CN113539909A - Fault detection method and device, terminal equipment and storage medium - Google Patents
Fault detection method and device, terminal equipment and storage medium Download PDFInfo
- Publication number
- CN113539909A CN113539909A CN202111077510.5A CN202111077510A CN113539909A CN 113539909 A CN113539909 A CN 113539909A CN 202111077510 A CN202111077510 A CN 202111077510A CN 113539909 A CN113539909 A CN 113539909A
- Authority
- CN
- China
- Prior art keywords
- manufacturing process
- signal
- standard
- fault detection
- semiconductor wafer
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L21/00—Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
- H01L21/67—Apparatus specially adapted for handling semiconductor or electric solid state devices during manufacture or treatment thereof; Apparatus specially adapted for handling wafers during manufacture or treatment of semiconductor or electric solid state devices or components ; Apparatus not specifically provided for elsewhere
- H01L21/67005—Apparatus not specifically provided for elsewhere
- H01L21/67242—Apparatus for monitoring, sorting or marking
- H01L21/67253—Process monitoring, e.g. flow or thickness monitoring
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Condensed Matter Physics & Semiconductors (AREA)
- General Physics & Mathematics (AREA)
- Manufacturing & Machinery (AREA)
- Computer Hardware Design (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Power Engineering (AREA)
- Testing Or Measuring Of Semiconductors Or The Like (AREA)
Abstract
The invention discloses a fault detection method, a fault detection device, terminal equipment and a storage medium, which are used for detecting faults in the manufacturing process of a semiconductor wafer, and the method comprises the following steps: acquiring monitoring signal data in the manufacturing process of a semiconductor wafer; performing feature extraction on the monitoring signal data to obtain a target generalized signal in the manufacturing process of the semiconductor wafer; and detecting whether the manufacturing process of the semiconductor wafer has faults or not according to the target summarizing signal. The invention can realize full-automatic feature extraction on the signal data acquired in the semiconductor wafer manufacturing process, reduce the dependence on expert experience in fault detection, and can keep the key features of the monitoring signal data in the semiconductor manufacturing process through the extracted target summary signal, thereby avoiding the loss of feature information and improving the fault detection precision in the semiconductor manufacturing process.
Description
Technical Field
The present invention relates to the field of semiconductor manufacturing technologies, and in particular, to a fault detection method and apparatus, a terminal device, and a storage medium.
Background
With the continuous development of sensors and information technologies, most modern semiconductor industries monitor the production process and production equipment by advanced process control to determine whether a fault exists in the production process of a semiconductor, and then preliminarily determine whether the manufactured semiconductor is qualified. However, due to the complexity and dynamics of the semiconductor manufacturing process, the existing fault detection method is too dependent on expert experience during modeling, and manual processing and feature extraction of collected data are required, so that the method is time-consuming and extremely high in cost. Moreover, the conventional feature extraction method based on expert experience inevitably causes information loss, thereby affecting the detection precision of the model.
Disclosure of Invention
The invention mainly aims to provide a fault detection method, a fault detection device, terminal equipment and a storage medium, and aims to solve the technical problems that the detection fault detection precision is influenced due to the fact that the characteristics are lost when the fault detection method in the semiconductor manufacturing process is manually extracted.
In addition, in order to achieve the above object, the present invention further provides a fault detection method for detecting a fault in a semiconductor wafer manufacturing process, the fault detection method comprising the steps of:
acquiring monitoring signal data in the manufacturing process of a semiconductor wafer;
performing feature extraction on the monitoring signal data to obtain a target generalized signal in the manufacturing process of the semiconductor wafer;
and detecting whether the manufacturing process of the semiconductor wafer has faults or not according to the target summarizing signal.
Optionally, the step of extracting features of the monitoring signal data to obtain a target summary signal in the semiconductor wafer manufacturing process includes:
calculating the dissimilarity between every two sequence values in the time sequence to obtain a dissimilarity matrix of the time sequence;
establishing an optimization matrix of the time series based on the dissimilarity matrix, wherein elements in the optimization matrix are used for determining whether one sequence value can represent the other sequence value between two corresponding sequence values according to the dissimilarity;
and calculating a cost function of the time sequence according to the optimization matrix, and performing feature extraction on the time sequence by using the cost function to obtain a target generalized signal in the manufacturing process of the semiconductor wafer.
Optionally, the step of performing feature extraction on the time series by using the cost function to obtain a target summary signal in the semiconductor wafer manufacturing process includes:
calculating an objective function for extracting the characteristics of the time series by using the cost function;
and carrying out optimization solution on the objective function according to the constraint conditions of the objective function to obtain the objective characteristic points of the time sequence, wherein the signals corresponding to the objective characteristic points are objective generalized signals in the semiconductor wafer manufacturing process, the constraint conditions of the objective function comprise causal constraints, and the causal constraints are that a first sequence value in the time sequence can only be represented by a second sequence value before the time point.
Optionally, the monitoring signal data is a time series, and before the step of performing feature extraction on the monitoring signal data, the method further includes:
acquiring standard signal data in the manufacturing process of a semiconductor standard wafer;
and performing dynamic time warping on the monitoring signal data based on the standard signal data, and aligning the time step between the monitoring signal data and the standard signal data.
Optionally, the step of detecting whether there is a fault in the manufacturing process of the semiconductor wafer according to the target summary signal includes:
acquiring a standard generalized signal of the standard wafer, wherein the standard generalized signal is obtained by performing feature extraction on the standard signal data;
and comparing the target summarized signal with the standard summarized signal to detect whether the manufacturing process of the semiconductor wafer has faults or not.
Optionally, the fault detection method further includes:
and when a characteristic design instruction is detected, modifying the characteristic points of the standard generalized signal according to the characteristic design instruction, updating the standard generalized signal according to the modified characteristic points, or generating a new version of the standard generalized signal.
Optionally, the fault detection method further includes:
calculating the importance index of each standard characteristic point in the standard generalized signal;
and screening the standard characteristic points according to the importance indexes of the standard characteristic points, and updating the standard generalized signals based on the screened standard characteristic points.
In addition, to achieve the above object, the present invention also provides a failure detection device, including:
the data acquisition module is used for acquiring monitoring signal data in the manufacturing process of the semiconductor wafer;
the signal summarization module is used for carrying out feature extraction on the monitoring signal data to obtain a target summarization signal in the manufacturing process of the semiconductor wafer;
and the fault detection module is used for detecting whether a fault exists in the manufacturing process of the semiconductor wafer according to the target summary signal.
Further, to achieve the above object, the present invention also provides an apparatus comprising: a memory, a processor and a program stored on the memory and executable on the processor, the program, when executed by the processor, implementing the steps of the method as described above.
Furthermore, to achieve the above object, the present invention also provides a computer-readable storage medium having a program stored thereon, which when executed by a processor, implements the steps of the method as described above.
Furthermore, to achieve the above object, the present invention also provides a computer program product comprising a computer program which, when being executed by a processor, realizes the steps of the method as described above.
Compared with the prior art, the fault detection method, the fault detection device, the terminal equipment and the storage medium provided by the embodiment of the invention have the advantages that monitoring signal data in the manufacturing process of the semiconductor wafer are acquired; performing feature extraction on the monitoring signal data to obtain a target generalized signal in the manufacturing process of the semiconductor wafer; and detecting whether the manufacturing process of the semiconductor wafer has faults or not according to the target summarizing signal. The method can realize full-automatic feature extraction, reduce the dependence on expert experience in fault detection, and can keep the key features of monitoring signal data in the semiconductor manufacturing process through the extracted target signal summarization, thereby avoiding the loss of feature information and improving the fault detection precision in the semiconductor manufacturing process.
Drawings
Fig. 1 is a schematic hardware structure diagram of an implementation manner of a terminal device according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart of a first embodiment of the fault detection method of the present invention;
fig. 3 is a functional block diagram of an embodiment of the fault detection apparatus of the present invention.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
In the following description, suffixes such as "module", "component", or "unit" used to denote elements are used only for facilitating the explanation of the present invention, and have no specific meaning in itself. Thus, "module", "component" or "unit" may be used mixedly.
The fault detection terminal (also called terminal, equipment or terminal equipment) in the embodiment of the invention can be a PC (personal computer), and can also be mobile terminal equipment with data processing and display functions, such as a smart phone, a tablet computer, a portable computer and the like.
As shown in fig. 1, the terminal may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, a communication bus 1002. Wherein a communication bus 1002 is used to enable connective communication between these components. The user interface 1003 may include a Display screen (Display), an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface). The memory 1005 may be a high-speed RAM memory or a non-volatile memory (e.g., a magnetic disk memory). The memory 1005 may alternatively be a storage device separate from the processor 1001.
Optionally, the terminal may further include a camera, a Radio Frequency (RF) circuit, a sensor, an audio circuit, a WiFi module, and the like. Such as light sensors, motion sensors, and other sensors. Specifically, the light sensor may include an ambient light sensor that may adjust the brightness of the display screen according to the brightness of ambient light, and a proximity sensor that may turn off the display screen and/or the backlight when the mobile terminal is moved to the ear. As one of the motion sensors, the gravity acceleration sensor can detect the magnitude of acceleration in each direction (generally, three axes), detect the magnitude and direction of gravity when the mobile terminal is stationary, and can be used for applications (such as horizontal and vertical screen switching, related games, magnetometer attitude calibration), vibration recognition related functions (such as pedometer and tapping) and the like for recognizing the attitude of the mobile terminal; of course, the mobile terminal may also be configured with other sensors such as a gyroscope, a barometer, a hygrometer, a thermometer, and an infrared sensor, which are not described herein again.
Those skilled in the art will appreciate that the terminal structure shown in fig. 1 is not intended to be limiting and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components.
As shown in fig. 1, a memory 1005, which is a kind of computer-readable storage medium, may include therein an operating system, a network communication module, a user interface module, and a computer program for failure detection.
In the terminal shown in fig. 1, the network interface 1004 is mainly used for connecting to a backend server and performing data communication with the backend server; the user interface 1003 is mainly used for connecting a client (user side) and performing data communication with the client; and the processor 1001 may be configured to invoke a fault detection program stored in the memory 1005 that, when executed by the processor, performs operations in the fault detection method provided by the embodiments described below.
Based on the hardware structure of the device, the invention provides various embodiments of the fault detection method.
In various embodiments of the present invention, key technical terms are used including:
wafer: refers to a chip used for manufacturing semiconductor circuits, and the manufacturing process of the wafer generally includes chemical vapor deposition and chemical mechanical polishing.
Signal summarization: a feature extraction mode is based on a tracking signal abstraction method, and the contour of a complete signal is fitted through a small number of selected key points.
It should be noted that in the semiconductor manufacturing process, there may be mechanical damage to the wafer used for manufacturing the semiconductor, and the reason for the mechanical damage is generally that the surface of the wafer is in an arc or discontinuous point distribution due to mechanical scratches caused by polishing or slicing. The mechanical damage is large or small, and usually affects the connectivity of the wafer circuit, which is a serious defect. However, such defects caused by improper operation of machines and the like can be corrected, and based on this, advanced process control theory has emerged as a defect solution for semiconductor manufacturing, which includes fault detection for monitoring and analyzing changes in production equipment or production process data, thereby detecting the presence of faults in the semiconductor manufacturing process and analyzing potential causes of faults.
With the development of sensing technology and IoT (Internet of Things), fault detection technology in semiconductor manufacturing process is developing towards intellectualization and digitization, and feature extraction is very important for collected mass data. Currently, common feature extraction methods include: extracting statistical characteristics such as mean value, standard deviation, extreme value and the like from the acquired original data, wherein the information loss is large, and the robustness of the detection model is difficult to ensure; extracting features from original data based on a dimensionality reduction algorithm in machine learning, such as principal component analysis, linear discriminant analysis and the like, wherein the dimensionality reduction method is based on the principle of maximum variance, information loss is easily caused, meanwhile, overfitting of a model is possibly caused, and partial algorithm parameters need to be manually defined; signal segmentation based on signal patterns and feature extraction algorithms are highly complex, requiring the number of segments and corresponding signal patterns to be determined with the aid of expert knowledge and manual operations. However, the above feature extraction methods all inevitably cause information loss, and some methods also require certain manual operations and combine with corresponding professional knowledge, which not only affects the accuracy of fault detection, but also affects the efficiency of feature extraction.
In order to solve the problems, the invention provides a fault detection method for a semiconductor wafer manufacturing process, which can realize automatic extraction of key features by summarizing acquired original data, thereby improving the feature extraction efficiency and ensuring the fault detection precision.
Specifically, referring to fig. 2, fig. 2 is a schematic flow chart of a first embodiment of the fault detection method of the present invention, in the first embodiment of the fault detection method of the present invention, the fault detection method includes:
step S10, acquiring monitoring signal data in the manufacturing process of the semiconductor wafer;
the fault detection method is used for detecting the semiconductor manufacturing process to determine whether faults exist in the semiconductor manufacturing process, and the semiconductor manufacturing process generally comprises the steps of carrying out chemical vapor deposition and chemical mechanical polishing on a wafer, wherein if faults exist in the wafer manufacturing process, the manufactured semiconductor is likely to have mechanical damage or other defects, so that unqualified products are produced. In the embodiment, a wafer manufacturing process is monitored by using a sensor technology and/or an IoT technology, monitoring signal data in the semiconductor wafer manufacturing process is collected, the monitoring signal data includes information such as temperature and pressure in the wafer manufacturing process, and whether a fault exists in the wafer manufacturing process is determined by monitoring data of various signals in the wafer manufacturing process. The acquisition of the monitoring signal data may be tracking and collecting the signal value of each monitoring signal in the wafer manufacturing process, or may be collecting every preset time interval, and the collected monitoring signal data is a time sequence. For example, if the total time required for the manufacturing process of one wafer is about 100 seconds, the signal values of the monitoring signals in the manufacturing process of the wafer are collected every 1 second, and after the manufacturing of the wafer is completed, the time series with the series length of 100 corresponding to each monitoring signal is obtained.
Step S20, performing feature extraction on the monitoring signal data to obtain a target generalized signal in the manufacturing process of the semiconductor wafer;
after the monitoring signal data in the manufacturing process of the semiconductor wafer is obtained, the obtained monitoring signal data is subjected to feature extraction, so that a target generalized signal of the monitoring signal data is extracted. The feature extraction of the monitoring signal data may specifically be to perform signal summarization on the monitoring signal data, and the signal summarization is described as an example below. The signal summarization is specifically to fit the monitoring signal data to obtain a fitted curve of the monitoring signal data, select some feature points from the curve, summarize the complete fitted curve of the monitoring signal data by using as few feature points as possible, and then describe the selected feature points by using a model or a function, wherein the model or the function is the target summarized signal, and the selected feature points are the key feature information of the monitoring signal data. For the manufacturing process of the semiconductor wafer, the amount of the collected data is huge, and if whether a fault exists in the manufacturing process of the semiconductor wafer is detected based on the collected data, a large amount of computing resources are consumed, data congestion may also be caused, and therefore the fault detection effect and/or the detection efficiency are/is affected.
Further, the refinement of step S20 includes:
step S2001, calculating the dissimilarity between every two sequence values in the time sequence to obtain a dissimilarity matrix of the time sequence;
as can be seen, the acquired monitoring signal data is a time series, the dissimilarity between every two sequence values in the time series is calculated to obtain a dissimilarity matrix of the time series, and a time series with a sequence length of m is assumed
For every two data points therein
And
the dissimilarity between them may beIs shown as
. The dissimilarity measure herein can be usedpThe norm distance, in terms of euclidean distance for example, dissimilarity can be expressed as (refer to the following equation 1):
then, calculating the dissimilarity between every two sequence values in the time series S, and expressing the calculated dissimilarity in a matrix form to obtain a dissimilarity matrix of the time series SDAs shown in the following equation 2:
step S2002, establishing an optimization matrix of the time series based on the dissimilarity matrix, wherein elements in the optimization matrix are used for determining, according to the dissimilarity, whether one sequence value can represent the other sequence value between two corresponding sequence values;
for matrixDThe goal is to find a small subset from the time series S to represent the entire time series S. Thus, based on dissimilarity
And associated assignment variables
Establishing an optimization matrix, which is shown in the following formula 3:
wherein the content of the first and second substances,
for representing timeIn the sequence S
Whether or not to be able to represent
The process of, if possible,
i.e. equal to the first eigenvalue and vice versa equal to the second eigenvalue, where the first and second eigenvalues are set to 1 and 0 for ease of calculation.
And step S2003, calculating a cost function of the time sequence according to the optimization matrix, and performing feature extraction on the time sequence by using the cost function to obtain a target generalized signal in the semiconductor wafer manufacturing process.
Based on equations 1-3 above, the cost function for representing all the sequence values in the time series S with one subset is (equation 4):
meanwhile, based on equation 4, the number of selected feature data points should be as small as possible, and therefore, the number of non-zero rows in the optimization matrix Z can be further expressed as (equation 5):
wherein
Is composed ofpNorm, I (-) is an indicator function when input
The particular eigenvalue may be 1 for a positive time equal to the particular eigenvalue. Combining the cost functions shown in equations 4 to 5 yields a useful goal for signal summarizationCalibration function (equation 6):
and (5) summarizing the signal of the time series S by using the objective function shown in the formula 6 to obtain an objective summarized signal in the manufacturing process of the semiconductor wafer.
Further, the refinement of step S2003 includes:
step A1, calculating an objective function for extracting the characteristics of the time series by using the cost function;
step A2, performing optimization solution on the objective function according to constraint conditions of the objective function to obtain objective feature points of the time sequence, wherein signals corresponding to the objective feature points are objective generalized signals in the semiconductor wafer manufacturing process, the constraint conditions of the objective function include causal constraints, and the causal constraints are that a first sequence value in the time sequence can only be represented by a second sequence value before the time point.
Referring to equation 6, when obtaining the target generalized signal, an objective function that can be used for signal generalization of time-series data in the semiconductor wafer manufacturing process is first calculated according to the cost function, and then the calculated objective function is used to solve the objective function optimally according to the constraint conditions of the objective function, so as to obtain a target feature point that can perform signal generalization on the time-series. The constraint condition of the objective function includes a causal condition, that is, in a time series, one sequence value can only be represented by a sequence value before the time point, and specifically, can be represented by the following formula 7:
the constraint conditions of the objective function can also comprise uniqueness constraint and continuity constraint, when the target feature points are selected, the uniqueness constraint indicates that the sequence value corresponding to each time point has only one representation, the continuity constraint indicates that the time point corresponding to each sequence value can only represent continuous time points, and the uniqueness constraint and the continuity constraint can be expressed by the following formulas 8-9:
and based on the constraint conditions, carrying out optimization solution on an objective function which can be used for carrying out signal summarization on the time sequence to obtain each representative point which can be used for summarizing the time sequence, namely an objective characteristic point, wherein a signal formed by the objective characteristic points is an objective summarized signal corresponding to the time sequence.
And step S30, detecting whether the manufacturing process of the semiconductor wafer has faults or not according to the target summary signal.
Further, after obtaining the target summary signal of the monitor signal data in the manufacturing process of the semiconductor wafer, whether a fault exists in the manufacturing process of the semiconductor wafer is detected according to the calculated target summary signal. Specifically, the data points in the target summary signal are analyzed to determine whether an abnormal data point exists in the manufacturing process of the semiconductor wafer, so as to detect whether a fault exists in the manufacturing process of the semiconductor wafer.
In the embodiment, the monitoring signal data in the manufacturing process of the semiconductor wafer is obtained; summarizing the monitoring signal data, and extracting a target summarized signal in the manufacturing process of the semiconductor wafer; and detecting whether the manufacturing process of the semiconductor wafer has faults or not according to the target summarizing signal. The method can realize full-automatic feature extraction, reduce the dependence on expert experience in fault detection, extract the key features of monitoring signal data in the semiconductor manufacturing process through signal summarization, avoid the loss of feature information and improve the fault detection precision in the semiconductor manufacturing process.
Further, on the basis of the above-described embodiment of the present invention, a second embodiment of the fault detection method of the present invention is proposed.
This embodiment is a step before step S20 in the first embodiment, and the difference between this embodiment and the above-described embodiment of the present invention is:
before step S20, the method further includes:
step B1, acquiring standard signal data in the manufacturing process of the semiconductor standard wafer;
and step B2, performing dynamic time warping on the monitoring signal data based on the standard signal data, and aligning the time step between the monitoring signal data and the standard signal data.
Based on the above embodiments, in this embodiment, the manufacturing process of one wafer often requires multiple sensors to monitor multiple signals, and the signal patterns monitored by each sensor are different, so all the sensor signals need to be signal-summarized to calculate the data point set, i.e. the target feature point, for summarizing the sensor signals.
The template for signal summarization obtained from the monitoring signal data of the standard wafer manufacturing process can be directly applied to the signal summarization of the monitoring signal data of other similar wafer manufacturing processes. However, the time lengths of the manufacturing processes of different wafers are not necessarily completely the same, so that the monitoring signal data of the manufacturing processes of different wafers may refer to the monitoring signal data of the manufacturing process of the standard wafer, but if the sequence length of the acquired monitoring signal data is not consistent with the sequence length of the monitoring signal data of the manufacturing process of the standard wafer, dynamic time normalization needs to be performed on the acquired monitoring signal data for aligning the time step length between the acquired monitoring signal data of the manufacturing process of the wafer and the monitoring signal data of the standard wafer. Specifically, whether the sequence length of the collected monitoring signal data is consistent with the sequence length of the standard signal data of the standard wafer or not is judged, if not, the collected monitoring signal data is firstly subjected to dynamic time warping, and then signal summarization is carried out to obtain a target summarized signal.
Further, based on the monitoring signal data after dynamic time warping, when the time step of the collected monitoring signal data is consistent with the time step of the standard signal data, the generalized signal of the standard wafer can be used for quickly determining the target feature point serving as the representative point in the collected monitoring signal data, so that the collected monitoring signal data is subjected to quick feature extraction, namely after the time step is aligned, according to the corresponding relation between the data point in the collected monitoring signal data and the data point in the standard signal data, the data point corresponding to the data point in the standard generalized signal is selected from the collected monitoring signal data and serves as the target feature point, and therefore the target generalized signal is obtained.
Further, based on the steps B1-B2, the refinement of the step S30 in the above embodiment includes:
step S3001, acquiring a standard generalized signal of the standard wafer, wherein the standard generalized signal is obtained by performing feature extraction on the standard signal data;
step S3002, comparing the target summarized signal with the standard summarized signal, and detecting whether there is a fault in the manufacturing process of the semiconductor wafer.
Based on the monitoring signal data of the standard wafer, the standard generalized signal can be obtained by summarizing the monitoring signal data of the standard wafer, and the target generalized signal can be obtained by quickly summarizing the acquired monitoring signal data based on the standard generalized signal. Meanwhile, based on the standard generalized signal, the target generalized signal of the collected monitoring signal data is compared with the standard generalized signal, so that whether abnormal data exists in the collected monitoring signal data is determined, and whether a fault exists in the manufacturing process of the semiconductor wafer is detected. If there are data points in the target summary signal that differ significantly from the standard summary signal, it can be determined that a fault exists in the wafer fabrication process. In the present embodiment, the standard wafer is a wafer that meets the standard selected from the historically manufactured wafers.
Further, after step S30, the method further includes:
and step C1, when a feature design instruction is detected, modifying the feature points of the standard generalized signal according to the feature design instruction, updating the standard generalized signal according to the modified feature points, or generating a new version of the standard generalized signal.
Further, in this embodiment, the standard generalized signal of the standard wafer is not fixed but variable, and supports the user to set or select the feature point in a customized manner, so as to modify the standard generalized signal of the standard wafer, thereby meeting the personalized wafer manufacturing requirements of the user. Specifically, when a user has a personalized wafer manufacturing requirement, in order to detect whether a fault exists in the manufacturing process of the customized wafer, the generalized signal of the standard wafer may be modified, so as to clearly determine the signal value of each monitoring signal in the manufacturing process of the customized wafer, when a feature design instruction of the user is detected, a feature point in the standard generalized signal of the standard wafer is modified according to the detected feature design instruction, and the standard generalized signal is modified according to the modified feature point, or a new version of the standard generalized signal is generated on the basis of the original standard generalized signal.
Further, in this embodiment, the steps of the fault detection method of the present invention further include:
step D1, calculating the importance index of each standard feature point in the standard generalized signal;
and D2, screening the standard characteristic points according to the importance indexes of the standard characteristic points, and updating the standard generalized signals based on the screened standard characteristic points.
In addition to the design and selection of the customized feature points, in this embodiment, signal summarization is performed on the monitoring signal data of the standard wafer, and when the standard summarized signal is generated, the importance indexes of the calculated standard feature points are also measured according to the calculated importance indexes, the importance of each standard feature point is ranked and screened according to the importance of each standard feature point, and the standard summarized signal is generated based on the screened standard feature points. It can be understood that, on the basis of not affecting the detection accuracy, in order to reduce the amount of calculation, feature points as few as possible are selected to generate the standard generalized signal, and therefore, after the standard generalized signal is generated, the importance of each standard feature point in the standard signal can be measured at any time, so that the standard generalized signal is updated.
The index for evaluating the importance of each standard feature point includes a pearson correlation coefficient, a chi-squared value, mutual information, a maximum information coefficient, and the like, and taking the maximum information coefficient as an example, the calculation method of the maximum information coefficient of each standard feature point is shown in the following formula 10:
wherein
Is indexed byiThe characteristic points of (a) are set,yis the label of the corresponding label, and the label is the label,
for the corresponding joint probability distribution,
in order to be a likelihood distribution,
is a priori. If the characteristic
And a sample labelyThe relationship between the two is not close to each other,
will be smaller, conversely, if the two are closely related, then
The value will be relatively large. In pair
Ranking is performedAnd then, the calculated feature points can be further screened, and only the first k important feature points are selected for constructing the detection model, namely, standard generalized signals are generated for machine learning so as to construct the fault detection model.
Specifically, a vectorized matrix input is constructed by using the screened feature points as standard feature points, and based on the matrix input, different types of machine learning algorithms can be used for FD modeling, including principal component analysis, support vector machines and the like, so as to construct a fault detection model.
In the embodiment, the standard generalized signal of the standard wafer is used as a reference basis to be compared with the target generalized signal of the collected monitoring signal data, so that whether a fault exists in the wafer manufacturing process is detected, and the detection precision and the detection efficiency of the fault can be improved. Furthermore, based on standard summarization signals, dynamic time normalization and rapid signal summarization can be carried out on collected monitoring signal data, full process automation of feature design is achieved, and compared with a semi-automatic or manual feature extraction and modeling process of a traditional fault detection model, efficiency is higher. Meanwhile, the method for abstracting and summarizing signals can replace the traditional manual signal segmentation and feature extraction, the dependence on expert experience is reduced, and the complexity of a fault detection model is reduced based on the method for abstracting and summarizing signals, so that the model training and online reasoning become more efficient. Furthermore, in this embodiment, a user is supported to self-define the standard generalization signal, the additional feature design of the user to the standard generalization signal meets the personalized customization requirement of the user, and the applicability and flexibility of the fault detection model are improved.
In addition, referring to fig. 3, an embodiment of the present invention further provides a fault detection apparatus, where the fault detection apparatus includes:
the data acquisition module 10 is used for acquiring monitoring signal data in the manufacturing process of the semiconductor wafer;
a signal summarization module 20, configured to perform feature extraction on the monitoring signal data to obtain a target summarization signal in the semiconductor wafer manufacturing process;
and the fault detection module 30 is used for detecting whether a fault exists in the manufacturing process of the semiconductor wafer according to the target summary signal.
Optionally, the signal summarization module 20 is further configured to:
calculating the dissimilarity between every two sequence values in the time sequence to obtain a dissimilarity matrix of the time sequence;
establishing an optimization matrix of the time series based on the dissimilarity matrix, wherein elements in the optimization matrix are used for determining whether one sequence value can represent the other sequence value between two corresponding sequence values according to the dissimilarity;
and calculating a cost function of the time sequence according to the optimization matrix, and performing feature extraction on the time sequence by using the cost function to obtain a target generalized signal in the manufacturing process of the semiconductor wafer.
Optionally, the signal summarization module 20 is further configured to:
calculating an objective function for signal summarization of the time series by using the cost function;
and carrying out optimization solution on the objective function according to the constraint conditions of the objective function to obtain the objective characteristic points of the time sequence, wherein the signals corresponding to the objective characteristic points are objective generalized signals in the semiconductor wafer manufacturing process, the constraint conditions of the objective function comprise causal constraints, and the causal constraints are that a first sequence value in the time sequence can only be represented by a second sequence value before the time point.
Optionally, the fault detection apparatus further includes a dynamic time warping module, configured to:
acquiring standard signal data in the manufacturing process of a semiconductor standard wafer;
and performing dynamic time warping on the monitoring signal data based on the standard signal data, and aligning the time step between the monitoring signal data and the standard signal data.
Optionally, the failure detection module 30 is further configured to:
acquiring a standard generalized signal of the standard wafer, wherein the standard generalized signal is obtained by performing feature extraction on the standard signal data;
and comparing the target summarized signal with the standard summarized signal to detect whether the manufacturing process of the semiconductor wafer has faults or not.
Optionally, the fault detection apparatus further includes a custom feature design module, configured to:
and when a characteristic design instruction is detected, modifying the characteristic points of the standard generalized signal according to the characteristic design instruction, updating the standard generalized signal according to the modified characteristic points, or generating a new version of the standard generalized signal.
Optionally, the fault detection apparatus further includes a feature screening module, configured to:
calculating the importance index of each standard characteristic point in the standard generalized signal;
and screening the standard characteristic points according to the importance indexes of the standard characteristic points, and updating the standard generalized signals based on the screened standard characteristic points.
In addition, an embodiment of the present invention further provides a computer-readable storage medium, where a fault detection program is stored, and when the fault detection program is executed by a processor, the fault detection program implements operations in the fault detection method provided in the foregoing embodiment.
In addition, an embodiment of the present invention further provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the computer program implements the operations in the fault detection method provided in the foregoing embodiments.
The embodiments of the apparatus, the computer program product, and the computer-readable storage medium of the present invention may refer to the embodiments of the fault detection method of the present invention, and are not described herein again.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity/action/object from another entity/action/object without necessarily requiring or implying any actual such relationship or order between such entities/actions/objects; the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
For the apparatus embodiment, since it is substantially similar to the method embodiment, it is described relatively simply, and reference may be made to some descriptions of the method embodiment for relevant points. The above-described apparatus embodiments are merely illustrative, in that elements described as separate components may or may not be physically separate. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of the invention. One of ordinary skill in the art can understand and implement it without inventive effort.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium (e.g., ROM/RAM, magnetic disk, optical disk) as described above and includes instructions for enabling a terminal device (e.g., a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the fault detection method according to the embodiments of the present invention.
The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (10)
1. A fault detection method for detecting faults in a semiconductor wafer manufacturing process, the fault detection method comprising the steps of:
acquiring monitoring signal data in the manufacturing process of a semiconductor wafer;
performing feature extraction on the monitoring signal data to obtain a target generalized signal in the manufacturing process of the semiconductor wafer;
and detecting whether the manufacturing process of the semiconductor wafer has faults or not according to the target summarizing signal.
2. The method of claim 1, wherein the monitoring signal data is a time series, and the step of performing feature extraction on the monitoring signal data to obtain the target summary signal in the semiconductor wafer manufacturing process comprises:
calculating the dissimilarity between every two sequence values in the time sequence to obtain a dissimilarity matrix of the time sequence;
establishing an optimization matrix of the time series based on the dissimilarity matrix, wherein elements in the optimization matrix are used for determining whether one sequence value can represent the other sequence value between two corresponding sequence values according to the dissimilarity;
and calculating a cost function of the time sequence according to the optimization matrix, and performing feature extraction on the time sequence by using the cost function to obtain a target generalized signal in the manufacturing process of the semiconductor wafer.
3. The method of claim 2, wherein the step of extracting the features of the time series using the cost function to obtain the target summary signal in the semiconductor wafer manufacturing process comprises:
calculating an objective function for extracting the characteristics of the time series by using the cost function;
and carrying out optimization solution on the objective function according to the constraint conditions of the objective function to obtain the objective characteristic points of the time sequence, wherein the signals corresponding to the objective characteristic points are objective generalized signals in the semiconductor wafer manufacturing process, the constraint conditions of the objective function comprise causal constraints, and the causal constraints are that a first sequence value in the time sequence can only be represented by a second sequence value before the time point.
4. The fault detection method of claim 1, wherein the monitoring signal data is in a time series, and wherein the step of extracting the characteristics of the monitoring signal data is preceded by the step of:
acquiring standard signal data in the manufacturing process of a semiconductor standard wafer;
and performing dynamic time warping on the monitoring signal data based on the standard signal data, and aligning the time step between the monitoring signal data and the standard signal data.
5. The method of claim 4, wherein the step of detecting whether the manufacturing process of the semiconductor wafer has a fault according to the target summary signal comprises:
acquiring a standard generalized signal of the standard wafer, wherein the standard generalized signal is obtained by performing feature extraction on the standard signal data;
and comparing the target summarized signal with the standard summarized signal to detect whether the manufacturing process of the semiconductor wafer has faults or not.
6. The fault detection method of claim 5, wherein the fault detection method further comprises:
and when a characteristic design instruction is detected, modifying the characteristic points of the standard generalized signal according to the characteristic design instruction, updating the standard generalized signal according to the modified characteristic points, or generating a new version of the standard generalized signal.
7. The fault detection method according to claim 5 or 6, characterized in that it further comprises:
calculating the importance index of each standard characteristic point in the standard generalized signal;
and screening the standard characteristic points according to the importance indexes of the standard characteristic points, and updating the standard generalized signals based on the screened standard characteristic points.
8. A fault detection device, characterized in that the fault detection device comprises:
the data acquisition module is used for acquiring monitoring signal data in the manufacturing process of the semiconductor wafer;
the signal summarization module is used for carrying out feature extraction on the monitoring signal data to obtain a target summarization signal in the manufacturing process of the semiconductor wafer;
and the fault detection module is used for detecting whether a fault exists in the manufacturing process of the semiconductor wafer according to the target summary signal.
9. A terminal device, characterized in that the terminal device comprises: memory, a processor and a fault detection program stored on the memory and executable on the processor, the fault detection program when executed by the processor implementing the steps of the fault detection method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that a fault detection program is stored on the computer-readable storage medium, which when executed by a processor implements the steps of the fault detection method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111077510.5A CN113539909A (en) | 2021-09-15 | 2021-09-15 | Fault detection method and device, terminal equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111077510.5A CN113539909A (en) | 2021-09-15 | 2021-09-15 | Fault detection method and device, terminal equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113539909A true CN113539909A (en) | 2021-10-22 |
Family
ID=78093259
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111077510.5A Pending CN113539909A (en) | 2021-09-15 | 2021-09-15 | Fault detection method and device, terminal equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113539909A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114781674A (en) * | 2022-06-16 | 2022-07-22 | 深圳市信润富联数字科技有限公司 | Method and device for positioning faults of wind power equipment, storage medium and electronic equipment |
| CN114859875A (en) * | 2022-07-07 | 2022-08-05 | 深圳市信润富联数字科技有限公司 | Fault management method, device, equipment and storage medium for multiple equipment |
| CN116046078A (en) * | 2023-03-31 | 2023-05-02 | 东莞市楷德精密机械有限公司 | Fault monitoring and early warning method and system for semiconductor cleaning equipment |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101976045A (en) * | 2010-08-25 | 2011-02-16 | 江苏大学 | Panel quality virtual measurement method and system for TFT-LCD etching process |
| US20120016643A1 (en) * | 2010-07-16 | 2012-01-19 | National Tsing Hua University | Virtual measuring system and method for predicting the quality of thin film transistor liquid crystal display processes |
| CN109670584A (en) * | 2018-12-27 | 2019-04-23 | 张婧 | A kind of fault diagnosis method and system based on big data |
| CN110096036A (en) * | 2018-01-29 | 2019-08-06 | 阿里巴巴集团控股有限公司 | A kind of determination method, device and equipment of equipment state |
| CN111368890A (en) * | 2020-02-26 | 2020-07-03 | 珠海格力电器股份有限公司 | Fault detection method and device and information physical fusion system |
| US20200333774A1 (en) * | 2018-10-09 | 2020-10-22 | Applied Materials, Inc. | Adaptive chamber matching in advanced semiconductor process control |
-
2021
- 2021-09-15 CN CN202111077510.5A patent/CN113539909A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120016643A1 (en) * | 2010-07-16 | 2012-01-19 | National Tsing Hua University | Virtual measuring system and method for predicting the quality of thin film transistor liquid crystal display processes |
| CN101976045A (en) * | 2010-08-25 | 2011-02-16 | 江苏大学 | Panel quality virtual measurement method and system for TFT-LCD etching process |
| CN110096036A (en) * | 2018-01-29 | 2019-08-06 | 阿里巴巴集团控股有限公司 | A kind of determination method, device and equipment of equipment state |
| US20200333774A1 (en) * | 2018-10-09 | 2020-10-22 | Applied Materials, Inc. | Adaptive chamber matching in advanced semiconductor process control |
| CN109670584A (en) * | 2018-12-27 | 2019-04-23 | 张婧 | A kind of fault diagnosis method and system based on big data |
| CN111368890A (en) * | 2020-02-26 | 2020-07-03 | 珠海格力电器股份有限公司 | Fault detection method and device and information physical fusion system |
Non-Patent Citations (1)
| Title |
|---|
| 刘强等: "数据驱动的工业过程运行监控与自优化研究展望 ", 《自动化学报》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114781674A (en) * | 2022-06-16 | 2022-07-22 | 深圳市信润富联数字科技有限公司 | Method and device for positioning faults of wind power equipment, storage medium and electronic equipment |
| CN114781674B (en) * | 2022-06-16 | 2022-09-16 | 深圳市信润富联数字科技有限公司 | Method and device for positioning faults of wind power equipment, storage medium and electronic equipment |
| CN114859875A (en) * | 2022-07-07 | 2022-08-05 | 深圳市信润富联数字科技有限公司 | Fault management method, device, equipment and storage medium for multiple equipment |
| CN116046078A (en) * | 2023-03-31 | 2023-05-02 | 东莞市楷德精密机械有限公司 | Fault monitoring and early warning method and system for semiconductor cleaning equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113539909A (en) | Fault detection method and device, terminal equipment and storage medium | |
| CN109612999A (en) | System and method for wafer map analysis | |
| US20090228217A1 (en) | Defect inspection method | |
| CN112767366A (en) | Image recognition method, device and equipment based on deep learning and storage medium | |
| US20070094196A1 (en) | Manufacture data analysis method and manufacture data analyzer apparatus | |
| CN105702595B (en) | The yield judgment method of wafer and the changeable quantity measuring method of wafer conformity testing | |
| JP2020198092A (en) | Method and system for unsupervised anomaly detection and cause explanation with majority voting for high-dimensional sensor data | |
| JP2009068946A (en) | Flaw sorting apparatus, flaw sorting method and program | |
| KR101998972B1 (en) | Method of analyzing and visualizing the cause of process failure by deriving the defect occurrence index by variable sections | |
| CN116453438A (en) | Display screen parameter detection method, device, equipment and storage medium | |
| CN116188432A (en) | Training method and device of defect detection model and electronic equipment | |
| CN113344994B (en) | Image registration method, device, electronic equipment and storage medium | |
| KR101782364B1 (en) | Vision inspection method based on learning data | |
| US8682058B2 (en) | Defect analysis method, apparatus, and recording medium using pattern dependence degree | |
| US20110013824A1 (en) | Inspection area setting method, inspection area setting apparatus, and computer program product | |
| CN113469151A (en) | Method, device, equipment and medium for screening sensor in industrial manufacturing | |
| CN116702005A (en) | Neural network-based data anomaly diagnosis method and electronic equipment | |
| CN113327204B (en) | Image calibration method, device, equipment and storage medium | |
| CN109891454A (en) | For sensing the method and system of the minor variations of process/equipment measurement data | |
| US20230325413A1 (en) | Error Factor Estimation Device and Error Factor Estimation Method | |
| JP4308113B2 (en) | Data analysis apparatus and method, and program | |
| CN112749781A (en) | Key point prediction method, system, electronic device and storage medium | |
| US20240095306A1 (en) | Information processing apparatus and information processing method | |
| JP2020101900A (en) | Image inspection apparatus, learning method for image inspection and image inspection program | |
| US11592807B2 (en) | Manufacturing defect factor searching method and manufacturing defect factor searching apparatus |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20211022 |