JPS63316434A - X-ray exposure - Google Patents

Abstract

PURPOSE:To execute a transcription operation with high accuracy by a method wherein a multiple exposure operation is executed by changing a gap length between a mask and an object to be exposed so that the disorder of a transcribed pattern due to an influence by Fresnel diffraction and a deterioration in the dimensional accuracy can be suppressed. CONSTITUTION:When a gap length between a mask and an object to be exposed is changed, a strength distribution of the Fresnel diffraction light on the object to be exposed is changed; when a multiple exposure operation is executed with this strength distribution, a distribution of a quantity of exposed light is made uniform. In the beginning, an irradiation duration necessary for an exposure operation is designated as T; a first exposure operation is executed for a duration of T/2 with a reference gap length Gi as shown in a position 1; then, a mask is shifted to a position 1' of the gap length multiplied by alpha; a second exposure operation is executed for the duration of T/2. When a multiple exposure operation is executed by changing the gap length between the mask and the object to be exposed, the unevenness of the irradiation strength due to an influence by diffraction and the disorder of a pattern shape can be reduced sharply.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to X-ray pattern transfer technology, and is particularly used for transferring high-density integrated circuit patterns onto semiconductor substrates.

[Conventional technology and its problems]

Pattern transfer methods using X-rays have the advantage of superior resolution compared to transfer methods using visible light to ultraviolet light, and superior productivity compared to pattern formation methods using electron beams. . Therefore, studies are being conducted at the research level as a pattern forming method for high-density integrated circuits with a line width of 015 μm or less, and device prototypes are being produced.

Seven problems with the X-ray exposure method are that exposure requires a relatively long time; for example, when an electron beam excitation type X-ray source is used, an exposure time of one minute or more is required. Gas discharge type plasma source and laser excitation type plasma source have also been proposed as more powerful X-ray sources. A synchrotron radiation source is a means for generating X-rays stronger than these sources.

Synchrotron radiation is the light emitted when electrons are accelerated to near the speed of light and a magnetic field is applied to make them move in a circular motion.The spectrum of synchrotron radiation ranges from the far-field region to X-rays and even R-rays. It extends to the territory. Among these, soft X-rays with wavelengths from several X to 20'A can be used for pattern transfer. X-rays with wavelengths shorter than this have high energy;
The range of electrons generated in the resist becomes long, making it unsuitable for forming fine patterns. On the other hand, X-rays with longer wavelengths have inferior substance-transmitting ability and cannot efficiently pass through the so-called membrane film that supports the absorber in the X-ray mask.

In order to obtain the wavelength range required for transfer, synchrotron radiation is usually reflected by a mirror and beryllium (Beryllium) is used.
) etc., and then irradiate the X-ray mask and substrate. The mirror attenuates short-wavelength components harmful to transfer, and in some cases evens out the intensity of the X-rays spatially. The window plays the role of suppressing the heat generated by the X-ray mask by cutting out long wavelength components.

Now, the resolution of the transfer method using X-rays depends on (1) penumbra filling caused by the finite size of the radiation source, and (ii) the flight of electrons generated by X-rays in the resist. Cheng, (iiD
Determined by X-ray diffraction. Of these, the penumbra blur (1) can be ignored when synchrotron radiation is used because the source is sufficiently far away. The range of the electron in (1i) is
For example, if the wavelength of X-rays is 7A, it is 0.7 μm or less, and there is no problem if the pattern size to be transferred is up to about 0.3 prn. For pattern transfer with a line width of about OJμm (
Diffraction in iii) has the greatest effect on resolution. The influence of diffraction will be specifically explained below.

FIG. 72 shows the state of diffraction occurring at the edges of the semi-infinitely expanding X-ray shielding film. The vertical axis shows the X-ray intensity on the X-ray exposed object separated from the shielding film and the gap G. The irradiation intensity at a sufficiently far distance from the edge is set to 7. In baton transfer using X-rays, the X-ray shielding film is an X-ray absorber in an X-ray mask, and the object to be exposed to X-rays corresponds to a substrate such as a semiconductor.

This phenomenon is well known as Fresnel diffraction, where ρ is the distance on the shielding film surface or the distance on the exposed object surface, and λ is the X-ray wavelength.
Then, u-1':': '' is characterized by the dimensionless number U, which is set as 11. The horizontal axis in Figure 12 shows the distance from the edge of the shielding film as the value of U. As can be seen from the figure, the intensity of X-ray irradiation due to diffraction is maximum at u = ut =, /, 22, and u = u2 = /
, becomes minimum at J'7. In actual dimensions, if G = gjμm and λ = lJh, the distance l from the edge of the shielding film is J = /, -2
, 2, fl=0. /lttm.

When u2 =/, ♂7, ft=0.2! Corresponds to pm. For example, the literature (N.

Atoda et al., Journal of Bacchieum Science and Technology, Volume B/-v, p. 1267, 3rd year.
Journal or Vacuum8science
and Technology, B/-1i, 1
2t7 (/ri♂3)).

This effect of diffraction indicates that the intensity of the X-rays can decrease even though the X-rays are shining, and in pattern transfer, the requirements for controlling the exposure amount become stricter. This also leads to deterioration of dimensional accuracy and disorder of pattern shape.

As shown in Figure 1.2, from the edge u = u2 =
Since the strength is minimum at /, ♂7, in a slit-shaped baton with another shielding film at the position u = 3.711, which is twice the width, the pattern center part will be separated by u2 from both sides.
Since it is at a distance of = /, ♂7, the decrease in intensity is emphasized. The simulation results when u = 3.71A are shown in Figure 73 (-), and the wavelength distribution of X-ray energy absorbed by the resist used in this simulation is shown in Figure 73 (-).
Figure price). The wavelength distribution in Fig. 13(b) shows that the acceleration voltage is 21 GeV, the magnetic field strength is 9.63 Gauss, the radius of curvature of the electron orbit is f, and x#j emitted from the tracing of A m is obliquely incident. It is incident on the sxc mirror at an angle Z of 2 degrees, and the window material of Be with a thickness of 2iμm and the S with a thickness of 2μm are
The energy transmitted through the iN membrane and absorbed by the PMMA resist with a thickness of 0.7 μm is calculated.

With such a wavelength distribution, the change in intensity due to the influence of diffraction is smoother than in the case of a single wavelength shown in Figure 1, but it is still large in the center as shown in Figure 73 (a). Strength decreases. Although the central region is the area where the X-rays strike, the intensity is about 0.72 times that of the incident X-rays. Figure 13 (The slit width of aJ is the wavelength λ
Assuming that the peak wavelength is &3A and the gap G is ≠3 μm, this corresponds to 015 μm.

In fact, when pattern transfer is performed using synchrotron radiation light that produces a wavelength distribution as shown in FIG. 73(b), the transferred resist pattern has a pattern corresponding to the central four parts of the intensity distribution shown in FIG. 73(a). A fine structure of the resist pattern appears.

Furthermore, a square punch pattern is influenced by the surrounding μ sides. Such patterns exist as contact holes in integrated circuit patterns. Figure 73(b)
The simulation results using the X-ray spectrum shown in Fig. 1μ are shown in Fig. 1μ. The first diagram is a diagram viewed from above, and the curves indicate isointensity lines of the irradiation intensity distribution of the diffracted light. The distance between the iso-intensity lines is O, ko. When the intensity of incident X-rays is 1, the intensity is 2 or more at some μ locations, while the intensity is significantly reduced to about 0.3 at the center. The pattern shape is also greatly disturbed. As described above, Fresnel diffraction has a large effect on pattern transfer with a line width of approximately 0.3 μm.

In order to reduce the adverse effects of Fresnel diffraction, conventional efforts have been made to optimize the selection of the gap length G between the mask and the exposed surface and the wavelength λ used for exposure. As is clear from equation f1), if K, for example, G and λ are each set to //2, the pattern size ρ will be set to //2 for the same U value, that is, while being affected by diffraction to the same degree. be able to.

As for the minimum value of G, it is technically possible to set it at 20 μm or less, but /) the mask and the exposed surface require strict flatness, and 2) the relative relationship between the mask and the exposed object. There are disadvantages such as easy contact when moving parallel to G -110-! Opm is a value that can be achieved relatively easily and with high reliability.

The minimum value of λ is determined by how far the range of electrons excited by X-rays is allowed. The range of the electron is Gr
If the length is allowed up to 0.1 μm, the minimum value of λ is 7A.

From the above, if the wavelength of x#j is 7λ and the gap length between the mask and the exposed surface is 0μm, then u = j,
The pattern size corresponding to 7≠ is 0. X-ray transfer with dimensions of this order will cause blurring of the pattern shape and deterioration of dimensional accuracy due to Fresnel diffraction.

The purpose of the present invention is to suppress the disturbance of the transferred pattern shape and the deterioration of the dimensional accuracy caused by the influence of Fresnel diffraction, o.
The object of the present invention is to provide a method for accurately transferring a pattern of rμm or less.

[Means for solving problems]

In the present invention, by changing the gap length between the mask and the exposed object,
It is designed to perform multiple exposure.

[For production]

When the gap length between the mask and the exposed object is changed, the intensity distribution of Fresnel diffracted light on the surface of the exposed object changes. By performing multiple exposures with these intensity distributions, the exposure amount distribution is made uniform.

〔Example〕

Hereinafter, the present invention will be explained in detail based on examples.

FIG. 7 is a diagram for explaining an embodiment of the exposure method of the present invention, where / and /' are X-ray masks, 2 is an exposed object, 3 is a holder that supports the exposed object, and G is a synchronizer. This is the incident X-ray extracted from Tron synchrotron radiation. As mentioned above, synchrotron radiation has short wavelength and long wavelength components of 8iC.
The wavelength distribution shown in FIG. 73(b) is used, which is cut by a plane mirror such as, and a window material made of Be or the like.

First, let T be the irradiation time required for exposure, then / in Figure 1.
The first exposure is performed for T/2 hours with a reference gap length Gi shown at the position. Next, move the mask to the position /' in Figure 1 with the gap length α times T/, 2
The λth exposure in the time tube is performed.

In this exposure method, as can be seen from equation (1), u == u
After exposure at i, exposure is performed at u=u1xS/7'''' with the aim of canceling out the influence α of diffraction.

α; tt, gap G1 corresponding to u1 = 3.711
FIG. 2 shows the simulation results when a slit pattern with a diameter of 3 μm and a width of 0.3 μm was transferred. Second
Figure 1 (b) shows the intensity distribution obtained by the second exposure, and Figure 1 (b) shows the intensity distribution obtained by the first exposure and the intensity distribution obtained by the second exposure shown in Figure 13 (aJ). This is the intensity distribution when double exposure is obtained from the sum of the distributions.Second
The distribution shape in Figure (b) is similar to Figure 73 (
It can be seen that this is significantly improved compared to the case of 2L).

The exposure method of the present invention is effective as described below even when the influence of diffraction is greatest in the second exposure. U is most affected by diffraction during the second exposure.
= 3.7 II, but α, =/, J
Then, the first exposure becomes ui =lA73.

If the gap G1 is 3 μm, the pattern width is 0.1
．． This corresponds to 3 μm. The intensity distribution of the first exposure is shown in Figure 3(a), and the intensity distribution of the seventh exposure with this distribution and u =
3.74(K) is the intensity distribution after double exposure in the second exposure.

In the distribution shown in Figure 3 (-), the minimum value in the exposed area is approximately O0, while in the double exposed distribution shown in Figure 3 (b),
The minimum value has been improved to about O9. This shows that even where the influence of diffraction is greatest in the second exposure, the uniformity of the intensity distribution is improved compared to the conventional exposure method.

Next, in order to see the improvement effect on the planar pattern shape, the simulation results of the square punched putter 7 with a -side length of 005 μm will be described. Figure μ shows that the first exposure was performed with the distribution of u=j, 71/l (gear tube diameter 3 μm) shown in the first diagram, and the second exposure was performed with the distribution of α=/, 4
This is an isointensity diagram showing the irradiation intensity distribution showing the result of double exposure when the distribution is u-=Z (gap μm). The interval between the iso-intensity lines is 0.2, and the thick lines among them indicate the iso-intensity lines whose intensity is /, 00.

In addition, in FIG. 5, the − side length is 0. This is the case of a square punching pattern of t3μm, and the first exposure is u=lA73(
The second exposure was carried out at
This figure shows the results of double exposure performed at = 3.74' (gap H7 μm). It can be seen that the irradiation intensity distribution in the case of double exposure shown in FIGS. 1 and 5 has an improved square planar shape compared to the case of single exposure shown in FIG.

Although it has been described above that a uniform dose distribution can be obtained by the present invention in the case of α=/, J, the value of α is not limited to this value. Calculations based on the above wavelength distribution show that a significant improvement effect is obtained when α<2.4. Since it varies depending on the window material, etc., α
The effect can be expected over a wide range of values. Furthermore, although the explanation has been made using double exposure with different gap lengths, triple or quadruple exposure may be used, and furthermore, it is also possible to continuously change the gap length while applying X-rays.

Next, a method and a transfer device for making the X-ray exposure method described above more effective will be explained.

With the X-ray exposure method using synchrotron radiation, it is difficult to produce a large-area, high-precision X-ray mask, and 2) it is difficult to obtain wide X-rays with a uniform intensity distribution. For example, transfer the μQMiN angle at once, move the exposed object 110 degrees in a plane perpendicular to the incident X-ray, expose the next area again, and repeat this operation to create a pattern on the entire surface of the exposed object. A so-called step-and-repeat method is used for transcription.

A conventionally proposed method for controlling the gap length during step-and-repeat is shown in FIG. In Figure A (a), the set gap length value is always maintained and exposure/movement is repeated. In FIG. 6, E represents exposure and M represents movement of the exposed object. Furthermore, when the gap length is as small as 20 pm, for example, a method has been proposed in which the gap length is increased only during movement to avoid contact between the exposed object and the mask, as shown in FIG. A(b).

The present invention, which performs multiple exposure by changing the gap length, is highly efficient.
The method to be implemented is shown in FIG. 7 using the case of double exposure.

In the figure, K and M are the same as in Figure 6. Figure 7 (
A) is when there is no need to consider contact between the exposed object and the mask when moving the exposed object. The object moves with the same gap length that was used for exposure immediately before moving, and after the movement is completed,
This is a method in which the first exposure is performed with the same gap length.

For each exposure position, exposure can be performed by changing the gap value seven times. FIG. 7B shows a case where it is necessary to take a large gap length when moving the exposed object, and the gap length when moving is made to match the long gap length when performing double exposure. After the movement is completed, the first exposure is performed with a short gap length, and the second exposure and movement are performed with a long gap length. By matching the gap length during movement with the long gap length for double exposure, it is possible to change the gap value twice for each exposure position, as in Fig. g(b).

Although the case of double exposure has been described as an example, the present invention can be applied to multiple exposures of three or more times or to cases where the gap length is continuously varied.

FIG. 7(C) shows an example in which the gap length is changed continuously and there is no need to consider contact between the exposed object and the mask.
FIG. 7D shows an example in which the gap length during movement is made to match the longest gap length used in exposure.

Next, specific means for controlling the gap length will be explained.

To perform X-ray exposure, it is necessary to adjust the gap and position between the mask and the wafer to a high He.

There are several theoretical proposals, such as the use of X-rays and capacitance, but due to the ease of equipment implementation, we have proposed two methods: ■ interference light detection using a mask, a diffraction grating on the eight faces of the sea urchin, and a laser beam, and ■ the use of an objective lens. It can be roughly divided into two types: image detection of the mask surface used and image detection of the wafer surface.

As an example of implementing the method (2), a gap detection method using a single grid is described in detail in, for example, the literature (Journal of Japan Society for Precision Engineering, May issue, 2016, p. 7/7).

In this method, as shown in the figure, a grating with a pitch P on a mask surface, a wafer reflection surface, and a laser beam that is incident perpendicularly to the surface are used. The intensity I+1 of the +7th order diffracted light changes as shown in Figure 1. This curve is an envelope wave with a period of p2/λ caused by the transmitted diffracted light generated between the diffraction grating and the reflective surface, and the λ/2 wave generated by the interference of the diffracted light reflected from the top surface of the mask.
It consists of periodic interference waves and reaches its maximum at a gap length G that satisfies λG==nP2 (n is an integer). Therefore, gap setting becomes possible by detecting this point.

If a He-Ne laser (λ=0.t32♂μm) is used and the grating pitch P is 3μm, Gp=/442μm.
The maximum value of the interference wave is obtained at a period of m.

Therefore, when applied to the present invention, the seventh exposure is
= jG, = ll-2, gap length in tpm, 2
By setting the gap length of the second exposure to 02=jGp=7/μm, the second gap value is set to the first exposure /, J 7
Can be doubled.

To apply this gap adjustment method, it is necessary to set the gap within a certain range, in the above example, within ±7 μm, but this can be achieved with mechanical precision. Alignment detection is possible using a double diffraction grating, as in the above-mentioned document.

In addition, as a diffraction grating, literature (Journal of Vacuum Science and Technology, /vol.
eum 5science and technology
If the method uses a two-dimensional Fresnel lens, such as the one proposed in gy, Vol. Since there is an inversely proportional relationship, different gap positions can be set using laser beams of different wavelengths.

In the alignment method (2) using an objective lens, gap length and position detection can be performed by optically moving the focal point position.

In the above explanation, the incident X-rays are assumed to be parallel. However, X-rays using synchrotron radiation have a divergence angle that depends on the distance from the light emitting point to the surface to be exposed when no mirror is used or only a plane mirror is used. When there is a divergence angle, a change in gap length appears as a change in position in pattern transfer. FIG. 10 is a diagram for explaining this change in position. l and /' are the X-ray masks, C is the exposed object, and G is the incident xlfs. Assuming that the side length of the transfer area on the X-ray mask is Ja, the maximum amount of change in gap value during multiple exposure is δ, and the distance from the exposed surface to the light source is D,
The deviation Δ in the transfer pattern position due to the change δ in the gap value is Δ=−× δ ・・・・・・
(2) Given by D.

For example, D = / Q m , a = 2 cm
When s δ=30 μm, Δ=lQnm. In order to eliminate such positional deviations due to changes in the gap length, it is sufficient to make the X-rays parallel, and the closer they are to parallelism, the smaller the amount of positional deviation will be. Simply, it is sufficient to move the exposed object away from the light emitting source, but then the XIs intensity per unit area becomes small.

In order to obtain parallel X-rays or X-rays with a small divergence angle without reducing the intensity of the X-rays, it is effective to use a curved mirror. Figure 1 is an example of obtaining parallel X-rays using a rotating parabolic mirror. In the transfer method using synchrotron radiation, curved mirrors are well known as a method of obtaining high intensity X-rays, and in the X-ray transfer device implementing the present invention, the gap can be reduced because the divergence angle can be reduced. It has the advantage that the positional deviation is small even when the length is changed.

In the above embodiments, a synchrotron/synchrotron radiation source was used as the incident X-ray source, but if a gas discharge type or laser excitation type plasma X-ray source is used, It goes without saying that the method of the present invention is effective in cases where resolution degradation due to diffraction is greater than degradation due to the range of electrons generated by absorption of radiation in the resist.

〔Effect of the invention〕

As explained above, the present invention has the advantage that unevenness in irradiation intensity and disturbances in pattern shape due to the effects of diffraction can be significantly reduced by performing multiple exposures by changing the gap length between the mask and the exposed object. be.

[Brief explanation of the drawing]

Figure 1 is a diagram for explaining the present invention in detail, Figure 2 (a
) is u = , Z ri (gap length z&rμm,
Slit width O0! Fig. 1 (b) is a diagram showing the irradiation intensity distribution of Fig. 1 (a) (corresponding to μm) (u = 2.ri6
) and Figure 73(a) are diagrams showing the irradiation intensity distribution when double exposure is performed with the distribution of (u = 3.7 Fig. 3(b) is a diagram showing the irradiation intensity distribution with a width of 0.1.3 μm (corresponding to a width of 0.1.3 μm), and Fig. 3(a) (u = 4473
) and the irradiation intensity distribution when double exposure is performed with a distribution where u = 3.7≠. Side length O1
! Figure 3 is an iso-intensity diagram showing the irradiation intensity distribution of a square cutout pattern when the first exposure is double exposed at u = j, ri 6 (corresponding to a gap length of 17 μm). The first exposure is u=lA73(Gee?/Product length u3pm)
, - equivalent to side length 0.63 μm), the second exposure was
= 3.7μ (equivalent to gap length t and rμm) Iso-intensity diagram showing the irradiation intensity distribution of a square punch pattern when double exposed (corresponding to gap length t and rμm). An explanatory diagram of the gap control method using (a) a method in which the exposed object is moved while keeping the gap length constant, and (b) a method in which the exposed object is moved by increasing the gap length only during movement (E is an exposure diagram) 7 is a diagram showing an example of a gap control method when performing step-and-repeat exposure, and FIG. 2 is a diagram illustrating a gap length control method using a diffraction grating. Figures 1 and 2 show + in the gap length control method using a diffraction grating.
Diagram showing the relationship between the seventh-order diffraction light intensity and the gap length, No. 10
The figure is a diagram to explain the positional shift of the transferred pattern due to a change in the gap length. Figure 72 is a diagram showing the irradiation intensity distribution of diffracted light generated at the edge of the X-ray shielding film that spreads semi-infinitely, and Figure 13 is a diagram showing the irradiation intensity distribution of diffracted light generated at the edge of the X-ray shielding film that spreads semi-infinitely. (b) A diagram showing the wavelength distribution of energy absorbed in the resist used to calculate the irradiation intensity distribution. Figure 1U shows the case where the influence of diffraction is the greatest (u
3.711) is an isointensity diagram showing the irradiation intensity distribution of diffracted light in a square cutout pattern. l, /'...xIm mask, λ...exposed object, 3.
...Holder, Hiroshi... Incident X-rays.

Claims (1)

[Claims] An X-ray exposure method characterized by setting the gap length between an X-ray mask and an exposed object to two or more different values and performing multiple exposure, or by performing exposure while continuously changing the gap length.