JPH0472376B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a fine pattern forming method using a charged particle beam, that is, a charged particle beam lithography method.

A conventional pattern forming method has been based on a mask original pattern transfer technique using an optical exposure mask.

However, as pattern dimensions become finer to the extent that they approach the wavelength of the light used for exposure, forming patterns using this method reaches its theoretical limits.

In recent years, charged particle beam irradiation technology has come to be used as a technology for forming finer patterns than conventional optical exposure technology. Charged particle beam irradiation technology has also been improved, from a method in which a charged particle beam with a round cross section narrowed to about 0.5 microns is used to fill in a pattern in one stroke, to a method in which a cross section of several microns to several tens of microns formed into a rectangular shape is used. Methods using charged particle beams with specific dimensions have come into widespread use, and the time required to fill in patterns has been significantly shortened.

However, if you try to form a pattern with a charged particle beam with a large current amount because it has a large cross-sectional dimension of several tens of microns, there will be a problem when using a charged particle beam with a small cross-sectional dimension of about 0.5 microns. Due to the energy of the irradiated particle beam, it began to produce unfavorable results in pattern formation.

This is because when charged particles are incident on a substrate and a charged particle beam layer formed on the substrate, the temperature rise in the charged particle beam layer is caused by the energy of the incident charged particles in the charged particle beam layer and the substrate. This is determined by the balance between irradiation energy and thermal diffusion, which converts into heat and diffuses heat in the particle beam sensitive layer and substrate.In the case of a charged particle beam incident with a large cross-section, the This is because the temperature rise at the center of the irradiated charged particle beam layer and the substrate increases in proportion to the cross-sectional size of the charged particle beam. Furthermore, this geometrical distribution of temperature rise has the effect of significantly smoothing out the irradiated charged particle pattern, which has a sharp rectangular cross-section. Usually, the charged particle beam layer is made of a polymeric material and thermally decomposes at several hundred degrees Celsius. A thermochemical reaction occurs even if it does not result in thermal decomposition. Normally, when patterning a charged particle beam layer with a charged particle beam, the voltage range is from several keV to several tens of keV.
An incident charged particle beam of keV is inelastically scattered in the charged particle beam layer and the substrate, and a few eV in the charged particle beam layer is scattered.
The area where the energy is distributed is considered to be the pattern formation area. The shape of the region where this energy of several eV is distributed almost faithfully reproduces the sharp rectangular cross-sectional shape of the irradiated charged particle beam, which is why charged particle beam lithography technology is used for forming fine patterns. . However, as mentioned above, the blurred pattern distribution of several hundred degrees Celsius
superimposed on the sharp distribution of the charged particle layer, resulting in only a blurred pattern and a significant deterioration in the resolution of the charged particle layer. In order to significantly shorten the pattern drawing time,
The advantage of increasing the cross-sectional size of a charged particle beam no longer satisfies the requirement of obtaining a highly accurate fine pattern, and has instead become an obstacle.

In describing the embodiments of the present invention below, in order to make the content of the explanation more clear, a case will be described in which an electron beam, which is widely used today, is used as a charged particle beam. This is for convenience of explanation and is not intended to limit the invention.

Examples of how the resist pattern shape and resist sensitivity of monodisperse polystyrene electron beam-sensitive resists differ depending on whether or not there is a rise in resist temperature due to electron beam irradiation are shown below. FIG. 1a is a cross-sectional photograph of a pattern of monodisperse polystyrene on a Si substrate after development.
The interval between the marker lines at the bottom of the photo is 0.5 μm. Figure 1b shows the Si substrate in Figure 1a.
I just changed it to a SiO ₂ substrate. All other conditions, such as electron beam irradiation conditions and development processing conditions, were kept the same. The electron beam irradiation conditions shown in FIGS. 1a and 1b are patterns for irradiation with a rectangular electron beam of 12.5 μm×12.5 μm.

The protruding film at the lower right of FIG. 1b was produced when the substrate and resist were cut to obtain a cross-sectional photograph, and is not relevant to this explanation. The portion relevant to this description is the depressed portion of the resist caused by electron beam irradiation and development processing, which is located at the center of FIGS. 1a and 1b.

For the SiO ₂ substrate, the thermal diffusivity of the substrate is two orders of magnitude lower than that of the Si substrate, so the accelerated energy of the irradiated electron beam cannot be rapidly thermally diffused, and the monodisperse polystyrene layer is heated, resulting in a decrease in sensitivity. changed to the high-sensitivity side, and the cross-sectional shape of the pattern became worse. Si substrate and
Figure 2 shows how much the sensitivity of monodisperse polystyrene to electron beams differs depending on the SiO ₂ substrate. Monodisperse polystyrene is a negative resist. That is, the larger the amount of electron beam irradiation, the higher the normalized residual film rate after development. The normalized residual film ratio is the film thickness after development divided by the film thickness before electron beam irradiation. In Figure 2, the curve 20 is
This is a sensitivity curve of monodisperse polystyrene provided on a SiO ₂ substrate, and curve 21 is a sensitivity curve of monodisperse polystyrene provided on a Si substrate. The sensitivity on the SiO ₂ substrate changes to the lower electron dose side by 15% compared to that on the Si substrate. The electron beam irradiation conditions and development conditions in FIG. 2 are the same as those under which FIG. 1 was obtained. The fact that the pattern shape and sensitivity differ depending on the substrate material as shown in Figures 1 and 2 is because the electron beam pattern has a large cross-sectional dimension of 10 μm or more, so it is necessary to draw the electron beam pattern under conditions of a large current. This is something that has become more obvious over time.
In the case of the conventional electron beam cross-sectional size of 1 μm or less, there was no problem as long as a normal electron beam pattern drawing device was used.

Now, in FIG. 1B, there are more stripes of resist between the resist patterns than in FIG. 1A, and it is impossible to etch the substrate using the resist patterns as a mask. As mentioned in the explanation of Figure 2, this means that if the irradiated electron beam dose with a residual film rate of 50% changes by 15%, it will become impossible to form a resist pattern as shown in Figure 1b. It shows. The design dimension of the resist pattern shown in FIG. 1, that is, the distance between the resist layers, is 1.5 μm. Even considering the case of pattern formation of 2 μm to 3 μm, if the irradiated electron beam dose changes by 20% with a 50% remaining film rate, the first
It is thought that pattern formation becomes impossible, similar to b in the figure.

Next, with the Si substrate described in Figures 1 and 2,
Let's consider how much temperature change occurred in the resist layer compared to the SiO ₂ substrate. The distribution when the irradiation electron beam is incident on the resist layer and the underlying substrate is TEEverhart
This has been elucidated by Mr. (TE Everhart
and PHHoff, J.Appl.Phys.42, 5837 (1971))
As a result, the heat generation distribution in the resist layer and the substrate can be determined, and by solving the thermal diffusion equation, the temperature increase distribution can be determined analytically or by computer simulation. To make the explanation more concrete, let us assume a case that can be approximated by a cylindrical coordinate system as shown in FIG. The heat generation distribution is limited to the range of depths Z ₁ and Z ₂ from the surface with the radius R indicated by the diagonal line in Figure 3, and the temperature rise at point A at the center of the heat generation distribution is calculated using the following integral formula (1). It can be expressed as

T (P, Z ₁ , Z ₂ , R, t ₀ , C, ρ, D) = 2P/Cρ∫ ^Z2 _Z1 ∫ ^R ₀ ∫ ^t0 ₀ 2πr (4πDt) ^-3/2 exp [-(r ² + z ² )/4Dt〕dt dr dz ……(1) However, T (degrees) is the rising temperature, P (W/cm ³ ) is the amount of heat generated per unit time and unit volume, and the heating area is the depth z ₁ from the substrate surface. From (cm)
Area of radius R (cm) up to z ₂ (cm).

t ₀ (sec) is the heat generation time.

D (cm ² /sec) is the thermal conductivity of the substrate, which is the specific heat C
(Thermal diffusivity divided by the product of Joule/(g・deg) and density ρ (g/cm ³ ).

z ₁ (cm) is stated in the explanation of P.

z ₂ (cm) is stated in the explanation of P.

R (cm) is stated in the explanation of P.

C (Joule/(g・deg)) is described in the explanation of D.

ρ (g/cm ³ ) is described in the explanation of D.

Let's simulate how much the temperature of the electron beam sensitive layer rises due to the irradiation electron beam according to this calculation formula. Usually, an electron beam-sensitive layer with a thickness of about 1 micron is provided on a substrate such as silicon or glass, and an acceleration energy of about 20 keV and 0.4 A /
A typical case is irradiation with an electron beam with a current density of about cm ² . (For example, JBX-6 variable rectangular electron beam lithography system manufactured by JEOL Ltd.) For convenience of calculation, a circularly shaped electron beam with a width of 14 μm in diameter was used.
Figure 4 shows an example of irradiation under the conditions of 20 keV and 0.4 A/cm ² . z ₁ and z ₂ of heat distribution in depth direction
were approximated as 1.5 μm and 2.5 μm, respectively. When the acceleration energy of the irradiated electron beam is 20 keV, most of the irradiated electron beam passes through the electron beam-sensitive layer made of polymer and becomes heat in the substrate. Indicates increased temperature on a certain substrate surface. Further, it is the temperature increase at the center of the shaping electron beam where the temperature increase is the highest. For comparison, cases 23 where the substrate is silicon with a thermal diffusivity of 8.7×10 ⁻¹ cm ² /sec and cases 22 where the substrate is glass with a thermal diffusivity of 8.1×10 ⁻³ cm ² /sec are shown. By continuous electron beam irradiation for 500 μsec, the surface of the glass substrate shows a temperature increase of about 300 degrees. If the optical transfer mask is the substrate,
A metallic chromium layer of approximately 800 Å thick is placed on top of the glass, but it is so thin that the metallic chromium layer only has an effect on reducing temperature rise by a few percent.

Therefore, when patterning a normal optical transfer mask with an electron beam, 22 is used for the glass substrate shown in Figure 4.
is compatible.

Since molded charged particle beam lithography is an excellent method for shortening drawing time, the purpose of this invention is to obtain appropriate charging by considering the relationship between the thermal diffusivity of the substrate material and the temperature rise of the charged particle beam layer. The object of the present invention is to provide an unprecedented charged particle beam lithography method by providing particle beam irradiation conditions and suppressing the temperature rise of a charged particle beam layer.

A feature of the charged particle beam lithography method according to the present invention is that, in order to prevent deterioration of the resolution of the charged particle beam layer due to a temperature rise in the charged particle beam layer caused by charged particle beam irradiation, The irradiation time during the particle beam irradiation conditions is limited, and the sensitivity curve of the charged particle beam layer (graph of the normalized residual film rate by development versus irradiated charged particle dose) is used as a reference for the limitation. The goal is to keep the amount of change in the irradiated charged particle dose, which corresponds to a 50% residual film rate, due to a temperature rise within 15%. The limiting method is characterized by the use of a method in which the charged particle beam flow is interrupted intermittently and the temperature of the charged particle beam layer, which has risen due to irradiation, is lowered on a time average basis.

Examples of the present invention are shown below. When a glass substrate is irradiated with an electron beam as shown in FIG. 4, the value of the substrate surface temperature rise increases as the irradiation time elapses. The optimal irradiation electron dose for polymethyl methacrylate (PMMA), which is a typical electron beam-sensitive material, is 200 μC/cm ² .
An irradiation current density of 0.4 A/cm ² requires an irradiation time of 500 μsec, and in that case, as shown in Figure 4, the temperature of the area irradiated with the electron beam rises to more than 300 degrees Celsius, causing PMMA to heat up. I ended up disassembling it,
It becomes impossible to form fine patterns in PMMA. For example, when the substrate surface temperature rises to 100 degrees, stop the electron beam irradiation, wait until the substrate surface temperature almost returns to the original, start electron beam irradiation again, and when the substrate surface temperature reaches 100 degrees again, Repeat the process of stopping irradiation.
When using PMMA, by repeating the above steps until the total electron beam irradiation time reaches 500μsec, it is possible to suppress the temperature rise of PMMA to 100 degrees or less, and also ensure the optimum irradiation current amount. can. Of course, if it is desired to reduce the temperature rise to 50 degrees, which is less than 100 degrees, this can be achieved by shortening the continuous electron beam irradiation time in the above-mentioned repeated operations. In order to explain the embodiment of the present invention in more detail, FIG. 5 shows the working contents. 5 in Figure 6
12 conceptually shows the rise curve of the substrate surface temperature in the flow from the 7th stage to the 12th stage when electron beam irradiation is performed according to the flowchart shown in the figure. This is a case where the electron beam irradiation time _t4 is divided into four times _t1 and the remaining time t, resulting in five intermittent electron beam irradiations. t ₁ was selected so that the maximum temperature rise on the substrate surface would be below the desired value of 24 within ₁ hour of each electron beam irradiation.
At t ₂ , when the electron beam irradiation is stopped for t ₂ hours, the substrate surface temperature almost returns to the original temperature, and even if electron beam irradiation is performed for the next t ₁ hour, the substrate surface temperature rise remains at the desired value of the maximum substrate surface temperature rise 24 The following have been selected. Although FIG. 5 shows a case where the determination at step 13 causes the process to return to step 5, it is also possible to return to step 1. This method is particularly effective when the material of the substrate to which the electron beam is irradiated differs depending on the location.
t ₁ hour determined in the second and third stages of Figure 5
t ₂ hours is determined, and a different t ₁ hour and
Selecting t ₂ hours is a characteristic method of the present invention. Normally, the number of types of substrate materials and electron beam-sensitive materials that are practically used in the semiconductor industry is not so large, so Figure 4 obtained from the integral formula (1) in the main text is calculated for each material, and the irradiation It is easy to tabulate t ₁ and t ₂ in FIG. 6 using the electron beam cross-sectional dimensions and irradiation current density as parameters.

In the description of the specific embodiments of the present invention, it has been simplified that the irradiated electron beam produces a heat distribution in a cylindrical substrate as shown in FIG. In reality, the distribution is blurred in the surface direction and R direction.

However, the effect thereof is on the order of several percent to change the results of the model calculations in the main text, and does not change the gist of the explanation of the embodiments of the present invention. In addition, when specifically creating the table of t ₁ and t ₂ in Fig. 6, the cross-sectional dimension of the irradiated electron beam is set to 1 μm, for example.
There is no practical problem even if it is divided into groups such as ~4 μm, 4 μm to 8 μm, and 8 μm to 12 μm. The reason for this is that it is possible to adopt a value with a margin in the desired value of the maximum temperature rise on the substrate surface.

[Brief explanation of the drawing]

Figures 1a and 1b are diagrams of the charged particle beam layer shown to visually explain the correlation between the heat of the charged particle beam drawing section and the generated pattern shape, which is the origin of the present invention. Scanning electron micrographs of the surface and cross section. FIG. 1a shows a case where the material is not affected by heat much, and FIG. 1b shows a case where it is subjected to a large amount of heat. Figure 2 shows the sensitivity curve of monodisperse polystyrene. FIG. 3 is a schematic diagram for explaining formula (1). Figure 4 shows the temperature increase on the substrate surface due to electron beam irradiation. FIG. 5 shows the procedure for implementing the present invention. FIG. 6 shows the temperature increase on the substrate surface in the irradiation area in the graphic drawing method of the present invention. In the figure, each symbol indicates the following. 1 to 14 are each step in the implementation procedure of the present invention. 20... Sensitivity curve of monodispersed polystyrene on a glass substrate. 21... Sensitivity curve of monodisperse polystyrene on silicon substrate. 22...Temperature rise when electron beam is irradiated onto the glass substrate. 23...Temperature rise when electron beam is irradiated onto silicon substrate.
24... Desired value of maximum temperature rise on the substrate surface. A...
The intersection of the extension line of the center of the heat distribution and the substrate surface. Z ₁
...Distance of heat distribution from the substrate surface. Z ₂ ... Distance of heat distribution from the substrate surface. R...radius of heat distribution. t ₁ ...Electron beam irradiation time. t ₂ ...Electron beam irradiation stop time. t ₃ ...Electron beam irradiation time. t...Electron beam irradiation time.

Claims (1)

[Claims] 1 In a charged particle beam lithography method in which a charged particle beam is intermittently irradiated onto a charged particle beam layer on a substrate to draw a figure, the temperature rise of the charged particle beam layer due to the irradiation of the charged particle beam The combination of the continuous irradiation time t ₁ , which is at or below the limit at which the desired shape cannot be obtained with high accuracy, and the time t ₂ , which is the time from irradiation interruption until the temperature returns to almost the original temperature, is used for substrate materials and charged particle beam materials. , the irradiation acceleration voltage, the current density of the charged particle beam, and the cross-sectional area of the charged particle beam are determined in advance as parameters, and based on these, the irradiation and interruption of the charged particle beam are repeated until the appropriate dose is reached, and the figure is created. A charged particle beam lithography method characterized by drawing.