Ion implanter control method and device, computer equipment and storage medium

Abstract

The invention relates to a control method and device of an ion implanter, computer equipment and a computer readable storage medium. The method comprises the following steps: acquiring an expected beam parameter output by the ion implanter, and detecting initial state data of a current operation state of the ion implanter; the expected beam parameters and the initial state data are input into a prediction neural network to calculate the air inflow of various gases of an initial air inflow proportion adjusting ion source; when the output beam parameters do not meet the expected beam parameters, state change data at the current moment are calculated, the state change data and the expected beam parameters are input into a correction neural network to calculate air inflow correction values of various gases of the ion source at the next moment, and meanwhile the air inflow of various gases of the ion source is calculated; the air inflow of the ion source is adjusted at the next moment; according to the technical scheme, the automatic control function is realized, the plasma beam output efficiency of the ion implanter is improved, and the control accuracy is improved.

Description

Technical Field

[0001] The present application relates to the technical field of ion implanters, and in particular to a control method, device, computer equipment and computer-readable storage medium for an ion implanter. Background Art

[0002] The application field of hydrogen-helium coaxial ion implanter is very wide. It can simultaneously produce two ion beams, hydrogen molecule ions H2+ and helium ions He+. It is used in the research and development and testing of nuclear energy engineering materials, the development of spacecraft surface materials, ion beam material modification, nuclear medicine and radioactive medicine research, and catalyst development in the energy field, etc. It covers key technologies in the field of new energy, medical care, environmental protection and other fields.

[0003] The ion source used in the hydrogen-helium coaxial ion implanter has more obvious characteristics than ordinary ion sources. For example, the ionization process of the ion source is complicated, and the ionization effect is affected by multiple factors such as gas concentration, temperature, and electric field in the cavity. It is difficult to accurately establish a corresponding model to describe the ionization process; the ionization of the ion source is easily affected by the external environment, and the ionization effect is greatly affected by the vacuum degree and impurity gas in the ion source cavity. The vacuum effect of the vacuum system is affected by factors such as the temperature and humidity of the external environment. Therefore, in different time periods, the same hydrogen-helium intake ratio will produce different proportions of beams. flow; the ionization effect of the ion source is different during long-term operation. The ion source will consume the impurity gas in the cavity while generating ions, and the impurity gas attached to the inner wall of the cavity will also be released during the operation of the ion source, affecting the ionization effect of the ion source; some ion implanters cannot detect the intensity of the beam and the ratio of each ion in the beam while outputting the beam to other equipment. Therefore, the ion implanter cannot continuously feedback the beam, which also brings certain difficulties to the beam adjustment. The change of the ionization effect of the ion source will change the ion beam intensity and the ratio of each ion in the beam generated by the implanter.

[0004] At present, the function of hydrogen-helium coaxial ion implantation machines in adjusting the beam is mainly based on the beam intensity and ion ratio fed back by the ion implantation machine control system, combined with experience to adjust the ion source intake. There are also preliminary plans for intelligent control, such as the use of fuzzy modular neural networks and particle swarm optimization algorithms. These control methods are inefficient and difficult to control. In addition, since the measurement and control equipment of some implantation machines will block the transmission of the plasma beam to subsequent devices during the measurement of the beam intensity and the ratio of each ion in the beam, the time for the implantation machine to transmit the beam to the subsequent equipment will be compressed, which reduces the efficiency of the equipment's output of the plasma beam. Summary of the invention

[0005] The purpose of the present application is to solve one of the above-mentioned technical defects and to provide a control method, device, computer equipment and computer-readable storage medium for an ion implanter to improve control efficiency and control accuracy.

[0006] A control method for an ion implanter, comprising:

[0007] Obtaining the desired beam parameters output by the ion implanter;

[0008] Control the ion implanter to output the ion beam and detect the initial state data of the current operating state of the ion implanter;

[0009] Inputting the desired beam parameters and initial state data into a prediction neural network to obtain an initial intake ratio of various gases in the ion source, and adjusting the intake amount of various gases in the ion source according to the initial intake ratio;

[0010] Detecting real-time status data of the ion implanter, and judging whether the output beam parameters of the ion implanter meet the requirements of the expected beam parameters according to the real-time status data;

[0011] If the output beam parameters of the ion implanter do not meet the requirements of the expected beam parameters, the state change data of the ion implanter at the current moment is calculated based on the real-time state data, the state change data and the expected beam parameters are input into the correction neural network to calculate the intake amount correction values ​​of various gases in the ion source at the next moment, the intake amounts of various gases in the ion source are calculated based on the intake amount correction values, and the intake amount of the ion source is adjusted at the next moment, so that the output beam parameters of the ion implanter meet the requirements of the expected beam parameters.

[0012] In one embodiment, the desired beam parameters include desired beam intensity and / or desired beam ion ratio;

[0013] The initial state data includes the gas pressure in the cavity of the ion implanter, the intensity of the current beam and / or the current beam ion ratio;

[0014] The state change data includes: a change in the gas pressure in the chamber of the ion implanter, a change in the beam intensity and / or a change in the beam ion ratio.

[0015] In one embodiment, the control method of the ion implanter further includes:

[0016] If the output beam parameters of the ion implanter meet the requirements of the expected beam parameters, the ion implanter is controlled to maintain the corresponding parameters to output the ion beam, and the output beam parameters of the ion beam generated by the ion implanter are regularly checked to see if they meet the requirements of the expected beam parameters.

[0017] In one embodiment, the state change data of the ion implanter at the current moment is calculated according to the real-time state data, and the state change data and the expected beam parameters are input into the correction neural network to calculate the intake amount correction values ​​of various gases of the ion source at the next moment, including:

[0018] According to the recorded real-time status data, the status change data of the ion implanter in multiple time periods are obtained as the status change data at the current moment, and the multiple status change data are input into the input layer of the correction neural network to calculate the intake amount correction value of various gases in the ion source at the next moment.

[0019] In one embodiment, the desired beam parameters and initial state data are input into a prediction neural network to obtain an initial intake ratio of various gases in the ion source, and the intake amount of various gases in the ion source is adjusted according to the initial intake ratio, including:

[0020] Determine the total gas intake volume of the ion source according to the desired beam intensity;

[0021] Inputting the desired beam current ion ratio and initial state data into a prediction neural network to obtain the initial intake ratio of various gases in the ion source and sending it to a gas regulation program;

[0022] The gas intake amount of various gases of the ion source is calculated according to the initial gas intake ratio and the total gas intake amount through the gas adjustment program, and the gas intake amount of various gases is controlled.

[0023] In one embodiment, the intake amount of various gases of the ion source is calculated according to the intake amount correction value, including:

[0024]

[0025] In the formula, is the gas intake volume at the current moment, is the gas intake at the last moment, Q is the total intake of all gases in the ion source, Δr e Indicates the proportional change of gas ions corresponding to the next moment of the corrected neural network output; k S is the correspondence between the desired gas feed ratio and the desired beam ion ratio, k W It is the correspondence between the change of gas intake ratio and the change of beam ion ratio.

[0026] In one embodiment, the prediction neural network includes: an input layer, a hidden layer and an output layer; wherein each parameter in the hidden layer is obtained by training the historical data of the operation of the ion implanter; the input layer accesses the initial state data and the expected beam parameters, and the output layer outputs the intake ratio of each gas in the ion source calculated by the hidden layer.

[0027] In one embodiment, the correction neural network includes: an input layer, a hidden layer and an output layer; wherein the parameters in the hidden layer are obtained by training the historical data of the ion implanter operation; the input layer is connected to the state change data and the expected beam parameters generated at the current moment after the ion implanter completes the initial intake ratio adjustment, and the output layer outputs the intake amount correction value of various gases of the ion source at the next moment calculated by the hidden layer.

[0028] In one embodiment, the control method of the ion implanter further includes:

[0029] Building a predictive neural network;

[0030] Obtain the historical records of the intake volume of various gases in the ion implanter at different time periods and the ion ratio of various ions in the ion beam;

[0031] The prediction neural network is trained with the state parameters and ion ratio of the ion implanter as input and the intake ratio of various gases as output.

[0032] In one embodiment, the control method of the ion implanter further includes:

[0033] Establishing a revised neural network;

[0034] Taking the ion ratio and the real-time status data of the ion implanter as input and the intake ratio of various gases as output;

[0035] The neural network is trained with the intake ratio of various gases at the current moment as the prediction result of the intake ratio of the next moment at the previous moment, and the parameters of the neural network are updated and corrected by back propagation based on the error between the prediction result and the actual data.

[0036] A control device for an ion implanter, comprising:

[0037] A parameter input module is used to obtain the desired beam parameters output by the ion implanter;

[0038] A state detection module is used to control the ion implanter to output the ion beam and detect the initial state data of the current operating state of the ion implanter;

[0039] An initial adjustment module, used for inputting the desired beam parameters and initial state data into a prediction neural network to obtain an initial intake ratio of various gases in the ion source, and adjusting the intake amount of various gases in the ion source according to the initial intake ratio;

[0040] A parameter judgment module is used to detect the real-time status data of the ion implanter and judge whether the output beam parameters of the ion implanter meet the requirements of the expected beam parameters according to the real-time status data;

[0041] The real-time adjustment module is used to calculate the state change data of the ion implanter at the current moment according to the real-time state data if the output beam parameters of the ion implanter do not meet the requirements of the expected beam parameters, input the state change data and the expected beam parameters into the correction neural network to calculate the intake amount correction value of various gases of the ion source at the next moment, calculate the intake amount of various gases of the ion source according to the intake amount correction value, and adjust the intake amount of the ion source at the next moment so that the output beam parameters of the ion implanter meet the requirements of the expected beam parameters.

[0042] A computer device, comprising:

[0043] one or more processors;

[0044] Memory;

[0045] One or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs are configured to execute the control method of the ion implanter.

[0046] A computer-readable storage medium stores a computer program, which realizes the control method of the ion implanter when executed by a processor.

[0047] The scheme of the above-mentioned embodiment obtains the desired beam parameters output by the ion implanter; controls the ion implanter to output the ion beam, and detects the initial state data of the current operating state of the ion implanter; inputs the desired beam parameters and the initial state data into the prediction neural network to obtain the initial intake ratio of various gases of the ion source, and adjusts the intake amount of various gases of the ion source; detects the real-time state data of the ion implanter, and if the output beam parameters of the ion implanter do not meet the requirements of the desired beam parameters, calculates the state change data of the ion implanter at the current moment according to the real-time state data, inputs the state change data into the correction neural network to calculate the intake amount correction value of various gases of the ion source at the next moment, calculates the intake amount of various gases of the ion source according to the intake amount correction value, and adjusts the intake amount of the ion source at the next moment, so that the output beam parameters of the ion implanter meet the requirements of the desired beam parameters; this technical scheme realizes the automatic control function, improves the efficiency of the ion implanter in outputting the plasma beam, improves the control accuracy, and can increase the time for the ion implanter to transmit the beam to the subsequent equipment.

[0048] Additional aspects and advantages of the present application will be partially given in the following description, which will become apparent from the following description, or will be understood through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0050] Figure 1 is a system framework diagram of a control method for an ion implanter of an example;

[0051] Figure 2 is a flow chart of a control method of an ion implanter according to an embodiment;

[0052] Figure 3 is an example of a beam regulation block diagram of an ion implanter;

[0053] Figure 4 is an example beam stability monitoring diagram;

[0054] Figure 5 is an example beam control flow chart;

[0055] Figure 6 This is a comparison chart of the training effects of different time steps of an example;

[0056] Figure 7 This is a comparison chart of the training process of LSTM and ANN;

[0057] Figure 8 This is a comparison chart of the prediction results of LSTM and ANN for recent data;

[0058] Fig. 9 This is a comparison chart of the prediction results of long-term data by LSTM and ANN;

[0059] Fig.10 is a schematic structural diagram of a control device for an ion implanter according to an embodiment;

[0060] Fig.11 is a structural block diagram of a computer device according to an embodiment. DETAILED DESCRIPTION

[0061] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present application.

[0062] Those skilled in the art will appreciate that, unless otherwise stated, the singular forms "a", "an", "said" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of the features, integers, steps, operations, but does not exclude the presence or addition of one or more other features, integers, steps, operations.

[0063] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those generally understood by those skilled in the art to which this application belongs. It should also be understood that terms such as those defined in general dictionaries should be understood to have meanings consistent with those in the context of the prior art, and will not be interpreted with idealized or overly formal meanings unless specifically defined as here.

[0064] The technical solution of the present application is applied in an ion implanter to control the ion implanter to generate an ion beam with an expected beam intensity and beam ion ratio, such as Figure 1 As shown, Figure 1 This is a system framework diagram of an example control method of an ion implanter, including a host computer and a slave computer. The host computer can be a computer device with a control software system installed, and the slave computer can be a PLC (Programmable Logic Controller). The ion implanter is controlled by the host computer and the slave computer. In order to quickly and stably obtain the desired ion beam flow, the technology of the present application provides an automatic beam adjustment method, which can adjust the hydrogen and helium gas intake in real time according to the current beam state, the desired beam intensity, and the desired ion ratio. The host computer can run the automatic beam adjustment program of the ion implanter. The automatic beam adjustment program is configured to automatically adjust the intake of various gases in the ion source of the ion implanter through a predictive neural network, a corrected neural network, and a gas adjustment program, so as to realize the automatic adjustment function of the ion beam output by the ion implanter.

[0065] The present application provides a control method for an ion implanter, which can be implemented on a host computer. Figure 2 As shown, Figure 2 The present invention is a flow chart of a control method of an ion implanter according to an embodiment, comprising:

[0066] Step S10, obtaining the desired beam parameters output by the ion implanter.

[0067] For example, after the ion implanter is normally started, the desired beam parameters may include desired beam intensity and desired beam ion ratio; if the operator wants the ion implanter to produce a specified beam intensity and beam ion ratio, the operator can input the intake amount of each gas in the ion source in the user interface of the host computer.

[0068] Step S20, controlling the ion implanter to output an ion beam, and detecting initial state data of the current operating state of the ion implanter.

[0069] Specifically, an ion beam will be generated after the ion implanter is turned on normally; during this process, the initial state data detected by each sensor can be received by the lower computer and sent to the upper computer; exemplarily, the initial state data may include the gas pressure in the cavity of the ion implanter, the intensity of the current beam, the current beam ion ratio, etc.

[0070] Step S30, inputting the desired beam parameters and initial state data into a prediction neural network to obtain initial intake ratios of various gases in the ion source, and adjusting intake amounts of various gases in the ion source according to the initial intake ratios.

[0071] Specifically, the upper computer can input the expected beam intensity, the expected beam ion ratio, the gas pressure in the cavity of the ion implanter, the current beam intensity, the current beam ion ratio, etc. into a pre-trained predictive neural network to obtain the initial intake ratio of various gases and send it to the lower computer. The lower computer then calculates the intake volume of various gases based on the initial intake ratio, and adjusts the intake volume of various gases in the ion source so that the state parameters of the ion beam output by the ion implanter are within the specified range.

[0072] In one embodiment, the scheme of adjusting the intake amount of various gases of the ion source in step S30 may include the following:

[0073] First, the total air intake of the ion source is determined according to the expected beam intensity; then the expected beam ion ratio and initial state data are input into the prediction neural network to obtain the initial air intake ratio of various gases in the ion source and send it to the gas regulation program; wherein the initial state data includes data reflecting the current state information of the ion implanter, such as the air pressure in the ion implanter cavity, the intensity of the current beam, the current beam ion ratio, etc.; finally, the air intake of various gases in the ion source is calculated according to the initial air intake ratio and the total air intake through the gas regulation program, and the air intake of various gases is controlled.

[0074] As in the technical solution of the above-mentioned embodiment, after the operator inputs the desired beam parameters of the ion implanter, such as the desired beam intensity and the desired ion ratio, the upper computer inputs the desired beam parameters into the input layer of the prediction neural network, and at the same time transmits the detected initial state data into the input layer of the prediction neural network. The prediction neural network automatically calculates the intake ratio of various gases, thereby calculating the intake volume of various gases. The intake volume of the ion source is adjusted by the lower computer, thereby improving the control efficiency and accuracy.

[0075] In one embodiment, the prediction neural network may include: an input layer, a hidden layer and an output layer; wherein the parameters in the hidden layer are obtained by training the historical data of the ion implanter operation; the input layer accesses the initial state data and the expected beam parameters, and the output layer outputs the intake ratio of each gas in the ion source calculated by the hidden layer.

[0076] In one embodiment, the process of training a prediction neural network may include the following:

[0077] Establish a prediction neural network; obtain the historical records of the intake volume of various gases in the ion implanter at different time periods and the ion ratio of various ions in the ion beam; train the prediction neural network with the state parameters and ion ratio of the ion implanter as input and the intake ratio of various gases as output.

[0078] Step S40, detecting the real-time status data of the ion implanter, and judging whether the output beam parameters of the ion implanter meet the requirements of the expected beam parameters according to the real-time status data.

[0079] Specifically, after the initial adjustment of the ion implanter, the lower computer can receive real-time status data of the ion implanter detected by various sensors and send it to the upper computer. The upper computer can judge whether the output beam parameters of the ion implanter meet the requirements of the expected beam parameters based on the real-time beam intensity and beam-ion ratio of the output ion beam. When comparing and judging, it can be compared whether the difference between the actual beam intensity and the expected beam intensity is within the set threshold range, and whether the difference between the actual beam ion ratio and the expected ion ratio is within the set threshold range.

[0080] Step S50, if the output beam parameters of the ion implanter do not meet the requirements of the expected beam parameters, calculate the state change data of the ion implanter at the current moment according to the real-time state data, input the state change data and the expected beam parameters into the correction neural network to calculate the intake amount correction value of various gases of the ion source at the next moment, calculate the intake amount of various gases of the ion source according to the intake amount correction value, and adjust the intake amount of the ion source at the next moment, so that the output beam parameters of the ion implanter meet the requirements of the expected beam parameters.

[0081] Exemplarily, the state change data of the ion implanter may include the change in the gas pressure in the cavity of the ion implanter, the change in the beam intensity, the change in the beam ion ratio, etc.; using a pre-trained correction neural network, the state change data and the desired beam parameters are input into the correction neural network, and the correction value of the intake volume of various gases in the ion source at the next moment is calculated, and then the gas intake volume is corrected and adjusted so that the output beam parameters of the ion implanter meet the required range of the desired beam parameters.

[0082] In one embodiment, when the correction value of the intake amount of various gases of the ion source at the next moment is calculated by the correction neural network, the following may be included:

[0083] According to the recorded real-time status data, the status change data of the ion implanter in multiple time periods are obtained as the status change data at the current moment, and the multiple status change data are input into the input layer of the correction neural network to calculate the intake amount correction value of various gases in the ion source at the next moment.

[0084] In one embodiment, the intake amount of various gases of the ion source is calculated according to the intake amount correction value in step S50, and the intake amount calculation formula includes the following:

[0085]

[0086] In the formula, is the gas intake volume at the current moment, is the gas intake at the last moment, Q is the total intake of all gases in the ion source, Δr e Indicates the proportional change of gas ions corresponding to the next moment of the corrected neural network output; k S is the correspondence between the desired gas feed ratio and the desired beam ion ratio, k W It is the corresponding relationship between the change of gas intake ratio and the change of beam ion ratio.

[0087] For example, when adjusting the ion implanter, the ratio of ions generated at the ion source is related to the ratio of the gas, taking hydrogen as an example:

[0088]

[0089] Δr q =kΔr e

[0090] Among them, r q is the proportion of hydrogen intake to the total intake, r e is the proportion of hydrogen ions in the total intake, Q is the total intake volume of hydrogen and helium, is the intake volume of hydrogen at the current moment, is the intake volume of hydrogen at the last moment; the prediction result of hydrogen by the prediction neural network can be expressed as:

[0091] Q 1 =Q 1S +Q 1W

[0092] Among them, Q 1 is the predicted hydrogen intake amount, Q 1S Q is the exact amount of hydrogen gas corresponding to the beam flow expected to be generated by the implanter. 1W The prediction error is that directly predicting the hydrogen intake will produce an error. The correction is made through the correction neural network. The above intake calculation formula can be used to calculate the hydrogen intake at the next moment. As the hydrogen intake ratio increases, the proportion of hydrogen ions will also increase. k is used to map the change in the proportion of hydrogen intake and the change in the proportion of hydrogen ions. The change in the ion ratio at the next moment is obtained by comparing the expected ion ratio at the next moment with the ion ratio at the current moment. e , the change in the proportion of hydrogen intake at the next moment is obtained through k mapping, and finally the hydrogen intake amount at the current moment is calculated based on the change in the current proportion of hydrogen intake Since the k value is predicted by the modified neural network, an error k will be generated. W , which will lead to an error in the prediction of the hydrogen intake amount, Δr e The smaller the value, the smaller the error in predicting the hydrogen intake amount. In the process of beam adjustment, the hydrogen and helium intake amounts are predicted and modified multiple times to continuously approach the desired beam ion ratio, thereby reducing Δr e value, making the prediction results more accurate.

[0093] In the scheme of the above embodiment, while using the neural network to continuously predict the air intake ratio of the ion source at the next moment at different moments, the ion implanter generates state change data at the corresponding moment, and the neural network records the state data at different moments, and predicts the air intake ratio at the next moment based on the state data at the previous multiple moments, thereby improving the accuracy of the prediction.

[0094] Exemplarily, in a hydrogen-helium coaxial ion implanter, it can be divided into an ion source component, a transmission component and an ion beam detection component according to its function; the ion source component may include various ion sources for generating plasma, the transmission component may include a mass analyzer, and the ion beam detection component may include a Faraday cage for detecting beam intensity. In the ion source component, a mixed ion beam is generated by feeding hydrogen and helium with different flow rates, and after separation and purification by the transmission component, the H2+ and He+ mixed ion beam with an impurity ion content of less than 2% is transmitted to the Faraday cage, the total current intensity is recorded, and the ratio of the H2+ and He+ ion beams is calculated by moving the beam clamping plate; by obtaining data such as the total beam current intensity of the ion implanter, the vacuum degree of the ion source extraction area, the hydrogen and helium intake amount, and the microwave power, the corresponding hydrogen and helium intake amount under the current state of the ion implanter is calculated and adjusted according to the user's desired beam intensity and ion ratio.

[0095] When in use, the operator can input the desired beam intensity and beam ion ratio in the user interface of the automatic beam adjustment program. The automatic beam adjustment program will determine whether the beam state generated by the ion implanter at the current moment meets the requirements, and record the current beam state and ion implanter state data. If the beam state at the current moment meets the requirements, the ion implanter outputs a beam that meets the requirements and checks the beam state regularly. If the beam state at the current moment does not meet the requirements, the neural network is used to calculate the ion source intake ratio at the next moment based on the state data at the current moment, and the gas intake volume is automatically adjusted according to the intake ratio of various gases, so that the ion source produces a beam state close to the user's expectations. For ease of description, the embodiment of the present application will take a hydrogen-helium coaxial ion implanter as an example to explain, and adjust the intake volume of the two gases, hydrogen and helium, input by the ion implanter, so as to control the output H2+ and He+ ion beam intensities.

[0096] In one embodiment, the correction neural network may include: an input layer, a hidden layer and an output layer; wherein the parameters in the hidden layer are obtained by training the historical data of the ion implanter operation; the input layer is connected to the state change data and the expected beam parameters generated at the current moment after the ion implanter completes the initial intake ratio adjustment, and the output layer outputs the intake amount correction value of various gases of the ion source at the next moment calculated by the hidden layer.

[0097] In one embodiment, the process of training the modified neural network may include the following:

[0098] Establish a correction neural network; take the ion ratio and real-time status data of the ion implanter as input and the intake ratio of various gases as output; train the neural network with the intake ratio of various gases at the current moment as the prediction result of the intake ratio at the next moment from the previous moment, and back-propagate and update the parameters of the correction neural network according to the error between the prediction result and the actual data.

[0099] In one embodiment, in step S50, if the output beam current parameters of the ion implanter meet the requirements of the desired beam current parameters, the ion implanter is controlled to maintain the corresponding parameters to output the ion beam current, and the output beam current parameters of the ion beam generated by the ion implanter are regularly detected to see whether they meet the requirements of the desired beam current parameters. In this step, if the actual beam current parameters meet the requirements of the desired beam current parameters, the current output ion beam current of the ion implanter is maintained, and the monitoring state is entered at the same time, and the ion beam current generated by the ion implanter is regularly detected to see whether it meets the requirements of the desired beam current parameters; if it does not meet the requirements, the corresponding processing flow is executed.

[0100] The technical solution of the above-mentioned embodiment realizes the automatic control function, improves the efficiency of the ion implanter in outputting the plasma beam, improves the control accuracy, and can increase the time for the ion implanter to transmit the beam to the subsequent equipment.

[0101] In order to make the technical solution of the present application more clear, more embodiments are described below using a hydrogen-helium coaxial ion implanter as an example.

[0102] In one embodiment, the prediction neural network and the correction neural network can use LSTM (Long Short-Term Memory), which is composed of multiple neurons, each of which is composed of a forget gate, an input gate, an output gate, and a cell state update. The data information selected by the input gate will be recorded, the data selected by the forget gate will be discarded, the output gate controls the output of the neuron, and the cell state update is used to transfer the data information recorded at the current moment to the neuron at the next moment.

[0103] In one embodiment, when detecting the status data of the ion implanter, the status data of the ion implanter can be obtained through sensors such as gas flow meters, barometers, voltmeters, ammeters, etc. installed on the ion implanter, and the analog signals of the sensors are converted into digital signals through the lower computer and transmitted to the neural network through the EPICS channel for application.

[0104] In one embodiment, the beam adjustment process of the ion source of the ion implanter can be divided into multiple time segments, and the parameters related to the beam adjustment in each time segment may include: hydrogen intake x1, helium intake x2, gas pressure at the ion source x3, ion beam intensity x4, actual current of the high voltage x5, ion ratio of the ion beam x6, and the expected beam ion ratio change in the next time period x7; the above data records the status information of the ion implanter in different time periods, and the status information in different time periods is transmitted to the neural network to predict the relationship between the change in the hydrogen intake ratio and the change in the beam ion ratio at the next moment.

[0105] like Figure 3 As shown, Figure 3 This is an example of an ion implanter beam regulation block diagram; the control system records the parameters related to beam regulation in the time periods T1, T2, ..., Tn, and Tn+1. These data record the state information of the ion implanter in different time periods. The information in different time periods is transmitted to the neural network to calculate the corresponding relationship k between the change in the gas intake ratio and the change in the beam ion ratio at the next moment. The calculation formula of k can be expressed as: k = f lstm (X); where k maps the change in gas intake ratio and gas ion ratio, and f lstm is a pre-trained neural network model, and X is a two-dimensional array containing the parameters related to beam regulation at different time segments.

[0106] In one embodiment, when training a neural network, actual data generated by an ion implanter can be used for training. Taking a hydrogen-helium coaxial ion implanter as an example, the ion implanter modifies the intake of hydrogen and helium in different time periods to obtain the ion ratio of the implanter beam. The intake ratio of hydrogen and helium at the current moment is the prediction result of the intake ratio at the previous moment for the next moment. During the training of the neural network, the model forwards once according to the input at the current moment to obtain the predicted intake of hydrogen and helium at the next moment, compares the predicted result with the actual data and calculates the error, and then updates the parameters of the network through back propagation.

[0107] In one embodiment, taking a hydrogen-helium coaxial ion implanter as an example, the processing of the gas adjustment program may be as follows:

[0108] The beam state generated by the ion implanter is related to the amount of hydrogen and helium gas input at the ion source. Hydrogen and helium can be expressed as:

[0109] Q 1 =r q Q (1)

[0110] Q 2 =(1-r q)Q (2)

[0111] Among them, Q 1 is the hydrogen intake volume, Q 2 is the helium intake volume, r q is the proportion of hydrogen in the total gas intake, Q is the total gas intake of the ion source, and r q The value of Q is calculated from the total intake volume Q 1 and Q 2 The total gas intake is determined by the expected beam intensity. The calculation is performed as follows:

[0112]

[0113] Where e represents the error between the expected beam intensity and the actual beam intensity, ΔQ is the change in the total gas intake. Since the total beam current of the ion source is affected by the gas ratio, the higher the rq, the more sensitive the total current is to the change of Q, and K P , K i is the defined gain coefficient. Since the beam intensity of the ion source is affected by the gas ratio, the higher the proportion of hydrogen, the more sensitive the beam intensity is to the change of the total gas intake. Therefore, K P , K i The calculation is based on the hydrogen ratio r q The size of C p , C i 、b p1 、b p2 、b i1 、b i2 , K d is a constant obtained based on experience; when r q When K increases P and K i It decreases accordingly, and the injection machine flow intensity can be controlled by adjusting the total air intake volume.

[0114] For example, in order to improve the operating stability of the ion source of the ion implanter, the automatic beam adjustment is turned off and on for 24 hours. Figure 4 As shown, Figure 4 This is an example of a beam stability monitoring diagram. The vertical axis in the figure is the monitored beam intensity at the end of the ion implanter. Figure a is a monitoring diagram when the automatic beam adjustment is turned off, and Figure b is a monitoring diagram when the automatic beam adjustment program is turned on. It can be seen that when the automatic beam adjustment is turned off, the beam intensity of the ion implanter fluctuates to a certain extent over time. After the automatic beam adjustment program is turned on, the beam of the ion implanter is stably maintained at 20.5μA for a long time.

[0115] The relationship between the change in ion ratio and the change in hydrogen intake ratio can be expressed as:

[0116] Δrq =k t Δr e (6)

[0117]

[0118] Q 1,t =Q 1,t-1 +k t Δr e Q (9)

[0119] Q 2,t =Q 2,t-1 +(1-k t Δr e )Q (10)

[0120] Among them, Δr q is the change in hydrogen intake ratio, Δr e is the change in the proportion of H2+ ions, and at a certain moment, k t Δr e Mapping to Δr q , E 1 is the H2+ ion current intensity, E is the total beam current intensity, and the H2+ ion ratio r is calculated. e . After obtaining Δr at time t q value, through the Q of the previous moment 1,t Calculate the current hydrogen intake volume Q 1,t Similarly, the helium intake volume Q at the current moment can be calculated 2,t Because k t It is not a constant value. Due to the influence of plasma state in different time periods, k t will change when using deep learning to predict k t The value of will produce errors, but according to formulas (9) and (10), when Δre becomes smaller, the errors in the prediction of Q1,t and Q2,t also decrease. Therefore, as the intake amount of hydrogen and helium is adjusted multiple times during the automatic beam adjustment process, the obtained ion ratio gradually approaches the expected value.

[0121] Due to Δr q With Δr e There are more complex mapping relationships between them. t The accuracy of the mapping relationship may not be enough, so deep learning alone is not enough to represent k t Predicting the k value may cause deviation and lead to unsatisfactory prediction results. Therefore, in practical applications, we can skip the k t The prediction of the value is directly achieved by using deep learning through Δr e For Δr q Make predictions.

[0122] like Figure 5 As shown, Figure 5 is an example of a beam control flow chart; when Δr e When the value is too large, deep learning is used to directly calculate the Δr at the next moment. q Make a prediction and then recalculate Δr e If Δr e If the value is within the allowable error range, then k t The value directly predicts Δr q Value, k t The value is set through debugging experience and is a fixed value between 0 and 1. Deep learning is mainly used to quickly reduce Δr e The range of values, when Δr e The value is small enough, through k t The intake ratio is adjusted multiple times to make the ion ratio produced by the ion source gradually approach the ratio expected by the user.

[0123] After completing the predicted calculation of the hydrogen and helium intake amounts, the gas regulation program sends the calculated results to the lower computer, which controls the gas flowmeter of hydrogen and helium to adjust the gas flow rate, thereby achieving control of the ion source gas intake.

[0124] For example, in response to the problems of cumbersome, difficult and low degree of automation in the manual adjustment of beam in current ion implanters, the present application proposes an automatic beam adjustment method for ion implanters based on LSTM networks; the LSTM network in deep learning is used to dynamically predict the change in the proportion of hydrogen gas intake in the ion source of the ion implanter, and the total gas intake is adjusted using PID control according to the predicted results, thereby realizing automatic regulation of the beam of the ion implanter.

[0125] For example, since the state of plasma is affected by a variety of uncontrollable and difficult to measure factors, the ion beam intensity and ion ratio generated by the ion source change; however, the change of the ion source state is slow, and the current state of the ion source can be mapped out through some parameters at the current moment, and the state of the ion source at the next moment can be predicted based on the change of the parameter value. Accordingly, in order to obtain the best intelligent beam adjustment scheme, the scheme of the present application can introduce ANN (Artificial Neural Network) and LSTM (Long Short-Term Memory) beam adjustment methods. Among them, ANN predicts the output value after adding an activation function based on the input parameters and different numbers of hidden layers. The neurons of the LSTM neural network are composed of an input gate, a forget gate, an output gate, a memory unit, and a hidden layer unit. In the process of adjusting the beam of the ion implanter, corresponding data will be generated at different time segments. These data are stored in the memory of the LSTM network after being screened by the input gate. The LSTM network will predict the next moment Δr based on these memories. q value.

[0126] When using the LSTM network to q When predicting the value, it is necessary to predict the next moment online according to the dynamic changes in the ion source state, so as to predict the output results in multiple subsequent time periods. The data structure of the network in LSTM is represented as follows:

[0127]

[0128] Among them, the matrix A p t p The data structure input into the LSTM network at the time is a two-dimensional matrix of n×m; n is the characteristic number representing the state of the ion source, and m is the step length of recording the data of the ion source state in different time periods. p ) is at t p The error between the actual ion ratio and the expected ion ratio of the ion source at the time, x(n,t p ) is expressed as t p Other characteristic parameters of the ion source at the time, such as the vacuum degree of the extraction area, the total gas intake, the total beam current intensity and other data. In the automatic beam adjustment process, according to the data matrix A in the previous m time periods p Prediction p+1 At time Δr q After adjusting the hydrogen and helium intake according to the predicted results, the ion implanter measures the new ion ratio and sets t p+1 The ion source status data at the moment is stored in matrix A p+1 Used for t p+2The step size m determines the LSTM network's prediction of the next moment's data based on the data of the previous m moments. The larger the step size, the smaller the error of the model training. Therefore, a larger step size should be selected.

[0129] For example, Figure 6 As shown, Figure 6 This is a comparison chart of training effects at different time steps. The vertical axis is the model loss Loss, and the horizontal axis is the number of iterations Epochs. When using data structures with step lengths of 1, 3, 5, and 7 for training, it can be clearly seen that with each increase in the step length, the input network data matrix will have to add a row, and the network's computational workload will also increase. When the step length increases from 5 to 7, the error decreases, but gradually converges to the same. Therefore, considering the compromise between computer memory and computing power consumption and computational accuracy, the final step length is 5.

[0130] The input parameters of the LSTM network include the intake volume of hydrogen and helium, the vacuum degree of the extraction area, the total beam intensity output by the ion implanter, the load of the extraction high-voltage power supply, the current hydrogen intake ratio, the current beam ion ratio, and the error between the current ion ratio and the expected ratio. The number of layers of the LSTM network is 2, the number of nodes is 128, the probability of dropout is 0.2, and the number of nodes output by the last layer of the network is 32. After that, the predicted value of the hydrogen intake ratio is output through a fully connected layer. The ion source randomly adjusts the intake volume of hydrogen and helium to generate the corresponding beam ion ratio value, and the data generated by the random adjustment is used for model training. The beam ion ratio at the next moment is taken as the expected value of the beam ion ratio at the previous moment. The LSTM network uses the data at the previous moment and the corresponding expected value to predict the hydrogen intake ratio at the next moment. The hydrogen intake ratio at the next moment in the training data is taken as the true value to calculate the error predicted by the LSTM network, and finally trained by the Adam optimization algorithm with a dynamically updated learning rate. In order to ensure the universality of the model, the prediction results of the LSTM network were verified using data from three weeks later and the prediction error was calculated.

[0131] ANN and LSTM networks are used for comparison. The hidden layer of ANN is 3 layers, the number of nodes is 34, 64, and 16 respectively, the probability of dropout is 0.2, and the ReLU activation function is used between each layer of the network. Figure 7 As shown, Figure 7 This is a comparison of the training process of LSTM and ANN. As can be seen from the figure, as the number of iterations increases, the error of the model prediction of ANN begins to gradually increase after a certain number of training, while the LSTM network can still maintain the error within a small range. LSTM network and ANN are used to predict Δrq respectively, and the data obtained from the test a few days later and the data obtained from the test three weeks later are predicted respectively. The prediction results are as follows Figure 8and Fig. 9 As shown. Among them, Figure 8 This is a comparison chart of the prediction results of recent data by LSTM and ANN. Fig. 9 This is a comparison of the prediction results of LSTM and ANN for long-term data; the red dot in the figure is Δr q The predicted values ​​are connected with a broken line. It can be seen that for recent data, the errors of LSTM network and ANN prediction are small because the ion source state does not change much. For long-term data, the accuracy of ANN prediction drops significantly due to the change of ion source state, while the prediction of LSTM network still maintains a high accuracy. Therefore, LSTM network is more suitable for automatic beam adjustment of ion implantation machine.

[0132] As in the above-mentioned embodiment, the technical solution of using ANN and LSTM networks to compare prediction accuracy, combined with the intelligent beam adjustment method, reduces the impact of changes in parameters such as ambient temperature and gas inlet bottle supply pressure on ion beam stability, and realizes unattended operation.

[0133] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0134] The following is a detailed description of the relevant embodiments of the control device for the ion implanter.

[0135] refer to Fig.10 As shown, Fig.10 is a schematic diagram of the structure of a control device of an ion implanter according to an embodiment, which may include:

[0136] The parameter input module 10 is used to obtain the desired beam parameters output by the ion implanter;

[0137] The state detection module 20 is used to control the ion implanter to output the ion beam and detect the initial state data of the current operation state of the ion implanter;

[0138] An initial adjustment module 30, for inputting the desired beam parameters and initial state data into a prediction neural network to obtain an initial intake ratio of various gases in the ion source, and adjusting the intake amount of various gases in the ion source according to the initial intake ratio;

[0139] The parameter judgment module 40 is used to detect the real-time status data of the ion implanter and judge whether the output beam parameters of the ion implanter meet the requirements of the expected beam parameters according to the real-time status data;

[0140] The real-time adjustment module 50 is used to calculate the state change data of the ion implanter at the current moment according to the real-time state data if the output beam parameters of the ion implanter do not meet the requirements of the expected beam parameters, input the state change data and the expected beam parameters into the correction neural network to calculate the intake amount correction value of various gases of the ion source at the next moment, calculate the intake amount of various gases of the ion source according to the intake amount correction value, and adjust the intake amount of the ion source at the next moment, so that the output beam parameters of the ion implanter meet the requirements of the expected beam parameters.

[0141] The control device of the ion implanter of this embodiment can execute a control method of the ion implanter provided in the embodiment of the present application. The implementation principle is similar. The actions performed by each module in the control device of the ion implanter in each embodiment of the present application correspond to the steps in the control method of the ion implanter in each embodiment of the present application. For the detailed functional description of each module in the control device of the ion implanter, please refer to the description in the corresponding control method of the ion implanter shown in the previous text, which will not be repeated here.

[0142] Embodiments of a computer device and a computer-readable storage medium are described below.

[0143] The present application provides a technical solution of a computer device for realizing functions related to a control method of an ion implanter. The computer device comprises:

[0144] One or more processors; a memory; one or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the one or more processors, and the one or more program configurations are used for the control method of the ion implanter of any embodiment.

[0145] like Fig.11 As shown, Fig.11 1 is a block diagram of a computer device of an embodiment, which may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc. The computer device may include one or more of the following components: a processing component 102, a memory 104, a power component 106, a multimedia component 108, an audio component 110, an input / output (I / O) interface 112, a sensor component 114, and a communication component 116.

[0146] The processing component 102 generally controls the overall operation of the computer device, such as operations associated with display, telephone calls, data communications, camera operations, and recording operations.

[0147] The memory 104 is configured to store various types of data to support operations in the computer device, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0148] The power supply assembly 106 provides power to the various components of the computer device.

[0149] The multimedia component 108 includes a screen that provides an output interface between the computer device and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). In some embodiments, the multimedia component 108 includes a front camera and / or a rear camera.

[0150] The audio component 110 is configured to output and / or input audio signals.

[0151] The input / output interface 112 provides an interface between the processing component 102 and the peripheral interface modules, which may be keyboards, click wheels, buttons, etc. These buttons may include but are not limited to: a home button, a volume button, a start button, and a lock button.

[0152] The sensor assembly 114 includes one or more sensors for providing various aspects of status assessment for the computer device. The sensor assembly 114 may include a proximity sensor configured to detect the presence of a nearby object without any physical contact.

[0153] The communication component 116 is configured to facilitate wired or wireless communication between the computer device and other devices. The computer device can access a wireless network based on a communication standard, such as Wi-Fi, a carrier network (such as 2G, 3G, 4G or 5G), or a combination thereof.

[0154] The computer-readable storage medium provided in the present application is used to implement functions related to the control method of an ion implanter. The computer-readable storage medium stores at least one instruction, at least one program, code set or instruction set. The at least one instruction, at least one program, code set or instruction set is loaded by a processor and executes the control method of an ion implanter of any embodiment. In an exemplary embodiment, the computer-readable storage medium can be a non-temporary computer-readable storage medium including instructions, such as a memory including instructions. For example, the non-temporary computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device.

[0155] The present application also provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A computer device, such as a processor of a computer device, reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device implements the control method of the ion implanter of any embodiment when executing.

[0156] The control scheme of the ion implanter in the above-mentioned embodiments can be executed by a host computer, which can be a computer device. The host computer outputs control parameters to a slave computer, which can be a PLC. The slave computer controls the intake amount of hydrogen and helium of the ion implanter. The operator runs the automatic beam adjustment program of the ion implanter on the computer device, waits for the ion implanter to start normally, and then inputs the desired beam parameters in the user operation interface of the automatic beam adjustment program. The automatic beam adjustment program automatically completes the adjustment of the ion implanter beam and feeds back the adjusted ion beam of the implanter on the user interface. The automatic beam adjustment program inputs the desired beam parameters into the input layer of the prediction neural network. The slave computer collects the status data of the ion implanter through various sensors. The collected status data is simply processed by the slave computer and then input into the In the automatic beam adjustment program of the upper computer, the automatic beam adjustment program analyzes and processes the collected status data and transmits it to the input layer of the prediction neural network. The prediction neural network calculates the intake ratio of each gas in the ion source and the intake volume. The lower computer controls the intake volume of the gas flow meter according to the calculated intake volume. When the ion beam flow output by the ion implanter changes, the automatic beam adjustment program can determine the need to correct the ion beam flow, and calculate the change in the intake ratio of various gases in the ion source through the various status information of the ion beam flow uploaded by the lower computer, and the corrected neural network. The automatic beam adjustment program calculates the intake volume according to the change in the intake ratio, and automatically adjusts the ion beam flow intensity and ion ratio of the ion implanter through the lower computer through the gas adjustment program, and completes the correction process by adjusting the intake volume of the ion source through the lower computer.

[0157] The technical solution of the embodiment of the present application greatly improves the efficiency of the control operation and has high control accuracy, and is suitable for promotion and use in ion implantation machines.

[0158] The above description is only a partial implementation method of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A control method for an ion implanter, characterized in that: include: Obtaining the desired beam parameters output by the ion implanter; Control the ion implanter to output the ion beam and detect the initial state data of the current operating state of the ion implanter; Inputting the desired beam parameters and initial state data into a prediction neural network to obtain an initial intake ratio of various gases in the ion source, and adjusting the intake amount of various gases in the ion source according to the initial intake ratio; Detecting real-time status data of the ion implanter, and judging whether the output beam parameters of the ion implanter meet the requirements of the expected beam parameters according to the real-time status data; If the output beam parameters of the ion implanter do not meet the requirements of the expected beam parameters, the state change data of the ion implanter at the current moment is calculated based on the real-time state data, the state change data and the expected beam parameters are input into the correction neural network to calculate the intake amount correction values ​​of various gases in the ion source at the next moment, the intake amounts of various gases in the ion source are calculated based on the intake amount correction values, and the intake amount of the ion source is adjusted at the next moment, so that the output beam parameters of the ion implanter meet the requirements of the expected beam parameters.

2. The control method of the ion implanter according to claim 1, characterized in that: The desired beam parameters include desired beam intensity and / or desired beam ion ratio; The initial state data includes the gas pressure in the cavity of the ion implanter, the intensity of the current beam and / or the current beam ion ratio; The state change data includes: a change in the gas pressure in the chamber of the ion implanter, a change in the beam intensity and / or a change in the beam ion ratio.

3. The control method of the ion implanter according to claim 2, characterized in that: Also includes: If the output beam parameters of the ion implanter meet the requirements of the expected beam parameters, the ion implanter is controlled to maintain the corresponding parameters to output the ion beam, and the output beam parameters of the ion beam generated by the ion implanter are regularly checked to see if they meet the requirements of the expected beam parameters.

4. The control method of the ion implanter according to claim 1, characterized in that: The state change data of the ion implanter at the current moment is calculated according to the real-time state data, and the state change data and the expected beam parameters are input into the correction neural network to calculate the gas intake correction values ​​of various gases in the ion source at the next moment, including: According to the recorded real-time status data, the status change data of the ion implanter in multiple time periods are obtained as the status change data at the current moment, and the multiple status change data are input into the input layer of the correction neural network to calculate the intake amount correction value of various gases in the ion source at the next moment.

5. The control method of the ion implanter according to claim 2, characterized in that: Inputting the desired beam parameters and initial state data into a prediction neural network to obtain an initial intake ratio of various gases in the ion source, and adjusting the intake amount of various gases in the ion source according to the initial intake ratio, including: Determine the total gas intake of the ion source according to the desired beam intensity; Inputting the desired beam current ion ratio and initial state data into a prediction neural network to obtain the initial intake ratio of various gases in the ion source and sending it to a gas regulation program; The gas intake amount of various gases of the ion source is calculated according to the initial gas intake ratio and the total gas intake amount through the gas adjustment program, and the gas intake amount of various gases is controlled.

6. The control method of the ion implanter according to claim 5, characterized in that: The intake amount of various gases of the ion source is calculated according to the intake amount correction value, including: Q 1t =Q 1t-1 +(k S +k W )Δr e Q In the formula, Q 1t is the gas intake volume at the current moment, Q 1t-1 is the gas intake at the last moment, Q is the total intake of all gases in the ion source, Δr e Indicates the proportional change of gas ions corresponding to the next moment of the corrected neural network output; k S is the correspondence between the desired gas feed ratio and the desired beam ion ratio, k W It is the corresponding relationship between the change of gas intake ratio and the change of beam ion ratio.

7. The control method of the ion implanter according to claim 2, characterized in that: The prediction neural network comprises: an input layer, a hidden layer and an output layer; wherein each parameter in the hidden layer is obtained by training the historical data of the ion implanter operation; the input layer accesses the initial state data and the expected beam parameters, and the output layer outputs the intake ratio of each gas in the ion source calculated by the hidden layer; The correction neural network includes: an input layer, a hidden layer and an output layer; wherein the parameters in the hidden layer are obtained by training the historical data of the ion implanter operation; the input layer is connected to the state change data and the expected beam parameters generated at the current moment after the ion implanter completes the initial intake ratio adjustment, and the output layer outputs the intake amount correction value of various gases of the ion source at the next moment calculated by the hidden layer.

8. The control method of the ion implanter according to claim 7, characterized in that: Also includes: Building a predictive neural network; Obtain the historical records of the intake volume of various gases in the ion implanter at different time periods and the ion ratio of various ions in the ion beam; The prediction neural network is trained with the state parameters and ion ratio of the ion implanter as input and the intake ratio of various gases as output.

9. The control method of the ion implanter according to claim 7, characterized in that: Also includes: Establishing a revised neural network; Taking the ion ratio and the real-time status data of the ion implanter as input and the intake ratio of various gases as output; The neural network is trained with the intake ratio of various gases at the current moment as the prediction result of the intake ratio of the next moment at the previous moment, and the parameters of the neural network are updated and corrected by back propagation based on the error between the prediction result and the actual data.

10. A computer device, characterized in that: The computer device comprises: one or more processors; Memory; One or more application programs, wherein the one or more application programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs are configured to execute the control method of the ion implanter according to any one of claims 1 to 9.

Images