JPH04206624A - Pulse count/analog signal conversion circuit and charged particle beam aligner - Google Patents

Abstract

PURPOSE:To enhance the accuracy and the speed of an electron beam aligner by a method wherein up-pulses and down-pulses are input from a pulse converter of a laser interferometer, an analog signal is output according to the difference in the number of the pulses and it is applied to a deflection system. CONSTITUTION:The output of a laser interferometer 47 is converted into pulses; their up-pulses UP and down-pulses DP are input to a conversion circuit 70; an analog signal is output according to the difference in the number of the pulses. At the circuit 70, one transistor is turned on at each up-pulses and one transistor is turned off at each pulse DP so as to prevent a glitch from being generated. This output is applied to a subdeflector, and a position feedback is performed without the shake and the dislocation of a beam. Thereby, the position feedback without the glitch can be performed by a simple constitution, and the accuracy and the speed of an electron beam aligner can be enhanced.

Description

[Detailed Description of the Invention] [Summary of the Invention] Regarding a circuit that converts the number of pulses into an analog signal and a charged particle beam exposure apparatus using the same, a conventional counter is used to obtain a deflection control signal for stage position feedback. and D
We have developed a circuit that directly converts the up pulses and down pulses output from the pulse converter of the laser l[+ into analog signals that indicate the difference in the number of these pulses, without using AC, and using this circuit, we can The purpose is to obtain a control signal, and the pulse number/analog signal conversion circuit has a first gate to which the up pulse is input, a second gate to which the down pulse is input, and the outputs of these gates. a flip-flop that is set and reset by , and a third gate that produces an output that opens the first gate and the second gate of the subsequent stage when the output of the flip-flop is 0 and the output of the subsequent flip-flop is 1; It has multiple gate groups consisting of, and furthermore,
1.0 of the output of each flip-flop of the plurality of gate groups
The structure includes a plurality of current circuits that output the same current and stop the output, and an addition circuit that takes the sum of the output currents of these current circuits, and also feeds back the movement of the stage to the charged particle beam deflection system. In a charged particle beam exposure system that does not disturb stage movement or drawing, an up pulse is output when the stage moves in one direction, and a down pulse is output when the stage moves in the opposite direction.These pulses are set and reset. multiple flip-flops (FF)
, a plurality of current circuits (R, Q) that output the same current from the output of the flip-flop, and an adder circuit (ADD) that takes the sum of the outputs of these current circuits, and the number of inputs of the up and down pulses. A pulse number/analog signal conversion circuit that outputs an analog signal according to the difference between the stages is configured to be used as a circuit that feeds back the movement of the stage to the charged particle beam deflection system.

[Industrial application field]

The present invention relates to a circuit that converts a pulse number into an analog signal and a charged particle beam exposure apparatus using the circuit.

In recent years, with the increase in the density of integrated circuits, new exposure methods using charged particle beams, particularly electron beams, and X-rays have been investigated and are not actually used, replacing photolithography, which has been the mainstream for forming fine patterns for many years. It's starting to look like this. A major feature of electron beam exposure is that it can form patterns using electron beams to form fine patterns on the order of microns or smaller.

Electron beam exposure methods include exposure methods using a point beam and a variable rectangular beam, and these are called single-stroke exposure methods, but they have extremely low throughput and are considered unusable for mass production of items such as memory. .

On the other hand, there is a method in which a pattern is formed as an opening in a stencil mask, and the beam is shaped and exposed by selectively passing through the pattern. This is called block exposure and is expected to speed up LSI manufacturing in the future.

[Conventional technology]

FIG. 4 shows an outline of an electron beam exposure apparatus using a stencil mask. 11 is an electron gun, 12 is a slit, 13
~18 is an electronic lens, 20 is a stencil mask, 21~
24 is a deflector, 3o is a wafer, and 3I is a stage.

The stage 31 is moved by a stage moving mechanism 32, and the amount of movement is measured by a laser interferometer 32. A digital/analog converter DAC and a buffer amplifier AMP are attached to the deflector, and a moving mechanism 34 is attached to the stencil mask 2o. A pattern control controller 35, a blanking control circuit 36, a deflection control circuit 37, a sequence controller 38, and a data memory 39 are used to control these.
, a processor 4o, an interface 41, and a magnetic disk 42.

The pattern controller 35 deflects the electron beam as shown by the dotted line, and moves the stencil mask 2o in the horizontal direction via the mask moving mechanism 34.
The electron beam passes through the desired pattern on the stencil mask and is shaped into the pattern. Since the same pattern appears repeatedly in a memory or the like, if the pattern is formed on a stencil mask, the pattern can be drawn all at once, increasing the processing speed. If the repetition rate is high (none, except that the entrance platform is drawn each time), an aperture for a variable rectangular beam is also formed in the stencil mask for this purpose.

Drawing on the wafer 1 with an electron beam is performed as shown in FIG. In this example, move from the bottom left edge to the top left edge, move one step knob to the right, then move from bottom to bottom, move one step knob to the right at the bottom end, and move from bottom to top. The order is... During this time, a stripe-like area about 2 mm wide is exposed, and this exposure area is divided into a 2 mm x 2 mm main field and subfields obtained by subdividing the main field. Movement of the exposure point within this main/subfield is performed by deflecting the electron beam by the main/sub deflection system.

FIG. 6 shows an outline of the stage drive mechanism, 32a, 32
b is a motor for driving the stage 31 in the X and Y directions; 33a;
, 33b is a laser length measuring device that detects the amount of movement in the X and Y directions, and 32c is a stage control section.

Exposure is performed by deflecting the electron beam without moving the stage. At this time, the stage is moved so as not to disturb the drawing pattern, and an outline of the control mechanism for this purpose is shown in FIG. In this figure, 45 is a pattern generator, 46
The pattern corrector 46, DAC, AMP, deflector 24
Drawing is performed within the subfield by deflecting the electron beam B together with a. Also, 47 is a pulse converter, 48
is a stage position counter, which in FIG. 6(a) works together with the DAC, AMP, and deflector 24b to deflect the electron beam B so that the projection position of the electron beam remains unchanged even if the stage 3I moves. .

With this latter system (this is the position feedback system and the former is the drawing system), even if drawing is performed while the stage is moving, the drawn object will not be deformed by the stage movement. Figure 7 (b
) differs from FIG. 7(a) in that the signals are combined into one after DAC, but the rest is the same as FIG. 7(a).

The output of the laser interferometer is converted into pulses by a pulse converter 47, this pulse is counted up/down by a stage position counter 48, and the counted value is converted into an analog signal by a DAC. When the stage 31 moves in one direction, the output signal of the DAC becomes as shown at 32 in FIG. On the other hand, a drawing-related signal is like S in the same figure.

The position feedback system ensures that the drawing is not disturbed by vibrations of the stage. The pulse converter 47 generates one pulse when the stage moves by 0.005 μm (=λ/120). Therefore, the moving speed of the stage is 50 mm/sec.
c, the output frequency is IOMH2, and the counter 48 counts up/down this.

As shown in FIG. 10, the stage position is returned to the main deflector and sub deflector. The count value of the counter 48 of the laser interferometer output is taken into the register 49 at the start of processing in the main field, and the count value is taken into the register 51.
The difference from the target value set at (this is a value that brings the beam to the center of the main field) is applied to the main deflector 64, thereby performing stage position feedback to the main deflection system. Further, the difference between the count value of the counter 48 and the counter count value taken into the register 49 is added to the sub-deflector 65, thereby performing stage position feedback to the sub-deflection system.

In detail, the above difference is made into analog by DAC and converted to analog by AMP.
is amplified and added to the main/sub deflector. And since this AMP can produce positive and negative outputs, this positive,
A glue margin register 50 is provided so that the negative range can be used. If we add the value of the glue allowance register in the main system and subtract it in the sub system, there is no beam deflection due to these as a whole, and the amplifier output is positive.

Both negative and negative ranges are available.

[Problem to be solved by the invention]

The count value of the counter 48 is converted into an analog signal by a DAC, but this poses a problem in that glitches appear in the output of the DAC. In other words, the DAC consists of a switch that turns on and off at each digit of binary digital data (1, 0) and weighted resistors, but for example, 01111...1 and the next 10000, old... In this case, a large number of switches are switched from on to off all at once, and the heaviest weighted resistor is switched on instead.
In particular, if a new resistor is turned on too late or too early, a large current change occurs, causing a glitch as shown in FIG. 9(a).

This is a problem when precisely exposing a fine pattern because it causes the beam to oscillate, causing a positional shift and disturbing the drawn pattern. Therefore, it may be possible to smooth the signal using a capacitor, but in this case, the result will be as shown in 9th IN (b), and the glitch cannot be removed. Furthermore, since the count of the stage position counter is asynchronous with the exposure clock, it is not possible to interrupt the exposure at the position where a glitch appears.

Also, with conventional DACs, it is difficult to achieve both accuracy and speed; for example, a DA with 12-bit accuracy and 10MHz speed
Manufacturing C is not easy. There is also the problem of bit mismatch. For these various reasons, conventional DACs will become increasingly finer in the future, with line widths of 0. 1μm, o, 05μm
, it is insufficient to expose LSI patterns as fine as 0.02 μm with high reliability and high speed.

Therefore, in order to obtain a deflection control signal for stage position feedback, the present invention directly converts up pulses and down pulses output from a pulse converter of a laser interferometer, without using a counter and DAC as in the conventional method. The object of the present invention is to develop a circuit that converts into an analog signal indicating the difference in numbers, and to use this circuit to obtain the deflection control signal.

[Means for Solving the Problems] As shown in FIG. 1, in the present invention, the up pulse UP and down pulse DN obtained by pulse converting the output of a laser interferometer are input, and the difference between these pulse numbers is calculated. A conversion circuit 70 that outputs an analog signal according to the deflection system is used in the deflection system.

The output of the pulse number/analog signal conversion circuit 70 is applied to the sub-deflector system in FIG. The main deflector system is the same as the conventional one (FIG. 10).

As shown in FIG. 2, the conversion circuit 70 includes a first gate G1 to which the up pulse UP is input, a second gate DN to which the down pulse DN is input, and a flip-flop that is set and reset by the outputs of these gates. A gate group consisting of a FF and a third gate G3 that generates an output that opens the first gate G1 and the second gate in the subsequent stage when the output of the flip-flop is 0 and the pressure of the subsequent flip-flop is 1. Provide the required number of output stages (N). Further, each of these flip-flops is provided with current circuits R and Q that output or do not output the same current I depending on the outputs of 1 and 0, and an adder circuit that calculates the sum of the outputs of these current circuits. A.D.
Equipped with D.

[Effect]

In the circuit shown in Fig. 2, each time the up pulse UP is input, the outputs of gate groups A, B, C, and D change from 1100 to 1110. The output of ADD increases by ■. When the down pulse DN is input, the opposite is true, and the outputs of A, B, C, and D are 1111, 1110, 110.
0,1000, and the adder output decreases by I.

In this way, the number of inputs of up pulses UP from adder ADD,
An analog signal (current or voltage) is obtained according to the number of input down pulses DN and, if these are input alternately, the difference in the number of these pulses.

When one up pulse is input, the number of transistors that are turned on increases by one, and when one down pulse is input, the number of transistors that are turned off increases by one. Only one transistor is switched on/off by pulse input, and the counter and D
A large number of transistors do not switch at the same time in a certain order of magnitude as in AC. Also, the output current ■ is all the same, that is, the resistance R is the same value in each current circuit, and like a DAC, it is 1, 2
゜4.8. ...and not weighted by 2'.

As a result, no glitch occurs in FIG.

Therefore, this pulse number/analog signal conversion circuit 70 is
When used in a charged particle beam exposure apparatus as shown in the figure, disturbances in drawing due to glitches can be avoided.

Glitches can be prevented from occurring by inserting/removing a resistor with the same resistance value each time a pulse is input without weighting the resistors. However, if this is simply attempted, the configuration of the decoder section that turns on/off the current circuit becomes extremely complicated, and it is virtually impossible to implement it in a multi-bit system. In this respect, the decoder section (gate group) in FIG. 2 is very simple as shown.

〔Example〕

The conversion operation of the circuit of FIG. 2 will now be described in detail. Gate group A
, B, C, etc. are 40 if they are 12 bits.
Since there are 96 steps, 4096 steps are provided correspondingly. Now, to talk about these 18 items, uppulse U
When P is input, the output of each group changes as shown in Table 1, and when down pulse DN is input, the output of each group changes as shown in Table 2.

Table 1 ooooooooooooooooooo.

Table 2 11111111 III I I ] 111 That is, since all the weights are the same (I), l increases by 1 each time UP is entered, and l decreases by 1 or 1 each time DN enters. now,
The i+1 bit from the right of Table 1.2 is C, the i bit is B
, when the i-1st bit is A, when the up pulse UP is input, B=O1, if A=1, B changes to 1, and when the down pulse DN is input, C=0. If B=1, B changes to 0, and in any other state, it becomes 1. O There is no change. The circuit in Figure 2 utilizes this point; if the output of the current stage (for example, B) is 0 and the output of the subsequent stage (A in this example) is 1, the gate G
3 becomes 1, gate G is opened, and input of up pulse UP to the current stage is permitted. If the output of the current stage is 1 or the output of the subsequent stage is O, the output of the gate 1'Gs is 0, which closes the gate Gl and prohibits input of the up pulse UPO to the current stage. For down pulse DN, the previous stage (C in this example)
If the output is O and the current stage output is 1, the output of the previous stage gate G3 is 1
Then, the gate G2 of the current stage is opened and the input of the DN to the current stage is permitted. If the output of the current stage is 0 or the output of the previous stage is 1, the output of the previous stage gate G3 becomes O, closing the current stage gate G2 and prohibiting input of DN to the current stage. This is Table 1.2
The output state can be as follows.

In FIG. 2, A=B=l and C=D=0, and in this case, the gate GI of stage 0 is open and accepts UP, and the gate G2 of stage B is open and ready to accept DN. In stage B, G1 is closed and input of UP is prohibited, and in stage 0, G2 is closed and input of DN is prohibited. G in A and D stages
1. G2 is closed and input of both UP and DN is prohibited. Assuming that the output is the Q output of the FF, the output of G1 sets the FF, and the output of G resets the FF. When the output is 1, the transistor Q is on, and when the output is 0, the transistor Q is off. Gate CI, G2 is AND gate, gate
Ga is an inhibit gate, but this can be changed as appropriate. For example, if the FF (7) Q output is input to G, G becomes an AND gate. Further, the transistor Q may be a pnp instead of the npn shown in the figure, and may be a MOS FET instead of a bipolar transistor.

In addition to the present invention, glitches can be eliminated without using a DAC consisting of weighted resistors and switches;
For 2 bits and 4096 steps, it is sufficient to provide 4096 circuits that output the same current and turn them on and off one by one each time a pulse is input. The circuit 80 in FIG. 3 counts up and down the counters UP and DN, decodes the counted value with a decoder, and
Turns on/off four current circuits. However, with this configuration, the circuit scale of the decoder becomes enormous. In other words, in the case of a memory decoder, the circuit scale is not large because it is sufficient to select one of 4094 wires, but in the case of a parallel current switch circuit, Table 1.
2, the number of current circuits that are turned on and off by pulse input is 1.
．． 2, 3.・・・・・・Increased to 4096 and then 409
6, 4095, 4094, etc. Since it decreases by 1, a decoder that makes such a selection is a memory decoder (such as a NOR gate that receives a selection of each bit of the address and its inverted bit). It is virtually impossible to configure it like this. In this regard, the conversion circuit 70 of the present invention
As is clear from Fig. 2, the configuration is extremely simple, and the circuit scale is sufficient to be mounted on one chip together with the parallel current circuit.

The pulse number/analog signal conversion circuit 70 in FIG. 1 is initially set to all 0s when the stage is progressing in the UP direction (the direction in which UP occurs), and in this state, the analog final feedback output is 1. Adjust so that The glue margin register 50 outputs 10V to the main deflector system. On the other hand, if it is proceeding in the DN direction, the initial setting is all set to 1, and in this state, the analog final feedback output is adjusted to be 10V. There is one in the glue register 50.
output. Then, to apply feedback at the position where the stage is stopped, initialize it to 100...0, and adjust so that the analog final feedback output is Ov in this state. At this time, the glue allowance register 50 contains OV.
output. SP in FIG. 1 is a set pulse that performs these operations.

FIG. 9(d) shows the output of the circuit of the present invention with a glitch, and FIG. 9('c) shows the output of the conventional circuit with respect to this.

〔Effect of the invention〕

As described above, according to the present invention, it is possible to provide an exposure apparatus that can perform position feedback without glitches and that does not cause positional deviation or blur due to beam fluctuation. Furthermore, unlike DAC, there is no bit misalignment, and exposure is possible with high precision and stability, while the stage is continuously moved.

Further, the pulse number/analog signal conversion circuit of the present invention has a simple circuit configuration and can be relatively compact even when extremely multi-stage output is required. This conversion circuit can be used not only in exposure equipment but also in various applications where glitch-free, high-precision conversion output is desired.

[BRIEF DESCRIPTION OF THE DRAWINGS] FIG. 1 is a block diagram of the exposure apparatus of the present invention, FIG. 2 is a circuit diagram of the main part of the conversion circuit of the present invention, and FIG. 3 is an explanatory diagram of the advantages of the circuit of FIG. Fig. 4 is an explanatory diagram of the electron beam exposure apparatus, Fig. 5 is an explanatory diagram of the wafer exposure procedure, Fig. 6 is an explanatory diagram of the stage movement system, Fig. 7 is an explanatory diagram showing a conventional example of position return, and Fig. 8 is an explanatory diagram of the stage movement system. The diagram is DA
FIG. 9 is an explanatory diagram of a glitch, and FIG. 10 is an explanatory diagram of position feedback to the main/sub deflection system. In Figure 1, 70 is a pulse number/analog signal conversion circuit,
In the figure, 01 to G3 are gates, FF is a flip-flop, and R
and Q are a resistor and a transistor that constitute a current circuit. Applicant: Fujitsu Limited

Claims (1)

[Claims] 1. First gate (G_1) to which up pulse is input
and the second gate (G_2) to which the down pulse is input.
and a flip-flop (FF) that is set and reset by the outputs of these gates, and when the output of the flip-flop is 0 and the output of the subsequent flip-flop is 1, the first gate and the second subsequent flip-flop A plurality of gate groups (N
), further comprising a plurality of (N) current circuits (R, Q) that output the same current (I) and stop outputting the same current (I) when the output of each flip-flop of the plurality of (N) gate groups is 1 or 0; A pulse number/analog signal conversion circuit characterized by comprising an adder circuit (ADD) that sums the output currents of these current circuits. 2. In a charged particle beam exposure system in which stage movement is fed back to the charged particle beam deflection system so that the stage movement does not disturb drawing, the up pulse output when the stage moves in one direction and the up pulse output in the opposite direction of the stage The down pulse to be output by movement is input,
Multiple flip-flops (FF) that are set and reset by these, multiple current circuits (R, Q) that output the same current at the output of the flip-flops, and an adder circuit (ADD) that takes the sum of the outputs of these current circuits. ), and a pulse number/analog signal conversion circuit that outputs an analog signal according to the difference in the number of inputs of the up and down pulses is used as a circuit that feeds back the movement of the stage to the charged particle beam deflection system. A charged particle beam exposure apparatus characterized by: