Polysilicon deposition nozzle control system based on fluid field control

Abstract

This invention relates to the field of polycrystalline silicon deposition technology, and more particularly to a polycrystalline silicon deposition nozzle control system based on fluid field control. The system includes: a fluid field monitoring module for real-time acquisition of fluid field distribution data and nozzle parameters at various locations within the polycrystalline silicon deposition chamber; a fluid analysis module for determining the eddy current distribution and / or dead zone distribution within the fluid field based on changes in the fluid field distribution data at various locations within the polycrystalline silicon deposition chamber over a preset time period, and for determining the defect change type; and a nozzle adjustment module for determining a number of associated nozzles based on the eddy current distribution and / or dead zone distribution within the fluid field, and for determining the nozzle adjustment method based on the defect change type and the number of associated nozzles. This invention improves the accuracy of nozzle control and ensures the uniformity of polycrystalline silicon deposition.

Description

Technical Field

[0001] This invention relates to the field of polycrystalline silicon deposition technology, and in particular to a polycrystalline silicon deposition nozzle control system based on fluid field control. Background Technology

[0002] In polycrystalline silicon chemical vapor deposition (CVD), the uniformity of the fluid field formed by the injection of reactive gas into the cavity through the deposition nozzle directly determines the quality of the deposited layer and the production yield. Affected by nozzle layout, gas source fluctuations, and cavity flow channel constraints, the cavity fluid field is prone to two types of defects: eddies and dead zones. Eddies induce turbulent gas mixing, leading to abnormal local deposition rates and byproduct inclusions; dead zones cause reactive gas stagnation, resulting in cavity contamination and material loss. Current nozzle control methods largely rely on manual experience, lacking quantitative identification methods for the number and distribution of eddies and dead zones. They also fail to establish a mapping logic between defect characteristics, nozzle types, and control methods. Homogeneous control is used for defects caused by single or multiple nozzles, and for frequently occurring stable or unstable defects, resulting in poor targeting, ineffective defect elimination, and difficulty in ensuring the long-term stability of the polycrystalline silicon deposition fluid field.

[0003] Chinese patent CN107488836B discloses a method for depositing polycrystalline silicon thin films. This method uses a vertical tube furnace with silane as the reactant gas. During deposition, the gas flow rates at the bottom and top nozzles are essentially the same, with the top nozzle having a higher flow rate than the middle nozzle. The gas flow rate at the middle nozzle increases sequentially from bottom to top. Because the reactant gas flows upwards and is gradually consumed during the flow, the sequentially increasing gas flow rate at the middle nozzle compensates for the gas consumed during the flow, thus improving the uniformity between wafers. Furthermore, increasing the gas flow rate at the top reduces the partial pressure of the hydrogen gas generated during the reaction, further enhancing the uniformity of the polycrystalline silicon thin film within the top wafer.

[0004] The existing technology has the following problems: it adopts a fixed nozzle design and adjusts the fluid field state only by changing the nozzle gas flow rate. It lacks analysis and identification of eddies and dead zone defects, which will lead to aggravation of fluid field disturbance and cause uneven local deposition of polycrystalline silicon. Summary of the Invention

[0005] To address this, the present invention provides a polycrystalline silicon deposition nozzle control system based on fluid field control, which overcomes the problem in the prior art that uses a fixed nozzle design and adjusts the fluid field state only by changing the nozzle gas flow rate, lacking analysis and identification of eddies and dead zone defects, which leads to aggravated fluid field disturbances and causes uneven local deposition of polycrystalline silicon.

[0006] To achieve the above objectives, the present invention provides a polycrystalline silicon deposition nozzle control system based on fluid field control, comprising:

[0007] The fluid field monitoring module is used to collect fluid field distribution data and nozzle parameters at various locations inside the polysilicon deposition chamber in real time. The fluid field distribution data includes flow velocity, fluid pressure and fluid concentration, and the nozzle parameters include nozzle length, injection angle and nozzle working status.

[0008] The fluid analysis module is used to determine the eddy current distribution and / or dead zone distribution in the fluid field based on the changes in fluid field distribution data at various locations inside the polysilicon deposition cavity within a preset time period, and to determine the defect change type, wherein the defect change type includes stable and frequently occurring type and unstable and frequently occurring type.

[0009] A nozzle adjustment module is used to determine a number of associated nozzles based on the eddy current distribution and / or dead zone distribution within the fluid field, and to determine the nozzle adjustment method based on the defect variation type and the number of associated nozzles, including:

[0010] Under the first relative condition, the nozzle length adjustment amount is determined based on the eddy current distribution and / or dead zone distribution in the fluid field, so as to adjust the nozzle length of the associated nozzle;

[0011] Under the second relative condition, an abnormal distribution area is determined based on the vortex distribution and / or dead zone distribution, so as to adjust the working state of the nozzles within the abnormal distribution area;

[0012] Under the third relative condition, a nozzle adjustment coefficient is determined based on the distribution of each associated nozzle in order to adjust the spray angle of the nozzles inside the polysilicon deposition chamber.

[0013] Furthermore, the fluid analysis module divides the preset time period into several monitoring cycles, and determines several abnormal circulation regions based on the changes in flow field velocity at various locations inside the polysilicon deposition cavity within the preset time period. Based on the flow field pressure and flow field concentration of any abnormal circulation region within any monitoring cycle, the module determines the local change characterization value corresponding to the abnormal circulation region within the monitoring cycle, so as to determine whether eddies exist in the abnormal circulation region within the monitoring cycle.

[0014] Furthermore, the fluid analysis module determines several candidate dead zone regions based on the fluid concentration changes at various locations inside the polysilicon deposition cavity within the preset time period, and determines whether a dead zone exists in the candidate dead zone region within any monitoring period based on the concentration change characterization value and pressure change characterization value of any candidate dead zone region within any monitoring period.

[0015] The concentration change characterization value is determined based on the comparison between the flow field concentration in the candidate dead zone region and the preset concentration during the monitoring period;

[0016] The pressure change characterization value is determined based on the comparison between the flow field pressure change rate of the candidate dead zone region and the preset pressure change rate during the monitoring period.

[0017] Furthermore, the fluid analysis module determines whether eddies exist in the abnormal circulation region within the monitoring period based on the comparison result between the local change characterization value corresponding to any abnormal circulation region within any monitoring period and the first preset change threshold.

[0018] Furthermore, the fluid analysis module determines whether a dead zone exists in the candidate dead zone region within the monitoring period based on the comparison results of the concentration change characterization value of any candidate dead zone region within any monitoring period with the second preset change threshold and the comparison results of the pressure change characterization value with the third preset change threshold.

[0019] Furthermore, the fluid analysis module determines the type of defect change based on the comparison between the number of abnormal monitoring cycles and the preset number;

[0020] The abnormal monitoring cycle is the monitoring cycle in which eddies and / or dead zones exist.

[0021] Furthermore, the nozzle adjustment module determines several associated nozzles based on the comparison results between the key vortex region and the radiation region of each nozzle, as well as the comparison results between the key dead zone region and the radiation region of each nozzle.

[0022] The key eddy region is a set of abnormal circulation regions where eddies exist within any of the monitoring periods;

[0023] The critical dead zone region is a set of candidate dead zone regions that exist within any of the monitoring periods;

[0024] The radiation area of ​​any of the nozzles is determined based on the nozzle parameters of any nozzle.

[0025] Furthermore, under the first relative condition, the nozzle adjustment module determines the regional radiation coefficient based on the comparison result between the key vortex region and the associated nozzle radiation region and / or the comparison result between the key dead zone region and the associated nozzle radiation region, and determines the nozzle length adjustment amount based on the regional radiation coefficient and the preset nozzle length.

[0026] The first relative condition is that the defect change type is stable and frequent and the number of associated nozzles is unique.

[0027] Furthermore, under the second relative condition, the nozzle adjustment module determines the abnormal distribution area based on the key vortex region and / or the key dead zone region, and adjusts the working state of all nozzles in the abnormal distribution area to the start state;

[0028] The working states include a start state and a stop state;

[0029] The second relative condition is that the defect change type is unstable and occurs frequently, and the number of associated nozzles is unique.

[0030] Furthermore, under the third relative condition, the nozzle adjustment module clusters each of the associated nozzles to determine several cluster groups, and determines the nozzle adjustment coefficient corresponding to each cluster group based on the number of associated nozzles in each cluster group, so as to adjust the nozzle injection angle in the key vortex region and / or key dead zone region corresponding to each cluster group.

[0031] The third relative condition is that the number of associated nozzles is not unique.

[0032] Compared with existing technologies, the advantages of this invention are as follows: By setting up a fluid field monitoring module, this invention can collect fluid field distribution data and nozzle parameters at various locations inside the polysilicon deposition cavity in real time. This allows for precise capture of dynamic changes in the flow field. The collected fluid field data corresponds one-to-one with the spatial location of the cavity. Combined with the coordinate information of the nozzle position, it can achieve precise positioning of flow field defects and associated nozzles, improving the accuracy of subsequent nozzle control. By setting up a fluid analysis module, based on the changes in fluid field distribution data at various locations inside the polysilicon deposition cavity within a preset time period, it can achieve quantitative and precise identification of eddy / dead zone distribution, effectively eliminating the interference of occasional flow field fluctuations. Furthermore, by accurately distinguishing between two types of defects—stable and frequently occurring defects and unstable and frequently occurring defects—based on the data change patterns, it clarifies the dynamic characteristics of defects, avoids using homogeneous control strategies for different types of defects, and provides a basis for differentiated control. By setting up a nozzle adjustment module, associated nozzles are precisely located based on the distribution of eddies / dead zones, avoiding indiscriminate control and ensuring process continuity. The nozzle adjustment method is matched with the defect change type and the number of associated nozzles to achieve directional and precise control, making the adjustment method highly compatible with defect characteristics, avoiding blind control, and improving the efficiency of flow field defect elimination. Three adjustment methods are designed to adapt to stable and frequent defects, unstable and frequent defects, and global defects associated with multiple nozzles, respectively, taking into account both local single-point defects and overall layout defects, to achieve quantitative control of nozzles, improve the accuracy of nozzle control, and ensure the uniformity of polysilicon deposition.

[0033] Furthermore, the fluid analysis module of this invention divides a preset time period into several monitoring cycles and performs eddy current determination on a cycle-by-cycle basis. This allows for precise capture of the temporal variation patterns of eddies within each cycle. Using flow field velocity changes as the basis for determining abnormal circulation regions, it accurately captures areas where the velocity vector exhibits a rotational trend, quickly completing the screening and delineation of abnormal circulation regions. This narrows the scope of subsequent precise eddy current determination, avoiding the omission of potential eddy current regions. For each abnormal circulation region within a monitoring cycle, it combines the two core parameters of flow field pressure and flow field concentration to comprehensively calculate local change characterization values. Based on these characterization values, it determines whether eddies exist, achieving quantitative and precise eddy current determination, reducing the false positive rate, and adapting to flow field characteristics under different process parameters. Using the local change characterization values ​​of abnormal circulation regions as the core basis for eddy current determination, and determining eddies based on these local change characterization values ​​and a first preset change threshold, enables quantitative and precise identification of eddy current characteristics.

[0034] Furthermore, the fluid analysis module of this invention quickly locates candidate dead zones by analyzing fluid concentration changes, achieving preliminary coarse localization of dead zones. Then, by combining the concentration change characterization value, pressure change characterization value, and corresponding preset thresholds for dual comparison and verification, the dead zone is accurately determined. Instead of using a single threshold for mixed judgment, the concentration change characterization value and pressure change characterization value are independently compared with the second and third preset thresholds, respectively. This allows for the adaptation to the differentiated judgment criteria of the two types of dead zones with stagnant flow and stable pressure, effectively distinguishing between gas stagnation dead zones and ordinary low-flow-rate areas, thus improving the accuracy and stability of dead zone identification.

[0035] Furthermore, the nozzle adjustment module of this invention identifies associated nozzles by comparing key vortex regions, key dead zones, and nozzle radiation regions. Adjustment is only applied to nozzles that cause flow field defects, reducing ineffective adjustments and ensuring continuous and stable operation of the polysilicon deposition process. For scenarios where defects occur frequently and stably with a single associated nozzle, the regional radiation coefficient is determined by comparing key vortex regions and / or key dead zones with the associated nozzle radiation region. Combined with a preset nozzle length, the nozzle length adjustment is quantified, accurately characterizing the overlap and depth of influence between the defect region and the nozzle radiation range. This ensures a high degree of match between the nozzle length adjustment and defect characteristics, thoroughly compensating for insufficient single-nozzle jet directionality and abnormal airflow diffusion, while avoiding abnormal airflow parameters caused by excessive nozzle length adjustment. For scenarios where defects occur frequently and stably with a single associated nozzle, a soft control method is adopted, defining abnormal distribution regions and activating nozzles within these regions. No hardware modifications to the nozzles are required; only the nozzle's operating state is adjusted. This provides a fast response, quickly replenishing diversions, stabilizing airflow fluctuations, suppressing the random generation of frequent and unstable defects, and activating only abnormal distribution regions. Nozzles within a given area do not alter the operating status of nozzles in other areas, minimizing interference with the overall fluid field and deposition process. For global defect scenarios where the number of associated nozzles is not unique, clustering is used to group associated nozzles into several clusters based on their degree of association and distribution location. Adjustment coefficients are then determined based on the number of nozzles within each group, rather than adjusting individual nozzles independently. On one hand, clustering can accurately identify the synergistic interference relationships among multiple nozzles, avoiding airflow interference to other nozzles caused by adjusting a single nozzle, thus achieving synergistic control of multiple nozzles. On the other hand, the adjustment coefficients are quantified based on the number of nozzles within each group, making nozzle spray angle adjustments more scientific, achieving a systematic improvement in fluid field uniformity, and reducing the complexity of multi-nozzle adjustments. By covering all three types of relative conditions, precise adaptation to defect scenarios is achieved, improving the accuracy of nozzle control and ensuring the uniformity of polysilicon deposition. Attached Figure Description

[0036] Figure 1 This is a structural block diagram of a polycrystalline silicon deposition nozzle control system based on fluid field control, according to an embodiment of the present invention.

[0037] Figure 2 A logic diagram for determining whether an abnormal circulation region has eddies within any monitoring period, as shown in this embodiment of the invention.

[0038] Figure 3 This is a logic diagram for determining whether a candidate dead zone exists within any monitoring period, as shown in this embodiment of the invention.

[0039] Figure 4 A logic diagram for determining the type of defect change in an embodiment of the present invention. Detailed Implementation

[0040] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0041] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0042] Please see Figures 1-4 As shown, this embodiment of the invention provides a polycrystalline silicon deposition nozzle control system based on fluid field control, comprising:

[0043] The fluid field monitoring module is used to collect fluid field distribution data and nozzle parameters at various locations inside the polysilicon deposition chamber in real time. The fluid field distribution data includes flow velocity, fluid pressure and fluid concentration, and the nozzle parameters include nozzle length, injection angle and nozzle working status.

[0044] In implementation, a three-dimensional rectangular coordinate system is established with the geometric center of the deposition chamber as the origin (X-axis: radial direction of the chamber, Y-axis: circumferential direction of the chamber, Z-axis: axial / height direction of the chamber). The interior of the chamber is divided into three-dimensional grid units of equal size (e.g., 5mm×5mm×5mm, which can be adjusted according to the chamber size / deposition accuracy) according to the process accuracy requirements. A unique spatial coordinate code is assigned to each grid unit (e.g., X01Y02Z03). The reference coordinates of all nozzles in this coordinate system (the center position of the nozzle outlet) are calibrated, and the coverage radiation range of each nozzle is recorded (determined by spray simulation based on the nozzle position, nozzle length, and spray angle of a single nozzle), forming a chamber space-nozzle position mapping table. The nozzle's working state includes an on-state and a off-state. The working states of each nozzle inside the cavity can be different. The initial working state of each nozzle can be set according to the actual situation. The nozzles are telescopic and adjustable, meaning the nozzle length can be adjusted. Several nozzles are evenly distributed inside the polysilicon deposition cavity. There are no specific limitations on the nozzle type and nozzle position. The actual implementer can set them according to the actual situation. The number of nozzles is positively correlated with the size of the internal space of the polysilicon deposition cavity. The working states of each nozzle are set at intervals at the beginning of deposition, meaning that the working states of adjacent nozzles are different. The nozzle length and spray angle of each nozzle in the on-state are the initial nozzle length and spray angle, respectively. The actual implementer can set the initial nozzle length based on the actual situation or the average nozzle length in the polysilicon deposition process that has passed the qualification test in historical data. The actual implementer can set the initial spray angle based on the actual situation or the average spray angle in the polysilicon deposition process that has passed the qualification test in historical data.

[0045] It is understandable that sensor arrays can be uniformly deployed inside or outside the cavity to achieve full coverage of monitoring within the cavity. There are no specific limitations on the equipment structure and methods for acquiring flow field velocity, fluid pressure, and fluid concentration. For example, a miniature thermal anemometer or ultrasonic flow velocity sensor can be used to collect flow field velocity, a miniature piezoresistive pressure sensor can be used to collect flow field pressure, and a laser Raman spectroscopy concentration sensor or an online gas chromatograph sensor can be used to collect fluid concentration.

[0046] The fluid analysis module is used to determine the eddy current distribution and / or dead zone distribution in the fluid field based on the changes in fluid field distribution data at various locations inside the polysilicon deposition cavity within a preset time period, and to determine the defect change type, wherein the defect change type includes stable and frequently occurring type and unstable and frequently occurring type.

[0047] Specifically, the fluid analysis module divides the preset time period into several monitoring cycles, and determines several abnormal circulation regions based on the changes in flow field velocity at various locations inside the polysilicon deposition cavity within the preset time period. Based on the flow field pressure and flow field concentration of any abnormal circulation region within any monitoring cycle, the module determines the local change characterization value corresponding to the abnormal circulation region within the monitoring cycle, so as to determine whether eddies exist in the abnormal circulation region within the monitoring cycle.

[0048] Specifically, the fluid analysis module determines whether eddies exist in the abnormal circulation region within any monitoring period based on the comparison result between the local change characterization value corresponding to any abnormal circulation region within any monitoring period and the first preset change threshold.

[0049] In implementation, for any three-dimensional mesh cell, the corresponding velocity rotation intensity is calculated based on the flow field velocity changes within that three-dimensional mesh cell over a preset time period. For example, the three-dimensional velocity field v = (v... x ,v y ,v z ), v x v is the velocity component along the x-axis. y v is the y-axis velocity component. z If the z-axis velocity component is given, then the corresponding velocity rotation intensity is... Where sqrt() is the preset square root determination function. Let be the partial derivative of the z-axis velocity with respect to the y-axis. Let be the partial derivative of the y-axis velocity with respect to the z-axis. Let be the partial derivative of the x-axis velocity with respect to the z-axis. Let be the partial derivative of the z-axis velocity with respect to the x-axis. Let be the partial derivative of the y-axis velocity with respect to the x-axis. Let x be the partial derivative of the x-axis velocity with respect to the y-axis. If, at any given time point, the velocity rotation intensity corresponding to a given 3D grid cell is greater than a preset rotation intensity, then that 3D grid cell is designated as a candidate grid cell. For any candidate grid cell, if the percentage of time during which the velocity rotation intensity corresponding to that candidate grid cell is greater than the preset rotation intensity is greater than a first preset percentage, then that candidate grid cell is designated as an anomalous grid cell. Anomalous grid cells are then clustered to determine several anomalous cluster groups. Anomalous grid cells within any anomalous cluster group are then merged to form a corresponding anomalous circulation region. In practice, implementers can set the preset rotation intensity based on the actual situation or historical data showing the minimum velocity rotation intensity of eddy regions that have passed compliance testing. The first preset percentage is set to a range of 30%–40%.

[0050] It is understood that the preset time period should cover at least one complete polysilicon deposition process cycle. The actual implementer can set it based on the actual situation or the polysilicon deposition process cycles that have passed the qualification test in historical data. Preferably, the preset time period is set to one polysilicon deposition process cycle. The actual implementer can set the number of monitoring cycles based on the actual situation. Preferably, the monitoring cycle value range can be set to 3s to 5s.

[0051] Understandably, for any monitoring period and any abnormal circulation region, on the one hand, the average flow field pressure of the abnormal circulation region during the monitoring period is calculated as the abnormal pressure, and the average flow field pressure inside the polysilicon deposition cavity during the monitoring period is calculated as the overall pressure. The ratio of the abnormal pressure to the overall pressure is determined as the flow field pressure characterization value. On the other hand, the eddy current concentration variation curve of the abnormal circulation region during the monitoring period is constructed, and the eddy current concentration variation rate at each node is calculated. The ratio of the minimum eddy current concentration variation rate to the maximum eddy current concentration variation rate is determined as the flow field concentration characterization value, and the product of the flow field pressure characterization value and the flow field concentration characterization value is determined as the local variation characterization value corresponding to the abnormal circulation region during the monitoring period.

[0052] It is understandable that, for any monitoring period and any abnormal circulation region, if the local change characteristic value corresponding to the abnormal circulation region within that monitoring period is greater than a first preset change threshold, then it is determined that eddies exist in the abnormal circulation region within that monitoring period; if the local change characteristic value corresponding to the abnormal circulation region within that monitoring period is less than or equal to the first preset change threshold, then it is determined that eddies do not exist in the abnormal circulation region within that monitoring period. Practitioners can set the first preset change threshold based on the actual situation or the average local change characteristic value corresponding to eddy regions that have passed the compliance test in historical data.

[0053] Specifically, the fluid analysis module of this invention divides a preset time period into several monitoring cycles and performs eddy current determination on a cycle-by-cycle basis. This allows for precise capture of the temporal variation patterns of eddies within each cycle. Using flow field velocity changes as the basis for determining abnormal circulation regions, it accurately captures areas where the velocity vector exhibits a rotational trend, quickly completing the screening and delineation of abnormal circulation regions. This narrows the scope of subsequent precise eddy current determination, avoiding the omission of potential eddy current regions. For each abnormal circulation region within a monitoring cycle, it comprehensively calculates local change characterization values ​​based on two core parameters: flow field pressure and flow field concentration. The presence of eddies is then determined based on these characterization values, achieving quantitative and precise eddy current determination, reducing the false positive rate, and adapting to flow field characteristics under different process parameters. Using the local change characterization values ​​of abnormal circulation regions as the core basis for eddy current determination, and determining eddies based on these local change characterization values ​​and a first preset change threshold, enables quantitative and precise identification of eddy current characteristics.

[0054] Specifically, the fluid analysis module determines several candidate dead zones based on the fluid concentration changes at various locations within the polysilicon deposition cavity during the preset time period, and determines whether a dead zone exists within any of the candidate dead zones during any monitoring period based on the concentration change characterization value and pressure change characterization value of any of the candidate dead zones within any monitoring period; wherein, the concentration change characterization value is determined based on the comparison result of the flow field concentration of the candidate dead zone within the monitoring period and the preset concentration; the pressure change characterization value is determined based on the comparison result of the flow field pressure change rate of the candidate dead zone within the monitoring period and the preset pressure change rate.

[0055] Specifically, the fluid analysis module determines whether a dead zone exists in the candidate dead zone region within any monitoring period based on the comparison results of the concentration change characterization value of any candidate dead zone region within any monitoring period with the second preset change threshold and the pressure change characterization value with the third preset change threshold.

[0056] In implementation, for each three-dimensional grid cell, the average flow field velocity of each time slice within a preset time period is calculated, and three-dimensional grid cells with an average flow field velocity less than the preset velocity are identified as candidate dead zones. For any monitoring cycle and any candidate dead zone, on the one hand, the ratio of the percentage of time during which the average flow field concentration of the candidate dead zone is greater than the preset concentration to the percentage of time during which the average flow field concentration is greater than the preset concentration is calculated and determined as the first comparison value. The dead zone flow field concentration change curve of the candidate dead zone within the monitoring cycle is constructed, and the dead zone flow field of each node is calculated. The concentration change rate is calculated, and the ratio of the minimum dead zone flow field concentration change rate to the maximum dead zone flow field concentration change rate is determined as the second comparison value. The product of the first comparison value and the second comparison value is determined as the concentration change characterization value. On the other hand, the dead zone flow field pressure change curve of the candidate dead zone region is constructed within the monitoring period, and the dead zone flow field pressure change rate of each node is calculated. The ratio of the minimum dead zone flow field pressure change rate to the maximum dead zone flow field pressure change rate is determined as the pressure change characterization value. The second preset time period percentage is set to a range of 50% to 60%.

[0057] It is understood that, for any monitoring period and any candidate dead zone, if the concentration change value corresponding to the candidate dead zone is greater than the second preset threshold and the pressure change value is greater than the third preset threshold during that monitoring period, then the candidate dead zone is determined to exist within that monitoring period. If the concentration change value corresponding to the candidate dead zone is less than or equal to the second preset threshold, or the pressure change value is less than or equal to the third preset threshold, then the candidate dead zone is determined not to exist within that monitoring period. Practitioners can set the second preset threshold based on the average concentration change value of dead zones that have passed the compliance test in historical data, or the third preset threshold based on the average pressure change value of dead zones that have passed the compliance test in historical data.

[0058] Specifically, the fluid analysis module determines the type of defect change based on the comparison between the number of abnormal monitoring cycles and the preset number;

[0059] The abnormal monitoring cycle is the monitoring cycle in which eddies and / or dead zones exist.

[0060] During implementation, if the number of abnormal monitoring cycles exceeds the preset number, the defect change type is determined to be a stable and frequently occurring type. If the number of abnormal monitoring cycles is less than or equal to the preset number, the defect change type is determined to be an unstable and frequently occurring type. Implementers can set the preset number based on actual conditions or 1 / 3 to 1 / 2 of the total number of monitoring cycles.

[0061] Specifically, the fluid analysis module of this invention quickly locates candidate dead zones by analyzing changes in fluid concentration, achieving preliminary coarse localization of the dead zones. Then, by combining the concentration change characterization value, pressure change characterization value, and corresponding preset thresholds for dual comparison and verification, the dead zone is accurately determined. Instead of using a single threshold for mixed judgment, the concentration change characterization value and pressure change characterization value are independently compared with the second and third preset thresholds, respectively. This allows for adaptation to the differentiated judgment criteria of the two types of dead zones: flow stagnation and pressure stability. It effectively distinguishes between gas stagnation dead zones and ordinary low-flow-rate areas, improving the accuracy and stability of dead zone identification.

[0062] The nozzle adjustment module is used to determine a number of associated nozzles based on the eddy current distribution and / or dead zone distribution in the fluid field, and to determine the nozzle adjustment method based on the defect change type and the number of associated nozzles.

[0063] Specifically, the nozzle adjustment module determines a number of associated nozzles based on the comparison results of the key vortex region and the radiation region of each nozzle, as well as the comparison results of the key dead zone region and the radiation region of each nozzle; wherein, the key vortex region is a set of abnormal circulation regions with vortices within any monitoring period; the key dead zone region is a set of candidate dead zone regions with dead zones within any monitoring period; and each of the nozzle radiation regions is determined based on the nozzle parameters of any nozzle.

[0064] In implementation, the coverage radiation range of each nozzle can be determined according to the above cavity space-nozzle position mapping table, and this range can be used as the nozzle radiation area corresponding to that nozzle. For any monitoring cycle with eddies, the ratio of the overlapping area of ​​the nozzle radiation area and the key eddy region of any nozzle in the start state to the area of ​​the nozzle radiation area is determined as the eddy current comparison value for that monitoring cycle. If the eddy current comparison value is greater than the preset comparison value, the nozzle corresponding to the nozzle radiation area is determined as the associated nozzle. For any monitoring cycle with dead zones, the ratio of the overlapping area of ​​the nozzle radiation area and the key dead zone region of any nozzle in the start state to the area of ​​the nozzle radiation area is determined as the dead zone comparison value for that monitoring cycle. If the dead zone comparison value is greater than the preset comparison value, the nozzle corresponding to the nozzle radiation area is determined as the associated nozzle. The implementer can set the preset comparison value based on the actual situation or 1 / 5 to 1 / 4 of the nozzle radiation area.

[0065] Specifically, the nozzle adjustment method includes,

[0066] Under the first relative condition, the nozzle length adjustment amount is determined based on the eddy current distribution and / or dead zone distribution in the fluid field, so as to adjust the nozzle length of the associated nozzle;

[0067] Specifically, under the first relative condition, the nozzle adjustment module determines the regional radiation coefficient based on the comparison results of the key vortex region and the associated nozzle radiation region and / or the comparison results of the key dead zone region and the associated nozzle radiation region, and determines the nozzle length adjustment amount based on the regional radiation coefficient and the preset nozzle length; wherein, the first relative condition is that the defect change type is stable and frequent and the number of associated nozzles is unique.

[0068] In practice, if only eddies exist, the average value of the eddy current comparison for each monitoring cycle is determined as the regional radiation coefficient. If only dead zones exist, the average value of the dead zone comparison for each monitoring cycle is determined as the regional radiation coefficient. If both eddies and dead zones exist, the average value of the eddy current comparison and the dead zone comparison for each monitoring cycle is determined as the regional radiation coefficient. It is understood that only monitoring cycles with eddies and / or dead zones are counted; monitoring cycles without eddies and / or dead zones are not included in the calculation of the regional radiation coefficient.

[0069] Understandably, the product of the regional radiation coefficient and the preset nozzle length is used to determine the nozzle length adjustment amount. The preset nozzle length can be set to 1 / 2 to 2 / 3 of the initial length of the corresponding associated nozzle. If the associated nozzle is associated with a critical vortex region, the initial length of the associated nozzle is reduced according to the nozzle length adjustment amount. That is, the adjusted nozzle length is the difference between the initial length of the associated nozzle and the nozzle length adjustment amount. If the associated nozzle is associated with a critical dead zone region, the initial length of the associated nozzle is increased according to the nozzle length adjustment amount. That is, the adjusted nozzle length is the sum of the initial length of the associated nozzle and the nozzle length adjustment amount.

[0070] Under the second relative condition, an abnormal distribution area is determined based on the vortex distribution and / or dead zone distribution, so as to adjust the working state of the nozzles within the abnormal distribution area;

[0071] Specifically, under the second relative condition, the nozzle adjustment module determines the abnormal distribution area based on the key vortex area and / or the key dead zone area, and adjusts the working state of all nozzles in the abnormal distribution area to the start state; wherein, the working state includes the start state and the stop state; the second relative condition is that the defect change type is unstable and frequently occurs and the number of associated nozzles is unique.

[0072] In implementation, the union of the critical eddy region and the critical dead zone region is determined as the abnormal distribution region.

[0073] Under the third relative condition, a nozzle adjustment coefficient is determined based on the distribution of each associated nozzle in order to adjust the spray angle of the nozzles inside the polysilicon deposition chamber.

[0074] Specifically, under the third relative condition, the nozzle adjustment module clusters the associated nozzles to determine several cluster groups, and determines the nozzle adjustment coefficient corresponding to each cluster group based on the number of associated nozzles in each cluster group, so as to adjust the nozzle injection angle in the key vortex region and / or key dead zone region corresponding to each cluster group; wherein, the third relative condition is that the number of associated nozzles is not unique.

[0075] In implementation, the Euclidean distance between any two associated nozzles is calculated. This distance represents the proximity of the two associated nozzles on the horizontal plane. The smaller the distance, the more easily the airflow interferes with each other and the more likely they should be grouped into the same cluster. The clustering distance threshold is set to 0.5 to 0.8 times the smallest radiation area radius among the radiation areas of each associated nozzle. In practical applications, a greedy nearest-neighbor clustering algorithm can be used to cluster each associated nozzle. For example, each associated nozzle can be treated as an initial cluster. The inter-group distance between any two clusters is calculated sequentially. The inter-group distance is the distance between the two closest nozzles in the two groups. If the distance between two groups is less than or equal to the clustering distance threshold, the two clusters are merged into a new cluster. This process continues until the distance between any two clusters is greater than the clustering distance threshold, at which point the clustering ends. This yields several clusters. This is existing technology and will not be elaborated further. For any cluster group, the ratio of the number of associated nozzles in the cluster group to the number of nozzles in the start-up state is determined as the nozzle adjustment coefficient. If the region corresponding to the associated nozzles in the cluster group is a critical vortex region, the adjusted spray angle of the corresponding associated nozzle is determined by multiplying the nozzle adjustment coefficient by the average initial spray angle of the corresponding associated nozzle, and the adjusted spray angle of the corresponding associated nozzle is determined by the difference between the initial spray angle and the spray angle adjustment. If the region corresponding to the associated nozzles in the cluster group is a critical dead zone region, the adjusted spray angle of the corresponding associated nozzle is determined by multiplying the nozzle adjustment coefficient by the average initial spray angle of the corresponding associated nozzle, and the adjusted spray angle of each associated nozzle is determined by the sum of the initial spray angle and the spray angle adjustment.

[0076] Specifically, the nozzle adjustment module of this invention identifies associated nozzles by comparing key vortex regions, key dead zones, and nozzle radiation regions. Adjustments are made only to nozzles that cause flow field defects, reducing ineffective adjustments and ensuring the continuous and stable operation of the polysilicon deposition process. For scenarios where defects occur frequently and stably and there is only one associated nozzle, the regional radiation coefficient is determined by comparing key vortex regions and / or key dead zones with the radiation regions of associated nozzles. Combined with a preset nozzle length, the nozzle length adjustment is quantified, accurately characterizing the overlap and depth of influence between the defect region and the nozzle radiation range. This ensures a high degree of match between the nozzle length adjustment and defect characteristics, effectively compensating for insufficient single-nozzle jet directionality and abnormal airflow diffusion, while avoiding abnormal airflow parameters caused by excessive nozzle length adjustment. For scenarios where defects occur frequently and stably and there is only one associated nozzle, a soft control method is adopted, defining abnormal distribution regions and activating nozzles within these regions. No hardware modifications to the nozzles are required; only the nozzle's operating state is adjusted. This provides a fast response, quickly replenishing diversions, stabilizing airflow fluctuations, suppressing the random generation of frequent and unstable defects, and activating only the abnormal distribution region. Nozzles within the domain do not alter the operating status of nozzles in other areas, minimizing interference with the overall fluid field and deposition process. For global defect scenarios where the number of associated nozzles is not unique, clustering is used to group associated nozzles into several clusters based on their degree of association and distribution location. Adjustment coefficients are then determined based on the number of nozzles within each group, rather than adjusting individual nozzles independently. On one hand, clustering can accurately identify the synergistic interference relationships among multiple nozzles, avoiding airflow interference to other nozzles caused by adjusting a single nozzle, thus achieving synergistic control of multiple nozzles. On the other hand, the adjustment coefficients are quantified based on the number of nozzles within each group, making nozzle spray angle adjustments more scientific and achieving a systematic improvement in fluid field uniformity while reducing the complexity of multi-nozzle adjustments. By covering all three types of relative conditions, precise adaptation to defect scenarios is achieved, improving the accuracy of nozzle control and ensuring the uniformity of polysilicon deposition.

[0077] This invention, by setting up a fluid field monitoring module, collects real-time fluid field distribution data and nozzle parameters at various locations inside the polysilicon deposition cavity. This allows for precise capture of dynamic changes in the flow field, with each collected fluid field data corresponding one-to-one with its spatial location within the cavity. Combined with the coordinate information of the nozzle positions, precise location of flow field defects and associated nozzles can be achieved, improving the accuracy of subsequent nozzle control. Furthermore, by setting up a fluid analysis module, based on the changes in fluid field distribution data at various locations inside the polysilicon deposition cavity within a preset time period, quantitative and precise identification of eddy / dead zone distribution is achieved. This effectively eliminates interference from occasional flow field fluctuations and accurately distinguishes between two types of defects: those that occur frequently but are stable, and those that occur frequently but are unstable. This clarifies the dynamic characteristics of defects, avoids using homogeneous control strategies for different types of defects, and provides a basis for differentiated control. By setting up a nozzle adjustment module, associated nozzles are precisely located based on the distribution of eddies / dead zones, avoiding indiscriminate control and ensuring process continuity. The nozzle adjustment method is matched with the defect change type and the number of associated nozzles to achieve directional and precise control, making the adjustment method highly compatible with defect characteristics, avoiding blind control, and improving the efficiency of flow field defect elimination. Three adjustment methods are designed to adapt to stable and frequent defects, unstable and frequent defects, and global defects associated with multiple nozzles, respectively, taking into account both local single-point defects and overall layout defects, to achieve quantitative control of nozzles, improve the accuracy of nozzle control, and ensure the uniformity of polysilicon deposition.

[0078] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A polysilicon deposition nozzle regulation system based on fluid field control, characterized in that, include: The fluid field monitoring module is used to collect fluid field distribution data and nozzle parameters at various locations inside the polysilicon deposition chamber in real time. The fluid field distribution data includes flow velocity, fluid pressure and fluid concentration, and the nozzle parameters include nozzle length, injection angle and nozzle working status. The fluid analysis module is used to determine the eddy current distribution and / or dead zone distribution in the fluid field based on the changes in fluid field distribution data at various locations inside the polysilicon deposition cavity within a preset time period, and to determine the defect change type, wherein the defect change type includes stable and frequently occurring type and unstable and frequently occurring type. A nozzle adjustment module is used to determine a number of associated nozzles based on the eddy current distribution and / or dead zone distribution within the fluid field, and to determine the nozzle adjustment method based on the defect variation type and the number of associated nozzles, including: Under the first relative condition, the nozzle length adjustment amount is determined based on the eddy current distribution and / or dead zone distribution in the fluid field, so as to adjust the nozzle length of the associated nozzle; Under the second relative condition, an abnormal distribution area is determined based on the vortex distribution and / or dead zone distribution, so as to adjust the working state of the nozzles within the abnormal distribution area; Under the third relative condition, a nozzle adjustment coefficient is determined based on the distribution of each associated nozzle in order to adjust the spray angle of the nozzles inside the polysilicon deposition chamber.

2. The fluid field controlled polysilicon deposition nozzle regulation system of claim 1, wherein, The fluid analysis module divides the preset time period into several monitoring cycles, and determines several abnormal circulation regions based on the changes in flow field velocity at various locations inside the polysilicon deposition cavity within the preset time period. Based on the flow field pressure and flow field concentration of any abnormal circulation region within any monitoring cycle, the module determines the local change characterization value corresponding to the abnormal circulation region within the monitoring cycle, so as to determine whether there are eddies in the abnormal circulation region within the monitoring cycle.

3. The fluid field controlled polysilicon deposition nozzle regulation system of claim 2, wherein, The fluid analysis module determines several candidate dead zones based on the fluid concentration changes at various locations inside the polysilicon deposition chamber within the preset time period, and determines whether a dead zone exists in any candidate dead zone within any monitoring period based on the concentration change characterization value and pressure change characterization value of any candidate dead zone within any monitoring period. The concentration change characterization value is determined based on the comparison between the flow field concentration in the candidate dead zone region and the preset concentration during the monitoring period; The pressure change characterization value is determined based on the comparison between the flow field pressure change rate of the candidate dead zone region and the preset pressure change rate during the monitoring period.

4. The polycrystalline silicon deposition nozzle control system based on fluid field control according to claim 3, characterized in that, The fluid analysis module determines whether eddies exist in the abnormal circulation region within any monitoring period based on the comparison result between the local change characterization value corresponding to any abnormal circulation region within any monitoring period and the first preset change threshold.

5. The polycrystalline silicon deposition nozzle control system based on fluid field control according to claim 4, characterized in that, The fluid analysis module determines whether a dead zone exists in the candidate dead zone region within any monitoring period based on the comparison results of the concentration change characterization value of any candidate dead zone region within any monitoring period with the second preset change threshold and the pressure change characterization value with the third preset change threshold.

6. The polycrystalline silicon deposition nozzle control system based on fluid field control according to claim 5, characterized in that, The fluid analysis module determines the type of defect change based on the comparison between the number of abnormal monitoring cycles and the preset number. The abnormal monitoring cycle is the monitoring cycle in which eddies and / or dead zones exist.

7. The polycrystalline silicon deposition nozzle control system based on fluid field control according to claim 6, characterized in that, The nozzle adjustment module determines several associated nozzles based on the comparison results between the key vortex region and the radiation region of each nozzle, as well as the comparison results between the key dead zone region and the radiation region of each nozzle. The key eddy region is a set of abnormal circulation regions where eddies exist within any of the monitoring periods; The critical dead zone region is a set of candidate dead zone regions that exist within any of the monitoring periods; The radiation area of ​​any of the nozzles is determined based on the nozzle parameters of any nozzle.

8. The polycrystalline silicon deposition nozzle control system based on fluid field control according to claim 7, characterized in that, The nozzle adjustment module, under the first relative condition, determines the regional radiation coefficient based on the comparison results of the key vortex region and the associated nozzle radiation region and / or the comparison results of the key dead zone region and the associated nozzle radiation region, and determines the nozzle length adjustment amount based on the regional radiation coefficient and the preset nozzle length. The first relative condition is that the defect change type is stable and frequent and the number of associated nozzles is unique.

9. The polycrystalline silicon deposition nozzle control system based on fluid field control according to claim 8, characterized in that, Under the second relative condition, the nozzle adjustment module determines the abnormal distribution area based on the key vortex region and / or the key dead zone region, and adjusts the working state of all nozzles in the abnormal distribution area to the start state. The working states include a start state and a stop state; The second relative condition is that the defect change type is unstable and occurs frequently, and the number of associated nozzles is unique.

10. The polycrystalline silicon deposition nozzle control system based on fluid field control according to claim 9, characterized in that, Under the third relative condition, the nozzle adjustment module clusters the associated nozzles to determine several cluster groups, and determines the nozzle adjustment coefficient corresponding to each cluster group based on the number of associated nozzles in each cluster group, so as to adjust the nozzle injection angle in the key vortex region and / or key dead zone region corresponding to each cluster group. The third relative condition is that the number of associated nozzles is not unique.

Images