate measurements on the surface of the object. Alternatively, the map may be generated by a surface scanning operation (the operation being performed, for example, by system 100B of FIG. 1 ).

[0099] Figure 7A diagrammatic representation of a machine in the example form of a computing device 700 is described, in which a set of instructions for causing the machine to perform any one or more of the methods discussed herein can be executed. In an alternative embodiment, the machine can be connected (e.g., networked) with other machines in a local area network (LAN), an intranet, an extranet, or the Internet. The machine can operate as a server or client machine in a client and server network environment, or as a peer machine in a peer (or distributed) network environment. The machine can be a personal computer (PC), a tablet computer, a set-top box (STB), a personal digital assistant (PDA), a cellular phone, a web appliance, a server, a network router, a switch or a bridge, or any machine capable of executing a set of instructions (in sequence or otherwise) that specify the actions to be taken by the machine. Further, although only a single machine is shown, the term "machine" should also be considered to include a collection of any machines (e.g., computers) that execute a set (or multiple sets) of instructions to perform any one or more of the methods discussed herein, either individually or collectively. In embodiments, computing device 700 may correspond to one or more of server machine 370 , server machine 380 , or prediction server 312 described herein.

[0100] The example computing device 700 includes a processing device 702, a main memory 704 (e.g., read only memory (ROM), flash memory, dynamic random access memory (DRAM) such as synchronous dynamic random access memory (SDRAM), etc.), a static memory 706 (e.g., flash memory, static random access memory (SRAM), etc.), and a secondary memory (e.g., a data storage device 728) that communicate with each other via a bus 708.

[0101] The processing device 702 may represent one or more general-purpose processors such as a microprocessor, a central processing unit, etc. In more detail, the processing device 702 may be a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, a processor implementing other instruction sets, or a processor implementing a combination of instruction sets. The processing device 702 may also be one or more special-purpose processing devices such as an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), a network processor, etc. The processing device 702 may also be or include a system-on-chip (SoC), a programmable logic controller (PLC), or other types of processing devices. The processing device 702 is configured to execute processing logic to perform the operations discussed herein.

[0102] The computing device 700 may further include a network interface device 722 for communicating with a network 764. The computing device 700 may also include a video display unit 710 (e.g., a liquid crystal display (LCD) or a cathode ray tube (CRT)), an alphanumeric input device 712 (e.g., a keyboard), a cursor control device 714 (e.g., a mouse), and a signal generating device 720 (e.g., a speaker).

[0103] The data storage device 728 may include a machine-readable storage medium (or more specifically, a non-transitory computer-readable storage medium) 724 having stored thereon one or more sets of instructions 726 embodying any one or more of the methods or functions described herein. A non-transitory storage medium refers to a storage medium other than a carrier wave. The instructions 726 may also reside completely or at least partially within the main memory 704 and / or within the processing device 702 during execution of these instructions by the computer device 700, the main memory 704 and the processing device 702 also constituting computer-readable storage media.

[0104] Although the computer-readable storage medium 724 is shown as a single medium in the example embodiment, the term "computer-readable storage medium" should also be considered to include a single medium or multiple media (such as a centralized or distributed database and / or associated caches and servers) that store the one or more sets of instructions. The term "computer-readable storage medium" should also be considered to include any medium that can store or encode a set of instructions for execution by a machine and cause the machine to perform any one or more of the methods of the present disclosure. Therefore, the term "computer-readable storage medium" should be considered to include (but not limited to) solid-state memory and optical and magnetic media.

[0105] The foregoing description sets forth many specific details of examples such as specific systems, components, methods, etc., so that several embodiments of the present disclosure are well understood. However, it will be appreciated by those skilled in the art that at least some embodiments of the present disclosure may be implemented without these specific details. In other cases, well-known components or methods are not described in detail, or these components or methods are presented in a simple block diagram format to avoid unnecessarily obscuring the present disclosure. Therefore, the specific details set forth are merely exemplary. Specific implementations may differ from these exemplary details and are still considered to be within the scope of the present disclosure.

[0106] References throughout this specification to "one / a embodiment" mean that a particular feature, structure, or characteristic described in conjunction with the embodiment is included in at least one embodiment. Therefore, the phrase "in one / a embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment. In addition, the term "or" is intended to mean an inclusive "or" rather than an exclusive "or". When the term "about" or "approximately" is used herein, the term is intended to mean that the accuracy of the nominal value presented is within ±10%.

[0107] Although the operations of the methods herein are shown and described in a particular order, the order of operations of each method may also be changed so that certain operations may be performed in reverse order so that certain operations may be performed at least partially in parallel with other operations. In another embodiment, instructions or sub-operations of different operations may be performed in an intermittent and / or alternating manner.

[0108] It should be understood that the above description is intended to be illustrative, not restrictive. For those skilled in the art, many other embodiments will be apparent after reading and understanding the above description. Therefore, the scope of this disclosure will be determined with reference to the attached claims and the entire scope of equivalents given by such claims.

Claims

1. A system, comprising: a radiation source configured to emit a radiation beam; a first optical sensor configured to detect a first intensity of a first portion of the radiation beam reflected from a surface of an object; a second optical sensor configured to detect a second intensity of a second portion of the radiation beam scattered by the surface of the object; as well as A processing device is communicatively coupled to the first optical sensor and the second optical sensor, wherein the processing device is configured to determine at least one of a roughness of the surface of the object or an emissivity of the surface of the object based on a comparison of the first intensity and the second intensity.

2. The system of claim 1, wherein the radiation source comprises a mid-infrared supercontinuum laser.

3. The system of claim 1, further comprising: a mirror configured to direct the radiation beam toward the object; as well as A reflective objective is configured to receive the second portion of the radiation beam scattered by the surface of the object and to direct the second portion of the radiation beam scattered by the surface of the object toward the second optical sensor.

4. The system of claim 3, wherein the reflective objective comprises a Schwarzschild objective.

5. The system of claim 3, further comprising: A beam splitter is arranged along the optical axis between the radiation source and the mirror, wherein the first part of the radiation beam reflected from the surface of the object is reflected from the mirror, returned through one or more lenses, and directed to the first optical sensor by the beam splitter.

6. The system of claim 5, further comprising: a third optical sensor, wherein the beam splitter is configured to direct a portion of the radiation beam emitted by the radiation source to the third optical sensor, wherein the third optical sensor is configured to detect a third intensity of the portion of the radiation beam, and wherein the processing device is configured to normalize the detected first intensity and the detected second intensity based on the detected third intensity.

7. The system of claim 6, further comprising: A polarizing filter is disposed between the radiation source and the beam splitter along the optical axis, and the polarizing filter is configured to polarize the radiation beam emitted from the radiation source.

8. The system of claim 1, further comprising: One or more lenses are configured to focus the radiation beam, wherein the radiation beam is focused to a spot size having a diameter less than about 200 microns on the surface of the object.

9. The system of claim 8, further comprising: A rotatable mirror is configured to direct the radiation beam emitted by the radiation source toward the one or more lenses, wherein the rotatable mirror is configured to cause the radiation beam to periodically move across the surface of the object in response to rotation of the rotatable mirror.

10. The system of claim 9, wherein the system detects at least one of emissivity or roughness with a measurement system variation of less than 0.1%.

11. A method comprising: emitting a radiation beam from a radiation source; detecting, by a first optical sensor, a first intensity of a first portion of the radiation beam reflected from a surface of a chamber component of a processing chamber; detecting, by a second optical sensor, a second intensity of a second portion of the radiation beam scattered by the surface of the chamber component; as well as At least one of a roughness of the surface of the chamber component or an emissivity of the surface of the chamber component is determined based on a comparison of the first intensity and the second intensity via a processing device communicatively coupled to the first optical sensor and the second optical sensor.

12. The method of claim 11, wherein the radiation source comprises a mid-infrared supercontinuum laser.

13. The method of claim 11, further comprising: detecting, via a third optical sensor, a third intensity of a portion of the radiation beam emitted by the radiation source, the portion of the radiation beam emitted by the radiation source being directed toward the third optical sensor by a beam splitter disposed along an optical axis between the radiation source and a mirror, wherein the mirror is configured to direct the radiation beam toward the chamber component; And based on the detected third intensity, the detected first intensity and the detected second intensity are normalized.

14. The method of claim 11, further comprising: inputting data associated with at least one of the emissivity or the roughness of the surface of the chamber component into a model; as well as An output is received from the model, the output comprising predicted substrate process results, wherein the predicted substrate process results correspond to future substrates to be processed using the chamber components.

15. The method of claim 14, wherein the model comprises a trained machine learning model.

16. The method of claim 11, further comprising: Training a machine learning model to produce a trained machine learning model, wherein the machine learning model is trained with data inputs, the data inputs comprising one or more of: historical chamber component surface roughness data, historical chamber component emissivity data, data corresponding to the roughness of the surface of the chamber component, and data corresponding to the emissivity of the surface of the chamber component, wherein the data inputs are labeled with corresponding target output data, the target output data comprising historical substrate process result data corresponding to substrates processed with one or more historical chamber components.

17. The method of claim 11, further comprising: The surface of the component is scanned with the radiation beam by periodically moving the radiation beam across the surface of the component in response to rotation of a rotatable second mirror.

18. A non-transitory machine-readable storage medium comprising instructions that, when executed by a processing device, cause the processing device to perform operations comprising: receiving data associated with at least one of an emissivity or a roughness of a surface of a chamber component of a processing chamber; inputting the data associated with at least one of the emissivity or the roughness of the surface of the chamber component into a trained machine learning model; and An output is received from the trained machine learning model, the output comprising a predicted substrate process result, wherein the predicted substrate process result corresponds to a future substrate to be processed in the processing chamber using the chamber component.

19. The non-transitory machine-readable storage medium of claim 18, wherein the trained machine learning model is trained with data inputs comprising one or more of: historical chamber component surface roughness data, historical chamber component emissivity data, data corresponding to the roughness of the surface of the chamber component, and data corresponding to the emissivity of the surface of the chamber component, wherein the data inputs are labeled with corresponding target output data, the target output data comprising historical substrate process result data corresponding to substrates processed with one or more historical chamber components.

20. The non-transitory machine-readable storage medium of claim 18, wherein receiving data associated with at least one of the emissivity or the roughness of the surface of the chamber component comprises: receiving first sensor data from a first optical sensor, the first sensor data indicating a first intensity of a first portion of a radiation beam reflected from the surface of the chamber component; as well as receiving second sensor data from a second optical sensor, the second sensor data indicating a second intensity of a second portion of the radiation beam scattered by the surface of the chamber component, Wherein at least one of the emissivity or the roughness of the surface is based on a comparison of the first intensity and the second intensity.

Images