KR102252144B1 - Electronic apparatus for identifying operation of plasma and operating method thereof - Google Patents

Abstract

The present invention relates to an electronic apparatus for checking operation of plasma to efficiently check an operation state of the plasma and an operation method thereof. According to the present invention, a method for operating an electronic apparatus for checking operation of plasma comprises the following steps: acquiring a value of at least one parameter for plasma in a specific time unit; checking operation information corresponding to each of a plurality of time sections of the specific time unit on the basis of applying the value of at least one parameter acquired in the specific time unit to a first machine learning algorithm; checking operation information corresponding to each of the one or more operation sections including at least one of the plurality of time sections and distinguished from each other on the basis of operation information corresponding to each of the plurality of time sections; and checking final operation information of the plasma on the basis of the operation information corresponding to each of the one or more operation sections.

Description

Electronic device for checking plasma operation and its operation method {ELECTRONIC APPARATUS FOR IDENTIFYING OPERATION OF PLASMA AND OPERATING METHOD THEREOF}

The present disclosure relates to an electronic device for confirming an operation of a plasma using at least one of at least one parameter for plasma and operation information corresponding to a specific time period during a plasma process, and an operation method thereof.

Various equipments are used to manufacture semiconductor devices. For example, plasma equipment is used as an equipment for depositing a material film on a wafer or etching a material film formed on a wafer. Such plasma equipment may deposit or etch a material film by forming plasma in a sealed chamber in a vacuum state and injecting a reactive gas.

In order to check the operation of the plasma equipment, for example, an on/off state of the plasma equipment, a separate measuring instrument is mounted on the plasma equipment. Separate instruments include, for example, fiber sensors. The fiber sensor detects light generated during plasma ignition through the light receiving unit to check the operation state of the plasma.

In this way, since a separate measuring device is required in addition to the plasma equipment to check the operation state of the plasma equipment, additional costs may be incurred, as well as lowering the efficiency of using the plasma equipment and causing trouble. Accordingly, there is a need for a method to more easily check the operating state of the plasma equipment.

The problem to be solved by this embodiment is an electronic device that makes it easier and more effective to check the operation of plasma by using at least one parameter of plasma and operation information corresponding to a specific time period of the plasma process, and its operation. It is in providing a way. In addition, it is possible to check the operation of the plasma using parameters acquired during the process without installing a separate additional sensor. In addition, it is universally applied to various processes, so that the operation of plasma can be checked.

The technical problem to be achieved by the present embodiment is not limited to the technical problems as described above, and other technical problems may be inferred from the following embodiments.

According to a first embodiment, a method of operating an electronic device for checking an operation of a plasma includes acquiring a value of at least one parameter for the plasma in a specific time unit, and at least one Checking motion information corresponding to each of a plurality of time sections of the specific time unit based on applying a parameter value to a first machine learning algorithm, and based on motion information corresponding to each of the plurality of time sections, the Checking motion information corresponding to each of the one or more motion sections that include at least one of a plurality of time sections and are separated from each other, and determine the final motion information of the plasma based on motion information corresponding to each of the one or more motion sections. I can confirm.

According to a second embodiment, in a computer-readable non-transitory recording medium recording a program for executing an operation method of an electronic device for checking plasma operation on a computer, the operation method is applied to the plasma in a specific time unit. Acquiring a value of at least one parameter for and corresponding to each of a plurality of time intervals of the specific time unit based on applying the value of at least one parameter obtained in the specific time unit to a first machine learning algorithm Checking motion information, and checking motion information corresponding to each of one or more motion sections that are separated from each other and include at least one of the plurality of time sections based on motion information corresponding to each of the plurality of time sections. And, it is possible to check the final operation information of the plasma based on operation information corresponding to each of the one or more operation sections.

According to a third embodiment, an electronic device for checking plasma operation includes a memory for storing a command and a processor, and the processor is connected to the memory, and at least one parameter for the plasma in a specific time unit Acquire a value of, and check operation information corresponding to each of a plurality of time intervals of the specific time unit based on applying the value of at least one parameter obtained in the specific time unit to a first machine learning algorithm, and the Based on the motion information corresponding to each of the plurality of time sections, the motion information corresponding to each of the at least one motion section that includes at least one of the plurality of time sections and is separated from each other is checked, and the motion information corresponding to each of the one or more motion sections is checked. The final operation information of the plasma may be checked based on the operation information.

Details of other embodiments are included in the detailed description and drawings.

According to the present disclosure, the electronic device for checking the operation of the plasma and the operation method thereof include at least one of at least one parameter for the plasma and operation information corresponding to a specific time period of the plasma process to check the final operation information of the plasma. By doing so, it is possible to more efficiently check the operating state of the plasma using the plasma facility itself.

The effects of the invention are not limited to the effects mentioned above, and other effects that are not mentioned will be clearly understood by those skilled in the art from the description of the claims.

1 is a conceptual diagram illustrating a method of operating an electronic device for checking plasma operation according to an exemplary embodiment.
2 is a functional block diagram of an electronic device for checking a plasma state according to an exemplary embodiment.
3 is a flowchart illustrating a method of operating an electronic device for checking a plasma state according to an exemplary embodiment.
4 is a diagram illustrating at least one parameter used in an electronic device according to an exemplary embodiment.
5 is a diagram illustrating an example of motion information corresponding to a plurality of time periods of an electronic device according to an exemplary embodiment.
6 is a diagram illustrating an example of one or more operation periods of an electronic device according to an exemplary embodiment.
7 is a diagram illustrating an example of data used for learning a machine learning algorithm for confirming final operation information of plasma in an electronic device according to an embodiment.
8 is a diagram illustrating a neural network according to an embodiment.

As for terms used in the embodiments, general terms that are currently widely used as possible are selected while considering functions in the present disclosure, but this may vary according to the intention or precedent of a technician working in the field, the emergence of new technologies, and the like. In addition, in certain cases, there are terms arbitrarily selected by the applicant, and in this case, the meaning will be described in detail in the corresponding description. Therefore, the terms used in the present disclosure should be defined based on the meaning of the term and the overall contents of the present disclosure, not a simple name of the term.

When a part of the specification is said to "include" a certain component, it means that other components may be further included rather than excluding other components unless specifically stated to the contrary.

The expression of "at least one of a, b, or c" described throughout the specification is'a alone','b alone','c alone','a and b','a and c','b and c' ', or'a, b, and c all' may be encompassed.

The'electronic device' mentioned below may be referred to as an electronic device, and may be implemented as a computer or portable terminal capable of accessing a server or other electronic device through a network. Here, the computer includes, for example, a notebook equipped with a web browser, a desktop, a laptop, and the like, and the portable terminal is, for example, a wireless communication device that guarantees portability and mobility. , International Mobile Telecommunication (IMT), Code Division Multiple Access (CDMA), W-Code Division Multiple Access (W-CDMA), and all kinds of communication-based terminals such as LTE (Long Term Evolution), smartphones, tablet PCs, etc. It may include a handheld-based wireless communication device.

Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings so that those of ordinary skill in the art may easily implement the present disclosure. However, the present disclosure may be implemented in various different forms and is not limited to the embodiments described herein.

Hereinafter, exemplary embodiments of the present disclosure will be described in detail with reference to the drawings.

1 is a conceptual diagram illustrating a method of operating an electronic device for checking plasma operation according to an exemplary embodiment.

Referring to FIG. 1, the electronic device may acquire at least one parameter for plasma. For example, the electronic device may acquire the first parameter 111 to the n-th parameter 112. Each of the parameters acquired by the electronic device may be acquired in a specific time unit. For example, the first parameter 111 to the n-th parameter 112 may be obtained every 300 ms from the plasma facility. If the parameters are acquired for 900 ms, the electronic device may acquire a total of 3n parameter values by acquiring each of the first parameter 111 to the n-th parameter 112 in units of 300 ms, that is, three times.

In an embodiment, the at least one parameter is the pressure of the plasma chamber, the angle of the valve, the current applied to the plasma heating device, the temperature of the plasma heating device, the amount of process gas injected into the chamber, and the chamber. It may include at least one of a pressure of a gas injected to make it atmospheric pressure, a temperature of a heater chuck, and a temperature of a chamber. For example, the first parameter 111 may correspond to the pressure of the plasma chamber, and the n-th parameter 112 may correspond to the temperature of the plasma chamber.

The electronic device may apply a parameter acquired in a specific time unit to the first machine learning algorithm 121. In this case, the electronic device may check operation information corresponding to each of a plurality of time intervals of a specific time unit from the first machine learning algorithm 121. For example, when a parameter is acquired in units of 300ms for 900ms, the electronic device will have three time periods: a first time period from 0ms to 300ms, a second time period from 300ms to 600ms, and a third time period from 600ms to 900ms. It is possible to check the operation information of the plasma for each time period. For example, the plasma operation information is information on whether the plasma operation is being performed, and may be displayed in the form of, for example, on or off. For another example, the plasma operation information is all on indicating that both stages are in plasma ignition operation when the plasma processing chamber is plural (ex: two stages), and some on indicating that some of the stages are in operation ( Alternatively, it may appear in a form such as one on) or all off indicating that not all of the stages are in operation.

The first machine learning algorithm 121 may include, for example, a tree-based machine learning algorithm (eg, LightGBM). Tree-based machine learning algorithms can use the tree structure to set criterion values for independent variables and predict dependent variables through the set criterion values. Accordingly, the first machine learning algorithm 121 may predict plasma operation information based on the parameter being input.

Meanwhile, in an embodiment, a parameter input to the first machine learning algorithm 121 may be one-dimensional data. In this case, the electronic device may predict motion information by applying the acquired parameter to the first machine learning algorithm 121 in response to the parameter being acquired. That is, by obtaining a parameter for each specific time unit and applying it to the first machine learning algorithm 121, motion information based on the corresponding parameter may be predicted for each specific time. For example, when a parameter is acquired in the first time section, the parameter is applied to the first machine learning algorithm 121 to predict motion information for the first time section, and then, when the parameter is acquired in the second time section, it is determined. 1 It is possible to predict motion information for the second time interval by applying it to the machine learning algorithm 121.

In the case of analyzing motion information through the first machine learning algorithm 121 in real time in response to the acquisition of parameters as described above, the entire data is not converted into one-dimensional data, but is based on one-dimensional data input in a specific time unit. Since temporal characteristics are maintained through analysis, it is possible to prevent performance degradation of electronic devices.

Accordingly, the electronic device may check operation information for each of a plurality of time sections (eg, a first time section, a second time section, and a third time section) of a specific time unit. The electronic device may check one or more operation periods including at least one of the plurality of time periods based on the confirmation of operation information for each of the plurality of time periods.

The one or more operation sections may include, for example, at least one of a section previously designated in relation to the progression of the plasma process, for example, a process preparation section, a process finishing section, and an entire process section. Here, the process preparation section may correspond to a section for making the inside of the chamber into a vacuum state, the process finishing section may correspond to a section for releasing the vacuum state inside the chamber, and the process preparation section and the process finishing section may correspond to the inside of the chamber and It may contain a lot of relevant data. Here, the process preparation section may correspond to, for example, the initial k (pre-designated positive number) sections of the plurality of time sections, the process finishing section may correspond to the last k sections of the plurality of time sections, and the entire process section may correspond to the entire plurality of time sections. . When the plasma state is predicted using motion information corresponding to the divided operation sections such as the process preparation section, the process finishing section, and the entire process section, the accuracy of the plasma state prediction is improved compared to using only the motion information of the entire process section. The plasma state prediction method according to the process change may have versatility. However, such a pre-designated section is not fixed as a process preparation section/process finishing section/complete process section, and may be subdivided differently when a process (or recipe) is changed. For a more specific example related to this, refer to FIG. 6.

The electronic device may check motion information corresponding to one or more motion sections. For example, when one or more operation periods include a first operation period through an m-th operation period, the electronic device includes first operation information 131 corresponding to the first operation period through m-th operation information corresponding to the m-th operation period (132) can be found.

The electronic device may apply motion information corresponding to each of one or more motion sections to the second machine learning algorithm 141. The second machine learning algorithm 141 may include, for example, a rule-based machine learning algorithm. Specifically, for example, the second machine learning algorithm 141 uses motion information corresponding to each of one or more motion sections as an input value and the final motion information as a correct answer value (output value). When motion information is input to the second machine learning algorithm 141, it may be a pre-learned rule that outputs final motion information. The second machine learning algorithm may be pre-learned based on previously acquired input values and output values, and when motion information corresponding to one or more motion sections is input to the second machine learning algorithm, the final motion is based on the previously learned rules. Information can be printed. For a specific example of the input value and the correct answer value (output value), reference may be made to FIG. 7.

Accordingly, the electronic device may check the final operation information of the plasma using the rule according to the second machine learning algorithm 141. The final operation information may include, for example, information indicating whether a plasma is actually ignited or a plasma is generated in the chamber.

2 is a functional block diagram of an electronic device for checking plasma operation according to an exemplary embodiment. Although components related to the present embodiment are illustrated in FIG. 2, it is not limited thereto, and other general-purpose components may be further included in addition to the components illustrated in FIG. 2.

The electronic device 200 may include a memory 210 and a processor 220. Each element shown in FIG. 2 refers to a unit that processes at least one function or operation, which may be implemented by hardware or software, or a combination of hardware and software.

According to an embodiment, the electronic device 200 of FIG. 2 may be implemented as a server, a computer, or a terminal, and the present specification is not limited by the implementation method of the electronic device 200.

The memory 210 may store various data related to the electronic device 200. For example, the memory 210 may store at least one instruction for the operation of the electronic device 200. In this case, the processor 220 to be described later may perform various operations based on instructions stored in the memory 210. For another example, the memory 210 may store information on a machine learning algorithm used for the operation of the electronic device 200 or information on a parameter. In this case, the processor 220 may perform an operation using information stored in the memory 210. However, the information stored in the memory 210 is not limited to the above-described example, and various types of information may be stored in the memory 210.

The processor 220 may control the overall operation of the electronic device 200. For example, the processor 220 may control the operation of the electronic device 200 by controlling elements of the electronic device 200 based on a command stored in the memory 210.

The processor 220 may be composed of one or more cores, and a central processing unit (CPU) of the electronic device 200, a general purpose graphics processing unit (GPGPU), and a tensor processing unit (TPU). : Tensor processing unit), such as data analysis, may include a processor for deep learning. The processor 220 may estimate the density of plasma according to an embodiment of the present disclosure by reading a computer program stored in the memory 210. According to an embodiment of the present disclosure, the processor 220 may perform an operation for learning a neural network. The processor 220 is a neural network such as processing input data for learning in deep learning (DN), extracting features from the input data, calculating errors, and updating the weights of the neural network using backpropagation. You can perform calculations for learning. At least one of the CPU, GPGPU, and TPU of the processor 220 may process the learning of the neural network. In addition, in an embodiment of the present disclosure, data may be processed using a neural network learned by using a plurality of processors together. Also, a program executed in the electronic device 200 according to an embodiment of the present disclosure may be a CPU, a GPGPU, or a TPU executable program. In an embodiment of the present disclosure, data processed using a neural network may include all types of data acquired at an industrial site. For example, it may include an operation parameter of a device for production of a product in a product production process, sensor data obtained by an operation of the device, and the like. For example, a temperature setting of an equipment in a specific process, a wavelength of a laser, etc. in the case of a process using a laser may be included in the type of data processed in the present disclosure. For example, the processed data includes lot equipment history data from a management execution system (MES), processing tool recipes, processing tool test data, probe test data, electrical test data, combined measurement data, and diagnostics. Data, remote diagnosis data, post-processing data, and the like may be included, and the present disclosure is not limited thereto. As a more specific example, raw processing tool data obtained from a semiconductor fab, equipment interface information, process metrology information, operation test information, etc. may be included, but the present disclosure is not limited thereto. Does not. The description of the types of data described above is only an example, and the present disclosure is not limited thereto.

The processor 220 may obtain a value of at least one parameter for plasma in a specific time unit. A specific time unit and at least one parameter may be specified in advance. For example, the specific time unit is 300 ms, and at least one parameter is the pressure of the plasma chamber, the valve angle, and the current applied to the plasma heating device (or the percentage of the current applied to the total current that can be applied to the plasma heating device (%). Value), the temperature of the plasma heating device, the current applied to the heater chuck (or the percentage (%) value of the current currently applied among the total current that can be applied to the chuck heating device), the temperature of the heater chuck, the process injected into the chamber It may include at least one of the amount of gas, the amount of gas injected to bring the chamber to atmospheric pressure, and the temperature of the chamber. Here, the valve angle may represent the position (or information indicating the position) of the valve for adjusting the pressure in the chamber to the set value through a device that adjusts the pressure in the chamber to the set value of the user called APC (adaptive pressure control). . That is, the valve angle may represent a position value for the degree of opening and closing in order to control the pressure through the valve. The amount of the process gas injected into the chamber may represent the amount being injected in real time when the process gas is being injected into the chamber. The pressure of the gas injected to bring the chamber to atmospheric pressure may represent a supply pressure of _{a gas (eg, N 2} gas) injected to bring the chamber from a vacuum to atmospheric pressure after completion of the plasma process.

In an embodiment, the processor 220 may check (or acquire) at least one parameter through equipment related to the operation of the plasma (eg, equipment used for a plasma process). For example, the plasma chamber pressure can be controlled by the APC angle value, and the current and temperature applied to the heater chuck can be obtained from the heater chuck. The amount of the process gas injected into the chamber and the pressure of the injected gas to bring the chamber to atmospheric pressure may be obtained through a gas injection device (or a gas monitoring device). The temperature of the chamber can be obtained from the chamber.

The processor 220 may check operation information corresponding to each of a plurality of time intervals of a specific time unit based on applying the value of at least one parameter acquired in a specific time unit to the first machine learning algorithm.

For example, the processor 220 may obtain the value of at least one parameter for each specific time unit. When the parameter value is obtained, the processor 220 may input the parameter value to the first machine learning algorithm in real time. In this case, the value of at least one parameter may be input to the first machine learning algorithm at a specific time interval.

Here, the first machine learning algorithm may be an algorithm learned to check plasma operation information corresponding to a specific time unit based on at least one parameter of a specific time unit. Accordingly, the processor 220 may check plasma operation information for each of a plurality of time intervals having a specific time unit. A specific example related to this may refer to FIGS. 4 and 5.

In an embodiment, the first machine learning algorithm may include a tree-based machine learning algorithm. The tree-based machine learning algorithm may include, for example, LightGBM, a kind of gradient boosting decision tree model, but is not limited thereto.

The processor 220 includes at least one of a plurality of time intervals based on operation information corresponding to each of the plurality of time intervals, and may check operation information corresponding to each of one or more operation intervals that are separated from each other. The one or more operation periods are, for example, a first section including at least one time section before the first time point among a plurality of time sections, and a second section including at least one time section after the second time point among the plurality of time sections. It may include at least one of a section and a third section including all of the plurality of time sections. In this case, the first viewpoint and the second viewpoint may be predetermined viewpoints. For another example, at least one operation section may correspond to an initial k (a positive number previously designated) sections, a process finishing section may correspond to a final k section among a plurality of time sections, and an entire process section may correspond to the entire plurality of time sections. For a more specific example related to this, refer to FIG. 6.

In an embodiment, the processor 220 may check motion information corresponding to the motion section based on motion information corresponding to each time section included in the motion section. When one or more operation intervals include a first operation interval and the first operation interval includes at least three or more time intervals among a plurality of time intervals, the processor 220 stores at least three motion information corresponding to the first operation interval. It can be checked to correspond to information indicated by a number of motion information corresponding to each of the above time intervals. Specifically, for example, the first motion section includes a first time section, a second time section, and a third time section, and motion information of each of the first time section and the second time section is on, and When the motion information is off, the processor 220 may check motion information corresponding to the first motion period as a value indicated by the plurality of motion information, that is, on. The processor 220 may check the final operation information of the plasma based on operation information corresponding to each of one or more operation sections. For example, when one or more operation periods include three operation periods, the final operation information of the processor 220 may be determined as information indicated by a plurality of operation information corresponding to the three operation periods. More specifically, for example, when the motion information of the first motion section is on and the motion information of the second motion section and the third motion section is off, the processor 220 may check the final motion information as off.

In an embodiment, the processor 220 may check the final motion information of the plasma based on applying motion information corresponding to each of the one or more motion sections to the second machine learning algorithm. For example, if the processor 220 inputs motion information indicated by each of one or more motion sections into the second machine learning algorithm, the second machine learning algorithm may output final motion information of the plasma according to a previously learned rule. In this case, the second machine learning algorithm may be learned to check the final operation information of the plasma based on the operation information of the plasma for each of the one or more operation sections.

In an embodiment, the second machine learning algorithm may include a rule-based machine learning algorithm. In this case, the processor 220 learns the second machine learning algorithm by using various data on motion information corresponding to the motion section, for example, motion information for each motion section and a large amount of data representing final motion information corresponding thereto. You can do it. An example of such data may be referred to FIG. 7.

In an embodiment, the processor 220 may check plasma operation setting information. The plasma operation setting information may be set by a user or another electronic device as information for setting whether or not to operate the plasma, and may be stored in advance in the electronic device 200. The plasma operation setting information may include, for example, on or off type information.

In an embodiment, the processor 220 may check whether the final motion information predicted through the second machine learning algorithm corresponds to the motion setting information. For example, when the final motion information is predicted to be on through the second machine learning algorithm, the processor 220 may check whether the plasma motion setting information is also on.

In an embodiment, when the final operation information corresponds to the operation setting information, the processor 220 may request the start of a semiconductor process using plasma. The processor 220 may provide an alarm when the final operation information does not correspond to the operation setting information. The alarm may be provided using at least one of text, image, and sound, for example. In this case, the electronic device 200 may allow the user to check the plasma process by recognizing that there is an error in the operation state of the plasma.

3 is a flowchart illustrating a method of controlling an electronic device for checking plasma operation according to an exemplary embodiment. Each step of the method illustrated in FIG. 3 may be performed in a different order from that illustrated in the drawings depending on the case. Hereinafter, content overlapping with the previously described content may be omitted.

Referring to FIG. 3, in operation 310, the electronic device may obtain a value of at least one parameter for plasma in a specific time unit. For example, when a specific time unit is 300 ms and at least one parameter includes a first parameter and a second parameter, the electronic device may acquire the first parameter and the second parameter for the plasma every 300 ms. In this case, the first parameter and the second parameter may be obtained through equipment used to perform the plasma process. In this case, since the plasma operation can be performed without additional equipment for confirming the plasma operation, process efficiency can be improved.

In operation 320, the electronic device may check operation information corresponding to each of a plurality of time intervals of a specific time unit based on applying a value of at least one parameter acquired in a specific time unit to the first machine learning algorithm.

In an embodiment, the electronic device may apply the value of the at least one parameter to the first machine learning algorithm in response to obtaining the value of the at least one parameter in a specific time unit. In this case, a value of at least one parameter may be input to the first machine learning algorithm every specific time. The entire time interval for checking the operation of the plasma may correspond to a plurality of time intervals, and in this case, the plurality of time intervals may be understood as dividing the entire time interval into a specific time unit.

In operation 330, the electronic device may check operation information corresponding to each of one or more operation intervals, which include at least one of the plurality of time intervals, based on operation information corresponding to each of the plurality of time intervals.

In an embodiment, each of the one or more operation intervals may include at least three or more time intervals. In this case, the electronic device may determine motion information corresponding to the motion section as motion information indicated by a plurality of the at least three time sections.

In operation 340, the electronic device may predict final operation information of the plasma based on operation information corresponding to each of one or more operation periods (eg, a process preparation period, a process completion period, and an entire process period).

In an embodiment, the electronic device may confirm final motion information of plasma by applying motion information corresponding to each of one or more motion sections to the second machine learning algorithm.

4 is a diagram illustrating at least one parameter used in an electronic device according to an exemplary embodiment. Specifically, FIG. 4 is a conceptual diagram for explaining an operation in which a parameter is acquired in a specific time unit.

Referring to FIG. 4, parameters measured from equipment used in the plasma process may be measured in a specific time unit, and thus may be expressed as a two-dimensional matrix having a time axis and a parameter axis as shown in FIG. 4. However, when it is divided into a specific time unit, since it is possible to check only the parameter value corresponding to the specific time unit, it can be converted into one-dimensional data (parameter value). In this case, the specific time unit may correspond to the length of the rectangular time axis corresponding to the first time section 410 and the n-th time section 420 of FIG. 4. Assuming that the time axis gradually increases from left to right, the parameter may be obtained first in the first time period 410 and later in the n-th time period 420.

In an embodiment, the times of each of the first time section 410 to the nth time section 410 may be divided. For example, each of the first time section 410 to the nth time section 410 may be a section for a different time. Specifically, for example, the first time section 410 corresponds to a time section from 09:43:41 seconds to less than 43 seconds, and the second time section (not shown) is from 09:43:43 seconds to less than 45 seconds It may correspond to the time interval up to. However, the present invention is not limited thereto, and in some cases, at least some of the first time section 410 to the nth time section 410 may include the same time. Specifically, for example, if the first time section 410 corresponds to a time section ranging from 09:43:41 seconds to less than 43 seconds, the second time section (not shown) is from 09:43:42 seconds to 44 seconds It may correspond to a time period up to less than.

In an embodiment, motion information may be classified for each time period. For example, as shown in FIG. 4, the operation information of the first time section 410 is confirmed as on. Operation information of the n-th time period 420 may be identified as off.

5 is a diagram illustrating an example of motion information corresponding to a plurality of time periods of an electronic device according to an exemplary embodiment. Specifically, FIG. 5 shows an example in which motion information for all a plurality of time sections is provided in a graph as motion information for each of a plurality of time sections of a specific time unit is identified.

Referring to FIG. 5, a plurality of time intervals may include approximately 2000 time intervals, and a specific time unit may be 105 ms (unit indication is omitted on the graph). That is, operation information for each of 2000 time intervals is checked in units of 105 ms, so that information in the form of a graph may be provided as shown in FIG. 5 shows that motion information for 2000 time intervals is identified as one of all on, 1 on (one on), and all off by the first machine learning algorithm.

In FIG. 5, when there are a plurality of plasma ignition devices, the plasma operation information is all on (all on) indicating that all ignition devices are operating, and partial on (or one on) indicating that some of the ignition devices are operating. , Or all off, indicating that not all of the ignition devices are operating.

In this case, on a graph in which the x-axis is the time axis and the y-axis is the motion information axis, the position of the motion information corresponding to each time section is displayed, so that graph information as shown in FIG. 5 may be provided.

6 is a diagram illustrating an example of one or more operation sections of an electronic device according to an exemplary embodiment.

The graph of FIG. 6 may correspond to the graph of FIG. 5. Referring to FIG. 6, one or more operation periods may include a first operation period 610, a second operation period 620, and a third operation period 630. The first operation period 610 includes, for example, k initial (eg, 40) time periods, and the second operation period 620 includes k time periods in the late (or late) period, and the third The operation period 630 may include all of the plurality of time periods (eg, about 2000).

Motion information corresponding to each of the one or more motion sections may be identified based on motion information included in each motion section. For example, motion information corresponding to the first motion section may be determined as information indicated by a number of motion information included in the first motion section, and the second motion is information indicated by a number of motion information included in the second motion section. Motion information corresponding to the section may be determined, and motion information corresponding to the third motion section may be determined as information indicated by a plurality of motion information included in the third motion section. For another example, the value of motion information corresponding to all on is 0, the value of motion information corresponding to 1 on is 1. The value of motion information corresponding to all off is 2, and the average of all motion information for each section As a result, motion information corresponding to the motion section may be determined. If the average value does not appear as an integer, motion information may be determined through a method such as rounding, but is not limited thereto.

7 is a diagram illustrating an example of data used for learning a machine learning algorithm for confirming final operation information of plasma in an electronic device according to an exemplary embodiment.

Referring to FIG. 7, for each of one or more operation periods, that is, a first operation period, a second operation period, and a third operation period, a rule for operation information and a correct answer corresponding thereto may be prepared in advance. The rule for the correct answer (output) may correspond to the final operation information of the plasma.

In an embodiment, rules may be provided for every number of cases according to the number of values that one or more motion sections and motion information may have. For example, if at least one motion section is a first motion section, a second motion section, and a third motion section, and motion information is all on, one on, and all off, a total of 27 cases can be derived. Rules can be established for each of the numbers. In this case, a correct answer value (output value) may be pre-designated for each data for each of the 27 cases, and learning of the second machine learning algorithm may be performed based on this. Specifically, the correct answer may be determined mainly based on the third operation section (eg, the entire process section), but there may be a case where the motion information corresponding to the third operation section and the correct answer are different. In this case, the correct answer may be determined in consideration of motion information for the first motion section and the second motion section. Therefore, rather than using only motion information for the entire process section, motion information through rules according to the second machine learning algorithm can derive more accurate values.

According to an embodiment, when the operation section is divided into n according to a change in a process or a recipe, the second machine learning algorithm may be learned in advance based on motion information for the n operation sections. In this way, a universal model that can be applied to a changed process can be created.

8 is a diagram illustrating a neural network according to an embodiment.

In the present specification, the first machine learning algorithm may be a tree-based machine learning algorithm, and the second machine learning algorithm may be a rule-based machine learning algorithm. As a specific example, the first machine learning algorithm may be implemented as LightGBM that grows a tree through a leaf-wise method, but is not limited thereto.

In the present specification, the following contents may be applied to the first machine learning algorithm and the second machine learning algorithm. In this specification, a neural network, a neural network, and a model may be used with the same meaning. A neural network can be made up of a set of interconnected computational units, which can generally be referred to as “nodes”. These “nodes” may also be referred to as “neurons”. The neural network is composed of at least one or more nodes. The nodes (or neurons) that make up neural networks can be interconnected by one or more “links”.

In a neural network, one or more nodes connected through a link may relatively form a relationship between an input node and an output node. The concepts of an input node and an output node are relative, and any node in an output node relationship with respect to one node may have an input node relationship in a relationship with another node, and vice versa. As described above, the input node to output node relationship can be created around the link. One or more output nodes may be connected to one input node through a link, and vice versa.

In the relationship between the input node and the output node connected through one link, the value of the output node may be determined based on data input to the input node. Here, a node interconnecting the input node and the output node may have a weight. The weight may be variable, and in order for the neural network to perform a desired function, it may be changed by a user or an algorithm. For example, when one or more input nodes are interconnected to one output node by respective links, the output node includes values input to input nodes connected to the output node and a link corresponding to each input node. The output node value may be determined based on the weight.

As described above, in a neural network, one or more nodes are interconnected through one or more links to form an input node and an output node relationship in the neural network. The characteristics of the neural network may be determined according to the number of nodes and links in the neural network, the association relationship between the nodes and the links, and a weight value assigned to each of the links. For example, when there are the same number of nodes and links, and two neural networks having different weight values between the links exist, the two neural networks may be recognized as being different from each other.

A neural network may be configured including one or more nodes. Some of the nodes constituting the neural network may configure one layer based on the distances from the initial input node. For example, a set of nodes with a distance n from the initial input node, n layers can be configured. The distance from the first input node may be defined by the minimum number of links that must be traversed to reach the node from the first input node. However, the definition of such a layer is arbitrary for explanation, and the order of the layer in the neural network may be defined in a different way than that described above. For example, the layer of nodes may be defined by the distance from the last output node.

The first input node may mean one or more nodes to which data is directly input without going through a link in a relationship with other nodes among nodes in the neural network. Alternatively, in a relationship between nodes based on a link in a neural network network, it may mean nodes that do not have other input nodes connected by a link. Similarly, the final output node may mean one or more nodes that do not have an output node in a relationship with other nodes among nodes in the neural network. In addition, the hidden node may refer to nodes constituting a neural network other than the first input node and the last output node. In the neural network according to an embodiment of the present disclosure, the number of nodes in the input layer may be the same as the number of nodes in the output layer, and the number of nodes decreases and then increases again as the input layer proceeds to the hidden layer. I can. In addition, the neural network according to another embodiment of the present disclosure may be a neural network in which the number of nodes of an input layer is less than the number of nodes of an output layer, and the number of nodes decreases as the number of nodes progresses from the input layer to the hidden layer. have. In addition, the neural network according to another embodiment of the present disclosure may be a neural network in which the number of nodes of an input layer may be greater than the number of nodes of an output layer, and the number of nodes increases as the number of nodes progresses from the input layer to the hidden layer. I can. The neural network according to another embodiment of the present disclosure may be a neural network in the form of a combination of the aforementioned neural networks.

A deep neural network (DNN) may mean a neural network including a plurality of hidden layers in addition to an input layer and an output layer. Deep neural networks can be used to identify potential structures in data. For example, to identify potential structures such as photos, texts, videos, voices, music, etc. (e.g., what objects are in photos, what are the content and emotions of the text, what are the content and emotions of the voice, etc.) I can. Deep neural networks include convolutional neural networks (CNNs), recurrent neural networks (RNNs), auto encoders, Generative Adversarial Networks (GANs), and restricted Boltzmann machines (RBMs). boltzmann machine), deep belief network (DBN), Q network, U network, and Siam network. The description of the above-described deep neural network is only an example, and the present disclosure is not limited thereto.

The neural network may be learned by at least one of supervised learning, unsupervised learning, and semi supervised learning. Learning of neural networks is to minimize output errors. In learning of a neural network, iteratively inputs training data to the neural network, calculates the output of the neural network for the training data and the error of the target, and transfers the error of the neural network from the output layer of the neural network to the input layer in a direction to reduce the error. This is a process of updating the weight of each node of the neural network by backpropagation in the direction. In the case of supervised learning, learning data in which the correct answer is labeled in each learning data is used (ie, labeled learning data), and in the case of unsupervised learning, the correct answer may not be labeled in each learning data. That is, for example, in the case of supervised learning related to data classification, the training data may be data in which a category is labeled with each training data. Labeled training data is input to a neural network, and an error may be calculated by comparing an output (category) of the neural network with a label of the training data. As another example, in the case of unsupervised learning related to data classification, an error may be calculated by comparing input training data with a neural network output. The calculated error is backpropagated in the neural network in the reverse direction (ie, from the output layer to the input layer), and the connection weight of each node of each layer of the neural network may be updated according to the backpropagation. A change amount may be determined according to a learning rate in the connection weight of each node to be updated. The computation of the neural network for the input data and the backpropagation of the error can constitute a learning cycle (epoch). The learning rate may be applied differently according to the number of repetitions of the learning cycle of the neural network. For example, a high learning rate is used in the early stages of training of a neural network, so that the neural network quickly secures a certain level of performance to increase efficiency, and a low learning rate can be used in the later stages of training to increase accuracy.

In the learning of a neural network, in general, the training data may be a subset of actual data (that is, data to be processed using the learned neural network), and thus, errors in the training data decrease, but errors in the actual data occur. There may be increasing learning cycles. Overfitting is a phenomenon in which errors in actual data increase due to over-learning on learning data. For example, a neural network learning a cat by showing a yellow cat may not recognize that it is a cat when it sees a cat other than yellow, which may be a kind of overfitting. Overfitting can cause an increase in errors in machine learning algorithms. Various optimization methods can be used to prevent such overfitting. In order to prevent overfitting, methods such as increasing training data, regularization, and dropout in which some nodes of the network are omitted during the training process may be applied.

The electronic device and its operation method according to the present exemplary embodiment can check the plasma operation using parameters obtained from the plasma processing facility itself, so that the plasma process can be managed more efficiently and easily.

This embodiment can be represented by functional block configurations and various processing steps. These functional blocks may be implemented with various numbers of hardware or/and software configurations that perform specific functions. For example, the embodiment is an integrated circuit configuration such as memory, processing, logic, a look-up table, etc., capable of executing various functions by controlling one or more microprocessors or other control devices. Can be hired. Similar to how components can be implemented as software programming or software elements, this embodiment includes various algorithms implemented with a combination of data structures, processes, routines or other programming components, including C, C++, Java ( Java), an assembler, or the like may be implemented in a programming or scripting language. Functional aspects can be implemented with an algorithm running on one or more processors. In addition, the present embodiment may employ a conventional technique for electronic environment setting, signal processing, and/or data processing. Terms such as “mechanism”, “element”, “means”, and “composition” can be used widely, and are not limited to mechanical and physical configurations. The term may include the meaning of a series of routines of software in connection with a processor or the like.

The above-described embodiments are merely examples, and other embodiments may be implemented within the scope of the claims to be described later.

Claims (15)

In the method of operating an electronic device for checking the operation of plasma,
Obtaining a value of at least one parameter for the plasma in a specific time unit,
Checking motion information corresponding to each of the plurality of time intervals of the specific time unit based on applying the value of the at least one parameter acquired in the specific time unit to a first machine learning algorithm; and
Checking motion information corresponding to each of the one or more motion sections that are separated from each other and including at least one of the plurality of time sections based on motion information corresponding to each of the plurality of time sections; and
And checking final operation information of the plasma based on operation information corresponding to each of the one or more operation sections.
The method of claim 1,
The checking of the final motion information of the plasma includes confirming the final motion information of the plasma based on applying motion information corresponding to each of the one or more motion sections to a second machine learning algorithm.
The method of claim 1,
The first machine learning algorithm is learned to check plasma operation information corresponding to the specific time unit based on at least one parameter of the specific time unit.
The method of claim 1,
The first machine learning algorithm comprises a tree-based machine learning algorithm.
The method of claim 2,
The second machine learning algorithm is learned to check the final operation information of the plasma for the plurality of time intervals based on the operation information of the plasma for each of the one or more operation intervals.
The method of claim 2,
The second machine learning algorithm comprises a rule-based machine learning algorithm.
The method of claim 1,
The at least one operation section includes a first section including at least one time section prior to the first time point among the plurality of time sections, and a first section including at least one time section after the second time section among the plurality of time sections. The operation method comprising at least one of two sections and a third section including all of the plurality of time sections.
The method of claim 1,
The method of operation, wherein the at least one parameter is identified from a facility related to the operation of the plasma.
The method of claim 1,
The at least one parameter is a pressure of a plasma chamber, a valve angle, a current applied to a plasma heating device, a temperature of the plasma heating device, an amount of process gas injected into the chamber, and the chamber. The operating method comprising at least one of a pressure of a gas injected to bring the gas to an atmospheric pressure state, a temperature of a heater chuck, and a temperature of the chamber.
The method of claim 1,
When the one or more operation periods include a first operation period and the first operation period includes at least three or more time periods among the plurality of time periods, the operation information corresponding to the first operation period is the at least three time periods. The operation method corresponding to information indicated by a plurality of operation information corresponding to each of the at least two time intervals.
The method of claim 1,
Checking the operation setting information of the plasma; and
Further comprising the step of checking whether the final operation information corresponds to the operation setting information.
The method of claim 11,
And requesting to start a semiconductor process using the plasma when the final operation information corresponds to the operation setting information.
The method of claim 11,
Providing an alarm when the final operation information does not correspond to the operation setting information.
A non-transitory recording medium that can be read by a computer in which a program for executing the method of claim 1 is recorded on a computer.
In the electronic device for checking the operation of plasma,
A memory for storing instructions,
Including a processor,
The processor is connected to the memory,
Obtaining a value of at least one parameter for the plasma in a specific time unit,
Checking motion information corresponding to each of a plurality of time intervals of the specific time unit based on applying the value of at least one parameter acquired in the specific time unit to a first machine learning algorithm,
Based on the motion information corresponding to each of the plurality of time sections, check motion information corresponding to each of the one or more motion sections that are separated from each other and include at least one of the plurality of time sections,
The electronic device, which checks the final operation information of the plasma based on operation information corresponding to each of the one or more operation sections.

Images