JP2014074698A - Optical defect detection classification method and optical defect detection classification apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an optical defect detection classification method capable of excluding a defect repaired in a previous step from defects detected in a present step and capable of improving defect detection/classification accuracy.SOLUTION: In the optical defect detection classification method, repaired data of a defect (205) concerned with first processing which is repaired in a previous step A-2 are excluded from defect data of defects (205, 206, 207) detected by optically inspecting an inspection target (L) applied to second processing (B-1) and the defect data left after the exclusion are extracted as defect data of the defects generated in a present step. Consequently second detection of the repaired defect (205) as a defect can be avoided, useless increases in a detection frequency of defects and defect sorts can be suppressed and detection accuracy and classification accuracy of defects can be improved.

Description

  The present invention relates to an optical defect detection and classification method and an optical defect detection and classification apparatus.

  2. Description of the Related Art Conventionally, an optical defect detection and classification device is used in a manufacturing process of a liquid crystal display device or a TFT (thin film transistor) element.

  In this optical defect detection and classification apparatus, the substrate as the inspection object is irradiated with light, the reflected light or transmitted light from the substrate is measured, and the contrast comparison with the reflected light or transmitted light in a normal state is performed. The defect detected on the substrate is detected by comparing the image feature quantity of the CCD image of the defect with the image feature quantity of the representative image of various defects stored in the apparatus. It is automatically classified.

  In actual factories, the above classification results are used to correct defects and detect abnormal processes at an early stage, thereby helping to improve process yield (direct rate).

  The optical defect detection / classification apparatus recognizes a difference between the normal part and “optical” as a defect.

  However, when correcting a defect detected in the previous process, use a technique such as laser drawing / laser cutting to correct the electrical circuit by connecting the disconnected part or cutting the leaked part (short-circuited part). However, it is difficult to achieve the same appearance as the normal part. In general, the optical defect detection and classification apparatus described above recognizes and detects a difference between the "normal" part and the "optical" as a defect, and detects the defect based on the image feature amount extracted from the captured image of the detected defect. Therefore, the corrected defect portion is recognized and detected as a defect in the next process inspection.

  That is, there is a problem that, as the process goes to the subsequent process, the number of detected defects and the defect type tend to increase due to the influence of the defect corrected in the previous process, and the defect detection accuracy may decrease.

JP 2008-32929 A JP 2010-127748 A JP 2010-14436 A

  Accordingly, an object of the present invention is to provide an optical defect detection and classification method and an optical defect detection and classification method that can exclude defects repaired in the previous process from defects detected in the subsequent process, and improve defect detection and classification accuracy. To provide an apparatus.

In order to solve the above problems, an optical defect detection and classification method according to the present invention irradiates light on an inspection object that has been subjected to a predetermined process in the current process, and reflects or transmits light on the inspection object. The optical information of the inspection object is obtained from the optical information of the inspection object, and the optical information of the inspection object is compared with the optical information of the inspection object when the processing is normally performed to detect the defect of the inspection object. ,
Based on the results of classifying the defects detected in the previous process for each defect type, excluding defect data for defects resulting from the processing performed on the inspection object in the previous process from the defect data of the detected defects, The defect data remaining after the exclusion is extracted as defect data for defects that occurred in the current process,
The defect data of defects generated in the current process is classified for each defect classification type.

  According to the defect detection classification method of the present invention, defect data resulting from the processing performed on the inspection object in the previous process is excluded from the defect data detected in the current process, and the defect data remaining after the exclusion is excluded. Since it is extracted as defect data of a defect that has occurred in the current process, it is possible to avoid detecting a defect that has been repaired in the previous process as a defect again in the current process, and to improve the accuracy of defect detection and classification.

1 is a schematic diagram showing a schematic configuration of a defect detection and classification apparatus 100 according to an embodiment of the present invention. 2 is a block diagram of the defect detection and classification apparatus 100. FIG. It is a typical process drawing explaining the process in which the defect which generate | occur | produces in the manufacturing process of a TFT element with the said defect detection classification device 100 is detected. It is a typical process drawing explaining the process in which the defect which generate | occur | produces in the manufacturing process of a TFT element with the said defect detection classification device 100 is detected. 5 is a flowchart for explaining a process of detecting defects generated in the TFT element manufacturing process by the defect detection and classification apparatus 100. It is a flowchart following the flowchart of FIG. It is a flowchart following the flowchart of FIG.

  Hereinafter, the present invention will be described in detail with reference to the illustrated embodiments.

  FIG. 1 is a schematic diagram showing a schematic configuration of a defect detection and classification apparatus 100 according to an embodiment of the present invention. The defect detection and classification apparatus 100 includes a backlight 10, an inspection table 20, an imaging device 31 equipped with a CCD area sensor (hereinafter referred to as “area sensor”) 30 and an actuator 40, and a computer 50. Is done.

  The defect detection and classification apparatus 100 captures a substrate L on which a TFT (Thin Film Transistor) element is imaged from a plurality of imaging positions by an area sensor 30, and uses a pixel shift using image data captured at each imaging position to increase the resolution. And a defect due to an electrical disconnection or short circuit in the substrate L, a minute defect due to contamination, etc. are detected on the basis of the high resolution image.

  The backlight 10 includes a light source that can emit light. In this embodiment, the backlight 10 is configured as a surface light source that emits light in a planar manner upward. The inspection table 20 has a flat placement surface S on which a substrate L as a processing object is placed. In this embodiment, the inspection table 20 is mounted on the backlight 10 so that the placement surface S is horizontal or substantially horizontal. Therefore, by driving the backlight 10, the substrate L placed on the inspection table 20 can be irradiated with light from below.

  The area sensor 30 serving as an imaging element is disposed above the inspection table 20 so as to face the mounting surface S of the inspection table 20 and is configured to be able to image the substrate L mounted on the inspection table 20. . The area sensor 30 starts or stops the imaging operation according to the imaging timing control signal transmitted from the computer 50. The area sensor 30 is configured to output image data of a captured image to the computer 50.

  The actuator 40 that is a moving device changes the relative positional relationship between the substrate L placed on the inspection table 20 and the area sensor 30 by moving the area sensor 30. In this embodiment, the actuator 40 moves the area sensor 30 in a first direction and a second direction, which are pixel arrangement directions in the area sensor 30, and a third direction perpendicular to the first direction and the second direction. It is comprised so that movement is possible along. The actuator 40 moves the area sensor 30 to a predetermined position in accordance with an imaging position control signal transmitted from the computer 50.

  As shown in the block diagram of FIG. 2, the computer 50 includes a defect detection unit 61, a defect extraction unit 62, and a defect classification unit 63. The defect detection unit 61 includes a storage unit 51, an inspection unit 52, and a comparison unit 53, and the defect extraction unit 62 includes an extraction unit 54. The defect classification unit 63 includes a classification unit 55 and a classification storage unit 56.

  First, pattern image data such as a gate formed on the substrate L imaged by the area sensor 30 is stored in the inspection unit 52. Next, the comparison unit 53 compares the pattern image data stored in the inspection unit 52 with the reference image data stored in advance in the storage unit 51, and evaluates the comparison result. The reference image data is image data when a gate or the like is normally formed on the substrate L.

  When the comparison unit 53 determines that the pattern image data and the reference image data are different from each other as a result of the image data comparison by the comparison unit 53, an image portion different from the reference image data in the pattern image data is defective. The detection result is output from the comparison unit 53 to the defect extraction unit 62.

  The extraction unit 54 of the defect extraction unit 62 converts the defect data of the previous process and the non-correctable defect data of the previous process into the above-described defect based on the correction result classification result from the correction process 57 of the previous process by the method described later. Excluded from the detection result, only the defect data in the current process is extracted and output to the defect classification unit 63.

  Finally, the classification unit 55 of the defect classification unit 63 compares the defect data in the current process input from the extraction unit 54 with the representative image data stored for each defect classification in the classification storage unit 56, and compares the current data. The defect data in the process is classified into the most suitable defect type and output as the final defect classification result.

  Hereinafter, referring to the process diagrams of FIGS. 3 and 4 and the flowcharts of FIGS. 5, 6, and 7, the process of actually forming, inspecting, and correcting the gate layer as the lowest layer on the substrate L (process A) -1, step A-2), forming, inspecting and correcting the first source layer on the substrate L (step B-1, step B-2), and forming the second source layer on the substrate L The defect detection and classification operation according to the above embodiment will be described while sequentially explaining the formation, inspection, and correction processes (process C-1 and process C-2). In an actual TFT element, there are many layers such as an insulating film and a semiconductor film in addition to the gate layer and the source layer, which are omitted in FIG.

  First, a process for forming, inspecting, and correcting a gate layer will be described.

  In step S1, a gate layer is formed on the substrate L by sputtering or the like, and in step S2, photolithography and etching (dry etching or wet etching) are sequentially performed to form the gate electrode 201 (step A-). 1).

  Subsequently, in step S <b> 3, image data obtained by imaging the substrate L from above the substrate L by the area sensor 30 of the imaging device 31 is output from the imaging device 31 to the inspection unit 52 in the computer 50 and stored. Further, the comparison unit 53 compares the captured pattern image data stored in the inspection unit 52 with the reference image data stored in advance in the storage unit 51, and compares the reference image in the pattern image data. An image portion different from the data is output from the comparison unit 53 to the defect extraction unit 62 as a defect detection result, that is, defect data (step A-1). Here, the reference image data is reference image data when the gate electrode 201 is normally formed on the substrate L, and the defect data as the defect detection result represents a broken portion 203 as a defect portion. Yes.

  The defect data representing the disconnection unit 203 is input from the comparison unit 53 to the extraction unit 54. In an example of this embodiment, the gate layer is the lowermost layer, and there is no pre-process of the process of forming the gate electrode 201. For this reason, in step S4, the correction result classification result in which the defect is corrected in the correction process 57 of the previous process is not input to the extraction unit 54. Therefore, in step S4, the extraction unit 54 outputs the defect data from the comparison unit 54 to the classification unit 55 of the defect classification unit 63 as it is.

In step S5, the classification unit 55 of the defect classification unit 63 compares the defect data from the extraction unit 54 with the representative image data for each defect classification stored in the classification storage unit 56. The defect data is classified into the defect type representing the disconnection portion 203 as the final defect classification result, and the defect type data representing the disconnection of the gate electrode 201 is output as the final defect classification result as a result of this classification.
Next, in the correction step A-2, laser drawing or the like is performed on the disconnected portion 203 using a laser device (not shown) or the like based on the defect type data output from the defect classification unit 63. The connecting portion 205 is formed. Thereby, the electrical connection of the gate electrode 201 is recovered (step S6).

  At the stage of this step S6 (correction process A-2), the correction type (corrected, uncorrectable, unnecessary correction) is given to each final defect classification result by the correction process operator, and defect correction data ( It is stored in the storage unit 51 as defect coordinates and correction type (corrected, uncorrectable, unneeded correction). As a specific example, “corrected” is assigned to the connection portion 205 of the gate electrode 201 as the correction type.

  The defect correction data is used for determining duplicate detection by position information comparison and excluding duplicate defects in the next inspection step B-1.

  In the inspection step A-1, the comparison unit 53 compares the captured pattern image data with the reference image data to detect a defect, as shown in the inspection step A-1 in FIG. In some cases, there are pseudo-defects a to h that are removed in a cleaning process (wet or dry cleaning process) before correction, such as a foreign object. In this case, the pseudo defects a to h are actually removed in the cleaning process before the correction, and the pseudo defects a to h themselves disappear at the correction time in the correction process A-2 in FIG. . Therefore, “correction unnecessary” is given to the pseudo defects a to h as the correction type by the correction process operator.

  Next, a process (process B-1 and B-2) for forming, inspecting, and correcting the first source layer after forming, inspecting, and correcting the gate layer (gate electrode 201) will be described.

  First, in step S7, a first source layer is formed on the substrate L by sputtering on the entire surface, and the process proceeds to step S8 where photolithography and etching (dry etching or wet etching) are performed in order, and the first source layer is formed. A source electrode 204 is formed (step B-1).

  Subsequently, the process proceeds to step S <b> 9, and image data obtained by imaging the substrate L from above the substrate L is output to the inspection unit 52 in the computer 50 by the area sensor 30 of the imaging device 31 and stored. Further, the comparison unit 53 compares the captured image data stored in the inspection unit 52 with the reference image data stored in advance in the storage unit 51, and compares the reference image data in the image data with the reference image data. Different image portions are output as defect detection results, that is, defect data, from the comparison unit 53 to the defect extraction unit 62 (step B-1). Here, the reference image data is reference image data when the gate electrode 201 and the first source electrode 204 are normally formed on the substrate L.

  In step S9 of this process B-1, duplicate defect exclusion processing based on the correction results classification results in the previous processes A-1 and A-2 has not yet been performed. Therefore, the defect data output from the comparison unit 53 to the defect extraction unit 62 includes not only the defect data of the first source layer (first source electrode 204) formed in the step B-1, but also the gate electrode. The defect data corrected in the previous step A-2 and the defect data that cannot be corrected are included. In an example of this embodiment, not only the defective portions 206 and 207 of the first source electrode 204 to be originally detected in this step B-1, but also the connection portion corrected in the previous step A-2 (step S6). 205 image data is also included.

  Next, proceeding to step S10, the extraction unit 54 of the defect extraction unit 62 determines the previous process from the defect data from the comparison unit 53 based on the classification result of the correction results in the previous process A-2 (gate electrode correction process). The corrected defect data and the non-correctable defect data are excluded, and only defect data in the current process is extracted. Specifically, the extraction unit 54 includes the defective portions 206 and 207 of the first source electrode 204 and the connection portion 205 corrected in the previous step A-2 (step S6). The connection part 205 which is a corrected defect is excluded, and only the defect parts 206 and 207 of the first source electrode 204 are output to the defect classification part 63 (step B-1).

  In step S11, the classification unit 55 of the defect classification unit 63 compares the defect data from the extraction unit 54 with the representative image data for each defect classification stored in the classification storage unit 56. Then, using the defect data as a final defect classification result, defect type data representing the disconnection portion 206 and the leak portion 207 is output (step B-1).

  Next, in step S12, based on the defect type data output from the classification unit 55, a laser device (not shown) or the like is used to perform laser drawing or the like on the disconnection unit 206 to connect the connection unit 208. Then, laser processing is performed on the leak portion 207 to cut the short-circuit portion, thereby forming the leak repair portion 209. Thereby, the disconnection part 206 and the leak part 207 are each electrically repaired (step B-2).

  In step S12 (process B-2), “corrected” is assigned as the correction type by the correction process operator to the connection unit 208 and the leak repair unit 209.

  Next, the process proceeds to step S13, and after forming, inspecting, and correcting the gate electrode 201 and the first source electrode 204, a process of forming, inspecting, and correcting the second source electrode 211 (process C-1, C-2) ) Will be described.

  First, in step S13, a second source layer is formed on the substrate L by sputtering or the like, and the process proceeds to step S14, where photolithography and etching (dry etching or wet etching) are performed in order, and the second source layer is formed. A source electrode 211 is formed (step C-1).

  In step S15, the area sensor 30 of the imaging device 31 outputs and stores image data obtained by imaging the substrate L from above the substrate L to the inspection unit 52 in the computer 50. Further, the comparison unit 53 compares the captured image data stored in the inspection unit 52 with the reference image data stored in advance in the storage unit 51, and compares the reference image data in the image data with the reference image data. Different image portions are output as defect detection results, that is, defect data, from the comparison unit 53 to the defect extraction unit 62 (step C-1). Here, the reference image data is reference image data when the gate electrode 201, the first source electrode 204, and the second source electrode 211 are normally formed on the substrate L.

  In step S15 of the process C-1, duplicate defect exclusion processing based on the correction result classification results in the previous processes A-1, A-2 and B-1, B-2 has not yet been performed. Therefore, the defect data output from the comparison unit 53 to the defect extraction unit 62 includes not only the defect data of the second source layer (second source electrode 211) formed in this step C-1, but also the gate electrode. The corrected defect data and the uncorrectable defect data in the pre-process A-1 for forming 201 and the pre-process B-1 for forming the first source electrode 204 are included. In an example of this embodiment, not only the defective portion 211 of the second source electrode 211 that should be originally detected in the step C-1, but also the connecting portion 205 corrected in the previous step A-2 (step S6) The image data of the connection unit 208 and the leak repair unit 209 corrected in the process B-2 (step S12) is also included.

  Next, proceeding to step S16, the extraction unit 54 of the defect extraction unit 62 determines the correction results in the previous process A-2 (gate electrode correction process) and B-2 (first source electrode correction process). Based on the defect data from the comparison unit 53, the corrected defect data and the non-correctable defect data in the previous process are excluded, and only the defect data in the current process is extracted. Specifically, the extraction unit 54 includes the defect part 212 of the second source electrode 211 and the connection part 205 and the process B-2 (step S12) corrected in the previous process A-2 (step S6). ), The connection parts 205 and 208 and the leak repair part 209 which are the corrected defects in the previous process are excluded from the connection part 208 and the leak repair part 209 corrected in step), and only the defect part 212 of the second source electrode 211 is defective. It outputs to the classification | category part 63 (process C-1).

  In step S17, the classification unit 55 of the defect classification unit 63 compares the defect data from the extraction unit 54 with the representative image data for each defect classification stored in the classification storage unit 56. Then, using the defect data as a final defect classification result, defect type data representing the disconnection portion 212 is output (step C-1).

  Next, in step S18, based on the defect type data output from the classification unit 55, laser connection or the like is performed on the disconnection unit 212 using a laser device (not shown) or the like to connect the connection unit 213. Form. Thereby, the said disconnection part 212 is repaired (process C-2).

  In step S18 (process C-2), “corrected” is assigned to the connection unit 213 as the correction type by the correction process operator.

  As described above, according to the defect detection apparatus 100 of this embodiment, the defect extraction unit 62 compares and refers to the defect detection result (defect data) input from the comparison unit 53 and the classification result of the correction results of the previous process. To do. As a result, the defect extraction unit 62 can output only defect data in the current process, excluding defect data caused by a repaired part repaired in the previous process or a defect classified as unrepairable. Therefore, it is possible to improve the defect detection accuracy and classification accuracy by suppressing the number of defect detections and defect types from increasing unnecessarily.

  In the above embodiment, the defect is detected and imaged using the light transmitted by the backlight 10, but in the case of a device in which a plurality of layers are stacked as in the TFT element manufacturing process, reflection is performed as necessary. Light (incident light) and transmitted light may be selectively used, and a configuration may be adopted in which defect detection and imaging are mainly performed using reflected light (epiped light).

  In the above embodiment, in steps S3, S9, and S15, the comparison unit 53 performs defect detection by comparing the captured image data with the reference image data of the normal part. However, the inspection object is a TFT element. In the case of an object composed of a periodic repetitive pattern like this, even if a defect is detected as follows, considering that most of the pattern formed on the substrate is basically a normal part. Good. That is, the image data obtained by imaging the surface of the substrate L by the area sensor 30 is internally processed, and the image data at the position to be inspected and the peripheral portion separated from this position by n times the repetition period (n is a natural number) It is also possible to determine that the image data at a position different from the image data of the peripheral portion is a defect. In this case, it is not always necessary to capture the reference image (normal part) in advance and store it in the storage unit 51 in advance.

  In the above embodiment, the defect detection apparatus for detecting defects in the TFT element manufacturing process has been described. However, the present invention can also be applied to a manufacturing process of a liquid crystal panel, a color filter, etc. The present invention can be applied to a manufacturing process in which a defect repaired part manufactured by the above process and repaired in a certain process may be detected when a defect is detected in a subsequent process.

DESCRIPTION OF SYMBOLS 10 Backlight 20 Inspection stand 30 Area sensor 31 Imaging device 40 Actuator 50 Computer 51 Storage part 52 Inspection part 53 Comparison part 54 Extraction part 55 Classification part 56 Classification storage part 61 Defect detection part 62 Defect extraction part 63 Defect classification part 100 Defect detection Classification device 201 Gate electrodes 203, 206, 212 Disconnected portion 204 First source electrodes 205, 208, 213 Connection portion 207 Leak portion 209 Leak repair portion 211 Second source electrode L Substrate S Placement surface a to h Foreign matter (pseudo) defect)

Claims (4)

The inspection object that has been subjected to a predetermined process in the current process is irradiated with light, optical information of the inspection object is obtained from reflected light or transmitted light from the inspection object, and optical information of the inspection object And the optical information of the inspection object when the above processing is normally performed, and the defect of the inspection object is detected,
Based on the result of classifying the defects detected in the previous process for each defect type, the defect data of the defects caused by the processing performed on the inspection object in the previous process is excluded from the defect data of the detected defects. , The defect data remaining after the exclusion is extracted as defect data of defects that occurred in the current process,
An optical defect detection and classification method, wherein defect data of defects generated in the current process is classified for each defect classification type. The optical defect detection and classification method according to claim 1,
The defect data of the defects detected in the previous process are classified into corrected defect data of already corrected defects, uncorrectable defect data of defects that could not be corrected, and correction-unnecessary defect data that did not require correction. And
Excluding the corrected defect data and the non-correctable defect data from the defect data of the defect detected in the current process, and extracting the defect data remaining after the exclusion as the defect data of the defect generated in the current process A characteristic optical defect detection and classification method. The optical defect detection and classification method according to claim 1 or 2,
An optical defect detection and classification method, wherein optical information of an inspection object when the above processing is normally performed is stored in a storage unit in advance. The inspection object that has been subjected to a predetermined process in the current process is irradiated with light, optical information of the inspection object is obtained from reflected light or transmitted light from the inspection object, and optical information of the inspection object And the optical information of the inspection object when the above processing is performed normally, detect the defect of the inspection object, and output the defect data of the defect,
The processing performed on the inspection object in the previous process from the defect data of the detected defect based on the result of classifying the defects detected in the previous process for each defect type when the defect data is input from the defect detection unit A defect extraction unit that excludes defect data due to defects, extracts defect data remaining after exclusion as defect data of defects generated in the current process, and outputs the defect data;
An optical defect detection classification comprising: a defect classification unit for inputting defect data of defects generated in the current process from the defect extraction unit, and classifying and outputting the defect data for each defect classification type apparatus.

Images