JP2011181935A - Collector for euv light source - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method and an apparatus for removing debris from a reflecting surface of an EUV (Extreme ultraviolet) collector in an EUV light source. <P>SOLUTION: A current source (DC voltage source) 220 may be connected to a collector mirror 150, a metallic backing material (aluminum or nickel) for the mirror 150. The mirror 150 may be heated to a temperature higher than the temperature of ambient gas (helium gas) filling the inside of a chamber 26. In another aspect of debris cleaning, introduction of RF from an antenna schematically shown by an RF frequency voltage 230 and 232 in the chamber 26 may be incorporated. The RF may be connected to the mirror 150 or the metallic backing material, and in this case, a dark shield made of an appropriate conductive material and connected to the earth potential may be formed by covering the back face of the collector mirror 150, and is separated from the mirror by an insulator (air gap) and a voltage (DC from the DC power source 220 connected to the mirror 150 as well). <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to the field of EUV (soft x-ray) light generation for application as a semiconductor integrated circuit lithography exposure exposure source, and in particular to a light collector for such an apparatus.

[Related applications]
This application is a continuation-in-part of US application Ser. No. 10 / 409,254, filed Apr. 8, 2003, the disclosure of which is incorporated herein by reference.

  The ever-increasing need to apply smaller critical dimensions to manufacture semiconductor integrated circuits shifts from deep ultraviolet (DUV) light to extreme ultraviolet (EUV) light, also called soft X-rays. There is a need. To generate such light at an effective energy level that allows sufficient throughput in an EUV lithography tool (eg, a stepper scanner or scanner), eg, for an acceptable lifetime during major part replacement There are various proposals for apparatus and methods.

  For example, there are proposals to use lithium, for example, to generate light with a wavelength centered at 13.5 nm, which is introduced and / or irradiated to form a plasma, and the plasma Excites a lithium atom into a state, and decay from that state results in the majority of EUV photons having an energy distribution centered around 13.5 nm. This plasma may be formed by a discharge using a high density plasma concentration electrode in the vicinity of a solid or liquid lithium source, for example as described in the following patents and references: ie, July 2003 US Pat. No. 6,586,757 entitled “Plasma Concentrated Light Source Using Active Buffer Gas Control” issued to Melynchuk et al. On the 1st, and the above-mentioned patent filed on 8 April 2003 US patent application Ser. No. 10 / 409,254; US Pat. No. 6,566, entitled “Plasma Concentrated Light Source with Tandem Elliptical Mirror Unit” issued May 20, 2003 to Rauch et al. , 668; and the United States entitled “Plasma Concentrated Light Source with Improved Pulsed Power System” issued to Partlo et al. On May 20, 2003. No. 6,566,668, which is assigned to the assignee of the present application, and the applications, patents and other references cited therein, all of which are incorporated herein by reference. As well as other representative patents or published patent applications such as “X-ray exposure apparatus” published on January 24, 2002 with Amemlya et al. As inventor. US Published Application 2002-0009176A1, the disclosure of which is incorporated herein by reference. In addition, as described in the following patents and published US applications, the plasma can cause a target, eg, a liquid metal (eg, lithium) droplet, or a liquid or solid target metal (eg, lithium) to drop in the droplet. May be induced, for example, by irradiating a focused laser beam onto the target: published on September 4, 2001 to Kondo et al. US Pat. No. 6,285,743 entitled “Methods and Apparatus for Soft X-Ray Generation”; “Methods for Generating Extreme Short Wave Radiation” issued to Bisschops on Dec. 10, 2002; US Pat. No. 6,493,423 entitled; “EUV, XUV and X-ray wavelength light sources made from laser plasma” published on October 3, 2002 with Richardson as inventor U.S. Published Application 2002-0141536A1 entitled "Laser Plasma Source for Extreme Ultraviolet Lithography Using Water Drop Targets" issued April 23, 2002 to Richardson. U.S. Pat. No. 6,307,913 entitled “X-Ray, Extreme Ultraviolet and Ultraviolet Radiation Shaped Light Source” issued to Foster et al. Incorporated herein by reference as part of this specification).

From the formation of such plasma and generation of EUV light from the obtained plasma, the amount of energy in the EUV light that is desired to be generated in a desired band is relatively enormous, for example, 100 W / cm ^2. It is necessary to ensure that the collection is performed as efficiently as possible. This efficiency does not degrade significantly, i.e. it can maintain such high efficiency over a relatively long period of operation, e.g. very high pulse repetition rate (> 4 KHz) for an effective 100% duty cycle It is necessary to be able to maintain effective operation for one year. There are many difficult problems to meet these goals, but that aspect will be addressed in the description of aspects of the invention relating to collectors for EUV light sources.

  Some problems that need to be addressed in a viable design include the following: molybdenum (Mo or Mo, for example, through the outer coating of ruthenium (Ru), eg perpendicular to the incident reflecting mirror Diffusion of Li into a layer of a multilayer structure, such as alternating layers of (Moly) and silicon (Si), and its effect on the lifetime of, for example, primary and / or secondary collectors; for example, the chemical reaction between Li and Si, and For example, its effect on the lifetime of the primary and / or secondary collector; for example out-of-band radiation from a laser that produces irradiation for ignition to form a plasma (eg deep UV resist type is brought into lithography in the EUV range, Suppose that such out-of-band light scattered from the target can result in efficient exposure of the resist. For example, 248 nm radiation from a KrF excimer laser that needs to be kept low to avoid any effect on resist exposure; achieving 100 W delivery of output light energy to the intermediate focus; And the secondary collector has a lifetime of at least 5 G pulses; the required conversion efficiency with a given light source, eg a given target (eg a target droplet, or a target in a droplet, or other target), and a high temperature It is to achieve the required lifetime of the multilayer mirror with the operating temperature and out-of-band radiation centered around a wavelength of eg 13.5 nm.

  By utilizing multilayer reflection, normal incident reflection (NIR) mirrors can be constructed for EUV wavelengths of interest, for example, around 5-20 nm, near 11.3 nm, or 13.0-13.5 nm Is well known. The nature of such mirrors depends on the composition, number, order, crystallinity, surface roughness, interdiffusion, period and thickness ratio, the amount of annealing, etc. for some or all of the layers involved And, for example, depending on whether a diffusion barrier is used, what material and thickness the barrier layer is, and its influence on the composition of the layers separated by the barrier layer. For example, Braun, et al., “Multi-component EUV multi-layer mirrors”, Proc. SPIE 5037 (2003) (“Braun”); Feigl, et al., “Heat resistance of EUV multi-layer mirrors” for long-time applications, "Microelectronic Engineering 57-58, p.3-8 (2001) (" Feigl "); based on application 10 / 847,744 filed May 1,2002, 2002 Issued to Barbee, Jr. et al. Claims US Patent No. 6,396,900 ("Barbee") entitled "Multilayer Film with Sharp and Stable Interface for Use"; and Japanese Application filed April 21, 1992 “Multilayer film reflection for soft x-rays” issued to Itoh et al. On June 7, 1994, based on application No. 45,763 filed on April 14, 1993. As described in US Pat. No. 5,319,695 entitled “Body”.

Itoh can be used on a substrate to form a multilayer film composed of materials with different X-ray reflectivity, for example, a silicon layer, a molybdenum layer, and a hydrogenated interface layer formed between each pair of adjacent layers. Describes silicon (Si) and molybdenum (Mo) deposited alternately. Barbee describes a thin layer of a third compound, such as boron carbide (B ₄ C) ₂ , located at both interfaces (Mo-on-Si interface and Si-on-Mo interface). This third layer consists of boron carbide and other carbon and boron based compounds and is characterized as having low absorption at EUV and soft X-ray wavelengths. Thus, a multilayer film comprising alternating layers of Mo and Si includes an intermediate layer of boron carbide (eg, B ₄ C) and / or a boron-based compound between each layer. This intermediate layer can change the surface (interface) chemistry, which can result in increased reflectivity and increased thermal stability, for example, for Mo / Si, interdiffusion is prevented or reduced, Brings about the desired effect. Barbee also describes changing the thickness of the third layer from the Mo-on-Si interface to the Si-on-Mo interface. Barbee also states that the sharpness of the Mo-on-Si interface will typically be about 2.5 times worse than the sharpness of the Si-on-Mo interface; however, Mo-on-Si Due to the deposition of the B ₄ C intermediate layer at the interface, the sharpness of such an interface is comparable to that of the Si-on-Mo interface. Braun describes using carbon barrier layers to reduce interdiffusion at the Mo-on-Si interface to improve reflectivity while improving thermal stability and reducing internal strain. Yes. In Braun, when changing the thickness at the Mo-on-Si interface and the Si-on-Mo interface, the Mo-Si boundary usually forms MoSi ₂ , and the morphology of the Mo and / or Si layer is Special mention is made, for example, that it can be influenced by a carbon-containing barrier layer. In addition, Braun focuses on the effect of barrier layer formation on the interfacial roughness of the Mo-Si interface without the barrier layer. Braun reports a reflectivity of 70.1% at λ = 13.3 nm using a Mo / SiC multilayer. The use of Mo / Si / CB ₄ C, even with annealing, has also been described for reducing internal distortion compared to multilayers, which uses such multilayer mirrors for curved mirrors. Affects ability. Braun also has trade-offs between interlayer contrast, shock reflection and absorption in multi-layer configurations, such as NbSi layers with low but low contrast absorption in Nb, and high contrast but Ru layer absorption. It describes high Ru / Si, both of which have lower properties than Mo / Si multilayer stacks. Braun also describes the theoretical utility of using three layers, such as Mo / Si / Ag, or Mo / Si / Ru, which theoretically have high reflectivity. , The embodiment of Ag is limited in the Mo layer by the voids in the Ag layer at the desired thickness and the calculated best reflectivity at λ = 13.5 nm of the Mo / Si / C / Ru multilayer The theoretical reflectance cannot be achieved because the thickness hinders crystallization in the Mo layer. However, Braun also responds to the theoretically calculated reflectance expectations, possibly due to the initial Mo layer deposition surface roughness that the Mo / Si / C / Ru multilayer stack extends upward through the stack. I find it impossible. Feigl describes the effect of elevated temperatures below 500 ° C on the structural stability of Mo / Mo ₂ C / Si / Mo ₂ C multilayer stacks, including the use of Mo / Si and ultra-thin Mo ₂ C barrier layers. ing. Feigl, for example by annealing Mo and Si at temperatures above 200 ° C., prevents this barrier layer from forming MoSix internal diffusion layers, and Mo / No ₂ C / Si / Mo ₂ C and Mo ₂ / It is mentioned that the Si system remains stable up to 600 ° C. The former system with ultra-thin Mo ₂ C barrier layer and multiple layers (MoSi ₂ is also suggested but not tested) and the latter are formed by replacing Mo in a multilayer system with Mo ₂ C . According to Feigl, the reflectivity of the Mo ₂ C / Si system remains above 0.8 through 600 ° C., whereas the Mo / No ₂ C / Si / Mo ₂ C system is 0.7% at the above temperature. It gradually decays to a slightly lower value, and decreases to about 0.7 at 400 ° C.

  The Applicant proposes certain other materials for the barrier layer, as well as other potential improvements of the multilayer stack for EUV applications.

  A method and apparatus for removing debris from a reflective surface of an EUV collector in an EUV light source is disclosed, comprising a reflective surface comprising a first material and a debris comprising a second material and / or a compound of a second material. The system and method comprises a controlled sputtering ion source (the ion source may include a gas containing atoms of the sputtering ion material); and excitation of the atoms of the sputtering material to an ionized state. The ionized state of the selected energy peak having a high probability of sputtering the second material and a very low possibility of sputtering the first material. Selected to have a surrounding distribution. The guidance mechanism may include an RF or microwave guidance mechanism. The gas is maintained at a pressure that partially determines the selected energy peak, and the guidance mechanism includes the second material from the reflective surface at a rate greater than the inflow rate of plasma debris atoms of the second material. An inflow of ions of the sputtering ion material may be formed that forms a sputter density of the material. The sputter rate may be selected for a predetermined desired lifetime of the reflective surface. The reflective surface may be capped. The collector may comprise an elliptical mirror and a debris shield that may include a radially extending channel. The first material may be molybdenum, the second material may be lithium, and the ionic material may be helium. The system may include a heater for evaporating the second material from the reflective surface. The guide mechanism may be connected to the reflective surface between ignition and ignition. The reflective surface may have a barrier layer. The collector may be a spherical mirror combined with the viewing angle of the incident reflective shell, which may act as a spectroscopic filter due to the choice of layer material for the multilayer stack on the reflector shell. . The sputtering can be combined with heating, the latter removing lithium, the former removing lithium compounds, and the sputtering was due to ions generated in the plasma rather than excited gas atoms. It's okay.

FIG. 1 shows a schematic diagram representing an overall broad concept for a laser-generated plasma EUV light source according to one aspect of the present invention. Fig. 4 schematically illustrates the operation of a system controller according to one aspect of an embodiment of the present invention. FIG. 6 is a side view of an embodiment of an EUV light collector according to one aspect of the present invention when viewed from an irradiation ignition point toward an embodiment according to an embodiment of the present invention. FIG. 2B shows a cross-sectional view of the embodiment of FIG. 2A along line 2B in FIG. 2A. 6 illustrates another alternative embodiment of a vertical angle incident collector according to one aspect of the present invention. 1 shows a schematic diagram of a normal angle incident collector debris management system according to one aspect of the present invention. FIG. 5A-5C illustrate the timing of providing RF and / or DC current to the collector wash signal / collector mirror in accordance with one aspect of one embodiment of the present invention. FIG. 4 shows a schematic cross-sectional side view of an embodiment of the present invention with respect to the viewing angle of an incident collector. FIG. 4 shows a schematic cross-sectional side view of an embodiment of the present invention with respect to the viewing angle of an incident collector. FIG. 5 shows plots of incident angle incidence reflectivity at various reflective surfaces at an incident angle of 5 ° and a predetermined associated wavelength. Figure 5 shows a plot of the incident angle incidence reflectivity at a given associated wavelength for various reflection surfaces for 15 °. FIG. 6 shows a schematic diagram of another embodiment of a collector according to one aspect of the invention. A calculated number of lithium atoms per droplet useful for explaining one aspect of an embodiment of the present invention vs. The diameter of the droplet is shown. The calculated inflow of lithium atoms into the mirror surface vs. useful for explaining one aspect of one embodiment of the present invention. Indicates the radius of the mirror. A calculated required lithium thickness sputter rate vs. useful for explaining one aspect of an embodiment of the present invention. Indicates the mirror diameter. A necessary ratio of molybdenum sputter rate to lithium sputter rate to obtain a one year lifetime using 300 pairs of multi-layer coated mirrors, useful for explaining one aspect of an embodiment of the present invention, vs. Indicates the mirror radius. FIG. 6 shows sputter yields for lithium, silicon, and molybdenum using helium ions useful for describing one aspect of one embodiment of the present invention. FIG. 5 shows normalized helium ion energy with sputter yields for lithium, silicon and molybdenum useful in describing one aspect of one embodiment of the present invention. FIG. 3 shows the current density of helium ions along with sputter yields for lithium, silicon and molybdenum useful in explaining one aspect of one embodiment of the present invention. FIG. 5 shows total helium ion sputtering rates for lithium, silicon, and molybdenum useful for describing one aspect of one embodiment of the present invention. FIG. 4 shows normalized lithium ion energy with sputter yields for lithium, silicon and molybdenum useful in explaining one aspect of one embodiment of the present invention. Radiant power density for a black body, useful for describing one aspect of an embodiment of the present invention vs. Indicates temperature. 1 shows a schematic diagram of one aspect of one embodiment of the present invention. FIG. FIG. 4 shows experimental results for the stopping power of helium and argon buffer gases against tin and lithium ions according to aspects of an embodiment of the present invention. FIG. FIG. 4 shows experimental results for the stopping power of helium and argon buffer gases against tin and lithium ions according to aspects of an embodiment of the present invention. FIG. FIG. 6 shows further experimental results of helium and argon buffer gas stopping powers for both lithium and tin according to aspects of an embodiment of the present invention. FIG. FIG. 6 shows further experimental results of helium and argon buffer gas stopping powers for both lithium and tin according to aspects of an embodiment of the present invention. FIG. FIG. 6 shows further experimental results of helium and argon buffer gas stopping powers for both lithium and tin according to aspects of an embodiment of the present invention. FIG. FIG. 6 shows further experimental results of helium and argon buffer gas stopping powers for both lithium and tin according to aspects of an embodiment of the present invention. FIG. FIG. 6 shows further experimental results of helium and argon buffer gas stopping powers for both lithium and tin according to aspects of an embodiment of the present invention. FIG.

  Referring to FIG. 1, there is shown a schematic diagram of the overall broad concept of an EUV light source, eg, a plasma generated EUV light source 20 according to one aspect of the present invention. The light source 20 may include a pulsed laser system 22, such as a gas discharge laser, such as an excimer gas discharge laser, such as a KrF or ArF laser operating at high power and a high pulse repetition rate, and may be, for example, US Pat. The laser system may be a MOPA configuration as shown in 191, 6,549,551, and 6,567,450. The laser may be, for example, a solid state laser, such as a YAG laser. The light source 20 may also include a target delivery system 24 that delivers a target in the form of, for example, a droplet, a solid particle, or a solid particle contained within the droplet. The target is delivered by the target delivery system 24 to an irradiation site 28, also known as an ignition site or fireball, for example, inside the chamber 26. Embodiments of the target delivery system 24 are described in further detail below.

  A laser pulse delivered from the pulsed laser system 22 through the window (not shown) of the chamber 26 along the laser optical axis 55 to the irradiation site is associated with the target arrival caused by the target delivery system 24. As will be described in more detail below, properly focused to form an ignition or fireball that, depending on the target material, has X-rays (or soft) with certain characteristics including the wavelength of the generated X-ray light. X-ray (EUV)) emitting plasma, the type and amount of debris emitted from the plasma during or after ignition.

  The light source may also include a collector 30, such as a reflector (eg, a shortened elliptical system) with an aperture for laser light to enter the ignition site 28. Embodiments of the collector system are described in further detail below. The collector 30 may be, for example, an elliptical mirror having a first focal point at the ignition site 28 and a second focal point (also referred to as the intermediate focal point 40) at the so-called intermediate point 40, where EUV light is output from the light source. And input to an integrated circuit lithography tool (not shown), for example. The system 20 may also include a target position detection system 42. The pulse system 22 includes, for example, a magnetic reactor switching pulse compression for the oscillator laser system 44 and a pulse power timing monitor system 56 for the oscillator laser system 44 and a pulse power timing monitor system 56 for the amplifier laser system 48, for example. Master oscillator, for example comprised of a dual chamber gas discharge laser system having an oscillator laser system and an amplifier laser system 48, with a magnetic circuit switching pulse compression and timing circuit 52 for the amplification circuit system 48 and an amplification laser system 48 A power amplifier (MOPA) may be included. The pulse power system may include a power source for generating laser output from, for example, a YAG laser. The system 20 may also include an EUV light source controller system 60, which may also include a target position detection feedback system 62 and an ignition control system 64, for example, along with a laser beam positioning system 66.

  The target position detection system may include a plurality of drop imagers 70, 72 and 74 that provide inputs relating to the position of the target drop (eg, relating to the ignition site), and these inputs are used as the target position detection feedback system. Which can, for example, calculate target position and trajectory, from which target errors can be calculated even on an average basis rather than per droplet, which is then input to the system controller 60 Provided as, for example, the controller can provide, for example, a laser position and direction correction signal to, for example, a laser beam positioning system 66, which controls the laser position and direction changer 68, for example. Use this to focus the laser beam Can be changed to a different ignition point 28.

  The imager 72 may be aimed, for example, along the imaging line 75, eg, aligned with the desired trajectory path of the target droplet from the target delivery mechanism 92 to the desired ignition site 28, and the imagers 74 and 76 may be, for example, It may be aimed along imaging lines 76 and 78 that alone intersect the desired trajectory path at several points 80 along the path before the desired firing site 28.

  The target delivery control system 90 is responsive to a signal from the system controller 60, for example, changing the emission point of the target droplet 94 as it is emitted by the target delivery mechanism 92 to reach the desired ignition site 28. You may correct | amend the error in a droplet.

  The EUV light source detector 100 in the vicinity of the intermediate focus 40 also provides feedback to the system controller 60 that can indicate, for example, the timing of the laser pulses and errors in collection, so that the target droplets are effective and efficient. Proper interception may be performed at the correct location and time for LPP / EUV light generation.

  Referring now to FIG. 1A, further details of the controller system 60 and associated monitoring and control systems 62, 64 and 66 shown in FIG. 1 are schematically shown. The controller receives, for example, a plurality of position signals 134, 136 and a trajectory signal 136 from a target position detection feedback signal correlated to a system clock signal provided to a system component on the clock bus 115, for example, by a system clock 116. The controller 60 can, for example, calculate the pre-arrival tracking and timing system 110 that can calculate the actual position of the target at some point in system time, the actual trajectory of the target droplet at a certain system time. Target trajectory calculation system 112, irradiation site time and spatial error calculation system capable of calculating temporal and spatial error signals compared to some spatial and temporal desirable points for firing 114 may be included.

  Controller 60 then provides time error signal 140 to firing control system 64 and spatial error signal 138 to laser beam positioning system 66. The firing control system 64 calculates and provides a resonant charge start signal 122 to the resonant discharge portion 118 of the laser oscillator 44 magnetic reactor switching pulse compression and timing circuit 50, and also includes, for example, a PA magnetic reactor switching pulse compression and timing circuit. 52 may be provided with a resonant charge start signal 124, both of which may be the same signal. Also, a trigger signal 130 is provided for the oscillator laser 44 magnetic reactor switching pulse compression and timing circuit 50 and a trigger signal 132 is provided for the compression circuit portion of the amplifier laser system 48 magnetic reactor switching pulse compression and timing circuit 52. May be. These may not be the same signal, and some may be calculated from the temporal error signal 140 and inputs from the optical out detectors 54 and 56 (for oscillator laser system and amplifier laser system, respectively).

  The spatial error signal may be provided to a laser position and orientation control system 66, which provides, for example, a firing point signal and a line-of-sight signal to a laser beam positioner, which is, for example, a laser system amplifier at the time of firing. By changing one or both of the output position of the laser 48 and the aiming direction of the laser output beam, the laser may be positioned to change the focal point for the ignition site.

  2A and 2B, a side view of the collector 30 looking inside the collector mirror 150 and a cross-sectional view of the rotationally symmetric collector mirror 150 taken along section line 2B of FIG. 2A are shown, respectively. (Although the cross-sectional view will be the same along any radial axis in FIG. 2A).

  As shown in FIG. 2A, the elliptical collector mirror 150 is circular in cross section looking through the mirror, which is the largest cross section of the mirror (as shown in FIG. 2A, almost at the focal point 28 of the elliptical mirror 150). And does not prevent the target droplet 94 from reaching an ignition point designed to be at the focal point 28. However, the mirror may extend further towards the intermediate focus and may have appropriate holes (not shown) in the mirror to allow the target droplet to pass to the focus. . The elliptical mirror also passes through the collection optics 156 so that the LPP laser beam 154 collected through the mirror 150 allows entry to the firing point 28 where it is desired to be the focal point of the elliptical mirror. For example, the aperture 152 shown in a circle in FIG. 2A may be provided. The aperture 152 also depends on the type of control system used, within the requirement (if any) to change the beam optical path to correct the focus of the laser beam 154 on the ignition site, eg, generally It can be processed into a rectangular beam profile.

  2A and 2B show a debris shield 180 according to an aspect of one embodiment of the present invention. The debris shield 180 may be made of a plurality of thin plates 182 made of, for example, molybdenum foil, defining a narrow flat channel 184 extending radially outward from a desired ignition site and extending radially through the debris shield 180. The description of FIG. 2A is very schematic and does not follow a certain scale, and the channel is actually made as thin as possible. Preferably, the foil plate 182 is made thinner than the channel 184 so that it hardly blocks X-rays emitted from the plasma formed by the ignition of the target droplet 94 by the laser beam focused at the ignition site 28. be able to.

  2B, the function of the channel 182 in the debris shield can be seen. A single radial channel can be seen in FIG. 2B, and this same channel will be seen in any cross section through the rotational symmetry axis of the debris shield in the collector mirror 150 and debris shield 180 channels. Each ray 190 of EUV light (and other light energy) emitted from the ignition site 28 travels radially outward from the ignition site 28 and passes through a respective channel in the debris shield 180, as shown in FIG. 2B. In addition, if desired, this will spread completely to the reflective surface of the collector mirror 150. When impinging on the surface of the elliptical mirror 150 at any angle of incidence, the light ray 190 will be reflected in the same channel 180 as the reflected light ray 192 collected on the intermediate focus 40 shown in FIG.

  Referring now to FIG. 3, another embodiment according to one aspect of an embodiment of the present invention is shown. In this embodiment, the debris shield 180 is not shown for the sake of simplicity, and this embodiment will be described in more detail below, when appropriate, for example, similar to the elliptical collector mirror shown in FIGS. 2A and 2B. Can be utilized with or without a debris shield, as described above. In this embodiment, a second collector reflecting mirror is added, which is centered on the focal point of the ignition site 28, ie, the elliptical mirror 150, from the collector mirror 150 to the intermediate focus 40 (shown in FIG. 1). A cross-section of a spherical mirror 202 with an aperture 210 for passing light into The collector mirror 150 functions as described above for FIGS. 2A and 2B with respect to the light ray 190 emitted from the ignition point 28 toward the collector mirror 150. The light beam 204 emitted from the ignition site 28 away from the collector mirror 150 and colliding with the spherical mirror 202 is reflected in the opposite direction through the focal point of the elliptical collector mirror, and is as if it was emitted from the focal point 28 of the elliptical mirror 150. In this way, it passes over the elliptical collector mirror 150 and will therefore also be collected into the intermediate focus 40. As described in connection with FIGS. 2A and 2B, it will be apparent that this will occur with or without the debris shield 180 present.

  Referring now to FIG. 4, another aspect of debris management according to an embodiment of the present invention is schematically illustrated. FIG. 4 shows a collector mirror 150 connected to a current source, such as a DC voltage source 220. This current can be an embodiment of the present invention, such as maintaining the reflector at a selected temperature for evaporating the deposited lithium, for example. Another concept for removing lithium from the first collector mirror is to use helium or hydrogen ion sputtering. The small mass of these ions leads to very low sputter yield, for example for the molybdenum and / or silicon layers of EUV multilayer mirrors made with Mo / Si layers, when maintained at low energies (eg <50 eV). Can do.

  Returning to FIG. 4, a debris cleaning apparatus in accordance with an aspect of an embodiment of the present invention is shown. As illustrated in FIG. 4, the current source (eg, DC voltage source) 220 may be connected to, for example, a collector mirror 150, eg, a metallic backing material (eg, aluminum or nickel) for the mirror 150. Thus, the mirror 150 may be heated to a temperature higher than the temperature of the ambient gas (eg, helium gas) that fills the inside of the EUV light source chamber 26. In accordance with another embodiment of the present invention, other heating of the reflector may occur, for example heating by radiant heating from a heating lamp (not shown) in the container 26.

  Another aspect of debris cleaning may be incorporated, such as RF introduction from, for example, an RF frequency voltage 230 as shown in FIG. 4 and an antenna schematically shown at 232 in chamber 26 of FIG. In fact, similar to the DC shown in FIG. 4, the RF may be connected to a mirror 150 or a metal backing (not shown), in this case made of a suitable conductive material and connected to ground potential. A dark shield (not shown) may be formed over the back surface of the collector mirror 150, which may be an insulator (eg, an air gap) and a voltage (eg, DC from the DC power supply 220 also connected to the mirror 150). ) To separate from the mirror.

As shown in FIG. 5A to FIG. 5C, during predetermined ignition of LPP at times t ₁ , t ₂ , t ₃ , during ignition occurs at t ₁ , t ₂ , t ₃ , and at the time of ignition For a short time on either side, the RF may be replaced by a DC voltage, and during such time, until the next occurrence of the DC voltage during the next ignition, at least immediately after the ignition, if not completely. RF is used. Also, the DC from the power source 220 may be at a positive potential during each ignition time (possibly coexisting with a continuous voltage from the RF source 230) and during such a positive pulse. Then, it may be at an undesired potential.

  The voltage applied to the collector mirror 150 means, on the one hand, to evaporate the lithium released from the plasma upon and after the ignition of a target droplet such as metal debris, for example lithium or other target metal material. . Also, metal elements such as K, Fe, Na, which appear due to impurities in the lithium target droplet itself and which are similarly deposited on the collector mirror 150 after ignition, can also be evaporated. .

  This RF means that a localized ionic plasma of excited He atoms is formed in the vicinity of the collector mirror 150 surface, and these excited ions in the localized plasma are collected by the collector. It is intended to impact lithium atoms or lithium compounds on the mirror 150 to sputter them off the mirror surface. This embodiment of the invention contemplates, for example, balancing between the evaporation mechanism and the sputtering mechanism, for example, RF <500 W power (as dictated by federal regulations for RF frequency sputtering, If at 13.65 MHz, the mirror temperature should be maintained at or near some desired temperature, and if RF increases (eg> 500 W at 13.65 MHz), the corresponding temperature will be Could be reduced.

  6A and 6B, a side view of an embodiment of the invention relating to another collector device is shown. As shown in FIGS. 6A and 6B, the collector 225 may be composed of, for example, a plurality of nested shells, such as an elliptical reflective shell and a parabolic reflective shell (eg, the parabolic shape in FIG. 6A). Shells 230 and 240, elliptical shells 250 and 260) form different compartments. Parabolic shells, such as 230 and 240, are each composed of a first parabolic reflective surface 233,242 and a second parabolic reflective surface 234,244. The elliptical sections 250 and 260 may include elliptical reflective surfaces 252 and 262, for example. In FIG. 6B, another embodiment with two additional parabolic shell sections 232 and 236 is shown, which includes, for example, a first parabolic reflective surface 231 and a second parabolic reflective surface 234. The second section 236 includes, for example, a first parabolic reflective surface 237, a second parabolic reflective surface 238, and a third parabolic reflective surface 239.

  Each reflective shell 230, 240, 250 and 260 has light emitted from the ignition point 21 between them, typically within a 11 ° -55 ° spherical section from the axis of rotation 310 aligned with the focal point of the collector 225 reflective shell. And the shells 230, 240, 250 and 260 are also generally symmetric about the axis of rotation 310. By way of example, the embodiment of FIG. 6A shows an embodiment in which all light is introduced into at least one of the shells 230.240, 250, and 260 in the portion of the sphere just described. In the case of parabolic shell sections 230 and 240, they are incident on the first reflecting surface 233, 242 and reflected towards the intermediate focal point 40, or from the respective second reflecting surface 234, 244 to the intermediate focal point. Reflected. In the case of elliptical shell sections 250, 260, for example, the ellipse formed by the reflective surfaces 252, 262 each has a first focal point at the ignition point 28 and a second focal point at the intermediate focal point 40, All light introduced into each shell 250, 260 is reflected to the intermediate focus.

  Depending on the material of each reflective surface 233, 234, 242, 244, 252, and 262, the angle of incidence of a particular ray, and the number of reflections in a given shell section 230, 240, 250, and 260, a certain average reflection efficiency is achieved. Depending on the structure of these shells, a certain percentage of available light is introduced into each compartment 230, 240, 250 and 260, with an average total of 65% in the shell compartment 230, as shown in FIG. 6A. 19% is reflected and collected with efficiency, 17% is reflected and collected with an average total efficiency of 75% in the shell compartment 240, and 43% is reflected and collected with an average total efficiency of 80% in the shell compartment 250 In the shell compartment 260, 21% would be reflected and collected with an average total efficiency of 91%.

  FIG. 6B shows another embodiment with the addition of two parabolic shell sections 232, 236. These added compartments can serve, for example, to collect up to about 85% light from the axis of rotation, and at least one of the added compartments can be a first reflective surface 237, a second reflective surface. 238 and a third reflective surface 239. As can be seen from FIG. 6B, for example, a light ray 290 emitted from a light source or ignition point is introduced into the parabolic reflection shell section 236 and reflected as a light ray 292 to the second reflective surface 238 and then as a light ray 294 as a third light source. A light beam 296 that is reflected to the reflective surface 239 and then collected can then be formed. Similarly, a ray 300 is introduced into the parabolic reflector shell 236 at the other end of the shell opening, reflected as a ray 320 at the first reflecting surface, reflected as a ray 304 at the second reflecting surface 237, The light beam 306 may be collected at the very end of the third reflective surface. For example, in one of the parabolic shell sections, such as section 240, light ray 280 is introduced into this section 240, reflected from the first parabolic reflecting surface 242 as light ray 282, and reflected as a condensed light ray 283. It must be reflected from the very edge of the surface, and another ray 284 must be introduced into the compartment and reflected from the second reflecting surface as a concentrated ray 286. In the case of one of the elliptical shell sections (eg, 250), the light beam 308 emitted from the ignition point may be introduced into the shell section 250 and reflected from the elliptical reflective surface 252 as the collected light beam 309, and the light beam 318. May be introduced into the shell compartment 250 on the opposite side of the light beam 308 and reflected as a focused light beam 319.

  Referring now to FIG. 7 and FIG. 8, (1) a single layer ruthenium reflective surface, (2) a 14 nm thick single Mo layer and a 4 nm single layer for viewing angles of 5 ° and 15 °, respectively. Mo / Si bilayer stack with one Si layer, (3) For a 10-cycle multilayer Mo / Si stack with a pitch of 9.2 nm and Mo / Si thickness ratio of 22.5: 1, for example with 40 multilayer stacks A plot of the glancing angle of the incident reflectance is shown. In each reflector with Mo / Si, a molybdenum substrate is assumed. If the spectroscopic purity is part of the specification for the light to be delivered, the collector is a nested shell collector, eg in the embodiment of FIGS. 6A and 6B, which is advantageous for reflectivity, for example, near the selected center wavelength. By using the reflection characteristics, it is possible to adjust to a certain wavelength having a certain predetermined bandwidth extension.

  FIG. 9 illustrates aspects of one embodiment of the present invention. In this embodiment, the collector assembly 330 may include, for example, a portion of a spherical mirror reflective surface 332 that directs light originating from the ignition point 28, for example, three nested ellipses at the nested elliptical shell collector 334. It may be a multilayer stack of normal incidence angles that reflect to one of the shell sections 336, 338 and 340. Each of the shell sections 336, 338 and 340 may have a reflective surface 366, 368, 369 inside the respective shell 360, 362 and 364. As shown in FIG. 9, for example, shell section 336 may receive light from rim section 370 of spherical mirror 332, shell section 338 may receive light from an intermediate section of spherical mirror 332, and shell section 340 may be spherical. Light reflected from the central portion of the mirror 322 is received.

  The shell sections 336, 338 and 340 may be coated with multiple layers of Mo / Si rather than the conventionally proposed thick single layer Ru. In accordance with an aspect of one embodiment of the present invention, there are two reflections at a viewing angle of about 5 ° to 15 °, eg, once from a spherical mirror, and once each shell (eg, having an elliptical reflective surface). Reflection from the shell. This can significantly reduce, for example, a significant amount of out-of-band EUV radiation, assuming 13.5 is the desired band, for example. For example, a Wolter-type configuration of Ru mirrors remains highly reflective for both 13.5 nm and 11 nm at 5 ° and 15 ° incident viewing angles, whereas the Mo / Si stack The incident reflection viewing angle coating can be much more selective, especially at about 15 °, as shown in FIGS.

  The embodiments described above do not have the spatial purity of, for example, a grating spectral purity filter, as proposed in the art, but in the reflectance and maintenance of in-band EUV radiation It has significant advantages over other proposed solutions (eg grating filters).

  A lithium LPP EUV light source according to an aspect of an embodiment of the present invention could use a solid stream of liquid lithium or a lithium droplet source. For droplet sources, the sum of atoms per droplet can be calculated, and for solid streams, it can be assumed that only the material in the focused light beam constitutes the droplet upon ignition, but from a debris perspective Adjacent materials in the flow can also form debris, especially when impacted by low energy laser radiation in the skirt of the energy distribution of the focused laser beam.

液滴あたりの原子数の計算は、例えばリチウムの密度およびその原子量から生じる。液滴の質量は次式で表される：

ここで、ρ_lithium＝０．５３５ｇ／ｃｍ³がリチウムの密度であり、その結果は下記の通りである。

ここで、液滴の直径はセンチメータ単位であり、得られる質量はグラム単位である。従って、液滴中の原子数は、液滴質量をリチウムの原子量で乗算し、単位を適正に変換することによって与えられる。

ここで、Ｍ_{lithium atom}＝６．９４１ａｍｕであり、従って、

である。ここで、液滴の直径はセンチメータ単位である。液滴の直径をセンチメータからマイクロメータに変換すると、次式が与えられる：

Since it is considered desirable for the droplet source to have a droplet size that matches the focused light beam, both types of target sources should be considered to have the same droplet size given by the droplet diameter d _droplet Can do. Thus, the droplet volume is given by:

The calculation of the number of atoms per droplet results, for example, from the density of lithium and its atomic weight. The mass of the droplet is expressed as:

Here, ρ _lithium = 0.535 g / cm ³ is the density of lithium, and the result is as follows.

Here, the diameter of the droplets is in centimeters and the resulting mass is in grams. Thus, the number of atoms in the droplet is given by multiplying the droplet mass by the atomic weight of lithium and converting the units appropriately.

Where M _{lithium atom} = 6.941 amu,

It is. Here, the diameter of the droplet is in centimeters. Converting the droplet diameter from centimeters to micrometers gives the following formula:

Number of atoms per droplet vs. The droplet size is shown in FIG. FIG. 10 also shows the number of 13.5 nm photons contained in one 40 mJ pulse, for example. The 40 mJ pulse example assumes a 10% conversion efficiency to 4π steradians and 400 mJ laser pulses. The number of 13.5 nm photons per pulse is given by:
N _Photons = 13.5 nm optical pulse energy (mJ) / E _Photon (eV) · 1.6 × 10 ⁻¹⁶ (mJ / eV) {7};
Here, the energy of the 13.5 nm photon is 91.6 eV. The number of photons obtained for a 40 mJ pulse is 2.72 × 10 ¹⁵ . For example, a 50 μm droplet has one lithium atom per 13.5 nm photon. Usually, one could assume a plurality of photons emitted from each emitting element. This assumption will allow the use of droplet diameters smaller than 50 μm. For example, smaller droplet diameters may be important because the use of lithium in the collector optics and the deposition rate of lithium is proportional to the cube of the droplet diameter.

In accordance with a possible aspect of embodiments of the present invention, assuming no lithium recovery exists, for example, the calculation of the annual usage of lithium is given by multiplying the number of pulses per year by the amount per pulse. As an example, assuming a repetition rate RR, and a duty cycle DC, the resulting mass usage is given by, for example:
Mass per year = M _droplet * RR * 60 sec / min * 60 min / hr * 24 hr / day * 365 day / yr * DC {8}.
That is,
Mass per year = 8.83 × 10 ⁻⁶ · d ³ _droplet · RR · DC {9};
Here, the droplet diameter is in micrometers, and the resulting mass is in grams. For example, a system operating at 6 kHz without lithium recovery and operating at a 100% duty cycle with a droplet diameter of 50 μm will have 6,622 grams or approximately 12.3 liters of lithium capacity for 1 year full operation. Will consume. Under similar conditions, a droplet diameter of 25 μm would consume only 828 grams or about 1.5 liters of lithium.

Assuming that the lithium droplet expands uniformly in all directions when heated by the laser pulse, the atomic flux decreases with the square of the distance from the laser / droplet interaction point (ignition site). Will. The number of atoms released from the interaction point per second is the number of atoms per droplet multiplied by the repetition rate:
Total atomic emission = 2.43 × 10 ¹⁰ · d ³ _droplet · RR {10};
Here, the droplet diameter is in micrometers and RR is the laser repetition rate in Hz.

The atomic flux (number of atoms / cm ² ) through the surface of the phantom sphere centered on the ignition site would be the total atomic emission divided by the surface area in centimeters:
Atomic flux = 1.93 × 10 ⁹ (d ³ _droplet · RR / r ² _sphere ) {11}; The resulting flux is in units of atoms / cm ² . FIG. 11 shows the rate of lithium influx onto the mirror surface vs. 6 assuming a repetition rate of 6 kHz and a duty cycle of 100%. The mirror radii for several droplet diameters [ie (1) 25 μm, (2) 55 μm, (3) 100 μm, and (4) 200 μm] are shown.

  In order to maintain a high mirror reflectivity, the influx of lithium onto the mirror surface can be exceeded by, for example, the lithium sputter rate caused by the incidence of helium ions. In addition, due to the long lifetime of the mirror, the sputtering rate of molybdenum with these same ions (eg, helium ions) must be slower by an order greater than that of, for example, lithium.

  For example, to achieve a one year lifetime for a multilayer coated collector mirror, the required ratio of the sputtering rates of the first and second metals, eg, the ratio of molybdenum to lithium, for example, the first 200 layer pairs are corroded. It can be calculated, for example, by assuming the use of a multilayer stack of 300 layers, for example, so that 100 good layer pairs remain comfortable and effective and high reflectivity is maintained. Also, a sputtering rate of the silicon layer is assumed that is higher than the sputtering rate of the first metal (eg, molybdenum layer) and thus provides a negligible contribution to the mirror lifetime.

A typical EUV mirror has, for example, a molybdenum layer with a molybdenum layer thickness of 2.76 nm and a silicon layer so that 200 pairs for sacrificial corrosion give a molybdenum corrosion of, for example, 552 nm before the end of the lifetime of the mirror. Can consist of a pair of For a useful lifetime of one year, the sputtering rate of molybdenum must be 552 nm / year or less, that is, 1.75 × 10 ⁻⁵ nm / sec or less.

The lithium sputtering rate in atoms per cm ² per ^second (equal to the lithium inflow rate derived above) is given by the following atomic number density and atomic weight per mass density: With proper unit conversion, the lithium monolayer thickness is converted to nm / sec:
Atomic number density = ρ _lithium (g / cm ³ ) / [M _{lithium atom} (amu) · {1.6605 × 10 ⁻²⁴ g / 1 amu}] {12};
Here, ρ _lithium = 0.535 g / cm and M _{lithium atom} = 6.941 amu. The resulting atom number density for lithium is 4.64 × 10 ²² atoms / cm ³ . If this number of lithium atoms is aligned in a cube of 1 cm on each side, the number of atoms per cm along the edge is the square root of the atom number density, ie 3.58 × 10 ⁷ atoms / cm. I will. The resulting monolayer thickness is 2.78 × 10 ⁻⁸ cm, or 0.278 nm. Thus, the number of atoms per cm in the monolayer is the square of the number of atoms per cm along the edge, ie 1.28 × 10 ¹⁵ atoms / cm ² .

The number of atoms (eg, lithium) removed by sputtering per second must match the inflow rate given in Equation 11. Thus, the number of monolayers removed per second is equal to the inflow rate divided by the number of primitives per cm ² in the monolayer. The thickness removal rate is the single layer removal rate multiplied by the single layer thickness. That is,
Thickness removal rate = single layer thickness (nm) × [inflow rate (atoms / cm ² s) / number of atoms in single layer (atoms / cm ² )] {13}.
Using the value for lithium:
Lithium thickness removal rate = 2.17 × 10 ⁻¹⁶ · lithium inflow rate (atoms / cm ² s) {14}.
The unit obtained is nm / sec. The lithium inflow rate shown in FIG. 11 translates into the required lithium thickness sputter rate shown in FIG. 12 for the same 1-4 droplet size, repetition rate and duty cycle. This result further emphasizes the need for small droplet size and large mirror deformation. Otherwise, the required sputter rate may be impractical.

The required thickness sputter rate for lithium can be compared to the maximum allowable thickness sputter rate for molybdenum, eg, 1 year collector life. The data of FIG. 12 divided into the maximum allowable molybdenum sputter rate, 1.75 × 10 ⁻⁵ nm / sec, is shown in FIG. 13 for the same 1-4 droplet size, repetition rate and duty cycle. .

  The question is what is needed to obtain a molybdenum sputter rate that is four orders of magnitude lower than the lithium sputter rate. Sputter yields of lithium and molybdenum when attacked by helium ions are described, for example, in W. Eckstein, “Calculated Sputtering, Reflection and Range Values”, June 24, 2002. This sputter yield data vs. Ion energy of (3) Lithium into Mo at Eth = 52.7 eV, (2) Lithium into Si at Eth = 10.1 eV, (1) Helium into Li Along with the data, it is shown in FIG. As can be seen, a low selected helium ion energy results in an acceptable lithium sputter yield and substantially no molybdenum sputter yield. However, problems can arise from the fact that the incident ion energy cannot be completely controlled. That is, the energy spectrum of incident helium ions is not a delta function. It is the spread of ion energy that must be evaluated when determining the sputtering difference between lithium and molybdenum.

  References to RF induction (RFI) plasma include, for example, J. Hopwood, “Ion Bombardment Energy Distributions in a Radio Frequency Induction Plasma”, Applied Physics Letters, Vol. 62, No. 9 (March 1,1993), pp 940. There is an example of forming an ion energy distribution which is, for example, a FWHM Gaussian shape of 2.5 eV as described in -942.

The peak of the ion energy distribution can be adjusted by appropriately selecting, for example, the electric field strength and the helium pressure. For example, by selecting a peak ion energy of 20 eV, helium ions have a high sputter yield for lithium but safely have energy below the sputtering threshold of molybdenum. FIG. 15 shows a plot of normalized ion energy distribution centered on 20 eV and 2.5 eV FWHM (1 is log scale, 2 is linear scale): (3) lithium, (4) silicon, and ( 5) Shown with sputter yield for molybdenum. It can be seen that there are very few helium ions with energies above the sputtering threshold of molybdenum. In order to determine the sputtering rate of molybdenum under these conditions, it is necessary to calculate the influx of helium ions necessary to keep the mirror surface clean from lithium atoms. Since the bulk of the helium ion energy distribution falls within a region of substantially constant lithium sputter yield, a constant sputter yield of 0.2 atoms per ion can be assumed.
Helium ion inflow (number of ions / cm ² s)
= Lithium inflow (atoms / cm ² s) / sputter yield (atoms / ion) {15}. Therefore, the helium ion density must be five times the lithium inflow density value shown for various conditions in FIG.

  This helium ion influx expressed in Equation 15 may be considered minimal, for example, assuming that lithium does not deposit completely uniformly. In this case, a higher total sputter rate may be required to ensure that no lithium islands are generated. On the other hand, other researchers have shown that the emission of material from the LPP plasma tends to move towards the laser source. Thus, the system can be configured to irradiate a lithium droplet from a direction away from the collector or through the collector mirror aperture so that much of this debris does not strike the collector mirror. Thus, the total lithium load on the mirror may be reduced from the total theoretical amount that strikes the mirror.

従って、正規化されたガウス関数の積分は、√（π（ＦＷＨＭ）²／２ｌｎ（４））であり、ヘリウムイオンのピーク電流密度は次式によって与えられる。
ピークヘリウム電流密度（イオン数／ｃｍ²ｓ／ｅＶ）
＝ヘリウムイオン流入（イオン数／ｃｍ²ｓ）／√（π（ＦＷＨＭ）²／２ｌｎ（４）） ｛１７｝．
ミラー半径が１０ｃｍで２５μｍの液滴の場合をとれば、合計１．８８×１０¹⁵リチウム原子／ｃｍ²ｓをスパッタするために、ピークヘリウム電流密度は、１ｅＶ当り３．３８×１０¹⁵イオン／ｃｍ²でなければならない。このヘリウム電流密度分布（１）は、（２）シリコンスパッタ密度、および（３）リチウムスパッタ密度、並びに経験的に決定された（４）リチウム、（５）シリコン、および（６）モリブデンのスパッタ収量、並びにこれら関数とイオン電流密度の積と共に、ｌｏｇスケールで図１６にプロットされている。この分析の驚くべき有益な結果は、モリブデンについてのピークスパッタ密度が、３．５×１０^-205原子／ｃｍ²／ｅＶ（グラフ上には図示せず）と信じ難いほど小さい値であることを示している。事実、ピークシリコンスパッタ密度でさえ、リチウムの場合よりも３桁以上小さい。 Knowing the total flux of helium ions and assuming a Gaussian energy distribution with a peak of 20 eV and a FWHM of 2.5 eV, the normalized Gaussian distribution is √ (2πσ ² ), where σ ² Gives the variance of the distribution associated with the FWHM by:

Therefore, the integral of the normalized Gaussian function is √ (π (FWHM) ² / 2ln (4)), and the peak current density of helium ions is given by:
Peak helium current density (number of ions / cm ² s / eV)
= Helium ion inflow (number of ions / cm ² s) / √ (π (FWHM) ² / 2ln (4)) {17}.
Taking the case of a 25 μm droplet with a mirror radius of 10 cm, the peak helium current density is 3.38 × 10 ¹⁵ ions / eV per eV to sputter a total of 1.88 × 10 ¹⁵ lithium atoms / cm ² s. Must be cm ² . This helium current density distribution (1) shows (2) silicon sputter density and (3) lithium sputter density and empirically determined sputter yields of (4) lithium, (5) silicon, and (6) molybdenum. , As well as the product of these functions and ion current density, are plotted on a log scale in FIG. The surprising and beneficial result of this analysis is that the peak sputter density for molybdenum is incredibly small, 3.5 × 10 ⁻²⁰⁵ atoms / cm ² / eV (not shown on the graph). Show. In fact, even the peak silicon sputter density is three orders of magnitude lower than that of lithium.

The integral of these sputter densities over all helium ion energies gives the total sputter rate. These integrals are shown in FIG. 17 as dashed curves for (1) lithium and (2) silicon, respectively. This integrated lithium sputter density is 1.88 × 10 ¹⁵ atoms / cm ² s, which matches the lithium inflow rate. The integrated silicon sputter density is 9.17 × 10 ¹⁰ atoms / cm ² s. The integrated molybdenum sputter density is 1.16 × 10 ⁻²⁰⁵ atoms / cm ² s. Thus, the differential sputter rate between molybdenum and lithium is so low that only a small number of layers need to be used, for example, for the collector mirror, for example much less than the 300 base pair mirror concept previously anticipated . A single molybdenum layer will last more than a year under these conditions and this sputter yield model. This property could be further improved by using a debris shield between the ignition point and the collector primary or secondary mirror, but as can be seen from these results, the debris shield is at least completely for the lithium target. May be omitted. This type of stimulated plasma induced ionization sputtering (especially for lithium targets) of debris derived from EUV optics, as can be seen from the above, other target types (eg moving tapes or other types of mobile solid targets) System) could also be made possible. Helium ion sputtering removes lithium atoms from the collector mirror at a sufficient rate, while molybdenum can be configured to sputter at a sufficiently slow rate for a life much longer than one year.

In the embodiment of the invention being discussed, sputtering of molybdenum by, for example, lithium ions must also be considered. This is because, for example, there are debris-forming lithium ions from the ignition plasma that do not reach the surface of the optics but are available to the sputtering plasma and are accelerated towards the mirror surface with the same energy distribution as the helium ions. Because. The literature also provides data on the sputter yield of lithium and molybdenum with lithium ions. This data is shown in curve 1 for lithium at Eth = 36.3 eV with the same normalized lithium ion energy distribution used for helium ions in FIG. In order to calculate the molybdenum sputter density from lithium, the total lithium ion inflow must be known. Unlike this calculation for Helium (Equation 15), it is not clear what the total lithium inflow will be, but a conservative choice would be the total lithium atom inflow generated by the LPP ignition plasma. Using Equation 17 and using the assumption of a 25 μm droplet and a 10 cm mirror radius, 1.88 × 10 ¹⁵ lithium ions / cm ² s would be incident on the mirror. The peak lithium ion current density is 7.06 × 10 ¹⁵ lithium ions / cm ² s per eV, assuming 2.5 eV FWHM spread in the incident ion energy, which is the sputter for molybdenum. When multiplied by yield and integrated over the total ion energy, a total molybdenum sputter density of 2.54 × 10 ⁻⁴⁸ atoms / cm ² s is given. This is much higher than for helium ions, but still much lower than the rate required for a useful lifetime of one year.

The molybdenum sputter density with lithium ions can be converted to a thickness loss rate by using Equations 12 and 13. For molybdenum
ρ _moly = 10.2 g / cm ³
M _{moly atom} = 95.94 amu = 1.59 × 10 ⁻²² g
Molybdenum atom number density = 6.40 × 10 ²² atoms / cm ³
Molybdenum single layer thickness = 2.50 × 10 ⁻⁸ cm = 0.250 nm
Molybdenum single layer atomic density = 1.59 × 10 ¹⁵ atoms / cm ²
It is.

Thus, the sputter thickness rate for molybdenum is 3.99 × 10 ⁻⁶⁴ nm / sec, or 1.25 × 10 ⁻⁵⁶ nm / year when attacked by lithium atoms. This also leads to the conclusion that the beneficial results described above of the sputtering plasma ionization cleaning of the EUV optics, for example by helium ion sputtering, are feasible even when using, for example, lithium sputtering of molybdenum.

  An additional beneficial result is a reconsideration of using the previously proposed eg ruthenium capping layer on eg a multi-layer mirror. Ruthenium capping layers have been proposed to prevent EUV assisted oxidation of the first silicon layer in the Mo / Si stack. Since the molybdenum layer is rapidly oxidized when exposed to room air, multilayer mirrors usually terminate with a silicon layer rather than a molybdenum layer. Prior to the above analysis regarding sputtering plasma cleaning of EUV optics, Applicant has for example, for a multi-layer mirror terminated with silicon, the first silicon layer is etched to expose the first molybdenum layer or ruthenium capping layer, Considered with the expectation that if this approach would be adopted, oxidation of the first molybdenum layer would be avoided. The ultra-slow corrosion rate of molybdenum and the anticipated low corrosion rate for ruthenium allows the use of a ruthenium capping layer that is expected to last for the useful lifetime of the mirror. This does not result in the loss of the first silicon layer, nor does it need to be bothered about what the sputtered silicon atoms produce, nor does it cause oxidation problems for the exposed molybdenum layer. The sputter yield of ruthenium with lithium and helium is expected to be similar to that of molybdenum, but ruthenium still has to be determined because it has a higher atomic weight than molybdenum.

The minimum RF power required to form the desired sputtering plasma at or near the optical surface can be calculated, for example, by assuming that all helium ions that are formed impact the collector mirror, which is necessary Would underestimate the RF power, but should give an approximate estimate of the magnitude. Each helium ion impinging on the collector mirror requires 24.5 eV to ionize, and according to the above example of one embodiment of the invention, an average kinetic energy of 20 eV when it reaches the collector mirror. Must have. Multiplying these two energy values by the required influx of helium ions (9.40 × 10 ¹⁵ ions / cm ² s from FIG. 15) gives the plasma power. Converting the energy unit from eV to J gives a minimum plasma power density of 66.9 mW / cm ² . Multiplying half the mirror surface area (628 cm ² ) with a radius of 10 cm yields a minimum total plasma power of 42 W. If conservatively assuming that only 1% of the plasma power is used effectively, the required plasma power is 4.2 kW, especially considering the very large area that this power can be consumed. Is acceptable. This estimate of plasma power is comparable to the previous assumption of 400 mJ per pulse at 6 kHz LPP laser power (2.4 kW), and assuming a collector mirror defining π steradians, it is half this laser power, ie Will be exposed to 1.2 kW. The thermal load from the LPP is the same as that of the plasma cleaning. The sum of these two powers is 5.4 kW, resulting in a power density on the mirror of 8.6 W / cm ² . Applicant believes that a collector mirror exposed to a power density of 10 W / cm ² or less is easily cooled, for example in a water channel along the back of the mirror or between a grounded shield and the mirror. Believe.

ここでの温度はケルビン単位で、得られる電力密度はＷ／ｃｍ²単位であり、これは図１９にプロットされている。この入射電力の全てを放射させるためには５００℃を越える温度が必要とされるであろうから、当該多層スタックへの損傷を防止するためには、集光ミラーの能動冷却が必要とされるように思われる。 If the plasma power effect is 10% greater, the total power density on the mirror is only 2.6 W / cm ² , allowing the mirror to be radiatively cooled according to Stefan's radiation law. This law states that the radiated power from a black body per square meter of temperature T is given by

The temperature here is in Kelvin and the resulting power density is in W / cm ² , which is plotted in FIG. In order to radiate all of this incident power, temperatures above 500 ° C. will be required, so active cooling of the collector mirror is required to prevent damage to the multilayer stack. Seems to be.

  Referring now to FIG. 20, a damaged EUV optical system (eg, one that has lost its reflectivity) is regenerated, for example, by material (eg, carbon and / or carbon-based molecules) deposited on the reflective surface. An apparatus and method according to an embodiment of the invention is schematically shown for the deposited material to be coated on a contaminant introduced into an EUV plasma chamber, for example, or on an EUV device field reflective surface From sputtered or photonically removed from the layers of the multilayer reflective stack. As can be seen in FIG. 20, the photochemical cleaning apparatus 400 can include a chamber that can be fitted with a collector holding jig 402 adapted to hold the collector for cleaning. Also, the light from the light source 410 simulates the light coming from the supply point at the collector focal point (eg, the ignition point 28 described above) so that the collector 404 is illuminated with light from the target ignition site. Along with the collector holding jig 402 and the light source 410, for example, a source of photon energy (for example, a DUV light source 410) may be included.

According to one embodiment of the present invention, for example, chamber 401 may be purged by use of nitrogen that is first supplied to the chamber through an N ₂ valve and then exhausted from chamber 401 using a gas outlet valve, followed by Alternatively, a fluorine-containing gas (for example, molecular F ₂ or NF ₃ ) may be introduced. The collector 404 is then irradiated by a light source (eg, DUV light in the λ range of 160-300 nm) from a 193 nm KrF excimer laser, for example, in a high power MOPA configuration of about 40 W with a pulse repetition rate of about 4 kHz. . This can serve to induce the production of, for example, a fluorine-based carbon material (eg CF ₄ ) in the gas phase, which is then chambered through the gas outlet valve 420 under a second nitrogen purge. 401 can be discharged.

  Another example of a KrF • DUV light source could be a commercially available DUV lamp, such as a KrCl • DVU lamp.

  Applicants have stated that deposition of about 3.5 nm thick carbon atoms on an EUV optics, eg, a collector reflective surface, can reduce reflectivity by about 5% and 10 nm deposition can be reduced by about 14%. Predict. Such a deposit thickness level is removed from the reflective surface of the collector optics, for example, under treatment with fluorine for a selected time using the selected concentration and DUV light level listed above. It is expected that This process could also employ replenishment of the fluorine supply using a gas flow control valve (not shown), for example, to maintain the desired fluorine concentration during the cleaning process.

Applicants also here to help improve the thermal stability and reflectivity of Mo / Si reflective stacks optimized for EUV light reflectivity, eg, 13.5 nm, according to aspects of embodiments of the present invention. It is proposed that other types of barrier materials may be used in the multilayer reflective mirror stack. In order to facilitate the smoothing of very thin (eg 1 nm) barrier layers that are compatible with eg Mo / Si and possibly MoSi ₂ , for example, retaining an appropriate level of transparency to 13.5 nm light, , A carbide selected from the group consisting of ZrC, NbC, SiC, a boride selected from the group consisting of ZrB ₂ , NbB ₂ , a disilicide selected from the group consisting of ZrSi ₂ , NbSi ₂ , and a nitride The use of an interdiffusion barrier layer comprising BN, ZrN, NbN and Si ₃ N ₄ is proposed. Other such layers could include yttrium, scandium, strontium compounds and / or pure forms of these metals. Of the above, the carbides and borides described above are preferred because of their ability to form smoother diffusion barrier layers using such materials.

In accordance with an aspect of one embodiment of the present invention, Applicants envision a multilayer stack comprising MoSi ₂ / Si, Mo ₂ C / Si, Mo / C / Si / C, and Mo / X / Si / X. Here, the first two are MLMs, and MoSi ₂ or Mo ₂ C is used in place of Mo, which is usually used for Mo / Si mirror coating without an interdiffusion barrier. The other two are used in conjunction with so-called interdiffusion barriers, where C means carbon and X is a suitable material containing further compounds (such as the borides, disilicides and nitrides described above as X materials) Means. Nitride is the presently preferred embodiment by Applicants for interdiffusion barrier layers in applications according to embodiments of the present invention. MoSi ₂ / Si has been described in a paper by Y. Ishii et al. Entitled “Thermal Resistance of Mo / Si, MoSi ₂ / Si, and Mo ₅ Si ₃ / Si Multilayer Soft X-ray Mirrors” (J. Appl. Phys. 78, (1995) p.5227).

  Helium is highly transparent to EUV, which is a good choice for buffer gas (90% transmission is typical). Based on the partial pressure required for efficient sputtering (only a few mTorr), the permeation of the helium buffer gas is approximately 100%. Possible collector multilayer surfaces include, for example, 300 coating pairs instead of the usual 90 pairs. This extra pair will not improve the reflectivity over 90 pairs, but these extra layers are used when the top layer is eroded and removed if needed. With 300 layer pairs of mirrors, the difference in sputtering rate between the lithium and the mirror need not be so high that a single mirror layer lasts several months at a time, for example. Instead, there may be, for example, an extra 210 layer pair beneficial for sustainable mirror erosion.

Lithium compounds that may be generated in the LPP vessel, such as LiH, LiOH, Li ₂ CO _3, etc., may have melting points in excess of 600 ° C. and therefore may not evaporate from the mirror. These may in some cases form a shell over the lithium deposited on the mirror surface. However, they can be sputtered very efficiently, for example, by a sputtering ion plasma containing ionized He atoms, or can be sputtered by fast lithium ions released from the plasma and the atomic form of lithium itself. Will.

  The sputter rate required to stop lithium deposition in advance could be much higher for EUV light sources than for example specified in the literature implemented in modern deposition and etching mechanisms. At least part of the reason is, for example, due to a combined approach to keep lithium away from the mirror surface. In accordance with an aspect of one embodiment of the present invention, Applicant uses evaporation to remove the bulk of lithium while very light sputter rate to remove unavoidable lithium and carbon compounds deposited on the mirror surface. Is assumed to be used. However, even very light sputtering plasmas that impinge on at least the primary and secondary reflective surfaces could also have beneficial carbon and other lithium compound removal properties. The use of this concept for non-intermediate focal points, such as illuminator reflective surfaces and projection reflective surfaces, will prove beneficial in removing debris that happens to reach the reflective surface of the lithography tool. In the lithography tool itself, for example, because the deposition rate is low, the thermal load and sputtering rate may be low enough for this to be effective.

  Lithium and lithium compounds sputtered with lithium released from the plasma and not collected on the reflective surface extend from a cold finger (not shown) contained in the EUV light source container, such as the inner wall of the container and from the collector It can be trapped in a cooled form of water cooled fins or plates that extend out of the optical path to the intermediate focus.

For example, in the case of tin as the source element, it may be possible to use, for example, a metal hydride (eg SnH ₄ ), which is vapor at room temperature, with a hydrogen plasma to clean the collector in a tin-based LPP light source. unknown. Hydrogen has a high 13.5 nm transmission and the resulting SnH ₄ could be pumped out rather than trapped on a cold finger like lithium.

  Applicants have tested the stopping power of, for example, helium and argon against both tin and lithium ions. The results are shown in FIGS. 21A and 21B. The two graphs have the same data that differ only in scale. Lines 500, 502 and 503 for tin at different measurement distances from the source plasma (96.5 cm, 61 cm and 32.5 cm, respectively) are the helium buffer for the solid line and the argon buffer for the dashed line. Line 506 is for lithium. If pressure * distance product scaling is applied, these three sets of tin data will generally overlap each other's top.

  Applicants have also observed that for a given gas pressure, argon has a stopping power that is at least 10 times higher than helium. Also, lithium can be stopped with less buffer gas than tin. Also, when scaled to the true working distance (-10 cm) of the LPP collector, the required buffer pressure would need to be in the range of about 10 mT for tin, even in the case of argon. Let's go. Since xenon and tin have approximately the same atomic weight, Applicants expect that the required buffer pressure for xenon LPP will also be in the range of 10 mT. Such a high buffer pressure can present EUV self-absorption problems for xenon and tin. However, this is not the case for lithium because of both the low buffer pressure requirements and low EUV absorption of lithium.

  For example, if you continue to test the stopping power of a buffer against fast ions, for example using a Faraday cup that collects and measures ions at known distances through known aperture sizes at increasing buffer gas pressures, If this Faraday cup signal decreases, a measurement of the ion stopping force is given. The results for tin and lithium are shown in FIGS. 22A-22E. 22A and 22B show raw Faraday cup signals vs. tin and lithium, respectively. Shows time. In FIGS. 22C and 22D, these signals are plotted against ion energy using time-of-flight for tin and lithium, respectively. In FIG. 22E, the area under these curves is plotted against the product of buffer gas pressure * distance, with the lower plot line (1) being tin and the upper plot line ( 2) is lithium.

  The surprising result of this analysis by the applicant is that the final graph shows that the Faraday cup signal vs. for both tin and lithium. It was to show that the P * D product of the buffer gas was approximately the same for both elements. Applicant has stated that this analysis did not actually measure the loss of ions captured by the Faraday cup, but instead measured neutralization of ions by buffer gas (so-called electron capture by ions). I am convinced that it can be explained. If the ion is neutralized, it will not register in the Faraday cup. This is because, for example, tin has a higher electron capture cross section than lithium, especially considering that tin ions are highly charged (7-11 times ionized) and that lithium can only be ionized at most 3 times. This can be explained from the fact that it can be included. The result of the stopping force shown in FIG. 22E can be considered as an evaluation of the problem of the buffer gas stopping force in that the stopping force cannot be better than the predicted value by these curves.

By looking at the observed value of the stopping force as the upper limit, it is possible to calculate the required pressure of the argon buffer gas to extend the collector mirror lifetime to 100 B pulses. Starting from the Engineering Test Standard (ETS) established by EUV LLC, which reported that one multi-layer mirror pair using xenon LPP and a collector distance of 12 cm was corroded every 15M pulses, and Assuming that the reflectivity of the multilayer mirror does not degrade significantly until 10 layer pairs are removed, the ETS collector mirror will have a lifetime of 150M pulses compared to the requirement of 100B pulses. This leads to the conclusion that a 666X reduction in the corrosion rate is necessary. In FIG. 22E, a P * D product of about 500 mT * cm would be required to achieve this level reduction. For example, a working distance of 12 cm creates a need for an argon pressure of 42 mT, for example. This also concludes that lithium is a better target than xenon, for example, because 42mT buffer pressure is not enough for xenon LPP due to the strong EUV absorption of xenon trapped in argon buffer. Can bring. However, in the case of lithium, this amount of buffer pressure is not a problem for lithium absorption. Tin may also be sufficient depending on, for example, the vapor pressure of SnH ₄ and the rate of evaporation from the collector mirror surface. Thus, it appears that a relatively large buffer gas pressure is required, which leads to the conclusion that xenon is not a good target (and so will tin), but lithium looks best.

  Applicants have also noted that before the effect of heating the collector reflective surface is achieved, the material to be vaporized must have a certain thickness (eg, 50 mm, eg, 10 monolayer) before the published vapor pressure is achieved. Even though it is adversely affected by the fact that the material (e.g. lithium) may be difficult to evaporate directly from the mirror surface, the transmittance of such a thickness of lithium on the mirror surface is about 95. It was determined that the layers on such mirrors would not be significantly impaired from the overall CE, eg 13.5 nm. In addition, such an “evaporation-free” layer of lithium may actually be beneficial in that it can protect the collector mirror from the onslaught of fast lithium atoms and ions. This lithium layer will be sputtered instead of the molybdenum layer of the multilayer mirror. Since xenon is a gas, it will not form such a protective layer, and the tin layer will only transmit 52% due to its very high EUV absorption.

For example, for an ion energy of about 1 keV (spatter rate tends to saturate above this energy level), if the sputter yield of lithium for molybdenum is much lower than the sputter yield of xenon for molybdenum:
Incident ion
Target material Lithium Xenon
Lithium 0.21? ?
Molybdenum 0.081 1.45
Xenon will sputter molybdenum at a rate 18 times higher than lithium. This difference alone will give a collector lifetime of 2.7 B pulses without any other changes. A thin layer of steady state lithium without evaporation will give the remaining 37X reduction in sputter rate. If not, the EUV-LLC concept of producing mirrors with ~ 100 extra sacrificial layer pairs can add a 10X, for example 27B pulse increase in lifetime, which results in low corrosion from lithium. In combination, a collector lifetime of 100B pulses could be given.

  Applicants have also tested the effectiveness of collector mirror electrostatic protection. This concept has been proposed in the literature to generate an electric field between the light source LPP and the collector mirror so as to climb the potential well as high energy ions move toward the mirror. This potential well can be deep enough to lose all of their kinetic energy before they reach the mirror. In fact, they turn around the potential well and are driven away under the potential well. However, Applicants have found that attempts to do this by running an electrical connection through the vessel to the collector mirror are invalid due to a target bias that drops to near zero when pulsing the laser. This was determined to be the result of the high peak current required to maintain the bias voltage and the large wire required (thus reducing all of the voltage along the inductance of the wire). In order to correct this problem, the applicant installed a capacitor inside the vacuum vessel and built a low inductance bus work between the ground and the target plate. Inductance was measured by placing the copper sheet up and around the target and attaching it to ground. By charging the capacitor to a low voltage and discharging it by pressing the copper sheet against the target, Applicants measured the ring voltage waveform and inferred the inductance. The result was a half-cycle discharge waveform of 104nH at 697ns. This discharge period is much longer than the laser pulse and subsequent EUV emission, and there is concern about whether this bias can be maintained during the critical period (~ 20 ns) when ions are formed and exit the plasma region. Arise. However, the applicant has determined that such a short time scale is not important. The important thing is for example. Maintaining or reestablishing the target bias within a time scale that is short compared to the migration time of ions from the target to the mirror. In the current configuration, the ion transfer time is 2.5 μs, so a circuit half period of 0.7 μs should be sufficient.

  In testing this device, Applicant was surprised to discover that a full 0.47 μF capacitance drained from its −1000 V on almost exactly the same time scale as when measuring inductance using a copper strap. It was. Applicants have determined that the laser pulse initiates a discharge between the target plate and the vessel wall. This discharge completes the circuit between the high voltage end of the capacitor and ground and drains the capacitor as if a copper strap was placed across it. Clearly, the events that develop during and immediately after the laser pulse are that a plasma is formed at the target point and that the plasma emits a large amount of hard UV and EUV radiation across the vessel. . Most of the energy of these photons exceeds the work function of the metal inside the container, so photoelectrons are generated on all the metal surfaces. These photons also have sufficient energy to ionize any gas atoms present in the vessel. In this case, argon is used as a buffer gas and is easily ionized by hard UV and EUV radiation generated by LPP. And eventually, electrons and ions are formed in the LPP and in the outward flow into the volume of the container. An exception is for ions attracted to a biased target plate. They strike the plate and form secondary electrons. In essence, a discharge formation occurs between two metal plates held at a potential between each other as if the device is a laser triggered discharge switch.

  In addition, there are several possibilities for effective electrostatic repulsion, but this is more complicated and not actually electrostatic. The idea is to pulse the bias so that it exists only after the initial event of the laser pulse. In just a few hundred ns most of the electrons will hit the container wall and of course radiation will occur. At this time, it would be possible to repel or attract the ions by applying a bias away from the collector mirror.

  Those skilled in the art will recognize that the preferred embodiments and aspects of the invention described above are not exclusive, and that other embodiments to the embodiments described above may be used without departing from the spirit and scope of the invention disclosed herein. It will be appreciated that variations and additions can be made. Therefore, the claims should not be construed as limited to the embodiments and aspects disclosed above, but the described elements and equivalents of the described elements It should be construed to be included in the spirit. By way of example, other target materials and multi-layer reflective coating materials may be ions induced by the formation of a sputtering plasma in the vicinity of the light reflective surface (this ion may be other than helium, for example H, N or O). ), Which may have the same relationship as described above, allowing continuous cleaning. Also, for example, the heating mechanism for the reflective surface could be a heating lamp directed at the reflective surface. Those skilled in the art will appreciate other such modifications and additions.

Claims (10)

A method of regenerating a collector for an EUV light source having a reflective surface contaminated with debris, comprising:
A method comprising photochemically cleaning the reflective surface of the collector under irradiation from an ultraviolet light source in a cleaning chamber containing a gas containing a carbon oxidant.   The method according to claim 1, wherein the irradiation from the ultraviolet light source is performed by a light source irradiated from a supply point substantially at a point light source position corresponding to a plasma ignition site of a normally used EUV light source of the collector.   The method of claim 1, wherein the ultraviolet light source is a DUV light source. The method of claim 1, wherein the gas containing the carbon oxidant comprises molecular F ₂ . The method of claim 1, wherein the gas containing carbon oxidant comprises NF ₃ .   The method according to claim 1, wherein the photochemical cleaning is performed using light having a wavelength in a range of 160 to 300 nm.   The method of claim 1, wherein the ultraviolet light source is a DUV lamp.   The method of claim 1, comprising purging the cleaning chamber to evacuate the fluorine-based carbon material in the gas phase. An apparatus for regenerating a collector for an EUV light source having a reflective surface contaminated with debris,
A cleaning chamber;
An ultraviolet light source,
A source of a gas containing a carbon oxidant that photochemically cleans the reflective surface of the collector under irradiation from the ultraviolet light source in the cleaning chamber;
A device characterized by comprising:   The apparatus according to claim 9, wherein the ultraviolet light source generates light having a wavelength in a range of 160 to 300 nm.

Images