CN1673992A - A method for establishing defect data bank - Google Patents
A method for establishing defect data bank Download PDFInfo
- Publication number
- CN1673992A CN1673992A CN 200410031730 CN200410031730A CN1673992A CN 1673992 A CN1673992 A CN 1673992A CN 200410031730 CN200410031730 CN 200410031730 CN 200410031730 A CN200410031730 A CN 200410031730A CN 1673992 A CN1673992 A CN 1673992A
- Authority
- CN
- China
- Prior art keywords
- defect
- classification
- semiconductor fabrication
- database
- defects
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 65
- 230000007547 defect Effects 0.000 title claims description 213
- 238000004519 manufacturing process Methods 0.000 claims abstract description 70
- 239000004065 semiconductor Substances 0.000 claims abstract description 52
- 230000008569 process Effects 0.000 claims abstract description 24
- 238000001514 detection method Methods 0.000 claims abstract description 15
- 238000012795 verification Methods 0.000 claims abstract description 7
- 230000002950 deficient Effects 0.000 claims description 34
- 238000005070 sampling Methods 0.000 description 8
- 238000004458 analytical method Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 239000012467 final product Substances 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 239000000047 product Substances 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
Images
Abstract
The fault data base establishing method includes providing one wafer, passing the wafer through the first semiconductor manufacture process to form several faults on the wafer, performing fault detection to detect these faults, providing one preset fault data base with fault classifying parameters corresponding one second semiconductor manufacture process, performing automatic fault classification on the detected faults based on the preset fault data base so as to classify the fault areas into several kinds of fault types, and final verification step to verify the fault classifying accuracy of the automatic fault classification with the results of manual fault classification.
Description
Technical field
The invention provides a kind of method for building up of defect database, refer in particular to a kind of automatic defect classification (automatic defect classification, ADC) defect database method for building up of system of being applied to.
Background technology
In semiconductor fabrication; tend to because some are unavoidable former thereby generate tiny particulate or defective; and along with constantly dwindling of component size in the semiconductor fabrication and improving constantly of circuit integration; these atomic little defectives or particulate also are on the rise to the influence of integrated circuit quality; therefore for keeping the stable of product quality; usually when carrying out each described semiconductor fabrication; also must carry out defects detection at the semiconductor element of being produced; to analyze the basic reason that causes these defectives according to the result who detects; could further avoid or reduce generation of defects afterwards, to reach the purpose that promotes semiconductor fabrication yield and fiduciary level by the adjustment of procedure parameter.
Please refer to Fig. 1, Fig. 1 is the schematic flow sheet of defects detection and analysis in the routine techniques.As shown in Figure 1, to earlier online product wafer be taken a sample 12, carrying out a defects detection 14 according to obtained sample wafer, generally speaking, defects detection more than 14 utilizes board to carry out a defects detection in the mode of full-wafer scanning, still can carry out defective afterwards and detect 18 again, utilize instrument such as SEM that described defective is done the observation of thin portion, analyze 22 to carry out defect cause according to the result who is observed at detected defective.And need cost great amount of manpower and time owing to defective detects 18 again, therefore can not carry out defective one by one to detected defective on practice detects 18 again, usually carry out in the mode of sampling, and be to make the result of sampling analysis have higher representativeness, most likely after finishing defects detection 14, carry out a classification of defects 16 earlier, detected defective in the defects detection 14 is divided into different defect types, again in each defect type, the defective that some is taken out in sampling is a sample, carry out defective and detect 18 again, analyze 22 to carry out defect cause, to find out the main cause that causes these defectives, manage adjustment by procedure parameter afterwards again and overcome and avoid or reduce generation of defects.
In routine techniques, classification of defects 16 carries out in artificial mode mostly, therefore often also require a great deal of time and manpower, automatic classification of defects (automaticdefect classification but also develop gradually now, ADC) technology, utilize machine to replace the classification work of manually carrying out, for instance, the board of many defects detection is all with an automatic defect classification feature, when finishing defects detection 14, directly carry out classification of defects 16, and will detect with classification results and offer online slip-stick artist, carry out follow-up defective for it and detect 18 again and analyze 22 with defect cause.
Generally speaking, these classification of defects boards all are connected to a defect database, control automatic defect classification board and remove to carry out suitable classification of defects by being stored in a classification of defects parameter in the described defect database, detected defective is divided into the several deficiencies type, in other words, whether these classification of defects boards can correctly carry out classification of defects 16, whether look closely its classification of defects parameter appropriately decides, yet since various manufacture processes the defective that may cause often have sizable difference, therefore for different manufacture processes, the definition mode or the mode classification of various defect types often also are not quite similar, that is to say before formal online use, essential elder generation is at the process of desire to carry out defect analysis, set up a defect database, and set suitable classification of defects parameter, these classification of defects boards just can carry out correct classification.
Please refer to Fig. 2, Fig. 2 is the method for building up synoptic diagram of a defect database in the routine techniques.As shown in Figure 2, take a sample 32 earlier, and the sample wafer of obtaining is carried out defects detection, classification of defects and defective detect again, collect defect information 34, after all types of defectives are all collected certain information, just can set up a defect database, and utilize these defect information to train the classification of defects board according to collected defect information, to set classification of defects parameter 36, can utilize the classification of defects board to come described sample wafer is carried out the automatic defect classification afterwards.Certainly, before the formal online use 40 of described defect database, still can carry out checking 38 steps at least once, utilize the artificial defect sorting result to verify that the automatic defect sorting result compares, whether correct to differentiate described defect database.
As previously mentioned, in routine techniques, essential will passing through earlier collected defect information 38, can set up a defect database, and be the accuracy rate that increases the automatic defect classification, often need to utilize a large amount of defect sample of SEM observation, with as the foundation of setting up defect database.Generally speaking, each defect type probably all needs 30 to 50 defect sample, can set up a defect database and also correctly set the classification of defects parameter of board, therefore, if with a semiconductor fabrication that can produce six kinds of different defect types is example, so rough 200 to 300 defect sample of observation that just need, this will expend the plenty of time, often only the foundation of a defect database just needs nearly two months time, and many times, the number of some defect type may be lacked especially, often need to carry out several batches sampling and could obtain enough sample numbers and set the classification of defects parameter, this also can significantly increase the degree of difficulty of setting up defect database.
Along with the progress of semiconductor industry manufacture process and the consideration of economic benefit, the diameter of wafer was marched toward 12 inches by 8 inches of past, the live width size also enters below 0.13 micron even 0.1 micron by 0.18 micron of the past, at this by testing in the process of producing in batches, often need manufacture process is repeatedly significantly changed and adjustment, but conventional defect database method for building up often needs considerable time to set up to be finished, this will significantly slow down the speed that defect cause is analyzed, and need the long time can meet the requirement of volume production, therefore, press for a kind of method for building up of defect database rapidly and accurately now, to address the above problem.
Summary of the invention
Fundamental purpose of the present invention is to provide a kind of database building method of automatic defect categorizing system, to solve the problem in the routine techniques.
Most preferred embodiment of the present invention discloses a kind of defect database method for building up, one wafer at first is provided, it experiences one first semiconductor fabrication, and in this first semiconductor fabrication, form a plurality of defectives, then carry out a defects detection, to detect this a plurality of defectives, one default defect database is provided again, include a classification of defects parameter corresponding to one second semiconductor fabrication in the defect database, and according to default defect database, detected described defective is carried out the automatic defect classification, defect area is divided into multiple different defect type, carry out a verification step at last again, verify the classify accuracy of automatic defect classification by an artificial defect sorting result to each defect type, wherein second semiconductor fabrication has identical manufacture process type with first semiconductor fabrication, and be the last manufacture process of first semiconductor fabrication, or first semiconductor fabrication have identical pattern-making or defect kenel with second semiconductor fabrication.
Because defect database method for building up of the present invention utilizes existing defect database to be the basis, upgrade defect database by replenishing of suitable checking and defect information again, to promote the accuracy of automatic defect classification, so can significantly shorten the Time Created of defect database.
Description of drawings
Fig. 1 is a defect cause analytical approach synoptic diagram.
Fig. 2 is the method for building up synoptic diagram of a normal defect database.
Fig. 3 is the method for building up synoptic diagram of a defect database among the present invention.
The reference numeral explanation
12 samplings of 10 defect control methods
14 defects detection, 16 classifications of defects
18 defectives detect 22 defect cause analyses again
Defect information is collected in 32 samplings 34
36 set 38 checkings of classification of defects parameter
40 lines use 110 samplings
112 samplings 114 provide a default defect database
116 set 118 checkings of classification of defects parameter
120 collect defect information 122 upgrades defect database
124 online uses
Embodiment
Please refer to Fig. 3, Fig. 3 is the method for building up synoptic diagram of a defect database among the present invention.As shown in Figure 3, at first, take a sample 112, obtain a sample wafer, this wafer experiences one first semiconductor fabrication, and forms a plurality of defectives in this first semiconductor fabrication, one default defect database 114 then is provided, this preset database is corresponding to one second semiconductor fabrication, and sets sorting parameter 116 according to this default defect database, then goes to carry out the automatic defect classification with this sorting parameter.Then will carry out a checking 118 steps, to differentiate the accuracy of this automatic defect classification.
As previously mentioned, checking 118 utilizes artificial mode that the defective on the above-mentioned sample wafer is carried out classification of defects equally, to obtain a correct classification results, again with its with before utilize board to carry out the automatic defect sorting result to contrast, when automatic defect sorting result and manual sort's result coincide, this defect database gets final product online use 124, is applied to online classification of defects board; And when automatic defect sorting result and artificial defect sorting result have part difference, then can judge whether this difference is important by the operator, for instance, suppose in the automatic defect sorting result, though the number of detected all types of defectives is lower than the artificial defect sorting result, but still can effectively detect some unusual a large amount of defectives, and when satisfying the demand that defect cause analyzes, even the automatic defect classification is slightly variant with the artificial defect sorting result, also the classification of defects parameter of this defect database can be regarded as correctly, and it is applied to online classification of defects board.
Yet when difference between the two excessive and can not accept the time, so only need to collect defect information 120 with aforementioned manner again, and upgrade defect database 122 according to collected defect information and get final product, generally speaking, when only the degree of discrimination is not good on a certain defect type as if described defect database, then only need to collect defect information 120 and get final product, and only need to add the defect sample of 5 to 10 these defect types usually, can significantly increase the degree of discrimination of this defect type at this defect type.Certainly, after the renewal of finishing defect database, the classification of defects parameter after this defect database resets therewith will reset the classification of defects parameter again, and verify 118 once more, till can be classified exactly to the defective that described first semiconductor fabrication is caused.
It should be noted that if second semiconductor fabrication and first semiconductor fabrication are close manufacture process, then can significantly promote the differentiation accuracy of default defect database at the beginning.In preferred embodiment of the present invention, second semiconductor fabrication is the last manufacture process of first semiconductor fabrication, and second semiconductor fabrication has identical manufacture process type with first semiconductor fabrication, or first semiconductor fabrication have similar pattern-making or defect type to second semiconductor fabrication, for instance, if first semiconductor fabrication is one 0.13 microns gate etch manufacture processes under the manufacture process, second semiconductor fabrication may be the gate etch manufacture process (identical manufacture process type) under 0.15 micron manufacture process before so, or the etch process of gate lateral wall (similar pattern-making), therefore should be quite approximate corresponding to the defect database of first semiconductor fabrication with defect database corresponding to second semiconductor fabrication, so default defect database has quite high accuracy, often do not need that many times defect database is done any adjustment and can satisfy the demand that defect cause is analyzed, and under most of situations, only need collect the defect information of a small amount of defect sample, can successfully set up out the defect database of corresponding described first semiconductor fabrication, therefore often only need just can finish in 3 to 5 days the foundation of a defect database.
Compare with the method for building up of defect database in the routine techniques, defect database method for building up of the present invention utilizes an existing defect database, cooperate a defect database update mechanism to adjust the classification of defects parameter of described defect database again, so can in the extremely short time, set up a reliable and correct defect database, with the analysis of acceleration defect cause, and promote the manufacture process fiduciary level.
The above only is preferred embodiment of the present invention, and all equivalences of carrying out according to claim of the present invention change and revise, and all should belong to covering scope of the present invention.
Claims (12)
1. the method for building up of a defect database, described defect database includes a classification of defects parameter, carries out the automatic defect classification for a classification of defects board, and the method for building up of described defect database comprises at least:
One wafer is provided, and described wafer experiences one first semiconductor fabrication, and forms a plurality of defectives in described first semiconductor fabrication;
Carry out a defects detection, to detect described defective;
One defect database is provided, includes a classification of defects parameter corresponding to one second semiconductor fabrication in the described defect database;
According to described defect database, detected described defective is carried out the automatic defect classification, described defect area is divided into multiple different defect type;
Described defective is carried out artificial defect classification, described defect area is divided into multiple different defect type; And
Carry out a verification step, utilize described artificial defect sorting result to verify the classify accuracy of described automatic defect classification each defect type.
2. the method for claim 1, wherein when the classify accuracy of described automatic defect classification was not good, described method also included a defect database step of updating.
3. method as claimed in claim 2, wherein said defect database step of updating comprises:
According to described artificial defect sorting result, collect a defect information at the not good defect type of accuracy in the described automatic defect classification;
Revise described defect database according to described defect information; And
Again carry out described verification step.
4. the method for claim 1, wherein said second semiconductor fabrication is the last manufacture process of described first semiconductor fabrication, and described second semiconductor fabrication has identical manufacture process type with described first semiconductor fabrication.
5. the method for claim 1, wherein said first semiconductor fabrication has similar pattern-making or defect type to described second semiconductor fabrication.
6. an automatic defect sorting technique comprises:
One wafer is provided, and described wafer experiences one first semiconductor fabrication, and forms a plurality of defectives in described first semiconductor fabrication;
Carry out a defects detection, to detect described defective;
One defect database is provided, includes a classification of defects parameter corresponding to one second semiconductor fabrication in the described defect database; And
Carry out automatic defect classification according to described defect database, being the number of drawbacks type with described classification of defects.
7. method as claimed in claim 6, wherein said method also includes a verification step, to verify the classify accuracy of described automatic defect classification to each defect type.
8. method as claimed in claim 7, wherein said verification step includes the following step:
Described defective is carried out artificial defect classification; And
Utilize described artificial defect sorting result to verify the classify accuracy of described automatic defect classification.
9. method as claimed in claim 7, wherein when the classify accuracy of described automatic defect classification was not good, described method also included a defect database step of updating.
10. method as claimed in claim 7, wherein said defect database step of updating comprises:
According to described artificial defect sorting result, collect a defect information at the not good defect type of accuracy in the described automatic defect classification;
Revise described defect database according to described defect information; And
Again carry out described verification step.
11. method as claimed in claim 6, wherein said second semiconductor fabrication is the last manufacture process of described first semiconductor fabrication, and described second semiconductor fabrication has identical manufacture process type with described first semiconductor fabrication.
12. method as claimed in claim 6, wherein said first semiconductor fabrication has similar pattern-making or defect type to described second semiconductor fabrication.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200410031730 CN1673992A (en) | 2004-03-24 | 2004-03-24 | A method for establishing defect data bank |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200410031730 CN1673992A (en) | 2004-03-24 | 2004-03-24 | A method for establishing defect data bank |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN1673992A true CN1673992A (en) | 2005-09-28 |
Family
ID=35046545
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200410031730 Pending CN1673992A (en) | 2004-03-24 | 2004-03-24 | A method for establishing defect data bank |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1673992A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101477065B (en) * | 2009-01-08 | 2011-02-09 | 西安电子科技大学 | IC defect classification method based on defect boundary value change frequency |
| CN102156136A (en) * | 2011-03-14 | 2011-08-17 | 浙江展邦电子科技有限公司 | Method for detecting PCB negative film |
| CN102637258A (en) * | 2012-04-01 | 2012-08-15 | 首钢总公司 | Method for creating online surface quality detection system defect library |
-
2004
- 2004-03-24 CN CN 200410031730 patent/CN1673992A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101477065B (en) * | 2009-01-08 | 2011-02-09 | 西安电子科技大学 | IC defect classification method based on defect boundary value change frequency |
| CN102156136A (en) * | 2011-03-14 | 2011-08-17 | 浙江展邦电子科技有限公司 | Method for detecting PCB negative film |
| CN102156136B (en) * | 2011-03-14 | 2015-12-02 | 浙江展邦电子科技有限公司 | A kind of PCB negative detecting method |
| CN102637258A (en) * | 2012-04-01 | 2012-08-15 | 首钢总公司 | Method for creating online surface quality detection system defect library |
| CN102637258B (en) * | 2012-04-01 | 2014-05-28 | 首钢总公司 | Method for creating online surface quality detection system defect library |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US6886153B1 (en) | Design driven inspection or measurement for semiconductor using recipe | |
| US9335277B2 (en) | Region-of-interest determination apparatus, observation tool or inspection tool, region-of-interest determination method, and observation method or inspection method using region-of-interest determination method | |
| US7062081B2 (en) | Method and system for analyzing circuit pattern defects | |
| US7739064B1 (en) | Inline clustered defect reduction | |
| US20030022401A1 (en) | Semiconductor device inspection method | |
| US8170707B2 (en) | Failure detecting method, failure detecting apparatus, and semiconductor device manufacturing method | |
| JP2009010405A (en) | Method and device for detecting local outlier | |
| DE112018001534T5 (en) | METHODS AND SYSTEMS FOR AVERAGE INLINE TESTING OF COMPONENTS AND LATENT RELIABILITY OF DEFECTIVE DETECTION | |
| US6399953B1 (en) | Scanning electronic microscope and method for automatically observing semiconductor wafer | |
| CN106372667A (en) | Method for detecting adverse state of inclined sleeve part screws of high-speed train overhead line system | |
| CN113763312A (en) | Detecting defects in semiconductor samples using weak marks | |
| US6807454B2 (en) | Method for automatically controlling defect -specification in a semiconductor manufacturing process | |
| CN109285791A (en) | Design layout-based rapid online defect diagnosis, classification and sampling method and system | |
| CN112907561A (en) | Notebook appearance flaw detection method based on deep learning | |
| JP4080087B2 (en) | Analysis method, analysis system, and analysis apparatus | |
| US7071011B2 (en) | Method of defect review | |
| CN1673992A (en) | A method for establishing defect data bank | |
| CN109426013B (en) | Method and device for analyzing and detecting and repairing defects of color film substrate | |
| CN100375258C (en) | Method for detecting again fault | |
| KR101897582B1 (en) | Method of design-based defect classification and system thereof | |
| CN116363426A (en) | Automatic detection method for health state of highway antiglare shield | |
| US20050177264A1 (en) | Method of building a defect database | |
| JP7015235B2 (en) | Range-based real-time scanning electron microscope invisible binner | |
| CN110120035A (en) | A kind of tire X-ray defect detection method differentiating defect grade | |
| DE10101203A1 (en) | Method to detect and classify scratch on semiconductor wafer; involves forming list of faulty cells in co-ordinate system to obtain standard deviation, which is compared with number of faulty cells |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |