JPS592372B2 - Electron beam exposure method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an electron beam exposure method particularly suitable for repeatedly exposing the same pattern at high speed.

There is a method called a two-stage DAC method that performs electron beam exposure at high speed and with high precision.

This consists of two DACs: a positioning DAC (digital-to-analog converter) that determines the starting point of a subfield within the deflection field, and a scanning DAC that performs scanning within the subfield.
The deflection of the electron beam is controlled by the sum of the outputs of these DACs. The relationship between the deflection field and the subfields is, for example, as shown in FIG. 1a.
That is, the subfield SF' is obtained by dividing the deflection field DF into a plurality of vertical and horizontal lines at regular intervals, and scanning by the electron beam is performed in subfield SF' units. P
is a pattern or a group of patterns drawn by an electron beam. In this method, the pattern P tends to span two or more subfields SF', and in this case, it is necessary to change the subfields. Subfields are changed by inputting the new starting point coordinate data to the positioning DAC, and while the scanning DAC is fast because the number of bits of the input digital data is, for example, 10 bits, the positioning DAC is Since the number is, for example, 14 bits and the notching time is long, there is a drawback that the exposure time becomes long if the subfields are changed frequently. For the fixed subfield method shown in Figure 1a, the subfield S is as shown in Figure 1b.
A so-called floating subfield method, in which the coordinates of the starting point E of F are arbitrarily selected, can be considered. In this method, pattern P can be rationally incorporated into one subfield SF,
Since the need for two subfields to draw one pattern can be largely avoided, the exposure time can be shortened. There are a raster scan method and a vector scan method for drawing a pattern.The former has a simple electron beam deflection control system but requires a long exposure time, and the latter has advantages and disadvantages opposite to the former.

FIG. 2 shows an example of drawing a chevron pattern in a magnetic bubble memory using the vector scan method. In this case, the chevron pattern P is approximately regarded as a collection of many rectangular areas are sequentially scanned with an electron beam to complete the exposure of one sieve pattern P. This method requires a large number of starting point coordinates and length data to specify each rectangular area A, so an increase in transfer time due to an increase in the amount of data transfer and an increase in the number of settlings of the positioning DAC will increase the exposure time. let In order to improve this point, the present inventors previously proposed an electron beam exposure method called a pattern package method.

As shown in Fig. 3, this method considers a simple-shaped (rectangular) area B that circumscribes an arbitrary-shaped pattern P to be drawn,
This area B is scanned with an electron beam along a reciprocating scanning line C, and the electron beam is turned on and off (on and off within pattern P) according to electron beam on-off information (referred to as package data) stored in memory in advance. In the area D outside the pattern P, it is turned off) and drawn. In this way, the data required to draw pattern P is only the starting point information and size information of simple shape area B, so the amount of data transferred and the required exposure time are significantly reduced compared to the method shown in Figure 2. be done. This pattern package method is particularly effective when the same pattern is repeatedly exposed to different positions of the deflection field. If this pattern package method is combined with the floating subfield method shown in FIG. 1b, it is expected that the exposure time will be further shortened. However, problems remain with simple combinations. That is, as shown in FIG.
, and length Ay, and one simple rectangular area B is made to correspond to one subfield SF, many simple rectangular areas Bl, B2, B3 are formed in the deflection field DF as shown in FIG. 4b. . . . exists, it is necessary to change the input of the positioning DAC for each region B and thus for each pattern P to output the starting point coordinates for each subfield SF1, SF2, . . . When there are many patterns P, the desired exposure time becomes long. FIG. 5 is a schematic block diagram showing the main parts of a beam exposure apparatus employing the exposure method shown in FIG. 4, with electron beam deflection means etc. being omitted.

In the figure, 1 is a central processing unit (CPU), 2 is a disk memory, and 3 is an interface. The features of the exposure device are pattern package memory (PPM) 4,
It is located after the starting point register 5 and high-speed counter circuit 6. The memory PPM4 stores electron beam on/off information (package data) PD within a simple rectangular area B. This data PD is read out in synchronization with the raster scan operation of the deflection system, and is sent to the blanker B through the beam blanking driver 7.
LK is driven (electron beam is turned on only in the pattern P portion). The start point register 5 receives digital information (Xn, Yn) indicating the start point E of the subfield SF, temporarily stores it, and stores it at low speed and high resolution (for example, 14 bits).
The positioning DAC 8 outputs the starting point information F converted into an analog value. The high-speed counter circuit 6 receives width and length information Ax, ay of area B, and in the x direction, O
The up-down count is continued repeatedly between and Ax. The area between O and Ax (or vice versa) corresponds to the scanning line C in FIG. 3, and the number of scanning lines is defined by the y-direction information Ay. High speed low resolution (e.g. 10 bit) scanning DAC9
Since G converts the output of the counter circuit 6 into an analog value, the converted output G becomes a repeating triangular wave as shown in FIG. The final output applied to the electron beam deflection amplifier DEF is a superposition of the outputs F and G of DACs 8 and 9, and by changing the starting point information (Xn, Yn), the DC level of the output F of the positioning DAC 8 changes. death,
Accordingly, as shown in Fig. 4b, the scanning starting points El, E2, E
3... is shifted. FIG. 6 is a waveform diagram assuming that the rectangular areas Bl, B2, B3, etc. in FIG. A predetermined settling time ST is required for switching the output F. This time ST
is one of the factors that lengthens the exposure time of the entire deflection field DF, but the time ST itself cannot be shortened unless the resolution is reduced. Therefore, in order to shorten the exposure time of the entire deflection field DF, it is effective to reduce the number of times the time ST occurs. For this purpose, it is conceivable to include a plurality of simple shape regions B in one subfield SF, as shown in FIG. 7, for example.

The size of the subfield SF is determined by the scan DACODA convertible limit, but in this example, if the pattern P in area B is a chevron pattern of a magnetic bubble memory, one S
It is possible to include a large number of B and F within F. Once the on/off information of subfields having a plurality of patterns P as shown in FIG. 7 is stored in the pattern package memory, the start point E of the subfield
By simply determining , it is possible to expose a plurality of patterns P, seven in this example, without changing the starting point and therefore without having to spend a long settling time ST. However, in this case, in addition to the non-exposed portion D shown in FIG. 3, it is necessary to scan the non-exposed portion 1 between the simple shape areas B, which increases the time required for exposure. The present invention aims to further improve this point and further shorten the exposure time.

In the present invention, the unnecessary scanning (raster scan) of the non-exposed portion 1 shown in FIG. 7 is stopped, and vector scan is introduced as shown in FIG. B3... is made to exist discretely. And each area Bl, B2,...
...'s origins Jl, J2... are high-speed scanning DACs
It is set by introducing the starting point shift amount to . FIG. 9 is a diagram illustrating the exposure method of the present invention.

As shown in this figure, in the present invention, the deflection area DF (this indicates the deflection limit of the electron beam, and the positioning DACOD
Set subfields SF1, SF2, etc. with an arbitrary point as the starting point within (the A conversion limit coincides with the tenon), and create multiple simple shape regions within this subfield with an arbitrary point as the origin. Set B. Subfield SF,
, SF2,... scanning starting point El, E2,...
... is determined by the low-speed positioning DAC, and the start point of the simple shape region in the subfield is determined by the shift circuit, but the origin position of regions Bl, B2, ... in the subfield SF may be arbitrary. , one of them, for example, J1 in FIG. 8, does not necessarily have to coincide with the scanning start points El, E2, . . . of the subfields SF1, SF2, . In the electron beam exposure method of the present invention, the low-speed positioning DAC is used only for shifting the subfield SF within the deflection field DF, so the number of occurrences of the settling time ST is minimized. Moreover, within one subfield SF, raster scanning is performed only within each simple figure area B, and high-speed DA is used between each area B.
Since it is shifted quickly in the C system (the settling time of a high-speed DAC can be ignored compared to that of a low-speed DAC),
The exposure time of the entire deflection field DF is significantly reduced. FIG. 10 shows an embodiment of the present invention, and the same parts as in FIG. 5 are given the same reference numerals.

The main difference between this example and FIG. 5 is that a simple shape area starting point memory 10 and a high speed starting point shift circuit 11 are added.
The memory 10 stores each simple shape area Bl, B2,
The coordinate information (Xn, yn) of the origins Jl, J2, . . . of . high speed counter 6
The width and length information given to Ax, ay are the same, but since the initial value 1NT is given from the shift circuit 11 based on (Xn, yn), the up-down count is different from that of (in the y direction, between Yn-Yn+Ay). As a result, Xn,
This is equivalent to applying a bias corresponding to yn, and the output G of the high-speed scanning DAC 9 changes as shown in FIG.
Note that since the information (Xn, yn) given to the starting point register 5 is the starting point coordinates of the subfields SF1, SF2, . . . , the output F of the positioning DAC 8 changes in subfield units. Also, the output PDG of memory PPM4
is the on/off information of the simple shape area. By combining such DAC outputs F and G, the electron beam deflected by the deflection amplifier DEF can expose a desired pattern at high speed within the deflection field DF through two types of shifts shown in FIGS. 8 and 9. . FIG. 11 is a diagram illustrating a deflection operation procedure using the circuit of FIG. 10. FIG. 12 shows another embodiment of the invention.

In this example, the starting point register 5 in FIG. 10 is replaced with a low-speed starting point shift circuit 12, and the simple rectangular starting point memory 10 is replaced with a high-speed matrix data memory 13. In FIG. 8, all simple figure areas Bl, B2,...
. . . are set in a matrix shape, and in FIG. 9, all subfields SF1, SF2, . . .
are arranged in a matrix, the start point information (XO, yO) of the first subfield SFl and the interval and order data of the subfields (DX, dY, Nx, Ny) are supplied to the shift circuit 12. The output is arranged in a matrix and becomes the starting point information of the subfield. In addition, the origin information (XO, yO) of the first simple rectangular area B1 and the interval and repetition number data (Dx, dy, nx, ny
) is given to the memory 13, its output is arranged in a matrix and becomes the starting point information of area B. It is clear that if either one of the arrangements is not in a matrix state, the same arrangement as shown in FIG. 10 may be applied. As described above, according to the present invention, there is an advantage that electron beam exposure using a two-stage DAC method can be made as fast as possible.

[Brief explanation of the drawing]

1A and 1B are explanatory diagrams of the electron beam exposure procedure using the two-stage DAC method, FIG. 2 is an explanatory diagram of the exposure procedure using the divided rectangular method, FIG. 3 is an explanatory diagram of the pattern package method, and FIG. b is an explanatory diagram of electron beam exposure that combines the pattern package method and floating subfield method, and FIG.
Figure 6 is a waveform diagram for explaining its operation; Figure 7 is an explanatory diagram for exposing multiple patterns in one subfield; Figures 8 and Figure 9 is an explanatory diagram showing the basics of the exposure method of the present invention.
Fig. 11 is a main part block diagram showing one embodiment of the present invention.
The figure is an operation waveform diagram, and FIG. 12 is a main part block diagram showing another embodiment of the present invention. In the figure, P is a pattern to be drawn, B, Bl, B2...
... is a simple shape region, SF, SFl, SF2...
... is a floating subfield, DF is a deflection field, 4
is pattern package memory, 8 is low-speed positioning DAC
, 9 is a high-speed scanning DAC, and 11 is a high-speed system starting point shift circuit.

Claims (1)

[Scope of Claims] 1. In an electron beam exposure method in which exposure is performed by deflecting an electron beam by the sum of the output of a low-speed, high-precision positioning DA converter and the output of a high-speed, low-precision scanning DA converter, the scanning DA converter A plurality of desired patterns to be drawn within a subfield determined by the scanning limit of the device are stored in a pattern package memory, and multiple desired patterns are drawn at different positions within one subfield by introducing starting point data into the scanning DA converter. The same desired pattern is drawn, and the positioning DA
An electron beam exposure method characterized in that by inputting starting point data to a converter, the subfield is shifted to an arbitrary position within an electron beam deflection area and the writing is continued. 2. The desired pattern consists of on/off information of the electron beam, and its outline is made into a simple shape to define the raster scan area, and by introducing the starting point data of each raster scan area into the scanning DA converter, the same pattern can be created within the subfield. 2. The electron beam exposure method according to claim 1, wherein a plurality of desired patterns are drawn by raster scanning.