JPH10173274A - Excimer laser device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain an excimer laser device which is further improved in exposure accuracy and optical processing accuracy by controlling halogen gas supply aiming at a partial halogen gas pressure that can minimize the output fluctuation of the laser device. SOLUTION: Laser pulse oscillation is performed in a laser chamber 2 by exciting a laser gas by discharge and the emitted pulsed light is narrowed in band by means of a band-narrowing unit 6. Then the pulsed light narrowed in band returns to the chamber 2 and is outputted as an oscillated laser beam L through a partially transmissive mirror 7 after the light is amplified in the chamber 2. Part of the outputted laser beam L is sampled through beam splitters 8 and 13 and made incident to a light receiving element 14. A CPU 15 detects the light energy Ei (i=1, 2, 3,...) of each pulse oscillation based on the received light output of the element 14 whenever the pulse oscillation is performed and controls a laser power source circuit 16 and a gas supplying device 17. Therefore, a uniform pulsed light output can be obtained when halogen gas supply is controlled by giving the highest priority to the suppression of the fluctuation of each output energy.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for outputting a laser beam to a processing apparatus for performing predetermined exposure or processing on a semiconductor, a polymer material, an inorganic material, or the like using the laser beam, and a laser chamber for the laser beam. In particular, the present invention relates to an excimer laser device that performs laser pulse oscillation by filling the inside thereof with a laser gas containing a halogen gas. The present invention relates to an improvement for always obtaining a uniform pulse energy value when performing a burst mode operation that is repeatedly performed.

[0002]

2. Description of the Related Art In a field using ultraviolet light, such as a semiconductor exposure apparatus, strict exposure control is required to maintain the resolution of a circuit pattern at a certain level or more. . However, an excimer laser used as a light source of a semiconductor exposure apparatus has a variation in pulse energy for each pulse due to a so-called pulse discharge excitation gas laser, and it is necessary to reduce the variation in order to improve the accuracy of exposure control. There is.

On the other hand, a semiconductor exposure apparatus alternately repeats exposure and stage movement. That is, FIG.
Although the semiconductor wafer W on which the C chips TP are arranged is shown, in the exposure of the stepper method, when an exposure process of irradiating a large number of continuous pulse lights to one IC chip TP on the semiconductor wafer W is completed. Then, the wafer W or the optical system is moved so that the next unirradiated IC chip TP is irradiated with the continuous pulse light, and after this stage movement, the same light irradiation as described above is performed. Then, while performing such exposure and stage movement alternately, all the ICs on the semiconductor wafer W
When the exposure of the chip TP is completed, the exposed wafer W is carried out, the next wafer W is set at the irradiation position, and the same light irradiation as described above is repeated.

As described above, in a semiconductor exposure apparatus, exposure and stage movement are alternately repeated.
As shown in FIG. 19, the operating state of the excimer laser, which is the light source of the exposure apparatus, is inevitably a continuous pulse oscillation operation in which laser light is continuously pulsed a predetermined number of times, and a pulse oscillation is stopped for a predetermined time. It becomes a burst mode in which the oscillation pause time t is repeated.

That is, in the burst mode operation shown in FIG. 19, the oscillation stop time t corresponds to the time required for moving the stage in the semiconductor exposure apparatus, and the oscillation stop time t is not necessarily constant due to various causes. Does not. For example, the oscillation stop time when exchanging wafers is much longer than that when moving the stage between IC chips, and the oscillation stop time when moving between IC chips in the same row is replaced with a different row. The oscillation stop time when moving between IC chips is completely different. Further, when the number, arrangement, and the like of the IC chips mounted on the wafer are different, the oscillation stop time is also changed. There are various other causes for changing the oscillation stop time. Note that FIG. 19 shows the energy intensity of each pulse when the excitation intensity (power supply voltage, discharge voltage) is fixed to a constant value.

When the length of the oscillation stop time t changes in the burst operation as described above, this change causes a large change in the output of each laser pulse as shown in FIG. That is, when the oscillation stop time t is short, the influence of the past laser oscillation remains as an increase in the gas temperature, turbulence of the gas in the laser chamber, a local increase in the electrode temperature, and the like, and the oscillation stop time t is reduced. If it is long, the effect of the past laser oscillation has disappeared on the laser. For this reason, as shown in FIG. 19, even when the discharge power supply voltage of the laser is constant, the output energy decreases when the pause time is short, and increases when the pause time is long. The laser output greatly changes according to the oscillation stop time.

On the other hand, as described above, since the excimer laser is a pulse discharge pumped gas laser, it is difficult to always oscillate with a constant pulse energy. This is because (1) the density disturbance of the laser gas occurs in the discharge space due to the discharge, making the next discharge non-uniform and unstable, and (2) the discharge electrode due to this non-uniform discharge. There is a case where a local temperature rise occurs on the surface, which deteriorates the next discharge and makes the discharge uneven and unstable.

In particular, the tendency is remarkable at the beginning of the continuous pulse oscillation period, and as shown in FIG. 20, in the spike region including the first few pulses after the elapse of the oscillation pause period t, the spike region is relatively high at first. The pulse energy is obtained, and thereafter, the pulse energy gradually decreases, so-called a so-called spiking phenomenon appears. At the end of this spike region, the pulse energy goes through a plateau region followed by a relatively high level of stable values before entering the steady region.

As described above, in the excimer laser apparatus of the burst mode operation, the above-described variation in energy for each pulse lowers the accuracy of the exposure control, and the spiking phenomenon further significantly increases the variation. There is a problem that accuracy is greatly reduced.

Therefore, the present applicant takes advantage of the property that the energy of a pulse oscillated increases as the excitation intensity (charge voltage, discharge voltage) increases, and makes use of the property of the first pulse of continuous pulse oscillation in burst mode. The discharge voltage (charge voltage) is reduced, and the discharge voltage of the subsequent pulses is gradually increased. In other words, the discharge voltage is changed for each pulse to prevent the initial energy increase due to the spiking phenomenon. Various patents have been filed for inventions relating to control for preventing the occurrence of loose spiking (Japanese Patent Application No. Hei.
191056, JP-A-7-106678 (Japanese Patent Application No. 5-249483) and the like.

That is, according to the above-mentioned prior art, the energy of each pulse of the continuous pulse oscillation is considered in consideration of various parameters such as the oscillation suspension time t and the power lock voltage (power supply voltage determined according to the deterioration of the laser gas). Is stored in advance in a table for each pulse of the continuous pulse oscillation, and the actual pulse energy Ei (i = 1, 2,...) During the current continuous pulse oscillation is detected. Then, the detected value Ei is compared with the pulse energy target value Er, and the prestored discharge voltage data for each pulse is corrected and updated based on the comparison result. This correction voltage data is used as discharge voltage data in the next burst cycle.

In the discharge voltage correction control, the pulse energy Ei at the time of laser oscillation is detected based on the discharge voltage data Vi stored in the table, and the difference ΔE (= Ei−Er) from the target energy Er is calculated. A correction discharge voltage value ΔV (= G · ΔEG: gain constant) is calculated in accordance with the difference ΔE, and the discharge voltage data Vi stored in the table is corrected by the correction discharge voltage value ΔV. Discharge voltage data Vi ′ (= Vi + ΔV) is obtained.

However, according to the above prior art,
Since spike killer control is performed not only in the spike region but also in the plateau region and the steady region, the effect of suppressing variation in pulse energy in regions other than the spike region is not sufficient.

This is because in the initial pulse of the continuous pulse,
The effect of the laser oscillation pause (the laser stabilizes) remains strong, and the output power becomes larger than in other regions even when the same discharge voltage is applied, but the laser oscillation pauses in the plateau and stable regions thereafter. It is considered that the influence of the pulse oscillation until immediately before (such as an increase in the electrode temperature and disturbance of the laser gas) is more strongly affected.

Further, in the above-mentioned prior art, since the spike killer control is executed for all the pulses of the continuous pulse oscillation, the amount of stored data for the spike killer control is increased, and a large memory capacity is required. There is also a problem that it takes time to read data.

In a halogen gas-filled excimer laser, laser oscillation is performed by filling a laser gas containing a halogen gas into a laser chamber. Therefore, such an excimer laser requires an operation. As the process proceeds, the halogen gas is consumed due to the evaporation of the electrode material and the chemical reaction with the material constituting the laser chamber, and the laser output decreases. Therefore, conventionally, the following control has been performed to compensate for a decrease in laser output due to consumption of halogen gas.

That is, the output of the laser increases with an increase in the excitation intensity (charging voltage, power supply voltage). Therefore, conventionally, the laser output is detected and the power supply voltage value is controlled in accordance with the detection. Stabilizes the laser output. This control is usually called power lock control.

However, even with this control, if the operation is continued for a long time, the oscillation efficiency is reduced due to consumption of the halogen gas, and a predetermined output cannot be maintained unless the power supply voltage (power lock voltage) is gradually increased. .

To solve such a problem, Japanese Patent Application Laid-Open No. 3-166
In the publication No. 783, attention is paid to the fact that there is a laser gas pressure value that maximizes the oscillation efficiency (ratio of output laser energy to input power) for each power supply voltage value. As the voltage increases, the power supply voltage and the laser gas pressure are controlled so that the oscillation efficiency maintains the maximum value.

In other words, this prior art focuses on the oscillation efficiency of the laser and attempts to control the charging voltage and the laser gas pressure so that the oscillation efficiency always maintains a maximum value.

The method according to the prior art is an effective method when an excimer laser is used for processing in which it is most important to increase the laser output as much as possible.

However, when an excimer laser is used in a stepper type or step & scan type reduction projection exposure apparatus, the problem is how to increase the laser output of each pulse (to increase the oscillation efficiency). The most important purpose is to obtain a pulse light having a uniform output.

That is, according to the above-mentioned prior art, since the power supply voltage control and the laser gas supply control are not performed with a view to obtaining a uniform laser output, it is impossible to improve the exposure accuracy by one. It is.

The present invention has been made in view of such circumstances, and an excimer laser in which the pulse energy of all the pulses of continuous pulse oscillation is always made uniform to further improve the accuracy of light exposure and light processing. It is intended to provide a device.

[0025]

According to the present invention, a laser gas containing a halogen gas is sealed in a laser chamber, and a pulse discharge is performed in the laser chamber to excite the laser gas to generate pulsed laser oscillation. A burst mode operation in which a continuous oscillation operation of continuously performing a pulse oscillation of a laser beam for a predetermined number of times and a stop operation of suspending the pulse oscillation for a predetermined oscillation suspension time are alternately performed in one burst cycle. In an excimer laser device that performs control such that each output energy of the pulse oscillation falls within a predetermined target value range, a pulse number indicating a pulse order within one burst period, and a plurality of different oscillation pauses. Using the time as a parameter, the variation in the output of each pulse oscillation falls within a predetermined allowable value range. And voltage data table means for storing in advance respective initial power supply voltage, and oscillation-stopping time measuring means for measuring the oscillation-stopping time for each burst period,
Oscillation control for reading a power supply voltage value corresponding to the measured oscillation suspension time and a pulse number from the power supply voltage data table means every one burst cycle, and performing pulse oscillation according to the read power supply voltage value Means, monitoring means for sequentially monitoring the output of each pulse of continuous oscillation in association with a pulse number, gas replenishing means for replenishing the laser gas into the laser chamber, and each pulse laser based on the monitor output of the monitoring means. A variation calculating means for determining a variation in output energy of the oscillation light;
A gas replenishing control unit for replenishing the halogen gas by controlling the gas replenishing unit so that the calculated variation falls within a predetermined target value range.

According to the present invention, in the configuration in which the power supply voltage control for preventing the spike is performed in consideration of the oscillation stop time and the pulse number, the output energy of each pulse oscillation light is not prioritized, but the output energy of each pulse oscillation light is adjusted. The halogen gas supply control is performed with priority given to suppressing the variation. That is, the halogen gas supply control is performed aiming at the halogen gas partial pressure at which the variation in the laser output is a minimum value or a value near the minimum value.

Therefore, according to the present invention, the variation in the output of each pulse oscillation light can be suppressed to a minimum, and the excimer laser device of the present invention is applied to a light source for a reduced projection exposure apparatus for performing reduced projection exposure of a semiconductor. By doing so, it becomes possible to perform a highly accurate exposure process.

According to the present invention, a laser gas containing a halogen gas is sealed in a laser chamber, and a pulse discharge is performed in the laser chamber to excite the laser gas to perform pulsed laser oscillation. A burst mode operation is repeatedly performed in which a continuous oscillation operation in which pulse oscillation is continuously performed and a stop operation in which the pulse oscillation is paused only for a predetermined oscillation pause time are performed in one burst cycle. In an excimer laser device that performs control such that each output energy falls within a predetermined target value range, the variation in the output of each pulse oscillation is set to a predetermined allowable value using the time elapsed since the start of oscillation and the oscillation pause time as parameters. Voltage data table means for storing in advance the initial values of the power supply voltage falling within the range An oscillation pause time measuring means for measuring the oscillation pause time for each burst cycle, a clock means for measuring an elapsed time from the start of oscillation of the pulse oscillation for each pulse oscillation, and Oscillation control for reading from the power supply voltage data table means a power supply voltage value corresponding to the measured oscillation suspension time and corresponding to the elapsed time from the start of the measured oscillation, and performing pulse oscillation according to the read power supply voltage value. Means, monitor means for sequentially monitoring the output of each pulse of continuous oscillation, gas supply means for supplying the laser gas into the laser chamber, and output energy of each pulsed laser oscillation light based on the monitor output of the monitor means. A variation calculating means for determining the variation, and the gas replenishment so that the calculated variation falls within a predetermined target value range. So that comprises a gas supply control means for supplying a halogen gas by controlling the stage.

According to the invention, in the configuration in which the power supply voltage control for preventing the spike is performed in consideration of the oscillation stop time and the elapsed time from the start of the oscillation, the output energy of each pulse oscillation light is not prioritized. The halogen gas supply control is performed by giving top priority to the suppression of the variation of each output energy.

Therefore, according to the present invention, the variation in the output of each pulsed light can be suppressed to a minimum, and the excimer laser device of the present invention is applied to a light source for a reduced projection exposure apparatus for performing reduced projection exposure of a semiconductor. By doing so, it becomes possible to perform a highly accurate exposure process.

[0031]

Embodiments of the present invention will be described below in detail with reference to the accompanying drawings.

FIG. 4 shows a configuration in which the excimer laser device according to the present invention is applied to a stepper light source for performing a reduced projection exposure process of a semiconductor circuit pattern. That is, 1 is a narrow-band excimer laser, and 50 is a stepper as a reduction projection exposure apparatus.

In FIG. 4, the laser chamber 2 of the excimer laser 1 has a discharge electrode and the like (not shown), and the laser chamber 2 is filled with a halogen gas such as F2, a rare gas such as Kr, and a buffer gas such as Ne. The laser gas is excited by the discharge between the discharge electrodes to perform laser pulse oscillation. The emitted pulse light is applied to the band narrowing unit 6 (in this case, the prism beam expander 3,
4, including the grating 5), is returned to the laser chamber 2, is amplified again, and is output as the oscillation laser light L via the partially transmitting mirror 7.
Part of the output light returns to the laser chamber 2 again, and laser oscillation occurs.

A part of the oscillated laser light L is sampled by the beam splitter 8, and then enters the etalon spectroscope 9 through the light diffusion plate 18, where the lens 19
To form a concentric fringe pattern. The etalon spectroscope 9 also receives reference light of a known wavelength in advance, and the CPU 11 compares the reference light formed on the light receiving element 10 with the fringe pattern of the laser light L to determine the wavelength of the output laser light L. And the spectrum width. The CPU 11 outputs the measured wavelength and spectrum width data to the wavelength controller 12. The wavelength controller 12 adjusts and controls the laser oscillation wavelength by changing the angle of the grating 5 based on the input wavelength and spectrum width data, thereby changing the light incident angle on the grating 5 which is a wavelength selection element.

On the other hand, a part of the laser beam transmitted through the beam splitter 8 is further sampled by the beam splitter 13 and is incident on the light receiving element 14. Further, the remaining laser light is transmitted through the slit 31 to the exposure device 50.
Emitted to The slit 31 also functions as a shutter that blocks the output laser light.

The CPU 15 detects the light energy Ei (i = 1, 2, 3,...) Of each pulse oscillation based on the light receiving output of the light receiving element 14 every time the pulse oscillation is performed, and based on the output Ei. To control the laser power supply circuit 16 and the gas supply device 17. The detected pulse energy Ei is stored in a table in the CPU 15 as the energy Pj, i of the i-th pulse in the j-th pulse group.

The laser power supply circuit 16 discharges by giving a potential difference Vi between the discharge electrodes according to the voltage data applied from the CPU 15. In the laser power supply circuit 16, after charging is once performed by a charging circuit (not shown), discharging is performed by operation of a switching element such as a GTO or a thyratron.

FIG. 5 shows various specific examples of the gas supply device 17.

5A to 5D, two gas cylinders 20 and 21 are used.
F2, Kr, and Ne are α: b: c (α = n · a, n>
The other gas cylinder 21 is filled at a molar ratio of 1).
Is filled with Kr and Ne in a molar ratio of b: c.

That is, when the laser gas is injected into the laser chamber 2 (when the gas is initially charged into the vacuum laser chamber, or when the gas is replenished halfway when the output variation σ is outside the allowable range), By injecting a predetermined amount of gas from one of the gas cylinders 20 and 21 into the laser chamber 2, the F 2 gas injected from the gas cylinder 20 is diluted by the gas injected from the other cylinder 21, and as a result, Is an ideal mixture ratio a: b: c. It should be noted that the same effect can be obtained by replenishing gas only from the gas cylinder 20 when replenishing gas in the middle.

If the total pressure of the gas in the laser chamber rises excessively during gas replenishment, the exhaust valve 22 is opened and a part of the gas is exhausted to reduce the total pressure in the laser chamber to a predetermined pressure. They are adjusted so that they can be maintained. Also,
The total gas pressure measuring device 40 measures the total gas pressure in the laser chamber 2 and is used for controlling to maintain the total pressure at a predetermined pressure.

In FIG. 9A, the on-off valve 2
The supply of gas is controlled by 3 and 24, and the gas flow rate is adjusted by adjusting the opening and closing time of the on / off valves 23 and 24.

In FIG. 9B, sub-tanks 25 and 26 are provided in the gas supply path,
On / off valves 27 and 28 are provided on the downstream side of 26.

In FIG. 9C, mass flow controllers (mass flow controllers) 29 and 30 are provided in the gas supply path. This mass flow controller 2
Numerals 9 and 30 control the amount of gas passing so that the mass flow rate becomes a desired constant value. In the case of the configuration of FIG. 9 (c), the gas flow rate can be controlled with high accuracy by setting the flow rates of the mass flow controllers 29, 30 to be constant and adjusting the opening / closing time of the on / off valves 23, 24. become. The gas flow rate may be controlled only by the mass flow controllers 29 and 30 without the on / off valves 23 and 24.

In the configuration of the gas supply device 17, if the distance between the inlet for injecting the laser gas into the laser chamber and the outlet for discharging the laser gas out of the laser chamber is made as far as possible, It is possible to reduce the amount of exhausting the new gas as it is.

As shown in FIG. 19, the laser beam L continuously oscillates for a predetermined number of times at a predetermined cycle, and stops the continuous pulse oscillation for a predetermined time after the continuous oscillation operation. It is output intermittently by a burst mode operation in which a stop operation (oscillation suspension time) is alternately repeated. The following signals are input to the CPU 15 from the exposure device 50: a burst signal BS (see FIG. 6); a laser oscillation synchronization signal (external trigger) TR (see FIG. 6); a target pulse energy value Er laser oscillation synchronization signal TR functions as a trigger signal for each pulse during continuous pulse oscillation in the excimer laser device 1. The burst signal BS functions to start the continuous oscillation operation in the excimer laser device 1 at the rising edge (burst on) and to stop the continuous oscillation operation at the falling edge (burst off). A predetermined time t1 after the first laser oscillation synchronization signal TR is generated, and the last laser oscillation synchronization signal TR
Is set to be burst-off after a predetermined time t2 from the occurrence of.

Based on these input signals, the CPU 15 uses a preset discharge voltage data table in a spike area including the first predetermined number Nf (for example, about 10) of pulses during continuous pulse oscillation. The control is executed, and in the subsequent region, the pulse voltage (feedback control) for controlling the discharge voltage with reference to the discharge voltage of the pulse already generated in the burst cycle is executed without using the discharge voltage data table. The details will be described later. Further, the CPU 15 measures the oscillation suspension time t for each burst cycle. The measured oscillation suspension time is used for the table control.

The exposure device 50 is provided with a beam splitter 52 for sampling a part of the laser light L incident through the slit 51, and the sampling light is incident on an optical monitor module 54 via a lens 53. .
The light monitor module 54 detects the energy Ei 'per pulse of the incident laser light L, and inputs the detected energy value Ei' to the exposure apparatus controller 55. The laser light that has passed through the beam splitter 52 is used for a reduced projection exposure process.

The exposure apparatus controller 55 controls the laser oscillation synchronization signal T in addition to the reduction projection exposure processing, the movement control of the stage on which the wafer is mounted, and the exchange control of the wafer.
R, burst signal BS and target pulse energy value E
An operation such as transmitting r to the excimer laser 1 is performed.

First, the discharge voltage control performed in this embodiment will be described with reference to FIG.

FIG. 7 shows the output change of each laser pulse within one burst period when the discharge voltage is kept constant. As described above, the first pulse is the oscillation pause time immediately before that. As a result, the laser is stabilized, so that its output is the largest, and the second and subsequent pulses repeatedly oscillate under the influence of the immediately preceding laser oscillation, so that the output of each pulse gradually decreases. And in this case the thirteenth
After the first pulse, since the laser is in a steady state, almost the same light energy is output.

Such characteristics of the output laser pulse appear almost in common over all burst periods. Therefore, in this embodiment, one burst period is divided into two control areas A and B based on the pulse number, and the discharge voltage control is performed separately for each of the control areas A and B.

In this case, the control area A has a pulse number of 1 to 1.
The pulse in the spike region 2 is set as a control target, and the control region B
Indicates that the pulses in the plateau region and the steady region after the thirteenth generation are to be controlled.

In the control area A, discharge voltage control is performed based on discharge voltage data stored in a discharge voltage data table (built-in the CPU 15) obtained in advance. That is, in the table control, the pulse energy Ei at the time of laser oscillation is detected based on the discharge voltage data Vi (i = 1, 2,... N) stored in the discharge voltage data table, and the difference ΔEi (= Ei−Er), and calculates a corrected discharge voltage value ΔVi (= G · Δ) according to the difference ΔEi.
EG: gain constant), the discharge voltage data Vi stored in the table is corrected by the corrected discharge voltage value ΔVi, and the corrected discharge voltage data Vi ′ (= Vi)
+ ΔV).

On the other hand, in the control area B, control is performed without using the discharge voltage data table. That is, in the control region D, the past (for example, immediately before) of the pulse
From the relationship between the excitation intensity (discharge voltage) of the pulse oscillation and the pulse energy value, the excitation intensity (charge voltage) required to make the pulse energy value of the pulse coincide with the target value Er is obtained, and the obtained excitation intensity is obtained. The control for causing the pulse oscillation by is performed. This control is referred to as pulse control (feedback control).

FIG. 8 shows the discharge voltage data set and stored in the discharge voltage data table. The horizontal axis represents the pulse number (1 ≦ i ≦ 12), and the vertical axis represents the oscillation pause time (t1, t2). , t
3, t4, ...).

That is, in the discharge voltage data table,
Using the pulse number i and the oscillation pause time t as parameters, each discharge voltage data Vi (tk) (1 ≦ i ≦ 12, 1
≤ k) is stored in advance. For example, V1 (t20) is discharge voltage data having a pulse number of 1 and an oscillation pause time corresponding to t20, and V6 (t50) is discharge voltage data having a pulse number of 6 and an oscillation pause time corresponding to t50.

Next, a method for preparing the discharge voltage data table shown in FIG. 8 in advance will be described.

When preparing such a discharge voltage data table, the shutter (slit 31 in FIG. 4) between the excimer laser 1 and the reduction projection exposure device 50 is closed so that laser light does not enter the reduction projection exposure device 50. After the state
For example, three representative pause times T as shown in FIG.
The excimer laser 1 is actually oscillated using a test operation pattern including 1, T2, and T3 (T1 <T2 <T3). The three representative pause times T1, T2, and T3 are the three oscillation pauses that occur most frequently when the exposure apparatus 50 is operated, and are appropriate according to the exposure apparatus 50 and the wafer pattern that is frequently exposed. Try to set a certain time. In FIG. 8, t1 = T1, t21 = T2,
It is assumed that t51 = T3.

The discharge voltage during this operation is shown in FIG.
As shown in (b), V1 (T1), V2 (T1),... VN (T1), V1 (T1) are sequentially arranged from the top pulse for each of the three pause times T1, T2, and T3.
2), V2 (T2),... VN (T2), V1 (T3), V2 (T3),.
3), and the optical power of each pulse detected as a result of the operation is E1 (T1) for each of the three rest periods T1, T2, and T3.
1), E2 (T1),..., EN (T1), E1 (T2), E2 (T2),.
2), E1 (T3), E2 (T3),... EN (T3). At the time of this test operation, as shown in FIG.
A signal TR 'having the same timing as the laser oscillation synchronization signal TR (see FIG. 6) sent from 0 is generated by the CPU 15 in a pseudo manner, and continuous pulse oscillation is executed in synchronization with the pseudo signal TR'.

At the time of this test operation, Nf pulses (12 pulses belonging to the control area A) for which discharge voltage data is to be created among pulses within one burst period
In each case, laser oscillation is repeated while adjusting the discharge voltage until the following expression (1) is satisfied.

| Ei (T1) −Er | <ΔEa, | Ei (T2) −Er | <ΔEb, and | Ei (T3) −Er | <ΔEc, where 1 ≦ i ≦ k, ΔEa, ΔEb, and ΔEc are Threshold value Er: target value (1) That is, according to the above equation (1), for the first pulse, | E1 (T1) −Er | <ΔEa and | E1 (T2) −Er | < ΔEb and | E1 (T3) −Er |
Laser oscillation is repeated while adjusting the discharge voltages V1 (T1), V1 (T2), and V1 (T3) until <ΔEc is satisfied. In this way,
When V1 (T1), V1 (T2), and V1 (T3) satisfying the above equation are obtained, a discharge voltage data table for the first pulse is created based on these three discharge voltage data.

That is, as shown in FIG. 10, when the discharge voltages V1 (T1), V1 (T2) and V1 (T3) corresponding to the three typical oscillation pause times T1, T2 and T3 are obtained, Connect three points with a straight line, and when t ≦ T1, substitute V = V1 (T1),
When t ≧ T3, V = V1 (T3) is substituted to set the relationship between the oscillation suspension time t and the discharge voltage V. For example, in the table shown in FIG.
1 = T1, t21 = T2, t51 = T3, and by executing the test operation, first, three oscillation suspension times t1 (= T1
1), discharge voltage V corresponding to t21 (= T2), t51 (= T3)
1 (t1), V1 (t21), and V1 (t51) are obtained, and these three data are linearly interpolated to obtain a value between these three points (T1 <t1).
<T2, T2 <t <T3) to obtain discharge voltage data at each point.

As for the second pulse, | E2 (T1) −Er | <ΔEa, | E2 (T2) −Er | <ΔEb, and | E2 (T3) −Er |
Laser oscillation is repeated while adjusting the discharge voltages V2 (T1), V2 (T2), and V2 (T3) until <ΔEc is satisfied.
When 2 (T1), V2 (T2) and V2 (T3) are obtained, a discharge voltage data table for the second pulse is created in the same manner as described above based on these three discharge voltage data.

In the same manner, the same adjustment oscillation is repeatedly executed for the first Nf pulses belonging to the control region A, and the discharge voltage data for these Nf pulses is obtained.

Next, the operation for correcting the discharge voltage data table shown in FIG. 8 will be described.

This correction operation differs between when the actual oscillation pause time t coincides with the three typical oscillation pause times T1, T2 and T3 and when they do not coincide.

First, the oscillation pause time t of the burst cycle
Will be described when T1 does not match T1, T2, and T3.

As described above, the first N of the burst period
With respect to the f pulses, discharge voltage data corresponding to the oscillation pause time t of the burst cycle is read from the table, and laser oscillation is performed with a discharge voltage corresponding to the read discharge voltage data. As a result, it is assumed that Ei (1 ≦ i ≦ Q) is obtained as an actual output of each pulse. next,
For each pulse, a difference ΔEi (= Ei−Er) between the target energy Er and the detected value Ei of each pulse is calculated, and this difference is compared with a predetermined threshold value ΔE.

As a result of this comparison, when a pulse of | ΔEi |> ΔE is generated, the discharge voltage having a pulse number of | ΔEi |> ΔE among the Nf discharge voltage data corresponding to the oscillation pause time t Correction update only data (Vi '
= Vi + ΔEi · G).

For example, in FIG. 8, it is assumed that the current oscillation suspension time is t20, and as a result of the oscillation, the pulse number is 2
Assuming that the relationship of | ΔEi |> ΔE occurs in the pulses of pulses 4 and 4, power supply voltage data V2 (t
Only 20) and V4 (t20) are corrected and updated according to the following equation.

V2 (t20) ′ = V2 (t20) + Gb · (E2−Er) V4 (t20) ′ = V4 (t20) + Gc · (E4−Er) That is, the measured oscillation suspension time t is T1, T2. , T3
Does not coincide with the oscillation pause time t in the discharge voltage data of the pulse in which | ΔEi |> ΔE has occurred.
Is corrected and updated. As a result, the discharge voltage data has a singular point (see point J in FIG. 10) that breaks the linear interpolation relationship shown in FIG.

Next, the measured oscillation pause time t is T1,
The operation at the time of coincidence with any one of T2 and T3 will be described.

In such a case, for the pulse of the pulse number in which | ΔEi |> ΔE has occurred, only the discharge voltage data corresponding to the oscillation stop time (in this case, any of T1, T2, and T3) is used. Instead of correcting and updating, all the discharge voltage data (of the entire oscillation stop time) corresponding to the pulse number are corrected and updated.

For example, when the oscillation suspension time t is T2 (= t21)
It is assumed that the relationship of | ΔEi |> ΔE occurs in the pulse having the pulse number 5 in the oscillation in the case where

In this case, all the discharge voltage data V5 (t1) to V5 (t53) corresponding to the pulse number 5 are corrected and updated according to the following equation.

V5 (t1) ′ = V5 (t1) + G · (E5−Er) :: V5 (t53) ′ = V5 (t53) + G · (E5−Er): The dashed line in FIG. This is shown on a graph.

The above is the details of the discharge voltage data table control performed in the control area A.

Next, the pulse control executed for the pulses included in the control area B shown in FIG. 7 will be described.

In this pulse control, first, the discharge voltage V of the pulse is determined using the pulse energy value Ei of the already output pulse in the current burst cycle and the discharge voltage Vi at that time.

Regarding the pulse number to be referred to, (1) the pulse energy value Ei of the pulse immediately before the pulse and the charging voltage Vi at that time, and (2) the N of the pulse (for example, N =
2, N = 3, etc.) The pulse energy value E of the previous pulse
i, the charging voltage Vi at that time, (3) the average value of the pulse energies Ei to Ei + n of n pulses having a pulse number smaller than the pulse number of the pulse, and the corresponding charging voltages Vi to Vi There is an average value of + n.

Next, using the read pulse energy value Ei and charge voltage Vi, a charge voltage value V for making the pulse energy the target value Er is calculated.

That is, the read data (Ei,
The pulse energy value Ei of Vi) is changed to the target energy value Er.
When Ei = Er, the charging voltage value V = Vi
And when Er> Ei, V = Vi + G · ΔVc (Δ
Vc: a predetermined set value), and if Er <Ei, V
= Vi−G · ΔVc.

The above is the details of each pulse control performed on the pulses in the control area D.

FIG. 11 shows a KrF excimer laser in which the laser power supply voltage V and the gas partial pressure PGA in the laser chamber are kept substantially constant.
It shows the output laser light energy E and the variation (standard deviation) σ of the output laser light energy corresponding to F2. That is, the horizontal axis represents the fluorine gas partial pressure PF2 (proportional to the molar concentration of F2 gas in the laser chamber), and the vertical axis represents the output laser light energy E and the variation (standard deviation) σ of the output laser light energy.

According to FIG. 11, the laser output E is F
(2) It takes the maximum value when the partial pressure is the predetermined value Pe (the efficiency is maximum), and monotonically increases at a partial pressure lower than this partial pressure value Pe,
At a partial pressure higher than this partial pressure value Pe, it decreases monotonically.

On the other hand, the standard deviation σ of the laser output takes the minimum value when the F2 partial pressure is at the predetermined value Pf, and monotonically decreases when the partial pressure is lower than the partial pressure value Pf, and is higher than the partial pressure value Pf. The partial pressure increases monotonically.

In the characteristics shown in FIG. 11, the phenomenon noticed by the present inventor is that Pe ≠ Pf. In the present invention, the output variation σ is minimized even if the laser output (oscillation efficiency) is somewhat sacrificed. The F2 gas supply control is performed by setting the fluorine partial pressure value Pf to be the highest priority target value.

In FIG. 11, σc is an allowable limit value (allowable upper limit value) of the output variation, and the F2 partial pressure corresponding to the allowable limit value σc has two values, PMIN and PMAX.

Next, the output variation σ of the excimer laser
Are the laser power supply voltage V, the total gas pressure PGA in the laser chamber, and the halogen gas partial pressure PF2 in the laser chamber.
Change in response to a change in such a parameter.

FIG. 12 shows the fluorine partial pressure value Pf and the total gas pressure PGA which minimize the variation σ when the power supply voltage V is set to three different values V1, V2, V3 (V1>V2> V3).
According to FIG. 12, the fluorine partial pressure value Pf which minimizes the variation σ even when the total pressure value PGA increases until the gas total pressure PGA reaches the predetermined value PG1 is approximately When PGA> PG1, the fluorine partial pressure value Pf shows a decreasing tendency in accordance with an increase in the total pressure PGA. This property is common across the power supply voltage values V1, V2, V3. Further, according to FIG. 12, the fluorine partial pressure value Pf that minimizes the variation σ increases in accordance with the increase in the power supply voltage V.

FIG. 13 shows three types of power supply voltages V.
The graph shows output variation characteristics when 1, V2, and V3 (V1>V2> V3). According to FIG. 13, when the power supply voltage V increases, the fluorine partial pressure value Pf that minimizes the variation σ increases (Pf3 <Pf2 <Pf1),
Further, it can be seen that the variation value σ itself decreases as the power supply voltage V increases (σ1 <σ2 <σ3). It can also be seen that the difference in the Pf value itself corresponding to the same power supply voltage difference is smaller in the region where the power supply voltage is high than in the region where the power supply voltage is low (Pf3−Pf2 <Pf2−Pf1).

FIG. 14 shows that the total gas pressure PGA is calculated using five different values P1, P2, P3, P4, P5 (P1>P2>P3>P4> P4).
5 shows the output variation characteristics in the case of 5). According to FIG. 14, as the total pressure PGA increases, the fluorine partial pressure value Pf that minimizes the variation σ decreases (Pfa <Pfb <Pfc), and the variation value σ itself increases. (Σa>σb> σc). Also, when the total pressure PGA becomes a certain value or less (in this case, PGA <P3),
It has also been found that the variation value σ and the fluorine partial pressure value Pf that minimizes the variation σ are almost the same and do not change.

As described above, according to FIGS.
In order to keep the variation σ as small and constant as possible, the power supply voltage V should be large, and the total gas pressure P
It turns out that smaller GA is better.

Next, a method of obtaining the output variation (standard deviation) σ will be described with reference to FIGS.

As described above, in the stepper type semiconductor exposure apparatus, the exposure and the stage movement are alternately repeated, so that the operating state of the excimer laser, which is the light source of the exposure apparatus, is necessarily A burst mode as shown in FIG.

FIG. 15 is an enlarged view of the pulse train in one burst period shown in FIG. The energy of each pulse light is Ei (i = 1, 2,...), And the number of pulses in one set for obtaining the variation σ is Ns.

In this case, a standardized value ε obtained by dividing the standard deviation σ as the variation data by the average value EA of the pulse output is used.
(= 3 · σ / EA) is used. That is, the standard deviation σ and the output average value EA are set for each set including the Ns pulses.
Is calculated, and the calculated standard deviation σ and output average value EA are calculated.
Is used to calculate the variation data ε.

The standard deviation σ is obtained as follows.

First, the integrated value ET of the light energy of Ns pulses is obtained according to the following equation. Note that Σ (i = 1, Ns)
Is a symbol meaning that integration is performed from i = 1 to i = Ns.

ET = Σ (i = 1, Ns) Ei = E1 + E2 + E3 +... + ENs Next, the average value EA of these Ns pulsed light outputs is calculated according to the following equation.

EA = ET / Ns Next, using the average value EA obtained above, the standard deviation σ of these Ns pulses is obtained according to the following equation (2).

[0103] (2) As described above, in the case of the stepper method, 1 to Ns, Ns + 1
Σ is determined for each set of ２2Ns, 2Ns + 1、２3Ns.

Next, using the standard deviation σ and the output average value EA obtained above, an output variation ε is obtained according to the following equation.

Ε = 3 · σ / EA (3) Next, a method of obtaining the standard deviation σ by the step & scan method will be described.

That is, in the stepper system, exposure is performed while the stage is stopped, whereas in the step-and-scan system, exposure is performed while moving the stage, which has the advantage that a large area can be exposed. .

That is, in this step & scan method, a single pulse laser (an elongated sheet called a sheet beam) is applied so that a predetermined number N0 of pulse lasers are set at all points on the IC chip TP. Processing is performed while shifting the irradiation area of the pulse laser beam on the workpiece by a predetermined pitch every time a beam having a rectangular cross-sectional shape is incident. That is, as shown in FIG. 16, the irradiation area of each sheet beam (the area indicated by E1, E2, E3,...) Is smaller than the area of the IC chip TP, and these pulsed laser beams are sequentially arranged at a predetermined pitch ΔP Are scanned while being superimposed, and a predetermined number N0 of sheet beams are incident on each point, and the entire surface of the IC chip TP is exposed.

For example, in FIG. 16, N0 = 4, point A is exposed by the integrated energy of four pulsed laser beams E1, E2, E3 and E4, and point B is exposed to four pulsed laser beams E2, E2. The exposure is performed by the integrated energy of E3, E4 and E5. The following points C are similarly exposed by the integrated energy of the four pulsed laser beams.

Therefore, when the standard deviation σ is obtained by such a step-and-scan method, the number of pulses Ns = No in one set when the standard deviation σ is obtained, and the above (2)
The standard deviation may be obtained using an equation. In addition,
Variations may be obtained by using the total number of sheet beams irradiated to one IC chip as one set.

The operation of the first embodiment of the present invention will be described below with reference to the flowcharts of FIGS.

In the first embodiment, all the pulse oscillations within one burst period are set as objects for obtaining the variation σ, and the halogen is set so that the variation σ thus obtained becomes a value near the minimum value σmin. The gas supply is controlled. Further, as described above, the power supply voltage Vi is controlled by the table control in the control area A including the first about 10 pulses, and the power supply voltage Vi is controlled by the pulse control in the subsequent control area B.

Here, in order to control the output variation σ to be always smaller than σc, the F2 partial pressure value PF2 must be equal to PM.
It is necessary to control so as to be between IN and PMAX.

However, since the output variation σ is not linear as shown in FIG. 11, when the output variation value σ becomes close to σc during the above control, this state indicates that the fluorine partial pressure is PMIN. Without determining which state is closer to PMAX or PMAX, it cannot be determined whether or not to supply F2 gas. That is, when the partial pressure of F2 is smaller than PMIN, it is necessary to supply the F2 gas. When the partial pressure of F2 is larger than PMAX, there is no need to supply the F2 gas.

Therefore, if attention is paid to the fact that the amount of F2 gas decreases with the progress of laser pulse oscillation, the above-mentioned problem can be solved by performing the following control.

That is, after the laser gas is replenished into the laser chamber or after all the laser gas in the laser chamber is replaced with a new laser gas, the F2 gas is decreasing as the laser pulse oscillation progresses. If the output variation σ is monitored after the gas replenishment or the gas exchange, it is possible to determine which position of the σ curve in FIG.

For example, at the time of gas replenishment or gas exchange,
If the partial pressure of F2 in the laser chamber is set to a value between PMAX and Pf (for example, a partial pressure slightly lower than PMAX), the halogen gas decreases as the number of laser oscillations increases. Correspondingly, the variation value σ moves along the arrow F on the σ curve in FIG. That is, since σ continues to decrease from the value corresponding to the partial pressure value slightly lower than PMAX and reaches the minimum value σMIN and then starts increasing, the F2 gas is supplied when σ = σc is reached thereafter. By doing so, the output variation value σ can be controlled to σc or less.

The details of the above control will be described below with reference to FIG.

First, the operator determines the number Nf of pulses included in the control area A, the number N of pulses in a burst cycle, the number Ns of data for obtaining variation, the target value Er of pulse energy, and the allowable limit value of the variation value. St is set to an appropriate value and stored (step 100). With this setting, the CPU 15 calculates the count value ET of the integrating counter that sequentially integrates the output energy Ei of each oscillation pulse,
A counter value k for counting the number of pulses for obtaining the variation σ is initialized to 0 (step 100).

The variation limit allowable limit value St is a threshold value for comparison with the output variation ε, and is a value corresponding to σc in FIG. St is set to a predetermined numerical value, for example, 7%.

Next, the operator sets the total gas pressure PGA and the halogen gas partial pressure Pf that can keep the output variation ε as small as possible from the relationships shown in FIGS. 4 to 8 (step 110). Then, the laser gas is initially charged into the laser chamber (to fill the vacuum laser chamber after passivation processing with a new gas) so that the total gas pressure PGA and the halogen gas partial pressure Pf set as described above are obtained.

Next, the CPU 15 initializes the counter value i of the counter for counting the current pulse oscillation number in the burst cycle to 0, and measures the oscillation pause time t in the burst cycle (step 110). Also, C
The PU 15 initializes a flag FLAG, which is referred to to determine whether to execute the gas supply subroutine, to 0 (Step 120).

Next, after incrementing the pulse counter value i by 1 (step 130), the CPU 15 compares the pulse count value i with a predetermined value N + 1 to determine whether or not the burst cycle has ended (step 140). If not, the pulse count value i is then compared with a predetermined value Nf to determine whether to perform the above-described table control or to perform each pulse control (step 15).
0).

When i ≦ Nf, the CPU 15 reads the power supply voltage data Vi corresponding to the measured oscillation suspension time t and the pulse number i from the power supply voltage data table shown in FIG. The power supply voltage value Vi is output to the laser power supply circuit 16 (step 160). As a result, pulse oscillation by the power supply voltage Vi is performed by the laser power supply circuit 16 (step 180). The output energy Ei of the pulse oscillation is monitored by the light receiving element 14 and input to the CPU 15. The CPU 15 stores this monitor value Ei.

In this embodiment, the above-mentioned correction of the table data is performed during the oscillation suspension time. That is, during the oscillation suspension time, the CPU 15 compares the monitor value Ei of each pulse oscillation of the previous burst cycle with the target value Er (Δ
If Ei = Ei−Er), | ΔEi |> ΔE, the table data corresponding to the pulse number i is corrected. In addition,
At the time of this correction, the measured oscillation suspension time t is equal to the T1, T
If it matches any of T2 and T3, the pulse number i
Are corrected and updated, and if they do not match, only the table data corresponding to the oscillation suspension time t in the table data corresponding to the pulse number i is corrected. Of course, this table correction may be performed during pulse oscillation.

Next, after incrementing the counter value k by 1 (step 200), the CPU 15 adds the measured output energy Ei to the previous pulse energy integrated value ET,
The integrated counter value ET is updated with the addition result ET + Ei (step 210). Next, the CPU 15 determines whether or not the pulse count value i matches the set value Ns (step 170), and if not, repeats the procedures of steps 130 to 210 until they match.

On the other hand, in the determination at step 150, i> Nf
In this case, the CPU 15 executes the above-described pulse control (step 170).

That is, in this case, the power supply voltage is controlled by the following equation using the immediately preceding oscillation history.

Vi = Vi-1 + G * (Er-Ei-1) Vi: power supply voltage of the current pulse Vi-1: power supply voltage of the immediately preceding pulse G: gain Er: target value of pulse light energy Ei-1: immediately before After that, the pulse oscillation operation proceeds and the pulse count value i
Is equal to Ns, the CPU 15 calculates the standard deviation σ of these Ns oscillation pulses according to the above equation (2), and calculates the average value EA of the outputs of the Ns oscillation pulses.
(= ET / Ns) and standardized output variation ε (= 3 · σ / EA) is obtained by dividing the calculated standard deviation σ by the output average value EA (steps 230 and 24).
0).

Next, the CPU 15 compares the calculated output variation ε with the set allowable limit value St. As a result of this comparison, if the output variation ε falls within the range of the set value St (ε ≦ St), it is not necessary to supply the halogen gas.
Is set to 0, and the procedure of steps 130 to 210 is repeated.

However, in the determination at step 250,
If ε> St holds, it is determined whether or not the flag FLAG = −1. If the flag FLAG is -1, it is determined that the F2 partial pressure is equal to or higher than PMAX, and the procedure shifts to step 130 without performing halogen gas supply. That is, when the flag FLAG = −1, F
2 By continuing pulse oscillation without replenishing gas,
The F2 gas is naturally reduced along the arrow Q in FIG. 11 (the laser oscillation causes the F2 gas to react with a material such as a laser electrode to be converted into a fluoride, and the F2 gas itself decreases), and the output fluctuation due to the natural decrease of the F2 gas. ε is made smaller than the set value St.

In step 250, the flag FLAG = -1
If halogen gas replenishment is performed when
Since the F2 partial pressure increases, the output variation ε further increases along the arrow R in FIG.

Next, the flag FLAG
If not, the flag FLAG is set in the next step 260.
= 1 is set, and if the flag FLAG is not 1, the flag FLAG is set to 1 (step 300).
The gas supply subroutine shown in FIG. 2 is executed (step 310). This gas supply subroutine will be described later.

On the other hand, if the flag FLAG is 1 in step 270, the variance εk-1 of the previous set calculated before is compared with the variance εk of the current set calculated this time, and εk-1> ε
In the case of k, it is determined that the variation ε along the arrow S in FIG. 11 has been reduced by the previous gas replenishment, and the procedure shifts to step 310 to execute the gas replenishment subroutine, thereby further replenishing the gas. Is executed (step 280).

However, at step 280, εk-1
If ≦ εk is satisfied, it is determined that the variation ε along the arrow R in FIG. 11 has increased, and after setting the flag FLAG = −1 (step 290), without replenishing the halogen gas, The procedure moves to step 130. That is, in this case, the pulse oscillation is continued without replenishing the F2 gas, whereby the F2 gas is naturally reduced along the arrow Q in FIG. 11, and the variation ε is reduced from the set value St by the natural reduction of the F2 gas. Make it smaller.

Thus, according to the control procedure shown in FIG.
When ε ≦ St (when the determination in step 250 is NO), the F2 gas is not supplied.

When it is determined that ε> St and PF2 ≧ PMAX (the determination in step 260 is YE
At the time of S or when the determination at step 280 is NO),
Wait for the natural decrease of F2 gas without supplying F2 gas.

However, when ε ≧ St and PF2 <PMIN
Is determined (the determination in step 270 is N
At the time of O or when the determination at step 280 is YES), the supply control of the F2 gas is executed until ε <St.

In the control procedure shown in FIG. 1, when ε exceeds the maximum allowable value St, an abnormal signal is output, the abnormal signal is transmitted to the semiconductor exposure apparatus, and the emission side of the excimer laser is output. May be closed to close the exposure process. In this case, after that, when ε <St, the shutter may be opened to restart the exposure processing.

Next, the gas supply subroutine will be described with reference to FIG.

This gas replenishment subroutine includes two methods, a method of not exhausting gas and a method of exhausting gas to maintain the total pressure within a predetermined range.

That is, when the F2 partial pressure of one of the gas cylinders 20 of the gas replenishing device 17 shown in FIG. 5 is increased to about 5%, as described above, both gas cylinders 20,
The gas supply is performed by using the gas supply 21. In this case, since the gas supply amount is small, the exhaust step can be omitted. On the other hand, when the F2 partial pressure of the gas cylinder 20 is set to be as low as about 1%, a large amount of gas is supplied using only the gas cylinder 20, and the increase in the total pressure is reduced by exhaust.

The gas replenishment subroutine shown in FIG. 2A is a case in which no gas is exhausted. When the gas replenishment subroutine is started, the halogen gas is supplied using the two gas cylinders 20 and 21. That is, the halogen gas partial pressure in the laser chamber is a low value of 0.3% or less, and the gas supply amount in this case is small. Therefore, the gas supply amount is offset by the halogen gas consumption amount, and the total gas pressure is reduced. The gas exhaust process is omitted because it is considered to be substantially constant. However, even in such a case, when the number of gas replenishments increases, the total gas pressure rises as the gas exceeds the pressure P3 in FIG. 14 and reaches the pressure P2. (Σc
→ σb), even if the halogen gas is supplied, the variation σ does not decrease, but increases. That is, the range of the value of σ is not affected by the total pressure PGA (P in FIG. 14).
If the halogen gas replenishment control is performed at F2 ≦ P3), the value of σ can be reduced by replenishing the halogen gas. However, if the replenishment of the halogen gas is repeated, the total pressure rises and finally It has the characteristic of the pressure P2 in FIG.

In this embodiment, since the variation σ (or ε) is monitored, if the phenomenon that σ (or ε) rises when the halogen gas is replenished is determined, the above total pressure can be determined. Can be grasped. Therefore, when such a phenomenon of increase in the total pressure occurs, if the laser oscillation is continued and the halogen gas partial pressure inside the laser chamber is continuously reduced, the value of σ decreases along the arrow U in FIG. Eventually, it will fall into a range smaller than σc, and at that time, the above-described control for replenishing the halogen gas according to the value of σ may be started again.

Next, the gas supply subroutine shown in FIG. 2 (b) is a case where a gas exhaust step is performed. When the gas supply subroutine is started, first, a part of the gas in the laser chamber is exhausted (step 400). That is, by performing gas exhaustion before gas replenishment, a gas containing laser-oscillated impurities is exhausted. Next, F2, Kr, Ne are supplied by the gas replenishing device 17 shown in FIG.
The halogen gas F2 is supplied into the laser chamber 2 by supplying a predetermined amount of the mixed gas into the laser chamber 2 (step 410). After the replenishment, the total pressure PGA in the laser chamber is measured by the total pressure measuring sensor 40 (step 420), and when the measured value becomes larger than the predetermined set pressure Pga1 (step 43).
0), and the gas in the laser chamber 2 is exhausted (step 440).

That is, if the total gas pressure in the laser chamber rises too much, as can be seen from FIG. 14, the variation value σ itself also rises, and σ may exceed the allowable upper limit σc. If the total gas pressure in the laser chamber rises too much, the fluorine partial pressure PF2 will decrease, and the laser oscillation efficiency (output laser light energy E) will drop extremely (see FIG. 11). Therefore, FIG.
According to the gas replenishment routine described above, in order to eliminate the above-mentioned phenomenon, a part of the gas is exhausted from the laser chamber, and the total gas pressure is always kept within a predetermined set pressure Pga1 (for example, FIG. 1).
4 (about P3).

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

In the second embodiment, a step 195 is added between step 190 and step 200 in the flowchart of FIG. 1, and otherwise, FIG.
The procedure is exactly the same.

That is, in the second embodiment, with respect to the pulse oscillation of the control region A in which the pulse number i is included in the range from 1 to Nf, the variation value ε is not calculated and the target is not to be supplied with the halogen gas. The variation value ε is calculated only for the pulse oscillation included in the control region B in which the pulse number i is larger than Nf, and is set as a target to be supplied with the halogen gas. This is because in the spike region of the control region A, although the spike killer control based on the table data is performed, the output is not stable and the variation value itself is large as compared with the control region B. The control area A is ignored.

Incidentally, the following implementation is also possible as the halogen gas supply control.

That is, while monitoring the output variation ε, the halogen gas is continuously (continuously) supplied little by little into the laser chamber without interruption so that the value of ε does not greatly deviate from the minimum value. Also in this control, the total pressure continues to increase, and eventually exceeds the value P3 in FIG. 14, so that the supply of the halogen gas causes only the increase in ε. Therefore, in this case, when the above state is detected, a part of the gas is exhausted from the inside of the laser chamber to lower the total pressure, and preferably the total pressure is reduced to a value P3 or less in FIG. What should I do?

Alternatively, when the halogen gas is supplied, the gas is exhausted at the same time, so that the total gas pressure in the laser chamber can hardly be changed.

In the voltage data table shown in FIG. 8, the voltage data is stored in correspondence with the pulse number i and the oscillation suspension time t.
As shown in (1), the elapsed time Td from the start of oscillation may be used instead of the pulse number i.

That is, in the case of the voltage data table using the pulse number i as a parameter, the stored voltage data is adapted to these frequencies in consideration of the frequency of the laser oscillation synchronization signal TR to be used or the maximum repetition frequency of the laser device. Is set to a value that Therefore, when laser pulse oscillation is performed at another repetition frequency, the spiking prevention control cannot be accurately performed unless a voltage data table that stores voltage data suitable for the repetition frequency is used.

As described above, in the case of the voltage data table in which the voltage data is stored corresponding to the pulse number i, when trying to correspond to various repetition frequencies, each voltage corresponding to these various frequencies is different. It is necessary to store data individually, which causes an increase in cost in terms of memory.

Therefore, in the table of FIG. 17, voltage data is stored corresponding to the elapsed time Td from the start of oscillation and the oscillation pause time t.

That is, as shown in FIG. 17, the horizontal axis of the voltage data table sets the elapsed time Td from the start of oscillation at a predetermined time interval tg, and (0, 1tg) for each of these elapsed times. , 2tg,...), Power supply voltage data for preventing spiking are stored separately. In the same manner as described above, the oscillation quiescent time t is set on the vertical axis, and different power supply voltage data is set for each oscillation quiescent time.

In this voltage data table, at the time of the first pulse oscillation, the voltage data corresponding to Td = 0 is used. During pulse oscillation, a time interval from the reception of the oscillation synchronization signal TR to the reception of the next oscillation synchronization signal is measured (the basic unit of time measurement operation is set to tg). The required voltage data is read from the voltage data table based on the time measurement result. For example, assuming that the time interval between the reception of the first TR signal and the reception of the next TR signal is 3tg, the voltage data corresponding to Td = 3tg is used. Similarly, the time interval between the reception of the second TR signal and the reception of the next third TR signal is 3tg.
In this case, the voltage data corresponding to Td = 6tg is used.

In the case of the voltage data table shown in FIG. 17, data correction as performed in the voltage data table shown in FIG. 8 may be performed. That is, when the measured oscillation pause time t matches any of the representative pause times T1, T2, and T3 shown in FIG. 10, the pulse whose output deviation is larger than the predetermined set value ΔE is used. Corrects and updates all the discharge voltage data (of the entire oscillation stop time) corresponding to the elapsed time Td.
When the measured oscillation stop time t does not coincide with the representative pause times T1, T2, T3, only the discharge voltage data corresponding to the elapsed time Td and corresponding to the oscillation stop time is corrected and updated. When using the voltage data table shown in FIG. 17 to perform table control only in the spike region (control region A), the power supply voltage for only the pulse for the first predetermined time in one burst cycle is stored in the table. What is necessary is just to memorize a value.

In the above embodiment, the value ε obtained by dividing the standard deviation σ of Ns oscillation pulses by the average value EA (= ET / Ns) of Ns oscillation pulses is used as the output variation value. However, the standard deviation σ may be used as the output variation.

In the above-described embodiment, the pulse control is performed in the control area B. However, in the control area B, the discharge voltage control using the discharge voltage data table is performed similarly to the control area A. Is also good.

In the above embodiment, when the table is corrected, the measured pause time t is set to the representative pause time T1, T1.
2. All the table data corresponding to the pulse number is corrected when it matches T3. However, even when the measured pause time t does not match the representative pause times T1, T2, and T3, all the table data are corrected. May be corrected. Further, when the measured pause time t matches the representative pause time T1, T2, T3, only the table data corresponding to the representative pause time may be corrected and updated.

Further, in the above embodiment, as shown in FIG. 10, when the oscillation pause time t is smaller than the representative pause time T1, the voltage data corresponding to the representative pause time T1 is substituted. In such a case, the voltage data used in the last pulse of the immediately preceding burst cycle may be used for the first pulse oscillation at the head.

In the above embodiment, when the average value EA of the pulse light power falls below a predetermined value, the output is increased by supplying a mixed gas of a rare gas and a buffer gas into the laser chamber. You may do so.
That is, since the total pressure PGA and the average value EA of the pulsed light power have a positive correlation, a decrease in EA is suppressed by supplying a mixed gas of a rare gas and a buffer gas to increase the total pressure PGA. To At this time, by supplying a mixed gas of a rare gas and a buffer gas from the gas cylinder 21 shown in FIG. 5, gas can be supplied without changing the ratio of the rare gas and the buffer gas in the laser chamber.

[Brief description of the drawings]

FIG. 1 is a flowchart showing a first embodiment of the present invention.

FIG. 2 is a flowchart showing a gas supply subroutine.

FIG. 3 is a flowchart showing a second embodiment of the present invention.

FIG. 4 is a block diagram illustrating a configuration example of an excimer laser device.

FIG. 5 is a block diagram showing various configuration examples of a gas supply device.

FIG. 6 is a time chart of a burst signal and a laser oscillation synchronization signal.

FIG. 7 is a diagram showing a divided control area within one burst period.

FIG. 8 is a diagram showing stored contents of a power supply voltage data table.

FIG. 9 is a time chart showing an operation state when a power supply voltage data table is created.

FIG. 10 is a diagram showing a method of creating power supply voltage data stored in a power supply voltage data table.

FIG. 11 is a graph showing a relationship between output energy and output energy with respect to fluorine partial pressure.

FIG. 12 is a graph showing a relationship between a total gas pressure and a fluorine partial pressure value that minimizes output variation.

FIG. 13 is a graph showing the relationship between the partial pressure of fluorine and output variation using the power supply voltage as a parameter.

FIG. 14 is a graph showing the relationship between the partial pressure of fluorine and output variation using the total gas pressure as a parameter.

FIG. 15 is a view for explaining a method of obtaining a variation value within one burst period.

FIG. 16 is a diagram illustrating a step-and-scan method.

FIG. 17 is a diagram showing stored contents of another power supply voltage data table.

FIG. 18 is a plan view showing an IC chip on a wafer.

FIG. 19 is a diagram showing a pulse energy waveform in burst operation when the charging voltage is kept constant.

FIG. 20 is an enlarged view showing a pulse energy waveform in one burst cycle.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Excimer laser apparatus 2 ... Laser chamber 3, 4 ... Prism beam expander 5 ... Grating 6 ... Band narrowing unit 7 ... Partial transmission mirror 8, 13 ... Beam splitter 9 ... Etalon spectroscope 10, 14 ... Light receiving element 11, 15 CPU 12 Wavelength controller 17 Gas replenishing device 20, 21 Gas cylinder 23, 24 On-off valve 25, 26 Sub tank 29, 30 Mass flow controller 50 Exposure device

Claims (38)

[Claims] A laser gas containing a halogen gas is sealed in a laser chamber, and a pulse discharge is performed in the laser chamber to excite the laser gas to perform pulsed laser oscillation. A burst mode operation is repeatedly performed in which a continuous oscillation operation in which pulse oscillation is performed and a stop operation in which the pulse oscillation is paused for a predetermined oscillation suspension time are alternately performed as one burst cycle. An excimer laser device that performs control so that the pulse oscillation falls within a predetermined target value range, wherein a pulse number indicating a pulse order within one burst cycle and a plurality of different oscillation pause times are used as parameters to output the respective pulse oscillations. The initial value of the power supply voltage whose variation falls within a predetermined allowable value range is stored in advance. Voltage data table means, oscillation pause time measuring means for measuring the oscillation pause time for each burst cycle, and a pulse corresponding to the measured oscillation pause time from the power supply voltage data table means for each burst cycle. Oscillation control means for reading a power supply voltage value corresponding to the number and performing pulse oscillation according to the read power supply voltage value; monitoring means for sequentially monitoring the output of each pulse of continuous oscillation in association with the pulse number; Gas supply means for replenishing the laser gas into the laser chamber; variation calculation means for determining a variation in output energy of each pulsed laser oscillation light based on a monitor output of the monitoring means; and a range in which the calculated variation is a predetermined target value range. Gas replenishment control means for controlling the gas replenishment means so as to enter thereinto replenish the halogen gas, Excimer laser device equipped with 2. A method according to claim 1, wherein one burst period is divided into a first region corresponding to a first-half spike region and a second region belonging to a second half, and said variation calculating means varies a pulse included in said first region. The excimer laser device according to claim 1, wherein the excimer laser device is excluded from a target for which is obtained. 3. The power supply voltage data table obtains, in advance, an initial value of the power supply voltage corresponding to a representative plurality of oscillation pause times for each pulse number, and obtains an initial value of the plurality of power supply voltages in a straight line for each pulse number. 2. The excimer laser device according to claim 1, wherein the excimer laser device is created by interpolation. 4. The power supply voltage data table according to claim 1, wherein said power supply voltage data relating to an oscillation pause time larger than a largest value among said representative plurality of oscillation pause times is stored in said power supply voltage data table. 4. The excimer laser device according to claim 3, wherein the power supply voltage data corresponding to the largest value is substituted. 5. The power supply voltage data table according to claim 1, wherein said power supply voltage data relating to an oscillation pause time smaller than a minimum value of said representative plurality of oscillation pause times is stored in said power supply voltage data table. 4. The excimer laser device according to claim 3, wherein the power supply voltage data corresponding to the smallest value is substituted. 6. When the measured oscillation pause time is smaller than the smallest value among the representative plurality of oscillation pause times, the last pulse of the immediately preceding burst cycle is determined for the first pulse oscillation. 4. The excimer laser device according to claim 3, wherein pulse oscillation is performed using a power supply voltage value used for oscillation as a power supply voltage. 7. A difference between an output value of each pulse monitored by said monitor means and a target value of said output energy is determined for each pulse, and for a pulse whose difference exceeds an allowable limit, the pulse of the pulse is determined. 2. The excimer laser device according to claim 1, further comprising table correction means for correcting and updating a storage power supply voltage value of said voltage data table means corresponding to a number and said measured oscillation pause time using said difference. 8. The table correction means, when the oscillation suspension time measured by the oscillation suspension time measuring means coincides with any of the plurality of representative oscillation suspension times, each table monitored by the monitor means. For a pulse in which the difference between the output value of the pulse and the target value exceeds the allowable limit, all the storage power supply voltage values of the voltage data table means corresponding to the pulse number of the pulse are respectively corrected according to the difference. The excimer laser device according to claim 7, wherein the excimer laser device is updated. 9. The power supply voltage data table means includes:
The power supply voltage value for only a first predetermined number of pulses in a burst period is stored, and pulse oscillation control by the oscillation control means is performed only for the first predetermined number of pulses. 2. The excimer laser device according to 1. 10. The power supply voltage data table means includes:
A power supply voltage value is stored only for the first predetermined number of pulses in one burst period, and the pulse oscillation control by the oscillation control means and the storage power supply data correction control by the table correction means are executed by the first predetermined number of pulses. 8. The method according to claim 7, wherein the processing is performed only for the pulses of
The excimer laser device as described in the above. 11. With respect to pulses generated after the first predetermined number of pulses, a monitor value of at least one pulse already output within a current burst cycle obtained from the monitor means and a power supply voltage at that time 11. The excimer laser device according to claim 9, wherein a power supply voltage value at the time of the current pulse oscillation is calculated based on the value, and pulse oscillation is performed based on the power supply voltage value. 12. The excimer laser device according to claim 1, wherein said variation measuring means includes variation calculation means for calculating output variation of each pulse oscillation light based on a monitor output of said monitor means. 13. The variation calculation means calculates a standard deviation and an average value of output energies of a predetermined number of pulsed light beams, and sets a value obtained by dividing the standard deviation by an average value as an output variation value. 13. The excimer laser device according to item 12. 14. A variation in output energy of each pulsed laser oscillation light takes a predetermined minimum value corresponding to a partial pressure of a halogen gas in the laser chamber, and a limit value of a target value range for the variation. Has two values, a first value smaller than the halogen gas partial pressure value corresponding to the minimum value, and a second value larger than the halogen gas partial pressure value corresponding to the minimum value. When the calculated output variation is out of the target value range, the control unit determines whether the current halogen gas partial pressure is smaller than the first value or the current halogen gas partial pressure is smaller than the first value. 2. The excimer laser device according to claim 1, wherein it is determined whether the partial pressure value is larger than the second value, and only when the former is the former, the gas controller is controlled to supply the halogen gas. 15. An exhaust unit for exhausting a laser gas in the laser chamber, wherein the control unit causes the exhaust unit to perform an exhaust operation before the halogen gas is supplied by the gas supply unit. The excimer laser device as described in the above. 16. The apparatus according to claim 1, further comprising an exhaust unit for exhausting the laser gas in the laser chamber, wherein the control unit causes the exhaust unit to perform an exhaust operation while the halogen gas is supplied by the gas supply unit. The excimer laser device as described in the above. 17. A gas total pressure measuring means for measuring a total pressure of a laser gas in the laser chamber, wherein the control means controls a total gas pressure in the laser chamber based on a measured value of the gas total pressure measuring means. 17. The excimer laser device according to claim 15 or 16, wherein gas supply by said gas supply means and gas exhaustion by said exhaust means are controlled such that does not exceed a predetermined upper limit. 18. The gas replenishing means for supplying a buffer gas and a rare gas, and when the average value of the output energy of the pulse light falls below a predetermined value, the gas replenishing means enters the laser chamber. The excimer laser device according to claim 1, wherein a mixed gas of a buffer gas and a rare gas is supplied. 19. The excimer laser device according to claim 1, wherein said control means outputs an abnormal signal when said calculated variation falls outside a predetermined target value range. 20. A laser gas containing a halogen gas is sealed in a laser chamber, and a pulse discharge is performed in the laser chamber to excite the laser gas to perform pulsed laser oscillation. A burst mode operation is repeatedly performed in which a continuous oscillation operation in which pulse oscillation is performed and a stop operation in which the pulse oscillation is paused for a predetermined oscillation suspension time are alternately performed as one burst cycle. An excimer laser device that executes control such that the value falls within a predetermined target value range. The variation in the output of each pulse oscillation falls within a predetermined allowable value range, using the time elapsed since the start of oscillation and the oscillation pause time as parameters. Voltage data table means for storing in advance respective initial values of the input power supply voltage; Oscillation pause time measurement means for measuring time for each burst cycle; clock means for measuring the time elapsed from the start of the pulse oscillation for each pulse oscillation; and power supply voltage data for each burst cycle. Oscillation control means for reading a power supply voltage value corresponding to the measured oscillation suspension time and corresponding to the elapsed time from the measured oscillation start from the table means, and performing pulse oscillation according to the read power supply voltage value; Monitoring means for sequentially monitoring the output of each pulse of oscillation; gas replenishing means for replenishing the laser gas into the laser chamber; and variation for obtaining variation in output energy of each pulsed laser oscillation light based on the monitor output of the monitoring means. Calculating means, controlling the gas replenishing means so that the calculated variation falls within a predetermined target value range. An excimer laser device, comprising: gas supply control means for replenishing a halogen gas by means of: 21. A burst period is divided into a first region corresponding to a first-half spike region and a second region belonging to the second half, and the variation calculating means varies a pulse included in the first region. 21. The excimer laser device according to claim 20, wherein the excimer laser device is excluded from an object for which is obtained. 22. The power supply voltage data table, wherein an initial value of the power supply voltage corresponding to a plurality of typical oscillation pause times is obtained in advance for each of a plurality of elapsed times from the start of different oscillations,
21. The excimer laser device according to claim 20, wherein the excimer laser device is created by linearly interpolating the initial values of the plurality of power supply voltages for each elapsed time from the start of oscillation. 23. The power supply voltage data table, wherein the power supply voltage data relating to an oscillation pause time larger than the largest value among the representative plurality of oscillation pause times is stored in the power supply voltage data table. The power supply voltage data corresponding to the largest value is substituted.
Excimer laser device mounted on 2. 24. The power supply voltage data table, wherein the power supply voltage data relating to an oscillation suspension time smaller than a minimum value of the representative plurality of oscillation suspension times is stored in the power supply voltage data table. The power supply voltage data corresponding to the smallest value is substituted.
Excimer laser device mounted on 2. 25. When the measured oscillation pause time is smaller than the smallest value among the representative plurality of oscillation pause times, the last pulse of the immediately preceding burst cycle is determined for the first pulse oscillation. 23. The excimer laser device according to claim 22, wherein pulse oscillation is performed using a power supply voltage value used for oscillation as a power supply voltage. 26. A difference between an output value of each pulse monitored by said monitor means and a target value of said output energy is obtained for each pulse, and for a pulse whose difference exceeds an allowable limit, oscillation of said pulse is performed. 21. The excimer laser device according to claim 20, further comprising table correction means for correcting and updating the storage power supply voltage value of the voltage data table means corresponding to the elapsed time from the start and the measured oscillation pause time using the difference. 27. When the oscillation pause time measured by the oscillation pause time measuring unit coincides with one of the plurality of representative oscillation pause times, the table correction means may include:
Regarding the pulse in which the difference between the output value of each pulse monitored by the monitoring means and the target value exceeds an allowable limit, all the storages in the voltage data table means corresponding to the elapsed time from the start of oscillation of the pulse. 3. The power supply voltage value is corrected and updated according to the difference.
6. The excimer laser device according to 6. 28. The power supply voltage data table means,
The power supply voltage value is stored only for the pulse for the first predetermined time in one burst cycle, and the pulse oscillation control by the oscillation control means is executed only for the pulse included in the first predetermined time. The excimer laser device according to claim 20, wherein: 29. The power supply voltage data table means,
The power supply voltage value is stored only for the pulse for the first predetermined time in one burst cycle, and the pulse oscillation control by the oscillation control means and the storage power supply voltage data correction control by the table correction means are controlled by the first predetermined time. 27. The excimer laser device according to claim 26, wherein the operation is performed only for a pulse included in time. 30. With respect to the pulse generated after the first predetermined time, the monitor value of at least one pulse already output within the current burst cycle obtained from the monitor means and the power supply voltage value at that time are obtained. 30. The excimer laser device according to claim 28, wherein a power supply voltage value at the time of the current pulse oscillation is calculated based on the calculated value, and pulse oscillation is performed based on the power supply voltage value. 31. An excimer laser device according to claim 20, wherein said variation measuring means comprises variation calculating means for calculating output variation of each pulse oscillation light based on a monitor output of said monitoring means. 32. The variation calculating means obtains a standard deviation and an average value of output energy of a predetermined number of pulsed oscillation lights set in advance, and sets a value obtained by dividing the standard deviation by an average value as an output variation value. 31. An excimer laser device according to item 31. 33. A variation in output energy of each pulsed laser oscillation light takes a predetermined minimum value corresponding to a partial pressure of a halogen gas in the laser chamber, and a limit value of a target value range for the variation. Has two values, a first value smaller than the halogen gas partial pressure value corresponding to the minimum value, and a second value larger than the halogen gas partial pressure value corresponding to the minimum value. When the calculated output variation is out of the target value range, the control means determines whether the current halogen gas partial pressure is smaller than the first value or the current halogen gas partial pressure is smaller than the first value. 21. The excimer laser device according to claim 20, wherein it is determined whether or not the partial pressure value is larger than the second value, and only when the former is the former, the gas control means is controlled to supply the halogen gas. 34. An exhaust unit for exhausting a laser gas in the laser chamber, wherein the control unit causes the exhaust unit to perform an exhaust operation before the halogen gas is supplied by the gas supply unit. The excimer laser device as described in the above. 35. An exhaust unit for exhausting a laser gas in the laser chamber, wherein the control unit causes the exhaust unit to perform an exhaust operation while the halogen gas is supplied by the gas supply unit. The excimer laser device as described in the above. 36. A gas total pressure measuring means for measuring a total pressure of a laser gas in the laser chamber, wherein the control means controls a gas total pressure in the laser chamber based on a measurement value of the gas total pressure measuring means. 36. The excimer laser device according to claim 34, wherein the gas supply by the gas supply unit and the gas exhaust by the exhaust unit are controlled so that the gas does not exceed a predetermined upper limit. 37. The gas supply means for supplying a buffer gas and a rare gas, and when the average value of the output energy of the pulse light falls below a predetermined value, the gas supply means enters the laser chamber. 21. The excimer laser device according to claim 20, wherein a mixed gas of a buffer gas and a rare gas is supplied. 38. An excimer laser device according to claim 20, wherein said control means outputs an abnormal signal when said calculated variation falls outside a predetermined target value range.