In-situ cleaning and monitoring method of vapor deposition chamber for APCVD tray system

Abstract

This invention provides an in-situ cleaning and monitoring method for the vapor deposition chamber of an APCVD tray system, relating to the field of semiconductor technology. The method includes: performing ellipsometric spectral scanning at multiple locations on the inner wall of the chamber using an ellipsometer to calculate the initial deposit thickness; calculating the flow field distribution based on the thickness and chamber geometry to identify areas with insufficient cleaning rates; setting a compensating gas injection port upstream of these areas; and periodically scanning during the cleaning process to determine whether the deposit has been removed to the substrate interface by analyzing the thickness evolution curve and changes in optical constants. This method can accurately monitor the cleaning process, prevent excessive substrate etching, and improve cleaning uniformity and efficiency.

Description

Technical Field

[0001] This invention relates to the field of semiconductor technology, and in particular to a method for in-situ cleaning and monitoring of the vapor deposition chamber of an APCVD tray system. Background Technology

[0002] During semiconductor manufacturing and thin film growth, byproducts and deposits gradually accumulate on the inner walls of vapor deposition chambers. These residues can affect the stability of subsequent processes and product quality. To ensure production continuity and product consistency, the chambers need to be cleaned regularly. Traditionally, the cleaning of vapor deposition chambers mainly employs plasma cleaning technology. This involves introducing reactive gases, such as those containing fluorine, and generating plasma under radio frequency or microwave excitation. The reactive particles react with the deposits on the inner walls of the chamber, converting them into volatile substances that are then removed.

[0003] As semiconductor device dimensions continue to shrink and process requirements become increasingly stringent, traditional chamber cleaning technology faces multiple challenges. The internal structure of the chamber is complex, and the thickness and composition of deposits vary significantly in different locations. Traditional cleaning methods use uniform time and parameters, which cannot accurately clean the deposits in different locations within the chamber. This results in some areas being under-cleaned while others are over-cleaned, which not only reduces equipment utilization but may also damage the chamber substrate.

[0004] The existing cleaning process lacks effective real-time monitoring methods, and mainly relies on experience to determine the cleaning time or roughly judge the endpoint through light emission spectrum. It is impossible to obtain the actual removal status of deposits at various locations on the inner wall of the chamber, resulting in a highly blind cleaning process and difficulty in accurately grasping the cleaning endpoint. This not only affects the cleaning efficiency but also increases the consumption of greenhouse gases such as fluorides. Summary of the Invention

[0005] This invention provides an in-situ cleaning and monitoring method for the vapor deposition chamber of an APCVD tray system, which can solve the problems in the prior art.

[0006] A first aspect of the present invention provides a method for in-situ cleaning and monitoring of the vapor deposition chamber of an APCVD tray system, comprising:

[0007] Initial elliptic spectral scanning was performed at multiple specific locations on the inner wall of the vapor deposition chamber using an elliptic measurement device. The initial deposit thickness at each specific location was calculated based on the elliptic parameters. Based on the initial deposit thickness and the chamber geometry, the flow field distribution of the reactive gas and the concentration distribution of active particles within the chamber were calculated to identify areas with insufficient cleaning rates.

[0008] A compensating gas injection port is set at the upstream flow field location of the region with insufficient cleaning rate, and the compensating gas injection flow rate is calculated based on the initial sediment thickness and target cleaning time of the region with insufficient cleaning rate.

[0009] During the cleaning process, the elliptic measurement device performs periodic elliptic spectral scanning at each characteristic location. Based on the time-series data of the elliptic parameters, the surface film thickness evolution curve and optical constant change trajectory at each specific location are calculated. The thickness decay acceleration parameter is obtained by performing second derivative analysis on the thickness evolution curve. When the thickness decay acceleration parameter changes from a positive value to a negative value, it indicates that the cleaning rate has reached its peak and then begins to decrease. At the same time, it is verified whether the optical constant has entered the intrinsic optical constant range of the substrate material. If both conditions are met, it is determined that the deposit at that location has been removed to the substrate interface. The plasma cleaning is then terminated and inert gas protection is activated to prevent substrate etching.

[0010] Initial ellipsometry scanning was performed at multiple specific locations on the inner wall of the vapor deposition chamber using an ellipsometry measurement device. The initial sediment thickness at each specific location was calculated based on the ellipsometry parameters, including:

[0011] Based on the chamber geometry information and the preset spatial sampling strategy, multiple specific locations are determined on the inner wall of the chamber. These multiple specific locations include the location near the main gas inlet, the location near the gas outlet, and the location evenly distributed on the side wall of the chamber.

[0012] The light source module of the ellipsometric measurement device emits a polarized beam at each specific position. The polarized beam is reflected by the surface at the specific position and then received by the detection module to obtain information on the polarization state change of the reflected light.

[0013] Based on the polarization state change information, the ellipticity parameters at each specific position are calculated, and the ellipticity parameters include the amplitude ratio and the phase difference;

[0014] The relationship between the refractive index and extinction coefficient of the sediment material as a function of wavelength was calculated based on the ellipticity parameters.

[0015] Based on the wavelength relationship between the ellipticity parameter and the refractive index and extinction coefficient, inversion calculations are performed to obtain the initial sediment thickness and initial sediment optical constant at each specific location. The initial sediment optical constant includes the initial refractive index and the initial extinction coefficient.

[0016] Based on the initial deposit thickness and chamber geometry information, the flow field distribution of the reactive gas and the concentration distribution of active particles within the chamber are calculated, and areas with insufficient cleaning rates are identified, including:

[0017] Based on the chamber geometry information, the gas flow rate and velocity at the main gas inlet are set as the inlet boundary conditions for flow field calculation, and the gas outlet pressure is set as the outlet boundary condition for flow field calculation.

[0018] Based on the inlet and outlet boundary conditions, the velocity vector and pressure distribution of the reactant gas are calculated by solving the partial differential equations describing the mass and momentum transport of the fluid.

[0019] Based on the velocity vector, combined with the diffusion coefficient of active particles and the convective transport effect of the reactant gas, the transport trajectory and residence time of active particles from the main gas inlet to the inner wall of the chamber under the drive of the flow field are calculated to obtain the concentration distribution of active particles.

[0020] The initial sediment thickness at each specific location is multiplied by the concentration of active particles, and then divided by the sediment material density to obtain the sediment reduction per unit time at each specific location. The sediment reduction per unit time is then determined as the theoretical cleaning rate at each specific location.

[0021] Identify specific locations where the initial sediment thickness is greater than a preset thickness threshold and the theoretical cleaning rate is less than a preset rate threshold, and define the area where the specific location is located as the area with insufficient cleaning rate.

[0022] Based on the velocity vector, combined with the diffusion coefficient of the active particles and the convective transport effect of the reactant gas, the transport trajectory and residence time of the active particles from the main gas inlet to the inner wall of the chamber under the drive of the flow field are calculated, and the concentration distribution of the active particles is obtained as follows:

[0023] An initial active particle release point is set at the main gas inlet, and an initial active particle concentration and initial release rate are set. The displacement of the active particles under the convection of the reaction gas is calculated based on the velocity vector, and the displacement is equal to the product of the velocity vector and the time step.

[0024] The diffusion displacement of active particles driven by the concentration gradient is calculated based on the diffusion coefficient of active particles. The diffusion displacement is positively correlated with the concentration gradient and the diffusion coefficient. The displacement generated by convection and the diffusion displacement generated by diffusion are vector superimposed to obtain the total displacement vector of active particles in a single time step.

[0025] Based on the total displacement vector, the movement path of the active particles from the main gas inlet to each specific position is gradually tracked to form the transport trajectory of the active particles;

[0026] The cumulative time step taken by active particles to reach a specific location is statistically analyzed to obtain the residence time of the active particles.

[0027] Based on the transport trajectory and the residence time, combined with the generation rate of active particles in the plasma excitation region and the consumption rate during the transport process, the net increase of active particles is calculated, and the net increase of active particles is taken as the active particle concentration.

[0028] A compensating gas injection port is set at the upstream flow field location of the region with insufficient cleaning rate. The compensating gas injection flow rate is calculated based on the initial sediment thickness and target cleaning time in the region with insufficient cleaning rate, including:

[0029] Determine the flow path of the reaction gas from the main gas inlet to the insufficient cleaning rate area; select a position on the flow path that is upstream of the insufficient cleaning rate area and whose distance from the insufficient cleaning rate area meets a preset distance condition, and set a compensation gas injection port at this position;

[0030] The initial sediment thickness at each specific location within the insufficient cleaning rate area is obtained, and the average initial sediment thickness within the insufficient cleaning rate area is calculated.

[0031] Divide the average initial sediment thickness by the target cleaning time to obtain the target cleaning rate required for the area where the cleaning rate is insufficient.

[0032] Based on the difference between the current theoretical cleaning rate and the target cleaning rate in the area with insufficient cleaning rate, calculate the required increase in active particle concentration to be added to the area with insufficient cleaning rate.

[0033] The total amount of active particles required to be replenished per unit time is calculated based on the increase in the concentration of active particles and the volume of the area where the cleaning rate is insufficient.

[0034] The total amount of active particles required to be replenished per unit time is divided by the active particle generation efficiency to obtain the injection flow rate of the compensation gas.

[0035] Based on the inversion of elliptic parameter time-series data, the surface film thickness evolution curves and optical constant variation trajectories at specific locations are calculated. Second-order derivative analysis of the thickness evolution curves yields the thickness decay acceleration parameters, including:

[0036] During the cleaning process, the elliptic measurement device performs periodic elliptic spectral scanning at each specific location at preset time intervals to obtain the elliptic parameters at each specific location at different times, forming elliptic parameter time series data.

[0037] The elliptic parameters at each time step in the elliptic parameter time series data are taken as observations, and the surface film thickness and optical constant at each time step are taken as parameters to be determined. The surface film thickness and optical constant values ​​at each specific location at each time step are obtained by inversion calculation.

[0038] The surface film thickness values ​​at each specific location at each time point are arranged in chronological order to form the surface film thickness evolution curve at each specific location;

[0039] Arrange the optical constant values ​​at each specific location at each time in chronological order to form the trajectory of the optical constant change at each specific location;

[0040] The first derivative of the surface film thickness evolution curve is calculated to obtain the rate of change of thickness over time at each specific location;

[0041] The second derivative of the thickness change rate over time is calculated to obtain the thickness decay acceleration parameter at each specific location. The thickness decay acceleration parameter characterizes the trend of the cleaning rate over time.

[0042] A second aspect of the present invention provides an in-situ cleaning and monitoring system for the vapor deposition chamber of an APCVD tray system, comprising:

[0043] The first unit is used to perform initial elliptic spectral scanning on multiple specific locations on the inner wall of the vapor deposition chamber of the APCVD tray system using an elliptic measurement device, calculate the initial deposit thickness at each specific location based on the elliptic parameters, and calculate the flow field distribution of the reactive gas and the concentration distribution of active particles in the chamber based on the initial deposit thickness and the chamber geometry information, and identify areas with insufficient cleaning rate.

[0044] The second unit is used to set a compensating gas injection port at the upstream flow field location of the region with insufficient cleaning rate, and to calculate the compensating gas injection flow rate based on the initial sediment thickness and the target cleaning time in the region with insufficient cleaning rate.

[0045] The third unit is used to perform periodic elliptic spectral scanning at each characteristic location during the cleaning process using the elliptic measurement device. Based on the time-series data of the elliptic parameters, the surface film thickness evolution curve and optical constant change trajectory at each specific location are calculated by inversion. The thickness decay acceleration parameter is obtained by performing second derivative analysis on the thickness evolution curve. When the thickness decay acceleration parameter changes from a positive value to a negative value, it indicates that the cleaning rate has reached its peak and then begins to decrease. At the same time, it is verified whether the optical constant has entered the intrinsic optical constant range of the substrate material. If both conditions are met, it is determined that the deposit at that location has been removed to the substrate interface, the plasma cleaning is terminated, and inert gas protection is activated to prevent substrate etching.

[0046] A third aspect of the present invention,

[0047] An electronic device is provided, comprising:

[0048] processor;

[0049] Memory used to store processor-executable instructions;

[0050] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0051] Fourth aspect of the present invention,

[0052] A computer-readable storage medium is provided, having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0053] The beneficial effects of this application are as follows:

[0054] Based on the initial sediment thickness and chamber geometry information, the distribution of the reaction gas flow field and the concentration of active particles are calculated, which can accurately identify areas with insufficient cleaning rate and solve the technical problem that traditional methods cannot effectively identify areas that are difficult to clean.

[0055] By setting up a compensating gas injection port in the upstream flow field of the region with insufficient cleaning rate and calculating a reasonable compensating gas injection flow rate, accurate compensation for the difficult cleaning region was achieved, significantly improving cleaning uniformity and efficiency.

[0056] By monitoring the thickness decay acceleration parameters through periodic elliptic spectral scanning combined with second-order derivative analysis, the cleaning progress can be judged in real time and the cleaning endpoint can be accurately identified, effectively avoiding substrate damage caused by over-cleaning.

[0057] Simultaneously verifying whether the optical constants fall within the intrinsic optical constant range of the substrate material provides a dual verification mechanism for determining the cleaning endpoint, significantly improving the accuracy of the cleaning endpoint determination. Attached Figure Description

[0058] Figure 1 This is a schematic flowchart of the in-situ cleaning and monitoring method for the vapor deposition chamber of the APCVD tray system according to an embodiment of the present invention;

[0059] Figure 2 A schematic diagram of the process for obtaining thickness decay acceleration parameters. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0061] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0062] refer to Figure 1 and Figure 2 The in-situ cleaning and monitoring method for the vapor deposition chamber of the APCVD tray system according to an embodiment of the present invention includes:

[0063] Initial elliptic spectral scanning was performed at multiple specific locations on the inner wall of the vapor deposition chamber of the APCVD tray system using an elliptic measurement device. The initial deposition thickness at each specific location was calculated based on the elliptic parameters. Based on the initial deposition thickness and the chamber geometry information, the flow field distribution of the reactive gas and the concentration distribution of active particles in the chamber were calculated to identify areas with insufficient cleaning rate.

[0064] A compensating gas injection port is set at the upstream flow field location of the region with insufficient cleaning rate, and the compensating gas injection flow rate is calculated based on the initial sediment thickness and target cleaning time of the region with insufficient cleaning rate.

[0065] During the cleaning process, the elliptic measurement device performs periodic elliptic spectral scanning at each characteristic location. Based on the time-series data of the elliptic parameters, the surface film thickness evolution curve and optical constant change trajectory at each specific location are calculated. The thickness decay acceleration parameter is obtained by performing second derivative analysis on the thickness evolution curve. When the thickness decay acceleration parameter changes from a positive value to a negative value, it indicates that the cleaning rate has reached its peak and then begins to decrease. At the same time, it is verified whether the optical constant has entered the intrinsic optical constant range of the substrate material. If both conditions are met, it is determined that the deposit at that location has been removed to the substrate interface. The plasma cleaning is then terminated and inert gas protection is activated to prevent substrate etching.

[0066] For example, the method of this application can be applied to an APCVD silicon wafer tray system that integrates in-situ purification and monitoring functions, which includes:

[0067] Integrated in-situ purification unit: The tray system includes a special base. This base integrates a gas distributor and a miniature plasma generator. During process breaks or maintenance modes, a fluorinated cleaning gas (such as NF3 or CF4) can be introduced into the bottom area of ​​the tray, generating low-temperature plasma to perform in-situ dry etching and cleaning of the tray bottom and sidewalls, removing deposits.

[0068] Pallet status monitoring system: Integrates a wireless passive surface acoustic wave temperature / pressure sensor on the pallet body or base, or uses an infrared thermometer and camera integrated into the device through an optical window on the base to monitor the surface temperature distribution and surface deposit coverage of the pallet in real time (judged by changes in color or reflectivity).

[0069] Modular quick-change design: The tray and base are connected by a precision snap or magnetic assisted docking structure, and are equipped with standardized air circuits and (optional) circuit interfaces to achieve quick disassembly and assembly, facilitating maintenance and replacement.

[0070] Furthermore, in this application, the APCVD (Atmospheric Pressure Chemical Vapor Deposition) tray system refers to an equipment system for performing chemical vapor deposition under atmospheric pressure conditions. This system includes a vapor deposition chamber, a tray transport mechanism, a gas supply system, and a temperature control system. In the APCVD tray system, the substrate is placed on a tray, and the deposition process of the thin film material is completed within the vapor deposition chamber through continuous or intermittent transport of the tray.

[0071] In this application, during the initial stage of the cleaning process, the cleaning rate gradually increases due to the sufficient contact between the active particles in the plasma and the sediment surface. At this time, the thickness decay acceleration parameter is positive, indicating that the cleaning rate is in an accelerating upward phase. When the cleaning progresses to the middle stage, the cleaning rate reaches its maximum value and then begins to slow down. At this point, the thickness decay acceleration parameter changes from a positive value to a negative value. This transition point marks the beginning of the decline in the cleaning rate after reaching its peak, and typically occurs when most of the sediment thickness has been removed and the sediment is close to the substrate interface.

[0072] Optical constants, including refractive index *n* and extinction coefficient *k*, are intrinsic parameters of a material's response to light. Different materials have different ranges of optical constants. During the cleaning process, as deposits are gradually removed, the optical constants detected by the ellipsometer gradually transition from the optical constants of the deposit material to those of the substrate material. Specifically:

[0073] The optical constants of sediment materials are typically related to their chemical composition and microstructure; for example, the refractive index of silicon oxide sediments usually falls within a specific range. The intrinsic optical constants of the substrate material are known physical properties; for example, the refractive index and extinction coefficient of a quartz substrate have definite numerical ranges at specific wavelengths. When the optical constants obtained from inversion calculations fall within the range of the intrinsic optical constants of the substrate material, it indicates that the ellipsoidally polarized beam has penetrated the remaining extremely thin sediment layer, and the optical response of the substrate material has been detected.

[0074] In practical applications, relying solely on either the thickness attenuation acceleration parameter or any one of the optical constants can lead to misjudgment.

[0075] Relying solely on the negative thickness attenuation acceleration parameter may result in a decrease in the washing rate before the sediment is completely removed, such as temporary fluctuations in the concentration of active particles. Relying solely on the optical constant to enter the substrate range may result in the detection of substrate signals even when sediment remains due to the greater penetration depth of the ellipsoidal beam.

[0076] Therefore, this invention employs a collaborative verification mechanism based on dual criteria:

[0077] The first criterion is that the thickness decay acceleration parameter changes from a positive value to a negative value, indicating that the cleaning kinetics have entered the final stage. The second criterion is that the optical constant enters the range of the intrinsic optical constant of the substrate material, indicating that the detection signal is dominated by the optical response of the substrate material. Only when both of the above criteria are met simultaneously can it be determined that the deposit at that specific location has been removed to the substrate interface.

[0078] Once it is determined that the sediment has been removed to the substrate interface, the system immediately performs the following operations:

[0079] Terminate plasma cleaning. Turn off the RF or microwave power supply to stop plasma excitation and prevent active particles from continuing to etch the exposed substrate material.

[0080] Initiate inert gas protection. Immediately introduce inert gas (such as nitrogen or argon) into the vapor deposition chamber to rapidly displace residual active particles and reactive gases from the chamber, while simultaneously forming an inert gas protective layer on the substrate surface.

[0081] Adjust the chamber pressure and temperature. Reduce the chamber pressure to a safe value, while controlling the chamber temperature to cool down slowly to avoid thermal stress damage to the substrate material due to sudden temperature changes.

[0082] The above process ensures that the substrate material is not etched due to excessive cleaning while the deposits are completely removed, thus protecting the structural integrity and surface quality of the inner wall of the vapor deposition chamber of the APCVD tray system.

[0083] In one optional implementation, an initial ellipticity spectral scan is performed at multiple specific locations on the inner wall of the vapor deposition chamber using an ellipticity measurement device. The initial sediment thickness at each specific location is then calculated based on the ellipticity parameters, including:

[0084] Based on the chamber geometry information and the preset spatial sampling strategy, multiple specific locations are determined on the inner wall of the chamber. These multiple specific locations include the location near the main gas inlet, the location near the gas outlet, and the location evenly distributed on the side wall of the chamber.

[0085] The light source module of the ellipsometric measurement device emits a polarized beam at each specific position. The polarized beam is reflected by the surface at the specific position and then received by the detection module to obtain information on the polarization state change of the reflected light.

[0086] Based on the polarization state change information, the ellipticity parameters at each specific position are calculated, and the ellipticity parameters include the amplitude ratio and the phase difference;

[0087] The relationship between the refractive index and extinction coefficient of the sediment material as a function of wavelength was calculated based on the ellipticity parameters.

[0088] Based on the wavelength relationship between the ellipticity parameter and the refractive index and extinction coefficient, inversion calculations are performed to obtain the initial sediment thickness and initial sediment optical constant at each specific location. The initial sediment optical constant includes the initial refractive index and the initial extinction coefficient.

[0089] Based on the chamber's geometric structure and a pre-defined spatial sampling strategy, several specific locations are determined on the chamber's inner wall. The chamber is typically cylindrical or rectangular. Representative measurement points are selected on the inner wall according to airflow distribution patterns and deposition uniformity requirements. These specific locations include those near the main gas inlet, near the gas outlet, and evenly distributed locations on the chamber's sidewalls. For a cylindrical chamber, four points can be selected within a 5-centimeter radius around the main gas inlet, three points near the outlet, one point every 45 degrees circumferentially on the chamber sidewalls, and one point every 10 centimeters axially. Pre-marking these locations ensures the ellipsometer can accurately align with the measurement points.

[0090] The ellipsometric measurement device controls the light source module to emit a polarized beam at each specific location. The device includes a light source module, a polarizer, an analyzer, and a detector module. The light source module typically uses a xenon lamp or a helium-neon laser, producing a continuous spectrum with wavelengths ranging from 400 to 800 nanometers. After passing through the polarizer, the light becomes linearly polarized, with the incident angle typically set to 60 to 70 degrees, close to Brewster's angle, providing maximum measurement sensitivity. The incident light is reflected from the sample surface at a specific location, changing its polarization state; the reflected light is received by the analyzer and detector module. The detector module uses a photomultiplier tube or photodiode array to record the intensity changes of the reflected light at different wavelengths. For locations within the chamber that are difficult to measure directly, a specially designed extended optical path system, such as a combination of optical fiber and mirrors, can be used to guide the beam into the designated location inside the chamber.

[0091] Based on the received reflected light information, polarization state change information is obtained, and ellipsometric parameters at each specific location are calculated. Polarization state change information is obtained by comparing the polarization state differences between the incident and reflected light. During the calculation, Fourier analysis is used to convert the modulated signal received by the detector into polarization parameters. The ellipsometric parameters include the amplitude ratio Ψ and phase difference Δ, representing the ratio and phase difference of the p-polarized and s-polarized reflection coefficients, respectively. For each measurement location, the values ​​of Ψ and Δ at different wavelengths are recorded to form spectral ellipsometric data.

[0092] The relationship between the refractive index and extinction coefficient of sediment materials as a function of wavelength was calculated based on ellipticity parameters. First, a basic optical model was assumed for the sediments, such as a homogeneous monolayer structure and a substrate of a known material (e.g., silicon or stainless steel). The optical properties of the sediments were described using either the Cauchy model or the Forouhi-Bloomer dispersion model. The Cauchy model is suitable for describing transparent or weakly absorbing materials, and its formula is simple and easy to fit; while the Forouhi-Bloomer model is more suitable for describing the optical properties of semiconductors or dielectric materials. The relationship between the refractive index and extinction coefficient of the material as a function of wavelength was obtained by fitting the ellipticity parameter spectral data using the least squares method.

[0093] Inversion calculations are performed based on the wavelength relationship between ellipticity parameters, refractive index, and extinction coefficient to obtain the initial sediment thickness and initial sediment optical constants at specific locations. The inversion calculation employs the Levenberg-Marquardt nonlinear least squares algorithm, finding the optimal fitting parameters through multiple iterations. An optical model is established, including the substrate, sediment layer, and any potential surface roughness layer, with initial guesses for sediment thickness and optical constants. The root mean square error between the theoretically calculated ellipticity parameters and the measured data is compared. Iteration stops when the error is less than a preset threshold (e.g., 0.001) or the maximum number of iterations is reached (e.g., 100), and the final sediment thickness and optical constants are output.

[0094] To improve measurement accuracy, multi-angle measurements can be performed at different incident angles, and the data can be combined for comprehensive analysis. For complex multilayer structures, more complex optical models can be constructed using the effective medium approximation theory, or a gradient refractive index model can be introduced to describe the compositional variations within the deposited layer. In practical applications, for different deposited materials (such as silica, silicon nitride, or metal oxides), a suitable optical dispersion model should be selected to obtain more accurate information on thickness and optical constants.

[0095] The initial sediment thickness distribution map obtained by the above methods can be used to evaluate the uniformity of deposition within the chamber, providing a basis for subsequent process optimization and chamber cleaning and maintenance, and effectively improving the stability of the vapor deposition process and product yield.

[0096] In one optional implementation, based on the initial deposit thickness and chamber geometry information, the flow field distribution of the reactive gas and the concentration distribution of active particles within the chamber are calculated, and areas with insufficient cleaning rates are identified, including:

[0097] Based on the chamber geometry information, the gas flow rate and velocity at the main gas inlet are set as the inlet boundary conditions for flow field calculation, and the gas outlet pressure is set as the outlet boundary condition for flow field calculation.

[0098] Based on the inlet and outlet boundary conditions, the velocity vector and pressure distribution of the reactant gas are calculated by solving the partial differential equations describing the mass and momentum transport of the fluid.

[0099] Based on the velocity vector, combined with the diffusion coefficient of active particles and the convective transport effect of the reactant gas, the transport trajectory and residence time of active particles from the main gas inlet to the inner wall of the chamber under the drive of the flow field are calculated to obtain the concentration distribution of active particles.

[0100] The initial sediment thickness at each specific location is multiplied by the concentration of active particles, and then divided by the sediment material density to obtain the sediment reduction per unit time at each specific location. The sediment reduction per unit time is then determined as the theoretical cleaning rate at each specific location.

[0101] Identify specific locations where the initial sediment thickness is greater than a preset thickness threshold and the theoretical cleaning rate is less than a preset rate threshold, and define the area where the specific location is located as the area with insufficient cleaning rate.

[0102] Based on the initial sediment thickness and chamber geometry information, the process of calculating the flow field distribution of the reactive gas and the concentration distribution of active particles within the chamber to identify areas with insufficient cleaning rates can be implemented as follows:

[0103] Acquire information about the chamber's geometry, including its three-dimensional dimensions, the location of internal components, and the position, shape, and size of the gas inlet and outlet. The chamber geometry can be modeled using computer-aided design software to create a three-dimensional digital model. Simultaneously, collect initial deposit thickness data at specific locations on the chamber's inner wall. This can be done using non-contact measuring devices such as laser scanners or optical thickness gauges, with measurement point intervals set to 5 mm, to generate a deposit thickness distribution map.

[0104] For setting boundary conditions in the flow field calculation, based on the chamber geometry, the gas flow rate and velocity are set as inlet boundary conditions at the main gas inlet. For example, for a fluorine-based cleaning gas using nitrogen as the carrier gas, the inlet velocity can be set to 10 m / s and the flow rate to 200 standard cubic centimeters per minute. Pressure is set as the outlet boundary condition at the gas outlet, typically a negative pressure of 300 Pascals. For the chamber walls, a no-slip boundary condition is set, meaning the fluid velocity at the walls is zero.

[0105] The partial differential equations describing mass and momentum transport in fluids were solved using computational fluid dynamics methods to calculate the velocity vector and pressure distribution of the reactant gas. Specifically, the Navier-Stokes equations were used for calculation, numerical discretization was performed using the finite volume method, and the SIMPLE algorithm was employed for solving. It should be noted that the methods used to solve the partial differential equations are well-known to those skilled in the art, and therefore will not be elaborated further. The chamber space was divided into grid cells. The grid size was set smaller in regions with large airflow variations, typically 0.5 to 2 mm; and larger in regions with smaller variations, typically 2 to 5 mm. The convergence criterion during the solution process was a residual value less than 10^(-4).

[0106] After obtaining the velocity vector field, the transport trajectory and residence time of the active particles under the flow field are calculated by combining the diffusion coefficient of the active particles with the convective transport effect of the gas. For fluorine-based active particles, the diffusion coefficient can be set to 0.05 cm² / s. The convection-diffusion equation is used for solution:

[0107] The rate of change of active particle concentration over time equals the sum of the diffusion term and the convection transport term, minus the consumption term. The diffusion term is proportional to the second derivative of the concentration gradient; the convection transport term is proportional to the dot product of the velocity vector and the concentration gradient; and the consumption term is proportional to the product of the reaction rate constant, the active particle concentration, and the sediment surface area. By solving this equation, the active particle concentration distribution at various locations within the chamber can be obtained.

[0108] The initial sediment thickness at each specific location is multiplied by the concentration of active particles at that location, and then divided by the sediment material density to obtain the sediment reduction rate per unit time at each specific location. For example, if the initial sediment thickness at a location is 2 micrometers, the concentration of active particles is 10^15 particles per cubic centimeter, the sediment is aluminum fluoride with a density of 2.9 grams per cubic centimeter, then the sediment reduction rate per unit time at that location can be calculated as a specific value, which is the theoretical cleaning rate for that location.

[0109] Finally, areas with insufficient cleaning rates are identified. A preset thickness threshold of 1 micrometer and a preset cleaning rate threshold of 0.1 micrometers per minute are set. All specific locations are iterated over; if the initial sediment thickness at a location is greater than 1 micrometer and the theoretical cleaning rate is less than 0.1 micrometers per minute, then the area at that specific location is identified as an area with insufficient cleaning rates.

[0110] In practical applications, areas with insufficient cleaning rates can be further visualized, such as by marking these areas in red on the 3D model of the chamber, to visually show the locations requiring special attention. Simultaneously, for these areas, the gas inlet flow distribution can be adjusted or auxiliary gas inlets can be added to increase the concentration of active particles in these areas, thereby improving cleaning efficiency.

[0111] This method is particularly effective for large semiconductor manufacturing chambers. For example, in a plasma etching chamber with a diameter of 500 mm and a height of 400 mm, this method identified an area with insufficient cleaning rate near the top edge of the chamber. After adjusting the nozzle distribution of the top gas spray head, the theoretical cleaning rate in this area increased by 45%, and the actual cleaning time was reduced by approximately 30%, demonstrating the effectiveness of the calculation method.

[0112] In one optional implementation, based on the velocity vector, and combining the diffusion coefficient of the active particles with the convective transport effect of the reactant gas, the transport trajectory and residence time of the active particles from the main gas inlet to the inner wall of the chamber under the drive of the flow field are calculated to obtain the active particle concentration distribution, including:

[0113] An initial active particle release point is set at the main gas inlet, and an initial active particle concentration and initial release rate are set. The displacement of the active particles under the convection of the reaction gas is calculated based on the velocity vector, and the displacement is equal to the product of the velocity vector and the time step.

[0114] The diffusion displacement of active particles driven by the concentration gradient is calculated based on the diffusion coefficient of active particles. The diffusion displacement is positively correlated with the concentration gradient and the diffusion coefficient. The displacement generated by convection and the diffusion displacement generated by diffusion are vector superimposed to obtain the total displacement vector of active particles in a single time step.

[0115] Based on the total displacement vector, the movement path of the active particles from the main gas inlet to each specific position is gradually tracked to form the transport trajectory of the active particles;

[0116] The cumulative time step taken by active particles to reach a specific location is statistically analyzed to obtain the residence time of the active particles.

[0117] Based on the transport trajectory and the residence time, combined with the generation rate of active particles in the plasma excitation region and the consumption rate during the transport process, the net increase of active particles is calculated, and the net increase of active particles is taken as the active particle concentration.

[0118] When simulating the concentration distribution of active particles in a plasma reaction chamber, it is first necessary to obtain the velocity vector field of the reactant gas. Then, by combining the diffusion characteristics of the active particles with the convective transport effect of the reactant gas, the transport trajectory and residence time of the active particles in the flow field are calculated, and finally the concentration distribution of the active particles is determined.

[0119] Initial active particle release points are set at the main gas inlet location; these release points can be uniformly distributed across the inlet cross-section. For specific plasma processes, such as oxygen atoms or oxygen ions in oxygen plasma, an initial active particle concentration C0 (e.g., 10⁻⁶) is set.15 pcs / cm 3 ) and initial release rate R0 (e.g., 10) 5 (plasma particles / second). The selection of initial concentration and release rate depends on parameters such as plasma power, gas composition, and operating pressure.

[0120] Based on the pre-acquired velocity vector field, the displacement of active particles under the convection of the reactant gas is calculated. At each time step Δt (typically chosen as 10), the displacement is calculated. -5 Up to 10 -3 Within a second (to ensure computational accuracy), the displacement ΔXc of an active particle due to gas convection can be expressed as the product of the velocity vector v and the time step Δt, i.e., ΔXc = v·Δt. Here, v represents the gas velocity vector at the current position of the active particle, including the vx, vy, and vz components in three-dimensional space.

[0121] Simultaneously, reactive particles are also subject to diffusion driven by the concentration gradient. The diffusion displacement ΔXd is related to the concentration gradient grad(C) and the diffusion coefficient D. For specific reactive particles, such as oxygen atoms in argon gas, the diffusion coefficient D can be determined according to the kinetic theory of gases, with a typical value of 10. -2 Up to 10 -1 cm 2 / s, varying with pressure and temperature. Calculations of diffusion displacement must consider the direction and magnitude of the concentration gradient; in regions with high concentration gradients, the diffusion effect is more significant.

[0122] By vector superimposing the displacement ΔXc generated by convection and the diffusion displacement ΔXd generated by diffusion, the total displacement vector of the active particles within a single time step is obtained as ΔX = ΔXc + ΔXd. This vector superposition takes into account the combined effect and directionality of the two transport mechanisms, thus more accurately reflecting the actual motion of the active particles.

[0123] The motion path of the active particles is tracked step-by-step based on the total displacement vector ΔX. Starting from the initial release point, the position of the active particles is updated to X(t+Δt)=X(t+ΔX) after each time step Δt. By continuously updating the position coordinates, a complete transport trajectory of the active particles from the main gas inlet to each region of the chamber is formed. To improve computational efficiency, an adaptive time step can be used, employing a smaller time step in regions with drastic velocity changes.

[0124] During the tracking of the movement of active particles, the cumulative time step taken for each active particle to reach a specific position is statistically analyzed to obtain the residence time τ = ∑Δt. Residence time is an important parameter for assessing the likelihood of interaction between active particles and the substrate surface; the longer the residence time, the higher the reaction probability.

[0125] The net increase in active particles is calculated by combining the generation rate G of active particles in the plasma-excited region with the consumption rate L during transport. The generation rate G depends on the plasma power density, electron energy distribution function, and gas molecule collision cross-section, and can be determined by a plasma chemical reaction kinetic model. The consumption rate L includes two parts: bulk deactivation (e.g., collisions with other gas molecules) and surface deactivation (e.g., contact with the chamber wall).

[0126] The net increase in active particles, ΔC = G·τ - L·τ, represents the change in active particle concentration during transport. Ultimately, the active particle concentration, C = C0 + ΔC, represents the active particle concentration at a specific location under the combined effects of flow field and diffusion.

[0127] In practical applications, such as semiconductor etching processes, this method can be used to calculate the concentration distribution of fluorine or chlorine radicals within the chamber and predict etching uniformity; in thin film deposition processes, it can calculate the concentration distribution of active precursor molecules and optimize deposition uniformity; in surface treatment applications, it can simulate the oxygen atom concentration distribution and predict the surface oxidation or functionalization effect.

[0128] The active particle concentration distribution calculated by the above method can be used to guide the design of plasma reactors, optimization of process parameters, and research on reaction mechanisms, thereby improving the uniformity and repeatability of plasma treatment and reducing process development costs.

[0129] In one optional embodiment, a compensating gas injection port is set at the upstream flow field location of the insufficient cleaning rate region, and the compensating gas injection flow rate is calculated based on the initial sediment thickness and target cleaning time of the insufficient cleaning rate region, including:

[0130] Determine the flow path of the reaction gas from the main gas inlet to the insufficient cleaning rate area; select a position on the flow path that is upstream of the insufficient cleaning rate area and whose distance from the insufficient cleaning rate area meets a preset distance condition, and set a compensation gas injection port at this position;

[0131] The initial sediment thickness at each specific location within the insufficient cleaning rate area is obtained, and the average initial sediment thickness within the insufficient cleaning rate area is calculated.

[0132] Divide the average initial sediment thickness by the target cleaning time to obtain the target cleaning rate required for the area where the cleaning rate is insufficient.

[0133] Based on the difference between the current theoretical cleaning rate and the target cleaning rate in the area with insufficient cleaning rate, calculate the required increase in active particle concentration to be added to the area with insufficient cleaning rate.

[0134] The total amount of active particles required to be replenished per unit time is calculated based on the increase in the concentration of active particles and the volume of the area where the cleaning rate is insufficient.

[0135] The total amount of active particles required to be replenished per unit time is divided by the active particle generation efficiency to obtain the injection flow rate of the compensation gas.

[0136] In semiconductor processing, the cleaning of deposits within the cavity often faces the problem of uneven cleaning, especially in areas where the cleaning rate may be insufficient. This embodiment provides a method for compensating for areas with insufficient cleaning rates by setting a compensating gas injection port at an appropriate location and calculating the required injection flow rate to solve this problem.

[0137] Determining the flow path of the reactant gas from the main gas inlet to the region with insufficient purging rate can be achieved through computational fluid dynamics (CFD) simulation. By simulating the gas flow within the cavity, a detailed flow field distribution from the main gas inlet to the region with insufficient purging rate can be obtained. In the CFD simulation, parameters such as the cavity geometry, the location and flow rate of the main gas inlet, temperature, and pressure conditions can be set. Then, the Navier-Stokes equations can be solved to obtain the velocity, pressure, and temperature field distributions within the cavity. Based on these distributions, the gas flow path can be plotted, determining how the reactant gas flows from the main gas inlet to the region with insufficient purging rate.

[0138] After determining the flow path, a location upstream of the insufficient cleaning rate region and at a distance meeting a preset condition must be selected as the location for the compensation gas injection port. The preset distance condition can be determined based on the gas diffusion characteristics. Typically, it needs to ensure that the compensation gas has sufficient distance to mix uniformly with the main gas, while preventing excessive loss of active particles before reaching the target region. For example, in a typical plasma cleaning chamber, this distance might be set 10 to 30 centimeters upstream of the boundary of the insufficient cleaning rate region, depending on the chamber size and gas flow velocity.

[0139] Next, the initial deposit thickness at specific locations within the insufficient cleaning rate area is obtained. This can be achieved by measuring multiple sampling points using an optical thickness gauge before cavity maintenance, or by collecting data through the built-in cavity monitoring system. Assuming n sampling points are selected within the insufficient cleaning rate area, the measured thicknesses are d1, d2, ..., The average initial sediment thickness is then calculated as the arithmetic mean of these thickness values.

[0140] Based on the obtained average initial sediment thickness and the preset target cleaning time, the target cleaning rate required for areas with insufficient cleaning rate is calculated. For example, if the average initial sediment thickness is 2 micrometers and the target cleaning time is 20 minutes, then the target cleaning rate is 0.1 micrometers / minute.

[0141] Next, it is necessary to determine the current theoretical cleaning rate for the area with insufficient cleaning rate. This can be determined through experimental measurement or calculation based on existing cleaning models. Assuming the current theoretical cleaning rate is 0.06 micrometers / minute, there is a difference of 0.04 micrometers / minute compared to the target cleaning rate of 0.1 micrometers / minute. This difference represents the cleaning rate that needs to be increased by compensating gas injection.

[0142] There is usually a proportional relationship between the cleaning rate and the concentration of active particles, which can be established based on experimental data or theoretical models. For example, assuming that under given conditions, the cleaning rate increases by 0.01 micrometers per minute for every 1 × 10^10 particles / cubic centimeter increase in the concentration of active particles, then to increase the cleaning rate by 0.04 micrometers per minute, an increase of 4 × 10^10 particles / cubic centimeter in the concentration of active particles is required.

[0143] Based on the calculated increase in active particle concentration and the volume of the area with insufficient cleaning rate, calculate the total amount of active particles required to be replenished per unit time. If the volume of the area with insufficient cleaning rate is 5000 cubic centimeters, then the total amount of active particles required to be replenished per unit time is 4 × 10^10 particles / cubic centimeter × 5000 cubic centimeters = 2 × 10^14 particles / unit time.

[0144] Finally, the required compensation gas injection flow rate is calculated based on the active particle generation efficiency of the compensation gas. The active particle generation efficiency depends on the type of compensation gas, the injection method, and the energy conditions within the cavity. For example, if oxygen is used as the compensation gas, under specific radio frequency power and pressure conditions, one standard cubic centimeter (sccm) of oxygen can produce 1 × 10^12 active oxygen atoms. Therefore, the required compensation gas injection flow rate is 2 × 10^14 atoms / unit time ÷ 1 × 10^12 atoms / sccm = 200 sccm.

[0145] The compensation gas injection flow rate calculated by the above method can accurately compensate for the active particles required in areas with insufficient cleaning rate, thereby achieving uniform cleaning within the target cleaning time, effectively solving the problem of uneven cavity cleaning, and improving the utilization efficiency and product yield of semiconductor processing equipment.

[0146] In one optional implementation, the surface film thickness evolution curve and optical constant change trajectory at each specific location are calculated by inversion based on elliptic parameter time-series data. Second-derivative analysis of the thickness evolution curve yields the thickness decay acceleration parameters, including:

[0147] During the cleaning process, the elliptic measurement device performs periodic elliptic spectral scanning at each specific location at preset time intervals to obtain the elliptic parameters at each specific location at different times, forming elliptic parameter time series data.

[0148] The elliptic parameters at each time step in the elliptic parameter time series data are taken as observations, and the surface film thickness and optical constant at each time step are taken as parameters to be determined. The surface film thickness and optical constant values ​​at each specific location at each time step are obtained by inversion calculation.

[0149] The surface film thickness values ​​at each specific location at each time point are arranged in chronological order to form the surface film thickness evolution curve at each specific location;

[0150] Arrange the optical constant values ​​at each specific location at each time in chronological order to form the trajectory of the optical constant change at each specific location;

[0151] The first derivative of the surface film thickness evolution curve is calculated to obtain the rate of change of thickness over time at each specific location;

[0152] The second derivative of the thickness change rate over time is calculated to obtain the thickness decay acceleration parameter at each specific location. The thickness decay acceleration parameter characterizes the trend of the cleaning rate over time.

[0153] Based on elliptic measurement technology, this method acquires the evolution trajectory of surface film thickness and changes in optical constants by real-time monitoring of elliptic parameters during the thin film cleaning process, and obtains the thickness attenuation acceleration parameter through second-order derivative analysis. This method can be applied to fields such as semiconductor manufacturing and optical thin film cleaning efficiency evaluation, providing an effective means for quantitatively evaluating the cleaning process.

[0154] An ellipsometry device was used to monitor the thin film surface in real time. The cleaned sample was placed on the sample stage of the ellipsometry device, and the measurement wavelength range was set to 300 nm to 800 nm, using a continuous scanning mode. An appropriate measurement time interval, typically 5 to 30 seconds, was set based on the sample characteristics and the expected duration of the cleaning process. During the cleaning process, the ellipsometry device periodically scanned specific locations on the sample surface at preset time intervals, acquiring the ellipsometry parameters Ψ and Δ values ​​at different times. These parameters, changing over time, formed ellipsometry parameter time-series data, recording the dynamic changes in the thin film properties during the cleaning process.

[0155] For each specific location, multiple measurement points can be pre-marked on the sample surface, such as a 3×3 grid of 9 points, to ensure consistency of measurement locations. Each scan will obtain a set of ellipsoidal parameter data for these locations at a specific time. For example, for the measurement of the i-th location at time tj, the obtained ellipsoidal parameters can be represented as Ψi(tj) and Δi(tj), where i represents the location number and j represents the time series number.

[0156] Next, based on the acquired ellipticity parameter time-series data, the film thickness and optical constants at each time point are inverted and calculated. First, an optical model suitable for the film under test is established; for example, the Cauchy dispersion model can be used for single-layer transparent films, and the Tauc-Lorentz model can be used for semiconductor materials. For the ellipticity parameters Ψi(tj) and Δi(tj) at each measurement time tj, the film thickness di(tj) and optical constants ni(tj) and ki(tj) are used as parameters to be determined. The difference between the theoretical calculation values ​​and the experimental measurements is minimized using the least squares method.

[0157] The Levenberg-Marquardt algorithm is used in the inversion calculation process. Reasonable initial values ​​and parameter constraints are set, and the calculation is iterative until convergence. When the root mean square error is less than a preset threshold (usually 0.01) or the maximum number of iterations (e.g., 100) is reached, the best fit result for that moment is considered to have been obtained. For each specific position i at time tj, the corresponding film thickness value di(tj) and optical constant values ​​ni(tj) and ki(tj) are obtained.

[0158] The thickness values ​​obtained from the inversion calculation are arranged in chronological order to form thin film thickness evolution curves. For each specific location i, its thickness evolution curve can be represented as di(t), t∈[t1, t2, ..., tm], where m represents the total number of measurements. These curves visually demonstrate the changing trend of the thin film thickness at each location during the cleaning process, reflecting the spatial distribution characteristics of the cleaning effect. Similarly, the optical constant values ​​ni(tj) and ki(tj) at each time point are arranged in chronological order to form optical constant variation trajectories, denoted as ni(t) and ki(t). The optical constant variation trajectories can reveal the changes in the composition or structure of the thin film material during the cleaning process.

[0159] The thickness evolution curve is subjected to first derivative calculation to obtain the thickness change rate at each location. The first derivative is calculated using the central difference method. The formula for the thickness change rate vi(tj) at location i at time tj is: vi(tj) = [di(tj+1) - di(tj-1)] / [tj+1 - tj-1]. The thickness change rate characterizes the cleaning rate; the larger the negative value, the faster the film is removed at that location.

[0160] Further calculations using the second derivative of the thickness change rate yielded the thickness decay acceleration parameter. Using the central difference method, the thickness decay acceleration ai(tj) at position i at time tj was calculated: ai(tj) = [vi(tj+1) - vi(tj-1)] / [tj+1 - tj-1]. The thickness decay acceleration parameter reflects the trend of the cleaning rate over time; a positive value indicates an increasing cleaning rate, while a negative value indicates a decreasing cleaning rate.

[0161] By analyzing the thickness decay acceleration parameter, key transition points in the cleaning process can be identified. For example, when the parameter changes from a positive to a negative value, it indicates that the cleaning rate has reached its peak and begins to decline; when the parameter approaches zero, it indicates that the cleaning rate is stabilizing. These characteristics can be used to optimize cleaning process parameters, such as determining the optimal cleaning time and adjusting the cleaning agent concentration.

[0162] In practical applications, spatial analysis can be performed on the thickness attenuation acceleration parameters at multiple specific locations to generate a two-dimensional distribution map of the acceleration parameters, visually displaying the spatial non-uniformity of the cleaning effect. For areas with poor cleaning results, the cleaning time can be appropriately extended or the cleaning agent formula can be adjusted.

[0163] This method provides quantitative evaluation indicators for the cleaning process through ellipsometry parameter time-series data inversion and second-derivative analysis, offering a scientific basis for the optimization and quality control of thin-film cleaning processes. Compared with traditional single-point measurement or offline detection methods, this method enables real-time monitoring and dynamic analysis of the cleaning process, improving the accuracy and efficiency of process control.

[0164] The present invention provides an in-situ cleaning and monitoring system for the vapor deposition chamber of an APCVD tray system, comprising:

[0165] The first unit is used to perform initial elliptic spectral scanning on multiple specific locations on the inner wall of the vapor deposition chamber of the APCVD tray system using an elliptic measurement device, calculate the initial deposit thickness at each specific location based on the elliptic parameters, and calculate the flow field distribution of the reactive gas and the concentration distribution of active particles in the chamber based on the initial deposit thickness and the chamber geometry information, and identify areas with insufficient cleaning rate.

[0166] The second unit is used to set a compensating gas injection port at the upstream flow field location of the region with insufficient cleaning rate, and to calculate the compensating gas injection flow rate based on the initial sediment thickness and the target cleaning time in the region with insufficient cleaning rate.

[0167] The third unit is used to perform periodic elliptic spectral scanning at each characteristic location during the cleaning process using the elliptic measurement device. Based on the time-series data of the elliptic parameters, the surface film thickness evolution curve and optical constant change trajectory at each specific location are calculated by inversion. The thickness decay acceleration parameter is obtained by performing second derivative analysis on the thickness evolution curve. When the thickness decay acceleration parameter changes from a positive value to a negative value, it indicates that the cleaning rate has reached its peak and then begins to decrease. At the same time, it is verified whether the optical constant has entered the intrinsic optical constant range of the substrate material. If both conditions are met, it is determined that the deposit at that location has been removed to the substrate interface, the plasma cleaning is terminated, and inert gas protection is activated to prevent substrate etching.

[0168] A third aspect of the present invention,

[0169] An electronic device is provided, comprising:

[0170] processor;

[0171] Memory used to store processor-executable instructions;

[0172] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0173] Fourth aspect of the present invention,

[0174] A computer-readable storage medium is provided, having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0175] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0176] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for in-situ cleaning and monitoring of the vapor deposition chamber of an APCVD tray system, characterized in that, include: Initial ellipticity spectral scanning was performed at multiple specific locations on the inner wall of the vapor deposition chamber of the APCVD tray system using an ellipticity measurement device. The initial deposition thickness at each specific location was calculated based on the ellipticity parameters. Based on the initial deposition thickness and the chamber geometry information, the flow field distribution of the reactive gas and the concentration distribution of reactive particles within the chamber were calculated to identify areas with insufficient cleaning rates. The multiple specific locations include the locations near the main gas inlet, the locations near the gas outlet, and the locations uniformly distributed on the sidewalls of the chamber. A compensating gas injection port is set at the upstream flow field location of the region with insufficient cleaning rate, and the compensating gas injection flow rate is calculated based on the initial sediment thickness and target cleaning time of the region with insufficient cleaning rate. During the cleaning process, the elliptic measurement device is used to perform periodic elliptic spectrum scanning at each specific location. Based on the time series data of the elliptic parameters, the surface film thickness evolution curve and optical constant change trajectory at each specific location are calculated. The thickness decay acceleration parameter is obtained by performing second derivative analysis on the thickness evolution curve. When the thickness decay acceleration parameter changes from a positive value to a negative value, it indicates that the cleaning rate has reached its peak and then begins to decrease. At the same time, it is verified whether the optical constant has entered the intrinsic optical constant range of the substrate material. If both conditions are met, it is determined that the deposit at that location has been removed to the substrate interface. The plasma cleaning is then terminated and inert gas protection is activated to prevent substrate etching. Optical constants, including refractive index n and extinction coefficient k, are intrinsic parameters of a material's response to light. Inversion calculations were performed based on the wavelength relationship between ellipticity parameters, refractive index, and extinction coefficient to obtain the initial sediment thickness and initial sediment optical constants at each specific location. The inversion calculations employed the Levenberg-Marquardt nonlinear least squares algorithm, which iteratively searched for the optimal fitting parameters.

2. The method according to claim 1, characterized in that, Initial ellipsometry scanning was performed at multiple specific locations on the inner wall of the vapor deposition chamber using an ellipsometry measurement device. The initial sediment thickness at each specific location was calculated based on the ellipsometry parameters, including: Based on the chamber geometry information and the preset spatial sampling strategy, multiple specific locations are determined on the inner wall of the chamber. The light source module of the ellipsometric measurement device emits a polarized beam at each specific position. The polarized beam is reflected by the surface at the specific position and then received by the detection module to obtain information on the polarization state change of the reflected light. Based on the polarization state change information, the ellipticity parameters at each specific position are calculated, and the ellipticity parameters include the amplitude ratio and the phase difference; The relationship between the refractive index and extinction coefficient of the sediment material as a function of wavelength was calculated based on the ellipticity parameters. Based on the wavelength relationship between the ellipticity parameter and the refractive index and extinction coefficient, inversion calculations are performed to obtain the initial sediment thickness and initial sediment optical constant at each specific location. The initial sediment optical constant includes the initial refractive index and the initial extinction coefficient.

3. The method according to claim 1, characterized in that, Based on the initial deposit thickness and chamber geometry information, the flow field distribution of the reactive gas and the concentration distribution of active particles within the chamber are calculated, and areas with insufficient cleaning rates are identified, including: Based on the chamber geometry information, the gas flow rate and velocity at the main gas inlet are set as the inlet boundary conditions for flow field calculation, and the gas outlet pressure is set as the outlet boundary condition for flow field calculation. Based on the inlet and outlet boundary conditions, the velocity vector and pressure distribution of the reactant gas are calculated by solving the partial differential equations describing the mass and momentum transport of the fluid. Based on the velocity vector, combined with the diffusion coefficient of active particles and the convective transport effect of the reactant gas, the transport trajectory and residence time of active particles from the main gas inlet to the inner wall of the chamber under the drive of the flow field are calculated to obtain the concentration distribution of active particles. The initial sediment thickness at each specific location is multiplied by the concentration of active particles, and then divided by the sediment material density to obtain the sediment reduction per unit time at each specific location. The sediment reduction per unit time is then determined as the theoretical cleaning rate at each specific location. Identify specific locations where the initial sediment thickness is greater than a preset thickness threshold and the theoretical cleaning rate is less than a preset rate threshold, and define the area where the specific location is located as the area with insufficient cleaning rate.

4. The method according to claim 3, characterized in that, Based on the velocity vector, combined with the diffusion coefficient of the active particles and the convective transport effect of the reactant gas, the transport trajectory and residence time of the active particles from the main gas inlet to the inner wall of the chamber under the drive of the flow field are calculated, and the concentration distribution of the active particles is obtained as follows: An initial active particle release point is set at the main gas inlet, and an initial active particle concentration and initial release rate are set. The displacement of the active particles under the convection of the reaction gas is calculated based on the velocity vector, and the displacement is equal to the product of the velocity vector and the time step. The diffusion displacement of active particles driven by the concentration gradient is calculated based on the diffusion coefficient of active particles. The diffusion displacement is positively correlated with the concentration gradient and the diffusion coefficient. The displacement generated by convection and the diffusion displacement generated by diffusion are vector superimposed to obtain the total displacement vector of active particles in a single time step. Based on the total displacement vector, the movement path of the active particles from the main gas inlet to each specific position is gradually tracked to form the transport trajectory of the active particles; The cumulative time step taken by active particles to reach a specific location is statistically analyzed to obtain the residence time of the active particles. Based on the transport trajectory and the residence time, combined with the generation rate of active particles in the plasma excitation region and the consumption rate during the transport process, the net increase of active particles is calculated, and the net increase of active particles is taken as the active particle concentration.

5. The method according to claim 1, characterized in that, A compensating gas injection port is set at the upstream flow field location of the region with insufficient cleaning rate. The compensating gas injection flow rate is calculated based on the initial sediment thickness and target cleaning time in the region with insufficient cleaning rate, including: Determine the flow path of the reaction gas from the main gas inlet to the insufficient cleaning rate area; select a position on the flow path that is upstream of the insufficient cleaning rate area and whose distance from the insufficient cleaning rate area meets a preset distance condition, and set a compensation gas injection port at this position; The initial sediment thickness at each specific location within the insufficient cleaning rate area is obtained, and the average initial sediment thickness within the insufficient cleaning rate area is calculated. Divide the average initial sediment thickness by the target cleaning time to obtain the target cleaning rate required for the area where the cleaning rate is insufficient. Based on the difference between the current theoretical cleaning rate and the target cleaning rate in the area with insufficient cleaning rate, calculate the required increase in active particle concentration to be added to the area with insufficient cleaning rate. The total amount of active particles required to be replenished per unit time is calculated based on the increase in the concentration of active particles and the volume of the area where the cleaning rate is insufficient. The total amount of active particles required to be replenished per unit time is divided by the active particle generation efficiency to obtain the injection flow rate of the compensation gas.

6. The method according to claim 1, characterized in that, Based on the inversion of elliptic parameter time-series data, the surface film thickness evolution curves and optical constant variation trajectories at specific locations are calculated. Second-order derivative analysis of the thickness evolution curves yields the thickness decay acceleration parameters, including: During the cleaning process, the elliptic measurement device performs periodic elliptic spectral scanning at each specific location at preset time intervals to obtain the elliptic parameters at each specific location at different times, forming elliptic parameter time series data. The elliptic parameters at each time step in the elliptic parameter time series data are taken as observations, and the surface film thickness and optical constant at each time step are taken as parameters to be determined. The surface film thickness and optical constant values ​​at each specific location at each time step are obtained by inversion calculation. The surface film thickness values ​​at each specific location at each time point are arranged in chronological order to form the surface film thickness evolution curve at each specific location; Arrange the optical constant values ​​at each specific location at each time in chronological order to form the trajectory of the optical constant change at each specific location; The first derivative of the surface film thickness evolution curve is calculated to obtain the rate of change of thickness over time at each specific location; The second derivative of the thickness change rate over time is calculated to obtain the thickness decay acceleration parameter at each specific location. The thickness decay acceleration parameter characterizes the trend of the cleaning rate over time.

7. An in-situ cleaning and monitoring system for the vapor deposition chamber of an APCVD tray system, used to implement the method as described in any one of claims 1-6, characterized in that, include: The first unit is used to perform initial elliptic spectral scanning of multiple specific locations on the inner wall of the vapor deposition chamber of the APCVD tray system using an elliptic measurement device, calculate the initial deposit thickness at each specific location based on the elliptic parameters, and calculate the flow field distribution of the reactive gas and the concentration distribution of active particles in the chamber based on the initial deposit thickness and the chamber geometry information, and identify areas with insufficient cleaning rate. The multiple specific locations include the location near the main gas inlet, the location near the gas outlet, and the uniformly distributed location on the side wall of the chamber. The second unit is used to set a compensating gas injection port at the upstream flow field location of the region with insufficient cleaning rate, and to calculate the compensating gas injection flow rate based on the initial sediment thickness and the target cleaning time in the region with insufficient cleaning rate. The third unit is used to perform periodic elliptic spectral scanning at specific locations during the cleaning process using the elliptic measurement device. Based on the time-series data of the elliptic parameters, the surface film thickness evolution curve and optical constant change trajectory at each specific location are calculated by inversion. The thickness decay acceleration parameter is obtained by performing second derivative analysis on the thickness evolution curve. When the thickness decay acceleration parameter changes from a positive value to a negative value, it indicates that the cleaning rate has reached its peak and then begins to decrease. At the same time, it is verified whether the optical constant has entered the intrinsic optical constant range of the substrate material. If both conditions are met, it is determined that the deposit at that location has been removed to the substrate interface, the plasma cleaning is terminated, and inert gas protection is activated to prevent substrate etching.

8. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 6.

Images