Reliable, modular, production quality narrow-band high rep rate ArF excimer laser

[0003] This Application is a Continuation-In-Part of Ser. No. 09/041,474, Reliable, Modular, Production Quality Narrow Band KrF Excimer Laser, filed Mar. 11, 1998 now U.S. Pat. No. 5,991,324; Ser. No. 08/995,832, Excimer Laser Having Pulse Power Supply with Fine Digital Regulation, filed Dec. 22, 1997 now U.S. Pat. No. 5,982,795; Ser. No. 08/896,384 Wavelength Reference for Excimer Laser, filed Jul. 18, 1997 now U.S. Pat. No. 5,978,391; Ser. No. 08/939,611, Protective Overcoat for Replicated Diffraction Gratings, filed Sep. 29, 1997 now U.S. Pat. No. 6,162,495; Ser. No. 08/947,474, Narrow Band Excimer Laser, filed Oct. 10, 1997 now U.S. Pat. No. 5,982,800; Ser. No. 09/034,870, Pulse Energy Control for Excimer laser, filed Mar. 4, 1998 now U.S. Pat. No. 6,005,879; Ser. No. 09/082,139, Narrow Band Excimer Laser with Gas Additive, filed May 20, 1998 now U.S. Pat. No. 6,014,398 Ser. No. 09/157,067, Reliable Modular Production Quality Narrow Band High Rep Rate Excimer Laser, filed Sep. 18,1998 now U.S. Pat. No. 6,128,323; Ser. No. 09/162,341, Line Narrowing Apparatus with High Transparency Prism Beam Expander filed Sep. 28, 1998 now U.S. Pat. No. 5,978,409; Ser. No. 09/165,593, Wavelength System for an Excimer Laser filed Oct. 2, 1998 now U.S. Pat. No. 5,978,394; Ser. No. 09/206,526, Wavelength Reference for Laser, filed Dec. 7, 1998; Ser. No. 09/211,825, High Pulse Rate Power System with Resonant Power Supply filed Dec. 15, 1998 now U.S. Pat. No. 6,028,872; Ser. No. 09/217,340 Durable Etalon Based Output Coupler filed Dec. 21, 1998 now U.S. Pat. No. 6,137,821 all of which are incorporated herein by reference. This invention relates to lasers and in particular to narrow-band ArF excimer lasers.

[0004] Krypton-Fluoride (KrF) excimer lasers are currently becoming the workhorse light source for the integrated circuit lithography industry. The KrF laser produces a laser beam having a narrow-band wavelength of about 248 nm and can be used to produce integrated circuits with dimensions as small as about 180 nm. The Argon Fluoride (ArF) excimer laser is very similar to the KrF laser. The primary difference is the laser gas mixture and a shorter wavelength of the output beam. Basically, Argon replaces Krypton and the resulting wavelength of the output beam is 193 nm. This permits the integrated circuit dimensions to be further reduced to about 120 nm. A typical prior-art KrF excimer laser used in the production of integrated circuits is depicted in FIG. 1 and FIG. 2. A cross section of the laser chamber of this prior art laser is shown in FIG. 3. A pulse power system 2 powered by high voltage power supply 3 provides electrical pulses to electrodes 6 located in a discharge chamber 8. Typical state-of-the art lithography lasers are operated at a pulse rate of about 1000 Hz with pulse energies of about 10 mJ per pulse. The laser gas (for a KrF laser, about 0.1% fluorine, 1.3% krypton and the rest neon which functions as a buffer gas) at about 3 atmospheres is circulated through the space between the electrodes at velocities of about 1,000 inches per second. This is done with tangential blower 10 located in the laser discharge chamber. The laser gases are cooled with a heat exchanger 11 also located in the chamber and a cold plate (not shown) mounted on the outside of the chamber. The natural bandwidth of the excimer lasers is narrowed by line narrowing module 18. Commercial excimer laser systems are typically comprised of several modules that may be replaced quickly without disturbing the rest of the system. Principal modules include:

[0005] Laser Chamber Module,

[0006] Pulse Power System with: high voltage power supply module,

[0007] commutator module and high voltage compression head module,

[0008] Output Coupler Module,

[0009] Line Narrowing Module,

[0010] Wavemeter Module,

[0011] Computer Control Module,

[0012] Gas Control Module,

[0013] Cooling Water Module

[0014] Electrodes 6 consist of cathode 6A and anode 6B. Anode 6B is supported in this prior art embodiment by anode support bar 44 which is shown in cross section in FIG. 3. Flow is clockwise in this view. One corner and one edge of anode support bar 44 serves as a guide vane to force air from blower 10 to flow between electrodes 6A and 6B. Other guide vanes in this prior art laser are shown at 46, 48 and 50. Perforated current return plate 52 helps ground anode 6B to the metal structure of chamber 8. The plate is perforated with large holes (not shown in FIG. 3) located in the laser gas flow path so that the current return plate does not substantially affect the gas flow. A peaking capacitor comprised of an array of individual capacitors 19 is charged prior to each pulse by pulse power system 2. During the voltage buildup on the peaking capacitor, two preionizers 56 weakly ionize the lasing gas between electrodes 6A and 6B and as the charge on capacitors reach about 16,000 volts, a discharge across the electrode is generated producing the excimer laser pulse. Following each pulse, the gas flow between the electrodes of about 1 inch per millisecond, created by blower 10, is sufficient to provide fresh laser gas between the electrodes in time for the next pulse occurring one millisecond later.

[0015] In a typical lithography excimer laser, a feedback control system measures the output laser energy of each pulse, determines the degree of deviation from a desired pulse energy, and then sends a signal to a controller to adjust the power supply voltage so that energy of subsequent pulses are close to the desired energy. In prior art systems, this feedback signal is an analog signal and it is subject to noise produced by the laser environment. This noise can result in erroneous power supply voltages being provided and can in turn result in increased variation in the output laser pulse energy.

[0016] These excimer lasers are typically required to operate continuously 24 hours per day, 7 days per week for several months, with only short outages for scheduled maintenance. One problem experienced with these prior-art lasers has been excessive wear and occasional failure of blower bearings.

[0017] A prior art wavemeter utilizes a grating for coarse measurement of wavelength and an etalon for fine wavelength measurement and contains an iron vapor absorption cell to provide an absolute calibration for the wavemeter. This prior art device focuses the coarse signal from the grating onto a linear photo diode array in the center of a set of fringes produced by the etalon. The center fringes produced by the etalon are blocked to permit the photo diode array to detect the coarse grating signal. The prior-art wavemeter cannot meet desired speed and accuracy requirements for wavelength measurements.

[0018] A need exists in the integrated circuit industry for a modular, reliable, production line quality ArF excimer laser in order to permit integrated circuit resolution not available with KrF lasers.

[0019] The present invention provides a reliable modular production quality ArF excimer laser capable of producing at repetition rates in the range of 1,000 to 2,000 Hz laser pulses with pulse energies greater than 5 mJ with a full width half, maximum bandwidth of about 0.6 pm or less and wavelength stability of less than 0.1 pm. Using this laser as an illumination source, stepper or scanner equipment can produce integrated circuit resolution of 0.12 pm or less. Replaceable modules include a laser chamber; a modular pulse power system; and a line narrowing module.

[0020] Important improvements have been provided in the pulse power unit to produce faster charging. These improvements include an increased capacity high voltage power supply, an improved commutation module that generates a high voltage pulse from the capacitors charged by the high voltage power supply and amplifies the pulse voltage 28 times with a very fast voltage transformer having a secondary winding consisting of a single four-segment stainless steel rod. A novel design for the compression head saturable inductor (referred to herein as a “pots and pans” design) greatly reduces the quantity of transformer oil required and virtually eliminates the possibility of oil leakage which in the past has posed a hazard.

[0021] Improvements in the laser chamber permitting the higher pulse rates and improved bandwidth performance include the use of a single preionizer tube.

[0022] Improvements in the resonance cavity of preferred embodiments of the present invention include a line narrowing module with CaF prism beam expanders and a grating specially coated for UV damage resistance. Preferred embodiments comprised output couplers having substantially increased reflectivity over prior art designs.

[0023] A newly designed wavemeter including a computer processor programmed with an algorithm for controlling wavelength measurements and computing wavelengths at rates sufficient for feedback control of the wavelength of the output laser beam at rates of 1,000 Hz or faster. In a preferred embodiment a vapor cell with platinum vapor providing a reference absorption line for system calibration.

[0024] Other embodiments of the present invention include ceramic bearings. Optionally magnetic bearings may be utilized. Reaction forces on the bearings may be reduced by providing an aerodynamic contour on the anode support bar. Other improvements include use of acoustic baffles for laser chambers producing disruptive acoustic shock waves.

[0025] FIG. 1 is a drawing of a prior art commercial excimer lithography laser.

[0026] FIG. 2 is a block diagram showing some of the principal elements of a prior art commercial excimer lasers used for integrated circuit lithography.

[0027] FIG. 3 is a drawing of the laser chamber of the FIG. 2 laser.

[0028] FIGS. 4A and 4B are drawings of a preferred embodiment of the present invention.

[0029] FIG. 5 is a drawing showing a blower drive unit including magnetic bearings.

[0030] FIGS. 6 and 6A are cross section drawings of laser chambers of preferred embodiments of the present invention.

[0031] FIG. 7 is a drawing showing features of a preferred preionizer tube.

[0032] FIG. 8A is a block diagram of a pulse power system of the preferred embodiment of the present invention.

[0033] FIG. 8B is a simplified circuit diagram of the above preferred embodiment.

[0034] FIG. 8C is a combination block diagram, circuit diagram of a high voltage power supply which is part of the above preferred embodiment.

[0035] FIG. 8D is a prospective assembly drawing of a pulse transformer used in the above preferred embodiment.

[0036] FIG. 8E is a drawing of a primary winding of a pulse transformer used in the above preferred embodiment.

[0037] FIGS. 8F1, 8F2 and 8F3 are time line charts showing pulse compression using the above preferred embodiment.

[0038] FIGS. 8G1 and 8G2 are drawing showing two views of a saturable inductor.

[0039] FIGS. 8H1 and 8H2 shows the mounting of a compression head in a preferred embodiment.

[0040] FIGS. 9A and 9B are drawings describing a preferred heat exchanger design.

[0041] FIG. 10 is a drawing showing features of an improved wavemeter.

[0042] FIG. 10A is a graph of an ArF excimer laser broadband spectrum.

[0043] FIG. 10B is an ArF narrowband spectrum.

[0044] FIG. 10C shows the relationship between tuning mirror position and output wavelength for an ArF excimer laser.

[0045] FIG. 10D is a block diagram showing the principal elements for controlling wavelength of an excimer laser.

[0046] FIG. 10E depicts a 1024-pixel photo diode array.

[0047] FIG. 10F describes the light patterns on the FIG. 5 photo diode array used for making coarse and fine wavelength measurements.

[0048] FIGS. 10G1 and 10G2 are views of an etalon.

[0049] FIG. 10H is a drawing of an etalon assembly.

[0050] FIG. 10J is a graph showing a calibration procedure.

[0051] FIG. 10K is a drawing of a vapor cell.

[0052] FIG. 10I is a graph showing absorption in the vapor cell.

[0053] FIGS. 11A and 11B are graphs describing the functioning of the FIG. 10 wavemeter.

[0054] FIGS. 12A through 12E show various anode support bar designs.

[0055] FIG. 13 describes a preferred enclosure cooling system.

[0056] FIGS. 14A, 14B and 14C show preferred blower blade structure designs.

[0057] FIGS. 15A and 15B show grating cross section.

[0058] FIG. 16 shows a big manifold gas supply system.

[0059] A preferred embodiment of the present invention can be described by reference to the drawings.

[0060] A front view of a preferred embodiment of the present invention is shown in FIG. 4 respectively. This drawing emphasizes the modular nature of their particular invention which allows very quick replacement of modules for repair, replacement and maintenance. The principal features of this embodiment are listed below corresponding to the reference numbers shown on FIG. 4.

[0061] 201 Laser enclosure

[0062] 202 Gas module

[0063] 203 Cooling water supply module

[0064] 204 AC/DC distribution module

[0065] 205 Control module

[0066] 206 Line narrowing module

[0067] 207 Compression head

[0068] 208 High voltage pulse power supply module

[0069] 209 Commutator module for pulse power supply

[0070] 210 Metal fluoride trap

[0071] 211 Laser chamber

[0072] 213 Wavemeter module

[0073] 214 Automatic shutter

[0074] 216 Output coupler

[0075] 217 Blower motor

[0076] 218 Metal fluoride trap power supply

[0077] 219 Status lamp

[0078] 220 24 volt power supply

[0079] 221 Chamber window

[0080] 222 Gas control flexible connection

[0081] 224 Vent box

[0082] A preferred embodiment of the present invention is an improved version of the laser described in FIGS. 1, 2 and 3. This preferred embodiment includes the following improvements:

[0083] 1) A single tube larger preionizer replaces the prior-art combination of a two-tube preionizer to provide improved efficiency, better preionization and improved laser gas flow between the electrodes;

[0084] 2) A silicon-free fan blade which may be a one-piece machined blade;

[0085] 3) Acoustic baffles are provided to minimize adverse effects of acoustic shock waves resulting from the electric discharges;

[0086] 4) The solid-state pulse power system has been modified to produce faster rise time, providing more consistent pulses, and improved laser efficiency at higher voltages;

[0087] 5) More precise control of the charging voltage of the pulse power system;

[0088] 6) The reflectivity of the output coupler has been increased to about 30%, substantially decreasing the bandwidth of the output pulse;

[0089] 7) Fused silica prisms have been replaced with CaF prisms to provide much better thermal stability;

[0090] 8) MgF₂coated grating;

[0091] 9) An improved wavemeter providing much more precise measurement of nominal wavelength and bandwidth is provided;

[0092] 10) Improved wavemeter by using an elastomer free etalon design and a new diffuser; and

[0093] 11) A computer controller programmed with a new algorithm providing a much improved control of pulse energy and burst energy.

[0094] As shown in FIG. 6, a single larger preionizer tube 56A has replaced the two-preonizer tubes 56 shown in FIG. 3. The single tube preionizer is fabricated in accordance with the description in U.S. Pat. No. 5,719,896, issued Feb. 17, 1998, which is incorporated herein by reference. Applicants have discovered that one preionizer tube is not only sufficient, but very surprisingly provides improved performance over the two-preionizer design. In this embodiment the preionizer is located upstream of the electrodes. Applicants do not fully understand the reason for the improved performance. Applicants have determined that the one tube preionizer improves in the pulse-to-pulse stability by providing improved spacial stability of the discharge.

[0095] Referring now to FIG. 7, this preionizer utilizes an integrated tube design, having bushing element 180 with anti-tracking grooves 170 incorporated therein as an integral component of the tube. The diameter of the rod portion 145 and the OD of the bushing portion 180 of the preionizer is ½ inch. The inside conductor rod 146 has a diameter of {fraction (7/37)} inch and the connecting wire extending through the bushing section to make a ground connection is about {fraction (1/16)} inch diameter. Prior preionizer tube designs utilized a two-diameter design, with the rod portion at about ¼ inch diameter and the bushings at about 1 inch diameter. This necessitated, for manufacturing purposes, a bonding process to join the bushing component with the tube component. The constant diameter, thicker tube design is contrary to conventional design rules, which would predict a reduction in ionization due to lower capacitances. In most designs, the tube thickness is dependent upon the dielectric strength of the material selected. Those skilled in the art will recognize that the prior art conventional preionizer tube design technique is to select a material with the highest dielectric strength and determining a wall thickness to match this capacity. For example, a sapphire material is known to have a dielectric strength ranging from 1200 volts/mil to 1700 volts/mil. Therefore, a dielectric thickness of 0.035 inches thick, provides a safety factor of 2 if the laser operates at 25 kV. This design yields a lower capacitance; however, the actual effect of this reduced capacitance on laser operation was discovered to be negligible, with a surprising increase in the measured geometric irradiation of the electrode gap. Because of the constant diameter, thicker tube wall, integral bushing design, a single piece of material can be machined to provide anti-tracking grooves 170. Because of the single piece construction, there is no need to use ultra-pure (i.e., 99.9%) polycrystalline translucent aluminum oxide ceramic, although Applicants continue to use the ultra-pure material. There is no requirement to perform the difficult surface polishing of tube geometries in preparation for diffusion bonding to artificially create the integral relationship between bushing 180 and tube 145. In fact, it has been determined that high purity is not as important a property as porosity of the material. It has been found that the greater the porosity, the more the dielectric strength is reduced. As a result, a commercial grade ceramic, preferably with purity of at least 99.8% and low porosity, such as that manufactured by Coors Ceramics Company under the material No. AD-998E and having a dielectric strength of 300 volts/mil may be used. Bushings 180, having anti-tracking grooves 170 disposed therein, as previously described, act to prevent high voltage tracking axially along the surface of the tube from the cathode to the ground plane 160.

[0096] As explained above, Applicants have discovered that a single preionizer works dramatically better than two preionizers, and as explained above the first preferred embodiment places the single preionizer system of the electrodes. Applicants have also experimented with the single preionizer located downstream and has discovered that at certain blower speeds this arrangement produces substantially better pulse energy stability than the upstream arrangement on the two tube arrangement.

[0097] Improvements have been made to the chamber to improve the efficiency of the laser. A single piece cathode insulator 55A comprised of alumina, Al₂O₃insulates the cathode from the upper chamber structure as shown in FIG. 6A. In a prior art design, eight separate insulators were needed to avoid insulator cracking due to thermal expansion stress in the insulator. This important improvement permitted the head portion of the chamber to be made shorter which significantly reduced the distance between cathode 83 the peaking capacitor 82. The individual capacitors 54A forming the peaking capacitor array 82 were moved horizontally in closer to the cathode as compared to the prior art. To reduce thermal expansion difference between the single piece insulator and the chamber structure the upper chamber 8A was fabricated from ASTM A3C steel which has a coefficient of thermal expansion closer to Al₂O₃than aluminum. The bottom section 8B of chamber 8 is aluminum, but Applicants have determined that the difference in thermal expansion between ASTM A3C steel and aluminum is not a problem.

[0098] Prior art cathodes for commercial lithography lasers were typically supported by a cathode support bar 53 as shown in FIG. 3. In this preferred embodiment, the cathode support bar was eliminated and the cathode 83 was made slightly thicker and mounted directly on the single piece insulator 55A. The cathode 83 is connected to the high voltage side 82A of peaking capacitor 82 by 15 feed through rods 83A and connecting nuts 83B. In the preferred embodiment, a new anode support bar 84A is substantially more massive than prior art anode support bars and comprises fins 84B located in the gas flow region. Both of these features minimize temperature variations of the anode.

[0099] Applicants have discovered that prior art elastomer seals reacted with fluorine gas to produce contaminants in the laser gas which degraded laser performance. A preferred embodiment of the present invention uses all metal seals to seal the laser chamber. The preferred metal seals are tin plated inconel 1718 seals. In the case of heat exchanger 11, a metal bellows has been provided which permits thermal expansion and contraction and allows metal seals to be used to seal cooling water inlets.

[0100] Applicants have also discovered that elements of stainless steel also react with fluorine to produce contaminants in the laser gas. Therefore, in this preferred embodiment, prior art stainless steel current return structures and gas flow vanes have been replaced with monel current returns 250 and monel flow vanes 252 and 254.

[0101] Applicants have discovered that a significant cause of distortion of the quality of laser beams produced by narrow-band excimer lasers operating at 1000 Hz or greater is acoustic shock waves created by the electric discharge of one pulse which reflects from elements of chamber structure back to the space between the electrodes and distorts the laser beam of the next pulse occurring 1.0 millisecond later. An embodiment described herein (see FIG. 6A) substantially minimizes this effect by providing angled, grooved acoustic baffles 60 and 62 on both sides of the laser chamber. These baffles absorb a portion of the acoustic energy and reflect a portion of the acoustic energy down into the lower region of the laser chamber away from the electrodes. In this preferred embodiment, the baffles consist of a machined metal structure with grooves 0.1 mil wide, 0.3 mil deep spaced at 0.2 mil intervals; a 0.3 mil deep groove is shown at 61 in baffle 60 in FIG. 6A. These baffles have been shown by actual testing to substantially reduce pulse quality distortion caused by acoustic shock waves.

[0102] Applicants have also discovered that acoustic shock effects can be minimized by reducing streamers in the electric discharge. In fact, in a preferred embodiment of the present invention changes made in the chamber head (discussed above) and the new preionizer designed reduced acoustic shock so that acoustic baffles were not needed.

[0103] This preferred embodiment of the present invention includes major improvements in the prior art gas circulator which has greatly improved laser performance. These improvements are in the construction of a braze free blower blade structure. A new non-symmetrical blade arrangement which greatly decreases resonance effects and improved bearings.

[0104] Silicon Free Fan Blade Structure

[0105] Applicants have discovered that a brazing material commonly used in blower blade construction was the primary source of SiF₆in the laser chamber. This gas significantly degraded laser performance for KrF lasers but was a total disaster for ArF lasers. Applicants have identified four solutions to this problem. First the blade structure was machined in segments from a solid block of material (in this case aluminum). Another solution was to cast the blade structure in segments. The segments then are welded together using electron beam welding in which no new material is added. It is also feasible to fabricate the blade structure by joining blades to a frame structure but in this case the joining is by electron beam welding instead of the prior art brazing process. The fourth method is to join the blade to a frame structure using a soldering process using a silicon free solder. Aluminum 6061 is used as the base material for all of the component pieces. These parts are then copper-plated in prelude to the soldering process. With all of the parts assembled, the fan is then soldered together using a low temperature solder, typically 91% tin (Sn) and 9% Zinc (Zn) in a vacuum furnace. This solder is chosen due to its lack of silicon and its ability to work with copper plated aluminum. The assembled and soldered fan is then nickel-plated. This method of construction yields a non-silicon fan that is inexpensive to manufacture.

[0106] Reducing Resonance Effects

[0107] Prior art blower blade structures consisted of a tangential blower with 23 longitudinal blades. These blades were mounted symmetrically at the circumference of the structure. Substantial resonance effects were measured both with respect to fan parameters and actual laser performance. Perturbations in the laser beam were shown to correspond to acoustic waves at 23 times the rotating frequency of the fan. Adverse affects on bearing performance were also measured corresponding to 23 times the fan's rotating frequency.

[0108] Improvements in fan structure design call for a non symmetrical blade arrangement such as that shown in FIG. 14A. An alternative as shown in FIG. 14B where the fan blade structure is formed of 16 separate machined or cart segments with each segment having 23 blades is to rotate each segment by 360°/(15×23) or about 1° relative to the adjacent segment. Another improvement which is made relatively easy in the machine or cast approach to fan blade structure fabrication is to form the blades into air foils as shown at 320 in FIG. 14C. Prior art blades were stamped and a cross section of the two of the stamped blades are shown for comparison at 314. The direction of rotation is shown at 318 and 330 represents the circumference of the blade structure. Whereas conventional blades are uniform in thickness, airfoil blades have a tear shape profile including a rounded leading edge, a thickened midsection and a tapered trailing edge.

[0109] Embodiments of the present invention will be made available with one of two alternative bearing improvements over the prior art.

[0110] Ceramic Bearings

[0111] A preferred embodiment of the present invention includes ceramic bearings. The preferred ceramic bearings are silicon nitride lubricated with a synthetic lubricant, preferably perfluoropolyalkylether (PFPE). These bearings provide substantially greater life as compared to prior art excimer laser fan bearings. In addition, neither the bearings nor the lubricant are significantly affected by the highly reactive fluorine gas.

[0112] Magnetic Bearings

[0113] Another preferred embodiment of the present invention comes with magnetic bearings supporting the fan structure as shown in FIG. 5. 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 in which the rotor 129 of the motor and the rotors 128 of at least two bearings are sealed within the gas environment of the laser cavity and the motor stator 140 and the coils 126 of the magnetic bearing magnets are located outside the gas environment. This preferred bearing design also includes an active magnetic thrust bearing 124 which also has the coils located outside the gas environment.

[0114] As shown in FIG. 3, prior art gas flow from blower 10 was forced to flow between electrodes 6A and 6B by anode support bar 44. However, Applicants have discovered that the prior art designs of support bar 44 such as that shown in FIG. 3 produced substantial aerodynamic reaction forces on the blower which were transferred to the blower bearings resulting in chamber vibration. Applicants suspect that these vibrational forces are responsible for blower bearing wear and possibly occasional bearing failures. Applicant has tested other designs, several of which are shown in FIGS. 12A-12E, all of which reduced the aerodynamic reaction forces by distributing over a longer time period, the reaction force resulting each time a blade passes close to the edge of support bar 44. One of Applicants preferred anode support bar design is shown in FIG. 6A at 84A. This design has substantially greater mass which minimizes anode temperature savings. The total mass of the anode and the anode support bar is about 3.4 Kg. Also, this design comprises fins 84B which provides added cooling for the anode. Applicants tests have indicated that both the acoustic baffles and the aerodynamic anode support bar tend to reduce slightly the gas flow so that is gas flow is limited, the utilization of these two improvements should involve a trade-off analysis. For these reasons two improvements are shown on FIG. 6A and not FIG. 6.

[0115] A preferred pulse power system is manufactured in four separate modules as indicated in FIGS. 8A and 8B, each of which becomes an important part of the excimer laser system and each of which can be quickly replaced in the event of a parts failure or in the course of a regular preventative maintenance program. These modules are designated by Applicants: high voltage power supply module 20, commutator module 40, compression head module 60 and laser chamber module 80.

[0116] High Voltage Power Supply Module

[0117] High voltage power supply module 20 comprises a 300 volt rectifier 22 for converting 208 volt three phase plant power from source 10 to 300 volt DC. Inverter 24 converts the output of rectifier 22 to high frequency 300 volt pulses in the range 100 kHz to 200 kHz. The frequency and the on period of inverter 24 are controlled by the HV power supply control board 21 in order to provide course regulation of the ultimate output pulse energy of the system. The output of inverter 24 is stepped up to about 1200 volts in step-up transformer 26. The output of transformer 26 is converted to 1200 volts DC by rectifier 28 which includes a standard bridge rectifier circuit 30 and a filter capacitor 32. DC electrical energy from circuit 30 charges 8.1 μF C_ocharging capacitor 42 in commutator module 40 as directed by HV power supply control board 21 which controls the operation of inverter 24 as shown in FIG. 8A. Set points within HV power supply control board 21 are set by laser system control board 100.

[0118] The reader should note that in this embodiment as shown in FIG. 8A that pulse energy control for the laser system is provided by power supply module 20. The electrical circuits in commutator 40 and compression head 60 merely serve to utilize the electrical energy stored on charging capacitor 42 by power supply module 20 to form at the rate of 2,000 times per second an electrical pulse, to amplify the pulse voltage and to compress in time the duration of the pulse. As an example of this control, FIG. 8A indicates that processor 102 in control board 100 has controlled the power supply to provide precisely 700 volts to charging capacitor 42 which during the charging cycle is isolated from the down stream circuits by solid state switch 46. The electrical circuits in commutator 40 and compression head 60 will upon the closure of switch 46 very quickly and automatically convert the electrical energy stored on capacitor 42 into the precise electrical discharge pulse across electrodes 83 and 84 needed to provide the next laser pulse at the precise energy needed as determined by processor 102 in control board 100.

[0119] Commutator module 40 comprises C_ocharging capacitor 42, which in this embodiment is a bank of capacitors connected in parallel to provide a total capacitance of 8.1 μF. Voltage divider 44 provides a feedback voltage signal to the HV power supply control board 21 which is used by control board 21 to limit the charging of capacitor 42 to the voltage (called the “control voltage”) which when formed into an electrical pulse and compressed and amplified in commutator 40 and compression head 60 will produce the desired discharge voltage on peaking capacitor 82 and across electrodes 83 and 84.

[0120] In this embodiment (designed to provide electrical pulses in the range of about 3 Joules and 16,000 volts at a pulse rate of 1000 Hz pulses per second), about 250 microseconds (as indicated in FIG. 8F1 ) are required for power supply 20 to charge the charging capacitor 42 to 800 volts. Therefore, charging capacitor 42 is fully charged and stable at the desired voltage when a signal from commutator control board 41 closes solid state switch 44 which initiates the very fast step of converting the 3 Joules of electrical energy stored on charging capacitor C_ointo a 16,000 volt discharge across electrodes 83 and 84. For this embodiment, solid state switch 46 is a IGBT switch, although other switch technologies such as SCRs, GTOs, MCTs, etc. could also be used. A 600 nH charging inductor 48 is in series with solid state switch 46 to temporarily limit the current through switch 46 while it closes to discharge the C_ocharging capacitor 42.

[0121] Pulse Generation Stage

[0122] The first stage of high voltage pulse power production is the pulse generation stage 50. To generate the pulse the charge on charging capacitor 42 is switched onto C₁8.5 μF capacitor 52 in about 5 μs as shown on FIG. 8F2 by closing IGBT switch 46.

[0123] First Stage of Compression

[0124] A saturable inductor 54 initially holds off the voltage stored on capacitor 52 and then becomes saturated allowing the transfer of charge from capacitor 52 through 1:23 step up pulse transformer 56 to C_p-1capacitor 62 in a transfer time period of about 550 ns, as shown on FIG. 8F3, for a first stage of compression 61.

[0125] The design of pulse transformer 56 is described below. The pulse transformer is extremely efficient transforming a 700 volt 17,500 ampere 550 ns pulse rate into a 16,100 volt, 760 ampere 550 ns pulse which is stored very temporarily on C_p-1capacitor bank 62 in compression head module 60.

[0126] Compression head module 60 further compresses the pulse.

[0127] Second Stage of Compression

[0128] An L_p-1saturable inductor 64 (with about 125 nH saturated inductance) holds off the voltage on 16.5 nF C_p-1capacitor bank 62 for approximately 550 ns then allows the charge on C_p-1to flow (in about 100 ns) onto 16.5 nF Cp peaking capacitor 82 located on the top of laser chamber 80 and which is electrically connected in parallel with electrodes 83 and 84 and preionizer 56A. This transformation of a 550 ns long pulse into a 100 ns long pulse to charge Cp peaking capacitor 82 makes up the second and last stage of compression as indicated at 65 on FIG. 8A.

[0129] About 100 ns after the charge begins flowing onto peaking capacitor 82 mounted on top of and as a part of the laser chamber module 80, the voltage on peaking capacitor 82 has reached about 14,000 volts and discharge between the electrodes begins. The discharge lasts about 50 ns during which time lasing occurs within the optical resonance chamber of the excimer laser. The optical resonance chamber described in detail below is defined by a line narrowing package 86 comprised in this example by a 3-prism beam expander, a tuning mirror and an eschelle grating and an output coupler 88. The laser pulse for this laser is a narrow band, 20 to 50 ns, 193 nm pulse of about 5 mJ and the repetition rate is 1000 pulses per second. The pulses define a laser beam 90 and the pulses of the beam are monitored by photodiode 92, all as shown in FIG. 8A.

[0130] The signal from photodiode 94 is transmitted to processor 102 in control board 100 and the processor uses this energy signal and preferably other historical pulse energy data (as discussed below in the section entitled Pulse Energy Control Algorithm) to set the command voltage for the next and/or future pulses. In a preferred embodiment in which the laser operates in a series of short bursts (such as 100 pulse 0.5 second bursts at 1000 Hz separated by a dead time of about 0.1 second) processor 102 in control board 100 is programmed with a special algorithm which uses the most recent pulse energy signal along with the energy signal of all previous pulses in the burst along with other historical pulse profile data to select a control voltage for the subsequent pulse so as to minimize pulse-to-pulse energy variations and also to minimize burst-to-burst energy variations. This calculation is performed by processor 102 in control board 100 using this algorithm during a period of about 35 μs. The laser pulses occurs about 5 μs following the T_ofiring of IGBT switch 46 shown on FIG. 8F3 and about 20 μs are required to collect the laser pulse energy data. (The start of the firing of switch 46 is called T_o) Thus, a new control voltage value is thus ready (as shown on FIG. 8F1) about 70 microseconds after the firing of IGBT switch 46 for the previous pulse (at 2,000 Hz the firing period is 500 μs). The features of the energy control algorithm are described below and are described in greater detail in U.S. patent application Ser. No. 09/034,870 which is incorporated herein by reference.

[0131] This preferred embodiment is provided with electronic circuitry which recovers excess energy onto charging capacitor 42 from the previous pulse. This circuitry substantially reduces waste energy and virtually eliminates after ringing in the laser chamber 80.

[0132] The energy recovery circuit 57 is comprised of energy recovery inductor 58 and energy recovery diode 59, connected in series across Co charging capacitor 42 as shown in FIG. 8B. Because the impedance of the pulse power system is not exactly matched to that of the chamber and due to the fact that the chamber impedance varies several orders of magnitude during the pulse discharge, a negative going “reflection” is generated from the main pulse which propagates back from the chamber towards the front end of the pulse generating system. After the excess energy has propagated back through the compression head 60 and the commutator 40, switch 46 opens up due to the removal of the trigger signal by the controller. The energy recovery circuit 57 reverses the polarity of the reflection which has generated a negative voltage on the charging capacitor 42 through resonant free wheeling (a half cycle of ringing of the L-C circuit made up of the charging capacitor 42 and the energy recovery inductor 58) as clamped against reversal of current in inductor 58 by diode 59. The net result is that substantially all of the reflected energy from the chamber 80 is recovered from each pulse and stored on charging capacitor 42 as a positive charge ready to be utilized for the next pulse. FIG. 8F1, 2 and 3 are time line charts showing the charges on capacitor Co, C₁, C_p-1and Cp. The charts show the process of energy recovery on Co.

[0133] In order to completely utilize the full B-H curve swing of the magnetic materials used in the saturable inductors, a DC bias current is provided such that each inductor is reverse saturated at the time a pulse is initiated by the closing of switch 46.

[0134] In the case of the commutator saturable inductors 48 and 54, this is accomplished by providing a bias current flow of approximately 15A backwards (compared to the directional normal pulse current flow) through the inductors. This bias current is provided by bias current source 120 through isolation inductor LB1. Actual current flow travels from the power supply through the ground connection of the commutator, through the primary winding of the pulse transformer, through saturable inductor 54, through saturable inductor 48, and through isolation inductor LB1 back to the bias current source 120 as indicated by arrows B1.

[0135] In the case of compression head saturable inductor, a bias current B2 of approximate 5A is provided from the second bias current source 126 through isolation inductor LB2. At the compression head, the current splits and the majority B2-1 goes through saturable inductor Lp-164 and back through isolation inductor LB3 back to the second bias current source 126. A smaller fraction of the current B2-2 travels back through the HV cable connecting the compression head 60 and the commutator 40, through the pulse transformer secondary winding to ground, and through a biasing resistor back to the second bias current source 126. This second smaller current is used to bias the pulse transformer so that it is also reset for the pulsed operation. The mount of current which splits into each of the two legs is determined by the resistance in each path and is intentionally adjusted such that each path receives the correct amount of bias current.

[0136] In this embodiment, we refer to the flow of pulse energy through the system from a standard three-phase power source 10 to the electrodes and to ground beyond electrode 84 as “forward flow” and this direction as the forward direction. When we refer to an electrical component such as a saturable inductor as being forward conducting we mean that it is biased into saturation to conduct “pulse energy” in a direction toward the electrodes. When it is reverse conducting it is biased into saturation to conduct energy in a direction away from the electrodes toward the charging capacitor. The actual direction of current flow (or electron flow) through the system depends on where you are within the system. The direction of current flow is now explained to eliminate this as a possible source of confusion.

[0137] By reference to FIGS. 8A and 8B, in this preferred embodiment Co capacitor 42 is charged to (for example) a positive 700 volts such that when switch 46 is closed current flows from capacitor 42 through inductor 48 in a direction toward C₁capacitor 52 (which means that electrons are actually flowing in the reverse direction). Similarly, the current flow is from C₁capacitor 52 through the primary side of pulse transformer 56 toward ground. Thus, the direction of current and pulse energy is the same from charging capacitor 42 to pulse transformer 56. As explained below under the section entitled “Pulse Transformer” current flow in both the primary loops and the secondary loop of pulse transformer 56 is toward ground. The result is that current flow between pulse transformer 56 and the electrodes during the initial portion of the discharge (which represents the main portion [typically about 80 percent] of the discharge) is in the direction away from the electrodes toward transformer 56. Therefore, the direction of electron flow during the main discharge is from ground through the secondary of pulse transformer 56 temporarily onto C_p-1capacitor 62 through inductor 64, temporarily onto Cp capacitor 82, through inductor 81, through electrode 84 (which is referred to as the discharge cathode) through the discharge plasma, through electrode 83 and back to ground. Thus, between pulse transformer 56 and the electrodes 84 and 83 during the main discharge electrons flow in the same direction as the pulse energy. Immediately following the main portion of the discharge, currents and electron flow are reversed and the reverse electron flow is from ground up through the grounded electrode 84, though the discharge space between the electrodes to electrode 83 and back through the circuit through transformer 56 to ground. The passage of reverse electron flow through transformer 56 produces a current in the “primary” loops of transformer 56 with electron flow from ground through the “primary” side of pulse transformer 56 (the same direction as the current flow of the main pulse) to ultimately charge Co negative as indicated qualitatively in FIG. 8F2. The negative charge on Co is reversed as shown in FIG. 8F2 and explained above in the section entitled Energy Recovery.

[0138] A more detailed circuit diagram of the power supply portion of the preferred embodiment is shown in FIG. 8C. As indicated in FIG. 8C, rectifier 22 is a 6 pulse phase controlled rectifier with a plus 150v to −150V DC output. inverter 24 is actually three invertors 24A, 24B and 24C. Invertors 24B and 24C are turned off when the voltage on 8.1 μF Co charging capacitor 42 is 50 volts less than the command voltage and inverter 24A is turned off when the voltage on Co 42 slightly exceeds the command voltage. This procedure reduces the charge rate near the end of the charge. Step up transformers 26A, 26B and 26C are each rated at 7 kw and transform the voltage to 1200 volt AC.

[0139] Three bridge rectifier circuits 30A, 30B and 30C are shown. The HV power supply control board 21 converts a 12 bit digital command to an analog signal and compares it with a feedback signal 45 from Co voltage monitor 44. When the feedback voltage exceeds the command voltage, inverter 24A is turned off as discussed above, Q2 switch 34 closes to dissipate stored energy within the supply, Q3 isolation switch 36 opens to prevent any additional energy leaving the supply and Q1 bleed switch 38 closes to bleed down the voltage on Co 42 until the voltage on Co equals the command voltage. At that time Q1 opens.

[0140] The principal components of commutator 40 and compression head 60 are shown on FIGS. 8A and 8B and are discussed above with regard to the operation of the system. In this section, we describe details of fabrication of the commutator.

[0141] In this preferred embodiment solid state switch 46 is an P/N CM 1000 HA-28H IGBT switch provided by Powerex, Inc. with offices in Youngwood, Penn.

[0142] Inductors 48, 54 and 64 comprise saturable inductors similar to those described in U.S. Pat. Nos. 5,448,580 and 5,315,611. A top and section view of a preferred saturable inductor design is shown respectively in FIGS. 8G1 and 8G2. In the inductors of this embodiment, flux excluding metal pieces such as 301, 302, 303 and 304 are added as shown in FIG. 8G2 in order to reduce the leakage flux in the inductors. The current input to this inductor is a screw connection at 305 to a bus also connected to capacitor 62. The current makes four and one half loops through vertical conductors. From location 305 the current travels down a large diameter conductor in the center labeled 1A, up six smaller conductors on the circumference labeled 1B, down 2A, up 2B, down all of the flux excluder elements, up 3B, down 3A, up 4B and down 4A, and the current exits at location 306. Where a pot like housing 64A serves as a high voltage current lead. The “lid” 64B of the saturable inductor is comprised of an electrical insulator 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, oil insulated components are limited to the saturable inductors and the oil is contained in the pot-like oil containing metal housing 64A which is, as stated above, the high voltage connection output lead. All seal connections are located 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 excluding metal components are in the middle of the current path through the inductor, the voltage allowing a reduction in the safe hold-off spacing between the flux exclusion metal parts and the metal rods of the other turns. Fins 307 are provided to increase heat removal.

[0143] Capacitor banks 42, 52 and 62 are all comprised of banks of commercially available off-the-shelf capacitors connected in parallel. These capacitors are available from suppliers such as Murata with offices in Smyrna, Ga. Applicants preferred method of connecting the capacitors and inductors is to solder or bolt them to positive and negative terminals on special printed circuit board having heavy nickel coated copper leads in a manner similar to that described ins U.S. Pat. No. 5,448,580.

[0144] Pulse transformer 56 is also similar to the pulse transformer described in U.S. Pat. Nos. 5,448,580 and 5,313,481; however, the pulse transformers of the present embodiment has only a single turn in the secondary winding and 23 separate primary windings. A drawing of pulse transformer 56 is shown in FIG. 8D. Each of the 23 primary windings comprise an aluminum spool 56A having two flanges (each with a flat edge with threaded bolt holes) which are bolted to positive and negative terminals on printed circuit board 56B as shown along the bottom edge of FIG. 8D. Insulators 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 1{fraction (1/16)} inches long with a 0.875 OD with a wall thickness of about {fraction (1/32)} inch. The spool is wrapped with one inch wide, 0.7 mil thick Metglas™ 2605 S3A and a 0.1 mil thick mylar film until the OD of the insulated Metglas™ wrapping is 2.24 inches. A prospective view of a single wrapped spool forming one primary winding is shown in FIG. 8E.

[0145] The secondary of the transformer is a single stainless steel rod mounted within a tight fitting insulating tube of electrical glass. The winding is in four sections as shown in FIG. 8D. The stainless steel secondary shown as 56D in FIG. 8D is grounded to a ground lead on printed circuit board 56B at 56E and the high voltage terminal is shown at 56F. As indicated above, a 700 volt pulse between the + and − terminals of the primary windings will produce a minus 16,100 volt pulse at terminal 56F on the secondary side for a 1 to 23 voltage transformation. This design provides very low leakage inductance permitting extremely fast output rise time.

[0146] The Cp capacitor 82 is comprised of a bank of twenty-eight 0.59 nf capacitors mounted on top of the laser chamber pressure vessel. The electrodes 83 and 84 are each solid brass bars about 28 inches long which are separated by about 0.5 to 1.0 inch. In this embodiment, the top electrode 83 is the cathode and the bottom electrode 84 is connected to ground as indicated in FIG. 8A.

[0147] This preferred embodiment of the present invention includes a compression head mounting technique shown in FIGS. 8H1 and 8H2. FIG. 8H1 is a side section view of the laser system showing the location of the compressor head module 60 in relation to electrodes 83 and 84. This technique was designed to minimize the impedance associated with the compression lead chamber connection and at the same time facilitates quick replacement of the compression head. As shown in FIGS. 8H1 and 8H2 the ground connection is made with an approximately 28 inch long slot tab connection along the back side of the compression head as shown at 81A in FIG. 8H1 and 81B in FIG. 8H2. The bottom of the slot tab is fitted with flexible finger stock 81C. A preferred finger stock material is sold under the trade name Multilam®.

[0148] The high voltage connection is made between a six-inch diameter smooth bottom of saturable inductor 64 and a mating array of flexible finger stock at 89 in FIG. 8H1. As above, a preferred finger stock material is Multilam®. This arrangement permits the replacement of the compression head module for repair or preventative maintenance in about five minutes.

[0149] This preferred embodiment comprises a fluorine control system which permits operation within a chosen sweet spot without the use of a fluorine monitor. This embodiment can be described by reference to FIG. 16.

[0150] Laser chamber 1 comprises about 20.3 liters of laser gas. Nominally as described above, the constituents are 3.5 percent argon, 0.1 percent fluorine and the remainder neon. The 0.1 percent fluorine represents a volume of about 0.0023 liters or 2.3 ml of fluorine at 3.5 atm. In mass terms the nominal amount of fluorine in the laser chamber is about 97 mg. The partial pressure of the pure fluorine is about 360 Pa, pure fluorine (corresponding to about 36 kPa of the 1% fluorine mixture). During normal operations with the laser operating at a duty factor of about 40 percent (which is typical for a lithography laser) fluorine is depleted at a rate of about 3.9 mg per hour (this corresponds to about 4% of the fluorine in the chamber per hour). In terms of partial pressure of pure fluorine, this normal depletion rate of fluorine is about 14 Pa per hour. To make up for this depletion using the 1% fluorine gas mixture, a volume of the mixture equivalent to about 1.4 kPa per hour is added to the chamber.

[0151] The fluorine depletion rate for the laser is far from constant. If the laser fan is operating but no lasing is taking place the fluorine depletion rate is cut approximately in half. If the fan is shutdown the fluorine depletion rate is cut to about ¼ the 40% duty factor depletion rate. At 100% duty factor the depletion rate is about double the 40% duty factor depletion rate.

[0152] A preferred embodiment of the present invention comprises software which seals all gas pressure readings to a common temperature eliminates temperature effects when comparing gas functions such as bleeds or injects that occur at different times or on different systems. Gas pressures at arbitrary temperatures may also be scaled to produce the equivalent pressures at operating temperature. Temperature compensation can be added into gas-function software to produce more consistent gas pressures under operating conditions.

[0153] For a fixed number of moles of gas in a closed volume, the ratio P/T (absolute pressure over absolute temperature) is constant. If we define pressure at a reference temperature (P_refand T_ref, respectively) and then change the temperature, we have

[0154] P_ref/T_ref=P/T

[0155] so that either

[0156] P_ref=P*T_ref/T  (1)

[0157] or

[0158] P=P_ref*T/T_ref  (2)

[0159] In both cases, the pressure P at temperature T corresponds to the pressure P_refat temperature T_ref. Equation (1) scales a pressure reading back to the reference temperature T_ref, for example in comparison of pressures recorded at different temperatures. Equation (2) scales a desired pressure at T_refto the equivalent pressure at the current temperature T, which would be used in setting pressure targets for gas functions. For example, assume the desired pressure from a gas fill is 360 kPa at 50° C. (323.2° K), and the chamber temperature is 27° C. (300.2° K). Equation (2) yields a corrected target pressure of P=(360 kPa)(300.2)/(323.2)=334 kPa. If the chamber is filled to this initial pressure at 27° C., the pressure at a temperature of 50° C. will be 360 kPa, as desired.

[0160] By reference to FIG. 16, this process is accomplished as follows. Laser chamber is filled to about 3.6 atmospheres (360 kPa) with the lasing gas mixture discussed above (i.e., 0.1% Fl, 3.5% Ar and 96% Ne). This produces lasing in an optically-resonant cavity consisting of reflecting element 20 and output coupler 22 as a result of periodic discharges between electrodes 21. Laser beam 26 exits the laser through output coupler 88, and a small portion of the beam is extracted with beam splitter 524 and directed to pulse energy monitor 528 which is a fast photodiode. A feedback signal is sent through circuit 530 to high voltage control processor 100 which regulates the high voltage charging system 20 which provides a controlled high voltage input in the range of about 600 volts to about 1100 volts (depending primarily on the fluorine concentration, the time since the last refill and the age of the laser components) to voltage compression and amplification circuit 40-60 which compresses and amplifies the high voltage from high voltage charging system 20 to very short discharge pulses in the range of about 11,000 volts to 20,000 volts across electrodes 83 and 84.

[0161] The magnitude of the discharge voltage pulse needed to produce a desired pulse energy such as 5 mJ depends substantially on the fluorine concentration and for this type of laser. The operating voltage range should preferably be determined by the performance of a test (discussed above) in which the fluorine concentration is allowed to vary over the possible operating range with a corresponding variation in the discharge voltage and pulse parameters measured to determine the “sweet spot” for the particular laser. Once the sweet spot has been determined (for example, 602 volts to 608 volts) the laser can be controlled automatically within the sweet spot using the system described in FIG. 16.

[0162] Sufficient fluorine is injected into the chamber to permit the laser to produce the desired pulse energy with a charging voltage lower than the upper limit of the sweet spot voltage. With valves 510, 506, 504, 515, 517 and 512 closed, the laser is operated with the high voltage control 100 controlling the charging voltage to produce the desired pulse energy until fluorine depletion causes the charging voltage to increase to the upper limit of the sweet spot voltage, in this case 608 volts. Gas control processor 536 monitors the high voltage values utilized by high voltage control 100 and when 608 volts is reached, processor 536 opens valve 506 then reads the chamber pressure from a pressure transducer 507 after its signal stabilizes. (In this embodiment, an average of 10,000 control voltage values are averaged to establish the discharge voltage. At 1,000 Hz this takes 10 seconds.) Stabilization of the pressure requires about 10 seconds. If the chamber pressure is higher than the pre-inject target pressure (e.g., 290 kPa), gas control processor 536 using a predetermined value of ΔP/ΔT will calculate the time necessary then the processor will cause valve 512 to open for time necessary to reduce the chamber pressure to 290 kPa then it closes valve 512. Valve 512 may be blipped in little steps. Processor 536 waits 10 seconds for the pressure to stabilize and reads the chamber pressure again from a signal from transducer 507. The process may be repeated but if it is the laser operator is informed to check the predetermined ΔP/ΔT timing value for valve 512. Also the processor can be programmed to update values of ΔP/ΔT based on measurements during bleed periods.

[0163] After the preinject chamber pressure is determined to be about 290 kPa or less. Valve 506 is closed and valves 515, 510 and 504 are opened and inject bottle 502 is pressurized to $\Delta \mathrm{kPa}\approx \frac{\left(\mathrm{INJECT}\mathrm{SIZE}\right)\left(\mathrm{CHAMBER}\mathrm{VOLUME}\right)}{\left(\mathrm{INJECT}\mathrm{BOTTLE}\right)\left(\mathrm{PLUS}\mathrm{MANIFOLD}\mathrm{VOLUME}\right)}$

[0164] in excess of the chamber pressure with halogen rich gas from gas bottle 514. This gas bottle 14 contains a mixture of 1% Fl, 1% Kr and 98% Ne. Valves 515 and 504 are then closed and valve 506 is opened permitting gas from inject bottle 502 and manifold 537 to flow into laser chamber 1. The volume of gas in bottle 502 and manifold 537 is 0.567 liters and the gas will flow until the pressure is equalized in the chamber, manifold and inject bottle. Assuming an initial chamber pressure of 290 kPa, the new pressure (neglecting possible temperature effects) will be about: $\begin{array}{c}P=\frac{20.3l}{20.867l}\left(290\mathrm{kPa}\right)+\frac{0.567l}{20.867l}\left(310\mathrm{kPa}\right)\\ P=290.54\end{array}$

[0165] The increase in pressure is 0.54 kPa which implies that about $\left(\frac{\mathrm{.54}\mathrm{kPa}}{290\mathrm{kPa}}\right)\left(20.3l\right)=37\mathrm{ml}$

[0166] of the halogen rich gas mixture (at about 290 kPa) was added to the chamber.

[0167] Since the halogen rich gas source is only 1% fluorine, the actual volume of fluorine added is only about 0.37 ml (again at the chamber pressure of 290 kPa). A volume of 0.37 ml of fluorine at 290 kPa represents a mass of fluorine of 1.5 mg. Since in this example, the normal depletion rate of fluorine is in the range of about 3.3 mg/hr. this addition of fluorine would be sufficient to compensate for depletion for a period of 27 minutes so that to maintain substantially constant fluorine concentration with injections of this magnitude, injection would occur at intervals in the range of about 27 minutes. Processor 536 is programmed to begin the injections each time the high voltage control voltage (which preferably is averaged over about 10,000 pulses) reaches the upper limit of the sweet spot voltage range when the laser is operating.

[0168] The system may include flow restrictor 505, which preferably includes a needle valve, which can be manually adjusted to control the rate of flow from inject bottle 502 into chamber 1. If desired, it is adjusted to spread out the flow over about 70 to 90% of the interval between injections. By so doing this procedure approaches a continuous injection system but is much simpler. Alternatively flow restrictor 505 could be a computer controlled needle valve with which continuous injection could be even more closely approximated.

[0169] The process described above basically replaces depleted fluorine on an almost continuous basis. Since the fluorine gas source is only 1% fluorine it also replaces a portion of the Ar and Ne in the chamber on an almost continuous basis. Nevertheless, even though a portion of the laser gas is being substantially continuously replaced, operation in this mode results in a build up of contaminants in the laser gas which reduces the efficiency of the laser. This reduction in efficiency requires an increase in the voltage and/or an increase in the fluorine concentration to maintain the desired pulse energy. For this reason, normal practice with prior art systems suggest that periodically the laser be shutdown for a substantially complete gas exchange. This substantially complete gas exchange is referred to as a refill. These periods may be determined based on number of laser pulses such as 100,000,000 pulses between refills, or refill times may be determined based on calendar time since the last refill or a combination of pulses and calendar time. Also the refill times may be determined by the magnitude of the charging voltage needed for a desired output at a particular fluorine concentration. Preferably after a refill a new test for the “sweet spot” should be run. Also, periodically in between fills the sweet spot test should be performed so that if the sweet spot changes the operator will know where the new sweet spot is.

[0170] A refill may be accomplished using the system shown in FIG. 16 as follows. With valves 510, 506, 515, 512, 517, and 504 closed, valves 506 and 512 are opened, vacuum pump 513 is operated and the laser chamber is pumped down to an absolute pressure of less than 13 kPa. (A direct pump down line may be provided between the chamber 1 and vacuum pump 513 to permit a quick pump down.) Valve 512 is closed. Valve 516 is opened and 3.5% Ar, 96.5% Ne buffer gas from buffer gas bottle 516 is added to the chamber to fill it to a pressure equivalent to 262 kPa at 50° C. (Note that for this 20.3 liter laser chamber, temperature correction can be approximated using a ΔP/ΔT correction of 1 kPa/° C. for a chamber temperature deviation from 50° C. So if the chamber temperature is 23° C. it would be filled to 247 kPa.) Valve 517 is closed and valve 515 is opened and a quantity of the 1% Fl, 3.5% Ar, 96.4% Ne mixture from halogen rich gas bottle 514 is added to chamber 1 to fill it to a pressure equivalent to 290 kPa at 50° C. (Note the same temperature correction discussed above can be used.) This will produce a gas mixture in the chamber of approximately 0.1% Fl, 3.5% Ar and 96.4% Ne. When the chamber is heated to about 50° C. the pressure will be about 3 atm or 290 kPa.

[0171] In a preferred embodiment, a fast photo detector (such as a photo-diode) can be utilized to measure periodically (such as for each pulse or once per second) the temporal pulse shape to determine pulse width or the integral square pulse width (i.e., [∫f(t) dt]²/∫f²(t) dt where f(t) is the temporal pulse wave form). Many measurements can be averaged to improve accuracy. Applicants have discovered that at constant pulse energy the pulse width and especially the integral square pulse width is a strong function of fluorine concentration. Therefore, measurements of pulse width or the integral square pulse width can be used in a feedback scheme to regulate fluorine injections on a continuous or semi-continuous basis to maintain fluorine concentration within a desired range or “sweet spot”. The photo detector shown at 196 in FIG. 10 can be used to monitor pulse width. In addition to serving as a wavelength calibration detector. Or a separate detector could be provided.

[0172] Because O₂in is the presence of high energy, UV photons can be extremely detrimental to optical components. Applicants have developed an N₂purge system that is greatly improved over prior art systems. All optical components associated with the laser that are outside the chamber are purged with nitrogen. This nitrogen system is operated at a pressure that is during operation of the laser only about 10 pascals in excess of atmospheric pressure. This small pressure differential is preferred to avoid a pressure distortion effect on the optical components. Components purged include the line narrowing module, the output coupler, the wavemeter and the shutter assembly.

[0173] Seals are provided at all potential leakage sites output ports connecting of {fraction (1/16)}-inch id. tubes about 6 feet long are provided. The flow through the output ports is monitored to assure proper functioning of the purge system. Preferred flow rates of about 41/minute through the {fraction (1/16)}-inch id. 6 foot long tube is a preferred flow rate to correspond to the desired N₂pressure differential.

[0174] In this preferred embodiment of the present invention the reflectivity of the output coupler has been approximately tripled from about 10% which was typical of prior art narrow-band excimer lasers to about 30%. This is to provide more feedback within the laser cavity to reduce the laser bandwidth and to improve laser efficiency.

[0175] Applicants have developed a grating highly resistant to photon damage. A preferred method for fabricating this grating is described in detail in U.S. patent application Ser. No. 08/939,611, filed Sep. 29, 1997, which is incorporated herein by reference. The first steps in the fabrication of the gratings are similar to the well-known technique of making replica gratings from master or submaster gratings. This technique produces a grating having an epoxy substrate with a thin aluminum reflective coating, which may be cracked or has relatively thick grain boundaries containing oxides and hydroxides of aluminum and typically is also naturally coated with an aluminum oxide film. The grating is subsequently recoated in a vacuum chamber with a thin, pure, dense aluminum overcoat and then also in the vacuum the aluminum overcoat is coated with a thin film of MgF₂. The grating is especially suited for use for wavelength selection in an ArF laser operating producing an ultraviolet laser beam at a wavelength of about 193 nm. The oxygen free aluminum overcoat prevents the ultraviolet light from causing damage by stimulating chemical reactions in grating materials under the aluminum grating surface or in the aluminum oxide film. The MgF2 additionally prevents oxidation on the surface of the aluminum overcoat. A cross-sectional views of the grating are shown in FIGS. 15A and 15B.

[0176] The increase in the reflection of the output coupler and the potential increase in the repetition rate had the effect of substantially increasing the light passing through the line-narrowing module. The additional heat generated by absorption of this additional illumination causes thermal distortion in prior art beam expanding fused silica prisms. To solve this problem prior art fused silica prisms were replaced with calcium fluoride prisms. The three beam expanding CaF prisms 437A, B and C are shown in FIG. 15A. Calcium fluoride has higher thermal conductivity and can handle the additional energy without unacceptable distortion.

[0177] Thermal lensing in CaF prisms is substantially lower than that for fused silica for two reasons. CaF prisms have thermal conductivity about 10 times that of fused silica and absorption at UV wavelength about one half that of fused silica. The advantages of CaF over fused silica have for some time been recognized by stepper makers but Applicants were to the best of their knowledge the first to apply the new high quality CaF technology to and the production use of CaF prism beam expanders.

[0178] Prism 437A is a 1-inch prism mounted with an angle of incidence of 74.0 degrees, prism 437B is a 1-inch prism mounted at 74.0 degrees, and prism 437C is a two inch prism mounted at 74.0 degrees. The magnification of the beam expansion system is 21 and the dispersion of the line narrowing system is 1.09 mr/pm.

[0179] FIG. 10A shows the approximate natural broadband spectrum of a high pulse rate ArF excimer laser. As shown in FIG. 10A, the FWHM bandwidth is about 472 pm. This particular laser may be operated at a rate of up to 1000 Hz, and the typical pulse energy is about 25 mJ per pulse when operated broadband. This broadband spectrum is generally not useful for integrated circuit lithography which typically requires bandwidths of less than 1.0 pm.

[0180] The laser may be narrow banded, using well known prior art techniques. Narrowbanding produces an output spectrum such as that shown in FIG. 10B. In this case the FWHM bandwidth is greatly reduced (by a factor of almost 800) to about 0.6 pm, and the pulse energy is reduced (by a factor of about 5) to about 5 mJ. As a result, the intensity of the pulse at the desired narrow band is very greatly increased as indicated in FIG. 10B.

[0181] As shown in FIG. 10D, the laser 30 may be tuned to operate at any wavelength within the ArF broadband spectrum using tuning mirror 36 in line-narrowing module 31. In a preferred embodiment, the laser is tuned by pivoting mirror 36 with stepper motor 39 so as to slightly change the angle at which the laser beam (expanded by prisms 37A, 37B and 37C) is incident on grating 38. The relationship between wavelength and mirror position as measured by steps of stepper motor 39 is shown in FIG. 10C where one full step of the stepper motor produces a change of about 0.15 pm in the nominal narrowband output center wavelength. The stepper motor scan of a few millimeters is sufficient to scan the output wavelength of laser 30 throughout 500 pm tuning range from about 193,100 pm to about 193,600 pm. Note that the relationship between mirror position and wavelength is not exactly linear, but in the narrow tuning range of this laser, a linear relationship can be assumed and in this preferred embodiment linearity is not required.

[0182] FIG. 10 shows the layouts of a preferred wavemeter unit 120A an absolute wavelength reference calibration unit 190, and a wavemeter processor 197.

[0183] The optical equipment in these units measure pulse energy, wavelength and bandwidth. These measurements are used with feedback circuits to maintain pulse energy and wavelength within desired limits. The equipment calibrates itself by reference to an atomic reference source on the command from the laser system control processor.

[0184] As shown in FIG. 10, the output beam from output coupler 32 (as shown in FIG. 10D) intersects partially reflecting mirror 170, which passes about 95.5% of the beam energy as output beam 33 and reflects about 4.5% for pulse energy, wavelength and bandwidth measurement.

[0185] About 4% of the reflected beam is reflected by mirror 171 to energy detector 172 which comprises a very fast photo diode 69 which is able to measure the energy of individual pulses occurring at the rate of 1000 per second. The pulse energy is about 5 mJ, and the output of detector 69 is fed to a computer controller which uses a special algorithm to adjust the laser charging voltage to precisely control the pulse energy of future pulses based on stored pulse energy data in order to limit the variation of the energy of individual pulses and the integrated energy of bursts of pulses.

[0186] About 4% of the beam which passes through mirror 171 is reflected by mirror 173 through slit 177 to mirror 174, to mirror 175, back to mirror 174 and onto echelle grating 176. The beam is collimated by lens 178 having a focal length of 458.4 mm. Light reflected from grating 176 passes back through lens 178, is reflected again from mirrors 174, 175 and 174 again, and then is reflected from mirror 179 and focused onto the left side of 1024-pixel linear photo diode array 180. The spatial position of the beam on the photo diode array is a coarse measure of the relative nominal wavelength of the output beam.

[0187] Linear Photo diode array 180 is depicted in greater detail in FIG. 10E. The array is an integrated circuit chip comprising 1024 separate photo diode integrated circuits and an associated sample and hold readout circuit. The photo diodes are on a 25 micrometer pitch for a total length of 25.6 mm (about one inch). Each photo diode is 500 micrometer long. Photo diode arrays such as this are available from several sources. A preferred supplier is Hamamatsu. In our preferred embodiment, we use a Model S3903-1024 which can be read at the rate of 2×10⁶pixels/sec on a FIFO basis in which complete 1024 pixel scans can be read at rates of 2000 Hz or greater.

[0188] The coarse wavelength optics in wavemeter module 120 produces a rectangular image of about 0.25 mm×3 mm on the left side of photo diode array 180. The ten or eleven illuminated photo diodes will generate signals in proportion to the intensity of the illumination received and the signals are read and digitized by a processor in wavemeter controller 197. Using this information and an interpolation algorithm controller 197 calculates the center position of the image.

[0189] This position (measured in pixels) is converted into a coarse wavelength value using two calibration coefficients and assuming a linear relationship between position and wavelength. These calibration coefficients are determined by reference to an atomic wavelength reference source as described below. For example, the relationship between image position and wavelength might be the following algorithm:

[0190] λ=(2.3 pm/pixel) P+191,625 pm

[0191] where

[0192] P=coarse image central positions

[0193] About 95% of the beam which passes through mirror 173 as shown in FIG. 10 is reflected off mirror 182 through lens 183 onto a diffuser at the input to etalon assembly 184. The beam exiting etalon 184 is focused by a 458.4 mm focal length lens in the etalon assembly and produces interference fringes on the middle and right side of linear photo diode array 180 after being reflected off two mirrors as shown in FIG. 10.

[0194] The spectrometer must measure wavelength and bandwidth substantially in real time. Because the laser repetition rate may be 1000 H_zor higher, it is necessary to use algorithms which are accurate but not computationally intensive in order to achieve the desired performance with economical and compact processing electronics. Calculational algorithm therefore preferably should use integer as opposed to floating point math, and mathematical operations should preferably be computation efficient (no use of square root, sine, log, etc.).

[0195] The specific details of a preferred algorithm used in this preferred embodiment will now be described. FIG. 11B is a curve with 5 peaks as shown which represents a typical etalon fringe signal as measured by linear photo diode array 180. The central peak is drawn lower in height than the others. As different wavelengths of light enter the etalon, the central peak will rise and fall, sometimes going to zero. This aspect renders the central peak unsuitable for the wavelength measurements. The other peaks will move toward or away from the central peak in response to changes in wavelength, so the position of these peaks can be used to determine the wavelength, while their width measures the bandwidth of the laser. Two regions, each labeled data window, are shown in FIG. 11B. The data windows are located so that the fringe nearest the central peak is normally used for the analysis. However, when the wavelength changes to move the fringe too close to the central peak (which will cause distortion and resulting errors), the first peak is outside the window, but the second closest peak will be inside the window, and the software causes the processor in control module 197 to use the second peak. Conversely, when the wavelength shifts to move the current peak outside the data window away from the central peak the software will jump to an inner fringe within the data window. The data windows are also depicted on FIG. 10F.

[0196] The steps involved are as follows:

[0197] 1. After a laser shot, the photo diode array output is electronically read out and digitized. Data points are separated by an interval physically determined by the spacing of the photo diode array elements, in this case 25 micrometer pitch.

[0198] 2. The digital data is searched to find the peak intensity value in the data window. The previous peak location is used as a starting point. Small regions are searched left and right of the starting point. The search region is extended by small intervals left and right until a peak is found. If the peak is outside the data window, the search will automatically continue until the other peak is found.

[0199] 3. Based on the intensity of the peak, a 50% level is computed as shown in FIG. 11A. The 0% level is measured periodically between the pulses. Based on the computed 50% level, points are examined right and left of the peak until the data points which border the 50% level are found. A linear interpolation is computed between pairs of points, which border the 50% level to find the leftand right half-maximum positions, labeled A, and B in FIG. 11A. These positions are computed to a fraction of a pixel such as {fraction (1/16)}, using an integer data format.

[0200] 4. Steps 2 and 3 are duplicated for the two data windows, giving a total of four interpolated 50% positions. As indicated FIG. 11B, two diameters are computed. D1 is the inner fringe diameter while D2 is the outer fringe diameter.

[0201] 5. An approximation to the wavelength is determined by the coarse wavelength circuit, as described in the preceding section “Coarse Wavelength Measurement.”

[0202] The inner and outer fringe diameters D1 and D2 (in units of pixels) are each converted to wavelength by the following equations:

[0203] λ=λ₀+Cd(D²−D₀²)+N·FSR

[0204] where λ=wavelength corresponding to diameter D

[0205] λ₀=calibration wavelength

[0206] D₀=diameter corresponding to wavelength λ₀

[0207] Cd=calibration constant dependant on the optical design

[0208] FSR=free spectral range of the etalon

[0209] N=integer,=0, ±1, ±2, ±3 . . .

[0210] The values λ₀, K₁, FSR, and D₀are determined and stored at the time of calibration. The value for N is chosen such that:

[0211] |λ−λ_c|≦½FSR

[0212] where λ_c=coarse wavelength determination.

[0213] For example, in a preferred embodiment, we select a reference wavelength λ₀=193,436.9 pm (corresponding to an absorption line of a platinum hollow cathode lamp). At this wavelength, the fringe diameter D₀might be found to be 300 pixels. Cd is a constant which can either be directly measured or calculated from the optical design. In our preferred embodiment, Cd=−9.25×10⁻⁵pm/pixel². Thus, for example, with the laser operating at a different wavelength, the fringe diameter may be measured to be 405 pixels. The possible wavelengths computed by equation (1) are: $\begin{array}{c}\lambda =193,436.9\mathrm{pm}-9.25\times {10}^{-5}\mathrm{pm}\text{/}{\mathrm{pixel}}^2\left[{\left(405\right)}^2-{\left(300\right)}^2\right]+N·\mathrm{FSR}\\ =193,443,7+N·\mathrm{FSR}\end{array}$

[0214] If the free spectral range FSR=20 pm, then the possible values for λ include:

[0215] 193,403.7 pm N=−2

[0216] 193,423.7 N=−1

[0217] 193,443.7 N=0

[0218] 193,463.7 N=+1

[0219] 193,483.7 N=+2

[0220] If the coarse wavelength is determined to be λ_c=193,401.9, for example, then the processor will select the value λ=193,403.7 pm (N=−2) as the solution in the closest agreement with λ_c.

[0221] The inner and outer fringe diameters D₁and D₂as shown in FIG. 9B are each converted into wavelengths λ₁and λ₂, respectively. The final value which is reported for the laser wavelength is the average of these two calculations: $\lambda =\left(\frac{{\lambda }_1+{\lambda }_2}{2}\right)$

[0222] The bandwidth of the laser is computed as (λ₂−λ₁)/2. A fixed correction factor is applied to account for the intrinsic width of the etalon peak adding to the true laser bandwidth. Mathematically, a deconvolution algorithm is the formalism for removing the etalon intrinsic width from the measured width, but this would be far too computation-intensive, so a fixed correction Δλε is subtracted, which provides sufficient accuracy. Therefore, the bandwidth is: $\Delta \lambda =\left(\frac{D_2-D_1}{2}\right)-\mathrm{\Delta \lambda }\epsilon$

[0223] Δλε depends on both the etalon specifications and the true laser bandwidth. It typically lies in the range of 0.1-1 pm for the application described here.

[0224] In this preferred embodiment, wavemeter 120 is calibrated with the optical equipment shown in atomic wave reference unit 190 as shown in FIG. 10.

[0225] The approximately 5% portion of the beam passing through mirror 182 is reflected from mirror 186 through lens 188 and into atomic wavelength reference unit 190. The light passes through diffuser 191, reflects off mirror 192 and is focused by lens 193 to a focal point in the center of vapor cell 194, and is focused again by lens 195 onto photo diode 196. Calibration is accomplished by scanning the wavelength of the laser (with tuning mirror 36 as shown in FIG. 10D) while keeping the output energy of the laser constant, as monitored and controlled by energy detector 69. In this preferred embodiment, the scanning and calibration sequence is automated and programmed into the control electronics for the laser. The wavelength range of the scan is chosen so as to include an absorption wavelength of the platinum vapor cell 194. For example, in this preferred embodiment, the strong absorption at 193,436.9 pm is used, and the laser is programmed to scan from about 193,434 pm to 193,440 pm. When the laser wavelength coincides with the absorption wavelength, a substantial reduction in signal (10-50%) is seen by the photo diode 196. During the scan, two corresponding sets of data are taken, the signal from the photo diode 196, and the wavelength as measured by the wavemeter 120. A representative set of data is shown in FIG. 10J, where the signal from the photo diode 196 is plotted against the wavelength as reported by the wavemeter 120. The processor analyzes the photo diode data and determines the apparent wavelength λ₁which corresponds to the center of the absorption dip. Since the true wavelengths ₀of the atomic absorption reference is known with precision, the calibration error (λ₁−λ₀) can be calculated. This error is then used to automatically correct the calibration constants used by both the coarse and fine wavelength algorithms.

[0226] Conventional etalon mounting schemes typically employ an elastomer to mount the optical elements to the surrounding structure, to constrain the position of the elements but minimize forces applied to the elements. A compound commonly used for this is room-temperature vulcanizing silicone (RTV). However, various organic vapors emitted from these elastomers can deposit onto the optical surfaces, degrading their performance. In order to prolong etalon performance lifetime, it is desirable to mount the etalon in a sealed enclosure that does not contain any elastomer compounds.

[0227] A preferred embodiment includes an improved etalon assembly shown at 184 in FIG. 10. In this etalon assembly is shown in detail in FIGS. 10G1 and 10G2, the fused silica etalon 79 itself is comprised of a top plate 80 having a flange 81 and a lower plate 82, both plates being comprised of premium grade fused silica. The etalon is designed to produce fringes having free spectral range of 20.00 pm at 193.35 nm when surrounded by gas with an index of refraction of 1.0003 and a finesse equal to or greater than 25. Three fused silica spacers 83 with ultra low thermal expansion separate the plates and are 934 micrometer ±1 micrometer thick. These hold the etalon together by optical contact using a technique well known in the optics manufacturing art. The reflectance of the inside surfaces of the etalon are each about 88 percent and the outside surfaces are anti-reflection coated. The transmission of the etalon is about 50 percent.

[0228] The etalon 79 is held in place in aluminum housing 84 only by gravity and three low force springs 86 pressing the flange against three pads not shown but positioned on 120 degree centers under the bottom edge of flange 81 at the radial location indicated by leader 85. A clearance of only 0.004 inch along the top edge of flange 81 at 87 assures that the etalon will remain approximately in its proper position. This close tolerance fit also ensures that if any shock or impulse is transferred to the etalon system through the mounting, the relative velocities between the optical components and the housing contact points will be kept to a minimum. Other optical components of etalon assembly 184 include diffuser 88, window 89 and focusing lens 90 having a focal length of 458.4 mm.

[0229] The diffuser 88 may be a standard prior art diffuser commonly used up-stream of an etalon to produce a great variety of incident angles needed for the proper operation of the etalon. A problem with prior art diffusers is that about 90 percent of the light passing through the diffuser is not at a useful angle and consequently is not focused on the photo diode array. This wasted light, however, adds to the heating of the optical system and can contribute to degradation of optical surfaces. In an alternative embodiment a diffractive lens array is used as the diffuser 88. In this case a pattern is produced in the diffractive lens array which scatters the light thoroughly but only within an angle of about 5 degrees. The result is that about 90 percent of the light falling on the etalon is incident at useful angles and a much greater portion of the light incident on the etalon is ultimately detected by the photo diode array. The result is the light incident on the etalon can be greatly reduced which greatly increases optical component life. Applicants estimate that the incident light can be reduced to less than 10% of prior art values with equivalent light on the photo diode array.

[0230] Details of platinum vapor cell 194 are described by reference to FIG. 10K. This cell is a modified series L2783 hollow cathode lamp tube similar to the one described in U.S. Pat. No. 5,450,202. A glass envelope with UV-transmitting windows 314 and 316 contains neutral gas, neon. The major difference is that the hollow cathode of the cell in this preferred embodiment comprises a very thin platinum “hollow T” shaped sleeve which covers the surface of the cathode 320 facing anode 318 and the inside surface of hollow cathode 320. A DC source of about 150 volts energizes the cell creating a plasma containing platinum ions which are generally contained within the hollow cathode as a vaporous material as shown in FIG. 10K.

[0231] In another preferred embodiment of this invention, focusing element 193 may be used with a suitable aperture to allow a collimated portion of the beam to pass through the cathode 198, through optical element 195 and onto detector 196. In this embodiment, measures must be taken to avoid optical interference effects due to reflections from parallel surfaces of the windows in cell 194. Such measures might include providing a small wedge angle between the inner and outer window surfaces on both the entrance and exit windows of cell 194.

[0232] Since the platinum vapor provides two distinctive absorption lines within the tuning range of the ArF laser, both lines are available if needed to improve the accuracy of the calibration. Procedures could be established to use both lines on each calibration. But preferably the second line could be checked only occasionally if the occasional checks show the calibration at one line calibrates the laser accurately at the second line.

[0233] Tuning of laser 30 of FIG. 10D to a desired wavelength may be performed in various well known ways, which may be mechanically, optically, although tuning the laser optically, as shown in FIG. 10D, is preferred. Any of these known mechanisms for tuning a laser in response to a control signal may constitute the wavelength adjustment mechanism.

[0234] Although a specific structure is shown for detecting platinum absorption lines other suitable embodiments may be employed. These may be implemented using different optical setups, as would be understood by those skilled in the art after reviewing this disclosure.

[0235] Preferred embodiments of the present invention which is especially useful for operation at repetition rates in excess of 1000 Hz, includes a unique cooling technique shown in FIG. 17 for cooling an excimer laser.

[0236] Components of the laser are contained in enclosure 240 which is maintained on the inside at a slight vacuum produced by a blower mounted in a vent as shown at 224 in FIGS. 17 and 4A. The cabinet comprises filtered intake port 241 near the top of the cabinet and a few small leakage sources, such as around gasketed doors, so that the flow of room air through the laser enclosure is about 200 ft³/min which is not nearly sufficient to remove the beat produced by heat producing components of the laser.

[0237] The very great majority (roughly 90 percent) of the waste heat produced by the laser (roughly 12 kw at 100% duty factor) is removed by a chilled water system as shown in FIG. 17.

[0238] In this embodiment the major beat sources in the laser are the high voltage supply 20, the commutator 40, the compression head 60 and the laser chamber 80. For the chamber a water cooled heat exchanger is located inside the chamber and heat is transferred from circulating laser gas to the heat exchanger to the cooling water. Another heat exchanger (not shown) is mounted on an outside surface of the chamber. For the rest of the major heat producing components cooling water is piped to the location of the component and one or more fans force air through a water-to-air heat exchanger onto the component as shown in FIG. 17. For the compression head the circulation is contained as shown, but for the HVPS and the commutator the circulation is onto the component then through other portions of the enclosure to also cool other components before being recirculated back to the heat exchangers.

[0239] Dividing pans 242 and 243 guide the general ventilation air from filter 241 through a path shown by open headed arrows 244 to vent 224.

[0240] This cooling system contains no ducts and except for a water line feeding the heat exchangers inside of and attached to the laser chamber there is no water line connection to any laser component. Since all components (other than the laser chamber) are cooled by air blown about inside the enclosure, there are no cooling connections to make a break when installing and replacing components. Also, the lack of need for ducting greatly increases useable components and working space inside the enclosure.

[0241] The embodiment of the present invention includes a computer controller program with a new algorithm, which substantially reduces prior art variations in pulse energy and total integrated burst energy. The improved equipment and software and a preferred process for reducing energy sigma and burst dose variation is described below.

[0242] As stated in the background section of this specification, the burst mode is a typical mode of operation of an excimer laser used for the light source of a stepper machine in the lithographic production integrated circuits. In this mode the laser is operated to produce “a burst” of pulse at the rate of 1000 Hz for about 110 milliseconds to produce 110 pulses to illuminate a section of a wafer. After the burst the stepper moves the wafer and the mask and once the move is complete which takes typically a fraction of a second the laser produces another 110 pulse burst. Thus, normal operation is bursts of about 110 milliseconds followed by dead times of a fraction of a second. At various times, longer dead time periods will be provided so that other operations can be performed. This basic process continues 24 hours a day, 7 days per week, for several months with the laser typically producing several millions of bursts per day. In the above burst mode, it is usually important that each section of the wafer received the same illumination energy on each burst. Also, chip makers want the pulse to pulse variation to be minimized.

[0243] This preferred embodiment of the present invention accomplishes these objectives with equipment and software which monitors the energy of each pulse (pulse N−1) then controls the energy of the next pulse (pulse N) based on the results of a:

[0244] 1) a comparison of the measured energy of pulse N−1 with a target pulse energy and

[0245] 2) a comparison of the accumulated dose of the burst through pulse N−1 to a target pulse dose through pulse N−1.

[0246] In the typical ArF excimer laser we have been discussing the energy of the first 30-40 pulses is typically less stable than the rest of the burst due to transient effects in the laser gas. After about 40 ms following the first pulse, the pulse energy at constant voltage is relatively constant. In dealing with these early perturbations, Applicants have separated the burst into two time-wise regions, the first region (consisting of a number of the earlier pulses, for example, 40 pulses) called the “K” region and a second region (consisting of the pulses which follow the K region) which Applicants, in this specification, refer to as the “L” region.

[0247] This embodiment of the present invention utilizes prior art excimer laser equipment for pulse energy control. Pulse energy of each pulse of each burst is measured by photodiode 69 as shown in FIG. 10. The overall response time of this photodiode and its sample and hold circuit, including time required to reset the circuit, is less than substantially less than 500 microseconds. The accumulated signal resulting from each approximately 15 ns pulse is stored a few microseconds after the pulse is over and this signal is read six times and the average is stored by computer controller 22 approximately 1.0 microsecond after the beginning of the pulse. The accumulated energy of all the previous individual pulses in a burst is referred to as the burst dose value. Computer controller utilizes the signal representing the pulse energy of pulse N along with target pulse energy and the burst dose value in order to specify the high voltage for the pulse N+1. This calculation requires about 200 microseconds. When the value of high voltage for N+1 is determined, computer controller sends a signal to the high voltage command (VCMD) 120 of the high voltage power supply as shown in FIG. 9 establishing the charging voltage for pulse N+1 that takes a few microseconds. Computer controller 22 commands the high voltage power supply to charge up capacitor Co to the specified voltage. (At high repetition rates in excess of 2000 Hz it may be desirable to start the charging before the calculation is complete.) The charging requires about 250 microseconds so that Co is fully charged and ready to go when it receives a trigger signal for pulse N+1 from trigger circuit 13 as shown in FIG. 2 at 0.5 millisecond after the trigger signal from pulse N. On the trigger signal, capacitor C₀discharges its approximately 700 volts into the magnetic compression circuit shown in FIG. 8B over a period of about 5 microseconds and the pulse is compressed and amplified by the magnetic compression circuit to produce a discharge voltage on capacitor Cp of about 16,100 volts which discharges across electrodes 6 in about 100 ns producing a laser pulse of about 5 mJ and about 15 ns in duration.

[0248] A special preferred process for adjusting the charging voltage to achieve substantially desired pulse energies when operating in a burst mode is described below.

[0249] The process utilizes two voltage adjustment algorithms. The first algorithm applies to the first 80 pulses and is called the KPI algorithm. The second algorithm called the PI algorithm applies to pulses after pulse number 40. This time period after the 80^thpulse is herein called the “L region” of the burst. The initials “PI” refer to “proportional integral” and the “K” in “KPI” refers to the “K region” of the burst.

[0250] The K region comprises pulses 1 through k, where k=40 for this preferred embodiment. The algorithm for setting the charging voltage for pulse N is:

[0251] V_N=(V_B)_N−(V_C)_N−1N=1,2 . . . k

[0252] where: $\begin{array}{c}{V}_N=\mathrm{charging}\mathrm{voltage}\mathrm{for}N\text{'}\mathrm{th}\mathrm{pulse}\\ {\left(V_B\right)}_N=\mathrm{an}\mathrm{array}\mathrm{of}k\mathrm{stored}\mathrm{voltages}\mathrm{which}\mathrm{represent}\mathrm{the}\\ \mathrm{current}\mathrm{best}\mathrm{estimate}\mathrm{of}\mathrm{the}\mathrm{voltage}\mathrm{required}\mathrm{to}\\ \mathrm{produce}\mathrm{the}\mathrm{target}\mathrm{energy}{E}_T\mathrm{for}\mathrm{the}N\text{'}\mathrm{th}\mathrm{pulse}\mathrm{in}\\ \mathrm{the}K\mathrm{region}.\mathrm{This}\mathrm{array}\mathrm{is}\mathrm{updated}\mathrm{after}\mathrm{each}\mathrm{burst}\\ \mathrm{according}\mathrm{to}\mathrm{the}\mathrm{equation}\mathrm{below}.\\ {\left(V_C\right)}_{N-1}=a\mathrm{voltage}\mathrm{correction}\mathrm{based}\mathrm{the}\mathrm{energy}\mathrm{error}\mathrm{of}\mathrm{the}\\ \mathrm{previous}\mathrm{pulse}\mathrm{and}\mathrm{on}\mathrm{the}\mathrm{energy}\mathrm{errors}\mathrm{which}\mathrm{occurred}\\ \mathrm{for}\mathrm{the}\mathrm{previous}\mathrm{pulses}\mathrm{in}\mathrm{the}\mathrm{burst},\mathrm{up}\mathrm{to}\mathrm{pulse}N-1\\ =\sum _{i=1}^{N-1}\frac{(A·{\epsilon }_1+B·D_i}{\left(dE/dV\right)}\end{array}$

[0253] By definition, (V_c)₀=0 $\begin{array}{c}A,B=\mathrm{fractions}\mathrm{typically}\mathrm{between}0\mathrm{and}1,\mathrm{which}\mathrm{in}\mathrm{this}\\ \mathrm{preferred}\mathrm{embodiment}\mathrm{both}A\mathrm{and}B\mathrm{are}0.5\\ {\epsilon }_i=\mathrm{the}\mathrm{energy}\mathrm{error}\mathrm{of}\mathrm{the}i\text{'}\mathrm{th}\mathrm{pulse}\\ =E_i-E_T,\mathrm{where}{E}_i\mathrm{is}\mathrm{the}\mathrm{energy}\mathrm{for}\mathrm{the}i\text{'}\mathrm{th}\mathrm{pulse},\mathrm{and}\\ E_T\mathrm{is}\mathrm{the}\mathrm{target}\mathrm{energy}\\ D_i=\mathrm{the}\mathrm{cumulative}\mathrm{dose}\mathrm{error}\mathrm{of}\mathrm{the}\mathrm{burst},\mathrm{including}\mathrm{all}\\ \mathrm{pulses}\mathrm{from}1\mathrm{through}i\\ =\sum _{k=1}^i{\epsilon }_k\\ dE/dV=a\mathrm{rate}\mathrm{of}\mathrm{change}\mathrm{of}\mathrm{pulse}\mathrm{energy}\mathrm{with}\mathrm{charging}\\ \mathrm{voltage}.(\mathrm{In}\mathrm{this}\mathrm{embodiment},\mathrm{one}\mathrm{or}\mathrm{more}\mathrm{values}\mathrm{of}\\ dE/dV\mathrm{is}\mathrm{determined}\mathrm{experimentally}\mathrm{during}\mathrm{each}\\ \mathrm{burst}\mathrm{and}a\mathrm{running}\mathrm{average}\mathrm{of}\mathrm{these}\mathrm{values}\mathrm{is}\mathrm{used}\\ \mathrm{for}\mathrm{the}\mathrm{calculation})\end{array}$

[0254] The stored values (VB)N are updated during or after each burst according to the following relation: ${\left(V_B\right)}_N^{M+1}={\left(V_B\right)}_N^M-C·\left(\frac{{\epsilon }_N}{dE/dV}-{\left(V_C\right)}_N\right),$

[0255] where the index M refers to the burst number

[0256] C=a fraction typically between 0 and 1, which in this preferred embodiment is 0.3.

[0257] The L region comprises pulses k+1 to the end of the burst (for a preferred embodiment, pulse numbers 41 and higher). The algorithm for setting the charging voltage for pulse N is: $V_N=V_{N-1}-\frac{\left(A·{\epsilon }_{N-1}+B·D_{N-1}\right)}{\left(dE/dV\right)}N=k+1,k+2,\dots$

[0258] where:

[0259] V_N=charging voltage for N'th pulse

[0260] V_N−1=charging voltage for N−1'st (previous) pulse

[0261] The variables A, B, ε₁, D₁, and dE/dV are defined as before.

[0262] A new value for dE/dV is determined periodically, in order to track the relatively slow changes in the characteristics of the laser. In the preferred embodiment, dE/dV is measured by varying or dithering the voltage in a controlled manner during two successive pulses in the L region. For these two pulses, the normal PI energy control algorithm is temporarily suspended and replaced by the following:

[0263] For pulse j: $V_j=V_{j-1}-\frac{\left(A·{\epsilon }_{j-1}+B·D_{j-1}\right)}{\left(dE/dV\right)}+V_{\mathrm{Dither}}$

[0264] where V_Dither=a fixed voltage increment, typically a few volts

[0265] For pulse j+1:

[0266] V_j+1=V_j−2·V_Dither

[0267] After pulse j+1, dE/dV is calculated: $dE/dV=\frac{\left(E_{j+1}-E_j\right)}{2·V_{\mathrm{Dither}}}$

[0268] The calculation of dE/dV can be very noisy, since the expected energy changes due to the dithering voltage can be of the same magnitude as the normal energy variation of the laser. In the preferred embodiment, a running average of the last 50 dE/dV calculations is actually used in the PI and KPI algorithms.

[0269] The preferred method for V_Ditherchoosing is to specify a desired energy dither E_Dither, typically a few percent of the energy target E_T, and then use the current (averaged) value for dE/dV to calculate V_Dither: $V_{\mathrm{Dither}}=\frac{E_{\mathrm{Dither}}}{\left(dE/dV\right)}$

[0270] Pulse j+2 (immediately following the two dithered pulses) is not dithered, but has the special value: $V_{j-2}=V_{j-1}+V_{\mathrm{Dither}}-\frac{\left(A·\left({\epsilon }_{j+1}+E_{\mathrm{Dither}}\right)+B·D_{j+1}\right)}{\left(dE/dV\right)}\left(\mathrm{pulse}j+2\right)$

[0271] This special value for V_j+2is corrected for both the applied voltage dither and the expected energy dither from pulse j+1.

[0272] Many variations on the algorithm described above are possible. For example, dE/dV can be determined in the L region as well as the K. The dithering can be performed once per burst, or several times. The dithering sequence may be performed at a fixed pulse number j as described above, or it may be initiated for a randomly chosen pulse number which varies from one burst to the next.

[0273] The reader should recognize that A, B and C are convergence factors, which could have many other values. Higher values than those specified above could provide quicker convergence but could lead to increased instability. In another preferred embodiment, A={square root over (2+L B)}. This relationship is developed from a recognized technique to produce critical damping. B could be zero in which case there would be no dose correction; however, A should not be zero because it provides a dampening term for the dose conveyance portions of the algorithm.

[0274] If the determined value of dE/dV becomes too small the above algorithm could cause over correction. Therefore a preferred technique is to arbitrarily double dE/dV if the energy sigma value exceeds a threshold. Default values of V and dE/dV are provided for the first pulse of a burst. D is set to zero at the start of each burst. The default dE/dV is set at about three times the expected dE/dV to avoid initial over correction.

[0275] An alternate method of determining dE/dV without the dither referred to above is to merely measure and store the energy and voltage values during laser operation. (Measured rather than specified voltage values can also be used.) These data can be used to determine dE/dV as a function of V for constant pulse energy. The reader should note that each individual value of dE/dV would contain fairly large uncertainties because the elements of the value are differences of measurements having significant uncertainties. However, averaging large numbers of dE/dV values can reduce these uncertainties. The dither exercise to determine dE/dV does not have to be made on each burst but instead could be done periodically such as once every M bursts. Or the measurement of dE/dV could be replaced by a calculation performed by the computer or the value of dE/dV could be inserted manually by the operator of the previous pulse for the calculation of V_N+1. An alternate approach would be to use the actual measured value for V_Nfor this control system. Also the value of V_BINare calculated from specified values, not actual measure values in the above-described embodiment. An obvious alternative would be to use measured voltage values. E_Tis normally a constant value such as 10 mJ but it does not have to be constant. For example, E_Tof the last ten pulses could be smaller than the nominal pulse energy so that percentage deviations from target E_Tfor these pulses would have a smaller effect on the integrated pulse dose. Also, it may be preferable in some situations to program computer controller 22 to provide E_Tvalues that vary from burst to burst.

[0276] Although this very narrow-band ArF excimer laser has been described with reference to a particular embodiment, it is to be appreciated that various adaptations and modifications may be made. For example, many alternative embodiments are discussed in the patent applications listed in the first sentence of this specification, all of which have been incorporated herein by reference. An etalon output coupler could be used to provide additional line narrowing. The invention is to be limited only by the appended claims.