CN108062500A - A kind of characteristic recognition method and device based on CDSEM - Google Patents
A kind of characteristic recognition method and device based on CDSEM Download PDFInfo
- Publication number
- CN108062500A CN108062500A CN201610984604.3A CN201610984604A CN108062500A CN 108062500 A CN108062500 A CN 108062500A CN 201610984604 A CN201610984604 A CN 201610984604A CN 108062500 A CN108062500 A CN 108062500A
- Authority
- CN
- China
- Prior art keywords
- cdsem
- feature
- substrate
- identified
- identification informations
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/69—Microscopic objects, e.g. biological cells or cellular parts
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of characteristic recognition method and device based on CDSEM, the described method includes：The CDSEM identification informations of training substrate are obtained by CDSEM, known feature is provided on the trained substrate, the CDSEM identification informations include CDSEM images and its corresponding CDSEM indicatrixes；According to the CDSEM identification informations of the trained substrate and it is described known to feature neutral net is trained so that training after neutral net be suitable for substrate to be identified carry out feature recognition.The solution of the present invention can carry out more accurate identification based on CDSEM to the feature of Semiconductor substrate.
Description
Technical field
The present invention relates to semiconductor manufacturing facility technical field, more particularly to a kind of characteristic recognition method based on CDSEM
And device.
Background technology
Critical size scanning electron microscope (Critical Dimension Scanning Electronic
Microscope, CDSEM) it is a kind of critical size (CD) for being used to measure pattern in Semiconductor substrate in semiconductor processing
Instrument, operation principle are：The electron beam irradiated from electron gun is converged by collector lens, is arrived through perforate (aperture)
Up on the pattern of measure object, capture the secondary electron released using detector and transform it into electric signal, obtain two dimension
CDSEM images, and generate its corresponding CDSEM indicatrixes, using CDSEM images and its corresponding CDSEM indicatrixes as
The high-precision critical size for measuring measure object in basis.
Existing CDSEM can measure the critical size of the feature of Semiconductor substrate, but in some cases simultaneously
Can not accurately the feature of double of conductor substrate be identified.For example, characteristic size is smaller, raised and recess feature size phase
When approximation, existing CDSEM is poor to the feature recognition accuracy of Semiconductor substrate.
The content of the invention
Present invention solves the technical problem that it is that more accurate identification is carried out to the feature of Semiconductor substrate based on CDSEM.
In order to solve the above technical problems, the embodiment of the present invention provides a kind of characteristic recognition method based on CDSEM, including：
The CDSEM identification informations of training substrate are obtained by CDSEM, known feature, the CDSEM are provided on the trained substrate
Identification information includes CDSEM images and its corresponding CDSEM indicatrixes；According to the CDSEM identification informations of the trained substrate
And the known feature is trained neutral net so that the neutral net after training be suitable for substrate to be identified into
Row feature recognition.
Optionally, the characteristic recognition method based on CDSEM further includes：Substrate to be identified is obtained by CDSEM
CDSEM identification informations；It is treated according to the neutral net after the CDSEM identification informations of the substrate to be identified and the training to described
Identify that substrate carries out feature recognition.
Optionally, the feature includes：Protrusion and recess；It is described that substrate progress feature recognition to be identified is included：Really
Fixed feature to be identified is protrusion or recess.
Optionally, the CDSEM identification informations that training substrate is obtained by CDSEM include：Described in being obtained by CDSEM
The CDSEM identification informations of the known feature in sampling substrate using as log-on message, the sampling substrate be from
It is chosen in multiple training substrates comprising identical known feature；It is obtained by CDSEM batches the multiple comprising identical
CDSEM identification informations corresponding with the log-on message in the training substrate of known feature.
Optionally, obtaining the CDSEM identification informations includes：It is obtained respectively with the different measurement patterns of CDSEM described
CDSEM identification informations, the measurement pattern include raised measurement pattern and recess measurement pattern；Institute is obtained by CDSEM batches
Stating CDSEM identification informations corresponding with the log-on message in multiple training substrates comprising identical known feature includes：With
Raised measurement pattern and recess measurement pattern obtain respectively in the multiple training substrate comprising identical known feature with institute
State the corresponding CDSEM identification informations of log-on message；After obtaining the CDSEM identification informations, the feature recognition based on CDSEM
Method further includes：The CDSEM identification informations got with the measurement pattern being consistent with the known feature are added just
True label, the CDSEM identification informations got with the measurement pattern not being inconsistent with the known feature add mistake
Label.
Optionally, according to the CDSEM identification informations and the known feature neutral net is trained including：Root
The CDSEM identification informations and the corresponding label training neutral net that different measurement patterns obtain according to this so that institute
It states the neutral net after training to be suitable for identifying feature to be identified, with selection and the feature phase to be identified on the substrate to be identified
The measurement pattern matched somebody with somebody carries out dimensional measurement to the feature to be identified.
The embodiment of the present invention also provides a kind of specific identification device based on CDSEM, including：First identification information obtains single
Suitable for obtaining the CDSEM identification informations of training substrate by CDSEM, known feature, institute are provided on the trained substrate for member
Stating CDSEM identification informations includes CDSEM images and its corresponding CDSEM indicatrixes；Training unit, suitable for according to the training
The CDSEM identification informations of substrate and the known feature are trained neutral net, so that the nerve net after training
Network is suitable for carrying out feature recognition to substrate to be identified.
Optionally, the specific identification device based on CDSEM further includes：Second discrimination information acquisition unit, suitable for logical
Cross the CDSEM identification informations that CDSEM obtains substrate to be identified；Recognition unit, suitable for being known according to the CDSEM of the substrate to be identified
Neutral net after other information and the training carries out feature recognition to the substrate to be identified.
Optionally, the feature includes：Protrusion and recess；It is described that substrate progress feature recognition to be identified is included：Really
Fixed feature to be identified is protrusion or recess.
Optionally, first discrimination information acquisition unit includes：Log-on message acquiring unit obtains suitable for passing through CDSEM
The CDSEM identification informations of the known feature in the sampling substrate are taken so that as log-on message, the sampling serves as a contrast
Bottom is chosen from multiple training substrates comprising identical known feature；Batch acquiring unit, suitable for passing through CDSEM batches
Amount obtains CDSEM identifications letter corresponding with the log-on message in the multiple training substrate comprising identical known feature
Breath.
Optionally, obtaining the CDSEM identification informations includes：It is obtained respectively with the different measurement patterns of CDSEM described
CDSEM identification informations, the measurement pattern include raised measurement pattern and recess measurement pattern；The batch acquiring unit is suitable for
With raised measurement pattern and recess measurement pattern obtain respectively in the multiple training substrate comprising identical known feature with
The corresponding CDSEM identification informations of the log-on message；The specific identification device based on CDSEM further includes：Label addition is single
Member is correct suitable for being added to the CDSEM identification informations got with the measurement pattern being consistent with the known feature
Label, with the label for the CDSEM identification informations addition mistake that the measurement pattern not being inconsistent with the known feature is got.
Optionally, the training unit be suitable for according to different measurement patterns obtain the CDSEM identification informations and
The corresponding label training neutral net so that the neutral net after the training is suitable for identifying feature to be identified, with selection
Dimensional measurement is carried out to the feature to be identified with the measurement pattern that the feature to be identified on the substrate to be identified matches.
Compared with prior art, the technical solution of the embodiment of the present invention has the advantages that：
By obtaining the CDSEM identification informations of training substrate, CDSEM identification informations are including CDSEM images and its accordingly
CDSEM indicatrixes, and train and known feature is provided on substrate, therefore according to the CDSEM identification informations of the trained substrate
And the known feature, neutral net can be trained, so that the neutral net after training is suitable for to be identified
Substrate carries out feature recognition.Existing CDSEM can directly measure the size of the feature of Semiconductor substrate, but due to surveying
Tested feature is not distinguished accurately before amount, and different features has corresponding suitable measurement method, if
Measurement method corresponding with feature cannot be chosen, measurement result is present with larger deviation.The embodiment of the present invention is served as a contrast using training
The CDSEM identification informations and known feature at bottom are trained neutral net, and the neutral net after training can be more accurate
Judgement feature to be identified so that CDSEM can be carried out further on the basis of the feature to Semiconductor substrate is identified
Measurement, so as to promote the measurement accuracy rate of CDSEM.
Description of the drawings
Fig. 1 is a kind of flow chart of the characteristic recognition method based on CDSEM in the embodiment of the present invention；
Fig. 2 is a kind of schematic diagram based on CDSEM indicatrixes；
Fig. 3 is a kind of flow chart of specific implementation of step S11 in Fig. 1；
Fig. 4 is a kind of structure diagram of the specific identification device based on CDSEM in the embodiment of the present invention.
Specific embodiment
As previously mentioned, existing CDSEM can measure the critical size of the feature of Semiconductor substrate, but can not
Accurately the feature of Semiconductor substrate is identified.
Through inventor the study found that being measured using existing CDSEM, different measurement patterns can be selected to be surveyed
When feature in amount, selected measurement pattern and Semiconductor substrate is inconsistent, measurement result is present with larger deviation, and shows
Some CDSEM can not accurately be identified the feature of Semiconductor substrate, thus the accuracy of measurement result have it is to be hoisted.
In embodiments of the present invention, by obtaining the CDSEM identification informations of training substrate, CDSEM identification informations include
CDSEM images and its corresponding CDSEM indicatrixes, and train and known feature is provided on substrate, therefore according to the training
The CDSEM identification informations of substrate and the known feature, can be trained neutral net, so that the god after training
It is suitable for carrying out more accurate feature recognition to substrate to be identified through network.And then CDSEM can so as to semiconductor
The feature of substrate further measured on the basis of more accurate identification, so as to promote the measurement of CDSEM
Accuracy rate.
It is understandable for above-mentioned purpose, feature and advantageous effect of the invention is enable to become apparent, below in conjunction with the accompanying drawings to this
The specific embodiment of invention is described in detail.
Fig. 1 is a kind of flow chart of the characteristic recognition method based on CDSEM in the embodiment of the present invention, can include following step
Suddenly：
In step S11, the CDSEM identification informations of training substrate are obtained by CDSEM, are provided on the trained substrate
The feature known, the CDSEM identification informations include CDSEM images and its corresponding CDSEM indicatrixes；
In step s 12, according to the CDSEM identification informations of the trained substrate and it is described known to feature to nerve net
Network is trained, so that the neutral net after training is suitable for carrying out feature recognition to substrate to be identified.
Training substrate can be provided with the Semiconductor substrate of known feature, and in specific implementation, feature can include
Raised (line) and recess (space), carrying out feature recognition to substrate to be identified can include：Determine that feature to be identified is
Protrusion or recess.
In specific implementation, CDSEM identifies that image can be captured the secondary electron released using detector and become
Two-dimentional CDSEM images are obtained after being changed to electric signal, CDSEM indicatrixes can be curve corresponding with CDSEM images, such as
It can include curve directly corresponding with CDSEM images, to the curve after curve progress smoothly and/or to the curve derivation
The curve obtained afterwards.
Curve directly corresponding with CDSEM images in CDSEM indicatrixes and the curve after derivation are shown in Fig. 2
Schematic diagram.Wherein, upside figure is that can correspond to one with the direct homologous thread of CDSEM images, two of which peak fractions
Either the measurement of a concave both sides of the edge critical size is typically to measure a protrusion or a concave both sides to protrusion
Width between edge；Downside figure is obtained curve after derivation.
In a non-limiting example, protrusion can be the photoresist for being covered in Semiconductor substrate.Need what is illustrated
It is that photoresist is only example, protrusion can also be other appropriate materials, such as various dielectric materials.
In the prior art, when directly being measured using CDSEM, typically by being obtained to CDSEM indicatrix measurements
Critical size.But due to that ought be raised or concave difference between two peak values, measurement result be different；And CDSEM cannot
Accurate distinguishing characteristic, particularly when the edge of protrusion is more sharp, error rate higher, therefore need the feature to measurement object
It distinguishes.
Referring to Fig. 3, the CDSEM identification informations of trained substrate can be obtained as follows：
In step S31, the CDSEM that the known feature in the sampling substrate is obtained by CDSEM knows
Other information is using as log-on message, the sampling substrate is chosen from multiple training substrates comprising identical known feature
's；
In step s 32, by CDSEM batches obtain in the multiple training substrate comprising identical known feature with
The corresponding CDSEM identification informations of the log-on message.
In specific implementation, it is known that feature can be raised or recess, multiple instructions for including identical known feature
It can be with a batch of multiple substrates to practice substrate.A substrate can be chosen as sampling substrate, using same as obtaining in batches
The reference of the CDSEM identification informations of other substrates of batch.
In an infinite example, protrusion or a concave edge can be selected, with different measurement patterns pair
This feature obtains corresponding to the CDSEM indicatrixes at the edge as log-on message after measuring.It can be included in log-on message
Primitive curve and it is smooth after curve, CDSEM images can also be included.
In specific implementation, obtaining the CDSEM identification informations can include：With the different measurement patterns of CDSEM point
The CDSEM identification informations are not obtained；Wherein, measurement pattern can be configured by CDSEM, can include protrusion measurement mould
Formula and recess measurement pattern.
Raised measurement pattern is suitable for measuring the critical size of protrusion, and notching pattern is suitable for concave critical size
It measures, during measurement, CDSEM identification informations can be obtained.Recess can also be measured using boss pattern,
Conversely, protrusion can also be measured using notching pattern, but measurement result has relatively large deviation.
Specifically, to same feature, the CDSEM curves measured under different measurement patterns are different；More
Specifically, under different measurement patterns, the mode of the derivation of pair curve directly corresponding from CDSEM images can be different, obtain
Curve after the derivation arrived is different.And critical size is obtained with reference to the curve after asking, therefore measured under different measurement patterns
Critical size difference is obtained, if using with treating that the measurement pattern that the feature of side is not inconsistent measures, obtained measurement result deviation
It can be bigger.
Corresponding, the curve included in foregoing log-on message can be multigroup image similar to Fig. 2, i.e. upside in Fig. 2
The curve after a peak value and downside derivation in curve, due to being measured with different measurement patterns to known feature
It is different during obtained result, therefore multigroup curve can be included in log-on message.Label can also be included in log-on message, to refer to
Show whether measurement pattern matches with the feature of sampling substrate.The CDSEM images of sampling substrate can also be included in log-on message,
And the raised and corresponding gray scale of recess.
Step S32 can include in Fig. 3：The multiple include is obtained with raised measurement pattern and recess measurement pattern respectively
CDSEM identification informations corresponding with the log-on message in the training substrate of identical known feature.
After the CDSEM identification informations are obtained, the characteristic recognition method based on CDSEM can also include：To with
CDSEM identification informations that the measurement pattern being consistent with the known feature is got add correct label, with institute
State the label for the CDSEM identification informations addition mistake that the measurement pattern that known feature is not inconsistent is got.
CDSEM identifications letter corresponding with the log-on message is being obtained from the training substrate comprising identical known feature
Can be using log-on message as reference, to choose corresponding CDSEM identification informations during breath.
It specifically, can be according to CDSEM images and protrusion and the feature known to corresponding gray scale acquisition that is recessed
Shape and position, feature known to discovery, measures known feature, using log-on message as reference on training substrate
After being compared, CDSEM identification informations corresponding with log-on message can be chosen from the curve that measurement obtains.
When it is known be characterized as protrusion when, can to obtained under raised measurement pattern CDSEM identification informations addition just
True label, in the label of the CDSEM identification informations addition mistake obtained under measurement pattern that is recessed；Conversely, as known spy
Levy for recess when, can be to adding correct label in the CDSEM identification informations that are obtained under measurement pattern of be recessed, in protrusion survey
The label of the CDSEM identification informations addition mistake obtained under amount pattern.So far, can obtain training the required instruction of neutral net
Practice data, wherein CDSEM images, CDSEM indicatrixes and corresponding label can be included.
With continued reference to Fig. 1, in specific implementation, step S12 can include：According to being obtained with different measurement patterns
CDSEM identification informations and the corresponding label training neutral net so that the neutral net after the training is suitable for identification
Feature to be identified, the measurement pattern to be matched with selection with the feature to be identified on the substrate to be identified is to the spy to be identified
Sign carries out dimensional measurement.
It specifically, can be with CDSEM images and its corresponding CDSEM indicatrixes and corresponding label to nerve net
Network is trained so that neutral net is receiving the corresponding CDSEM images of feature to be identified and its corresponding CDSEM features song
During line, neutral net is trained using above-mentioned training data so that the neutral net after the training, which is suitable for identifying, to be waited to know
Other feature, so as to the measurement pattern to be matched based on recognition result selection with the feature to be identified on the substrate to be identified
Dimensional measurement is carried out to the feature to be identified, further obtains more accurate measurement result.
In specific implementation, the characteristic recognition method based on CDSEM can also include：
Step S13 obtains the CDSEM identification informations of substrate to be identified by CDSEM；
Step S14, according to the neutral net after the CDSEM identification informations of the substrate to be identified and the training to described
Substrate to be identified carries out feature recognition.
The process of the CDSEM identification informations of substrate to be identified is obtained by CDSEM with obtaining training substrate by CDSEM
The process of CDSEM identification informations is similar, and details are not described herein.
In embodiments of the present invention, by obtaining the CDSEM identification informations of training substrate, CDSEM identification informations include
CDSEM images and its corresponding CDSEM indicatrixes, and train and known feature is provided on substrate, therefore according to the training
The CDSEM identification informations of substrate and the known feature, can be trained neutral net, so that the god after training
It is suitable for carrying out feature recognition to substrate to be identified through network.Existing CDSEM can be directly to the ruler of the feature of Semiconductor substrate
It is very little to measure, but due to not distinguished accurately to the feature being tested before measuring, and different features has and corresponds to therewith
Suitable measurement method, if measurement method corresponding with feature cannot be chosen, measurement result is present with larger deviation, therefore existing
The accuracy that some CDSEM measure the critical size of the feature of Semiconductor substrate has to be hoisted.The embodiment of the present invention carries
For a kind of characteristic recognition method based on CDSEM so that CDSEM can accurately be known in the feature to Semiconductor substrate
Not, and further measured on this basis, so as to promote the measurement accuracy rate of CDSEM.
The embodiment of the present invention also provides a kind of specific identification device based on CDSEM, and structure diagram includes referring to figure：
First discrimination information acquisition unit 41, it is described suitable for obtaining the CDSEM identification informations of training substrate by CDSEM
Known feature is provided on training substrate, the CDSEM identification informations include CDSEM images and its corresponding CDSEM features
Curve；
Training unit 42, suitable for the CDSEM identification informations according to the trained substrate and it is described known to feature to god
It is trained through network, so that the neutral net after training is suitable for carrying out feature recognition to substrate to be identified.
In specific implementation, the specific identification device based on CDSEM can also include：
Second discrimination information acquisition unit 43, suitable for passing through the CDSEM identification informations that CDSEM obtains substrate to be identified；
Recognition unit 44, suitable for the nerve net after the CDSEM identification informations according to the substrate to be identified and the training
Network carries out feature recognition to the substrate to be identified.
In specific implementation, the feature includes：Protrusion and recess；It is described that feature recognition bag is carried out to substrate to be identified
It includes：Determine feature to be identified for protrusion or recess.
In specific implementation, the first discrimination information acquisition unit 41 can include：
Log-on message acquiring unit (not shown), suitable for pass through CDSEM obtain it is described known in the sampling substrate
The CDSEM identification informations of feature are using as log-on message, the sampling substrate is comprising identical known spy from multiple
It is chosen in the training substrate of sign；
Batch acquiring unit (not shown), it is the multiple comprising identical known feature suitable for being obtained by CDSEM batches
Training substrate in CDSEM identification informations corresponding with the log-on message.
In specific implementation, obtaining the CDSEM identification informations includes：It is obtained respectively with the different measurement patterns of CDSEM
The CDSEM identification informations are taken, the measurement pattern includes raised measurement pattern and recess measurement pattern；
The batch acquiring unit is suitable for obtaining the multiple include respectively with raised measurement pattern and recess measurement pattern
CDSEM identification informations corresponding with the log-on message in the training substrate of identical known feature；
The specific identification device based on CDSEM further includes：Label adding device 45, suitable for with it is described known
CDSEM identification informations that the measurement pattern that feature is consistent is got add correct label, with the known feature
The label for the CDSEM identification informations addition mistake that the measurement pattern not being inconsistent is got.
In specific implementation, the training unit 42 is suitable for being identified according to the CDSEM obtained with different measurement patterns
Information and the corresponding label training neutral net so that the neutral net after the training is suitable for identifying spy to be identified
Sign, to select to carry out ruler to the feature to be identified with the measurement pattern that the feature to be identified on the substrate to be identified matches
Very little measurement.
The specific implementation of the specific identification device based on CDSEM in the embodiment of the present invention and advantageous effect may refer to base
In the characteristic recognition method of CDSEM, this is not repeated.
Although present disclosure is as above, present invention is not limited to this.Any those skilled in the art are not departing from this
It in the spirit and scope of invention, can make various changes or modifications, therefore protection scope of the present invention should be with claim institute
Subject to the scope of restriction.
Claims (12)
1. a kind of characteristic recognition method based on CDSEM, which is characterized in that including：
The CDSEM identification informations of training substrate are obtained by CDSEM, known feature is provided on the trained substrate, it is described
CDSEM identification informations include CDSEM images and its corresponding CDSEM indicatrixes；
According to the CDSEM identification informations of the trained substrate and it is described known to feature neutral net is trained so that
Neutral net after must training is suitable for carrying out feature recognition to substrate to be identified.
2. the characteristic recognition method according to claim 1 based on CDSEM, which is characterized in that further include：
The CDSEM identification informations of substrate to be identified are obtained by CDSEM；
According to the neutral net after the CDSEM identification informations of the substrate to be identified and the training to the substrate to be identified into
Row feature recognition.
3. the characteristic recognition method according to claim 1 based on CDSEM, which is characterized in that the feature includes：Protrusion
And recess；It is described that substrate progress feature recognition to be identified is included：Determine feature to be identified for protrusion or recess.
4. the characteristic recognition method according to claim 1 based on CDSEM, which is characterized in that described to be obtained by CDSEM
The CDSEM identification informations of training substrate include：
The CDSEM identification informations of the known feature in sampling substrate are obtained by CDSEM using as log-on message,
The sampling substrate is chosen from multiple training substrates comprising identical known feature；
It is obtained by CDSEM batches corresponding with the log-on message in the multiple training substrate comprising identical known feature
CDSEM identification informations.
5. the characteristic recognition method according to claim 4 based on CDSEM, which is characterized in that obtain the CDSEM identifications
Information includes：The CDSEM identification informations are obtained respectively with the different measurement patterns of CDSEM, and the measurement pattern includes convex
Play measurement pattern and recess measurement pattern；
It is obtained by CDSEM batches corresponding with the log-on message in the multiple training substrate comprising identical known feature
CDSEM identification informations include：
The multiple training substrate for including identical known feature is obtained with raised measurement pattern and recess measurement pattern respectively
In CDSEM identification informations corresponding with the log-on message；
After obtaining the CDSEM identification informations, the characteristic recognition method based on CDSEM further includes：
The CDSEM identification informations addition got with the measurement pattern being consistent with the known feature is correctly marked
Label, with the label for the CDSEM identification informations addition mistake that the measurement pattern not being inconsistent with the known feature is got.
6. the characteristic recognition method according to claim 5 based on CDSEM, which is characterized in that identified according to the CDSEM
Information and the known feature neutral net is trained including：According to the CDSEM obtained with different measurement patterns
Identification information and the corresponding label training neutral net so that the neutral net after the training, which is suitable for identifying, to be waited to know
Other feature, with select with the measurement pattern that matches of feature to be identified on the substrate to be identified to the feature to be identified into
Row dimensional measurement.
7. a kind of specific identification device based on CDSEM, which is characterized in that including：
First discrimination information acquisition unit, suitable for obtaining the CDSEM identification informations of training substrate, the training lining by CDSEM
Known feature is provided on bottom, the CDSEM identification informations include CDSEM images and its corresponding CDSEM indicatrixes；
Training unit, suitable for the CDSEM identification informations according to the trained substrate and it is described known to feature to neutral net
It is trained, so that the neutral net after training is suitable for carrying out feature recognition to substrate to be identified.
8. the specific identification device according to claim 7 based on CDSEM, which is characterized in that further include：
Second discrimination information acquisition unit, suitable for passing through the CDSEM identification informations that CDSEM obtains substrate to be identified；
Recognition unit, suitable for the neutral net after the CDSEM identification informations according to the substrate to be identified and the training to institute
It states substrate to be identified and carries out feature recognition.
9. the specific identification device according to claim 7 based on CDSEM, which is characterized in that the feature includes：Protrusion
And recess；It is described that substrate progress feature recognition to be identified is included：Determine feature to be identified for protrusion or recess.
10. the specific identification device according to claim 7 based on CDSEM, which is characterized in that first identification information
Acquiring unit includes：
Log-on message acquiring unit, the CDSEM suitable for obtaining the known feature in sampling substrate by CDSEM know
Other information is using as log-on message, the sampling substrate is chosen from multiple training substrates comprising identical known feature
's；
Batch acquiring unit, suitable for being obtained by CDSEM batches in the multiple training substrate comprising identical known feature
CDSEM identification informations corresponding with the log-on message.
11. the specific identification device according to claim 10 based on CDSEM, which is characterized in that obtain the CDSEM and know
Other information includes：The CDSEM identification informations are obtained respectively with the different measurement patterns of CDSEM, and the measurement pattern includes
Raised measurement pattern and recess measurement pattern；
The batch acquiring unit is suitable for obtaining respectively with raised measurement pattern and recess measurement pattern the multiple comprising identical
CDSEM identification informations corresponding with the log-on message in the training substrate of known feature；
The specific identification device based on CDSEM further includes：Label adding device, suitable for with the known feature phase
The CDSEM identification informations that the measurement pattern of symbol is got add correct label, with what is be not inconsistent with the known feature
The label for the CDSEM identification informations addition mistake that measurement pattern is got.
12. the specific identification device according to claim 11 based on CDSEM, which is characterized in that the training unit is fitted
According to the CDSEM identification informations and the corresponding label training neutral net obtained with different measurement patterns, make
It obtains the neutral net after the training to be suitable for identifying feature to be identified, to select and the feature to be identified on the substrate to be identified
The measurement pattern to match carries out dimensional measurement to the feature to be identified.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610984604.3A CN108062500B (en) | 2016-11-09 | 2016-11-09 | CDSEM-based feature identification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610984604.3A CN108062500B (en) | 2016-11-09 | 2016-11-09 | CDSEM-based feature identification method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108062500A true CN108062500A (en) | 2018-05-22 |
| CN108062500B CN108062500B (en) | 2020-05-08 |
Family
ID=62137878
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610984604.3A Active CN108062500B (en) | 2016-11-09 | 2016-11-09 | CDSEM-based feature identification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108062500B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210325314A1 (en) * | 2020-04-15 | 2021-10-21 | Shanghai Huali Integrated Circuit Corporation | Image processing method and device for cdsem |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090103799A1 (en) * | 2007-10-19 | 2009-04-23 | Samsung Electronics Co., Ltd. | Pattern matching method in manufacturing semiconductor memory devices |
| CN105336635A (en) * | 2014-07-17 | 2016-02-17 | 中芯国际集成电路制造(上海)有限公司 | CD-SEM apparatus correction method, CD-SEM apparatus application method and CD-SEM apparatus |
-
2016
- 2016-11-09 CN CN201610984604.3A patent/CN108062500B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090103799A1 (en) * | 2007-10-19 | 2009-04-23 | Samsung Electronics Co., Ltd. | Pattern matching method in manufacturing semiconductor memory devices |
| CN105336635A (en) * | 2014-07-17 | 2016-02-17 | 中芯国际集成电路制造(上海)有限公司 | CD-SEM apparatus correction method, CD-SEM apparatus application method and CD-SEM apparatus |
Non-Patent Citations (2)
| Title |
|---|
| BYUNGWHAN KIM ET AL.: "Neural Network Recognition of Scanning Electron Microscope Image for Plasma Diagnosis", 《NEURAL NETWORK RECOGNITION OF SCANNING ELECTRON MICROSCOPE IMAGE》 * |
| 陈智刚 等: "注塑模设计中侧凹特征的自动识别", 《中国铸造装备与技术》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210325314A1 (en) * | 2020-04-15 | 2021-10-21 | Shanghai Huali Integrated Circuit Corporation | Image processing method and device for cdsem |
| US11573188B2 (en) * | 2020-04-15 | 2023-02-07 | Shanghai Huali Integrated Circuit Corporation | Image processing method and device for CDSEM |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108062500B (en) | 2020-05-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP4772153B2 (en) | Apparatus and method for verifying the location of a region of interest in a sample in an image generation system | |
| CN107516624B (en) | Sample position calibration method and device | |
| EP2666050B1 (en) | Microscope slide coordinate system registration | |
| CN104657711B (en) | A kind of readings of pointer type meters automatic identifying method of robust | |
| US10068126B2 (en) | Microscope slide coordinate system registration | |
| JP2002508069A (en) | Measurement mark, method of detecting measurement mark, and method of measuring object | |
| CN103994786A (en) | Image detecting method for arc ruler lines of pointer instrument scale | |
| JP4905013B2 (en) | Appearance inspection apparatus, appearance inspection method, height measurement method, and circuit board manufacturing method | |
| JP2003513251A5 (en) | ||
| US20110262043A1 (en) | Pattern matching method and image processing device | |
| CN112864037A (en) | Wafer measuring method, device, medium and electronic equipment | |
| JP2018205247A (en) | X-ray diffraction analysis method and device | |
| US20120025073A1 (en) | Orientation imaging using wide angle convergent beam diffraction in transmission electron microscopy | |
| CN106017328B (en) | A kind of polymorphic type line width measuring method and device | |
| CN110633620A (en) | Pointer instrument scale identification method and electronic equipment | |
| TW201707123A (en) | Substrate pre-alignment method | |
| CN111486787A (en) | Test positioning method and test positioning system | |
| CN108062500A (en) | A kind of characteristic recognition method and device based on CDSEM | |
| CN111815580B (en) | Image edge recognition method and small module gear module detection method | |
| CN114518070A (en) | Rapid positioning method and system for scanning probe microscope combined system | |
| KR20030005301A (en) | Method and device for processing images | |
| CN106996911B (en) | The localization method of two-dimensional detection paste solder printing | |
| CN109683449A (en) | Evaluation method determines method, lithographic equipment and non-transitory computer-readable storage media | |
| US7132843B2 (en) | Method and apparatus for positioning a test head on a printed circuit board | |
| JPH04269614A (en) | Pattern-position detecting method and executing apparatus thereof |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |