JP5817383B2 - Pattern measurement method - Google Patents

Description

  The present invention relates to a pattern measuring method, and more particularly to a method for measuring a pattern from a secondary electron scanning image acquired from a scanning electron microscope.

  In recent years, with the miniaturization of semiconductor circuit patterns, the resolution of an optical exposure apparatus has reached its limit. The dimensions of the design pattern (mask dimension) and the dimensions of the transfer pattern (resist pattern) formed by transferring the design pattern onto the resist. The magnitude of the difference that occurs between (processed dimensions) cannot be ignored. In order to reduce the difference between the mask dimension and the processing dimension, a technique of previously adding a correction pattern to the photomask design data, that is, proximity effect correction (Optical Proximity Correction (OPC)) is used. ing. In general, a photomask pattern formed by the OPC technique is called an OPC pattern, and the dimensions of the OPC pattern are often measured in the photomask manufacturing process.

  When measuring the pattern dimensions of a semiconductor device, scan the semiconductor device with charged particles, obtain a scanned image from secondary electrons emitted from the sample surface, and analyze the image to determine the pattern dimension of the semiconductor device. taking measurement. Patent Document 1 listed below describes a technique for measuring the dimensions of an OPC pattern using a scanning electron microscope.

JP 2009-36572 A

  When measuring the pattern dimension of a photomask, an operator sets a measurement target region (Region Of Interest (ROI)) based on design data, creates an automatic measurement recipe, and performs measurement.

  Since a general photomask pattern not subjected to OPC processing is a rectangle having a constant width dimension, the ROI set in the automatic measurement recipe does not deviate from the desired pattern to be measured, and automatic measurement is possible. It was possible. However, the OPC-processed photomask pattern is often a polygon whose width dimension is not constant but has a stepped shape, or a shape with rounded corners, making it difficult to set an appropriate ROI. It is. Therefore, when the measurement is not performed properly, the operator has to manually set an appropriate ROI each time, and there is a problem that it takes a lot of time for the measurement.

  Patent Document 1 describes a technique for measuring the dimension of an OPC pattern, and sets a range until the amount of change in the difference between the contour line data of the measurement pattern and the design data exceeds a certain threshold as a measurement range. And a method of measuring is disclosed. However, there is a problem that an effective measurement area cannot be set when overlaying fails because an effective method for matching the contour line data and the design data is not described. Further, when the image quality of the scanned image is poor, there is a case where the outline data is not properly extracted, and there is a problem that an appropriate measurement region cannot be set in this case as well.

In order to solve the above-described problem, the invention according to claim 1 is directed to a pattern measurement method for measuring pattern dimensions at a plurality of locations from a single pattern image captured by a scanning electron microscope. A setting step of setting a measurement target region including the plurality of locations , a measurement step of detecting an edge position of a pattern included in the measurement target region and measuring the pattern dimension in the measurement target region, and the measurement target An estimation step of estimating the pattern dimensions of the plurality of locations from the measured values of the pattern dimensions in the region, and the estimation step includes a first threshold value group respectively corresponding to the plurality of locations from the design data of the pattern And setting the measured value to extract a plurality of regions respectively corresponding to the plurality of locations from the measured value; and Calculating a change amount of the measurement value of the turn dimension, extracting a second measurement target region whose absolute value of the change amount is equal to or less than a second threshold, one of the plurality of regions, and the first A step of estimating an average value of the measurement values of the pattern dimensions of the regions included in both of the two measurement target regions as a measurement value of the pattern dimensions at a location corresponding to any one of the plurality of locations; It is characterized by including.

The invention according to claim 2 is the pattern measurement method according to claim 1, wherein in the setting step, a range including all of the plurality of locations and larger than the plurality of locations is set as the measurement target region. Features.

  According to a third aspect of the present invention, in the pattern measurement method according to the first or second aspect, in the measurement step, a measurement box is set at both ends of the measurement target region, and the pattern is within the range of the measurement box. A line profile in a direction orthogonal to the edge of the pattern is acquired from the pattern image at predetermined intervals, the edge position is detected at each predetermined interval from a differential profile peak obtained by differentiating the line profile, and the predetermined interval is detected. The pattern dimension is measured every time.

  According to the present invention, the dimension of the OPC pattern of the photomask can be easily and accurately measured.

It is a figure which shows the observation image 10 of the photomask pattern 11. FIG. It is a figure explaining the process which sets a measurement object area | region to a photomask pattern. It is a figure which shows the observation image 30 of the photomask pattern 31. FIG. It is a figure which shows the result of having measured the line width of the photomask pattern for every micro area | region. It is the figure which processed the measurement result of line width statistically, and approximated it with the curve. It is a figure which shows the result of having measured the line width of the photomask pattern for every micro area | region. It is a figure showing the variation | change_quantity between adjacent dimension values. It is a figure which shows the observation image 80 of the photomask pattern 81. FIG. It is a figure which shows the result of having measured the line width of the photomask pattern for every micro area | region. It is a figure showing the variation | change_quantity between adjacent dimension values. It is a figure which shows the SEM image 110 which imaged the photomask pattern 111. FIG. It is a figure which shows the result of having measured the line | wire width of the photomask pattern 111 for every micro area | region. It is a figure showing the variation | change_quantity between adjacent dimension values.

  Hereinafter, a pattern measuring method according to the present invention will be described in detail with reference to embodiments shown in the drawings. First, a general method for measuring the OPC pattern will be described. Then, the pattern measuring method according to the present invention will be described.

  FIG. 1 is a view showing an observation image 10 of the photomask pattern 11. In the following drawings, the pattern dimension in the x-axis direction is called the length, and the pattern dimension in the y-axis direction is called the width. Here, an example of a general OPC pattern is shown. In order to correct the optical proximity effect, the width dimension near the center of the pattern is narrowed. In FIG. 1, an observation image 10 is an observation image that is scanned based on secondary electrons emitted from the surface by, for example, irradiating a pattern with a charged particle beam. An SEM (Scanning Electron Microscope) observation image corresponds to this.

  In FIG. 1, a portion whose width dimension near the center is narrowed is a measurement target of the pattern dimension. In order to measure the width dimension of the measurement object, it is necessary to detect the upper and lower edge portions. Therefore, as shown in FIG. 2, the ROI 12 is set to match the measurement target region. FIG. 2 is a diagram illustrating processing for setting a measurement target region in a photomask pattern. Since the width of the photomask pattern 11 gradually decreases toward the center, the corners are rounded. Since this portion is not a measurement target, the ROI 12 is set so as not to cover this region. There are measurement boxes 13 and 14 at both ends of the ROI 12, and they are set to an appropriate size so that the pattern edge enters the area of each measurement box.

  Next, in the measurement box, a line profile in the direction orthogonal to the edge is acquired from the scanned image, and the pattern edge position is determined. Generally, a line profile is generated by averaging the line profiles acquired for each pixel in a measurement box. As a method for determining the pattern edge position, there is a method in which a differential profile obtained by differentiating the line profile is generated and a differential peak is used as an edge. If the positions of the pattern edges at both ends are determined, the distance between them is measured as the width dimension. As described above, in order to measure the dimension of the OPC pattern, the ROI must be set to an appropriate size and measured.

  Hereinafter, a method capable of measuring the pattern dimension in the measurement target range without setting the ROI size in detail for the measurement target pattern will be described.

<Embodiment 1>
FIG. 3 is a diagram showing an observation image 30 of the photomask pattern 31. This OPC pattern is also a pattern in which the width dimension becomes narrow near the center, and this region becomes the measurement range (desired range). First, the ROI 32 is set to an appropriate size for the OPC pattern. The ROI 32 is desirably set to a size slightly larger so that the measurement target pattern is included, that is, a range that includes all the desired range and is larger than the desired range. Since the size of the ROI 32 is appropriate, this ROI includes regions other than the pattern of the measurement range. This indicates a portion with rounded corners, a region with a wide width dimension, or the like.

  Next, line profiles are acquired in the measurement boxes at both ends, pattern edges are determined, and width dimensions are measured. In general, the average value of the entire ROI region is measured as the width dimension, but in the present invention, the width dimension is measured by dividing each minute area. That is, the line profile in the direction orthogonal to the pattern edge within the measurement box is acquired from the pattern image at predetermined intervals (for each minute region), and the differential of the line profile is differentiated from the peak of the differential profile at predetermined intervals. The edge position is detected, and the pattern dimension (line width) for each predetermined interval is measured.

  FIG. 4 is a diagram illustrating a result of measuring the line width of the photomask pattern for each minute region. In FIG. 4, the horizontal axis represents the X coordinate, and the vertical axis represents the width dimension. The graph in FIG. 4 is not actually a continuous value but a line width value for each predetermined interval. In FIG. 4, the width dimension is small near the center of the X coordinate. FIG. 5 is a diagram in which the measurement result of the line width is statistically processed and approximated by a curve. The photomask pattern 31 is broadly divided into two regions: a region having a short line width near the center and a region having a long line width near both ends. Therefore, in FIG. 5, a data group appears in two, D1 having a short line width and D2 having a long line width. These D1 and D2 correspond to D1 and D2 shown in FIG. Since the photomask pattern 31 has a short line width as the measurement target area, the average value of the data group of D1 is estimated as the pattern dimension. Actually, the range of ± several nm from the peak value of the data group may be averaged, and the user may determine the averaging range as appropriate. According to this method, since it is not necessary to set the size of the ROI in detail, manual operation by the operator is unnecessary, and the width dimension of the OPC pattern can be measured easily and accurately.

<Embodiment 2>
In the second embodiment of the present invention, a modification of the method described in the first embodiment will be described. Since the steps up to and including FIG. 4 are the same as those of the first embodiment, only the subsequent steps will be described below.

  FIG. 6 is a graph similar to FIG. 4, and FIG. 7 is a diagram showing the amount of change between adjacent dimension values (adjacent at predetermined intervals) from the line width data of FIG. Th1 in FIG. 6 and E1 in FIG. 7 each represent a threshold value necessary for measuring a desired dimension value. The threshold value Th <b> 1 (first threshold value) is a threshold value that is necessary when measurement is limited to a desired measurement region when there are a short region and a wide region such as the photomask pattern 31. The threshold value Th1 may be determined in advance from design data. The threshold value E1 (second threshold value) is a threshold value necessary for limiting the measurement to an area where the amount of change between adjacent dimension values is small. The threshold E1 may be set to an appropriate value depending on the degree of roughness. According to this example, the amount of change between adjacent dimension values is larger than the threshold value E1 in the region where the roundness of the corners and the dimension value change greatly. Therefore, the region exceeding the threshold value E1 can be removed from the measurement region, and the influence of the roundness of the corner can be removed.

  In the photomask pattern 31 of FIG. 3, the measurement target includes a region whose value in FIG. 6 is smaller than the threshold Th1. Therefore, first, Xa in FIG. 6 can be limited as the measurement target region. Next, a region where the value of the change amount in FIG. 7 is smaller than the threshold ± E1 is a measurement target region. In FIG. 7, Xb, Xc, and Xd are applicable. Since the measurement target region is a region within the range of Xa and included in any of Xb, Xc, and Xd, Xc can be limited as the final measurement target region in this example. Among the measurement results of FIG. 6, the average value of the line widths within the range of Xc is output as the measurement result. According to this method, the width dimension can be measured accurately and easily even with an OPC pattern having a complicated shape.

<Embodiment 3>
In the third embodiment of the present invention, a case where line widths at a plurality of locations are measured within one OPC pattern will be described.

  FIG. 8 is a view showing an observation image 80 of the photomask pattern 81. The ROI 82 is set so that the entire OPC pattern is included. Next, line profiles are acquired in the measurement boxes 83 and 84 at both ends, the pattern edge is determined, and the width dimension is measured. FIG. 9 is a diagram illustrating a result of measuring the line width of the photomask pattern for each minute region. A threshold value is set in advance from the design data so that the width dimension of the measurement region can be measured. Th2, Th3, and Th4 in FIG. 9 each represent a threshold value for measurement with a measurement region limited. In this example, since there are four areas to be measured in the OPC pattern 81, three threshold values are set. FIG. 10 is a diagram showing the amount of change between adjacent line widths from the line width data of FIG. The threshold E2 is set to an appropriate value depending on the degree of roughness.

  Here, for example, when measuring the measurement area A1, the average value of the width dimensions of the area where the value in FIG. 9 is smaller than Th2 and the value in FIG. 10 is smaller than E2 may be output as the measurement result. The region corresponds to Xj in FIG. In the case of the measurement region A2, the value in FIG. 9 is a region between Th2 and Th3 and the value in FIG. 10 is smaller than E2. The region corresponds to Xk in FIG. Similarly, in the other measurement regions, the measurement region may be determined from the width dimension threshold value and the change amount threshold value, and the average value of the width dimensions of the region may be output as the measurement result. Therefore, according to this method, it is possible to measure the width dimension of a plurality of locations only by setting one ROI for an OPC pattern having a plurality of measurement locations.

  Specific examples of the pattern measuring method of the present invention will be described below.

  FIG. 11 is a diagram showing an SEM image 110 obtained by imaging the photomask pattern 111. The photomask pattern 111 has a narrow line width near the center due to OPC, and this portion is a measurement target region. The ROI 112 was set so that the measurement target region of the photomask pattern 111 was included. Next, measurement was performed by dividing the line width into minute regions. Here, the measurement width is 20 pixels and the measurement interval is 5 pixels. Appropriate values for the measurement width and the measurement interval may be determined from the resolution of the SEM image. FIG. 12 is a diagram illustrating a result of measuring the line width of the photomask pattern 111 for each minute region. The horizontal axis represents the X coordinate (the left end of the ROI 112 is the origin), and the vertical axis represents the measured line width. Th5 in FIG. 12 is a threshold for limiting the measurement region, and is set to 200 nm from the design data. Since the measurement target area is smaller than the threshold Th5, the range of about 50 pixels to 160 pixels in the X direction corresponds to the measurement target area. FIG. 13 is a diagram illustrating the amount of change between adjacent dimension values. Here, the threshold value E3 is set to ± 2.0 nm. A region smaller than the threshold value E3 in the range of the measurement target region is determined as a final measurement target region. Here, 70 to 140 pixels in the X direction are measurement target areas. Next, when the average value of the measurement values of the line width of the measurement target region was calculated as a measurement result, it was calculated to be 152.6 nm. From the above, it was possible to calculate the line width of the measurement target region of the photomask pattern 111 without setting the ROI size in detail.

  Since the pattern measurement method of the present invention can accurately and easily measure the line width of a complex OPC pattern without setting the ROI in detail, it can be used in the field of measuring pattern dimensions of semiconductors, photomasks, nanoimprints, and the like. Expected to be used.

10, 30, 80, 110 ... pattern observation image 11, 31, 81, 111 ... photomask pattern 12, 32, 82, 112 ... ROI
Th1, Th2, Th3, Th4, Th5... Dimensional threshold value Xa, Xb, Xc, Xd, Xj, Xk ... Range to be measured region

Claims (3)

In the pattern measurement method for measuring the pattern dimensions of a plurality of locations from a single pattern image captured by a scanning electron microscope,
A setting step for setting a measurement target region including the plurality of locations for the pattern image;
A measurement step of detecting an edge position of a pattern included in the measurement target region and measuring the pattern dimension in the measurement target region;
Estimating the pattern dimensions of the plurality of locations from the measured values of the pattern dimensions in the measurement target region, and
The estimation step includes
Setting a first threshold value group corresponding to each of the plurality of locations from the design data of the pattern, and extracting a plurality of regions respectively corresponding to the plurality of locations from the measurement values;
Calculating a change amount of the measurement value of the pattern dimension in the measurement target region, and extracting a second measurement target region whose absolute value of the change amount is a second threshold value or less;
The average value of the measurement values of the pattern dimensions of the areas included in both of the plurality of areas and the second measurement target area corresponds to any one of the plurality of areas. Including a step of estimating as a measurement value of the pattern dimension of the location,
A pattern measuring method characterized by the above. The pattern measurement method according to claim 1, wherein in the setting step, a range including all of the plurality of locations and larger than the plurality of locations is set as the measurement target region.   In the measurement step, measurement boxes are set at both ends of the measurement target region, line profiles in a direction perpendicular to the edge of the pattern within the measurement box are obtained at predetermined intervals from the pattern image, and 3. The pattern measurement according to claim 1, wherein the edge position at each predetermined interval is detected from a peak of a differential profile obtained by differentiating a line profile, and the pattern dimension is measured at each predetermined interval. Method.

Images