JP3288794B2 - Charge beam correction method and mark detection method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a charged beam drawing technique for drawing a fine pattern such as an LSI on a sample, and more particularly to a displacement, rotation, magnification, and the like of a shaped beam projected on the sample by a character projection method. The present invention relates to a method for correcting astigmatism and the like and a method for detecting a mark formed on a sample.

[0002]

2. Description of the Related Art Conventionally, an electron beam exposure apparatus using a variable shaped beam such as a rectangle or a triangle has been used to draw a desired pattern on a sample such as a semiconductor wafer. In such an electron beam exposure apparatus, when drawing a pattern having a complicated shape, the number of figures per unit area increases, so that there is a problem that the drawing time increases in proportion to this and the throughput is significantly reduced. .

As a method for solving such a problem, L
Several types of complex figure patterns repeatedly used on SI chips, etc. are processed and formed on an aperture mask, and each pattern figure on the aperture mask is selectively irradiated with a beam. There has been proposed an exposure apparatus that employs a method of projecting an image on a reduced scale (hereinafter, referred to as a “character projection method”). The pattern having the complicated shape is, for example, a pattern obtained by combining a plurality of rectangles and triangles, or a pattern having uneven portions.

[0004] In such an apparatus, a beam pattern or the like having a complicated shape obtained by combining a plurality of rectangles or triangles is used.
Since irradiation can be performed in a single operation, it is possible to greatly speed up drawing. However, the character projection method uses LSI
In order to draw a pattern, a method of correcting the position, magnification, astigmatism, and the like of a beam having a complicated shape such as a character beam with high accuracy is essential.

The details of the method of correcting a character beam will be described with reference to the drawings. FIG. 3 shows a schematic configuration diagram of the electron beam exposure apparatus. Reference numeral 10 denotes a sample chamber.
A sample stage 1 on which a sample 11 such as a semiconductor wafer is mounted
2 are accommodated. The sample stage 12 can be moved in the X direction and the Y direction by the sample stage drive circuit 31 that has received a command from the computer 30. The deflectors 25 and 26 are constituted by beam scanning deflectors for scanning a character beam on the sample 11 by a deflection signal from the deflection control circuit 33.

An electron detector 37 for detecting reflected electrons and the like from the sample 11 is provided in the sample chamber 10. The electron detector 37 is used for detecting a reflected electron from a mark of a fine gold particle located on the sample 11. The material and shape of the mark are not limited as long as the beam shape can be measured by the reflected beam reflected electrons or the like, and there is no problem if the height of the mark is substantially the same level as the sample surface.

Reference numeral 20 denotes an electron optical column, 21 denotes an electron gun, 22a to 22e denote electron lenses, 23 denotes a blanking deflector, 24 denotes a beam forming deflector, 27a and 27b denote beam forming apertures, and 28 denotes an axis. Matching coil, 30 is a computer, 32 is a laser measuring system, 33 is a deflection control circuit, 3
4 is a blanking control circuit, and 35 is a variable shaping beam size control circuit.

Next, the character projection system will be described with reference to FIGS. A rectangular hole is formed in the first aperture 27a. The image of this hole can be formed at an arbitrary position on the second aperture 27b of the character aperture by the lens 22c and the shaping deflector 24.

An LSI is provided on the character aperture 27b.
Since a plurality of frequently used characters 30a to 30d are machined in the chip, the beam can be shaped into a desired character by changing the beam deflection direction with the shaping deflector 24. Here, it is molded to 30a,
By irradiating the shaped character beam to a desired position on the sample 11 by the deflectors 25 and 26, L
The drawing shows how an SI pattern is drawn. 30e is
By superimposing the beam with the rectangular aperture of 27a,
This is an aperture for forming a rectangular or triangular beam.

In such an apparatus, the deflector 25,
When the voltage at 26 is constant, the irradiation position on the sample greatly changes depending on the arrangement coordinates of the character figure. Therefore, the fixed point of each character beam (the reference position of the beam)
It is necessary to define a different offset voltage (return voltage) for each character beam so that it matches on the sample surface.

Next, a method of measuring the two-dimensional distribution of the character beam will be described with reference to FIGS. First, the offset voltage of the deflectors 25 and 26 in FIG. 3 is set to 0 volt to form a character beam 30 on the sample surface 11, and this beam is scanned on the fine gold particles 31 in the X direction. Next, the beam is slightly shifted in the Y direction by the deflectors 25 and 26 and scanned again in the X direction. By repeating such scanning, a two-dimensional distribution of a character beam shape as shown in FIG. 6A can be obtained. However, since the resolution of the beam is finite here, the edge of the beam to be measured has a blurred shape as shown in FIG.

FIG. 7 shows the design data of the size, shape, inclination and position of the character beam on the sample.
Correlation between the aperture design data binarized by the presence or absence of the beam as shown in FIG. 7B and the measurement data as shown in FIG. 7A while changing the position, magnification and rotation of the aperture design data A function is obtained, and the point at which the correlation becomes maximum is defined as the position, magnification, and rotation angle of the measured beam. As for the astigmatism condition, the amount of astigmatism can be determined by changing the way of blurring in the direction of astigmatism generation, performing a process of blurring the aperture design data, and calculating a correlation with the blurring process. Based on the magnification, distortion (shape) and inclination of the measurement data obtained in this way, feedback is made to the exposure apparatus to correct each value.

More specifically, first, the design data A in FIG. 7 and the measured data A 'are matched on a computer. This can be achieved by applying a return (offset) voltage corresponding to each character beam to the deflectors 25 and 26. Similarly, the correction amount of the rotation angle required for the aperture 27b can be determined from the angle between a relatively long line segment in the character beam, for example, the line segment AB and the line segment A'B '. Further, the amount of magnification correction required for the objective lens can be determined from the difference in length. At this time, the vertices (for example, vertices C and C ') of the aperture design data and the measurement data may not match due to the astigmatism of the optical system. Direction and amount can be determined.

By feeding back the correction amounts thus obtained to the exposure apparatus, even a beam having a complicated shape can be measured with high accuracy, and as a result, the drawing accuracy can be improved.

The resolution of a beam image detected by the above method is determined by the size of a mark. Therefore, in order to accurately measure the complicated uneven shape of the character beam, it is necessary to use a mark sufficiently smaller than the unevenness. However, as the size of the mark becomes smaller, the S / N of the detection signal deteriorates and the position detection accuracy decreases, so that it is necessary to effectively perform noise processing on the captured image data. Such noise processing causes a problem that the processing time increases as the data amount increases.

On the other hand, in the electron beam exposure apparatus, at the time of overlay drawing, a mark provided on a substrate is beam-scanned from above a resist to obtain a mark waveform. At this time, local charge-up occurs depending on the material of the resist, the mark waveform becomes asymmetric, noise increases according to the thickness of the film, and deposits adhere to the edge of the mark. The waveform of the detected mark edge may be disturbed. For this reason, the mark detection accuracy has deteriorated.

With respect to noise, an average addition process or a method of taking in a waveform and then performing a filtering process is generally applied. In the case of a mark waveform disturbance (non-target, etc.), the portion is removed to calculate a mark position. I was However, there is a problem that such processing and calculation take a long time.

[0018]

As described above, in a method for performing high-precision beam calibration from a two-dimensional beam pattern shape measured using minute marks on a sample surface, noise must be effectively removed. However, the current situation is that this increases the time required for the entire beam calibration and, consequently, reduces the operation rate of the exposure apparatus. Therefore, there has been a demand for a method for calibrating the position, magnification, astigmatism, and the like of a beam having a complicated shape such as a character beam with high speed and high accuracy. In addition, when the shape of the mark waveform to be detected is distorted or noise increases during overlay drawing,
There is a problem that mark detection accuracy is deteriorated.

The present invention has been made in consideration of the above circumstances, and has as its object to provide a charged beam exposure apparatus capable of generating not only a rectangle and a rectangle but also a beam having a complicated shape such as a character beam. It is an object of the present invention to provide a charged beam correction method capable of performing high-speed and high-accuracy calibration of beam magnification, rotation, displacement, and astigmatism in an apparatus.

Another object of the present invention is to prevent the mark detection accuracy from deteriorating due to the collapse of the mark waveform and the increase in noise at the time of overlay drawing, and to perform the mark detection at high speed and with high accuracy. It is an object of the present invention to provide a mark detection method capable of detecting a mark.

[0021]

The gist of the present invention is to use a pattern obtained by convolving the design data of the beam pattern with a noise filter as a reference pattern.

That is, according to the present invention (claim 1), a shaped beam is generated by irradiating an aperture mask having a predetermined aperture shape with a charged beam, and the shaped beam is deflected and drawn at a desired position on a sample. A charged beam exposure apparatus for measuring a two-dimensional intensity distribution of a beam by scanning the beam at least once on a minute mark, calculating a correlation coefficient between the intensity distribution and a reference pattern, and calculating the calculation result. In the method of adjusting and correcting various parameters of the electron optical system based on the above, a pattern obtained by convolving the beam design data and the noise filter function as a reference pattern is calculated in advance, and the pattern is used at the time of adjustment. And

Further, according to the present invention (claim 2), a mark formed on a substrate is irradiated with a charged beam or light to measure the shape of the mark, and a correlation coefficient between the measured mark shape and a reference pattern is determined. In the method of calculating and obtaining the mark position based on the calculation result, it is necessary to previously calculate, as a reference pattern, data obtained by convolving the design data of the mark and the noise filter function as a template, and to use the template when measuring the mark. Features.
Here, preferred embodiments of the present invention include the following. (1) Use a pattern obtained by convolving the distribution corresponding to the electron reflectance of the minute mark with the design data as the reference pattern. (2) When calibrating the beam based on the calculated shift amount,
Correction by moving the mechanism of the charged beam exposure system. (3) Either a linear one-dimensional mark or a dot-like two-dimensional mark is used as the minute mark.

(4) Instead of using, as a template, data obtained by convolving the design data of the mark and the noise filter function, a mark waveform detected in a state where the resist is not applied, or calculated from an acceleration voltage and a mark material. Create a template using the mark waveform.

[0025]

According to the present invention (claim 1), the reference pattern of the beam can be provided with a noise filter function. Therefore,
By simply calculating the cross-correlation coefficient between the pattern with the noise filter function and the beam shape measured by the fine particles,
High-speed and high-precision correction of the character beam to the size, shape and direction according to the design data becomes possible.

Here, when comparing the design data with the measurement data, instead of previously calculating a pattern obtained by convolving the design data with the noise filter function as in the present invention, noise processing is performed on the measurement data as in the prior art. , A highly accurate correction is possible. However, calculation for noise processing requires extra time, and when performing noise processing on measurement data, it is necessary to wait for the time required for noise processing at the time of measurement. On the other hand, in the present invention, since the calculation for the noise processing can be performed in advance, such a waste of time can be omitted.

According to the present invention (claim 2), in the method for determining the mark position from the correlation coefficient between the measurement mark and the reference pattern, the reference pattern is provided with a noise filter function and has a high correlation with the measured value. Can be obtained as ideal data. Therefore, if such a template is calculated or measured in advance, even if noise, charge-up, or deterioration of the detected mark waveform due to mark processing or film formation occurs, a high-precision mark that is not affected by such deterioration can be obtained. Detection becomes possible.

[0028]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The details of the present invention will be described below with reference to the illustrated embodiments. (First embodiment)

First, the same parts as those in the conventional example are denoted by the same reference numerals and detailed description thereof will be omitted. The method of obtaining the size, shape inclination and position of the shaped beam different from the conventional example will be described in detail with reference to FIG.

FIG. 1A shows the size, shape, inclination, and amount of positional deviation measured from the aperture design data and the correlation coefficient of the measured two-dimensional distribution of the beam using the binarization template of the conventional method. FIG. On the other hand, FIG. 1B is a block diagram showing a measuring method according to the present embodiment. In this block diagram, a pattern 102 (hereinafter, referred to as an optimization template) obtained by convolving the design data 100 and the noise filter 101 is obtained in advance. This eliminates the need for noise filter processing as in the conventional example, and significantly reduces the time required for beam calibration. Here, if a function close to the electron reflectance of the fine particles is adopted as the distribution of the noise filter, the coincidence rate between the optimized template and the measured two-dimensional part is increased, and the detection accuracy is further improved.

Next, the above-described optimization template 102
A method of calibrating the beam using the method will be described with reference to FIG. First, an optimization template 102 is created by the convolution of the aperture design data 100 and the noise filter 101 (step a). Next, the deviation amount of the beam position is detected at the peak position of the correlation coefficient calculated from the optimization template 102 and the two-dimensional beam distribution. Then, the deflector 25,
Calibration is performed with a return voltage of 26. Thereafter, the above operation is repeated until the amount of deviation is eliminated (step b).

When calibrating the direction of the beam, the two-dimensional distribution of the beam is measured while rotating the second aperture 27b in FIG. 3, and the peak of the correlation coefficient calculated from the distribution and the optimization template is maximized. Thus, the rotation amount of the second aperture 27b is determined (step c). Calibration of astigmatism and magnification is performed on the astigmatic coil and the objective lens in the same manner as in the direction calibration (step d) (step e).
Finally, it is checked whether the calibration has been performed by the operations of (step b) to (step e), and if any part has been calibrated, the process returns to (step b) again.

As described above, according to the present embodiment, by using the optimization template having the functions of the aperture design data and the noise filter as the reference data used when calculating the correlation, the beam distribution can be measured every time. The noise processing required for (1) can be omitted. As a result, the size, shape, and inclination of the beam can be quickly and accurately calibrated with respect to the shaped beam of any shape including the rectangular beam and the character projection beam.

It is needless to say that the above-described correction method can be applied to an ordinary rectangular beam and also to an exposure apparatus having one or a plurality of shaping apertures. Also,
The above-described correction method does not depend on the method of measuring the two-dimensional distribution of the beam. That is, the present invention can be applied to a two-dimensional distribution measured using a one-dimensional mark. Further, in the present embodiment, the obtained beam size, shape and tilt are fed back to various parameters of the electron optical system to perform correction. For example, the tilt is corrected by moving a rotary table of a sample stage on which a sample is mounted. It is clear that this may be done. (Second Embodiment) Next, as a second embodiment of the present invention,
A method for detecting a mark formed on a sample will be described.

FIG. 8 is a block diagram for measuring a mark position from a mark design data and a correlation coefficient of a two-dimensional distribution of a measured mark waveform using a conventional binarization template. Detection data obtained by actually measuring the mark is used as input data, and the data is passed through a noise filter to remove noise. Then, the degree of correlation is calculated from the detection data subjected to the noise filter processing and the reference data (template) obtained from the design data, and the mark position is detected at the peak position of the correlation coefficient.

FIG. 9 is a block diagram showing a measuring method according to this embodiment. In this block diagram, a pattern 3 (optimized template) obtained by convolving the design data 1 and the noise filter 2 is obtained in advance.
This eliminates the need for noise filtering at the time of measurement as in the conventional example, and can significantly reduce the time required for mark detection.

Next, a mark detection method using the above-described optimization template 3 will be described with reference to FIG.
First, the optimization template 3 is calculated from the convolution of the mark design data 1 and the noise filter 2 as described above. Then, optimization template 3
And the mark position is detected at the peak position of the correlation coefficient calculated from the detected mark and the three-dimensional profile (input data) of the detected mark.

At this time, the calculation for creating the optimization template may be performed in advance, and the calculation for the noise processing is not required in the actual measurement. Therefore, the occupation time of the electron beam exposure apparatus in the mark detection can be reduced, and the operation rate of the apparatus can be improved.

As described above, according to the present embodiment, the mark position can be determined by using the optimization template having both the design data of the mark and the function of the noise filter as reference data necessary for calculating the degree of correlation. The noise processing required for each measurement can be omitted. As a result, the position of the mark can be detected at high speed and with high accuracy for marks of any shape. In this example, the beam is two-dimensionally scanned to obtain a three-dimensional mark waveform, but it is of course applicable to a waveform obtained by one-dimensional (line) scanning. (Third embodiment)

In the mark detection using the correlation coefficient, it is ideal that the point of high precision is to use data very similar to the detected mark waveform (that is, high correlation) as a template.

Therefore, in this embodiment, the mark width, depth,
Based on the material, the thickness of the material, the acceleration voltage of the electron beam, and the like, the energy of the electrons reflected on the substrate on which the mark is formed is determined, and the mark waveform is calculated from the energy. Then, using this mark waveform as an optimization template, mark detection is performed in the same manner as in the previous embodiment. (Fourth embodiment)

When a mark covered with a resist or an insulating film is measured in the overlay drawing, the mark waveform becomes asymmetric due to local charge-up as shown in FIG. As shown in (1), there is a case where a deposit adheres to the side wall of the mark during film formation and the waveform of the detected edge deteriorates. Even in such a case, the above-described second and third embodiments are effective, but a more effective method will be described.

In this embodiment, a typical mark waveform in a state where there is no resist or deposit is actually measured and obtained. Using this waveform as an optimization template, a correlation coefficient with measurement data is obtained to calculate a mark position. Thus, data having a higher correlation than a waveform obtained by calculation can be used as a template. As a matter of course, the data obtained by such measurement is convolved with a sufficient average addition process or a noise filter function, and the data obtained as a result can be used as a template to improve the accuracy.

In the second to fourth embodiments, the case where the electron beam exposure apparatus is used has been described. However, the present invention is not limited to this. Is applied to detect the image of the mark.

[0045]

As described in detail above, the present invention (claim 1)
According to the above, by using an optimization template having a combination of aperture design data and a noise filter function as reference data used when calculating a correlation degree, it is possible to omit noise processing required every time a beam distribution is measured. Can be. As a result, the size, shape, and inclination of the beam can be calibrated at high speed and accurately with respect to the shaped beam of any shape including the rectangular beam and the character projection beam, so that a charged beam exposure apparatus with a high operation rate can be realized.

According to the present invention (claim 2), by using an optimization template having both the design data of the mark and the noise filter effect as the reference data used when calculating the correlation coefficient, it is necessary to detect the mark. Noise processing can be omitted. Further, by using the simulation result of the mark waveform detected in the optimization template or the measured value of the mark waveform in a state where the substance covering the mark is removed, even if the mark is covered with a resist or a film, The mark position can be accurately detected.

[Brief description of the drawings]

FIG. 1 is a block diagram for explaining a first embodiment of the present invention.

FIG. 2 is a flowchart for explaining a beam calibration method according to the first embodiment.

FIG. 3 is a schematic configuration diagram showing an electron beam exposure apparatus used in a conventional calibration method.

FIG. 4 is a schematic diagram for explaining a character projection beam system.

FIG. 5 is a schematic diagram for explaining a beam scanning method.

FIG. 6 is a schematic diagram showing a measured beam intensity distribution.

FIG. 7 is a schematic diagram showing a method for comparing measured data and aperture design data.

FIG. 8 is a block diagram for explaining a conventional example of detecting a mark position using a correlation coefficient.

FIG. 9 is a block diagram for explaining a method of calculating an optimization template according to the second embodiment.

FIG. 10 is a block diagram for explaining a mark detection method in a second embodiment.

FIG. 11 is a signal waveform diagram for explaining the fourth embodiment and showing that a mark waveform is disturbed by the influence of a resist or a deposit.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 ... Sample chamber 11 ... Sample 12 ... Sample stand 20 ... Electron optical column 21 ... Electron gun 22a-22e ... Lens 23-26 ... Deflector 27a, 27b ... Beam shaping aperture 30 ... Computer 31 ... Sample stand drive circuit 32 ... Laser measuring system 33 ... Deflection control circuit 34 ... Blanking control circuit 35 ... Variable shaping beam size control circuit 100 ... Design data 101 ... Noise filter 102 ... Optimization template 1 ... Design data 2 ... Noise filter 3 ... Optimization template

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-4-123419 (JP, A) JP-A-4-124810 (JP, A) JP-A-61-116852 (JP, A) JP-A-2- 250311 (JP, A) JP-A-3-266415 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) H01L 21/027 G03F 7/20

Claims (2)

(57) [Claims] A charged beam exposure apparatus for irradiating an aperture mask having a predetermined aperture shape with a charged beam to generate a beam having the aperture shape, deflecting the beam, and writing the beam at a desired position on a sample. Using the beam, the beam is scanned at least once on a minute mark to measure a two-dimensional intensity distribution of the beam, perform a convolution calculation between the intensity distribution and a reference pattern, and calculate the convolution.
Based obtains beam position, a method of complement <br/> positive beam position by adjusting the various parameters <br/> electron optical system from the sought beam position, the plane of the beam as the reference pattern Convolution meter for design data showing shape and noise filter function
The calculated data is calculated in advance as a corrected calculated reference pattern, and when the beam position is obtained, the correction meter is used.
A charged beam correction method, wherein a beam position is directly obtained from a result of a convolution calculation using a calculated reference pattern . 2. A mark formed on a substrate is irradiated with a charged beam or light to measure the shape of the mark, and a convolution calculation between the measured mark shape and a reference pattern is performed.
A method for obtaining a mark position based on the result , wherein the design data indicating the planar shape of the mark and a noise filter function are used as the reference pattern by a convolution meter.
The calculated data is calculated in advance as a corrected reference pattern, and when the mark is measured, the corrected calculated reference pattern is used.
A mark position using a pattern .