Opacity of neutral and low ion stages of Sn at the wavelength 13.5 nm used in extreme-ultraviolet lithography M. Lysaght, 1 D. Kilbane, 2 N. Murphy, 1 A. Cummings, 1 P. Dunne, 1 and G. O’Sullivan 1 1 Department of Physics, University College Dublin, Dublin 4, Ireland 2 Center For Laser Plasma Research, Dublin City University, Glasnevin, Dublin 9, Ireland Received 12 January 2005; published 15 July 2005 Current research on sources for extreme ultraviolet lithography EUVLhas converged on the use of discharge or laser produced plasmas containing xenon, tin, or lithium with tin showing by far the most promise. Because of their density, radiation transport from these plasmas is a major issue and accurate photoabsorption cross sections are required for the development of the plasma models needed to optimize conditions for source operation. The relative EUV photoionization cross sections of Sn I through Sn IV have been measured and from a comparison with the results of many body calculations, the cross section has been estimated to be close to 11 Mb in each species at 13.5 nm 91.8 eV, the wavelength of choice for EUVL. DOI: 10.1103/PhysRevA.72.014502 PACS numbers: 32.30.Jc, 32.80.Fb The wavelength of choice for extreme ultraviolet lithog- raphy EUVLhas been selected as 13.5 nm, based on the availability of Mo/ Si multilayer mirrors with reflectivity ap- proaching 72% within a bandwidth of approximately 0.5 nm at this wavelength. A wide variety of sources based either on pulsed discharges or laser-produced plasmas LPPsare cur- rently under investigation as potential sources. Most recent work has concentrated on three elements, Xe, Li, and Sn, as source materials. Xenon has the advantage that its inert gas- eous nature can eliminate particulate debris, though the maximum conversion efficiency CEattained to date is ap- proximately 0.8% for discharges 1and 1.2% for LPPs 2. The emission from a xenon plasma at 13.5 nm originates from 4p 6 4d 8 4p 6 4d 7 5p transitions in a single ion stage, Xe 10+ 3and calculations based on steady state collisional- radiative model CR, in which collisional ionization is bal- anced by three-body and radiative recombination 4, predict that the maximum attainable concentration of Xe 10+ is only of the order of 40% 5. More recently a CE value of 2% has been measured using the Lyman line of Li III 6. However emission from plasmas containing tin is potentially more in- tense than either 7and CE values of up to 3% have already been reported 8with a number of possible schemes for debris elimination proposed. These include the use of laser evaporated cavity confined tin vapor 9,10, mass limited tin doped droplets 11, liquid tin jets 12, and tin doped ceram- ics and glasses whose hardness limits the amount of particu- late debris 13. In the EUV spectrum of Sn, both atomic structure calculations and experiment demonstrate that the strongest lines result from 4p 6 4d N 4p 5 4d N+1 +4p 6 4d N-1 4 f 1 N 6transitions that form a line group near 13.5 nm 14. Configuration interaction CIbetween the 4p 5 4d N+1 and 4p 6 4d N-1 4 f configurations is known to be important and results in a spectral narrowing of the transition array 7,14. More importantly, it is known that such transitions in adja- cent ion stages, i.e., differing only in 4d subshell occupancy, tend to overlap in energy so that in the most extreme cases resonant emission from 10 ion stages can overlap in energy to yield an unresolved transition array UTA15. Such UTAs based on 4p 6 4d N 4p 5 4d N+1 +4p 6 4d N-1 4 f are the strongest features in EUV spectra 16and derive their inten- sity from the transfer of oscillator strength from 4d f photoionization resonances in the neutrals to 4d 4 f transi- tions in the particular ions 17. The emission within the required 2% bandwidth originates from Sn VIII through Sn XIII so in a 35–40 eV plasma, all ions present can con- tribute to the in-band emission resulting in theoretical CE values close to 5–6% 8,18. The reason why this maximum CE has not been attained by experiment to date is that it is achieved in the plasma core and radiation trapping by low ion stages considerably reduces the observed intensity. In order to determine the maximum intensity that can be ex- tracted it is necessary to have reliable information on the photoabsorption cross sections for the low ion stages of tin found in the plasma periphery. This letter directly addresses this problem and data are reported for the cross sections for neutral through four times ionized tin within the bandwidth of interest. The dual laser plasma DLPtechnique in which the ions produced by one laser pulse are backlit by EUV continuum radiation from a second laser pulse was used to obtain the experimental data 19. The spectra of Sn II through Sn IV were recorded photoelectrically with a 1024-element silicon photodiode array which was fiber-optically coupled to an image-intensified microchannel plate assembly, mounted on a 2.2 m grazing incidence vacuum spectrograph using a 1200 grooves/ mm grating to give a photon energy resolution of E / E of 1000. A 690 mJ, 15 ns Nd:YAG laser pulse was focused onto a pure tin target using a cylindrical lens to give a line plasma 12 mm in length while an 810 mJ, 15 ns Nd:YAG pulse tightly focused onto a tungsten target gener- ated the backlighting continuum. The spectrum of Sn I was recorded under similar conditions using a Schwob–Fraenkel 2.2 m grazing incidence spectrometer at a spectral resolution,  0.42 Å. By altering the interlaser time delay between the formation of the absorbing plasma column and the con- tinuum emitting plasma, the laser irradiance on the absorbing target, or probing different plasma regions, it was possible to obtain absorption spectra dominated by a single ion stage. At an interlaser time delay of 500 ns and greater the absorption was mainly due to neutral tin, though it proved extremely difficult to obtain an appreciable population of this stage. At a delay of 300 ns the plasma was dominated by Sn II and at PHYSICAL REVIEW A 72, 014502 2005 1050-2947/2005/721/0145024/$23.00 ©2005 The American Physical Society 014502-1