JPH0436360B2 - - Google Patents

Description

[Detailed Description of the Invention] This invention relates to synchrotron radiation (hereinafter simply referred to as synchrotron radiation).
For industrial use of charged particle orbital synchrotron radiation, such as in exposure equipment (referred to as SR), X-ray projection is devised so that the X-ray components contained in the synchrotron radiation can be obtained with a uniform intensity distribution over a large area. It is related to the device. The content of the present invention will be explained below by taking a pattern exposure apparatus using X-rays included in SR as an example.

The intensity distribution of X-rays on the sample surface of an SR exposure device is, for example, synchrotron radiation obtained from an electron storage ring with an electron energy of 600 MeV and an orbital curvature radius of 2 m.
When used for exposure at a distance of 10 m, the X-ray component with a wavelength of 10 Å is concentrated in a narrow range with a half-width of 5 mm in the vertical direction, as shown in Figure 1. The strength is uniform in the horizontal direction. There are two methods for uniformly printing a pattern on the entire surface of a silicon wafer, for example 10 cm in diameter, using X-rays having such an intensity distribution.

The first method, as shown in FIG. 2, is a method in which a mask 1 and a wafer 2 are integrally moved continuously in a direction perpendicular to the long side of a rectangle of a cross section of an X-ray beam 3 (vertical direction). However, this method requires a mask 1 with a diameter of 10 cm, which is the same size as the wafer 2, and with an X-ray mask that retains the pattern with a thin film of 2 to 3 μm thick, the entire area is patterned with submicron precision. is difficult to form.

In the second method, as shown in FIG. 3, the unit exposure field 4 that is exposed at one time on the wafer 2 is limited to an area of about one LSI chip (for example, 1 cm square).
In the mask 1, a pattern is created only in the unit transfer area 4', which is the window corresponding to the unit exposure field 4, and the relative positions of the X-ray beam 3 and the mask 1 are fixed, and only the wafer 2 is exposed to these areas. This is a step-and-repeat method in which exposure is repeated by moving one field relative to the field. According to this method, the area of the holding film of the mask 1 is reduced because only the unit transfer area 4' is required, and the drawbacks of the first method are solved, but the X-ray intensity distribution within one field is made uniform. must be done.

Conventionally proposed methods for making the X-ray intensity distribution uniform in the step-and-repeat method are as follows: (1) During exposure of one field, the wafer 2 and the mask 1 are integrated, and the X-ray beam 3 is The intensity distribution is made uniform by continuously moving it within one field relative to the field.

(2) As shown in Fig. 4, the SR beam is applied to the reflecting mirror 5, and the reflecting mirror 5 is oscillated to scan the 1-unit exposure field 4 with the reflected beam, thereby calculating the temporal integrated intensity. to make them uniform.

(3) As shown in FIG. 5, the SR beam is diverged by a convex mirror 6 to make the intensity distribution gentler and closer to uniformity.

This is a method such as

However, in method (1), the movement mechanism for the wafer 2 includes step movement of the wafer 2 alone in units of one exposure field for step-and-repeat, as well as a movement mechanism that is integrated with the mask 1 within the unit exposure field 4. It is necessary to provide a dual motion function of continuously moving the reflector 5, which has the disadvantage of complicating the mechanism.
There is a disadvantage that a separate mechanical movement mechanism must be introduced for the oscillating movement of the oscilloscope. Additionally, both methods (1) and (2) have the disadvantage that the precision and stability of mechanical motion govern the uniformity and reproducibility of the beam intensity distribution. Method (3) does not have the same mechanically moving parts as methods (1) and (2), so it avoids the same kind of drawbacks, but because the beam is expanded, the unit exposure field is 4.
The disadvantage is that the beam power density per unit area within the beam is lower than the original power density at the center of the beam, and the exposure time for one field becomes longer.

This invention was developed in view of the above points, and the beams on both sides, which are unnecessary in one field exposure, are guided away from their original direction of travel by fixed reflectors, and the beams are guided to the center. By superimposing the rectilinear beam of the beam and these reflected beams on the sample surface, the beam intensity distribution within the field is made to be uniform at a level equivalent to or higher than the high power density at the center of the beam. It is. The present invention will be explained below with reference to the drawings.

FIG. 6 is a conceptual diagram of an embodiment of the present invention.
Figure 6 shows the SR X-ray beam viewed diagonally from above.The cross section of the beam on the sample surface is centered at O (at the same height as the electron orbital plane), has a width of 8 cm in the horizontal direction, and a width of 8 cm in the vertical direction. They are rectangular parallelepipeds L ₁ , L ₂ , L ₃ , and L ₄ with a height of 1 cm. A case will be explained in which a central 1 cm square portion ABCD of this beam is used as a unit exposure field 4 in step-and-repeat exposure.

As shown in Figure 1, for example, the X-ray intensity decreases as you move from the center O to R, and at the halfway point
I'm getting used to it. If you go straight there, it will be a square
The beam β ₁ that was supposed to be incident on the upper half EFPQ of square EFGH adjacent to ABCD is reflected by the reflector to change its course, and is reflected in the lower half of square ABCD.
Inject it into the CDNM part. The beam path, the arrangement of the reflecting mirrors, etc. will be described in detail later. square
The intensity distribution of the original rectilinear beam β ₀ along the central axis SOR of ABCD is as shown in FIG. 1, and is represented by curves a and a' in FIG. 7. to this
As a result of superimposing the intensity distribution of the reflected beam β ₁ incident on the CDNM (curve b in FIG. 7), the intensity distribution between the SOs becomes as shown in curve c in FIG. 7, which is a substantially uniform distribution. The range of variation is ±4% and may be considered a uniform distribution for exposure purposes. Similarly, for the upper half ABMN of square ABCD, if the beam β ₂ that should be incident on the lower half GHQP of square EFGH is reflected and made incident on the lower half GHQP of square EFGH, by symmetrical operation to the above, then by superposition, square ABCD upper half of
Even with ABMN, a substantially uniform intensity distribution represented by curve c' in FIG. 7 can be obtained.

As mentioned above, since the beam intensity is always uniform in the horizontal direction, this means that an X-ray distribution that can be considered sufficiently uniform for the purpose of exposure has been achieved within the area of square ABCD. The correction for the mirror reflectance not being 100% will be described later.

Next, the reflecting mirror used in this invention will be explained. X-rays with a wavelength of about 10 A can be reflected only by total internal reflection due to oblique incidence below the critical angle. An example of a reflector structure is one in which gold is deposited to a thickness of 1000A on the surface of a completely flat quartz plate. This reflective mirror surface has a wavelength of 10A.
If a line is incident at an oblique angle such that the angle of incidence (oblique angle) measured from the mirror surface is 2°, the incident intensity will be
80% of X-rays are reflected.

An example of the arrangement of the reflecting mirrors is shown in the plan view of FIG. 8a and the side view of FIG. 8b. The actual glancing angle is 2 degrees, but it is exaggerated to 10 degrees in the figure to make it easier to understand. Therefore, the dimensional relationships are not necessarily accurate. The following description will be made using the same coordinate axes as those shown in FIG.

Part β ₁₁ of the beam is reflected by mirrors M ₁₁ and M' ₁₁ and moved down by 5 mm in the -z direction, and part β ₁₂ is reflected by mirrors M ₁₂ and M' ₁₂ and also moved down by 5 mm in the -z direction. Similarly, β ₂₁ is raised by 5 mm in the z direction due to reflection by mirrors M ₂₁ and M' ₂₁ , and β ₂₂ is raised by reflection by mirrors M ₂₂ and M' ₂₂ . Therefore, the upper half of the beam β ₁ (β ₁₁
+β ₁₂ ) and the lower half β ₂ (β ₂₁ +β ₂₂ ) are interchanged to form a beam, which is reflected by mirror M ₃ in the −y direction and is incident on the area of square ABCD on the second surface of the wafer. Strictly speaking, the length of the side in the y direction AB and
Although the CD becomes longer by 0.06%, it is a negligible amount for exposure purposes.

This beam β is reflected three times by the mirror, so
Strength is reduced to 50%. Therefore, the straight beam β ₀
The beam β′ (see FIG. 6) adjacent to the beam β on the opposite side with respect to the mirrors M ₁₁ , M′ ₁₁ , . . .
M ₂₂ , M′ ₂₂ are reflected by mirrors arranged symmetrically with respect to the zx plane, and the upper half β′ ₁ is also
If the lower half β′ ₂ is incident on the CDMN and the area ABMN, the curve C in _FIG .
A nearly uniform distribution, denoted C′, is achieved.

Beams β and β' are incident on the two wafer surfaces at an angle (incidence angle) of 4° from the perpendicular, and the penumbra blurring of the transferred pattern due to this is due to the mask
When the wafer spacing is 5μm, it is 0.35μm, and 1 to
Accuracy is sufficient for transferring submicron patterns.

In addition, interference fringes are generated by the superposition of the straight beam β ₀ and the reflected beams β and β′, but when the angle of incidence on the two wafer surfaces is 4°, the interference fringe interval caused by X-rays with a wavelength of 10 Å is 140Å (0.014μm),
Fine enough to not affect pattern transfer.

In the above treatment, SR has been treated as parallel light, but wafer 2 is
Assuming that the SR is placed at a distance of 10 m, the position where the SR enters the first reflecting mirror is 9 m 45 cm from the light emitting point, and the beam spread at this position is a maximum of 0.03°, so the error due to this will reduce the accuracy described above. does not affect the estimate.

Note that the arrangement of the reflecting mirrors for realizing the superposition of the beams whose principle is shown in FIG. 6 is not limited to the example described here. It is also clear that by appropriately selecting the area of a portion of the overlapping beams, the beam intensity can be made uniform over an area larger than 1 cm square.

As explained in detail above, the present invention is capable of directing X-ray beams other than those irradiated onto the required area of the X-ray beam of synchrotron radiation, typified by a pattern transfer device, by adjusting the angle between the beam center axis and the mirror surface. By providing a reflector that irradiates X-rays with a uniform intensity distribution to the required area by reflecting the X-rays under oblique incidence conditions below the critical angle and superimposing them on the required area, the required area that has not been used in the past can be reduced. It is possible to effectively utilize X-ray beams other than the above, and to realize a uniform spatial distribution of soft X-ray components within a certain area, which has important industrial significance.

[Brief explanation of drawings]

Figure 1 shows the distribution of the 10A wavelength component included in the SR from an electron storage ring with an electron energy of 600 MeV and an orbital curvature radius of 2 m in the direction perpendicular to the electron orbital plane, measured at a distance of 10 m from the emission point. The diagram is
Figure 3 shows the principle of the parallel movement method, which exposes the entire wafer by SR. Figure 3 shows the principle of the step-and-repeat method, and Figure 4 shows the principle of the step-and-repeat method. Figure 5 is a principle diagram of a method in which the SR beam is scanned using a convex mirror.
The figure shows the principle of the method of obtaining a uniform intensity distribution by reflection and superimposition of SR beams in this invention, and Figure 7 shows the almost uniform intensity distribution of the SR beam achieved by this method. 8 are diagrams showing an example of the arrangement of reflecting mirrors to realize this. In the figure, 1 is the mask, 2 is the wafer, 3 is the SR beam, 4 is the unit exposure field on the wafer surface in the step-and-repeat method, 4' is the corresponding unit transfer area on the mask, and 5 is the neck. A swinging reflecting mirror, 6 is a convex mirror, β(β ₁ , β ₂ ,
β ₁₁ , β ₁₂ , ..., β ₂₂ ), β ₀ , β' are part of the SR beam, M ₁₁ , M ₁₂ , ..., M ₂₂ , M' ₁₁ , M' ₁₂ , ...,
M′ ₂₂ and M ₃ represent mirrors in this invention.

Claims (1)

[Claims] 1 In an X-ray projection device that irradiates a required area with a portion of the projected X-ray beam emitted as orbital synchrotron radiation of charged particles, , a reflecting mirror is provided to irradiate X-rays with a uniform intensity distribution to the required area by reflecting the X-rays under oblique incidence conditions such that the angle formed by the beam center axis and the mirror surface is equal to or less than the critical angle, and superimposing the X-rays on the required area. An X-ray projection device characterized by: