JP4292409B2 - TFT array inspection apparatus and TFT array inspection method - Google Patents

Description

  The present invention relates to a TFT array substrate inspection apparatus for inspecting a TFT array substrate such as a liquid crystal display or an organic EL display.

  An electron beam method and an optical method are known as TFT array inspection devices. In this conventional TFT array inspection apparatus, a defect of the TFT array is detected by irradiating the TFT substrate with an electron beam or light, measuring the potential state of the TFT substrate, and detecting pixels having an abnormal potential on the TFT substrate. Examples of TFT array inspection plate devices using electron beams include Patent Documents 1 and 2, and examples of TFT array inspection devices using light include Patent Documents 3 and 4.

  In the TFT array inspection apparatus, two-dimensional measurement data corresponding to the pixel arrangement of the TFT substrate to be inspected is acquired by the above-described method, and defective pixels are extracted from the two-dimensional measurement data, and the coordinates of the defective pixel and the defect are detected. Perform defect inspection such as species classification.

  FIG. 11 is a profile example in which the signal intensity at each point on a certain line is plotted with respect to two-dimensional measurement data corresponding to the pixel arrangement of the TFT substrate to be inspected. The intensity of this profile represents the signal intensity of each pixel on the line. In addition, the numerical value in a figure is an example.

  Conventionally, defective pixels are extracted from this profile by comparing the signal intensity of the profile with a threshold value. Since the signal intensity of a defective pixel is lower than that of a normal pixel, a pixel whose signal intensity is lower than a threshold value can be detected as a defective pixel.

  The defect types are classified into point defects that exist as isolated points, and line defects that are a set of points that are continuous in the horizontal direction (x direction) or the vertical direction (y direction).

  Line defects can be extracted as a set of detected point defects. 12 and 13 are a flowchart and an explanatory diagram for explaining a processing procedure for extracting a line defect.

  FIG. 13A shows the point defect detection result for each segment of the pixel position of the TFT substrate. The hatched segments in FIG. 13A indicate segments where defects have been detected. This point segment is detected by comparing the signal intensity with a threshold value (step S101). Whether this segment is a line segment forming a line defect or a point segment forming a point defect can be determined by the following steps S102 to S104.

  With respect to the point segments in FIG. 13A, the point segments continuous in the x direction or the y direction are combined. FIG. 13B shows a state in which a connecting segment is formed by connecting the point segments. In the figure, an example is shown in which point segments continuous in the x direction are combined. Next, a combined segment having a predetermined length or more among the combined segments is also set as a line segment. In FIG.13 (b), the connection segment of length 3 or more is made into the line segment (step S103).

  Furthermore, a segment having a distance between a point segment or a line segment on the same line that is a certain distance or less is combined to form a combined segment. FIG. 12C shows an example in which adjacent segments whose distance between segments is 2 or less are combined (step S104).

  Among line defects, it is known that certain line defects appear in such a manner that line segments intersect each other in the horizontal direction (x direction) or the vertical direction (y direction). In the case of this type of line defect, the defect point that is a defect factor is an intersection of two line segments.

  Therefore, the defect point of this line defect is usually specified by obtaining the intersection of the two line segments.

JP 2000-3142 A JP-A-11-265678 Japanese Patent No. 3994948 Japanese Patent No. 3275103

  As shown in FIG. 11, the accuracy of defect point extraction is greatly affected by variations in the signal intensity of the measurement signal and the S / N ratio. Therefore, when the signal intensity of the measurement signal varies or the S / N is insufficient, the continuity of the extracted defect points is reduced, and the portion to be classified as a line defect is classified as a large number of point defects. It will be.

  Further, at the defect point of the cross line defect, either the horizontal (x direction) or vertical (y direction) line segment constituting the line defect is caused by factors such as the pixel state (defect state) or the structure of the TFT substrate. In some cases, the S / N greatly decreases, and it becomes difficult to extract line segments. The above tendency becomes more remarkable when a difference in signal intensity between a defective pixel and a normal pixel approaches due to factors such as measurement system noise and a sufficient signal intensity difference cannot be obtained.

  In such a case, as for the line segment which should originally be extracted, only one of the horizontal direction (x direction) or the vertical direction (y direction) is extracted, and the other one is not extracted. As a result, the defect point cannot be specified, and the inspection performance deteriorates.

  Therefore, the present invention has been made to solve the above-mentioned problems and to improve the accuracy of line defect extraction in a TFT array inspection apparatus, and to accurately detect cross-line defects even in measurement data that is resistant to noise and has insufficient S / N. The purpose is to specify.

  In order to solve the above-described object, the present invention obtains a feature amount representing the feature of each line in the horizontal direction and the vertical direction of measurement data, and obtains a line defect based on the feature amount. In conventional line defect extraction, point defects are extracted by evaluating the intensity at each point on each line of measurement data, and line defects are extracted by combining these point defects. Susceptible to / N. On the other hand, in the present invention, a signal distribution profile is formed by obtaining an amount characterizing each line of measurement data, and line defects are extracted based on the signal distribution profile. Since it is formed by a plurality of measurement data included in the line, it is characterized by being hardly affected by noise and S / N.

  That is, noise and S / N affect each point of data in the conventional line defect extraction, whereas the line defect extraction of the present invention affects a plurality of data on the line. The impact on accuracy is reduced.

  In the present invention, for two-dimensional measurement data obtained by driving each pixel of a TFT substrate, a signal distribution profile representing the characteristics of the same line is obtained from the measurement data of each line in the x direction and the y direction, and the signal distribution profile is obtained. Extract lines that contain line defects.

  The signal distribution profile can be obtained by obtaining, for each line in one direction of the two-dimensional measurement data, any one of the average intensity, integrated value, and variance value of the measurement data along the other direction. Even when noise is added or the signal intensity fluctuates, the influence on the signal distribution profile is reduced by obtaining the average intensity, integrated value, and variance value of the measurement data.

  Further, according to the present invention, a defect point of a line defect is obtained from the extracted line using a point where these lines intersect as a candidate point, and a correlation value difference between measurement data at the candidate point in the x direction and the y direction.

  In the vicinity of the defect point where the cross-line defect exists, the measurement data in the x direction and the y direction both exhibit the same characteristics. On the other hand, in the vicinity of a defect point where a line defect other than a cross line defect exists, one direction includes a line defect and the other direction does not include a line defect. Therefore, the measurement data represents different characteristics depending on the direction. In the present invention, a point where the extracted lines intersect is set as a candidate point, and the candidate point is a defect point of a line defect from the difference between the correlation value in the x direction and the correlation value in the y direction of the measurement data at the candidate point. It is determined whether or not. When the correlation value difference is small, since the characteristics of the measurement data in both directions are similar, it is determined that the point is a cross line defect point. On the other hand, when the correlation value difference is large, since the characteristics of the measurement data in both directions are different, it is determined not to be a cross-line defect point. This is because the cross-line defect is estimated to have common signal characteristics in the x direction and the y direction, and the correlation value in each direction is estimated to be close.

  The TFT array inspection apparatus of the present invention is a TFT array inspection apparatus that inspects a TFT array based on two-dimensional measurement data obtained by driving each pixel of the TFT substrate. The data processing means for obtaining the defect point is provided. The data processing means obtains a signal distribution profile representing the characteristics of the line from the measurement data of each line in the x direction and the y direction, obtains the y direction position of the defect point from the signal distribution profile representing the line characteristics in the x direction, and y The x-direction position is obtained from the signal distribution profile representing the line characteristics in the direction, and the defect point of the line defect is obtained from the x-direction position and the y-direction position.

  The data processing means forms a signal distribution profile by obtaining one of the average intensity, integrated value, and variance value of the measurement data along the other direction for each line in one direction of the two-dimensional measurement data. be able to. For the average intensity, the integrated value, and the variance value, values representing the characteristics of each line are calculated for each line, and the influence of fluctuations in individual signal values due to noise and signal intensity variations is reduced. As a result, as in the conventional array inspection in which individual measurement data is compared with threshold values, line defects due to noise and signal intensity variations can be prevented from being leaked and detected with high accuracy.

  In addition, the position detection of the defect point by the data processing means of the present invention can be performed by subtracting the baseline of the signal distribution profile from the signal distribution profile in each direction to obtain a difference profile and obtaining the peak position of this difference profile. it can.

  Further, the data processing means of the present invention obtains a correlation value in each direction at the intersection of the x-direction position and the y-direction position obtained by detecting the position of the defect point, and determines the defect point of the line defect from the difference between the correlation values. Identify.

  According to the TFT array inspection apparatus of the present invention, the line defect extraction accuracy can be improved. In addition, cross-line defects can be accurately identified even in measurement data that is resistant to noise and has insufficient S / N.

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

  FIG. 1 is a diagram for explaining the outline of the TFT array inspection of the present invention. In FIG. 1, two-dimensional measurement data 10 is two-dimensional measurement data corresponding to the pixel array of the TFT substrate to be inspected, and can be obtained by the technique using the above-described electron beam or light. With respect to the two-dimensional measurement data 10, a signal distribution profile representing the characteristics of the line is obtained using the measurement data of each line in the horizontal direction (x direction) and the vertical direction (y direction). In FIG. 1A, a signal distribution profile 11x is a signal distribution profile representing line characteristics in the horizontal direction (x direction), and a signal distribution profile 11y is a signal distribution profile representing line characteristics in the vertical direction (y direction). .

  The signal distribution profile 11x can be created by obtaining an average intensity, an integrated value, or a variance value for the signal intensity on each line in the x direction of the two-dimensional measurement data 10, and obtaining this value along the y direction. . The signal distribution profile 11y is created by obtaining an average intensity, an integrated value, or a variance value for the signal intensity on each line in the y direction of the two-dimensional measurement data 10, and obtaining this value along the x direction. Can do. In FIG. 1A, the creation of this signal distribution profile is indicated by a circled number 1.

  The signal distribution profile 11 (11x, 11y) represents the characteristics of each line, and when the signal intensity of a certain line is reduced due to a line defect, the value corresponding to the line on the signal distribution profile 11 is reduced. . At this time, noise and signal intensity variations included in the two-dimensional measurement data are canceled in the formation of the signal distribution profile, and the influence on the value of the signal distribution profile is reduced.

  When a line defect exists and defective pixels are concentrated on a specific line, a peak appears at a corresponding position in the signal distribution profile. Further, the peak height of the signal distribution profile corresponds to the number of defective pixels in the corresponding line. On the image, it can be observed as the contrast between the line and the line of another pixel.

  In FIG. 1A, the value by the line 20 in the two-dimensional measurement data 10 appears as a peak 12x on the signal distribution profile 11x, and the value by the lines 21 and 22 in the two-dimensional measurement data 10 is on the signal distribution profile 11y. Respectively appear as a peak 12y1 and a peak 12y2. Here, the line 20 and the line 22 are caused by a cross line defect, and the line 21 is caused by a line defect other than the cross line defect.

  A candidate point for a cross-line defect can be obtained from the peak position of the signal distribution profile (circle numeral 2 in the figure). In the example shown in FIG. 1A, from the intersection of the peak 12x of the signal distribution profile 11x and the peaks 12y1 and 12y2 of the signal distribution profile 11y, P1 and P2 (as indicated by broken lines in the figure) as candidate points 13 of the cross line defect. Circle) is required.

  A defect point 14 (FIG. 1B) of a cross line defect is extracted from the candidate points P1 and P2 (circle numeral 3 in the figure). The defect point 14 can be obtained from the difference between the correlation values in the horizontal direction (x direction) and the correlation value in the vertical direction (y direction) for each candidate point P1 and P2, respectively.

  Next, a process for extracting a defect point of a cross line defect according to the present invention will be described. FIG. 2 is a flowchart for explaining cross point extraction processing for cross line defects, FIG. 3 is a diagram for explaining signals in the extraction processing, and FIG. 4 explains processing for specifying defect points. FIG.

  First, a signal distribution profile is created from two-dimensional measurement data. The signal distribution profile can be obtained by calculating an average value, a cumulative value, and a variance value for each line in the horizontal direction (x direction) and the vertical direction (y direction). FIG. 3A shows an example of average value profile data (step S1).

  Cross point extraction processing is performed in steps S2 to S11 below using the signal distribution profile obtained in step S1.

  The signal distribution profiles in the horizontal direction (x direction) and the vertical direction (y direction) are read (step S2). Since the read signal distribution profile data includes noise, the noise is removed by smoothing processing. The smoothing process can be performed by, for example, a weighted moving average process, but other processing methods may be applied. FIG. 3B shows the profile data after the smoothing process (step S3).

  Next, baseline data is extracted by applying a one-dimensional median filter process to the smoothed profile data. FIG. 3C shows the baseline data (step S4).

  In two-dimensional measurement data, the signal intensity of a normal image may vary depending on the location on the image. In such measurement data, when baseline data is obtained by median processing, a reference signal intensity at each position on the image can be obtained from the baseline data.

  Therefore, by subtracting the baseline data obtained in step S4 from the smoothing data obtained in step S3, a signal intensity distribution that compensates for variations due to locations on the image can be obtained. FIG. 3D shows this difference profile data (step S5).

  In the difference profile data, the defect point appears as a peak position on the difference profile data. The peak position can be extracted by differentiating the differential profile data (step S6) and obtaining a zero crossing point on the differential profile data. FIG. 3E shows differential profile data. The zero crossing points on the differential profile data include those due to noise as well as those due to defect points.

  In order to remove noise from the zero crossing points on the differential profile data, for example, the standard deviation σ of the difference profile data is used, and the intensity of the difference profile data corresponding to the zero crossing point on the differential profile data is a predetermined intensity (for example, 3) is extracted as a peak position (step S7).

  The extracted peak position represents the position of the line including the line defect. Here, at the defect point where the cross line defect exists, the line defect in the horizontal direction (x direction) and the line defect in the vertical direction (y direction) intersect, and at the defect point where the line defect which is not the cross line defect exists, The line defect in the horizontal direction (x direction) does not intersect with the line defect in the vertical direction (y direction).

  Therefore, the peak position does not necessarily represent a defect point of a cross line defect, and may represent a line defect that is not a cross line defect. Therefore, the detected peak position becomes a candidate point of the cross point where the defect point of the cross line defect exists.

  Therefore, when the peak position is detected, it is determined in steps S9 and S10 whether or not the peak position is a cross line defect (step S8).

  FIG. 4 is a diagram for explaining a procedure for extracting a defect point of a cross-line defect from a peak point candidate point.

  FIG. 4A shows two-dimensional measurement data similar to that shown in FIG. In FIG. 4A, the points where the line 20 obtained from the peak position 12x of the signal distribution profile 11x intersects with the lines 21 and 22 obtained from the peak positions 12y1, 12y2 of the signal distribution profile 11y are candidate points P1, P2. It becomes.

  Here, at each candidate point P1, P2, a correlation value is obtained in each of the horizontal direction (x direction) and the vertical direction (y direction), and the candidate point is crossed based on the difference between the correlation values (correlation value difference). It is determined whether or not the defect point is a line defect.

  The correlation value can be obtained from the average value (x direction average value, y direction average value) of a predetermined number of pixels in the horizontal direction (x direction) and the vertical direction (y direction) with each candidate point as the center. . FIG. 4B shows an example of obtaining a correlation value difference at the candidate point P1. The average value in the x direction can be obtained by selecting a predetermined number of positive and negative pixels in the x direction with the candidate point P1 as the center and calculating the average value of the intensity of the pixel. Similarly, the average value in the y direction can be obtained from the average value of the intensity of the pixels by selecting a predetermined number of positive and negative pixels in the y direction centered on the candidate point P1.

  FIG. 4C shows an example in which the correlation value difference at the candidate point P2 is obtained. Similarly to the candidate point P1, the correlation values of the x-direction average value and the y-direction average value are obtained and the correlation is obtained. A value difference is obtained (step S9).

  The correlation value difference between the candidate point P1 and the correlation value difference between the candidate points P2 serves as an index for determining whether the candidate point is a defect point of a cross line defect. It is determined as a defect point. This determination can be made, for example, by comparison with a predetermined value set in advance (step S10).

  If the peak position is not detected in step S8, it is determined that there is no intersection and it is determined that there is no defect of a cross line defect (step S11).

  5 to 9 are examples of measurement data. FIG. 5 is an example of two-dimensional measurement data and a signal distribution profile, and corresponds to FIG. 1 (a). In the two-dimensional measurement data 10 shown in FIG. 5, normal pixels are displayed as gray dots because the intensity varies to some extent, and defective pixels are displayed as dark dots. The darkness of the defective pixel is given a gradation depending on its intensity.

  FIG. 6 shows a signal intensity distribution on a line in the vertical direction (y direction) indicated by a broken line in FIG. As shown in FIG. 6, in the signal intensity distribution on the line, since the difference in signal intensity between the normal pixel and the defective portion is small, when the influence by noise is large, the normal portion and the defective portion depending on the threshold value. Is difficult to separate, and line defects cannot be extracted.

  FIG. 7 shows a profile of measurement data (displayed with Raw) and a profile after median filter processing (displayed with Median). As shown in the figure, in actual measurement data, the signal intensity of normal pixels may vary depending on the location on the image. Therefore, by obtaining a baseline using a median filter, a reference signal intensity at each position is obtained, thereby compensating for variations in signal intensity depending on locations on the image.

  8 and 9 show a differential profile (displayed by Difference), a differential profile (displayed by Derivative), and a threshold value (displayed by 3Sigma) for discriminating between noise components. FIG. 9 is an enlarged view of a part of FIG. A peak point is extracted by detecting the zero crossing point of the differential profile, and the peak and noise are selected using the standard deviation σ of the difference profile. Thereby, the position shown by the arrow in FIG. 9 can be specified as a cross point (defect point).

  When there is a line defect other than the cross line defect, the defect point of the cross line defect can be specified using the correlation value difference as shown in FIG.

  FIG. 10 is a diagram for explaining a configuration example of the TFT array inspection apparatus of the present invention. The TFT array inspection apparatus 1 includes a measurement probe source 6 that irradiates a TFT substrate 2 to be inspected with a measurement probe such as an electron beam or light, a detector 7 that detects the potential of the TFT substrate 2, and inspection control. Means 3 are provided. The inspection control unit 3 controls an inspection signal forming unit 4 that applies an inspection signal to the TFT substrate 2 and a data processing unit 5 that performs a defect inspection based on the measurement voltage.

  The TFT substrate 2 includes pixel electrodes arranged in an array, a driving circuit for the TFT substrate, and wiring including a power supply line, and the inspection signal forming means 4 is used for other data lines and scan signal lines of the power supply line. An inspection signal having a predetermined pattern corresponding to the defect inspection item is applied to the wiring. The inspection signal pattern to be applied is selected by the inspection control means 3 according to the defect inspection item.

  The data processing means 5 performs line defect extraction processing 5A, and performs line defect determination processing 5B on the extracted line defects.

  In the line defect extraction process 5A, the signal distribution is obtained by obtaining an average value, a cumulative value, or a dispersion value of the measurement data of each line in the x direction and the y direction with respect to two-dimensional measurement data obtained by driving each pixel of the TFT substrate. A signal distribution profile creation process 5A1 for creating a profile and a profile data evaluation process 5A2 for extracting a line defect by evaluating the obtained signal distribution profile data are performed.

  In the signal distribution profile of the present invention, in addition to obtaining the line feature amount by the average intensity of each line in the horizontal and vertical directions of the measurement data, the line feature amount may be obtained by an integrated value or a variance value instead of the average intensity. .

  In addition to the liquid crystal array substrate of the present invention, it can be applied to inspection of an organic EL array substrate.

It is a figure for demonstrating the outline | summary of the TFT array test | inspection of this invention. It is a flowchart for demonstrating the extraction process of the cross point of the cross line defect of this invention. It is a figure for demonstrating the signal in the extraction process of this invention. It is a figure for demonstrating the process which specifies the defect point of this invention. It is an example of two-dimensional measurement data and a signal distribution profile. It is a figure which shows signal intensity distribution on the line of a perpendicular direction (y direction). It is a figure which shows the profile after a profile of measurement data, and a median filter process. It is a figure which shows a difference profile and a differential profile. It is a figure which shows a difference profile and a differential profile. It is a figure for demonstrating one structural example of the TFT array test | inspection apparatus of this invention. It is the example of a profile which plotted the signal strength of each point on a line. It is a flowchart for demonstrating the process sequence which extracts a line defect. It is explanatory drawing for demonstrating the process sequence which extracts a line defect.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... TFT array inspection device, 2 ... TFT substrate, 3 ... Inspection control device, 4 ... Inspection drive circuit, 5 ... Data processing means, 5A ... Line defect extraction processing, 5A1 ... Signal distribution profile creation processing, 5A2 ... Profile data evaluation Processing, 5B: Line defect determination processing, 6: Inspection probe source, 7: Detector, 10: Two-dimensional measurement data, 11, 11x, 11y ... Signal distribution profile, 12, 12x, 12y1, 12y2 ... Peak, 13 ... Candidate points, 14 ... defect points, 20-22 ... lines.

Claims (3)

In a TFT array inspection apparatus that inspects a TFT array based on two-dimensional measurement data obtained by driving each pixel of a TFT substrate,
A data processing means for obtaining a defect point of a line defect from two-dimensional measurement data of a TFT substrate is provided.
The data processing means obtains a signal distribution profile representing the characteristics of the line from measurement data of each line in the x direction and the y direction,
Find the y-direction position of the defect point from the signal distribution profile representing the line characteristics in the x-direction,
Obtain the x-direction position from the signal distribution profile representing the line characteristics in the y-direction,
Obtaining a candidate point of a cross-line defect from an intersection of the x-direction position and the y-direction position;
Obtain a correlation value in the x direction and y direction for each candidate point,
A TFT array inspection apparatus, wherein a defect point of a cross-line defect is obtained based on a magnitude of a correlation value difference obtained from the difference between the correlation values . The data processing means includes
A correlation value in the x direction is obtained from an average value of the intensities of a predetermined number of pixels in the x direction around the candidate point;
A correlation value in the y direction is obtained from an average value of intensities of a predetermined number of pixels in the y direction around the candidate point,
2. The TFT array inspection apparatus according to claim 1, wherein a difference between a correlation value in the x direction and a correlation value in the y direction obtained at each candidate point is obtained as a correlation value difference . About two-dimensional measurement data obtained by driving each pixel of the TFT substrate,
A signal distribution profile representing the characteristics of the line is obtained from the measurement data of each line in the x direction and the y direction,
Obtaining a line including a line defect from the signal distribution profile ;
A cross line defect candidate point is defined as a cross line defect candidate point, and a correlation value difference between the x-direction correlation value and the y-direction correlation value of the measurement data at the candidate point is obtained, and the cross-line defect defect is determined from the correlation value difference. A method for inspecting a TFT array, characterized by obtaining a point.

Images