Method for determining or predicting state of chamber, method for correcting measurement error in chamber, and apparatus for monitoring state of chamber

Abstract

The present disclosure discloses a method by which a chamber monitoring apparatus monitors a state of a chamber. In particular, the method may comprise: generating a plurality of pieces of state data by measuring the state of the chamber; acquiring a plurality of pieces of state similarity data by calculating similarities between each of the plurality of pieces of state data and reference state data; acquiring a plurality of state similarity patterns by matching the plurality of pieces of state similarity data with process section data regarding sections in which a plurality of processes are respectively performed; and determining whether or not the chamber normally operates or predicting a point in time for an abnormal operation of the chamber, by using two or more state similarity patterns.

Description

Method for judging or predicting the state of a chamber, method for correcting measurement error for the chamber, and device for monitoring the state of the chamber

[0001] The present disclosure relates to a method for determining or predicting the state of a chamber, a method for correcting a measurement error for the chamber, and a device therefor.

[0002] Specifically, the present disclosure relates to a method for predicting whether a chamber is malfunctioning, aging, or requiring replacement of components, and / or the timing of failure, aging, or replacement of components. Furthermore, the present disclosure relates to a method for correcting measurement errors that may occur when attaching or detaching a monitoring device to a chamber and / or when measurements are performed between different chambers.

[0003] In semiconductor and display manufacturing and inspection processes, it's crucial to ensure that equipment is within verified normal operating ranges and that process conditions are properly set. While monitoring specific plasma physical quantities can provide some insight into the ongoing process, directly verifying equipment assembly before the process begins or ensuring that there are no abnormalities during the process remains challenging. Ultimately, test runs are the only option, requiring verification of process results, which incurs additional costs and time. Furthermore, even when process results are incorrect, identifying the source of the problem can be challenging.

[0004] Furthermore, the method of conducting a test run and checking the process results as described above can only monitor the current status of the equipment and cannot predict the extent to which the equipment is aging or when at least one part of the equipment will fail.

[0005] However, discovering equipment failures after they occur can disrupt manufacturing processes like semiconductor and display manufacturing for days or even months, potentially causing significant delays in production schedules and unexpected declines in production volume. Furthermore, reducing measurement errors when assessing equipment condition is considered crucial for accurately determining the timing of equipment deterioration and failure.

[0006] The present disclosure may provide for determining whether a chamber is aging or failing.

[0007] The present disclosure may provide for predicting the aging and failure point of a chamber.

[0008] The present disclosure may provide a method for reducing measurement error when performing multiple state measurements for one chamber.

[0009] The present disclosure may provide a method for reducing measurement errors when measuring states for multiple chambers.

[0010] The problems to be solved in this disclosure are not limited to the problems described above, and problems not mentioned can be clearly understood by a person having ordinary skill in the technical field to which this disclosure pertains from this disclosure.

[0011] According to an embodiment of the present disclosure, a method for monitoring a state of a chamber by a chamber monitoring device is provided, wherein: a process cycle including a plurality of processes occurs at least once within the chamber; reference state data for monitoring the state of the chamber is obtained; wherein the reference state data is first state data of the chamber measured in an idle section of the chamber or second state data of the chamber measured in a section corresponding to any one of the plurality of processes; transmitting a radio wave of a specific frequency range into the chamber M times and receiving a radio wave reflected from the inside of the chamber M times; wherein M is a natural number equal to or greater than 2; measuring the state of the chamber based on the radio waves received M times to generate a plurality of third state data; calculating similarities between each of the plurality of third state data and the reference state data, and obtaining a plurality of state similarity data using the similarities; By matching the plurality of state similarity data with the process section data for the sections in which each of the plurality of processes is performed, a plurality of state similarity patterns configured in process cycle units are obtained; and among the plurality of state similarity patterns, two or more state similarity patterns are used to determine whether the chamber is operating normally or to predict the time point of abnormal operation of the chamber.

[0012] A chamber monitoring device for monitoring the state of a chamber according to the present disclosure, comprising: - wherein a process cycle including a plurality of processes occurs at least once within the chamber; - an antenna for transmitting radio waves of a specific frequency range into the interior of the chamber and receiving radio waves reflected within the interior of the chamber; a bracket for fixing the antenna to the outside of the chamber; a signal processing unit for applying an electric signal to the antenna and obtaining an electric signal from the antenna; and a control unit for controlling the signal processing unit and generating state data of the chamber based on the electric signal obtained by the signal processing unit; wherein the control unit obtains reference state data for monitoring the state of the chamber, - wherein the reference state data is first state data of the chamber measured in an idle section of the chamber or second state data of the chamber measured in a section corresponding to any one of the plurality of processes;The signal processing unit is controlled to obtain an electric signal according to the M received electric signal by causing the antenna to transmit the radio wave M times and receive the radio wave reflected inside the chamber M times, wherein M is a natural number equal to or greater than 2, to measure the state of the chamber based on the obtained electric signal to generate a plurality of third state data, to calculate similarities between each of the plurality of third state data and the reference state data, to obtain a plurality of state similarity data using the similarities, and to match the plurality of state similarity data with process section data for sections in which each of the plurality of processes is performed, thereby obtaining a plurality of state similarity patterns configured in units of process cycles, and among the plurality of state similarity patterns, two or more state similarity patterns can be used to determine whether the chamber is operating normally or to predict the time point of abnormal operation of the chamber.;

[0013] In accordance with an embodiment of the present disclosure, a method for determining an error correction function for correcting a measurement error between parameter sets caused by attachment or detachment of P chamber monitoring devices (P is a natural number greater than or equal to 1) according to the present disclosure, wherein the parameter sets are one of (i) an S-parameter set, (ii) an H-parameter set, (iii) a Y-parameter set, (iv) a Z-parameter set, and (v) a parameter set derived from one of the S-parameters, the H-parameters, the Y-parameters, and the Z-parameters, wherein: attaching a first chamber monitoring device among the P chamber monitoring devices within a first region outside a chamber in a first process state, and then transmitting a radio wave of a specific frequency range into the interior of the chamber, thereby generating a reference parameter set of the chamber; repeatedly performing a first operation N times (N is a natural number greater than or equal to 2) for each of the P chamber monitoring devices, thereby generating L first parameter sets, wherein L is a value obtained by multiplying P by N; A method comprising: determining a first function including a plurality of first error correction parameters; updating the plurality of first error correction parameters with a plurality of second error correction parameters by repeating a second operation until difference information between an output parameter set obtained by substituting any one of the L first parameter sets into the first function and the reference parameter set is included within a predetermined range; and determining a second learning model including the plurality of second error correction parameters as the error correction function; wherein the second operation comprises: obtaining a first output parameter set by substituting any one of the L first parameter sets into the first function, wherein the any one of the L first parameter sets is a first parameter set that has not been previously substituted into the first function;And obtaining first difference information between the first output parameter set and the reference parameter set, and modifying the plurality of first error correction parameters based on the first difference information; wherein the first operation may include: detaching the chamber monitoring device from the chamber, reattaching it within the first region, and then transmitting radio waves of the specific frequency range into the interior of the chamber, wherein the chamber is in the first process state; and receiving radio waves reflected from the interior of the chamber, and generating a first parameter set based on the received radio waves.

[0014] In a method for determining an error correction function for correcting a measurement error between parameter sets corresponding to each of a plurality of chambers using P chamber monitoring devices (P is a natural number greater than or equal to 1) according to an embodiment of the present disclosure, wherein the parameter set is one of (i) an S-parameter set, (ii) an H-parameter set, (iii) a Y-parameter set, (iv) a Z-parameter set, and (v) a parameter set derived from one of the S-parameters, the H-parameters, the Y-parameters, and the Z-parameters, wherein, among the P chamber monitoring devices, a first chamber monitoring device is attached within a first region outside a first chamber in a first process state, and then a radio wave of a specific frequency range is transmitted into the first chamber, thereby generating a reference parameter set of the first chamber; A method for generating L sets of first parameters by repeatedly performing a first operation N times (N is a natural number greater than or equal to 2) for each of P chamber monitoring devices, wherein L is a value obtained by multiplying P by N; determining a first function including a plurality of first error correction parameters; updating the plurality of first error correction parameters with a plurality of second error correction parameters by repeating a second operation until difference information between an output parameter set obtained by substituting any one of the L sets of first parameters into the first function and the reference parameter set is included within a predetermined range; and determining a second learning model including the plurality of second error correction parameters as the error correction function; wherein the second operation obtains a first output parameter set by substituting any one of the N sets of first parameters into the first function, wherein any one of the sets of first parameters is a first parameter set that has not been previously substituted into the first function;And obtaining first difference information between the first output parameter set and the reference parameter set, and modifying the plurality of first error correction parameters based on the first difference information; wherein the first operation may include: attaching a chamber monitoring device within a first area of ​​a second chamber selected from among the plurality of chambers, and then transmitting radio waves of the specific frequency range into the interior of the second chamber, wherein the second chamber is in the first process state, and the second chamber is a chamber that has not been previously selected from among the plurality of chambers; and receiving radio waves reflected from the interior of the second chamber, and generating a first parameter set based on the received radio waves.

[0015] A method for determining an error correction function for correcting a measurement error between parameter sets caused by attachment or detachment of chamber monitoring devices using P chamber monitoring devices (P is a natural number greater than or equal to 1) according to one embodiment of the present disclosure and a measurement error between parameter sets corresponding to each of Q chambers (Q is a natural number greater than or equal to 2), wherein the parameter sets are one of (i) an S-parameter set, (ii) an H-parameter set, (iii) a Y-parameter set, (iv) a Z-parameter set and (v) a parameter set derived from one of the S-parameter, the H-parameter, the Y-parameter and the Z-parameter, wherein, after attaching a first chamber monitoring device among the P chamber monitoring devices within a first region outside a first chamber in a first process state, a radio wave of a specific frequency range is transmitted into the first chamber, thereby generating a reference parameter set of the first chamber; A first operation is repeatedly performed Q times (Q is a natural number greater than or equal to 2) for each of P chamber monitoring devices to generate L sets of first parameters, wherein the first operation is performed N times for each of the Q chambers, and L is a product of P, Q, and N; A first learning model including a plurality of first error correction parameters is determined; A second operation is repeated until difference information between an output parameter set obtained by inputting any one of the L sets of first parameters into the first learning model and the reference parameter set is included within a predetermined range, thereby updating the plurality of first error correction parameters into a plurality of second error correction parameters; And A second learning model including the plurality of second error correction parameters is determined as the error correction function;, wherein the second operation comprises: inputting one of the L first parameter sets into the first learning model to obtain a first output parameter set, wherein the one of the first parameter sets is a first parameter set that has not been previously input into the first learning model; and obtaining first difference information between the first output parameter set and the reference parameter set, and modifying the plurality of first error correction parameters based on the first difference information; and wherein the third operation comprises: detaching a chamber monitoring device from the chamber, reattaching it within the first region, and transmitting radio waves of the specific frequency range into the interior of the chamber, wherein the chamber is in the first process state; And receiving a radio wave reflected from inside the chamber, and generating a first parameter set based on the received radio wave; In accordance with an embodiment of the present disclosure, a chamber monitoring device may be provided for monitoring a first chamber using a parameter set, wherein the parameter set is one of (i) an S-parameter set, (ii) an H-parameter set, (iii) a Y-parameter set, (iv) a Z-parameter set, and (v) a parameter set derived from one of the S-parameter, the H-parameter, the Y-parameter, and the Z-parameter; obtaining a reference parameter set for monitoring the first chamber; attaching the chamber monitoring device within a first area outside the first chamber; transmitting a radio wave of a specific frequency range into the interior of the first chamber, wherein the first chamber is in one of a plurality of process states that can be performed within the first chamber and an idle (ILDE) state; receiving a radio wave reflected from inside the first chamber, and generating a first parameter set based on the received radio wave; By substituting the first parameter set into the error correction function, a second parameter set in which the first parameter set is corrected is obtained;Comparing the second parameter set with the reference parameter set to calculate a similarity for each of the second parameters included in the second parameter set; and obtaining monitoring information of the first chamber based on the similarity for each of the second parameters; wherein the error correction function comprises generating a plurality of second parameter sets for the at least one chamber by repeatedly attaching and detaching at least one chamber monitoring device to and from each of the at least one chamber at least once, and sequentially using each of the plurality of second parameter sets to modify error correction parameters included in the error correction function, wherein one of the plurality of second parameter sets is input to the error correction function, and the error correction parameters are modified until an output parameter set satisfies a certain condition.

[0016] According to the present disclosure, the state of a chamber can be monitored at low cost and time by transmitting radio waves into the interior of the chamber and receiving and analyzing radio waves reflected from the interior of the chamber.

[0017] According to the present disclosure, the accuracy of chamber status monitoring measurement can be improved by reducing measurement errors that may occur while monitoring the status of each chamber by repeatedly attaching and detaching the monitoring device to each chamber at least once.

[0018] The effects of the invention of the present application are not limited to the effects described above, and effects not mentioned can be clearly understood by a person having ordinary skill in the technical field to which the present application belongs from the present application.

[0019] FIG. 1 is a block diagram of a monitoring device according to one embodiment.

[0020] Figures 2 and 3 are schematic diagrams of a monitoring device installed in a chamber according to one embodiment.

[0021] Figure 4 is a graph of the S11 parameter, which is an example of state data.

[0022] FIG. 5 is a flowchart of a method for generating reference state data according to one embodiment of the present disclosure.

[0023] Figure 6 is a flowchart of a method for calculating the similarity between reference state data and state data.

[0024] Figure 7 is a graph showing actual measured similarity values ​​according to the present disclosure.

[0025] FIG. 8 is a flowchart of a method for predicting an abnormal operation point of a chamber or determining whether a chamber is operating normally by calculating a similarity pattern according to the present disclosure.

[0026] Figure 9 relates to a method for determining a similarity pattern by matching process data and a similarity graph.

[0027] FIG. 10 is a method for determining a similarity pattern according to an embodiment of the present disclosure.

[0028] FIG. 11 is for explaining predicting the abnormal operation point of a chamber according to an embodiment of the present disclosure.

[0029] Figures 12 and 13 are for explaining how to determine whether a chamber is operating normally according to an embodiment of the present disclosure.

[0030] FIG. 14 is for explaining the consistency between the result of determining whether a chamber is operating normally according to an embodiment of the present disclosure and whether an actual chamber is operating normally.

[0031] FIGS. 15 to 17 are for explaining the correspondence between the predicted time of abnormal operation of the chamber and the actual time when the chamber operates abnormally according to an embodiment of the present disclosure.

[0032] Figures 18 and 19 are for explaining another embodiment of determining abnormal operation of a chamber.

[0033] Figures 20 to 22 illustrate another embodiment of predicting the abnormal operation point of a chamber.

[0034] Figures 23 to 26 are provided to explain measurement errors that occur when attaching or detaching a monitoring device to at least one chamber.

[0035] FIG. 27 is intended to illustrate examples of learning models that can be used for measurement error correction according to the present disclosure.

[0036] FIGS. 28 to 31 are for explaining a method of generating an error correction function according to an embodiment of the present disclosure.

[0037] Figures 32 to 36 are for explaining a method of generating chamber monitoring information after correcting errors in state data using an error correction function and the error correction results obtained thereby.

[0038] Figure 37 is a block diagram of a monitoring system according to one embodiment.

[0039] Figure 38 is a schematic diagram of a monitoring system installed in process equipment according to one embodiment.

[0040] Since the embodiments described in this disclosure are intended to clearly explain the spirit of the present disclosure to a person having ordinary skill in the art to which the present disclosure pertains, the present disclosure is not limited to the embodiments described in this disclosure, and the scope of the present disclosure should be interpreted to include modified or altered examples that do not depart from the spirit of the present disclosure.

[0041] The terms used in this disclosure have been selected from widely used, common terms, taking into account the functions of this disclosure. However, these terms may vary depending on the intentions of those skilled in the art, customs, or the emergence of new technologies. However, if a specific term is defined and used with an arbitrary meaning, the meaning of that term will be described separately. Therefore, the terms used in this disclosure should be interpreted based on the actual meaning of the term and the overall content of this disclosure, rather than simply the name of the term.

[0042] The drawings of the present disclosure are intended to facilitate explanation of the present disclosure, and shapes depicted in the drawings may be exaggerated as necessary to aid understanding of the present disclosure, and thus the present disclosure is not limited by the drawings.

[0043] In the present disclosure, if a detailed description of the configuration or function of a notice related to the present disclosure is deemed to obscure the gist of the present disclosure, a detailed description thereof will be omitted as necessary. Furthermore, unless otherwise specified, numbers (e.g., "first," "second," etc.) used in the description of the present disclosure are merely identifiers used to distinguish one component from another.

[0044]

[0045] According to the present disclosure, a monitoring device can be provided. The monitoring device can transmit radio waves into a chamber of process equipment, such as semiconductor process equipment or display process equipment, and receive radio waves reflected from the chamber, thereby monitoring the geometrical state and / or electrical characteristics of the chamber. More specifically, the chamber includes a plurality of parts, such as a lower electrode on which a substrate, such as a wafer, is placed, an upper electrode facing the lower electrode, pins for supporting the substrate, and a baffle. The geometrical state of the chamber can be defined by a combination of the geometrical states of each part, such as the position and shape.

[0046] Alternatively, for example, as plasma is generated within the chamber, the load within the chamber may change, or as a process cycle progresses within the chamber, the environment within the chamber may change according to each process. In this case, a monitoring device can monitor the electrical characteristics according to the changing load or environment within the chamber.

[0047] Even when the same radio wave is transmitted into the chamber, the received radio wave may differ depending on the geometrical state of the chamber. For example, if the geometrical state of some parts is changed, such as by changing the position or shape of some parts, even when the same radio wave is transmitted into the chamber, the radio wave received after the change may differ from the radio wave received before the change. In other words, the way the radio wave is reflected inside the chamber may vary depending on the geometrical state of the chamber. In other words, the received reflected wave may reflect the geometrical state of the chamber.

[0048] Through this, the monitoring device can monitor chambers where no process is in progress (as well as chambers where a process is in progress). For example, the monitoring device can monitor chambers where no process is in progress, such as by monitoring the geometrical condition inside the chamber to determine the assembly status of process equipment during manufacturing of process equipment or to determine the results after preventive maintenance (PM) of process equipment. For another example, the monitoring device can monitor chambers where a process is in progress, such as by monitoring the geometrical condition inside the chamber to check whether the process is progressing normally at each process step or to predict the timing of preventive maintenance.

[0049] The monitoring target of the device and method for monitoring the state of the chamber disclosed by the present disclosure is, as described above, whether the equipment has been properly assembled before the process begins, or whether there is any abnormality in the state of the equipment between and during the process. That is, the monitoring target of the device and method disclosed by the present disclosure may include the geometric shape of each part inside the chamber, a level of wear-out of each part, a relative positional relationship between each part, unnecessary deposition of by-products generated by the process for the target substrate through the chamber on the inner wall of the chamber and on the surface of each part inside the chamber, etc. In addition, the monitoring target of the device and method disclosed by the present disclosure may include the state of materials and energy supplied inside the chamber to be monitored (for example, active species, etching gas, inert gas, RF power, or plasma thereof introduced into the chamber to perform a necessary process, etc.).

[0050]

[0051] In a method for monitoring a state of a chamber by a chamber monitoring device according to an embodiment of the present disclosure, - wherein a process cycle including a plurality of processes occurs at least once within the chamber - the chamber monitoring device obtains reference state data for monitoring the state of the chamber, - wherein the reference state data is first state data of the chamber measured in an idle section of the chamber or second state data of the chamber measured in a section corresponding to any one of the plurality of processes - transmitting a radio wave of a specific frequency range into the chamber M times and receiving a radio wave reflected from the inside of the chamber M times - wherein M is a natural number greater than or equal to 2 -; The method may include measuring the state of the chamber based on the M received radio waves to generate a plurality of third state data, calculating similarities between each of the plurality of third state data and the reference state data, obtaining a plurality of state similarity data using the similarities, matching the plurality of state similarity data with process section data for sections in which each of the plurality of processes is performed, thereby obtaining a plurality of state similarity patterns configured in units of process cycles, and determining whether the chamber is operating normally or predicting an abnormal operation time of the chamber using two or more state similarity patterns among the plurality of state similarity patterns.

[0052] At this time, predicting the abnormal operation point of the chamber; may include: selecting two or more state similarity data from among the plurality of state similarity data, wherein one state similarity data corresponding to the first process is selected from each of the plurality of state similarity patterns; determining a function for predicting the abnormal operation based on the two or more state similarity data, wherein the function is determined based on a time variable; and predicting the first time point at which the result value of the function calculated based on the time variable becomes lower than a preset threshold similarity as the abnormal operation point of the chamber.

[0053]

[0054] Additionally, the chamber monitoring device may repeatedly predict the abnormal operation point of the chamber each time an additional state similarity pattern is acquired.

[0055] In addition, the chamber monitoring device may determine whether the chamber is operating normally by selecting a first state similarity pattern that is most recently acquired among the plurality of state similarity patterns, selecting a second state similarity pattern that is different from the first state similarity pattern among the plurality of state similarity patterns, calculating difference values ​​between state similarity data of the first state similarity pattern and state similarity data of the second state similarity pattern according to each process and corresponding time point within each process, and determining that the chamber is operating abnormally in a process corresponding to a difference value that exceeds a threshold similarity difference value among the calculated difference values.

[0056] In addition, the chamber monitoring device may further include determining a threshold similarity difference value based on a gap between the first state similarity pattern and the second state similarity pattern among the plurality of state similarity patterns to determine whether the chamber is operating normally.

[0057] In addition, the chamber monitoring device may calculate difference values ​​between state similarity data of the first state similarity pattern and state similarity data of the second state similarity pattern, including correcting the state similarity data of the first state similarity pattern based on the state similarity data of the second state similarity pattern, and calculating difference values ​​between the corrected state similarity data of the first state similarity pattern and the state similarity data of the second state similarity pattern.

[0058] In addition, the chamber monitoring device may correct the state similarity data of the first state similarity pattern, including: selecting two or more state similarity data from among the plurality of state similarity data, wherein one state similarity data corresponding to the first process is selected from each of the plurality of state similarity patterns; determining a function for predicting the abnormal operation based on the two or more state similarity data, wherein the function is determined based on a time variable; and correcting the state similarity data of the first state similarity pattern based on the state similarity data of the second state similarity pattern and the function.

[0059] Additionally, the second state similarity pattern may be the state similarity pattern that is acquired first among the plurality of state similarity patterns.

[0060] In addition, the chamber monitoring device may obtain the reference state data by transmitting radio waves of the specific frequency range into the chamber N times in an idle section of the chamber or a section corresponding to any one of the plurality of processes, receiving radio waves reflected inside the chamber N times, wherein N is a natural number; measuring the state of the chamber N times based on the radio waves received N times to generate a plurality of fourth state data, and obtaining the reference state data based on the plurality of fourth state data.

[0061] In a chamber monitoring device for monitoring the state of a chamber according to an embodiment of the present disclosure, - wherein a process cycle including a plurality of processes occurs at least once within the chamber - the chamber monitoring device includes: an antenna for transmitting radio waves of a specific frequency range into the interior of the chamber and receiving radio waves reflected from the interior of the chamber; a bracket for fixing the antenna to the exterior of the chamber; a signal processing unit for applying an electric signal to the antenna and obtaining an electric signal from the antenna; a control unit for controlling the signal processing unit and generating state data of the chamber based on the electric signal obtained by the signal processing unit;

[0062] The control unit obtains reference state data for monitoring the state of the chamber, wherein the reference state data is first state data of the chamber measured in an IDLE section of the chamber or second state data of the chamber measured in a section corresponding to any one of the plurality of processes; The signal processing unit is controlled to obtain an electric signal according to the M received electric signal by causing the antenna to transmit the radio wave M times and receive the radio wave reflected inside the chamber M times, wherein M is a natural number greater than or equal to 2, to measure the state of the chamber based on the obtained electric signal to generate a plurality of third state data, to calculate similarities between each of the plurality of third state data and the reference state data, to obtain a plurality of state similarity data using the similarities, and to obtain a plurality of state similarity patterns configured in units of process cycles by matching the plurality of state similarity data with process section data for sections in which each of the plurality of processes is performed, and among the plurality of state similarity patterns, two or more state similarity patterns can be used to determine whether the chamber is operating normally or to predict the time point of abnormal operation of the chamber.

[0063] At this time, the control unit selects two or more state similarity data from among the plurality of state similarity data, - wherein the control unit selects one state similarity data corresponding to the first process from each of the plurality of state similarity patterns -; determines a function for predicting the abnormal operation time based on the two or more state similarity data, - the function is determined based on a time variable -, and can predict a time when a result value of the function calculated based on the time variable becomes less than a preset threshold similarity as the abnormal operation time of the chamber.

[0064] Additionally, the control unit can repeatedly predict the abnormal operation point of the chamber each time an additional state similarity pattern is acquired.

[0065] In addition, the control unit may select a first state similarity pattern that is most recently acquired among the plurality of state similarity patterns, select a second state similarity pattern that is different from the first state similarity pattern among the plurality of state similarity patterns, calculate difference values ​​between state similarity data of the first state similarity pattern and state similarity data of the second state similarity pattern according to each process and corresponding time point within each process, and determine that the chamber is operating abnormally in a process corresponding to a difference value that exceeds a threshold similarity difference value among the calculated difference values.

[0066] Additionally, the control unit can determine the threshold similarity difference value according to the gap between the first state similarity pattern and the second state similarity pattern among the plurality of state similarity patterns.

[0067] In addition, the control unit can correct the state similarity data of the first state similarity pattern based on the state similarity data of the second state similarity pattern, and calculate difference values ​​between the corrected state similarity data of the first state similarity pattern and the state similarity data of the second state similarity pattern.

[0068] In addition, the control unit may select two or more state similarity data from among the plurality of state similarity data, wherein one state similarity data corresponding to the first process is selected from each of the plurality of state similarity patterns, determine a function for predicting the abnormal operation time based on the two or more state similarity data, wherein the function is determined based on a time variable, and correct the state similarity data of the first state similarity pattern based on the state similarity data of the second state similarity pattern and the function.

[0069] Additionally, the second state similarity pattern may be the state similarity pattern that is acquired first among the plurality of state similarity patterns.

[0070] In addition, the control unit may transmit radio waves of the specific frequency range into the chamber N times in an idle section of the chamber or a section corresponding to any one of the plurality of processes, receive radio waves reflected inside the chamber N times - where N is a natural number -, measure the state of the chamber N times based on the radio waves received N times, generate a plurality of fourth state data, and obtain the reference state data based on the plurality of fourth state data.

[0071]

[0072] In a method for determining an error correction function for correcting a measurement error between parameter sets caused by attachment or detachment of chamber monitoring devices using P chamber monitoring devices (P is a natural number greater than or equal to 1) according to an embodiment of the present disclosure, wherein the parameter set is one of (i) an S-parameter set, (ii) an H-parameter set, (iii) a Y-parameter set, (iv) a Z-parameter set, and (v) a parameter set derived from one of the S-parameter, the H-parameter, the Y-parameter, and the Z-parameter, wherein, after attaching a first chamber monitoring device among the P chamber monitoring devices within a first region outside a chamber in a first process state, transmitting a radio wave of a specific frequency range into the interior of the chamber to generate a reference parameter set of the chamber, and repeatedly performing a first operation N times (N is a natural number greater than or equal to 2) for each of the P chamber monitoring devices to generate L first parameter sets, wherein, L is a value obtained by multiplying P by N, wherein, a plurality of The method may include determining a first function including first error correction parameters, and repeating a second operation until difference information between an output parameter set obtained by substituting any one of the L first parameter sets into the first function and the reference parameter set is included within a predetermined range, thereby updating the plurality of first error correction parameters into a plurality of second error correction parameters, and determining a second learning model including the plurality of second error correction parameters as the error correction function.

[0073] At this time, the second operation may include obtaining a first output parameter set by substituting any one of the L first parameter sets into the first function, wherein the any one of the first parameter sets is a first parameter set that has not been previously substituted into the first function; and obtaining first difference information between the first output parameter set and the reference parameter set, and modifying the plurality of first error correction parameters based on the first difference information.

[0074] Additionally, the first operation may include detaching the chamber monitoring device from the chamber, reattaching it within the first region, transmitting radio waves of the specific frequency range into the interior of the chamber, wherein the chamber is in the first process state; receiving radio waves reflected from the interior of the chamber, and generating a first parameter set based on the received radio waves.

[0075] Additionally, the first process state may be an IDLE state of the chamber.

[0076] Additionally, the first process state may be a state in which any one of a plurality of processes that can be performed within the chamber is performed.

[0077] Additionally, the first region may be a region corresponding to a viewport formed in the chamber.

[0078] In addition, the chamber monitoring device may generate the reference parameter set by transmitting the radio wave M times (M is a natural number greater than or equal to 1) without detaching the chamber monitoring device from the first region, receiving the radio wave reflected inside the chamber M times, obtaining M sets of second parameters based on each of the M received radio waves, and generating the reference parameter set of the chamber using the M sets of second parameters.

[0079] Additionally, the first learning model may be based on a single layer perceptron.

[0080] Additionally, the first learning model may be based on a multi-layer perceptron.

[0081] In a method for determining an error correction function for correcting a measurement error between parameter sets corresponding to each of a plurality of chambers using P chamber monitoring devices (P is a natural number greater than or equal to 1) according to an embodiment of the present disclosure, wherein the parameter set is one of (i) an S-parameter set, (ii) an H-parameter set, (iii) a Y-parameter set, (iv) a Z-parameter set, and (v) a parameter set derived from one of the S-parameter, the H-parameter, the Y-parameter, and the Z-parameter, wherein, among the P chamber monitoring devices, a first chamber monitoring device is attached within a first region outside a first chamber in a first process state, and then a radio wave of a specific frequency range is transmitted into the first chamber to generate a reference parameter set of the first chamber, and a first operation is repeatedly performed N times (N is a natural number greater than or equal to 2) for each of the P chamber monitoring devices to generate L first parameter sets, wherein, L is a value obtained by multiplying P by N. -, determining a first function including a plurality of first error correction parameters, and repeating a second operation until difference information between an output parameter set obtained by substituting any one of the L first parameter sets into the first function and the reference parameter set is included within a predetermined range, thereby updating the plurality of first error correction parameters to a plurality of second error correction parameters, and determining a second learning model including the plurality of second error correction parameters as the error correction function.

[0082] At this time, the second operation may include obtaining a first output parameter set by substituting any one of the L first parameter sets into the first function, wherein the any one of the first parameter sets is a first parameter set that has not been previously substituted into the first function; obtaining first difference information between the first output parameter set and the reference parameter set, and modifying the plurality of first error correction parameters based on the first difference information.

[0083] Additionally, the first operation may include attaching a chamber monitoring device within a first region of a second chamber selected from among the plurality of chambers, transmitting radio waves of the specific frequency range into the interior of the second chamber, wherein the second chamber is in the first process state, and the second chamber is a chamber not previously selected from among the plurality of chambers; receiving radio waves reflected from the interior of the second chamber, and generating a first parameter set based on the received radio waves.

[0084] Additionally, the first process state may be an IDLE state.

[0085] Additionally, the first process state may be a state in which any one process among a plurality of processes that can be performed in each of the plurality of chambers is performed.

[0086] Additionally, the first region may be a region corresponding to a viewport of each of the plurality of chambers.

[0087] In addition, the chamber monitoring device may generate the reference parameter set by transmitting the radio waves into the interior of the first chamber M times (M is a natural number greater than or equal to 1) without detaching the chamber monitoring device from the first chamber, receiving radio waves reflected from the interior of the first chamber M times, obtaining M sets of second parameters based on each of the M received radio waves, and generating the reference parameter set of the chamber using the M sets of second parameters.

[0088] Additionally, the first learning model may be based on a single layer perceptron.

[0089] Additionally, the first learning model may be based on a multi-layer perceptron.

[0090] Additionally, the first chamber may be included in the plurality of chambers.

[0091] Additionally, the first chamber may not be included in the plurality of chambers.

[0092] In one embodiment of the present disclosure, a method for determining an error correction function for correcting a measurement error between parameter sets caused by attachment or detachment of chamber monitoring devices using P chamber monitoring devices (P is a natural number greater than or equal to 1) and a measurement error between parameter sets corresponding to each of Q chambers (Q is a natural number greater than or equal to 2), wherein the parameter sets are one of (i) an S-parameter set, (ii) an H-parameter set, (iii) a Y-parameter set, (iv) a Z-parameter set, and (v) a parameter set derived from one of the S-parameter, the H-parameter, the Y-parameter, and the Z-parameter, wherein, after attaching a first chamber monitoring device among the P chamber monitoring devices within a first region outside a first chamber of a first process state, a radio wave of a specific frequency range is transmitted into the first chamber, thereby generating a reference parameter set of the first chamber; A first operation is repeatedly performed Q times (Q is a natural number greater than or equal to 2) for each of P chamber monitoring devices to generate L sets of first parameters, wherein the first operation is performed N times for each of the Q chambers, and L is a product of P, Q, and N; A first learning model including a plurality of first error correction parameters is determined; A second operation is repeated until difference information between an output parameter set obtained by inputting any one of the L sets of first parameters into the first learning model and the reference parameter set is included within a predetermined range, thereby updating the plurality of first error correction parameters into a plurality of second error correction parameters; And A second learning model including the plurality of second error correction parameters is determined as the error correction function;and the second operation may include: inputting one of the L first parameter sets into the first learning model to obtain a first output parameter set, wherein the one of the first parameter sets is a first parameter set that has not been previously input into the first learning model; and obtaining first difference information between the first output parameter set and the reference parameter set, and modifying the plurality of first error correction parameters based on the first difference information; and the third operation may include: detaching a chamber monitoring device from the chamber, reattaching it within the first region, and transmitting radio waves of the specific frequency range into the interior of the chamber, wherein the chamber is in the first process state; and receiving radio waves reflected from the interior of the chamber, and generating a first parameter set based on the received radio waves.

[0093] In an embodiment of the present disclosure, a method for a chamber monitoring device to monitor a first chamber using a parameter set, wherein the parameter set is one of (i) an S-parameter set, (ii) an H-parameter set, (iii) a Y-parameter set, (iv) a Z-parameter set and (v) a parameter set derived from one of the S-parameters, the H-parameters, the Y-parameters and the Z-parameters, wherein the chamber monitoring device obtains a reference parameter set for monitoring the first chamber, attaches the chamber monitoring device within a first area outside the first chamber, and transmits radio waves of a specific frequency range into the interior of the first chamber, wherein the first chamber is in one of a plurality of process states that can be performed within the first chamber and an idle (ILDE) state, receives radio waves reflected from the interior of the first chamber, generates a first parameter set based on the received radio waves, and substitutes the first parameter set into an error correction function to generate a second parameter set in which the first parameter set is corrected. It may include obtaining, comparing the second parameter set with the reference parameter set, calculating a similarity for each of the second parameters included in the second parameter set, and obtaining monitoring information of the first chamber based on the similarity for each of the second parameters.

[0094] At this time, the error correction function may be generated by repeatedly attaching and detaching at least one chamber monitoring device to and from each of at least one chamber at least once to generate a plurality of second parameter sets for the at least one chamber, sequentially using each of the plurality of second parameter sets to modify error correction parameters included in the error correction function, and inputting one of the plurality of second parameter sets into the error correction function to modify the error correction parameters until an output parameter set satisfies a certain condition.

[0095] Additionally, the first chamber may be included in at least one chamber.

[0096] Additionally, the first chamber may not be included in the at least one chamber.

[0097] Additionally, the monitoring information may be used to determine whether the first chamber is operating normally or to predict the point in time when the first chamber is operating abnormally.

[0098] Additionally, the above-described condition may be that the difference information between the output parameter set and the reference parameter set is included within a predetermined range.

[0099]

[0100] In one embodiment of the present disclosure, a method for monitoring a state of a chamber by a chamber monitoring device comprises: obtaining process section data for identifying a process section of each of a plurality of processes performed within the chamber, wherein the plurality of processes include a first process and a second process different from the first process; confirming a process currently being performed within the chamber based on the process section data; obtaining first reference state data corresponding to the first process from a storage unit based on the confirmed process being the first process, wherein the first reference state data is state data of the chamber measured in advance in the first process; transmitting a radio wave of a specific frequency range into the chamber during a first process section during which the first process is performed, and receiving a radio wave reflected from the inside of the chamber; measuring a state of the chamber based on the radio wave received within the first process section, thereby obtaining first state data; calculating a first similarity using the first state data and the first reference state data; Based on the first similarity, determining whether the chamber is operating normally in the first process; If it is determined that the first process has ended and a new process has started in the chamber based on the process section data, confirming the newly started second process; Obtaining second reference state data corresponding to the second process from the storage unit, wherein the second reference state data is state data of the chamber measured in advance in the second process; Transmitting radio waves of the specific frequency range into the interior of the chamber during the second process section and receiving radio waves reflected from the interior of the chamber; Measuring the state of the chamber based on the radio waves received within the second process section to obtain second state data;It may include calculating a second similarity using the second reference state data instead of the second state data and the first reference state data; and determining whether the chamber is operating normally in the second process based on the second similarity.

[0101] Here, determining whether the chamber is operating normally in the first process may include determining that the chamber is operating normally in the first process based on the first similarity exceeding a preset threshold similarity; and determining that the chamber is operating abnormally in the first process based on the first similarity being lower than or equal to the preset threshold similarity.

[0102] In addition, determining whether the chamber is operating normally in the second process may include determining that the chamber is operating normally in the second process based on the second similarity exceeding a preset threshold similarity; and determining that the chamber is operating abnormally in the second process based on the second similarity being lower than or equal to the preset threshold similarity.

[0103] In addition, receiving a radio wave reflected from inside the chamber during the first process section; may include transmitting a radio wave of the specific frequency range into the inside of the chamber M times during the first process section and receiving a radio wave reflected from inside the chamber M times, where M is a natural number; obtaining the first state data; may include measuring the state of the chamber based on the radio wave received M times within the first process section, thereby obtaining M first state data; calculating the first similarity; may include calculating M first similarities using the first reference state data and each of the M first state data; and determining whether the chamber is operating normally in the first process; may include determining that the chamber is operating abnormally in the first process based on whether at least one of the M first similarities is lower than or equal to a preset threshold similarity.

[0104] In addition, receiving a radio wave reflected from inside the chamber during the second process section; may include transmitting a radio wave of the specific frequency range into the inside of the chamber M times during the second process section and receiving a radio wave reflected from inside the chamber M times, where M is a natural number; obtaining the second state data; may include measuring the state of the chamber based on the radio wave received M times within the second process section, thereby obtaining M pieces of second state data; calculating the second similarity; may include calculating M pieces of second similarity using the second reference state data and each of the M pieces of second state data; and determining whether the chamber is operating normally in the second process; may include determining that the chamber is operating abnormally in the second process based on whether at least one of the M pieces of second similarity is lower than or equal to a preset threshold similarity.

[0105] In addition, the method comprises: transmitting radio waves of the specific frequency range into the chamber Q or more times in each of a fourth process section in which the first process is performed and a fifth process section in which the second process is performed, and receiving radio waves reflected inside the chamber Q or more times, wherein Q is a natural number, the fourth process section is a process section in which the first process is performed before the first process section, and the fifth process section is a process section in which the second process is performed before the second process section; measuring the state of the chamber based on the radio waves received Q or more times within the fourth process section, thereby obtaining N fourth state data, and obtaining P fifth state data of the state of the chamber based on the radio waves received Q or more times within the fifth process section, wherein N and P are each a natural number, and N is the same as or different from P; obtaining the first reference state data based on the N fourth state data, and obtaining the second reference state data based on the P fifth state data; And it may further include storing the first reference state data and the second reference state data in the storage unit.

[0106] In one embodiment of the present disclosure, a chamber monitoring device for monitoring a state of a chamber includes: an antenna for transmitting a radio wave of a specific frequency range into the interior of the chamber and receiving a radio wave reflected from the interior of the chamber; a signal processing unit for applying an electric signal to the antenna and obtaining an electric signal from the antenna; a storage unit for storing (i) reference state data of the chamber and (ii) process section data for identifying a process section of each of a plurality of processes performed within the chamber, wherein the plurality of processes include a first process and a second process different from the first process; and a control unit for controlling the signal processing unit and generating state data of the chamber based on the electric signal obtained by the signal processing unit., wherein the control unit, based on the process section data, confirms the process currently being performed in the chamber, and, based on the confirmed process being the first process, obtains first reference state data corresponding to the first process from the storage unit, wherein the first reference state data is state data of the chamber measured in advance in the first process, transmits the radio wave into the interior of the chamber during the first process section in which the first process is performed, and controls the signal processing unit so that the radio wave reflected from the interior of the chamber is received, measures the state of the chamber based on the radio wave received within the first process section, and obtains first state data, calculates a first similarity using the first state data and the first reference state data, and determines whether the chamber is operating normally in the first process based on the first similarity, and, based on the process section data, when it is determined that the first process is terminated and a new process is started in the chamber, confirms the newly started second process, and obtains second reference state data corresponding to the second process from the storage unit, wherein the second reference The status data may be status data of the chamber previously measured in the second process, - transmitting the radio waves into the interior of the chamber during a second process section in which the second process is performed, controlling the signal processing unit to receive radio waves reflected from the interior of the chamber, measuring the status of the chamber based on the radio waves received within the second process section to obtain second status data, calculating a second similarity using the second reference status data instead of the second status data and the first reference status data, and determining whether the chamber is operating normally in the second process based on the second similarity.;

[0107] Here, the control unit can determine that the chamber is operating normally in the first process based on the first similarity exceeding a preset threshold similarity, and can determine that the chamber is operating abnormally in the first process based on the first similarity being less than or equal to the preset threshold similarity.

[0108] In addition, the control unit may determine that the chamber is operating normally in the second process based on the second similarity exceeding a preset threshold similarity, and may determine that the chamber is operating abnormally in the second process based on the second similarity being lower than or equal to the preset threshold similarity.

[0109] In addition, the control unit controls the signal processing unit to transmit the radio wave into the chamber M times during the first process section, and to receive the radio wave reflected inside the chamber M times - where M is a natural number -, to measure the state of the chamber based on the radio wave received M times within the first process section, to obtain M first state data, and to calculate M first similarities using the first reference state data and each of the M first state data, and to determine that the chamber operates abnormally in the first process based on the fact that at least one of the M first similarities is lower than a preset threshold similarity.

[0110] In addition, the control unit controls the signal processing unit to transmit the radio wave into the chamber M times during the second process section, and to receive the radio wave reflected inside the chamber M times - where M is a natural number -, to measure the state of the chamber based on the radio wave received M times within the second process section, to obtain M pieces of second state data, and to calculate M pieces of second similarity using the second reference state data and each of the M pieces of second state data, and to determine that the chamber operates abnormally in the second process based on the fact that at least one of the M pieces of second similarity is lower than a preset threshold similarity.

[0111] In addition, the control unit controls the signal processing unit so that the radio waves are transmitted into the chamber Q or more times in each of the fourth process section in which the first process is performed and the fifth process section in which the second process is performed, and the radio waves reflected inside the chamber are received Q or more times, wherein Q is a natural number, the fourth process section is a process section in which the first process is performed before the first process section, and the fifth process section is a process section in which the second process is performed before the second process section, and based on the radio waves received Q or more times within the fourth process section, the state of the chamber is measured to obtain N fourth state data, and based on the radio waves received Q or more times within the fifth process section, the state of the chamber is obtained P fifth state data, wherein each of N and P is a natural number, and N is the same as or different from P, the first reference state data is obtained based on the N fourth state data, and the second reference state data is obtained based on the P fifth state data, and the first reference state data is Data and the second reference state data can be stored in the storage unit.

[0112]

[0113] Meanwhile, in describing embodiments of the present disclosure, terms used may be defined as follows.

[0114] - Process State: Multiple processes are performed within the chamber according to the process cycle. In addition, a state in which no process is performed within the chamber may be referred to as idle. That is, the process state according to the embodiment of the present disclosure may mean that among the multiple processes or the idle state, any one of the multiple processes is being performed within the chamber, or no process is being performed. For example, the idle state may mean a state in which no process is performed within the chamber. As another example, the first process state (or the second to Nth process states) may mean a state in which the first process state (or the second to Nth process states) included within the multiple processes is performed within the chamber.

[0115] - IDLE State: This may be a process state in which no process is in progress in the chamber, such as when assembly of process equipment is completed during equipment manufacturing or when preventive maintenance of process equipment is completed. It may also refer to a state in which no process is in progress within the chamber after one process cycle is completed and before the next process cycle begins. In the present disclosure, the IDLE state may also be understood and interpreted as a process state.

[0116] - Process Cycle: Multiple processes are performed within the chamber according to the process cycle. Meanwhile, an idle period during which the chamber is idle before the multiple processes are performed may be included. In other words, the process cycle in the present disclosure may be a concept encompassing both the idle period and the period during which multiple processes are performed. For example, if the multiple processes are comprised of the first through Nth processes, the process cycle may mean "an idle period + a period during which the first through Nth processes are performed."

[0117] - Status data: Status data may be a set of parameters reflecting the geometric and electrical status inside the chamber. For example, parameters that may be utilized as status data may be i) S-parameters, ii) H-parameters, iii) Y-parameters, iv) Z-parameters, v) a combination of at least two parameters selected from S-parameters, H-parameters, Y-parameters and Z-parameters, vi) parameters derived from S-parameters, H-parameters, Y-parameters and Z-parameters, or vii) parameters derived from a combination of at least two parameters selected from S-parameters, H-parameters, Y-parameters and Z-parameters. The monitoring device may measure the frequency characteristics inside the chamber to derive a set of parameters, and may use the sets of parameters to determine what status the chamber is in (e.g., whether it is in a normal operating state).

[0118] - Normal / Abnormal Operation: This classification distinguishes between normal and abnormal operation of the chamber. Here, abnormal operation may refer to at least one of chamber failure, aging, replacement, replacement of parts, and occurrence of an abnormality. For example, whether a chamber is operating normally means determining whether the chamber is currently operating normally or is operating abnormally (e.g., at least one of chamber failure, aging, replacement, replacement of parts, and occurrence of an abnormality).

[0119] Additionally, predicting the point of abnormal chamber operation can mean predicting the point at which the chamber changes from normal operation to abnormal operation. For example, predicting the point at which abnormal operation is required means predicting the point at which chamber replacement, chamber component replacement, or chamber inspection is necessary.

[0120]

[0121] Fig. 1 is a block diagram of a monitoring device according to one embodiment. Referring to Fig. 1, the monitoring device (100) may include an antenna (110), a signal processing unit (120), a communication unit (130), a control unit (140), and a storage unit (150).

[0122]

[0123] The monitoring device (100) can transmit and receive radio waves through the antenna (110). The antenna (110) can receive an electrical signal and transmit the radio waves. The antenna (110) can receive the radio waves and convert them into electrical signals. The monitoring device (100) may include one antenna (110). Alternatively, the monitoring device (100) may include two or more antennas (110). In this case, some of the two or more antennas (110) may be for transmitting radio waves, and the rest may be for receiving radio waves. Alternatively, each of the two or more antennas (110) may be for transmitting and receiving radio waves at different locations.

[0124] The monitoring device (100) can apply an electric signal to the antenna (110) through the signal processing unit (120) and obtain an electric signal from the antenna (110). The signal processing unit (120) can apply an electric signal of a specific frequency range to the antenna (110). The signal processing unit (120) can obtain an electric signal of a specific frequency range from the antenna (110). The signal processing unit (120) can control the frequency and amplitude of a radio wave to be output through the antenna (110) and sense information about the frequency and amplitude of a radio wave received through the antenna (110). More specific details about this will be described later.

[0125] The monitoring device (100) can generate status data through a signal processing unit (120). More specific details about this will be described later.

[0126] The monitoring device (100) can communicate with the outside world through the communication unit (130). For example, the communication unit (130) can transmit status data, monitoring information, etc. to the outside world.

[0127] The communication unit (130) can perform wired or wireless communication. The communication unit (130) may be, for example, a wired / wireless Local Area Network (LAN) module, a WAN module, an Ethernet module, a Bluetooth module, a Zigbee module, a USB (Universal Serial Bus) module, an IEEE 1394 module, a Wi-Fi module, an EtherCAT module, a DeviceNet module, or a combination thereof, but is not limited thereto.

[0128] The monitoring device (100) can generate monitoring information through the control unit (140). The control unit (140) can generate monitoring information based on status data. More specific details regarding this will be described later.

[0129] The control unit (140) may be implemented as a computer or a similar device using hardware, software, or a combination thereof. In terms of hardware, the control unit (140) may be one or more processors. Alternatively, the control unit (140) may be provided as processors that are physically separated and cooperate through communication. Examples of the control unit (140) include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), a state machine, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a radio-frequency integrated circuit (RFIC), or a combination thereof. In terms of software, the control unit (140) may be provided in the form of a program that drives the hardware control unit (140).

[0130] Meanwhile, the operations of all embodiments described in the present disclosure may be controlled and executed by the control unit (140). For example, operations including at least one of Embodiments 1 to 5 described in the present disclosure may be executed under the control of the control unit (140). In other words, the control unit (140) may control the overall operation of the monitoring device (100).

[0131] The monitoring device (100) can store various data and programs in the storage unit (150). For example, the storage unit (150) can store status data generated by the signal processing unit (120). As another example, the storage unit (150) can store monitoring information generated by the control unit (140).

[0132] The storage unit (150) may be, for example, a nonvolatile semiconductor memory, a hard disk, a flash memory, an SSD (Solid State Drive), a RAM (Random Access Memory), a ROM (Read Only Memory), an EEPROM (Electrically Erasable Programmable Read-Only Memory), or other types of (tangible) nonvolatile recording media or a combination thereof, but is not limited thereto.

[0133] The monitoring device (100) may further include a fixing member (160). The fixing member (160) may fix the antenna (110) to a position around the chamber. For example, the fixing member (160) may fix the antenna (110) so that the antenna (110) and the chamber have a predetermined distance. The predetermined distance may be, for example, 1 mm, 3 mm, 5 mm, 7 mm, or 1 cm, but is not limited thereto. For another example, the fixing member (160) may fix the antenna (110) so that the antenna (110) and the chamber come into contact with each other. The fixing member (160) may be, for example, a bracket, but is not limited thereto.

[0134] The monitoring device (100) may further include an output unit (170). For example, the output unit (170) may be a display. The monitoring device (100) may display monitoring information through the display. As another example, the output unit (170) may be a speaker. The monitoring device (100) may output an alarm through the speaker.

[0135] The monitoring device (100) may further include an electromagnetic wave shield (180). The electromagnetic wave shield (180) prevents or reduces the influence of external electromagnetic waves on the antenna (110). For example, the electromagnetic wave shield (180) may be placed outside the chamber to surround the antenna (110). The electromagnetic wave shield (180) may be provided with various materials capable of shielding electromagnetic waves.

[0136] The monitoring device (100) may be provided in an integrated form. For example, the monitoring device (100) may be provided in an integrated form with an antenna (110), a signal processing unit (120), a communication unit (130), a control unit (140), a storage unit (150), and a fixing unit (160).

[0137] Alternatively, the monitoring device (100) may be provided in a separate form. For example, the monitoring device (100) may be provided in a form in which the antenna (110) and the remaining components are separated. A description of the integrated monitoring device and the separate monitoring device will be provided later with reference to FIG. 3.

[0138] Not all of the components illustrated in FIG. 1 are essential components of the monitoring device, and at least some of the components illustrated in FIG. 1 may be omitted. In addition, the monitoring device may additionally include components not illustrated in FIG. 1.

[0139] The monitoring device can be implemented using a network analyzer such as a vector network analyzer (VNA).

[0140] For example, the signal processing unit, communication unit, control unit, and storage unit can be implemented using a network analyzer.

[0141] For another example, the signal processing unit may be implemented using a network analyzer. In this case, the monitoring device may include separate devices for implementing a communication unit, control unit, and storage unit in addition to the network analyzer.

[0142]

[0143] Monitoring devices can be installed in the chamber.

[0144] Figures 2 and 3 are schematic diagrams of a monitoring device installed in a chamber according to one embodiment.

[0145] Referring to FIG. 2, the antenna may be located outside the chamber (10). For example, the antenna may be located outside the viewport (11) formed in the chamber (10).

[0146]

[0147] The antenna may be installed outside the chamber (10) by a fixing member. For example, the antenna may be installed outside the chamber (10) by a fixing member so as to have a predetermined distance from the chamber (10). For another example, the antenna may be installed outside the chamber (10) by a fixing member so as to be in contact with the chamber (10).

[0148] Although not shown, the antenna may be located inside the chamber. For example, the antenna may be located inside a viewport formed in the chamber. The antenna may be installed inside the chamber by a fixing member. However, if the antenna is located inside the chamber, problems such as contamination of the chamber by the antenna or contamination of the antenna during the process may occur, which may be disadvantageous compared to when the antenna is located outside the chamber.

[0149] Referring to FIG. 3, in the case of an integrated monitoring device (100), the monitoring device (100) may be used in a form in which it is mounted in a chamber (10). For example, this may mean that all components of the monitoring device (100) are created to be non-detachable, so that the entire monitoring device (100) is attached to the outside of the chamber.

[0150] In the case of a detachable monitoring device, the monitoring device may be used in a form where only some of its components are mounted in a chamber, and the remaining components are connected to the components mounted in the chamber via wires. For example, the monitoring device may be used in a form where the antenna is mounted in the chamber, and the remaining components are connected to the antenna via wires. The connecting wires may be, for example, coaxial cables. For example, in the case of a detachable monitoring device, only some of the components (e.g., the antenna) may be attached to the outside of the chamber, and the remaining components may be separated from some of the components and connected to some of the components via wires or wirelessly. In other words, in the case of a detachable monitoring device, it may mean that at least some of the components of the monitoring device are designed to be detachable, and each component is designed to be assembled depending on the purpose of use.

[0151] Here, "assembleable" may mean that some components are assembled so that they are directly connected to the rest of the components without connecting wires, such as coaxial cables. Furthermore, as described above, some components may be connected to the rest of the components via wires, or they may be connected wirelessly. For example, if some components are connected wirelessly to the rest of the components, each of the components may further include a wireless communication unit for wireless communication.

[0152] Although FIGS. 2 and 3 illustrate that the antenna is positioned relative to the viewport, the antenna may also be positioned relative to a region of the chamber that is permeable to radio waves, such as ceramic, in addition to the viewport. For example, a separate port for mounting the antenna may be formed in the chamber in addition to the existing viewport, and the antenna may be positioned relative to the separate port.

[0153] Additionally, although the monitoring device is described as including one antenna in FIGS. 2 and 3, as described above, the monitoring device may include two or more antennas. In this case, some of the two or more antennas may be mounted in one area of ​​the chamber, and the rest may be mounted in another area of ​​the chamber.

[0154]

[0155] Below, the chamber condition monitoring method is described in more detail.

[0156]

[0157] Monitoring devices can generate status data.

[0158] The status data may be related to radio waves incident from the antenna into the chamber (hereinafter, "incident waves") and radio waves reflected from the chamber and received by the antenna (hereinafter, "reflected waves"). The status data may be related to a voltage applied to the antenna and a voltage output from the antenna.

[0159] Status data can correspond to a specific frequency range. Status data can be generated for a specific frequency range.

[0160] State data can be defined using incident and reflected waves. For example, state data can be defined using the ratio of incident and reflected waves, such as the ratio of reflected waves to incident waves. State data can be, for example, but is not limited to, S-parameters, H-parameters, Y-parameters, Z-parameters, or parameters that can be calculated therefrom.

[0161] In some embodiments, the state data may be S-parameters.

[0162] In some other embodiments, the state data may be H-parameters.

[0163] In some other embodiments, the state data may be a Y-parameter.

[0164] In some other embodiments, the state data may be Z-parameters.

[0165] In some other embodiments, the state data may be a parameter derived from at least one selected from an S-parameter, an H-parameter, a Y-parameter, and a Z-parameter.

[0166] Additionally, in some other embodiments, the state data may be a combination of at least two parameters selected from among S-parameters, H-parameters, Y-parameters and Z-parameters.

[0167] Additionally, in some other embodiments, the state data may be a parameter derived from a combination of at least two parameters selected from among S-parameters, H-parameters, Y-parameters and Z-parameters.

[0168] That is, the state data may be i) an S-parameter, ii) an H-parameter, iii) a Y-parameter, iv) a Z-parameter, v) a combination of at least two parameters selected from an S-parameter, an H-parameter, a Y-parameter and a Z-parameter, vi) a parameter derived from an S-parameter, an H-parameter, a Y-parameter and a Z-parameter, or vii) a parameter derived from a combination of at least two parameters selected from an S-parameter, an H-parameter, a Y-parameter and a Z-parameter.

[0169] For example, state data can be represented as an n X 2 matrix as follows:

[0170] (1st frequency) (magnitude of reflected wave / magnitude of incident wave)

[0171] (second frequency) (magnitude of reflected wave / magnitude of incident wave)

[0172] ...

[0173] (nth frequency) (magnitude of reflected wave / magnitude of incident wave)

[0174] As another example, state data can be represented as an n X 3 matrix as follows:

[0175] (1st frequency) (magnitude of incident wave) (magnitude of reflected wave)

[0176] (second frequency) (magnitude of incident wave) (magnitude of reflected wave)

[0177] ...

[0178] (nth frequency) (magnitude of incident wave) (magnitude of reflected wave)

[0179] Here, n is the number of frequencies at which the status data is measured, the first frequency is the lower limit of the frequency at which the status data is measured, the nth frequency is the upper limit of the frequency at which the status data is measured, and the first to nth frequencies are the frequency ranges at which the status data is measured.

[0180] Meanwhile, when the monitoring device (100) measures status data, the status data can be measured by transmitting radio waves corresponding to each of the first to n-th frequencies at least once. For example, after measuring parameters while transmitting radio waves of the first frequency m times, the parameters can be measured while transmitting radio waves of the second frequency m times, and repeating this process. In this case, the parameters can be measured while transmitting radio waves of the n-th frequency m times. At this time, the intensity of the radio waves of the first to n-th frequencies during the m transmissions can be different or the same for each transmission.

[0181] As another example, parameters can be measured by transmitting once from the first frequency wave to the nth frequency wave, and then measuring the parameters by transmitting once again from the first frequency wave to the nth frequency wave, thereby measuring the parameters by transmitting m times from the first frequency wave to the nth frequency wave. In this case, the wave intensity of each transmitted frequency can be different or the same for each transmission.

[0182] Meanwhile, n here can be determined based on the resolution of the frequency to be measured, i.e., the interval between frequencies. For example, when measuring state data between 3 GHz and 8.5 GHz, if n is 55, frequency characteristics can be measured in 0.1 GHz increments. As another example, if n is 11, frequency characteristics can be measured in 0.5 GHz increments.

[0183] The status data may include parameter values ​​for a specific frequency range. In the above example, (magnitude of reflected wave / magnitude of incident wave), (magnitude of incident wave), (magnitude of reflected wave), etc. may be parameter values. The number of parameter values ​​included in the status data may be, for example, 500 or more, 1000 or more, 1500 or more, or 2000 or more, but is not limited thereto. The status data may include multiple state data belonging to a specific frequency range, and multiple peaks belonging to the specific frequency range. The number of peaks included in the status data may be, for example, 5, 10, 20, 50, or 100 or more, but is not limited thereto.

[0184] State data can correspond to a specific point in time. State data can be generated at a specific point in time. For example, a specific point in time could be a point in time when the process is not in progress, such as when assembly of process equipment is completed during equipment manufacturing or when preventive maintenance of process equipment is completed. Alternatively, a specific point in time could be a point in time when the process is in progress, such as at a specific process step.

[0185] Fig. 4 is a graph of the S11 parameter, which is an example of state data. Fig. 4 shows the S11 parameter in dB units for the frequency range from 3 GHz to 8.5 GHz. In Fig. 4, dozens of peaks can be seen, but the number of peaks may vary depending on the criteria for counting peaks. For example, in Fig. 4, peaks are used to distinguish characteristics of state data or to distinguish patterns of state data, and the criteria may vary depending on the purpose. For example, a peak may be defined as a value at which the radio intensity in dB in each frequency range is above a certain level, or may refer to a frequency at which the radio intensity of a previous frequency range increases or decreases by a certain level compared to the radio intensity of a current frequency range. In addition, a peak may refer to a frequency at which the radio intensity of a previous frequency range increases or decreases by a certain level compared to the radio intensity of a current frequency range (i.e., the slope of the graph) by a certain level.

[0186]

[0187] Status data can reflect the geometric state of the chamber. For example, if the geometric state of a part, such as its position or shape within the chamber, changes, the status data after the change may differ from the status data before the change. For another example, the status data when some parts are damaged may differ from the status data when they are not damaged. For another example, the status data when foreign matter is present within the chamber may differ from the status data when the foreign matter is not present.

[0188] Status data can also reflect the electrical status of the chamber. For example, since conductivity and permittivity vary depending on the material properties of the chamber or part, status data can reflect the electrical status of the chamber. For another example, if a polymer film or other material forms on the chamber's internal walls as the process progresses, the permittivity of the walls will change, and this can also reflect the electrical status of the chamber.

[0189] That is, the status data may change depending on the process status within the chamber as the process cycle repeats. For example, the status data measured in the first process state of the process cycle and the status data measured in the second process state may change as the process state changes, even if there is no change in the position or shape of the parts contained within the chamber.

[0190] In one embodiment, the monitoring device can generate status data by transmitting radio waves into the chamber at different frequencies within a specific frequency range and receiving radio waves reflected within the chamber. For example, the monitoring device can generate a first parameter value by transmitting a radio wave of a first frequency into the chamber and receiving the radio waves reflected within the chamber, and can generate a second parameter value by transmitting a radio wave of a second frequency into the chamber and receiving the radio waves reflected within the chamber. By performing this for all frequencies, the monitoring device can generate status data corresponding to a specific frequency range.

[0191]

[0192] [Example 1: Method for Obtaining Reference State Data]

[0193] In order to monitor a chamber, there may be a chamber in a process state (e.g., an idle state) that serves as a reference for chamber monitoring, and the chamber in the current process state may be monitored by comparing the chamber in the reference process state with the chamber in the current process state. In this way, the process state of the chamber that serves as the reference for chamber monitoring may be referred to as a reference state. For example, the reference state may be a golden chamber state. In another example, the reference state may be a process state of a chamber of interest to the user. Chambers of interest to the user may be, for example, a chamber in a good assembly state, a chamber that has completed preventive maintenance, a chamber with good process results, a chamber in a defective state, a chamber in a state requiring preventive maintenance, or a chamber in an accident state.

[0194] Additionally, the reference state may be a state in which the chamber process is not in progress, such as a state in which the assembly of process equipment is completed during equipment manufacturing or a state in which preventive maintenance of process equipment is completed. Hereinafter, the process state in which the chamber process is not in progress, as described above, is referred to as the IDLE state. Meanwhile, the IDLE state may also refer to a process state in which no process is in progress within the chamber after one process cycle is completed and before the next process cycle begins.

[0195] Meanwhile, the reference state may be a process state other than the idle state. For example, when P processes are included in a process cycle, the reference state may be any one of the P processes. In other words, in order to generate reference state data according to an embodiment of the present disclosure, reference state data may be generated in any one of the P processes. In addition, reference state data may be generated for each of the P processes. For example, from the first reference state data for the internal state of the chamber in the first process to the P-th reference state data for the internal state of the chamber in the P-th process, data may be generated.

[0196] The monitoring device can generate state data of the chamber in reference state (hereinafter referred to as “reference state data”).

[0197] FIG. 5 is a flowchart of a method for generating reference state data according to one embodiment.

[0198]

[0199] Referring to FIG. 5, a method for generating reference state data may include a step of preparing a chamber having a plurality of parts (S501), a step of transmitting radio waves into the interior of the chamber and receiving radio waves reflected from the interior of the chamber (S503), a step of generating state data of the chamber using the received radio waves (S505), a step of determining whether a certain number of state data has been acquired (S507), and a step of generating reference state data based on the certain number of state data (S509).

[0200] In step (S501), the monitoring device may be installed in a chamber having a plurality of parts. At this time, the process state of the chamber may be the aforementioned reference state. For example, the process state of the chamber may be any one of P processes.

[0201] In step (S503), the monitoring device can transmit radio waves of a specific frequency range into the interior of the chamber in the reference state and receive radio waves reflected from the interior of the chamber.

[0202] The above specific frequency range can be determined in consideration of the characteristics of the monitoring target of the device and method for monitoring the process state of the chamber disclosed by the present disclosure.

[0203] For example, as described above, the monitoring target of the device and method for monitoring the process status of the chamber disclosed by the present disclosure is whether the equipment (chamber) is properly assembled before the process as described above, or whether there is an abnormality in the inner wall of the equipment or the surface of various parts arranged inside the equipment due to the process performed between processes and after the process is completed. That is, the target to be monitored by the device and method disclosed by the present disclosure may include the geometric shape of each part inside the chamber, a level of wear-out of each part, a relative positional relationship between each part, unnecessary deposition of by-products generated by the process for the target substrate through the chamber on the inner wall of the chamber and the surface of each part inside the chamber, etc. Accordingly, the range of the specific frequency used in the device and method disclosed by the present disclosure may be determined as a frequency band that is advantageous for monitoring the geometric shape of each part inside the chamber, the relative position between each part, the degree of wear of each part, or the unnecessary deposition of byproducts generated by the process on the inner wall of the chamber and the surface of each part inside the chamber. Meanwhile, the range of the specific frequency may be determined as a frequency band that is not affected or is relatively less affected by the state of materials and energy supplied inside the chamber while the process on the substrate is in progress in the chamber to be monitored (e.g., active species, etching gas, inert gas, RF power, or plasma thereof introduced into the chamber to perform the required process). In other words, the range of the specific frequency may be determined as a frequency band that can relatively completely measure the process state inside the chamber without being relatively affected by the state of materials and energy supplied inside the chamber while the process is performed inside the chamber.

[0204] As another example, a specific frequency range can be determined as a frequency band that can measure the state of the supplied material and energy, depending on the state of the supplied material and energy. For example, when monitoring electrical / physical properties within a chamber by measuring frequency characteristics, the state of the supplied material and energy must be measured, and a frequency range in which the electrical / physical properties change according to the measured state must be selected. In other words, a specific frequency range can be determined as a frequency band that is affected by the state of the supplied material and energy.

[0205] According to some embodiments, the specific frequency range may be determined as a frequency band having a wavelength of 1 mm to 1000 mm. That is, the specific frequency range may be 300 MHz to 300 GHz.

[0206] According to some other embodiments, the specific frequency range may be determined as a frequency band having a wavelength of 10 mm to 500 mm. That is, the specific frequency range may be 600 MHz to 30 GHz.

[0207] According to some other embodiments, the specific frequency range may be from 1 GHz to 20 GHz.

[0208]

[0209] A specific frequency range may depend on the size of the space inside the chamber and / or the size of the parts contained within the chamber.

[0210] The lower limit of a specific frequency range may depend on the size of the chamber interior space. For example, the larger the chamber interior space, the smaller the lower limit may be.

[0211] The upper limit of a particular frequency range may depend on the size of the parts contained within the chamber. For example, the smaller the part, the higher the upper limit may be.

[0212] In step (S505), the monitoring device (100) can generate status data of the chamber in the reference state using the received radio waves. Meanwhile, in step (S507), it can be determined whether a preset number of status data has been acquired. If the preset number of status data has not been acquired, the monitoring device can repeat steps S503 to S505 until the preset number of status data has been acquired.

[0213] At this time, the preset constant number may be a natural number greater than or equal to 1. If the preset constant number is 1, the monitoring device (100) can determine the state data of the chamber generated through a single measurement as reference state data. If the preset constant number is 2 or more, the preset constant number of state data can be repeatedly acquired, and reference state data can be generated based on the preset constant number of state data (S509). For example, the average value, deviation value, standard deviation value, mode, median, etc. of the preset number of state data can be calculated as reference state data. Meanwhile, if the state data is a set of measurement parameters for each frequency such as the S11 parameter, the monitoring device (100) can calculate the average value, deviation value, standard deviation value, mode, and median of each measurement parameter, and set the set of calculated measurement parameters as reference state data.

[0214] Meanwhile, when measuring a preset number of status data, the reference state of the chamber must be the same. For example, when the chamber is in an IDLE state, if the monitoring device (100) measures the status data of the chamber for the first time, the chamber must remain in an idle state until all of the preset number of status data are measured. For example, when the chamber is in a process state performing a specific process, if the monitoring device (100) measures the status data of the chamber for the first time, the chamber must remain in the corresponding process state until all of the preset number of status data are measured.

[0215] Meanwhile, when measuring a preset number of state data, the chamber to be measured may be replaced each time the measurement is made, as long as the reference state is the same. However, generally, the monitoring device (100) will measure a preset number of state data in the same reference state for one chamber to generate reference state data.

[0216] In addition, the above-described steps S501 to S509 may be performed for each process. For example, if P processes are performed within a chamber, steps S501 to S509 may be performed for each of the P processes, thereby generating reference status data for each of the P processes. In other words, reference status data may be generated for each process in order to monitor the process status of the chamber and whether there are any abnormalities in each process.

[0217]

[0218] [Example 2: Method for Calculating Similarity Between Reference State Data and State Data]

[0219] Monitoring information may include information regarding the similarity between the reference state data and the state data measured by the monitoring device (100). The higher the similarity, the more likely the current process state is to be understood or determined to be similar to the reference state. Alternatively, if the similarity exceeds a similarity threshold, the current process state may be understood or determined to be the reference state. In other words, by referring to the similarity, the current process state of the chamber can be determined.

[0220] The monitoring information may include a scatter plot of status data against reference status data. The scatter plot allows the user to visually confirm the relationship between the reference status data and the status data. For example, the scatter plot may be a grid representation of the similarity between the status data measured at a predetermined period or according to a measurement trigger signal indicated by the monitoring device (100) and the reference status data, expressed as points. In other words, the scatter plot may be a graph representation of the similarity between the status data and the reference status data at a predetermined period or according to a measurement trigger signal indicated by the monitoring device (100).

[0221] Monitoring information may include a history of the chamber's past occurrences of a reference condition. This history may include, but is not limited to, the date, time, and number of times the reference condition occurred.

[0222] Monitoring information may include a description of the reference state. The description may, for example, be a description of what process state the reference state represents, but is not limited thereto. It may also include various other information related to the reference state.

[0223]

[0224] Referring to FIG. 6, the method may include a step (S601) of calculating a similarity between state data and reference state data, and a step (S603) of generating monitoring information based on the similarity. Here, the state data may be a set of parameters measured by the monitoring device (100) to determine the process state of the corresponding chamber at the time when the chamber is measured. In step S603, the monitoring device (100) may generate monitoring information based on the similarity. For example, according to [Example 3], the monitoring device (100) may determine state similarity data based on the calculated similarity. For example, the monitoring device (100) may compare each of the state data measured by each of a plurality of measurement trigger signals indicated to the monitoring device (100) at predetermined intervals with reference data to calculate a plurality of similarities, generate a plurality of state similarity data based on the plurality of similarities, and generate monitoring information based on the plurality of state similarity data.

[0225] In step (S601), the monitoring device (100) can calculate the similarity between the status data and the reference status data.

[0226] Similarity can be calculated through, but is not limited to, a graph similarity algorithm such as average, sum of squares (SOS), cosine similarity, correlation integral technique, or convolution technique.

[0227] Below we describe some examples of calculating similarity.

[0228] As an example of calculating similarity, the monitoring device (100) can calculate similarity by utilizing the difference between reference state data and state data within a specific frequency range. For example, referring to [Mathematical Formula 1] below, the monitoring device (100) can calculate similarity as the sum of the squares of the differences between the parameter values ​​of the reference state data and the parameter values ​​of the state data within a specific frequency range.

[0229]

[0230] Here, S11 (reference state) is a parameter value of reference state data, S11 (process state) is a parameter value of state data measured by the monitoring device (100), and N is the total number of data points.

[0231] As another example of calculating similarity, referring to [Mathematical Formula 2], the monitoring device (100) can calculate similarity using the cosine similarity between reference state data and state data within a specific frequency range.

[0232]

[0233] Here, S11 (reference state) represents the parameter values ​​of the reference state data in vector form, and S11 (process state) represents the parameter values ​​of the state data in vector form.

[0234] Meanwhile, if the above [Mathematical Formula 2] is expressed in the form of similarity %, it is as shown in [Mathematical Formula 3] below. In addition, [Mathematical Formula 3] also expresses the vector form of [Mathematical Formula 2] expressed in scalar form.

[0235]

[0236] [Example 3: Method for Determining and Predicting Chamber Process Status Using Similarity Patterns]

[0237] 1. Problems with prior art

[0238] As described above, the process status of the target chamber can be determined based on the similarity calculated by comparing the reference state data with the state data measured by the monitoring device (100). However, when the process cycle is repeated multiple times within the target chamber, a phenomenon is discovered where the similarity gradually decreases even when the chamber performs the same process in normal operation.

[0239] Figure 7 shows that the similarity gradually decreases as the process cycle is repeated multiple times. The X-axis of the graph illustrated in Figure 7 represents time, and the Y-axis represents the similarity between the reference state data and the measured state data.

[0240] Additionally, Fig. 7(a) shows the state similarity data on the first day of measurement, and Fig. 7(b) shows the state similarity data on the 160th day of measurement. The solid lines in Fig. 7(a) and Fig. 7(b) are lines connecting similarities in the same process, and are intended to show how the similarity changes over time.

[0241] Referring to Figure 7(a), it can be seen that the similarity decreases relatively quickly over time on Day 1 even though less than a day has passed, and referring to Figure 7(b), it can be seen that it decreases gradually on Day 16. In addition, the state similarity data on Day 1 have a relatively higher similarity than the dotted line located between 98% and 97.5% similarity, and can be seen that there is a relatively large difference from the dotted line. On the other hand, the state similarity data on Day 1 have a similarity that is almost close to the dotted line located between 98% and 97.5% similarity, and can be seen that it is lower than the dotted line.

[0242] As can be seen from FIGS. 7 (a) and (b), the similarity between the reference state data and the state data varies over time even in the same process. Therefore, it may be difficult to accurately determine which process is being performed in the chamber or what the current process state of the chamber is based solely on this similarity measurement. In other words, when there is a 'difference' in the similarities between the state data measured at each of two points in time and the reference state data, it is not possible to distinguish whether this 'difference' is caused by the difference between the process state of the chamber when measuring the reference state data and the process state of the chamber when measuring the measured state data, or by a change (or deterioration) in the geometric / electrical characteristics of the chamber.

[0243] Furthermore, a significant difference in the similarity between the state data and the reference state data for each process may indicate that the chamber is not performing its functions properly. In other words, a decrease in the similarity between the state data and the reference state data over time may not indicate a failure of a component or specific part of the target chamber, but rather the chamber's aging over time. However, as time passes and the chamber ages, causing it to malfunction and no longer function properly, effective preventive maintenance (PM) becomes difficult. Therefore, a method is needed to predict when a chamber will no longer function properly and to enable efficient PM.

[0244] Accordingly, in [Example 3] according to the present disclosure, the monitoring device (100) can acquire state similarity patterns in which the similarity between state data and reference state data changes over time. In [Example 3], we will examine a method in which the monitoring device (100) uses these state similarity patterns to determine whether the chamber is currently operating normally and to predict the point in time when it is operating abnormally.

[0245] Before explaining, a state similarity pattern can refer to the trend of similarity changes among multiple process states within a process cycle as they pass through the process cycles. For example, a state similarity pattern is generated for each process cycle, and can include state similarity data for each of the multiple processes.

[0246] For example, the monitoring device (100) can calculate similarities between state data measured at predetermined intervals or by each of a plurality of measurement trigger signals directed to the monitoring device (100) and reference state data, and can obtain a plurality of state similarity data using the calculated similarities. In addition, the monitoring device (100) can obtain state similarity patterns using the state similarity data according to the method described below.

[0247] In other words, the state similarity pattern can be understood as a set of multiple state similarity data generated based on the similarities between the reference state data and the chamber state data during one process cycle.

[0248]

[0249] 2. Example 3-1: Method for Obtaining State Similarity Patterns

[0250] FIG. 8 is a schematic illustration of a method for determining whether a chamber is operating normally or predicting the timing of abnormal operation of the chamber according to Example 3. Referring to FIG. 8, a monitoring device (100) may be attached to the exterior of the chamber (S801). For example, the monitoring device (100) may be attached to the viewport (11) of the chamber (10).

[0251] The monitoring device (100) transmits radio waves having a specific frequency range into the chamber (10), and can obtain status data of the chamber (10) based on radio waves reflected and received from the inside of the chamber (10) (S803).

[0252] For example, the monitoring device (100) can repeatedly measure status data according to a predetermined cycle or a plurality of measurement trigger signals directed to the monitoring device (100). For example, the predetermined cycle may be in units of several seconds or tens of seconds. For example, the measurement trigger signal may be generated by the control unit (14) of the monitoring device (100) and transmitted to the signal processing unit (120), or may be received from an external device through the communication unit (130) and transmitted to the signal processing unit (120). Whenever the signal processing unit (120) receives a measurement trigger signal, the monitoring device (100) can measure status data.

[0253] At this time, the monitoring device (100) can repeatedly measure the status data according to a predetermined cycle or a plurality of measurement trigger signals directed to the monitoring device (100) while attached to the outside of the chamber. However, the present invention is not limited thereto, and the monitoring device (100) can be detached from the chamber (10) and then reattached according to a predetermined cycle or a plurality of measurement trigger signals directed to the monitoring device (100) to repeatedly measure the status data. Alternatively, the monitoring device (100) can repeatedly measure the status data according to a predetermined cycle while attached to the outside of the chamber, but in some cases, the monitoring device (100) can be detached from the chamber (10) once aperiodically and then reattached to measure the status data. In this case, after the monitoring device (100) is reattached to the chamber (10), the status data can be measured based on the received measurement trigger signal, or the status data can be measured by transmitting the measurement trigger signal to the signal processing unit (120) by attaching the monitoring device (100) to the chamber (10).

[0254] In this case, the formation of the state similarity pattern described below can be measured by connecting to the state similarity data measured before attachment or detachment. In addition, after correcting the measurement error due to attachment or detachment of the monitoring device (100), the state similarity pattern can be measured by connecting to the state similarity data measured before attachment or detachment. Meanwhile, the method for correcting the measurement error due to attachment or detachment will be discussed in detail in [Example 5] described below.

[0255] The monitoring device (100) can calculate the similarity between the reference state data and the state data acquired according to [Example 1] (S805). At this time, the similarity can be calculated according to [Example 2]. In addition, the similarity can be calculated immediately or in real time when the state data is secured according to a predetermined cycle or a plurality of measurement trigger signals instructed to the monitoring device (100). For example, the monitoring device (100) can measure the state data in a state where it is once attached to the outside of the chamber (10), and calculate the similarity between the measured state data and the reference state data according to the predetermined cycle or a plurality of measurement trigger signals instructed to the monitoring device (100). In addition, the monitoring device can generate a plurality of state similarity data using the similarities calculated according to the predetermined cycle or a plurality of measurement trigger signals instructed to the monitoring device (100). One state similarity data can include a similarity value between the state data corresponding to one point in time and the reference state data.

[0256] The monitoring device (100) can form a plurality of similarity patterns using a plurality of state similarity data (S807). For example, as shown in FIG. 7, a similarity pattern per process cycle can be found using a state similarity graph, which is a set of a plurality of state similarity data. In other words, the monitoring device (100) can determine a state similarity pattern per process cycle from a state similarity graph for a set of state similarity data, thereby determining a plurality of state similarity patterns per process cycle.

[0257] The monitoring device (100) can predict the abnormal operation point of the chamber (10) using a plurality of state similarity patterns. For example, the monitoring device (100) performs graph fitting for a plurality of state similarity patterns (S809) and can predict the abnormal operation point of the chamber (10) using the fitted graph and the threshold similarity (S811). The specific execution process of S809 to S811 will be described in detail in [Example 3-2].

[0258] Meanwhile, the monitoring device (100) can determine whether the chamber is operating normally using a plurality of state similarity patterns. For example, the monitoring device (100) compares the most recent state similarity pattern among the formed state similarity patterns with a comparison state similarity pattern that is not the most recent state similarity pattern (S813), and determines whether the chamber (10) is operating normally based on the comparison result (S815). The specific execution process of S813 to S815 will be described in detail in [Example 3-3].

[0259]

[0260] Below, the process of finding a state similarity pattern in step S807 based on multiple state similarity data in step S805 and forming multiple state similarity patterns will be described.

[0261]

[0262] (1) A method for finding state similarity patterns by comparing them with process section data.

[0263] The monitoring device (100) can generate a state similarity graph based on multiple state similarity data as described in step S805. Then, by comparing the state similarity graph with the process section data, the transition points between each process can be recognized, and accordingly, the transition points between each process can be recognized. For example, referring to FIG. 9, a state similarity graph that appears as multiple processes included in a process cycle are performed can be seen. Within each process, state similarity data can be measured multiple times, and a state similarity graph can be drawn based on this. Meanwhile, the actual state similarity graph can take various forms. In particular, referring to the 4th process state and the 6th process state, the same similarity is not measured in one process state, but different similarities are measured as the process progresses.

[0264] However, in the present disclosure, for the convenience of explanation, the state similarity graph is schematically described as a rectangular graph as in FIGS. 10 to 12. In addition, as in FIG. 9, the number of process states included in one process cycle may be eight or more, but for the convenience of explanation, as in FIGS. 10 to 12, it is assumed and described that one process cycle has four process states. However, it is clear that the interpretation is not limited to four or eight process states included in one process cycle. In other words, the number of processes included in a process cycle according to an embodiment of the present disclosure is not limited to a specific number, and two or more processes are determined according to the process process without any limitation in number and included in one process cycle.

[0265] While the monitoring device (100) calculates the similarity according to a predetermined cycle or a plurality of measurement trigger signals directed to the monitoring device (100), the process cycle is repeated multiple times, and the point in time at which each process is switched while the process cycle is repeated multiple times can be recorded in the process section data. Meanwhile, such process section data is stored in the storage unit (150) of the monitoring device (100) or stored in an external storage device such as a USB, so that the monitoring device (100) can read the process section data stored in the external storage device.

[0266] The monitoring device (100) matches the point in time at which each process is switched stored in the process section data with the state similarity graph, thereby finding the point in time at which each process is switched and the point in time at which one process cycle ends and the next process cycle begins in the state similarity graph. For example, as shown in Fig. 9, the point in time at which each process state is switched and the point in time at which the process cycle is switched can be found by matching the start point of each process state recorded in the process section data with the state similarity graph.

[0267] In addition, the monitoring device (100) can determine the state similarity graph during the time from the start to the completion of one process cycle as a state similarity pattern using the above-described method. That is, if the process cycle is performed repeatedly multiple times, the state similarity graph for each process cycle can be recognized as each state similarity pattern. Meanwhile, the state similarity pattern may include an idle section included between each process cycle or before the first process cycle. For example, the state similarity pattern may mean a state similarity graph during an idle section and a process cycle occurring immediately after the idle section. In addition, an idle section may mean that the chamber is in an idle state.

[0268] Hereinafter, for the convenience of explanation, unless otherwise specified, a state similarity pattern should be interpreted as meaning a state similarity graph during a process cycle that includes an idle section and multiple processes occurring immediately after the idle section. In addition, in the following description, for the convenience of explanation, it is assumed and described that the process cycle includes processes 1 through 4, but any one of the first through fourth processes may mean an idle section or an idle state. For example, the first process may mean an idle section or an idle state. In addition, in the present disclosure, for the convenience of explanation, it is described that the process cycle includes processes 1 through 4, but it is not limited thereto, and the number of processes that may be included in the process cycle may vary depending on the type of work performed by the chamber (10).

[0269] The monitoring device (100) compares the process section data and the state similarity graph whenever at least one state similarity data is determined, and when each process cycle is completed, reads the state similarity graph corresponding to each process cycle as a similarity pattern and stores it in the storage unit (150).

[0270] The monitoring device (100) can collect a plurality of state similarity patterns by performing the method described above each time the process cycle is repeated and storing the recognized state similarity patterns in the storage unit (150).

[0271]

[0272] (2) A method for determining a state similarity pattern by collecting state similarity data until N transition states occur.

[0273] Let's look at another way to determine state similarity patterns.

[0274] Referring to FIG. 10, the monitoring device (100) can repeatedly calculate the similarity according to a predetermined cycle or a plurality of measurement trigger signals directed to the monitoring device (100) until N transitions occur, and can generate a plurality of state similarity data using the calculated similarity. In addition, the monitoring device (100) can generate a state similarity graph based on the plurality of state similarity data.

[0275] At this time, N may be a value preset in the monitoring device (100). For example, if the monitoring device (100) generates state similarity data until 100 transition states occur, continuously draws the state similarity data, and when it appears that a similar similarity trend appears in units of 4 sections, the monitoring device (100) may determine that there are 4 processes and determine state similarity patterns corresponding to processes 1 to 4. In this case, the monitoring device (100) may determine state similarity patterns in units of 4 transitions, form a plurality of state similarity patterns, and store them in the storage unit (150). The monitoring device (100) may generate state similarity data based on the similarity calculated according to a predetermined cycle or a plurality of measurement trigger signals directed to the monitoring device (100), and may repeatedly determine state similarity patterns in units of 4 transitions, and repeatedly store them in the storage unit (150).

[0276]

[0277] 3. Example 3-2: Method for predicting abnormal operation point of a chamber using state similarity pattern

[0278] Now, with reference to FIG. 11, let us examine how the monitoring device (100) predicts the abnormal operation point of the chamber (10).

[0279] Referring to FIG. 11, the monitoring device (100) can repeatedly calculate similarity according to a predetermined cycle or a plurality of measurement trigger signals directed to the monitoring device (100), thereby generating a plurality of state similarity data. In addition, the monitoring device (100) can repeatedly generate a state similarity graph based on the plurality of state similarity data.

[0280] In addition, the monitoring device (100) can obtain a plurality of state similarity patterns according to [Example 3-1]. Referring to Fig. 11, it can be seen that the plurality of state similarity patterns are repeated according to a similar pattern. However, it can be seen that the absolute state similarity data of the plurality of state similarity patterns is gradually decreasing. In other words, in Fig. 11, the value of the state similarity data of the first process included in the first state similarity pattern is higher than the value of the state similarity data of the first process included in the second state similarity pattern. Similarly, the value of the state similarity data of each of the second to fourth processes included in the first state similarity pattern is higher than the value of the state similarity data of each of the second to fourth processes included in the second state similarity pattern. However, the first state similarity pattern and the second state similarity pattern may exhibit a similar / identical trend. That is, for example, the difference between the state similarity data values ​​of each of the first to fourth processes included in the state similarity pattern and the state similarity data values ​​of each of the second to fourth processes included in the second state similarity pattern corresponding thereto may be the same / similar.

[0281] In addition, the monitoring device (100) can set at least one reference point in each of the plurality of state similarity patterns acquired according to [Example 3-1]. For example, referring to FIG. 11, among the process states (i.e., idle sections and processes) included in each process cycle, one can be selected, and at least one reference point can be determined in the selected process state (idle section or process). At this time, at least one reference point selected in each state similarity pattern can be selected within the same process state (e.g., the same idle section). In addition, at least one reference point selected in each state similarity pattern can be selected as the same corresponding time point within the same process state.

[0282] For example, the monitoring device (100) may select a center point of the first process state (e.g., an idle section) of each state similarity pattern, and connect the center points of the first process states of each state similarity pattern to perform graph fitting using an exponential function in the form of A exp(-bx) + C. Meanwhile, examples of performing graph fitting are not limited to the above-described contents. For example, the first process state may be replaced with any process from the second process state to the Nth process state, and the center point may be replaced with a start point, an end point, or a specific point in time in the selected process state. In addition, the graph fitting performed by the monitoring device (100) is not limited to graph fitting using an exponential function, and various graph fitting methods such as graph fitting using a linear function, graph fitting using a polynomial function, and graph fitting using a logarithmic function may be applied. In addition, the function used in the graph fitting may use time as a variable. In other words, since the state similarity graph is generated based on the similarities between the state data of the chamber (10) measured over time and the reference state data and a plurality of state similarity data generated using the similarities, the function used for graph fitting may also have to be a time-dependent function.

[0283] In addition, although the above-described example describes selecting one point in time from one selected process state, multiple points in time may be selected. For example, graph fitting may be performed by connecting the start point and the end point of the first process state of each state similarity pattern. In addition, in the above-described example, not only one process state is selected, but multiple process states may be selected. For example, the monitoring device (100) may perform graph fitting by connecting the start point of the first process state of each state similarity pattern and the start point of the second process state, and may also perform graph fitting by connecting the start point and the end point of the first process state of each similarity pattern and the start point and the end point of the second process state.

[0284] When graph fitting is performed, the monitoring device (100) can compare the fitted graph with the threshold similarity. Then, the point where the fitted graph and the threshold similarity meet is determined as the predicted time point of abnormal operation of the chamber (10), and the user can be notified through the communication unit (130) or the output unit (not shown) that abnormal operation of the chamber (10) is predicted to occur at that time point. That is, the monitoring device (100) can predict the times when the value of the fitted graph becomes lower than the threshold similarity as the predicted time point of abnormal operation of the chamber (10). Therefore, the monitoring device (100) can determine the first time point when the value of the fitted graph becomes lower than the threshold similarity as the predicted time point of abnormal operation of the chamber (10).

[0285] Meanwhile, the critical similarity can be set differently depending on the specifications of the chamber (10) and the process margin.

[0286] Meanwhile, the graph fitting and prediction of the abnormal operation time of FIG. 11 can be continuously performed at regular intervals. Referring to FIG. 7, when looking at the state similarity graph of the region immediately after the monitoring device (100) starts monitoring the chamber (10) or starts PM, it can be seen that the slope of the state similarity graph is steep. Therefore, if the abnormal operation time of the chamber (10) is predicted in the time region immediately after the monitoring device (100) starts monitoring the chamber (10) or starts PM, the time can be predicted to be relatively early. However, it can be seen that the slope of the state similarity graph becomes gentler as the operation of the chamber (10) becomes more stable as it goes toward the initial region and the progress region where the chamber (10) operates. Therefore, the abnormal operation time of the chamber (10) can be predicted to be relatively slow as it goes toward the initial region and the progress region where the chamber (10) operates. Additionally, since the process may malfunction due to a specific event while the chamber (10) is operating, and the chamber (10) or the process may change unexpectedly due to a specific event, there is an aspect in which it is impossible to continuously predict how changes will occur. Therefore, the monitoring device (100) must continuously perform graph fitting to predict the abnormal operation time of the chamber (10) in order to increase the accuracy of predicting the abnormal operation time.

[0287] Meanwhile, the period in which the monitoring device (100) performs graph fitting to predict the abnormal operation point of the chamber (10) may be several seconds or several tens of seconds. Alternatively, the period in which the monitoring device (100) performs graph fitting to predict the abnormal operation point of the chamber (10) may be one day or several days. For example, the monitoring device (100) may perform graph fitting to predict the abnormal operation point of the chamber (10) every time it measures similarity according to a predetermined period or a plurality of measurement trigger signals directed to the monitoring device (100), or may perform graph fitting every time one state similarity pattern is acquired to predict the abnormal operation point of the chamber (10).

[0288]

[0289] 4. Example 3-3: Method for determining whether a chamber is operating normally using a state similarity pattern

[0290] Hereinafter, with reference to FIGS. 12 and 13, a method for determining whether a chamber (10) is operating normally will be examined. In order to determine whether a chamber (10) is operating normally, two state similarity patterns among a plurality of state similarity patterns acquired by the monitoring device (100) can be compared.

[0291] At this time, since it is to determine the current state of the chamber, the monitoring device (100) can determine one of the two state similarity patterns as the most recently acquired state similarity pattern.

[0292] The remaining state similarity pattern may be any one of the multiple similarity patterns, other than the most recent state similarity pattern. Hereinafter, the remaining state similarity pattern for comparison with the most recent state similarity pattern is referred to as the comparison state similarity pattern.

[0293]

[0294] (1) Comparative similarity pattern selection method

[0295] The comparison state similarity pattern may be the state similarity pattern initially acquired after the monitoring device (100) measures the similarity of the chamber (10). The initial state similarity pattern may be the state similarity pattern when the state of the chamber (10) is at its best since the monitoring device (100) began measuring the similarity. Therefore, by comparing the state of the chamber (10) when its state was at its best with the current state of the chamber (10), it is possible to determine to what extent the current state of the chamber (10) has deteriorated.

[0296] Meanwhile, the comparative state similarity pattern may be an average state similarity pattern calculated by averaging the state similarity patterns from the initial state similarity pattern to the Pth state similarity pattern. As seen in Fig. 7, the similarity decreases relatively rapidly immediately after the monitoring of the chamber (10) state begins. Therefore, by using the average state similarity pattern calculated through the average when the chamber (10) has stabilized to a certain extent and when the state of the chamber (10) is at its best as the comparative state similarity pattern, it is possible to determine how much the current state of the chamber (10) has deteriorated on average.

[0297] In addition, the comparison state similarity pattern may be the state similarity pattern closest to the most recent similarity pattern. In other words, it may be the second most recent state similarity pattern. Since the monitoring device (100) repeatedly determines whether the chamber (10) is operating normally whenever a state similarity pattern is acquired, if the second most recent state similarity pattern is used as the comparison state similarity pattern, it is easy to determine at what point a sudden breakdown or abnormal state of the chamber (10) occurred.

[0298] Alternatively, the comparison state similarity pattern may be selected as any state similarity pattern among the state similarity patterns other than the most recent state similarity pattern.

[0299]

[0300] (2) A method for determining whether the chamber (10) is operating normally by comparing the comparison state similarity pattern with the most recent state similarity pattern.

[0301] The monitoring device (100) can compare the state similarity data of the comparison state similarity pattern and the state similarity data of the most recent state similarity pattern for each process state, and can recognize that an abnormal factor (e.g., a breakdown, etc.) has occurred in a process that includes state similarity data in which the difference between the state similarity data is greater than a preset threshold similarity difference value.

[0302] That is, the method for recognizing abnormal operation of a chamber described with reference to FIG. 12 may mean a method for recognizing abnormal operation of a chamber by comparing the relative size difference between each state similarity data included in the most recent state similarity pattern and each state similarity data included in the comparison state similarity pattern.

[0303] At this time, the threshold similarity difference value that serves as the basis for comparison is set in the following manner, and whether the chamber (10) is operating normally can be determined in the following manner.

[0304]

[0305] 1) A method for determining whether a chamber (10) is operating normally by using different “critical similarity differences”.

[0306] Depending on which state similarity pattern is determined by the comparison state similarity pattern, the threshold similarity difference values ​​may need to be set differently to determine whether the chamber (10) is operating normally. For example, as shown in Fig. 7, even if the chamber (10) maintains normal operation rather than abnormal operation, the overall similarity may gradually decrease as the process cycle is repeated.

[0307] Therefore, even if the chamber (10) is operating normally, if the comparison state similarity pattern is the first state similarity pattern, the difference between the most recent state similarity pattern and the state similarity data will be relatively large, and if the comparison state similarity pattern is the second most recent state similarity pattern, the difference between the most recent state similarity pattern and the state similarity data will be relatively small. Therefore, the threshold similarity difference value set depending on which state similarity pattern the comparison state similarity pattern is must be set differently.

[0308] Figure 12(a) shows the case where the second most recent state similarity pattern is a comparison state similarity pattern, and Figure 12(b) shows the case where the first state similarity pattern is a comparison state similarity pattern, where the threshold similarity difference value is set.

[0309] For example, referring to FIG. 12(a), if the second most recent state similarity pattern is the comparison state similarity pattern, the threshold similarity difference value is set to a relatively small value, so that the chamber (10) can be classified as being in abnormal operation even if the difference between the state similarity data of the comparison state similarity pattern and the most recent state similarity pattern is small. On the other hand, referring to FIG. 12(b), if the first state similarity pattern is the comparison state similarity pattern, the threshold similarity difference value is set to a relatively large value, so that the chamber (10) can be classified as being in abnormal operation only if the difference between the similarity of the comparison state similarity pattern and the most recent state similarity pattern is relatively large.

[0310] Therefore, if the comparison state similarity pattern is not fixed to any of the examples described above and can be changed, a plurality of threshold similarity difference values ​​must be set. For example, if the comparison state similarity pattern is selected as the first state similarity pattern, the second most recent state similarity pattern, or the average state similarity pattern depending on the state measurement situation of the chamber (10), a plurality of threshold similarity difference values ​​must be set according to each of the state similarity patterns that can be selected. In addition, the monitoring device (100) must determine whether the chamber (10) is operating normally using the threshold similarity difference value corresponding to the selected state similarity pattern.

[0311] For example, the monitoring device (100) may determine the threshold similarity difference value based on the pattern interval between the selected comparison state similarity pattern and the most recent state similarity pattern. For example, the monitoring device (100) may determine the threshold similarity difference value based on the number of state similarity patterns (i.e., pattern interval) included between the selected comparison state similarity pattern and the most recent state similarity pattern. Alternatively, the monitoring device (100) may determine the threshold similarity difference value based on the time interval between the selected comparison state similarity pattern and the most recent state similarity pattern.

[0312] At this time, the threshold similarity difference value may be selected from among multiple preset threshold similarity difference values ​​based on the time interval or pattern interval. Alternatively, the threshold similarity difference value may be calculated based on the time interval or pattern interval using a specific algorithm or formula.

[0313]

[0314] 2) Method of judgment using a single "critical similarity difference value"

[0315] A monitoring device (100) sets a threshold similarity difference value, and when a comparison state similarity pattern is selected, the selected comparison state similarity pattern can be approximately corrected with the most recent state similarity pattern. In addition, by comparing the most recent corrected state similarity pattern with the selected comparison state similarity pattern, it is possible to determine whether the chamber (10) is operating normally.

[0316] For example, as can be seen in Fig. 13, the most recent state similarity pattern can be corrected by utilizing a fitted graph obtained in the process of performing [Example 3-2] or a function (using a time variable) corresponding to the fitted graph.

[0317] Fig. 13 shows that the comparison state similarity pattern is selected as the initial state similarity pattern. Referring to Fig. 13, the monitoring device (100) can correct the most recent state similarity pattern to be generally similar to the initial state similarity pattern using a fitted graph or a corresponding function. For example, the most recent state similarity pattern can be moved to a point corresponding to the initial state similarity pattern according to the fitted graph or corresponding function. Alternatively, the difference value between the most recent state similarity pattern and the reference point (e.g., the start point, the end point, or the center point of each selected process) used for graph fitting or function derivation within the initial state similarity pattern can be added to the state similarity data included in the most recent state similarity pattern to be generally corrected to be similar to the most recent state similarity pattern.

[0318] The monitoring device (100) calculates the difference between the state similarity data of the most recent corrected state similarity pattern and the first state similarity pattern, and can recognize that an abnormal operation of the chamber (10) has occurred in a process state that includes state similarity data having a difference greater than a set threshold similarity difference value.

[0319] Meanwhile, although the above-described embodiment was written assuming that the comparison state similarity pattern is the initial state similarity pattern, the principle of the above-described embodiment can be applied even if the comparison state similarity pattern is an arbitrary state similarity pattern. For example, the monitoring device (100) can compare the comparison state similarity pattern with the most recent state similarity pattern after overall correcting the most recent state similarity pattern to correspond to an arbitrary state similarity pattern using a fitted graph or a correspondence function. In addition, the monitoring device (100) can recognize that an abnormal operation of the chamber (10) has occurred in a process state including state similarity data in which the difference between the state similarity data is greater than a set threshold similarity difference value.

[0320]

[0321] Meanwhile, in the above-described embodiment, the period for comparing the most recent state similarity pattern with the comparative state similarity pattern may be the same as the period for forming a single state similarity pattern. For example, the monitoring device (100) may select a comparative state similarity pattern each time a state similarity pattern is acquired, thereby determining whether the chamber (10) is operating normally.

[0322]

[0323] 5. Experimental data

[0324] Fig. 14 shows the experimental data described below, which shows the consistency and accuracy of whether the chamber (10) was actually abnormal when the monitoring device (100) determined whether the chamber (10) was operating abnormally using the state similarity pattern according to the above-described [Example 3]. Referring to Fig. 14(a), as a result of continuously measuring the similarity data of the fourth process state while going through the process cycle based on the state similarity data of the fourth process state when the chamber is normal, it was found that there was a significant difference between the state similarity data of the fourth process state when it was normal and the state similarity data of the fourth process state when it was normal. Accordingly, as a result of actually confirming the fourth process, it was confirmed that there was a problem in the fourth process.

[0325] Figure 14(b) shows the state similarity data measured while artificially changing the process state every 10 minutes. Looking at Figure 14(b), in the process where the chamber was initially normal, the state similarity data was measured higher than the critical similarity. However, when the process state was artificially made abnormal 10 minutes later, the state similarity data was measured significantly lower than the critical similarity.

[0326] In this way, when artificially changing the process status at 10-minute intervals, it was confirmed that the status similarity data continuously changed, and it was observed that the status similarity data dropped significantly in the section where the change was abnormal.

[0327] In addition, FIGS. 15 to 17 are intended to explain the accuracy of predicting the abnormal operation time of the chamber (10) according to the above-described [Example 3].

[0328] Figure 15 shows a graph fitted using an exponential function on a state similarity graph according to [Example 3]. Figure 15 shows a graph fitted based on a state similarity graph drawn by measuring state similarity data for approximately 160 to 170 days, and calculating the point in time at which abnormal behavior is predicted by comparing it with a threshold similarity.

[0329] The results predicted that the chamber under test would become abnormal at 221.7 days. However, a significant process difference was actually detected at 222 days, prompting PM.

[0330] The experimental results showed that the chamber could accurately predict abnormal behavior at least two months before it began. This confirmed that it was possible to prepare in advance for immediate PM in case the chamber exhibited abnormal behavior.

[0331] Figure 16 illustrates how the state similarity graph and graph fitting using the exponential function approach the critical similarity as measurements on the chamber continue. Figure 16(a) is a graph for the first day of measurements on the target chamber, and Figure 16(b) is a graph for the 160th day of measurements on the target chamber.

[0332] In Figure 16, graph fitting was performed based on the first process state. Referring to Figure 16(a), the fitted graph shows a significant difference from the critical similarity. However, referring to Figure 16(b), the fitted graph shows a significantly smaller difference from the critical similarity.

[0333] In other words, as the measurement time passes, the point at which the fitted graph meets the critical similarity can be precisely measured, and thus the point at which the target chamber will begin to behave abnormally can be predicted.

[0334] Figure 17 shows a wider view by changing the X-axis scale of Figure 16 from 'minutes' to 'hours'. Figure 17(a) is a graph of the first day of measurement for the target chamber, and Figure 17(b) is a graph of the 160th day of measurement for the target chamber. Referring to Figure 17, the difference in the fitted graph explained through Figure 16 and the relationship between the fitted graph and the critical similarity can be seen more clearly.

[0335]

[0336] [Example 4: Method for judging and predicting chamber process status using process-specific reference status data]

[0337] Hereinafter, in Example 4, a method for determining whether a chamber is operating normally or predicting the point in time when the chamber is operating abnormally is described in a different way from Example 3.

[0338] In Example 4, the monitoring device (100) generates reference state data for each process performed within the chamber. In addition, the monitoring device (100) uses process section data to determine the process currently being performed within the chamber, loads reference state data of the determined process, and calculates the similarity between the loaded reference state data and the currently measured state data. In addition, the monitoring device (100) determines abnormal operation of the chamber or predicts the point in time of abnormal operation of the chamber based on the calculated similarity.

[0339] For convenience of explanation, Example 4 exemplifies five process states occurring within a chamber. That is, one process cycle is described as consisting of five processes. However, as described above, Example 4 should not be construed as being limited to five processes within a process cycle, and it should be noted that Example 4 can be applied if more than one process is performed within a process cycle.

[0340] Below, the operation in this embodiment 4 will be examined in more detail.

[0341]

[0342] 1. Example 4-1: Method for determining whether a chamber is operating normally using process-specific reference status data

[0343] Figures 18 and 19 are for explaining a method of determining whether a chamber is operating normally using process-specific reference state data according to Example 4-1.

[0344] In Example 4-1, as can be seen in FIG. 18, the monitoring device (100) uses process section data to identify the current process section or process state of the chamber (10). The monitoring device (100) measures the state data and compares the measured state data with the reference state data of the identified process to calculate a similarity.

[0345] Here, the process interval data includes data that can identify the start and end times of each process. Alternatively, the process interval data includes data that can identify the length or time of the interval during which each process is performed. For example, the monitoring device (100) can use the process interval data to identify the order in which processes are performed within a process cycle and the time interval during which each process is performed. Accordingly, as described above, the monitoring device (100) can recognize the transition points between each process through the process interval data, and compare the start point of the process cycle with the transition point between processes to identify the process currently being performed within the chamber (10). Accordingly, the monitoring device (100) can select reference state data of the identified process and compare it with the current state data. In other words, the monitoring device (100) uses the process interval data to identify the point in time when a process is changed, and based on the point in time when the change occurs, changes the reference state data to the reference state data of the changed process, and compares the reference state data with state data measured periodically / aperiodically to calculate a similarity.

[0346] If the process is being performed normally, the similarity between the reference state data of the current process and the state data measured by the monitoring device (100) will be very high. However, if the process is being performed abnormally or an abnormality occurs in the chamber (10), the similarity between the reference state data of the current process and the state data measured by the monitoring device (100) will be relatively low. Using this, in Example 4-1, it is possible to identify whether the chamber (10) is operating abnormally.

[0347] For example, referring to FIG. 18, in the first process section, the similarity between the reference state data of the first process and the state data measured by the monitoring device (100) in the first process section is measured, in the second process section, the similarity between the reference state data of the second process and the state data measured by the monitoring device (100) in the second process section is measured, and in the third process section, the similarity between the reference state data of the third process and the state data measured by the monitoring device (100) in the third process section is measured.

[0348] And, in the fourth process section, the similarity between the reference state data of the fourth process and the state data measured by the monitoring device (100) in the fourth process section is measured, and in the fifth process section, the similarity between the reference state data of the fifth process and the state data measured by the monitoring device (100) in the fifth process section is measured.

[0349] Referring to Fig. 18, a similarity of over 99.8%, which is close to 100%, was calculated in the first process section, the second process section, the fourth process section, and the fifth process section, which indicates that the chamber (10) is performing normal operations in the first and second processes. However, in the third process section, it can be seen that a similarity lower than the critical similarity was calculated in at least a portion of the third process section. This indicates that the chamber (10) is performing abnormal operations in the third process.

[0350] Now, referring to FIGS. 18 and 19, a method for determining whether the chamber (10) is operating abnormally will be described in detail. According to FIGS. 18 and 19, a monitoring device (100) can be attached to the outside of the chamber (S1901). For example, the monitoring device (100) can be attached to the viewport (11) of the chamber (10). Thereafter, the monitoring device (100) determines the process currently being performed in the chamber based on the process section data. For example, referring to FIG. 18, if the monitoring device (100) determines that the current point in time is the first process section based on the process section data, the monitoring device (100) selects the reference state data of the first process. If the current point in time is included in the second process section based on the process section data, the monitoring device (100) selects the reference state data of the second process, and if the current point in time is included in the third process section, the monitoring device selects the reference state data of the third process (S1903). In addition, the monitoring device (100) transmits radio waves having a specific frequency range into the chamber (10) and can obtain status data of the chamber (10) based on the radio waves reflected and received from the inside of the chamber (10) (S1905). Here, the method for obtaining status data of the chamber (10) has already been described through Example 3 and S803 of FIG. 8, and thus a detailed description thereof will be omitted.

[0351] The monitoring device (100) can calculate the similarity between the selected reference state data and the acquired state data (S1907). At this time, the method for calculating the similarity has also been described in Embodiments 2 and 3 and S805 of FIG. 8, and thus a detailed description thereof will be omitted.

[0352] The monitoring device (100) determines whether the calculated similarity is below a preset threshold similarity (S1909). If the calculated similarity exceeds the preset threshold similarity, the monitoring device (100) determines that the process within the chamber (100) is being performed normally, and repeats the process from S1903.

[0353] If the calculated similarity is lower than a preset threshold similarity, the chamber (10) is judged to be operating abnormally and the user can be notified to take appropriate action (S1911).

[0354]

[0355] 2. Example 4-2: Method for predicting abnormal operation point of a chamber using process-specific reference state data

[0356] FIGS. 20 to 22 are for explaining a method for predicting abnormal operation time of a chamber using process-specific reference state data according to Example 4-2. Referring to FIG. 20, as described in Example 3, even when measuring the similarity between the same reference state data (e.g., reference state data of the first process) and the state data measured by the monitoring device (100) in the same process (e.g., the first process), it can be seen that the similarity gradually decreases.

[0357] That is, even if the similarity between the state data measured at the time and the reference state data for the process is calculated using reference state data for the process at the time when the monitoring device (100) measures the state data, rather than using a similarity pattern as in Example 3-2, the similarity for the same process may gradually decrease.

[0358] Accordingly, in Example 4-2, a method is examined for predicting a point in time when a chamber (10) will operate abnormally by measuring similarity based on at least one reference point within the same process while multiple process cycles are repeated. That is, a method is examined for predicting a point in time when a chamber (10) will operate abnormally by performing graph fitting based on at least one reference point within the same process section, without using a similarity pattern as in Example 3-2.

[0359] Figure 21 shows an example in which a monitoring device (100) measures state data in a first process within each cycle during multiple process cycles, and measures the similarity between the measured state data and reference state data of the first process.

[0360] Referring to FIG. 21, the monitoring device (100) measures similarities in the same first process and performs graph fitting on the similarities as in Example 3-2. The monitoring device (100) can determine the point in time at which the fitted graph meets the threshold similarity as the point in time at which abnormal operation of the chamber is predicted. Meanwhile, the graph fitting can be performed each time the similarity is measured in the first process of each cycle. Alternatively, the graph fitting can be performed each time N similarities are calculated by calculating the similarity in the first process of each cycle for N cycles.

[0361] In addition, in Example 4-2, graph fitting is described as being performed using the similarities between the process status data and the state data of the first process, but this is for the convenience of explanation and understanding, and graph fitting and abnormal operation of the chamber (10) are not necessarily predicted based on the first process, and it is obvious to a person skilled in the art that graph fitting can be performed by selecting any one process included in the process cycle.

[0362] FIG. 22 is for explaining the operation of a monitoring device (100) that predicts the abnormal operation point of a chamber according to Example 4-2.

[0363] In Fig. 22, S2201 to S2207 are the same as S1901 to S1907 of Fig. 19 related to Example 4-1, so redundant descriptions are omitted.

[0364] The monitoring device (100) sets at least one reference point and performs graph fitting using the set at least one reference point (S2209). For example, the monitoring device (100) may select one of the process states included in each process cycle (e.g., the first process) and set at least one reference point in the selected process state. Meanwhile, the method for the monitoring device (100) to set at least one reference point has been described in Example 3-2, and thus, this may be applied. Therefore, redundant descriptions will be omitted.

[0365] In addition, the monitoring device (100) can perform graph fitting using the similarity calculated from at least one reference point of the selected one process state (e.g., the first process) for each process cycle. For example, the monitoring device (100) can perform graph fitting to connect the similarities of at least one reference point included in the first process of each process cycle using a specific function. For example, the specific function may be an exponential function in the form of A exp(-bx) + C as mentioned in Example 3-2, but is not limited thereto, and various functions may be used as mentioned in Example 3-2. A specific graph fitting method may apply the graph fitting method described in Example 3-2, and thus a detailed description thereof will be omitted. The monitoring device (100) can predict the first time point at which the fitted graph value becomes less than or equal to a threshold similarity as the abnormal operation time point of the chamber (10) (S2211).

[0366] Meanwhile, graph fitting and prediction of abnormal operation points can be continuously performed at regular intervals. Therefore, after performing S2211, the monitoring device (100) returns to S2203 and can perform subsequent procedures. Since the graph fitting and prediction cycle of abnormal operation points overlap with the description of Example 3-2, the description of Example 3-2 applies, and a detailed description is omitted.

[0367]

[0368] [Example 5: Method for correcting measurement error of monitoring device (100)]

[0369] 1. Problems with prior art

[0370] There is a problem in that the parameters (e.g., S11 parameters) for measuring frequency characteristics used by the monitoring device (100) according to the present disclosure to measure the state of the chamber (10) are very sensitive to the position of the chamber (10) to which the monitoring device (100) is attached, and thus the measured parameters are measured differently depending on the position to which the monitoring device (100) is attached.

[0371] For example, referring to FIG. 23, if a set of parameters (i.e., state data) for the same chamber (10) in the same state (e.g., idle state) is repeatedly measured using the same monitoring device (100), each of the repeatedly measured sets of parameters (i.e., state data) must all include the same parameter values, so that all of the sets of parameters (i.e., state data) must indicate that the chamber (10) is always in the same state.

[0372] However, in reality, as described above, the plurality of parameter sets measured by the monitoring device (100) are measured differently depending on the relative attachment position of the monitoring device (100). For example, referring to FIG. 23, let us assume that the parameter sets are measured while the monitoring device (100) is attached and detached to the viewport (11) of the chamber (10) that is in an idle state. Since the chamber (10) is in the same idle state, the user expects that even if the parameter sets (i.e., state data) are measured while repeatedly attaching and detaching the monitoring device (100) to the viewport (11), the parameter corresponding to the same frequency will always have the same value within the plurality of parameter sets (i.e., state data).

[0373] However, as a result of actual measurement, it was found that the first parameter value measured by attaching the antenna (110) of the monitoring device (100) to the position 110-1 within the viewport (11) and measuring it, the second parameter value measured by detaching the antenna (110) from 110-1 and reattaching it to the position 110-2 and measuring it, and the third parameter value measured by detaching the antenna (110) and reattaching it to the position 110-3 and measuring it have different values. Meanwhile, although 110-1, 110-2 and 110-3 appear to have a difference that can be confirmed with the naked eye in the drawing, this is for the convenience of explanation, and in reality, the difference between the center points of 110-1, 110-2 and 110-3 is several mm to several um or several tens of um. In other words, even if the attachment location of the monitoring device (100) has a difference of several mm, several um, or several tens of um depending on the attachment or detachment, the first parameter value to the third parameter value have different differences from each other, so it can be seen that the measurement results of multiple parameter sets are significantly affected even by a minute attachment error of the monitoring device (100).

[0374] At this time, the first parameter value, the second parameter value, and the third parameter value are values ​​for parameters corresponding to the same frequency.

[0375] The result values ​​for this can be seen in Figs. 24 to 26.

[0376] FIG. 24 and FIG. 25 show state similarity data generated between reference state data and state data without attaching or detaching the monitoring device (100), and FIG. 19c shows state similarity data generated by calculating the similarity between reference state data and state data of the monitoring device (100) according to the above-described [Example 3] while attaching or detaching the monitoring device (100) to the chamber (10) in an idle state, and expressing a state similarity graph using the generated state similarity data.

[0377] Referring to Fig. 24, the monitoring device (100) was attached to the chamber (10) without being detached, and then the state similarity data was continuously measured. As can be seen in Fig. 24, when the state similarity data is measured multiple times while attached to the chamber (10), the measured state similarity data changes only within 0.001%. In other words, the difference between the measured state similarity data does not exceed 0.001%. That is, when the state similarity data is measured multiple times while the monitoring device (100) is attached to the chamber (10), the difference between the state similarity data is so small that it can be ignored. Fig. 25 is a graph of Fig. 24 scaled down along the Y axis. Referring to Fig. 25, it can be seen that the state similarity data measured multiple times is almost constant.

[0378] Referring to Fig. 26, although the chamber (10) is in the same idle state, when state similarity data is generated by repeatedly attaching and detaching, it can be found that different state similarity data is generated each time the chamber is attached and detached. At this time, Fig. 26 and Fig. 25 have the same Y-axis scale. Referring to Fig. 26, it can be seen that the state similarity data is measured differently each time the monitoring device (100) is repeatedly attached and detached and measured. Specifically, when the monitoring device (100) is repeatedly attached and detached and measured, even if the chamber (10) is in the same state, the measured state similarity data changes within a maximum of 0.03%. In other words, the difference between the measured state similarity data can be up to 0.03%. This is a significant difference compared to a difference of only within 0.001%.

[0379] In other words, even if the state of the chamber (10) is the same, the monitoring device (100) may determine that the state of the chamber (10) is different depending on the position where the antenna (110) is attached within the viewport (11), which may lower the measurement accuracy.

[0380] To solve this, the antenna (110) must always be attached to the same location even if attachment and detachment of the antenna (110) is repeated within the viewport (11).

[0381] However, this is realistically very difficult. Since the attachment of the antenna (110) is done directly by a person or machine, it is practically impossible to attach it to the same location every time without even an inch of error.

[0382] Therefore, a method is needed for the monitoring device (100) to compensate for measurement errors resulting from the attachment or detachment of the antenna (110).

[0383] Furthermore, the above-described problem can be equally applied when performing state measurements on multiple chambers (10). Here, the multiple chambers (10) may be identical products manufactured through the same manufacturing process. However, even if they are identical products, the internal environments of the multiple chambers may differ slightly due to tolerances. Accordingly, even if the monitoring device (100) performs state measurements on multiple chambers that are all in the same state (e.g., idle state), the resulting values ​​of the measured parameter sets may differ. Therefore, a method for correcting errors in state measurements between multiple chambers is also necessary.

[0384] Furthermore, in order to repeatedly perform state measurements for multiple chambers, the monitoring device (100) must be continuously attached and detached to different chambers while performing state measurements. Therefore, the error correction due to attachment and detachment described above can occur equally for multiple chambers. Therefore, a method capable of correcting both state measurement and attachment and detachment errors between different chambers is also required.

[0385]

[0386] 2. Example 5-1: Method for determining an error correction function to correct a measurement error caused by repeated attachment and detachment of a monitoring device (100) to one chamber (10).

[0387] In an embodiment according to the present disclosure, an error correction function is determined using a deep learning model (hereinafter, 'learning model'), and the determined error correction function is used to correct the state data measured by the monitoring device (100), and then the corrected state data is compared with reference state data to calculate similarity, and state similarity data can be generated.

[0388] According to an embodiment of the present disclosure, a learning model may use either a single-layer perceptron neural network such as FIG. 27(a) or a multi-layer perceptron neural network such as FIG. 27(b). The single-layer perceptron neural network of FIG. 27(a) is a learning model that consists of an input layer and an output layer, and no other layers exist between the input layer and the output layer. The input data value input to each node of the input layer is converted / corrected into an output data value multiplied by at least one weight corresponding to each node. At this time, the input data value may also be converted / corrected into an output value multiplied by at least one weight and then a certain bias value added.

[0389] The multilayer perceptron neural network of Figure 27 (b) is a deep learning model with one or more additional layers between the input and output layers. Figure 27 (b) shows an example of a learning model with two hidden layers between the input and output layers.

[0390] The input data value input to each node of the input layer is converted into first intermediate data that is multiplied by at least one weight corresponding to each node of the first hidden layer (by adding a bias value) and then input to the second hidden layer. The first intermediate data input to the second hidden layer is converted into second intermediate data that is multiplied by at least one weight corresponding to each node of the second hidden layer (by adding a bias value) and then input to the output layer. The second intermediate data input to the output layer can be finally converted / corrected into an output value that is multiplied by at least one weight corresponding to each node of the output layer (by adding a bias value).

[0391] Hereinafter, the present disclosure assumes and explains the error correction function using a single-layer perceptron neural network. However, the concept of the present disclosure can also be extended and interpreted to encompass multi-layer perceptron neural networks. For example, multiple W and b values, as described below, can be set, and the parameters of each of the multiple W and b values ​​can be modified as they pass through the multi-layer perceptron neural network.

[0392] Now, with reference to FIGS. 28 to 31, let us examine a method for determining an error correction function according to an embodiment of the present disclosure.

[0393] Referring to FIG. 28, the monitoring device (100) can arbitrarily set parameter values ​​of nodes existing in each of at least one layer of the learning model (S2801).

[0394] Fig. 29(a) shows an example of a learning model. Referring to Fig. 29(a), a learning model for correcting the state data measured by the monitoring device (100) may include an error correction matrix and an error correction bias, and the error correction matrix may include w for n nodes. 11 From w nn It can include error correction parameters from b1 to b. In addition, the error correction bias is nIt may include error correction parameters up to .

[0395] In step S2801, the monitoring device (100) w 11 From w nn Error correction parameters from b1 to b n The error correction parameters can be set arbitrarily.

[0396] The monitoring device (100) can measure state data by repeating attachment and detachment to the chamber (10) M times while the chamber (10) is in the same state (e.g., idle state), thereby preparing M sets of state data (S2803).

[0397] For example, referring to FIG. 30, a chamber (10) in a specific process state (e.g., an idle state) is prepared (S3001), and a monitoring device (100) is attached to the outside of the chamber (10) (S3003). The monitoring device (100) can transmit radio waves of a specific frequency range into the chamber (10) and receive radio waves reflected from the inside of the chamber (10). In addition, the monitoring device (100) can measure state data of the chamber (10) using the received radio waves and store the measured state data in the storage unit (150) (S3005). The monitoring device (100) is detached from the chamber (10) (S3007), and S3003 to S3007 are repeatedly performed M times, so that the monitoring device (100) can prepare M state data.

[0398] The monitoring device (100) can input the first state data among the M state data into the learning model (S2805). The monitoring device (100) can obtain the first output state data based on the first state data from the learning model (S2807).

[0399] For example, referring to FIG. 29(b), each of the M state data measured by the monitoring device (100) may include n parameters, and each of the n parameters may be input to each of the n nodes of the learning model.

[0400] When each of the n parameters is input through each node, each of the n parameters is input into the w of the error correction matrix. 11 From w nn Multiplied by the error correction parameters up to and the error correction bias b1 to b n In addition to the error correction parameters, it is output as output status data.

[0401] The monitoring device (100) calculates difference information between the first output state data and the reference state data. For example, the monitoring device (100) can calculate difference values ​​between corresponding parameters among n parameters included in the first output state data and n parameters included in the reference state data, and generate difference information based on the calculated difference values. For example, the difference information can be determined using the MSE (mean square error) of [Mathematical Formula 4] below.

[0402]

[0403] Here, z is the reference state data, and y is the output state data. That is, i can have values ​​from 1 to n, z1 is the first parameter value among n parameters included in the reference state data, and z n is the nth parameter value among the n parameters included in the reference state data. Similarly, y1 is the first parameter value among the n parameters included in the output state data, and y n is the nth parameter value among the n parameters included in the output status data.

[0404] However, the method of using MSE and [Mathematical Formula 4] are only examples for calculating difference information and should not be interpreted as being limited thereto. For example, in addition to [Mathematical Formula 4], the monitoring device (100) may determine difference information using the mean error, which is the average of difference values, or the mean standard error, which is the standard error value of difference values. As another example, the monitoring device (100) may determine difference information using the median error, which is the center value of difference values, or the mode error, which is the mode of difference values.

[0405] If the difference information is not included in the predetermined range (No), the monitoring device (100) can update the parameters of the nodes present in each of at least one layer of the learning model (S2813). For example, if the MSE value is not included in the predetermined range, the monitoring device (100) can update the parameters of the error correction matrix w 11 From w nn Error correction parameters and error correction bias b1 to b n The error correction parameters can be updated up to .

[0406] For example, the monitoring device (100) uses the partial differential value of the difference information (e.g., MSE) as in [Mathematical Formula 5] to calculate the w of the error correction matrix. 11 From w nn Error correction parameters and error correction bias b1 to b n The error correction parameters can be updated up to .

[0407]

[0408] Here, lr (learning rate) is an adjustment rate for adjusting error correction parameters and can be preset.

[0409] The monitoring device (100) returns to step S2805, inputs the second state data among the M state data into the learning model with updated error correction parameters, and proceeds to step S2109 to generate difference information between the second output state data and the reference state data. In addition, if the difference information is not included in a predetermined range, the monitoring device (100) updates the error correction parameters again based on the difference information (S2813), and returns to step S2805 to input the third state data into the learning model.

[0410] In this way, the monitoring device (100) repeats steps S2805 to S2813 until the difference information is included in a predetermined range, and if the difference information is included in the predetermined range (Yes), it determines an error correction function using the error correction matrix and error correction bias updated until then (S2811).

[0411] Alternatively, the monitoring device (100) may repeatedly perform steps S2805 to S2813 a preset number of times and determine an error correction function using the error correction matrix and error correction bias when the preset number of repetitions is completed. In this case, step S2809 may be omitted. In other words, the monitoring device (100) may repeatedly perform steps S2805, S2807, and S2813 a preset number of repetitions without determining whether the difference information is included within a preset range.

[0412] Alternatively, the monitoring device (100) may repeatedly perform S2805 to S2813 for a preset number of repetitions and determine an error correction function using an error correction matrix and an error correction bias when the preset number of repetitions is completed. However, if the difference information is included within a preset range before the preset number of repetitions is reached, the error correction function may be determined using the error correction matrix and error correction bias updated until then.

[0413]

[0414] 3. Example 5-2: Method for determining a measurement error correction function by repeated measurements in multiple chambers.

[0415] As described above, even if the monitoring device (100) acquires status data for multiple chambers manufactured through the same manufacturing process in the same process state (e.g., idle state), different status data may be acquired due to tolerances, etc.

[0416] Therefore, in order to compensate for these measurement errors, it is necessary to determine an error compensation function to compensate for the measurement errors between multiple chambers.

[0417] Basically, the monitoring device (100) can also determine an error correction function for compensating for measurement errors between multiple chambers through the same principles and processes as in [Example 5-1]. In other words, the description of [Example 5-1] can be easily understood by replacing "measuring status data by repeating attachment and detachment for one chamber" with "measuring status data for different chambers."

[0418] Therefore, any content that overlaps with [Example 5-1] will be omitted.

[0419] In other words, the error correction function determination method described in FIG. 21 can be used in the same manner in the present embodiment. However, the only difference from [Example 5-1] is the process of preparing M state data in step S2803. In the description of [Example 5-1] described above, step S2803 was described with reference to FIG. 30.

[0420] On the other hand, in the present embodiment, the process of preparing M state data in step S2803 may proceed according to FIG. 31.

[0421] For example, referring to FIG. 31, in order to prepare M state data according to the present embodiment, M chambers in a specific process state (e.g., idle state) can be prepared (S3101).

[0422] Among the M chambers, a monitoring device (100) can be attached to the first chamber (S3103). The monitoring device (100) can transmit radio waves of a specific frequency range into the first chamber and receive radio waves reflected from the first chamber.

[0423] In addition, the monitoring device (100) can measure the status data of the first chamber using the received radio waves and store the measured status data in the storage unit (150) (S3105). By detaching the monitoring device (100) from the first chamber (S3107) and repeatedly performing S3103 to S3107 M times, the monitoring device (100) can prepare M status data.

[0424] When the monitoring device (100) acquires M state data according to the above-described method, the monitoring device (100) can determine an error correction function by repeatedly performing S2805 to S2813 of FIG. 28 until the difference information between the output state data output from the learning model and the reference state data is included within a predetermined range (S2811).

[0425] In this embodiment, the method by which the monitoring device (100) determines the error correction function, i.e., the method of performing S2805 to S2813 and S2811, is the same as that described in [Example 5-1], so the description will be omitted.

[0426] Meanwhile, if the number of chambers that are the subject of learning in [Example 5-2] is sufficiently large, it can also be used to correct measurement errors in chambers other than the chambers that are the subject of learning.

[0427]

[0428] 4. Example 5-3: Generating a detachment error correction function for each of a plurality of chambers, thereby generating an error correction function to which Example 5-1 and Example 5-2 can be applied simultaneously.

[0429] Meanwhile, in order to create an error correction function that can simultaneously correct the detachment measurement error of [Example 5-1] and the measurement error between different chambers of [Example 5-2], M state data can be secured as follows.

[0430] First, P chambers can be prepared. At this time, all P chambers can be products of the same model produced through the same process. However, as described above, process errors may exist. In addition, all P chambers for collecting M status data are in the same first process state (e.g., idle state). Among the P chambers, a monitoring device (100) is attached to the first chamber, and the status data of the first chamber is measured. Then, while the first chamber is in the same first process state, the monitoring device (100) can be repeatedly attached and detached from the first chamber Q times, thereby measuring Q status data for the first chamber.

[0431] Next, the monitoring device (100) is attached to the second chamber among the P chambers, and the status data of the second chamber is measured. Then, while the second chamber is in the same first process state, the monitoring device (100) is repeatedly attached and detached from the second chamber Q times, and Q status data for the second chamber can be measured.

[0432] In the same manner as described above, the monitoring device (100) can obtain Q status data for each of the P chambers. That is, the monitoring device (100) can collect PxQ status data in the manner described above. At this time, PxQ = M.

[0433] When the monitoring device (100) acquires M state data according to the above-described method, the monitoring device (100) can determine an error correction function by repeatedly performing S2805 to S2813 of FIG. 28 until the difference information between the output state data output from the learning model and the reference state data is included within a predetermined range (S2811).

[0434] In this embodiment, the method by which the monitoring device (100) determines the error correction function, i.e., the method of performing S2805 to S2813 and S2811, is the same as that described in [Example 5-1], so the description will be omitted.

[0435]

[0436] 5. Example 5-4: Method for correcting errors using an error correction function

[0437] Now, let us look at a method of correcting state data using at least one error correction function generated using at least one of [Example 5-1], [Example 5-2], and [Example 5-3] with reference to FIG. 32.

[0438] A monitoring device (100) is attached to the outside or inside of a target chamber. The monitoring device (100) can transmit radio waves of a specific frequency range into the inside of the target chamber and receive radio waves reflected from the target chamber. The monitoring device (100) can measure status data (e.g., a set of S parameters as a response to each frequency) using the received radio waves (S3201).

[0439] The monitoring device (100) can input status data into an error correction function and obtain corrected status data from the error correction function (S3203).

[0440] For example, an error correction function for correcting state data may use at least one of the error correction functions that can be generated using at least one of [Example 5-1], [Example 5-2], and [Example 5-3].

[0441] As a specific example, the monitoring device (100) may use one of the following four methods to obtain the corrected status data.

[0442] (1) In order to correct an error according to a chamber to which a monitoring device (100) is attached, state data can be input into the first error correction function generated by [Example 5-2], and corrected state data can be obtained from the first error correction function.

[0443] (2) In order to correct the error according to the attachment position of the monitoring device (100), the state data can be input into the second error correction function generated by [Example 5-1], and the corrected state data can be obtained from the second error correction function.

[0444] (3) In order to correct both the error according to the attachment position of the monitoring device (100) and the error according to the chamber to which the monitoring device (100) is attached, the state data can be input into the third error correction function generated by [Example 5-3], and the corrected state data can be obtained from the third error correction function.

[0445] (4) As another example, in order to correct both the error according to the attachment position of the monitoring device (100) and the error according to the chamber to which the monitoring device (100) is attached, the first error correction function generated by [Example 5-1] and the second error correction function generated by [Example 5-2] may be used.

[0446] For example, referring to FIG. 21, the monitoring device (100) inputs the state data into the first error correction function generated by [Example 5-2] (S3203-1) to correct the measurement error due to the tolerance of the plurality of chambers in the measured state data, and obtains the first output state data from the first error correction function (S3203-3). In addition, the monitoring device (100) inputs the state data into the second error correction function generated by [Example 5-1] (S3203-5) to correct the measurement error due to attachment / detachment of the first corrected state data, and obtains the second output state data from the second error correction function (S3203-7). The monitoring device (100) can determine the second output state data as the final corrected state data.

[0447] At this time, it is not allowed to input the status data measured by the monitoring device (100) into the second error correction function to first obtain the second output status data, and then input the second output status data into the first error correction function to obtain the first output status data. This is because first correcting the process error between chambers and then correcting the attachment / detachment for one chamber can further increase the accuracy of the correction of the measurement error.

[0448]

[0449] The monitoring device (100) can obtain similarity by comparing the corrected state data obtained by any one of (1) to (4) in step S3203 described above with reference state data, and can obtain state similarity data using the obtained similarity (S3205).

[0450] The monitoring device (100) can generate monitoring information of the target chamber based on the state similarity data (S3207).

[0451] At this time, steps S3205 and S3207 may be performed based on Embodiment 3. For example, at least one error correction function is generated according to at least one of the embodiments described in Embodiment 5, and the monitoring device (100) may input the state data of the target chamber measured into the error correction function to obtain corrected state data. Then, when performing at least one of [Embodiment 3-1], [Embodiment 3-2], and [Embodiment 3-3], the monitoring device (100) may use the state data described in [Embodiment 3-1], [Embodiment 3-2], and [Embodiment 3-3] as the corrected state data to perform at least one of [Embodiment 3-1], [Embodiment 3-2], and [Embodiment 3-3].

[0452] In other words, the status data measured by the monitoring device (100) can be corrected using at least one of [Example 5-1] to [Example 5-3], and the corrected status data based on at least one of [Example 3-1] to [Example 3-3] can be used to determine whether the target chamber is operating normally or to predict the time of abnormal operation.

[0453] That is, at least one of [Example 5-1] to [Example 5-3] and at least one of [Example 3-1] to [Example 3-3] can be applied together in sequence, and in this case, the “state data” of [Example 3-1] to [Example 3-3] should be understood as the “corrected state data” of [Example 5-1] to [Example 5-3].

[0454]

[0455] 6. Experimental data

[0456] Figures 33 to 36 show that when state data is measured while repeatedly attaching and detaching the monitoring device (100) to the chamber (10) without correction according to [Example 5-1], an error correction function is created through a learning model, and the measurement error of the measured state data is corrected by the created error correction function.

[0457] Meanwhile, in Fig. 33, the vertical axis shows the value of state similarity data, and the horizontal axis shows the number of repeated attachment / detachment.

[0458] Figures 33 and 34 are graphs showing the similarity calculated by repeatedly attaching and detaching a monitoring device (100) to a chamber (10) without compensation while measuring state data, comparing the state data with reference state data, and state similarity data based thereon, and the same graphs are displayed on different scales. At this time, the chamber (10) is always in the same state (e.g., idle state).

[0459] As can be seen in FIG. 33 and FIG. 34, it can be seen that the state similarity data becomes different simply by attaching and detaching the monitoring device (100) to the chamber (10).

[0460] Figure 35 is a graph representing state similarity data obtained based on the corrected state data after correcting the state data measured by the monitoring device (100) using an error correction function generated by sequentially inputting 30 state similarity data into the learning model. Figure 35 has the same Y-axis scale as Figure 33.

[0461] Looking at Figure 35, it can be seen that the state similarity data is measured significantly more uniformly compared to Figure 33.

[0462] Figure 36 is a graph showing state similarity data obtained based on the corrected state data after correcting the state data measured by the monitoring device (100) using an error correction function generated by sequentially inputting 70 state similarity data into the learning model.

[0463] Looking at Figure 36, we can see that the state similarity data is more uniform than Figures 33 and 35.

[0464] That is, the more state data is trained through the learning model, the more accurately the measurement error due to attachment and detachment is corrected by the error correction function.

[0465] Likewise, in [Example 5-2] and [Example 5-3] as well, the more the number of state data learned through the learning model increases, the more accurately the measurement errors caused by multiple chambers will be corrected by the error correction function. This is something that can be easily predicted by a person skilled in the art.

[0466]

[0467] [Monitoring System]

[0468] A monitoring system may be provided comprising one or more of the aforementioned monitoring devices. The monitoring system may be installed in process equipment comprising one or more chambers.

[0469] Figure 37 is a block diagram of a monitoring system according to one embodiment.

[0470] Referring to Fig. 37, the monitoring system may include one or more monitoring devices (100) and management devices (200). Here, the monitoring device (100) is the monitoring device (100) described above, so a redundant description thereof will be omitted.

[0471] The monitoring system may include a management device (200). The management device (200) may manage the monitoring device (100).

[0472] The management device (200) may include a communication unit (210), a storage unit (220), and a control unit (230).

[0473] The management device (200) can communicate with the outside world through the communication unit (210). For example, the management device (200) can obtain monitoring information from the monitoring device (100) through the communication unit (210). As another example, the management device (200) can transmit monitoring information to an external device, such as process equipment or a fab, through the communication unit (210). Duplicate descriptions of similar parts to the communication unit (210) of the monitoring device (100) are omitted.

[0474] The management device (200) can store various data and programs in the storage unit (220). For example, the storage unit (220) can store monitoring information. As another example, the storage unit (220) can store reference status data. Duplicate descriptions of similar parts to the storage unit (220) of the monitoring device (100) are omitted.

[0475] The control unit (230) can process and perform calculations on various types of information within the management device (200). The control unit (230) can control other components constituting the management device (200). Duplicate descriptions of similar parts to the control unit (230) of the monitoring device (100) are omitted.

[0476] FIG. 38 is a schematic diagram of a monitoring system installed in process equipment according to one embodiment, in which the monitoring device is integrated. Referring to FIG. 38, each chamber (10) of the process equipment (1) may be equipped with a monitoring device (100). The monitoring device (100) may generate status data of the chamber (10) in which it is installed. The monitoring device (100) may generate monitoring information based on the status data. Alternatively, the monitoring device (100) may transmit status data to a management device (200), and the management device (200) may generate monitoring information.

[0477]

[0478] Meanwhile, an error correction function can also be created using multiple monitoring devices (100). Even when multiple monitoring devices (100) measure a single chamber, measurement errors may occur due to tolerances or circuit characteristics between the multiple monitoring devices (100), and thus, there may be a need to correct such measurement errors.

[0479] For example, when Embodiment 5-1 is performed by exchanging different monitoring devices (100) for each attachment / detachment for one chamber (or every certain number of attachment / detachment operations (e.g., 1000 times)), error correction of multiple monitoring devices (100) is possible. In addition, if the parameter is sufficiently large, error correction of monitoring devices other than the multiple monitoring devices (100) is also possible. For example, if the number of monitoring devices (100) is R and the number of attachment / detachments of each monitoring device (100) is S, each monitoring device (100) can be attached / detached to a chamber, thereby preparing R x S status data. Accordingly, error correction according to attachment / detachment of multiple monitoring devices (100) can be performed using the R x S status data.

[0480] In addition, as mentioned in Example 5-2, by using a plurality of monitoring devices (100) to measure state similarity data for a plurality of chambers for each of the plurality of monitoring devices (100) to create a data set, error correction for the plurality of monitoring devices (100) and the plurality of chambers is possible. For example, if the number of monitoring devices (100) is R and the number of chambers is T, by using each of the monitoring devices (100) to measure state data for S chambers (10), R x T state data can be prepared. Accordingly, error correction for the plurality of chambers of the plurality of monitoring devices (100) can be performed using the R x T state data.

[0481] In addition, as mentioned in Example 5-3, by applying the method of creating a data set through attachment / detachment for one chamber while exchanging multiple monitoring devices (100) to different chambers, an error correction function for multiple monitoring devices (100) and multiple chambers can be generated, thereby enabling error correction. For example, if the number of monitoring devices (100) is R, the number of chambers (10) is T, and the number of attachment / detachments of one monitoring device (100) per chamber (10) is S, R x T x S state data can be prepared. Accordingly, error correction for multiple chambers of multiple monitoring devices (100) can be performed using R x T x S state data.

[0482] In this case, if the number of the plurality of monitoring devices (100) and / or the plurality of chambers is sufficiently large, measurement errors for monitoring devices (100) and chambers other than the plurality of monitoring devices (100) and / or the plurality of chambers can also be corrected.

[0483]

[0484] [Implementation examples of examples]

[0485] Meanwhile, as mentioned in the description of the above-described embodiments, the above-described embodiments can each be applied independently, but at least one of the above-described embodiments can also be implemented in combination.

[0486] For example, after generating reference state data according to [Example 1], an error correction function may be generated according to [Example 5], and the state data measured by the monitoring device (100) may be substituted into the error correction function to obtain corrected state data. In addition, the monitoring device (100) may utilize the corrected state data as the state data of [Example 3] or [Example 4], and may generate a state similarity pattern and a state similarity graph using [Example 2]. In addition, the monitoring device (100) may determine whether the target chamber is operating normally or predict the time of abnormal operation according to [Example 3] or [Example 4].

[0487] Meanwhile, based on the above, an error correction function created for a plurality of monitoring devices according to the monitoring system described in FIGS. 27 and 28 can also correct measurement errors as a combination of [Example 1] to [Example 5], determine whether the target chamber is operating normally, or predict the time of abnormal operation.

[0488]

[0489] The method according to an embodiment may be performed by processing logic including hardware, firmware, software, or a combination thereof. The method according to an embodiment may be performed by a processor executing code stored in a non-transitory computer-readable medium. Examples of the non-transitory computer-readable medium include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, and hardware devices specifically configured to store and execute program instructions such as ROMs, RAMs, and flash memories.

[0490]

[0491] Although the present disclosure has been described above based on examples, the present disclosure is not limited thereto, and it is obvious to those skilled in the art that various changes or modifications can be made within the spirit and scope of the present disclosure, and therefore, it is made clear that such changes or modifications fall within the scope of the appended patent claims.

Claims

1. In a method for a chamber monitoring device to monitor the state of a chamber, - Here, a process cycle including multiple processes occurs at least once within the chamber - Obtaining reference state data for monitoring the state of the chamber, wherein the reference state data is first state data of the chamber measured in an IDLE section of the chamber or second state data of the chamber measured in a section corresponding to any one of the plurality of processes; Transmitting radio waves of a specific frequency range M times into the interior of the chamber, and receiving radio waves reflected inside the chamber M times, where M is a natural number greater than or equal to 2; Measure the state of the chamber based on the M received radio waves to generate a plurality of third state data; Calculating similarities between each of the plurality of third state data and the reference state data, and obtaining a plurality of state similarity data using the similarities; By matching the above plurality of state similarity data with the process section data for the sections in which each of the above plurality of processes is performed, a plurality of state similarity patterns configured in process cycle units are obtained; and Among the plurality of state similarity patterns, determining whether the chamber is operating normally or predicting the time of abnormal operation of the chamber by using two or more state similarity patterns; Chamber condition monitoring method.

2. In paragraph 1, Predicting the abnormal operation point of the above chamber; Selecting two or more state similarity data from the above plurality of state similarity data, wherein one state similarity data corresponding to the first process is selected from each of the above plurality of state similarity patterns; Determining a function for predicting the abnormal behavior based on the two or more state similarity data, wherein the function is determined based on a time variable; and Predicting the first point in time at which the result value of the above function calculated according to the time variable becomes lower than the preset threshold similarity as the point in time of abnormal operation of the chamber; Chamber condition monitoring method.

3. In paragraph 2, Predicting the abnormal operation point of the above chamber; This is performed repeatedly as additional state similarity patterns are acquired. Chamber condition monitoring method.

4. In paragraph 1, Determining whether the above chamber is operating normally; Among the above multiple state similarity patterns, the most recently acquired first state similarity pattern is selected; Among the plurality of state similarity patterns, a second state similarity pattern different from the first state similarity pattern is selected; Calculating difference values ​​between state similarity data of the first state similarity pattern and state similarity data of the second state similarity pattern according to each process and corresponding time point within each process; and Including, among the above-described calculated difference values, determining that the chamber is operating abnormally in a process corresponding to a difference value exceeding the critical similarity difference value; Chamber condition monitoring method.

5. In paragraph 4, Determining whether the above chamber is operating normally; Further comprising: determining the threshold similarity difference value according to the interval between the first state similarity pattern and the second state similarity pattern among the plurality of state similarity patterns; Chamber condition monitoring method.

6. In paragraph 4, Calculating difference values ​​between state similarity data of the first state similarity pattern and state similarity data of the second state similarity pattern; Correcting the state similarity data of the first state similarity pattern based on the state similarity data of the second state similarity pattern; and Comprising: calculating difference values ​​between state similarity data of the above-mentioned corrected first state similarity pattern and state similarity data of the above-mentioned second state similarity pattern; Chamber condition monitoring method.

7. In paragraph 6, Correcting the state similarity data of the above first state similarity pattern; Selecting two or more state similarity data from the above plurality of state similarity data, wherein one state similarity data corresponding to the first process is selected from each of the above plurality of state similarity patterns; Determining a function for predicting the abnormal behavior based on the two or more state similarity data, wherein the function is determined based on a time variable; and Compensating the state similarity data of the first state similarity pattern based on the state similarity data of the second state similarity pattern and the function; Chamber condition monitoring method.

8. In paragraph 4, The above second state similarity pattern is the state similarity pattern that is acquired first among the plurality of state similarity patterns. Chamber condition monitoring method.

9. In paragraph 1, Obtaining the above reference status data; In an idle section of the chamber or a section corresponding to one of the plurality of processes, transmitting a radio wave of the specific frequency range into the chamber N times, and receiving a radio wave reflected inside the chamber N times, where N is a natural number; Measure the state of the chamber N times based on the N received radio waves to generate a plurality of fourth state data; and Acquiring the reference state data based on the plurality of fourth state data; including; Chamber condition monitoring method.

10. In a chamber monitoring device that monitors the state of the chamber, - Here, a process cycle including multiple processes occurs at least once within the chamber - An antenna that transmits radio waves of a specific frequency range into the interior of the chamber and receives radio waves reflected from the interior of the chamber; A bracket for fixing the antenna to the outside of the chamber; A signal processing unit that applies an electric signal to the antenna and obtains an electric signal from the antenna; A control unit that controls the signal processing unit and generates status data of the chamber based on an electric signal obtained by the signal processing unit; The above control unit, Obtaining reference state data for monitoring the state of the chamber, wherein the reference state data is first state data of the chamber measured in an IDLE section of the chamber or second state data of the chamber measured in a section corresponding to any one of the plurality of processes; Controlling the signal processing unit so that the antenna transmits the radio wave M times and receives the radio wave reflected inside the chamber M times to obtain an electric signal according to the radio wave received M times, wherein M is a natural number greater than or equal to 2; By measuring the state of the chamber based on the acquired electric signal, multiple third state data are generated, Computing similarities between each of the plurality of third state data and the reference state data, and obtaining multiple state similarity data using the similarities, By matching the above plurality of state similarity data with the process section data for the sections in which each of the above plurality of processes is performed, a plurality of state similarity patterns configured in process cycle units are obtained, and Among the above multiple state similarity patterns, two or more state similarity patterns are used to determine whether the chamber is operating normally or to predict the time of abnormal operation of the chamber. Chamber monitoring device.

11. In Article 10, The above control unit, Among the plurality of state similarity data, two or more state similarity data are selected, - wherein the control unit selects one state similarity data corresponding to the first process from each of the plurality of state similarity patterns -; determining a function for predicting the abnormal operation time based on the two or more state similarity data, wherein the function is determined based on a time variable, and The point in time when the result value of the above function calculated according to the time variable becomes lower than the preset threshold similarity is predicted as the point in time when the chamber begins to operate abnormally. Chamber monitoring device.

12. In paragraph 11, The above control unit, Repeatedly predicting the abnormal operation point of the chamber as additional state similarity patterns are acquired. Chamber monitoring device.

13. In paragraph 10, The above control unit, Among the above multiple state similarity patterns, the most recently acquired first state similarity pattern is selected, Among the above multiple state similarity patterns, a second state similarity pattern different from the first state similarity pattern is selected, The difference values ​​between the state similarity data of the first state similarity pattern and the state similarity data of the second state similarity pattern are calculated according to each process and corresponding time points within each process, and Among the above calculated difference values, in the process corresponding to a difference value exceeding the critical similarity difference value, it is determined that the chamber is operating abnormally. Chamber monitoring device.

14. In paragraph 13, The above control unit, Among the plurality of state similarity patterns, determining the threshold similarity difference value according to the interval between the first state similarity pattern and the second state similarity pattern. Chamber monitoring device.

15. In paragraph 13, The above control unit, Correcting the state similarity data of the first state similarity pattern based on the state similarity data of the second state similarity pattern, Calculating difference values ​​between the state similarity data of the above-mentioned corrected first state similarity pattern and the state similarity data of the above-mentioned second state similarity pattern. Chamber monitoring device.

16. In paragraph 15, The above control unit, Among the above plurality of state similarity data, two or more state similarity data are selected, wherein, one state similarity data corresponding to the first process is selected from each of the above plurality of state similarity patterns. determining a function for predicting the abnormal behavior based on the two or more state similarity data, wherein the function is determined based on a time variable, and Correcting the state similarity data of the first state similarity pattern based on the state similarity data of the second state similarity pattern and the function. Chamber monitoring device.

17. In paragraph 13, The above second state similarity pattern is the state similarity pattern that is acquired first among the plurality of state similarity patterns. Chamber monitoring device.

18. In paragraph 10, The above control unit, In an idle section of the chamber or a section corresponding to one of the plurality of processes, transmitting a radio wave of the specific frequency range into the chamber N times and receiving a radio wave reflected inside the chamber N times, where N is a natural number; Based on the N received radio waves, the state of the chamber is measured N times to generate a plurality of fourth state data, and Obtaining the reference state data based on the above plurality of fourth state data, Chamber monitoring device.

Images