JPS6145858B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method for determining a reference point for pattern depiction in electron beam exposure, and more particularly to a method for confirming a position in two stages by detecting alignment marks.

Lithography using electron beams has much better resolution than that using ultraviolet rays, so
It is used to form ultra-fine patterns for LSI etc. Electron beam lithography differs from ultraviolet or X-ray lithography in that it does not use a mask. That is, the electron beam is scanned by an electron beam deflection device to expose the necessary areas to light, and this is carried out in response to instructions from a computer which has been given a program in advance.
In that case, selection of a reference point for generating a deflection signal for pattern depiction by electron beam scanning becomes an important issue. This is important when a new pattern is drawn on a semiconductor substrate that has undergone a process using one pattern at a position determined in accordance with the previous pattern.

Therefore, during the first pattern drawing, a figure indicating the reference point for the second and subsequent pattern drawings is also drawn, and in the second and subsequent electron beam exposures, the reference point is used. It is necessary to take measures such as detecting the electron beam and moving the irradiated object to the correct position based on the information, or generating an electron beam deflection signal for pattern depiction in accordance with the detection.

For this purpose, conventionally, as shown in FIG. 1, for example, L-shaped marks 2 are placed at the four corners of the level patterns 1 of the IC arranged in a matrix. The wafer is pre-aligned with relatively high precision (so-called Briar Line, hereinafter this name will be used), the L-shaped mark is detected through trial irradiation, and the desired pattern is drawn based on that information. method was adopted. In this case, while the IC pattern has a size of several mm square, the length of one stroke of the L-shaped mark is about several tens of micrometers, but if the position detection mark itself is small like this, A trial scan must be performed over a wide area to pick up the pattern, which not only increases the time required, but also causes the electron beam to irradiate areas that should not be irradiated with the electron beam, making it difficult to obtain the desired pattern. There is also a risk of damage to the Therefore, in order to avoid this, highly accurate pre-alignment must be performed.

Silicon wafers are usually provided with notches called facets or orientation flats to indicate the crystal orientation within the plane of the wafer. Although it is possible to achieve alignment within 1/1000 to 1/2000 radian in the circumferential direction (hereinafter referred to as the θ direction), it is difficult to improve the accuracy of the Briar line in the radial direction, especially in the direction parallel to the facet. be.

The present invention solves this problem by providing one or several relatively large alignment marks and detecting the alignment marks after relatively rough pre-alignment. This paper presents a method to solve this problem by moving the body to its normal position.

FIG. 2 shows one embodiment of the present invention, in which 11 is silicone ware, 12 is a matching mark, and 13 is, for example, an FC pattern. As is clear from the diagram, the overall size of this alignment mark is
It is similar to an IC pattern, specifically 1 to 2 mm.
It is about the size of a square. The advantage of using such a large alignment mark is that even with a relatively rough pre-alignment, electron beam irradiation for position detection can be limited within that area. On the other hand, by scanning such a wide area with an electron beam, specific points within the area,
For example, if you try to detect the center point, the total scanning length will be large and the time required will be long.
In order to solve this problem, it is necessary to give the alignment mark pattern special properties. This point will be discussed in detail later.

It is clear that by using two such alignment marks, the accuracy of alignment in the θ direction can be made extremely high. In electron beam exposure, the area on which a pattern can be drawn by deflecting the electron beam is approximately a few square millimeters, and beyond that it must be mechanically fed, so alignment in the θ direction is performed at this stage. Correctly aligning the pattern repeat direction with the mechanical feed direction facilitates subsequent precise alignment.

Once it is determined how much the wafer has deviated from its original position, the wafer is moved in the X and Y directions and rotated in the θ direction to place it in the correct position based on this information. Next, the areas where a pattern is to be depicted are sequentially sent to the electron beam irradiation range to depict a predetermined pattern.
At this time, as shown in Figure 1, if a minute alignment mark is provided for each unit pattern and the electron beam deflection signal for pattern depiction is corrected by detecting the position, the pattern can be drawn correctly at the desired position. I can draw.

As mentioned above, providing alignment marks for each pattern has been conventionally done, but while the conventional technology requires precise pre-alignment by hand, the method of the present invention uses electron beam irradiation. The high-precision part of pre-alignment is performed by position detection, which not only requires relatively rough manual pre-alignment, but also enables the size of alignment marks provided for each pattern to be smaller. , the time required for detecting the alignment mark at that stage is shortened, and furthermore, no damage to the pattern occurs. In particular, two or more first alignment marks 12 are provided, and θ is
By also performing directional alignment, the electron beam irradiation area is moved from the first alignment mark 12 to the second alignment mark 2 provided for each pattern, and the second alignment mark 2 is detected. The time required is significantly reduced. In other words, unless alignment is performed not only in the X and Y directions but also in the θ direction at the pre-alignment stage, it will not be possible to miniaturize the alignment marks 2 provided for each pattern and shorten the time required for detection. .

Next, the graphical characteristics of the relatively large alignment mark used for alignment in the first stage will be explained.
As already mentioned, if the alignment mark is made larger, the accuracy of pre-alignment may be lower, but a wider range must be scanned before confirming a specific point, which increases the time required. In order to avoid this, the pattern must have such a characteristic that it is possible to know the position of a specific point within the pattern, such as the center point, simply by scanning a portion of the pattern. On the other hand, there are restrictions as described below regarding the constituent elements of the alignment mark.

Detection of reflected electrons is commonly used as a means of obtaining information regarding the irradiation position of the electron beam. The amount of reflected electrons varies depending on the material and shape of the object irradiated with the electron beam. Therefore, the alignment mark is made by attaching a different substance to the surface of the object to be irradiated, or by changing the shape of the surface by means such as etching.

When the irradiation target is a silicon wafer, providing a metal layer on the surface is effective in changing the amount of backscattered electrons, but it is not stable against thermal or chemical treatments performed after patterning. Therefore, it is necessary to select materials that do not have a negative effect on the characteristics of the IC.
Considering the disadvantages of providing a step of selectively depositing a metal layer solely for the purpose of creating an alignment mark, the use of a metal layer alignment mark is a less advantageous method.

Further, although silicon dioxide is a thermally and chemically stable material, the amount of backscattered electrons generated is not much different from that of silicon, making it difficult to take advantage of the difference in materials. Therefore, it is preferable that the alignment mark is formed by the unevenness of the silicon itself or the unevenness caused by the selective deposition of silicon dioxide.

Furthermore, the pattern detection method has the following restrictions. That is, in order to detect reflected electrons, a detection element, such as a photodiode, is usually placed surrounding the electron beam so that the function does not have anisotropy. The pulse-like signal that appears when scanning a step-like part of the irradiated object is the most reliable information, and it is difficult to detect the difference in height between flat surfaces or the direction of a slope. Not suitable. Since a resist is applied to the surface of the object to be irradiated, this tendency is particularly noticeable. Therefore, it is desirable that the position detection pattern be drawn by lines of this step.

An example of a pattern that satisfies these conditions and has the required characteristics is shown in FIG. As is clear from the figure, the steps created by selectively etching the substrate form a plurality of concentric squares centered on point 0.
The distance between one step, which is one side of these squares, and the step next to it satisfies the condition d _o < d _o-1 (n=1, 2, 3,...), and d _o (n=0,1,2,
) is used as a known value to detect the center point. The procedure for detecting the coordinates of center point 0 by scanning this pattern in the X and Y directions of the orthogonal coordinate system will be described below. Here, in the X direction,
The Y direction is assumed to almost match the direction of the horizontal lines and vertical lines of the pattern. That is, when the electron beam is irradiated in either the X direction or the Y direction, the electron beam should not scan the horizontal and vertical axes of the pattern at the same time. This is achieved by using the wafer orientation crack.
This can be easily achieved by pre-aligning the direction.

It is now assumed that trial irradiation is started from point A in FIG. 3 in the positive direction of X. After detecting the first signal at the border between the region of width d ₂ and the region of width d ₁ , return to the starting point,
Scanning is performed in the negative direction of X, and a second signal is detected at the boundary with the area of width _d3 . After such a scan, it is found that the region including the starting point has a width of d ₂ and is separated from the center by d ₀ /2+d ₁ in the positive or negative direction of X. Then, for example, by continuing to scan in the negative direction of X, we know that the width of the next region is d ₃ , so that the It is known that it is the X coordinate of a point plus d ₀ /2+d ₁ . Next, the Y coordinate of point 0 is determined by scanning in the Y direction through center point 0.

The example given here is one of the simplest ways to find the center point, but no matter which point you start from, basically as long as you have two pieces of data about the distance in the X or Y direction. , X coordinate of center point 0 or Y
This pattern has the property of determining coordinates.

Furthermore, it is also an important issue to quickly determine whether or not this trial irradiation is performed within the alignment pattern area, but this is also possible according to the pattern of this embodiment. That is, no matter which point within the pattern you start scanning, the
If scanning of ±1/2d ₀ is performed in each direction, at least two lines should be detected. Therefore, by placing the execution of this scan at the beginning of the program and confirming that the scan start point is a point within the alignment pattern, the number of determinations in the subsequent program can be reduced. If the width d ₂ is detected during this starting point confirmation work as in the above example, it is sufficient to proceed directly to the work of determining the center position.

The above theory is based on the premise that the alignment pattern is formed with extremely high accuracy. However, in reality, it is inevitable that the accuracy of pattern formation is considerably inferior to the resolution of electron beam exposure. Therefore, the various scans described above must be performed while taking into account errors during pattern formation.

That is, the first scan of ±1/2d ₀ is ±(1/2d ₀ +α)
The difference between d _o and d _o+1 should be set sufficiently large to avoid errors in determining whether the detected width d is d _o or d _o+1. Consideration is required.

In addition, to determine the coordinates of the center point, instead of using the initially calculated values as they are, we actually measure the X or Y coordinates of each side of the square in the center of the pattern based on those values, and then calculate them again from that data. The error can be reduced by determining the coordinates of the center or by averaging the measured values at several locations.

In principle, it is possible to detect the center point using this method even if there is only one square, but
In order to shorten the required time, it is necessary to use a pattern having the characteristics as in the above embodiment.

In the above embodiment, the difference in the length of the sides of the squares arranged concentrically is made smaller toward the outside, but on the other hand, even if the difference in length of the sides of the squares is made larger toward the outside as shown in FIG. can be found. Furthermore, in the case where the pre-alignment of the wafer has high accuracy in the Y direction, for example, a configuration in which rectangles long in the X direction are arranged concentrically as shown in FIG. 5 can be used. Furthermore, it is also possible to use a method of calculating the coordinates of the center point by tracing one circumference of a concentric circle pattern as shown in FIG. 6.

When alignment marks are formed by etching or selective deposition of silicon dioxide, it is extremely difficult to realize a pattern in which lines intersect.
For example, it is almost impossible to draw an X mark with a step line formed by etching, and from this point of view as well, it can be said that the use of concentric figures is advantageous. In order to shorten the required time and simplify scanning distance calculations, trial irradiation for positioning may be performed not continuously but discontinuously at minute intervals, that is, in the form of dotted lines. The method of the invention can be implemented as is. This is because the step part of the alignment pattern formed on the silicon wafer is actually a slope, the electron beam has a thickness, and the spacing between spots in discontinuous scanning does not significantly exceed the resolution. This is due to a number of reasons.

As described above, according to the method of the present invention, even with a relatively rough pre-alignment, a desired figure can be drawn at an accurate position using an electron beam.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing a conventional arrangement of alignment marks, FIG. 2 is a diagram showing an example of alignment pattern arrangement in the present invention, and FIGS. 3 to 6 are alignment patterns used in the present invention. 1 and 13 are target IC patterns, 2 is an alignment mark, 11 is a semiconductor wafer,
Reference numeral 12 indicates an area where an alignment pattern is provided.

Claims (1)

[Scope of Claims] 1. In an electron beam exposure positioning method performed prior to irradiating an irradiation area of an irradiation object with an electron beam and performing exposure, a predetermined position on the irradiation object outside the irradiation area is provided. a plurality of rectangular similar figures spaced apart by a distance of a step of scanningly irradiating electrons to the first alignment mark and changing the position of the irradiated object based on the information obtained thereby regarding the position of the irradiated object in the X direction, Y direction, and circumferential direction; a second alignment mark that is smaller than the first alignment mark and is provided in the vicinity of a position where the electron beam irradiation area is to depict a target figure in the irradiation area by electron beam irradiation; irradiating the second alignment mark with an electron beam in a scanning manner, and setting a reference point for drawing the figure based on information about the position of the irradiated object obtained thereby; A positioning method for electron beam exposure, characterized by comprising a step of determining the position.