CN114807635B - Method for preparing high-purity indium by multi-channel array directional solidification based on machine learning - Google Patents
Method for preparing high-purity indium by multi-channel array directional solidification based on machine learning Download PDFInfo
- Publication number
- CN114807635B CN114807635B CN202210489018.7A CN202210489018A CN114807635B CN 114807635 B CN114807635 B CN 114807635B CN 202210489018 A CN202210489018 A CN 202210489018A CN 114807635 B CN114807635 B CN 114807635B
- Authority
- CN
- China
- Prior art keywords
- directional solidification
- indium
- purity
- machine learning
- purity indium
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000007711 solidification Methods 0.000 title claims abstract description 97
- 230000008023 solidification Effects 0.000 title claims abstract description 97
- 229910052738 indium Inorganic materials 0.000 title claims abstract description 84
- APFVFJFRJDLVQX-UHFFFAOYSA-N indium atom Chemical compound [In] APFVFJFRJDLVQX-UHFFFAOYSA-N 0.000 title claims abstract description 84
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000010801 machine learning Methods 0.000 title claims abstract description 40
- 238000000746 purification Methods 0.000 claims abstract description 19
- 230000008569 process Effects 0.000 claims abstract description 18
- 238000002790 cross-validation Methods 0.000 claims abstract description 7
- 239000002994 raw material Substances 0.000 claims abstract description 5
- 239000010453 quartz Substances 0.000 claims description 22
- VYPSYNLAJGMNEJ-UHFFFAOYSA-N silicon dioxide Inorganic materials O=[Si]=O VYPSYNLAJGMNEJ-UHFFFAOYSA-N 0.000 claims description 22
- 230000008018 melting Effects 0.000 claims description 18
- 238000002844 melting Methods 0.000 claims description 18
- 239000007789 gas Substances 0.000 claims description 15
- 239000012535 impurity Substances 0.000 claims description 13
- 230000001681 protective effect Effects 0.000 claims description 7
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 claims description 6
- 238000007418 data mining Methods 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 5
- 238000005266 casting Methods 0.000 claims description 4
- 239000008367 deionised water Substances 0.000 claims description 4
- 229910021641 deionized water Inorganic materials 0.000 claims description 4
- 238000001035 drying Methods 0.000 claims description 4
- 238000005868 electrolysis reaction Methods 0.000 claims description 4
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Chemical compound O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 4
- 239000001257 hydrogen Substances 0.000 claims description 3
- 229910052739 hydrogen Inorganic materials 0.000 claims description 3
- 238000011068 loading method Methods 0.000 claims description 3
- 229910052757 nitrogen Inorganic materials 0.000 claims description 3
- 238000007789 sealing Methods 0.000 claims description 3
- 238000004140 cleaning Methods 0.000 claims description 2
- 230000026058 directional locomotion Effects 0.000 claims description 2
- 239000011261 inert gas Substances 0.000 claims description 2
- 238000005406 washing Methods 0.000 claims description 2
- 125000004435 hydrogen atom Chemical class [H]* 0.000 claims 1
- 239000004065 semiconductor Substances 0.000 abstract description 8
- 239000007769 metal material Substances 0.000 abstract description 3
- 238000005457 optimization Methods 0.000 abstract description 3
- 238000004857 zone melting Methods 0.000 description 7
- 238000002474 experimental method Methods 0.000 description 6
- 239000000463 material Substances 0.000 description 6
- 238000010438 heat treatment Methods 0.000 description 5
- 238000004519 manufacturing process Methods 0.000 description 4
- 239000000155 melt Substances 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 238000004506 ultrasonic cleaning Methods 0.000 description 4
- 150000001875 compounds Chemical class 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000002360 preparation method Methods 0.000 description 3
- 238000005292 vacuum distillation Methods 0.000 description 3
- UFHFLCQGNIYNRP-UHFFFAOYSA-N Hydrogen Chemical compound [H][H] UFHFLCQGNIYNRP-UHFFFAOYSA-N 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000011010 flushing procedure Methods 0.000 description 2
- 238000013178 mathematical model Methods 0.000 description 2
- GPXJNWSHGFTCBW-UHFFFAOYSA-N Indium phosphide Chemical compound [In]#P GPXJNWSHGFTCBW-UHFFFAOYSA-N 0.000 description 1
- 238000003723 Smelting Methods 0.000 description 1
- 229910045601 alloy Inorganic materials 0.000 description 1
- 239000000956 alloy Substances 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000009792 diffusion process Methods 0.000 description 1
- 239000002019 doping agent Substances 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- WPYVAWXEWQSOGY-UHFFFAOYSA-N indium antimonide Chemical compound [Sb]#[In] WPYVAWXEWQSOGY-UHFFFAOYSA-N 0.000 description 1
- RPQDHPTXJYYUPQ-UHFFFAOYSA-N indium arsenide Chemical compound [In]#[As] RPQDHPTXJYYUPQ-UHFFFAOYSA-N 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 239000007791 liquid phase Substances 0.000 description 1
- 230000033001 locomotion Effects 0.000 description 1
- 229910052751 metal Inorganic materials 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 238000011403 purification operation Methods 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000007670 refining Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000003756 stirring Methods 0.000 description 1
- 238000005092 sublimation method Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 238000009849 vacuum degassing Methods 0.000 description 1
Classifications
-
- C—CHEMISTRY; METALLURGY
- C22—METALLURGY; FERROUS OR NON-FERROUS ALLOYS; TREATMENT OF ALLOYS OR NON-FERROUS METALS
- C22B—PRODUCTION AND REFINING OF METALS; PRETREATMENT OF RAW MATERIALS
- C22B58/00—Obtaining gallium or indium
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B22—CASTING; POWDER METALLURGY
- B22D—CASTING OF METALS; CASTING OF OTHER SUBSTANCES BY THE SAME PROCESSES OR DEVICES
- B22D27/00—Treating the metal in the mould while it is molten or ductile ; Pressure or vacuum casting
- B22D27/04—Influencing the temperature of the metal, e.g. by heating or cooling the mould
- B22D27/045—Directionally solidified castings
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B22—CASTING; POWDER METALLURGY
- B22D—CASTING OF METALS; CASTING OF OTHER SUBSTANCES BY THE SAME PROCESSES OR DEVICES
- B22D46/00—Controlling, supervising, not restricted to casting covered by a single main group, e.g. for safety reasons
-
- C—CHEMISTRY; METALLURGY
- C22—METALLURGY; FERROUS OR NON-FERROUS ALLOYS; TREATMENT OF ALLOYS OR NON-FERROUS METALS
- C22B—PRODUCTION AND REFINING OF METALS; PRETREATMENT OF RAW MATERIALS
- C22B9/00—General processes of refining or remelting of metals; Apparatus for electroslag or arc remelting of metals
- C22B9/006—General processes of refining or remelting of metals; Apparatus for electroslag or arc remelting of metals with use of an inert protective material including the use of an inert gas
-
- C—CHEMISTRY; METALLURGY
- C22—METALLURGY; FERROUS OR NON-FERROUS ALLOYS; TREATMENT OF ALLOYS OR NON-FERROUS METALS
- C22B—PRODUCTION AND REFINING OF METALS; PRETREATMENT OF RAW MATERIALS
- C22B9/00—General processes of refining or remelting of metals; Apparatus for electroslag or arc remelting of metals
- C22B9/16—Remelting metals
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Landscapes
- Engineering & Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Mechanical Engineering (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Manufacturing & Machinery (AREA)
- Materials Engineering (AREA)
- Metallurgy (AREA)
- Organic Chemistry (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Strategic Management (AREA)
- Artificial Intelligence (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Tourism & Hospitality (AREA)
- Development Economics (AREA)
- Quality & Reliability (AREA)
- Medical Informatics (AREA)
- Operations Research (AREA)
- Computing Systems (AREA)
- Game Theory and Decision Science (AREA)
- Mathematical Physics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Manufacture And Refinement Of Metals (AREA)
- Crystals, And After-Treatments Of Crystals (AREA)
Abstract
A method for preparing high-purity indium by multi-channel array directional solidification based on machine learning is characterized in that 5N indium is prepared by an electrolytic method, and the raw materials are placed into a vacuum chamber for array multi-channel directional solidification purification, so that high-purity indium products with the purity of 6N and above which are uniformly arranged are obtained. According to the invention, a plurality of machine learning prediction models are established by combining a machine learning method, the accuracy of the model is evaluated through ten-fold cross validation, different machine learning models are compared and evaluated, the optimal machine learning model is screened out, the optimal experimental parameter range of directional solidification of the high-purity indium is predicted, the process parameter optimization of directional solidification of the high-purity indium is realized more quickly, and the method has the advantages of high selectivity, high purification efficiency and strong controllability, and can provide high-quality high-purity metal materials for the semiconductor industry.
Description
Technical Field
The invention belongs to the technical field of purification and directional solidification of high-purity materials, and particularly relates to a method for preparing high-purity indium by multi-channel array directional solidification based on machine learning assistance.
Background
High purity indium plays a key role in the indium industry, and is mainly used for manufacturing semiconductor compounds, high purity alloys, doping agents of semiconductor materials and the like due to excellent light permeability and strong electrical conductivity. In the field of photoelectrons, indium and its compound semiconductors have wide application, and are mainly synthesized into indium-based III-V compound semiconductors such as indium antimonide (InSb), indium phosphide (InP), indium arsenide (InAs) and the like, and as laser light sources for optical fiber communication, heterojunction solar cell materials, infrared detection and magneto-optical devices, the purity requirements on indium metal are also higher and higher, generally 6N and above, and the impurity content is required to be lower than 1ppm, even lower than 0.1ppm. The preparation methods of high-purity indium are numerous, and mainly include sublimation method, directional solidification method, vacuum distillation method, extraction method, electrolytic method and the like. The electrolytic refining can realize 4N-5N purification, and the vacuum distillation method and the directional solidification can realize 4N-6N purification; the directional solidification method has the advantages of convenient operation, higher efficiency, low cost and no environmental protection pressure, and is particularly suitable for preparing high-purity indium.
Because the diffusion rate of impurities is slow in the directional solidification purification, the moving speed of each melting zone needs to be strictly controlled in the completion of the directional solidification purification process, and the purification operation needs to be repeated for a plurality of times to reach the designated purity. The directional solidification method for preparing the high-purity material is described in some patent documents, and Chinese patent CN 107858523A realizes the purification of the high-purity indium by utilizing horizontal zone melting, and the purity of the prepared high-purity indium reaches 6N and above; chinese patent CN 102392294A utilizes horizontal zone melting technology, low vacuum conditions, material carrying boat, etc. to integrate various purification methods of vacuum distillation, vacuum degassing and zone melting smelting together, thereby realizing the preparation of high purity semiconductor material by a set of equipment and process. The key to improving directional solidification efficiency is the control of the melt zone, including the width of the melt zone, the stirring of the melt, the speed of movement of the melt zone, and its stability. The results obtained by different researchers also differ due to the complexity of the parameters and the interplay between the parameters. Scholars Hao Xin et al propose that various mathematical models can find parameters such as optimum number of operations, fuse length, etc. The change rule of solidification efficiency with the number of times and the length of the fuse area is researched by the method of the mathematical model and experiments by the Spim. The method is used for researching the rule of influence of technological parameters of qualitative solidification and purification of high-purity indium on the purification effect, establishing the internal relation between the technological parameters and the product performance (purity), and has important significance for realizing stable mass production of the high-purity indium of 6N5 and above. By combining a large number of reliable experiments with data mining by means of a machine learning method, mining an internal relation or experience rules, guiding the experiments by using numerical simulation optimization results, generating a certain amount of data base by means of high-throughput multi-channel array directional solidification equipment, establishing a model by means of machine learning, predicting an optimal test parameter range, obtaining optimal test parameters, and using the optimal test parameters for reverse prediction guiding production, the directional solidification method with high selectivity, high purification efficiency and strong controllability for providing high-quality high-purity metal materials for the semiconductor industry can be provided.
Disclosure of Invention
The invention aims to provide a method for preparing high-purity indium by multi-channel array directional solidification based on machine learning, which utilizes multi-channel array directional solidification equipment to carry out high-throughput experiments to generate a large amount of practical and reliable experimental data, utilizes data mining to assist in analyzing a directional solidification process, establishes a multi-factor coupled machine learning model, predicts and optimizes experimental process parameters within a certain range, and is used for optimizing a high-purity indium directional solidification process.
The technical scheme adopted by the invention is as follows:
a method for preparing high-purity indium by multi-channel array directional solidification based on machine learning is to prepare 5N indium by an electrolytic method as a raw material, and put the raw material into a vacuum chamber for array multi-channel directional solidification purification to obtain high-purity indium products with the purity of 6N and above which are uniformly arranged.
Further, the method for preparing high-purity indium by multi-channel array directional solidification based on machine learning comprises the following steps:
s1, directional solidification is carried out, wherein the process is as follows:
s1.1, preparing 5N indium by electrolysis, uniformly loading the 5N indium into 9 quartz boats, placing the 9 quartz boats into a quartz tube of a three-by-three array type multichannel directional solidification furnace, sealing the quartz tube, starting a power supply system, vacuumizing, and introducing protective gas, wherein the protective gas is hydrogen or nitrogen or inert gas, and the gas flow is 0.5-1L/min;
s1.2, setting the width of a three-by-three array type multi-channel heater of a three-by-three array type multi-channel directional solidification furnace to be 10-60 mm, directionally moving at the speed of 10-50 mm/h, directionally solidifying for 1-6 times, and controlling the temperature of a melting area to be 120-250 ℃; obtaining 6N high-purity indium;
s1.3, stopping the furnace, taking out high-purity indium samples in nine groups of quartz boats, removing 10-30% of tails, washing with deionized water, ultrasonically cleaning for 5-10 min, drying, remelting and casting into ingots, and respectively measuring the purity of indium under directional solidification according to different process parameters;
s2, constructing a high-purity indium directional solidification data set: collecting and recording the moving speed, the heater width, the solidification times, the melting zone temperature of the directional solidification furnace obtained in the step S1 and the purity of the product indium under corresponding parameters, and constructing a high-purity indium directional solidification data set for subsequent data mining;
s3, constructing a machine learning prediction model of the high-purity indium directional solidification process: taking the moving speed of the directional solidification furnace, the width of the heater, the solidification times, the temperature of a melting zone and the purity of corresponding indium in the step S1 as characteristic variables, taking the total impurity content of a product as a target variable, training and modeling by using different machine learning algorithms, comparing and evaluating different machine learning models through ten-fold cross validation, and selecting the model with the smallest error as a final prediction model;
s4, setting moving speed, heater width, solidification times and melting zone temperature parameters, inputting the set parameters into a final prediction model in the step S3, predicting all possible parameter combinations by using the final prediction model, and selecting the parameter with the lowest predicted total impurity content as the optimal parameter.
Compared with the prior art, the invention has the following beneficial technical effects:
1. according to the invention, indium purification under different technological parameter conditions is realized by using multichannel array directional solidification equipment, and multistage heating is realized by controlling the temperature of the heater, so that the heating temperature gradient distribution is more uniform and the temperature field is more stable. Meanwhile, parameters such as the moving speed, the width of the heater, the solidification times, the temperature of the melting area, the gas flow and the like can be sampled at fixed points and monitored in real time, and the data can be visualized through a computer, collected and normalized to construct a database.
2. According to the method, a plurality of machine learning prediction models are established by combining a machine learning method, the accuracy of the model is evaluated through ten-fold cross validation, different machine learning models are compared and evaluated, the optimal machine learning model is screened out, the optimal experimental parameter range of directional solidification of the high-purity indium is predicted, the process parameter optimization of the directional solidification of the high-purity indium is realized more quickly, and the method for preparing the high-purity indium through the directional solidification with high selectivity, high purification efficiency and strong controllability is provided.
3. The method has the advantages of high selectivity, high purification efficiency and strong controllability, and can provide high-quality high-purity metal materials for the semiconductor industry.
Drawings
FIG. 1 is a schematic diagram of a method for preparing high purity indium by multi-channel array directional solidification;
FIG. 2 is a flow chart of a high throughput directional solidification experiment and machine learning prediction.
Detailed Description
In order to more clearly illustrate the present invention, the present invention will be further described with reference to examples and drawings. It is to be understood by persons skilled in the art that the following detailed description is illustrative and not restrictive, and that this invention is not limited to the details given herein.
Referring to fig. 1, the device adopted by the method of the invention comprises a three-by-three array type multi-channel directional solidification furnace 6, an electric control device 10 of the three-by-three array type multi-channel directional solidification furnace, a computer 3 and a power module 1 connected with the computer through an interface 2, wherein the computer can perform remote control and data transmission. The collected data results may be saved as structured CSV format or exported directly to Excel. The three-by-three array type multichannel directional solidification furnace and the electrical control device thereof are all devices in the prior art. The distance between the heaters in the three-by-three array type multichannel directional solidification furnace is 180-240 mm, and the radial heat radiation width of the heating ring is 5-8 mm. The solidifying furnace is internally provided with a plurality of clapboards and a plurality of array-type distributed quartz tubes, and valves on the clapboards and heaters thereof are mutually communicated or independently closed. Sliding doors are arranged on two sides of the outer shell of the solidification furnace, and high-temperature-resistant glass windows for observing the inside of the furnace are arranged on the doors.
Example 1
The method for preparing the high-purity indium by multi-channel array directional solidification based on machine learning comprises the following steps:
s1, purifying indium under the condition of different directional solidification speeds, wherein the process is as follows:
s1.1, preparing 5N indium by an electrolysis method in the prior art, uniformly loading the 5N indium into 9 quartz boats, placing the 9 quartz boats into a quartz tube 12 of a three-by-three array type multi-channel directional solidification furnace 6, closing the quartz tube, and starting an electric control device. Opening a control cabinet button 8 of the electric control device, setting technological parameters such as zone melting speed, zone melting times, gas flow and the like on a panel 9, opening a vacuum system of the three-by-three array type multi-channel directional solidification furnace and a deflation valve 11 of a protective gas system 5, vacuumizing, and introducing protective gas nitrogen, wherein the gas flow is 1L/min; heating the quartz tube by a heating coil 7;
s1.2, the widths of three-by-three array type multi-channel heaters of the three-by-three array type multi-channel directional solidification furnace are respectively set to be 10mm, 40mm and 60mm, the widths of each heater are respectively arranged to move in an array type directional manner at the speeds of 10mm/h, 20mm/h, 30mm/h, 40mm/h and 50mm/h, directional solidification is respectively carried out for 1-6 times, and the temperature of a melting zone can be set to be 120-250 ℃. The directional solidification of the high-purity indium under different conditions can be realized by changing different process parameters. The purification process strengthens the axial distribution of impurities and strengthens the liquid phase mass transfer rate to finally achieve the purification effect, thus obtaining 6N high-purity indium;
s1.3, stopping the furnace, taking out high-purity indium samples in nine groups of quartz boats, removing 10-30% of tails, flushing with deionized water, placing into an ultrasonic cleaning machine for ultrasonic cleaning for 5-10 min, drying, re-casting, and respectively detecting the purity of indium under directional solidification according to different process parameters to obtain high-purity indium with purity of 6N and above;
s2, constructing a high-purity indium directional solidification data set: collecting and recording the moving speed, the heater width, the solidification times, the melting zone temperature of the directional solidification furnace obtained in the step S1 and the purity of the product indium under corresponding parameters, and constructing a high-purity indium directional solidification data set for subsequent data mining;
s3, constructing a machine learning prediction model of the high-purity indium directional solidification process: taking the moving speed of the directional solidification furnace, the width of the heater, the solidification times, the temperature of a melting zone and the purity of corresponding indium in the step S1 as characteristic variables, taking the total impurity content of a product as a target variable, training and modeling by using different machine learning algorithms, comparing and evaluating different machine learning models through ten-fold cross validation, and selecting the model with the smallest error as a final prediction model;
s4, setting moving speed, heater width, solidification times and melting zone temperature parameters, inputting the set parameters into a final prediction model in the step S3, predicting all possible parameter combinations by using the final prediction model, and selecting the parameter with the lowest predicted total impurity content as the optimal parameter.
Example 2
The method for preparing high-purity indium by multi-channel array directional solidification based on machine learning is used for purifying indium under the condition of different directional solidification times, and comprises the following steps:
(1) Putting 5N indium prepared by electrolysis into 9 quartz boats, putting the 9 quartz boats into a quartz tube in a three-by-three array type multichannel directional solidification furnace, sealing the quartz tube, starting an electric control device, vacuumizing, and introducing protective gas hydrogen with the gas flow of 0.5L/min;
(2) Setting the widths of three-by-three array type multi-channel heaters of the three-by-three array type multi-channel directional solidification furnace to be 20mm and 30mm respectively, and carrying out directional movement at the speeds of 20mm/h and 30mm/h respectively, wherein the directional solidification is carried out for 1-6 times respectively, and the temperature of a melting area is 180-200 ℃;
(3) And (3) after the operation is finished, stopping the furnace, taking out a high-purity indium sample in the quartz boat, removing 15% of tail, flushing with deionized water, placing into an ultrasonic cleaning machine for ultrasonic cleaning for 8min, drying, and detecting after remelting casting to obtain high-purity indium with purity of 6N and above.
Example 3
The preparation of the high-purity indium is optimized by machine learning continuously based on the high-purity indium prepared in the embodiment 2, and the machine learning prediction of a high-purity indium database is performed by the following method:
s2, constructing a high-purity indium directional solidification data set: collecting and recording the moving speed, the heater width, the solidification times, the melting zone temperature of the directional solidification furnace obtained in the step S1 and the purity of the product indium under corresponding parameters, and constructing a high-purity indium directional solidification data set shown in the table 1 for subsequent data mining;
s3, constructing a machine learning model of the high-purity indium directional solidification process: based on the process parameters of the above embodiment, including heater moving speed, heater width, solidification times, melting zone temperature, gas flow rate, etc., and total impurity content of the product (high purity indium purity=1—total impurity content), high purity indium machine learning modeling and application were performed according to the flow shown in fig. 2 (fig. 2 is modified, please check whether it is correct). Taking the moving speed of the directional solidification furnace, the width of a heater, the solidification times, the temperature of a melting zone and the purity of corresponding indium as characteristic variables, taking the total impurity content of a product as a target variable, normalizing the data, and storing the normalized data into a high-purity indium directional solidification database shown in table 1.
TABLE 1 high purity indium directional solidification data set
80% of the data in the database are randomly extracted as training data, and the remaining 20% are used as test data. Four commonly used machine learning Regression algorithms are adopted, including support vector Regression (support vector Regression, SVR), gradient ascending Regression (gradient boosting Regression, GBR), ridge Regression (RT), regression Tree (RT) are adopted to model training sets respectively, root mean square error (root mean square error, RMSE) of ten-fold cross validation is calculated respectively, and as shown in table 2, a machine learning algorithm model constructed by the GBR algorithm with the minimum root mean square error is selected as a final prediction model;
table 2 ten fold cross validation errors for different machine learning algorithms
| Machine learning algorithm | Root Mean Square Error (RMSE) | Determining coefficient (R) 2 ) |
| SVR | 0.20 | 0.95 |
| GBR | 0.08 | 0.99 |
| Ridge | 0.26 | 0.93 |
| RT | 0.15 | 0.97 |
S4, machine learning prediction is carried out based on the data set in the table 1, the set parameters are input into the final prediction model in the step S3, values are uniformly interpolated in the adjustable numerical intervals of technological parameters such as zone melting speed, zone melting temperature, solidification times and gas flow, high-flux virtual experiment parameter combinations are constructed through cross combination, each high-flux virtual experiment parameter combination is predicted by using the final prediction model, and parameters with the lowest predicted impurity total content are selected as optimal parameters for actual production and experiment.
The percentages stated in the present invention are percentages by mass unless otherwise indicated.
Claims (1)
1. The method for preparing high-purity indium by multi-channel array directional solidification based on machine learning is characterized in that 5N indium is prepared by an electrolytic method as a raw material, and the raw material is placed into a vacuum chamber for array multi-channel directional solidification purification to obtain high-purity indium products with the concentration of 6N and above which are uniformly arranged; the method comprises the following steps:
s1, directional solidification is carried out, wherein the process is as follows:
s1.1, preparing 5N indium by electrolysis, uniformly loading the 5N indium into 9 quartz boats, placing the 9 quartz boats into a quartz tube of a three-by-three array type multichannel directional solidification furnace, sealing the quartz tube, starting a power supply system, vacuumizing, and introducing protective gas, wherein the protective gas is hydrogen, nitrogen or inert gas, and the gas flow is 0.5-1L/min;
s1.2, setting the width of a three-by-three array type multi-channel heater of a three-by-three array type multi-channel directional solidification furnace to be 10-60 mm, carrying out directional movement at the speed of 10-50 mm/h, carrying out directional solidification for 1-6 times, and setting the temperature of a melting area to be 120-250 ℃; obtaining 6N high-purity indium;
s1.3, stopping the furnace, taking out high-purity indium samples in nine groups of quartz boats, removing 10-30% of tails, washing with deionized water, ultrasonically cleaning for 5-10 min, drying, remelting and casting into ingots, and respectively measuring the purity of indium under directional solidification according to different process parameters;
s2, constructing a high-purity indium directional solidification data set: collecting and recording the moving speed, the heater width, the solidification times, the melting zone temperature of the directional solidification furnace obtained in the step S1 and the purity of the product indium under corresponding parameters, and constructing a high-purity indium directional solidification data set for subsequent data mining;
s3, constructing a machine learning prediction model of the high-purity indium directional solidification process: taking the moving speed of the directional solidification furnace, the width of the heater, the solidification times, the temperature of a melting zone and the purity of corresponding indium in the step S1 as characteristic variables, taking the total impurity content of a product as a target variable, training and modeling by using different machine learning algorithms, comparing and evaluating different machine learning models through ten-fold cross validation, and selecting the model with the smallest error as a final prediction model;
s4, setting moving speed, heater width, solidification times and melting zone temperature parameters, inputting the set parameters into a final prediction model in the step S3, predicting all possible parameter combinations by using the final prediction model, and selecting the parameter with the lowest predicted total impurity content as the optimal parameter.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210489018.7A CN114807635B (en) | 2022-05-06 | 2022-05-06 | Method for preparing high-purity indium by multi-channel array directional solidification based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210489018.7A CN114807635B (en) | 2022-05-06 | 2022-05-06 | Method for preparing high-purity indium by multi-channel array directional solidification based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114807635A CN114807635A (en) | 2022-07-29 |
| CN114807635B true CN114807635B (en) | 2024-03-05 |
Family
ID=82510626
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210489018.7A Active CN114807635B (en) | 2022-05-06 | 2022-05-06 | Method for preparing high-purity indium by multi-channel array directional solidification based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114807635B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106407544A (en) * | 2016-09-12 | 2017-02-15 | 南京航空航天大学 | Method for establishing stiffness prediction model of IC10 directional solidification material |
| CN111286755A (en) * | 2020-04-10 | 2020-06-16 | 云南锡业集团（控股）有限责任公司研发中心 | Method for preparing high-purity indium by electrolysis-regional method |
| CN113343524A (en) * | 2021-06-01 | 2021-09-03 | 西安建筑科技大学 | Fe-Al-Ta ternary alloy directional solidification process optimization method based on simulation |
| CN113802151A (en) * | 2021-10-27 | 2021-12-17 | 上海大学 | Multi-channel electrolysis device for high-purity indium electrorefining and machine learning process optimization method |
-
2022
- 2022-05-06 CN CN202210489018.7A patent/CN114807635B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106407544A (en) * | 2016-09-12 | 2017-02-15 | 南京航空航天大学 | Method for establishing stiffness prediction model of IC10 directional solidification material |
| CN111286755A (en) * | 2020-04-10 | 2020-06-16 | 云南锡业集团（控股）有限责任公司研发中心 | Method for preparing high-purity indium by electrolysis-regional method |
| CN113343524A (en) * | 2021-06-01 | 2021-09-03 | 西安建筑科技大学 | Fe-Al-Ta ternary alloy directional solidification process optimization method based on simulation |
| CN113802151A (en) * | 2021-10-27 | 2021-12-17 | 上海大学 | Multi-channel electrolysis device for high-purity indium electrorefining and machine learning process optimization method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114807635A (en) | 2022-07-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113802151B (en) | Multi-channel electrolysis device for high-purity indium electrorefining and machine learning process optimization method | |
| US4265859A (en) | Apparatus for producing semiconductor grade silicon and replenishing the melt of a crystal growth system | |
| CN102934239B (en) | The height of polycrystal silicon ingot used for solar batteries exports manufacturing equipment | |
| CN114807635B (en) | Method for preparing high-purity indium by multi-channel array directional solidification based on machine learning | |
| CN102520616A (en) | Partial decoupling unminimized model prediction function control method in oil refining industrial process | |
| CN104407649B (en) | A kind of temperature control pressue device for gas hydrates crystal growth optical observation | |
| CN112859771A (en) | Quality prediction method, preparation method and system of gallium oxide based on deep learning and model-guiding method | |
| CN114934198B (en) | Method for preparing high-purity indium based on machine learning optimization vacuum distillation | |
| CN117959902A (en) | Gas purification function early warning system and method based on data feedback | |
| CN100575843C (en) | Polycrystalline silicon reducing furnace water-cooling double glass viewing mirror | |
| CN112834700A (en) | Quality prediction method, preparation method and system of high-resistance gallium oxide based on deep learning and guided mode method | |
| US5154795A (en) | System for setting analysis condition for a thermal analysis of a fluid inside an apparatus | |
| CN107829112A (en) | A kind of aluminium electrolysis cell anode carbon block current balance type control system and method | |
| CN103154296A (en) | Method for producing metal material and metal material | |
| CN117126988A (en) | Intelligent temperature-control heat treatment equipment for high-chromium castings | |
| CN107640749B (en) | High-purity tellurium dioxide and preparation method thereof | |
| CN210560721U (en) | Vacuum deposition system sample warm table | |
| CN218059139U (en) | Low-oxygen ultrahigh-purity arsenic rod forming device | |
| CN210778645U (en) | Cooling atmosphere circulating air duct | |
| CN210736550U (en) | Special shaping annealing device of glass beverage bottle production | |
| CN221275952U (en) | Water cooling screen for accelerating crystallization speed of crystal bar and single crystal furnace | |
| CN114007980B (en) | Method for producing polycrystalline silicon | |
| CN118504787B (en) | Hydrogen production amount prediction method of microwave-enhanced solar thermochemical hydrogen production system | |
| CN218893463U (en) | Continuous graphite purification device | |
| CN104445878A (en) | Device for forming glass product |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |