CN117671376A

CN117671376A - Annular defect identification method and system

Info

Publication number: CN117671376A
Application number: CN202311684453.6A
Authority: CN
Inventors: 苏远丽; 金肖明; 杨鑫; 杜余峰; 王明照; 沈明杰; 马佳豪
Original assignee: Xinrate Intelligent Technology Suzhou Co ltd
Current assignee: Xinrate Intelligent Technology Suzhou Co ltd
Priority date: 2023-12-08
Filing date: 2023-12-08
Publication date: 2024-03-08

Abstract

The invention relates to the technical field of semiconductors, in particular to a method and a system for identifying annular defects, comprising the following steps: performing defect classification on the defect graph by adopting a defect classification model to obtain at least one graph set; acquiring actual production information, wherein the actual production information comprises maintenance level of a machine and the number of graphs of a graph set; selecting a recommended graphic processing mode according to actual production information, wherein the graphic processing mode comprises the following steps of: stacking and fitting; when the actual production information accords with the first processing condition and the second processing condition, the recommended graphic processing modes are respectively overlapping and fitting; performing graphic processing on the graphic set by selecting a recommended graphic processing mode to obtain a new defect graphic; and judging the fault type which causes the generation of the annular defect according to the new defect pattern. The invention provides a graphic processing scheme based on machine and product data fusion guidance, so that the defect positioning efficiency is improved and the reliability of a positioning result is ensured.

Description

Annular defect identification method and system

Technical Field

The invention relates to the technical field of semiconductors, in particular to a method and a system for identifying annular defects.

Background

During the fabrication of semiconductor devices, each process flow may create undesirable structures on the wafer, which may cause the on-chip circuitry to fail to function properly, known as wafer defects. In order to ensure the wafer yield and productivity, it is important to maintain the processing equipment of the wafer. The maintenance of the existing wafer equipment mainly has two major difficulties:

1. maintenance of equipment failure with hysteresis

Existing equipment maintenance schemes are typically: and performing defect detection on the wafer by using the defect detection model, and further reversely pushing out whether the machine has faults by using a defect detection result. However, the scheme of using the defect to reversely push whether the machine has faults is that firstly, serious hysteresis exists in fault discovery, secondly, a judging mechanism is often required to be independently constructed for different types of machines or processes so as to monitor whether the operation of the corresponding machine is in a normal state, and the monitoring difficulty and the cost are very high.

For example, CN202311213205.3 discloses an intelligent monitoring and early warning system for wafer production. The early warning system can discover the problem of possible dust abnormality increase by using multiple treatments such as an edge detection algorithm, a double-branch defect detection model, an equipment fault prediction model and the like. Further, the failure of the machine is reversely deduced from the dust phenomenon. However, the analysis of the singleness problem is very limited in the types of faults and machine types that can be monitored, and it is difficult to find marks from the wafer at the beginning of the occurrence of the hidden trouble.

Also for example, CN202310893076.0 discloses a distributed parallel computing based FDC trace analysis method and storage medium. However, the method often uses testing, diagnosis and other operations to analyze the failure cause only after the failure problem is highlighted.

In fact, in the processing process, the data volume of different types of machines is huge and complex, and parameters with different properties are included, so that the difficulty in data acquisition, analysis and supervision is very high.

2. Automatic identification or tracing of defects is difficult

In order to find the hidden trouble problem of the machine at an earlier stage, the early maintenance of the machine is also quite critical. In fact, in actual production, the step of using defects to locate directly to a specific machine tool often requires manual analysis by a field engineer.

Currently, there are also schemes that attempt to perform automated localization by comparing defect data with equipment parameters. For example, CN201310119893.7 discloses a method and system for automatically detecting mechanical scratches. The method locates the device that generated the defect by matching the defect data with the generating device. As another example, CN202111547662.7 discloses a method of detecting robot arm scratches. However, due to the complexity of the wafer processing flow and the concealment of early defects (e.g., which may be difficult to clearly appear on a wafer photo), it is also difficult for an automated inspection method to locate an accurate tool at a first time.

Therefore, a method for monitoring and early warning the wafer quality and the machine maintenance in time when the fault is earliest is needed.

Disclosure of Invention

The invention aims to provide a method and a system for identifying annular defects, which partially solve or alleviate the defects in the prior art and can monitor and early warn the quality of a wafer and the maintenance of a machine in time when the fault is earliest.

In order to solve the technical problems, the invention adopts the following technical scheme:

in a first aspect of the present invention, there is provided a method for identifying an annular defect, including the steps of:

s201, obtaining a plurality of first defect graphs of a plurality of sample wafers, and performing defect classification on the plurality of first defect graphs by adopting a defect classification model to obtain at least one first graph set, wherein the first graph set comprises: at least one defect pattern having a ring defect;

s202, acquiring actual production information, wherein the actual production information comprises: the maintenance grade of at least one machine and the number of patterns of the first pattern set, which are experienced by the sample wafer in the processing process, are acquired through the machine maintenance monitoring system;

s203, selecting a recommended graphic processing mode according to the actual production information, wherein the graphic processing mode comprises the following steps: stacking and fitting; when the actual production information accords with the first processing condition, the recommended graphic processing mode is a superposition graph; and/or when the actual production information accords with the second processing condition, the recommended graphic processing mode is fitting; wherein,

The first processing condition means that the maintenance level is less than or equal to a set maintenance level and the number of graphics is greater than or equal to a first set number; the second processing condition means that the maintenance level is greater than a set maintenance level and the number of patterns is less than a first set number;

s204, performing graphic processing on the first graphic set by using a recommended graphic processing mode to obtain a second defect graphic;

s205, judging the fault type which causes the annular defect to be generated according to the second defect graph.

In some embodiments, the fault types include: fault machine information; s205 includes the steps of:

obtaining shape data of the defect through the second defect graph, wherein the shape data comprises one or more of the following: defect length, defect radian, defect radius, and gap between defect and wafer center;

acquiring a preset machine characteristic library associated with the defect graph; the machine characteristic library comprises: a first set of quantized features comprising one or more of the following defect feature data: the rotation radius of the mechanical arm, the rotation speed of the mechanical arm, the distance between the mechanical arm and the wafer center and the position of an adsorption foot of the adsorption equipment are respectively associated with corresponding wafer machine information;

And matching the shape data with the machine characteristic library, and identifying the machine opposite to the current shape data as a corresponding fault machine when the shape data is matched with the corresponding defect characteristic data.

In some embodiments, the machine feature library further comprises: a set of graphical features, the set of graphical features comprising: and defect characteristic patterns corresponding to defects caused by at least one machine or equipment parts in the machine.

In some embodiments, the step of S201 further comprises: and marking defect points in the defect graphs in the first graph set, wherein the defect points are points identified as annular defects by the defect classification model.

In some embodiments, before S203, the method further comprises the step of:

selecting a plurality of defect sample groups from the marked defect points, wherein one defect sample group comprises: one or more defect points;

calculating the distances between the defect sample groups and the wafer centers; when the difference value between the distance between the defect sample group and the circle center and the standard distance is larger than a first difference value threshold value, identifying the corresponding defect sample group as an error group;

and when the number of the error groups is greater than an error threshold value, screening the corresponding first defect graph from the first graph set to obtain a new first graph set.

In some embodiments, S203 comprises:

when the actual production information does not accord with the first processing condition or the second processing condition, a first prompt signal is sent to a user;

and receiving a first selection signal issued by a user, the first selection signal comprising: and the graphics processing mode information is used for determining a recommended graphics processing mode according to the first selection signal.

Another aspect of the present invention provides a system for identifying an annular defect, including:

the annular defect classification module is configured to obtain a plurality of first defect patterns of a plurality of sample wafers, and perform defect classification on the plurality of first defect patterns by adopting a defect classification model so as to obtain at least one first pattern set, wherein the first pattern set comprises: at least one defect pattern having a ring defect;

the actual production information acquisition module is configured to acquire actual production information, and the actual production information comprises:

the maintenance grade of at least one machine and the number of patterns of the first pattern set, which are experienced by the sample wafer in the processing process, are acquired through the machine maintenance monitoring system;

the processing mode selection module is configured to select a recommended graphic processing mode according to the actual production information, and the graphic processing mode comprises: stacking and fitting; when the actual production information accords with the first processing condition, the recommended graphic processing mode is a superposition graph; and/or when the actual production information accords with the second processing condition, the recommended graphic processing mode is fitting; wherein,

the graphic processing module is configured to select a recommended graphic processing mode to perform graphic processing on the first graphic set so as to obtain a second defect graphic;

and the fault judging module is configured to judge the fault type causing the annular defect according to the second defect graph.

In some embodiments, the fault types include: fault machine information; the fault judging module comprises:

a shape data acquisition unit configured to acquire shape data of the defect through the second defect pattern, the shape data including one or more of: defect length, defect radian, defect radius, and gap between defect and wafer center;

the feature library acquisition unit is configured to acquire a preset machine feature library associated with the defect graph; the machine characteristic library comprises: a set of quantized features comprising one or more of the following defect feature data: the rotation radius of the mechanical arm, the rotation speed of the mechanical arm, the distance between the mechanical arm and the wafer center and the position of an adsorption foot of the adsorption equipment are respectively associated with corresponding wafer machine information;

And the first matching unit is configured to match the defect characteristic data set with the machine characteristic library, and identify a machine opposite to the shape data as a corresponding fault machine when the defect characteristic data is matched with the corresponding defect characteristic data.

In some embodiments, the annular defect classification module is further configured to mark defect points in the defect patterns in the first pattern set, wherein the defect points are points identified as annular defects by the defect classification model.

In this embodiment, the operation condition (such as the risk level of the machine) and the actual defect condition (such as the number of the classified defect patterns) of the machine are introduced to automatically select the processing mode of the defect patterns. And further, the defect identification or analysis mode can be flexibly adjusted on the basis of ensuring the defect positioning reliability in the face of different actual production scenes.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. Like elements or portions are generally identified by like reference numerals throughout the several figures. In the drawings, elements or portions thereof are not necessarily drawn to scale. It will be apparent to those of ordinary skill in the art that the drawings in the following description are of some embodiments of the invention and that other drawings may be derived from these drawings without inventive faculty.

FIG. 1 is a flow chart of a shift scheduling method according to an exemplary embodiment of the invention;

FIG. 2 is a schematic diagram of a system module of a shift scheduling method according to an exemplary embodiment of the present invention;

FIG. 3 is a flow chart of a method for identifying annular defects according to an exemplary embodiment of the invention;

FIG. 4 is a flow chart illustrating a method for identifying linear defects according to an exemplary embodiment of the present invention;

FIG. 5 is a schematic diagram of a sample pattern of ring defects;

FIG. 6 is a schematic diagram of another ring defect sample;

FIG. 7 is a schematic diagram of a sample graph of a radiation defect;

FIG. 8 is a schematic diagram of a sample pattern of scratch defects;

fig. 9 is a schematic diagram of a sample pattern of yet another scratch defect.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. It will be apparent that the described embodiments are some, but not all, embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

In this document, suffixes such as "module", "component", or "unit" used to represent elements are used only for facilitating the description of the present invention, and have no particular meaning in themselves. Thus, "module," "component," or "unit" may be used in combination.

The terms "upper," "lower," "inner," "outer," "front," "rear," "one end," "the other end," and the like herein refer to an orientation or positional relationship based on that shown in the drawings, merely for convenience of description and to simplify the description, and do not denote or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus should not be construed as limiting the invention. Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.

The terms "mounted," "configured to," "connected," and the like, herein, are to be construed broadly as, for example, "connected," whether fixedly, detachably, or integrally connected, unless otherwise specifically defined and limited; the two components can be mechanically connected, can be directly connected or can be indirectly connected through an intermediate medium, and can be communicated with each other. The specific meaning of the above terms in the present invention will be understood in specific cases by those of ordinary skill in the art.

Herein, "and/or" includes any and all combinations of one or more of the associated listed items.

Herein, "plurality" means two or more, i.e., it includes two, three, four, five, etc.

It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus 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 apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

As used in this specification, the term "about" is typically expressed as +/-5% of the value, more typically +/-4% of the value, more typically +/-3% of the value, more typically +/-2% of the value, even more typically +/-1% of the value, and even more typically +/-0.5% of the value.

In this specification, certain embodiments may be disclosed in a format that is within a certain range. It should be appreciated that such a description of "within a certain range" is merely for convenience and brevity and should not be construed as a inflexible limitation on the disclosed ranges. Accordingly, the description of a range should be considered to have specifically disclosed all possible sub-ranges and individual numerical values within that range. For example, a rangeThe description of (c) should be taken as having specifically disclosed sub-ranges such as from 1 to 3, from 1 to 4, from 1 to 5, from 2 to 4, from 2 to 6, from 3 to 6, etc., as well as individual numbers within such ranges, e.g., 1,2,3,4,5, and 6. The above rule applies regardless of the breadth of the range.

Herein, a machine (or processing machine) is also referred to as an apparatus (or processing equipment), which may refer to any one or more processing modules or devices on a production processing line of wafers.

Herein, "failure" may also be referred to as a machine failure or an equipment failure, and refers to an equipment or a process problem that may affect a process such as processing, detecting, etc. a wafer, thereby affecting a yield or productivity of the wafer. For example, the fault may be a mechanical fault, such as a scratch defect on the wafer caused by loosening of a mechanical arm of the machine; for another example, the fault may be a process fault, such as a problem with the process recipe settings of the tool, resulting in insufficient wafer throughput.

Herein, a failure may refer to a specific failure problem (e.g., loosening of screws on a certain machine), or may refer to the same type of failure (e.g., occurrence of problems with process recipes on associated machines).

A defective dot, as used herein, refers to a dot on a wafer where an anomaly exists and may cause a circuit failure problem (e.g., a pixel anomaly on a wafer pattern).

Herein, the ring-shaped defect refers to a defect formed by gathering a plurality of defect points in a ring shape (e.g., a circular ring) or an approximately ring shape state on a wafer, as shown in fig. 5 and 6.

Herein, the linear defect (or linear defect) refers to a defect formed by linearly collecting a plurality of defect points on a wafer. The straight line defect includes one or more of the following types: linear scratches, linear particle defects, and radial particle defects. The linear scratches may be one scratch (as shown in fig. 8 and 9) or multiple scratches on the wafer; the linear particle defects are typically particle defects that are linearly aggregated; the radial particle defects are typically particle defects that diverge in multiple directions (as shown in fig. 7, which is a plurality of particle defects that diverge radially across the wafer).

In large wafer processing plants, it is often involved in the simultaneous operation of multiple wafer processing lines. And because of the types and the number of the machine stations, the difficulty of monitoring the running state of the machine stations is very high, so that the machine stations are very likely to be missed in the early stage of the fault (for example, the experience of a field engineer is insufficient, the fault is misjudged as normal running), further the influence degree of the fault is enlarged, and the wafer yield is seriously influenced.

At present, the machine fault is usually deduced reversely through the defect graph, but at the earliest stage that the fault may be about to occur or has occurred, the actual defect reflected on the wafer may not be obvious, as shown in fig. 8 and 9, and the defect is actually incomplete. At this time, if the working experience of the worker is relatively small, the defect is likely to be passed, so that the adverse effect of the malfunction is gradually enlarged. Alternatively, automated classification models have difficulty locating defective patterns for them.

In order to reduce the influence of faults on the operation of the production line as much as possible, the invention proposes a scheme capable of maintaining or checking the hidden trouble of faults possibly existing before wafer defects are generated or at the early stage when the number of the defects is relatively small.

Example 1

As shown in FIG. 1, the invention provides a scheduling method and system for performing early maintenance on a machine. The invention firstly provides a scheduling method capable of carrying out early maintenance on a machine based on dynamic and static combined production line data, which comprises the following steps:

s101, determining the static maintenance level of at least one machine according to a static database; wherein, the static database comprises: setting maintenance time of a plurality of machine stations; the static maintenance grade of the machine can be classified into a first grade, a second grade and a third grade according to the interval between the current time and the set maintenance time;

s102, determining a first dynamic maintenance level of the machine according to a first dynamic database; the first dynamic database includes one or more of the following operational data: voltage operation data, current operation data and standard data corresponding to the voltage operation data and the current operation data; the working data are used for representing the change relation of the working value of the machine along with the working time; correspondingly, S102 includes:

acquiring at least one piece of working data of the machine and standard data corresponding to the working data;

when the difference between the working value (such as a voltage value or a current value) of one data point in the working data and the standard data is monitored to be larger than a preset difference threshold (namely a second difference threshold), the data point is regarded as an abnormal point;

Determining a first dynamic maintenance level according to the abnormal point; when the number of abnormal points belongs to a first set threshold value, the first dynamic maintenance level is one level; when the number of abnormal points belongs to a second set threshold value, the first dynamic maintenance level is two levels; when the number of abnormal points belongs to a third set threshold value, the first dynamic maintenance level is three-level;

s103, determining the comprehensive maintenance level of the corresponding machine according to the static maintenance level and the first dynamic maintenance level; when the static maintenance level is one level and the first dynamic maintenance level is one level, the comprehensive maintenance level is one level; when the static maintenance level is a second level and the first dynamic maintenance level is a second level, the comprehensive maintenance level is a second level; when the static maintenance level is three-level and the first dynamic maintenance level is three-level, the comprehensive maintenance level is three-level;

s104, sorting the maintenance sequence of the machine according to the comprehensive maintenance level. At this time, a scheduling scheme obtained according to the comprehensive maintenance level ranking can be obtained.

In some embodiments, when the static maintenance level is one level and the first dynamic maintenance level is two levels,

The comprehensive maintenance level is two levels;

when the static maintenance level is one level and the first dynamic maintenance level is three levels, the comprehensive maintenance level is three levels;

when the static maintenance level is two-level and the first dynamic maintenance level is one-level, the comprehensive maintenance level is one-level;

when the static maintenance level is two-level and the first dynamic maintenance level is three-level, the comprehensive maintenance level is three-level;

when the static maintenance level is three-level and the first dynamic maintenance level is one-level, the comprehensive maintenance level is two-level;

and when the static maintenance level is three-level and the first dynamic maintenance level is two-level, the comprehensive maintenance level is three-level.

For example, in some embodiments, the static database includes: the set maintenance time (or recommended maintenance time) of the plurality of machine stations preset by the user; and time grading rules corresponding to each machine station when static maintenance grades are graded.

For example, for different types of machine, the next set maintenance time of the machine can be predefined in combination with the service life and the conventional maintenance period. And, according to the interval between the current time and the set maintenance time, the static maintenance level of the machine can be at least divided into a first level, a second level and a third level (which is equivalent to dividing the static operation risk of the machine into a low risk, a medium risk and a high risk).

For example, in some embodiments, when a certain machine is greater than or equal to 20 days from the set maintenance time, the static maintenance level is one level, when the set maintenance time is less than 20 days and greater than or equal to 10 days from the set maintenance time, the static maintenance level is two level, and when the set maintenance time is less than 10 days, the static maintenance level is three level.

The set maintenance time may also be a safe service period (or average service life) of the machine or a device component of the machine.

For another example, in some embodiments, the safe service period of the mechanical arm a in a certain machine is 1 year, the static maintenance level of the first 1-6 months may be defined as one level, the static maintenance level of the first 7-9 months may be defined as two levels, and the static maintenance level of the first 10-12 months may be defined as three levels. It will be appreciated that the specific temporal ranking rules may be freely adjusted by the user.

In some embodiments, the voltage operating data used to determine the first dynamic maintenance level may be an SPC control Chart of voltages (e.g., voltage monitoring is implemented using SPC Chart). An abnormality is considered to occur in a data point in an SPC control chart when a large difference between the voltage value (corresponding to the operating value) and the control point (corresponding to the standard data) is detected. And, the first dynamic maintenance class can be classified into a first class, a second class, and a third class (corresponding to a sequential classification into a low risk, a medium risk, and a high risk) according to the abnormal number of data points.

Alternatively, in other embodiments, SPC control charts of current flow can be used to implement the hierarchical processing of the first dynamic maintenance level.

The standard data may be data selected or input by a user in combination with actual production conditions.

In some embodiments, before S103, the method further comprises the step of:

s104, determining a second dynamic maintenance level of the machine according to a second dynamic database, wherein the second dynamic database comprises: product data of the wafer, wherein S104 includes: selecting product data associated with the machine from the second dynamic database as dynamic adjustment data, and classifying the second dynamic maintenance grade into a first grade, a second grade and a third grade according to the wafer quality according to the dynamic adjustment data;

s105, verifying or adjusting the first dynamic maintenance level according to the second dynamic maintenance level; wherein,

when the second dynamic maintenance level is smaller than or equal to the first dynamic maintenance level, maintaining the original level of the second dynamic maintenance level; when the second dynamic maintenance level is greater than the first dynamic maintenance level, the first dynamic maintenance level is adjusted according to the second dynamic maintenance level; wherein when the second dynamic maintenance level is two or three stages and the first dynamic maintenance level is one stage, the first dynamic maintenance level is adjusted to be two stages; when the second dynamic maintenance level is three-level and the first dynamic maintenance level is two-level, the first dynamic maintenance level is adjusted to three-level.

In some embodiments, the second dynamic maintenance level may be graded according to the number of defects of the wafer. For example, inputting a wafer into a defect classification model (e.g., an ADC classification model) may initially identify the number of defects or types of defects that may be present on the wafer. The second dynamic maintenance class is classified into a first class, a second class and a third class in order from small to large according to the number of defects which may exist.

Alternatively, in other embodiments, the yield information of the wafer may be obtained from the product data, and the risk of the second dynamic maintenance level may be obtained directly in reverse from the yield data.

In some embodiments, the product data includes one or more of the following: annular defects, scratch defects, radial particle defects, and linear particle defects; the machine in the static database is marked with a corresponding machine label; the machine label comprises: the first machine label is used for executing rotation operation in the wafer processing process of a machine corresponding to the first machine label; the second machine label is used for clamping the wafer in the wafer processing process; the third machine label is used for executing air blowing operation in the wafer processing process; wherein, aiming at the machine stations marked with different labels, different defect data (or product data) have different preset recommendation priorities;

Correspondingly, in S104, dynamic adjustment data corresponding to the machine label is selected according to the recommended priority of the product data.

For example, in some embodiments, for a first type of machine labeled with a first machine label, the annular defect is associated with a first recommended priority, and the scratch defect, the radial particle defect, and the linear particle defect are associated with a second recommended priority. In this case, the product data of the annular defect is preferably selected as dynamic adjustment data for the first type of machine.

For example, in some embodiments, for a second type of machine labeled with a second machine label, the scratch defect is associated with a first recommended priority, and the annular defect, the radial particle defect, and the linear particle defect are associated with a second recommended priority. In this case, the product data of the scratch defect is preferably selected as dynamic adjustment data for the second type of machine.

For example, in some embodiments, for a third type of machine labeled with a third machine label, a radial particle defect, a linear particle defect is associated with a first recommended priority, and a scratch defect, a ring defect is associated with a second recommended priority. In other words, in the third type of machine, it is preferable to select radial particle defects and product data of the radial particle defects as dynamic adjustment data.

For another example, in some embodiments, when two or more tags are associated with a machine, multiple product data may be selected as dynamic adjustment data.

For example, in some embodiments, the first machine label corresponds to one or more of the following machines: a rotating machine, a polishing mucous membrane machine, etc.

For example, in some embodiments, the machine corresponding to the second machine tag has one or more of the following: wafer dicing machine.

For example, in some embodiments, the third machine label corresponds to one or more of the following: a dryer station, a blower station, etc.

In the invention, factors related to the operation of the machine are selected in a limited way from the angles of static data and dynamic data respectively, and the limited data are subjected to multiple fusion by utilizing a multi-stage evaluation scheme. And the operation condition of the machine is subjected to multistage evaluation by utilizing multistage fusion of dynamic and static data so as to timely maintain and repair the machine at the initial stage of fault generation even before the fault is truly reflected on the wafer graph. In addition, the invention can improve the maintenance efficiency in a mode of advancing the maintenance time under the condition of not additionally increasing personnel configuration by the rapid evaluation mechanism, and simultaneously enables the limited personnel configuration to exert larger maintenance effectiveness.

In order to improve the linkage between the static data and the dynamic data, the invention also synchronously defines the machine type, the defect type and other angles, and mainly selects the linear defect (particularly, the scratch defect or the granular defect) and the annular defect as key identification objects so as to assist in improving the reliability and the accuracy of the dynamic maintenance grade evaluation of the machine.

For example, in some embodiments, taking an annular defect as an example, when it is detected that a wafer passing through a certain station 1 (which has a first station tag associated therewith) forms an annular defect on the defect, and the annular defect is deemed to be generated by the station 1, then the second dynamic maintenance level of the station is identified as three levels. When the annular defect on the wafer is not matched with the machine 1, the second dynamic maintenance level of the machine 1 can be identified as a second level; otherwise, the second dynamic maintenance level of the machine may be identified as a level.

Correspondingly, the invention also provides a scheduling system (or a machine maintenance monitoring system) for performing early maintenance on the machine, as shown in fig. 2, including:

a static rating module 10 configured to determine a static maintenance level of at least one machine from the static database; wherein, the static database comprises: setting maintenance time of a plurality of machine stations; the static maintenance grade of the machine is divided into a first grade, a second grade and a third grade according to the interval between the current time and the set maintenance time;

A first dynamic rating module 11 configured to determine a first dynamic maintenance level of the machine according to a first dynamic database; the first dynamic database includes one or more of the following operational data: voltage operation data, current operation data and standard data corresponding to the voltage operation data and the current operation data; the working data are used for representing the change relation of the working value of the machine along with the working time; correspondingly, the first dynamic rating module comprises:

a data acquisition unit 111 configured to acquire at least one piece of work data of the machine, and standard data corresponding to the work data;

an anomaly monitoring unit 112 configured to treat a data point of the working data as an anomaly point when it is detected that a difference between a working value of the data point and the standard data is greater than a preset difference threshold;

a dynamic rating unit 113 configured to determine a first dynamic maintenance level from the outlier; when the number of abnormal points belongs to a first set threshold value, the first dynamic maintenance level is one level; when the number of abnormal points belongs to a second set threshold value, the first dynamic maintenance level is two levels; when the number of abnormal points belongs to a third set threshold value, the first dynamic maintenance level is three-level;

The comprehensive rating module 13 is configured to determine a comprehensive maintenance level of the corresponding machine platform according to the static maintenance level and the first dynamic maintenance level; when the static maintenance level is one level and the first dynamic maintenance level is one level, the comprehensive maintenance level is one level; when the static maintenance level is a second level and the first dynamic maintenance level is a second level, the comprehensive maintenance level is a second level; when the static maintenance level is three-level and the first dynamic maintenance level is three-level, the comprehensive maintenance level is three-level;

the maintenance ordering module 14 is configured to order the maintenance orders of the machine according to the comprehensive maintenance level.

In some embodiments, when the static maintenance level is a first level and the first dynamic maintenance level is a second level, the integrated maintenance level is a second level; when the static maintenance level is one level and the first dynamic maintenance level is three levels, the comprehensive maintenance level is three levels; when the static maintenance level is two-level and the first dynamic maintenance level is one-level, the comprehensive maintenance level is one-level; when the static maintenance level is two-level and the first dynamic maintenance level is three-level, the comprehensive maintenance level is three-level; when the static maintenance level is three-level and the first dynamic maintenance level is one-level, the comprehensive maintenance level is two-level; and when the static maintenance level is three-level and the first dynamic maintenance level is two-level, the comprehensive maintenance level is three-level.

In some embodiments, further comprising:

a second dynamic rating module 15 configured to determine a second dynamic maintenance level of the machine according to a second dynamic database, wherein the second dynamic database comprises: the second dynamic rating module is further configured to select product data associated with the machine from the second dynamic database as dynamic adjustment data, and divide the second dynamic maintenance class into a first class, a second class and a third class according to the quality of the wafer according to the dynamic adjustment data;

a first dynamic rating modification module 16 configured to verify or adjust the first dynamic maintenance level in accordance with the second dynamic maintenance level; when the second dynamic maintenance level is smaller than or equal to the first dynamic maintenance level, the original level of the second dynamic maintenance level is maintained; when the second dynamic maintenance level is greater than the first dynamic maintenance level, the first dynamic maintenance level is adjusted according to the second dynamic maintenance level; wherein when the second dynamic maintenance level is two or three stages and the first dynamic maintenance level is one stage, the first dynamic maintenance level is adjusted to be two stages; when the second dynamic maintenance level is three-level and the first dynamic maintenance level is two-level, the first dynamic maintenance level is adjusted to three-level.

In some embodiments, the product data includes one or more of the following: annular defects, linear defects, ray defects; the machine in the static database is marked with a corresponding machine label; the machine label comprises: the first machine label is used for executing rotation operation in the wafer processing process of a machine corresponding to the first machine label; the second machine label is used for clamping the wafer in the wafer processing process; the third machine label is used for executing air blowing operation in the wafer processing process;

correspondingly, the second dynamic rating module is further configured to select dynamic adjustment data corresponding to the machine label according to the recommendation priority of the product data.

In some embodiments, for a first type of machine, the first dynamic database further comprises: rotational speed operation data, and standard data corresponding to the rotational speed operation data.

Exemplary identification schemes for straight line defects, annular defects will be described below:

example two

In the early stages of the development of the machine fault, the form of the fault problem on the wafer (or the defect pattern) may not be obvious, and in this case, if the engineer is less experienced, the wafer with the fault may be ignored, so that the hidden trouble is eliminated. Moreover, due to the incomplete nature of the defect pattern, it is difficult to obtain accurate results when analyzing using defect recognition or comparison patterns. Correspondingly, the invention provides a method for quickly identifying defects, which is used for assisting in evaluating the dynamic maintenance grade of a machine, as shown in fig. 3, and takes annular defects as an example, and comprises the following steps:

S201, obtaining a plurality of first defect graphs of a plurality of sample wafers, and performing defect classification on the plurality of first defect graphs by adopting a defect classification model to obtain at least one first graph set, wherein the first graph set comprises: at least one defect pattern having a ring defect; at this time, the first graphic set is preliminarily identified as having the same annular defect;

the defect classification model can be obtained by training a plurality of defect graph sets of a plurality of annular defects of different types through a neural network modeling method. Such as a defect classification model, preferably a ring defect classification model, may be an ADC classification model.

S202, acquiring actual production information, wherein the actual production information comprises: acquiring maintenance level (equivalent to risk level) of at least one machine experienced by the sample wafer in the processing process and/or the number of patterns of the first pattern set through a machine maintenance monitoring system;

in some embodiments, the tool maintenance monitoring system may be the scheduling system described above, which may automatically rank the maintenance level of the historical processing tools of the wafer.

For example, in some embodiments, a "maintenance level" may refer to an integrated maintenance level that is determined in accordance with both the static maintenance level and the first dynamic maintenance level, as described above, as well as the first dynamic maintenance level, as described above.

Alternatively, in other embodiments, the tool maintenance monitoring system may be automatically entered by the user. For example, when the user considers that the machine has a fault, the first input signal can be directly input to the machine maintenance monitoring system; wherein the first input signal comprises: machine information, and maintenance level of the machine (e.g., when a user considers that the machine has a fault, the maintenance level may be directly adjusted to three levels); the machine maintenance monitoring system then determines a maintenance level of the machine in response to the input first input signal.

For another example, when the user does not find that the machine has a fault temporarily, the second input signal may be directly input to the machine, where the maintenance level in the second input signal is two or one.

The fault type may refer to a machine type that causes a fault to occur.

Herein, "overlay" may also be referred to as "graphic fusion" and refers to the superposition of at least two graphics (or, alternatively, of a marking element on a graphic) to form a new graphic. The mode selected by the stacking process can be one or more of the following modes: pixel level stacking, region level stacking, mixed mode stacking, mixed channel stacking methods, and the like.

Herein, "fitting" refers to employing data enhancement methods (e.g., pattern inversion, rotation, magnification, cropping, etc.) to fit calculations to complete defect geometries (e.g., complete ring defects) using local defect geometries (e.g., defect points of defects).

Acquiring a preset first machine characteristic library associated with the defect graph; the first machine characteristic library comprises: a first set of quantized features comprising one or more of the following defect feature data: the rotation radius of the mechanical arm, the rotation speed of the mechanical arm, the distance between the mechanical arm and the wafer center and the position of an adsorption foot of the adsorption equipment are respectively associated with corresponding wafer machine information;

and matching the shape data with the machine characteristic library, and identifying the machine opposite to the shape data as a corresponding fault machine when the shape data is matched with the corresponding defect characteristic data.

In some embodiments, the machine feature library further comprises: a first set of graphical features, the set of graphical features comprising: and defect characteristic patterns corresponding to defects caused by at least one machine or equipment parts in the machine.

In some embodiments, when the machine is not matched, a second prompt signal may also be sent to the user for judgment by human intervention.

It can be understood that an engineer can construct a machine feature library in advance by combining engineering experience, wherein the quantized features in the machine feature library can be obtained from specification parameters of the machine, the defect feature pattern can be obtained from historical production data, and typical defect feature patterns possibly generated by the machine are bound with the machine so as to facilitate the positioning of machine faults.

In some embodiments, the step of S201 further comprises: and marking defect points in the defect graphs in the first graph set, wherein the defect points are points identified as annular defects by the defect classification model. In this embodiment, the marked defect point corresponds to the marking element.

In some embodiments, the first processing condition refers to the number of graphics being greater than or equal to a first set number, and the second processing condition refers to the number of graphics being less than the first set number.

Alternatively, in other embodiments, the first processing condition refers to the maintenance level (corresponding to the risk level) being less than or equal to a set maintenance level (e.g., when the maintenance level is one level); the second process condition means that the maintenance level is greater than a set maintenance level (e.g., the maintenance level is two or three levels).

In this embodiment, the operation condition (such as the risk level of the machine) and the actual defect condition (such as the number of the classified defect patterns) of the machine are introduced to automatically select the processing mode of the defect patterns. And further, the defect identification or analysis mode can be flexibly adjusted in the face of different actual production scenes.

In addition, at the initial stage of fault hidden danger, the number of defects is relatively limited, defects of the defect patterns are large, and the probability of misjudgment of the manual or automatic matching model is high. The application provides a graphic processing scheme based on machine and product data fusion guidance, which can improve the defect positioning efficiency to a certain extent and ensure the reliability of the defect positioning result.

In some embodiments, when the classification accuracy of the defect classification model is relatively low, before S203, the method further includes the steps of:

wherein, the standard pitch may refer to an average pitch value of a plurality of defect sample groups;

alternatively, each defect sample group may be calculated to obtain a plurality of pitches, and when the number of defect sample groups belonging to the same pitch is the largest, the corresponding pitch is taken as the standard pitch.

In some embodiments, S203 comprises:

The embodiment provides a man-machine cooperative semi-automatic identification mode, which can be used for carrying out automatic graphic processing selection and graphic processing through a rapid evaluation mode (such as setting of first and second processing conditions), and can be used for capturing key problems sharply and introducing manual borrowing. And therefore, the defect graph is rapidly and accurately processed by utilizing a man-machine cooperative mode.

In this embodiment, the defect types are initially classified, and the matching between the defects and the machine is quickly performed by using a semi-automatic mode, so as to assist the user in quickly determining the risk degree of the fault or hidden trouble of the machine. Furthermore, the actual wafer product data can be quickly fed back to the machine maintenance stage to quickly maintain and grade the running state of the machine.

For example, in some embodiments, the method further comprises the step of:

obtaining the defect number of defects generated by one machine;

determining a second dynamic maintenance level of the machine according to the defect number (which can be used for preliminarily reacting the wafer quality);

when the defect number of one machine table belongs to a first defect threshold range, the second dynamic maintenance level of the machine table is one level; when the defect number of one machine table belongs to a second defect threshold range, the second dynamic maintenance level of the machine table is two-level, and when the defect number of one machine table belongs to a third defect threshold range, the second dynamic maintenance level of the machine table is three-level.

In this embodiment, according to the rapid identification and positioning of the ring-shaped defect, in addition to assisting the user (such as an engineer) in primarily classifying the defect, a referential rating factor (such as a second dynamic maintenance level) can be provided to a scheduling system (or a machine maintenance monitoring system) for machine maintenance, so as to implement comprehensive evaluation on the maintenance sequence of the machine from a plurality of maintenance factors such as a static machine maintenance period, an actual machine running state, and final wafer product quality, so that the failure or hidden danger of the machine can be rapidly checked under the condition that the selected rating factor is relatively limited.

It will be appreciated that one or more of the method steps of the present embodiments may also be applicable to the identification of straight line defects.

The embodiment also correspondingly provides a ring defect identification system, which comprises:

the annular defect classification module 21 is configured to obtain a plurality of first defect patterns of a plurality of sample wafers, and perform defect classification on the plurality of first defect patterns by adopting a defect classification model to obtain at least one first pattern set, wherein the first pattern set comprises: at least one defect pattern having a ring defect;

the actual production information acquisition module 22 is configured to acquire actual production information, and the actual production information includes: the method comprises the steps of obtaining maintenance grade of at least one machine and the number of patterns of a first pattern set, wherein the maintenance grade is experienced by a sample wafer in the processing process, through a machine maintenance monitoring system;

A processing mode selection module 23, configured to select a recommended graphics processing mode according to the actual production information, where the graphics processing mode includes: stacking and fitting; when the actual production information accords with the first processing condition, the recommended graphic processing mode is a superposition graph; and/or when the actual production information accords with the second processing condition, the recommended graphic processing mode is fitting; wherein,

a graphics processing module 24 configured to select a recommended graphics processing mode to perform graphics processing on the first graphics set to obtain a second defect graphics;

the fault judging module 25 is configured to judge a fault type causing the ring defect according to the second defect pattern.

In some embodiments, the fault types include: fault machine information; the failure determination module 25 includes:

A shape data acquisition unit 251 configured to acquire shape data of the defect through the second defect pattern, the shape data including one or more of: defect length, defect radian, defect radius, and gap between defect and wafer center;

a feature library obtaining unit 252 configured to obtain a preset machine feature library associated with the defect pattern; the machine characteristic library comprises: a set of quantized features comprising one or more of the following defect feature data: the rotation radius of the mechanical arm, the rotation speed of the mechanical arm, the distance between the mechanical arm and the wafer center and the position of an adsorption foot of the adsorption equipment are respectively associated with corresponding wafer machine information;

the first matching unit 253 is configured to match the defect feature data set with the machine feature library, and identify a machine opposite to the defect feature data as a corresponding faulty machine when the defect feature data matches the corresponding defect feature data.

In some embodiments, the system further comprises: a filtering module configured to select a plurality of defect sample groups from the marked defect points, and one defect sample group includes: one or more defect points; calculating the distances between the defect sample groups and the wafer centers; when the difference value between the distance between the defect sample group and the circle center and the standard distance is larger than a first difference value threshold value, identifying the corresponding defect sample group as an error group; and when the number of the error groups is greater than an error threshold value, screening the corresponding first defect graph from the first graph set to obtain a new first graph set.

In some embodiments, the system further comprises: the user adjusting module is configured to send a first prompt signal to a user when the actual production information does not accord with the first processing condition or the second processing condition; and receiving a first selection signal issued by a user, the first selection signal comprising: and the graphics processing mode information is used for determining a recommended graphics processing mode according to the first selection signal.

In some embodiments, the system further comprises: the second dynamic rating module is further configured to obtain the defect number of the defects generated by one machine; determining a second dynamic maintenance level of the machine according to the defect number (which can be used for preliminarily reacting the wafer quality); when the defect number of one machine table belongs to a first defect threshold range, the second dynamic maintenance level of the machine table is one level; when the defect number of one machine table belongs to a second defect threshold range, the second dynamic maintenance level of the machine table is two-level, and when the defect number of one machine table belongs to a third defect threshold range, the second dynamic maintenance level of the machine table is three-level.

Example III

The invention also provides a rapid defect identification method, which is described by taking a straight line defect as an example, as shown in fig. 4, and comprises the following steps:

s301, acquiring a plurality of first defect patterns of a plurality of sample wafers, and performing defect classification on the plurality of first defect patterns by adopting a linear defect classification model to obtain at least one second pattern set, wherein the second pattern set comprises: at least one defect pattern having a straight line defect; wherein the straight line defect includes one or more of the following types: linear scratches, linear particle defects, and radial particle defects;

In some embodiments, the linear defect classification model may be obtained by selecting a defect pattern set of a plurality of different types of linear defects as a sample and training the defect pattern set through a neural network modeling method.

In some embodiments, the linear defect classification model may also be directly selected from the existing ADC classification model.

S302, acquiring a pre-stored second feature database associated with the first defect graph, wherein the feature database comprises: a second set of quantization features, comprising: workpiece size of equipment components in at least one machine; and/or a second set of graphical features comprising: a defect signature of a wafer defect caused by at least one machine or equipment component of the machine;

for example, in some embodiments, the workpiece size may be the width of the clamp.

For example, in some embodiments, the defect signature may be a typical defect photograph taken by a user from historical production data.

S303, matching the corresponding defect graph with the second characteristic database, and determining a fault type which causes the linear defect to be generated according to at least one machine when the corresponding defect graph is matched with the at least one machine; wherein,

the matching process comprises the following steps: when the linear defect is a linear scratch, obtaining defect characteristics of a corresponding defect graph, wherein the defect characteristics comprise: the distance between the scratch and the circle center, and/or the scratch length, and/or the distance between at least two scratches; matching the defect feature with a second quantitative feature set according to the defect feature;

Alternatively, the matching process includes: when the linear defect is a linear particle defect, the linear particle defect; and performing similarity comparison on the defect graph and the second graph feature set by using an image recognition model (or a similarity comparison model), and considering that the defect graph and the second graph feature set are matched when the similarity between the defect feature graph and the defect graph is larger than the preset similarity.

In some embodiments, when the defect pattern is successfully matched with the multiple machine platforms, a List of machine platforms can be provided for the engineer according to the specific result of similarity comparison, and the List is arranged from high to low according to the similarity.

In some embodiments, when matching to at least two machines, at least: when the first machine and the second machine are used, the method further comprises the steps of:

acquiring a positioning point of the wafer when the wafer enters the first machine, wherein the positioning point is used for representing the processing angle of the wafer entering the machine;

determining a forming azimuth of the linear defect according to the locating point, wherein the forming azimuth refers to the direction from the starting position to the ending position of the linear defect;

obtaining a plurality of sample defect groups from the linear defects, and calculating defect point density of each sample defect group; wherein one sample defect group comprises: one or more defect points;

When the defect density of the sample defect group gradually increases along the forming direction, the linear defect is primarily considered to be generated by the first machine.

If not, the linear defect can be primarily determined to be caused by the second machine or other machines.

Or, the positioning point of the second machine can be further used for judging whether the current linear defect is related to the second machine.

For example, as shown in fig. 8, in some embodiments, when a wafer enters the machine 2, positioning points (such as notches provided at the edge of the wafer) are marked on the wafer, and a coordinate system is constructed according to the directions of the positioning points so as to facilitate a user (such as an engineer) to determine whether the wafer pattern is offset. The Y-axis may be established in the pointing direction of the notch and the X-axis may be established in a direction parallel to the pointing direction. At this time, if it is observed that the defect density (e.g., the distribution density of particles) of the wafer gradually decreases along the Y-axis direction, it is considered that the current defect may be caused by the machine 2.

In some embodiments, when the second graphic set does not match the machine, then the method comprises the steps of:

s304, acquiring actual production information, wherein the actual production information comprises: the maintenance grade of at least one machine which is experienced by the sample wafer in the processing process and the number of the patterns of the second pattern set are obtained through a machine maintenance monitoring system;

S305, selecting a recommended graphic processing mode according to the actual production information, wherein the graphic processing mode comprises the following steps: stacking and fitting; when the actual production information accords with the first processing condition, the recommended graphic processing mode is a superposition graph; and/or when the actual production information accords with the second processing condition, the recommended graphic processing mode is fitting; wherein,

s306, selecting a recommended graphic processing mode to perform graphic processing on the second graphic set so as to obtain a third defect graphic.

In some embodiments, in S301, the method further comprises the steps of: and marking defect points in the defect graphs in the second graph set, wherein the defect points are points which are identified as linear defects by the defect classification model.

In this embodiment, the comprehensive determination is preferably performed by using static data and dynamic data (such as machine operation data or preliminary defect classification results), so that the actual production information is used to guide the rapid identification of defects.

In some embodiments, when the actual production information does not meet the first processing condition or the second processing condition, a first prompt signal is sent to a user;

In this embodiment, for the critical stage (graphics processing stage) of the line defect recognition, a semi-automatic mode is used to select the graphics processing mode, so as to perform semi-automatic graphics processing. The defect pattern analysis method can be used for rapidly analyzing and identifying the defect pattern by taking actual production data as guidance, and simultaneously can timely introduce manual borrowing under special conditions so as to ensure the efficiency and accuracy of defect identification.

The embodiment also correspondingly provides a system for identifying the linear defect, which comprises the following steps:

the linear defect classification module 31 is configured to obtain a plurality of first defect patterns of a plurality of sample wafers, and perform defect classification on the plurality of first defect patterns by adopting a linear defect classification model to obtain at least one second pattern set, where the second pattern set includes: at least one defect pattern having a straight line defect; wherein the straight line defect includes one or more of the following types: linear scratches, linear particle defects, and radial particle defects;

A second feature data acquisition module 32 configured to acquire a pre-stored second feature database associated with the first defect map, wherein the feature database comprises: a second set of quantization features, comprising: workpiece size of equipment components in at least one machine; and/or a second set of graphical features comprising: a defect signature of a wafer defect caused by at least one machine or equipment component of the machine;

a linear defect matching module 33 configured to match a corresponding defect pattern with the second feature database, and when at least one machine is matched, determine a fault type that causes the linear defect to be generated according to the machine; wherein,

alternatively, the matching process includes: when the linear defect is a linear particle defect, the linear particle defect; and comparing the similarity between the defect pattern and the second pattern feature set by using an image recognition model, and when the similarity between the defect feature pattern and the defect pattern is greater than the preset similarity, considering that the defect feature pattern and the defect pattern are matched.

In some embodiments, the straight line defect matching module 33 includes:

the first positioning unit 331 is configured to obtain, when the first positioning unit is matched with at least two machines, positioning points of the wafer when the wafer enters the first machine, where the positioning points are used to represent a processing angle of the wafer entering the machine, and the at least two machines include: the first machine table and the second machine table;

a second positioning unit 332, configured to determine a forming direction of the linear defect according to the positioning point, where the forming direction refers to a direction from a start position to an end position of the linear defect;

a second matching unit 333 configured to acquire a plurality of sample defect groups from the linear defects, and calculate defect point densities of the respective sample defect groups; wherein one sample defect group comprises: one or more defect points; when the defect density of the sample defect group gradually increases along the forming direction, the linear defect is primarily considered to be generated by the first machine.

In some embodiments, the system further comprises:

a production information acquisition module configured to acquire actual production information when the second graphic set is not matched to a machine, and the actual production information includes: the maintenance grade of at least one machine and the number of patterns of the second pattern set, which are acquired by the machine maintenance monitoring system and used for the sample wafer in the processing process;

The graphic processing mode selection and acquisition module is configured to select a recommended graphic processing mode according to the actual production information, and the graphic processing mode comprises: stacking and fitting; when the actual production information accords with the first processing condition, the recommended graphic processing mode is a superposition graph; and/or when the actual production information accords with the second processing condition, the recommended graphic processing mode is fitting; wherein,

and the graphic processing module is configured to select a recommended graphic processing mode to perform graphic processing on the second graphic set so as to obtain a third defect graphic.

In some embodiments, the linear defect classification module marks defect points in the defect patterns in the second pattern set, wherein the defect points are points identified as linear defects by the defect classification model.

The present invention also provides a computer device comprising: a processor; and a memory for storing the processor-executable instructions; wherein the processor is configured to invoke the instructions stored in the memory to perform the method of any of the embodiments described above.

The invention also provides a computer program product configured to store computer readable instructions which, when executed, cause a computer to perform the method described in any of the embodiments above.

The invention also provides a computer readable storage medium having stored thereon computer program instructions which when executed by a processor implement the method as described in any of the above embodiments.

From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) comprising several instructions for causing a computer terminal (which may be a mobile phone, a computer, a server, or a network device, etc.) to perform the method according to the embodiments of the present invention.

The embodiments of the present invention have been described above with reference to the accompanying drawings, but the present invention is not limited to the above-described embodiments, which are merely illustrative and not restrictive, and many forms may be made by those having ordinary skill in the art without departing from the spirit of the present invention and the scope of the claims, which are to be protected by the present invention.

Claims

1. A method of identifying an annular defect, comprising the steps of:

s203, selecting a recommended graphic processing mode according to the actual production information, wherein the graphic processing mode comprises the following steps: stacking and fitting; when the actual production information accords with the first processing condition, the recommended graphic processing mode is a superposition graph; and/or when the actual production information accords with the second processing condition, the recommended graphic processing mode is fitting; wherein the first processing condition means that the maintenance level is less than or equal to a set maintenance level and the number of figures is greater than or equal to a first set number; the second processing condition means that the maintenance level is greater than a set maintenance level and the number of patterns is less than a first set number;

2. The method for identifying an annular defect according to claim 1, wherein the fault type comprises: fault machine information; s205 includes the steps of:

3. The method for identifying an annular defect according to claim 2, wherein the machine feature library further comprises: a set of graphical features, the set of graphical features comprising: and defect characteristic patterns corresponding to defects caused by at least one machine or equipment parts in the machine.

4. The method for identifying an annular defect according to claim 1, wherein S201 further comprises the steps of: and marking defect points in the defect graphs in the first graph set, wherein the defect points are points identified as annular defects by the defect classification model.

5. The method of claim 4, further comprising the step of, prior to S203:

selecting a plurality of defect sample groups from the marked defect points, wherein one defect sample group comprises: one or more defect points; calculating the distances between the defect sample groups and the wafer centers; when the difference value between the distance between the defect sample group and the circle center and the standard distance is larger than a first difference value threshold value, identifying the corresponding defect sample group as an error group;

6. The method for identifying an annular defect according to claim 1, wherein S203 comprises:

when the actual production information does not accord with the first processing condition or the second processing condition, a first prompt signal is sent to a user; and receiving a first selection signal issued by a user, the first selection signal comprising: and the graphics processing mode information is used for determining a recommended graphics processing mode according to the first selection signal.

7. A system for identifying annular defects, comprising:

the actual production information acquisition module is configured to acquire actual production information, and the actual production information comprises: the maintenance grade of at least one machine and the number of patterns of the first pattern set, which are experienced by the sample wafer in the processing process, are acquired through the machine maintenance monitoring system;

8. The system of claim 7, wherein the fault type comprises: fault machine information; the fault judging module comprises:

9. The system of claim 8, wherein the machine feature library further comprises: a set of graphical features, the set of graphical features comprising: and defect characteristic patterns corresponding to defects caused by at least one machine or equipment parts in the machine.

10. The system of any of claims 7-9, wherein the annular defect classification module is further configured to mark defect points in the defect patterns in the first pattern set, wherein a defect point is a point identified as an annular defect by the defect classification model.