An etching process optimization method and system for high aspect ratio trench type memory IC

Abstract

The application discloses an etching process optimization method and system for a high-depth-width-ratio trench type storage IC, relates to the storage IC etching technical field, and has the following specific steps: collecting chamber internal polymer deposition thickness data, wafer table rotation speed gradient data, trench transverse line width fluctuation detection data, exhaust pumping speed segmented adjustment data to form a multi-source data set; when the trench depth-width-ratio fluctuation amplitude exceeds a device factory control threshold, the multi-source data set is subjected to time sequence calculation to generate a trench transverse line width fluctuation deviation interval; and the effect is that the two types of defects, i.e., trench sidewall deposition imbalance and trench bottom deposition imbalance, can be distinguished, corresponding pumping speed compensation margins are matched according to the deviation interval, reverse deviation of the trench size caused by excessive adjustment or insufficient adjustment of the pumping speed is avoided, the polymer deposition imbalance degree can be quantified, and the precision of line width deviation determination and compensation parameter matching is improved.

Description

Technical Field

[0001] This invention relates to the field of etching technology for memory ICs, and more specifically, to an optimized etching process method and system for high aspect ratio trench memory ICs. Background Technology

[0002] Trench-type integrated circuit memory relies on a high aspect ratio trench structure to achieve high-density memory cell arrangement. Trench dry etching is the core process that determines the uniformity of device dimensions and the yield of finished products. During the etching process, fluorocarbon polymers are continuously generated and deposited on the trench sidewalls, trench bottom, and inner walls of the etching chamber. Imbalance in polymer deposition thickness directly changes the trench lateral linewidth, causing excessive dispersion in the aspect ratio of the entire wafer trench. Once the fluctuation exceeds the factory control threshold, the electrical performance of the corresponding memory cell fails, directly reducing the production yield.

[0003] Existing processes rely solely on fixed segmented pumping speed settings, making it difficult to correlate four types of timing data: chamber polymer deposition thickness, wafer stage rotation speed, trench linewidth fluctuation, and segmented exhaust pumping speed. They can only passively adjust pumping speed parameters after detecting linewidth deviations, lacking a multi-source data coupling timing calculation mechanism. This makes it impossible to accurately classify trench sidewall deposition imbalance and trench bottom deposition imbalance as two different defect types. Using only fixed pumping speed compensation ranges, they cannot match appropriate compensation margins based on linewidth deviation ranges, frequently resulting in excessively large pumping speed adjustments causing reverse dimensional deviations in the trench, or insufficient adjustments failing to smooth out linewidth fluctuations. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide an optimized etching process method and system for high aspect ratio trench memory ICs.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] An optimization method for etching process of high aspect ratio trench memory ICs, the method includes the following specific steps:

[0007] A multi-source dataset is formed by collecting data on polymer deposition thickness inside the chamber, wafer stage rotation speed gradient, trench lateral linewidth fluctuation detection, and exhaust pumping speed segment adjustment.

[0008] When the trench aspect ratio fluctuation exceeds the device factory control threshold, time-series calculations are performed on the multi-source dataset to generate the trench lateral linewidth fluctuation deviation range. The trench polymer deposition imbalance type is determined based on the trench lateral linewidth fluctuation deviation range, and the pumping speed compensation margin is determined based on the trench polymer deposition imbalance type.

[0009] Based on the difference between the standard transverse linewidth of the trench and the measured transverse linewidth in the difference parameter, the segmented pumping speed adjustment amount required to smooth the linewidth fluctuation within the limited deposition correction time is calculated, and the first pumping speed control coefficient of the current batch of wafers is calculated based on the segmented pumping speed adjustment amount.

[0010] The complete trench etching process is divided into two stages: the core process for stabilizing linewidth and the auxiliary process for tolerating linewidth. The vacuum energy consumption of the core process for stabilizing linewidth and the auxiliary process for tolerating linewidth is calculated to obtain the second pumping speed control coefficient for the current batch of wafers.

[0011] The stratified compensation adjustment process is performed on the segmented exhaust pumping speed of the chamber based on the first pumping speed control coefficient and the second pumping speed control coefficient.

[0012] Preferably, the exhaust pumping speed segment adjustment data includes the main etching pumping speed setting, the passivation deposition pumping speed setting, and the cleaning and purging pumping speed setting.

[0013] Preferably, the pumping speed compensation margin is the exhaust flow compensation parameter corresponding to the difference between the standard transverse line width of the trench and the measured transverse line width.

[0014] Preferably, when the trench aspect ratio fluctuation exceeds the device's factory control threshold, a time-series calculation is performed on the multi-source dataset to generate the trench lateral linewidth fluctuation deviation range, specifically including the following steps:

[0015] Based on the preset multi-source dataset, abnormal data caused by etching process start-up and shutdown and external interference are removed to obtain regularized time-series data;

[0016] Based on regular time series data, the continuous change trend of the trench lateral linewidth with the etching process is determined to obtain the linewidth evolution trend.

[0017] The linewidth steady-state reference is obtained by matching the acceptable range of linewidth values ​​according to the device's factory control standards; the abnormal fluctuation component is obtained by calibrating the linewidth evolution trend with the linewidth steady-state reference.

[0018] The range of values ​​for the horizontal linewidth fluctuation deviation of the trench is generated by defining the value range based on the global floating boundary of the abnormal fluctuation component.

[0019] Preferably, the type of polymer deposition imbalance in the trench is determined based on the fluctuation range of the trench's lateral linewidth, and the pumping speed compensation margin is determined based on the type of polymer deposition imbalance in the trench. This specifically includes the following steps:

[0020] The deposition imbalance process attributes are obtained by matching the chamber deposition accumulation characteristics under the corresponding imbalance state according to the trench polymer deposition imbalance type.

[0021] Based on the trench lateral linewidth fluctuation deviation range corresponding to the deposition imbalance process attributes, the degree of imbalance between linewidth fluctuation and polymer deposition is matched.

[0022] The process deviation correction level is determined based on the degree of correlation imbalance, and the basic compensation range of the etching chamber exhaust pumping speed is matched according to the process deviation correction level.

[0023] By combining the process stability of high aspect ratio trench etching, the basic compensation range is modified for steady-state process adaptation to obtain the pumping speed compensation margin suitable for the current wafer batch.

[0024] Preferably, based on the difference between the standard transverse linewidth of the trench and the measured transverse linewidth in the difference parameter, the segmented pumping speed adjustment amount required to smooth the linewidth fluctuation within a limited deposition correction time is calculated, specifically including the following steps:

[0025] The line width fluctuation correction requirement is calculated based on the pumping speed compensation margin, the line width difference between the standard transverse line width of the trench and the measured transverse line width.

[0026] Based on the linewidth fluctuation correction requirements, a time-series correction benchmark is formed by matching the preset limited deposition correction time. Based on the time-series correction benchmark, the correction weight of each exhaust process in the overall trench etching process is determined to obtain the correction ratio of each process.

[0027] The basic adjustment amplitude of each process is obtained by combining the modified ratio of each process with the pumping speed compensation margin. Based on the deposition smoothing characteristics of high aspect ratio trench etching, the basic adjustment amplitude is steadily adapted and corrected to obtain the segmented pumping speed adjustment amount.

[0028] Preferably, the first pumping speed control coefficient for the current batch of wafers is calculated based on the segmented pumping speed adjustment amount, specifically including the following steps:

[0029] The process adaptability of each stage of the pumping speed adjustment is determined based on the matching characteristics between the segmented pumping speed adjustment and the polymer deposition smoothing of each segment of the etching process.

[0030] Based on the process adaptability, the segmented pumping speed adjustment is calibrated to obtain a standard adjustment amplitude that is suitable for the current etching conditions.

[0031] The process characteristics of the current batch of wafers are used to match the process characteristics with the completed conditions of the baseline process to obtain the batch process offset.

[0032] The standard adjustment amplitude is corrected by combining the batch process offset, and the first pumping speed control coefficient is obtained to match the batch etching requirements of the wafer.

[0033] Preferably, the complete trench etching process is divided into two steps: a core step for stabilizing linewidth and an auxiliary step for tolerating linewidth. Specifically, this includes the following steps:

[0034] Based on the adjustment amplitude according to the process standard, screen the strength of the correction effect of the pumping speed adjustment on the line width fluctuation in each pumping speed segment process, and determine the line width influence sensitivity of each pumping speed segment process.

[0035] Based on the sensitivity of linewidth influence, the polymer deposition molding effect weight of each pumping speed segment process is correlated, and the key processes for groove linewidth molding accuracy and morphological uniformity are screened out.

[0036] The critical processes with stringent requirements for linewidth control accuracy and high demand for fluctuation correction are marked as core processes for linewidth stability.

[0037] Conventional processes with minimal interference in the final linewidth forming of the groove and sufficient tolerance for process errors are designated as linewidth tolerance auxiliary processes.

[0038] Preferably, the second pumping speed control coefficient for the current batch of wafers is obtained by calculating the vacuum energy consumption of the core process for linewidth stabilization and the auxiliary process for linewidth tolerance, specifically including the following steps:

[0039] The process energy consumption characteristics are obtained by collecting the vacuum energy consumption loss status of the core process with stable linewidth and the auxiliary process with linewidth tolerance during operation.

[0040] The energy consumption characteristics are matched with the energy consumption baseline state of conventional standard processes. The line width forming weights of the core process for line width stability and the auxiliary process for line width tolerance are combined to determine the degree of interference of energy consumption of each process on the steady state of the etching process.

[0041] Based on the degree of interference, the energy consumption correction weights are preset for the core process and the auxiliary process respectively. Combined with the preset process energy consumption stability standard and energy consumption correction weights, the energy consumption loss characteristics of the process are balanced and adapted to obtain the energy consumption balance benchmark.

[0042] Based on the energy consumption balance benchmark, a second pumping speed control coefficient is generated to match the etching conditions of the current batch of wafers.

[0043] An etching process optimization system for high aspect ratio trench-type memory ICs includes:

[0044] Data acquisition module: Collects polymer deposition thickness data inside the chamber, wafer stage rotation speed gradient data, trench lateral linewidth fluctuation detection data, and exhaust pumping speed segment adjustment data to form a multi-source dataset;

[0045] Judgment module: When the trench aspect ratio fluctuation exceeds the device factory control threshold, the module performs time-series calculations on the multi-source dataset to generate the trench lateral linewidth fluctuation deviation range. Based on the trench lateral linewidth fluctuation deviation range, the module determines the trench polymer deposition imbalance type and determines the pumping speed compensation margin based on the trench polymer deposition imbalance type.

[0046] Calculation Module 1: Based on the difference between the standard transverse linewidth of the trench and the measured transverse linewidth in the difference parameter, calculate the segmented pumping speed adjustment amount required to smooth the linewidth fluctuation within the limited deposition correction time, and calculate the first pumping speed control coefficient of the current batch of wafers based on the segmented pumping speed adjustment amount.

[0047] Calculation Module 2: Divide the pumping speed segment of the complete trench etching process into a linewidth stabilization core process and a linewidth tolerance auxiliary process, and calculate the vacuum energy consumption of the linewidth stabilization core process and the linewidth tolerance auxiliary process to obtain the second pumping speed control coefficient of the current batch of wafers.

[0048] Execution module: Performs stratified compensation adjustment on the segmented exhaust pumping speed of the chamber according to the first pumping speed control coefficient and the second pumping speed control coefficient.

[0049] Compared with existing technologies, this invention has the following advantages: It constructs a multi-source time-series dataset by collecting four types of data: chamber polymer deposition thickness, wafer stage rotation speed gradient, trench lateral linewidth fluctuation, and segmented exhaust pumping speed. When the trench depth-to-width ratio fluctuation exceeds the control threshold, it performs time-series calculations to generate linewidth fluctuation deviation ranges. This allows for the differentiation between two types of defects: trench sidewall deposition imbalance and trench bottom deposition imbalance. Based on the deviation range, it matches the corresponding pumping speed compensation margin, avoiding the defects of the process not being able to couple multi-source time-series data and the compensation amplitude being fixed and singular. It also avoids reverse deviations in trench dimensions caused by over- or under-adjustment of pumping speed, quantifying the degree of polymer deposition imbalance and improving the accuracy of linewidth deviation judgment and compensation parameter matching. By using the difference parameter between the standard trench lateral linewidth and the measured lateral linewidth, combined with the pumping speed compensation margin and the limited deposition correction time, it calculates the segmented pumping speed adjustment amount for each process step by step, relying on layered calculation logic to obtain... The first pumping speed control coefficient enables the smoothing of trench linewidth fluctuations within a preset fixed time, solving the problem of lacking a timing constraint correction benchmark and being unable to quantitatively solve the problem of adapting the pumping speed adjustment range. It accurately controls the total amount of vacuum exhaust correction, shortens the process time for the trench morphology to recover to the standard range, and stabilizes the uniformity of the trench aspect ratio of the entire wafer. By analyzing the sensitivity of the pumping speed to the linewidth, the etching pumping speed segmentation process is divided into a core process for linewidth stabilization and a linewidth tolerance auxiliary process. The vacuum energy consumption of the two types of processes is calculated to obtain the second pumping speed control coefficient. Based on the two sets of control coefficients, layered differentiated pumping speed compensation is achieved. Sufficient correction weights are allocated to the core processes of main etching and passivation deposition to ensure trench forming accuracy. The adjustment range of the cleaning and purging auxiliary processes is reduced to reduce ineffective vacuum energy consumption. This overcomes the shortcomings of synchronous equal amplitude compensation for all processes and excessive energy consumption, taking into account both the device size control accuracy and the energy-saving requirements of mass production equipment. Attached Figure Description

[0050] Figure 1 This is a flowchart illustrating an optimized etching process for a high aspect ratio trench-type memory IC, as provided in an embodiment of the present invention.

[0051] Figure 2 This is a schematic diagram of a module for optimizing the etching process of a high aspect ratio trench memory IC, as provided in an embodiment of the present invention. Detailed Implementation

[0052] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0053] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0054] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places throughout this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0055] Reference Figures 1-2 As shown.

[0056] The embodiments further illustrate the etching process optimization method and system for high aspect ratio trench memory ICs proposed in this invention.

[0057] An optimization method for etching process of high aspect ratio trench memory ICs, the method includes the following specific steps:

[0058] A multi-source dataset is formed by collecting data on polymer deposition thickness inside the chamber, wafer stage rotation speed gradient, trench lateral linewidth fluctuation detection, and exhaust pumping speed segment adjustment.

[0059] When the trench aspect ratio fluctuation exceeds the device factory control threshold, time-series calculations are performed on the multi-source dataset to generate the trench lateral linewidth fluctuation deviation range. The trench polymer deposition imbalance type is determined based on the trench lateral linewidth fluctuation deviation range, and the pumping speed compensation margin is determined based on the trench polymer deposition imbalance type.

[0060] Based on the difference between the standard transverse linewidth of the trench and the measured transverse linewidth in the difference parameter, the segmented pumping speed adjustment amount required to smooth the linewidth fluctuation within the limited deposition correction time is calculated, and the first pumping speed control coefficient of the current batch of wafers is calculated based on the segmented pumping speed adjustment amount.

[0061] The complete trench etching process is divided into two stages: the core process for stabilizing linewidth and the auxiliary process for tolerating linewidth. The vacuum energy consumption of the core process for stabilizing linewidth and the auxiliary process for tolerating linewidth is calculated to obtain the second pumping speed control coefficient for the current batch of wafers.

[0062] The stratified compensation adjustment process is performed on the segmented exhaust pumping speed of the chamber based on the first pumping speed control coefficient and the second pumping speed control coefficient.

[0063] The system first calculates the overall control weight using the formula: Overall Control Weight = First Pumping Speed ​​Control Coefficient × Second Pumping Speed ​​Control Coefficient. This dimensionless value serves as the unified correction benchmark for tiered adjustments. For example, if the first pumping speed control coefficient is 0.448 and the second pumping speed control coefficient is 0.9417, substituting these values ​​into the calculation yields: Overall Control Weight = 0.448 × 0.9417 ≈ 0.4219. The overall control weight is then matched to the original pumping speed level for each process segment. The calculation formula is: Corrected Segmented Pumping Speed ​​Level = Original Segmented Pumping Speed ​​Level + Segmented Pumping Speed ​​Adjustment Amount × Overall Control Weight. All pumping speed values ​​are uniformly measured in liters per second to eliminate calculation errors caused by unit conversion.

[0064] Layered adjustment strictly distinguishes between the core linewidth stabilization process and the auxiliary linewidth tolerance process, applying differentiated compensation levels. The main etching process and passivation deposition process belong to the core linewidth stabilization process, and these two types of processes directly determine the uniformity of the trench's transverse linewidth. Sufficient pumping speed compensation is fully applied using the comprehensive control weight. The cleaning and purging process belongs to the auxiliary linewidth tolerance process, used only for chamber cleaning and not directly changing the trench morphology. The system adds an auxiliary attenuation coefficient to the comprehensive control weight to reduce the adjustment range. The auxiliary attenuation coefficient is fixed at 0.3, and the calculation formula is: Auxiliary process corrected pumping speed level = Original cleaning and purging pumping speed level + Cleaning and purging segmented pumping speed adjustment amount × Comprehensive control weight × 0.3. This reduces meaningless changes to the vacuum pumping power of the auxiliary process, saving equipment energy.

[0065] For example, the current main etching pumping speed is 35 liters per second, with a corresponding segmented pumping speed adjustment of 7 liters per second. Substituting these values ​​into the formula, the corrected main etching pumping speed is approximately 37.95 liters per second (35 + 7 × 0.4219). The passivation deposition pumping speed is 18 liters per second, with a segmented pumping speed adjustment of 5 liters per second. The corrected passivation deposition pumping speed is approximately 20.11 liters per second (18 + 5 × 0.4219). The cleaning and purging pumping speed is 60 liters per second, with a segmented pumping speed adjustment of -5 liters per second. Adding the attenuation coefficient, the corrected cleaning and purging pumping speed is approximately 60 + (-5) × 0.4219 × 0.3 ≈ 59.37 liters per second.

[0066] After the layered compensation adjustment is completed, the three stages of the etching process run simultaneously with the corrected pumping speed. Within the preset deposition correction time, the chamber exhaust flow rate is adjusted evenly to balance the polymer deposition thickness on the trench sidewalls and bottom, quickly narrow the fluctuation range of the trench lateral linewidth, and stabilize the depth-to-width ratio within the factory control threshold range. At the same time, the overall energy consumption of the vacuum unit is controlled by the auxiliary process attenuation adjustment mechanism, achieving the dual process effect of improving chip yield and reducing production costs.

[0067] The exhaust pumping speed segment adjustment data includes the main etching pumping speed level, the passivation deposition pumping speed level, and the cleaning and purging pumping speed level.

[0068] The main etching pumping speed control regulates the vacuum exhaust rate during the silicon substrate removal stage of the trench. This process involves intense plasma etching reactions, generating a large amount of fluoropolymer precursors. If the speed setting is too low, byproducts accumulate in the chamber, causing polymer thickening on the trench sidewalls, resulting in a shrinkage of the lateral linewidth and an excessive aspect ratio. If the speed setting is too high, active plasma is removed, leading to insufficient trench depth. The formula for calculating this adjustment is: Main etching segment pumping speed adjustment = Standard main etching pumping speed baseline value - Real-time main etching speed setting value. All values ​​are converted to liters per second before being included in the calculation. This process is crucial for linewidth stability, and pumping speed changes have the highest weighting on linewidth. It allocates the largest energy consumption percentage when calculating the second pumping speed control coefficient and provides the largest correction during layered compensation. For example, if the measured main etching speed is 32 liters per second and the standard baseline value is 40 liters per second, substituting these values ​​into the formula yields: Main etching segment pumping speed adjustment = 40 - 32 = 8 liters per second. This value is directly included in the calculation of the first pumping speed control coefficient.

[0069] The passivation deposition pumping speed control regulates the exhaust rate during the passivation film formation stage on the trench sidewalls. This process only involves polymer deposition and does not etch the silicon substrate. A speed too low will result in an excessively thick passivation film at the top of the trench and a thinner film at the bottom, leading to sidewall deposition imbalance. A larger pumping speed compensation margin is needed. Conversely, a speed too high will remove deposited gas, resulting in insufficient passivation film thickness and lateral etching, causing trench lateral linewidth expansion. The corresponding calculation formula is: Passivation deposition segmented pumping speed adjustment = Standard passivation deposition pumping speed baseline value - Real-time passivation deposition speed adjustment value, with units uniformly expressed in liters per second. This process is also a core process for linewidth stabilization, with energy consumption weight second only to the main etching process, and the correction magnitude is secondary when adjusting in layers. For example, if the real-time passivation deposition speed is 16 liters per second and the standard baseline is 22 liters per second, the passivation deposition segmented pumping speed adjustment = 22 - 16 = 6 liters per second.

[0070] The cleaning and purging pumping speed setting only applies to the chamber cleaning process after etching a single batch of wafers. It removes residual polymer particles and does not directly change the current wafer trench morphology, only indirectly affecting the initial deposition state of the next batch. This process is classified as a linewidth tolerance auxiliary process. The calculation formula is: Cleaning and purging segmented pumping speed adjustment = Standard cleaning and purging pumping speed baseline value - Real-time cleaning and purging speed setting value, with units uniformly in liters per second. When solving for the second pumping speed control coefficient, the lowest energy consumption weight is assigned. During layered compensation, only minor adjustments are made, or even the original value is maintained, reducing ineffective vacuum energy consumption. If the real-time cleaning and purging speed setting is 62 liters per second and the standard baseline is 54 liters per second, the cleaning and purging segmented pumping speed adjustment = 54 - 62 = -8 liters per second.

[0071] After the three sets of pumping speed adjustments are calculated individually, they are superimposed. The total segmented pumping speed adjustment is calculated as follows: main etching segment pumping speed adjustment + passivation deposition segment pumping speed adjustment + cleaning and purging segment pumping speed adjustment. The total adjustment is combined with the limited deposition correction time to calculate the first pumping speed control coefficient, and then combined with the vacuum energy consumption loss values ​​of the three processes to calculate the second pumping speed control coefficient. The two sets of coefficients together generate a comprehensive adjustment weight, which is used to perform differentiated and layered compensation for the three pumping speed levels, balance the polymer deposition thickness in the trench, quickly smooth out lateral linewidth fluctuations, and simultaneously rely on the low-weight adjustment of auxiliary processes to control the overall vacuum energy consumption, taking into account both trench dimensional accuracy and mass production costs.

[0072] The pumping speed compensation margin is the exhaust flow compensation parameter corresponding to the difference between the standard transverse line width of the trench and the measured transverse line width.

[0073] The difference between the standard transverse linewidth and the measured transverse linewidth of the trench is calculated using the formula: Transverse linewidth difference = Standard transverse linewidth value - Measured transverse linewidth value. The original unit for the linewidth value is often nanometers, while the unit for exhaust flow rate is liters per second. Since these units cannot be directly calculated, a linewidth-flow rate conversion factor is needed to unify the units. This conversion factor, calibrated through multiple mass production tests, represents the basic exhaust flow rate correction range corresponding to each 1-nanometer linewidth deviation, and is measured in liters per second per nanometer. The calculation formula is: Basic flow rate offset = Transverse linewidth difference × Linewidth-flow rate conversion factor. After calculation, all calculations are unified to liters per second, eliminating calculation errors caused by unit inconsistencies.

[0074] The pumping speed compensation margin is matched based on the base flow rate offset. This margin is then superimposed with a fixed imbalance correction coefficient based on the previously determined trench polymer deposition imbalance type. The complete formula is: Pumping Speed ​​Compensation Margin = Base Flow Rate Offset × Imbalance Correction Coefficient. Different deposition imbalance types correspond to independent imbalance correction coefficient values. When deposition imbalance occurs at the bottom of the trench, plasma lateral erosion is more severe, resulting in a larger imbalance correction coefficient. When deposition imbalance occurs on the trench sidewall, uneven passivation film distribution has a lower impact, resulting in a smaller imbalance correction coefficient. After determining the deposition imbalance type, the system retrieves the matched coefficient and substitutes it into the calculation to obtain the pumping speed compensation margin adapted to the current wafer defect state.

[0075] For example, the standard lateral linewidth of a memory IC trench is 60 nanometers, while the actual measured lateral linewidth of the wafer is 52 nanometers. Substituting these values ​​into the formula, we get the lateral linewidth difference = 60 - 52 = 8 nanometers. The process calibration linewidth flow rate conversion factor is 0.3 liters per second per nanometer. The calculated basic flow rate offset is 8 × 0.3 = 2.4 liters per second. The system determines that the current defect belongs to the deposition imbalance at the bottom of the trench, and the corresponding imbalance correction factor is 1.5. The final pumping speed compensation margin is 2.4 × 1.5 = 3.6 liters per second. This value of 3.6 liters per second is the basic compensation benchmark for subsequent calculation of segmented pumping speed adjustment.

[0076] The pumping speed compensation margin is applied throughout the subsequent process control logic. Based on this margin, the system constrains the upper and lower limits of pumping speed adjustment for each segment of the process, preventing excessive pumping speed adjustment from causing reverse deviations in trench morphology, while also avoiding insufficient adjustment that fails to smooth linewidth fluctuations. This parameter transforms the abstract trench size deviation into a directly controllable vacuum exhaust flow rate value, providing a precise quantitative basis for correcting polymer deposition imbalances. It also provides a fixed compensation benchmark for solving the first pumping speed control coefficient, enabling quantitative adaptive correction of high aspect ratio trench etching processes.

[0077] When the trench aspect ratio fluctuation exceeds the device's factory control threshold, time-series calculations are performed on the multi-source dataset to generate the trench lateral linewidth fluctuation deviation range. This includes the following steps:

[0078] Based on the preset multi-source dataset, abnormal data caused by etching process start-up and shutdown and external interference are removed to obtain regularized time-series data;

[0079] Based on regular time series data, the continuous change trend of the trench lateral linewidth with the etching process is determined to obtain the linewidth evolution trend.

[0080] The linewidth steady-state reference is obtained by matching the acceptable range of linewidth values ​​according to the device's factory control standards; the abnormal fluctuation component is obtained by calibrating the linewidth evolution trend with the linewidth steady-state reference.

[0081] The range of values ​​for the horizontal linewidth fluctuation deviation of the trench is generated by defining the value range based on the global floating boundary of the abnormal fluctuation component.

[0082] The system performs anomaly data removal by retrieving complete multi-source datasets, including polymer deposition thickness data inside the etching chamber, wafer stage rotation speed gradient data, trench lateral linewidth fluctuation detection data, and exhaust pumping speed segment adjustment data. It removes jump data at the start and stop of the etching equipment according to preset filtering rules, while also filtering out interference sampling points caused by external temperature and pressure disturbances. The remaining time-series aligned, distortion-free values ​​are combined into regularized time-series data. For example, if a single batch of data acquires 120 time-series sampling points, 8 of which are anomalous values ​​at the start and stop of the equipment, and 12 are interference data caused by workshop temperature fluctuations, after removal, 100 sets of regularized time-series data are ultimately retained for subsequent calculations, avoiding interference from distorted data in the accuracy of subsequent trend judgment.

[0083] The system extracts the linewidth evolution trend based on the regularized time-series data. The regularized time-series data contains the trench lateral linewidth detection values ​​corresponding to continuous time nodes. The system compares the linewidth value difference of adjacent nodes in chronological order. The calculation formula is: single-step linewidth change = measured linewidth value of the next time node - measured linewidth value of the previous time node. All single-step linewidth changes are connected in chronological order to completely restore the continuous change curve of the trench lateral linewidth from the start to the end of etching. This curve is the linewidth evolution trend, which intuitively reflects the change law of continuous contraction or expansion of linewidth throughout the etching process.

[0084] To calculate abnormal fluctuation components, the system retrieves the fixed acceptable linewidth range within the device's factory control standards. The center value of this range serves as the linewidth steady-state reference. Then, the measured linewidth value at each time point within the linewidth evolution trend is subtracted from the steady-state reference value. The calculation formula is: Single-point fluctuation component = Measured horizontal linewidth value at the node - Steady-state reference value. All single-point fluctuation components are integrated to obtain a complete set of abnormal fluctuation components. This complete set of abnormal fluctuation components can quantify the degree to which the linewidth deviates from the standard size at each etching period. For example, if the steady-state linewidth reference is 50 nanometers, and the measured linewidth at a certain moment is 56 nanometers, substituting this into the calculation yields a single-point fluctuation component = 56 - 50 = 6 nanometers, indicating a positive deviation in linewidth at that moment.

[0085] A global floating boundary is defined to generate a deviation range. The system traverses all abnormal fluctuation components, extracts the minimum value as the lower boundary of the range, and extracts the maximum value as the upper boundary. The upper and lower boundaries together form the trench lateral linewidth fluctuation deviation range. This range can intuitively reflect the floating limit of the linewidth deviation of the current batch of wafers. Subsequently, the upper and lower limits of the range are used to determine the type of trench polymer deposition imbalance and match the corresponding pumping speed compensation margin. If the distribution range of all abnormal fluctuation components is from -4 nm to +7 nm, then the lower limit of the final generated trench lateral linewidth fluctuation deviation range is -4 nm and the upper limit is 7 nm. This fully quantifies the overall deviation level of linewidth fluctuation and provides a quantitative basis for subsequent pumping speed control calculations.

[0086] The type of polymer deposition imbalance in the trench is determined based on the fluctuation range of the trench's lateral line width. The pumping speed compensation margin is then determined based on this type of imbalance, specifically including the following steps:

[0087] The deposition imbalance process attributes are obtained by matching the chamber deposition accumulation characteristics under the corresponding imbalance state according to the trench polymer deposition imbalance type.

[0088] Based on the trench lateral linewidth fluctuation deviation range corresponding to the deposition imbalance process attributes, the degree of imbalance between linewidth fluctuation and polymer deposition is matched.

[0089] The process deviation correction level is determined based on the degree of correlation imbalance, and the basic compensation range of the etching chamber exhaust pumping speed is matched according to the process deviation correction level.

[0090] By combining the process stability of high aspect ratio trench etching, the basic compensation range is modified for steady-state process adaptation to obtain the pumping speed compensation margin suitable for the current wafer batch.

[0091] Based on the identified types of polymer deposition imbalance in the trenches, the polymer accumulation distribution patterns within the chamber under the corresponding imbalance conditions are retrieved to generate deposition imbalance process attributes. Different imbalance types correspond to independent deposition distribution characteristics. The process attribute for trench sidewall deposition imbalance is characterized by an overall higher polymer thickness on the sidewalls, while the process attribute for trench bottom deposition imbalance is characterized by excessive polymer accumulation at the bottom. The system binds the deviation range values ​​with the deposition distribution morphology to form standardized process attribute records. For example, when the upper limit of the deviation range is consistently greater than 6 nanometers, the system determines it to be trench bottom deposition imbalance and matches it with the deposition imbalance process attribute of excessive bottom accumulation.

[0092] The deposition imbalance process attributes are correlated with a preset standard deviation range threshold to calculate the degree of correlation between linewidth fluctuations and polymer deposition. The calculation formula is: Correlation Imbalance Degree = Maximum Absolute Value of Current Deviation Range ÷ Standard Range Threshold of the Same Type of Imbalance. The values ​​are uniformly converted to dimensionless values ​​to eliminate scale differences caused by nanometer units. For example, if the standard range threshold for bottom deposition imbalance is 5 nanometers, and the maximum absolute value of the current deviation range is 7.5 nanometers, the correlation imbalance degree is calculated as 7.5 ÷ 5 = 1.5. This value directly reflects the severity of the current deposition deviation from the standard state.

[0093] The process deviation correction levels are determined based on the degree of correlation imbalance. Each consecutive numerical range corresponds to an independent correction level, and the base compensation range for the venting speed of the etching chamber is then determined according to the level. A correlation imbalance degree less than 1 corresponds to the first-level correction level, with a base compensation range of 2 liters per second. A correlation imbalance degree between 1 and 2 corresponds to the second-level correction level, with a base compensation range of 4 liters per second. For example, a correlation imbalance degree of 1.5 belongs to the second-level correction level, corresponding to a base compensation range of 4 liters per second.

[0094] A steady-state coefficient for the trench etching process is introduced for adaptation and correction. This steady-state coefficient is calibrated using historical stability data from the batch wafer etching process, and is calculated as follows: Pump speed compensation margin = Basic compensation range × Steady-state coefficient. A steady-state coefficient less than 1 indicates poor batch process stability, requiring a smaller compensation range to avoid over-adjustment. A steady-state coefficient greater than 1 indicates stable and controllable batch fluctuations, allowing for a moderately larger compensation range to improve correction efficiency. If the current batch's steady-state coefficient is 0.9, combined with the aforementioned basic compensation range of 4 liters per second, the pump speed compensation margin is calculated to be 4 × 0.9 = 3.6 liters per second. This value serves as the baseline parameter for exhaust flow compensation for this batch of wafers.

[0095] It achieves a complete quantitative conversion from linewidth deviation range to directly adjustable pumping speed parameters, relies on layered mapping logic to distinguish the influence weight of different deposition defects, and makes adaptation adjustments in combination with batch process stability to avoid over-correction or under-correction problems caused by fixed compensation values, and provides accurate benchmark parameters for subsequent segmented pumping speed adjustment calculations.

[0096] Based on the difference between the standard transverse linewidth of the trench and the measured transverse linewidth in the difference parameter, the segmented pumping speed adjustment required to smooth the linewidth fluctuation within a limited deposition correction time is calculated, specifically including the following steps:

[0097] The line width fluctuation correction requirement is calculated based on the pumping speed compensation margin, the line width difference between the standard transverse line width of the trench and the measured transverse line width.

[0098] Based on the linewidth fluctuation correction requirements, a time-series correction benchmark is formed by matching the preset limited deposition correction time. Based on the time-series correction benchmark, the correction weight of each exhaust process in the overall trench etching process is determined to obtain the correction ratio of each process.

[0099] The basic adjustment amplitude of each process is obtained by combining the modified ratio of each process with the pumping speed compensation margin. Based on the deposition smoothing characteristics of high aspect ratio trench etching, the basic adjustment amplitude is steadily adapted and corrected to obtain the segmented pumping speed adjustment amount.

[0100] The calculation of linewidth fluctuation correction requirements is completed collaboratively based on the pumping speed compensation margin and the linewidth difference parameter. The basic calculation formula is: Linewidth fluctuation correction requirement = Linewidth difference parameter × Unit deviation flow conversion coefficient + Pumping speed compensation margin. The original unit of the linewidth difference is nanometers, and the unit of the pumping speed compensation margin is liters per second. The conversion coefficient unifies the units of both calculations and is calibrated by the mass production process, with the unit being liters per second per nanometer. For example, if the standard transverse linewidth of the trench is 50 nanometers, and the measured value is 57 nanometers, the linewidth difference parameter = 50 - 57 = -7 nanometers, the conversion coefficient is 0.4 liters per second per nanometer, and the pumping speed compensation margin is 3.2 liters per second. Substituting these values, we get the linewidth fluctuation correction requirement = -7 × 0.4 + 3.2 = 0.4 liters per second. This value represents the total amount of exhaust flow correction that needs to be completed.

[0101] The system matches the timing correction benchmark and calculates the correction ratio for each process. It retrieves a preset, limited deposition correction duration and binds it to the linewidth fluctuation correction requirement to generate a timing correction benchmark. This benchmark constrains all correction operations to be completed within a fixed timeframe. The system distinguishes between three venting processes: main etching, passivation, deposition, cleaning, and purging. Correction weights are allocated according to the impact of each process on the linewidth, calculated as follows: Single-process correction ratio = Single-process correction weight ÷ Total weights of all processes. The total weight of the three processes is set to 10, with main etching weight at 6, passivation deposition weight at 3, and cleaning / purging weight at 1. Therefore, the correction ratio for main etching is 6 ÷ 10 = 0.6, for passivation deposition at 3 ÷ 10 = 0.3, and for cleaning / purging at 1 ÷ 10 = 0.1.

[0102] The segmented pumping speed adjustment is calculated using the formula: Basic adjustment amplitude for a single process = Linewidth fluctuation correction requirement × Corresponding process correction ratio. Then, a deposition smoothing steady-state correction coefficient is introduced to complete the numerical correction. The complete formula is: Segmented pumping speed adjustment = Basic adjustment amplitude for a single process × Deposition smoothing steady-state correction coefficient. The steady-state correction coefficient conforms to the smoothing characteristics of polymer deposition in high aspect ratio trenches. For batches with severe fluctuations, the coefficient is greater than 1; for batches with stable processes, the coefficient is less than 1. For example, with a deposition smoothing steady-state correction coefficient of 1.2, the basic adjustment amplitude for main etching is 0.4 × 0.6 = 0.24 liters per second, the segmented pumping speed adjustment for main etching is 0.24 × 1.2 = 0.288 liters per second, the segmented pumping speed adjustment for passivation deposition is 0.4 × 0.3 × 1.2 = 0.144 liters per second, and the segmented pumping speed adjustment for cleaning and purging is 0.4 × 0.1 × 1.2 = 0.048 liters per second.

[0103] The segmented pumping speed adjustment corresponds to the exhaust correction range of the three etching processes. The values ​​are uniformly measured in liters per second, which ensures that the vacuum pumping speed of each stage is adjusted synchronously within the limited deposition correction time, gradually eliminating the deviation of the trench lateral line width. The values ​​are directly involved in the calculation of the first pumping speed control coefficient, realizing differentiated pumping speed correction for the entire process.

[0104] The first pumping speed control coefficient for the current batch of wafers is calculated based on the segmented pumping speed adjustment amount, specifically including the following steps:

[0105] The process adaptability of each stage of the pumping speed adjustment is determined based on the matching characteristics between the segmented pumping speed adjustment and the polymer deposition smoothing of each segment of the etching process.

[0106] Based on the process adaptability, the segmented pumping speed adjustment is calibrated to obtain a standard adjustment amplitude that is suitable for the current etching conditions.

[0107] The process characteristics of the current batch of wafers are used to match the process characteristics with the completed conditions of the baseline process to obtain the batch process offset.

[0108] The standard adjustment amplitude is corrected by combining the batch process offset, and the first pumping speed control coefficient is obtained to match the batch etching requirements of the wafer.

[0109] To determine the process adaptability of each pumping speed adjustment segment, the system retrieves the polymer deposition and smoothing adaptability characteristics of the three processes: main etching, passivation, deposition, cleaning, and purging. The segmented pumping speed adjustment amount for each process is compared with the process's adaptability threshold. The calculation formula is: Single-segment process adaptability = Segmented pumping speed adjustment amount ÷ Process adaptability threshold. This value is a dimensionless parameter used to determine whether the current adjustment range conforms to the deposition repair pattern of that process. For example, if the adaptability threshold for the main etching process is 0.5 liters per second, corresponding to a segmented pumping speed adjustment amount of 0.3 liters per second, the process adaptability for that segment is calculated as 0.3 ÷ 0.5 = 0.6. The closer the value is to 1, the higher the matching degree between the adjustment range and the process repair characteristics.

[0110] The segmented pumping speed adjustment is calibrated based on process adaptability. The calculation formula is: Standard adjustment amplitude per segment = Segmented pumping speed adjustment × Process adaptability per segment, completing the initial process adaptability correction for the adjustment amplitude of each process. For example, if the pumping speed adjustment for the main etching segment is 0.3 liters per second and the process adaptability is 0.6, the corresponding standard adjustment amplitude per segment is 0.3 × 0.6 = 0.18 liters per second. The total standard adjustment amplitude is obtained by calculating and summing the values ​​for the three processes separately. The calculation formula is: Total standard adjustment amplitude = Main etching standard adjustment amplitude + Passivation deposition standard adjustment amplitude + Cleaning and purging standard adjustment amplitude.

[0111] To calculate the batch process offset, the system extracts the process characteristics of the current batch of wafer etching, including chamber temperature, stage rotation speed, and initial polymer thickness. It then performs a difference calculation on each of these parameters compared to the corresponding parameters of the preset baseline process condition. The weighted sum of all differences is divided by the total parameter threshold of the baseline condition to obtain the batch process offset. This dimensionless value reflects the degree to which the overall process of the current batch deviates from the standard condition. For example, if the total sum of the comprehensive parameter differences for the current batch is 12, and the total parameter threshold of the baseline condition is 20, the batch process offset is calculated as 12 ÷ 20 = 0.6.

[0112] The first pumping speed control coefficient is calculated using the following formula: First pumping speed control coefficient = Total standard adjustment amplitude × Working condition correction coefficient corresponding to batch process deviation. The working condition correction coefficient is linearly calibrated with the deviation, with a deviation of 0.6 matching a correction coefficient of 1.4. For example, if the total standard adjustment amplitude of the three processes is 0.32 liters per second, substituting the values, we can obtain the first pumping speed control coefficient = 0.32 × 1.4 = 0.448. This coefficient will participate in the subsequent chamber layer compensation adjustment, prioritizing the process requirements for smoothing the fluctuation of the groove linewidth, accurately constraining the basic correction amplitude of the pumping speed of each segment process, and avoiding secondary deviations in the groove morphology caused by excessive or insufficient adjustment amplitude.

[0113] The complete trench etching process is divided into two stages: a core process for stabilizing linewidth and an auxiliary process for tolerating linewidth. Specifically, the process includes the following steps:

[0114] Based on the adjustment amplitude according to the process standard, screen the strength of the correction effect of the pumping speed adjustment on the line width fluctuation in each pumping speed segment process, and determine the line width influence sensitivity of each pumping speed segment process.

[0115] Based on the sensitivity of linewidth influence, the polymer deposition molding effect weight of each pumping speed segment process is correlated, and the key processes for groove linewidth molding accuracy and morphological uniformity are screened out.

[0116] The critical processes with stringent requirements for linewidth control accuracy and high demand for fluctuation correction are marked as core processes for linewidth stability.

[0117] Conventional processes with minimal interference in the final linewidth forming of the groove and sufficient tolerance for process errors are designated as linewidth tolerance auxiliary processes.

[0118] The sensitivity of linewidth impact for each pumping speed segment is calculated. The system retrieves the standard adjustment amplitude for each process and compares the dimensional change in trench transverse linewidth under the same pumping speed adjustment value. The calculation formula is: Single-process linewidth impact sensitivity = Linewidth dimensional change ÷ Standard adjustment amplitude for process. This value is a dimensionless parameter; the larger the value, the stronger the corrective effect of the pumping speed change on linewidth fluctuations. For example, the standard adjustment amplitude for the main etching process is 0.5 liters per second, corresponding to a linewidth change of 5 nanometers. Substituting this into the calculation, the linewidth impact sensitivity is 5 ÷ 0.5 = 10. The standard adjustment amplitude for the cleaning and purging process is 0.5 liters per second, corresponding to a linewidth change of 0.2 nanometers. The corresponding sensitivity is 0.2 ÷ 0.5 = 0.4, which visually demonstrates the difference in the impact of the two types of processes on linewidth.

[0119] The system assigns weights to polymer deposition molding processes based on their sensitivity to linewidth influence. A fixed weight value corresponds to a preset sensitivity range; processes with a sensitivity greater than 5 are assigned a high molding weight, while those with a sensitivity less than 2 are assigned a low molding weight. For example, the main etching process, with a sensitivity of 10 (greater than 5), is assigned a molding weight of 0.6; the passivation deposition process, with a sensitivity of 7, is assigned a molding weight of 0.3; and the cleaning and purging process, with a sensitivity of 0.4 (less than 2), is assigned a molding weight of 0.1. The molding weight directly reflects the contribution of this process to the formation of the passivation film on the trench sidewalls and the determination of the trench's lateral dimensions. A higher weight value indicates a more critical impact on the trench linewidth molding accuracy and morphological uniformity. The system selects processes with higher weight values ​​as critical processes.

[0120] The core processes for linewidth stability are identified through screening. These critical processes have stringent precision control standards; even minor deviations in linewidth can lead to excessive aspect ratios in devices. They also require higher precision for process fluctuation correction and are thus uniformly designated as core processes for linewidth stability. For example, the main etching process and passivation deposition process have forming weights of 0.6 and 0.3 respectively. Both are critical processes and are therefore classified as core processes for linewidth stability. Subsequent vacuum energy consumption calculations will assign them higher weights, and more significant parameter corrections will be performed during pumping speed layer compensation.

[0121] The linewidth tolerance auxiliary processes, which have low forming weights, have minimal interference with the final trench linewidth forming. Even with slight fluctuations in pumping speed parameters, they will not cause the trench size to exceed the limit, providing ample process tolerance. Therefore, all of them are marked as linewidth tolerance auxiliary processes. The cleaning and purging process has a forming weight of only 0.1 and only affects the chamber cleaning after etching. It does not participate in the trench forming process, so it is classified as a linewidth tolerance auxiliary process. Subsequent energy consumption calculations allocate a low weight to it, and pumping speed adjustments are only made slightly to reduce the ineffective energy consumption output of the vacuum system.

[0122] After the process division is completed, the system separately calculates the vacuum energy consumption loss generated by the core process with stable linewidth and the auxiliary process with tolerance linewidth. Based on the energy consumption ratio, the second pumping speed control coefficient is solved, realizing the differentiated pumping speed control logic of prioritizing the correction of the core process and energy-saving control of the auxiliary process.

[0123] The second pumping speed control coefficient for the current batch of wafers is obtained by calculating the vacuum energy consumption of the core process for linewidth stability and the auxiliary process for linewidth tolerance. This process includes the following steps:

[0124] The process energy consumption characteristics are obtained by collecting the vacuum energy consumption loss status of the core process with stable linewidth and the auxiliary process with linewidth tolerance during operation.

[0125] The energy consumption characteristics are matched with the energy consumption baseline state of conventional standard processes. The line width forming weights of the core process for line width stability and the auxiliary process for line width tolerance are combined to determine the degree of interference of energy consumption of each process on the steady state of the etching process.

[0126] Based on the degree of interference, the energy consumption correction weights are preset for the core process and the auxiliary process respectively. Combined with the preset process energy consumption stability standard and energy consumption correction weights, the energy consumption loss characteristics of the process are balanced and adapted to obtain the energy consumption balance benchmark.

[0127] Based on the energy consumption balance benchmark, a second pumping speed control coefficient is generated to match the etching conditions of the current batch of wafers.

[0128] The vacuum energy consumption status of the core process for linewidth stabilization and the auxiliary process for linewidth tolerance is collected throughout the entire operation. Two types of values ​​are extracted for each process: real-time power of the vacuum unit and process duration, combined to form the process energy consumption characteristics. Power is measured in watts (W) and duration in seconds (s). After standardizing the conversion of units, the total energy consumption of a single process is calculated using the formula: Total energy consumption of a single process = Real-time power of the vacuum unit × Process duration. For example, the main etching core process has a power of 450 watts and a duration of 120 seconds, so the total energy consumption of this process is 450 × 120 = 54,000 watt-seconds. The cleaning and purging auxiliary process has a power of 100 watts and a duration of 60 seconds, so the total energy consumption is 100 × 60 = 6,000 watt-seconds. The energy consumption characteristics of both types of processes are fully recorded for subsequent comparison.

[0129] The standard process energy consumption benchmark is matched and the degree of energy consumption interference is determined. The benchmark energy consumption value corresponding to the conventional standard process is retrieved, and the calculation formula is: Process energy consumption deviation value = Total energy consumption of a single process - Benchmark energy consumption value of the corresponding process. Then, the process linewidth forming weight is introduced for weighted conversion, and the calculation formula is: Degree of energy consumption interference = Process energy consumption deviation value ÷ Benchmark energy consumption value × Linewidth forming weight. This dimensionless value represents the intensity of energy consumption fluctuation on the etching steady state. The main etching forming weight is 0.6, and the benchmark energy consumption is 50,000 watt-seconds. Substituting these values, the degree of energy consumption interference is obtained as (54,000 - 50,000) ÷ 50,000 × 0.6 = 0.048. The cleaning and purging forming weight is 0.1, and the benchmark energy consumption is 6,500 watt-seconds. The degree of energy consumption interference is obtained as (6,000 - 6,500) ÷ 6,500 × 0.1 ≈ -0.0077. The degree of interference caused by the energy consumption fluctuation of the core process is significantly higher than that of the auxiliary process.

[0130] Energy consumption correction weights are assigned and the energy consumption balance benchmark is calculated. The higher the degree of energy consumption interference, the larger the energy consumption correction weight is assigned. The correction weight for core processes ranges from 0.6 to 0.9, and the correction weight for auxiliary processes ranges from 0.1 to 0.4. Combined with the preset process energy consumption stability standard, the calculation formula is: Energy Consumption Balance Benchmark = Total Benchmark Energy Consumption of All Processes ÷ (Energy Consumption Correction Weight of Core Processes + Energy Consumption Correction Weight of Auxiliary Processes). For example, if the correction weight for core processes is 0.8, the correction weight for auxiliary processes is 0.2, and the total benchmark energy consumption is 56,500 watt-seconds, the energy consumption balance benchmark = 56,500 ÷ (0.8 + 0.2) = 56,500 watt-seconds. This value is the balance reference standard for batch energy consumption control.

[0131] Based on the energy consumption balance benchmark and the current wafer etching conditions, a second pumping speed control coefficient is calculated. The formula is: Second pumping speed control coefficient = Energy consumption balance benchmark ÷ Total measured energy consumption of all processes in the current batch. The total measured energy consumption of the current batch equals the total energy consumption of the core process plus the total energy consumption of the auxiliary process. For example, the total measured energy consumption = 54000 + 6000 = 60000 watt-seconds. Substituting this into the calculation, the second pumping speed control coefficient is 56500 ÷ 60000 ≈ 0.9417. This coefficient is coupled with the first pumping speed control coefficient to adjust the pumping speed of each segment in a layered manner. Sufficient pumping speed correction range is reserved for core processes that have a significant impact on linewidth, while reducing excess vacuum energy consumption output of auxiliary processes, thus balancing trench dimensional accuracy and mass production energy-saving requirements.

[0132] An etching process optimization system for high aspect ratio trench-type memory ICs includes:

[0133] Data acquisition module: Collects polymer deposition thickness data inside the chamber, wafer stage rotation speed gradient data, trench lateral linewidth fluctuation detection data, and exhaust pumping speed segment adjustment data to form a multi-source dataset;

[0134] Judgment module: When the trench aspect ratio fluctuation exceeds the device factory control threshold, the module performs time-series calculations on the multi-source dataset to generate the trench lateral linewidth fluctuation deviation range. Based on the trench lateral linewidth fluctuation deviation range, the module determines the trench polymer deposition imbalance type and determines the pumping speed compensation margin based on the trench polymer deposition imbalance type.

[0135] Calculation Module 1: Based on the difference between the standard transverse linewidth of the trench and the measured transverse linewidth in the difference parameter, calculate the segmented pumping speed adjustment amount required to smooth the linewidth fluctuation within the limited deposition correction time, and calculate the first pumping speed control coefficient of the current batch of wafers based on the segmented pumping speed adjustment amount.

[0136] Calculation Module 2: Divide the pumping speed segment of the complete trench etching process into a linewidth stabilization core process and a linewidth tolerance auxiliary process, and calculate the vacuum energy consumption of the linewidth stabilization core process and the linewidth tolerance auxiliary process to obtain the second pumping speed control coefficient of the current batch of wafers.

[0137] Execution module: Performs stratified compensation adjustment on the segmented exhaust pumping speed of the chamber according to the first pumping speed control coefficient and the second pumping speed control coefficient.

[0138] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0139] 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 of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing etching process of high aspect ratio trench type memory IC, characterized in that, The method includes the following specific steps: A multi-source dataset is formed by collecting data on polymer deposition thickness inside the chamber, wafer stage rotation speed gradient, trench lateral linewidth fluctuation detection, and exhaust pumping speed segment adjustment. When the trench aspect ratio fluctuation exceeds the device factory control threshold, time-series calculations are performed on the multi-source dataset to generate the trench lateral linewidth fluctuation deviation range. The trench polymer deposition imbalance type is determined based on the trench lateral linewidth fluctuation deviation range, and the pumping speed compensation margin is determined based on the trench polymer deposition imbalance type. Based on the difference between the standard transverse linewidth of the trench and the measured transverse linewidth in the difference parameter, the segmented pumping speed adjustment amount required to smooth the linewidth fluctuation within the limited deposition correction time is calculated, and the first pumping speed control coefficient of the current batch of wafers is calculated based on the segmented pumping speed adjustment amount. The complete trench etching process is divided into two stages: the core process for stabilizing linewidth and the auxiliary process for tolerating linewidth. The vacuum energy consumption of the core process for stabilizing linewidth and the auxiliary process for tolerating linewidth is calculated to obtain the second pumping speed control coefficient for the current batch of wafers. The stratified compensation adjustment process is performed on the segmented exhaust pumping speed of the chamber based on the first pumping speed control coefficient and the second pumping speed control coefficient.

2. The etching process optimization method for a high aspect ratio trench-type memory IC according to claim 1, characterized in that, The exhaust pumping speed segment adjustment data includes the main etching pumping speed level, the passivation deposition pumping speed level, and the cleaning and purging pumping speed level.

3. The etching process optimization method for a high aspect ratio trench-type memory IC according to claim 2, characterized in that, The pumping speed compensation margin is the exhaust flow compensation parameter corresponding to the difference between the standard transverse line width of the trench and the measured transverse line width.

4. The etching process optimization method for a high aspect ratio trench-type memory IC according to claim 3, characterized in that, When the trench aspect ratio fluctuation exceeds the device's factory control threshold, time-series calculations are performed on the multi-source dataset to generate the trench lateral linewidth fluctuation deviation range. This includes the following steps: Based on the preset multi-source dataset, abnormal data caused by etching process start-up and shutdown and external interference are removed to obtain regularized time-series data; Based on regular time series data, the continuous change trend of the trench lateral linewidth with the etching process is determined to obtain the linewidth evolution trend. The linewidth steady-state reference is obtained by matching the acceptable range of linewidth values ​​according to the device's factory control standards; the abnormal fluctuation component is obtained by calibrating the linewidth evolution trend with the linewidth steady-state reference. The range of values ​​for the horizontal linewidth fluctuation deviation of the trench is generated by defining the value range based on the global floating boundary of the abnormal fluctuation component.

5. The etching process optimization method for a high aspect ratio trench-type memory IC according to claim 4, characterized in that, The type of polymer deposition imbalance in the trench is determined based on the fluctuation range of the trench's lateral line width. The pumping speed compensation margin is then determined based on this type of imbalance, specifically including the following steps: The deposition imbalance process attributes are obtained by matching the chamber deposition accumulation characteristics under the corresponding imbalance state according to the trench polymer deposition imbalance type. Based on the trench lateral linewidth fluctuation deviation range corresponding to the deposition imbalance process attributes, the degree of imbalance between linewidth fluctuation and polymer deposition is matched. The process deviation correction level is determined based on the degree of correlation imbalance, and the basic compensation range of the etching chamber exhaust pumping speed is matched according to the process deviation correction level. By combining the process stability of high aspect ratio trench etching, the basic compensation range is modified for steady-state process adaptation to obtain the pumping speed compensation margin suitable for the current wafer batch.

6. The etching process optimization method for a high aspect ratio trench-type memory IC according to claim 5, characterized in that, Based on the difference between the standard transverse linewidth of the trench and the measured transverse linewidth in the difference parameter, the segmented pumping speed adjustment required to smooth the linewidth fluctuation within a limited deposition correction time is calculated, specifically including the following steps: The line width fluctuation correction requirement is calculated based on the pumping speed compensation margin, the line width difference between the standard transverse line width of the trench and the measured transverse line width. Based on the linewidth fluctuation correction requirements, a time-series correction benchmark is formed by matching the preset limited deposition correction time. Based on the time-series correction benchmark, the correction weight of each exhaust process in the overall trench etching process is determined to obtain the correction ratio of each process. The basic adjustment amplitude of each process is obtained by combining the modified ratio of each process with the pumping speed compensation margin. Based on the deposition smoothing characteristics of high aspect ratio trench etching, the basic adjustment amplitude is steadily adapted and corrected to obtain the segmented pumping speed adjustment amount.

7. The etching process optimization method for a high aspect ratio trench-type memory IC according to claim 6, characterized in that, The first pumping speed control coefficient for the current batch of wafers is calculated based on the segmented pumping speed adjustment amount, specifically including the following steps: The process adaptability of each stage of the pumping speed adjustment is determined based on the matching characteristics between the segmented pumping speed adjustment and the polymer deposition smoothing of each segment of the etching process. Based on the process adaptability, the segmented pumping speed adjustment is calibrated to obtain a standard adjustment amplitude that is suitable for the current etching conditions. The process characteristics of the current batch of wafers are used to match the process characteristics with the completed conditions of the baseline process to obtain the batch process offset. The standard adjustment amplitude is corrected by combining the batch process offset, and the first pumping speed control coefficient is obtained to match the batch etching requirements of the wafer.

8. The etching process optimization method for a high aspect ratio trench-type memory IC according to claim 7, characterized in that, The complete trench etching process is divided into two stages: a core process for stabilizing linewidth and an auxiliary process for tolerating linewidth. Specifically, the process includes the following steps: Based on the adjustment amplitude according to the process standard, screen the strength of the correction effect of the pumping speed adjustment on the line width fluctuation in each pumping speed segment process, and determine the line width influence sensitivity of each pumping speed segment process. Based on the sensitivity of linewidth influence, the polymer deposition molding effect weight of each pumping speed segment process is correlated, and the key processes for groove linewidth molding accuracy and morphological uniformity are screened out. The critical processes with stringent requirements for linewidth control accuracy and high demand for fluctuation correction are marked as core processes for linewidth stability. Conventional processes with minimal interference in the final linewidth forming of the groove and sufficient tolerance for process errors are designated as linewidth tolerance auxiliary processes.

9. The etching process optimization method for a high aspect ratio trench-type memory IC according to claim 8, characterized in that, The second pumping speed control coefficient for the current batch of wafers is obtained by calculating the vacuum energy consumption of the core process for linewidth stability and the auxiliary process for linewidth tolerance. This process includes the following steps: The process energy consumption characteristics are obtained by collecting the vacuum energy consumption loss status of the core process with stable linewidth and the auxiliary process with linewidth tolerance during operation. The energy consumption characteristics are matched with the energy consumption baseline state of conventional standard processes. The line width forming weights of the core process for line width stability and the auxiliary process for line width tolerance are combined to determine the degree of interference of energy consumption of each process on the steady state of the etching process. Based on the degree of interference, the energy consumption correction weights are preset for the core process and the auxiliary process respectively. Combined with the preset process energy consumption stability standard and energy consumption correction weights, the energy consumption loss characteristics of the process are balanced and adapted to obtain the energy consumption balance benchmark. Based on the energy consumption balance benchmark, a second pumping speed control coefficient is generated to match the etching conditions of the current batch of wafers.

10. An etching process optimization system for a high aspect ratio trench-type memory IC, applied to the etching process optimization method for a high aspect ratio trench-type memory IC as described in claims 1-9, characterized in that, include: Data acquisition module: Collects polymer deposition thickness data inside the chamber, wafer stage rotation speed gradient data, trench lateral linewidth fluctuation detection data, and exhaust pumping speed segment adjustment data to form a multi-source dataset; Judgment module: When the trench aspect ratio fluctuation exceeds the device factory control threshold, the module performs time-series calculations on the multi-source dataset to generate the trench lateral linewidth fluctuation deviation range. Based on the trench lateral linewidth fluctuation deviation range, the module determines the trench polymer deposition imbalance type and determines the pumping speed compensation margin based on the trench polymer deposition imbalance type. Calculation Module 1: Based on the difference between the standard transverse linewidth of the trench and the measured transverse linewidth in the difference parameter, calculate the segmented pumping speed adjustment amount required to smooth the linewidth fluctuation within the limited deposition correction time, and calculate the first pumping speed control coefficient of the current batch of wafers based on the segmented pumping speed adjustment amount. Calculation Module 2: Divide the pumping speed segment of the complete trench etching process into a linewidth stabilization core process and a linewidth tolerance auxiliary process, and calculate the vacuum energy consumption of the linewidth stabilization core process and the linewidth tolerance auxiliary process to obtain the second pumping speed control coefficient of the current batch of wafers. Execution module: Performs stratified compensation adjustment on the segmented exhaust pumping speed of the chamber according to the first pumping speed control coefficient and the second pumping speed control coefficient.

Images