Search

CN-121985761-A - Intelligent regulation and control method and system for growth technological parameters of gallium nitride heteroepitaxial layer

CN121985761ACN 121985761 ACN121985761 ACN 121985761ACN-121985761-A

Abstract

The invention provides a method and a system for intelligently regulating and controlling growth process parameters of a gallium nitride heteroepitaxial layer, which relate to the technical field of semiconductor manufacturing and comprise the steps of collecting multilayer temperature distribution, gas phase component concentration, pressure fluctuation and substrate rotation speed data of a chamber; decomposing temperature gradient components, combining gas phase concentration and rotation speed, calculating to obtain space residence time distribution and local reaction rate distribution, generating a control sequence by adopting a multivariable decoupling compensation algorithm based on characteristic parameters of a key influence area, converting the control sequence into an execution instruction and synchronously executing the execution instruction, realizing the multi-parameter decoupling cooperative regulation of a chamber, and effectively improving the growth uniformity and the accurate control of an epitaxial layer.

Inventors

  • CHANG FENGYI
  • SONG WEI

Assignees

  • 中科(合肥)微电子研究院有限公司

Dates

Publication Date
20260505
Application Date
20260129

Claims (9)

  1. 1. The intelligent regulation and control method for the growth technological parameters of the gallium nitride heteroepitaxial layer is characterized by comprising the following steps: collecting multi-layer temperature distribution data, gas phase component concentration data, pressure fluctuation data and substrate rotation speed data in a gallium nitride epitaxial layer growth chamber; Decomposing the multilayer temperature distribution data into a vertical temperature gradient component and a radial temperature gradient component according to the height direction of the chamber, combining the gas phase component concentration data and the substrate rotation speed data, obtaining the space residence time distribution and the local reaction rate distribution of the reactant in the chamber through fluid dynamic coupling calculation, and identifying key influence areas causing uneven thickness of the epitaxial layer; Calculating independent control quantity of each executing mechanism by adopting a multivariable decoupling compensation algorithm based on the temperature gradient characteristic parameter, the concentration gradient characteristic parameter and the flow field characteristic parameter of the key influence area, wherein the multivariable decoupling compensation algorithm is used for separating the interference effect of temperature adjustment on gas flow, the disturbance effect of pressure adjustment on temperature distribution and the influence effect of flow adjustment on pressure stability by establishing a cross coupling matrix among parameters to generate a decoupled heating power compensation sequence, an air inlet flow compensation sequence and an air exhaust rate compensation sequence; The heating power compensation sequence is converted into a time sequence control instruction of a partition heater, the air inlet flow compensation sequence is converted into an opening instruction of a multi-channel mass flow controller, the exhaust rate compensation sequence is converted into a position instruction of a vacuum regulating valve, and the time sequence control instruction, the opening instruction and the position instruction are synchronously executed to realize decoupling cooperative regulation of multiple parameters of a chamber.
  2. 2. The method of claim 1, wherein decomposing the multilayer temperature distribution data into a vertical temperature gradient component and a radial temperature gradient component in a chamber height direction, combining the gas phase component concentration data and the substrate rotation speed data, obtaining a spatial residence time distribution and a local reaction rate distribution of a reactant substance in the chamber through hydrodynamic coupling calculation, and identifying a key influence region causing epitaxial layer thickness non-uniformity comprises: Performing space coordinate transformation on the multilayer temperature distribution data, calculating partial derivatives of the temperature of each measuring point to the height under a cylindrical coordinate system to obtain the vertical temperature gradient component, and calculating partial derivatives of the temperature to the radial distance to obtain the radial temperature gradient component; based on the vertical temperature gradient component and the radial temperature gradient component, establishing a convection diffusion equation describing a gas phase mass transfer process by combining the mole fraction distribution of each reaction precursor in the gas phase component concentration data; converting the substrate rotation speed data into tangential speed field distribution of gas in a chamber, and solving the tangential speed field distribution and the convection diffusion equation in a coupling way to obtain a three-dimensional concentration field evolution process of a reaction precursor in a chamber space; according to the three-dimensional concentration field evolution process, calculating the time of gas particles at each spatial position from an air inlet to the position, and obtaining the spatial residence time distribution; Calculating the reaction rate of each spatial position through a chemical reaction kinetic equation by combining the spatial residence time distribution, the vertical temperature gradient component and the radial temperature gradient component to obtain the local reaction rate distribution; And extracting a space region with the mean value fluctuation amplitude exceeding the mean value in the local reaction rate distribution, and marking the space region as the key influence region.
  3. 3. The method of claim 2, wherein converting the substrate rotational velocity data into a tangential velocity field distribution of the gas in the chamber, coupled with the convective diffusion equation, to obtain a three-dimensional concentration field evolution of the reactive precursor in the chamber space comprises: calculating tangential velocity distribution of gas in the boundary layer of the surface of the substrate due to viscous drag according to the substrate rotation speed data; Based on the tangential velocity distribution and the chamber geometry, obtaining tangential velocity field distribution and radial velocity field distribution of gas in the whole space of the chamber by solving a Navier-Stokes momentum conservation equation set; Substituting the tangential velocity field distribution and the radial velocity field distribution into a convection term of the convection diffusion equation, and setting a diffusion term coefficient according to a molecular diffusion coefficient of each reaction precursor; discretizing the convection diffusion equation by adopting a finite volume method, dividing a chamber space into a plurality of control body units, and obtaining the concentration value of the reaction precursor in each control body unit under each time step; reconstructing the concentration value of each time step according to the space coordinate and the time coordinate to form the three-dimensional concentration field evolution process describing the distribution rule of the reaction precursor along with the time and the space change rule.
  4. 4. The method of claim 1, wherein calculating the independent control amounts of each actuator using a multivariable decoupling compensation algorithm, wherein the multivariable decoupling compensation algorithm separates the disturbing effects of temperature regulation on gas flow, the disturbing effects of pressure regulation on temperature distribution, and the influencing effects of flow regulation on pressure stability by establishing a cross-coupling matrix between parameters, and generating the decoupled heating power compensation sequence, intake flow compensation sequence, and exhaust rate compensation sequence comprises: The temperature gradient characteristic parameter, the concentration gradient characteristic parameter and the flow field characteristic parameter of the key influence area are extracted to construct a multidimensional parameter vector representing the current state of the cavity; Based on the transfer function matrix, determining an influence coefficient of temperature regulation on gas flow velocity, an influence coefficient of pressure regulation on temperature distribution and an influence coefficient of flow regulation on pressure fluctuation by a system identification method, and constructing the cross coupling matrix; According to the main control channel gain matrix and the cross interference gain matrix, designing a decoupling compensator matrix so that the result of multiplying the decoupling compensator matrix and the cross coupling matrix is a diagonal matrix; And inputting the deviation of the multidimensional parameter vector and the target parameter vector into the decoupling compensator matrix, and calculating to obtain the decoupled heating power compensation sequence, the decoupled air inlet flow compensation sequence and the decoupled exhaust rate compensation sequence.
  5. 5. The method of claim 4, wherein performing a matrix decomposition operation on the cross-coupling matrix, extracting main diagonal elements to form a main control channel gain matrix, and extracting off-diagonal elements to form a cross-interference gain matrix comprises: Classifying elements of the cross-coupling matrix according to row and column indexes, wherein row indexes represent controlled parameter types, and column indexes represent control input types; Extracting elements with the same row index and column index from the cross coupling matrix, and arranging the elements according to the original position to form the main control channel gain matrix, wherein diagonal elements of the main control channel gain matrix respectively represent the direct control gain of heating power to temperature, the direct control gain of air inlet flow to concentration and the direct control gain of exhaust rate to pressure; and extracting elements with different row indexes and column indexes from the cross coupling matrix, and arranging the elements according to the original positions to form the cross interference gain matrix.
  6. 6. The method of claim 1, wherein converting the heating power compensation sequence to a time-series control command for a zone heater, converting the intake air flow compensation sequence to an opening command for a multi-way mass flow controller, and converting the exhaust gas rate compensation sequence to a position command for a vacuum regulator valve comprises: Acquiring a power response characteristic curve of the zone heater, a flow-opening mapping relation of the multi-path mass flow controller and a speed-position mapping relation of the vacuum regulating valve; According to the power compensation value of each time point in the heating power compensation sequence, combining the power response characteristic curve, determining the voltage value or the current value which is required to be applied by each partition heater at the corresponding time point through reverse table lookup or interpolation calculation, and generating the time sequence control instruction; According to the flow compensation value of each channel in the intake flow compensation sequence, combining the flow-opening mapping relation, determining the valve core opening angle or opening percentage of each path of mass flow controller through inverse function calculation, and generating the opening instruction; according to the rate compensation value in the exhaust rate compensation sequence, combining the rate-position mapping relation, and determining the valve opening position of the vacuum regulating valve or the step number of the stepping motor through numerical solution to generate the position instruction; performing time stamp synchronization processing on the time sequence control instruction, the opening instruction and the position instruction, and ensuring that the time difference of receiving instructions of all execution mechanisms is within a preset synchronization window; And sending the time sequence control instruction, the opening instruction and the position instruction which are subjected to the time stamp synchronization processing to a corresponding actuating mechanism driving unit through a control bus.
  7. 7. An intelligent regulation and control system for growth process parameters of a gallium nitride heteroepitaxial layer, for implementing the method as set forth in any one of claims 1 to 6, comprising: The first unit is used for collecting multi-layer temperature distribution data, gas phase component concentration data, pressure fluctuation data and substrate rotation speed data in the gallium nitride epitaxial layer growth chamber; The second unit is used for decomposing the multilayer temperature distribution data into a vertical temperature gradient component and a radial temperature gradient component according to the height direction of the chamber, combining the gas phase component concentration data and the substrate rotation speed data, obtaining the space residence time distribution and the local reaction rate distribution of the reaction substances in the chamber through hydrodynamic coupling calculation, and identifying a key influence area causing the uneven thickness of the epitaxial layer; The third unit is used for calculating the independent control quantity of each executing mechanism by adopting a multivariable decoupling compensation algorithm based on the temperature gradient characteristic parameter, the concentration gradient characteristic parameter and the flow field characteristic parameter of the key influence area, wherein the multivariable decoupling compensation algorithm is used for separating the interference effect of temperature regulation on gas flow, the disturbance effect of pressure regulation on temperature distribution and the influence effect of flow regulation on pressure stability by establishing a cross coupling matrix among parameters to generate a decoupled heating power compensation sequence, an air inlet flow compensation sequence and an air outlet rate compensation sequence; And the fourth unit is used for converting the heating power compensation sequence into a time sequence control instruction of the partition heater, converting the air inlet flow compensation sequence into an opening instruction of the multi-path mass flow controller, converting the exhaust rate compensation sequence into a position instruction of the vacuum regulating valve, and synchronously executing the time sequence control instruction, the opening instruction and the position instruction to realize decoupling cooperative regulation of multiple parameters of the chamber.
  8. 8. An electronic device, comprising: A processor; A memory for storing processor-executable instructions; Wherein the processor is configured to invoke the instructions stored in the memory to perform the method of any of claims 1 to 6.
  9. 9. A computer readable storage medium having stored thereon computer program instructions, which when executed by a processor, implement the method of any of claims 1 to 6.

Description

Intelligent regulation and control method and system for growth technological parameters of gallium nitride heteroepitaxial layer Technical Field The invention relates to the technical field of semiconductor manufacturing, in particular to an intelligent regulation and control method and system for growth process parameters of a gallium nitride heteroepitaxial layer. Background Gallium nitride (GaN) is used as a third-generation wide-bandgap semiconductor material, has the excellent characteristics of large forbidden bandwidth, high breakdown electric field, good thermal conductivity, high electron mobility and the like, and is widely applied to the fields of high-frequency high-power electronic devices, optoelectronic devices, microwave radio-frequency devices and the like. The quality of the growth of the gallium nitride epitaxial layer directly determines the performance and reliability of the final device, with epitaxial layer thickness uniformity being one of the key indicators for evaluating epitaxial quality. Currently, metal Organic Chemical Vapor Deposition (MOCVD) is a mainstream technology for industrially producing high-quality gallium nitride epitaxial layers, and the technology realizes atomic-level ordered deposition on the surface of a substrate by controlling parameters such as temperature, pressure, gas-phase reactant concentration, flow field and the like in a chamber. In the conventional gallium nitride epitaxial layer growth process, the regulation and control of process parameters mainly depend on experience setting and feedback adjustment of discrete sampling detection. The method can not grasp the complex temperature field, flow field and dynamic change of concentration field in the chamber in real time, and is difficult to accurately control the temperature gradient and gas-phase component distribution of the reaction area. Meanwhile, the existing control system often controls each process parameter such as temperature, pressure, air flow and the like as independent variables, and ignores the complex coupling relation existing between the process parameters and the air flow, so that disturbance influence on other parameters is inevitably generated when one parameter is adjusted. Disclosure of Invention The embodiment of the invention provides an intelligent regulation and control method and system for growth process parameters of a gallium nitride heteroepitaxial layer, which can solve the problems in the prior art. In a first aspect of the embodiment of the present invention, an intelligent control method for growth process parameters of a gallium nitride heteroepitaxial layer is provided, including: collecting multi-layer temperature distribution data, gas phase component concentration data, pressure fluctuation data and substrate rotation speed data in a gallium nitride epitaxial layer growth chamber; Decomposing the multilayer temperature distribution data into a vertical temperature gradient component and a radial temperature gradient component according to the height direction of the chamber, combining the gas phase component concentration data and the substrate rotation speed data, obtaining the space residence time distribution and the local reaction rate distribution of the reactant in the chamber through fluid dynamic coupling calculation, and identifying key influence areas causing uneven thickness of the epitaxial layer; Calculating independent control quantity of each executing mechanism by adopting a multivariable decoupling compensation algorithm based on the temperature gradient characteristic parameter, the concentration gradient characteristic parameter and the flow field characteristic parameter of the key influence area, wherein the multivariable decoupling compensation algorithm is used for separating the interference effect of temperature adjustment on gas flow, the disturbance effect of pressure adjustment on temperature distribution and the influence effect of flow adjustment on pressure stability by establishing a cross coupling matrix among parameters to generate a decoupled heating power compensation sequence, an air inlet flow compensation sequence and an air exhaust rate compensation sequence; The heating power compensation sequence is converted into a time sequence control instruction of a partition heater, the air inlet flow compensation sequence is converted into an opening instruction of a multi-channel mass flow controller, the exhaust rate compensation sequence is converted into a position instruction of a vacuum regulating valve, and the time sequence control instruction, the opening instruction and the position instruction are synchronously executed to realize decoupling cooperative regulation of multiple parameters of a chamber. Decomposing the multilayer temperature distribution data into a vertical temperature gradient component and a radial temperature gradient component according to the height direction of the chamber, combining the gas phas