JP4301261B2 - Pattern matching method and apparatus - Google Patents

Description

  The present invention relates to a scanning electron microscope, and more particularly to a pattern matching method and apparatus that can suitably perform alignment to an inspection position on a semiconductor integrated circuit.

  With the recent integration of semiconductor elements, a scanning electron microscope is used for observation and inspection of miniaturized circuit patterns. Scanning electron microscopes (hereinafter referred to as CD-SEMs) that measure the dimensions of specific patterns used in semiconductor manufacturing lines are being automated to reduce human dust generation and improve processing capacity, as with other devices. It has been.

  In order to automatically measure the target pattern on the wafer, the observation field is moved to an approximate position by moving the stage, the position of the measurement pattern is accurately determined from the observation field, and the field is moved to that position. Follow the procedure to measure. In order to perform automatic operation, the above sequence is stored as a file (hereinafter referred to as a recipe file). During automatic operation, the recipe file is read and the sequence is automatically executed. In order to detect an accurate position of the measurement pattern, an image portion (hereinafter referred to as a template) including a characteristic pattern serving as a guide is registered in advance, and is determined based on the distance from the pattern position detected by template matching.

  Japanese Patent Application Laid-Open No. 9-245709 discloses a technique in which a template serving as a guide is registered in advance and the position of a target measurement pattern is determined by matching using the template.

JP-A-9-245709

  The inspection position alignment of the sample by the template matching method has the following problems.

  First, in order to perform template matching, it is necessary to register a template as a guide in advance as described above. However, in order to register a template for template matching, it is necessary to introduce a sample such as a semiconductor wafer into the sample chamber and set an environment for observation. In addition, it takes a considerable amount of time to search for a pattern serving as a template.

  In addition, when performing alignment by template matching, the normalized correlation value was calculated for the image including the registered pattern and the front of the detected image. I was also calculating. For this reason, an erroneous position may be detected due to noise, sample charge-up, or non-uniform contrast.

  It is an object of the present invention to operate a scanning electron microscope with high accuracy and high speed by reducing processing relating to inspection alignment or input work.

  In order to achieve the above-mentioned object, the present invention detects a pattern based on electrons obtained by scanning an electron beam on a sample, and selects a desired pattern based on the detected pattern and a previously registered pattern. In a scanning electron microscope having a function of specifying a position, means for setting information on the type of pattern, the interval between a plurality of parts constituting the pattern, and the dimensions of the parts constituting the pattern, and the means obtained by the means The present invention provides a scanning electron microscope characterized by comprising means for forming a pattern image composed of a plurality of portions based on information.

  According to such a configuration, the template can be registered without setting the observation environment of the scanning electron microscope such as evacuation. In particular, a pattern that is arranged with a certain degree of regularity, such as a line pattern or a hole pattern formed on a semiconductor wafer, has the size and relative positional relationship of a plurality of portions forming the pattern. If there is information to specify, the pattern image can be specified.

  The present invention focuses on this point, and is configured so that necessary conditions are selectively input, and a pseudo pattern is formed based on the condition, so that an actual pattern image that has been performed so far is used. Therefore, it is not necessary to form an image using, and the working time can be greatly reduced.

  Furthermore, in order to achieve the above object, the present invention detects a pattern based on electrons obtained by scanning an electron beam on a sample, and selects a desired pattern based on the detected pattern and a previously registered pattern. In the scanning electron microscope having the function of specifying the position of the detected pattern, means for recognizing the number of detected patterns and / or intervals between a plurality of portions of the detected patterns, and recognized by the means A means for calculating an evaluation value based on a comparison between a number and the number of registered patterns, and / or an interval recognized by the means and a comparison between intervals between parts of the registered pattern The present invention provides a scanning electron microscope comprising means for calculating an evaluation value based on the degree of coincidence.

  For example, if a pattern consisting of multiple parts (line pattern or hole pattern) is crushed and line patterns or hole patterns are in contact, it is not necessary to measure or inspect the line or hole pattern. This pattern is difficult to match. According to the present invention, it is possible to determine whether or not it is appropriate as a target for length measurement or inspection by selectively evaluating the number and / or interval of each line pattern. Become.

  According to the present invention, it is possible to significantly reduce the processing or input work related to the template matching technique that has been used for inspection alignment so far, and to operate the scanning electron microscope with high accuracy and high speed. It becomes possible.

  An image obtained by a scanning electron microscope mainly includes information on the edge portion due to a large amount of secondary electron signals emitted from the edge portion of the pattern (edge effect). The edge portion of the pattern in the image is slightly smaller than the area of the entire image, but most of the pattern information of the sample is included. On the contrary, there is a lot of noise that has no pattern other than the edge portion or has little information to be used for alignment.

  A general template matching method uses a template to specify a pattern, but template matching obtains a normalized correlation value for all registered image parts and all detected images. Calculations were also made for missing parts (edgeless parts) and non-information noise parts. For this reason, an incorrect position may be detected due to noise, charge-up, or non-uniform contrast.

  In particular, in a semiconductor production line, a test wafer (hereinafter referred to as a conditioned wafer) in which conditions are changed for each chip for the purpose of obtaining optimum manufacturing conditions (for example, stepper exposure time, focus value change amount, etc.). ).

  A CD-SEM is used for the inspection of the conditioned wafer, but the pattern observed for each chip of the conditioned wafer becomes a pattern that greatly differs from the pattern shape as the deviation from the optimum manufacturing conditions. . Depending on the conditions, the resist may remain and the adjacent patterns may be connected, the pattern may fall down due to thinness, and the number may decrease. When the pattern shape or number changes in this way, it has been difficult to detect using template matching.

  Further, when registering the template, it is necessary to actually observe the sample pattern, take an image, and store the image portion. In addition, in the case of a conditional wafer, an operation to register a plurality of templates occurred. Skill is required to select a guide pattern and to determine the optimum magnification. Further, the success or failure of the detection is determined only by the normalized correlation value.

  For this reason, even if the number of patterns has changed, if the normalized correlation value is greater than or equal to the threshold value, the position where the pattern does not exist has been measured.

  The embodiment apparatus of the present invention can solve the above problem by performing processing related to template matching using information related to numerical values of pattern shapes and information related to positional relationships.

  Embodiments of the present invention will be described below with reference to the drawings.

  FIG. 1 is a schematic view of a scanning electron microscope according to the present invention. The electron beam 2 emitted from the electron gun 1 is deflected by the deflection coil 3, and then is narrowed down by the objective lens 4 and applied to the sample wafer 5 on the stage 6. The objective lens 4 is controlled by an objective lens control circuit 21 that can be controlled by the host computer 13. The scanning range and scanning position on the sample wafer 5 can be changed by a deflection signal generated by the deflection signal generator 8 and supplied to the deflection coil 3 amplified by the deflection amplifier 7.

  Secondary electrons emitted from the sample wafer 5 by the incidence of the electron beam 2 are converted into an analog electric signal by the detector 9, converted into a digital signal by the A / D converter 10, and then stored in the image memory 11. . The contents of the image memory 11 are always converted from a digital signal to an analog signal by the D / A converter 12 and applied to the grid as a luminance signal of the CRT 20. At this time, the A / D converter 10, the image memory 11, and the D / A converter 12 store the image signal after A / D conversion, and further, the timing signal for displaying the image after D / A conversion is a deflection signal. Received from the generator 8.

The deflection coil 19 of the CRT 20 is excited by the deflection amplifier 18 in accordance with the deflection signal from the deflection signal generator 8. The magnification is determined by the display width and scanning range of the CRT 20. The image processing device 16 is controlled by a signal from the host computer 13. The contents of the image memory 11 are transferred to the image memory 17 of the image processing device 16 for processing.

  FIG. 2 shows a sequence for registering pattern shape information (pattern type, pattern height, width, etc.) and positional relationship (interval between patterns, etc.) of the present invention. FIG. 3 is a diagram showing an example of a registration screen for registering shape information and positional relationships along the sequence of FIG.

  The host computer 13 displays a registration screen on the CRT 20 in accordance with a signal from the mouse 14 or the keyboard 15. The screen has a menu (301) for selecting the type of pattern, and a line pattern and a hole pattern can be selected (S2001).

The user inputs an input window (302a) having a line width Width (302), an input window (303a) having a length Height (303), an input window (304a) having a number Number (304), and a pitch Pitch (305) between line patterns. After input using the keyboard 15 in the input window (305a) (S2002), the OK button (306) is pressed. The host computer 13 obtains the magnification Mag (308) by the equation (1) based on the values from (302) to (305). Further, a pseudo pattern (309) displayed on the image when observed at the calculated magnification is created and stored in the CRT 20.
It is displayed as an image (307) together with Mag (308) (S2003). Display example
Shown in (a). The user confirms the input value by the pseudo pattern (309) (S2004). If there is a contradiction, steps (S2001 to S2004) are repeated. If there is no contradiction, the threshold value (311) Num. Accept. Of the number at the time of detection is input to the input window (311a) (S2005) and the registration button (310) is pressed. Upon receiving this signal, the host computer 13 receives the input information (302) to (305), the magnification (308), and the threshold value (313) Num. Accept. Is stored in the storage device 22 as a line detection file (312) in association with the recipe file 25 (S2006).

[Equation 1]
lwp = min (Width, Pitch-Width) (nm)
Mag. = Lpmin × rw / (pw × lwp) Equation (1)
However,
lwp: The smaller value of lw and lp−lw
rw: width of the image display area
pw: Display image width (pixel)
lpmin: Minimum processable width (pixel)

FIG. 4 shows a screen example (400) when a hole is selected. The user inputs an input window (401) for the hole diameter Diameter, an input window (402) for the number X-Num. In the x (column) direction, an input window (403) for the number Y-Num. In the y (row) direction, x ( After inputting using the keyboard 15 into the input window (404) of the pitch X-Pitch in the column) direction and the input window (405) of the pitch Y-Pitch in the y (row) direction (S2002), the OK button (406) Press. The host computer 13
A pseudo pattern (409) is created based on the values of (401) to (405). Further, the magnification Mag. (408) is obtained by Expression (2) and displayed on the CRT 20 together with the pseudo image (409) (S2003).

A display example is shown in FIG. The user confirms the input value with the pseudo image (407) (S2004). If there is a contradiction, steps (S2001 to S2004) are repeated. If there is no contradiction, the threshold value Num. Accept. Of the number at the time of detection is input to the input window (411) (S2005), and the registration button (410) is pressed. Upon receiving this signal, the host computer 13 stores the input information (401) to (404) and the magnification (408) in the storage device 22 in association with the recipe file 25 as a hole detection file (412) (S2006).

[Equation 2]
harea = max ((hnx−1) × hpx, (hny−1) × hpy) (nm)
mag = rw / (harea × 2) ... Formula (2)
However,
harea: (hnx-1) x hpx and (hny-1) x hpy, whichever is larger
rw: width of the image display area

  The pattern shape information and the positional relationship can be read out via the network 120 connected to the host computer 13. It is possible to read from the storage device 121 of the file server, the manufacturing condition file of the manufacturing device 123, or the storage device 123a or 124a of the manufacturing condition file of another manufacturing device 124.

  FIG. 5 shows a sequence for detecting a pattern using the pattern shape information and positional relationship of the present invention. The host computer 13 reads the recipe file 25 designated by the user with the mouse 14 or the keyboard 15 from the storage device 22 (S5001), and simultaneously detects the line detection file (312) (or hole detection file (412) associated with the recipe file 25. )) Is read out (S5002).

  Based on the measurement position information registered in advance in the recipe file 25, a drive signal is sent to the stage 6 and moved to the measurement position (S5003). At this time, the pattern shape information and the positional relationship are transferred to the image processing device 16. Further, the scanning range on the sample wafer 5 is determined according to the magnification (308) (S5004). An electron beam is irradiated onto the sample wafer 5 by the deflection signal generated by the deflection signal generator 8 based on this scanning range and amplified by the deflection amplifier 7 and supplied to the deflection coil 3, and the image memory 17 is operated according to the above procedure. An image is obtained (S5005).

In response to the detection processing signal from the host computer 13, the image processing device 16 executes detection processing (S 5006), and returns the number of detected patterns to the host computer 13. The host computer 13 compares with the number threshold value (313) (S5007), and if it is larger than the number threshold value (313), the image processing apparatus 16 executes pattern evaluation. The reason for executing the pattern evaluation when the number is larger than the number threshold is that noise or the like is taken into consideration, but the evaluation may be executed only when the number is the same.

  If the detected number is equal to or less than the number threshold (313), the pattern evaluation and measurement are skipped. The image processing apparatus 16 executes pattern evaluation according to this command (S5008), and returns an evaluation value to the host computer 13. The host computer 13 determines whether the pattern can be measured by comparing with the pattern evaluation threshold value of the recipe file (S5009). One aspect of this evaluation threshold is the number threshold (313) described above.

  If measurement is possible, a measurement edge is detected (S5100), measurement is performed (S5010), and measurement is skipped if it is determined that measurement is not possible. By configuring in this way, unnecessary calculations and scans can be avoided, and even if the number of patterns is reduced on a stepper conditioned wafer, the measurement pattern is detected and adjacent patterns are detected. Since it is possible not to measure a pattern that should not be measured, the useless processing can be reduced and the processing speed can be improved.

  Further, when detection is performed using the numerical information of the pattern shape and the positional relationship, it is possible to detect a pattern in which the shape such as the conditional pattern changes greatly and the number also changes. Furthermore, since numerical information of the pattern shape and the positional relationship are used, the template using the actual pattern image is not registered, so that the OFF line can be registered. In addition, it is not necessary to select a pattern as a guide that requires experience, and the registration work is simplified because numerical data is directly input. Since the operation is performed according to the threshold value input in advance based on the evaluation value of the pattern, the sequence can be optimized.

  When the above series of processing is completed, the next measurement position is read from the recipe file 25 (S5011). The host computer 13 performs measurement from the optimum measurement position based on the information of the manufacturing condition file from the manufacturing apparatus 123 and the manufacturing apparatus 124. The next measurement position is determined based on the number of detected patterns and the pattern evaluation value. For example, the direction in which the pattern shape deteriorates is determined from the number of detected patterns and the pattern evaluation value, and the pattern does not move to a measurement position outside the measurement position that is equal to or less than the threshold (S5012). This eliminates unnecessary processing and improves the processing speed.

  The image processing device 16 displays the result on the image memory 17 using the detected pattern position, pattern shape information, positional relationship, and mapping image. A display example of the result is shown in FIG. In the case of a line pattern, the result is displayed as in a line pattern image (800) in FIG. When the pattern to be detected is (801), the evaluation value a of the mapping image corresponding to each line pattern position obtained from the detected pattern position is displayed as in the graph (802). The height of each bar graph is a ratio based on the maximum value of the evaluation value a of the mapping image. In the case of the hole pattern image (803), a cross mark (804) is displayed as shown in FIG. The difference in the evaluation value of each pattern is reflected in the size of the cross mark.

  FIG. 6 shows detection processing in the image processing device 16 using the pattern shape information and the positional relationship. A pattern mask of a pattern shape to be detected is created from the pattern shape information pattern width (301) and pattern height (302) (S6001). The pattern mask is one pattern portion shown in FIGS. 4A and 4B. The evaluation value a is calculated while shifting the position of the pattern mask with respect to the entire image of the image memory 17 (S6002), and a mapping image is created in which the calculated evaluation value a is arranged corresponding to the pattern mask position (S6003). The evaluation value a is calculated by the formula (3), the formula (4), or the formula (5).

Here, P _ij is a density value at the point (X + i, Y + j) corresponding to the pattern mask of the line pattern image (800) or the hole pattern image (803), M _ij is a density value at the point (X + i + 1, Y + j + 1), N Is the number of pixels in the pattern mask.

Here, P _ij is the gradation value of the line pattern image (800) or hole pattern image (803), and N is the total number of pixels of the pattern mask.

Here, P _ij and P _avg are the average value in the pattern mask corresponding to the gradation value of the line pattern image (800) or hole pattern image (803), and N: the total number of pixels of the pattern mask.

Next, a positional relationship mask is created from the plurality of patterns (303) and the positional relationship (pitch (304)) (S6004). The line pattern positional relationship image (901) and the line pattern positional relationship mask (902) corresponding to FIG. 4 (a) and FIG. 4 (b), and the hole pattern positional relationship image (903) and the hole pattern positional relationship mask (904). ) Is shown in FIGS. 9 (a) and 9 (b). The total sum of the evaluation values a is obtained while shifting the position of the created positional relationship mask on the mapping image (S6005). The position where the total sum of evaluation values a is maximum is the position to be detected.
(S6006). In consideration of the decrease in the number of patterns in the positional relationship mask, the number of patterns is changed between the number of patterns (303) to the threshold value (311) (S6007), and (S6004) to (S6006) are repeated (S6008). With the above processing, the number of patterns and the detection position are obtained.

  A pattern shape evaluation procedure in the image processing apparatus 16 is shown in FIG. A portion having no pattern is selected from the mapping image created in (S6003) (S7001). For selection of a portion without a pattern, for example, a half value of the maximum value of the mapping image is set as a threshold value, and a portion below the threshold value is selected. The noise level (for example, variance) of the portion without the pattern is calculated by equation (5) (S7002).

Next, a portion where the pattern shape is likely to change is detected from the position detected in (S6006), and a signal level (for example, variance) of (S7003) is calculated (S7004). This part is often set between patterns. An evaluation value b (S / N ratio) is obtained from the noise level and signal level obtained in (S7002) and (S7003) (S7005) and returned to the host computer 13. For the hole pattern, the pattern position, the number, and the evaluation value b can be obtained in the same procedure.

An embodiment in which the scanning range of the electron beam is changed in accordance with the pattern shape from the input information will be described. This processing is used when a signal is obtained from a portion where the pattern shape is likely to change (S7004) or when a measurement edge is detected (S5100). When the signal is obtained from the portion where the pattern shape is likely to change, the host computer 13 determines the position detected in (S6006), the line width (302) and the pitch (305) (in the case of a hole pattern, the hole diameter (401), x, The scanning range is determined using the pitches (404, 405) in the y direction. An example of the scanning range of the line pattern and hole pattern is shown in FIG.

  If the measurement pattern is (1001) in the line pattern image (1000), the evaluation scanning range is at a position such as (1002). If the measurement pattern is (1004) in the hole pattern image (1003), the evaluation scanning range is (1005). FIG. 11 shows an example of the scanning range when detecting the measurement edge. If the measurement pattern is (1101) in the line pattern image (1100), the edge detection scanning range is at a position such as (1102). If the measurement pattern is (1104) in the hole pattern image (1103), the edge detection scanning range is (1105).

  According to the embodiment of the present invention, the observation magnification can be automatically determined by changing the area to be scanned on the sample to be detected from the numerical information of the pattern shape that changes the observation magnification. It is possible to automatically determine the optimum scanning direction for obtaining the evaluation value from the shape.

  Further, since unnecessary electron beam scanning is not performed, contamination caused by electron beams can be reduced. Since the numerical information and positional relationship of the pattern shape are detected by paying attention to the edges, they are not easily affected by the charge-up and contrast non-uniformity specific to the electron microscope image. In the registration work, the pattern shape and the positional relationship are input as numerical values, so that it is not necessary to detect a pattern and set a magnification that are a guide that requires experience. Also, since it is not necessary to actually observe the pattern on the wafer and register the template, a recipe file can be created offline.

It is the schematic of one Example of this invention. It is a figure explaining a registration sequence. It is a figure explaining the example of the registration screen which registers shape information and position information. It is a figure explaining the example of the registration screen which registers shape information and position information. It is a figure explaining the step which detects a pattern. It is a figure explaining the detection process step in an image processing device. It is a figure explaining the evaluation procedure of a pattern shape. It is a figure which shows the example of a display of the shape information of a pattern, and the evaluation result of positional relationship. It is a figure which shows the example of the positional relationship image of a hole pattern, and a positional relationship mask. It is a figure which shows the example of the scanning range of a hole pattern. It is a figure which shows the example of the scanning range in the case of detecting a measurement edge. It is a figure which shows the example which connected the scanning electron microscope of the present Example apparatus to the network.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Electron gun, 2 ... Electron beam, 3 ... Deflection coil, 4 ... Objective lens, 5 ... Sample wafer, 6 ... Stage, 7 ... Deflection amplifier 1, 8 ... Deflection signal generator, 9 ... Secondary electron detector, DESCRIPTION OF SYMBOLS 10 ... A / D converter, 11 ... Image memory, 12 ... D / A converter, 13 ... Host computer, 14 ... Mouse, 15 ... Keyboard, 16 ... Image processing device, 17 ... Image memory in image processing device, 18 ... CRT deflection amplifier, 19 ... CRT deflection coil, 20, 125b ... CRT,
DESCRIPTION OF SYMBOLS 21 ... Objective lens control circuit, 22 ... Host computer storage device, 25 ... Recipe file, 120 ... Network, 121, 122a, 123a, 124a, 125a ... Storage device, 122, 123, 124 ... Manufacturing device, 125 ... Computer.

Claims (7)

In a matching method for detecting a pattern based on electrons obtained by scanning an electron beam on a semiconductor wafer and performing matching between the detected pattern image and a pre-registered pattern image,
A threshold value is set in advance for the number of parts constituting the pattern, and the measurement pattern detected by the matching on the semiconductor wafer when the number of parts constituting the pattern falls below the threshold value A matching method characterized by skipping the measurement of the outer measurement position in. In claim 1,
The matching method, wherein the pattern is formed by a plurality of line patterns or a plurality of hole patterns. In a computer-readable recording medium recording a program for causing a computer to execute pattern matching between a pattern image of a semiconductor wafer obtained by a scanning electron microscope and a pattern image registered in advance,
In the program, a threshold value is set in advance for the number of a plurality of parts constituting the pattern, and the number of the plurality of parts constituting the pattern is detected by the matching when the number is less than the threshold value. A computer-readable recording medium storing a program, wherein the computer skips measurement of an outer measurement position of the measurement pattern on the semiconductor wafer. In claim 3,
The said pattern is comprised by the some line pattern or the some hole pattern, The computer-readable recording medium which recorded the program characterized by the above-mentioned. In a computer-readable recording medium recording a program for causing a computer to execute pattern matching between a pattern image of a semiconductor wafer obtained by a scanning electron microscope and a pattern image registered in advance,
To the program is set in advance the threshold with the number of the plurality of parts constituting the pattern, if the number of the plurality of parts constituting the pattern is below the threshold value, is detected by the matching so as not to perform the measurement of the outside of the measuring position is the pattern shape is deteriorated direction on the semiconductor wafer on the semiconductor wafer of the measurement pattern that, recording a program, characterized by controlling the scanning electron microscope Computer-readable recording medium. In claim 5,
The direction in which the shape of the measurement pattern is deteriorated is determined based on a comparison between a threshold value set in advance for the number of the plurality of portions constituting the pattern and the number of the portions constituting the detected pattern. The computer-readable recording medium which recorded the program characterized by the above-mentioned. In claim 6,
When the number of parts constituting the detected pattern falls below the threshold value, the direction outside the detected pattern in the semiconductor wafer is determined as a direction in which the pattern shape is deteriorated. A computer-readable recording medium on which a characteristic program is recorded.

Images