Radiant energy – Irradiation of objects or material
Reexamination Certificate
1999-06-30
2003-05-20
Lee, John R. (Department: 2881)
Radiant energy
Irradiation of objects or material
C250S492200, C250S492210, C250S492220, C250S398000
Reexamination Certificate
active
06566662
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates generally to a charged beam exposure system using electron beams etc for writing a pattern of a semiconductor integrated circuit such as an LSI etc on a sample such as a wafer and a mask at a high speed with a high accuracy, and more particularly to a charged beam pattern writing system contrived to enhance a throughput of the pattern writing by optimizing a stage speed.
2. Description of the Background Art
With scaling-up of a semiconductor device and as well fining of elements included therein, there widely spreads a pattern writing technology of writing a pattern on a wafer or mask by use of a charged beam exposure system.
In such a charged beam pattern writing technology, a range in which the charged beams can be deflected is limited, and the pattern is written by making the beams consecutively scan in an orthogonal direction (an X-direction) the reticle or the mask defined as a sample placed on the stage while deflecting the beams within a main deflection width (a Y-direction). Therefore, as shown in
FIG. 33
, for instance, a unit pattern writing area corresponding to a chip area is partitioned into frame fields F
1
-FN each defined by the main deflection beam deflection width. In
FIG. 33
, points P
1
-PN respectively denote frame originating points of the frame fields F
1
-FN. This type of pattern writing method is referred to as normal writing.
On the other hand, there has been carried out a pattern writing method of writing the same pattern a plurality of times while controlling a beam irradiation quantity of the charged beams for the purpose of enhancing a pattern dimensional accuracy, which is called a multi pattern writing. In the multi pattern writing, the pattern writing of a stripe field corresponding to a frame in the normal pattern writing, and a step movement is made in the direction (Y-direction) orthogonal to the stage moving direction. The above processes are repeated, thereby writing the pattern in all the stripe fields. This step movement corresponds to a width that is 1
umber-of-multi-writings of a frame width.
FIG. 34
shows a concept of the multi-writings described above, wherein double writings are shown in this example, and the stripe field is shifted by half a width of the frame field. In this figure, Ps
1
-Psm denote stripe originating points of stripe field S
1
-SM and a main-deflection beam deflection width equals to a stripe field width. The main-deflection beam deflection width equals to the stripe field width.
Further, the frame or stripe field is, if the necessity may arise, subdivided into sub-deflection fields in which the beams can be deflected by a sub-deflection beam deflector.
In the charged beam pattern writing method explained above, a pattern writing speed is constant in the frame or stripe field. Further, a patten writing time (for writing a desired pattern by controlling a beam position and a beam configuration) defined in the frame or stripe field must be a value enough to follow up the stage speed when writing the pattern in the frame or stripe field.
The following methods have hither been proposed for determining the stage speed which might satisfy the above condition.
(i) A stage speed for executing a process of writing the pattern in the frame field, is determined based on a total number of shots embraced in the frame field and a total number of sub-deflection fields (refer to Japanese Patent Application Laid-Open Publication No.1-152726).
(ii) A stage speed for executing a process of writing the pattern in the frame field is determined based on a virtual number of shots which is obtained by dividing a total areal size of the pattern embraced in the frame field by an average areal size of one shot, and on a total number of sub-deflection fields (refer to Japanese Patent Application Laid-Open Publication No.1-243520).
(iii) The frame field is virtually partitioned into segments each having a predetermined length, a number of shots is calculated per partitioned segment, a pattern writing time in the segment having a maximum number of shots, and a value obtained by dividing this pattern writing time by the length of the segment is set as a stage speed for executing the process of writing the pattern in the frame field (refer to Japanese Patent Application Laid-Open Publication No.2-5406).
There arise, however, the problems inherent in the conventional speed determining methods.
According to the conventional methods (i), (ii), the average stage speed is determined based on such a premise that the patterns are uniformly disposed in the frame field. In an actual LSI device pattern, however, a pattern layout in the frame field is not uniform, and a pattern density is not uniform, wherein a high pattern density area and a low pattern density area alternately exist. Generally, a longer pattern writing time is needed for writing the pattern in the high pattern density area.
Hence, at the speeds calculated by the conventional methods (i) and (ii), a writing error (caused when the pattern writing process is unable to follow up the stage speed) does not occur in the pattern writing in the low pattern density area, but occurs in the high pattern density area. The problem is that the stage speed is decreased, and the pattern writing process in the concerned frame field must be again carried out.
On the other hand, according to the method (iii), the frame field is virtually partitioned into block segments, and the speed is determined for the block segment, thus considering the problem derived from the low pattern density, which remains unsolved by the conventional methods (i) and (ii). Resultantly, however, the method (iii) is no better than obtaining the average speed in the block regions.
For example, as shown in
FIG. 36
, the frame is partitioned into five blocks each having a width W, thus virtually defining the blocks. Assuming an example in which a high pattern density region A extend to some parts of both of the blocks
3
,
4
, according to the method (iii), it follows that the pattern is written in the block at the stage speed (V
3
=W/T) determined in the block
3
.
However, though this speed V
3
is optimal to only the block
3
in terms of the pattern writing, actually the pattern writing is effected consecutively in the block
3
and the block
4
. Therefore, a stage speed optimal the pattern writing in the region A extending to both of the blocks
3
,
4
is given by V
34
=W/2T. Accordingly, there might be a possibility in which the error occurs in the pattern writing because of the pattern writing process being unable to follow up the stage movement in terms of the relationship of V
3
>V
34
. Hence, the problem is that the stage speed is decreased as in the case of the methods (i), (ii), and the pattern writing process must be re-executed.
Moreover, the stage speed in the stripe field in the multi pattern writings, on the occasion of determining it, can be calculated by the conventional methods (i), (ii), (iii). However, for instance, as shown in
FIG. 37
, a chip composed of five frame fields F
1
-F
5
undergoes the multi pattern writings of which the number is 4, with the result that the number of stripe fields is 23 given in the following formula.
Number-of-stripe-fields=number-of-frame-fields+(number-of-multi-pattern-writings−1)×(frame field+1)
Therefore, in this case, the process is executed for all the stripe fields (23 fields), wherein the number of writings is over N-times as large as the normal writings.
As described above, when writing the pattern at the stage speed obtained by the conventional methods, there might be a possibility in which the error occurs in the pattern writing. Further, a stage speed calculation processing time that is over N-times as large as the time in the normal writing, is needed in the multi pattern writing effected N times. This leads to an increase in the pattern writing time, i.e., a decline of throughput of the whole exposure system.
SUMMARY OF THE INVENTIO
Hara Shigehiro
Higurashi Hitoshi
Murakami Eiji
Lee John R.
Vanore David A.
LandOfFree
Charged beam exposure system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Charged beam exposure system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Charged beam exposure system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3078299