Shorter-wavelength extreme-UV sources below 10nm
In recent years, laser-produced dense plasmas have been attracting attention as high-efficiency, high-power sources of extreme UV (EUV) radiation. Sources with a wavelength less than 10nm are of particular interest for use in next-generation semiconductor lithography and for other applications, such as materials science and biological imaging. Manufacturer Cymer, for example, has already shipped a high-average-power 13.5nm engineering prototype to a semiconductor device company that would enable high-volume production at a power level of 80W.1 This source optimizes unresolved transition array (UTA) emission of highly ionized tin for high conversion efficiency (CE) of the input laser energy to the in-band (i.e., a bandwidth of about 2% around 13.5nm) EUV energy. Full recovery of the injected fuel is realized through ion deflection in a magnetic field. Low-density targets like tin further enable suppression of satellite (i.e., peripheral) emission. Full ionization, which helps to control debris and thus avoid damage to the source mirror, is attained with short-pulse CO2 laser irradiation.
Recently, the possibility of switching to an even shorter EUV wavelength of 6.Xnm was suggested.2 In fact, 6.Xnm beyond-EUV (BEUV) emission can be coupled with a molybdenum/boron carbide (Mo/B4C) or lanthanum/boron carbide (La/B4C) multilayer mirror whose reflectivity is currently 40% at 6.5–6.7nm (theoretical maximum >70%). The UTA emission exploited in tin is scalable to shorter wavelengths. The rare-earth element gadolinium (Gd) has a CE similar to that of tin, though at a higher plasma temperature, within a narrow spectral range centered near 6.7nm. However, no fundamental research has been reported on spectral behavior at 6.7nm and its dependence on various parameters, such as laser wavelength, initial target density, and dual-laser-pulse delay. EUV emission at this level could be tuned for use with a Mo/B4C multilayer mirror to power practical sources.
In a proof-of-principle experiment, we produced a source with peak emission around 6.5–6.7nm.3, 4 gadolinium and terbium (another rare-earth element) produce strong narrow-band emission, again attributable to thousands of resonance lines that merge to yield an n=4−n=4 UTA (where n is the principal quantum number), near 6.7nm. The spectral behavior of gadolinium and terbium plasmas is similar to that of tin because 4d open-shell ions are involved in each case. Figure 1(a) shows the ion population of a gadolinium plasma as a function of electron temperature, calculated in the steady-state collisional-radiative (CR) regime at an electron density of 1×1021cm−3. In such a plasma, the n=4−n=4 transitions in Gd12+ and Gd25+ ions form UTAs at electron temperatures of 50 and 120eV, respectively: see Figure 1(b) and (c).
We also observed variation of spectral behavior in gadolinium plasmas in the 6.7nm region when different laser wavelengths were used to change the critical electron densities. The corresponding EUV CEs were 1.1, 0.7, and 0.5% for wavelengths of 1064, 532, and 355nm. The intensity ratio of the resonant lines around 6.7nm to the satellite emission at wavelengths longer than 7nm decreased for the 532 and 355nm laser pulses compared with that at 1064nm. Even allowing for the presence of an underlying recombination continuum, satellite emission at wavelengths longer than 7nm increases with decreasing wavelength (see Figure 2). We attribute the decrease in 6.7nm emission to self-absorption in the denser, short-wavelength plasma.5 Because opacity effects on the resonance lines in gadolinium plasmas are large, it is important to generate a low-density plasma using a long-wavelength laser and/or a low-initial-target concentration of gadolinium. As with tin, the spectrum obtained with the low-initial-density target was narrower and more intense than that of the pure solid target. Consequently, the maximum CE proved to be about 1.8% in dual Nd:YAG (neodymium-doped yttrium aluminum garnet) laser-produced, low-density plasmas with a 30% initial target density of gadolinium (see Figure 3).6
Because its wavelength becomes shorter with increasing atomic number, the n=4−n=4 UTA can be used for other applications, such as transmission x-ray microscopy for imaging of, say, biological tissue (see Figure 4). We have conducted preliminary studies of the potential of bismuth as a BEUV source. Our calculations show that, at an electron temperature in the range 570–600eV, bismuth plasmas radiate strongly near 3.9nm. We have begun a number of experiments to explore how to optimize this emission in practice.
In summary, we have observed the spectral behavior of EUV CE around 6.7nm for shorter-wavelength EUV emission and measured it. The highest CE in this spectral region was 1.8%. Increasing the CE and spectral purity requires producing low-density plasmas, such as those obtained using CO2 lasers. For the near future, we plan to work on improving the efficiency of the BEUV source using high-temperature, low-density plasmas.
The author acknowledges many collaborators, in particular C. O'Gorman, T. Cummins, and D. Kilbane. Part of this work was performed under the auspices of the Ministry of Education, Culture, Science and Technology, Japan, and Utsunomiya University Distinguished Research Projects. The University College Dublin group acknowledges support from Science Foundation Ireland under Principal Investigator Programme research grant 07/IN.1/B1771.
Takeshi Higashiguchi is an associate professor. He received his PhD in engineering from Utsunomiya University. His research activities are focused on short-wavelength light sources, laser-plasma interactions, and plasma photonic devices.
Takamitsu Otsuka is a PhD candidate. He received his MS from Utsunomiya University. His research activity concentrates on developing shorter-wavelength EUV sources.
Noboru Yugami is a professor. He received his PhD in engineering from Osaka University. His research activities focus on plasma physics, laser-plasma interactions, and plasma photonic devices.
Weihua Jiang is a professor of electrical engineering. He received his PhD in energy and environmental science from Nagaoka University of Technology in 1991. His research activities include plasma physics, pulsed-power technology, and high-power microwaves.
Akira Endo has been active in the field of short-pulse, short-wavelength light sources, ranging from laser-Compton to UTA laser plasmas. He has managed research projects in these fields over the last two decades in Japan.
Padraig Dunne is an associate professor in the UCD School of Physics. He received his PhD from the National University of Ireland in experimental atomic spectroscopy. His research interests include laser-produced plasmas (LPPs) as sources of EUV radiation for lithography and other applications, LPP target development and analysis, EUV instrumentation, and LPP spectroscopy.
Bowen Li is a PhD candidate. He received his MS from North West Normal University, Lanzhou. His research focuses on shorter-wavelength EUV source development.
Gerry O'Sullivan is a professor. He received his PhD from UCD. His research activities have focused on EUV and soft x-ray spectroscopy of laser-produced plasmas, ionic photoabsorption, and EUV source development.
