KR102408711B1 - Method and apparatus for predicting fault types of wafers based on clusters - Google Patents

Abstract

The present invention provides a method for predicting a type of wafer failure based on cluster analysis performed by a computer, comprising: collecting wafer inspection data in the form of images or structured data; determining whether the wafer is defective based on the collected wafer inspection data; Separating and storing good wafer data and bad wafer data according to the determination result; clustering defect types with the collected defective wafer data; and analyzing the correlation between the clustered defect types and matching environmental data.

Description

Method and apparatus for predicting fault types of wafers based on clusters

The present invention relates to an apparatus and method for predicting wafer defect types using cluster analysis.

Since semiconductors are essential elements for various electronic products, in order to improve the productivity of semiconductors, it is necessary to improve facility diagnosis, process management, yield stabilization, etc., and to develop advanced process technologies. and is the most important in the semiconductor industry. The complex semiconductor wafer manufacturing process may have some defects and, depending on the type of defect, may cause failure in the production of the final product.

Accordingly, in the semiconductor wafer manufacturing process, producers produce good products through various inspection processes. For example, vibration, temperature, humidity, vibration, equipment status, etc. are continuously managed by the Internet of Things (IoT). In addition, the failure is predicted based on the upper and lower limits of the sensing value of the IoT sensor. However, it is difficult to categorize a lot of defective information with only the sensing value of the IoT sensor.

Therefore, there is a need for a technique capable of predicting the type of wafer defect in the semiconductor wafer manufacturing process.

In order to solve the above-described problems, it is possible to provide a method and an apparatus capable of predicting the type of wafer defect by cluster analysis of semiconductor inspection data.

According to the present invention for achieving the above object, there is provided a method for predicting a wafer defect type based on cluster analysis performed by computing, comprising: collecting wafer inspection data in the form of images or structured data; determining whether the wafer is defective based on the collected wafer inspection data; Separating and storing good wafer data and bad wafer data according to the determination result; clustering defect types with the collected defective wafer data; and analyzing a correlation between the clustered defect types and matching environmental data.

In an embodiment, the wafer data in the form of an image is a wafer map, and the wafer data in the form of the structured data includes information on a location where a defect of the wafer occurs.

In an embodiment, the separately storing the good wafer data and the bad wafer data may include storing the good wafer data in a storage unit and storing the bad wafer data in a bad list storage unit.

In an embodiment, the clustering of the defect types is characterized in that information about the location, angle, and thickness of the wafer included in the defective wafer data is clustered through a technique such as K-means or DBSCAN.

In one embodiment, the step of analyzing the correlation of the environmental data is by applying 4M1E (Man, Machine, Material, method & Environment) analysis to each cluster clustered with the location, angle, and thickness information of the wafer defect. It is characterized by generating a pattern for each defect type by analyzing the correlation.

A wafer defect type prediction apparatus based on another embodiment of the present invention includes: a data collection unit for collecting wafer inspection data in the form of images or structured data; a defect determination unit that determines whether the wafer is defective based on the collected wafer inspection data; a storage unit for storing data of non-defective wafers according to the determination result; a defect list storage unit for storing defective wafer data according to the determination result; a clustering unit for clustering defective types with the collected defective wafer data; and a defect pattern extracting unit analyzing a correlation between the clustered defect types and matching environmental data.

In an embodiment, the clustering unit is characterized in that the information about the location, angle, and thickness of the wafer included in the defective wafer data is clustered through a technique such as K-means or DBSCAN.

In one embodiment, the defect pattern extraction unit analyzes the correlation by applying 4M1E (Man, Machine, Material, method & Environment) analysis to each cluster clustered with the location, angle, and thickness information of the wafer defect by analyzing the correlation. It is characterized in that the pattern for each type is generated.

According to an embodiment of the present invention, it is possible to increase the efficiency of production management by predicting the type of defect according to the location of the defect of the wafer sensed in the semiconductor wafer manufacturing process.

In addition, by determining the correlation between the predicted defect types and environmental factors, it is possible to contribute to productivity improvement by reducing defect management costs and reducing defect rates.

In addition, the effects obtainable in the present disclosure are not limited to the aforementioned effects, and other effects not mentioned may be clearly understood by those of ordinary skill in the art to which the present disclosure belongs from the description below. .

1 is a block diagram illustrating an apparatus for predicting a wafer defect type according to an embodiment of the present disclosure.
2 is a flowchart illustrating a method for predicting a wafer defect type according to an embodiment of the present invention.
3 is a flowchart illustrating clustering according to an embodiment of the present invention.
4 is an example of a data sample for a defect type according to an embodiment of the present invention.
5 is an example of a data sample showing a clustering target of a defective type by lot (LoT) according to an embodiment of the present invention.
6 is a flowchart illustrating extraction of a bad pattern according to an embodiment of the present invention.

Since the present invention can have various changes and can have various embodiments, specific embodiments will be described in detail with reference to the drawings. However, this is not intended to limit the present invention to specific embodiments, it should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention. In describing each figure, like reference numerals have been used for like elements.

The terms used in the present application are only used to describe specific embodiments, and are not intended to limit the present invention. The singular expression includes the plural expression unless the context clearly dictates otherwise. In the present application, terms such as “comprise” or “have” are intended to designate that a feature, number, step, operation, component, part, or combination thereof described in the specification exists, but one or more other features It should be understood that this does not preclude the existence or addition of numbers, steps, operations, components, parts, or combinations thereof.

Unless defined otherwise, all terms used herein, including technical and scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related art, and should not be interpreted in an ideal or excessively formal meaning unless explicitly defined in the present application. does not

Throughout the specification and claims, when a part includes a certain element, it means that other elements may be further included, rather than excluding other elements, unless specifically stated to the contrary.

1 is a block diagram illustrating an apparatus for predicting a wafer defect type according to an embodiment of the present disclosure.

The wafer defect type prediction apparatus 100 includes a data collection unit 110 , a defect determination unit 120 , a storage unit 130 , a defect list storage unit 140 , a clustering unit 150 , and a defect pattern extraction unit 160 . includes

The data collection unit 110 collects wafer inspection data in the form of image data or structured data generated by the semiconductor inspection apparatus.

In one embodiment, the image data may be a wafer map. The shaping data may include defect location information of the wafer.

In an embodiment, the data collection unit 110 may collect wafer inspection data from a database of the semiconductor inspection apparatus. In another embodiment, the data collection unit 110 may collect wafer inspection data directly from the semiconductor inspection apparatus.

The defect determination unit 120 determines whether the wafer is defective based on the collected wafer inspection data.

The failure determination unit 120 determines whether the image data is defective, and determines whether the formatted data is defective.

In an embodiment, the failure determining unit 120 classifies information according to data types, and then divides data of each data type into good data and bad data.

In an embodiment, when the image data is a wafer map, the defect determination unit 120 may compare the image data with a pre-stored defect pattern and classify the image data into good data and bad data. In addition, it is possible to divide the data into good data and bad data based on the defective position information of the fixed data. This is not limited thereto as an example.

The storage unit 130 stores the data of the non-defective wafer according to the determination result. That is, the storage unit 130 stores the non-defective data of the first data type and the non-defective data of the second data type. The stored good wafer data is managed and utilized as information assets of the manufacturer.

The defective list storage unit 140 stores defective wafer data according to the determination result.

The clustering unit 150 groups defect types using the defective wafer data collected in the defect list storage unit 140 .

The defect pattern extractor 160 analyzes the correlation between the clustered defect types and matching environmental data. An example of the analyzed data is shown in Table 1 below.

breakage type broken person breakage factor Failure due to thermal and mechanical stresses and strains silicon chip die crack
Expansion destruction by moisture absorption
interfacial peel
package warpage
Breakage of the insulation layer in the device
Solder joint breakage
circuit board damage
crater breakage
wire breakage
breakage of metal wiring surface irregularities
Package thinning trend
High-temperature reflow
mechanical stress, impact
difference in coefficient of thermal expansion
periodic fatigue
Stress due to hygroscopic thermal expansion Failure at the metal joint interface intermetallic diffusion voids
Breakage in metal-to-metal junctions
short circuit in metal bump Difference in diffusion rate between metal junctions
bfittle intermetallic
metal ductility Surface contamination, chemical reaction surface contamination
surface discoloration
corrosion
metal oxidation
poor adhesion
Decreased solder wettability corrosive environment
temperature
relative humidity
Foreign matter in the process
surface roughness
Wettability Interfering Contaminants electrical or electrochemical transfer metal ion growth
movement in metal by electric current
Electroless metal crystal growth ionic impurities
Current density distribution
Densification of packages and wiring
temperature and humidity conditions

2 is a flowchart illustrating a method for predicting a wafer defect type according to an embodiment of the present invention.

The wafer defect type prediction apparatus of FIG. 1 performs the steps as shown in FIG. 2 .

In step S110, the data collection unit 110 collects measured data according to wafer inspection from various inspection equipment. The data to be collected has an image data format and a structured data format.

The image data may be a wafer map, and the shaping data may include information on defective locations of the wafer. The defect location information may be in the form of x,y coordinates of the defect.

The data collection unit 110 may collect equipment data.

In step S120 , the defect determination unit 120 determines whether the wafer is defective based on the collected wafer inspection data, and divides the wafer into good data and defective data. The processor unit 120 transmits and stores the good product data to the storage unit 130 , and transmits the defective data to the defective list extractor 140 . Accordingly, the storage unit 130 stores the data of the good wafer, and the defective list extractor 140 stores the bad wafer data.

In step S130 , the clustering unit 150 clusters each defect type based on the data stored in the defect list extracting unit 140 .

In an embodiment, the clustering unit 150 groups the defective position (Wafer_NO), angle, and thickness information of the wafer through a technique such as K-means or DBSCAN. With this, it is possible to discover key points of management at the manufacturing site based on 4M1E information on the location, angle, and thickness of wafer defects. The clustering method of such a clustering unit will be described later with reference to FIG. 3 .

In step S140 , the defect pattern extraction unit 160 analyzes the correlation between the clustered defect types and matching environmental data.

The defect pattern extraction unit 160 finds defects in a new pattern that did not exist before through the results of clustering, and analyzes the defect types and indicators with facility data and sensor data (temperature, humidity, particles, vibration, etc.) to create new Find the variables that affect the defect type.

3 is a flowchart illustrating clustering according to an embodiment of the present invention.

Referring to FIG. 3 , in step S131 , the clustering unit 150 performs primary clustering based on data stored in the defective list extraction unit 140 . In this case, the primary clustering criterion may be any one or more of thickness, angle, and position (x, y).

Next, in step S132, the secondary clustering is performed by analyzing the characteristics of each cluster based on the result of the primary clustering.

In step S133, defect types are divided according to the secondary clustering result, and newly collected defect types and indicators are generated. As shown in FIG. 4, the data for the generated defect type may include a wafer lot number, SEQ number, wafer ID, defect mode (BAD_MODE), and defect mode quantity (BAD_MODE_QTY) items. .

The clustering unit 150 may adopt a DBSCAN algorithm for clustering. The DBSCAN algorithm finds points in a dense region in the feature space. If data is contained within the specified distance (eps) from one randomly selected data point as much as the minimum number of samples (min_samples), this data point is classified as a core sample. If it is less than the minimum number of samples, it is labeled as noise. In this way, the cluster grows until there are no more key samples within the specified distance. Then, randomly select a point in a densely populated area that has not been visited, and repeat the same process for group evolution. Referring to FIG. 5 , an example of a cluster target of a defective type for each LoT can be confirmed. By clustering different types of defects by lot, you can create indicators of defects.

In step S134, it is determined whether the generated defect type and index can be a key management element for each cluster.

6 is a flowchart illustrating extraction of a bad pattern according to an embodiment of the present invention.

Referring to FIG. 6 , in step S141, environmental factors are collected by collecting equipment data and/or sensor data, and the extracted environmental factors are matched with new defect types and indicators and stored.

In step S142, the correlation between the plurality of feature values extracted from the clustering result and the environmental factor is analyzed. Correlation analysis can be performed using a chi-square test. The chi-square test is a statistical method based on the chi-square distribution, and is used to test whether the observed frequency is significantly different from the expected frequency.

The chi-square value is calculated as χ ² = Σ (observed value - expected value) 2 / expected value.

In step S143, a defect type pattern is extracted based on the result of the correlation analysis. Through One-Way-ANOVA analysis, new defect patterns are extracted when the results of clustering are considered together with environmental factors. Here, the independent variable, which is a nominal scale, becomes the cluster number, and the dependent variable, which is an interval scale or a ratio scale, becomes 4M1E. Here, 4M1E refers to the elements of the air conditioning environment such as man, machine, material, method, environment temperature and humidity of the clean room, and the concentration of particles (fine dust).

In step S144, a management key point is determined according to the defect type based on 4M1E. To this end, the management key points for each defect type may be stored in advance.

The type of defect may be foreign material contamination, chip breakage, metal oxidation, etc., and the key points for each management are in order, improvement of the air conditioning environment (Environment_ eg, fine dust) in the clean room, Machine) management and process management (Method) may be the cause of foreign substances in the process.

For example, if the defect type is foreign body contamination, based on 4M1E, sodium contamination by sweat of MAN, organic contamination by skin peeling, cosmetics, etc., particle contamination by movement, conversation, etc. Caution, rash of MACHINE As a management key point, it is possible to output contamination from energy sources such as plasma, process by-products, and contamination by chemistry, gas, etc. of materials.

By clustering the defect locations of wafers by the method and apparatus for predicting the type of wafer defect according to an embodiment of the present invention, the defect types that are repeatedly accumulated and the new defect types that could not be discovered before are displayed, thereby providing an optimal key according to the defect location. Management elements can be extracted.

In addition, the types of defects are divided into data containing specific information on defective wafers through wafer map, which is image data generated at wafer manufacturing and inspection sites, and structured data, which is information on the location of defects on wafers, and equipment data, etc. When analyzing environmental factors affecting

At the same time, the inefficiency of production management can be overcome because defects for which characteristics cannot be clearly identified can be classified by type through cluster analysis.

By monitoring environmental factors such as equipment data and sensor data, and at the same time analyzing the correlation between defect types based on 4M1E, which is a key element of manufacturing, the innumerable parameters of the wafer manufacturing site affect each defect type. The optimal management point can be suggested.

The device described above may be implemented as a hardware component, a software component, and/or a combination of the hardware component and the software component. For example, devices and components described in the embodiments may include, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA). , a programmable logic unit (PLU), microprocessor, or any other device capable of executing and responding to instructions, may be implemented using one or more general purpose or special purpose computers. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. A processing device may also access, store, manipulate, process, and generate data in response to execution of the software. For convenience of understanding, although one processing device is sometimes described as being used, one of ordinary skill in the art will recognize that the processing device includes a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that can include For example, the processing device may include a plurality of processors or one processor and one controller. Other processing configurations are also possible, such as parallel processors.

Software may comprise a computer program, code, instructions, or a combination of one or more thereof, which configures a processing device to operate as desired or is independently or collectively processed You can command the device. The software and/or data may be any kind of machine, component, physical device, virtual equipment, computer storage medium or apparatus, to be interpreted by or to provide instructions or data to the processing device. may be embodied in The software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored in one or more computer-readable recording media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc. alone or in combination. The program instructions recorded on the medium may be specially designed and configured for the embodiment, or may be known and available to those skilled in the art of computer software. Examples of the computer-readable recording medium include a hard disk, a magneto-optical medium such as a floppy disk, and program instructions such as a ROM, a RAM, a flash memory, etc. and hardware devices specially configured to perform Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like.

As described above, although the embodiments have been described with reference to the limited embodiments and drawings, various modifications and variations are possible from the above description by those skilled in the art. For example, the described techniques are performed in an order different from the described method, and/or the described components of the system, structure, apparatus, circuit, etc. are combined or combined in a different form than the described method, or other components Or substituted or substituted by equivalents may achieve an appropriate result.

Therefore, other implementations, other embodiments, and equivalents to the claims are also within the scope of the following claims.

110: data collection unit
120: bad judgment unit
130: storage
140: bad list storage unit
150: clustering unit
160: bad pattern extraction unit

Claims (9)

A method for predicting wafer defect types based on cluster analysis performed by a computer, comprising:
collecting wafer inspection data and environmental data in the form of images or structured data;
determining whether the wafer is defective based on the collected wafer inspection data;
Separating and storing good wafer data and bad wafer data according to the determination result;
clustering defect types with the collected defective wafer data; and
Analyzing the correlation between the clustered defect types and matching environmental data
including,
The environment data includes facility data and sensor data, and the sensor data includes any one or more of temperature, humidity, particles, and vibration,
The analysis step is
A method for predicting a wafer defect type based on clustering analysis, comprising analyzing a correlation between a plurality of feature values extracted from clustering results and environmental factors extracted from the facility data and the sensor data using a chi-square test.
According to claim 1,
The wafer data in the form of an image is a wafer map,
The wafer defect type prediction method based on cluster analysis, characterized in that the wafer data in the form of the shaping data includes information on the location where the wafer defect occurred.
3. The method of claim 2,
The step of separately storing the good wafer data and the bad wafer data includes:
The defective wafer data is stored in a storage unit, and the defective wafer data is stored in a defective list storage unit.
According to claim 1,
The step of clustering the defect types is
A method of predicting a wafer defect type based on clustering analysis, characterized in that the information about the location, angle, and thickness of the wafer included in the defective wafer data is clustered through techniques such as K-means and DBSCAN.
5. The method of claim 4,
The step of analyzing the correlation of the environmental data is
4M1E (Man, Machine, Material, method & Environment) analysis is applied to each cluster clustered by the location, angle, and thickness information where the wafer defect occurred, and the correlation is analyzed to generate a pattern for each defect type. A method of predicting types of wafer failure based on analysis.
a data collection unit for collecting wafer inspection data and environmental data in the form of images or structured data;
a defect determination unit that determines whether the wafer is defective based on the collected wafer inspection data;
a storage unit for storing data of non-defective wafers according to the determination result;
a defect list storage unit for storing defective wafer data according to the determination result;
a clustering unit for clustering defective types with the collected defective wafer data; and
Defect pattern extraction unit that analyzes the correlation between clustered defect types and matching environmental data
including,
The environment data includes facility data and sensor data, and the sensor data includes any one or more of temperature, humidity, particles, and vibration,
The defective pattern extraction unit
A wafer defect type prediction apparatus based on clustering analysis that analyzes the correlation between a plurality of feature values extracted from clustering results and environmental factors extracted from the facility data and the sensor data using a chi-square test.
7. The method of claim 6,
The wafer data in the form of an image is a wafer map,
The wafer defect type prediction apparatus based on cluster analysis, characterized in that the wafer data in the form of the shaping data includes information on the location where the wafer defect occurred.
7. The method of claim 6,
The clustering unit
A wafer defect type prediction device based on clustering analysis, characterized in that the information about the location, angle, and thickness of the wafer included in the defective wafer data is clustered through techniques such as K-means and DBSCAN.
9. The method of claim 8,
The defective pattern extraction unit
4M1E (Man, Machine, Material, method & Environment) analysis is applied to each cluster clustered by the location, angle, and thickness information where the wafer defect occurred, and the correlation is analyzed to generate a pattern for each defect type. Wafer defect type prediction device based on analysis.

Images