Sampling frequency calculation method, system, device and storage medium

Abstract

The application provides a spectrum sampling frequency calculation method, system, device and storage medium, relates to the semiconductor detection technical field, and the reference parameter and the time window set of the key event obtained from the historical knowledge base provide the prior basis for the configuration sampling frequency; the real-time adaptive factor is generated based on the collected spectrum data in each sampling period, which can reflect the instantaneous change of the current working condition; the target sampling frequency is configured by combining the timer time, the time window set matching result of the key event, the reference parameter and the real-time adaptive factor, the sampling frequency can be adaptively configured based on the key event, the on-demand allocation of the sampling resource is realized; the target sampling frequency is used for the next sampling period, a continuous closed-loop updating mechanism is formed, so that the optimal sampling strategy is dynamically generated without manual continuous intervention, and finally, the overall efficiency of data acquisition is greatly improved, and the reliable monitoring of the process core quality forming process is ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of semiconductor detection, and in particular to a method, system and device for calculating the sampling frequency of a spectrometer and a storage medium. BACKGROUND

[0002] In the epitaxial growth process of semiconductors, key parameters such as film thickness, composition and interface quality have a decisive influence on device performance. In order to realize real-time monitoring and process traceability of the epitaxial growth process, in the industry, in-situ optical monitoring means are usually introduced into the reaction cavity. Among them, in-situ monitoring based on reflectance spectroscopy is widely used in epitaxial process development and production process control due to its non-contact, fast response and rich information. In practical applications, the spectrometer needs to continuously collect reflectance spectrum data during epitaxial growth to support subsequent feature extraction, process determination and anomaly diagnosis.

[0003] In the traditional technology, common sampling strategies include fixed frequency sampling and adaptive sampling based on simple empirical rules. Fixed frequency sampling is simple to implement, but it is difficult to take into account the characteristics of significant changes in spectra between different process steps, and the problem of coexistence of data redundancy in non-critical stages and insufficient sampling resolution in critical stages often occurs. Although adaptive sampling based on empirical rules can switch the sampling frequency according to the process steps to some extent, the rules usually rely on manual presetting and static solidification, and have limited adaptability when facing formula iteration, equipment state drift or abnormal working conditions, and the maintenance cost is relatively high. SUMMARY

[0004] The purpose of the present application is to provide a method, system, device and storage medium for calculating the sampling frequency of a spectrometer to overcome the defect that the sampling frequency of the spectrometer in the traditional technology lacks adaptive scheduling capability for different processes.

[0005] In a first aspect, the present application provides a method for calculating the sampling frequency of a spectrometer, which is applicable to a semiconductor epitaxial growth detection device, the semiconductor epitaxial growth detection device comprising a spectrometer; the method comprising:

[0006] in response to a process change signal, initializing a timer and determining a current process;

[0007] based on the current process, querying the corresponding reference parameters and the time window set of key events from a pre-set historical knowledge base;

[0008] for each sampling period during the duration of the current process, obtaining the spectral data sampled by the spectrometer based on the current sampling frequency, and generating a real-time adaptation factor representing the degree of spectral anomaly based on the spectral data;

[0009] matching a timer time corresponding to a current sampling period with the time window set, and configuring a target sampling frequency according to a matching result, the reference parameter and the real-time adaptation factor;

[0010] using the target sampling frequency as a sampling frequency of a next sampling period of the optical spectrometer.

[0011] In one of the embodiments, the reference parameter includes a reference statistic and a reference sampling frequency.

[0012] The preset historical knowledge base includes the following steps:

[0013] acquiring historical detection data, the historical detection data including at least one of original reflectance spectrum time series data, process time series logs, wafer endpoint detection data and device electrical test data;

[0014] extracting key data from the historical detection data, the key data including process sequences and corresponding spectrum feature curves meeting a screening condition;

[0015] identifying key events in the key data to obtain a time window set and a stable time interval of the key events;

[0016] calculating a change rate sample set of the spectrum feature curves changing with time under each process sequence, and performing statistics on the change rate sample set to obtain a reference statistic;

[0017] performing statistics on historical sampling frequencies in the stable time interval to obtain a reference sampling frequency;

[0018] storing the reference sampling frequency, the reference statistic and the time window set of the key events in a preset data structure to obtain a historical knowledge base.

[0019] In one of the embodiments, the identifying the key events in the key data to obtain a time window set and a stable time interval of the key events includes:

[0020] performing time series alignment and pattern clustering operations on the key data to obtain a plurality of change templates corresponding to each process sequence; analyzing each of the change templates to identify key events therein to obtain a time window set of the key events corresponding to each change template; and a time interval outside the time window set is recorded as a stable time interval.

[0021] In one of the embodiments, the performing time series alignment and pattern clustering operations on the key data to obtain a plurality of change templates corresponding to each process sequence includes:

[0022] acquiring spectrum feature curves of different batches under a same process sequence;

[0023] performing non-linear time axis alignment on the spectrum feature curves by using a dynamic time warping algorithm to obtain a feature curve set;

[0024] performing clustering analysis on the feature curve set to divide the feature curve set into a preset number of pattern categories; for each pattern category, performing representative extraction on the feature curves belonging to the pattern category to generate a corresponding typical change pattern template, thereby obtaining a plurality of change templates.

[0025] In one embodiment, the change templates are analyzed to identify key events therein, and a time window set of key events corresponding to each change template is obtained, including:

[0026] performing numerical differentiation processing on each change template to obtain a change rate sequence and a change acceleration sequence representing the change of the change template over time within a process;

[0027] detecting local extreme points in the change rate sequence to determine key moments when the change in the change template is fastest or slowest;

[0028] detecting turning points where the sign changes in the change acceleration sequence to determine key moments when the change trend in the change template turns;

[0029] taking the key moments corresponding to the local extreme points and the key moments corresponding to the turning points as key event points; and expanding a corresponding key event time window in a preset direction with each key event point as the center, thereby obtaining a time window set of key events corresponding to each change template.

[0030] In one embodiment, the reference statistics include an arithmetic mean and a standard deviation;

[0031] The calculation of a change rate sample set of the spectrum feature curve over time under each process series, and the statistics of the change rate sample set to obtain reference statistics, include:

[0032] obtaining spectrum feature curves and their time information of different batches under the same process sequence;

[0033] Based on the spectrum feature difference value of adjacent sampling points and the sampling time interval, the spectrum feature change rate at each sampling point is calculated to form a change rate sample set under the process sequence;

[0034] statistical analysis of the change rate sample set is performed to calculate an arithmetic mean representing the normal change degree of the process sequence and a standard deviation representing the normal fluctuation range of the process sequence.

[0035] In one embodiment, the reference parameters include reference statistics;

[0036] The acquisition of the spectrometer based on the current sampling frequency sampling spectrum data, and based on the spectrum data generated to characterize the degree of real-time adaptation factor of spectral anomalies, including:

[0037] Acquisition of the spectrometer based on the current sampling frequency twice sampling spectrum data;

[0038] Extracting the feature wavelength value of two consecutive spectral data and its corresponding sampling time interval, and calculating the instantaneous change rate based on the feature wavelength value and the sampling time interval;

[0039] Based on two consecutive spectral data, the correlation of the two on the pre-set key waveband is calculated to obtain the stability index representing the stability of the spectrum shape; wherein the pre-set key waveband contains the stop band edge waveband;

[0040] Based on the deviation of the instantaneous change rate relative to the reference statistical quantity, and the spectral shape instability degree represented by the stability index, the real-time adaptation factor is generated, which is used to measure the degree of change rate deviation from the historical normal and the degree of decline of the spectrum shape stability.

[0041] In one of the embodiments, the real-time adaptation factor is generated based on the deviation of the instantaneous change rate relative to the reference statistical quantity, and the spectral shape instability degree represented by the stability index, including:

[0042] The expression of the real-time adaptation factor is calculated as:

[0043]

[0044]

[0045]

[0046] wherein, is the real-time adaptation factor; is the instantaneous change rate; and are the arithmetic mean and the standard deviation, respectively; is the stability index; is the Pearson correlation coefficient; is the key waveband; is the feature wavelength value; is the sampling time interval.

[0047] In one of the embodiments, the reference parameters include pre-set parameters, reference sampling frequency and reference statistical quantity;

[0048] The timer time corresponding to the current sampling period is matched with the time window set, and a target sampling frequency is configured according to a matching result, the reference parameter and the real-time adaptation factor, and the method comprises the following steps:

[0049] In the case where the timer time corresponding to the current sampling period falls within the time window set, a first frequency is determined based on the preset parameter, the reference statistical quantity and the real-time adaptation factor, otherwise a second frequency is determined based on the preset parameter, the reference sampling frequency and the real-time adaptation factor, the first frequency being higher than the second frequency;

[0050] The first frequency or the second frequency is taken as the target sampling frequency.

[0051] In one of the embodiments, the preset parameter comprises a key frequency and a key gain;

[0052] The first frequency is determined based on the preset parameter, the reference statistical quantity and the real-time adaptation factor, and the expression is as follows:

[0053]

[0054] wherein, is the key frequency; is the key gain.

[0055] In one of the embodiments, the preset parameter comprises a minimum threshold frequency and a stationary phase gain;

[0056] The second frequency is determined based on the preset parameter, the reference sampling frequency and the real-time adaptation factor, and the expression is as follows:

[0057]

[0058] wherein, is the reference sampling frequency; is the minimum threshold frequency, is the stationary phase gain.

[0059] In a second aspect, the present application provides a system for calculating the sampling frequency of a spectrometer, which is arranged in a semiconductor epitaxial growth detection device, and the semiconductor epitaxial growth detection device comprises a spectrometer; the system comprises:

[0060] A process interpreter is configured to acquire a process change signal published by an epitaxial equipment master in real time, initialize a timer and determine a current process in response to the process change signal, and based on the current process, query corresponding reference parameters and a time window set of key events from a preset historical knowledge base;

[0061] a real-time sensing adapter configured to acquire spectral data sampled by the spectrometer based on a current sampling frequency for each sampling period during a current process duration, and generate a real-time adaptation factor representing a degree of spectral abnormality based on the spectral data;

[0062] a decision maker configured to match a timer time corresponding to a current sampling period with the set of time windows, and configure a target sampling frequency according to a matching result, the reference parameter and the real-time adaptation factor; and generate a sampling instruction with the target sampling frequency as a sampling frequency of a next sampling period of the spectrometer;

[0063] an executor configured to generate a timer interrupt signal to trigger the spectrometer to perform a sampling action in response to the sampling instruction.

[0064] In a third aspect, the present application also provides a computer device. The computer device comprises a memory and a processor, the memory stores a computer program, and the processor implements the method steps in the first aspect when executing the computer program.

[0065] In a fourth aspect, the present application also provides a computer readable storage medium. The computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method steps in the first aspect.

[0066] The above-mentioned method, system, device and storage medium for calculating a sampling frequency of a spectrometer have at least the following advantages:

[0067] The present application queries corresponding reference parameters and a set of time windows of key events from a preset historical knowledge base based on a current process, to provide priori basis for subsequent configuration of a sampling frequency; in each sampling period during a current process duration, a real-time adaptation factor representing a degree of spectral abnormality is generated based on acquired spectral data, so that a sampling decision can reflect immediate changes of a current working condition; further, a timer time is matched with the set of time windows of key events, and a target sampling frequency is configured in combination with a matching result, reference parameters and the real-time adaptation factor, so that the sampling frequency can be adaptively configured based on key events, to realize on-demand allocation of sampling resources; finally, the target sampling frequency is used for a next sampling period, to form a continuous closed-loop updating mechanism, so that an optimal sampling strategy is dynamically generated without continuous human intervention, to finally realize a substantial improvement in overall efficiency of data acquisition while ensuring reliable monitoring of a core quality formation process of a process. BRIEF DESCRIPTION OF DRAWINGS

[0068] Figure 1 a flowchart of a method for calculating a sampling frequency of a spectrometer in one embodiment;

[0069] Figure 2A flowchart of a process for constructing a historical knowledge base in one embodiment;

[0070] Figure 3 A flowchart of a process for generating real-time adaptation factors in one embodiment;

[0071] Figure 4 A block diagram of a computing system for calculating a spectrometer sampling frequency in one embodiment;

[0072] Figure 5 An internal block diagram of a computer device in one embodiment. DETAILED DESCRIPTION

[0073] The present application can be implemented or applied in other different specific embodiments, and the details in the specification can be modified or changed based on different views and applications without departing from the spirit of the present application. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0074] Some exemplary embodiments of the present application are described for illustrative purposes, and it should be understood that the present application can be implemented in other ways not specifically shown in the drawings.

[0075] Referring to Figure 1 In one embodiment, the present application provides a method for calculating a spectrometer sampling frequency, which is applicable to a semiconductor epitaxial growth detection device including a spectrometer. The method specifically includes the following steps:

[0076] Step 102, in response to a process change signal, initializing a timer and determining a current process.

[0077] Specifically, an epitaxial growth process is usually composed of a plurality of process steps in sequence, each process step corresponding to different control conditions and exhibiting different spectral variation characteristics. The process change signal is an event signal output by a master control system of a semiconductor epitaxial device when a process step is switched during recipe execution, indicating that the epitaxial growth process enters a new process phase. Typically, the process change signal carries information such as step code (Step code), step number (Step ID), etc.

[0078] The timer is used to time the current process execution time. After detecting the process change signal, the timer is reset or the current time is taken as the starting time to obtain the starting reference time of the process step, and then the relative time within the process can be obtained during the process duration.

[0079] Step 104, based on the current process, query the corresponding reference parameters and the time window set of key events from the preset historical knowledge base.

[0080] Specifically, the historical knowledge base is used to store the spectrum behavior prior information corresponding to each process step in epitaxial growth, so as to quickly query the reference parameters and the time window set of key events according to the current process in the online stage. In the embodiment, the historical knowledge base is constructed based on historical detection data, wherein the historical detection data includes at least one of original reflectance spectrum time series data, process time sequence log, wafer endpoint detection data (EPD), and device electrical test data.

[0081] The above historical detection data is packaged as structured data packet according to batch, and the time stamp is strictly aligned.

[0082] Step 106, for each sampling period during the current process duration, obtain the spectrum data sampled by the spectrometer based on the current sampling frequency, and generate a real-time adaptation factor representing the degree of spectrum anomaly based on the spectrum data.

[0083] Specifically, a process step usually lasts for several seconds to several minutes, and the embodiment sets multiple sampling periods for each process step, each sampling period corresponding to one sampling frequency update and at least one spectrum collection. At the starting time of the current process, the sampling frequency of the previous process can be continued, or a preset sampling frequency can be used. The spectrometer samples the spectrum data based on the frequency, and performs feature extraction on the spectrum data to obtain a feature result representing the current process state. Based on the feature result, the deviation degree of the current spectrum dynamic relative to the historical normal is comprehensively evaluated, and a real-time adaptation factor is generated.

[0084] Step 107, match the timer time corresponding to the current sampling period and the time window set, and configure the target sampling frequency according to the matching result, the reference parameters and the real-time adaptation factor; the target sampling frequency is used as the sampling frequency of the next sampling period of the spectrometer.

[0085] Specifically, during the process duration, every time a sampling period comes, the current value of the timer is read as the relative time within the step corresponding to the sampling period. And match the relative time and the time window set of key events, if the relative time falls within the start and end range of any key event time window, it is considered that the current sampling period is in the key event stage; if it does not fall into any time window, it is considered that the current sampling period is in the stable stage. Further, according to the stage where the current sampling period is located, the corresponding target sampling frequency is configured.

[0086] The method for calculating the sampling frequency of the above spectrometer, based on the current process, queries the corresponding reference parameters and the time window set of key events from the preset historical knowledge base, providing prior basis for subsequent sampling frequency configuration; in each sampling period during the current process duration, based on the collected spectral data, a real-time adaptation factor representing the degree of spectral anomaly is generated, enabling the sampling decision to reflect the immediate changes of the current working condition; further, the timer time and the time window set of key events are matched, and the target sampling frequency is configured combining the matching result, the reference parameters and the real-time adaptation factor, enabling adaptive configuration of the sampling frequency based on key events, realizing on-demand allocation of sampling resources; finally, the target sampling frequency is used for the next sampling period, forming a continuous closed-loop updating mechanism, so as to dynamically generate the optimal sampling strategy without continuous human intervention, finally realizing substantial improvement of the overall efficiency of data acquisition while ensuring reliable monitoring of the core quality formation process of the process.

[0087] Please refer to Figure 2 Optionally, in the case that the reference parameters include reference statistics and reference sampling frequency, the manner of the preset historical knowledge base includes:

[0088] Step 202, historical detection data is acquired.

[0089] Step 204, key data is extracted from the historical detection data, the key data including process sequences meeting the screening conditions and corresponding spectral feature curves.

[0090] Step 206, key events in the key data are identified, obtaining a time window set of the key events and a stable time interval.

[0091] Step 208, a change rate sample set of the spectral feature curve changing with time under each process series is calculated, and the change rate sample set is statistically processed to obtain reference statistics.

[0092] Step 210, historical sampling frequencies in the stable time interval are statistically processed to obtain reference sampling frequencies.

[0093] Step 212, the reference sampling frequencies, the reference statistics and the time window set of the key events are stored according to a preset data structure to obtain a historical knowledge base.

[0094] Specifically, by setting the screening conditions, data packets meeting the screening conditions are screened out from the historical detection data for analysis, so as to identify the key event segments in each process step and summarize them into one or more time window sets of key events; meanwhile, in the non-key event segments, the sampling behavior and the spectral change characteristics are statistically processed, so as to learn the reference parameters for online scheduling. The screening conditions include at least one of meeting a preset yield threshold, EPD detection passing and key electrical indicators being qualified.

[0095] Further, the spectral feature curve is a time-varying feature sequence extracted from the original reflectance spectrum, and usually includes a stopband edge feature wavelength, a peak-valley position, or a fusion feature wavelength.

[0096] Optionally, the key events in the key data are identified to obtain a time window set and a stationary time interval of the key events, including:

[0097] The time series alignment and pattern clustering operations are performed on the key data to obtain a plurality of change templates corresponding to each process sequence; each change template is analyzed to identify key events therein to obtain a time window set of the key events corresponding to each change template; and a time interval outside the time window set is recorded as a stationary time interval.

[0098] Specifically, since there are slight differences in actual growth rates and response delays of different batches under the same process, the same physical process occurs in advance or lag in different batches, so that the spectral feature curves of each batch exist in the stretching effect on the time axis. Based on this, the time series alignment is performed on the above key data, so that the curve segments with the same change form correspond to the same relative position as much as possible after alignment.

[0099] The purpose of the clustering operation is to track different change forms that may exist under the same process, so as to obtain the change trajectory caused by the difference in device state, the difference in material batch, or the fine tuning of the formula.

[0100] Optionally, the time series alignment and pattern clustering operations are performed on the key data to obtain a plurality of change templates corresponding to each process sequence, including:

[0101] The spectral feature curves of different batches under the same process sequence are obtained; and a dynamic time warping algorithm is used to perform nonlinear time axis alignment on each spectral feature curve to obtain a feature curve set.

[0102] The clustering analysis is performed on the feature curve set to divide the feature curve set into a preset number of pattern categories; for each pattern category, the representative extraction is performed on the feature curves belonging to the pattern category to generate a corresponding typical change pattern template, so as to obtain a plurality of change templates.

[0103] Optionally, each change template is analyzed to identify key events therein to obtain a time window set of the key events corresponding to each change template, including:

[0104] Each change template is subjected to numerical differentiation processing to obtain a change rate sequence and a change acceleration sequence representing the change of the change template with respect to the time within the process.

[0105] Detecting local extreme points in the rate of change sequence to determine the key time when the change reaches the fastest or slowest in the change template.

[0106] Detecting turning points where the sign of the change acceleration sequence changes to determine the key time when the change trend turns in the change template.

[0107] Taking the key time corresponding to the local extreme point and the key time corresponding to the turning point as the key event point, and expanding the corresponding key event time window according to the preset direction to obtain the time window set of the key event corresponding to each change template.

[0108] Specifically, the present application analyzes each change template to identify the key event and form the time window set of the key event corresponding to the change template for each process step corresponding to the change template set, which is used for subsequent online sampling frequency configuration.

[0109] Calculating the first derivative of each change template to obtain the rate of change sequence, which is used to represent the degree of change of the change template over time. Calculating the second derivative of each change template to obtain the change acceleration sequence, which is used to represent the increasing or decreasing trend of the rate of change over time, thereby reflecting whether the change process is in an acceleration, deceleration or trend conversion state.

[0110] Further, to improve robustness, the present application can also perform smoothing processing on the rate of change sequence before calculating the change acceleration.

[0111] Further, the preset direction includes forward, backward, or forward and backward directions. For example, expanding according to the preset direction is to expand a preset time length forward and a preset time length backward from the key event point, thereby forming a time window covering the influence range of the key event. In another embodiment, expanding according to the preset direction can only expand forward or only expand backward to cover the premonitory segment before the key event occurs or the response segment after the key event occurs. Accordingly, the time outside the key event time window in a process step is the stable time interval.

[0112] Optionally, in the case where the reference statistics include the arithmetic mean and the standard deviation, calculating the rate of change sample set of the spectral feature curve over time under each process series, and performing statistics on the rate of change sample set to obtain the reference statistics, including:

[0113] Obtaining the spectral feature curves and their time information of different batches under the same process sequence.

[0114] Based on the spectral feature difference value of adjacent sampling points and the sampling time interval, calculating the spectral feature rate of change at each sampling point to form the rate of change sample set under the process sequence.

[0115] The statistical analysis is performed on the sample set of change rates to calculate an arithmetic mean value for representing the normal change degree of the process sequence and a standard deviation for representing the normal fluctuation range of the process sequence.

[0116] Specifically, the application first determines the identification information of the target process sequence, filters a plurality of historical batch data corresponding to the identification information from a historical database, and extracts the spectral feature curve and time information of each batch in the process sequence. The spectral feature curve is a sequence of features extracted from the original reflectance spectrum that changes over time; the time information includes the time stamp or relative time in the step corresponding to each sampling point, which is used to represent the time interval between adjacent sampling points.

[0117] For each historical batch, the spectral feature curve of the batch in the target process sequence is traversed in the order of sampling time, and adjacent sampling points are recorded as a group to calculate the change rate sample until all sampling points in the batch are completed, obtaining a sample set of change rates.

[0118] Further, all change rate samples in the sample set of change rates are added one by one to obtain a cumulative sum; then the cumulative sum is divided by the total number of samples to obtain an arithmetic mean value.

[0119] Further, the deviation between each change rate sample and the arithmetic mean value is calculated; the squares of the deviations are taken and added to obtain a sum of squared deviations; then the sum of squared deviations is divided by the total number of samples to obtain a variance; and finally the square root of the variance is obtained to obtain a standard deviation.

[0120] The reference sampling frequency is a recommended sampling frequency learned autonomously from historical detection data that meets the filtering conditions. Specifically, the effective sampling frequency in the stationary time interval is statistically or optimally calculated to obtain a frequency value that can track normal process changes and reduce redundant data in the non-critical stage, and the frequency is stored in the historical knowledge base as the reference sampling frequency of the step, which is used as a reference for sampling frequency configuration in the online stage.

[0121] Please refer to Figure 3 Optionally, in the case where the reference parameter further includes a reference statistical quantity, the spectral data sampled by the spectrometer based on the current sampling frequency is obtained, and a real-time adaptation factor representing the degree of spectral anomaly is generated based on the spectral data, including:

[0122] Step 302, obtaining spectral data sampled by the spectrometer based on the current sampling frequency twice in succession.

[0123] Step 304, extracting the characteristic wavelength values of the two continuous spectral data and their corresponding sampling time intervals, and calculating the instantaneous change rate based on the characteristic wavelength values and the sampling time intervals.

[0124] In step 306, based on the two continuous spectrum data, the correlation of the two in the preset key band is calculated to obtain a stability index representing the stability of the spectrum shape; wherein the preset key band includes the stop band edge band.

[0125] In step 308, based on the deviation degree of the instantaneous change rate relative to the reference statistical quantity and the spectrum shape instability degree represented by the stability index, a real-time adaptation factor is generated for simultaneously measuring the degree of deviation of the change rate from the historical normal and the degree of decline of the spectrum shape stability.

[0126] Optionally, generating the real-time adaptation factor based on the deviation degree of the instantaneous change rate relative to the reference statistical quantity and the spectrum shape instability degree represented by the stability index comprises:

[0127] The expression for calculating the real-time adaptation factor is:

[0128]

[0129]

[0130]

[0131] wherein, is the real-time adaptation factor; is the instantaneous change rate; and are the arithmetic mean and the standard deviation, respectively; is the stability index; is the Pearson correlation coefficient; is the key band; is the characteristic wavelength value; is the sampling time interval.

[0132] Optionally, in the case where the reference parameters include preset parameters, a reference sampling frequency and a reference statistical quantity, the timer time corresponding to the current sampling period and the time window set are matched, and the target sampling frequency is configured according to the matching result, the reference parameters and the real-time adaptation factor, comprising:

[0133] In the case where the timer time corresponding to the current sampling period falls within the time window set, a first frequency is determined based on the preset parameters, the reference statistical quantity and the real-time adaptation factor, otherwise a second frequency is determined based on the preset parameters, the reference sampling frequency and the real-time adaptation factor, the first frequency being higher than the second frequency; the first frequency or the second frequency is taken as the target sampling frequency.

[0134] Optionally, in the case where the preset parameters include a key frequency and a key gain, a first frequency is determined based on the preset parameters, the reference statistical quantity and the real-time adaptation factor, and the expression thereof is:

[0135]

[0136] wherein, is the critical frequency; is the critical gain.

[0137] Optionally, in the case that the preset parameters include the minimum threshold frequency and the stationary phase gain, the second frequency is determined based on the preset parameters, the reference sampling frequency and the real-time adaptation factor, and the expression is:

[0138]

[0139] wherein, is the reference sampling frequency; is the minimum threshold frequency, is the stationary phase gain.

[0140] The above method for calculating the sampling frequency of the optical spectrum analyzer, based on the current process, queries the corresponding reference parameters and the time window set of the critical event from the preset historical knowledge base, to provide prior basis for subsequent sampling frequency configuration; in each sampling period during the current process duration, the real-time adaptation factor representing the degree of spectral anomaly is generated based on the collected spectral data, so that the sampling decision can reflect the immediate changes of the current working condition; further, the timer time and the time window set of the critical event are matched, and the target sampling frequency is configured based on the matching result, the reference parameters and the real-time adaptation factor, so that the sampling frequency can be adaptively configured based on the critical event, and the on-demand allocation of sampling resources can be realized; finally, the target sampling frequency is used for the next sampling period, forming a continuous closed-loop updating mechanism, so that the optimal sampling strategy can be dynamically generated without continuous manual intervention, and finally the overall efficiency of data acquisition can be greatly improved while ensuring the reliable monitoring of the process core quality formation process.

[0141] It should be understood that, although each step in the flowchart involved in each embodiment as described above is displayed in sequence according to the arrow, these steps are not necessarily executed in sequence according to the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other sequences. Moreover, at least part of the steps in the flowchart involved in each embodiment as described above can include multiple steps or stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence of these steps or stages is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or steps or stages in other steps.

[0142] Based on the same inventive concept, the application further provides a system for calculating the sampling frequency of a spectrometer, which is applicable to the method for calculating the sampling frequency of a spectrometer. The system provides a solution to the problem in a similar manner to the solution provided in the method, and therefore the specific limitations in one or more apparatus embodiments provided below can be found in the limitations of the method provided above, which will not be described here again.

[0143] Please refer to Figure 4 In one embodiment, the system for calculating the sampling frequency of a spectrometer is arranged in a semiconductor epitaxial growth detection device, and the semiconductor epitaxial growth detection device includes a spectrometer. The system for calculating the sampling frequency of a spectrometer includes a process interpreter, a real-time perception adapter, a decision maker, and an executor.

[0144] The process interpreter is configured to acquire a process change signal published in real time by an epitaxial device master controller, initialize a timer and determine a current process in response to the process change signal, and query corresponding reference parameters and a time window set of key events from a preset historical knowledge base based on the current process.

[0145] The real-time perception adapter is configured to acquire spectral data sampled by the spectrometer based on a current sampling frequency for each sampling period during the duration of the current process, and generate a real-time adaptation factor representing the degree of spectral anomaly based on the spectral data.

[0146] The decision maker is configured to match the timer time corresponding to the current sampling period and the time window set, and configure a target sampling frequency according to the matching result, the reference parameters, and the real-time adaptation factor; take the target sampling frequency as the sampling frequency of the next sampling period of the spectrometer, and generate a sampling instruction.

[0147] The executor is configured to generate a timer interrupt signal in response to the sampling instruction to trigger the spectrometer to perform a collection action.

[0148] Optionally, the system for calculating the sampling frequency of a spectrometer further includes a database construction module.

[0149] The database construction module is configured to construct a preset historical knowledge base.

[0150] Optionally, in the case that the reference parameters include a reference statistical quantity and a reference sampling frequency, the database construction module constructs the preset historical knowledge base in the following manner: obtaining historical detection data; extracting key data from the historical detection data, the key data including process sequences and corresponding spectral feature curves that meet the screening conditions; identifying key events in the key data to obtain a time window set and a stable time interval of the key events; calculating a change rate sample set of the spectral feature curves under each process sequence with respect to time, and performing statistics on the change rate sample set to obtain the reference statistical quantity; performing statistics on historical sampling frequencies in the stable time interval to obtain the reference sampling frequency; and storing the reference sampling frequency, the reference statistical quantity, and the time window set of the key events according to a preset data structure to obtain the historical knowledge base.

[0151] Optionally, the database construction module identifies the key events in the key data to obtain a time window set and a stable time interval of the key events, including: performing time sequence alignment and pattern clustering operations on the key data to obtain a plurality of change templates corresponding to each process sequence; analyzing the change templates to identify the key events therein to obtain a time window set of the key events corresponding to each change template; and time intervals outside the time window set being recorded as the stable time interval.

[0152] Optionally, the database construction module performs time sequence alignment and pattern clustering operations on the key data to obtain a plurality of change templates corresponding to each process sequence, including: obtaining spectral feature curves of different batches under the same process sequence; performing nonlinear time axis alignment on the spectral feature curves by using a dynamic time warping algorithm to obtain a feature curve set; performing clustering analysis on the feature curve set to divide the feature curve set into a preset number of pattern categories; and for each pattern category, extracting a representative feature curve belonging to the pattern category to generate a corresponding typical change pattern template, thereby obtaining the plurality of change templates.

[0153] Optionally, the database construction module analyzes the change templates to identify the key events therein to obtain a time window set of the key events corresponding to each change template, including: performing numerical differentiation processing on each change template to obtain a change rate sequence and a change acceleration sequence representing the change of the change template with respect to time within the process; detecting local extreme points in the change rate sequence to determine key time points at which the change in the change template reaches the fastest or slowest; detecting turning points at which the sign changes in the change acceleration sequence to determine key time points at which the change trend in the change template changes; taking the key time points corresponding to the local extreme points and the key time points corresponding to the turning points as key event points; and expanding the key event points in a preset direction to form corresponding key event time windows, thereby obtaining the time window set of the key events corresponding to each change template.

[0154] Optionally, the database construction module, in the case that the reference statistics include an arithmetic mean and a standard deviation, calculates a sample set of variation rates of the spectral feature curve over time under each process sequence, and performs statistics on the sample set of variation rates to obtain the reference statistics, including: obtaining spectral feature curves and time information thereof of different batches under the same process sequence; calculating variation rates of spectral features at each sampling point based on spectral feature differences of adjacent sampling points and sampling time intervals, to form a sample set of variation rates under the process sequence; and performing statistical analysis on the sample set of variation rates to calculate an arithmetic mean for representing a normal variation degree of the process sequence, and a standard deviation for representing a normal fluctuation range of the process sequence.

[0155] Optionally, the real-time awareness adapter obtains spectral data sampled by the spectrometer based on the current sampling frequency, and generates a real-time adaptation factor representing a spectral abnormality degree based on the spectral data, including: obtaining spectral data sampled by the spectrometer based on the current sampling frequency for two consecutive times; extracting feature wavelength values and corresponding sampling time intervals of the two consecutive spectral data, and calculating instantaneous variation rates based on the feature wavelength values and the sampling time intervals; calculating correlations of the two consecutive spectral data on preset key wavebands to obtain stability indexes representing spectral shape stability, based on the two consecutive spectral data; wherein the preset key wavebands include stopband edge wavebands; and generating the real-time adaptation factor based on a deviation degree of the instantaneous variation rates relative to the reference statistics, and a spectral shape instability degree represented by the stability indexes, to measure a degree of deviation of the variation rates from a historical normal state and a degree of decline of spectral shape stability.

[0156] Optionally, the real-time awareness adapter generates the real-time adaptation factor based on a deviation degree of the instantaneous variation rates relative to the reference statistics, and a spectral shape instability degree represented by the stability indexes, including: calculating an expression of the real-time adaptation factor as:

[0157]

[0158]

[0159]

[0160] wherein, is the real-time adaptation factor; is the instantaneous variation rate; and are an arithmetic mean and a standard deviation, respectively; is the stability index; is a Pearson correlation coefficient; is a key waveband; is a feature wavelength value; is a sampling time interval.

[0161] Optionally, in the case that the reference parameters include the preset parameters, the reference sampling frequency and the reference statistics, the decision maker matches the timer time corresponding to the current sampling period and the time window set, and configures the target sampling frequency according to the matching result, the reference parameters and the real-time adaptation factor, including: in the case that the timer time corresponding to the current sampling period falls into the time window set, determining a first frequency based on the preset parameters, the reference statistics and the real-time adaptation factor, otherwise determining a second frequency based on the preset parameters, the reference sampling frequency and the real-time adaptation factor, the first frequency being higher than the second frequency; taking the first frequency or the second frequency as the target sampling frequency.

[0162] Optionally, in the case that the preset parameters include the key frequency and the key gain, the decision maker determines the first frequency based on the preset parameters, the reference statistics and the real-time adaptation factor, and the expression is:

[0163]

[0164] wherein, is the key frequency; is the key gain.

[0165] Optionally, in the case that the preset parameters include the minimum threshold frequency and the stationary phase gain, the decision maker determines the second frequency based on the preset parameters, the reference sampling frequency and the real-time adaptation factor, and the expression is:

[0166]

[0167] wherein, is the reference sampling frequency; is the minimum threshold frequency, is the stationary phase gain.

[0168] The above-mentioned spectrum sampling frequency calculation system queries the corresponding reference parameters and the time window set of the key event from the preset historical knowledge base based on the current process, providing prior basis for subsequent sampling frequency configuration; in each sampling period during the duration of the current process, the real-time adaptation factor representing the degree of spectral anomaly is generated based on the collected spectral data, so that the sampling decision can reflect the immediate changes of the current working condition; further, the timer time and the time window set of the key event are matched, and the target sampling frequency is configured in combination with the matching result, the reference parameters and the real-time adaptation factor, so that the sampling frequency can be adaptively configured based on the key event, and the on-demand allocation of sampling resources is realized; finally, the target sampling frequency is used for the next sampling period, forming a continuous closed-loop updating mechanism, so that the optimal sampling strategy is dynamically generated without the need for continuous human intervention, and finally the overall efficiency of data acquisition is greatly improved while ensuring the reliable monitoring of the core quality formation process of the process.

[0169] The modules in the above spectrum sampling frequency calculation system can be implemented by software, hardware, or a combination thereof. The modules can be embedded in a processor in a computer device or independent of the processor, or stored in a memory in the computer device to be invoked by the processor to perform operations corresponding to the modules.

[0170] Optionally, in an embodiment, the application further provides a semiconductor epitaxial growth detection device, comprising the spectrum sampling frequency calculation system and the spectrum provided in the above embodiments.

[0171] Specifically, the spectrum sampling frequency calculation system in the embodiment has the same structure and working principle as those in the above embodiments, and thus will not be described here for brevity.

[0172] The spectrum is configured to perform a collection action based on the timer interrupt signal output by the spectrum sampling frequency calculation system.

[0173] With the above scheme, the semiconductor epitaxial growth detection device calculates a real-time adaptation factor from the real-time collected spectrum data in each sampling period during the process duration, dynamically determines the sampling frequency of the next sampling period in combination with the reference parameter and the key event time window of the current step, and enables the spectrum to automatically increase the sampling density in the key event stage to completely capture the rapidly changing details, and automatically reduce the sampling density in the stable stage to reduce redundant data, so as to finally greatly improve the overall efficiency of data collection while ensuring reliable monitoring of the process core quality formation process.

[0174] In a feasible embodiment, a computer device is provided, which can be a terminal, and an internal structure diagram of the computer device can be as shown in Figure 5The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus. The communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to perform wired or wireless communication with external terminals. The wireless communication can be achieved through WIFI, mobile cellular network, NFC (Near Field Communication) or other technologies. The computer program is executed by the processor to implement the above-mentioned spectrum sampler frequency calculation method. The display unit of the computer device is configured to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0175] Those skilled in the art can understand that, Figure 5 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0176] In a feasible embodiment, a computer device is provided, which includes a memory and a processor. The memory stores a computer program. The processor executes the computer program to implement the method steps in the above-mentioned spectrum sampler frequency calculation method.

[0177] In a feasible embodiment, a computer readable storage medium is provided, which stores a computer program. The computer program is executed by a processor to implement the method steps in the above-mentioned spectrum sampler frequency calculation method.

[0178] In a feasible embodiment, a computer program product is provided, which includes a computer program. The computer program is executed by a processor to implement the method steps in the above-mentioned spectrum sampler frequency calculation method.

[0179] Any technical features in the above embodiments can be combined, and for the sake of brevity, not all possible combinations are described above, however, as long as the combinations of technical features do not have contradictions, they shall be considered within the scope of the present disclosure.

[0180] The above embodiments only express several implementation manners of the present application, and the description is relatively specific and detailed, but it shall not be understood as a limitation on the patent scope of the present application. It shall be pointed out that, for ordinary skilled persons in the art, several modifications and improvements can be made without departing from the concept of the present application, and these shall be within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. A method of calculating the sampling frequency of a spectrometer, characterized in that, The application is suitable for a semiconductor epitaxial growth detection device, which comprises a spectrometer; the method comprises: In response to a process change signal, initializing a timer and determining a current process; Based on the current process, querying corresponding reference parameters and a time window set of key events from a preset historical knowledge base; For each sampling period during the duration of the current process, obtaining spectral data sampled by the spectrometer based on a current sampling frequency, and generating a real-time adaptation factor representing the degree of spectral anomaly based on the spectral data; Matching the timer time corresponding to the current sampling period and the time window set, and configuring a target sampling frequency according to the matching result, the reference parameters and the real-time adaptation factor; Taking the target sampling frequency as the sampling frequency of the next sampling period of the spectrometer.

2. The method of claim 1, wherein, The reference parameters include reference statistics and reference sampling frequency; The manner of the preset historical knowledge base comprises: Obtaining historical detection data, which includes at least one of original reflectance spectrum time series data, process time series logs, wafer endpoint detection data and device electrical test data; Extracting key data from the historical detection data, which includes process sequences and corresponding spectral feature curves that meet the screening conditions; Identifying key events in the key data to obtain a time window set of key events and a stable time interval; Calculating a change rate sample set of the spectral feature curve changing with time under each process series, and statistically analyzing the change rate sample set to obtain reference statistics; Statistically analyzing historical sampling frequencies in the stable time interval to obtain reference sampling frequencies; Storing the reference sampling frequencies, reference statistics and time window set of key events in a preset data structure to obtain a historical knowledge base.

3. The method of claim 2, wherein, The identification of key events in the key data to obtain a time window set of key events and a stable time interval comprises: Performing time series alignment and pattern clustering operations on the key data to obtain a plurality of change templates corresponding to each process sequence; Analyzing each change template to identify key events therein to obtain a time window set of key events corresponding to each change template; time intervals outside the time window set are recorded as stable time intervals.

4. The method of claim 3, wherein, The time series alignment and pattern clustering operations on the key data to obtain a plurality of change templates corresponding to each process sequence comprise: Obtaining spectral feature curves of different batches under the same process sequence; Using a dynamic time warping algorithm to perform nonlinear time axis alignment on each spectral feature curve to obtain a feature curve set; Performing clustering analysis on the feature curve set to divide the feature curve set into a preset number of pattern categories; for each pattern category, extracting a representative feature curve belonging to the pattern category to generate a corresponding typical change pattern template, thereby obtaining a plurality of change templates.

5. The method of claim 3, wherein, The analysis of each change template to identify key events therein to obtain a time window set of key events corresponding to each change template comprises: Carrying out numerical differential processing on each of the change templates to obtain a change rate sequence and a change acceleration sequence representing changes of the change template over time within the process; Detecting local extreme points in the change rate sequence to determine key moments at which changes in the change template reach the fastest or slowest; Detecting turning points at which signs change in the change acceleration sequence to determine key moments at which change trends in the change template turn; Taking the key moments corresponding to the local extreme points and the key moments corresponding to the turning points as key event points, and expanding the key event time windows corresponding to the key event points in a preset direction to obtain a set of key event time windows corresponding to each of the change templates.

6. The method of claim 2, wherein, The reference statistics include an arithmetic mean and a standard deviation; The calculation of the change rate sample set of the spectral feature curve over time under each process series and the statistics of the change rate sample set to obtain the reference statistics include: Obtaining spectral feature curves and time information of different batches under the same process sequence; Based on the spectral feature difference of adjacent sampling points and the sampling time interval, calculating the spectral feature change rate at each sampling point to form a change rate sample set under the process sequence; Statistical analysis of the change rate sample set is carried out to calculate the arithmetic mean for representing the normal change degree of the process sequence and the standard deviation for representing the normal fluctuation range of the process sequence.

7. The method of claim 1, wherein, The reference parameter includes reference statistics; The acquisition of the spectral data sampled by the spectrometer based on the current sampling frequency and the generation of the real-time adaptation factor representing the spectral abnormality degree based on the spectral data include: Obtaining spectral data sampled by the spectrometer based on the current sampling frequency twice in succession; Extracting the characteristic wavelength values and their corresponding sampling time intervals of the two continuous spectral data, and calculating the instantaneous change rate based on the characteristic wavelength values and the sampling time intervals; Based on the two continuous spectral data, calculating the correlation of the two on a preset key wavelength band to obtain a stability index representing the spectral shape stability; wherein the preset key wavelength band includes a stop band edge wavelength band; Based on the deviation of the instantaneous change rate from the reference statistics and the spectral shape instability degree represented by the stability index, the real-time adaptation factor is generated to measure the degree of deviation of the change rate from the historical normal and the degree of decline of the spectral shape stability.

8. The method of claim 7, wherein, The generation of the real-time adaptation factor based on the deviation of the instantaneous change rate from the reference statistics and the spectral shape instability degree represented by the stability index includes: The expression for calculating the real-time adaptation factor is: wherein, is the real-time adaptation factor; is the instantaneous rate of change; and are the arithmetic mean and the standard deviation, respectively; is the stability index; is the Pearson correlation coefficient; is the key wavelength band; is the characteristic wavelength value; is the sampling time interval.

9. The method of claim 1, wherein, The reference parameter includes a preset parameter, a reference sampling frequency and reference statistics; The matching of the timer time corresponding to the current sampling period and the time window set, and the configuration of the target sampling frequency according to the matching result, the reference parameter and the real-time adaptation factor include: In a case where a timer time corresponding to a current sampling period falls into the set of time windows, a first frequency is determined based on the preset parameter, the reference statistical quantity and the real-time adaptation factor, otherwise a second frequency is determined based on the preset parameter, the reference sampling frequency and the real-time adaptation factor, the first frequency being higher than the second frequency; The first frequency or the second frequency is taken as the target sampling frequency.

10. The method of claim 9, wherein, The preset parameter comprises a key frequency and a key gain; The first frequency is determined based on the preset parameter, the reference statistical quantity and the real-time adaptation factor, and an expression thereof is: wherein is a key frequency; is a key gain.

11. The method of claim 9, wherein, The preset parameter comprises a minimum threshold frequency and a stationary phase gain; The second frequency is determined based on the preset parameter, the reference sampling frequency and the real-time adaptation factor, and an expression thereof is: wherein, is a reference sampling frequency; is a minimum threshold frequency, is a plateau gain.

12. A system for calculating the sampling frequency of a spectrometer, characterized in that, The system is arranged in a semiconductor epitaxial growth detection device, the semiconductor epitaxial growth detection device comprising a spectrometer; the system comprises: A process interpreter is configured to acquire a process change signal published by an epitaxial device master in real time, initialize a timer and determine a current process in response to the process change signal, and based on the current process, query a corresponding reference parameter and a set of time windows of key events from a preset historical knowledge base; A real-time perception adapter is configured to, for each sampling period during a current process duration, acquire spectrum data sampled by the spectrometer based on a current sampling frequency, and generate a real-time adaptation factor representing a degree of spectrum anomaly based on the spectrum data; A decision maker is configured to match a timer time corresponding to a current sampling period with the set of time windows, and configure a target sampling frequency according to a matching result, the reference parameter and the real-time adaptation factor; take the target sampling frequency as a sampling frequency of a next sampling period of the spectrometer, and generate a sampling instruction; An executor is configured to generate a timer interrupt signal in response to the sampling instruction, so as to trigger the spectrometer to perform a collection action.

13. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that, The processor executes the computer program to realize the steps of the method of any one of claims 1-11.

14. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the method of any one of claims 1-11.

Images