JP4148379B2 - Pulse laser emission timing signal control device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to pulse laser emission timing and control on the semiconductor exposure apparatus side in a pulse laser that excites a laser medium by performing pulse discharge at a predetermined repetition rate using a magnetic pulse compression circuit and performs pulse laser oscillation. The present invention relates to a light emission timing signal control device for a pulse laser that improves the synchronization accuracy with timing.
[0002]
[Prior art]
Conventionally, the use of an excimer laser has attracted attention as a light source of a reduction projection exposure apparatus (hereinafter referred to as a stepper apparatus) for manufacturing a semiconductor device. This is because the wavelength of the excimer laser is short (the wavelength of KrF is about 248.4 nm), which may extend the limit of light exposure to 0.5 μm or less. This is because many excellent advantages such as a deeper depth of focus, a smaller numerical aperture (NA) of the lens, a larger exposure area, and greater power can be expected compared to the i-line. is there.
[0003]
FIG. 1 shows a general configuration of a control system for the excimer laser 1 and the stepper device 10.
[0004]
The excimer laser 1 includes a laser chamber 2 in which discharge electrodes and the like are incorporated, a pulse power supply device 3 that applies a high-frequency voltage synchronized with the repetition frequency of pulse discharge between the discharge electrodes, and energy of laser light emitted from the laser chamber 2. A laser controller 5 that performs power supply voltage control, laser oscillation wavelength control, laser gas supply control, and the like based on the energy monitor 4 to be monitored, the energy command from the stepper device 10 side, the monitor value of the energy monitor 4 and the like. Etc. ,
The stepper device 10 includes a stepper controller 11 that transmits a pulse oscillation synchronization signal TR serving as a trigger signal for repetitive pulse oscillation, a target energy command for laser oscillation, and the like, a movable wafer table 12 on which a wafer is mounted, and the like. The wafer on the wafer table 12 is subjected to reduced projection exposure using the laser light incident from the excimer laser 1.
[0005]
FIG. In recent years, as the pulse power supply device 3, a device using a magnetic pulse compression circuit is often used to improve the durability of main switches such as thyratron and GTO. FIG. Shows an equivalent circuit of a general capacity transfer type magnetic pulse compression discharge apparatus. Also, FIG. In FIG. The waveforms of voltage and current in each part of the circuit are shown.
[0006]
this FIG. This discharge circuit is a two-stage magnetic pulse compression circuit that utilizes the saturation phenomenon of three magnetic switches AL0 to AL2 each consisting of a saturable reactor.
[0007]
First, since the energy command value is input from the stepper device 10 side before the first laser oscillation trigger signal is received, the laser controller 5 calculates the power supply voltage value necessary to output this energy. The voltage V0 of the high voltage power supply HV is adjusted based on the calculated value. At this time, the capacitor C0 is precharged with the charge from the high voltage power supply HV via the magnetic switch AL0 and the coil L1.
[0008]
Thereafter, when the first laser oscillation synchronization signal (trigger signal) TR from the stepper device 10 side is received, the main switch SW is turned on at the time of reception ( FIG. , Time t0). After the main switch SW is turned on, when the time product of the voltage VC0 of the capacitor C0 (time integrated value of the voltage VC0) reaches a limit value determined by the setting characteristics of the magnetic switch AL0, the magnetic switch AL0 is saturated at this time t1. The current pulse i0 flows through the loop of the capacitor C0, the magnetic switch AL0, the main switch SW, and the capacitor C1. The time τ0 from when the current pulse i0 starts to flow until it becomes 0 (time t2), that is, the charge transfer time τ0 until the charge is completely transferred from the capacitor C0 to the capacitor C1, ignores the loss due to the main switch SW and the like. In this case, the capacitance and inductance of the capacitor C0, the magnetic switch AL0, and the capacitor C1 are determined.
[0009]
Thereafter, when the time product of the voltage VC1 of the capacitor C1 reaches a limit value determined by the setting characteristics of the magnetic switch AL1, the magnetic switch AL1 is saturated at this time t3 and becomes low impedance. As a result, the current pulse i1 flows through the loop of the capacitor C1, the capacitor C2, and the magnetic switch AL1. The current pulse i1 becomes 0 at time t4 after passing through a predetermined transfer time τ1 determined by the capacitance and inductance of the capacitors C1 and C2 and the magnetic switch AL1.
[0010]
After this, when the time product of the voltage VC2 of the capacitor C2 reaches a limit value determined by the setting characteristics of the magnetic switch AL2, the magnetic switch AL2 is saturated at this time t5, whereby the capacitor C2, the peaking capacitor CP, the magnetic switch The current pulse i2 flows through the AL2 loop.
[0011]
Thereafter, the voltage VCp of the peaking capacitor Cp increases with the progress of charging. When this voltage VCp reaches a predetermined main discharge start voltage, the laser gas between the main electrodes 6 is broken down at time t6 to start main discharge. The The laser medium is excited by this main discharge, and laser light is generated after several nsec.
[0012]
Thereafter, the voltage of the peaking capacitor Cp rapidly decreases due to the main discharge, and returns to the state before the start of charging after a predetermined time.
[0013]
Such a discharge operation is repeatedly performed by the switching operation of the main switch 5 synchronized with the trigger signal TR, whereby pulse laser oscillation at a predetermined repetition frequency (pulse oscillation frequency) is performed.
[0014]
FIG. According to this magnetic compression circuit, the inductance of each stage of the charge transfer circuit composed of the magnetic switch and the capacitor is set so as to decrease as it goes to the subsequent stage, so that the peak values of the current pulses i0 to i2 increase sequentially. In addition, a pulse compression operation is performed so that the energization width is gradually reduced. As a result, a strong discharge is obtained in a short time between the main electrodes 6.
[0015]
By the way, in recent years, the exposure method on the stepper apparatus 10 side is shifting from a batch exposure method in which exposure is performed with the stage stopped to a step & scan method in which exposure is performed while moving the stage. The advantage of this step-and-scan method is that a large area can be exposed. In the future, the chip size tends to increase as the degree of integration increases, and the step-and-scan method will become mainstream.
[0016]
In this step & scan method, FIG. As shown in FIG. 2, the exposure processing is performed while moving the laser light or the wafer at a predetermined pitch ΔP while irradiating the IC chip 7 on the wafer with a laser beam called a sheet beam. The movement accumulated exposure amount of all points on the chip 7 (for example, FIG. Then, by setting the scanning pitch ΔP and the irradiation area of the sheet beam so that the movement integrated exposure amount at point A is equal to P1 + P2 + P3 + P4, uniform exposure can be performed at each point on the IC chip 7. To be
[0017]
Thus, in the step & scan method, exposure is performed while moving the stage (or laser light). For this reason, the actual light emission timing of each pulse laser on the excimer laser side and the movement control timing of the wafer (or laser light) on the stepper side are not completely synchronized, that is, the pulse laser light is not emitted. If the stage is not moved during the period, the stage is moved during the laser irradiation, resulting in a large variation in the exposure amount at each position.
[0018]
Thus, when performing step-and-scan exposure, it is necessary to accurately grasp the timing at which laser light emission actually occurs. For this reason, conventionally, the stepper device side outputs a laser oscillation synchronization signal TR. The time from the start to the actual laser oscillation is predicted using experience and actual measurement data, and various controls in the stepper device are synchronized based on this prediction.
[0019]
Also, in a batch exposure method stepper in which the stage on which the wafer is placed is stopped and the entire chip is irradiated with a single laser, conventionally, the laser irradiation position is moved from the IC chip position after the exposure to the next IC chip. The timing of starting and the timing of starting work are estimated in the same manner as described above using the laser oscillation synchronization signal TR.
[0020]
However, the previous FIG. In the magnetic compression circuit shown in FIG. 4, the time td from the time t0 when the trigger signal TR is input and the main switch SW is called to the time t6 when the laser light is actually generated (hereinafter referred to as the light emission delay time) is the current. It depends on the energization widths τ0, τ1, τ2 of the pulses i0, i1, i2 and the saturation times σ0, (τ0 + σ1), (τ1 + σ2) of the magnetic switches AL0-AL2.
[0021]
As described above, the energization width (charge transfer time) τ0, τ1, and τ2 is determined by the capacitance and inductance of the capacitors and magnetic switches included in each stage of the charge transfer circuit. It is greatly affected by temperature.
[0022]
Further, since the saturation time variations σ0, σ1, and σ2 are determined by the time product of the voltages applied to the magnetic switches AL0 to AL2, they are greatly affected by the voltage V0 of the high voltage power supply HV.
[0023]
Here, in the excimer laser, as described above, the power supply voltage V0 is not controlled to be constant, but is one of the control parameters for controlling the laser output to be constant. Has been. In other words, power lock control that controls the power supply voltage in consideration of a decrease in laser output due to a decrease in halogen gas, and a spike region that includes the first few pulses of continuous pulse oscillation increases the laser output compared to other regions. In order to eliminate the phenomenon, the power supply voltage V0 is variably controlled in consideration of various factors such as spike killer control for controlling the power supply voltage.
[0024]
Thus, in the excimer laser, since the power supply voltage V0 is one of the control parameters, it is impossible to keep it constant. For this reason, variations in the saturation time of the magnetic switch σ0, σ1, σ2 changes greatly due to the variable control of the power supply voltage.
[0025]
By the way, when the variations of the saturation time variations σ0, σ1, σ2 and the charge transfer times τ0, τ1, τ2 were examined by experiments, the following results were obtained.
[0026]
When the power supply voltage V0 is the maximum voltage,
σ0 ≒ 0
σ1 ≒ 0
σ2 ≒ 0
When the power supply voltage V0 is the minimum voltage
σ0 ≒ 800nsec
σ1 ≒ 500nsec
σ2 ≒ 200nsec
At start-up (low temperature)
τ0 ≒ 1.65μsec
τ1 ≒ 550nsec
τ2 ≒ 150nsec
During continuous operation (at high temperature)
τ0 ≒ 1.5μsec
τ1 ≒ 500nsec
τ2 ≒ 130nsec
Therefore, according to the above experiment, the light emission delay time td changes by a maximum of 1.5 μsec depending on the power supply voltage V0, and changes by a maximum of 220 nsec due to temperature drift.
[0027]
When the power supply voltage V0 is the maximum voltage, the saturation times σ0, σ1, and σ2 are almost zero so that the charge transfer time between the capacitors and the saturation time of the magnetic switch coincide, that is, σ0 to σ2 This is because the maximum value of the power supply voltage V0 is set so that becomes 0, and the power supply voltage control is performed within a range not exceeding this maximum voltage value. Therefore, in this case, the magnetic switch does not saturate during the charge transfer between the capacitors, that is, the situation where σ1 and σ2 take negative values does not occur. An increase in width is definitely prevented
[0028]
[Problems to be solved by the invention]
As described above, in the prior art, the actual light emission timing is predicted on the stepper device side based on the laser oscillation synchronization signal TR. However, since the actual light emission timing varies depending on the power supply voltage and temperature, the predicted light emission There is a problem that the timing is deviated from the actual light emission timing, and the laser light emission timing and the control timing on the stepper side cannot be well synchronized.
[0029]
On the other hand, there is a case in which continuous pulse oscillation at a constant interval is always desired in response to the pulse oscillation synchronization signal TR sent from the stepper device 10 side. There has been a problem that synchronization between the timing and the control timing on the stepper side is not well synchronized.
[0030]
The present invention has been made in view of such circumstances, and is a physical quantity timing that is before the actual laser emission timing and is closer to the actual laser emission timing than the laser oscillation synchronization signal sent from the semiconductor exposure apparatus. Is controlled by the laser side and the emission timing signal based on this is sent to the semiconductor exposure apparatus side to improve the synchronization accuracy between the laser emission timing and the control timing on the semiconductor exposure apparatus side. An object is to provide an apparatus.
[0031]
Further, in the present invention, by compensating for the change in the power supply voltage and the environmental temperature, even if the change in the power supply voltage and the environmental temperature occurs, the time is always a predetermined time before the actual light emission time of each pulse laser. Provided is a pulse laser emission timing signal control device that further improves the synchronization accuracy between the laser emission timing and the control timing on the semiconductor exposure apparatus side by transmitting the generated emission timing signal to the semiconductor exposure apparatus side. For the purpose.
[0032]
Furthermore, in the present invention, the time from when the pulse oscillation synchronization signal is received until the laser is actually emitted is always constant over each pulse, so that the laser emission timing and the semiconductor exposure apparatus side An object of the present invention is to provide a pulsed laser emission timing signal control device that improves the synchronization accuracy with the control timing.
[0033]
[Means for solving the problems and effects]
The pulse laser emission timing signal control device according to the first aspect of the present invention comprises a plurality of stages including a plurality of magnetic switches connected in series to a charging power source and a plurality of capacitors each connected in parallel to the charging power source. A magnetic pulse compression circuit that compresses current pulses into a plurality of stages by the plurality of stages of charge transfer circuits, and switching means that performs a switching operation for intermittently charging the power supply to the magnetic pulse compression circuit, A laser discharge electrode connected to the output terminal of the magnetic pulse compression circuit, and by turning on the switching means triggered by a laser oscillation synchronization signal having a predetermined repetition frequency received from the semiconductor exposure apparatus side, In a pulse laser that performs pulsed laser oscillation at a predetermined repetition frequency, the magnetic pulse A current detection means for detecting a rising point of a current pulse flowing through a charge transfer circuit of a predetermined stage set in advance in the compression circuit; a voltage command value for the charging power supply; and an actual state of the pulse laser from the time detected by the current detection means. Using a hyperbolic approximate relational equation with a delay time until the light emission time point, a waiting delay time obtained by subtracting a predetermined time set in advance from the delay time according to the voltage command value is calculated and output from the current detection means. Output control means for outputting to the semiconductor exposure apparatus side a delay signal obtained by delaying the detection signal delayed by the waiting delay time as a light emission timing signal of each pulse laser oscillation.
[0034]
According to the first aspect of the present invention, the rising point of the current pulse flowing through the charge transfer circuit of a predetermined stage set in advance is detected, and this detection signal is always used as a light emission timing signal a predetermined time before the actual light emission of the pulse laser. Since the output is made to the exposure apparatus side, the semiconductor exposure apparatus side predicts the actual laser light emission timing based on the input light emission timing signal, and synchronizes various controls such as wafer movement based on this prediction timing. There is an effect of being able to.
[0035]
According to the first aspect of the present invention, it is possible to absorb the variation due to the fluctuation of the power supply voltage in the time interval from the rise time of the current pulse detected by the current detection means to the actual light emission time of the laser, and the actual light emission time of the pulse laser. Since the timing adjustment is performed so that the light emission timing signal is always emitted at a time point that is a predetermined time earlier in advance, the synchronization accuracy between the actual light emission timing of the laser and the control timing on the semiconductor exposure apparatus side is improved. There exists an effect that it can improve further.
[0036]
In addition, according to the first aspect of the present invention, the waiting delay time is calculated using the hyperbolic approximate expression, so that the delay time can be output quickly with a simple configuration.
[0037]
According to a second aspect of the present invention, there is provided a pulse laser emission timing signal control device comprising a plurality of stages including a plurality of magnetic switches connected in series to a charging power source and a plurality of capacitors each connected in parallel to the charging power source. A magnetic pulse compression circuit that compresses current pulses into a plurality of stages by the plurality of stages of charge transfer circuits, and switching means that performs a switching operation for intermittently charging the power supply to the magnetic pulse compression circuit, Pulse oscillation having a predetermined repetition frequency received from the semiconductor exposure apparatus side, having a laser discharge electrode connected to the output terminal of the magnetic pulse compression circuit and a control means for outputting a voltage command value for the charging power source By turning on the switching means with a synchronization signal as a trigger, a pulse rate of a predetermined repetition frequency is obtained. In a pulse laser that oscillates, a reference delay time that is set in advance is a predetermined reference delay time that is greater than or equal to the maximum value of the fluctuation range of the actual light emission delay time from when the switching means is turned on until the laser oscillation starts A delay time calculation means for obtaining a difference between the set reference delay time and the actual light emission delay time of the pulse oscillation corresponding to the voltage command value output from the control means for each pulse oscillation; Delay means for delaying the pulse oscillation synchronization signal received from the semiconductor exposure apparatus by the difference time calculated by the delay time calculation means and outputting the delayed signal to the switching means; and a predetermined predetermined value in the magnetic pulse compression circuit Current detection means for detecting a rising point of a current pulse flowing through the charge transfer circuit of the second stage, a voltage command value for the charging power source and the current A waiting delay time obtained by subtracting a preset predetermined time from the delay time according to the voltage command value using a hyperbolic approximate relational equation with a delay time from the time point detected by the detecting means to the actual light emission time point of the pulse laser And an output control means for outputting a delay signal obtained by delaying the detection signal output from the current detection means to the semiconductor exposure apparatus side as a light emission timing signal of each pulse laser oscillation. It is characterized by.
[0038]
According to the second aspect of the present invention, a predetermined reference delay time that is not less than the maximum value of the fluctuation range of the actual light emission delay time from when the switching means is turned on until the laser oscillation is started is set in advance. The reference delay time is, for example, the time from when the switching means is turned on when the laser oscillation is performed with a predetermined voltage value equal to or lower than the lower limit value of the voltage command value until the laser oscillation is started. Then, the difference between the reference delay time and the actual light emission delay time of the pulse oscillation is obtained for each pulse oscillation, and the pulse oscillation synchronization signal received from the semiconductor exposure apparatus is delayed by this difference and output to the switching means. Thus, the time from when the laser oscillation pulse synchronization signal is received until the laser actually emits light is always made to coincide with the set reference delay time over each pulse. Since the time from when the signal is received until the laser actually emits light is always constant over each pulse, the laser emission timing and semiconductor exposure are not performed on the semiconductor device side without particularly difficult predictive control. There exists an effect which can fully synchronize with the control timing by the side of an apparatus.
[0039]
According to a second aspect of the present invention, a rising point of a current pulse flowing through a charge transfer circuit of a predetermined stage set in advance is detected, and this detection signal is always used as a light emission timing signal a predetermined time before the actual light emission of the pulse laser. Since the output is made to the semiconductor exposure apparatus side, the semiconductor exposure apparatus side predicts the actual laser light emission timing based on the input light emission timing signal, and synchronizes various controls such as wafer movement based on this prediction timing. There is an effect that can be taken.
[0040]
According to a second aspect of the present invention, variations due to fluctuations in the power supply voltage in the time interval from the rise time of the current pulse detected by the current detection means to the actual light emission time of the laser are absorbed, and the actual light emission time of the pulse laser is Since the timing adjustment is performed so that the light emission timing signal is always emitted at a time point that is a predetermined time earlier in advance, the synchronization accuracy between the actual light emission timing of the laser and the control timing on the semiconductor exposure apparatus side is improved. There exists an effect that it can improve further.
[0041]
In addition, according to the first aspect of the present invention, the waiting delay time is calculated using the hyperbolic approximate expression, so that the delay time can be output quickly with a simple configuration.
[0042]
According to a third aspect of the present invention, there is provided the pulse laser emission timing signal control device according to the second aspect, wherein the output control means normalizes the difference obtained by the delay time calculating means to a maximum value of the waiting delay time. And a counter for measuring the waiting delay time using a hyperbolic approximate relationship between the difference and the normalized waiting delay time.
[0043]
In the invention of claim 3, since the waiting delay time is measured using the difference already obtained by the delay time calculating means, the same effect as that of the invention of claim 2 can be obtained with a simple configuration.
Further, in the invention of claim 3, since the delay time is measured by the counter, the waiting delay time can be quickly measured and output, and thus, from the detection signal to the actual light emission, There is an effect that measurement can be reliably performed even in a short time.
[0044]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows a first embodiment of the present invention. In this first embodiment, the previous FIG. A two-stage magnetic pulse compression circuit similar to that shown in FIG. 6 is used, and the voltages VC0, VC1, VC2, VCp and current pulse waveforms i0, i1, i2 at each part are also used. FIG. It is the same as that shown in.
[0045]
That is, this embodiment circuit has a high voltage power source 1, a main switch SW, three magnetic switches AL0 to AL2, a coil L1, four capacitors C0, C1, C2, Cp, and a discharge electrode 6. A first-stage charge transfer circuit that forms a current loop i0 from the capacitor C0 to the magnetic switch AL0 and the main switch SW to the capacitor C1, and a current loop i1 from the capacitor C1 to the capacitor C2 through the magnetic switch AL1 A second stage charge transfer circuit is formed, and a third stage charge transfer circuit is formed which forms a current loop i2 from the capacitor C2 to the peaking capacitor Cp via the magnetic switch AL2.
[0046]
Since the inductance of the magnetic switches AL0 to AL2 at the time of saturation and non-saturation is set to decrease in order, and the capacitor capacity of each stage is set to be equal, FIG. As shown in FIG. 2, the peak values of the current pulses flowing through the charge transfer circuits at the respective stages are sequentially amplified, and the pulse widths are sequentially compressed.
[0047]
In the discharge circuit of FIG. 1 as well, the power supply voltage at which the charge transfer time between the capacitors and the saturation time of the magnetic switch coincide, that is, σ1 to σ2 becomes 0 is set as the maximum power supply voltage V0. The power supply voltage is controlled within the range not exceeding.
[0048]
Here, in the embodiment shown in FIG. 1, the rising point t3 of the current pulse i1 flowing in the charge transfer circuit of the stage immediately before the final stage, that is, the charge transfer circuit of the second stage. FIG. A current pulse detection sensor 20 for detecting a reference) is provided, and a detection signal Sa of the current pulse detection sensor 20 is transmitted to the stepper device 10 as a light emission timing signal Ht via the signal output unit 21.
[0049]
FIG. 2 shows the configuration of the control system of the excimer laser 1 and the stepper device 10. FIG. The light emission timing signal Ht output from the pulse power supply device 3 to the stepper device 10 is added to the configuration shown in FIG.
[0050]
This light emission timing signal is FIG. As shown in FIG. Therefore, the variation in delay time from when the light emission timing signal is generated until the laser emits light, according to the experimental results described in the section of the prior art,
When the power supply voltage V0 is the maximum voltage,
σ2 ≒ 0
When the power supply voltage V0 is the minimum voltage
σ2 ≒ 200nsec
At start-up (low temperature)
τ1 ≒ 550nsec
τ2 ≒ 150nsec
During continuous operation (at high temperature)
τ1 ≒ 500nsec
τ2 ≒ 130nsec
It becomes.
[0051]
Assuming that the period from the time t3 when the light emission timing signal is generated to the time t6 when the laser emits light is the light emission delay time td ', this td' changes by a maximum of 200 nsec depending on the power supply voltage and also changes by a maximum of 70 nsec due to the temperature drift. Become.
[0052]
Therefore, if the rising edge of the current pulse i1 is detected by the current pulse detection sensor 20 and this detection signal Sa is output to the stepper device 10 side as the light emission timing signal Ht, the light emission timing signal Ht is actually output after the light emission timing signal Ht is output. The variation in emission delay time td 'until the laser emits light (200 nsec in voltage, 70 nsec in temperature) is significantly smaller than the variation in emission delay time td in the conventional method (1.5 μsec in voltage, 220 nsec in temperature). Become.
[0053]
Therefore, if the light exposure timing signal Ht sent from the laser side is used as a reference to know the actual laser light emission timing on the semiconductor exposure apparatus side, the light emission timing can be known more accurately than before. Thus, the synchronization accuracy between the laser emission timing and the control timing on the stepper device side can be improved.
[0054]
Next, specific examples of the current pulse detection sensor 20 are shown in FIGS.
[0055]
In FIG. 3, a magnetic core (or coil) 22 as an inductance means is connected in series with the magnetic switch AL1. As shown in FIG. 3 (b), an induced voltage vs is generated in the magnetic core 22 by self-induction due to the current change of the current pulse i1, so that the rise time t3 of the current pulse i1 is determined by detecting the induced voltage vs. Can be detected.
[0056]
In FIG. 4, an induced electromotive voltage vs induced in the secondary winding 23 of the magnetic switch AL1 (see FIG. 4B) is detected, and the rising time t3 of the current pulse i1 is detected based on this voltage detection. I am doing so.
[0057]
In FIG. 5, the primary coil 24 is connected in series to the magnetic switch AL1, and the induced voltage vs of the secondary coil 25 generated by the current change of the current pulse i0 flowing through the primary coil 24 (see FIG. 5B). ) And the rise time t3 of the current pulse i1 is detected based on this voltage detection.
[0058]
In FIG. 6, an air core coil 27 is connected in a reset circuit 26 (a circuit for setting the BH characteristics of the magnetic switch to a predetermined initial state) of the magnetic switch AL1, and an induced voltage vs of the air core coil 27 is detected. Based on this voltage detection, the rise time t3 of the current pulse i1 is detected.
[0059]
In FIG. 7, resistors r1 and r2 are connected in parallel with the capacitor C1, and the rise time t3 of the current pulse i1 is detected based on detecting the divided voltage vs (see FIG. 7B). Yes.
[0060]
FIG. 8 shows a second embodiment of the present invention. In the second embodiment, the same two-stage magnetic pulse compression circuit as described above is used, and the voltages VC0, VC1, VC2, VCp and current pulse waveforms i0, i1, i2 at each part are also used. FIG. It is the same as that shown in.
[0061]
In the first embodiment, since the rising time t3 of the current pulse i0 is detected and this detection signal is used as the light emission timing signal, the light emission timing signal includes the time t3 to the light emission time t6. An error due to variations due to the power supply voltage and temperature in the period td ′ is included.
[0062]
Therefore, in this embodiment, the current pulse is absorbed so that the variation due to the power supply voltage and temperature during the light emission delay time td ′ is absorbed, and the light emission timing signal is always emitted at a time point a certain time before the actual light emission time point. After the timing of the detection signal Sa of the detection sensor 20 is adjusted, the detection signal Sa is output to the stepper device 10 side.
[0063]
For example, the transmission time point of the light emission timing signal Ht to the stepper device side is set to the detection time point t3 of the current pulse detection sensor 20 when the light emission delay time td ′ is the shortest. Assuming that the minimum value of the light emission delay time td 'is, for example, 500 nsec, the detection signal of the current pulse detection sensor 20 so that the light emission timing signal Ht is always sent 500 nsec before the time t6 when laser light emission is actually performed. The timing is adjusted.
[0064]
Therefore, when it is determined that the time point when the detection signal of the current pulse detection sensor 20 is output is 550 nsec before light emission, the detection signal is delayed by 50 nsec and then transmitted to the stepper device 10 side as a light emission timing signal. .
[0065]
In the embodiment of FIG. 8, such control is performed, and the configuration and operation will be described below.
[0066]
The current pulse detection sensor 20 detects the rising point t3 of the current pulse i1 as in the first embodiment. The detection signal Sa of the current pulse detection sensor 20 is input to the voltage compensation unit 40 and the temperature compensation unit 50.
[0067]
The current pulse detection sensor 30 detects the rising point t1 of the current pulse i0 flowing through the first-stage charge transfer circuit. The specific configuration is the same as that of the current pulse sensor shown in FIGS. The detection signal Sb of the current pulse detection sensor 30 is input to the temperature compensation unit 50.
[0068]
When the outputs Sa and Sb of the current pulse detection sensors 20 and 30 are directly input to the voltage compensator 40 and the temperature compensator 50 through normal electric cables, the high-speed generated from the magnetic pulse compression circuit. Electromagnetic coupling with the high voltage pulse causes a great amount of noise to be mixed into the detection outputs Sa and Sb, making subsequent signal processing very difficult. Therefore, in this case, the outputs Sa and Sb of the current pulse detection sensors 20 and 30 are converted into optical signals using a light emitting diode or the like, and the converted optical signals are converted to the voltage compensation unit 40 and the temperature compensation via an optical fiber. The data is input to the unit 50.
[0069]
The voltage compensator 40 compensates for variations in the light emission delay time td ′ due to fluctuations in the power supply voltage, and in this case, compensates for variations σ2 in the saturation time (τ1 + σ2) of the magnetic switch AL2 due to fluctuations in the power supply voltage.
[0070]
FIG. 9 shows an example of the voltage compensation unit 40, which includes a delay time conversion unit 41 and a delay unit 42. The delay time conversion unit 41 converts the power supply voltage command value into a voltage compensation delay time Tdy. For example, various power supply voltage values within the range from the minimum voltage to the maximum voltage, and the delay time for voltage compensation. It consists of a memory table in which the relationship with Tdy is set and stored.
[0071]
As described above, when the emission timing signal Ht is sent to the stepper device side as the detection time t3 of the current pulse detection sensor 20 when the light emission delay time td 'is the shortest, various power sources are used. The saturation time variation σ2 of the magnetic switch AL2 corresponding to the voltage may be obtained by measurement or calculation, and the correspondence between this σ2 and the power supply voltage may be stored in the memory table as it is.
[0072]
As shown in the input / output time chart in FIG. 10, the delay unit 42 delays the detection signal Sa of the current pulse detection sensor 20 by the delay time Tdy output from the delay time conversion unit 41, thereby emitting light. The power supply voltage compensation of the signal Ht is executed.
[0073]
Here, the detailed configuration of the delay time conversion unit 41 will be described.
[0077]
FIG. Of the delay time conversion unit 41 Specific examples of detailed configuration In this case, the delay time Tdy corresponding to the power supply voltage V0 is obtained by actually performing hyperbolic approximation calculation. That is, the delay time conversion unit 41 includes a Tdy calculation unit 41c, and the Tdy calculation unit 41c performs a hyperbolic approximation calculation and outputs the delay time Tdy corresponding to the input voltage command value V0. .
[0078]
FIG. These are the figures explaining the calculation content by the Tdy calculating part 41c. First, the actual light emission delay time td (from the external trigger signal TR to the actual light emission start time t6 ( FIG. See)
td = (a / V0) + b (1)
A hyperbolic approximation can be made ( FIG. reference).
[0079]
On the other hand, in the period from the external trigger signal TR to the time point t3 (the output Sa of the current pulse detection sensor 20), the hyperbola approximation expressed by the following equation (2) can be performed as well (td−td ′). That is,
td-td '= (c / V0) + d (2)
It is. Here, a to d are constants. Therefore, the delay time td ′ is
td '= td- (td-td')
= (Ac) / V0 + (bd) (3)
Thus, the delay time td ′ can also be approximated by a hyperbola. However, this delay time td ′ is obtained on the assumption that the environment such as the ambient temperature is the same, and corresponds to the delay time Tdy of only the portion that is voltage-compensated by the voltage compensator 40. Therefore, the delay time Tdy can be immediately output by hyperbolic approximation calculation corresponding to the voltage command value V0.
[0080]
Note that the parameters a to d in the above equation are the actual light emission delay times td1, td2, and (td1-td1 '), (td2) when laser oscillation is performed using two different power supply voltages V01, V02. -Td2 ') is measured and calculated in advance using these measured values and the power supply voltages V01 and V02.
[0081]
FIG. Shows a configuration for automatically generating the parameters a to d of the hyperbolic approximation equations (1) to (3).
[0082]
The light emission timing detection unit 62 outputs a light emission timing signal (t6) indicating an actual light emission time point by appropriately sampling a part of the emitted laser light. Based on the pulse oscillation synchronization signal TR and the light emission timing signal, the measuring unit 63 measures the actual light emission delay time td from when the pulse oscillation synchronization signal TR is applied to the main switch SW until the laser actually emits light. Further, based on the pulse oscillation synchronization signal TR and the output signal Sa, a delay time (td−td ′) from the time when the pulse oscillation synchronization signal TR is applied to the main switch SW to the time when the output Sa is obtained. Is measured and output to the storage unit 64.
[0083]
The storage unit 64 supplies various measured values td1, td2, td3,..., (Td1-td1 ′), (td2-td2 ′), (td3-td3 ′),. It is stored in association with the voltage command value V0 (V01, V02, V03,...). The calculation unit 65 calculates the parameters a to d a plurality of times using a plurality of sets of td values, (td−td ′) values and V 0 values stored in the storage unit 64, and obtains the average value thereof to obtain the final value. Parameter values a to d are obtained. Then, the parameter values a to d obtained in this way are set in the Tdy calculation unit 41c, and the parameters a to d of the hyperbolic approximation formula are periodically updated.
[0084]
In this way, the delay time Tdy can also be output by directly calculating and outputting using the hyperbolic approximate expression.
[0085]
Here, when the temperature compensation is not performed and the light emission timing signal Ht is transmitted a certain time before the light emission, FIG. As shown in FIG. 5, a delay time twait obtained by subtracting a predetermined time tconst may be output.
[0086]
That is,
twait = td'-tconst
= (Ac) / V0 + (bdt-const) (5)
The delay time twait is also a hyperbolic approximation formula, and as described above, the light emission timing signal Ht can be sent to the stepper device 10 side by a predetermined time tconst by performing the hyperbolic approximation calculation.
[0087]
In this way, the voltage compensation unit 40 in FIG. 8 outputs the timing signal Sa ′ in which the variation in the light emission timing caused by the power supply voltage is compensated.
[0088]
Next, the temperature compensation unit 50 in FIG. 8 compensates for variations in the light emission delay time td ′ due to temperature fluctuations in the magnetic pulse circuit. That is, in this case, temperature variations of the charge transfer time τ1 of the second-stage charge transfer circuit through which the current pulse i1 flows and the charge transfer time τ2 of the third-stage charge transfer circuit through which the current pulse i2 flows are compensated.
[0089]
FIG. Shows an example of the internal configuration of the temperature compensation unit 50. In this case, temperature compensation is performed without actually detecting the temperature.
[0090]
The time difference counter 51 counts the time difference between the detection signal Sa of the current pulse detection sensor 20 and the detection signal Sb of the current pulse detection sensor 30, so that the time from the rising point t1 of the current pulse i0 to the rising point t3 of the current pulse i1 is elapsed. The time (τ0 + σ1) is obtained, and this time difference data is output to the τ0 calculation unit 52 ( FIG. (See (a) to (c)).
[0091]
The τ0 calculation unit 52 estimates the elapsed time σ1 from the falling time point t2 of the current pulse i0 to the rising time point t3 of the current pulse i1 based on the voltage command value, and is estimated from the time difference data (τ0 + σ1) input from the time difference counter 51. By subtracting the time data σ1, the pulse width τ0 of the current pulse i0 is obtained. Therefore, the τ 0 calculation unit 52 has a memory table in which the correspondence between the power supply voltage and σ 1 is stored, for example.
[0092]
The delay time calculation unit 53 is configured to always generate various values of τ0 that can be taken corresponding to the change in the ambient temperature of the magnetic pulse compression circuit and the light emission timing signal Ht at a time point that is a certain time before the actual light emission time point. It has a memory table in which the correspondence with the temperature compensation delay time Tdy 'is stored. The delay time Tdy ′ takes into account fluctuations in τ1 and τ2 corresponding to temperature fluctuations, and the delay time Tdy ′ is set to be associated with the τ0 value. That is, if τ0 is determined, the values of τ1 and τ2 can be predicted, so if these τ1 and τ2 are used, how many seconds later the actual light emission time t6 is from the rise time t3 of the current pulse i1 (in this case) Can be predicted because it is only temperature compensation. Based on the predicted time τ1 + τ2 until the actual light emission time, the detection signal Sa of the current pulse detection sensor 20 is used to always generate the light emission timing signal Ht at a time point a certain time before the actual light emission time. The delay time Tdy ′ can be obtained by calculating how much time is delayed.
[0093]
The delay time calculation unit 53 selects a delay time Tdy ′ value corresponding to the τ0 value output from the τ0 calculation unit 52 from the memory table, and outputs the selected delay time Tdy ′ value to the delay unit 54.
[0094]
In the delay unit 54, FIG. As shown in the time chart in (d), the detection signal Sa of the current pulse detection sensor 20 (in this case, the output signal Sa ′ of the voltage compensation unit 40) by the delay time Tdy ′ output from the delay time calculation unit 53. To compensate the temperature compensation of the light emission timing signal Ht.
[0095]
In this way, the temperature compensation unit 50 of FIG. 8 outputs the light emission timing signal Ht in which the variation of the ambient temperature is compensated.
[0096]
of course, FIG. As shown in FIG. 4, temperature data as one of environmental parameters that are equivalently obtained from the current pulse detection sensors 20 and 30 may be acquired from the temperature sensor 70. The temperature sensor 70 detects the temperature of the insulating oil used in the saturable reactors AL0 to AL2, for example. In addition, it goes without saying that compensation may be made by detecting the environment on the pulse laser side.
[0097]
As described above, in the second embodiment, even if a change in the power supply voltage and the environmental temperature occurs by compensating for the change in the power supply voltage and the environmental temperature, a predetermined time set in advance from the actual light emission time of each pulse laser is always obtained. Since the light emission timing signal Ht generated just before is transmitted to the semiconductor exposure apparatus side, the synchronization accuracy between the laser light emission timing and the control timing on the semiconductor exposure apparatus side can be further improved. .
[0098]
In the above embodiment, the present invention is applied to a two-stage magnetic pulse compression circuit. However, the present invention may be applied to a three-stage or more magnetic pulse compression circuit. In the present invention, the rising point of the current pulse flowing through the charge transfer circuit in the stage immediately before the last stage of the magnetic pulse compression circuit is detected. However, the magnetic pulse compression after the laser synchronization trigger signal TR is detected. As long as the intermediate timing can be detected, the number of stages of the charge transfer circuit that detects the rising point of the current pulse may be set arbitrarily.
[0099]
In the above embodiment, the maximum value of the power supply voltage V0 is set so that the charge transfer time between the capacitors and the saturation time of the magnetic switch coincide, that is, σ0 to σ2 become 0, and this maximum voltage value However, the present invention can be applied to an apparatus that does not perform such power supply voltage control.
[0100]
In the above embodiment, temperature compensation and voltage compensation are performed on the laser side. However, data necessary for these compensations is transmitted from the laser side to the stepper side together with the light emission timing signal, and temperature compensation and voltage compensation are performed. You may make it carry out by the stepper side. In the above embodiment, the temperature compensation and the voltage compensation are performed on the laser side so that the light emission timing signal is always emitted a predetermined time before the actual light emission point. However, the detection signal of the current pulse detection sensor 20 is used as the light emission timing. As a signal, it may be sent to the stepper side, and numerical data indicating that actual light emission will occur after n seconds from this light emission timing signal may be sent to the stepper side.
[0101]
Furthermore, in the above embodiment, the laser apparatus is connected to a semiconductor exposure apparatus that performs step-scan exposure control, and a light emission timing signal generated from the laser apparatus is input to the step-scan semiconductor exposure apparatus. Even when the laser apparatus is connected to a batch exposure type semiconductor exposure apparatus, the light emission timing signal may be inputted to the batch exposure type semiconductor exposure apparatus. In the batch exposure method, the stage on which the wafer is placed is stopped and exposure is performed by irradiating the entire IC chip with a batch laser, and the laser irradiation position is changed from the IC chip position after the exposure to the next IC chip. When moving, the stage movement control must be performed. Thus, even in a batch irradiation type semiconductor exposure apparatus, if the timing for starting the stage movement and the timing for starting the work are determined by using the light emission timing signal sent from the laser apparatus side, the light emission of the laser is possible. It is possible to improve the synchronization accuracy between the timing and the control timing on the semiconductor exposure apparatus side.
[0102]
Further, in the above embodiment, voltage compensation and temperature compensation are performed based on the value of the voltage command value and the ambient temperature. However, the present invention is not limited to this, and when there is another factor that affects the delay time, Compensation should be performed in consideration of the influence of factors. For example, the temperature may be obtained by compensating both the temperature change of the insulating oil of the saturable reactor in the magnetic pulse compression circuit and the ambient temperature in the vicinity of the magnetic pulse compression circuit. Further, the compensation may be made in consideration of the influence caused by the laser oscillation mechanism itself.
[0103]
Next, a third embodiment will be described.
[0104]
FIG. FIG. 4 shows the configuration of the control system of the excimer laser 1 and the stepper device 10 according to the third embodiment of the present invention. The difference from the configuration shown in FIG. The pulse oscillation synchronization signal TR received from the 10 side is delayed according to the power supply voltage command V0 (details will be described later), and the delay signal TRL is input to the pulse power supply circuit 3.
[0105]
FIG. 2 shows an internal configuration example of the pulse power supply circuit 3 and the laser controller 5.
[0106]
The pulse power supply circuit 3 FIG. A two-stage magnetic pulse compression circuit similar to that shown in FIG.
[0107]
The voltage command calculation unit 80 of the laser controller 5 uses the energy monitor value Ea input from the energy monitor 4 as a feedback signal, and a voltage command necessary to output energy according to the energy command value E input from the stepper device 10 side. The value V0 is calculated, and the calculated value V0 is output to the high voltage power supply HV and the delay time calculation unit 82.
[0108]
In this case, it is assumed that the adjustment range of the voltage command value V0 is Vmin ≦ V0 ≦ Vmax.
[0109]
In this case, the charge transfer time between the capacitors C0 to C2 and Cp and the saturation time of the magnetic switches AL0 to AL2 coincide, that is, FIG. The maximum value Vmax of the power supply voltage V0 is set so that .sigma.0 to .sigma.2 becomes 0, and the power supply voltage control is performed within a range not exceeding the maximum voltage value Vmax. For this reason, the magnetic switch saturates in the middle of charge transfer between capacitors, that is, the situation where σ1 and σ2 take negative values does not occur, thereby reducing the peak value of the current pulse and increasing the conduction width This is definitely prevented.
[0110]
The configuration of the reference delay time setting unit 81, the delay time calculation unit 82, and the delay unit 83 uses each pulse as a time from when the laser oscillation pulse synchronization signal TR is received on the excimer laser 1 side until the laser actually emits light. This is for always matching.
[0111]
The reference delay time setting unit 81 is preset with a predetermined reference delay time Tds that is equal to or greater than the maximum value of the variation range of the light emission delay time from when the main switch SW is turned on until laser oscillation is actually started. . For example, when a predetermined voltage value Vs (≦ Vmin) equal to or smaller than the minimum voltage Vmin in the adjustment range Vmin ≦ V0 ≦ Vmax of the voltage command value V0 is determined, and laser oscillation is performed at this voltage Vs An oscillation delay time from when the main switch SW is turned on until laser oscillation is actually started is set as a reference delay time Tds. The set reference time Tds of the reference delay time setting unit 21 is input to the delay time calculation unit 82.
[0112]
Based on the input voltage command value V0, the delay time calculation unit 82 predicts and calculates the actual light emission delay time td when laser oscillation is performed with the voltage command value V0, and the reference delay time Tds and the actual light emission. A difference from the delay time td is obtained, and the obtained difference is output to the delay unit 83 as a delay time τ (= Tds−td). That is, the voltage command V0 output from the voltage command value calculation unit 80 is always larger than the power supply voltage Vs used to obtain the reference delay time Tds (V0 ≧ Vs), so the delay time calculation unit 82. The actual light emission delay time td that is predicted and calculated in (1) is always smaller than the reference delay time Tds, and the difference is calculated as the delay time τ. That is, as the power supply voltage V0 is increased, the saturation time of each of the magnetic switches AL0 to AL2 is shortened. Accordingly, the actual light emission delay time td is also shortened.
[0113]
The delay unit 83 delays the received pulse oscillation synchronization signal TR by the delay time τ input from the delay time calculation unit 82, and outputs the delayed signal TRL to the main switch SW.
[0114]
As a result, during each pulse oscillation, the time from when the pulse oscillation synchronization signal TR is received by the delay unit 83 until the actual laser emission occurs is the actual emission delay time td + delay time τ (= reference delay time Tds). ) And always coincides with the reference delay time Tds, which is a constant value.
[0115]
FIG. (a) shows the laser oscillation when the power supply voltage V0 is used as the reference delay time Tds setting voltage Vs. FIG. Shows the voltage waveform at each part of the magnetic pulse compression circuit of FIG. (d) shows voltage waveforms at various parts of the magnetic pulse compression circuit when the power supply voltage V0 is laser-oscillated at a predetermined voltage Va higher than the voltage Vs, and these time axes are common.
[0116]
As is apparent from the comparison of these voltage waveforms, when the initial charge voltage (command voltage) V0 increases, the voltage time product portion (S0, S1, S2) shrinks in the time axis direction, and each of the magnetic switches AL0 to AL1. Each saturation time is shortened.
[0117]
Therefore, FIG. As shown in (a), when the initial charge voltage V0 is small Vs, a reference delay time Tds is required from when the pulse oscillation synchronization signal TR is applied to the main switch SW until the laser actually emits light. But, FIG. As shown in (b), when the initial charge voltage V0 is large, Va requires only time td (<Tds) from when the pulse oscillation synchronization signal TR is applied to the main switch SW until the laser actually emits light. Not.
[0118]
FIG. The delay time calculation unit 82 predicts and calculates the actual light emission delay time td, subtracts the calculated actual light emission delay time td from the reference delay time Tds, and sets the difference (Tds−td) as the delay time τ. Output to. The delay unit 83 forms a signal TRL obtained by delaying the received pulse oscillation synchronization signal TR by the delay time τ, and outputs the delay signal TRL to the main switch SW. FIG. As shown in FIG. 8, the time from when the pulse oscillation synchronization signal TR is received by the laser controller 5 (by the delay unit 83) until the actual laser light emission occurs is always matched with the reference delay time Tds.
[0120]
Therefore, when laser oscillation is actually performed, the readout unit 91 reads out the delay time τ corresponding to the power supply voltage V0 and outputs the delayed time τ to the delay unit 83 for each pulse oscillation. The synchronization signal TR is delayed by a delay time τ. When the power supply voltage value V0 not included in the memory table 90 is input, the reading unit 91 has a delay time τ1 corresponding to two power supply voltages V01 and V02 (V01 <V0 <V02) close to the power supply voltage V0, The delay time τ corresponding to V0 is obtained by reading τ2 from the memory table 90 and performing linear interpolation using these.
[0121]
FIG. Is Specific example of the delay time calculation unit 82 In this case, the delay time τ corresponding to the power supply voltage V0 is obtained by actually performing hyperbolic approximation calculation.
[0122]
The td calculation unit 100 predicts and calculates the actual light emission delay time td corresponding to the power supply voltage command, and a program or circuit corresponding to the hyperbolic approximation formula (1) described above is set. That is,
td = (a / V0) + b (1)
It is.
[0123]
Note that the parameters a and b in the above equation are obtained by measuring the actual light emission delay times td1 and td2 when laser oscillation is performed using two different power supply voltages V01 and V02. Calculation is performed in advance using V01 and V02.
[0124]
The td calculation unit 100 calculates an actual light emission delay time td corresponding to the input power supply voltage command V 0 for each pulse oscillation based on the above equation (1), and outputs the calculated value td to the τ calculation unit 101. To do. The τ operation unit 101 subtracts the input real light emission delay time td from the set reference delay time Tds, and outputs the subtraction result τ (= Tds−td) to the delay unit 83.
[0125]
FIG. Shows a configuration for automatically generating the parameters a and b of the hyperbolic approximation formula (1).
[0126]
The light emission timing detection unit 102 outputs a light emission timing signal indicating the actual light emission time point by appropriately sampling a part of the emitted laser light. The td measuring unit 103 is FIG. As shown in FIG. 4, based on the pulse oscillation synchronization signal TR and the light emission timing signal, the actual light emission delay time td from when the pulse oscillation synchronization signal TR is applied to the main switch SW until the laser actually emits light is calculated. Measure and output the measured values to the V0-td storage unit 104.
[0127]
The V0-td storage unit 104 stores various measurement values td1, td2, td3,... Measured by the td measurement unit 103 in association with the power supply voltage command values V0 (V01, V02, V03,...) At that time. . The ab calculation unit 105 calculates the parameters a and b a plurality of times using a plurality of sets of td values and V0 values stored in the V0-td storage unit 104, and obtains an average value thereof to obtain a final parameter value. Find a and b. The parameter values a and b thus obtained are FIG. The parameters a and b of the hyperbola approximation formula (1) are periodically updated by inputting to the td calculation unit 100 shown in FIG.
[0128]
FIG. Shows the fourth embodiment of the present invention. In this case, in addition to the power supply voltage V0, the delay time τ 'described above is considered in consideration of variations in the actual light emission delay time td due to the ambient temperature of the magnetic compression circuit. The pulse oscillation synchronization signal TR is delayed by the determined delay time τ ′.
[0129]
That is, the light emission delay time includes not only the saturation time of the magnetic switches AL0 to AL2, but also the energization widths δ0, δ1, δ2 of the current pulses i0, i1, i2 ( FIG. However, the current-carrying widths (charge transfer times) δ0, δ1, and δ2 are determined by the capacitance and inductance of the capacitors and magnetic switches included in the charge transfer circuits at each stage. It is affected by the ambient temperature in the circuit.
[0130]
FIG. In this case, the reference delay time Tds set and stored in the reference delay time setting unit 81 is a predetermined voltage value Vs where the temperature is the predetermined reference temperature u0 and the power supply voltage V0 is equal to or lower than the minimum voltage Vmin as described above. The value when is is set. As described above, the delay time calculation unit 81 calculates the actual light emission delay time td when laser oscillation is performed with the voltage command value V0 based on the input voltage command value V0. A difference from the reference delay time Tds is obtained, and the obtained difference is output to the temperature compensation unit 111 as a delay time τ (= Tds−td).
[0131]
The temperature sensor 110 detects the ambient temperature u of the magnetic compression circuit, and outputs the detected temperature u to the temperature compensation unit 111.
[0132]
The temperature compensation unit 111 has a memory table that stores a correspondence relationship between a plurality of ambient temperatures u and delay times ε corresponding to these ambient temperatures u (this delay time ε takes into account only a temperature change). Yes. That is, in the state where the power supply voltage V0 is set to the reference delay time Tds setting voltage Vs, the actual light emission delay time td is measured by changing the temperature u in various ways, and the actual measurement value td and the power supply voltage V0 are used as the voltage. The difference ε (= Tds−td) from the reference delay time Tds when the temperature is Vs and the temperature is the predetermined reference temperature u0 is obtained, and the difference ε is stored in association with the ambient temperature u. I have to.
[0133]
Then, the temperature control unit 111 reads the delay time ε corresponding to the detected value u of the temperature sensor from the memory table, and sets the temperature with respect to the delay time τ considering only the power supply voltage input from the delay time calculation unit 21. The considered delay time ε is added, and the addition result τ ′ (= τ + ε) is output to the delay unit 83 as the final delay time τ ′.
[0134]
In the delay unit 83, the pulse oscillation synchronization signal TR is delayed by the delay time τ 'and applied to the main switch SW. Therefore, in this embodiment, changes in the power supply voltage and the ambient temperature are compensated, and when the pulse oscillation synchronization signal TR is received on the excimer laser side (or the pulse oscillation synchronization signal is transmitted by the stepper device 10). The time interval from the instant (time) until the actual laser emission occurs can always be constant over each pulse.
[0135]
Next, a fifth embodiment will be described.
[0136]
In the fifth embodiment, as shown in the third and fourth embodiments, the laser controller 5 is controlled to emit laser light after the reference delay time Tds after receiving the pulse oscillation synchronization signal TR from the stepper device 10 side. In addition, as shown in the first and second embodiments, a light emission timing signal Ht for notifying the stepper device 10 that laser light is emitted before a predetermined time tconst is sent to the stepper device 10 side. is there.
[0137]
FIG. These show the internal structure of the pulse power supply circuit 3 and the laser controller 5 which are the 5th Embodiment of this invention. FIG. In the pulse power supply circuit 3 and the laser controller 5, the same current pulse detection sensor 85 as that of the current pulse detection sensor 20 shown in the first embodiment is provided, and the magnetic pulse AL1 is turned on in the current pulse detection sensor 85. The rising time t3 of the current pulse i1 is detected and a detection signal Sa is output.
[0138]
The delay unit 83 is further provided with a signal delay unit 84. The signal delay unit 84 uses the delay time τ calculated by the delay time calculation unit 82 to emit a light emission timing signal before a certain time tconst of actual light emission. A light emission timing signal Ht obtained by delaying the detection signal Sa is output to the stepper device 10 side so that Ht is sent to the stepper device 10 side.
[0139]
FIG. These show the detailed configuration of the signal delay unit 84. FIG. The twait calculator 86 has a τ / N counter 87 described later, calculates a time twait based on the delay time τ input from the delay time calculator 82, and outputs the time twait to the Ht output delay unit 88. The Ht output delay unit 88 sends a light emission timing signal Ht obtained by delaying the detection signal Sa input from the current pulse detection sensor 85 by a time twait to the stepper device 10 side.
[0140]
FIG. Is FIG. It is a figure which shows the relationship between voltage command value V0 and time t similar to FIG. this FIG. As shown in FIG. 6, in order to actually emit light after the reference delay time Tds has elapsed from the pulse oscillation synchronization signal TR, a signal TRL delayed by the delay time τ from the reception of the pulse oscillation synchronization signal TR is output to the main switch SW. In order to generate the light emission timing signal Ht and output it to the stepper device 10 before the predetermined time tconst, the time twait from the time t3 must be calculated. The time twait is calculated by the twait calculation unit 86. The twait calculation unit 86 calculates the time twait from the delay time τ as follows.
[0141]
That is, FIG. As shown in FIG. 5, the delay time τ increases by a hyperbola approximation between the minimum voltage command value Vmin and the maximum voltage command value Vmax, and the time twait decreases by a hyperbola approximation.
[0142]
Therefore, when the maximum value τmax of the delay time τ is normalized to the maximum value twait (max) of the time twait, FIG. As shown in FIG. 5, the time twait and the normalized delay time τ / N have a relationship in which the value obtained by adding the respective values is always the time twait. In this case, it goes without saying that the normalized delay time τ / N also has a hyperbolic approximation characteristic. here,
N = twait (max) / τmax
It is. As a result,
twait = twait (max) −τ / N
If the time twait (max) at the minimum voltage command value Vmin and the delay time τmax at the maximum voltage command value Vmax are obtained in advance, the time wait (max) and the normalized value N are obtained. Thus, a time twait corresponding to this delay time τ is obtained from an arbitrary delay time τ.
[0143]
For example, FIG. When the delay time τx corresponding to the voltage command value Vx is input from the delay time calculation unit 82, the normalized delay time τx / N is subtracted from the time twait (max), and the time when the voltage command value Vx is reached. twait (x) is obtained. In this case, the twait calculator 86 does not need to know the value of the voltage command value Vx, and can immediately calculate the corresponding time twait (x) from only the input delay time τx.
[0144]
Here, the delay time τ is normalized to the time twait, but conversely, the time twait may be normalized to the delay time τ. Furthermore, each other may be normalized to a predetermined time intermediate between the delay time τmax and the time twait.
[0145]
Thus, in the fifth embodiment, the light emission timing signal Ht is efficiently output by effectively utilizing the delay time τ calculated by the delay time calculation unit 82.
[0146]
Next, a specific configuration of the τ / N counter 87 of the twait calculation unit 86 will be described.
[0147]
First, FIG. and FIG. A first specific configuration of the τ / N counter 87 of the twait calculation unit 86 will be described with reference to FIG. FIG. , The delay time τ is input to the timer 121 as an enable signal ( FIG. (See (a)), the timer 121 counts the clock signal in the enabled period, and holds the counted value by disabling the delay time τ (see FIG. FIG. (See (c)). The counted value is output to the divider 122. The divider 122 divides the counted value (equivalent to τ) by the normalized value N, and τ / N is output when the value is determined. The Thereafter, the twait calculation unit 86 subtracts τ / N from twait (max), and outputs the subtraction result to the Ht output delay unit 88 as twait.
[0148]
next, FIG. and FIG. A second specific configuration of the τ / N counter 87 of the twait calculation unit 86 will be described with reference to FIG. FIG. , The delay time τ is input to the m-ary counter 131, and the m-ary counter 131 generates an enable signal corresponding to 1 / m times based on the input clock signal ( FIG. (Refer to (c)), and output to the n-fold counter 132. The n-fold counter 132 generates an n-fold clock signal obtained by multiplying the clock signal by n, and counts the n-fold clock according to the input enable signal ( FIG. As a result, τ / N is output from the n-fold counter 132. Thereafter, the twait calculation unit 86 subtracts τ / N from twait (max), and outputs the subtraction result to the Ht output delay unit 88 as twait.
[0149]
In the second specific configuration, since there is no calculation time for division, τ / N can be output quickly.
[0150]
next, FIG. and FIG. A third specific configuration of the τ / N counter 87 of the twait calculation unit 86 will be described with reference to FIG. FIG. The delay time τ is input to one end of the AND circuit 141 ( FIG. (See (a)). On the other hand, the shift register 142 generates a logic signal that is generated by cyclically rotating the positive logic by 1 / N, and inputs it to the other end of the AND circuit 141 ( FIG. (See (b)). For example, when a positive logic of 1/5 is generated by a 10-stage shift register, “0010000000100” is repeatedly output. The AND circuit 141 ANDs the delay time τ and the signal from the shift register 142 ( FIG. (See (c)), this logical product is output to the timer 143 as an enable signal. The timer 143 measures the period enabled by the enable signal based on the clock signal and outputs the time data thus measured. As a result, the time data output from the timer corresponds to τ / N, and then the twait calculation unit 86 subtracts τ / N from twait (max), and uses the subtraction result as twait as the Ht output delay unit. 88 to be output.
[0151]
In the third specific configuration, τ / N can be obtained by thinning out the clock signal by 1 / N output from the shift register 142, and by simply changing the value of the shift register 142, τ / N can be quickly obtained. N can be output.
[0152]
next, FIG. A fourth specific configuration of the τ / N counter 87 of the twait calculation unit 86 will be described with reference to FIG. FIG. In the fourth specific configuration, the delay time τ is input to one end of the AND circuit 151 as in the third specific configuration. On the other hand, the shift register 152 generates a logic signal that is generated by circulating the positive logic equally at 1 / N, and inputs the generated logic signal to the other end of the AND circuit 151. The AND circuit 151 calculates a logical product of the delay time τ and the signal from the shift register 152 and outputs the logical product to the timer 153 as an enable signal. The timer 153 measures the period enabled by the enable signal based on the clock signal and outputs the time data thus measured.
[0153]
In the fourth specific configuration, a plurality of logic patterns PT1 to PTn set in the shift register 152 are further stored in the memory 154. The control unit 157 causes the selector 156 to select a desired logic pattern PT1 to PTn and loads it into the shift register 152. When this loading is completed, the control unit 157 causes the shift register 152 to perform a shift operation, and outputs a logic pattern to the AND circuit 151 as in the third specific configuration.
[0154]
Accordingly, the time data output from the timer corresponds to τ / N corresponding to the desired 1 / N logic pattern loaded in the shift register 152, and then the twait calculation unit 86 calculates the τ from twait (max) to τ. / N is subtracted, and the result of this subtraction is output to the Ht output delay unit 88 as twait.
[0155]
As a result, in the third specific configuration, for example, when 1 / N is 1/100, a 100-stage shift register is required. However, in the fourth specific configuration, a logical pattern of “0000000” is 9 Since it is only necessary to switch between and select one repetition of the logic pattern “1000000000”, it can be realized with a 10-stage shift register, and can be reduced in size and weight and flexible. .
[0156]
In the fifth embodiment described above, the second to fourth specific configurations with a short calculation time of τ / N are preferable, and the third and fourth specific configurations are easy to change and set N. A configuration is preferred.
[0157]
In the above embodiment, the delay unit 83 starts time counting when the pulse oscillation synchronization signal TR is input, and generates a trigger signal for outputting the pulse oscillation synchronization signal TR when the delay time τ elapses. Appropriate timer means can be formed. Also, the delay unit 83 compares the integrator that starts the integration operation when the pulse oscillation synchronization signal TR is input with the output of the integrator and the output τ corresponding to the delay time, and the comparison result matches. In this case, a comparator that generates a trigger signal for outputting the pulse oscillation synchronization signal TR may be used.
[0158]
In the above embodiment, the present invention is applied to a two-stage magnetic pulse compression circuit. However, the present invention may be applied to a three-stage or more magnetic pulse compression circuit.
[0159]
In the above embodiment, the maximum value of the power supply voltage V0 is set so that the charge transfer time between the capacitors and the saturation time of the magnetic switch coincide, that is, σ0 to σ2 become 0, and this maximum voltage value However, the present invention can be applied to an apparatus that does not perform such power supply voltage control.
[0160]
The present invention is applicable to a semiconductor exposure apparatus that performs exposure control of either a step scan method or a batch exposure method.
[0161]
Further, in the above embodiment, voltage compensation and temperature compensation are performed based on the value of the voltage command value and the ambient temperature. However, the present invention is not limited to this, and when there is another factor that affects the delay time, Compensation should be performed in consideration of the influence of factors. For example, the temperature may be obtained by compensating both the temperature change of the insulating oil of the saturable reactor in the magnetic pulse compression circuit and the ambient temperature in the vicinity of the magnetic pulse compression circuit. Further, the compensation may be made in consideration of the influence caused by the laser oscillation mechanism itself.
[Brief description of the drawings]
FIG. 1 is a circuit block diagram showing a first embodiment of the present invention.
FIG. 2 is a block diagram showing a configuration of a control system of an excimer laser and a stepper to which the present invention is applied.
FIG. 3 is a diagram showing a specific configuration of a current pulse detection sensor and current and voltage waveforms in that case.
FIG. 4 is a diagram showing another specific configuration of a current pulse detection sensor and current and voltage waveforms in that case.
FIG. 5 is a diagram showing another specific configuration of a current pulse detection sensor and current and voltage waveforms in that case.
FIG. 6 is a diagram showing another specific configuration of the current pulse detection sensor.
FIG. 7 is a diagram showing another specific configuration of a current pulse detection sensor and current and voltage waveforms in that case.
FIG. 8 is a circuit block diagram showing a second embodiment of the present invention.
FIG. 9 is a diagram illustrating an internal configuration example of a voltage compensation unit.
10 is a time chart of input / output signals of the delay unit of FIG. 9;
[ FIG. It is a diagram showing a specific detailed configuration of the voltage compensation unit.
[ FIG. It is an explanatory view for explaining that the relationship between the voltage command value and the delay time becomes a hyperbolic approximate curve.
[ FIG. FIG. 10 is a diagram showing a configuration for acquiring parameters of a hyperbolic approximate curve.
[ FIG. FIG. 10 is a diagram showing the relationship between a hyperbolic approximate curve and a certain time when a light emission timing signal is sent before a certain time before light emission.
[ FIG. It is a diagram showing an example of the internal configuration of a temperature compensation unit.
[ FIG. ] FIG. It is a time chart of the signal of each part of.
[ FIG. FIG. 11 is a circuit block diagram showing a modification of the second embodiment of the present invention in the case where temperature compensation is performed using a temperature sensor.
[ FIG. FIG. 10 is a circuit block diagram showing a third embodiment of the present invention.
[ FIG. FIG. 11 is a block diagram showing the configuration of an excimer laser and stepper control system to which the present invention is applied.
[ FIG. FIG. 11 is a time chart showing signals such as voltage waveforms and pulse oscillation synchronization signals at various parts of the magnetic compression circuit for illustrating the operation of the present invention.
[ FIG. It is a diagram showing another example of the internal configuration of the delay time calculation unit.
[ FIG. FIG. 10 is a diagram showing a configuration for automatically generating parameters of a hyperbolic approximate curve.
[ FIG. ] FIG. It is a time chart explaining the effect | action of.
[ FIG. FIG. 10 is a circuit block diagram showing a fourth embodiment of the present invention.
[ FIG. FIG. 10 is a circuit block diagram showing a fifth embodiment of the present invention.
[ FIG. FIG. 11 is a diagram showing a detailed configuration of the output signal delay unit 84;
[ FIG. It is a diagram showing the relationship of each time to the voltage command value V0 when the reference delay time Tds is set.
[ FIG. FIG. 11 is a diagram showing the relationship between time τ and time twait with respect to voltage command value V0.
[ FIG. ] FIG. Is a diagram when time τ is normalized with respect to time twait.
[ FIG. FIG. 11 is a diagram showing a first specific configuration of the τ / N counter 87 of the twait calculation unit 86.
[ FIG. ] FIG. It is a timing chart which shows the signal of each part of.
[ FIG. FIG. 11 is a diagram showing a second specific configuration of the τ / N counter 87 of the twait calculation unit 86.
[ FIG. ] FIG. It is a timing chart which shows the signal of each part of.
[ FIG. FIG. 11 is a diagram showing a third specific configuration of the τ / N counter 87 of the twait calculation unit 86.
[ FIG. ] FIG. It is a timing chart which shows the signal of each part of.
[ FIG. FIG. 11 is a diagram showing a fourth specific configuration of the τ / N counter 87 of the twait calculation unit 86.
[ FIG. It is a diagram showing the configuration of a conventional excimer laser and stepper control system.
[ FIG. It is a diagram showing a general magnetic compression circuit.
[ FIG. FIG. 12 is a diagram showing voltage and current waveforms at various parts of the magnetic compression circuit.
[ FIG. FIG. 10 is an explanatory view for explaining step-scan type reduced projection exposure.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... Excimer laser 2 ... Laser chamber 3 ... Pulse power supply device
4 ... Energy monitor 5 ... Laser controller 6 ... Main discharge electrode
7 ... IC 10 ... Stepper device 11 ... Stepper controller
12 ... Wafer table 20, 30 ... Current pulse detection sensor 22 ... Core
40 ... voltage compensation unit 41 ... delay time conversion unit 41a, 90 ... memory table
41b, 91 ... Read-out unit 41c ... Tdy operation unit 42, 54, 83 ... Delay unit
50, 111 ... temperature compensation unit 51 ... time difference counter 52 ... τ0 calculation unit
53 ... Delay time calculation unit 80 ... Voltage command value calculation unit
81 ... Reference delay time setting unit 82 ... Delay time calculation unit 84 ... Output signal delay unit
86... Twait calculation unit 87... Τ / N counter 100.
101 ... τ calculation unit 110 ... Temperature sensor AL0 to AL2 ... Magnetic switch
HV ... High voltage power supply SW ... Main switch C0 ~ C2a ... Capacitor
Cp ... Peaking capacitor

Claims (3)

A plurality of magnetic switches connected in series to the charging power supply and a plurality of capacitors each connected in parallel to the charging power supply constitute a multi-stage charge transfer circuit, and this multi-stage charge transfer circuit provides a current. A magnetic pulse compression circuit for compressing pulses into multiple stages;
Switching means for performing a switching operation for intermittently connecting the charging power source to the magnetic pulse compression circuit;
A laser discharge electrode connected to an output end of the magnetic pulse compression circuit;
Have
In a pulse laser that performs pulsed laser oscillation at a predetermined repetition frequency by turning on the switching means using a laser oscillation synchronization signal having a predetermined repetition frequency received from the semiconductor exposure apparatus side as a trigger,
Current detection means for detecting a rising point of a current pulse flowing through a charge transfer circuit of a predetermined stage set in advance in the magnetic pulse compression circuit;
Preliminarily set from the delay time corresponding to the voltage command value using a hyperbolic approximate relational expression between the voltage command value for the charging power source and the delay time from the time point detected by the current detection means to the actual light emission time point of the pulse laser. A delay time obtained by subtracting the predetermined time is calculated, and a delay signal obtained by delaying the detection signal output from the current detection means is output to the semiconductor exposure apparatus side as a light emission timing signal of each pulse laser oscillation. Output control means for
An emission timing signal control device for a pulse laser, comprising: A plurality of magnetic switches connected in series to the charging power supply and a plurality of capacitors each connected in parallel to the charging power supply constitute a multi-stage charge transfer circuit, and this multi-stage charge transfer circuit provides a current. A magnetic pulse compression circuit for compressing pulses into multiple stages;
Switching means for performing a switching operation for intermittently connecting the charging power source to the magnetic pulse compression circuit;
A laser discharge electrode connected to an output end of the magnetic pulse compression circuit;
Control means for outputting a voltage command value for the charging power supply;
Have
In a pulse laser that performs pulse laser oscillation at a predetermined repetition frequency by turning on the switching means using a pulse oscillation synchronization signal having a predetermined repetition frequency received from the semiconductor exposure apparatus side as a trigger,
A reference delay time setting means in which a predetermined reference delay time that is equal to or greater than a maximum value of a variation range of an actual light emission delay time from when the switching means is turned on until laser oscillation is started;
A delay time calculation means for obtaining a difference between the set reference delay time and the actual light emission delay time of the pulse oscillation corresponding to the voltage command value output from the control means for each pulse oscillation;
A delay means for delaying the pulse oscillation synchronization signal received from the semiconductor exposure apparatus by the difference time calculated by the delay time calculation means and outputting the delayed oscillation signal to the switching means;
Current detection means for detecting a rising point of a current pulse flowing through a charge transfer circuit of a predetermined stage set in advance in the magnetic pulse compression circuit;
Preliminarily set from the delay time corresponding to the voltage command value using a hyperbolic approximate relational expression between the voltage command value for the charging power source and the delay time from the time point detected by the current detection means to the actual light emission time point of the pulse laser. A delay time obtained by subtracting the predetermined time is calculated, and a delay signal obtained by delaying the detection signal output from the current detection means is output to the semiconductor exposure apparatus side as a light emission timing signal of each pulse laser oscillation. Output control means for
An emission timing signal control device for a pulse laser, comprising: The output control means includes
A counter that normalizes the difference obtained by the delay time calculating means to the maximum value of the waiting delay time and measures the waiting delay time using a hyperbolic approximate relationship between the difference and the normalized waiting delay time; The pulse laser emission timing signal control device according to claim 2.

Images