CN115293067A - MOCVD equipment growth result abnormity tracing method and system - Google Patents

MOCVD equipment growth result abnormity tracing method and system Download PDF

Info

Publication number
CN115293067A
CN115293067A CN202210968280.XA CN202210968280A CN115293067A CN 115293067 A CN115293067 A CN 115293067A CN 202210968280 A CN202210968280 A CN 202210968280A CN 115293067 A CN115293067 A CN 115293067A
Authority
CN
China
Prior art keywords
simulation
process parameters
result
cfd
growth
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.)
Pending
Application number
CN202210968280.XA
Other languages
Chinese (zh)
Inventor
王钢
王杰
裴艳丽
罗铁成
何宜聪
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Sun Yat Sen University
Original Assignee
Sun Yat Sen University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Sun Yat Sen University filed Critical Sun Yat Sen University
Priority to CN202210968280.XA priority Critical patent/CN115293067A/en
Publication of CN115293067A publication Critical patent/CN115293067A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/28—Design optimisation, verification or simulation using fluid dynamics, e.g. using Navier-Stokes equations or computational fluid dynamics [CFD]
    • C—CHEMISTRY; METALLURGY
    • C30—CRYSTAL GROWTH
    • C30B—SINGLE-CRYSTAL GROWTH; UNIDIRECTIONAL SOLIDIFICATION OF EUTECTIC MATERIAL OR UNIDIRECTIONAL DEMIXING OF EUTECTOID MATERIAL; REFINING BY ZONE-MELTING OF MATERIAL; PRODUCTION OF A HOMOGENEOUS POLYCRYSTALLINE MATERIAL WITH DEFINED STRUCTURE; SINGLE CRYSTALS OR HOMOGENEOUS POLYCRYSTALLINE MATERIAL WITH DEFINED STRUCTURE; AFTER-TREATMENT OF SINGLE CRYSTALS OR A HOMOGENEOUS POLYCRYSTALLINE MATERIAL WITH DEFINED STRUCTURE; APPARATUS THEREFOR
    • C30B25/00—Single-crystal growth by chemical reaction of reactive gases, e.g. chemical vapour-deposition growth
    • C30B25/02—Epitaxial-layer growth
    • C30B25/16—Controlling or regulating
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Chemical & Material Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Mathematical Physics (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • General Chemical & Material Sciences (AREA)
  • Molecular Biology (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Chemical Kinetics & Catalysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Crystallography & Structural Chemistry (AREA)
  • Materials Engineering (AREA)
  • Metallurgy (AREA)
  • Organic Chemistry (AREA)
  • Algebra (AREA)
  • Fluid Mechanics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a method and a system for tracing the abnormal growth result of MOCVD equipment, wherein the method comprises the following steps: carrying out CFD modeling on the MOCVD reaction cavity to generate a CFD model; determining a process parameter value range and performing a simulation experiment by using a CFD (computational fluid dynamics) model to obtain simulation process parameters and a simulation result; modeling according to the simulation process parameters and the simulation result to generate a neural network model; and matching the abnormal growth result with the process parameters based on the neural network model to obtain abnormal process parameters. The system comprises: the device comprises a CFD building module, an analog simulation module, a neural network building module and a matching module. By using the method and the device, the abnormal process parameters can be quickly and accurately found out, and the efficiency is improved. The method and the system for tracing the growth result abnormity of the MOCVD equipment can be widely applied to the field of MOCVD abnormity detection.

Description

MOCVD equipment growth result abnormity tracing method and system
Technical Field
The invention relates to the field of MOCVD (metal organic chemical vapor deposition) anomaly detection, in particular to a method and a system for tracing the anomaly of a growth result of MOCVD equipment.
Background
Metal Organic Chemical Vapor Deposition (MOCVD) is a process of transporting metal organic substances and reaction gases to the surface of a high-temperature substrate to perform thermal decomposition and chemical reaction, and finally forming a thin film. In the epitaxial growth process of the thin film, multidisciplinary knowledge inside the MOCVD cavity is crossed, the process parameters are more and mutually influenced, and the flowing state of internal gas is complex.
CFD (computational fluid dynamics) is a computer-aided research method for computational analysis by finite element analysis methods combined with multidisciplinary knowledge. The flow and temperature distribution of gas in the reaction cavity can be obtained through numerical simulation calculation, and the epitaxial growth rate and uniformity of the film can be obtained by combining with chemical reaction. At present, the accurate prediction of the growth rate and uniformity of the MOCVD film can be realized by accurate reaction cavity modeling and calculation models and matching with a perfect chemical reaction path, and the CFD is widely applied to the research of the design and growth process parameters of the MOCVD reaction cavity.
In the actual MOCVD epitaxial growth process, especially in industrial mass production, the growth process parameters are fixed after being determined, and once the growth result is abnormal, such as the growth rate and uniformity of a film are changed, the product batch is scrapped, and the loss is caused. Generally, the judgment is carried out through the experience of an engineer, if the problem cannot be found out, the process parameters need to be readjusted, and therefore time and labor are consumed, the cost is increased, and the efficiency is reduced.
Disclosure of Invention
In order to solve the above technical problems, an object of the present invention is to provide a method and a system for tracing the abnormal growth result of the MOCVD equipment, which can quickly and accurately find out the abnormal process parameters and improve the efficiency.
The first technical scheme adopted by the invention is as follows: a tracing method for MOCVD equipment growth result abnormity comprises the following steps:
carrying out CFD modeling on the MOCVD reaction cavity to generate a CFD model;
determining a process parameter value range and performing a simulation experiment by using a CFD (computational fluid dynamics) model to obtain simulation process parameters and a simulation result;
modeling according to the simulation process parameters and the simulation result to generate a neural network model;
and matching the abnormal growth result with the process parameters based on the neural network model to obtain abnormal process parameters.
Further, the step of determining a process parameter value range and performing a simulation experiment by using a CFD model based on an experimental design method to obtain simulation process parameters and a simulation result specifically comprises the following steps:
determining the value range of the process parameter;
setting a plurality of groups of simulation experiments by using an experiment design method in a process parameter value range to obtain simulation process parameters;
and carrying out simulation experiment by using the CFD model according to the simulation process parameters to obtain a simulation result.
Further, the step of matching the abnormal growth result with the process parameters based on the neural network model to obtain the abnormal process parameters specifically includes:
setting a plurality of groups of process parameter values based on a control variable method;
inputting the technological parameter values into the neural network model in sequence to obtain a prediction result;
comparing the prediction result with the abnormal growth result based on a mean square error formula to obtain a minimum mean square error value;
and calculating the minimum mean square error values of different process parameters based on a probability formula to obtain abnormal process parameters.
Further, the probability formula is specifically as follows:
Figure BDA0003795520390000021
in the above formula, P (x) i ) For abnormal growthThe probability of the result is that,
Figure BDA0003795520390000022
is the minimum mean square error value, x i Are process parameters.
Further, the method also comprises the step of verifying the CFD model, which comprises the following specific steps:
setting a plurality of groups of process parameter experimental values and carrying out growth experiments by using the MOCVD reaction chamber in sequence to obtain experimental results;
simulating the process parameter experiment value and sequentially utilizing the CFD model to carry out simulation and numerical simulation calculation to obtain a simulation result;
and calculating the experimental result and the simulation result based on the Pearson correlation coefficient, and verifying the CFD model.
Further, the experimental result is the growth rate of the semiconductor film, and the calculation steps are as follows:
obtaining a semiconductor film obtained after the experiment is finished and experiment time;
measuring the thickness of the semiconductor film;
and calculating the growth rate of the semiconductor film according to the thickness and the experimental time.
The second technical scheme adopted by the invention is as follows: an MOCVD equipment growth result abnormity traceability system comprises:
the CFD building module is used for carrying out CFD modeling on the MOCVD reaction cavity to generate a CFD model;
the simulation module is used for determining a process parameter value range and performing a simulation experiment by using the CFD model to obtain simulation process parameters and a simulation result;
the neural network construction module is used for modeling according to the simulation process parameters and the simulation result to generate a neural network model;
and the matching module is used for matching the abnormal growth result with the process parameters based on the neural network model to obtain the abnormal process parameters.
The method and the system have the beneficial effects that: the method comprises the steps of firstly, carrying out CFD modeling on the MOCVD reaction cavity to obtain a CFD model, so that the simulation experiment can be more attached; secondly, a CFD model is used for carrying out simulation experiment to obtain simulation process parameters and simulation results, so that the cost can be saved; then, a neural network model is constructed according to the simulation process parameters and the simulation results, and the mapping relation between the growth results and the process parameters is obtained; and finally, matching the abnormal growth result with the process parameters based on the neural network model, so that the abnormal process parameters can be quickly and accurately obtained, and the method can trace the abnormal growth result caused by the abnormal multi-parameter.
Drawings
FIG. 1 is a flow chart of steps of a method for tracing the abnormal growth result of MOCVD equipment according to the present invention;
FIG. 2 is a block diagram of the growth result anomaly tracing system of the MOCVD equipment;
FIG. 3 is a schematic illustration of a semiconductor thin film in accordance with an embodiment of the present invention;
FIG. 4 is a neural network model matching flow diagram in accordance with an embodiment of the present invention;
FIG. 5 is a schematic diagram of probability of abnormal process parameters according to an embodiment of the present invention.
Detailed Description
The invention is described in further detail below with reference to the figures and the specific embodiments. For the step numbers in the following embodiments, they are set for convenience of illustration only, the order between the steps is not limited at all, and the execution order of each step in the embodiments can be adapted according to the understanding of those skilled in the art.
Referring to fig. 1, the invention provides a method for tracing the abnormal growth result of MOCVD equipment, which comprises the following steps:
s1, carrying out CFD modeling on an MOCVD reaction cavity to generate a CFD model;
s2, verifying the CFD model, and knowing the accuracy of the CFD model through verifying the CFD model;
s2.1, setting a plurality of groups of process parameter experiment values and carrying out growth experiments by sequentially utilizing an MOCVD reaction chamber to obtain experiment results;
specifically, 13 sets of experimental values of process parameters were set as follows:
Figure BDA0003795520390000031
Figure BDA0003795520390000041
inputting the 13 sets of process parameter experimental values into the MOCVD reaction chamber in sequence for growth experiment to obtain 13 sets of experimental results; the experimental results were calculated as follows:
firstly, obtaining a semiconductor film obtained after an experiment is completed and experiment time, then dividing the semiconductor film into an inner circle, a middle circle and an outer circle as shown in fig. 3, selecting three points in each circle and measuring each point to obtain the thickness of the nine points, and finally calculating the growth rate of the semiconductor film according to the thickness and the experiment time, wherein the growth rate formula is the thickness divided by the experiment time.
S2.2, simulating the process parameter experiment value, and sequentially performing simulation and numerical simulation calculation by using a CFD model to obtain a simulation result;
specifically, the CFD model is used for carrying out 1:1, analog simulation.
And S2.3, calculating an experimental result and a simulation result based on the Pearson correlation coefficient, and verifying the CFD model.
Specifically, the pearson correlation coefficient formula is as follows:
Figure BDA0003795520390000042
in the above formula, x i As a result of the simulation, y i Results of the experiment, r XY For the correlation coefficient of the simulation results with the experimental results, here n =9.
r XY The closer to 1 the value of (A) is, the greater the correlation between the simulation result and the experimental result, and generally, the correlation coefficient reaches 0.6, which is a significant correlationIt is shown that the matching degree between the simulation result and the experiment result is high.
The pearson correlation coefficients of the experimental results and the simulation results are shown in the following table:
Figure BDA0003795520390000043
Figure BDA0003795520390000051
as can be seen from the above table, when the simulation result of the CFD model is compared with the experiment result of MOCVD, the correlation coefficient is over 0.6, which indicates that the correlation degree between the experiment result and the simulation result is high, and the calculation result of the CFD model is credible.
S3, determining a process parameter value range and performing a simulation experiment by using a CFD model to obtain simulation process parameters and a simulation result;
s3.1, determining the value range of the process parameters;
specifically, the process parameters have the following value ranges:
Figure BDA0003795520390000052
s3.2, setting a plurality of groups of simulation experiments by using an experiment design method in the process parameter value range to obtain simulation process parameters;
specifically, in order to obtain the growth results of all process parameters, as many sets of simulation experiments as possible are provided, and the present embodiment preferably provides 1000-2000 sets of simulation experiments.
And S3.3, carrying out a simulation experiment by utilizing the CFD model according to the simulation process parameters to obtain a simulation result.
And S4, modeling is carried out according to the simulation process parameters and the simulation results to generate a neural network model, and the neural network model can obtain the mapping relation between any process parameters and the growth results.
And S5, referring to FIGS. 4 and 5, matching the abnormal growth result with the process parameters based on the neural network model to obtain abnormal process parameters.
S5.1, setting a plurality of groups of process parameter values based on a control variable method;
specifically, assume a process parameter of x 1 ,x 2 ,x 3 ,...,x 10 And x 11 (ii) a Corresponding abnormal growth result is y 1 ,y 2 ,y 3 ,...,y 8 And y 9 。
By changing only x 1 Let x be 1 Global change is carried out in the value range of the process parameters, and other process parameters are fixed and unchanged;
by changing only x 2 Let x be 2 Global change is carried out in the value range of the process parameters, and other process parameters are fixed and unchanged;
by changing only x 3 Let x be 3 Global change is carried out in the value range of the process parameters, and other process parameters are fixed and unchanged;
repeating the above operation to change x in sequence 4 ,x 5 ,x 6 ,...,x 10 And x 11 。
S5.2, sequentially inputting the process parameter values into the neural network model to obtain a prediction result;
specifically, when only x is changed 1 In time, changing x can be predicted by neural network model 1 As a result of the growth of (2), obtaining a constantly changing x 1 Predicting the growth result;
when only x is changed 2 In time, changing x can be predicted by neural network model 2 As a result of the growth of (2), obtaining a constantly changing x 2 The predicted growth result of (2);
when only x is changed 3 In time, changing x can be predicted by neural network model 3 As a result of the growth of (2), obtaining a constantly changing x 3 Predicting the growth result;
repeating the above operation to obtain continuously changed x 4 ,x 5 ,x 6 ,...,x 10 And x 11 The predicted growth outcome of (1).
S5.3, comparing the prediction result with the abnormal growth result based on a mean square error formula to obtain a minimum mean square error value;
specifically, the mean square error formula is as follows:
Figure BDA0003795520390000061
in the above formula, Y i For neural network model prediction value, y i For abnormal growth results, here n =9.
For constantly changing x 1 The predicted growth result and the abnormal growth result are subjected to mean square error calculation to obtain the minimum mean square error value which is recorded as
Figure BDA0003795520390000071
Repeating the above operation for continuously changing x 2 ,x 3 ,x 4 ,...,x 10 And x 11 The mean square error calculation is carried out on the predicted growth result and the abnormal growth result,
Figure BDA0003795520390000072
and
Figure BDA0003795520390000073
and S5.4, calculating the minimum mean square error values of different process parameters based on a probability formula to obtain abnormal process parameters.
Specifically, the probability formula is as follows:
Figure BDA0003795520390000074
in the above formula, P (x) i ) Is the probability of an abnormal growth outcome,
Figure BDA0003795520390000075
is the minimum mean square error value，x i Are process parameters.
When the probability is higher, the possibility that the process parameter is abnormal is higher.
As can be seen from fig. 5, the probability of the abnormal growth result of the pressure in this embodiment is at most 40.2%, and therefore the abnormal process parameter of the abnormal growth result in this embodiment is the pressure.
As shown in fig. 2, a system for tracing the growth result abnormality of MOCVD equipment includes:
the CFD building module is used for carrying out CFD modeling on the MOCVD reaction cavity to generate a CFD model;
the simulation module is used for determining a process parameter value range and performing a simulation experiment by using the CFD model to obtain simulation process parameters and a simulation result;
the neural network construction module is used for modeling according to the simulation process parameters and the simulation result to generate a neural network model;
and the matching module is used for matching the abnormal growth result with the process parameters based on the neural network model to obtain abnormal process parameters.
The contents in the above method embodiments are all applicable to the present system embodiment, the functions specifically implemented by the present system embodiment are the same as those in the above method embodiment, and the beneficial effects achieved by the present system embodiment are also the same as those achieved by the above method embodiment.
While the preferred embodiments of the present invention have been illustrated and described, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims (7)

1. A tracing method for growth result abnormity of MOCVD equipment is characterized by comprising the following steps:
carrying out CFD modeling on the MOCVD reaction cavity to generate a CFD model;
determining a process parameter value range and performing a simulation experiment by using a CFD (computational fluid dynamics) model to obtain simulation process parameters and a simulation result;
modeling according to the simulation process parameters and the simulation result to generate a neural network model;
and matching the abnormal growth result with the process parameters based on the neural network model to obtain abnormal process parameters.
2. The MOCVD equipment growth result anomaly tracing method according to claim 1, wherein the step of determining a process parameter value range and performing a simulation experiment by using a CFD model based on an experiment design method to obtain simulation process parameters and a simulation result specifically comprises:
determining the value range of the process parameters;
setting a plurality of groups of simulation experiments by using an experiment design method in a process parameter value range to obtain simulation process parameters;
and carrying out simulation experiment by using the CFD model according to the simulation process parameters to obtain a simulation result.
3. The MOCVD equipment growth result abnormity tracing method according to claim 1, wherein the step of matching the abnormal growth result with the process parameters based on the neural network model to obtain the abnormal process parameters specifically comprises:
setting a plurality of groups of process parameter values based on a control variable method;
inputting the technological parameter values into the neural network model in sequence to obtain a prediction result;
comparing the prediction result with the abnormal growth result based on a mean square error formula to obtain a minimum mean square error value;
and calculating the minimum mean square error values of different process parameters based on a probability formula to obtain abnormal process parameters.
4. The MOCVD equipment growth result anomaly tracing method according to claim 3, wherein the probability formula is specifically as follows:
Figure FDA0003795520380000011
in the above formula, P (x) i ) Is the probability of an abnormal growth outcome,
Figure FDA0003795520380000012
is the minimum mean square error value, x i Are process parameters.
5. The MOCVD equipment growth result anomaly tracing method according to claim 1, further comprising verifying a CFD model, specifically as follows:
setting a plurality of groups of process parameter experimental values and carrying out growth experiments by sequentially utilizing the MOCVD reaction chamber to obtain experimental results;
simulating the process parameter experiment value and sequentially utilizing the CFD model to carry out simulation and numerical simulation calculation to obtain a simulation result;
and calculating an experimental result and a simulation result based on the Pearson correlation coefficient, and verifying the CFD model.
6. The MOCVD equipment growth result anomaly tracing method according to claim 5, wherein the experimental result is a growth rate of a semiconductor film, and the calculation steps are as follows:
acquiring a semiconductor film obtained after the experiment is finished and the experiment time;
measuring the thickness of the semiconductor thin film;
and calculating the growth rate of the semiconductor film according to the thickness and the experimental time.
7. The system for tracing the growth result abnormity of the MOCVD equipment is characterized by comprising the following components:
the CFD building module is used for carrying out CFD modeling on the MOCVD reaction cavity to generate a CFD model;
the simulation module is used for determining the value range of the process parameters and performing simulation experiments by using the CFD model to obtain simulation process parameters and simulation results;
the neural network construction module is used for modeling according to the simulation process parameters and the simulation result to generate a neural network model;
and the matching module is used for matching the abnormal growth result with the process parameters based on the neural network model to obtain the abnormal process parameters.
CN202210968280.XA 2022-08-12 2022-08-12 MOCVD equipment growth result abnormity tracing method and system Pending CN115293067A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210968280.XA CN115293067A (en) 2022-08-12 2022-08-12 MOCVD equipment growth result abnormity tracing method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210968280.XA CN115293067A (en) 2022-08-12 2022-08-12 MOCVD equipment growth result abnormity tracing method and system

Publications (1)

Publication Number Publication Date
CN115293067A true CN115293067A (en) 2022-11-04

Family

ID=83828038

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210968280.XA Pending CN115293067A (en) 2022-08-12 2022-08-12 MOCVD equipment growth result abnormity tracing method and system

Country Status (1)

Country Link
CN (1) CN115293067A (en)

Similar Documents

Publication Publication Date Title
Susto et al. Multi-step virtual metrology for semiconductor manufacturing: A multilevel and regularization methods-based approach
US8768500B2 (en) Manufacturing design and process analysis system
US8145337B2 (en) Methodology to enable wafer result prediction of semiconductor wafer batch processing equipment
US11586794B2 (en) Semiconductor processing tools with improved performance by use of hybrid learning models
Ma et al. Fault detection based on statistical multivariate analysis and microarray visualization
CN108369916B (en) Wafer point-to-point analysis and data presentation
Su et al. A processing quality prognostics scheme for plasma sputtering in TFT-LCD manufacturing
CN107909216B (en) Method for predicting actual production cycle of part
CN116802653A (en) Hybrid physical/machine learning modeling of processes
CN112433472B (en) Semiconductor production control method and control system
CN115293067A (en) MOCVD equipment growth result abnormity tracing method and system
CN109643671B (en) Self-repairing semiconductor wafer processing
Greeneltch et al. Design-aware virtual metrology and process recipe recommendation
KR20020082219A (en) Method for finding optimal set-points for machines and processes
CN117882017A (en) System and method for adaptive fault removal for semiconductor manufacturing equipment
Shahmohammadi et al. Sequential model-based design of experiments for development of mathematical models for thin film deposition using chemical vapor deposition process
Chao et al. Hybrid differential evolution for estimation of kinetic parameters for biochemical systems
WO1999046646A1 (en) Optimization method and device
US20220237451A1 (en) Semiconductor process prediction method and semiconductor process prediction apparatus for heterogeneous data
Zhang et al. Run-to-run control of thermal atomic layer deposition
CN118202367A (en) Coded substrate material identifier communication tool
JP2010141042A (en) Control system, control method and control program for process treatment apparatus, and program recording medium
TWI847049B (en) Methods with improved semiconductor processing performance by use of hybrid learning models
US20230057521A1 (en) System and method for diagnosing state of mechanical equipment
CN113588884B (en) Pollution source searching method based on AMC on-line monitoring system

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