JP2008022026A - Reliable, modular, production-quality narrow-band high repetition-rate f2 laser - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a reliable, modular, production-quality narrow-band high repetition-rate F<SB>2</SB>laser. <P>SOLUTION: A reliable, modular, production-quality excimer laser capable of producing 10 mJ laser pulses in the range of 1,000 Hz to 2,000 Hz or greater is provided. Replaceable modules include a laser chamber, a pulse power system including three modules, an optical resonator comprised of a line narrowing module and an output coupler module, a wavemeter module, an electrical control module, a cooling water module, and a gas control module. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  This application was filed on March 1, 1998 with US Patent Application Serial No. 09 / 271,041, Reliable, Modular, Production Quality Narrow-Band High Rep Rate ArF Excimer Laser, filed March 17, 1999. U.S. Patent Application Serial No. 09 / 041,474 Reliable, Modular, Production Quality Narrow-Band KrF Excimer Laser, and U.S. Patent Application Serial No. 08 / 995,832 Excimer Laser Having Pulse Power Supply with Fine, filed December 22, 1997 Digital Regulation, U.S. patent application serial number 08 / 896,384 filed July 18, 1997, Wavelength Reference for Excimer Laser, and U.S. patent application serial number 08 / 939,611 filed September 29, 1997, protective Overcoat for Replicated Diffraction Gratings, U.S. patent application serial number 08 / 947,474 filed March 4, 1998, Pulse Energy Control for Excimer Laser, and U.S. patent application serial number 09, filed May 20, 1998 / 082,139 Narrow Band Excimer Laser with Gas Additive and U.S. Patent Application Serial No. 09 / 157,067 filed September 18, 1998 Reliable, Modular, Production Quality Narrow Band High Rep Rate Excimer Laser, filed September 28, 1998 U.S. Patent Application Serial No. 09 / 162,341 Line Narrowing Apparatus with High Transparency Prism Beam Expander, U.S. Patent Application Serial No. 09 / 165,593 filed on October 2, 1998, Wavelength System for an Excimer Laser, December 1998 U.S. Patent Application Serial No. 09 / 206,526 filed on May 7, Wavelength Reference for Laser, and U.S. Patent Application Serial No. 09 / 211,825 filed December 15, 1998 High Pulse Rate Power System with Resonant Power Supply And US Patent Application Serial No. 09 / 217,340, Durable Etalon Based Output Coupler, filed December 21, 1998, all of which are incorporated herein by reference. The present invention relates to lasers, and more particularly to narrowband ArF excimer lasers.

KrF excimer lasers Krypton fluoride (KrF) excimer lasers are now a useful optical source for the integrated circuit lithography industry. A KrF laser produces a laser beam with a narrow band wavelength of about 248 nm and is used to make integrated circuits with dimensions as small as about 180 nm. Argon fluoride (ArF) excimer lasers are very similar to KrF lasers. The main difference is the laser gas mixing and the shorter wavelength of the output beam. Basically, argon replaces krypton, so that the wavelength of the output beam is 193 nm. This further reduces the size of the integrated circuit to about 120 nm. The F ₂ laser has long been recognized as a successor to KrF and ArF in the integrated circuit lithography industry because the pattern resolution can be substantially improved by the F ₂ beam at 157 nm. These F ₂ lasers are very similar to the KrF and ArF excimer lasers with some modifications and can be replaced with prior art KrF or ArF lasers to operate as F ₂ lasers. A typical prior art KrF excimer laser used in the manufacture of integrated circuits is shown in FIGS. A cross section of the laser chamber of this prior art laser is shown in FIG. A pulse power system 2 powered by a high voltage power supply 3 provides electrical pulses to electrodes 6 arranged in the discharge chamber 8. A typical state of the art lithography laser operates at a pulse frequency of about 1000 Hz with a pulse energy of about 10 mJ / pulse. At about 3 atmospheres (for a KrF laser, about 0.1% fluorine, 1.3% krypton, the rest being neon, which acts as a buffer gas), the laser gas can pass through the space between the electrodes at a rate of about 1000 inches / second Circulate through. This is done with a vertical blower 10 located in the laser discharge chamber. The laser gas is cooled by a heat exchanger 11 also disposed in the chamber and a cooling plate (not shown) attached to the outside of the chamber. The natural bandwidth of the excimer laser is narrowed by the line narrowing module 18. Commercially available excimer laser systems typically consist of various modules that can be quickly replaced without interfering with system downtime. The main modules include:
Laser chamber module Pulse power system with high voltage power supply module Commutator module and high voltage compression head module Output coupler module Line narrowing module Wavemeter module Computer control module Gas control module Cooling water module

  The electrode 6 includes a cathode 6A and an anode 6B. The anode 6B is supported in this prior art embodiment by an anode support bar 44 shown in the cross section of FIG. The flow is clockwise toward. One corner and one end of the anode support bar 44 serve as guide vanes for forcing air from the blower 10 to flow between the electrodes 6A and 6B. Other guide vanes in this prior art laser are shown at 46, 48 and 50. A perforated current return plate 52 helps ground the anode 6B to the metal structure of the chamber 8. The plate is perforated with large holes (not shown in FIG. 3) located in the laser gas flow path, and the current return plate does not substantially affect the gas flow. A peak capacitor consisting of an array of individual capacitors 19 is charged by the pulse power system 2 before each pulse. While the voltage builds up to the peak capacitor, the two play-on devices 56 weakly ionize the laser gas between the electrodes 6A and 6B, and when the capacitor charge reaches approximately 16,000 volts, the discharge of the electrode becomes an excimer. Generated to generate a laser pulse. For each next pulse, the gas flow between the electrodes of about 1 inch / millisecond produced by the blower 10 is sufficient to provide fresh laser gas between the electrodes just for the next pulse that occurs after 1 millisecond. is there.

  In a typical lithographic excimer laser, the feedback control system measures the laser energy of each pulse, determines the degree of deviation from the desired pulse energy, and powers the next pulse so that it is close to the desired energy. Send a signal to the controller to adjust the voltage. In prior art systems, this feedback signal was an analog signal and was subject to noise created by the laser environment. This noise can provide an incorrect power supply voltage and can subsequently cause an increase in the output laser pulse energy.

These excimer lasers are typically required to operate continuously for 24 hours a day, 7 days a week, months, with only a short outage scheduled for maintenance. One problem experienced with these prior art lasers is excessive wear and occasional failure of the blower bearing.
There is a need in the integrated circuit industry for a modular, reliable production line high quality F ₂ laser that can provide integrated circuit resolution not available with KrF and ArF lasers.

The present invention provides high reliability, modular, production, capable of producing laser pulses with pulse energies greater than 10 mJ with half-widths of about 1 pm or less at repetition rates in the range of 1000 to 2000 Hz or higher. Provide high quality F ₂ excimer laser. Preferred embodiments of the present invention may operate in the 1000-4000 Hz range with pulse energy in the range of 10-5 mJ with power output in the range of 10-40 watts. In using this laser as an illumination source, a stepper or scanner device produces an integrated circuit resolution of 0.1 μm or less. Replaceable modules include laser chambers and modular pulse power systems.

  Significant improvements over prior art excimer lasers have been provided for pulse power units to create faster charges. These improvements generate a high voltage pulse from an increased capacity high voltage power supply and a capacitor charged by the high voltage power supply and have a secondary winding consisting of a single four segment stainless steel rod. And an improved commutator module that amplifies the pulse voltage about 23 times with a fast voltage transformer. The new design for the compression head saturable inductor (referred to herein as the “pot and pan” design) significantly reduces the amount of transformer oil required, effectively eliminating the possibility of oil leaks that previously caused accidents. .

Improvements in laser chambers that allow higher pulse frequencies and improved performance include the use of a single play-on device tube.
In the preferred embodiment, the laser is tuned to the F ₂ 157.6 nm line using a set of two external prisms. In a second preferred embodiment, the laser operates in a broad band and the 157.6 nm line is selected outside the resonant cavity. In a third preferred embodiment, a line width of 0.2 pm is provided using an implanted seed.

Other embodiments of the present invention include ceramic bearings. Optionally, magnetic bearings can be utilized. Bearing reaction forces may be reduced by providing an aerodynamic curve of the anode support bar. Other improvements include the use of acoustic baffles for laser chambers that generate destructive acoustic shock waves.
It is preferred that the internal cavity beam line and the output beam line be completely sealed and purged with nitrogen.

First Preferred Embodiment A preferred embodiment of the present invention will be described with reference to the drawings.

Modular Laser Design A front view of a preferred embodiment of the present invention is shown in FIG. This figure highlights the modular nature of a particular invention that allows for very rapid replacement of modules for repair, replacement and maintenance. The main features of this embodiment are listed below in correspondence with the reference numbers shown in FIG.
DESCRIPTION OF SYMBOLS 201 Laser shielding 202 Gas module 203 Cooling water supply module 204 AC / DC power distribution module 205 Control module 206 Line narrow band module 207 Compression head 208 High voltage pulse power supply module 209 Commutator module 210 for pulse power supply 210 Fluoride metal trap 211 Laser chamber 213 Wavelength module 214 Automatic shutter 216 Output coupler 217 Blower motor 218 Metal fluoride trap power supply 219 Status lamp 220 24 volt power supply 221 Chamber window 222 Gas control flexible connection 224 Vent box

Preferred Embodiment The preferred embodiment of the present invention is an improved version of the laser described in FIGS. This preferred embodiment includes the following improvements:
1) The combination of the prior art of two tube play-on devices so that the single tube large play-on device provides better and better play-on and improved laser gas flow between the electrodes. Replace
2) A silicon-free fan blade is a one-piece machined blade;
3) The individual physics pulse power system has been modified to create faster rise times, providing more harmonized pulses and providing improved laser efficiency at higher voltages;
4) More precise control of the charge voltage of the pulse power supply device;
5) a computer control program with a new algorithm that provides improved control of pulse energy and burst energy;
6) The electrode space was reduced to 10 mm.

Chamber improvement
Single Play-On Tubes As shown in FIG. 6, a single large play-on device 56A replaces the two play-on devices tubes 56 shown in FIG. A single tube play-on device is manufactured in accordance with the description of US Pat. No. 5,719,896 issued Feb. 17, 1998, which is incorporated herein by reference. Applicants have found that not only one play-on device is sufficient, but very surprisingly provides improved performance over the two play-on device designs. In this embodiment, the play-on device is located upstream of the electrode. Applicants have determined that one tube play-on device improves the stability between pulses by providing improved spatial stability of the discharge.

  Referring now to FIG. 7, this play-on device utilizes an integrated tube design having a bushing element 180 with an anti-tracking groove 170 incorporated herein as an integral component of the tube. The OD of the rod portion 145 and the bushing portion 180 of the play-on device is 1/2 inch. The inner conductor rod 146 has a diameter of 7/37 inches, and the connecting wire extending through the bushing portion for connection to the ground is about 1/16 inch in diameter. The previous playonizer tube design utilized a two-diameter design where the rod portion was about 1/4 inch in diameter and the bushing diameter was about 1 inch. This required a bonding process for bonding the bushing component with the tube component for manufacturing purposes. A thinner tube design with a constant diameter is contrary to conventional design rules and can be expected to reduce ionization to make the volume smaller. In most designs, the tube thickness depends on the insulation strength of the selected material. One skilled in the art recognizes that the prior art play-on tube design technique is to select a material with high insulation strength and determine the wall thickness to match this capacity. For example, sapphire materials are known to have a dielectric strength ranging from 1200 volts / mil to 1700 volts / mil. Therefore, an insulation thickness of 0.035 inches provides a safety factor of 2 if the laser operates at 25 kV. This design produces a low capacity, however, it has been found that the actual effect of this reduced capacity of laser operation is negligible, and the measurement of electrode gap geometric radiation is surprisingly increased. Due to the constant diameter, thinner tube wall, integrated bushing design, a single piece of material is processed to provide an anti-tracking groove 170. Even if applicants continue to use indication purity materials, because of the single piece construction, it is not necessary to use ultra high purity (ie, 99.9%) polycrystalline translucent aluminum oxide ceramic. In order to artificially create an integral relationship between the bushing 180 and the tube 145, it is not necessary to perform difficult surface polishing of the tube geometry prepared for diffusion bonding. In fact, it has been found that high purity is not as important a feature as the porosity of the material. It has been found that the higher the porosity, the lower the insulation strength. As a result, commercial grade ceramics preferably have a purity of at least 99.8%, are low porosity, such as material number AD-998E as manufactured by Coors Ceramics Company, and have an insulation of 300 volts / mil. Used in strength. A bushing 180 having an anti-tracking groove 170 disposed therein as previously described operates to prevent axial high voltage tracking along the surface of the tube from the cathode to the ground 160.

  As mentioned above, Applicants have discovered that a single play-on device works dramatically better than two play-on devices, and as explained above, the first preferred embodiment is upstream of the electrodes. Place a single play-on device. Applicants have also experimented with a single play-on device located downstream and at a given blower speed, this arrangement is substantially better than the one placed upstream in the two-tube arrangement. Found to create stability.

High Efficiency Chamber Improvements have been made to the chamber to improve the efficiency of the laser. A single piece cathode insulator 55A comprising alumina, Al ₂ O ₃ insulates the cathode from the upper chamber structure as shown in FIG. 6A. In prior art designs, eight separate insulators were required to avoid insulator cracking due to the thermal expansion stress of the insulator. This significant improvement allows the chamber head portion to significantly reduce the distance between the cathode 83 of the peak capacity 82. The individual capacitors 54A forming the peak capacitance array 82 move closer to the cathode and horizontally compared to the prior art. To reduce the difference in thermal expansion between the single piece insulator and the chamber structure, the upper chamber 8A is made from ASTM A3C steel with a coefficient of thermal expansion closer to Al ₂ O ₃ than aluminum. Although the bottom 8B of the chamber 8 is aluminum, Applicants understand that the difference in thermal expansion between ASTM A3C steel and aluminum is not a problem. Both steel and aluminum parts are nickel coated.
A conventional cathode for a commercial lithographic laser is typically supported by a cathode support bar 53 as shown in FIG. In this preferred embodiment, the cathode support bar is eliminated and the cathode 83 is slightly thinned and mounted directly on the single piece insulator 55A. The cathode 83 is connected to the high voltage side 82A of the peak capacitor 82 by 15 feeds via a rod 83A and a connection nut 83B. In a preferred embodiment, the new anode support bar 84A is substantially larger than the prior art anode support bar and includes fins 84B disposed in the gas flow region. Both of these features minimize anode temperature fluctuations.

Metal Seal Applicants have discovered that prior art elastomer seals react with fluorine gas to produce laser gas contaminants that degrade laser performance. The preferred embodiment of the present invention uses an all metal seal to seal the laser chamber. A preferred metal seal is a tinned Inconel 1718 seal.

Monel Current Return and Blades Applicants have also discovered that stainless steel elements also react with fluorine to produce laser gas contaminants. Therefore, in this preferred embodiment, the prior art stainless steel current return structure and gas flow vanes are replaced with monel current return 250 and monel flow vanes 252 and 254.

Acoustic baffle Applicants have reflected a significant cause of distortion in the quality of the laser beam produced by narrowband excimer lasers operating at 2000 Hz or higher back from the chamber structure elements back to the space between the electrodes. An acoustic shock wave generated by the discharge of one pulse, which distorts the laser pulse of the next pulse that occurs after 0.5 milliseconds. The embodiment described here (see FIG. 6A) substantially minimizes this effect by acoustic baffles 63A and 64A that are angled and grooved on both sides of the laser chamber. These baffles absorb some of the acoustic energy and reflect some of the acoustic energy down towards the lower region of the laser chamber away from the electrodes. In this preferred embodiment, the baffle is machined into a metal structure with 0.1 mil wide, 0.3 mil deep, 0.2 mil grooves, and the 0.3 mil deep grooves are baffled in FIG. 6A. 63. These baffles have been shown by actual tests to substantially reduce pulse quality distortion caused by acoustic shock waves.

  Applicants have also discovered that the effects of acoustic shock are minimized by reducing streamers in the discharge. In fact, in the preferred embodiment of the present invention, the changes were made with a chamber head (discussed previously) and a new play-on device designed to reduce acoustic shock so that no acoustic baffle is required.

Fan Improvements This preferred embodiment of the present invention includes significant improvements in prior art gas circulators and greatly improves laser performance. These improvements are the construction of a blower blade structure without brazing. An asymmetric blade arrangement that greatly reduces the effects of resonance and an improved bearing.

Silicon-free fan blade structure Applicants have found that the brazing material commonly used in blower blade structures is the main source of SiF _{6 in} the laser chamber. This gas significantly degrades the laser performance of the KrF laser, but is also generally terrible for ArF and F ₂ lasers. The applicant has identified four solutions to this problem. First, the blade structure is partially processed from a solid block of material (in this case aluminum). Another solution is to partly cast the blade structure. The segments are then welded together using electron beam welding without adding new material. It is also suitable for manufacturing the blade structure by bonding the blade to the frame structure, in which case the bonding is done by electron beam welding instead of the prior art brazing process. A fourth method is to join the blade to the frame structure using a soldering process using silicon free solder. Aluminum 6061 is used as the base material for all component parts. These parts are then copper plated prior to the soldering process. When all parts are assembled, the fans are then soldered together in a vacuum furnace, typically with 91% tin (Sn) and 9% zinc (Zn), using low temperature solder. This solder is selected for its lack of silicon and its ability to work with copper-plated aluminum. The assembled and soldered fan is then nickel plated. This manufacturing method produces a non-silicon fan that is inexpensive to manufacture.

Reduction of resonance effects The blower blade structure of the prior art has a blower perpendicular to the 23 longitudinal blades. These blades are mounted symmetrically around the structure. The substantial resonance effect is measured in terms of both fan parameters and actual laser performance. The perturbation of the laser beam is shown corresponding to the acoustic wave at 23 times the rotational frequency of the fan. The adverse effect on bearing performance is also measured corresponding to 23 times the fan rotational frequency.

  Improvements in fan structure design require an asymmetric blade arrangement as shown in FIG. 14A. An alternative embodiment as shown in FIG. 14B where the fan blade structure is formed of 16 individually machined or cart segments with each segment comprising 23 blades is 360 ° / (15 × 23 ), Or by rotating each segment by about 1 ° relative to adjacent segments. Another improvement that can be made relatively easy in the casting approach or processing for fan blade structure manufacture is to form the blade in the airfoil as shown at 320 in FIG. 14C. The prior art blade is stamped and two cross sections of the stamped blade are shown at 314 for comparison. The direction of rotation indicated at 318 and 330 represents the blade structure circumference. Whereas conventional blades are of uniform thickness, airfoil blades have a tear profile that includes a peripheral lead end, a dense center, and a tapered trace end.

Bearing Improvements Embodiments of the present invention can use one of two other improved bearings over the prior art.
Ceramic Bearing A preferred embodiment of the present invention includes a ceramic bearing. A preferred ceramic bearing is silicon nitride lubricated with a chemically synthesized lubricating oil, preferably perfluoropolyalkylether (PFPE). These bearings provide a substantially longer life compared to prior art excimer laser fan bearings. Furthermore, neither the bearing nor the lubricating oil is significantly affected by the highly reactive fluorine gas.

Magnetic Bearing Another preferred embodiment of the present invention is made of a magnetic bearing that supports a fan structure as shown in FIG. In this embodiment, the shaft 130 supporting the fan blade structure 146 is in turn supported by an active magnetic bearing system and driven by a brushless DC motor 130, where the rotor 129 of the motor and the rotor 128 of at least two bearings are lasers. Sealed within the gas environment of the cavity, the motor starter 140 and the coil 126 of the magnetic bearing magnet are located outside the gas environment. This preferred bearing design also includes an active magnetic thrust bearing 124 that also has a coil located outside the gas environment.

Aerodynamic Anode Support Bar As shown in FIG. 3, prior art gas flow from the blower 10 is biased to flow between the electrodes 6A and 6B by the anode support bar 44. However, Applicants have discovered that the prior art design of the support bar 44 as shown in FIG. 3 moves to the blower bearing as a result of chamber vibration, creating a substantial aerodynamic reaction force on the blower. did. Applicants have speculated that these vibrational forces are responsible for blower bearing friction and can lead to occasional bearing failures. Applicants have tested other designs, some shown in FIGS. 12A-12E, all of which reduce the aerodynamic reaction force by distributing over time, the reaction force being near the end of the support bar 44 Each time the blade passes through. One of Applicants' preferred anode support bar designs is shown at 84A in FIG. 6A. This design has a substantially larger mass that minimizes anode temperature savings. The total amount of anode and anode support bar is about 3.4 kg. This design also includes fins 84B that provide additional cooling to the anode. Applicants' tests have shown that both the acoustic baffle and the aerodynamic anode support bar slightly reduce the gas flow so that the gas flow is limited, and the use of these two improvements involves a trade-off analysis. Two improvements on these reasons are shown in FIG. 6A but not in FIG.

Pulse power system
Functional Description of Four Pulse Power Modules A preferred pulse power system is manufactured in four separate modules as shown in FIGS. 8A and 8B, each being an important part of an excimer laser system, each of which is a part failure or normal It can be quickly replaced upon occurrence during a preventive maintenance program. These modules are a high voltage power supply module 20, a commutator module 40, a compression head module 60, and a laser chamber module 80 designed by the applicant.

High Voltage Power Supply Module The high voltage power supply module 20 includes a 300 volt rectifier 22 to convert 208 volt three phase plant power from the source 10 to 300 volt DC. The inverter 24 converts the output of the rectifier 22 into a high frequency 300 volt pulse in the range of 100 kHz to 200 kHz. The frequency and period of the inverter 24 are controlled by the HV power control board 21 to provide the ultimate output pulse energy course rule of the system. The output of inverter 24 is stepped up to about 1200 volts in step-up transformer 26. The output of the transformer 26 is converted to 1200 volts DC by a rectifier 28 having a standard bridge rectifier circuit 30 and a filter capacitor 32. The DC electrical energy from the circuit 30 causes 8.1 μFC ₀ to charge the capacitor 42 in the commutator module 40 as indicated by the HV power control board 21 that controls the operation of the inverter 24 as shown in FIG. 8A. Charge. The set point in the HV power supply control board 21 is set by the laser system control board 100.

  The reader should note that in this embodiment as shown in FIG. 8A, pulse energy control for the laser system is provided by the power supply module 20. The electrical circuit in the commutator 40 and compression head 60 forms electrical pulses at a rate of 2000 times / second, amplifies the pulse voltage, and compresses the charge capacitor 42 by the power supply module 20 in the time between pulses. It is simply useful for using stored electrical energy. As an example of this control, FIG. 8A shows that the processor 102 in the control board 100 controls the power supply to provide exactly 700 volts to the charging capacitor 42 that is isolated from downstream circuitry by the solid state physical switch 46 during the charging cycle. Indicates. The electrical circuits in the commutator 40 and compression head 60 near the closure of the switch 46 are required to provide the next laser pulse with the exact energy that needs to be determined by the processor 102 in the control board 100. And 84, very quickly and automatically converts the electrical energy stored in capacitor 42 into an accurate electrical discharge pulse.

Commutator Module The commutator module 40 has a C ₀ charge capacitor 42, which in this embodiment is a bank of capacitors connected in parallel to provide a total capacitance of 8.1 μF. The voltage divider 44 is formed into an electrical pulse that, when compressed and amplified by the commutator 40 and compression head 60, produces a voltage (“control voltage”) that produces the desired discharge voltage at the peak capacitor 82 and electrodes 83 and 84. The feedback voltage signal is provided to the HV power supply control board 21 used by the control board 21 to limit the charge of the capacitor 42.

In this embodiment (designed to provide electrical pulses in the range of about 3 joules and 16,000 volts at a pulse frequency of 2000 Hz pulses / second), about 250 microseconds (as shown in FIG. 8F1). Is required with respect to the power supply 20 to charge the charging capacitor 42 to 800 volts. Therefore, when the signal from the commutator control board 41 closes the solid state physical switch 44 showing a very early step of converting 3 joules of electrical energy stored in the charging capacitor C ₀ to 16,000 volts, the desired It is fully charged and stable with voltage. For this embodiment, the solid state physical switch 46 is an IGBT switch, but other switch technologies such as SCRs, GTOs, MCTs, etc. can also be used. The 600 nH charge inductor 48 is in series with the solid state physical switch 46 to temporarily limit the current through the switch 46 while the switch 46 is closed to discharge the C ₀ charging capacitor 42.

Pulse Generation Stage The first stage of high voltage pulse power generation is a pulse generation stage 50. To generate the pulse, the charge on the charging capacitor 42 is switched to the C ₁ 8.5 μF capacitor 52 at approximately 5 μs as shown in FIG. 8F2 by closing the IGBT switch 46.

First Stage of Compression Saturable inductor 54 initially holds the voltage stored in capacitor 52 off and then for a first stage of compression 61 for about 550 ns, as shown in FIG. In the transition time, the charge state can be transferred from the capacitor 52 to the C _p-1 capacitor 62 through the 1:23 step-up pulse transformer 56.

The design of the pulse transformer 56 is shown below. The pulse transformer makes a very effective transition from a 700 volt 17,500 ampere 550 ns pulse rate to a 16,100 volt 760 ampere 55 ns pulse and stores it very temporarily in the C _p-1 capacitor bank 62 in the compression head module 60. Is done.

Compression head module The compression head module 60 further compresses the pulses.
Second stage of compression (with approximately 125 nH saturable inductance) L _p-1 saturable inductor 64 holds the voltage off to 16.5 nFC _p-1 capacitor bank 62 for approximately 550 ns, and then the laser chamber 16.5 nFC _p peak capacitor 82 placed at the top of 80 (about 100 ns) allows charge to flow through C _p-1 and is electrically connected in parallel with electrodes 83 and 84 and play-on device 56A. The This conversion of a 550 ns long pulse to a 100 ns long pulse to charge the C _p peak capacitor 82 can create the second and final stages of compression as shown at 65 in FIG. 8A.

Laser chamber module After about 100 ns, the charge begins to flow to the peak capacitor 82 attached to the top as part of the laser chamber module 80, and the voltage of the peak capacitor 82 reaches about 14,000 volts, causing a discharge between the electrodes. Start. During the last approximately 50 ns discharge, lasing occurs in the optical resonance chamber of the excimer laser. The optical resonance chamber is configured by a line selection package 86 that is included in this example by an output coupler 88, along with two prism wavelength selectors and R-max mirrors shown as 86 in FIG. 8A. This laser pulse has a narrow band and a repetition rate of 20 to 50 ns, a pulse of 157 nm of about 10 mJ, and 2000 pulses / second. The pulses constitute a laser beam 90, and the pulses of the beam are monitored by a photodiode 92 as shown all in FIG. 8A.

Pulse Energy Control The signal from the photodiode 92 is forwarded to the processor 102 of the control board 100, which sets the energy signal and command voltage for the next and / or further pulses (pulse energy control). Preferably other historical pulse energy data (as discussed in a later section named Algorithm) is used. In a preferred embodiment, the laser operates in a series of short bursts (such as a burst of 100 pulses 0.5 seconds at 2000 Hz separated by a dead time of about 0.1 second), and the processor 102 of the control board 100 can All previous pulses in the burst along with other historical pulse profile data to select the control voltage for the next pulse so as to minimize the energy change of the It is programmed with a specific algorithm that uses the closest pulse energy signal together with the pulse energy signal. This calculation is performed by the processor 102 of the control board 100 using this algorithm for approximately 35 μs. The laser pulse occurs about 5 μS following the T ₀ firing of the IGBT switch 46 shown in FIG. 8F3, and about 20 μs is required to correct the laser pulse energy data. (The start of firing of switch 46 is referred to as T _0. ) Thus, the new control voltage value is thus approximately 70 milliseconds after firing of IGBT switch 46 with respect to the previous pulse (2000 Hz firing period is 500 μs). Prepared (as shown in FIG. 8F1). Features of the energy control algorithm are described below and are described in more detail in US patent application Ser. No. 09 / 034,870, which is incorporated herein by reference.

Energy recovery (Recovery)
This preferred embodiment provides an electrical circuit that recovers excess energy to the charge capacitor 42 from the previous pulse. This circuit substantially reduces unwanted energy and virtually eliminates it after ringing into the laser chamber 80.

The energy recovery circuit 57 includes an energy recovery inductor 58 and an energy recovery diode 59 and is connected in series across the C ₀ charge capacitor 42 as shown in FIG. 8B. Since the impedance of the pulse power system does not precisely match that of the chamber and the chamber impedance varies in various orders during the pulse discharge, negative behavior "reflection" is from the chamber towards the front end of the pulse generation system. It is generated from the main pulse that spreads back. After excess energy is spread back through the compression head 60 and commutator 40, the switch 46 is opened by the controller for removal of the trigger signal. The energy recovery circuit 57 is resonant freewheeling (half of the ringing of the L-C circuit is made from the jersey capacitor 42 and the energy recovery inductor 58) as fixed against the reversal of current in the inductor 58 by the diode 59. A plurality of reflections that generate a negative voltage on the charge capacitor 42 are inverted through the ring. The result obtained is that substantially all of the energy reflected from the chamber 80 is recovered from each pulse and stored in the charge capacitor 42 as a positive charge ready to be utilized for the next pulse. . 8F1, 2 and 3 are timeline charts showing the charge of the capacitors C ₀ , C ₁ , C _p-1 and C _p . Chart shows the process of energy recovery of C _0.

Magnetic Switch Bias To fully take advantage of the full BH curve swing of the magnetic material used in the saturable inductor, the DC bias current is set so that if each inductor is then saturated in reverse, the pulse is switched to switch 46. It is initialized by closing.

  In the case of commutator saturable inductors 48 and 54, this is achieved by providing a bias current flow of approximately 15A back through the inductor (compared to the normal pulse current flow direction). . This bias current is provided by the bias current source 120 via the insulated inductor LB1. The actual current flow is biased as indicated by arrow B1 through the commutator ground, through the primary winding of the pulse transformer, through the saturable inductor 54, and through the saturable inductor 48. Moving from the power source through the insulated inductor LB1 back to the current source 120.

  In the case of a compression head saturable inductor, a bias current B2 of approximately 5A is provided from the second bias current source 126 via the isolated inductor LB2. At the compression head, the current splits and the majority of B2-1 travels through the saturable inductor Lp-1 64 and back through the isolation inductor LB3 back to the second bias current source 126. A smaller part of the current B2-2 is a bias resistor that returns to the second bias current source 126 via the HV cable connected to the compression head 60 and the commutator 40, via the pulse transformer secondary winding to ground. To move back through. This second small current is used to bias the pulse transformer, which is also reset for pulse operation. The amount of current that splits in half is determined by the resistance in each path and is deliberately adjusted so that each path receives the correct amount of bias current.

Direction of Current In this embodiment, we pass pulse energy from the standard three-phase power supply 10 through the system to the electrode and, as the “forward direction”, across the electrode 84 to ground, this direction being the forward direction. When we refer to an electrical component such as a saturable inductor that is forward conducting, we mean that it is biased to saturate to guide "pulse energy" in the direction towards the electrode . When it is reverse conducting, it is biased to saturate to direct energy away from the electrode towards the charge capacitor. The actual direction of current (or electron flow) through the system depends on where you are in the system. In order to eliminate this as a possible cause of confusion, the current direction will now be described.

Referring to FIGS. 8A and 8B, in this preferred embodiment, C ₀ capacitor 42 is charged to (for example) +700 volts, and when switch 46 is closed, current is directed through inductor 48 in a direction toward C ₁ capacitor 52. From the capacitor 42 (which means that electrons actually flow in the opposite direction). Similarly, current flows from the C ₁ capacitor 52 through the primary side of the pulse transformer 56 toward ground. Therefore, the directions of current and pulse energy are the same from the charge capacitor 42 to the pulse transformer 56. As will be described in the item after “Pulse Transformer”, the current in both the primary and secondary loops of the pulse transformer 56 goes to ground. As a result, during the first part of the discharge (representing the main part of the discharge (typically about 80%)), the current between the pulse transformer 56 and the electrode is away from the electrode towards the transformer 56. Direction. Therefore, during the main discharge, the direction of electron flow is from ground to the Cp-1 capacitor 62 temporarily via the inductor 64 and temporarily to the Cp capacitor 82 via the secondary pulse transformer 56. The direction is such that it returns to ground via 81, via electrode 84 (referred to as discharge cathode), via electrode 83, and via electrode 83. Therefore, during the main discharge, electrons flow in the same direction as the pulse energy between the pulse transformer 56 and the electrodes 84 and 83. The main part of the discharge flows immediately, the current and the electron flow are reversed, and the reverse electron flow passes through the grounded electrode 84 to the electrode 83 via the discharge space between the electrodes. The current flows from the ground so as to return from 56 to the ground through the circuit. The path of reverse electron flow through the transformer 56 is such that the pulse transformer 56 will eventually charge (in the same direction as the current of the main pulse) to negatively charge C ₀ as qualitatively shown in FIG. 8F2. It creates a current in the “primary” loop of transformer 56 with the flow of electrons from ground through the “primary” side. The negative charge of C ₀ is reversed as shown in FIG. 8F2 as described in the item named “energy recovery” above.

Detailed description of pulse power components
Power Supply A more detailed circuit diagram of the power supply portion of the preferred embodiment is shown in FIG. 8C. As shown in FIG. 8C, the rectifier 22 is a 6-pulse phase controlled rectifier with pulses from 150 volts to -150 volts DC output. The inverter 24 is actually three inverters 24A, 24B and 24C. Inverters 24B and 24C are turned off when the voltage on the 8.1 μFC ₀ charge capacitor 42 is 50 volts below the command voltage, and inverter 24A is turned off when the voltage on C ₀ 42 slightly exceeds the command voltage. This procedure reduces the charge rate near the end of the charge. Setup transformers 26A, 26B and 26C each operate at 7 kw and convert the voltage to 1200 volts AC.

Three bridge rectifier circuits 30A, 30B and 30C are shown. The HV power supply control board 21 converts the 12-bit digital command into an analog signal and compares it with the feedback signal 45 from the C ₀ voltage monitor 44. When the feedback voltage exceeds the command voltage, the inverter 24A is turned off as described above, the Q2 switch 34 is closed to dissipate the energy stored in the power supply, and no further energy remains in the power supply. opened 36, the voltage of C ₀ closes Q1 bleed switch 38 to drop to narrow the voltage of C042 to equal the command voltage. At that time, Q1 is opened.

Commutator and Compression Head The major components of commutator 40 and compression head 60 are shown in FIGS. 8A and 8B and discussed above with respect to system operation. In this section, we will describe the details of commutator manufacturing.

Solid State Physical Switch In this preferred embodiment, the solid state physical switch 46 is manufactured by Powerex, Inc., an office in Youngwood, PA. P / N CM 1000 HA-28H IGBT switch provided by.

Inductors Inductors 48, 54 and 64 are saturable inductors similar to those described in US Pat. Nos. 5,448,580 and 5,315,611. A plan view and a cross-sectional view of a preferred saturable inductor design are shown in FIGS. 8G1 and 8G2, respectively. In the inductor of this embodiment, a flux that excludes metal pieces such as 301, 302, 303, and 304 is added as shown in FIG. 8G2 to reduce leakage flux in the inductor. The input of current to this inductor is a screw connection at 305 to the bus that is also connected to capacitor 62. The current makes a 4.5 loop through the vertical conductor. From position 305, the current travels down a central large diameter conductor named 1A, going up 6 surrounding small conductors named 1B, going down 2A, going up 2B, going down all flux exclusion elements Go up 3B, go down 3A, go up 4B, go down 4A, the current goes to location 306. A pot such as housing 64A serves as a high voltage current lead. The saturable conductor “lid” 64B includes an electrically insulating material such as Teflon. In prior art pulse power systems, oil leakage from oil-insulated electrical components has been a problem. In this preferred embodiment, the oil insulation component is limited to a saturable inductor and the oil is contained in a pot-like oil that includes a metal housing 64A that is a high voltage connection output lead as described above. All seal connections are placed above the oil level to substantially eliminate the possibility of oil leakage. For example, the lowest seal in inductor 64 is shown at 308 in FIG. 8G2. Since the flux removal metal component is in the middle of the current path through the inductor, the voltage can be lowered in keeping the space between the flux removal metal part and the metal rod of the other turn safely off. Fins 370 are provided to increase heat removal.

Capacitors Capacitor banks 42, 52 and 62 include all banks of capacitors connected in parallel that are available on the market. These capacitors are available from suppliers such as Murata with offices in Smyrna, Georgia. Applicants' preferred method of connecting capacitors and inductors is the positive and negative terminals of certain printed circuit boards having thick nickel-coated copper leads in a manner similar to that described in US Pat. No. 5,448,580. Against them with solder or bolts.

Pulse Transformer The pulse transformer 56 is also similar to the pulse transformer described in US Pat. Nos. 5,448,580 and 5,313,481, except that the pulse transformer of this embodiment comprises a secondary winding and 23 separate primary windings. It only turns once inside. A drawing of the pulse transformer 56 is shown in FIG. 8D. Each of the 23 primary windings is bolted to the positive and negative terminals of the printed circuit board 56B as shown along the bottom edge of FIG. 8D (each with a threaded bolt hole). An aluminum spool 56A having two flanges (with a flat end). Insulator 56C separates the positive terminal of each spool from the negative terminal of the adjacent spool. Between the flanges of the spool is a hollow cylinder with a wall thickness of about 1/32 inch and an outer diameter of 0.875 and a length of 1/16. The spool is wrapped with 0.7 mil thick Metglas 2605S3A 1 inch wide and 0.1 mil thick Mylar film until the outer diameter of the insulated Metglas wrapping is 2.24 inches . A perspective view of a single enclosing spool forming one primary winding is shown in FIG. 8E.

  The second transformer is a single stainless steel rod attached to a tight fitting that insulates the electrical glass tube. The winding is four parts as shown in FIG. 8D. The second stainless steel as shown at 56D in FIG. 8D is grounded at 56E to the ground lead of the printed circuit board 56B, and the high voltage terminal is shown as 56F. As described above, a 700 volt pulse between the + and-terminals of the primary winding produces a minus 16,100 volt pulse at the second side terminal 56F for the 1 to 23 volt transformer. This design provides a very low leakage inductance that allows a very fast output rise time.

Laser chamber pulse power components C _p capacitor 82 includes a bank of 0.59nf capacitor 28 mounted on top of the laser chamber pressure vessel. Electrodes 83 and 84 are each about 28 inches long solid brass bars spaced about 0.5 to 1.0 inches apart. In this embodiment, the top electrode 83 is a cathode and the bottom electrode 84 is connected to ground as shown in FIG. 8A.

Compression Head Mounting This preferred embodiment of the present invention includes the compression head mounting technique shown in FIGS. 8H1 and 8H2. FIG. 8H1 is a side view of the laser system showing the position of the compression head module 60 with respect to the electrodes 83 and 84. This technique is designed to minimize the impedance associated with the compression lead chamber connection, while at the same time facilitating rapid replacement of the compression head. As shown in FIGS. 8H1 and 8H2, grounding is made with a slot tab connection approximately 28 inches long along the back of the compression head as shown at 81A in FIG. 8H1 and 81B in FIG. 8H2. The bottom of the slot tab is attached with flexible finger stock 81C. A preferred fingerstock material is sold under the name Multilam®.

  A high voltage connection is made between the 6 inch diameter smooth bottom of saturable inductor 64 and the mate array of flexible finger stock at 89 in FIG. 8H1. As noted above, the preferred finger stock material is Multilam®. With this arrangement, the compression head module can be replaced in about 5 minutes for repair or preventive maintenance.

Gas Control Module This preferred embodiment is a fluorine control system that can operate within a selected sweet spot without the use of a fluorine monitor. This embodiment can be described by reference to FIG.

Fluorine depletion The laser chamber 1 contains about 20.3 liters of laser gas. As mentioned above, nominally the composition is 0.1 percent fluorine and the remainder is helium at a pressure of about 4 atmospheres. 0.1 percent fluorine represents 0.0023 liter or a volume of 2.3 mil fluorine at 4 atmospheres. The nominal mass of fluorine in the laser chamber is about 110 mg. The partial pressure of pure fluorine is about 411 Pa, pure fluorine (corresponding to about 41 kPa with 1% fluorine mixing). During normal operation in laser operation at a loading rate of about 40% (typical for lithographic lasers), fluorine is depleted at a rate of about 4.5 mg / hour (this is the amount of fluorine in the chamber per hour). Of about 4%). With respect to the partial pressure of pure fluorine, this normal depletion rate of fluorine is about 16 Pa / hour. To compensate for this deficiency using 1% fluorine gas mixing, a volume of mixing equal to about 1.6 kPa / hour must be added to the chamber.

  The fluorine depletion rate for the laser is not at all constant. If the laser fan is operating but no lasing has occurred, the fluorine depletion rate is cut approximately half. If the fan is shut down, the fluorine depletion rate is cut to about 1/4, resulting in a 40% loading rate depletion rate. At 100% loading rate, the depletion rate is about twice the 40% loading rate deficiency rate.

Gas Exchange The process described above basically replaces the fluorine that is almost continuously depleted. Since the fluorine gas source is only 1% fluorine, it also replaces part of the chamber He almost continuously. Otherwise, even if a portion of the laser gas is replaced substantially continuously, this mode of operation results in build-up of laser gas contaminants that reduce the efficiency of the laser. This reduction in efficiency requires an increase in voltage and / or an increase in fluorine concentration to maintain the desired pulse energy. For this reason, normal practice with prior art systems has been proposed to periodically shut down the laser in order to have a substantially complete gas exchange. This substantially complete gas exchange is called resupply. These periods are determined based on the number of laser pulses, such as 100,000,000 pulses during restocking, and the restocking time is based on the calendar time since the last restocking or pulse and calendar time combination. Determined. The replenishment time can be determined by the magnitude of the charge voltage that requires a desired output at a specific fluorine concentration. Preferably, after restocking, a new test for “sweet spots” should be performed. Also, the sweet spot test should be performed periodically during replenishment, and if the sweet spot changes, the operator can know where the new sweet spot is.

  Restocking can be accomplished using the system shown in FIG. 16 as shown below. Valves 510, 506, 515, 512, 517 and 504 are closed, valves 506 and 512 are open, vacuum pump 513 is open, and the laser chamber is lowered to an absolute pressure of 13 kPa. (A direct pump drawing may be provided between chamber 1 and vacuum pump 513 for quick drawing.) Valve 512 is closed. Valve 516 is open and 100% He buffer gas from buffer gas bottle 516 is applied to the chamber to fill the chamber to a pressure equal to 262 kPa at 50 ° C. (For this 20.3 liter laser chamber, the temperature correction can be approximated using a 1 kPa / ° C ΔP / ΔT correction for the chamber temperature deviation from 50 ° C. So if the chamber temperature is 23 ° C, it can be up to 247 kPa. The valve 517 is closed, the valve 515 is open, and the amount of mixture of 1% Fl, 99% He fills the chamber from the halogen-rich gas bottle 514 to a pressure equal to 290 kPa at 50 ° C. Added to. (Note that temperature correction should be made.) This provides gas mixing in a chamber of approximately 0.1% Fl and 99% He. When the chamber is heated to about 50 ° C., the pressure is about 4 atmospheres.

N ₂ purge system O ₂ strongly absorbs 157 nm light, so O ₂ must be removed from the beam path. Applicants have developed a significantly improved N ₂ purge system over prior art systems. All optical components related to the laser outside the chamber are purged with nitrogen. The nitrogen system is operated at the pressure during operation of the laser at only about 10 Pascals above atmospheric pressure. This small pressure difference is preferred to eliminate the effect of pressure distortion on the optical component. Purged components include line narrowing modules, output couplers, wavemeters and shutter assemblies.

  Seals are provided at all potential leak site output ports connected by an internal diameter 1/16 inch tube, approximately 6 feet long. The flow through the output port is monitored to ensure proper functioning of the purge system. A preferred flow rate of about 4 liters / minute through a 1/16 inch ID, 6 foot long tube is the preferred flow rate to accommodate the desired N2 pressure differential.

Laser Component Cooling A preferred embodiment of the present invention that is particularly useful for operation at excessive repetition rates of 1000-2000 Hz includes the unique cooling technique shown in FIG. 13 for cooling the excimer laser.

The laser components are contained in a perimeter 240 maintained inside with a slight vacuum created by a blower attached to the hole as shown at 224 in FIGS. 13 and 4A. The cabinet includes an intake port 241 filtered near the top of the cabinet and a small leak source such as around the gasket door, and the room air flow through the laser enclosure is about 200 cubic feet per minute; It is almost not enough to remove the heat created by the heat generating components of the laser.
Most of the unwanted heat (approximately 90%) produced by the laser (approximately 12 kw at 100% loading) is removed by the cold water system as shown in FIG.

In this embodiment, the most heat source of the laser is the high voltage power supply 20, the commutator 40, the compression head 60, and the laser chamber 80. With respect to the chamber, cooled heat exchanger water is placed in the chamber and heat is converted from the circulating laser gas to a cooling water heat exchanger. Another heat exchanger (not shown) is attached to the outer surface of the chamber. Most of the heat generation component cooling water is piped to the component location, and one or more fans energize the component via a water-air heat exchanger as shown in FIG. To do. For the compression head, circulation is included as shown, but for the HVPS and commutator, the other components are also cooled before recirculating back to the heat exchanger through the rest of the enclosure. Cycle is on the component as you do.
Split pans 242 and 243 guide general ventilation air from filter 241 through the path indicated by hole 224 from open heating arrow 244.

  This cooling system has no ducts except for the water supply line, the heat exchanger is internal and attached to the laser chamber, and there is no water supply line connected to the laser components. All components (not the laser chamber) are cooled by the air blown around the enclosure, and there is no cooling connection to interrupt when installing and replacing components. Also, the lack of the need for ducts significantly increases the available components and the working space inside the enclosure.

Pulse energy control algorithm
Mode of Operation—Chip Lithography Embodiments of the present invention include a computer control program with a new algorithm that substantially reduces prior art changes in pulse energy and total burst energy. Preferred devices and software, energy sigma, and preferred processes for reducing burst amount changes are described below.

  As mentioned in the prior art section of this specification, the burst mode is a typical mode of operation of an excimer laser used as an optical source for stepper processing in a lithographically manufactured integrated circuit. In this mode, the laser operates to create a “burst” of pulses at a frequency of 1000 Hz for approximately 110 milliseconds to generate 110 pulses to irradiate a portion of the wafer. After the burst, the stepper moves the wafer and mask, and once the movement is complete, typically over a fraction of a second, the laser generates another 110 pulse burst. Thus, normal operation is a burst of about 110 milliseconds following a fraction of a second of dead time. A longer dead time may be provided so that other operations can be performed at various times. This basic process typically lasts 24 hours a day, 7 days a week, months, with a laser that produces millions of bursts per day. In the burst mode described above, it is usually important that each part of the wafer receives the same irradiation energy in each burst. Chipmakers also want to minimize changes between pulses.

This preferred embodiment of the present invention has these objectives in an apparatus and software that monitors the energy of each pulse (pulse N-1) and then controls the energy of the next pulse (pulse N) based on the following results: Achieve:
1) Comparison of measured energy of pulse N-1 with target pulse energy 2) Comparison of the accumulated amount of bursts via pulse N-1 and the target pulse amount via pulse N-1.

  In a typical F2 excimer laser, we have discussed that the energy for the first 30-40 ms of a burst is typically less stable than the remaining burst due to laser gas transition effects. About 40 ms after the first pulse, the pulse energy is relatively constant at a constant voltage. In dealing with these early fluctuations, Applicants divided the burst into two wise regions; the first region (consisting of a number of early pulses, such as 40 pulses) is called the “K” region. , The second region (consisting of the pulse following the K region) is referred to by the applicant as the “L” region in this specification.

This embodiment of the present invention utilizes a prior art excimer laser device for pulse energy control. The pulse energy of each pulse of each burst is measured by a photodiode as shown in FIG. 8A. The holding circuit, including the overall response time of this photodiode, and its samples and the time required to reset the circuit, is substantially less than 500 milliseconds. The accumulated signal resulting from each approximately 15 ns pulse is stored a few milliseconds after the pulse is over, and this signal is read 6 times, and the average is approximately 1.0 milliseconds after the pulse begins by the computer controller 22. Stored. All the stored energy of the individual pulses before the burst is called the burst dose value. The computer controller utilizes a signal representing the pulse energy of pulse N along with the target pulse energy and a burst dose value to identify the high voltage for pulse N + 1. This calculation takes about 200 milliseconds. When the high voltage value for N + 1 is determined, the computer controller sends a signal to the high voltage command (VCMD) of the high voltage power supply as shown in FIG. 8A which establishes the charge voltage for pulse N + 1 which takes several microseconds. . Computer controller commands the high voltage power supply in order to charge up the capacitor C ₀ to a certain voltage. (For high repetition rates above 2000 Hz, it may be desired to start charging before the calculation is complete.) After the trigger signal from pulse N, it was shown in FIG. 2 at 0.5 ms. when receiving from the trigger circuit 13 a trigger signal related to the pulse N + 1 as charge it takes about 250 microseconds, C ₀ is fully charged ready to complete. With the trigger signal, capacitor C ₀ discharges approximately 700 volts into the magnetic compression circuit shown in FIG. 8B for a time of approximately 5 microseconds, and the pulse produces a laser pulse of approximately 10 mJ for a duration of approximately 15 ns. It is compressed and amplified by a magnetic compression circuit to provide a discharge voltage to a capacitor Cp of approximately 16,100 volts that discharges across electrode 6 at 100 ns.

Preferred Algorithm A particularly preferred process for adjusting the charge voltage to achieve a substantially desirable pulse energy when operating in burst mode is described below.
The process utilizes two voltage regulation algorithms. The first algorithm applies to the first 80 pulses and is called the KPI algorithm. A second algorithm, called the PI algorithm, is applied to pulses after pulse number 40. This time after the 80th is referred to herein as the “L region” of the burst. The initial “PI” is “Proportional Integral”, and “K” in “KPI” is the “K region” of the burst.

（Ｖ_C）₀＝０と定義することにより、
Ａ，Ｂ＝典型的には０と１の間の小数であり、この好ましい実施形態ではＡ及びＢの両方とも０．５である
ε_i＝ｉ番目のパルスのエネルギエラー
＝Ｅ_i−Ｅ_T、
ここで、Ｅ_iはｉ番目のパルスのエネルギであり、
Ｅ_Tは目標エネルギである
Ｄ_i＝１からｉまでの全てのパルスを含むバーストの累積ドーズエラー

dE/dV＝チャージ電圧を備えるパルスエネルギの変化の周波数。（この実施形態
では、ｄＥ／ｄＶの１又はそれ以上の値が各バースト中に実験的に決定
され、これらの値のランニング平均が計算に使用される） KPI Algorithm The K region encompasses 1 to k pulses, where k = 40 for the preferred embodiment. The algorithm for setting the charge voltage for pulse N is:
V _N = (V _B ) _N − (V _C ) _N−1 N = 1, 2,... K
here,
VN = Nth charge voltage (V _B ) _N = Requested to generate the Nth target energy ET in the K region
An array of k storing the voltage representing the best estimate of the voltage to be performed.
This array is updated after each burst according to the following equation:
(V _C ) _N-1 = Up to pulse N-1 occurred for the previous pulse in the burst
Voltage correction of previous pulse energy error based on energy error

By defining (V _C ) ₀ = 0,
A, B = typically a decimal number between 0 and 1, and in this preferred embodiment both A and B are 0.5 ε _i = energy error of i th pulse = E _i -E _T ,
Where E _i is the energy of the i th pulse,
E _T is the cumulative dose error of the burst containing all pulses from D _i = 1 to i which is the target energy

dE / dV = frequency of change of pulse energy with charge voltage. (In this embodiment, one or more values of dE / dV are experimentally determined during each burst and the running average of these values is used in the calculation)

ここで、インデックスＭはバーストナンバーのことである
Ｃ＝典型的には０と１の間の小数であり、
この好ましい実施形態では０．３である。 The stored value (V _B ) _N is updated during or after each burst according to the following relationship:

Here, the index M is a burst number, C = typically a decimal number between 0 and 1,
In this preferred embodiment, it is 0.3.

ここで、
Ｖ_N＝Ｎ番目のパルスに関するチャージ電圧
Ｖ_N-1＝Ｎ−１番目の（前の）パルスに関するチャージ電圧
変数Ａ，Ｂ，ε₁，Ｄ₁及びｄＥ／ｄＶは前に定義されている。 PI Algorithm The L region covers from pulse k + 1 to the end of the burst (in the preferred embodiment, pulse number 41 and beyond). The algorithm for setting the charge voltage for pulse N is:

here,
V _N = Charge voltage for the _Nth pulse V _N-1 = Charge voltage variables A, B, ε ₁ , D ₁ and dE / dV for the N−1th (previous) pulse have been defined previously.

ここで、Ｖ_Dither＝固定された電圧インクリメントであり、典型的には数ボルトである。
パルスｊ＋１に関して：
Ｖ_j+1＝Ｖ_j−２・Ｖ_Dither
ｊ＋１の後、ｄＥ／ｄＶは計算され：

Determination of dE / Dv New values for dE / dV are periodically determined to track relatively slow changes in laser features. In a preferred embodiment, dE / dV is measured by changing or dithering the voltage in a controlled manner during two consecutive pulses in the L region. For these two pulses, the normal PI energy control algorithm is temporarily interrupted and replaced by:
For pulse j:

Where V _Dither = a fixed voltage increment, typically a few volts.
For pulse j + 1:
V _{j + 1} = V _j -2 ・ V _Dither
After j + 1, dE / dV is calculated:

  The calculation of dE / dV is very noisy because the expected energy change due to the dithering voltage is of the same magnitude as the normal energy change of the laser. In the preferred embodiment, the running average of the last 50 dE / dV calculation is actually used in the PI and KPI algorithms.

パルスｊ＋２（直後に続く２つのディザリングパルス）はディザリングされないが、特定の値を有する：

Ｖ_j+2に関するこの特定の値は、適用された電圧ディザリングと、パルスｊ＋１から期待されるエネルギディザリングとの両方に関して補正される。 The preferred method for V _Dither selection is to identify the desired energy dither E _Dither, typically a few percent of the energy target E _T, and then the current relates dE / dV to calculate the V _Dither Use the (average) value of:

Pulse j + 2 (the two immediately following dithering pulses) is not dithered but has a specific value:

This particular value for V _{j + 2} is corrected for both the applied voltage dithering and the energy dithering expected from pulse j + 1.

  Many variations on the above algorithm are possible. For example, dE / dV can be determined in the L region as well as K. Dithering can be performed once per burst or several times. The dithering sequence can be performed with a fixed number of pulses as described above, or can be initialized to a randomly selected number of pulses that varies from one burst to the next.

The reader should be aware that A, B and C are convergence factors and can take many other values. Larger values than those specified above provide faster convergence, but can lead to increased instability. In another preferred embodiment, A = (2B) ^1/2 . This relationship was developed from a recognized technique to provide critical damping. B is zero in the absence of dose correction, but A should not be zero because it provides an attenuation term for the dose convergence portion of the algorithm.

  If the predetermined value of dE / dV is too small, the above algorithm should be overcorrected. Therefore, when the energy sigma value exceeds the threshold, the preferred technique is any double dE / dV. Initial values of V and dE / dV are provided for the first pulse of the burst. D is set to zero at the start of each burst. The default dE / dV is set to about 3 times the expected dE / dV to avoid initial overcorrection.

Another way to determine dE / dV without dithering as described above is to simply measure and store energy and voltage values during laser operation. (Measured rather than specific voltage values can also be used.) These data can be used to determine dE / dV as a function of V for a given pulse energy. The reader should note that each value of dE / dV contains a significant amount of uncertainty, since the value component is different from a measurement with significant uncertainty. However, a large average number of dE / dV values can reduce these uncertainties. Dithering to determine dE / dV should not be done in each burst, but instead can be done periodically, such as once every M bursts. Alternatively, the dE / dV measurement can be replaced by a calculation performed by a computer, or the dE / dV value can be manually inserted by the operator of the previous pulse for the calculation of V _{N + 1} . Another approach is to use the actual measured value for V _N for this control system. Also, the value of V _BIN is calculated from a specific value and is not an actual measured value in the embodiment described above. An obvious variant embodiment may use the measured voltage value. E _T is the normally constant value as 10 mJ, need not be constant. For example, the E _T of the last ten pulses smaller than the pulse energy of the nominal deviation percentage from the target E _T for these pulses, may provide a small effect integration was performed Parusudozu. It is also preferred in certain situations such as programming computer controller 22 to provide an _ET value that varies from burst to burst.

Single Line and Narrow Line Configuration FIG. 11A shows the preferred single line for the preferred F ₂ laser system.
In this configuration, one of the two large F2 lines is selected with a simple prism selector as shown. FIG. 11B shows a preferred line narrowing system in which the power oscillator is seeded by the master oscillator.

Prototype Unit A prototype F2 laser system unit was assembled and tested by the applicant and their fellow workers.
Prototype lasers are largely based on current-produced KrF and ArF lasers that incorporate various important improvements over prior art excimer laser systems, utilizing high efficiency chambers and solid state physical pulse power excitation. The discharge is corona play-on to minimize gas contamination. The entire optical beam path is nitrogen purged to avoid light absorption by oxygen and to avoid damage to optical components. All resonator optics are external to the laser chamber equipped with an angled chamber window. The gas mixture is 0.1% fluorine at 4 atm of helium and the electrode gap is reduced to 10 mm.
In this unit, Applicants have succeeded in demonstrating key parameters for a prototype lithography fluorine laser with only a slight change to the ArF laser current in the current development.

Experimental results The experimental results from the prototype unit are described below. Laser power is measured by a standard power meter and correlates with a piezoelectric joule meter. The contribution of the red atomic fluorine laser is subtracted, usually reaching less than 1% of the total energy. By emitting the beam delivery tube into the air, red radiation that strongly absorbs 157 nm light is measured.
The laser wavelength in this prototype unit operates in single line mode at 157.6 nm by tuning with a set of two external prisms. The laser can also be tuned to a 157.5 nm transition with reduced efficiency. No transition is observed at 156.7 nm. The laser spectrum as recorded by a 2 meter Jobin Yvon VUV spectrometer shows a measurement limit line width of 6 pm.

  In wide band (or multiline) operation, a maximum power of 12 W is obtained with a 1000 Hz repetition rate. The power increases linearly with the number of repetitions, without saturating. Single line mode behaves similarly but is 1/3 the energy. This energy reduction is due to the large cavity length increase in the present prism setup and can be significantly reduced. In burst mode, 5% 3σ stability was recorded. Only the starting burst transition of the output energy was observed. This is preferably compared to ArF lasers that exhibit large energy instabilities and burst transitions related to gas flow. From this it can be concluded that the manufactured fluorine lithography laser has better energy stability than the current dye ArF laser. The integral square pulse duration is 30 ns, reaching the performance of ArF laser.

  Careful selection of the laser chamber material and pre-on with corona discharge allows the laser to operate without supercleaning and 3M shots with halogen injection and minimal energy reduction over several hours.

  The dependence of the broadband laser power on the number of repetitions is displayed at the top of FIG. 10A. The bottom of FIG. 10A shows a slight drop off of the pulse energy with increasing frequency above 1000 Hz. The laser power increases almost linearly in frequency up to a power of 15 W at 1 kHz and about 19 W at 2000 Hz. Based on this linear relationship, it was estimated that the fluorine laser could be further changed to a few kHz operation and the gas flow was changed accordingly. Since Applicant has used helium as the buffer gas, only a fraction of the blower power of a standard neon-based laser is required and therefore there is no limit to faster flow rates.

  A good measure of energy stability is obtained by observing energy transitions in burst mode. Thus, the laser fires repeatedly during a burst and the average energy for each pulse position of the burst is recorded. Also, the average change in energy from burst to burst is calculated with respect to the number of pulses per burst. The results of the energy and stability curves for comparison with the fluorine laser and the narrow band ArF laser are displayed in FIGS. 10B1 and 2. Fluorine lasers show only small energy changes over 120 shot bursts. Energy stability shows an initial increase at the beginning of the burst and then stabilizes at the 3σ level of about 3%. In contrast, ArF lasers exhibit a large transition in energy and are about 7% 3σ instability. ArF lasers have a 0.5% dose stability in a 60 pulse window, so a fluorine laser is expected for at least the same dose stability. FIG. 10C includes pulse energy and 3σ values at 1000 Hz and 1900 Hz.

  The spectrum of a broadband fluorine laser as recorded by a VUV spectrometer is shown in FIGS. 10D1 and 2. It can be clearly seen that there are two transition lines at 157.52 nm and 157.63 nm. 87% of the laser energy is located on a wavelength line longer than 157.63 nm. No transition at 156.7 nm is observed. Single line mode operation at 157.63 nm is achieved by tuning with a set of two external prisms. The laser can also be tuned to the 157.52 nm transition line, but at a reduced efficiency. Also, enlarged views of the laser line at 157.63 nm are shown in FIGS. 10D1 and 2. The involved line widths of 1.14 pm FWHM and 2.35 pm 95% are measured. These line widths are much narrower than previously expected. Therefore, a fluorine laser line selected without further line narrowing is sufficient for all but not a sufficient reflective imaging system. The laser power versus single line laser repetition rate behavior shows the same linear increase as the wide band laser. However, the maximum power in this first experiment is limited to 4W. The reduced output power is caused by reflection losses in the line selection optics and excessively long cavity lengths.

  Horizontal and vertical beam characteristics were measured at a distance of 1 m from the laser. (See FIGS. 10E1 and 2). The beam exhibits a smooth profile with a high degree of symmetry. These profile types are easily measured by commonly used homogenizer techniques that produce very uniform illumination.

  Gas life assessment is derived by operating at a constant voltage without fluorine injection and recording the development of laser power versus shot number. There is no use of ultra-clean in these measurements. As shown in FIG. 10F, the laser power drops to 20% or less after 4 million laser shots and is at least comparable to ArF laser. Thus, from previous experience with ArF lasers, a gas lifetime of approximately 25 million shots can be estimated by providing periodic fluorine implantation data. This clearly shows the selection of compatible materials in the laser chamber and the results of using corona play on. For KrF and ArF lasers, a direct relationship between fluorine consumption and chamber lifetime was previously established. Therefore, we estimate the chamber lifetime of a fluorine laser to be on the same order as that of an ArF laser.

The lower graph in FIG. 15A shows the pulse energy versus F ₂ concentration, and the upper graph shows the maximum number of pulse repetitions, both at a blower speed of 2500 rpm. The upper graph of FIG. 15B shows the pulse shape as a function of time showing a FWHM of 16 ns, and the lower graph shows that the integrated square pulse width is about 37 ns.

Although this F ₂ laser system has been described with reference to particular embodiments, it will be apparent that various additions and modifications are possible. For example, many alternative embodiments are discussed in the patent applications listed in the first section of this specification, all of which are incorporated herein by reference. An etalon output coupler can be used to provide further line narrowing. The buffer gas may be neon instead of helium. The present invention should be limited only by the claims.

1 is a diagram of a prior art commercially available excimer lithography laser. FIG. 1 is a block diagram illustrating some of the main elements of a prior art commercial excimer laser used for integrated circuit lithography. FIG. FIG. 3 is a diagram of a laser chamber of the laser of FIG. FIG. 2 is a diagram of a preferred embodiment of the present invention. It is a figure which shows the blower drive unit containing a magnetic bearing. 1 is a cross-sectional view of a laser chamber according to a preferred embodiment of the present invention. 1 is a cross-sectional view of a laser chamber according to a preferred embodiment of the present invention. FIG. 6 is a diagram showing the characteristics of a preferred play-on device tube. 1 is a block diagram of a pulse power system according to a preferred embodiment of the present invention. FIG. 3 is a simplified circuit diagram of the preferred embodiment. It is a combination of a circuit diagram and a block diagram of a high voltage power supply that is a part of the preferred embodiment. It is an assembly perspective view of the pulse transformer used in the preferred embodiment. FIG. 4 is a diagram of a primary winding of a pulse transformer that would be the preferred embodiment. 6 is a timeline chart showing pulse compression using the preferred embodiment. 6 is a timeline chart showing pulse compression using the preferred embodiment. 6 is a timeline chart showing pulse compression using the preferred embodiment. It is a figure which shows a saturable inductor. It is a figure which shows a saturable inductor. Fig. 4 shows the mounting of a compression head in a preferred embodiment. Fig. 4 shows the mounting of a compression head in a preferred embodiment. FIG. 3 describes a preferred heat exchanger design. FIG. 3 describes a preferred heat exchanger design. 3 is a graph of test data obtained during an experiment with a prototype F ₂ laser. 3 is a graph of test data obtained during an experiment with a prototype F ₂ laser. 3 is a graph of test data obtained during an experiment with a prototype F ₂ laser. 3 is a graph of test data obtained during an experiment with a prototype F ₂ laser. 3 is a graph of test data obtained during an experiment with a prototype F ₂ laser. 3 is a graph of test data obtained during an experiment with a prototype F ₂ laser. 3 is a graph of test data obtained during an experiment with a prototype F ₂ laser. 3 is a graph of test data obtained during an experiment with a prototype F ₂ laser. 3 is a graph of test data obtained during an experiment with a prototype F ₂ laser. It shows two preferred F ₂ system configurations. It shows two preferred F ₂ system configurations. Various anode support bar designs are shown. Various anode support bar designs are shown. Various anode support bar designs are shown. Various anode support bar designs are shown. Various anode support bar designs are shown. 1 illustrates a preferred encircling cooling system. A preferred blower blade structure design is shown. A preferred blower blade structure design is shown. A preferred blower blade structure design is shown. It is a graph showing pulse energy versus F2 concentration and the maximum number of pulse repetitions. It is a graph which shows the pulse shape with respect to time. A large manifold gas supply system is shown.

Explanation of symbols

202 Gas Module 205 Control Module 206 Line Narrowing Module 207 Compression Head 208 High Voltage Pulse Power Supply Module 209 Commutator Module for Pulse Power Supply 211 Laser Chamber 213 Wavelength Module 216 Output Coupler 217 Blower Motor 221 Chamber Window

Claims (4)

A very narrow band high reliability, modular manufacturing high quality high repetition rate ArF excimer laser for generating a narrow band pulsed laser beam at a repetition rate of at least about 1000 Hz, comprising:
A. A rapidly replaceable laser chamber module including a laser chamber, the laser chamber module comprising:
1) two elongated electrodes;
2) a) fluorine,
b) an inert gas;
A laser gas consisting of
3) a gas circulator for circulating the gas between the electrodes at a rate of at least 2 cm / millisecond;
B. A modular pulse power system comprising at least one rapidly replaceable module, the system comprising a power supply, a pulse compression and amplification circuit, and a high voltage electrical power of at least 14000 volts across the electrodes at a frequency of at least about 1000 Hz. Pulse power control to produce pulses,
C. A laser pulse energy control system for controlling the voltage provided by the pulse power system, the control system generating a laser pulse energy monitor and a laser pulse having a pulse energy within a desired energy range And a computer processor programmed with an algorithm for calculating the electrical pulses required for the operation based on historical pulse energy data.   The laser of claim 1, wherein the two elongated electrodes comprise a cathode and an anode, the anode having cooling fins.   3. The laser of claim 2, wherein the anode and the anode support bar together have a mass of at least about 3.4 kg. A high reliability, modular manufacturing high quality, high repetition rate excimer laser for generating a narrowband pulsed laser beam at a repetition rate of at least about 1 kHz,
A. Quickly replaceable laser chamber module
1) two elongated electrodes;
2) a laser gas composed of fluorine and a buffer gas;
3) a gas circulation system for circulating the laser gas between the electrodes at at least 2 cm / millisecond, the gas circulation system comprising:
a) an unbrazed blade structure constituting the shaft;
b) a brushless motor for rotating the shaft;
c) a magnetic bearing for supporting the shaft, the motor and the bearing having a rotor attached to the shaft and sealed in an environment exposed to the laser gas, the motor And the bearing has a stator outside the laser gas environment;
B. A pulse power system substantially contained within at least one rapidly replaceable module, the pulse power system comprising:
1) a processor that periodically controls the high voltage power supply at a frequency of at least about 1000 Hz, and charging the charge capacitor with electrical energy to a predetermined pulse control voltage;
2) a compression and amplification circuit for connecting electrical energy stored in the charge capacitor to a high voltage electrical pulse of at least 14000 volts across the electrode;
laser.

Images