JPH01111326A - Electron beam exposure device - Google Patents

Abstract

PURPOSE:To accelerate an electron beam exposure and to improve its efficiency by providing a cash memory for temporarily storing pattern data read asynchronously with an exposure unit from a data memory at a location between the data memory and a pattern generator. CONSTITUTION:Pattern data are read asynchronously from an exposure unit 14, i.e., from a data memory 11 at a timing different from that of an exposure clock, and temporarily stored in a cash memory 12. A pattern generator 13 reads one pattern data from the memory 12, and processes necessary data for the exposure. Accordingly, a time from the access to the read of the data can be shortened, and the generator 13 can read asynchronously one pattern data by utilizing a period between an exposure and next exposure (a period that a blanking signal is OFF). Thus, a sample lithography with an electron beam can be performed efficiently at a high speed.

Description

[Detailed Description of the Invention] [Summary] A data memory that stores a plurality of pattern data, a pattern generator that performs data processing for each pattern data and generates the data necessary for exposure, and a pattern generator that processes the data for each pattern data and generates the data necessary for exposure. Regarding electron beam exposure equipment equipped with an exposure section that performs exposure at high speed, in order to perform electron beam exposure at high speed and increase efficiency, there is an asynchronous connection between the data memory and the pattern generation section. A cache memory is provided to temporarily hold the pattern data read out.

[Industrial application field]

The present invention provides a data memory for storing a plurality of pattern data and data processing for each ζ1 pattern data.

The present invention relates to an electron beam exposure apparatus that includes a pattern generation section that performs processing and generates data necessary for exposure, and an exposure section that exposes a sample with an electron beam according to this data.

More specifically, the present invention relates to data control of the above-mentioned electron beam exposure apparatus.

An electron beam exposure apparatus forms a pattern using an electron beam. A major feature of pattern formation using an electron beam is that it can draw fine patterns of less than a micron. Electron beam exposure apparatuses are broadly classified into Gaussian beam type and shaped beam type. These are also stage continuous movement methods that draw images while continuously moving the sample.

and step-and-beat methods in which the sample is moved by a predetermined amount for each drawing area. In any of these methods, shortening the exposure time is a major issue.

[Conventional technology]

FIG. 10 is a block diagram of an in-shell electron beam exposure apparatus. A central processing unit (hereinafter referred to as CPU) 1 controls the entire electron beam exposure apparatus. Disc 2 is
It stores various data and control information that match pattern data, which is graphic information to be drawn. Interface 3 is CPLJ
1 and the various parts described later. The data memory 4 is a large capacity memory that temporarily stores pattern data read from the disk 2. The pattern generation section 5 is
Data for one pattern is read out from the data memory 4 and divided into shots (exposure regions of electron beams that can be output at a time while maintaining a desired current density) to create data necessary for exposure. The pattern correction unit 6 corrects the amount of beam deflection according to the state (expansion/contraction, rotation, parallel shift, and height variation) of the drawing area of the sample.

An amplifier (AMP) 7a amplifies the blanking signal.

7b, 7-C, and 7d are a D/A converter (DAC) that converts the digital signal from the pattern correction section into an analog signal, and an amplifier (AMP) that amplifies the analog signal.
). The exposure unit 8 irradiates the sample 9 with an electron beam emitted from the electron gun 8a to draw a pattern.

FIG. 11 is a diagram for explaining in detail each part related to data control in the case of irradiating with a variable rectangular electron beam in the above configuration. The pattern generation section 5 is connected to the clock generation section 10.
According to the clock generated by the data memory 4, the first
1. One pattern data shown in FIG. 1(B) is read out and the operation code is interpreted. The operation code defines the pattern shape and shot time (beam-on time). The shot time is corrected for the proximity effect, so even for the same pattern group, the time the beam is turned on (i.e. the beam irradiation time) is different.
It is set for each pattern data. In addition, the pattern generation unit 5 calculates the starting point coordinates (X+, Y+) of one pattern shown in FIG. 11 (C1) and the size of the figure (X2, Y
2), one pattern is decomposed into shots as shown in FIG. 11(D) (in the example shown, it is decomposed into eight shots).
.

The pattern generator 5 then sends the beam position, beam (shot) size, and clock data (
The blanking signal @) shown in FIG. 11(E) is sent.

After these data are corrected by the pattern correction section 6, they are output to the corresponding sections at the next stage.

[Problem that the invention seeks to solve]

However, the conventional p-electron beam exposure apparatus described above has the following problems.

The beam position, beam size, and clock data created by the pattern generating section 5 are different for each pattern. Therefore, the processing time for one pattern data in the pattern generator 5 varies. For this reason, the pattern generation section 5 must perform sequential processing of reading one pattern data from the data memory 4 and, after completing this processing, receiving the next pattern data. Here, DRAM is generally used as the data memory 4, but it takes several 100 m5 (maximum 60 (h + S)) from receiving the read request from the pattern generator 15 to reading and confirming the data. Moreover, in the above sequential processing, since the 1I11111 clock of the entire electron beam exposure apparatus system uses a blanking signal (in other words, exposure clock), while the blanking signal is off (during this period, the 1I11111 clock of the entire electron beam exposure apparatus system is (determined according to the settling time, glitch removal of the D/A converter, processing time between patterns, etc.)
Data processing in the pattern generator 5 must also be stopped. Therefore, drawing takes a long time and is not efficient.

Therefore, an object of the present invention is to perform drawing by electron beam exposure at high speed and improve efficiency.

[Means for solving problems]

FIG. 1 is a block diagram of the principle of the present invention.

In the figure, a data memory 11 stores a plurality of pattern data. The cache memory 12 is a memory provided between the data memory 11 and the pattern generation section 13, and temporarily holds pattern data read from the data memory 11 in synchronization with the exposure section 14.

The pattern generating section 13 reads pattern data one by one from the cache memory 12, processes the data, and generates data necessary for exposure. The exposure section 14 exposes the sample to an electron beam according to this data.

[Effect]

The pattern data is read out from the data memory 11 asynchronously with the exposure unit 14 (that is, at a timing different from the exposure clock) and temporarily held in the cache memory 12. The pattern generating section 13 reads one pattern data from the cache memory 12 and performs data processing necessary for exposure.

Therefore, compared to the conventional pattern generation section which processes one pattern data and then accesses a large capacity data memory to read out the next one pattern data, in the present invention, the pattern generation section 13 processes one pattern data. Since one pattern data is read from 12, the time from access to data read can be shortened. Moreover, the pattern generating section 13 can read out one pattern data asynchronously with the exposing section 14 by using the time interval between one exposure and the next exposure (time when the blanking signal is off).

As a result, the sample can be drawn with the electron beam at high speed and efficiently.

〔Example〕

FIG. 2 is a block diagram of main parts of an embodiment of the present invention.

In the figure, a data memory 15 is a large capacity memory that temporarily stores pattern data read from a disk (corresponding to disk 2 in FIG. 10), not shown. The pattern generation section 16 includes a pitch memory 17, a cache memory 18, and a shot decomposition section 19. The pitch memory 17 is for converting one pattern data read from the data memory 15 so that the shot decomposition section 19 can easily perform shot decomposition. Of course, it is also possible to store one pattern data from the data memory 15 in the cache memory 18 without providing the pitch memory 17, but by providing the pitch memory 17,
The shot decomposition operation in the shot decomposition unit 19 can be made faster. The cache memory 18 temporarily stores one pattern data converted by the pitch memory 17. The capacity S of the cache memory 18 is such that it can store, for example, data for one subfield (typically 100 μIO of 0 small deflection areas), in other words, 100 pattern data.

The operation of reading one pattern data from the data memory 15, converting the data in the pitch memory 17, and storing it in the cache memory 18 is performed asynchronously with the exposure clock (blanking signal).

The shot decomposition unit 19 interprets the pattern shape and shot time and performs chicot decomposition using the converted one-pattern data read from the cache memory 18, and sends the beam position, beam (shot size), and clock data to the pattern correction unit 20. do.

Therefore, the shot decomposition section 19 corresponds to the function of the pattern generation section 5 described above with reference to FIGS. 10 and 11.

Furthermore, in terms of correspondence with the principle block diagram of the present invention shown in FIG. 1, the pattern generation section 13 in FIG. 1 corresponds to the shot decomposition section 19 of this embodiment. In this sense, in this embodiment, in addition to the shot decomposition unit 19, the pitch memory 17 and the cache memory 18 are combined into one pattern generation unit 1.
It is configured as 6.

The pattern correction section 20 corresponds to the pattern correction section 6 in FIGS. 10 and 11, and corrects the output data from the pattern generation section 16 according to the state of the drawing area of the sample, and /A converter and deflection amplifier (none of which are shown in FIG. 2).

Next, the operation of this embodiment will be explained with reference to FIGS. 3 to 5. Here, FIG. 3 is an explanatory diagram of the operation of the pitch memory,
FIG. 4 is a diagram showing the contents of the pitch memory, and FIG. 5 is an explanatory diagram of shot decomposition.

First, one pattern data 21 (FIG. 3) relating to one pattern shown in FIG. 5 is read from the data memory 15 and provided to the pitch memory 17. One pattern data consists of an operation code (defining the pattern shape and shot time), a pattern starting point (X+, Y+), and pattern size x 2° Y2. The pitch memory 17 stores the corresponding pitch (PX, PY) and number of shots (Nx, Nv) for each pattern size (X2, Y2).

Therefore, the pitch memory 17 stores the pattern size (X2
, Y2) and the number of shots (NX, NY) are output together with the operation code and starting point (X+, Y+). These data are illustrated as one pattern data 22 after conversion in FIG. The converted one-pattern data 22 is held in the cache memory 18 for one poem. These series of operations are asynchronous with the exposure zero check (blanking signal).

Therefore, it can operate even during exposure operation.

The shot decomposition unit 19 accesses the cache memory 18 and performs shot decomposition using the converted one pattern data 22.

1 pattern data has been converted in advance as described above, so the pitch (PX, PY) from the starting point (X+, Y+)
You only need to do this (NX, NY) times.

This process can be expressed as follows.

N=1 NY In this way, the shot decomposition process is greatly simplified.

In this way, shot decomposition is performed, and the output data of the shot decomposition section 19 is corrected by the pattern correction section 20, and then supplied to the exposure section via a D/A converter and a deflection amplifier, and drawing is performed. .

Next, a second embodiment of the present invention will be described with reference to a block diagram of the second embodiment shown in FIG.

In FIG. 6, the same elements as in FIG. 2 are indicated by the same numbers.

The feature of the second embodiment is that the cache memory 18 and the shot decomposition unit 19 shown in FIG.
, '19b, in that a switching section 23a was provided between the pitch memory 17 and the cache memories 18a and 18b, and a switching section 23b was provided between the shot decomposition sections 19a and 19b and the pattern correction section 20. . Each cache memory 18a and 18b has a capacity to store pattern data for one subfield. With the above configuration, for example, while the shot decomposition unit 19a is decomposing shots, the switching unit 23a can connect the bit memory 17 and the cache memory 18b, and pre-process pattern data for the next subfield. It becomes possible. Therefore, when changing subfields, the pattern generation process can be executed substantially without waste even during the necessary settling waiting time of the deflection amplifier.

FIG. 7 is a block diagram of a third embodiment of the present invention, in which the same components as in FIG. 6 are given the same reference numerals. The third embodiment is characterized by a shot decomposition section 19a in the second embodiment.

FIFO (f
1st-in first-out) memories 24a and 24b are provided. With this configuration, the outputs of the shot decomposition units 19a and 19b can be sequentially stored, and if there is a request from the pattern correction unit 20, the corresponding data can be output. In comparison, it is possible to achieve -1m advance processing, that is, to improve the processing speed.

Here, the maximum processing speed of the pattern correction section in the shell is 5
0 to 100n S/data, and the mining speed of the pattern generation part is 200 to 300TI8/data, so the second
It can be seen that performing advance processing using the configuration of the third embodiment is very effective. Although the second and third embodiments are examples of duplication, it is of course possible to triple or more.

FIG. 8 is a block diagram of a fourth embodiment of the present invention. In the figure, the same components as in the second embodiment shown in FIG. 6 are given the same reference numbers. The feature of the fourth embodiment is that the pitch memory 17 of the second embodiment is made 2M,
Pitch memories 17a and 17b are used as a switching unit 23a and cache memories 18a and 18b, respectively.
It was set up between. With this configuration, the pattern data read from the data memory 15 can be processed in parallel and output to the pattern correction section 20, thereby increasing the efficiency of the negative layer.

FIG. 9 is a block diagram of a fifth embodiment of the present invention. In the figure, the same reference numerals are given to the same components as in the third embodiment shown in FIG. 7. The feature of the fifth embodiment is that the pitch memory 17 of the third embodiment is duplicated,
Pitch memories 17a and 17b are provided between the switching section 23a and the cache memories 18a and 18b, respectively. With this configuration, data memory 1
5 can be processed in parallel and output to the pattern correction section 20, further improving efficiency.

Although the fourth and fifth embodiments have a duplex configuration, it is also possible to have a triple or more configuration.

〔Effect of the invention〕

As explained above, according to the present invention, a cache memory is provided between the data memory and the pattern generation section to temporarily hold the pattern data read out from the data memory asynchronously with the exposure clock. It can be done quickly and efficiently.

[Brief explanation of the drawing]

Fig. 1 is a block diagram of the principle of the present invention, Fig. 2 is a block diagram of main parts of an embodiment of the present invention, Fig. 3 is an explanatory diagram of the operation of the pitch memory, Fig. 4 is a diagram showing the contents of the pitch memory, FIG. 5 is an explanatory diagram of shot decomposition, FIG. 6 is a block diagram of the second embodiment, and FIG. 7 is a diagram of the third embodiment.
FIG. 8 is a block diagram of the fourth embodiment, and FIG. 9 is a block diagram of the fifth embodiment.
FIG. 10 is a block diagram of an electron beam exposure apparatus, and FIG. 11 is an explanatory diagram of conventional data processing. In the figure, 11... Data memory, 12... Cache memory, 13... Pattern generation section, 14... Exposure section, 15... Data memory, 16... Pattern generation section 17... Pitch memory; 18... Cache memory; 19... Shot decomposition unit; 20... Pattern correction unit. Patent applicant Fujitsu Ltd.

Claims (2)

[Claims] (1) Data memory that stores multiple pattern data (
11), and a pattern generator (13) that performs data processing for each pattern data and generates the data necessary for exposure.
and an exposure section (14) that exposes a sample to an electron beam according to the data, in which an exposure section (14) from the data memory (11) is inserted between the data memory (11) and the pattern generation section (13). An electron beam exposure apparatus characterized in that part (14) is provided with a cache memory (12) for temporarily holding pattern data read out asynchronously. (2) Electron beam exposure according to claim 1, wherein the cache memory (12) stores a starting point of one pattern data, a pitch for each shot, and a size of one pattern. Device.