Fundamental investigation on CO 2 laser-produced Sn plasma for an EUVL source Yezheng Tao*, Mark Tillack, Kevin Sequoia, Russel Burdt, Sam Yuspeh, and Farrokh Najmabadi University of California, San Diego 2008 International Workshop on EUV Lithography June 10-12 th, 2008. Maui, Hawaii, USA. 1
Contents 1. Backgrounds 2. CO 2 MOPA laser and diagnostics for fundamental EUVL source research 3. Results from CO 2 laser-produced Sn plasma Plasma density profile Effect of wavelength on mass of ablated material Effect of contamination of the target Effect of pulse duration 4. Discussions 2
1. Backgrounds - Is fundamental research still necessary for EUVL source? CO 2 laser-produced Sn plasma is one of the most promising candidates for HVM EUVL source due to its high efficiency, relatively low cost, and the ability to scale to high power. Issues remaining in CO 2 LPP EUVL source 1. Laser cost of ownership 2. CE- higher value is still possible, droplet 3. Debris and out-of-band radiation 4. Mass-limited Sn target operation Fundamental research is necessary 3
2. A CO 2 MOPA laser for EUV source fundamental research was developed at UCSD Oscillator. LPX-210 i Pre-Amp Questek- Nd:YAG laser Final AMP Lumonics LaserMark 3 TEA CO 2 lasers LPX-Osc, Questek-Pre-Amp, Lumonics-Final-Amp 1 Nd:YAG laser, trigger air-breakdown Synchronization DG 535-II from DG 535 I 4
Pulse durations from 15 to 110 ns are obtained by an external triggered air breakdown plasma Characters of the MOPA system Temporal shape of CO 2 laser pulse obtained using the plasma shutter at various delay times Beam size: φ12 mm Laser pulse durations: 15 to 110 ns Pulse energy: 100 mj (15 ns), 150 mj (25ns), 400 mj (60 ns), 550 mj (110ns) Focal spot: ~ 200 μm (F/10), 100 μm (F/5) Intensity: 1.5-2 10 10 W/cm 2, 8 10 10 W/cm 2 5
Comprehensive diagnostics for EUV, debris, and plasma CO 2 laser DG 535 I To DG 535 II ns Nd:YAG laser Laser pulse energy and temporal shape are monitored for each shot. E-Mon, FC, TGS, Interferometer, visible spectrometer and newly developed diagnostics are employed. 6
Temporally resolved in-band 13.5 nm EUV light detector was setup Diagnostics for temporal shape of in-band 13.5 nm EUV EUV mirror EUV PD laser EUV mirror: NTT, 13.5 nm, multi-layers Mo/Si EUV PD: IRD AXUV HS5 Transmission of Zr and Zr plus Rising time : 700 ps Mo/Si mirror Bandwidth: Broadband, 7-17 nm In-band 4% at 13.5 nm 7
Narrow-band EUV detector provides more accurate information about the dynamics of inband EUV emission Broadband light has a slower rising and falling slopes, and has a wider FWHM. This different comes from out-ofband emission. In-band provides more accurate information 8
In-band 13.5 nm EUV imaging laser Zr filter Mo/Si mirror CCD plasma EUV mirror: NTT, 13.5 nm, multi-layers Mo/Si CCD camera: Bandwidth: 4% at 13.5 nm Spatial resolution: 5 μm 9
A quarter- static electric energy analyzer was developed 10
3. The MOPA CO 2 laser can produce enough intensity to generate efficient 13.5 nm EUV light In-band CE vs. focus lens position Soft x-ray spectra under various intensities DOF ~ 1.5 mm CE dip located at the best focus and the spectral peak located at 13.5 nm reveals that the laser intensity of the MOPA CO 2 laser reaches the optimum intensity for EUV experiment. 11
Accurate time-resolved interferogram is obtained with a fine time synchronization of the MOPA laser 0 ns 30 ns 50 ns Data is under analyzing 200 μm 12
laser CO 2 laser ablates much less material as compared with that of Nd:YAG laser CO 2 laser Sn Plasma 200 μm YAG laser Sn Plasma For both cases, the probe beam is 532 nm green light. For Nd:YAG laser, the broad black region represents the region with a density above the n c (4 10 21 cm -3 ) for green light. For CO 2 laser, the opaque region is very small. The fringe shift gives out the density of the region around marked by the red line, ~10 19 cm -3. At least, 100 times less material is ablated by CO 2 laser as compared with Nd:YAG laser while CO 2 has a higher CE. Efforts to understand the fundamentals are necessary. 13
CO 2 laser is very sensitive to the contaminations E = 2 kev Single peak for 2 nd and after shots. Double peaks from fresh target surface The fast peak may come from the contaminations (C,O from pump) on the surface. This double peaks was never observed for Nd:YAG laser. For lower laser intensity, most of the laser energy goes into contaminations. Energetic Sn ions are produced by CO 2 laser. 14
In-band CE only weakly depends on pulse duration In-band CE, 2.8 % (2 π), is constantly obtained with CO 2 laser pulse with pulse durations from 25 to 110 ns. Long pulse could significantly simplify and reduce the cost of the CO 2 laser used in EUVL source. Larger pulse energy accompanying with long pulse makes it easier to realize mass-limited target for droplet target Long pulse may make it easier for alignment. Further effort is necessary to clarify the plasma physics dominating this effect. 15
The temporal shape of the in-band 13.5 nm EUV light follows that of laser pulse In-band 13.5 nm EUV light temporally follows laser pulse. For short pulse, EUV light is a little wider than laser, comes. For short pulse, EUV lasts for several ns even after the laser turns off For long pulse, EUV is shorter than laser. Even the short tail can contribute to a little EUV emission generation. So it is reasonable to get high CE even with a long pulse. 16
Similar soft x-ray spectra are observed for various pulse durations Spectra with 25 and 50 ns laser pulses are the same. Spectra from 100 ns pulse is similar with those of short pulse, except for a shift of spectral peak. The spectral shift of the peak comes from the temporal average. The low intensity tail of the long pulse contributes a lower T e, resulting a spectral shift towards long wavelength. This confirms the constant CE observed over a wide range of pulse durations. 17
Long pulse and the tails produces extra slow ions Long pulse produces extra slow ions, arising from the 3 times extra energy. More input energy, more EUV light, more slow ions. GOOD! Short tail produces extra slow ions. OK! Long tail produces a lot of extra slow ions. BAD! 18
l abs 4. Discussions The absorption length of laser light in plasma at various plasma densities for Nd:YAG and CO 2 lasers The absorption length for 1/e intensity attenuation of laser light in plasma is obtained from D.Attwood, Soft x-ray and EUV radiation. = n n c e v v g ei 1 n e / n Zn c 2 e ( kt For long pulse Nd:YAG laser, distributed laser absorption is significant. For CO 2 laser, most of the laser energy is always locally absorbed around the n c. e 3/ 2 ) Absorption length of laser light in plasma (kt e =30 ev, Z=10) Density Nd:YAG CO 2 (μm) (μm) n c /20 100 10000 n c /10 25 2500 n c /2 1 100 Several previous experiments have shown that CO2 laser-produced Sn plasma has a narrower spectrum than that of Nd:YAG laser, confirming the above estimation.
Summary 1. Capabilities to carry out fundamental researches for CO 2 laser-produced EUVL source have been develop at UCSD. 2. It was found that in-band CE weakly depends on pulse duration. CE 2.8 %, is obtained over pulse durations from 25 ns to 110 ns. 3. It was noted that CO 2 laser is sensitive to the contamination of the Sn target. 4. Accurate interferometry was obtained for CO 2 Sn plasma, data is under analyzing. 5. Interferogram shows that CO 2 laser ablates much less materials as compared with Nd:YAG laser while CO 2 has a higher CE. 6. Temporal shape of in-band 13.5 nm EUV emission was observed. It was shown that the temporal shape of the 13.5 nm EUV light follows that of laser pulse. 7. Long pulse only produces additional slow ions as compared with that of short pulse. Even a short tail may be acceptable. 20
Acknowledgements This work was supported by Cymer Inc. and by the University of California (UC) under the UC Industry-University Cooperative Research Program (ele06-10278). 21