System for determining type of wafer using machine learning and method thereof

Abstract

A substrate type determination system according to one embodiment may include: a sensor unit that measures the state of a substrate on which a process is performed and generates measurement data; a machine learning unit that receives learning data measured from a learning substrate of known type and machine learns the pattern of the learning data; and a type determination unit that determines the type of the substrate from the measurement data using a pre-trained machine learning unit.

Description

Technology Field

[0001] The following embodiment relates to a substrate type determination system and a method using machine learning. Background Technology

[0002] The manufacturing of semiconductor devices requires chemical mechanical polishing (CMP) operations, including polishing, buffing, and cleaning. Semiconductor devices are in the form of a multilayer structure, and transistor devices with diffusion regions are formed on the substrate layer. On the substrate layer, connecting metal lines are patterned and electrically connected to the transistor devices forming the functional components. As is known, the patterned conductive layer is insulated from other conductive layers by an insulating material such as silicon dioxide. As more metal layers and associated insulating layers are formed, the need to flatten the insulating material increases. If flattening is not performed, the fabrication of additional metal layers becomes substantially more difficult due to significant variations in surface morphology. Furthermore, since the metal line pattern is formed with an insulating material, metal CMP operations are performed to remove excess metal.

[0003] The CMP process includes a polishing process that flattens the surface of a substrate by physically abrading it. The polishing process is performed by physically abrading the substrate by rubbing it against a polishing pad with grooves formed on its surface. Meanwhile, the types of patterns formed on the surface of substrates vary, and there are also cases where no patterns are formed on the substrate. Therefore, a process recipe corresponding to the type of surface pattern of the substrate is applied. If the substrate pattern type and the process recipe type do not correspond, equipment problems or process accidents may occur. Conventionally, work was performed by an operator checking the substrate pattern type and inputting the corresponding process recipe; however, relying on such operator judgment posed a problem where errors could occur. Therefore, a substrate type determination system and method utilizing machine learning are required to automatically determine whether the substrate pattern type and the process recipe type match, thereby preventing process accidents.

[0004] The aforementioned background technology is one that the inventor possessed or acquired in the process of deriving the content of the disclosure of the present application, and it cannot be considered as prior art disclosed to the general public prior to the filing of this application. The problem to be solved

[0005] The objective of one embodiment is to provide a system and method for determining the type of substrate on which a process is performed using machine learning.

[0006] The objective of one embodiment is to provide a system and method for preventing process accidents by determining whether a determined substrate type matches a process type being performed. means of solving the problem

[0007] A substrate type determination system according to one embodiment may include: a sensor unit that measures the state of a substrate on which a process is performed and generates measurement data; a machine learning unit that receives learning data measured from a learning substrate of known type and machine learns the pattern of the learning data; and a type determination unit that determines the type of the substrate from the measurement data using a pre-trained machine learning unit.

[0008] The machine learning unit may include: a learning data acquisition unit that acquires learning data according to each type of a learning substrate of known type; a pattern extraction unit that receives each learning data from the learning data acquisition unit and extracts a pattern of each data from each learning data; and a machine learning model generation unit that generates a machine learning model by labeling the type of substrate corresponding to each extracted pattern.

[0009] The type determination unit may include a pattern matching unit that determines whether a pattern is matched by comparing the measurement data with a pattern extracted by the pattern extraction unit; and a result output unit that searches for a pattern determined to be matched by the pattern matching unit in the machine learning model and outputs a corresponding type as a result value.

[0010] It may further include a control unit that determines whether the type of the substrate and the type of process performed on the substrate match according to the result value of the result output unit.

[0011] The control unit may continue the process when it is determined that the type of the substrate and the type of the process match, and may stop the process or change the process being performed to a process corresponding to the type of the substrate when it is determined that the type of the substrate and the type of the process do not match.

[0012] The sensor unit may include at least one of an optical sensor, an eddy current sensor, a current sensor, a voltage sensor, and a torque sensor.

[0013] A method for determining a substrate type according to one embodiment may include: a machine learning step of receiving learning data measured from a learning substrate of known type and machine learning the pattern of the learning data; a process step of performing a process on the substrate; a measurement step of measuring the state of the substrate to generate measurement data; and a type determination step of determining the type of the substrate from the measurement data using a machine learning model that has been pre-trained through the machine learning step.

[0014] The machine learning step described above may include: a learning data acquisition step for acquiring learning data according to each type of a learning substrate of known type; a pattern extraction step for extracting a pattern of each data from each acquired learning data; and a machine learning model generation step for generating a machine learning model by labeling the type of substrate corresponding to each extracted pattern.

[0015] The type determination step may include a pattern matching step that determines whether a pattern is matched by comparing the measurement data with a pattern extracted in the pattern extraction step; and a result output step that searches for a pattern determined to be matched in the machine learning model and outputs a corresponding type as a result value.

[0016] It may further include a type matching determination step for determining whether the type of the substrate according to the output result value matches the type of the process performed in the process step.

[0017] If it is determined that the type of the substrate and the type of the process match, the process step is continued; if it is determined that the type of the substrate and the type of the process do not match, the process step is stopped, or the process performed in the process step is changed to a process corresponding to the type of the substrate and continued. The method may further include a control step. Effects of the invention

[0018] A substrate type determination system and method according to one embodiment can determine the type of substrate on which a process is performed using machine learning.

[0019] A substrate type determination system and method according to one embodiment can determine whether the determined substrate type matches the type of process being performed, thereby preventing process accidents caused by mismatch.

[0020] The effects of the substrate type determination system and method according to one embodiment are not limited to those mentioned above, and other unmentioned effects will be clearly understood by a person skilled in the art from the description below. Brief explanation of the drawing

[0021] FIG. 1 is a block diagram of a substrate type determination system according to one embodiment. FIG. 2 is a block diagram of a machine learning unit of a substrate type determination system according to one embodiment. FIG. 3 is a block diagram of a type determination unit of a substrate type determination system according to one embodiment. FIG. 4 is a flowchart of a method for determining a substrate type according to one embodiment. FIG. 5 is a flowchart of the machine learning steps for a substrate type determination method according to one embodiment. FIG. 6 is a flowchart of the type determination step for a substrate type determination method according to one embodiment. Specific details for implementing the invention

[0022] Hereinafter, embodiments are described in detail with reference to the attached drawings. However, various modifications may be made to the embodiments, and thus the scope of the patent application is not limited or restricted by these embodiments. It should be understood that all modifications, equivalents, and substitutions to the embodiments are included within the scope of the rights.

[0023] The terms used in the embodiments are for illustrative purposes only and should not be interpreted as intended to be limiting. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as "comprising" or "having" are intended to indicate the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0024] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the embodiments pertain. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application.

[0026] In addition, when describing with reference to the attached drawings, identical components are assigned the same reference numeral regardless of drawing symbols, and redundant descriptions thereof are omitted. In describing the embodiments, if it is determined that a detailed description of related prior art could unnecessarily obscure the essence of the embodiments, such detailed description is omitted.

[0027] In addition, terms such as first, second, A, B, (a), (b), etc., may be used when describing the components of the embodiments. These terms are intended only to distinguish the components from other components, and the nature, order, or sequence of the components is not limited by the terms. Where it is stated that a component is "connected," "combined," or "connected" to another component, it should be understood that the component may be directly connected or connected to the other component, but that another component may also be "connected," "combined," or "connected" between each component.

[0028] Components included in any one embodiment and components having common functions shall be described using the same names in other embodiments. Unless otherwise stated, the description in any one embodiment may also apply to other embodiments, and specific descriptions shall be omitted to the extent of overlap.

[0030] FIG. 1 is a block diagram of a substrate type determination system according to one embodiment. FIG. 2 is a block diagram of a machine learning unit of a substrate type determination system according to one embodiment. FIG. 3 is a block diagram of a type determination unit of a substrate type determination system according to one embodiment.

[0031] Referring to FIGS. 1 to 3, a substrate type determination system (1) according to one embodiment can determine the type (T) of a substrate (W) using machine learning. Here, the type (T) of the substrate (W) refers to the type of pattern formed on the surface of the substrate (W), and a case where no pattern is formed (non-patterned substrate) may also be included as one type. The substrate type determination system (1) can learn the pattern (P) of data corresponding to the type (T) of the substrate (W) through machine learning using learning data (LD) measured from a learning substrate (W) with a known type (T). The substrate type determination system (1) can determine whether a match exists by comparing the measurement data (SD) measured from the substrate (W) on which a process (e.g., a polishing process) is performed with the pre-learned pattern (P), and determine the type (T) corresponding to the matched pattern (P) as the type of the substrate (W). The substrate type determination system (1) determines whether the type (T) of the determined substrate (W) matches the type of the process being performed, thereby preventing process accidents caused by a mismatch between the substrate (W) type (T) and the process type.

[0033] A substrate type determination system (1) according to one embodiment may include a sensor unit (11), a machine learning unit (12), a type determination unit (13), and a control unit (14).

[0034] The sensor unit (11) can measure the state of the substrate (W). For example, the sensor unit (11) may be a sensor for measuring the polishing state or polishing endpoint of the substrate (W). For example, the sensor unit (11) may include at least one of an optical sensor, an eddy current sensor, a current sensor, a voltage sensor, and a torque sensor. The sensor unit (11) is provided on a platen and can measure information about the substrate (W) while rotating together with the platen. However, the type of sensor is not limited to this, and any type of sensor that measures the state of the substrate (W) may be acceptable. The sensor unit (11) can perform sensing on the substrate (W) on which the process is performed and generate measurement data (SD). The measurement data (SD) can be measured in real time by the sensor unit (11).

[0035] The machine learning unit (12) can receive learning data (LD) measured from a learning substrate (W) of known type (T). Here, learning data (LD) may refer to data measured by the sensor unit (11) performing sensing on the learning substrate (W) of known type (T). The machine learning unit (12) can machine learn the pattern (P) of the learning data (LD) using the learning data (LD). That is, the machine learning unit (12) can learn the pattern (P) of data generated according to each type (T) of the substrate (W) through supervised learning. The machine learning of the machine learning unit (12) can be performed in advance before the process is executed.

[0036] The machine learning unit (12) may include a learning data acquisition unit (121), a pattern extraction unit (122), and a learning model generation unit (123).

[0037] The learning data acquisition unit (121) can acquire each learning data (LD) according to each type (T) of the learning substrate (W) of which the type (T) is known. The learning data (LD) can be acquired for each type (T) of the substrate (W). For example, data acquired from a substrate (W) having a first type (T1) can be acquired as first learning data (LD1), and data acquired from a substrate (W) having a second type (T2) can be acquired as second learning data (LD2). The learning data acquisition unit (121) can acquire data measured through prior measurement or acquire data from a database.

[0038] The pattern extraction unit (122) can receive each learning data (LD) from the learning data acquisition unit (121). In the learning data (LD) detected in a specific type (T), a specific pattern (P) may be formed on the data due to that specific type (T). The pattern extraction unit (122) can extract the pattern (P) of the data from each received learning data (LD). The pattern (P) can be extracted for each learning data (LD). For example, the pattern extraction unit (122) can extract a first pattern (P1) from the first learning data (LD1) and extract a second pattern (P2) from the second learning data (LD2). For example, the pattern extraction unit (122) can extract the pattern (P) by performing FFT processing on the learning data (LD). The pattern (P) extraction by the pattern extraction unit (122) can be performed through machine learning from a plurality of learning data (LD) acquired from a substrate (W) of the same type (T).

[0039] The learning model generation unit (123) can generate a machine learning model (M) by labeling the type (T) of the substrate (W) corresponding to the extracted pattern (P). That is, for each specific pattern (P), the type (T) of the corresponding substrate (W) can be labeled and stored as a machine learning model (M). The machine learning model (M) may include information that a specific pattern (P) appears in the data detected in the specific type (T), in other words, information that if a specific pattern (P) appears in the detected data, the corresponding substrate (W) corresponds to the specific type (T). For example, the first machine learning model (M1) may store information about the first pattern (P1) and the first type (T1) corresponding thereto, and the second machine learning model (M2) may store information about the second pattern (P2) and the second type (T2) corresponding thereto.

[0040] The type determination unit (13) can determine the type (T) of the substrate (W) on which the process is performed from the measurement data (SD) using the pre-learned machine learning unit (12). The type determination unit (13) can compare the measurement data (SD) with a pre-learned pattern (P) and determine the type (T) corresponding to the matching pattern (P) as the type (T) of the substrate (W). In order to determine the type (T) of the substrate (W) quickly, the type determination unit (13) can determine the type (T) of the substrate (W) using the initial measurement data (SD). For example, the type determination unit (13) can determine the type (T) of the substrate (W) using only the measurement data (SD) measured by the sensor unit (11) within 3 cycles (number of rotations of the platen) from the start of the process.

[0041] The type determination unit (13) may include a pattern matching unit (131) and a result output unit (132).

[0042] The pattern matching unit (131) can determine whether the pattern (P) is matched by comparing the measurement data (SD) with the pattern (P) extracted from the pattern extraction unit (122). That is, the pattern matching unit (131) can determine the pattern (P) that matches the measurement data (SD). There may be multiple patterns (P) that match the measurement data (SD). For example, the pattern matching unit (131) can determine whether the pattern (P) is matched by performing FFT processing on the measurement data (SD). The task of matching the measurement data (SD) with a pre-learned pattern (P) by the pattern matching unit (131) can be performed through machine learning.

[0043] The result output unit (132) can search for a pattern (P) determined to be matched by the pattern matching unit (131) in a machine learning model (M) and output the corresponding substrate (W) type (T) as a result value. That is, the result output unit (132) can output the type (T) corresponding to the pattern (P) that matches the measurement data (SD) as the substrate (W) type (T). The output type (T) may represent the substrate (W) type (T) currently undergoing the process.

[0044] The control unit (14) can determine whether the type (T) of the substrate (W) and the type of process performed on the substrate (W) match according to the result value of the result output unit (132). If the control unit (14) determines that the type (T) of the substrate (W) and the type of process match, the process can be continued. If the control unit (14) determines that the type (T) of the substrate (W) and the type of process do not match, the process can be stopped and a notification can be generated. Alternatively, if the control unit (14) determines that the type (T) of the substrate (W) and the type of process do not match, the process being performed can be changed to a process corresponding to the determined type (T) of the substrate (W), the changed process can be performed, and a notification can be generated. With such a configuration, process accidents caused by a mismatch between the type (T) of the substrate (W) and the type of process can be prevented.

[0046] FIG. 4 is a flowchart of a substrate type determination method according to one embodiment. FIG. 5 is a flowchart of a machine learning step for a substrate type determination method according to one embodiment. FIG. 6 is a flowchart of a type determination step for a substrate type determination method according to one embodiment.

[0047] Referring to FIGS. 4 to 6, a substrate type determination method (2) according to one embodiment can determine the type of a substrate using machine learning and determine whether the determined type of the substrate matches the type of the process to control the process. In describing the substrate type determination method (2) with reference to FIGS. 4 to 6, content that overlaps with the above description is omitted as much as possible.

[0048] A substrate type determination method (2) according to one embodiment may include a machine learning step (21), a process step (22), a measurement step (23), a type determination step (24), a type matching determination step (25), and a control step (26).

[0049] The machine learning step (21) may be a step of receiving learning data measured from a learning substrate of known type and machine learning the pattern of the learning data.

[0050] The machine learning step (21) may include a learning data acquisition step (211), a pattern extraction step (212), and a learning model creation step (213).

[0051] The training data acquisition step (211) may be a step of acquiring training data according to each type of a known substrate. The pattern extraction step (212) may be a step of extracting a pattern of data from each acquired training data. The training model generation step (213) may be a step of generating a machine learning model by labeling the type of substrate corresponding to the extracted pattern.

[0052] The process step (22) may be a step of performing a process on a substrate. The type of process performed here may be primarily selected by the operator.

[0053] The measurement step (23) may be a step of measuring the state of the substrate and generating measurement data.

[0054] The type determination step (24) may be a step of determining the type of the substrate from the measurement data using a machine learning model that has been learned in advance through the machine learning step.

[0055] The type determination step (24) may include a pattern matching step (241) and a result output step (242).

[0056] The pattern matching step (241) may be a step of determining whether pattern matching is possible by comparing the measurement data with the pattern extracted in the pattern extraction step.

[0057] The result output step (242) may be a step of searching for a pattern determined to be matched in a machine learning model and outputting the type of the corresponding substrate as a result value.

[0058] The type matching determination step (25) may be a step for determining whether the type of substrate according to the output result value matches the type of process performed in the process step (22).

[0059] If the control step (26) determines that the type of the substrate and the type of the process match, the process step may continue. If the control step (26) determines that the type of the substrate and the type of the process do not match, the process step (22) may be stopped and a notification may be generated. Alternatively, if the control step (26) determines that the type of the substrate and the type of the process do not match, the process performed in the process step (22) may be changed to a process corresponding to the type of the substrate and proceed, and a notification may be generated.

[0061] Although the embodiments have been described above with reference to the limited drawings, those skilled in the art can apply various technical modifications and variations based on the above. For example, suitable results may be achieved even if the described techniques are performed in a different order than described, and / or if the components of the described system, structure, device, circuit, etc. are combined or assembled in a form different from described, or replaced or substituted by other components or equivalents.

[0062] Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims set forth below. Explanation of the symbols

[0063] 1: Board type determination system 11: Sensor section 12: Machine Learning Department 13: Type determination unit 14: Control unit

Claims

Claim 1 A sensor unit that measures the state of a substrate on which a process is performed and generates measurement data; a machine learning unit that receives learning data measured from a learning substrate of known type and machine learns the pattern of said learning data; and a type determination unit that determines the type of said substrate from said measurement data using a pre-trained machine learning unit, wherein the machine learning unit includes: a learning data acquisition unit that acquires learning data according to each type of a learning substrate of known type; a pattern extraction unit that receives each learning data from said learning data acquisition unit and extracts the pattern of said data from said learning data; and a machine learning model generation unit that generates a machine learning model by labeling the type of the substrate corresponding to each extracted pattern, and the type determination unit includes a pattern matching unit that determines whether pattern matching is possible by comparing said measurement data with the pattern extracted by said pattern extraction unit. A substrate type determination system comprising: a result output unit that searches for a pattern determined to be matched in the pattern matching unit in the machine learning model and outputs a corresponding type as a result value; further comprising a control unit that determines whether the type of the substrate and the type of the process performed on the substrate match according to the result value of the result output unit; wherein the control unit continues the process when it is determined that the type of the substrate and the type of the process match, and stops the process or changes the process being performed to a process corresponding to the type of the substrate when it is determined that the type of the substrate and the type of the process do not match, wherein the type refers to a type of pattern formed on the surface of the substrate. Claim 2 delete Claim 3 delete Claim 4 delete Claim 5 delete Claim 6 A substrate type determination system according to claim 1, wherein the sensor unit comprises at least one of an optical sensor, an eddy current sensor, a current sensor, a voltage sensor, and a torque sensor. Claim 7 The method comprises: a machine learning step of receiving training data measured from a training substrate of known type and machine learning the pattern of said training data; a process step of performing a process on said substrate; a measurement step of measuring the state of said substrate to generate measurement data; and a type determination step of determining the type of said substrate from said measurement data using a machine learning model pre-trained through said machine learning step, wherein the machine learning step comprises: a training data acquisition step of acquiring training data according to each type of a training substrate of known type; a pattern extraction step of extracting the pattern of said data from each acquired training data; and a machine learning model generation step of labeling the type of said substrate corresponding to each extracted pattern to generate a machine learning model, and the type determination step comprises: a pattern matching step of comparing said measurement data with the pattern extracted in said pattern extraction step to determine whether there is a pattern match. A method for determining a substrate type, comprising: a result output step for searching a pattern determined to be matched in a machine learning model and outputting a corresponding type as a result value; further comprising a type matching determination step for determining whether the type of the substrate and the type of process performed in the process step match according to the output result value; further comprising a control step for continuing the process step when it is determined that the type of the substrate and the type of the process match, and stopping the process step or changing the process performed in the process step to a process corresponding to the type of the substrate when it is determined that the type of the substrate and the type of the process do not match, wherein the type refers to a type of pattern formed on the surface of the substrate. Claim 8 delete Claim 9 delete Claim 10 delete Claim 11 delete

Images