JP3488707B2 - Pattern dimension measuring device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and apparatus for measuring a pattern size using a scanning probe such as a scanning electron microscope.

[0002]

2. Description of the Related Art Fine pattern dimensions are measured using a probe such as an electron beam, an ion beam, a light beam, or a mechanical probe. As a typical example, a case where the dimensions of a pattern processed on a wafer in a semiconductor manufacturing process are measured using a scanning electron microscope (length measurement SEM) will be described.

FIG. 1 is a schematic diagram showing the basic principle and configuration of a length measuring SEM. The electron beam 2 emitted from the electron gun 1 is narrowed down by the converging lens 3 and the objective lens 4 and focused on the wafer 5. At the same time, electron beam 2
Is bent by a deflector 6 to scan the wafer 5 two-dimensionally or one-dimensionally. Secondary electrons 7 are emitted from the wafer portion irradiated with the electron beam 2. The secondary electrons 7 are detected by the secondary electron detector 8 and converted into an electric signal, and then undergo processing such as amplification.

The secondary electron signal is used as a brightness modulation signal or a deflection signal of the display 9 such as a CRT. If the screen of the display 9 is two-dimensionally scanned in synchronization with the two-dimensional scanning of the electron beam 2 on the wafer surface and the luminance signal is modulated by the secondary electron signal, an SEM image is formed on the display 9. On the other hand, the main scanning signal of the electron beam 2 is displayed on the display 9
, And the secondary electron detection signal is the Y deflection signal of the display, the line profile is displayed on the display.

An example of a procedure for measuring a pattern dimension on a wafer using a length measuring SEM will be described below. The wafer 5 to be measured taken out from the wafer cassette 10 is pre-aligned with the orientation flat or notch of the wafer as a reference, and at the same time, the wafer number written on the wafer is read by a wafer number reader (not shown). The wafer number is unique to each wafer, and the recipe corresponding to the measured wafer 5 is read using the read wafer number as a key. The recipe defines the measurement procedure and measurement conditions of the wafer and is registered in advance for each wafer, and all the subsequent operations are performed according to this recipe.

After the wafer number is read, the wafer 5
Are transported into the sample chamber 11 held in vacuum and mounted on the XY stage 12. The wafer 5 loaded on the XY stage 12 is aligned using the alignment pattern formed thereon. The alignment pattern is the optical microscope 1 mounted on the upper wall of the sample chamber 11.
3 and the imaging device 14 are used to magnify the image about several hundred times. The captured alignment pattern is compared with a reference image registered in advance by the control device 15. The controller 15 aligns the wafer 5 by driving the drive motor 16 to correct the position coordinates of the XY stage 12 so that the captured pattern exactly overlaps the reference image.

After being aligned, the wafer 5 to be measured is moved to a predetermined measuring point. At the measurement point, the scanning electron beam 2 is irradiated and a high-magnification SEM image is formed. Using this SEM image, the pattern to be measured formed on the wafer is positioned with high accuracy. Similar to the alignment operation, the positioning is performed by comparing the SEM image of the length measurement point with the reference image and finely adjusting the electron beam scanning region so that the SEM image exactly overlaps the reference image. The measured pattern of the positioned wafer 5 is located substantially in the center of the screen, that is, immediately below the electron beam 2. In this state, the electron beam 2 scans the pattern to be measured one-dimensionally at the same scanning pitch (pixel size) as when forming the SEM image, and the pattern dimension is obtained from the obtained line profile according to a predetermined length measurement algorithm. When there are a plurality of measuring points on the wafer, the operation after the measuring point movement is repeated. When there are a plurality of wafers to be measured in the wafer cassette, the operation after the wafer is transferred to the sample chamber 11 is repeated.

There are various length measuring algorithms for obtaining the pattern size from the line profile, but generally, FIG.
The threshold method, the maximum slope value method, the linear approximation method, etc. shown in FIG. In FIG. 2, (a) schematically shows how the measured pattern 19 on the wafer 5 is scanned by the electron beam 2, and (b) shows the obtained line profile. Further, (c) is a threshold method, (d) is a linear approximation method, and (e) is a maximum slope value method.

In the threshold method, the position 2 where the slope portions at both ends of the line profile intersect a predetermined level (threshold value) L
1, 22 are obtained, and the distance W between the two positions 21, 22 is used as the dimension of the pattern 19. The threshold L is
It is usually defined as a percentage of the maximum height of the line profile. In the straight line approximation method, the slope portions at both ends of the line profile are linearly approximated by the least squares method, and the distance W between the intersections 23 and 24 of each straight line and the baseline is used as the pattern dimension. In the maximum inclination method, a tangent line is drawn on the slope portions at both ends of the line profile having the maximum inclination, and the intersection point 2 between the tangent line and the baseline is drawn.
The distance W between 5 and 26 is the pattern size.

[0010]

When positioning a pattern to be measured, a SEM image with a relatively low magnification is often used so that a desired pattern can be easily found from a large number of patterns. Also, when measuring the pattern dimension at a high magnification, the measurement may be performed while changing the magnification.

Even when the same pattern is measured, when the magnification is changed, there is a phenomenon that the measurement value and the measurement accuracy change depending on the magnification. For this reason, when performing observation at a low magnification for the above purpose, the measured value largely deviates from the actual size, and the variation in the measured value also increases. In addition, the linear approximation method and maximum slope method
The dependence of the measured value on the observation magnification is greater than that of the threshold method.

The present invention has been made in view of the above problems of the prior art, and accurately obtains the pattern dimension regardless of the setting of the length measurement algorithm for obtaining the pattern dimension from the observation magnification or the line profile. The purpose is to provide a possible method.

[0013]

FIG. 3 shows an example of measurement of SEM image magnification dependency of length measurement value, and FIG. 4 shows magnification dependency of variation (3σ) of length measurement value. The horizontal axis is the magnification. The figure shows the measured value by the maximum slope method and the measured value by the threshold method. 20%, 50%, 80% by threshold method
Represents the threshold level, and the graph shown as 20%, for example, is the measurement result when the threshold level is set to 20% of the maximum height of the line profile by the threshold method.

From FIGS. 3 and 4, it can be seen that when the threshold level is set to 20% by the maximum slope value method and the threshold value method, the measured value of the pattern width increases at a relatively low magnification and the measurement variation increases. . Also, the threshold level is set to 80
On the contrary, in the threshold method with%, the measured value of the pattern width becomes large when the measurement is performed at a high magnification. As a result of investigating the cause of the dependence of the measured value of the pattern size on the observation magnification, it was found that the observation magnification becomes smaller, the scanning pitch (pixel size) becomes larger, and the waveform dullness of the line profile increases. In particular, this tendency becomes remarkable when the scanning pitch is equal to or larger than the electron beam diameter.

FIG. 5 schematically shows the mechanism.
White circles drawn at the scanning pitch indicate sampling points by the electron beam 2, and the pixel size d is equal to the range covered by one sampling point. What is drawn as a line profile is a line profile created by data sampled at the scanning pitch. The numbers given as magnifications are for reference only.
As shown in (a), when the magnification is low, the waveform blunting of the line profile is large, and the pattern size W appearing in the line profile is larger than the actual pattern 19. As the magnification is increased in the order of (b), (c), and (d), the size W of the pattern appearing in the line profile changes.

[0016]

The present invention provides such a pattern dimension measurement value.
Based on the elucidation of the cause of the observation magnification dependence of
Scans the sample with a probe at a predetermined scan pitch.
And form a sample image using the scan signal obtained from the sample
The predetermined part of the pattern to be measured in the sample image.
In the pattern dimension measuring method, in which the dimension of a predetermined portion is obtained by scanning the obtained scanning signal with the threshold method,
The low magnification region is characterized in that the threshold value is set to a level of about 50%.

Further, the present invention is characterized in that an algorithm for obtaining a dimension by a threshold method after a straight line approximation or a maximum slope line is drawn is used as a length measurement algorithm. Further, the present invention is characterized in that the length measurement algorithm is switched to the threshold method in the low magnification region. According to the present invention, the dependence of the dimension measurement value on the observation magnification is reduced, so that it is possible to obtain a measurement value close to the actual dimension even when observing at a low magnification, and to perform measurement with high reproducibility.

[0019]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. Here, a case where the pattern dimension on the wafer is measured by using the length measurement SEM will be described as an example. FIG. 6 is an explanatory diagram of Reference Example 1 . This reference example 1
Then, when observing at a low magnification, the scanning pitch when positioning the measured pattern and the scanning pitch when determining the pattern dimension are automatically switched so that the scanning pitch when determining the pattern dimension is approximately the electron beam diameter. Make it fine.

Observation and measurement are performed according to the operation flow shown in the figure. First, the wafer to be measured is mounted on the XY stage (S1), and wafer alignment is performed using the alignment pattern on the wafer (S2). After that, the XY stage is moved to the length measurement point (S3), and the pattern to be measured is positioned using the specific pattern in the SEM image 61 so that the pattern p to be measured is located immediately below the electron beam (S4). In the SEM image 62 at the time when the positioning of the measured pattern p is completed, the measured pattern p is located at the center of the screen. Subsequently, as shown by enclosing with a broken line in the SEM image 63, the measured portion in the pattern p is designated (S
5).

When step 5 is completed, the scanning pitch is automatically switched so as to be approximately the same as the electron beam diameter (S6), and the pattern dimension is measured at this small scanning pitch (S7). The electron beam scanning area at this time becomes like 64. When the relative dimension of the pixel of the SEM image at the time of positioning in step 4 is represented by the area 66, the pixel dimension of the SEM image at the time of measuring the pattern dimension in step 7 is represented by the area 67.

As described above, regardless of the observation magnification when positioning the pattern to be measured, the scanning pitch is switched to a predetermined pitch suitable for accurately measuring the actual size of the pattern when measuring the pattern size, so that the same accuracy is always achieved. It is possible to measure the dimensions with, and it is possible to prevent the variation of the length measurement value due to the difference in magnification. FIG. 7 is an explanatory diagram of the first embodiment. In this embodiment, the threshold value is automatically set to a level of about 50% during observation with the low magnification and the threshold method.

FIG. 7A is a diagram for explaining pattern dimension measurement at high magnification by the threshold method. The pattern dimension measurement is performed in the designated broken line area in the SEM image shown on the left, and the line profile shown on the right is obtained. In the case of a high magnification, as can be seen from FIG. 3, since the dependency of the length measurement value on the magnification is small, even if the threshold level L1 is arbitrarily set, the variation of the length measurement value due to the magnification is small.

However, when the magnification is low, as can be seen from FIG. 3, the variation of the length measurement value due to the magnification is large at the threshold values of 20% and 80%, but the variation of the measurement value due to the magnification is small at the threshold value of 50%. Therefore, as shown in FIG. 7B, when measuring with the threshold method at a low magnification, the threshold L2 is automatically set to about 50% to minimize the fluctuation of the length measurement value due to the difference in magnification. can do.

FIG. 8 is an explanatory diagram of the second reference example . This ginseng
In consideration example 2 , when observing with the low magnification and the linear approximation method or the maximum slope value method, the measurement value is set to the value obtained by subtracting the predetermined value determined by the observation magnification from the value obtained from the linear approximation line or the maximum slope line. to correct. FIG. 8 is an example of the maximum inclination value method, and it is assumed that the pattern dimension obtained by the maximum inclination value method from the line profile is W1 when the magnification is 50,000 times. In this case, the pattern dimension W is calculated and displayed by the following equation.

W = W1−ΔW Here, ΔW is, as shown in FIG. 3, a length measurement value by the maximum slope value method at a magnification of 50,000 times and a length measurement value by the threshold method with a threshold value of about 50%. Is the difference. As described above, when the magnification is low, the variation of the length measurement value is small according to the threshold value method with the threshold value of about 50%. By converting to the measured value by
It is possible to obtain a length measurement value that does not depend on the magnification. As is clear from FIG. 3, since ΔW has a magnification dependency, the value of ΔW is obtained and stored in advance for each magnification, and the value of ΔW corresponding to the magnification is used in the above calculation.

The maximum slope value method has been described here.
Also in the linear approximation method, by converting the length measurement value to the length measurement value by the threshold value method in which the threshold value is set to about 50% by the same method,
It is possible to eliminate the magnification dependency of the length measurement value at low magnification. FIG. 9 is an explanatory diagram of the second embodiment. In this embodiment, as shown in FIG. 9A, a tangent line is drawn on the slope portion of the line profile with the maximum inclination, and the upper side is linearly approximated to linearly approximate the entire line profile to a trapezoidal shape. After that, as shown in FIG. 7B, the pattern dimension W is set by the threshold method in which the threshold value is about 50% with respect to the tangent line.
Ask for. It is clear from FIG. 5A that the measured value by this method reflects the actual size.

FIG. 10 is an explanatory diagram of the third embodiment. In this embodiment, the maximum slope value method or the linear approximation method is automatically switched to the threshold value method when measuring the length at a low magnification. Now, it is assumed that the measurement method is set to the maximum slope value method in the length measurement SEM. As shown in FIG. 10A, when measuring the length at a high magnification, the pattern dimension W is measured by the maximum inclination value method as set. If the magnification is changed to a low magnification for some reason, as shown in FIG. 10B, the waveform blunting of the line profile becomes large. Therefore, when the maximum slope value method is used to measure the length according to the setting, the error becomes large. Therefore, in this embodiment, when the measurement method is set to the maximum slope value method or the linear approximation method and the length is measured at a low magnification, the measurement method has a threshold value at which the dependency of the measurement value on the magnification is small is about 50.
The% threshold method shall be automatically switched.

Here, the length measurement SEM using the electron beam as the probe has been described as an example. However, the probe is a light beam (laser scanning microscope), an ion beam (FI
B) or a mechanical probe (scanning atomic force microscope). Although the case of one probe and one pixel has been described as an example, the present invention can be applied to a multi-probe and a multi-pixel type in addition to this.

[0030]

According to the present invention, the dependency of the length measurement value on the magnification can be reduced even in the low magnification region, and the length measurement accuracy can be improved.

[Brief description of drawings]

FIG. 1 is a diagram for explaining the principle and configuration of an SEM.

FIG. 2 is an explanatory diagram of a method for obtaining a pattern dimension from a line profile.

FIG. 3 is a diagram showing an example of magnification dependency of a length measurement value.

FIG. 4 is a diagram showing a relationship between variation in length measurement values and magnification.

5A and 5B are views for explaining a mechanism of magnification dependency of a dimension measurement value.

FIG. 6 is an explanatory diagram of Reference Example 1 .

FIG. 7 is an explanatory diagram of the first embodiment.

8 is an explanatory diagram of Reference Example 2. FIG.

FIG. 9 is an explanatory diagram of the second embodiment.

FIG. 10 is an explanatory diagram of the third embodiment.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Electron gun, 2 ... Electron beam, 3 ... Converging lens, 4 ... Objective lens, 5 ... Wafer, 6 ... Deflection coil, 7 ... Secondary electron, 8 ... Secondary electron detector, 9 ... Display, 10 ... Wafer Cassette, 11 ... Sample chamber, 12 ... XY stage, 1
3 ... Optical microscope, 14 ... Imaging device, 15 ... Control device, 1
6 ... Drive motor, 19 ... Pattern

─────────────────────────────────────────────────── ─── Continuation of the front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G01B 21/00-21/32 G01B 15/00-15/08

Claims (2)

(57) [Claims] 1. A sample is scanned by a probe at a predetermined scanning pitch, a sample image is formed using a scanning signal obtained from the sample, and a predetermined portion of a measured pattern in the sample image is scanned by the probe, In the pattern dimension measuring device for obtaining the dimension of the predetermined portion by processing the obtained scanning signal by the threshold method, when measuring the length at a magnification lower than the predetermined magnification, it does not depend on the set threshold level at that time and automatically. A pattern dimension measuring apparatus comprising means for setting a threshold value to about 50%. 2. A sample is scanned by a probe at a predetermined scanning pitch, a sample image is formed using a scanning signal obtained from the sample, and a predetermined portion of a pattern to be measured in the sample image is scanned by the probe, In the pattern dimension measuring apparatus for processing the obtained scanning signal according to a predetermined length measurement algorithm to obtain the dimension of the predetermined portion, when the magnification is changed to a lower magnification than the predetermined magnification, the maximum inclination method or the linear approximation method is used to measure the length. A pattern dimension measuring apparatus comprising means for automatically switching an algorithm to a threshold method.