X-Ray Lasers 2006: Proceedings of the 10th International Conference, August 20-25, Berlin, Germany (Springer Proceedings in Physics)

Citation preview

springer proceedings in physics 115

springer proceedings in physics 95 Computer Simulation Studies in Condensed-Matter Physics XVI Editors: D.P. Landau, S.P. Lewis, and H.-B. Sch¨uttler

105 Computer Simulation Studies in Condensed-Matter Physics XVIII Editors: D.P. Landau, S.P. Lewis, and H.-B. Sch¨uttler

96 Electromagnetics in a Complex World Editors: I.M. Pinto, V. Galdi, and L.B. Felsen

106 Modern Trends in Geomechanics Editors: W. Wu and H.S. Yu

97 Fields, Networks, Computational Methods and Systems in Modern Electrodynamics A Tribute to Leopold B. Felsen Editors: P. Russer and M. Mongiardo

107 Microscopy of Semiconducting Materials Proceedings of the 14th Conference, April 11–14, 2005, Oxford, UK Editors: A.G. Cullis and J.L. Hutchison

98 Particle Physics and the Universe Proceedings of the 9th Adriatic Meeting, Sept. 2003, Dubrovnik Editors: J. Trampeti´c and J. Wess

108 Hadron Collider Physics 2005 Proceedings of the 1st Hadron Collider Physics Symposium, Les Diablerets, Switzerland, July 4–9, 2005 Editors: M. Campanelli, A. Clark, and X. Wu

99 Cosmic Explosions On the 10th Anniversary of SN1993J (IAU Colloquium 192) Editors: J. M. Marcaide and K.W. Weiler

109 Progress in Turbulence 2 Proceedings of the iTi Conference in Turbulence 2005 Editors: M. Oberlack et al.

100 Lasers in the Conservation of Artworks LACONA V Proceedings, Osnabr¨uck, Germany, Sept. 15–18, 2003 Editors: K. Dickmann, C. Fotakis, and J.F. Asmus

110 Nonequilibrium Carrier Dynamics in Semiconductors Proceedings of the 14th International Conference, July 25-29, 2005, Chicago, USA Editors: M. Saraniti, U. Ravaioli

101 Progress in Turbulence Editors: J. Peinke, A. Kittel, S. Barth, and M. Oberlack

111

102 Adaptive Optics for Industry and Medicine Proceedings of the 4th International Workshop Editor: U. Wittrock 103 Computer Simulation Studies in Condensed-Matter Physics XVII Editors: D.P. Landau, S.P. Lewis, and H.-B. Sch¨uttler 104 Complex Computing-Networks Brain-like and Wave-oriented Electrodynamic Algorithms Editors: I.C. G¨oknar and L. Sevgi

Vibration Problems ICOVP 2005 Editors: E. Inan, A . Kiris

112 Experimental Unsaturated Soil Mechanics Editor: T. Schanz 113

Theoretical and Numerical Unsaturated Soil Mechanics Editor: T. Schanz

114 Advances in Medical Engineering Editor: T horste n M . B uzug 115 X-Ray Lasers 2006 Proceedings of the 10th International Conference, August 20-25, 2006, Berlin, Germany Editors: P.V. Nickles, K.A. Janulewicz

Volumes 69–9 4 are listed at the end of the book.

P.V. Nickles (Eds.)

K.A. Janulewicz

X-Ray Lasers 2006 Proceedings of the 10th International Conference, August 20–25 , 2006, B erlin , 2006, Germany

With 41 5 Figures

123

Dr. P.V. Nickles Max Born Institute for Nonlinear Optics and Short Pulse Spectroscopy Max Bornstrasse 2A D-12489 Berlin Germany Dr. K.A. Janulewicz Max Born Institute for Nonlinear Optics and Short Pulse Spectroscopy Max Bornstrasse 2A D-12489 Berlin Germany

_ A C.I.P. Catalogue record for this book is available from the Library of Congress. ISSN 0930-8989 ISBN-13 978-1-4020-6017-5 (HB) ISBN-13 978-1-4020-6018-2 (e-book)

_ Published by Springer, P.O. Box 17, 3300 AA Dordrecht, The Netherlands In association with Canopus Publishing Limited, 27 Queen Square, Bristol BS1 4ND, UK www.springer.com

All Rights Reserved © 2007 Springer No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work.

Overview of Tabletop X-Ray Laser Development at the Lawrence Livermore National Laboratory J. Dunn1, V. N. Shlyaptsev2, J. Nilsen1, R. F. Smith1, R. Keenan1, S. J. Moon1, J. Filevich3, J. Rocca3, A. J. Nelson1, J. R. Hunter1, M. C. Marconi3, Y. L. Li1*, A. L. Osterheld1, R. Shepherd1, H. Fiedorowicz4, A. Bartnik4, A. Ya. Faenov5, T.A. Pikuz5, P. Zeitoun6, S. Hubert7, S. Jacquemot7 and M. Fajardo8 1

Lawrence Livermore National Laboratory, Livermore, CA 94551, USA. University of California Davis-Livermore, Livermore, CA 94551, USA. 3 NSF ERC for Extreme Ultraviolet Science and Technology and Dept. of Elect. and Comp. Engineering, Colorado State Univ., Fort Collins, USA. 4 Inst. of Optoelectronics, Military Univ. of Technology, Warsaw, Poland. 5 Multicharged Ions Spectra Data Center of VNIIFTRI, Moscow, Russia. 6 Laboratoire d’Optique Appliquée, ENSTA, 91761 Palaiseau, France 7 Commissariat á l’Énergie Atomique, 91680 Bruyères-le-Chatel, France. 8 Centro de Fisica dos Plasmas, Inst. Superior Técnico, Lisbon, Portugal *Present address: Argonne National Laboratory, Argonne, IL 60439 USA 2

Summary. It is almost a decade since the first tabletop x-ray laser experiments were implemented at the Lawrence Livermore National Laboratory (LLNL). The decision to pursue the picosecond-driven schemes at LLNL was largely based around the early demonstration of the tabletop Ne-like Ti x-ray laser at the Max Born Institute (MBI) as well as the established robustness of collisional excitation schemes. These picosecond x-ray lasers have been a strong growth area for x-ray laser research. Rapid progress in source development and characterization has achieved ultrahigh peak brightness rivaling the previous activities on the larger facilities. Various picosecond soft-x-ray based applications have benefited from the increased repetition rates. We will describe the activities at LLNL in this area.

1 Introduction By the mid-1990s there were a number of remarkable advances in soft xray laser experimental and theoretical research that established the viability of tabletop pumped devices. The first demonstration of a tabletop x-ray laser was reported in 1994 using the collisional excitation scheme on a fast capillary discharge apparatus [1]. The Ne-like Ar 3p – 3s J = 0 – 1 line at

2

J. Dunn et al.

46.9 nm was shown to lase and within two years was operating in saturation [2]. On the laser-driven schemes, there were a variety of approaches described. A 10 Hz Pd-like Xe ion x-ray laser scheme with gL ~ 11 at 41.8 nm for 40 fs irradiation of a Xe gas cell using field-induced tunneling ionization followed by collisional excitation [3]. Observation of a Ne-like titanium 3p - 3s J = 0 - 1 transition at 32.6 nm transient collisional excitation scheme produced high gain g = 19 cm-1 and a gainlength product of 9.5 [4]. A laser-driven recombination scheme based on the H-like Li n = 2 – 1 transition was also demonstrated at 13.5 nm [5]. The rapid technology advances of compact, high peak power, short pulse lasers accelerated this work and subsequent laser-pumped soft x-ray amplifiers. This was the landscape of the tabletop x-ray lasers during 1996 when the decision at LLNL was made to start a small-scale experimental x-ray laser effort. This new approach for generating x-ray lasers with a higher repetition rate and smaller drive energy was a different direction compared to the renowned large-scale experiments using the two-beam Novette/Nova-based x-ray laser. The latter had been active for more than a decade but were now winding down as the plans for the National Ignition Facility were implemented. In this paper, we describe some of the main achievements in the tabletop x-ray laser research at LLNL over the last 10 years, as summarized in Fig. 1.

2 First Experimental Steps The transient collisional excitation scheme, as reported earlier by Shlyaptsev [6] and demonstrated by the MBI group [4], was the most appealing candidate for implementing at LLNL. It was first and foremost a collisional excitation scheme that had been well-studied by many laboratories in the Ne-like and Ni-like closed shell ion configuration. The use of the long nanosecond pulse to generate the initial plasma column and establish the ionization conditions had the advantages of creating the x-ray amplifier pre-conditions using a small amount of laser energy. After a given delay, when the plasma density profile expands and relaxes to aid the propagation of the x-ray laser and secondly the absorption of the picosecond laser pulse, the main short pulse was fired. The short pulse rapidly heated the plasma raising the electron temperature and strongly pumping the upper level of the inversion by monopole electron excitation. The use of the two separate laser pulses to optimize the conditions for the x-ray laser was the main reason that the drive energy could be reduced.

Overview of Tabletop X-Ray Laser Development

3

These were similar advantages to the pre-pulse technique that had been reported previously by Nilsen and colleagues [7].

Fig. 1. Time-line showing major activities and achievements on the tabletop x-ray laser work at LLNL in the years 1997 through 2006.

The tabletop experimental activities first started at LLNL in January 1997 when the short pulse, tabletop Janus Picosecond Laser was available. There were several stages involved before the x-ray laser experiments could begin. The laser at the time was a single high power short pulse beam producing ~3 J at 1054 nm wavelength in a 500 fs pulse with a repetition rate of 1 shot/4 minutes. The beam diameter, along with the optimization of other parameters, was increased to improve the available compressed energy to 7.5 J giving the laser a 15 TW peak power specification. Initial experiments were conducted with a ~1.5 ps short pulse. The long pulse came from the Janus laser: Rod shots of 5 J energy at

4

J. Dunn et al.

1064 nm in a 800 ps (FWHM) pulse could be fired at 1 shot/3 minutes. The two lasers were synchronized to within 80 ps rms jitter relative timing. The beams had orthogonal polarization and were combined into one beam path with a polarizer, the long pulse in transmission and the short pulse in reflection [8, 9]. The single beam line to the target chamber was focused with a 4-m focal length plano-concave cylindrical lens and a 0.61-m focal length on-axis parabola. The laser line focus was 1.2 cm × 80 µm (L × W) and the two laser beams were imaged and carefully overlapped at the target plane.

Fig. 2. First Ne-like ion x-ray laser spectrum observed for a 6 mm Ti target on 3 June 1997 using the Janus and Janus-ps lasers at LLNL [9].

The main diagnostic for the x-ray laser lines looking on-axis was a 1200 line mm-1 flat-field grating spectrometer with a back-thinned 1024 × 1024 charge-coupled device (CCD) detector to cover 13 - 35 nm. A gold-coated mirror collection optic imaged the plasma gain region with 1:1 magnification onto a 100 µm wide entrance slit. The instrument spatially resolved the position of the gain region. Additional instruments included a CCD x-ray slit camera with 25 µm spatial resolution for line focus uniformity and a CCD flat crystal KAP (001), 2d = 26.58 Å, spectrometer. All x-ray instruments used CCD detector systems that were LLNL designed, fabricated and optically calibrated. The first targets used to study the Ne-like ion scheme were polished low- to mid-Z metals including Ti, Fe, and Ge. There was additional theoretical support for lasing on Ti under a range of different picosecond pulse durations [10]. The Ge target did not demonstrate lasing with the laser energies available. Initially Ti did not lase when the two laser pulses

Overview of Tabletop X-Ray Laser Development

5

were separated by a delay of 800 ps peak-to-peak. Unexpectedly, some jitter in the laser synchronization delayed the short pulse to 1 ns which was sufficient to generate better plasma conditions for lasing [9]. Those first results are shown in Fig. 2 and further optimization of the x-ray signal was achieved with delays of 1.6 ns. The strongest lasing was on the 3p – 3s line at 32.6 nm with the 3d – 3p line also visible at ~5% intensity. Small signal gains of 24 cm-1 and gain length gL products approaching 15 for 1 cm targets were achieved [8]. Strong lasing was also observed on the Ne-like Fe 3p – 3s line at 25.5 nm. The main leap forward though was the demonstration of shorter wavelength lasing utilizing the Ni-like ion 4d – 4p scheme, specifically for Pd at 14.7 nm. The Ni-like schemes, with atomic number in the mid-forties, had not lased well under various pulse configurations on the larger lasers. There was a good match with the pslaser pumping conditions here and gains of 35 cm-1 and gL products of 12.5 were reported [11]. One major difference between the Ne-like and Nilike lasing was the lower output on the Ni-like schemes. This would be attributed to the shorter gain lifetime conditions for the Ni-like ion and would be improved by using traveling wave excitation.

3 Compact Multipulse Terawatt Laser Early in 1998, the tabletop laser was redesigned to accommodate an additional 50 mm rod amplifier to generate a long stretched pulse beam with 600 ps (FWHM) at the full aperture of 8.4 cm diameter. This allowed the tabletop laser to generate two full energy ~7 J beams for the x-ray laser on two standard laser tables. The laser was renamed the Compact Multipulse Terawatt (COMET) laser and was a stand-alone facility optimized for x-ray laser studies [12]. The vacuum compressor was upgraded with new 13” diameter, 1700 line mm-1, gold-coated gratings. A series of experiments were conducted over the next few years to both characterize and improve the x-ray laser parameters required for applications. The careful measurement of various Ni-like 4d – 4p transition wavelengths was carried out in lasing plasmas and compared with optimized level multiconfiguration Dirac-Fock code as well as high resolution spectra from non-lasing vacuum spark or laser-produced plasmas [13]. This was largely to gain better understanding of the Ni-like ion energy level structure (that is complex) as well as the precise transition wavelength knowledge required to develop and optimize the reflectivity of x-ray optics. The Ni-like x-ray lasers also seemed to be a good match with

6

J. Dunn et al.

the small ps-lasers and lower Z materials were observed to lase for the first time [14]. The next step was improving the output of a number of the Ne-like and Ni-like x-ray lasers. The Ne-like ion x-ray lasers studied to date were very close to saturation. A traveling wave excitation scheme was implemented before the final focusing optics utilizing a segmented reflection echelon that split the beam into delayed vertical slices [15]. This produced a segmented but contiguous line focus for both beams, initially with 5 steps then increased to 7 when the focus was lengthened to 1.6 cm. The traveling wave focus consisted of 0.22 cm long steps delayed by 7.7 ps relative to its neighbor. In addition the long pulse was defocused to 150 µm (FWHM) while retaining an 80 µm (FWHM) short pulse. This improved the x-ray laser propagation by reducing the negative effect of lateral density gradients. The combination of these two methods was very effective and increased the Pd x-ray laser output up to 100 times and into saturation for targets approaching 1 cm in length. Estimated output energy was now above 10 µJ and in saturation. Further optimization of the laser energy parameters and relative pulse delay gave narrow beam divergence ~2.5 mrad (FWHM) and improved x-ray brightness for the 14.7 nm laser. Nilike Mo laser at 18.9 nm was driven into saturation [15] and multilayercoated imaging optics allowed direct measurements of the near-field and far-field patterns of both the Mo and Pd lasers [16, 17]. The speckle pattern and various structures observed in the far-field x-ray beam were recently attributed to a combination of low spatial coherence, high longitudinal coherence and short pulse duration of these transient x-ray sources [18]. One of the projects in 2001 was to study gas puff targets heated by transient pumping. The argon laser successfully lased on both 3p – 3s and 3d – 3p lines at 46.9 nm and 45.1 nm, respectively [19]. However, further work in studying the ionization and absorption conditions during both laser pulses would be useful to optimize the amplification process.

4 Applications, New source Development and Characterization The Pd laser and the experimental setup described above became the workhorse for many of the studies up until the grazing incidence pumping (GRIP) method was adopted for the high repetition rate x-ray source development. In many cases the source characterization was in parallel with the application development. The plan for developing the x-ray laser interferometry of laser-produced plasmas based on the LLNL ps duration

Overview of Tabletop X-Ray Laser Development

7

14.7 nm line was discussed with Jorge Rocca’s group at Colorado State University and a second French team led by Philippe Zeitoun in 1999. In the former case, a skewed Mach-Zehnder interferometer using diffraction grating optics as beam splitters was developed from the design used at 46.9 nm [20]. The instrument was robust, had a high signal throughput and achieved uniform fringes with high visibility. This Diffraction Grating Interferometer (DGI) was an excellent match for the tabletop transient xray lasers and led to new picosecond resolution and micron spatial resolution data from laser-produced plasma phenomena [21 – 23]. The use of the interferometry on Pd plasmas when combined with the near-field imaging technique of the gain region yielded quantitative measurements of the density profile of the x-ray laser amplifier medium [24]. Fig. 3(a) and (b) show the effect of different short pulse durations on x-ray amplification for the same pre-pulse conditions used to generate the 14.7 nm Pd laser [15]. This study showed the beneficial effect on the x-ray laser output of using longer 6 – 13 ps short pulse drives together with optimized laser parameters [24]. The experiment used various, sophisticated x-ray laser diagnostics to understand and achieve better amplification conditions.

Fig. 3. The nearfield image of the Pd 14.7 nm x-ray laser, with both laser pulses fired, overlaid on a 2-D density map measured by 14.7 nm interferometry of the 600 ps heated Pd plasma at Δt = 700 ps. The short pulse would fire at this time. (a) Conditions for a 0.5 ps short pulse. (b) For a 13.4 ps short pulse where the x-ray laser output is substantially higher, more compact and less affected by refraction.

8

J. Dunn et al.

The second interferometry effort was based on a Michelson design using thin foil multilayer beamsplitters: A measurement of the longitudinal coherence of the Pd line operating in saturation was made for two different short pulse durations of ~6 ps, shown in Fig. 4(a), and ~13 ps [25]. This indicated that the Pd line was spectrally narrow approaching λ/Δλ of 5 × 104. This observation is consistent with the high gain determined from various transient pumping experiments over the last decade. Another experiment was performed to measure the x-ray laser pulse duration of the Pd line under various laser parameters [26]. Figure 4 (b) shows the duration of the 14.7 nm under similar pumping conditions of Fig. 4(a). The measured duration is typically 4 – 5 ps when heated by a 6 ps pulse. Shorter pulse durations are achieved below 3 ps when the optical laser pulse duration is reduced or the x-ray laser is operated out of saturation. This is in close agreement with results reported for the Ag x-ray laser [27]. The spectral and temporal measurements allow us to note that the psdriven Pd laser is operating in the few times transform-limited regime [28] in agreement with similar work as reported by other groups [29 – 31].

Fig. 4. (a) Fringe visibility of saturated Pd x-ray laser line heated by a 6 ps short pulse measured as a function of the differential arm separation of the Michelson interferometer. (b) Pulse duration of Pd 14.7 nm laser line measured with a fast 500 fs x-ray streak camera under similar heating and saturated output conditions.

We mention some recent activities. The Pd laser was shown to be a novel materials surface probe where the x-ray laser, incident on the surface, can photo-ionize the electrons in low energy valence and shallow core levels. The resultant kinetic energy of the ejected electrons is measured using time-of-flight spectroscopy to retrieve the original electron energy level structure [32]. This allows a ps snapshot of the electronic structure of the material while undergoing rapid changes [32]. A stable and high repetition soft x-ray laser source would have advantages for this type

Overview of Tabletop X-Ray Laser Development

9

of work since large photon numbers in single bursts are not necessary. High repetition rate ~10 Hz lasing has been demonstrated in our laboratory using the grazing incidence pumping geometry where the required laser pump energy can be reduced to the range of 150 mJ to ~1 J [33]. This scheme has been quickly adopted in numerous laboratories and is presented in various papers in this journal. A photoelectron microscopy tool based on high repetition, fs-heated vacuum ultraviolet radiation has been utilized for detailed surface chemistry studies [34]. This is one area where the higher photon energy and more monochromatic energy of the xray laser would make a strong application.

5 Conclusions We have given a brief overview of the last 10 years of research in tabletop x-ray lasers at LLNL. With the development of the Compact Multipulse Terawatt (COMET) laser as a dedicated x-ray laser facility a number of advances were made in the source development, including optimization of the output, optimization of key parameters and establishment of several applications. In this time several studies were conducted that broke new ground for ps-driven x-ray lasers: The gas puff x-ray laser at 46.9 nm illustrated the potential for a debris-less gain medium. The Grazing Incidence Pumping (GRIP) was shown to operate with laser energy as low as 150 mJ and is the route forward for the development of a high average power, high repetition rate 10 – 30 nm source. The collisional excitation scheme has been the continuous link through this research and will likely continue to in the near future.

Acknowledgments Work performed under the auspices of the US Department of Energy by the University of California Lawrence Livermore National Laboratory under Contract No. W-7405-Eng-48 and in part by US Department of Energy grant No. #DE-FG52-06NA26152, and in part, by the Ministry of Science and High Education of Poland under grant No 3 T11B 031 26 and by the NATO Collaborative Grant PST.CLG.979339.

10

J. Dunn et al.

References 1. 2. 3. 4. 5. 6. 7. 8. 9.

10. 11. 12.

13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28.

J. J. Rocca et al., Phys. Rev. Lett. 73(16), 2192 - 2195 (1994). J. J. Rocca et al., Phys. Rev. Lett. 77(8), 1476 - 1479 (1996). B. E. Lemoff et al., Phys. Rev. Lett. 74(9), 1574 - 1577 (1995). P. V. Nickles et al., SPIE Proceedings 2520, 373 - 378 (1995); P. V. Nickles et al., Phys. Rev. Lett. 78(14), 2748 - 2751 (1997). D. V. Korobkin et al., Phys. Rev. Lett. 77(26), 5206 - 5209 (1996). V. N. Shlyaptsev et al., SPIE Proceedings 2012, 111 - 118 (1993). J. Nilsen, B. J. MacGowan, L. B. Da Silva, and J. C. Moreno, Phys. Rev. A 48(6), 4682 - 4685 (1993). J. Dunn, A. L. Osterheld, R. Shepherd, W. E. White, V. N. Shlyaptsev, A. B. Bullock, and R. E. Stewart, SPIE Proceedings 3156, 114 - 121 (1997). J. Dunn, A. L. Osterheld, R. Shepherd, W. E. White, V. N. Shlyaptsev, and R. E. Stewart, "Table-Top Transient Collisional Excitation X-ray Laser Research at LLNL: Status June 1997", Lawrence Livermore National Laboratory, Livermore, CA UCRL-ID-127872 (1997). J. Nilsen, Phys. Rev. A 55(4), 3271 - 3274 (1997). J. Dunn, A. L. Osterheld, R. Shepherd, W. E. White, V. N. Shlyaptsev, and R.E. Stewart, Phys. Rev. Lett. 80, 2825 - 2828 (1998). COMET has 4 independent, synchronized beams: The two main beams, a third ~20 mJ (1ω or 2ω) short pulse probe beam and a fourth beam that uses energy from the long pulse arm for either 3 J in 600 ps or ~2 J compressed. Y. Li, J. Nilsen, J. Dunn, A. L. Osterheld, A. Ryabtsev, and S. Churilov, Phys. Rev. A 58, R2668 - 2671 (1998). J. Dunn et al., Opt. Lett. 24, 101 – 3 (1999). J. Dunn, Y. Li, A. L. Osterheld, J. Nilsen, J. R. Hunter, V. N. Shlyaptsev, Phys. Rev. Lett. 84, 4834 - 7 (2000). Y. L. Li, J. Dunn, J. Nilsen, T. W. Barbee, Jr., A. L. Osterheld, and V. N. Shlyaptsev, J. Opt. Soc. Am. B 17, 1098 - 1101 (2000). J. Nilsen et al., J. Opt. Soc. Am. B 20, 191 - 194 (2003). O. Guilbaud, A. Klisnick, K. Cassou, S. Kazamias, D. Ros, G. Jamelot, D. Joyeux, and D. Phalippou, Europhys. Lett. 74, 823 – 829 (2006). H. Fiedorowicz, A. Bartnick, J. Dunn, R. F. Smith, J. R. Hunter, J. Nilsen, A. L. Osterheld, and V. N. Shlyaptsev, Opt. Lett. 26, 1403 - 1405 (2001). J. Filevich, K. Kanizay, M. C. Marconi, J. L. A. Chilla, and J. J. Rocca, Opt. Lett. 25, 356 - 358 (2000). R. F. Smith et al., Phys. Rev. Lett. 89(6), 065004-1 (2002). J. Filevich et al., Appl. Opt. 43(19), 3938 - 46 (2004). J. Filevich et al., Phys. Rev. Lett. 94, 035005 (2005). R. F. Smith et al., Phys. Rev. E 72, 036404 (2005). R. F. Smith et al., Opt. Lett. 28, 2261 - 3 (2003). J. Dunn et al., SPIE Proceedings 5197, 51 - 59 (2003). A. Klisnick et al., Phys. Rev. A. 65, 033810 (2002). J. Dunn et al., SPIE Proceedings 5197, 43 - 50 (2003).

Overview of Tabletop X-Ray Laser Development 29. 30. 31. 32. 33. 34.

11

Y. Abou-Ali et al., Opt. Comm. 215, 397 - (2003). A. Klisnick et al., J. Quant. Spectrosc. Radiat. Transf. 99, 370 – 380 (2006). D. Benredjem et al., Phys. Rev. A 73, 063820 (2006). A. J. Nelson et al., Appl. Phys. Lett. 87, 154102 (2005). R. Keenan et al., Phys. Rev. Lett. 94, 103901 (2005). T. Munakata et al., Surf. Sci, 532, 1140 – 144 (2003).

This page intentionally blank

Recent Progress in X-Ray Laser Development and Applications at LIXAM A. Klisnick1, D. Ros1, G. Jamelot1, S. Kazamias1, M. Pitmann1, K. Cassou1, F. Plé1, O. Guilbaud1, I. R. Al’Miev1, C. Möller1, D. Benredjem1, J. Dubau1, D. Joyeux2, C.-G. Wahlström3, T. Kühl4, B. Rus5, H. Bercegol6, O. Larroche7, J.-P. Chambaret8, Ph. Zeitoun8 and P. Velarde9 1

LIXAM, CNRS, Univ Paris-Sud, UMR n° 8624, Orsay, F-91405 LCFIO, CNRS, Univ Paris-Sud, UMR n° 8501, Orsay, F-91405 3 Department of Physics, Lund University, Lund, S-22100 Lund 4 Gesellschaft fur Schwerionenforschung (GSI), Darmstadt, D-64291 5 Department of X-ray Lasers, Institute of Physics /PALS Centre, 18221 Prague 8, Czech Republic 6 CEA-CESTA/ DLP, BP 2, Le Barp, F-33114 7 CEA-DIF, BP 12, Bruyères-le-Chatel, F-91680 8 LOA, ENSTA, CNRS, Ecole Polytechnique, UMR n° 7639, Palaiseau, F-91761 9 Institute of Nuclear Fusion (DENIM) - Universidad Politécnica de Madrid, Spain 2

Summary. We present an overview of the recent progress achieved in the development of X-ray lasers and demonstration of applications by the LIXAM team, in collaboration with other groups. Lasing at 10 Hz was demonstrated at 18.9 nm in Ni-like Mo with a 30 µW average power, using transient pumping in the grazing incidence geometry. New numerical simulations have been developed to study spatial and temporal features of current and future X-ray lasers. The technique of interference microscopy with an X-ray laser beam was applied to the investigation of laser-induced damages in optical materials exposed to sub-ns laser pulses. Finally the construction of a 0.5 PW, 0.1 Hz Ti-Sa laser, the driver of the LASERIX facility, was successfully completed.

1 Introduction Since the last X-ray laser conference in Beijing, the research activity of the X-ray laser team at LIXAM has covered three different aspects of the development of X-ray lasers, namely fundamental studies, both experimental and theoretical; demonstration of applications; and the construction of an advanced laser driver for our future LASERIX facility.

14

A. Klisnick et al.

We have now started to explore the new potential opened by highenergy Ti:Sa lasers as a pump for transient X-ray lasers. This step is of particular importance for our future activity since the driver of LASERIX was chosen to be based of the Ti:Sa technology. We have carried out an extensive optimization and detailed characterization of the Grazing Incidence Pumping (GRIP) in a Ni-like Mo plasma, using the multiterawatt, 10 Hz facility at the Lund Laser Center, Sweden. The main results will be outlined in section 2. In section 3 we describe the development of new numerical simulations aimed at understanding and improving the spatial and temporal features of transient X-ray lasers. Section 4 deals with the development of a new application based on the technique of interference microscopy and applied to the investigation of laser-induced damages to optical surfaces. In section 5, we show that the construction of the Ti:Sa laser chain, the driver of LASERIX, has been completed and the expected performances have been successfully demonstrated. We outline the main directions for future work with this new laser.

2 Demonstration of a 10 Hz, 3µJ GRIP transient X-ray laser We have recently carried out an experimental campaign, involving collaboration with GSI and Lund University, at the Lund Laser Center (Sweden) using the multiterawatt laser facility. The general aim of this experiment was to demonstrate transient pumping at 10 Hz with a Ti:Sa driver. This is an important step towards the operation of our future LASERIX facility. Thanks to the 10 Hz repetition rate of the Lund laser system we were able to perform an extensive multiparameter scan for the optimization of a Ni-like Mo X-ray laser, in the Grazing Incidence Pumping (GRIP) geometry. This optimization work led to a substantial improvement of the output beam characteristics, in particular the average power that reached up to 30µW, and to an improved efficiency [1]. The main results of this optimization are presented in a companion paper [2]. Here we discuss a more specific goal of this experiment which was to get a detailed insight into the spatial behaviour of the X-ray laser pumped in the GRIP geometry, by implementing appropriate imaging diagnostics. The GRIP concept was initially proposed by Shlyaptsev [3] and first demonstrated by Keenan [4]. The main idea is to reduce the density at which the short pulse energy is absorbed into the preformed plasma by adjusting the grazing incidence angle φ, of the short pulse beam irradiating the preformed plasma. This allows to possibly match the position of the

Recent Progress in X-Ray Laser Development and Applications

15

plasma heated zone with the zone of maximum gain. As a result we expect that varying the angle φ will not only influence the X-ray laser beam output but also the spatial features of the source, like its size and position. As will be discussed in section 3, the knowledge of these features is essential because they largely influence the spatial profile of the beam. In order to investigate the influence of varying the GRIP angle on the spatial behaviour of the X-ray laser plasma we implemented two diagnostics, as shown in Fig . 1. The first diagnostic is a near-field imaging device giving the position and size of the XRL source at the plasma exit plane. The second diagnostic is a pinhole camera viewing the keV plasma emission from above, yielding the position and size of the zone heated by the short pulse. Both diagnostics were used together while varying the GRIP angle.

Fig. 1. Geometry of the experiment, showing the two main diagnostics used. Typical images obtained for the XRL source (left) and keV plasma emission (right) are also shown.

The results obtained are summarized in Fig. 2. Fig. 2a shows the estimated mean fluence of the XRL as a function of the GRIP angle. Each data point is averaged over ~10 shots. One can see that the data exhibits a clear optimum for a GRIP angle of 19°. The corresponding mean fluence is 0.33 J/cm2, while the integrated energy is 3µJ. Figure 2b shows the distance relative to the target surface of the XRL source, as a function of the GRIP angle. It can be seen that this distance gets smaller when the GRIP angle is increased, i.e. the XRL source is shifted towards higher densities, as expected from the theory. The data obtained from the keV

16

A. Klisnick et al.

pinhole camera also confirm this trend for the plasma heated zone, as discussed in [5]. The distance of the XRL source is minimum at the optimal angle of 19°. For larger angles the XRL is shifted to larger distances and this could be a result of increased refraction.

Fig. 2. Influence of varying the GRIP angle. (a) mean fluence of the XRL output at the exit plane; (b) distance of the XRL source from the target surface.

3 Numerical simulations of current and future X-ray lasers Three different works dealing with different aspects of the simulation of transient XRLs were performed recently. The first work arose from the observation of contrasted small-scale structures in the beam profile of TCE XRLs, routinely observed in our experiments [6] and also reported by other groups [7]. We also observed that the typical size and shape of those structures vary from shot to shot. In order to understand the origin of these structures and of their fluctuating shape, we developed a model described elsewhere [8, 9]. Figure 3 shows the beam profiles calculated by this model for three different sizes of the XRL source aperture, as shown in the inserts. These results account very well for the experimental observations. Further our model brings out a simple relationship between the angular size of the structures and the size of the XRL source: the angular size β is inversely proportional to the size of the source a. This result hence suggests that the shot-to-shot variation of the microstructures in the beam profile is due to shot-to-shot variation of the source size. Our model also predicts that the visibility of the microstructures decreases as the square root of the ratio between the duration and the coherence time of the XRL pulse [9]. The second simulation work, discussed in more detail in [10], was carried out in collaboration with LOA (France) and DENIM (Spain). A 2D-eulerian hydrocode ARWEN, developed by the Spanish group, was used to investigate the possibility of spatially shaping the preformed

Recent Progress in X-Ray Laser Development and Applications

17

Fig. 3. Beam profiles calculated from the model described in [9], using XRL source (shown in the insert of each image) with different horizontal width: (a) 10µm; (b) 20 µm; (c) 40 µm.

Fig. 4. 2D- electron density profiles calculated from the hydrocode ARWEN [11] at two successive stages of expansion, for two different transverse profiles of the pump laser beam: (left) Gaussian; (right) super-Gaussian.

plasma of transient X-ray lasers, by properly shaping the transverse profile of the pump laser beam. Figure 4 illustrates the possibility of substantially reducing the vertical refraction of the XRL beam and enlarging the vertical width of the gain zone, by using a super-Gaussian pump beam profile, instead of a Gaussian one. In the super-Gaussian case, the density profile is flatter at the two times of plasma expansion shown in Fig. 4. Such a profile will improve the conditions of propagation of the XRL beam along the amplifying plasma [11].

18

A. Klisnick et al.

Finally the third simulation work deals with the seeding of a transient Xray laser plasma with high-order harmonic radiation. This topic has recently received an increased interest, following recent successful experimental demonstrations [12, 13] and due to the potential major impact of this technique. To account for the complex problem of amplification of a broadband, femtosecond source into a narrow-band, picosecond amplifier, we have recently upgraded the numerical code COLAX. This code, initially described in [14], uses a Maxwell-Bloch formalism to calculate the temporal and spatial evolution of the electric field associated with the amplified radiation. In particular the Maxwell equation is coupled to the equation describing the time-dependency of the atom dipole polarisation density. Until now the code was used in the adiabatic approximation, where the polarisation is assumed proportional to the electric field at each time. We find that the adiabatic approximation, which is widely used, is not valid in our case, as illustrated in Fig. 5.

Fig. 5. Distribution of field intensity in the amplifying plasma, at a given time during the propagation of an harmonic pulse (narrow elongated feature in each image), which is incident from the left. Left image: time-dependent treatment of the polarization density. Right image: adiabatic approximation.

The left image of Fig. 5 shows the spatial distribution of amplified intensity at some time during the propagation of the harmonic pulse, calculated with a time-dependent treatment. One can see that amplification does not occur within the short harmonic pulse but later, in a wake induced by this pulse. This wake is totally absent in the adiabatic case shown in the right image of Fig. 5. Here amplification of harmonic radiation takes place within the harmonic pulse and ASE is the dominant feature at the forefront of the amplified radiation pulse. This result, discussed in more detail in

Recent Progress in X-Ray Laser Development and Applications

19

[15], is a first step towards more quantitative simulations that can be compared with experimental results. However it constitutes an important validation of the approach used in COLAX.

4 Interference microscopy for laser-induced damage studies We have carried out a new XRL interferometry experiment, using the PALS facility in Prague, in collaboration with IOP and CEA-CESTA. The main aim of this experiment was to apply the technique of interference microscopy to the investigation of laser-induced damages to optical surfaces exposed to 3ω laser pulses. The application part of the experimental set-up is shown in Fig. 6, while the X-ray laser beam is generated in another target chamber, not represented.

Fig. 6. Experimental set-up showing the XRL beam path to the sample and the interferometer. The damaging 3w beam is incident from the top on the front surface of the sample, while the XRL beam samples the rear surface.

The surface sampled by the XRL beam is the rear surface of a 10 mm thick beam-splitter, irradiated on the front face by a 3ω laser pulse. Several optical in-situ and post-shot inspection diagnostics were implemented but the main diagnostic was a microscope interferometer designed to map the target surface with a high spatial resolution [16]. The effect of the 3ω irradiation on the rear surface was studied by taking a sequence of three interferograms: (i) before irradiation as a reference, (ii) during or soon after the 3ω pulse, and (iii) 20 min later to detect any permanent damage. The results obtained, presented in more detail in [17], are summarized in

20

A. Klisnick et al.

Fig. 7, where lineouts along each of the three interferograms are shown. The main feature is apparent in the interferograms taken 5 ns after the 3ω irradiation. One can see that in the zone exposed to the 3ω laser beam, the interference fringes disappear, while they are totally recovered in the interferograms taken 20 minutes later. This transient loss of reflectivity of the sampled surface was rather unexpected and its interpretation is not fully understood yet. One possibility involves heating of the surface over a thin layer, leading to an increase of the surface roughness and hence to the observed loss of reflectivity.

Fig. 7. Lineouts across three interferograms taken successively, on the same sampled surface, at three stages of 3ω irradiation: before, during (t=5 ns) and after irradiation by the 3ω laser pulse. In the area exposed to the 3ω irradiation fringes are observed to disappear, while they are fully recovered in the post-mortem shot.

5 Conclusion and prospects In conclusion we have demonstrated a 3µJ XRL operating at 10 Hz, yielding a high average power of 30µW. Detailed information about the spatial behaviour of the source with respect to the GRIP angle were obtained, using spatially resolved X-ray diagnostics. New developments in the numerical simulations of transient X-ray lasers were carried out. The results obtained will be very important to guide future experimental works aimed at improving the quality of the XRL beam. The technique of interference microscopy was applied to the investigation of laser-induced damages in optical components. The results obtained show that the surface sampled by the XRL beam is subject to

Recent Progress in X-Ray Laser Development and Applications

21

transient alteration induced by the 3ω damaging laser beam, although the exact nature of the processes involved is not fully understood yet. All these works open promising prospects for the research program that will be carried out at the LASERIX facility. The current status and high potential of this facility is described in a companion paper [18]. The construction of the LASERIX laser driver was recently completed and the expected performances were demonstrated, namely an energy (before compression) of up 30 J at a record repetition rate of 0.1 Hz [19]. This performance was obtained because two major limiting factors were successfully overcome: (i) homogenization of the 2ω laser beams which are used to pump the final amplifier and (ii) parasitic lasing which can take place in the transverse direction of the large aperture TiSa crystal amplifier. Future work with the LASERIX laser includes the implementation of the GRIP pumping using the front-end at 2 J, 10 Hz. Our goal is then to investigate seeding with high-order harmonic radiation. The use of the full energy of LASERIX will be possible in mid-2007. This will then allow to extend pumping towards shorter wavelengths, possibly significantly below 10 nm, at a substantially improved repetition rate of 0.1 Hz.

Acknowledgments Financial support by the Access to Research Infrastructures activity in the 6th Framework Programme of the EU, Contract RII3-CT-2003-506350, LASERLAB Europe is gratefully acknowledged.

References 1. Cassou, K et al, 'Optimization towards ', accepted in Opt. Lett, 2006 2. Kazamias S. et al., ' A 10 Hz, 3 µJ transient X-ray laser pumped in grazing incidence geometry ', These Proceedings, 2006. 3. Shlyaptsev V. et al., in ′Soft x-ray laser and application V′, SPIE Proc. 5197, 221, 2003. 4. Keenan, R., 'High-Repetition-Rate Grazing-Incidence Pumped X-Ray Laser Operating at 18.9 nm', Phys. Rev Lett 94, 103901, 2005. 5. Zielbauer, B. et al., 'X-ray imaging of the heating zone of non-normal incidence pumped XRL plasma', These Proceedings, 2006.

22

A. Klisnick et al.

6. Klisnick, A. et al., in soft x-ray laser and application IV, Proc. SPIE Vol 4505, 75, 2001. 7. Dunn, J. et al., in soft x-ray laser and application IV, Proc. SPIE Vol 4505, 62, 2001. 8. Guilbaud, O. et al., ‘Origin of microstructures in picosecond X-ray laser beams’, These Proceedings, 2006. 9. Guilbaud, O. et al., ‘Origin of microstructures in picosecond X-ray laser beams’, Euro. Phys. Lett. 74, 823, 2006. 10. Cassou, K. et al., ‘2-D Hydrodynamic simulation of laser plasma generation for transiently pumped soft x-ray amplifier’, These Proceedings, 2006. 11. Cassou, K. et al., ‘Transverse spatial improvement of a transiently pumped soft X-ray amplifier’, to appear in Phys. Rev A, 2006. 12. Zeitoun, P. et al., ‘xxx’, Nature 431, 426, 2004. 13. Rocca, J. et al., ‘Advances in high repetition rate soft X-ray lasers’, These Proceedings, 2006. 14. Larroche, O. et al., Phys. Rev. A 62, 043815 2000. 15. Al Miev, I. et al., ‘Seeding high–order harmonic in a transient X-ray laser amplifier’, These Proceedings, 2006. 16. Rus, B. et al., ‘Advanced optical damage studies using X-ray laser interferometric microscopy’, in soft x-ray laser and application IV, Proc. SPIE Vol 4505, 75, 2005. 17. Jamelot, G. et al., ‘X-ray laser interferometric microscopy for advanced optical damage studies’, These Proceedings, 2006. 18. Ros, D. et al., ‘LASERIX : a multi X-Ray/XUV beamline high repetition-rate facility’, These Proceedings, 2006. 19. Plé, F. et al., ‘A high repetition-rate, petawatt class Ti:Sapphire laser dedicated to coherent X-UV generation at LASERIX’, These Proceedings, 2006.

A High Repetition-Rate, Petawatt Class Ti:Sapphire Laser Dedicated to Coherent X-UV Generation at LASERIX. F. Plé (1,2), M. Pittman (1), J-P. Chambaret (3) and G. Jamelot (1) 1

LIXAM, UMR n° 8624, Université Paris-Sud 11 / CNRS, Bâtiment 350, 91405 Orsay cedex, France 2 AMPLITUDE Technologies. CE2926. 91029 Evry cedex. France 3 LOA, UMR n°7639, ENSTA / CNRS / Ecole Polytechnique,Chemin de la Hunière. 91761 Palaiseau cedex. France

Summary. We have recently demonstrated amplification in a Ti:Sapphire laser system up to 30 Joules before compression at the record repetition rate of 0.1 Hz. This CPA Ti: Sapphire laser system is the first femtosecond petawatt class laser system running at a high repetition rate of 0.1Hz. Such performances have been obtained thanks to the particular design of the last booster amplifier. In this paper, we will present the last results obtained on this system and the different strategies we have carried out for counteracting the two most limiting factors for high energy amplification that are the parasitic lasing in the large aperture Ti: Sapphire crystal and the spatial homogenisation of the pump laser beams. This laser will be used as a driver of the LASERIX facility.

1 Introduction As high power amplification in CPA1 Ti:Sa laser system has continually evolved towards higher and higher energy levels, new technological issues have appeared. Two most limiting factors have arisen for very high energy amplification in Ti:Sa laser crystals: the pump laser beam homogenization and parasitic self-lasing2. Pump lasers profiles dedicated to such energy amplification are hard to be controlled. As the pump diameter increases, spatial effects in rods amplification induce large non-uniformities in the laser-beam intensity profile and present a big issue for crystal safety and amplification efficiency. An other issue is self parasitic lasing in high dimension Ti:Sa crystals. This parasitic effect is due to the formation of a transverse laser cavity in which the crystal faces act as mirrors1. Because of the geometrical characteristics of new large-aperture Ti:Sa amplifiers, the

24

F. Plé et al.

transverse gain could be very high and generate huge parasitic effects. These two points was the object of particular developments

2 The LASERIX driver The laser developed by the LIXAM as a driver for LASERIX3 is a classical CPA Ti:Sa laser system. It is based on a Femtolaser oscillator, a 1 KHz regenerative amplifier and 4 multipass amplifiers. The first three elements were provided by Thales Lasers and deliver a chirp pulse of 500 ps / 25 mJ with a spectral bandwidth of 25 nm FWHM. This part is seeded in two cryogenically booster amplifiers developed by Amplitude Technologies which provide a laser pulse of 2 joules to be injected in the last booster amplifier This last amplifier takes shape around one large aperture Ti:Sa crystal. It is 1.95 cm long (αl=3.5) and its diameter is 10 cm. This crystal is pumped by 8 Nd:YAG homogenised laser beams of 12.5 joules of energy and 30ns FWHM long delivered by Quantel. We are expecting 40 joules at the end of this last amplifier THALES LASER Öffner Stretcher

1kHz Regenerative Amplifier

AMPLITUDE TECHNOLOGIES 10 Hz - 4 Pass Preamplifier S

10Hz - 4 Pass Cryogenic Power Amplifier

10Hz - 4 Pass Cryogenic Power Amplifier P

P

P

P

P

P

P

P

J

Oscillator V

F emtolaser

azzler

D

Fig. 1. Front end of the CPA laser system of the LASERIX facility.

3 Experimental results Pumping laser beam shaping was performed by using a homogenizing system based on Micro-Lenses Array4 (MLA). Thanks to this method, we have succeeded at the same time to shape each elementary beam into homogenise super-Gaussian intensity profile and to up collimate the pumping laser beams from 2 cm to 6cm without using any afocal system. This particular method, dedicated to low coherent laser system, allows pumping the crystal with fluency greater than 2 J / cm² with superGaussian intensity profiles.

A High Repetition-Rate, Petawatt Class Ti:Sapphire Laser

25

Fig. 2. Image of the 60 mm diameter intensity pump profile measured on the crystal.

To limit parasitic self lasing effects, we have chosen to perform index matching with a high index of refraction liquid. In fact, the crystal we are using in the last stage of amplification is two highly doped in Ti3+ to use solid state index matching method. We had to match as better as possible the index of refraction of the Ti:Sa crystal to limit Fresnel refraction on the edge of the crystal5. Reflections on the two faces are settled by the AR coating.

Fig. 3. Image of the gain medium inside the Ti:Sa crystal. We observe a strong depletion of the gain owing to parasitic self lasing.

Finally, we flow the crystal with a derivative of diiodomethane (CH2I2) doped with an absorber for 800 nm radiation. Thank to this method and by seeding the IR pulse before the end of the pumping6, we succeeded in amplifying up to 30 joules with a transverse gain higher than 2000 and

26

F. Plé et al.

75 joules of pumping energy instead of the 100 joules theoretically available. Shot to shot energy at 0.1 Hz 35

Energy (Joules)

30 25 20 15 10 5 0 1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

Fig. 4. Shot to shot energy at the end of the last booster amplifier.

Preliminary pulse compression has been made for demonstration with a two gratings compressor by sampling some part of this energy. SPIDER measurements of the pulse duration give results in the range of 40 fs. Two big vacuum compressors are currently under development for high-energy compression

4 Conclusion We have developed a femtosecond Petawatt class laser system running at 0.1 Hz as a laser driver of the future LASERIX X-ray laser facility. We plan now to amplify with all the pumping energy to achieve the energy level require for the project and to compress half of this energy as soon as the laser should be installed in its dedicated building.

Acknowledgments This work was supported by the Conseil Général de l’Essonne and the Ministère de l'Education Nationale within the framework of the CPER 2000-2006. LASERIX is a X-ray laser facility of the University Paris-Sud 11 that will be installed in a new building of the LOA at the Ecole Nationale Supérieure des Techniques Avancées (ENSTA) in Palaiseau.

This page intentionally blank

A High Repetition-Rate, Petawatt Class Ti:Sapphire Laser

27

The contribution of the LIXAM, and particularly of Annie Klisnick and David Ros to the choice of titanium:sapphire technology for the LASERIX driver was highly appreciated. We warmly thank researchers and engineers from LOA for their fruitful collaboration.

References 1. D. Strickland and G. Mourou, : Compression of Amplified Chirped Optical Pulses, Opt. Comm. 56, 219, 1985. 2. F. G. Patterson, J. Bonlie, D. Price, and B. White : Suppression of parasitic lasing in large-aperture Ti:Sapphire laser amplifiers, Opt. Lett. 24, 14, 1999. 3. D. Ros, G. Jamelot, M. Pittman, F. Plé, S. Kazamias, A. Klisnick, J-C. Lagron, K. Cassou, O. Guilbaud, J-P. Chambaret, S. Sebban, P. Zeitoun : LASERIX : a multi X-Ray/XUV beamline high repetition-rate facility, in these Proceedings 4. A. DiChiara, E.A. Chowdhury, G. Ongadi, and B.C. Walker : TEM00 terawatt amplification by use of micro-optic spatial mode conversion, Opt. Letters, 28, 21, 2003. 5. M. P. Kalashnikov, V. Karpov, H. Schnönnagel, and W. Sandner : 100-terawatt titanium-sapphire laser system, Laser physics 12, Nr.2, 2002. 6. V. Chvykov, V. Yanovsky, S. W. Bahk, G. Kalintchenko, and G. Mourou : Suppression of parasitic lasing in multi-pass Ti:Sapphire amplifiers, CLEO conference, Baltimore (2003).

Recent Advances of the X-Ray Laser Research at APRC K. Nagashima, M. Kishimoto, T. Kawachi, N. Hasegawa, M. Tanaka, Y. Ochi, M. Nishikino, K. Sukegawa, H. Yamatani, Y. Kunieda and Y. Kato X-Ray Laser Group, Advanced Photon Research Center, Quantum Beam Science Directorate, Japan Atomic Energy Agency

Summary. We are developing plasma x-ray lasers pumped by ultra-short pulse lasers and have concentrated our efforts on developing a silver x-ray laser (13.9 nm) with spatially full coherence by means of a double target method. Recently, we have succeeded in developing a unique new pumping laser system, which is a CPA laser consisting of OPCPA and zigzag-slab type glass amplifiers. The OPCPA was adopted for improving the controllability of pumping pulses instead of a regenerative amplifier and the zigzag-slab amplifiers were adopted for increasing the repetition rate.

1 Introduction Collisional excitation scheme is the main current in plasma x-ray laser (XRL) research. This scheme succeeded in lasing in 1984 for the first time by using a huge-scale laser facility [1]. Then, many efforts have been made in order to reduce the pumping energy [2, 3]. Two advanced schemes were demonstrated by using CPA ultra-short pulse lasers. They were transient collisional excitation (TCE) scheme [4] and optical field ionization (OFI) scheme [5]. The TCE scheme was a method similar to the previous collisional excitation using a pre-pulse technique [2] and succeeded in reducing the pumping energy less than 10 J [4, 6]. Recently, a further advanced scheme, which is called grazing incidence pumping (GRIP), succeeded in reducing the pumping energy less than 1 J [7]. It means that we can now get XRL light by using commercial-based Ti:sapphire CPA lasers. So, it can be said that the trend of reducing the pumping energy has reached a satisfactory level. However, the beam quality of XRL stays at a poor level. An essential cause for the poor beam quality is that the XRL is light by amplified spontaneous emission (ASE) and the spatial mode is not

30

K. Nagashima et al.

selected in almost cases. Some groups tried to improve the beam quality by means of double pass amplification [8], double target configuration [9] and waveguide with a curved target [10]. An XRL beam with spatially full coherence beam was generated by the double target configuration [11], where an XRL seeder from the first target was amplified in an XRL amplifier on the second target. Recently, a higher harmonic seed was amplified in an XRL medium (the medium was generated by the OFI scheme) and a temporally full coherent beam was generated [12]. We think that this new trend of improving beam quality is essential in XRL research from now on.

2 Concept of MOPA The key concept for improving the beam quality is MOPA (masteroscillator/power-amplifier), which means amplification of seeding light. The seeding light is expected to be spatially and temporally full coherent, if it is possible. The several configurations of MOPA are considered as shown in Fig. 1. Seeder

Amplifier

(a)

Amplifier

Seeder

(b) Seeder

Mirror Amplifier

(c)

Mirror Spatial filter Amplifier

Coherent seeder

(d)

Fig. 1. Configurations of MOPA

Figure 1 (a) is a case where the sizes of the media are equal between the seeder and the amplifier. In this case (the seeding light has some divergence), the amplifier works as a spatial filter and spatially full coherence is realized by adjusting the distance between the seeder and the

Recent Advances of the X-Ray Laser Research at APRC

31

amplifier [11]. Figure 1 (b) is a case where the size of the amplifier is larger than that of the seeder. If the seeder is coherent, it is the simplest configuration of MOPA. In this case, the most important problem is how to make a large-scale amplifier. Figure 1 (c) is a case where the seeder is not coherent satisfactorily and the spatial mode is selected by a spatial filter. Figure 1 (d) is a case of optics-free, which is equivalent to (c). A candidate of the seeder is higher harmonic light (HHL) or XRL itself. The HHL has higher spatial and temporal coherences compared with those of XRL. In general, the spectral width of HHL is considerably broad compared with that of XRL. So, when the HHL seed is amplified in the XRL amplifier, the spectral range is limited by the bandwidth of the XRL. The amplifier works as a spectral filter and the pulse length of the amplified beam is determined by Fourier transform limit. In this sense, it seems that a combination of the HHL seeder and the XRL amplifier is the most suitable for improving temporal coherence. It is thought that the XRL seeder is better for improving only the spatial coherence.

3 Progress of Our Research We have developed TCE-XRLs using a high power CPA laser (Ti:sapphire front-end and rod-type Nd:glass power amplifier) [13]. An original point of our experimental device is an use of coaxial pulses (pre-pulse and main pulse). The two successive pulses are focused with a line shape by the same focusing optics. So, we have no additional optics for the pre-pulse. XRL seeder

Seeding beam XRL amplifier

Main pulse

Amplified beam

Pre-pulse

Traveling wave pumping

Fig. 2. Configuration of double target method

The saturated amplification was achieved by travelling wave pumping using a 6-step mirror [14]. Our original approach for improving the beam quality is double target method, where an XRL seeder and an XRL amplifier are generated on the two targets (see Fig. 2) and the XRL amplifier works as a spatial filter. The diffraction-limited beam was generated by this method, and it was found that the beam was spatially full

32

K. Nagashima et al.

coherent from Young’s double slit experiments (the spatial coherent length was longer than the beam diameter) [15]. For the present pumping laser system, the pumping conditions are equal between the two targets. The optimum condition for the amplifier was obtained when the gain was not maximal. It means that the optimum condition for the amplifier is different from that for the seeder. Therefore the pumping condition should be controlled individually between the seeder and the amplifier.

HHL seeder XRL beam

Gas jet

Ti:sapphire laser

XRL amplifier

Amplified HHL

Main pulse

Pre-pulse

Traveling wave pumping

Fig. 3. Configuration of amplification of higher harmonic light

We have tried amplification of HHL seeder [16]. The experimental configuration is shown in Fig. 3. In the experiment, the 29th HHL of Ti:sapphire laser (the wavelength 780 nm, the pulse length 80 fs) was generated in argon gas and the HHL was amplified in an XRL medium of neon-like manganese (the wavelength λ = 26.9 nm). The spectral width (FWHM) of the HHL was measured to be 0.12 nm and the pulse length was less than 80 fs. On the other hand, the spectral width of the XRL was estimated to be Δλ/λ ~ 10-4 (where λ is the wavelength). Therefore, it is thought the pulse length of the amplified HHL was determined by Fouriertransform limit and was about λ 2/2cΔλ ~ 0.5 ps (where c is light speed). In this case, the XRL amplifier works as a spectral filter and the amplified HHL has fully temporal coherence.

4 New Pumping Laser System The present pumping laser system consists of a Ti:sapphire oscillator, a pulse stretcher, a Ti:sapphire regenerative amplifier, a pulse controller, Nd:glass rod-type amplifiers and two pulse compressors [13]. The pulse controller makes a pre-pulse and a main pulse coaxially. This pumping beam is divided into two beams before the final rod amplifiers. The two beams irradiate two targets, which makes the XRL seeder and the XRL

Recent Advances of the X-Ray Laser Research at APRC

33

amplifier. There are some problems in the present pumping system. Firstly, an unexpected pre-pulse is generated in the regenerative amplifier. It makes the pulse controllability low. Secondly, the two beams have an identical pumping condition in spite that the different pulse conditions are expected between the seeder and the amplifier. Thirdly, the repetition rate of the system is very low. It is limited to 0.001 Hz by the cooling time of the glass rod. OPCPA amp. 10 Hz, 3 mJ

Pulse stretcher

Ti:sap oscillator 80 MHz, 3 nJ

Pulse controller.1

Zigzag slab amp.1 0.1 Hz, 10 J

Pulse compressor.1

Pulse controller.2

Zigzag slab amp.2 0.1 Hz, 10 J

Pulse compressor.2

Fig. 4. Block diagram of new pumping laser system

Now we are installing a new pumping laser system, which is able to solve the above-mentioned problems. The main advances are a change of the front-end amplifier from the regenerative amplifier to an OPCPA amplifier and a change of the main amplifier from the Nd:glass rod-type amplifiers to Nd:glass zigzag-slab-type amplifiers. The components of the new laser system are shown in Fig. 4. The OPCPA amplifier generates no unexpected pre-pulse and no ASE. So, the pulse controllability is improved considerably. The zigzag-slab-type amplifier can make the repetition rate up to 0.1 Hz, because the thermal effects are compensated in the zigzag-slab geometry. Furthermore, the pulse controller is improved and it can make two pre-pulses and one main pulse. The pumping beam is divided into two beams before the pulse controllers and the two pulse controllers can make individual pumping conditions for the two beams. So, we can optimize the pumping pulses for the XRL seeder and the XRL amplifier individually. We have confirmed the stable operation of the zigzag-slab-type amplifier at 0.1 Hz with the output energy of about 10 J.

5 Preliminary experiments using the new laser system A spatial pattern of the pumping beam from the zigzag-slab-type amplifier is square and has fairly good uniformity. So, it is supposed that the pattern

34

K. Nagashima et al.

after focusing to a line shape is improved compared with that from the rodtype amplifier. We carried out an experiment for lasing at 13.9 nm by temporarily replacing the rod-type amplifier to the zigzag-slab-type amplifier. The result is plotted in Fig. 5 (single target experiment). The small signal gain coefficient is 41.9 and the output energy is saturated at a gain-length product of 16.3, which is an appropriate value. The spatial divergence of the XRL beam was about 3 mrad. It is about a half of the previous value. The reason of this low divergence may be due to the improved focusing.

XRL Output Energy (mJ)

10 1 10-1 10-2

g ~ 41.9 gL ~ 16.3

10-3

0

0.2 0.4 0.6 0.8 Target Length (cm)

1.0

Fig. 5. XRL lasing at 13.9 nm using the zigzag-slab-type amplifiers

As above-mentioned, the pulse controllability is improved by replacing the regenerative amplifier to the OPCPA amplifier, because there are no prepulse and no ASE. We examined a pre-pulse effect using the OPCPA amplifier and the new pulse controller. The experiment was carried out for lasing at 13.9 nm. The pumping pulses consisted of a pre-pulse with 200 ps pulse length and a main pulse with 5 ps pulse length. The interval between the two pulses was 2.75 ns and the total pumping energy was about 10 J. From the experiments, it was found that the XRL beam direction and the output energy were strongly dependent on the pre-pulse. The results are plotted in Fig. 6. The horizontal beam direction was plotted in Fig. 6a as a function of energy ratio of the pre-pulse over the main pulse. The plotted values are relative angles from the direction at Pre/Main = 4.3 %. Increasing the pre-pulse energy ratio, an expansion speed of the preplasma increases and the scale length of the plasma density becomes longer. So, the refraction of the XRL beam due to the density gradient decreases with increasing the pre-pulse energy ratio. It is thought that the

Recent Advances of the X-Ray Laser Research at APRC

35

8 7 6 5 4 3 2 1 0

700 XRL beam horizontal direction (a)

0

0.01 0.02 0.03 0.04 0.05 Energy ratio of Pre / Main

XRL output energy [ nJ]

Relative angle [ mrad]

expansion speed is equal to the sound velocity and is proportional to the square root of the pre-pulse energy. It is consistent with the plotted values. Since the effective gain area increases with increasing the density scale length, it is reasonable that the XRL output energy increases with the prepulse energy ratio as shown in Fig. 6b.

600

(b)

500 400 300 200 100 0

0

0.01 0.02 0.03 0.04 0.05 Energy ratio of Pre / Main

Fig. 6. Effects of pre-pulse on the XRL beam direction and the output energy

6 Conclusions We have succeeded in lasing at 13.9 nm with spatially full coherence by means of the double target method, which is a MOPA concept using the XRL seeder and the XRL amplifier. Now, the trend of XRL research is toward improvement of the beam quality. We expect to make advances in this trend by developing the new pumping laser system, which consists of the OPCPA amplifier and the zigzag-slab-type glass amplifiers. It was proved that the OPCPA has a capability for improving the pulse controllability and the zigzag-slab amplifier has a capability for stable operation with the output energy of 10 J and the repetition rate of 0.1 Hz.

References 1. 2. 3. 4. 5. 6. 7.

D. L. Matthews et al., Phys. Rev. Lett. 54, 110 (1985) J. Nilsen et al., Phys. Rev. A 48, 4682 (1993) S. Basu et al., Appl. Rev. B 57, 303 (1993) P. V. Nickles et al., Phys. Rev. Lett. 78, 2748 (1997) B. E. Lemoff et al., Phys. Rev. Lett. 74, 1574 (1995) J. Dunn et al., Phys. Rev. Lett. 80, 2825 (1998) R. Keenan et al., Phys. Rev. Lett. 94, 103901 (2005)

36 8. 9. 10. 11. 12. 13. 14. 15. 16.

K. Nagashima et al. A. Carillon et al., Phys. Rev. Lett. 68, 2917 (1992) C. L. S. Lewis et al., Opt. Commun. 91, 71 (1992) R. Kodama et al., Phys. Rev. Lett. 73, 3215 (1994) M. Tanaka et al., Opt. Lett. 28, 1680 (2003) Ph. Zeitoun et al., Nature 431, 426 (2004) T. Kawachi et al., Appl. Opt. 42, 2198 (2003) T. Kawachi et al., Phys. Rev. A 66, 033815 (2002) M. Nishikino et al., Phys. Rev. A 68, 061802 (2003) N. Hasegawa et al., Inst. Phys. Conf. Ser. 186, 273 (2004)

X-Ray Lasing Using the GRIP Scheme J.E. Balmer, M. Grünig, C. Imesch and F. Staub Institute of Applied Physics, University of Bern

Summary. We report on x-ray laser experiments using the grazing-incidence pumping (GRIP) scheme with 8-ps duration, 1054-nm pump pulses and energies up to 3 J. Intense x-ray lasing from targets of Mo, Pd, Ag, and Sn has been obtained by optimizing the prepulse configurations (long pulse – short pulse – grazing incidence).

1 Introduction Important progress towards higher efficiency, reduced size, and higher repetition rate of soft-x-ray lasers has been achieved in recent years by adopting the transient-collisional-excitation (TCE) scheme as proposed by Afanasiev and Shlyaptsev [1]. In this scheme, a first nanosecond-duration laser pulse at 1011-1012 W/cm2 generates the plasma and the required closed-shell (Ne-like or Ni-like) ionization conditions. A second, much shorter (picosecond) laser pulse at 1014-1015 W/cm2 then rapidly heats the preformed plasma and generates a transient population inversion. If the fast temperature rise is of the order of the time scale of the collisional redistribution of the excited levels, it allows efficient pumping without perturbing the ionization. Very high x-ray laser gain coefficients (>100 cm−1) have been predicted with the possibility of saturated gain for target lengths of less than 1 cm. Experimentally, saturated operation of TCE x-ray lasers with gain coefficients as high as 62 cm−1 has been demonstrated by several groups [2]. The advantage of this scheme is that less than 10 J of laser energy from a chirped-pulse amplification (CPA) laser are sufficient to drive the inversion. Very recently, a novel irradiation geometry, the grazing-incidence pumping (GRIP) technique has been introduced [3], which has resulted in a further reduction of the pump pulse energy to ~1 J for saturated lasing at wavelengths down to 13.2 nm [4]. In the GRIP scheme, a preformed plasma is generated, much as in the TCE scheme, by irradiating the target with a long (~200 ps) laser pulse at normal incidence. The plasma is allowed to expand for a given time to

38

J. Balmer et al.

allow the electron density gradients to relax. The short (~ps) pump pulse then irradiates the expanding plasma plume at a grazing angle. The angle of incidence θ is chosen in such a way that the turning point density of the 2 pump radiation, ne - given by refraction as ne = necθ , where nec is the critical density for the pump laser wavelength - coincides with the density for which maximum gain is obtained for a given x-ray laser wavelength. Absorption of the pump energy thus occurs directly and very efficiently into the gain region. In this paper we report on the first phase of implementation of the GRIP scheme into the Nd:glass laser system used for x-ray laser experiments at our laboratory. The laser has recently been upgraded into a CPA (chirpedpulse amplification) system and currently produces a maximum of 6 J on target in a compressed pulse of 8 ps duration. One or several prepulse(s) can be introduced either along a separate beam line or, optionally, along the GRIP beam line. At the available shot rates of 5/hour for the 3 J experiments and 2/hour for the 6 J experiments, the optimization of the various prepulse configurations becomes a rather formidable task. Nonetheless, intense x-ray lasing was observed from Ni-like plasmas of Mo, Pd, Ag, and Sn at 18.9, 14.7, 13.9, and 11.9 nm, respectively, using a single prepulse irradiating the target along the GRIP beam line.

2 Experimental setup The experiments were conducted using the CPA upgrade to the existing Nd:glass laser system operated at our laboratory for a number of years. At the front end, the upgrade includes a commercial 100 mW, 200 fs Nd:glass oscillator followed by a double-pass, single-grating pulse stretcher and a diode-pumped Nd:glass regenerative amplifier producing a pulse energy of ~1 mJ. Due to gain narrowing during the amplification process, the bandwidth and the duration of the pulses are reduced to ~2 nm and 550 ps, respectively, at this point. The pulses are further amplified in the 5-stage main amplifier chain consisting of flashlamp-pumped Nd:glass rods having diameters of 10, 16, 25, 45, and 90 mm. The maximum energy output of the system is limited by the damage threshold of the 1740 line/mm compressor gratings, given by the manufacturer as 250 mJ/cm2. With the laser operated at the current beam diameter of 80 mm and for an angle of incidence of 61° on the gratings, this translates into a maximum energy of ~25 J incident on the first grating. Taking into account the 4-pass geometry of the compressor, the diffraction efficiency of the gratings of ~90 %, and the 2:1 peak-to-average ratio of

X-Ray Lasing Using the GRIP Scheme

39

the beam fluence distribution, a maximum of ~8 J was available for x-ray laser experiments in this phase. The experimental setup used for the x-ray laser experiments is shown schematically in Fig. 1. The main pulse beam was focused at a grazing angle of 22.5° using a multilayer-coated f = 91.4 cm parabolic mirror. This angle has been imposed, as shown in the figure, by the availability of an entrance port on the existing target chamber. The line focus was measured at low power to be 30 μm in width and 7.5 mm (FWHM) in length. The experience from previous experiments with our system implies that the line focus width will be approximately doubled at high power. A longduration (500 ps) prepulse was introduced along a separate beam line and focused by a single plan-convex lens tilted at the appropriate angle to produce a line focus of similar dimensions. An additional prepulse could be propagated along the main pulse beam line, thus producing a compressed 8-ps prepulse as an option. The main diagnostics of the plasma emission was an on-axis, time-integrating spectrometer that consists of a 1200 lines/mm, aberration-corrected, concave Hitachi grating (radius of curvature: 5649 mm), working at a grazing-incidence angle of 3 ° . The grating disperses the incident radiation on a 40 mm diameter P20 phosphor screen, which was imaged to a cooled CCD camera with a pixel size of 23 μm square. The wavelength coverage of the spectrometer was ~12 nm with a spectral resolution of ~0.2 nm.

3 J/8 ps + pp

On-axis parabola f = 91cm

< 4 J/500 ps

22.5° XRL

Spectrometer/ CCD

Fig. 1. Experimental setup for x-ray laser experiments at a fixed angle of incidence of 22.5°.

40

J. Balmer et al.

3 Soft-x-ray laser results In a first phase of GRIP experiments, a number of prepulse configurations were investigated in order to find the optimum prepulse conditions. The targets used were 25 mm wide slabs of Mo and Pd, having a diamondmachined surface finish. The targets were positioned in such a way that the line focus hit the surface close to the edge on the spectrometer side. Overfilling of the target on this side to avoid cold, absorbing plasma did not seem to be of major importance. A list of prepulse configurations for which x-ray lasing was observed is given in Table1. We note that in all the cases investigated, the existence of a short (compressed) prepulse was found to be beneficial to x-ray lasing; in fact it turned out that best results were obtained with a single prepulse incident along the GRIP path. This configuration was thus used for further optimization with respect to the prepulse-to-main pulse time interval. Figure 2 shows a set of on-axis spectra from targets of Mo, Pd, Ag, and Sn, irradiated with a 120 mJ prepulse preceding the 3 J main pulse by 13 2 2.4 ns. This translates into irradiances of ~8×10 W/cm for the main 12 2 pulse and ~4×10 W/cm for the prepulse. Lasing in Sn at 11.9 nm was found to be very weak under these irradiation conditions; most likely due to the fact that the 22.5° angle of incidence is too far from optimum for this element (see discussion below). In order to find the optimum delay conditions between prepulse and main pulse, a systematic study was conducted to investigate the x-ray laser output intensity as a function of the prepulse-to-main pulse time interval for a range of prepulse amplitudes, as shown in Figure 3 for the case of a Pd target. As can be seen, lasing on the 14.7-nm line is observed for prepulse amplitudes of 0.5-16 % and main pulse delays of up to 11 ns. Table 1 Prepulse configurations with x-ray lasing observed

prepulse(s)

main pulse

delay

Target

0.8 J/15 ps, along prepulse path (w/o pulse compression) 4 J/500 ps, along prepulse path + 0.15 J/8 ps along GRIP path 0.15 J/8 ps, along GRIP path + 0.5 J/8 ps along GRIP path 0.5 J/8 ps, along GRIP path

2 J/15 ps

300-700 ps 300 ps 5 ns 5 ns 800 ps 800 ps

Mo

3 J/8 ps 3 J/8 ps 3 J/8 ps

Pd Pd Pd

X-Ray Lasing Using the GRIP Scheme

41

λ

2nd order

Mo

Pd

Ag

Sn

Mo Pd Ag Sn (x50)

Intensity (counts)

8000 6000 4000 2000 0 5

10

15

20

25

30

Wavelength (nm)

Fig. 2. On-axis spectra from targets of Mo, Pd, Ag, and Sn irradiated at a fixed angle of incidence of 22.5°.

7000

2.8% 4.5% 8% 16% 0.5%

Intensity (counts)

6000 5000 4000 3000 2000 1000 0 0

2

4

6

8

10

12

Main pulse delay (ns)

Fig. 3. 14.7 nm laser intensity from a Pd target vs. main pulse delay, for a range of prepulse amplitudes.

This page intentionally blank

42

J. Balmer et al.

This is in distinct difference to the results of Refs. 3 and 4, where substantially narrower temporal windows of ~50 ps and ~800 ps have been reported for lasing in Mo and Ag, respectively. Of course, those experiments were driven either just above the threshold for lasing (Mo) or just above saturation (Ag), with main pulse energies of ~100 mJ and ~1 J, respectively. With a main pulse energy of 3 J as in our case, lasing should be expected to be well into the saturation regime and thus less critical with respect to the conditions of the preformed plasma. This has not been confirmed yet by measurement, but a comparison with previous data [5] suggests that a maximum value of the x-ray pulse energy of ~3 μJ has been attained in our current experiments, a factor of 4-5 higher than the values quoted in Ref. 4 at saturation, for a comparable pump pulse duration.

4 Discussion As shown above, lasing in Sn at 11.9 nm was found to be very weak under the current irradiation conditions or, more specifically, at the fixed 22.5° angle of incidence. (In the meantime, intense lasing has been obtained also from Sn targets, however, at increased main pulse energy of 6 J). The optimum angle of incidence for Sn has been reported to be 23° for a Ti:sapphire pump laser wavelength of 800 nm. Since the optimum angle scales with wavelength as θNd = θTi λNd /λTi ≈ 1.32θTi, this translates into an optimum angle of 30° for the 1054 nm Nd laser wavelength used in our case. Experiments at this angle will be conducted upon installation of a new target chamber in the very near future. An interesting point in the results shown in Fig.3 is the fact that, with the exception of the lowest energy prepulse (0.5 %), all the curves exhibit a double-humped dependence of the x-ray laser intensity on the time interval between prepulse and main pulse. This behaviour is not understood yet and will require more work in the future. A possible explanation could be that two temporal regimes exist for lasing, one during the ionizing phase of the plasma and the other during the recombining phase of the interaction. Between these two regimes, the plasma is overionized and gain is reduced. However, this is rather speculative and will require simulations to be performed, which is outside the scope of this work. Also, the irradiation history has been checked for spurious prepulses -5 and it was found that the pulse contrast ratio is better than ~10 for times down to the 1-ns response time of the photodiode used.

X-Ray Lasing Using the GRIP Scheme

43

4 Summary In summary, we have demonstrated intense x-ray lasing from nickel-like plasmas of Mo, Pd, Ag, and Sn by utilizing a short (8 ps) prepulse irradiating the target at a grazing angle of incidence of 22.5°, i.e., collinear with the 3 J, 8 ps main pulse. Saturated lasing has not been proved yet, but comparison with signal levels achieved in previous work suggests that saturation has well been reached in these experiments. This is supported by the fact that a large range of prepulse energies and time intervals gives rise to intense x-ray lasing, as has been demonstrated in the case of Pd.

References 1.

2.

3.

4.

5.

Yu.V. Afanas'ev and V.N. Shlyaptsev: 'Formation of a population inversion of transitions in Ne-like ions in steady-state and transient plasmas', Sov. J. Quant. Electron. 19, 1606, 1989. M.P. Kalashnikov et al.: 'Saturated operation of a transient collisional x-ray laser', Phys. Rev. A 57, 4778, 1998; J. Dunn et al.: 'Gain saturation regime for laser-driven tabletop, transient Ni-like ion X-ray lasers', Phys. Rev. Lett. 84, 4834, 2000. V.N. Shlyaptsev et al.: “Numerical study of transient and capillary x-ray lasers and their applications”, Proc. SPIE, Vol. 5197, 221-228, 2003; R. Keenan et al., “Efficient pumping scheme for high-average brightness collisonal x-ray lasers”, ibid., 213-220. Y. Wang et al.: “Demonstration of high-repetition-rate tabletop soft-x-ray lasers with saturated output at wavelengths down to 13.9 nm and gain down to 10.9 nm”, Phys. Rev. A 72, 053807-1-7, 2005; J.J. Rocca et al.: “Saturated 13.2 nm high-repetition-rate laser in nickel-like cadmium”, Opt. Lett. 30, 2581-2583, 2005. J.E. Balmer et al.: “Towards full characterization of nickel-like soft-x-ray lasers”, Proc. SPIE, Vol. 4505, 93-99, 2001.

Development of 0.1-Hz Repetition Driver Laser for the TCE X-Ray Lasers Y. Ochi, N. Hasegawa, T. Kawachi, K. Nagashima, M. Kishimoto, M. Tanaka, M. Nishikino, K. Sukegawa, H. Yamatani and Y. Kunieda Kansai Photon Science Institute, Japan Atomic Energy Agency

Summary. We developed new power amplifier using a zigzag slab Nd:glass to generate the TCE x-ray lasers more frequently. The chirped laser light at the wavelength of 1053 nm was amplified to > 10 J and then compressed to be ~1.6 ps at repetition rate of 0.1 Hz. By using the amplified laser light, gain saturated x-ray laser with the Ni-like silver at wavelength of 13.9 nm was obtained.

1 Introduction Developments of the chirped pulse amplification (CPA) [1] laser and the transient collisional excitation scheme [2] enabled us to generate the soft x-ray lasers with laboratory-scale laser system. There have been much progresses on x-lay lasing with a several materials using the CPA laser amplified by Nd:glass power amplifiers [3-6]. In particular for the Ni-like silver laser at the wavelength of 13.9 nm, saturated amplification with energy out put of 25 μJ due to the traveling pumping was achieved [6]. Furthermore, spatially full-coherent beam with diffraction limited beam divergence of 0.2 mrad was demonstrated by means of the oscillatoramplifier scheme using double targets [7]. Now our interest is moving to application studies using the highly coherent soft x-ray beam. However, the present repetition of soft x-ray laser generation is limited to be 1 shot/20 minutes by cooling time of glass rods used for the typical power amplifier. This repetition is not sufficient for making practical application study. Herein we report the development of new power amplifier for the CPA laser using zig-zag slab Nd:glass which provides > 10 J energy light at 0.1 Hz repetition. The zigzag slab is suitable for high repetition operation because the thermal distortion effects in the glass are principally canceled by propagating the zigzag pass. Specification of the power amplifier system is presented in the next

46

Y. Ochi et al.

section. The result of x-ray lasing experiment with the Ni-like silver is discussed in section 3.

2 Development of power amplifier using zigzag slab Nd:glass Figure 1 shows a schematic design of the power amplifier system. First, the p-polarized laser light from a pre-amplifier is shaped to be 10 mm square by a serrated aperture (SA). Then the laser light propagates through a thin plate polarizer (PL-1) and the polarization is changed to be "s" by passing through combination of a Faraday rotator (FR) and a half wave plate (H-WP). The image at the SA is transferred onto a mirror-2 (M-2) by means of a spherical lens pair (SLP-1). A vacuum tube ended by optical windows is placed between the lens pair in order to avoid the air breaking down around the focus point. Without a high-voltage supply to a pockels cell (PC), the polarization rotates 90 degrees (sÆp) by making a round trip of combination of a quarter wave plate (Q-WP) and PC. Then the image on the M-2 is relayed onto a mirror-3 (M-3) by means of a SLP-2. During high-voltage is supplied to the PC, the p-polarization is kept and then the image relay between the M-2 and the M-3 is repeated. The first zigzag slab amplifier (ZSA-1) is placed between the M-2 and the M-3, therefore 10x10 mm2 seed light is amplified whenever it passes the ZSA-1. The ZSA-1 consists of a silca-phosphate laser glass doped with 1.0 wt% Nd and 8 flash lamps. The dimension of the laser glass is 15 mm in width, 17 in height, and 99 mm in length. Preliminary result of the amplification test shows that the stored energy density of 0.3 J/cm3 is available. Here the laser light is amplified up to ~1 J which is determined by the typical damage threshold of the optical coating, i.e. 2 J/cm2. Then the amplified light is extracted by switching the PC voltage off. The original image of the SA is transferred to nearby a mirror-4 (M-4) by means of the SLP-1. Then the image is vertically extended to be 10x90 mm2 and transferred to the second zigzag slab amplifier (ZSA-2) by means of cylindrical lens pairs (CLP-1, 2). The ZSA-2 consists of the silca-phosphate laser glass doped with 1.0 wt% Nd and 24 flash lamps. The dimension of the laser glass is 15 mm in width, 110 in height, and 205 mm in length. The ZSA-2 provides >10 J by a double-pass amplification. After the ZSA-2, the amplified light is horizontally extended to be 90x90 mm2 and the image is transferred to the compressor by means of another cylindrical lens pair.

This page intentionally blank

Development of the 0.1-Hz Repetition Driver Laser for the TCE X-Ray

CLP-1

47

to compressor

CLP-2

s

M-4 p

FR

p

H-WP s

SA PL-1

ZSA-2 M-1

s s

pre-amp.

s

SLP-1

s

Q-WP M-2 PC

multi-pass amplification M-3

p

PL-2

SLP-2

ZSA-1

Fig. 1. Schematic design of the power amplifier. SA, PL, FR, H- or Q- WP, SLP, M, PC, CLP, and ZSA represent a soft aperture, a polarizer, a Faraday rotator, a half or quarter wave plate, a spherical lens pair, a flat mirror, a pockels cell, a cylindrical lens pair, and a zigzag slab amplifier, respectively. The italic signs mean polarization directions of the laser light.

3 X-ray lasing experiment

Intensity (arb. unit)

By using the laser pulse amplified by the zigzag slab amplifiers, we made x-ray lasing experiment with Ni-like silver. It is noted that the shot repetition was not 0.1 Hz in this experiment. The shot interval was about 1 minute owing to the target alignment and data acquirement. The laser pulse consisted a main pulse of 1.6 ps duration and 1/8 intensity pre-pulse at 1.1 ns before the main pulse. The laser pulse focused on the silver target in the line shape of 20 μm width and 0.8 cm length. The total irradiation energy was 7.3±1.3 J, corresponding to the intensity of 2.5x1015 W/cm2 for the main pulse and 3.2x1014 W/cm2 for the pre-pulse. 400

Ni-like silver

300 200 100 7

13.9 Wavelength (nm)

Fig. 2. Typical spectrum of the Ni-like silver laser.

23

48

Y. Ochi et al.

Typical spectrum obtained by a flat field spectrometer with a backilluminated CCD camera is shown in Fig.2. It can be seen that the Ni-like silver laser line is dominantly strong. In this case, the output energy of the Ni-like silver is about 1 μJ and the beam divergence in the direction perpendicular to the target was ~3 mrad. Intensity (arb. unit)

104 103 102

10 1

0

0.2

0.4

0.6

0.8

1.0

Target length (cm)

Fig. 3. Gain property of the Ni-like silver laser.

Figure 3 shows the gain property of the Ni-like silver for the target length from 0.25 cm to 0.78 cm. The output energy exponentially increases until the target length of 0.4 cm and linearly in following region. The small signal gain and the gain-length product is 41.9 and 16.3, respectively. Consequently, the gain saturated amplification of the Ni-like silver is demonstrated by the laser light amplified by the zigzag slab Nd:glass amplifiers.

4 Summary We have developed new power amplifier of the XRL driver laser by using the zigzag slab Nd:glass. With this power amplifier, the laser light was amplified up to >10 J at repetition rate of 0.1 Hz and compressed to be ~1.6 ps. The gain saturated Ni-like silver laser was demonstrated by using this laser system. We will construct one more laser line for applying the oscillator-amplifier scheme. Then we will provide the fully coherent XRL beam for application researches at 0.1-Hz repetition.

Development of the 0.1-Hz Repetition Driver Laser for the TCE X-Ray

49

References 1. Strickland, D. and Mourou, G. Opt. Commun., 56, 219, 1985. 2. Afanasiev, Y. A. and Shlyaptsev, V.N., Sov. J. Quantum Electron, 19, 1606, 1989. 3. Nickles, P.V., N. Shlyaptsev, V., Kalachnikov, M., Schnürer, M., Will, I., Sandner, W., Phys. Rev. Lett., 78, 2748, 1997. 4. Dunn, J., Osterheld, A.L., Shepherd, R., White, W.E., Shlyaptsev, V.N., and Stewart, R.E., Phys. Rev. Lett., 80, 2825, 1998. 5. Klisnick, A., Zeitou, Ph., Ros, D., Carillon, A., Fourcade, P., Hubert, S., Jamelot, G., Lewis, C.L.S., MacPhee, A., O’Rourcke, R., Keenan, R., Nickles, P.V., Janulewicz, K., Kalachinikov, M., Warwick, J., Chanteloup, J.C., Migus, A., Salmon, E., Saurtret, C., and Zou, J.P., Opt. Soc. Am. B, 17, 1093. 2000. 6. Kawachi, T., Kado, M., Tanaka, M., Sasaki, A., Hasegawa, N., Kilpio, A. V., Namba, S., Nagashima, K., Lu, P., Takahashi, K., Tang, H., Tai, R., Kishimoto, M., Koike, M., Daido, H. and Kato, Y., Phys. Rev. A, 66, 033815, 2002. 7. Nishikino, M., Tanaka, M., Nagashima, K., Kishimoto, M., Kado, M., Kawachi, T., Sukegawa, K., Ochi, Y., Hasegawa, N. and Kato, Y., Phys. Rev. A, 68, 061802(R), 2003.

Modulation of Plasma Density to Achieve Highly Efficient X-Ray Lasers With Good Beam Quality J. Zhao, Q. L. Dong and J. Zhang* Beijing National Laboratory for Condensed Matter Physics, Institute of Physics, Chinese Academy of Sciences, Beijing 100080, China

Summary. A plasma with a modulated density profile at the required electron density for high gain operation could be obtained using a normally incident 300 ps laser pulse followed by a grazing incident 300 ps laser pulse. Sudden heating of the modulated plasma region by another grazing incident 300 fs laser pulse would then yield a highly efficient x-ray laser beam with a defecting angle of 2 mrad and a divergence angle of 3 mrad. Saturation operation of the x-ray laser beam at 32.6 nm could be achieved with a total pump energy of less than 100 mJ.

1 Pump scheme to modulate the plasma density Since the first demonstration of x-ray lasing in Ne-like Se ions [1,2], the pumping process for soft x-ray lasers has gradually been transfered into two steps. The first is to generate a plasma with an abundant population of Ne-like or Ni-like ions. After a certain delay time, a main pulse then irradiates the plasma to generate soft x-ray lasers. To improve the pump efficiency, much effort has been made on both of these two aspects. To solve the problem of refraction due to the steep electron density gradient in the plasma, the pre-pulse (or multi-pulse) technique was introduced with a long delay time [3-5]. The driver energy has also been reduced dramatically to a few joules by short-pulse CPA pumping [6,7]. The pumping efficiency has been further improved by the grazing-incidence pumping (GRIP) scheme, which allows the main pulse to dump more energy in the gain region of the plasma [8-13]. Plasma with a shallowgradient electron density profile and long spatial gain region is crucial to increase the absorption and pumping efficiency of the main pulse. There are several techniques to optimize the plasma to more efficiently generate x-ray lasers. The exploding foil technique was firstly used to produce a flat electron density profile with a scale length of more than 100 µm [1-3]. Refraction was minimized using this technique. However, only

52

J. Zhao, Q. L. Dong and J. Zhang

very high energy pump lasers can burn through the thin foil and a large fraction of the laser energy is lost through radiation and conduction, so that it is practical only on large ICF laser systems with kJ energy output. The multi-pulse technique can generate a large, uniform plasma with a flat stage in the electron profile using three normal incidence pulses [4]. The flat stage has the advantage of little refraction and therefore the generated x-ray laser beam can be more efficiently amplified. However, the electron density in the flat stage (about 7×1020 cm−3 in [4]) is much higher than that required for a specific x-ray lasing (say 2×1020cm−3 for Ne-like Ti ions). Furthermore, the three-pulse configuration is also rather complicated for practical applications.

30 0p s

Delay T ime 5ns

300ps

30 0f s

Gain Region X-ray laser output Target

Fig. 1. The pulse configuration for modulating the plasma density to generate a plasma with a shallow gradient of electron density at the gain region, suitable for pumping Ne-like x-ray lasers.

We propose a new technique to modulate the plasma density to form a flat, uniform electron density distribution at a required electron density (gain region) for a specific x-ray laser, schematically as shown in Fig. 1. The modulation is accomplished by a grazing-incidence 300 ps laser pulse, irradiating an initial plasma (with no Ne-like ions) generated by a normal incidence 300 ps weak laser pulse. Because of a longer path length, the energy of the grazing incidence 300 ps pulse is more efficiently used to ionize the plasma to the ground state of Ne-like (or Ni-like) ions. Each time partition of the 300 ps pulse is refracted and dumps its energy at a predetermined electron density [11]. As the heated electrons expand forward, a flat stage with a shallow gradient electron density is generated. By changing the incidence angle of the second 300 ps pulse and the delay time between the two 300 ps pulses, the electron density at the flat stage can be optimized as the gain region for the desired soft x-ray laser, the uniformity and width of the gain region can be modulated, respectively.

Modulation of Plasma Density to Achieve Highly Efficient X-Ray Lasers

53

The modulated plasma pumped by a grazing incidence main pulse with a 300 fs duration can then generate highly efficient x-ray lasers with a good beam quality. To test the method for realizing the modulation of the plasma density, we consider, as an example, a Ne-like Ti x-ray laser at 32.6 nm. A modified 1D Lagrangian hydrodynamic code MED103 coupled with an atomic physics data package is used to obtain the time evolution of the laser-plasma interaction and the gain coefficients. In addition to the 2D ray-tracing code, which calculates the variation of the output intensity versus the plasma length, a 3D ray-tracing code is developed to calculate the near-field and far-field distributions of the output intensity of the x-ray laser beam.

2 Density modulation for the Ne-like Ti x-ray laser The wavelength of the pump laser is 800 nm. The pump geometry is the standard line focus on a 100 µm thick slab target. The pump laser pulses have a Gaussian profile. The purpose to modulate the plasma density before the main pump laser pulse is to generate a wide uniform flat electron density distribution at 2.0×1020 cm−3, which is the gain region for the Ne-like Ti x-ray laser. The duration of the two long pump pulses are fixed to 300 ps.

(a)

(b)

Fig. 2. The electron density profiles at the peak abundance time of Ne-like ions under (a) different incident angles from the surface of the target when the delay time is 5 ns (b) different intervals when the incident angle is 20◦.

54

J. Zhao, Q. L. Dong and J. Zhang

A 300 ps weak pulse of 1×1011W/cm2 is normally focused on the slab target to generate an initial plasma (with no Ne-like ions). By changing the grazing incidence angle of the second 300 ps pulse and the interval between the two pulses, the profile of plasma density can modulated. The peak intensity of the second pulse is simultaneously optimized, to ionize the plasma to the Ne-like state without over-ionization. Fig. 2 gives the electron density profiles at the peak abundance time of Ne-like ions under (a) different incident angles to the surface of the target when the delay time is 5 ns (b) different intervals when the incident angle is 20◦. The smaller incidence angle of the second pulse is used, the more uniform electron density in the plasma is generated. The second pulse with an incidence angle of 20◦ has the turning point at about 2.0×1020 cm−3, which is the required electron density for the Ne-like Ti x-ray laser, gives the widest gain region about 70 µm. The longer time the plasma expands before the second pulse, the wider the flat stage of electron density distribution is, as shown in Fig. 2 (b). But if the interval is too long, for example 7 ns, the electron density at the flat stage will be further reduced.

Moving

1

9 4

4

9

1

4 1

9

Fig. 3. The time evolution of electron density profiles during the second pulse from 5200 ps to 5550 ps. The peak time is 5360 ps. The distribution of the ground state Ne-like Ti ions at the end time of the second pulse is also plotted.

Modulation of Plasma Density to Achieve Highly Efficient X-Ray Lasers

55

In our simulation, the optimum plasma density profile is produced by the second pulse, which arrives 5 ns later with a peak intensity of 2.8×1011 W/cm2 and a grazing incidence angle of 20◦. To understand the mechanism of the modulation of the electron density profile by the grazing incidence 300 ps pulse, the time evolution of the electron density profile is given in Fig. 3. The surface of the target is located at 100 µm. The second pulse is divided into many time partitions. Each partition of the second pulse is reflected and dumps most of its energy at the turning point of ne=nc×sin2(θ), where nc=1.1×1021/λ2 (cm−3) is the critical density for the wavelength of the pump laser [11]. The electron density of 2.0×1020cm−3 corresponds to an incidence angle of 20◦. Heated by the earlier partitions of the second pulse, electrons move away from the target surface. Thus, at the intensity ascending side of the 300 ps pulse, the turning point of the laser beams moves fast forward along with the electrons, a deep valley of electron density is generated. At the descending side, another shallow valley ∼20 µm (which is important for generating good beam quality of xray laser) is produced.

Fig. 4. The time evolution of electron temperature in the plasma. Till the end of the second pulse the maximum temperature is reached, which represents the highest fraction of Ne-like ions.

Fig. 4 shows the time evolution of the electron temperature, which indicates the absorption of the second pulse in the plasma. The peak temperature also moves forward along the axial distance. At the end of the second pulse, the maximum temperature is reached corresponding to the highest fraction of Ne-like Ti ions in Fig. 3. Because of the increased path

56

J. Zhao, Q. L. Dong and J. Zhang

5900

5900

5800

5800

5700

0.7

5600 5500 0.5

5400 5300 180

Time ( ps )

Time ( ps )

length and energy absorption of the grazing incidence 300 ps pulse, a plasma with a high fraction of Ne-like Ti ions is efficiently produced.

230

280

Distance ( μm )

5700

0.7

5600 5500 0.5

5400 5300 160 180 200 220 240 260

330 0.3

(a)

Distance ( μm )

0.3

(b)

Fig. 5. The spatiotemporal distribution of the ground state Ne-like ions fraction, generated by (a) the optimized grazing incident 300 ps pulse with an incident angle of 20◦ and (b) the normal incident 300 ps pulse. The fraction in the plot is from 70 % (black) to 30 % (white) by steps of 20 %.

We also investigate the distribution of the ground state Ne-like ions generated by the optimized 300 ps pulse with a peak intensity of 2.8×1011 W/cm2 and a grazing incidence angle of 20◦ in Fig. 5 (a), compared to the case where the second 300 ps pulse irradiates into the initial plasma at a normal incidence angle, shown in Fig. 5 (b). For the normal incident 300 ps pulse, most energy is dumped at the critical surface of the plasma, a peak intensity of 5.0×1011 cm−3 is required to get the same population of Ne-like ions as the grazing incident 300 ps pulse. The grazing incident case has shown a wider distribution of Ne-like ions with a less pump energy.

3 The main 300 fs pump laser pulse As the consequence of the modulation, a plasma with a fairly uniform, long-extent (∼80 µm) gain region at the electron density of 1.0×1020 cm−3 ∼ 2.0×1020 cm−3 is generated, with the same local position as the high fraction of Ne-like Ti ions. It is one of the advantages for the grazing incidence main pump pulse to dump its energy at the gain region and reach a high population inversion, simultaneously. This plasma with a shallow electron density gradient has advantages not only for the propagation and amplification of the x-ray beam, but also for the propagation and absorption of the main pump pulse. An efficient x-ray laser beam with a

Modulation of Plasma Density to Achieve Highly Efficient X-Ray Lasers

1

57

2

Fig. 6. The electron density and gain profiles at 1 ps after the onset of the main pulse. The first deep valley of electron density is generated by the ascending side of the second pulse, while the second shallow valley by the descending side.

good beam quality can be expected. As the last step of the pumping of the Ne-like Ti x-ray laser, the plasma is suddenly heated up by the following main CPA laser pulse with a 300 fs duration, shown in Fig. 1. The main pulse irradiates the plasma at the same incidence angle (turning point) as the 300 ps pulse. This configuration ensures that most energy is absorbed in the region with a high fraction of Ne-like ions. The peak intensity of the main pump pulse is optimized to be 3.0×1014 W/cm2. The optimized delay time between the main pulse and the second pulse is 240 ps. The spatial distribution of the gain coefficient is shown in Fig. 6. Maximum gain coefficient happens at the turning point of the main pulse, which is in the second shallow valley of electron density distribution, generated by the descending side of the second pulse. This valley of ∼20 µm can be used as a channel for the propagation of the main pump pulse (high pump efficiency) and x-ray laser beams (high amplification efficiency and good beam quality) with no refraction.

4 Characteristics of the output of the x-ray laser beam Using 2D and 3D ray-tracing codes developed by ourselves, the characteristics of the output profile of the Ne-like Ti x-ray laser beam can be investigated in detail. In the 3D ray-tracing code, the variation of the plasma density, gain coefficient in the pump laser (radial) direction is given by MED103, whereas a Gaussian profile is assumed for the transverse direction. The time dependence of gain coefficients is not taken into account, because of the inherent travelling wave pump of the grazing incidence pumping scheme. The refraction and saturation effects in both

58

J. Zhao, Q. L. Dong and J. Zhang

8

1.6x10

4

1.4x10

0

1.2x10

-4

1.0x10

-8

8.0x10

205

210

215

220

Radial ( μm)

(a)

225

6.0x10 4.0x10

11 11 11 11 10 10

6

Vertical ( mrad )

Transverse ( μm)

Fig. 7. Ray-tracing results of the output of x-ray laser intensity versus the plasma length at incident angles of 18◦, 20◦ and 22◦.

4 2 0 -2 -4 -6 -6 -4 -2 0

10

2

4

6

Horizontal ( mrad )

11 1.6x10 11 1.4x10 11 1.2x10 11 1.0x10 10 8.0x10 10 6.0x10 10 4.0x10

(b)

Fig. 8. The patterns of (a) near field and (b) far field distributions of the output intensity of the Ne-like Ti soft x-ray laser beam at 32.6 nm, generated by a 0.15 cm long target. The intensity is depicted by a grey scale with an arbitrary unit.

radial and transverse directions are considered. About 1×107 rays are traced to calculate the output profiles of the x-ray laser beam. Figure 7 gives the 2D ray-tracing results of the output intensity of the x-ray laser beam versus the plasma length at incidence angles of 18◦, 20◦ and 22◦. Optimized output of the intensity is obtained with an incidence angle of 20◦. Saturation is achieved at a plasma length of 0.15 cm. The focal width is assumed to be 30 µm. The near and far field distributions of the output intensity at 32.6 nm, generated by a 0.15 cm long target, are calculated by the 3D ray-tracing code, as shown in Fig. 8. The target surface is located at 100 µm. The active region is ∼20 µm wide at the radial direction. Good

Modulation of Plasma Density to Achieve Highly Efficient X-Ray Lasers

59

output beam quality of the x-ray laser beam is shown with a deflecting angle of 2 mrad. Horizontal divergence (FWHM) is ∼3 mrad, vertical divergence (FWHM) is about 10 mrad. The pump energy of the three pump laser pulses are 13.5, 37.8 and 40.5 mJ, respectively. The total energy to pump the Ne-like Ti x-ray laser beam is only about 92 mJ, which is less than half of the pump energy with an optimized double pulse pump scheme used in Ref. [14].

5 Conclusion By modulating the plasma density with a 300 ps normal incidence pulse followed by a 300 ps grazing incidence second pulse, an optimized plasma with a large, uniform gain region for x-ray lasers, is generated. Suddenly heated by a 300 fs grazing incidence main pulse, an efficient x-ray laser with a good beam quality can be reached, because little refraction occurs during the propagation of the main pump pulse and x-ray beams. The 3D ray-tracing calculation shows a horizontal divergence (FWHM) of 4mrad and a deflecting angle of 2 mrad. For the Ne-like Ti x-ray laser at 32.6 nm, saturation is achieved with a total pump energy only about 92 mJ.

Acknowledgements This work is jointly supported by the National Natural Science Foundation of China (under Grant Nos. 10474137, 10574161, 10510490) and the National Hi-tech ICF Program.

References 1

2 3

4 5

D. L. Matthews, P. L. Hagelstein, M. D. Rosen, M. J. Eckart, N. M. Ceglio, A. U. Hazi, H. Medecki, B. J. MacGowan, J. E. Trebes, B. L. Whitten, et al., Phys. Rev. Lett 54, 110 (1985). M. D. Rosen, Phys. Rev. Lett 54, 106 (1985). L. B. Da Silva, B. J. MacGowan, D. L. Matthews, M. D. Rosen, H. A. Baldis, G. D. Enright, B. LaFontaine, and D. M. Villeneuve, Proceedings of the SPIE The International Society for Optical Engineering 1229, 128 (1990). J. Nilsen and J. C. Moreno, Phys. Rev. Lett 75, 2453 (1995). J. Zhang, A. G. MacPhee, J. Lin, E. Wolfrum, R. Smith, C. Danson, M. H. Key, C. L. S. Lewis, D. Neely, J. Nilsen, et al., Science 276, 1097 (1997).

60 6

7

8 9

10 11 12 13 14

J. Zhao, Q. L. Dong and J. Zhang P. J. Warwick, C. L. S. Lewis, M. P. Kalachnikov, P. V. Nickles, M. Schnurer, A. Behjat, A. Demir, G. J. Tallents, D. Neely, and E. Wolfrum, J. Opt. Soc. Am. B 15, 1808 (1998). M. P. Kalachnikov, P. V. Nickles, M. Schnurer, W. Sandner, V. N. Shlyaptsev, C. Danson, D. Neely, E. Wolfrum, J. Zhang, A. Behjat, et al., Phys. Rev. A 57, 4778 (1998). D. Alessi, B. M. Luther, Y. Wang, M. A. Larotonda, M. Berrill, and J. J. Rocca, Opt. Express 13, 2093 (2005). J. J. Rocca, Y. Wang, M. A. Larotonda, B. M. Luther, D. Alessi, M. Berrill, A. Weith, M. C. Marconi, C. S. Menoni, and V. N. Shlyaptsev, Soft X-Ray Lasers and Applications VI 5919, 591901 (pages 12) (2005). J. Dunn, R. Keenan, and V. N. Shlyaptsev, Soft X-Ray Lasers and Applications VI 5919, 591905 (pages 8) (2005). R. Keenan, J. Dunn, P. K. Patel, D. F. Price, R. F. Smith, and V. N. Shlyaptsev, Phys. Rev. Lett 94, 103901 (2005). J. Tummler, K. A. Janulewicz, G. Priebe, and P. V. Nickles, Phys. Rev. E 72, 037401 (2005). B. M. Luther, Y. Wang, M. A. Larotonda, D. Alessi, M. Berrill, M. C. Marconi, J. J. Rocca, and V. N. Shlyaptsev, Opt. Lett 30, 165 (2005). J. Zhao, Q. L. Dong, F. Yan, and J. Zhang, Phys. Rev. A 73, 033816 (2006).

Origin of Microstructures in Picosecond X-Ray Laser Beams O. Guilbaud1, A. Klisnick1, K. Cassou1, S. Kazamias1, D. Ros1, G. Jamelot1, D. Joyeux2 and D. Phalippou2 1

LIXAM, Laboratoire d'Interaction du rayonnement X Avec la Matière, Univ Paris-Sud, CNRS, UMR n° 8624, Orsay, F-91405, France. 2

LCFIO, Laboratoire Charles Fabry de l'Institut d'Optique, CNRS, UMR n° 8501, Orsay, F-91403 France.

Summary. Beam sections of transient collisional excitation (TCE) and optical field ionization (OFI) soft X-ray lasers exhibit highly contrasted intensity modulations forming complex patterns. We present model calculations that reproduce these experimental observations both qualitatively and quantitatively. The low spatial coherence, high longitudinal coherence and short pulse duration of the source are shown to control the visibility of the beam structures. Besides, simple relationships relating the spatial and temporal characteristics of the emitting source to the main features of the beam small-scale structures can be derived.

1 Introduction Important progress has been achieved in reducing the pumping energy of plasma based soft x-ray lasers (SXRL). Optical Field Ionization (OFI) [0] and Transient Collisional Excitation (TCE)[0] x-ray lasers are efficient enough to be pumped by a 1J Ti:Sapphire laser facility at 10 Hz. Compact devices that can be operated at high repetition rates are know available for a wide range of applications. Both schemes, OFI and TCE (including recent upgrades as Grazing Incidence Pumping methods [0]), have several common properties. In both cases, very high gain values (10 – 100 cm-1) are obtained during 10 ps to 30 ps after the pumping laser pulse intensity peak, leading to the emission of a short (a few ps) and intense soft x-ray pulse in the range of 40 nm to 6 nm. Because they are based on a single path amplification of spontaneous emission, their spatial coherence is low. On the other hand, because of the moderate ion temperature reached

62

O. Guilbaud et al.

during the peak gain, the Doppler broadening is small leading to high values of the temporal coherence. Many works have indeed pointed out the proximity between the coherence time and the pulse duration values of these SXRL, and despite their ASE nature, the emitted XUV pulses are close to the Fourier transform limit [0][0].

Fig. 1 Experimental far field images of Ni-like Silver TCE soft x-ray laser. (a) and (b) were obtained in the experiment described in reference [0]. (c) was obtained in conditions described in reference [0].

These schemes were also observed to generate much less uniform beams than QSS SXRL (see for example [0]). The far-field images of TCE or OFI x-ray lasers exhibit complex patterns with small-scale and highlycontrasted structures. Figure 1 presents some examples of experimental far-field images of a Ni-like silver TCE x-ray laser [0] [0]. In some cases, strongly distorted interference fringes were observed as shown in figure 1c. In most cases however the observed far-field images have the appearance of a speckle pattern with a typical angular size of 0.1 to 0.5 mrad and randomly distributed over the beam section as illustrated in figure 1a and 1b. From classical theory we know that speckle patterns result from the diffraction of a spatially coherent beam by a random phase or transmission medium. This interpretation cannot be directly applied to SXRL. Although the amplifying plasma is probably not homogeneous and could indeed include small-scale dephasing structures, the spatial coherence of the SXRL beam in the plasma is far too low to induce any speckle pattern in the far field. We have recently developed a model explaining the formation of the two types of observed far-field structures [0]: fringes and speckles. We have showed that these structures are apparent as a result of the specific optical properties of TCE SXRL which were recently experimentally

Origin of Microstructures in Picosecond X-Ray Laser Beams

63

investigated: a low spatial coherence, a high longitudinal coherence associated with a short pulse duration. After a brief summary of the model, we will discuss the predictions of our model regarding the beam structure of TCE and OFI SXRL. Finally we will present simple formulas that relate the main features of the beam small-scale structures to spatial and temporal properties of the emitting source.

2 Model description 2.1 General We briefly describe in this section the main characteristics of our model. The notations used are presented in figure 2. The amplifying plasma is modeled by a cylindrical rod of length L along the axis z. The gain coefficient is assumed to be independent of z. The refraction of the SXRL beam induced by electron density gradients is not included. Hence, our model can only be used to investigate the spatial structure of the beam but is not appropriate to predict the overall shape or the deflection angle. Finally, the electromagnetic field associated to the SXRL radiation is described in the framework of the scalar approximation.

Fig. 2 (a) Geometry of the model. (b) Soft x-ray laser field modulus and phase on the exit plane. (c) far-field intensity distribution.

64

O. Guilbaud et al.

The time-integrated intensity I(X,Y) detected at a distance D from the source can be deduced from the field instantaneous complex amplitude A(x, y,t) , on the exit aperture, by the Huygens-Fresnel principle: T /2

I (X ,Y ) =

∫

dt ∫∫ A( x, y, t ) e iϕ ( x , y ,t ) e

−i

2π ( ( x − X ) + ( y −Y )) λ D

dxdy (1)

−T / 2

where T is the integration time of the detector, much greater than the pulse duration τ XUV . To account for the ASE origin of the emitted field, we consider that the phase front ϕ ( x, y, t ) of the field at the exit aperture, and at a given time, is composed of small cells of uniform phase, randomly distributed as shown in figure 2b. The typical size of each cell is given by the spatial coherence length in the exit plane of the laser, and can have distinct values along x and y. An order of magnitude of the shape and size of the spatial coherence length can be deduced from the profile of spontaneous emission at the opposite end of the plasma rod, using the Van Cittert-Zernicke theorem. The field on the exit aperture is indeed the superposition of random phase fields initially emitted by the lasing ions through spontaneous decay, and amplified with propagation along the plasma rod. Because of the exponential growth of intensity with plasma length, only emitters at the opposite end of the plasma rod will significantly contribute to the field emitted at the exit aperture. To simplify the presentation the modulus of the field amplitude will be taken constant over the exit aperture (Figure 2b). The instantaneous far-field intensity pattern I(X,Y,t) at the detection plane will result from the interferences between the field amplitudes emitted by the numerous small regions of uniform phase at the exit aperture. It will thus exhibit properties similar to a speckle pattern that would have been produced by the diffraction of a spatially coherent beam through a random phase aperture analog to the one described by ϕ ( x, y , t ) . The experimental images being time-integrated, we shall now consider the evolution of the SXRL field over the whole pulse duration. To simplify 2

the presentation, the instantaneous intensity A(x, y,t) at the exit aperture is taken constant over the pulse duration. The phase front of the field ϕ ( x, y, t ) can be considered constant over a time shorter than τ C , the coherence time of the field. However, because of the chaotic nature of the

Origin of Microstructures in Picosecond X-Ray Laser Beams

65

ASE field, ϕ ( x, y, t ) randomly change over time scale greater than τ C . In summary, on a time interval of the order of τ C , the phase ϕ ( x, y, t ) and hence the far field instantaneous intensity I(X,Y,t) can be taken constant, but will change many times from one random distribution to a completely different one over the pulse duration. If we define N as the ratio between the pulse duration τ XUV and the coherence time τ C , then the integrated far-field intensity distribution can be approximated by an incoherent superposition of N different and independent instantaneous intensity farfield patterns.

3 Predictions of the model 3.1 Size of the far-field micro-structures In the previous section 2.2, we have emphasised the analogy existing between the instantaneous far-field intensity and classical speckles patterns. From this analogy we can deduce that the average shape and size of the micro-structures observed in the far-field are given by the diffraction pattern of the limiting diaphragm, namely here the exit aperture. If the emitting pupil of the SXRL source is elliptical, with ax and ay being its dimensions along the x and y axis respectively, an order of magnitude of the angular dimensions β X and β Y of a microstructure is thus given by:

β X ≈ λ a X and β Y ≈ λ aY .

3.2 Contrast of the far-field micro-structures A simple relation determining the contrast of the speckle structures can be derived from the temporal structure of the field considered in section 2.3. For N=1, The far-field integrated intensity I(X,Y) is a pure speckle pattern. It can be shown from classical theory of speckles that its spatial average

I

1

is equal to the spatial standard deviation σ 1 =

contrast of the speckles V = σ 1

I

1

I2

1

− I

2 1

. The

is thus equal to 1. For larger values

of N, the contrast V decreases as the inverse square root of N:

V =σN

I

N

= Nσ1 ( N I 1) = 1

N.

(2)

66

O. Guilbaud et al.

The contrast of the far-field structures is inversely proportional to the squared root of N. In the framework of our model, this feature explains why contrasted structures are observed in TCE or OFI x-ray laser beams (N close to unity), whereas they are smoothed out in QSS lasers, which have a longer pulse duration (N>30).

4 Applications to TCE soft x-ray lasers We have applied the model to three particular cases where the near field intensity distribution is supposed to be a) almost circular, b) strongly elliptic, c) composed of two sources. In the three cases, we have taken N=3, in order to be as close as possible of the Ni-like silver TCE x-ray laser properties measured in [0]: τ XUV ~ 5 ps and τ C ~2 ps.

Fig. 3. Simulated far-field patterns. N=3. Near-field intensity distribution is shown in the insert of each image.

The simulated far-field patterns are displayed in figure 3. The corresponding near field intensity distribution is shown in the insert of each figure. Figure 3a presents a contrasted and randomly distributed pattern of circular microstructures. This pattern is similar to the intensity distribution observed for some TCE x-ray lasers shots (see for example figure 1a). It presents also some dramatic similarity with some OFI x-ray laser footprint [0]. The more elongated near-field distribution considered in the insert of figure 3b, reproduces more precisely the general exit aperture shape of a TCE x-ray lasers. The far-field distribution resulting from this hypothesis presents evident similarity with figure 3a except in the general shape of the micro-structures. This comparison illustrates the relation derived in section 3.1 establishing a link between the shape of the exit aperture and the shape

Origin of Microstructures in Picosecond X-Ray Laser Beams

67

of the microstructure. Moreover the kind of structures obtained in figure 3b is often observed in TCE x-ray laser footprint as presented in figure 1b. Finally, we have performed simulations assuming a near-field structure composed of two separated sources. This type of near field shape has been experimentally observed in TCE x-ray lasers by Tanaka et al. [0] and by Guilbaud et al. [0] The numerical result is displayed in figure 2c. The beam profile is covered by distorted fringes of constant interfringe. This type of pattern is similar to the experimental images observed by Klisnick et al. [0] on a Ni-like silver TCE x-ray laser and displayed in figure 1a. Note that in the simulations, we did not assume any particular phase relation between the two sources, i.e, no mutual coherence. We have also performed different simulations for different values of N. We observed that for high N values, the structure vanishes, as expected from equation (2).

5 Conclusions In summary, we have presented model calculations that account qualitatively and quantitatively for the complex patterns observed in the beam cross-section of transient and OFI X-ray lasers. We show that the presence of the highly contrasted structures, speckles or fringes, is controlled by the specific optical properties of these sources: a low spatial coherence, a high longitudinal coherence, and above all, a short pulse duration. Our model reveals simple relationships between the main features of the far-field structures and the source characteristics. The contrast of the structures is inversely proportional to the square root of N, the ratio of the pulse duration to the coherence time. Moreover, the size and shape of the speckles is related by diffraction to the spatial characteristics of the source ones through a Fourier transform. This hence means that the size and shape of the XRL exit aperture on the one hand, and the proximity of the pulse to the Fourier transform limit on the other hand, can be extracted from the experimental far-field images of the beam. Finally, our model predicts that small-scale structures will affect the beam even though the amplifying plasma is perfectly homogeneous. As a result, a better uniformity of the beam, which is of great importance for future applications, can only be obtained by improving the spatial coherence at the source plane, e.g. by the injection of high order harmonics at the entrance of the plasma amplifier rod [0,0].

68

O. Guilbaud et al.

Acknowledgements We would like to thank Gilles Maynard (LPGP), Stéphane Sebban and Islam Bettaibi (LOA) for fruitful discussions on this work.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

Lemoff B.E. et al., Phys. Rev. Lett. 74, 1995, 1995 Nickles P.V. et al., Phys. Rev. Lett. 78, 2743, 1997 Keenan R. et al, Phys. Rev. Lett. 94, 103901, 2005 Smith R.F. et al., Opt. Lett. 28, 1, 2003 Guilbaud O. et al., Eur. Phys. J. D 40, 125, 2006 Rus B. et al., Phys. Rev. A 55, 3858, 1997 Klisnick A. et al., in soft x-ray laser and application IV, Proc. SPIE Vol 4505, 75, 2001 Guilbaud O. et al., Euro. Phys. Lett. 74, 823, 2006 Bettaibi I. et al., private communication Tanaka M. et al., in Proceedings of the international conference on VUV radiation physics, Surface Rev. Lett. 9, 641, 2002 Guilbaud O. et al., in soft X-ray laser and applications V, Proc. of SPIE Vol. 5197, 17, , 2003 Zeitoun P. et al., Nature 431, 426, 2004 Wang Y. et al., Phys. Rev. Lett., In press, 2006

Focusing System for a Non-Normal Incidence Pumped, Sub-10 nm TCE XRL D. Ursescua, C. Bischoffa,b, T. Kühla,c and B. Zielbauera,c,d a

Gesellschaft für Schwerionenforschung mbH, Darmstadt, Germany; Fachhochschule Friedberg, Friedberg, Germany; cMainz University, Mainz, Germany; dMax Born Institute, Berlin, Germany b

Summary. An extended ray-tracing study of the main pumping pulse (MP) focusing optics was made in order to develop a transient collisionally excited (TCE) x-ray laser (XRL) set-up for sub 10 nm wavelength by using non-normal incidence pumping. The study takes into account limitations determined by the large beam dimensions of the high energy MP. A setup using two off-the-shelf spherical mirrors is proposed, corresponding to a possible realization at the PHELIX laser facility.

1 Motivation Development of the x-ray lasers requires the generation of a well defined line-focus with a high aspect ratio. With the development of TCE systems, one more parameter became essential, the travelling wave speed (TW) of the focusing system. Several methods were proposed, including "misalignment" of the optical pulse compressor, echelons, and combinations of spherical and parabolic mirrors. A simple set-up using the aberrations (astigmatism and wave-front tilt) from a single spherical mirror can be used to generate an appropriate set-up (e.g. [1]) as first demonstrated experimentally in [2]. With the GRIP scheme [3], the incidence angle of the pulse on the target comes into play. This parameter produces important changes in the behaviour of the XRL [2-7]. The basic analytical relations describing the properties of this setup are presented in section 2. However, with the development of XRL toward shorter wavelengths, such a GRIP set-up is no more useful for two reasons: on one side the pumping beams are more energetic and, as a consequence of the damage threshold of the optics, the beams have larger radii thus generating significantly longer line foci. On the other hand the incidence angle of the pumping pulses are no more favourable in the GRIP case for sub 10 nm XRL,

This page intentionally blank

70

D. Ursescu et al.

smaller incidence angles for the pumping pulses on target being more appropriate [6,7]. As a consequence, other focusing systems have to be developed taking into account the TW, the incidence angle of the pulse on the target and also very important, the price of the optical components. This paper reports in section 3 a possible choice for the focusing system using low-price optics, as needed for future experiments at PHELIX laser facility [8].

2 Line focus with one spherical mirror Analytical relations can be obtained for line focus length and travelling wave speed in the case of a single focusing spherical mirror. The focusing system is represented in the fig. 1. A parallel laser beam is coming along lines L1M1 and L2M2 to the spherical mirror with the centre in O and radius OA = OM1 = OM2 = R = 2f. Using the notations in the picture one can write the line focus length:

Fig. 1. Representation of the main pulse focusing system used for deriving the analytical formula of the travelling wave and for the line focus length

and the average TW speed as:

The local TW can then be obtained as a limit of the above relation:

Focusing System for a Non-Normal Incidence Pumped, Sub-10 nm TCE XRL 71

3 Line focus generation using two spherical mirrors As it is clear from the relations above, one cannot control the line focus length and the incidence angle in an independent way by using only one spherical mirror for a given beam diameter. One way to do that is to use one off-axis parabolic mirror and one spherical mirror as reported in [9]. However, this approach is expensive when the beam diameter of the optics in use increases. This happens when one pumps shorter wavelengths XRL using picosecond pulses as it is prepared now at PHELIX facility. The alternative we analyse here is to use two off-the-shelf spherical mirrors, introducing in this way the additional degrees of freedom needed. The analytical approach becomes in this case difficult and an ray tracing program (Optica package for Mathematica) was used to analyse the optimal setup, taking into account the geometrical constraints. The pump laser beam has a beam diameter of 200 up to 250 mm and a target chamber with a diameter of 1600 mm. The optimal angle searched is 45°, as predicted in [6] for Sm. To check all possible geometrical setups, a simulation with Mathematica was programmed to scan the complete parameter space and to find the according parameters as a function of the angles of the mirrors, their focal lengths, diameters and distance (see fig. 2).

Fig. 2. Set-up with two spherical mirrors

The results obtained for a variation of the angle of the second spherical mirror are presented in fig. 3 for values between 0° and 60°. One can

72

D. Ursescu et al.

identify two regions of interest, separated by an interval where the parameters are producing clipping of the focused beam on the first mirror. The TW obtained in this way is below 2 c, and the incidence angle on the target can be varied between 20° and 60°. The parameter which has to be mastered in this case is the line-focus length which varies from 5 mm to 55 mm.

Fig. 3. Variation of the second angle, with first angle (between first mirror and first reflection) of 190 degree.

After optimization of the incidence angles on the mirrors, of the distance between the mirrors and of the focal lengths an optimal focusing system was obtained with the following parameters: focal length1=2054 mm, first angle=193°, distance=540 mm, focal length2=1524 mm, second angle 2=36.5°. In this case the TW =1.47 c, length of line focus=10.44 mm, incidence angle on target=42. 9° and the line focus width= 40 microns. In conclusion there were presented here focusing systems needed for obtaining TW at a given angle using one and two spherical mirrors. Optimal parameters for a sub-10 nm XRL are in reach with the two spherical mirrors system. The method to generate a line-focus with a given TW and incidence angle proposed in this paper is general and can be used to obtain other combinations of parameters as requested by the experiments.

Focusing System for a Non-Normal Incidence Pumped, Sub-10 nm TCE XRL 73

References 1. Neumayer, P.; Alvarez, J.; Mos, B.B.; Borneis, S.; Brück, K.; Gaul, E.W.; Häfner, C.; Janulewicz, K.A.; Kuehl, T.; Marx, D.; Reinhard, I.; Tomaselli, M.; Nickles, P.V.; Sandner, W.; Seelig, W.: 'X-ray laser spectroscopy on lithium-like ions', Soft X-Ray Lasers and Applications IV, SPIE, 4505, 236242, 2001 2. Neumayer, P.; Seelig, W.; Cassou, K.; Klisnick, A.; Ros, D.; Ursescu, D.; Kuehl, T.; Borneis, S.; Gaul, E.; Geithner, W.; Haefner, C. & Wiewior, P. 'Transient collisionally excited X-ray laser in nickel-like zirconium pumped with the PHELIX laser facility', Appl. Phys. B, 78, 957-959, 2004 3. Keenan, R.; Dunn, J.; Patel, P.K.; Price, D.F.; Smith, R.F. & Shlyaptsev, V.N.: 'High-Repetition-Rate Grazing-Incidence Pumped X-Ray Laser Operating at 18.9 nm', Phys. Rev. Lett., 94, 103901, 2005 4. Kühl, Th.; Ursescu, D.; Bagnoud, V.; Javorkova, D.; Rosmej, O.; Zimmer, D.; Ros, D.; Cassou, K.; Kazamias, S.; Klisnick, A.; Zielbauer, B.; Janulewicz, K.; Nickles, P.; Pert, G.; Neumayer, P.; Dunn, J. : 'A Non-Normal Incidence Pumped Ni-like Zr XRL for Spectroscopy of Li-like Heavy Ions at GSI/FAIR', these proceedings 5. Kazamias, S.; Cassou, K.; Ros, D.; Plé, F.; Jamelot, G.; Klisnick, A.; Lundh, O.; Lindau, F.; Persson, A.; Wahlström, C. G.; de Rossi, S.; Joyeux, D.; Zielbauer, B.; Ursescu, D.; Kühl, Th.: 'A 10 Hz, 3 µJ transient X-ray laser pumped in grazing incidence geometry', these proceedings 6. Pert, G. J.: 'Optimizing the performance of nickel-like collisionally pumped xray lasers', Phys. Rev. A, 73, 033809, 2006 7. Ursescu , D.; Zimmer, D.; Kühl, Th.; Zielbauer, B.; Pert, G. J.: 'Gain generation in the critical density region of a TCE XRL', these proceedings 8. Neumayer, P.; Bock, R.; Borneis, S.; Brambrink, E.; Brand, H.; Caird, J.; Campbell, E.; Gaul, E.; Götte, S.; Häfner, C.; Hahn, T.; Heuck, H.; Hoffmann, D.; Javorkova, D.; Kluge, H.; Kühl, T.; Kunzer, S.; Merz, T.; Onkels, E.; Perry, M.; Reemts, D.; Roth, M.; Samek, S.; Schaumann, G.; Schrader, F.; Seelig, W.; Tauschwitz, A.; Thiel, R.; Ursescu, D.; Wiewior, P.; Wittrock, U. & Zielbauer, B.: 'Status of PHELIX laser and first experiments', Laser and Particle Beams, 23, 385-389, 2005 9. King, R.E.; Pert, G. J.; McCabe, S. P.; Simms, P.A.; MacPhee, A.G.; Lewis, C.L.S.; Keenan, R.; O'Rourke, R.M.N.; Tallents, G.J.; Pestehe, S.J.; Strati, F.; Neely, D. & Allott, R.: 'Saturated x-ray lasers at 196 and 73 Ǻ pumped by a picosecond travelling-wave excitation', Phys. Rev. A, 64, 053810, 2001

X-Ray Imaging of the Heating Zone of Non-Normal Incidence Pumped XRL Plasma B. Zielbauer1,2,3, D. Ursescu2,3, T. Kühl2,3, S. Kazamias4, K. Cassou4, D.Ros4, A. Klisnick4, F. Plé4, S. de Rossi5, D. Joyeux5, O. Lundh6, F. Lindau6, A. Persson6 and C.-G. Wahlström6 1Max Born Insitut, Berlin, Germany; 2Gesellschaft für Schwerionenforschung (GSI), Darmstadt, Germany; 3Universität Mainz, Mainz, Germany; 4Université Paris XI, Paris, France; 5Laboratiore Charles Fabry (IOTA), France; 6Lund Institute of Technology, Lund, Sweden

Summary. Soft x-ray emission, above 600 eV, from a grazing incidence pumped Ni-like Mo x-ray laser (GRIP-XRL) [1] plasma was investigated. Using a pinhole camera looking along the target surface, perpendicular to the direction of the XRL emission, spatially-resolved information was obtained with a resolution which was limited only by the pinhole size of 10 µm. The relative distance from the target surface to the plasma zone heated by the picosecond pulse was investigated for different GRIP angles, energy ratios and delays between the plasma producing and the plasma heating pulses.

1 Motivation The optimization of the absorption of the compressed pump pulse within the plasma created by the long pulse is an important way to improve the performance of a plasma-based x-ray laser. One significant recent development following this idea is the so-called GRIP scheme [2]. It yields control of the depth in which the main pulse is absorbed within the plasma by varying its incidence angle in a strongly non-normal (“grazing”) regime. To learn more about the influence of various setup parameters on the position of the absorption region, a pinhole-type camera system with suitable parameters was developed and used as part of an extensive x-ray laser campaign. Under the assumption that the position of the strongest absorption is linked to the area of the strongest soft x-ray emission within the plasma, the changes of this position can be studied.

76

B. Zielbauer et al.

2 Setup The pinhole camera was positioned directly above the target, to achieve the highest sensitivity for the distance between the target and the plasma, see figure 1. The keV emission from the plasma was coming upwards along the target surface through a 1.4 mm opening in the target holder. Since the entire target and its holder could be moved up and down by 40 mm to change the target surface, also the pinhole height had to be changeable under vacuum in order to make use of the highest possible magnification. To facilitate the alignment under vacuum, the pinhole was located on a changer which in turn could be moved in the horizontal plane to compensate for movements resulting from changes of the pinhole height. For the position measurements, a pinhole diameter of 10 µm was used together with a filter consisting of 1 µm polypropylene and 200 nm aluminum positioned about one millimeter above the pinhole, blocking radiation below about 600 eV. The magnification of the imaging system was determined by the distances plasma/pinhole and pinhole/CCD chip and was about 19 for the 17 degree measurements and about 17 for the higher angles. From these parameters, a position sensitivity of less than two microns can be expected. The x-ray laser setup was based on a 4 mm wide Mo slab target which was hit perpendicularly by two uncompressed pulses (300 ps each, 480 mJ in total) after which the plasma was pumped by the main pulse (5 ps, 500 mJ) under various incidence angles. For a detailed description, see ref. [1].

Fig. 1. Schematic side view of the pinhole camera setup.

X-Ray Imaging of the Heating Zone of Non-Normal Incidence Pumped

77

3 Experimental results Each x-ray laser shot produced a CCD image similar to the one given in figure 2. After several steps of smoothing and averaging, vertical cross section lines were extracted, showing the averaged intensity distribution along an axis perpendicular to the target surface. An example of these plots is given in figure 3 for a GRIP angle of 17°, showing the averaged intensity over the distance to the target surface for all measured delays. The distances are to be taken only relative within each separate GRIP angle series since the setup used here did not allow for an absolute reference of the distance to the target and thus the position reference was lost due to magnification changes between different GRIP angle series. +150 µm

Prepulse Main pulse

pos. of long pulse only line (0 µm)

Plasma XRL

Target

-250 µm -750 µm

+750 µm

approx. center of plasma line (0 µm)

Fig. 2. Typical CCD image showing a field of view of about 1.5 mm around the center of the plasma line. 60

delay [ps]

200 300 400 500 600 700 800

50

40

intensity [a.u.]

30

20

10

0

-10 60

40

20

0

-20

-40

-60

distance from target surface relative to long pulse only line [µm]

Fig. 3. Vertical cross section plot for the 17° GRIP angle delay scan series.

78

B. Zielbauer et al.

One interesting feature can be seen by plotting the positions of the intensity peaks of the cross section lines over the respective delay times: linear fits (see figure 4) for each GRIP angle show clearly a decreasing slope with increasing GRIP angle. Based on the assumption that with increasing delay between the plasma-forming long pulse and the main pulse, the plasma is hit by the main pulse at a later stage in its development, one can interpret the slopes as expansion velocities. These velocities are plotted in figure 5, showing a clear decrease with increasing GRIP angle. This can be seen as an indirect demonstration that increasing the GRIP angle leads to absorption at higher electron density, closer to the target surface. 20 17° 19°

15

20° 21° distance 10 from target surface, referred to 5 long-pulseonly shots [µm] 0

-5

-10

200

300

400 500 600 delay long pulse/short pulse [ps]

700

800

Fig. 4. Emission peak positions for the measured delays of each GRIP scan. The linear fit lines are only given for 17° and 21°.

4 Comparison to simulation The measurement described above can be interpreted as a pump/probe experiment from which the dynamics of the hottest plasma region can be deduced as a function of the main pulse incidence angle. To analyze the effects, XRL plasma hydrodynamics modelling was performed using the EHYBRID code with inclusion of the short pulse incidence angle as in [3]. The simulation provides the temperature, average charge state, plasma expansion velocity and density distributions as functions of the distance to the target for different time instances. Since the turning point electron density can be estimated as being proportional to ncsin2α, a velocity can be deduced by evaluation the movement of points of equal density with increasing time. As expected, the turning point density expansion velocity

X-Ray Imaging of the Heating Zone of Non-Normal Incidence Pumped

79

is relatively high as presented in figure 5 using squares. The expansion velocity increases with decreasing electron density. The velocities predicted by EHYBRID are higher than the velocities of the x-ray emission peak measured in the experiment presented in figure 5 using circles. One possible interpretation for this could be that the timeintegrated x-ray emission peak is not placed at the turning point density region, but rather in the denser part of the plasma. speed from linear fits simulation (density)

50

4

simulation (nT )

40 speed 30 [km/s] 20 10 0

17°

18°

19° GRIP angle

20°

21°

Fig. 5. Comparison of the velocities derived from the keV emission peak measurements and the two modeling approaches

To support this assumption, the time-integrated plasma emission was simulated using the output of the EHYBRID code. The plasma emission was evaluated as being proportional with the electron density and with the electron temperature at power 4, as for the case of black body radiation. The simulated speed of the x-ray emission peak in this case (see figure 5, triangles) is much closer to the experimental result and qualitatively reproduces the measurements for the higher GRIP angles within a factor of 2. The result indicates, however, that the plasma is heated at densities higher than the turning point density which corresponds to a given incidence angle.

5 Conclusion The x-ray plasma emission of a GRIP Mo XRL was investigated using a versatile pinhole camera with variable magnification and adequate resolution. The x-ray plasma emission was studied in the plane along the plasma line focus and perpendicular to the target. Different main pulse angles were used and the delay of the pre-pulse to main pulse was varied

80

B. Zielbauer et al.

over several hundreds of picoseconds. In this way, the influence of the incidence angle on the heating of the plasma was studied. In addition, plasma modelling performed with the EHYBRID code was used to analyse the experimental results using a black body approximation of the plasma emission. The comparison shows good qualitative agreement between simulation and experiment. Furthermore, the measured velocities inferred from the data indicate that the absorption density increases with the GRIP angle, as expected from theory.

References 1. K. Cassou et al.: "A 10 Hz, 3 µJ transient x-ray laser pumped in grazing incidence geometry", this conference 2. Keenan, R.; Dunn, J.; Patel, P. K.; Price, D. F.; Smith, R. F.; Shlyaptsev, V. N.: 'High-Repetition-Rate Grazing-Incidence Pumped X-Ray Laser Operating at 18.9 nm", Phys. Rev. Lett., 94, 103901, 2005 3. Pert, G. J.: ′Optimizing the performance of nickel-like collisionally pumped xray lasers′, Phys. Rev. A, 73, 033809, 2006.

Formation of an Elongated Plasma Column From a Gas Puff Target Irradiated With a Laser Beam Focused Using an Axicon H. Fiedorowicz, A. Bartnik, R. Jarocki, R. Rakowski and M. Szczurek Institute of Optoelectronics, Military University of Technology, Warsaw, Poland

Summary. Elongated plasma columns can be formed by irradiation a gas puff target with a laser beam focused in line using an axicon. The plasma columns created from high-density gas puff targets based on a double-stream approach have been studied. The results of investigations with the use of optical and X-ray imaging and X-ray spectroscopy are presented.

1 Introduction There is a considerable interest in formation of elongated plasma columns in gases that could be used for guiding high-intensity laser pulses. The propagation of high-power laser pulses in plasmas may have widespread importance in a number of areas including X-ray lasers, high-harmonic generation, laser fusion, and laser plasma acceleration of charged particles. Relativistic and charge-displacement self-channeling in an under-dense plasma require of using intense ultra-short laser pulses. Another approach that uses a breakdown laser spark in a gas target can be realized using low intensity nanosecond laser pulses. It was demonstrated that plasma columns can be formed by irradiation elongated gas puff targets with nanosecond laser pulses focused to a line with an axicon [1]. This paper reports investigations on formation of an plasma column in a new high-density gas puff target [2] by focusing of nanosecond laser pulses from a Nd:glass laser using an axicon. The plasma columns have been studied with optical and X-ray imaging and X-ray spectroscopy techniques.

82

H. Fiedorowicz et al.

2 High-density gas puff target The target is based on a double-stream gas puff target approach [3]. It is formed by pulsed injection of working gas through a nozzle in a form of a slit, which is surrounded with two additional outer slit nozzles. The nozzles are supplied with gases from two separate electromagnetic valves mounted in the common body. Schematics of the nozzle setup and the valve system are presented in Fig. 1.

a)

b)

Fig. 1. Schematic of the double nozzle setup (a) and the valve system (b).

The stream of working gas injected through the central nozzle is confined by the outer stream of gas (helium), thus the central stream poses high density of gas at relatively large distance from the nozzle output. It allows efficient absorption of laser radiation in the plasma. Moreover, the central stream has a steep density gradient.

a)

b)

Fig. 2. Typical X-ray shadowgrams of the ordinary xenon gas puff target (a) and the double-stream xenon/helium gas puff target (b).

Formation of an Elongated Plasma Column From a Gas Puff Target

83

3 Gas puff target characteristics The targets were characterized with the use of the X-ray backlighting technique. Laser plasma X-ray source based on a gas puff target was used for backlighting. Typical X-ray shadowgrams of the ordinary xenon gas puff target (without helium) and the double-stream xenon/helium gas puff target are presented in Fig. 2. The corresponding target characteristics (gas density spatial profiles) for the ordinary xenon target and the double-stream xenon/helium gas puff target have been evaluated. The profiles obtained from the presented shadowgrams are shown in Fig. 3.

a)

b)

Fig. 3. Gas density profiles for the ordinary xenon gas puff target (a) and the double-stream xenon/helium gas puff target (b).

Fig. 4. Schematic of the experimental setup to study plasma columns produced from elongated high-density gas puff targets.

84

H. Fiedorowicz et al.

4 Experimental arrangements Schematic of the experimental setup to study plasma columns produced from elongated high-density gas puff targets is presented in Fig. 4. Laser pulses with energies up to 10 J in 1 ns FWHM from a Nd:glass laser were focused onto the target using an axicon. The axicon axis was parallel to the nozzle and placed at the distance of about 1 mm from the nozzle. The gas puff valve and the axicon mounted in the experimental chamber are shown in Fig. 5.

Fig. 5. The gas puff valve and the axicon mounted in the experimental chamber.

Fig. 6. Typical images of the plasma columns produced from 9-mm-long xenon gas puff targets.

5 Experimental results Typical images of the plasma columns produced from 9-mm-long xenon gas puff targets taken with a CCD photo-camera are presented in Fig. 6. X-

Formation of an Elongated Plasma Column From a Gas Puff Target

85

ray images of the plasma columns obtained with the use of a pinhole camera coupled to the X-ray CCD camera and the X-ray intensity distributions along the images are shown in Fig. 7. Uniform plasma columns were formed from 9 mm-long and 4 mm-long xenon targets. In the case of other gases (krypton, argon, oxygen) the columns were formed for 4 mm targets.

Fig. 7. X-ray images of the plasma columns obtained with the use of a pinhole camera coupled to the X-ray CCD camera and the X-ray intensity distributions for the xenon and krypton targets.

X-ray spectra measurements performed with the use of a transmission grating spectrograph made possible to estimate the electron temperature of the plasma. Typical spectra for the xenon and oxygen targets are presented in Fig. 8. The temperature of about 100 eV was obtained in the case of the oxygen target.

Fig. 8. Typical X-ray spectra for the xenon and oxygen gas puff targets

86

H. Fiedorowicz et al.

6 Conclusions Preliminary experiments on formation of elongated plasma columns from a high-density gas puff target irradiated with a nanosecond laser pulse focused to a line using an axicon were performed. Uniform plasma columns up a length of 9 mm could be produced.

References 1. Bartnik, A., Fiedorowicz, H., Kostecki, J., Szczurek, M., Morozov, A., Suckewer, S., Formation of elongated laser sparks in gas puff targets by nanosecond laser pulses, Proceedings of SPIE vol. 3156 Soft X-ray Lasers and Applications II (SPIE Press, Bellingham, 1997) 2. Fiedorowicz, H., Bartnik, A., Jarocki, R., Rakowski, R., Szczurek, M., Enhanced x-ray emission in the 1-keV range from a laser-irradiated gas puff target produced using the double-nozzle setup, Appl.Phys.B 70, 305-309, 2000 3. Fiedorowicz, H., Bartnik, A., Jarocki, R., Kostecki, J., Rakowski, R., Szczurek, M., Inst. Phys. Conf. Ser. No 186 Paper presented at 9th Int. Conf. X-ray Lasers, Beijing China, May 24-28, 2004©2005 IOP Publishing Ltd.

This page intentionally blank

Observation of Density Dip in a Pre-Formed Plasma: Toward Axial Pumping X-Ray Laser T. Kawachi, N. Hasegawa, M. Nishikino, Y. Ochi, M. Tanaka, T. Utsumi, and A. Sasaki X-ray Laser Group, Quantum Beam Science Directoratee, Japan Atomic Energy Agency (JAEA) Summary. We report the observation of electron density dip in a laser-produced molybdenum plasma pumped by double laser pulses which are linearly focused on the molybdenum slab target. Interferometer based on Wollaston prism was employed to diagnose the plasma, and we obtained interference pattern together with backlighting image of the plasma. The electron density dip was clearly formed at the time of 1~1.5 ns after the second pulse, and the electron density of the dip was estimated to be around 2x1018 cm-3 ~ 4x1018 cm-3. The existence of the density dip was also verified by the backlighting image, which showed that the probe beam was guided in the plasma column with the plasma length of up to 4.3 mm. The experimental result implies that the present plasma can be used as the pre-formed plasma for the axial pumping x-ray laser.

1 Introduction Ozaki et al. [1]. reported several years ago demonstration of directional xray laser beam at a wavelength of 18.9 nm of the nickel-like molybdenum ions, in which a molybdenum slab target was irradiated by line-focused pre-pulse of the laser in the normal direction, and the following main (heating) pulse with 150 fs-duration, an energy of 150 mJ was injected into the plasma column in the direction of longitudinal axis. Their explanation of the experimental result was that an electron density dip (or flat region) could be generated in the pre-formed plasma, and the main (heating) pulse in the longitudinal direction was guided in the pre-plasma to achieve substantial gain-length product. However, the existence of the density dip or flat region was not verified experimentally yet. Since the axial pumping scheme extends the possibility of x-ray lasers with very narrow beam divergence and improves the pumping efficiency of the GRIP (grazing incidence pumping) scheme [2], it is valuable to investigate the possibility of generating electron density dip in a pre-formed plasma. In this report, we employed the methods of interferometer and back lighting imaging

88

T. Kawachi et al.

technique to diagnose the pre-formed plasma, and found the electron density dip could be generated under double-pulse irradiation. Before the description of the details of the experiment, the related former works are described. The waveguiding of the pumping pulse has been studied before in the research fields of x-ray lasers and laser acceleration: In the laser-driven plasmas, Milchberg et al. found “light pipe” was generated in the gas target irradiated by double pulses irradiation. In this experiment, the first pulse makes density gradient in the radial direction, and the following second pulse was guided in the electron density structure [3,4]. In the discharge plasmas, Hosokai et al. demonstrated waveguiding of the terawatt laser pulse in the capillary discharge plasma column [5]. In the x-ray laser research, quite recently, Tommasini et al., measured density profile of the pre-formed plasma pumped by line-focused single pulse of TiSa laser. However, no density dip could be observed under the single pulse irradiation [6]. Therefore this work is located to the following attempt by use of double pulses irradiation.

2 Experiment and Discussion Experiment was conducted by use of JAEA CPA Nd:glass laser. The maximum total output energy of this laser is 20 J with 1 ps duration, however, since our objective was to diagnose the pre-formed plasma, we used only the pre-amplifier of this system to control the output energy to be 200 mJ ~ 500 mJ. The output consisted of two pulses. The duration of the first- and the second pulse was 10 ps, separated by 1.3 ns, and the energy ratio of the first pulse to the second pulse was 1:3.3. A portion of the uncompressed laser pulse was used as the probe beam: the duration was ~ 450 ps (bandwidth of 1.8 nm), and the energy was ~ 2 mJ. The probe beam was injected into the plasma in the longitudinal direction. The diameter of the probe beam was adjusted to be 5 mm at the position of the plasma. Then the probe beam went through the imaging lens, changed the polarization direction to 45 degree with respect to the horizontal axis by the first polarizer, was divided into two polarization components by Wollaston prism, WP, and changed the polarization direction again to the horizontal direction by the second polarizer. Finally the two components were merged by the steering mirrors at the focal position of the imaging lens, where the portion of the probe beam including the information of the plasma and that went through the vacuum region made interference pattern, and we could obtain the phase information of the probe beam at the edge of the pre-formed plasma. In this set up, the magnification factor

Observation of Density Dip in a Pre-Formed Plasma

89

was around 9, which was assured by use of 120 µm in diameter copper wire as the test sample. The spatial resolution was ~ 10 µm. We set the delay line in the optical path of the probe beam, and adjust the timing of the probe beam with respect to the pumping pulse. It should be noted that by cutting one of the probe beam of the interferometer, we could obtain the backlighting image of the plasma. In the following context, the time of the probe beam was measured from the peak of the second pumping pulse. Figure 1 shows the interferometer patterns together with the backlighting image of the pre-formed plasma at the timing of t = -300 ps; (a), t = +1000 ps; (b) and t = +1500 ps; (c), respectively. The total pumping laser energy was set to be 190-210 mJ, i.e., the pre-pulse energy and main-pulse energy was ~ 45-53 mJ and 145-157 mJ, respectively, which corresponds to the intensity on the target of 4x1012 W/cm2 and 1.2x1013 W/cm2, respectively.

(a)

(b)

(c)

Fig.1. Image of the plasma taken with an interferometer. Upper column is the interference pattern and the lower is the back lighting image.

In Fig.1 (a), the plasma image is the same with that pumped by single pulse (See ref. [6]). The interference pattern shows no unique feature in the density gradient, whereas at t=-1000 ps ~1500 ps, complicated patterns is shown in the interference patterns. The backlighting images at the same timing show the probe beam is guided in the plasma with the length of 1 mm. In order to assure the waveguiding, we change the thickness of the target from 1 mm to 4.3 mm, and the probe beam was clearly guided in the plasma as shown in Fig. 2. We tried to estimate the spatial profile of the electron density in Fig. 1. In the present experimental condition, the wavelength of the probe beam is 1 µm, and the thickness of the target or plasma is ~ 1 mm. This implies

90

T. Kawachi et al.

that one phase shift corresponds to the 2x1018 cm-3 of the electron density. We compare the interference pattern in Fig. 1 (c) with that without plasma, and the spatial profiles of the phase shift are derived. The data analysis showed that at t = +1500 ps, the density dip is generated: the electron density of the bottom of the dip is ~ 2x1018 cm-3, and that of the outer region was around 4x1018 cm-3. The interference patterns at t = +1000 ps is more complicated than that at t = + 1500 ps, which indicates that the density dip may be generated in higher density region by at least one-phase shift ( > 4x1018 cm-3), however, the density is higher than the range of the present interferometer. We estimate the possibility of axial pumping x-ray laser by using this pre-formed plasma. Consider the situation that the circularly polarized axial pumping (heating) laser pulse is injected into the pre-formed plasma. If the intensity is sufficiently high, the ions increase their ionization stage up to the nickel-like (z = 14). Our 1D hydrodynamics simulation (HYADES) shows the average of the charge state of the ions in the preformed plasma at the time of t = +1500 ps is z ~ 3, therefore the after the axial pumping pulse, the electron density of the pre-formed plasma is increased up to around ~ 1x1019 cm-3. This electron density is not so high compared with optimum electron density for the nickel-like molybdenum laser (> 5x1019 cm-3), however is enough to obtain substantial lasing in this scheme. In order to obtain the strong lasing in this scheme, the axial pumping pulse should be injected in earlier timing: Indeed hydrodynamics simulation shows that after the second (heating) pulse, plasma expands in the vacuum with keeping its spatial profile, therefore in earlier time region, electron density of the dip is expected much higher than at t =+1500 ps.

Fig. 2. The backlighting image of the plasma column. The probe beam was guided in the plasma with the length of up to 4.3 mm.

This tendency is consistent with present result, at t = 1000 ps, the electron density of the dip is higher than that at t = +1500 ps. In order to use earlier

Observation of Density Dip in a Pre-Formed Plasma

91

time region, it is required that the precise control of the intensity of the pumping pulses and their homogeneity on the target. In summary, we have demonstrated to generate electron density dip in the pre-formed plasma by use of double pulses irradiation. The plasma parameter at the time of t = 1.5 ns measured from the second pre-pulse timing, i.e. electron density and the ionization stage of the ions, indicate that the axial pumping x-ray laser in the nickel-like molybdenum ions is possible by use of this pre-formed plasma. This work is auspices by Japan Society for the Promotion of Science (JSPS).

References 1. T. Ozaki, R. A. Ganeev, A. Ishizawa, T. Kanai, and H. Kuroda, Phys. Rev. Lett. 89, 253902 (2002). 2. B. M. Luther et al., Opt. Lett. 30, 165 (2005). 3. C. G. Durfee III and H. M. Milchberg, Phys. Rev. Lett. 71, 2409-2412 (1993). 4. T.R. Clark and H. M. Milchberg, Phys. Rev. Lett. 78, 2373 (1997). 5. T. Hosokai, M. Kando, H. Dewa, H. Kotaki, S. K. N. Hasegawa, K. Nakajima and K. Horioka, Opt. Lett., 25, 10 (2000). 6. R. Tommasini, K. Eidmann, T. Kawachi and E. E. Fill, Phys. Rev. E 69, 066404 (2004).

This page intentionally blank

Experimental Studies on Transient Ni-Like Ag X-Ray Laser Pumped With the Picosecond Pulsed Laser Facility at NLHPLP J. R. Sun1) , Ch. Wang1) , Z. H. Fang1) , W. Wang1) , J. Xiong1) , J. Wu1) , S. Z. Fu1) , Y. Gu1) , S. J. Wang1) , G. P. Zhang2) , W. D. Zheng2) , G. L. Huang3) , F. Y. Guan3) , X. L. Xie3) 1)

Shanghai Institute of Laser Plasma, Shanghai, 201800, P. R. China Beijing Institute of Applied Physics and Computational Mathematics , Beijing, 100088, P. R. China 3) Shanghai Institute of Optics and Fine Mechanics, Shanghai, 201800, P. R. China 2)

Summary. The results of experimental studies on transient Ni-like Ag soft X-ray laser with the picosecond pulsed laser facility at the National Laboratory of High Power Laser and Physics (NLHPLP), China, has been reported in this paper. An somewhat intense Ni-like Ag X-ray laser beam at 13.9 nm with output energy 5~10 nJ was obtained from the solid flat targets under the joint irradiation of a long pulse laser beam of several hundred picosecond duration and another 1ps ultra-short pulsed laser.

1 Introduction Since X-ray laser has widespread application prospects in physics, biology, chemistry, materials science and initial confinement fusion (ICF), X-ray laser research has been intensively pursued by the community in recent years. Many laboratories have got the saturated output of X-ray laser from 3.5 nm to 50 nm [1-7] and try to demonstrate the soft X-ray laser amplification [8-15]. In order to find an approach to the table top X-ray laser devices of high efficiency and intense output in the short wavelength region, the most important issue in this X-ray laser development is to employ a proper compact pumping system. Recently, several experiments carried out in various laboratories have confirmed that saturated X-ray lasers pumped by transient collisional excitation (TCE) may be obtained with moderated pump laser energy requirements [16-22].

94

J. R. Sun et al.

In the generation of transient X-ray laser, the double-pulsed pumping scheme is often be used, where a long laser pre-pulse of nano- or subnanosecond duration forms the plasma on the target before the arriving of another short main pulse of pico- or sub-picosecond duration. To regulate properly the delay between two pumping pulses, the short main pulse can rapidly heat the pre-plasma and generate high gain output of the transient X-ray laser. In this paper we report the experimental demonstration of a Ni-like Ag X-ray laser at wavelength of 13.9 nm with a pumping energy ~15 J only. The experiments in this paper were characterized by employing a pre-prepulse before the pre-pulse to optimize the pre-plasma behavior, where the intensity ratio of the two pulses was 5~10 % and the time interval between them ranged 1 ~ 4 ns. A somewhat intense Ni-like Ag X-ray laser at 13.9 nm with output energy 5~10 nJ has been obtained under the irradiation on the solid flat targets with a long pulsed laser beam of 102 ps duration and an 1 ps ultra-short pulsed one jointly.

2 Experiments The experiments have been performed on the picosecond laser facility at the National Laboratory of High Power Laser and Physics (NLHPLP), Shanghai, China. This laser facility outputs two laser beams, in which the maximum energy of laser beam output from the compressor is ~15 J and with pulse duration ~1 ps, another long-pulse laser beam is of ~300 ps duration and ~25 J maximum output energy. The two beam lines provide respectively the pre-pulse and the main pulse with adjustable delay between them. The experimental setup is shown in Fig. 1. The long pre-pulse through the right window enters the chamber to irradiate the target. The horizontal line-focusing system for the pre-pulse consists of protruding cylindrical lens, glass windows and principle focus lens. On the other hand, the short main pulse passes the left window into the chamber to irradiate the target after a delay time. Its horizontal line-focusing system consists of plane reflector, concave cylindrical lens and the off-axial parabolic reflector. Two horizontal focus lines overlap after the off-axial non-spherical mirror and are with an incident angle of 25 degree towards the target. The both horizontal line-focusing systems are shown in Fig 1 (b) in detail. The linear spot size for the two focus lines is about 50 μm×10 mm. The laser intensity distribution along the line focus is asymmetry because of the single cylinder line focus system used. Therefore only the

Experimental Studies on Transient Ni-Like Ag X-Ray Laser

95

central part of line focus could be available. The slab targets were prepared with sliver coating of thickness 1 μm on polished glass substrates. The maximum length of sliver targets was 6 mm to assure the laser ablation plasma to cover the symmetrical central part of the line focus. The spectral composition of the X-ray output was observed with a flat field grating spectrometer consisting of a concave varied-spacing diffraction grating of 1200 grooves/mm and with an acceptance angle 12 mrad in the vertical direction. The detector was a back-illuminated X-ray charge-coupled device (CCD). The position of spectrometer and CCD detector in the horizontal direction could be changed to detect the X-ray output from different locations on the target. A Mo filament of 75 μm in diameter set at the spectrometer split was used to determine the refraction angle of X-ray laser beam.

Fig. 1. Experimental setup of (a) schematics (b) optical path

two

laser

pulse

driven

X-ray

laser

96

J. R. Sun et al.

3 Results and Conclusion Fig. 2 shows the image recorded by the CCD detector in an experiment where the 6 mm-long silver target was irradiated by two laser pulses separated by 120 ps. The pre-pulse duration was 414 ps, energy ~1.3 J, the energy ratio of pre-pre-pulse to that of pre-pulse ~ 7.3 % and the interval of them 2.95 ns. The main pulse duration was equal to 1 ps and energy ~ 4.1 J. The irradiances of pre-pulse and main pulse on the target were 5.0×1011 Wcm-2 and 7.0×1014 Wcm-2, respectively.

Fig. 2 CCD detector recorded radiation image from the silver coating target

The X-ray output spectrum was measured and treated, and is shown in Fig. 3, whereas its spatial angle distribution in the horizontal direction is shown Fig. 4. Considering the reflective angle was ~10 mrad and the divergence angle ~4 mrad, this spectrum has a certain dependence upon the angle off the target normal, hence a high intensity and a well directed behaviour. In order to measure the spectral wavelength, an Al pass filter with central wavelength of 17.04 nm was placed in the beam path. The spectral line wavelength of 13.9±0.1 nm was measured by the spectrometer. It can be concluded that this spectrum demonstrates the X4

1.2x10

4

1.4x10

NO: 050111020 1.2x10

Intensity (CCD Count)

1.0x10

CCD count

NO: 050111020

4

4

3

8.0x10

3

6.0x10

3

4.0x10

3

2.0x10

4

1.0x10

3

8.0x10

3

6.0x10

3

4.0x10

3

2.0x10

0.0

0.0

8

10

12

14

16

18

20

Wavelength (nm)

Fig. 3 Experimental spectrum of the Ni-like Ag X-ray laser output

0

2

4

6

8

10

12

Angle from target surface (mrad)

Fig. 4 The spatial angle distribution of the Ni-like Ag X-ray laser spectrum

ray output to be Ni-like Ag X-ray laser. The estimated X-ray output energy was 5~10 nJ. In conclusion, the Ni-like Ag x-ray laser output at 13.9 nm with energy 5~10 nJ has been successfully obtained in the experiments employing

Experimental Studies on Transient Ni-Like Ag X-Ray Laser

97

silver coating targets irradiated with three laser beams consisting of a long pre-pulse (102 ps in duration) added a pre-pre-pulse and an 1 ps ultra-short main pulse. The total pump energy used in the experiment was about 5 J. The better X-ray laser output could be achieved if the experimental setup with a line-focusing system capable of quasi-travelling wave irradiation on the target is developed. This approach has been considered as an important step in the development of X-ray laser and its applications in China.

Acknowledgements The authors are greatly obliged to stuff of the SG-Ⅱgroup at NLHPLP for their excellent efforts and contributions in the experiments.

References 1. Matthews D L, Hagelstein P L, Rosen M D et al , Phys. Rev. Lett. 54, 110, 1985 2. Wang S J, Gu Y, Zhou G L et al, Chin. Phys. Lett. 8, 618, 1991 3. Carillon A, Chen H Z, Dhez P et al, Phys. Rev. Lett. 68, 2917, 1992 4. Koch J A, MacGowan B J, Da Silva L B et al, Phys. Rev. Lett. 68, 3291, 1992 5. Rocca J J, Clark D P, Chilla L A et al, Phys. Rev. Lett. 77, 1476, 1996 6. Zhang J, MacPhee A G, lin J et al, Science 276, 1097, 1997 7. Wang C, Wang W, Wu J et al, Acta Phys. Sin. 53, 3752, 2004 8. Trebes J E, Brown S B, Campbell E M et al, Science 238, 517, 1987 9. DaSilva L B, Trebes J E, Balhorn R et al, Science 258, 269, 1992 10. Ress D, DaSilva L B, London R A et al, Science 265, 514, 1994 11. DaSilva L B, Barbee T W, Cauble R et al, Phys. Rev. Lett. 74, 3991, 1995 12. Rocca J J, Moreno C H, Marconi M C et al, Opt. Lett. 24, 420, 1999 13. Wang C, Gu Y, Fu S Z et al, Acta Phys. Sin. 51, 847, 2002 14. Filevich J, Rocca J J, Jankowska E et al, Phys. Rev. E 67, 056409, 2003 15. Wang C, Wang W, Sun J R et al, Acta Phys. Sin. 54, 202, 2005 16. Nickles P, Shlyaptsev V N, Kalachnikov M et al, Phys. Rev. Lett. 78, 2748, 1997 17. Dunn J, Osterheld A L, Shepherd R et al, Phys. Rev. Lett. 80, 2825, 1998 18. Kalachnikov M.P., Nickles, P.V., Schnürer, M., et al, Phys.Rev. A 57, 4778, 1998 19. Dunn J, Li Y, Osterheld A L et al, Phys. Rev. Lett. 84, 4834, 2000 20. Kuba J, Klisnick A, Ros D et al, Phys. Rev. A 62, 043808, 2000 21. King R E, Pert G J, McCabe S P et al, Phys. Rev. A 64, 053810, 2001 22. Klisnick A, Kuba J, Ros D et al, Phys. Rev. A 65, 033810, 2002

This page intentionally blank

X-Ray Laser Source Using Ceramic Target Y.Takemura1, N.Yamaguchi2 and T.Hara1 1

X-ray Laser & Plasma Engineering Laboratory, Toyota Technological Institute, 2-12-1 Hisakata, Tempaku, Nagoya 468-8511, Japan 2 Physics lab., Liberal Arts and Science, Graduate School of Medicine and Pharmaceutical Sciences, University of Toyama, 2630 Sugitani, Toyama 930-0194, Japan

Summary. Ceramic plates (AlN and alumina) have been used as targets for laserplasma x-ray sources for the first time. Radiation intensity of soft x-ray spectra from Li-like ions (Al XI) for the AlN target was almost the same as in the case of pure aluminium target. It was found that the ceramic target can be irradiated at the same place up to 70 times or more, radiating x-rays with nearly the same intensity as in the case when a virgin surface was irradiated. Craters formed on the target surface after irradiations were measured 3-dimensionally and their hollow volumes below the initial surface were analyzed. The ablated mass for AlN or alumina target was less than 1/6 compared with that for pure Al target. We succeeded in development of the laser-plasma x-ray source for a long operating period, maintaining stable x-ray output.

1 Introduction We are investigating table-top x-ray lasers based on the recombination plasma scheme. In this scheme, the well-known lasing line is Al XI 3d-4f transition line at 15.47 nm [1]. For researches on x-ray lasers and also on laser-plasma x-ray sources, pure aluminum slab or tape target is often used. After laser irradiation, some craters are created on the target surface or a part of aluminum tape is ablated, giving much particle debris. It causes one of the technical issues for development of practical laser-plasma x-ray sources. On the other hand, ceramic materials containing Al like AlN would produce little debris because of its high sublimating decomposition temperature (2450 °C), whereas the melting point of Al metal is 660 °C. Though ceramic materials could be useful as target materials, there are no attempts to apply ceramics for laser-plasma x-ray sources. We have carried out experiments using a ceramic target, which is AlN or Al2O3. Soft x-ray spectra were measured and surface profiles after laser

100

Y. Takemura, N.Yamaguchi and T.Hara

irradiation were observed. Characteristics of soft x-ray radiation and debris production will be described in this report.

2 Experimental Setup 2.1 Irradiation System and Target A 100 ps laser pulse from a mode-locked YAG oscillator, 1064 nm, was transformed into a linearly polarized 16-pulse-train through an optical stacker and a delay line component. The interval of each pulse was 200 ps. A YAG laser system with four-stage amplifiers delivered output energy of about 0.7 J. A Nd:Glass amplifier having a 25-mm-diameter rod was added to the YAG laser system, and the output laser energy was increased up to 2.1 J. In our experiments, line-shaped plasmas are produced on a target by using a line-focusing optics. Intense x-ray could be generated from the line-plasma, because the x-ray intensity is proportional to its plasma length for the optically thin radiation. Another merit of this type of source is that a very few debris could come along the surface of the target which is the direction of x-ray beam [2]. The line-focusing lens system consists of a five segmented prism array, a beam expander and a cylindrical lens assembly [3]. The prism splits the incident laser beam into five beamlets. Each beamlet is then separately focused on a line and all these individual line foci overlap at the same place, which forms a common-line focus. The intensity variations in the incident beam are averaged, causing an overall homogeneous illumination along the focal line. Besides the irradiation pattern has a small scale modulation due to interference between divided laser beams that results to form an array of micro-dot on a target [4]. The irradiation pattern has an irradiation area of 50 µm × 11 mm with microdots of 50 µm in diameter and 140 µm-pitch spacing. Because of the interference pattern in the focused beam, the power density was 1011− 1012 W/cm2. Targets used were aluminium (1 mm thick), sintered AlN (2 mm thick) and alumina (2 mm thick) plates. 2.2 Diagnostics X-ray emission from the plasma was analyzed using a space resolving flatfield grazing incidence soft x-ray spectrograph with a toroidal mirror as a

X-Ray Laser Source Using Ceramic Target

101

pre-focusing optics. The detector was an x-ray CCD camera (SX-TE/CCD 512TKB, Princeton Instrum.) using a back-illuminated CCD of 512 × 512 pixels with 24 × 24 µm pixel size. This spectrograph was designed to image x-rays from a source with a 1× magnification in a horizontal direction at the entrance slit and with a 3× magnification in a sagittal direction at the flat-field output plane. Surface profiles after laser irradiation were measured by using a 3D digital optical microscope (VHX-200/100F, KEYENCE) and a laser confocal microscope (VK-9500, KEYENCE). (a)

(b)

Intensity (A.U.)

Al XI 15.47nm Al V Al V

Al V

Al XI 15.47nm Al XI

Al XI

NV NV

Al IV Al V Al V

Al IV

Al V

NV

N IV NV

Al IV

12

13

14 15 Wavelength (nm)

16

17 12

13

14 15 Wavelength (nm)

16

17

(c) O VI

O VI or Al XI

Al V

Intenisty (A.U.)

OV Al XI or O IV O V OV O VI

12

13

14 15 16 Wavelength (nm)

17

Fig. 1. Typical soft x-ray spectra from (a) Al, (b) AlN and (c) alumina plasmas. Emission lines of the Li-like Al ions, labeled as Al XI, are intense in the AlN plasma.

3. Soft X-ray spectra from Laser-produced Plasmas Soft x-ray spectra were observed for each target material in the wavelength region from 12 nm to 17 nm. Typical spectra are shown in Fig. 1. In each

102

Y. Takemura, N.Yamaguchi and T.Hara

spectrum, line spectra have been originated from ions of composed materials of each target. Emission lines from highly ionized Al, especially Li-like Al, were still observed with about the same intensity in the AlN plasma as those in the pure-Al plasma. On the other hand, Al spectra were weak in the alumina plasma. We performed multiple shots irradiation experiments where the laser beam was focused on the same position of the target many times. When an Al slab target was used, craters were formed on the target surface aligning in the focused line. The depth of the craters was around 8 µm in a single shot, and then the crater became deeper as increase of times of irradiation. Therefore, Al targets cannot be used to produce x-ray radiation for multiple shots on the same place efficiently, because the effective power density decreases. On the other hand, ceramics suffered weak damage after a number of laser irradiations. This leads to capability of multiple use of the same target area. Figure 2 shows the trend of radiated intensity of the 15.47 nm line as a function of number of irradiations for AlN target and Al target. 8

2 10

ALN

Intensity (A.U.)

pure aluminum

8

1 10

0 0

10

20 30 40 50 60 Number of irradiations

70

Fig. 2. The trend of the 15.47 nm line intensity as a function of the number of irradiations for the case of AlN target.

4 Surface Modifications by Laser Irradiation Surface profiles of the target were measured three dimensionally. Small craters were formed by multiple irradiations even on the ceramic targets. The depth of the crater was 8 µm/shot, 1.4 µm/shot or 0.9 µm/shot, for the

X-Ray Laser Source Using Ceramic Target

103

pure Al, AlN or alumina target, respectively. It was clear that the rate of crater formation on the ceramic target is much lower than that on the Al slab target. 250

Ablated mass (μg)

200

150

100 Pure Aluminium Alumina ALN

50

0 0

20

40 60 80 Number of irradiations

100

Fig. 3. The amount of target mass evaluated from the crater volumes as a function of the number of irradiations. Those give the maximum value of the ablated mass from the target.

0.8 First shot After 10000 shots

Intensity (V)

0.6

0.4

0.2

0

-0.2 -50

0

50 Time (ns)

100

150

Fig. 4. The intensity of x-ray pulses measured by a fast response x-ray detector (AXUV HS1) at the different timing during continuous operation.

The above results lead that the amount of debris would be small in the case of ceramic target. The volume of craters was calculated from the 3D data of surface profile with respect to the initial flat surface. This gives us

This page intentionally blank

104

Y. Takemura, N.Yamaguchi and T.Hara

the maximum mass of the target material ablated. The amount of target mass evaluated from the hollow volume of the craters is plotted as a function of the number of irradiations in Fig. 3. Though the ablated mass for the ceramic targets is estimated to be in the order of several µg/shot, which is seemed to be rather large, it can be considered that the amount of debris from the ceramic target is 1/6~1/7 as that from Al slab target.

5 Laser-Plasma X-ray Source using Ceramic target In order to use laser-plasma x-ray source for a long operating period, we have constructed a target stage which holds a ceramic target. The ceramic plate can be moved back and forth along one axis on the stage, where the pitch of translation and the timing can be programmed by stepping motor controller. The target stage worked stably for 2 hours under 10 Hz-laserirradiation on the target surface. We have performed a test experiment on the stability of x-ray generation. X-ray pulses measured by a fast response x-ray detector (AXUV HS1) are shown in Fig. 4, where the signals show time behaviour of x-ray intensity generated at the different shot during continuous operation. The present results indicate that the stability of x-ray generation is fairly good.

6 Conclusions Ceramic targets containing Al, such as AlN and alumina, have been investigated as a target for soft x-ray source. In the case of AlN, emission lines from highly ionized Al, especially Li-like Al, were still observed with about the same intensity those in the pure-Al plasma. Moreover, it was found that 15.47 nm x-ray intensity for each laser shot was almost constant for 70 shots and more in the multiple shots irradiation experiments. It would be possible to construct a long-lived and compact laser-plasma xray source with less debris using a peace of AlN plate.

References 1. N. Yamaguchi, T. Hara T. Ohchi, C. Fujikawa and T. Sata: Jpn. J. Appl.Phys. 36, 51, 1999.

X-Ray Laser Source Using Ceramic Target

105

2. P. Abraha, Y. Hisada, K. Takamoto, N. Yamaguchi, and T. Hara: Jpn.J. Appl. Phys,. 42, 1491, 2003. 3. N. Yamaguchi, T. Ohchi, C. Fujikawa, A. Ogata, Y. Hisada, K. Okasaka, T. Hara, T. Tsunashima and Y. Iizuka: Rev. Sci. Instrum. 70, 1285, 1999. 4. N. Yamaguchi, C. Fujikawa, T. Ohchi and T. Hara: Jpn. J. Appl. Phys,.39 5268, 2000

The Generation of X-Ray Laser Radiation on SOKOLP Laser Facility A.V. Andriyash, O.V. Chefonov, D.A. Dmitrov, D.S. Gavrilov, A.G. Kakshin, E.A. Loboda, V.A. Lykov, E.P. Magda, V. Yu. Politov, A.V. Potapov, V.A. Pronin, G.N. Rykovanov, V.N. Sukhanov, A.S. Tischenko, A.A. Ugodenko, D.A. Vikhlyaev and A.L. Zapysov Russian Federal Nuclear Center – named after academician E.N.Zababakhin All-Russia Research Institute of Technical Physics (RFNC-VNIITF, Snezhinsk, Russia)

Summary. The transient collisional excitation (TCE) scheme was used to obtain generation of X-ray laser radiation on the 3p-3s transitions of the Ne-like Ti-ions at 10 TW SOKOL-P laser developed at RFNC-VNIITF. The Nd-laser light was focused in a line with length from 2 up to 8 mm and width of 30 microns. Two successive pulses irradiated the polished Ti-slab. The duration of prepulse was equal to 400 ps and a pumping pulse had duration of 4 ps; the delay between pulses equals to 1.5 ns. The Nd-laser energy was about of 10 J and the ratio of energy in prepulse and basic pulses was equal to 1:2. The grating spectrometer equipped with a focusing mirror and CCD detector was used for measurement of the X-ray laser line at 32.6 nm. A small signal gain about of 30 cm-1 was obtained in experiments with target length from 2 up to 4 mm. The 32.6 nm line radiation with energy about of 1 μJ and divergence of 9 mrad were registered in SOKOL-P experiments with target length of 8 mm.

Introduction The transient collisional excitation (TCE) scheme of X-ray laser (XRL) makes it possible to attain extra-high peak gain factors gmax≥100 cm-1 [1]. This is promising for the development of tabletop XRL as well as progress in shorter wavelengths. X-ray lasing on 3p-3s transition of Ne-like titanium ions with TCE was first demonstrated at Max Born Institute [2]. In those experiments the flat titanium targets were irradiated by a combination of a 1.5 ns prepulse (energy up to 7 J) and a main pulse of 0.7 ps (energy up to 4 J). The experimental value of XRL gain was measured to be g=19±1.4 cm-1. Similar experimental results, but with the higher gain factor (g=24 cm-1) were obtained at LLNL on the “Janus” facility [3].

108

A. V. Andriyash et al.

A numerical simulation of hydrodynamics and level-by-level ion kinetics for the XRL active media under powerful picoseconds heating has been carried out [4]. This simulation shows that the active media length dependence of output XRL energy is determined, first of all, not by the peak value of the gain, but by the refraction of XRL beam in the strong density-gradient amplifying region, and by the effect of radiation lag. This dependence results in obtaining of certain effective gain factor that does not reflect the g values really attainable in the XRL plasma. The gain value g ~ 50 cm-1 was estimated in [4] for the XRL on 3p-3s transitions of Nelike Ti ion.

1 Experimental Setup The experiments were conducted on the laser facility SOKOL-P [5]. The facility is a picosecond chirped pulse amplification laser whose maximum power is about 10 TW. In these TCE experiments, a portion of energy of the chirp-modulated amplified pulse tch=400 ÷ 440 ps long was used as a prepulse creating the laser active medium. The duration of the basic pumping pulse tp is varied with a base compressor adjustment. The compressor was set for pulse duration tp=4 ps. The picosecond pulse lag relative to prepulse was maintained constant and equal to 1.5 ns. The ratio of prepulse and main pulse energies was 1:2. This ratio was also maintained constant; the prepulse energy was 2.5±0.3 J, the main pumping pulse energy was 6.3±1.0 J. The accuracy of position coincidence of prepulse and main pulse beams focused into a thin line on the target is no worse than ±3 μm. The focusing optics consists of a toroidal mirror mounted in a vacuum target chamber and a meniscus lens located in the air in front of the LiF input window. This system ensures the laser beam focusing into a thin line with a maximum length of 10 mm. To eliminate the non-uniform irradiation of the target, the parallel beam was blinded and the target length was no longer than 8 mm. In this case the droop of intensity on the target edges was below than 15% of its value in the center of the focal line. X-ray images from the pinhole-camera enabled the control of the focal line width and irradiation uniformity as well. The target position in the pumping laser beam was not changed in experiments and the focal line was equal to 35-40 μm in width. The laser light intensity on the target surface was 2⋅1012 W/cm2 in prepulse and 6⋅1014 W/cm2 in the main ps pulse. The grazing incidence spectrograph with flat diffraction grating

The Generation of X-Ray Laser Radiation on SOKOL-P Laser Facility

109

(600 mm-1) was used to register spectra in the wavelength region 120÷400 Å (see Fig.1).

Fig. 1. Layout of grazing incidence spectrograph: 1 – target, 2 – focusing mirror, 3 –spectral slot, 4 – diffraction grating, 5 – CCD-matrix

A focusing spherical mirror was mounted between the target and the grating. The mirror collected the target radiation in a plane perpendicular to the target surface in an angle of 0.035 radians. The target center was in the meridional focal plane of the mirror, which formed almost parallel beam of x-rays. The width of this beam for x-ray laser radiation is determined by XRL divergence and by the collection angle of the mirror for all other registered spectral lines. Thus, this diagnostic allows analysis of XRL beam divergence in the plane perpendicular to the target surface. The spectrum is recorded with PI-SX: 400 x-ray CCD-camera. The spectral resolution λ/Δλ for the XRL wavelength λg=326 Å was about 140. Special attention was given to the thorough adjustment of the spectrograph relative to the axis of the x-ray laser beam. The error of the spectrograph slot alignment about the beam axis was approximately ±3 mrads.

2 Experimental Results Figure 2 shows a spectrogram obtained in the experiment with the 6-mm target. The bright spectral line with wavelength λ=326 Å dominates in the spectrogram. The target length was varied from 2 to 8 mm. Figure 3 shows the dependence of x-ray laser intensity on the target length. As far as

110

A. V. Andriyash et al.

pumping energy during the experiments was not strictly constant (experimental scattering ±16%), the experimental points on the plot are

Fig. 2. Spectrogram of the shot with 6 mm target length

Fig. 3. Dependence of XRL intensity on target length

arrow-accompanied to point trends in the variation of the recorded spectra in the case where the pumping energy would be of a nominal value of 6.3 J. For 2 mm targets, the XRL output was founded out to be below the diagnostic threshold sensitivity. The downward direction of the arrow means that the given experimental point is actually the upper limit of XRL output.

This page intentionally blank

The Generation of X-Ray Laser Radiation on SOKOL-P Laser Facility

111

Fitting our data for target lengths of 2-4 mm with the Linford formula and taking into account the above mentioned corrections due to pumping energy instability, gives a gain value of 30±5 cm-1, that is somewhat lower than the simulation gain value of ~ 50 cm-1 [4]. The calculated results seem to be overestimated because of the imperfect model that was used for the XRL radiation transport under strong influence of refraction. The maximum energy of the X-ray laser beam was registered about of 1 μJ and its divergence was 9±3 mrads in experiments with target length of 8 mm.

Conclusion The X-ray laser generation on the 3p-3s transitions of the Ne-like Ti-ions has been studied in experiments performed on 10 TW SOKOL-P laser facility developed at RFNC-VNIITF. The TCE scheme was implemented through the sequential irradiation of polished Ti-slab with 0.4 ns and 4 ps laser pulses focused into a line. Pumping laser energy density per unit line length was about of 0.7 J/mm. The experimental small signal gain of the 32.6 nm line radiation was 30±5 cm-1. The maximum energy of the XRLradiation about of 1 μJ and its divergence of 9±3 mrads were registered in experiments with the Ti- target length of 8 mm.

References 1. Afanasiev, Yu. V., Shlyaptsev, V.N.: Quantum Electronics, 16, 2499, 1989. 2. Nickles, P. V., Shlyaptsev, V. N. et al.: Phys. Rev. Lett., 78, 2748, 1997. 3. Dunn, J., Osterheld A. L., Sheperd R. et al.: In: Proc. SPIE Meeting (San Diego, California), 1997; Dunn J. et al.: Phys. Rev. Lett., 80, 2825, 1998. 4. Politov V. Yu., Lykov V. A., Shinkarev M. K.: Quantum Electronics, 30, 1037, 2000. 5. Dmitrov D. A., Fomichev L. A., Kakshin A. G. et al.: Proc. 28th European Conference on Laser Interaction with Matter, 591-599, 2004.

Electron Temperature Measurements of Model NeLike X-Ray Laser Plasmas From Resonance keV Spectra J. Nejdl*, B. Rus, J. Kuba*, T. Mocek, M. Kozlová, J. Polan and P. Homer *Faculty of Nuclear Sciences and Physical Engineering, Czech Technical University in Prague, Czech Republic Department of X-ray Lasers, Institute of Physics/ PALS Centre, Prague 8, Czech Republic

Summary. We present results of temperature measurements of Ne-like Zn plasmas created by an intense laser pulse, using slab targets. Pulses with typical energy of 50 J in 300 ps were focused to a line or point foci, in which the pumping pulse energy was varied from shot to shot. The electron temperature in dependence on the pulse energy and irradiation intensity is studied. The temperature is inferred spectroscopically from a set of selected Ne-like spectral lines. The experimental results are compared with the numerical simulations made by the hydrodynamic code EHYBRID.

1 Introduction Electron temperature is one of the crucial parameters of plasma based X-ray lasers, especially of those using collisional excitation pumping. Therefore, the knowledge of the dependence of temperature on various pumping conditions could bring us better understanding of the processes involved in setting up optimal conditions for the lasing action, and to help us increasing pump efficiency of these devices. During the experimental campaign at PALS Centre in 2006, we obtained time-averaged values of the electron temperature. These were measured from time integrated resonance spectra in the range of 8 to 12 Å yielded by KAP spectrometers using X-ray CCD camera as the detector.

114

J. Nejdl et al.

2 Experimental setup

Pumping beam

We used two nearly identical KAP crystal 2d=26.6Å spectrometers, equipped with a narrow slit covered by 6 or 18 µm Al filter, and a 16-bit X-ray CCD camera Andor DX440 with 2048×512 pixel resolution, cooled down to 10°C. The configuration of both spectrometers is outlined in Fig.1. An do

KAP crystal

rD x4 4

0

Andor Dx440

p Pre Target

Slit covered by Al filter

(Top view)

Target

e eb u ls

am

Main beam

(Side view)

a)

b)

Fig. 1. Setup of the spectrometers: a) short line plasma – the spectrometer slit is covered by 6µm Al filter, b) X-ray laser plasma– the slit is covered by 18µm thick Al filter. The line focus is orientated horizontally in both cases.

3 Examples of the obtained spectra

2p53d-2p6

2s2p63p-2s22p6

2p53s-2p6

9

9.5

10

10.5

λ [Å]

11

11.5

12

Fig. 2. Spectrum of the Ne-like Zn laser plasma and its lineout with calibrated wavelength axis, with some Ne-like lines labeled. The plasma is generated by a prepulse and the main pulse, the latter producing irradiance ~3x1013 Wcm-2.

Electron Temperature Measurements of Model Ne-Like X-ray Laser

115

Figure 2 shows a spectrum of a ~16-mm long section the zinc X-ray laser plasma (geometry corresponding to Fig.1b), created on a 3 cm long target by a weak prepulse and the main pulse separated by 10 ns; the intensity contrast of these pulses is ~7x10-4. Figure 3 shows examples of spectra generated by the second type of plasmas, produced using narrow slab targets (see Fig.1b) and different foci. a)

b) c) Fig. 3. Spectra of short zinc plasmas serving as testing system for X-ray laser amplifiers, produced by: a) Spot focus with 800µm diameter; b) 150µm wide line focus on 3mm long target; c) 150µm wide line focus on 1mm long target.

4 Technique of data analysis The electron temperature was estimated from the intensity ratio of two Nelike lines whose upper level populations are assumed to be mutually in thermodynamic equilibrium, while reabsorption of these lines in the plasma is similar. In this case, the intensity ratio of such lines is given by:

⎛ E − E1u I 2 A2 λ1 g 2 = exp⎜⎜ − 2u I1 A1 λ 2 g1 kTe ⎝

⎞ ⎟⎟ , ⎠

where A1, λ1, g1, E1u and A2, λ2, g2, E2u are the spontaneous emission probabilities, the wavelengths, statistical weights, and the energies of the upper level of each transition, respectively. In the below analysis, the Al filter transmissions and CCD spectral efficiency was further accounted for.

116

J. Nejdl et al.

5 Results We analysed the intensity ratio between the following spectral line pairs: 2s22p53s(3/2,1/2)J=1 → 2s22p6J=0 (λ=11.77Å)

(1)

2s22p53s(1/2,1/2)J=1 → 2s22p6J=0 (λ=11.52Å)

(2)

2s22p53d(3/2,5/2)J=1 → 2s22p6J=0 (λ=10.66Å)

(3)

and 2

5

2

2s 2p 3d(1/2,3/2)J=1 → 2s

2p6J=0

(λ=10.46Å)

(4)

Since the first pair of lines (3s-2p) is assumed to have non-negligible reabsorption, the intensity ratio I2/I1 of these lines is multiplied by a constant in order to adjust the deduced temperature to that obtained from the lines (3d-2p). For the X-ray laser plasma, this constant is 3/5, whereas for the plasma of test amplifier it amounts to 5/7. 5.1 Investigation of the plasma of the zinc X-ray laser Figures 4 and 5 show the electron temperature dependence on the prepulse energy, and the temperature along the ~16-mm section of the line focus. 60 50

T e [eV]

40 30

estimated from 3s-2p lines estimated from 3d-2p lines

20 10 0 1.7

1.9

2.1 2.3 2.5 Energy of prepulse [J]

2.7

2.9

Fig. 4. Electron temperature of the X-ray laser plasma as a function of the prepulse energy, estimated both from the 3s-2p and 3d-2p lines.

Electron Temperature Measurements of Model Ne-Like X-ray Laser

117

70 60

Te [eV]

50 40 30 20 10 0 0

2

4

6

8

10

12

14

16

Distance along the line focus z [mm]

Fig. 5. Electron temperature distribution along the line focus of the X-ray laser plasma, estimated from the 3d-2p lines.

5.2 Investigation of short plasmas relevant to X-ray laser amplifiers The temperature of 3-mm long zinc plasmas, produced by sequence of a weak prepulse (contrast 7x10-3) followed after 5.5 ns by the main pulse, is ~4 eV higher than in plasmas produced by the main pulse alone. Figure 6 shows lateral distribution of the temperature for both the prepulse-main pulse sequence and for single pulse pumping. 40

Te [eV]

30

20

3s-2p, main pulse only 10

3d-2p, main pulse only 3s-2p, prepulse + main pulse

0 1.0E+13

1.5E+13

2.0E+13

2.5E+13

3.0E+13

Main pulse intensity [Wcm-2]

Fig. 6. Electron temperature of the 3mm zinc plasmas depending on the main pulse intensity; the data include both prepulse and prepulse-free shots.

118

J. Nejdl et al.

Fig. 7. Lateral electron temperature distribution (across the line focus) of short line plasmas, estimated from the 3d-2p lines. Shown are cases of prepulse pumping (#27_0804, 3mm plasma) and single pulse pumping (#13_0713, 1mm plasma).

6 Discussion The higher electron temperature of plasmas when using prepulse (see Fig. 6) is likely to be attributed to more efficient absorption of the main pulse. The upper level of line 2 of our analysis, 2s22p53s(1/2,1/2)J=1, is the lower level of the lasing transition 2s22p53p(1/2,1/2)J=0 → 2s22p53s(1/2,1/2)J=1. In consequence, the thermodynamic equilibrium between the upper levels of line 1 and 2 could be perturbed, resulting in higher electron temperature deduced than the actual value; however, this effect is assumed to be negligible. It turns out that the experimentally estimated values of time-integrated electron temperature correspond fairly well to the electron temperature of the corona, given by simulations done by the EHYBRID code and followed by a special postcode. The compared time-integrated value is calculated by

T ( x) =

∫ T ( x, t ) ⋅ n ( x, t )dt , ∫ n ( x, t )dt e

e

e

where Te and ne are the electron temperature and the electron density.

Electron Temperature Measurements of Model Ne-Like X-ray Laser

119

Acknowledgments This work benefited from support of the Czech Science Foundation grant No. 202/05/2316, from the Centers of Fundamental Research project LC528 of the Czech Ministry of Education, and from NEST Specific Targeted Research Project of the EU, contract No. 12843. We are grateful to G. J. Pert (University of York) for providing us the EHYBRID code.

References 1. Rus, B. et al: Investigation of Zn and Cu prepulse plasmas relevant to collisional excitation x-ray lasers, Physical Review A 56, 4229, 1997 2. Zhang, H. et al: Collision Strengths and Oscillator Strengths for Excitation to the n=3 and 4 Levels of Neon-like Ions, Atomic Data and Nuclear Data Tables, Vol. 37, p. 17, 1987.

Development of Plasma X-Ray Amplifiers Based on Solid Targets for the Injector-Amplifier Scheme M. Kozlová1, B. Rus1, T. Mocek1, J. Polan1, P. Homer1, M. Stupka1, M. Fajardo2, D. De Lazzari2 and P. Zeitoun3 1 - Department of X-ray Lasers, Institute of Physics/ PALS Centre, Prague 8, Czech Republic 2 - Centro de Física dos Plasmas, Instituto Superior Técnico, Lisbon, Portugal 3 - Laboratoire d'Optique Appliquée, ENSTA - Ecole Polytechnique, Palaiseau, France

Summary. Results of experimental studies aimed at generation and diagnostics of advanced soft X-ray amplifiers, produced from solid targets, are presented. 2D profiles of electron density of short plasma columns, generated by ~300-ps laser pulses under various illuminating conditions, were investigated by near-field distribution of the plasma self-emission, and by X-ray laser backlighting at 21 nm, accessing in the given geometry electron densities of 1022 cm-3. The obtained data indicate that by employing line focus with concave intensity profile it is possible to generate laterally highly uniform plasma columns of width ~500 µm, potentially suitable as amplifiers with negligible lateral refraction. By X-ray laser backlighting we further probed the morphology and gain region of test Zn plasmas, pumped by a sequence of a loosely focused weak prepulse and tightly focused main pulse, separated by 5.5 ns. The data clearly show the beneficial role of the prepulse in lateral homogenization of the plasma, and reveal narrow ~50-µm gain region.

1 Introduction The scheme in which HHG seed is strongly amplified in a separated amplifier was demonstrated using gas-cell OFI plasmas, at the wavelength 41.8 nm [1]. In this approach, the energy extractable from the amplifier is limited by the saturation intensity, and therefore boosting the output beyond µJ level requires denser plasmas produced from solid targets [2]. Feasibility of this approach was recently demonstrated by amplification of HHG seed in Ne-like Ti plasma at 32.6 nm [3]. In this work, we studied 2D transverse density profile of line plasmas potentially exploitable as X-ray amplifiers, at conditions relevant to the

122

M. Kozlová et al.

pedestal pulse in transient X-ray lasers. Plasmas produced using conventional “flat top” line focus, as well as using laterally modulated illumination profile, were examined. The density profiles were investigated by high-resolution imaging of the XUV self-emission, and by backlighting with X-ray laser at 21.2 nm [4]; the latter technique is able accessing ne up to 2.4×1024 cm-3, though in practice this value is lower due to refraction of the X-ray beam. The ionization balance and electron temperature of the plasmas was measured by resonance keV spectroscopy, the results of which are presented elsewhere in these Proceedings [5]. In order to probe dense and emissive plasma by X-ray laser backlighting, spatial filtering in the X-ray beam surrogate focal plane was employed. This technique allows exploiting the directionality of the X-ray laser beam and largely reduces contribution of self-emission of the probed plasma into the detected signal.

2 Experimental setup The experiment was conducted at the PALS Centre, benefiting for the backlighting purposes from availability of the neon-like zinc X-ray laser at 21.2 nm. The investigated plasmas were produced by the auxiliary 148-mm

Fig. 1. Experimental scheme. Targets consisting of 1- or 3-mm wide slabs are irradiated either by a single pulse or by sequence prepulse / main pulse produced in the dogleg. The pulses are focused down to a ~150 µm × 4 mm line by f=120 cm spherical / f=400 cm cylindrical lens doublet; the prepulse is separately defocused by additional f=3000 cm lens. The X-ray laser at 21.2 nm, upon probing the plasma, is reflected by f=253 mm off-axis MoSi multilayer parabolic mirror and passes though a pinhole located at the focal plane of the mirror.

Development of Plasma X-Ray Amplifiers Based on Solid Targets

123

diameter beam, delivering 80 J at the fundamental wavelength (1.315 µm) in 300 ps (FWHM) pulses. A doublet of spherical f=120 cm and cylindrical f=400 cm lenses is employed to produce 4-mm long line focus (with minimal lateral size ~150 µm) overfilling 1 or 3 mm slab targets. The investigated plasmas are produced either by a single pulse or by sequence of a weak prepulse followed by the main pulse. In the latter case, dogleg arrangement (see Fig. 1) was employed: the incident beam is split by a 10/90 % (T/R) beamsplitter into two beams that are recombined back by a 20/80 % beamsplitter. The dogleg makes it possible changing the delay between the prepulse and the main beam from 2 to 10 ns. Additional lens in the optical path of the prepulse is employed to produce prepulse focus larger than the focus of the main pulse. The studied plasmas were further monitored by a keV spectrometer [5] and their dimensions were measured by means of VIS emission using an optical telemicroscope. The output plane of the plasma was imaged to a back-illuminated X-ray CCD camera PI-MTE (2048×2048, 13.5 µm pixels). The imaging was carried out by off-axis parabolic (f=253 mm) MoSi multilayer mirror, with magnification 8. A small pinhole (100 µm to 1 mm) was placed in the focal plane of the X-ray laser, acting effectively as spatial filter strongly improving the contrast of the plasma self-emission to the X-ray laser probe.

3 Experimental results In chapter 3.1, we show the effect of spatial filtering on the resolution of the imaged object, and demonstrate its ability to attenuate plasma emission. Results of probing plasmas that were produced comparatively by narrow line focus and by a “hollow” line focus are shown in chapter 3.2. In chapter 3.3, zinc plasmas produced with and without weak prepulse are investigated, and from local amplification/absorption information about the gain region is obtained. 3.1 Spatial filtering in X-ray laser plasma probing A simplified principle of this technique is shown in Figure 2a. Its use needs appropriately selected pinhole size, adequate to the interval of spatial frequencies pertinent to the details at the object to be imaged. The required pinhole size may be estimated using equation 1 (assuming paraxial approximation), adopting here an arbitrary criterion of at least three

124

M. Kozlová et al.

Fig. 2a. Principle of spatial filtering: rays of a (near) parallel beam pass freely through a pinhole, located in the focal plane of the lens/mirror, while majority of (near) isotropic emission from the probed object is stopped by the pinhole.

Fig. 2b. Intensity distribution in the focal plane of the probing 21.2-nm beam backlighting a rectangular obstacle of width D=1 mm.

maxima transmitted. In this experiment, D=1 mm, f =253 mm, λ=21.2 nm, which gives intensity distribution displayed in Figure 2b. I(y) = I0 sinc2(πDy/λf)

(1)

The first minimum corresponding to y =50 µm, the target edge appears blurred when 100-µm diameter pinhole is employed (Fig. 3a). When the pinhole size is increased to 300 µm so as to pass three diffraction maxima, sharp image of the target edge is produced (Fig. 3b). Reduction of the plasma self-emission, resulting in dramatic improvement of the contrast probe /plasma emission at the detector, is demonstrated in Figure 4. Here, a 300-µm pinhole placed in the focus of the imaging parabola, removes completely the plasma self-emission. The presence of the plasma is apparent only by the bent contours of the filtering mesh that acts here as a casual sensor of wavefront distortion of the probing X-ray beam.

Development of Plasma X-Ray Amplifiers Based on Solid Targets

125

Fig. 3. Effect of spatial filtering on the image of the target backlighted by X-ray laser at 21.2 nm: (a) 100 μm hole removes high-frequency details of the target edge, (b) 300 μm hole appropriately bandpassing target edge details (the chequered pattern is due to the mesh supporting thin Al filter in front of the CCD).

Fig. 4. Plasma column with length of 1 mm, produced with ~30 J and probed by the X-ray laser at 21.2 nm (the chequered pattern corresponds to a fine mesh supporting the Al filter located ~50 cm in front of the CCD camera): (a) shot taken without spatial filtering; (b) shot with spatial filtering, using a 300-µm diameter pinhole. The distorted contours of the mesh near the target are due to refraction of the XRL beam in the probed plasma; the contours are redrawn in the bottom image.

126

M. Kozlová et al.

3.2 Shaping the lateral plasma profile of the amplifier In searching for generation of laterally uniform line plasmas with width much larger than 100 µm, we studied comparatively target irradiation by the “conventional” narrow line focus, and by laterally concave (with intensity dip on the axis) line focus. The data, obtained with Fe targets, are shown in Figure 5. For the narrow focus, producing 4×1013 Wcm-2 along a single ~150 µm line, the coronal plasma XUV emission roughly reproduces the lateral profile of the driving intensity. For the concave focus, producing two 13 -2 ~1.6×10 Wcm peaks ~500 µm apart with intensity modulation axis/peak of ~3, the modulation of the XUV intensity, and hence modulation of the coronal plasma, is smaller than modulation of the driving intensity.

Fig. 5. Lateral profiles of the XUV self-emission of a 1-mm long Fe plasma produced by (a) narrow line focus, and (b) hollow focus. The bottom images show the front VIS plasma emission visualizing spatial distribution of the driving intensity (note different scales in the lateral and axial directions).

Both plasmas were axially probed by the X-ray laser, accessing large electron densities near the target. The obtained data are shown in Figure 6. The probed plasmas appear here as thin dark regions from which the X-ray laser is “pushed out”, and which extend beyond both sides of the target edge original position. These dark regions extend for the narrow and for the concave focus respectively 18 µm and 13 µm above the target surface. Most importantly, the data reveal that the concave focus produces laterally very homogeneous plasma of width ~500 µm, in stark contrast to the narrow

Development of Plasma X-Ray Amplifiers Based on Solid Targets

127

focus geometry resulting in laterally rapidly varying density profile. This indicates that plasmas employing concave line focus represent a promising approach for producing laterally refraction-free X-ray laser amplifiers.

Fig. 6. Images of the 1-mm long Fe plasmas, backlighted along their axis by the 21.2-nm X-ray laser: (a) plasma produced by the narrow focus, (b) plasma produced by the hollow focus. Both images were obtained using a 1-mm pinhole. The images are shown with different horizontal and vertical scale to improve the data visibility.

As Fe plasma is highly transparent at 21.2 nm even at low temperatures [6], we attribute the dark region to refraction of the X-ray beam. To assess the electron density, we used ray tracing to follow rays propagating near the target. This tracing assumes exponential electron density fall-off, ne=ne0exp(-x/L), where ne0 is density to be inferred, and L is adopted to have a value compatible with the extent of the detected plasma (> E S then we obtain the maximum energy available in the extraction process E AVAIL = g 0VE S , where V is a volume of the active/amplifying medium. This means, that the saturation fluence and its relation to the input energy fluence determine the energy extractable from the medium. Simple estimates, assuming an amplifying medium with the dimensions 5 mm×20 µm×20µm and a volume of 2×10-6 cm3, show that the input energy fluence necessary for efficient energy extraction should correspond to the total energy about 100 nJ. This is at the moment not achievable in the spectral range of our interest between 10 and 20 nm.

2. MBI project Analysis of the situation and the development prospects caused the MBIgroup to pursue its own way towards a beamline-like facility available to external users. It was decided to focus the attention on Ni-like soft X-ray laser working at 13.9 nm. The expected output parameters are collected in Table 1. The work towards this final goal is organized in two topics. The first one is experimental work on the dynamics and kinetics of XRL active medium irradiated in

On the Way Towards a High Repetition Rate X-Ray Laser

165

Table 1 Output parameters of the 100 Hz XRL planned at MBI

Parameter Pulse length/repetition rate Wavelength Coherence (spat./longit.)

Energy Peak/average brightness/ ph/(s*mrad*mm2*0.1%Bw Linewidth

Injector - amplifier project ≤ 1 ps/ 100 Hz 10 - 20 nm full/~0.5 mm 0.1 mW ~ 5 - 10 µJ 1026 /1015 300 fs). The general architecture of the Ti:Sa laser is schematically represented in figure 2.

Fig. 2. Schematic view of the LASERIX driver architecture

194 D. Ros et al.

The front-end is designed as a customized laser based on standard modules, developed by French companies (THALES LASER and AMPLITUDE TECHNOLOGIES). It is composed by two parts, one for the shaping and pre-amplification of the oscillator pulse, the other for two cryogenic amplifiers. The output energy at the front-end is more than 2J at the 10Hz repetition rate. The front-end beam (2.5 J) is then injected in the main amplifier, which is composed by the large Ti:Sa crystal (diameter 100 mm), shown in figure 3. The crystal is pumped by a 4-module Nd:glass laser delivering 100 Joules of 2ω green light, developed by the French laser company QUANTEL. The energy deposition on each side of the crystal is homogenized using lens arrays. The crystal is held in a mount in which a special liquid is circulating all around to cool the crystal and limit the transverse lasing. After 4 successive passes through the crystal, the expected output before compression is ≅ 40 Joules at the repetition rate of 10 Hz. Basically, as shown in figure 2, the 40-joule beam is divided in two parts, respectively 20 Joules of 500 ps and 10 Joules of 50 fs-1 ps (after compression). Besides, two more beams are offered at the final stage. Thus, the zero-order rejected by the compressor may be itself compressed to give a beam of ≅ 1 J in 50 fs. Besides, a weak part of the energy at the exit of the front-end, ≅50 mJ in 50 fs at the repetition rate of 10Hz, can be offered to the users

Fig. 3. The large titanium-doped sapphire crystal (diameter : 100 mm) of the LASERIX driver amplifier is shown on the left. On the right image, we can see the crystal and its mount, including a liquid all around the mount.

LASERIX: A Multi X-Ray/XUV Beamline High Repetition-Rate Facility

195

4 Description of the beam-lines possibility Owing to our choice of the Ti:Sa technology for the driver, LASERIX will offer the largest experimental possibilities. Thus, as shown in figure 4, the main beams represented in the part (a) give a long pulse and a short pulse leading to TCE X-ray lasers at 0.1 Hz repetition rate. Beside these two beams, several infrared beam-lines will be useable for applications in the IR range and/or for producing auxiliary XUV beamlines. For instance, the zero-order issuing from the compressor will be itself compressed to produce an auxiliary beam of 1 Joule in 50 fs, as shown in part (b) of figure 4. In addition the energy of the uncompressed beam will be generally larger than needed, so a part of it will be available. Finally, leaks taken in the front-end offer the possibility to have other auxiliary IR beam-lines, as shown in part (c) of the figure 4. Such beams can be used again for applications at 10Hz repetition rate. But, particularly they offer possibility to produce XUV beams by high-order harmonic generation (HHG). Such beams may be amplified in TCE plasmas to produce intense XRL beams of high optical quality [10]. All these beam-lines open a large field of applications, especially because they all are synchronized with the main beam

Fig. 4. Configuration of the main beam-lines in the experimental area.

196 D. Ros et al.

5 Applications of X-ray lasers and LASERIX project X-ray lasers can be employed to excite matter, to diagnose solid surfaces, or probe high-density plasmas. Indeed, the unique XRL properties (very large brightness, coherence, short duration) add to good beam collimation, and make XRLs attractive with respect to the other soft XUV sources. Owing to their high brightness, XRLs are well suited as diagnostic tools for various purposes, in particular in microscopy and interferometry [11,12,13]. As a widely used diagnosing tool in the research of laser-produced plasma, interferometry has many advantages in the accurate measurement of the plasma electron density because it directly gives refraction index mapping from the interference pattern processing. Indeed, due to the high intensity and short wavelength, X-ray lasers applied as a probing beam for the interferometry provides better penetration into high density plasma (1019–1022 cm-3) with small refraction, which is a main limitation of interferometry in the UV range for high-density gradient plasma diagnosis. In this paper, we just ullustrate the interest of X-ray lasers for laserproduced plasmas investigations from the results of an X-ray interference microscopy diagnosis using a wavefront-division imaging Fresnel bimirror interferometer [14] and a transient picosecond Ni-like Ag X-ray laser [15].

Fig. 5. Experimental set-up of the Fresnel bi-mirror X-ray interference microscopy.

The configuration of the experimental setup is shown in Figure 5. The X-ray laser illuminated the sample plasma and the bi-mirror separated the

LASERIX: A Multi X-Ray/XUV Beamline High Repetition-Rate Facility

197

wave front after the plasma and produced an enlarged overlapped interference field. The diffraction-limited ellipsoidal imaging mirror gave a magnified image of the sample plasma onto the thinned backlighted CCD. The spatial resolution in the sample plasma plane was limited by the magnification and the pixel size to ~ 1.75 µm. The diagnosed plasma was formed by irradiating a 1 mm-long Al slab target with a 1 J, 1.2 ns IR (1053 nm) pulse. A line focus system produced a plasma column of ~ 100 µm in width and 6 mm in length (intensity within the line focus: 1.4×1011 Wcm-2). The X-ray laser probed the plasma along the column axis. The intense plasma self-emission was observed close to the target surface and mixed with the interference fringes. However this low frequency signal can be removed by using a Fourier filtering in order to let appear the interference field. The density map may be then easily deduced from the fringe pattern, as shown by figure 6, which describes the successive steps of the analysis leading from the raw data to the electron density map. For the plasma length of 1 mm, the electron density, ne, is related to the fringe shift as ne ≈1.583×1020 δfringe [cm-3]. Due to the limitation of the spatial resolution, the minimum detectable limit of the electron density is ~ 3×1019 cm-3.

Fig. 6. Main steps of results and analysis of laser-produced plasmas investigations. 1) Image of the sample plasma, reflected by the reference part of the bimirror, without X-ray laser. 2) Interference field with superimposed plasma image, recorded for the same shot. 3) Interference field after removing plasma emission by a Fourier treatment. 4) Electron density map deduced from 3.

Pictures of the figure 6 correspond to a delay of 1 ns between the plasma heating pulse and the X-ray laser. The peak electron density measured is 1020 cm−3 at the position comprised between 10 and 15 μm away from the target surface. In some zones of plasmas diagnosed later, typically 2 and 3 ns after the peak of the heating pulse, inverted fringe shifts were observed, leading to refraction index larger than 1. That means that the plasma is no more a classical dilute electron gas and that the refraction in-

198 D. Ros et al.

dex is not mainly governed by the electron density, but by free-bound and bound-bound transitions. Thus, these results will contribute to the validation of the 1D and 2D hydrodynamic codes and a better understanding of the physics of laserproduced plasma [16]. They could be performed in a near future using LASERIX facility.

6 Summary and perspectives. Thus, the LASERIX facility is now built. The laser driver is running and the choice of the Ti:Sa is experimentally successful confirmed. It will offer the possibility to highly increase the repetition rate of the X-ray lasers and then will be very useful for applications. A virtual (3D) of the experimental area is represented in Figure 7.

Fig. 7. 3D virtual view of the LASERIX facility. The main chamber producing TCE X-ray lasers is represented in the center of the room.

This laser facility will be located in a new building, available in November 2006, at the Laboratoire d’Optique Appliquée (Ecole Polytechnique, Palaiseau, France). Thus, the LASERIX facility will be installed during the first semester of 2007 and first applications investigations will be performed during the second semester of 2007. Among the different applications, we plane to purchase the density diagnostics of plasmas produced by the low-energy beam. The possibility to vary the duration of the IR beam producing the plasma, to vary the wavelength of the X-ray laser,

LASERIX: A Multi X-Ray/XUV Beamline High Repetition-Rate Facility

199

and, above all, the large repetition rate (0.1 Hz) will be very useful to realize full-scale studies. That has been demonstrated at the 1st LASERIX Workshop [17], held in Orsay on 2-3 February 2006.

7 Acknowledgments LASERIX is an X-ray laser facility of the Université Paris-Sud. The financial support of the Conseil Général de l’Essonne and the Ministère de la Recherche under the Contrat de Plan Etat-Régions 2000-2006 is gratefully acknowledged. We are indebted to Patrick Georges (LCFIO), Pascal d’Oliveira (CEADRECAM) and Claude Rouyer (CEA-CESTA) for valuable discussions.

References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

D.L. Matthews et al, Phys. Rev. Lett. 54, 110, 1985 L.I. Gudzenko et al., Sov. Phys. Doklady 10, 147, 1965 R.C. Elton, Appl. Optics 14, 97, 1975 D. Ros et al, Laser ans Particle Beams Journal, 20, 23,.2002 B. Rus et al., Phys. Rev. A 55, 3858-73, (1997) S. Sebban et al., Phys. Rev. Letters, 86, 3004-7, (2001) P.V. Nickles et al., Phys. Rev. Lett. 78, 2748, 1997 A. Klisnick et al., J.O.S.A. B 17, 1093, 2000 G. Jamelot et al., in X-Ray Lasers 2004, IOP Conf Series N° 186, 677 (2005) P. Zeitoun et al., Nature 431, 426, (2004) D. Di Cicco et al., Opt. Lett. 17, 157, (1992). R.E. Burge, et al., Opt. Lett. 18, 66, (1993). D.H. Kalantar et al., Phys. Rev. Lett. 76, 3574, (1996). D. Joyeux et al., J. Phys IV France 11, Pr2, 511, (2001). A. Klisnick et al, J.O.S.A. B 17, 1093, (2000). H. Tang et al., Appl. Phys. B 78, 975, (2004). The CD ROM of contributions to the workshop is available : [email protected]

Seeding High Order Harmonic in a Transient X-Ray Laser Amplifier I. R. Al’Miev*, O. Larroche+, A. Klisnick*, C. Moller*, D. Benredjem*, S. Kazamias-Moucan*, O. Zabaydullin* and J. Dubau* *)

LIXAM, Laboratoire d’Interaction du rayonnement X Avec la Matiere, UMR 8624, Bat. 350, Universite Paris-Sud, 91405 Orsay, Cedex, France +) CEA-DIF, Boite postale 12, F-91680 Bruyeres-le-Chatel France

Summary. The seeding of high-order harmonic (HOH) radiation in a transient Xray laser amplifier was numerically investigated using the code COLAX. The code was recently modified to account (i) for the travelling-wave irradiation of a preformed plasma pumped in the transient regime, and (ii) for the injection of a HOH pulse at the X-ray laser wavelength, and at one boundary of the plasma amplifier. We show that the time-dependent description of the polarisation (TDP) leads to significantly different results in comparison to the widely used adiabatic description. Namely, with TDP, HOH produces a wake in the intensity distribution, which is not only amplified but lasts much longer than the HOH pulse. This could explain recent experimental observations in the spectral domain.

1 Context and motivation It is well known that a conventional optical laser exhibits few principal problems associated, firstly, with necessity of creating a high quality seed, secondly, with maintaining the quality over the amplification region, and, thirdly, with reducing the level of the self-emission of the amplifier. The area of X-ray lasers was investigated in a number of papers. Starting from the first successful demonstration of X-ray laser, the world has seen the lasing even at the wavelength 3.5 nm [2] that is just inside the waterwindow. One of new promising routes of the X-ray laser research is the amplification of HOH pulse inside the X-ray laser. The importance of this topic could be explained by the potential improvement of the properties of X-ray laser, for example, coherence. One of the first reports on the ampli-

202 I. R. Al’Miev et al.

fication of the HOH was published by Ditmire et al [3]. In their work they reported an amplification of the 21st harmonic of neodymium chirpedpulse-amplification laser pulse in Ne-like Ga laser plasma. Substantial progress has been made by Ph.Zeitoun et al [4] in 2004 with successful demonstration of seeding HOH in Ni-like Kr OFI amplifier at a wavelength of 32.8 nm. The authors claimed an amplification of 25th harmonic from Ar gas by a factor of 103. We note that the work in similar area has been made by Japanese scientists Kawachi et al [5], and Hasegawa et al [6]. They made an attempt to produce shorter wavelength radiation using transient collisional excitation (TCE) scheme. If one would like to obtain larger energy output and reach shorter lasing wavelengths then, indeed, TCE X-ray lasers, based on the use of solid target, is a valuable medium for the potential amplification. Our goal is to simulate propagation of short high-order harmonic pulse within Ni-like Ag laser-produced plasma created by the traveling wave irradiation. We note that the HOH pulse differs from X-ray laser pulse in that the spectral width of HOH is much broader than that of the X-ray laser pulse. That is why we have used and upgraded the code COLAX that is appropriate to resolve numerically short-, of order of 10 fs, and long-duration, of order of 10 ps, electric field amplitudes. Conditions of the simulations match the experimental ones reported by Klisnick et al [7].

2 Simulated experiment To perform the simulations we considered such scheme where Ag target is initially irradiated by the pre-pulse of smaller intensity, creating a preplasma. Then, it is irradiated by the main pulse of larger intensity, creating conditions that are optimal for the population inversion. Main pulse takes the form of the travelling wave irradiation. The angle of the travelling wave (TW), that is related to the velocity of TW irradiation, can be varied in the present version of the code; for the particular calculations we fixed it as 45° that corresponds to the velocity of the TW, which is equal to the speed of the light in the vacuum. As soon as the amplified spontaneous emission is generated, the HOH pulse is injected. It interacts with the laser-plasma amplifier, and produces some radiation output. To simulate Nilike Ag laser-produced plasma we used the code EHYBRID [8]. The following parameters were used as an input into EHYBRID. Intensity of the pre-pulse was taken to be 1012 W cm-2; intensity of the main pulse was 1015 W cm-2; the duration of the pre-pulse was 600 ps; the duration of the main pulse was 400 fs; the time interval between the pre- and main pulses was

Seeding High Order Harmonic in a Transient X-Ray Laser Amplifier

203

250 ps. The HOH pulse has Gaussian temporal and spatial shape of 15 fs temporal and of 30 µm spatial widths. The time of the injection of HOH pulse is a parameter in the code – it can be varied. For the particular calculations we fixed it as 2 ps with respect to the start of the travelling wave irradiation.

3 The code COLAX: present status and upgrade The code COLAX was originally written by Larroche et al. and has been used to model X-ray laser signal build-up [1]. The evolution of the electric field is described by Maxwell equation, which is transformed into paraxial envelope form,

∂E ± ∂E ± ic ∂ 2 E ± iω + = + (ε R E ± + 4πP± ) , c ⋅ ∂t ∂y 2ω ∂x 2 2c

(1)

where E+ and E- are the complex, slowly varying amplitudes of waves propagating in the positive and reverse directions. The Eq.(1) is selfconsistently bound with the equation describing the evolution of the polarisation,

∂P± = −γP± − iωDE ± + Γ± , ∂t

(2)

where γ has a meaning of the dipole de-phasing factor, and D is the reduced, non-dimensional population inversion between the upper and the lower X-ray lasing levels,

D=

d2 1 ( ρu − ρl ) = γGλ , hω 8π 2ω

(3)

where d is the absolute value of the non-diagonal matrix element of the dipole operator, ρ u ( ρ l ) are the dimensional populations of the upper, and lower levels, respectively, and the gain G takes into account the effect of the saturation,

G=

G0 1+

I

,

(4)

I sat

with G0 describing the small-signal gain, I is the electric field intensity, and Isat is the saturation intensity. Noise term, Γ+ , in the Eq.2 effectively describes spontaneous emission. To perform simulations we have upgraded COLAX. Travelling wave irradiation was taken into account in the following way. Local conditions of

204 I. R. Al’Miev et al.

plasma were determined by the angle of the travelling wave and EHYBRID time-history of plasma. EHYBRID is 1.5 dimensional code. It provided us with the cell position, electron density, electron and ion temperatures, populations of the upper and lower X-ray lasing levels, smallsignal gain, dipole de-phasing factor, and the saturation intensity as functions of the Lagrangian cell positions and time. By fixing plasma conditions along transverse axis at fixed y-coordinate, conditions at other ycoordinates are calculated, using space and time interpolation procedure, taking into account time delay of the front irradiation of the travelling wave. Then, such interpolated quantities are used as an input parameters in the code COLAX. To account for the HOH pulse injection boundary conditions as functions of time were modified.

4 Results Typical output of COLAX is shown in Fig.1. It shows the snapshot of intensity distribution as a function of the longitudinal, y, and transverse, x, distances at t = 8 ps. Longitudinal axis is the one, along which the X-ray radiation propagates. It has dimension of 3mm. The dimension of the

Fig. 1. Intensity distribution as a function of y- and x-coordinates at t = 8 ps with respect to the start of the travelling wave irradiation.

Seeding High Order Harmonic in a Transient X-Ray Laser Amplifier

205

transverse axis is 140 µm, while the initial position of the target surface is located at x = -70 µm . Intensity is given in log-scale. The narrow feature at y = 1.85 mm corresponds to the HOH pulse. The HOH pulse has a Gaussian spatial profile along the x-axis. The peak of the HOH profile is at x = -33 µm. The Fig.1 refers to the case when the electric field wave propagates towards the plasma output in “+”-direction, that is

I + = c E+

a)

2

/(8π ) , where c = 3 ⋅ 1010 cm ⋅ s −1 .

b)

Fig. 2. Intensity snapshots at a) t = 4.5 ps and b) t = 11.5 ps

This figure allows making important conclusions. Firstly, the HOH pulse is weakly amplified. Secondly, high-order harmonic induces the wake, or the tail, that is apparent behind the pulse. Fig.2 shows the evolution of the intensity at two successive times, as the electric field propagates towards the plasma output. It is clear, that the amplification of the intensity takes place behind the HOH pulse in the area with 10-20 µm transverse range where the gain has the maximum values. Secondly, the length of the amplified area is not conserved – it becomes longer as the pulse propagates further. To understand the nature of such distribution we ran COLAX including only the HOH pulse, while arbitrarily setting spontaneous emission to zero. Results are shown in Fig. 3a. One can see that, indeed, the HOH pulse induces the tail. In Fig 3b no HOH was injected, i.e. only spontaneous emission is amplified. Comparison of Fig.1 and Figs.3 shows that the

206 I. R. Al’Miev et al.

a)

b)

Fig. 3. Intensity distribution that includes a) only the HOH pulse, and b) only amplified spontaneous radiation; t = 8 ps.

Fig. 4. Intensity distribution assuming adiabatic model of COLAX calculations; t = 8 ps.

Seeding High Order Harmonic in a Transient X-Ray Laser Amplifier

207

total contribution of the HOH and the ASE leads to the little increase of the intensity in the area where the intensity has a maximum. Present version of the code COLAX allows making the electric field, and, respectively, intensity calculations in the adiabatic approximation, where the left hand side of the Eq.(2) equals to zero, that is

∂P+ = 0 . In ∂t

such approximation the polarisation is proportional to the electric field, and to the population inversion, according to the following equation,

P+ =

χ ω E + = −i DE + , 4π γ

(5)

where ω is the X-ray laser frequency. Such an approximation is widely used, and in particular it is implicit in the numerical codes based on radiative transfer description. Running COLAX in the adiabatic approximation mode we obtained significantly different results, both qualitatively and quantitatively. Fig. 4 shows the intensity snapshot calculated using this model, that is to be compared to Fig. 1. Firstly, the high-order-harmonic pulse is much more amplified in comparison to the time-dependent case. This can be seen in the centre of the pulse – the HOH pulse wings in the transverse direction have larger intensity values than in the case of the TD description. Secondly, the adiabatic approximation does not predict the existence of the electric field wake. Thirdly, the overall shape of ASE area differs from that of time-dependent calculations. Fourthly, the region of the maximum intensity is situated in the ASE feature. It is 20 times larger than that of the time-dependent case.

5 Conclusion In conclusion, we incorporated the travelling-wave model into the code COLAX. High-order harmonic seed was included as time- and spacedependence of the boundary conditions. Calculations demonstrate that, firstly, it is important to include time-dependent treatment of the polarisation, secondly, the high-order harmonic pulse leads to the wake that is amplified in time, and, thirdly, the wake lasts much longer than the HOH pulse. That can explain the experimental observations [4] in the spectral domain: namely, that, contrary to common belief, the HOH pulse is not amplified as a short, broadband pulse, but rather triggers a long-lasting (and thus possibly narrow-band) wake where most of the XRL energy goes. This fact is the principal difference between time-dependent and adiabatic models. Optimisation of the injection of the high-order harmonic

208 I. R. Al’Miev et al.

pulse and the extension of the population inversion calculations using time-dependent rate equations will be the asset in other paper.

References 1. Larroche, O., Ros, D., Klisnick, A., Sureau, A., Möller, C., Guennou, H., “Maxwell-Bloch modelling of x-ray-laser-signal buildup in single- and doublepass configurations”, Phys. Rev. A 62, 043815 (13 pages), 2000. 2. MacGowan, B.J., Maxon, S., Hagelstein, P.L., Keane, C.J., London, R.A., Matthews, D.L., Rosen, M.D., Scofield, J.H., Whelan, D.A., “Demonstration of soft x-ray amplification in nickel-like ions”, Phys. Rev. Lett. 59, 2157-2160, 1987. 3. Ditmire, T., Hutchinson, M.H.R., Key, M.H., Lewis, C.L.S., MacPhee, A ., Mercer, I., Neely, D., Perry, M.D., Smith, R.A., Wark, J.S., Zepf, M., “Amplification of xuv harmonic radiation in a gallium amplifier”, Phys. Rev. A 51, R4337-R4340, 1995. 4. Zeitoun, Ph., Falvre, G., Sebban, S., Mocek, T., Hallou, A., Fajardo, M., Aubert, D., Balcou, Ph., Burgy, F., Douillet, D., Kazamias, S., de Lacheze-Murel, G., Lefrou, T., le Pape, S., Mercere, P., Merdji, H., Morlens, A.S., Rousseau, J.P., Valentin, C., “A high-intensity highly coherent soft X-ray femtosecond laser seeded by a high harmonic beam”, Nature 431, 426-429, 2004. 5. Kawachi, T., Nagashima, K., Kishimoto, M., Hasegawa, N., Tanaka, M., Ochi, Y., Nishikino, M., Kawazome, H., Tai, R.Z., Namikawa, K., Kato, Y., “Recent Progress in X-ray Laser Research in JAERI”, Proc. of SPIE 5919, 59190L (11 pages), 2005. 6. Hasegawa, N., Kilpio, A.V., Nagashima, K., Kawachi, T., Kado, M., Tanaka, M., Namba, S., Takahashi, K., Sukegawa, K., Peixiang, L., Huajing, T., Kishimoto, M., Renzhong, T., Daido, H., Kato, Y., “Higher harmonics generation for the high coherent x-ray laser”, Proc. of SPIE, 4505, 204-210, 2001. 7. Klisnick, A., Guilbaud, O., Ros, D., Cassou, K., Kazamias, S., Jamelot, G., Lagron, J.-C., Joyeux, D., Phalippou, D., Lechantre, Y., Edwards, M., Mistry, P., Tallents, G., “Experimental study of the temporal coherence and spectral profile of the 13.9 nm transient X-ray laser”, JQSRT 370-380, 2006. 8. Pert, G.J., “Model calculations of XUV gain in rapidly expanding cylindrical plasmas”, J. Phys. B 9, 3301-3315, 1976.

Feasibility of 3.4 nm Laser Pumped by Ultraintense RBS Laser S. Suckewer, Y. Avitzour*, W. Cheng, J. Ren and S. Li Princeton University, Princeton, NJ, 08544, USA

Summary. Our presentation consisted of two parts. In the first part we presented main theoretical results on the plasma and pumping conditions required to generate gain at 3.4 nm in H-like CVI ions in transition from the first excited state to ground state. Transient population inversion is generated during the recombination process. It was shown that high gain (up to G ~200 cm-1) can be achieved using currently available compact lasers. In the second part we presented a new type of compact laser generating ultra-short and ultra-intensive pulses via Raman Backscattering (RBS) amplification and compression in plasma. We achieved large (up to 1000) “seed” amplification and its compression from ~1 psec down to 150 fsec in only 2 mm long plasma pumped with ~ 1014 W/cm2 pulses. We also presented very recent experiments on amplification in 2 passes setup. Such RBS amplifier and compressor is expected to provide in not too far future intensities ~1020 W/cm2 at high repetition rate in compact (university type) system, which would be ideal pump not only for 3.4 nm laser but even for shorter wavelength lasers.

1. Introduction Remarkable progress has been made in the development of soft x-ray lasers (SXLs) since their first demonstration in 1984 in so called collisional scheme [1] and recombination scheme [2]. Each of these schemes evolved into sub-schemes like transient gain generation (see e.g. [3 – 5], and very recently grazing incidence pumping [6, 7] in the collisional scheme, and generation gain in recombination scheme in transitions to ground states of ions pumped by very high intensity fsec-type of lasers [8 – 10]. Advances in their performance (wavelength range, gain and beam energy) have been achieved together with simplifications of the technology required to generate gain. These achievements have, in part, been stimulated by the first applications of soft x-ray lasers to soft x-ray microscopy, microholography, very high plasma density measurements and their potential applications to semiconductor surface studies, nano-lithography, dynami-

210

S. Suckewer et al.

cal response of semiconductors and biological cells, and a number of other applications. A key element here is the cost and availability of these

Fig. 1. Generic recombination soft x-ray laser scheme with lasing to the ground state.

devices, and intensive efforts are being made to develop compact soft xray lasers that are suitable and convenient for applications in academic and industrial research laboratories. Excellent examples of the progress in this direction are demonstrated by the achievement of lasing actions in the region 40–50 nm in ArIX in a capillary discharge[11] and in Xe IX with femtosecond laser [12]. However, right now, the most urging issue, in our opinion, is the development of compact soft X-ray laser, operating at repetition rate >1Hz and at much shorter wavelengths than collisional schemes offer.

Fig. 2. Scaling of the wavelengths versus ion charge Z- 1, of H-like ions for 32 and 2-1 lasing transitions

Feasibility of 3.4 Nm Laser Pumped by Ultraintense RBS Laser

211

In Fig. 2 is shown the scaling of the wavelengths versus ion charge (Z1) of H-like ions for 3–2 and 2–1 transitions in recombination scheme. One may see fast convergence to region of shorter wavelengths even for quite moderate Z. Especially attractive is scaling for 2 – 1 transitions with Z reaching wavelength ~ 1 nm already for the H-like NeX. Because fast decreasing the wavelength and high quantum efficiency for 2 – 1 transition in H-like ions it is possible to achieve lasing within the “water window” (2.3 – 4.4 nm) at quite high repetition rate using relatively compact lasers as pumps as is shown in Table 1 for CVI ions at 3.4 nm. (Description of the computer codes and results of calculations are discussed in the next sections). Consideration of plasma and pumping laser parameters for gain creation in transition to ground state of H-like ions Ultrashort laser pump pulses are crucial to realize lasing to the ground state of ions in the soft X-ray spectral region in order to provide the population inversion between the excited and ground states (see Fig. 1 for generic recombination scheme for generation population inversion to ground states and Fig.3 for transitions to ground state in C VI ions). Fig.1 illustrates the idea of the process of recombination to ground state. First, the ions are fully stripped from their electrons by an intense ultrashort laser pulse. Then, the ions are recombined by three-body recombination, The rate of such recombination is proportional to Ne2 and n4 (Ne is electron density, and n is the principal quantum number) Therefore the transition from Fast collisional recombinations 18.2 nm

H-like CVI ion

3.4 nm

Fig. 3. Recombination scheme for C VI ion

Lasing to ground state (with fsec pumping)

212

S. Suckewer et al.

C6+ to H-like CVI occurs primarily to the states with high n (Fig. 3). Collisional and radiative transitions to level n = 2 occurs faster than to ground level n = 1, creating population inversion between n = 2 and n = 1 (inversion is also created between levels n = 3 and n = 2, which lasts much longer than between levels n = 2 and n = 1). The ionization process has to be very fast. Rapid ionization generates minimal heating of the plasma, which is crucial for a recombination laser (in order for the recombination process to be dominated by collisional relaxation processes). In addition, the ultrashort pumping pulses are necessary because of the very short radiative life time, τ , of the first excited levels of ions and the decrease of this time with Z as Z-4. For example, τ = 26 psec for Li III and τ = 1.6 psec for C VI [13]. In fact, this time is significantly shorter because at high electron density, which is required for high gain, collisional life time is much shorter than radiative time. In addition, as will be discussed in the next section, short pumping time is required because high gain exist before Maxwellization of electron energy distribution takes place. Hence, only the powerful femtosecond-type laser makes it possible to realize transient recombination systems with lasing to the ground state in ions. Ions are initially totally stripped of all their electrons by optical-field-ionization (OFI). This process was predicted theoretically and calculations were performed based on Keldysh theory [1416]. The required power densities (intensities) and energies of fsec-type of laser for effective ionization of H-like CVI ions are shown in Table 1. Table 1. Required energies and intensities to reach C6+

Important result towards generating lasing action in “water window” by transitions to ground state of H-like ions was demonstration of the gain at 13.5 nm in H-like Li III ions in transition to ground state [8]. This experiment followed by demonstration the lasing action in the same transitions at 2 Hz repetition rate [9]. In [9] the lasing was generated with only a 50 mJ in 250 fsec, 0.25 μm KrF laser beam. This relatively high efficiency was achieved by longitudinal pumping and high quantum efficiency due to lasing to the ground state. Importance of these experiments was that they in principle “opened the road” to compact SXLs at much shorter wave-

Feasibility of 3.4 Nm Laser Pumped by Ultraintense RBS Laser

213

lengths using H-like ions with higher atomic number Z . Later, using ~1 μm , 300 fsec Nd/Glass laser and longer plasma column (14 mm instead of 5 mm) gain-length product was slightly increased to GL ≈ 7.2 from GL ≈ 6.5 for a 5 mm long plasma in an experiment with 0.25 μm indicating a tendency of decreasing gain G with an increasing wavelength of pumping beam. This result, which was confirmed in a similar experiment using 0.8 μm Ti/Sapphire laser [17], seems to imply higher Optical Field Ionization (OFI) efficiency at shorter wavelength, in contradiction to the theory of tunneling ionization. Only recently, based on computer modeling of the ionization (including tunneling, multiphoton and collisional ionizations) and recombination processes combined with calculation of electron distribution function from Fokker-Planck equation and radial motion of particles in plasmas (from MHD code) we were able to understand this phenomena and show that there is no contradiction. The reason for seemingly contradiction with the theory is higher deposition of above threshold ionization (ATI) energy to electrons when OFI laser is operating at longer wavelength. Such electron heating is decreasing rate of 3-body recombination of Li3+ to Li III hence, of course, has very negative effect on gain generation [18]. Even more puzzling was very large fluctuations of 13.5 nm output radiation intensity in all these experiments. We have attributed such fluctuations to gain G fluctuations due to some non-uniformity of the plasma column. Therefore we have dedicated a lot of effort to improve the uniformity of the plasma. However, we did not see any improvement in 13.5 nm intensity reproducibility with improvement of plasma uniformity. Again, only recent computer calculations has shown that all these problems were related to plasma heating during optical field ionization (primarily tunneling ionization) leading to too high electron temperature at the time of threebody recombination.

3. Computer modeling results for gain at 3.4 nm in CVI Recently, we have developed an elaborate numerical model to characterize recombination gain in the 2 - 1 transition of LiIII at 13.5 nm [18]. The model describes the effects of different experimental parameters on the gain. We were able to explain the gain observed in the earlier mentioned experiments [9, 10] and showed that it is possible to achieve high gain on this transition, especially when mixing the plasma with hydrogen. These calculations were extended into C VI at 3.4 nm [19,20].

214

S. Suckewer et al.

Recombination gain relies on having fully stripped ions in a relatively cold plasma. The ionization mechanism that is used to achieve this plasma is tunneling ionization by ultra-short and ultra-intense laser pulses (with pulse-duration of ≤ 100 fsec and intensities in the range of 1017 - 1019 W/cm2 depending on the element being used). Due to the short pulseduration, minimal heating is produced during the ionization. However, when calculating the average energy that is absorbed during the ionization process we find that the absorbed energy still corresponds to an electron temperature that would not allow for population inversion in the transition to ground state to be generated during the recombination process. But taking into account the actual phase-space distribution function of the plasma, including effects from both the non-Maxwellian nature of the distribution function and the spatial expansion and cooling of the plasma after ionization, we have shown that high gain is indeed feasible in the LiIII 2 - 1 transition [18]. In addition, we have shown very recently that the gain can be enhanced and become less stringently-dependent on exactly matching the required values of the experimental parameters, if hydrogen is mixed into the plasma [19,20]. The process of gain generation can be divided into three stages: (a) Ionization and heating, (b) Expansion and cooling and (c) Recombination and gain. We shall repeat briefly here the principles of our model (described in detail in [18]), and discuss more deeply the adjustments and enhancements that were required in order to apply the model to the C VI ion. (a) Ionization and heating: Ionization is achieved by OFI using an intense laser pulse. The ionization rate is calculated by the tunneling rate of an electron under the influence of the laser electric field treated in semiclassical approximation [16].The classical motion of the ionized electrons in the laser field after ionization yields the so-called residual energy that is absorbed by the electrons during the ionization process. This energy is due to the phase mismatch between the ionized electrons and the oscillating laser electric field. It is proportional to the quiver energy of the electrons, εq, at the time of ionization. Though the residual energy is in fact much smaller than the quiver energy, it would be high enough to prevent recombination gain to occur if the plasma were Maxwellian. However, the ionization process yields a plasma with a highly non-Maxwellian electron distribution function (EDF) as was shown in [18,19].Qualitatively, one can understand the properties of the OFI-EDF by realizing that most of the electrons are ionized at the peak of the laser oscillating electric field and continue to move in phase with it. Therefore, the vast majority of the electrons in the plasma absorb very little residual energy, and only a small fraction of the electrons are ionized off the peak of the laser electric field and absorb high energy. These highly energetic electrons contribute much

Feasibility of 3.4 Nm Laser Pumped by Ultraintense RBS Laser

215

to the overall electron average energy, but have a relatively low probability of participating in collisional processes. This effect gives rise to enhanced three-body recombination and electron-impact de-excitation rates. In other words, the non- Maxwellian nature of the distribution function causes the “effective recombination temperature" of the plasma to be much lower then the temperature of the corresponding Maxwellian plasma, given by 2/3 of the average energy. The “effective recombination temperature" can be defined by comparing the actual three-body recombination rate to the same rate in a Maxwellian plasma. The effective recombination temperature is always much lower for OFI-plasma than for a Maxwellian plasma with the same average energy. The ionization was simulated by the iPIC (Ionization Particle in Cell) code. For a sufficiently high electron density (Ne≥ 5 1019 cm-3), and a fixed electron temperature, gain is higher for higher Z ions. Therefore, we can expect to get better results for CVI (Z = 6) than for LiIII (Z = 3). However, the required intensity for the ionization of CVI ions is almost two orders of magnitude higher than the intensity required to ionize LiIII ions, and with it, the average energy of the electrons. The effects of the residual energy on the recombination gain are reduced significantly when taking into account the non-Maxwellian nature of the distribution function and by adding hydrogen to the plasma Adding hydrogen supplies cold electrons to the plasma since the hydrogen atoms are ionized by the front of the pump pulse (or by the pre-pulse) and absorb very little residual energy. These electrons then participate in the recombination process and enhance the gain. In the CVI case however, higher densities are required to achieve gain (about an order of magnitude higher than that required for LiIII), and collisions during the ionization process become more significant since the collision frequency scales linearly with the density. Unlike the residual heating, collisional heating affects the electrons ionized from carbon (Celectrons) and the electrons ionized from hydrogen (H-electrons) in the same way. One way to counter this effect is to use a shorter pulse for the ionization. The overall collisional heating is roughly proportional to the product νcolτ , where νcol is an average collision frequency, and τ is the pulse-duration. Therefore, increasing the density (and with it the collision frequency) by a factor of 5-10 and reducing the pulse-duration by the same factor should have very little net effect on the collisional heating. Using shorter pump pulses requires the use of a slightly higher intensity, since the ionization has to be completed in a shorter time, and full ionization is crucial to achieve high gain. However, even with the higher intensity the shorter pulses contain lower energy, which means that using shorter pulses would be more energy-efficient.

216

S. Suckewer et al.

(b) Expansion and cooling: The expansion and cooling (along with Maxwellization) process is simulated by numerically solving the FokkerPlanck (FP) equation for the distribution function that is calculated by the iPIC code with an implementation of the SPARK code [18]. Since the OFI-EDF was generated by the iPIC code in planar geometry and since the charge separation and space-charge oscillations may not have cylindrical symmetry (recall that the pump beam is linearly polarized), the straightforward conversion between planar and cylindrical was not appropriate here and the planar geometry option of the FP solver code was used. Finally, we note that due to the higher density and the short time scales at which gain occurs (usually less than 1 ps after ionization), the expansion cooling plays a less important role here since very little expansion can happen in these time scales. (The main importance of solving the FP equation is for calculating the Maxwellization process, which has to be taken into account to get realistic results.) Therefore, in contrast to the LiIII case, there is no need for a tight focus of the pump pulse in order to achieve gain (other than for achieving the required intensity). This property of the gain may make it easier to achieve longer gain channels for gain saturation.

Pump diameter: 10µm

Pump diameter: 30µm

C VI 3.4 nm

C VI 3.4 nm

b)

a) Plasma radius

b) Plasma

radius

Fig. 4. Gain coefficient (in cm-1) for CVI 2→1 transition with initial conditions of nC=1019cm -3 and nH=1020cm-3 for 2 different pumping beam(OFI) diameters, both with intensity I ~ 10!9 W/cm2

This page intentionally blank

Feasibility of 3.4 Nm Laser Pumped by Ultraintense RBS Laser

217

(c) Recombination and gain: The recombination process was simulated by solving the rate equations governing the process, taking into account three-body recombination, electron-impact ionization, electronimpact excitation and de-excitation, radiative recombination and radiative relaxation (spontaneous emission). The plasma was assumed to be optically thin, hence no photoexcitation processes were considered. The values of the rate coefficients in the rate equations were obtained by integrating the cross-sections of the different interactions over the actual nonMaxwellian, time-dependent, electron distribution function that was obtained from the iPIC and FP codes. The small-signal gain coefficient G was then calculated from broadening for the cold ions (the ions are assumed to be fixed during the whole process hence the ions temperature was taken to be the initial plasma temperature, i.e., Ti ≈1 eV) and the Stark broadening estimated for the Lyman-α line by the lesser of the widths given by the quasi-static linear Stark effect (Holtsmark theory) and electron impact broadening. In Figs 4a,b, are shown results for gain G (in cm-1) calculations for CVI 2–1 transition at 3.4 nm for 2 different diameters of pumping laser beam (50 fsec, 1019 W/cm2 ) versus plasma channel diameters and time of gain evolution. The maximum gain is reaching very high values in range G ≈ 150 - 200 cm -1, which is very encouraging result.

4. Raman Backscattering (RBS) Amplification and Compression The second part of our presentation was devoted to a new type of ultrashort and ultra-intensive compact laser system based on Raman Backscattering (RBS) amplification and compression in a few millimeter long plasma. In that presentation we briefly described the research and recent results with emphasis on multi-passes approach for maximizing system efficiency. The principle of RBS amplification in counterpropagating geometry is shown in Fig.5. Moderately intense, but long, laser pulses can be scattered into short very intense counterpropagating pulses in a plasma through stimulated Raman backscattering (RBS). Raman amplification of ultrashort pulses in a plasma is based on the three-wave interaction between the counterpropagating “seed” and “pump” pulses and the plasma waves[21] as is shown schematically in Fig.6. The plasma waves, with frequency ωpe (ωpe2= 4πe2ne/me), are ponderomotively driven by the periodic intensity pattern produced by the

218

S. Suckewer et al.

Raman amplifier

ω pump

ω pump ω ω ω + ω pe = ω seed

“Pump”

“Seed”

“Seed”

ω seed

ωpe

Plasma wave

Fig. 5. Principle of Raman backscattering amplification.

interference between the pump pulse and the seed (subpicosecond) pulse. If the frequency detuning between the two pulses matches the plasma frequency ωpe, i.e., ωpe = ωpump - ωseed, then the seed pulse can be amplified through the RBS instability of the pump.

ω

S eed

P um p

pum p

ω

se e d

P la s m a w a v e

ω p e 2 = 4 π n e e 2 /m R e s o n a n t C o n d itio n s: ω pum p = ω se e d k

pum p

=

k

se e d

+ +

ω k

e

pe pe

Fig.6. Schematic of three waves mixing

The RBS amplification can be divided into linear and nonlinear regimes. In the linear regime, the pump depletion is negligible and the gain is inde pendent of the seed intensity. The seed pulse is amplified and the pulse duration is increased due to the narrow bandwidth of the linear amplification (which is approximately twice the growth rate). The nonlinear regime is characterized by significant pump depletion and simultaneous temporal compression of the amplified pulse, as illustrated in Fig. 7. The front of the pulse is amplified while the tail part interacts with a depleted pump and is not amplified as much as the front. In this regime the efficiency of energy transfer from the pump to the seed can reach ~ 90% when ωpump ≥ 10

Feasibility of 3.4 Nm Laser Pumped by Ultraintense RBS Laser

219

ωpe according to theoretical predictions [22] through the following relation η = 1 - ωpe/ωpump. This could happen when the pump pulse duration is

Pump

P

Depleted pump

Seed

S Amplified Ampliseed

Plas ma

Fig. 7. Resonant Raman non-linear regime (pump depletion)

sufficiently long to match the length of the plasma and obtain nearly full pump depletion. Operating in this nonlinear regime of RBS amplification in the plasma is of highest interest. In Fig. 8 is shown schematically experimental setup on which nonlinear regime was reached [23]. The pump entered the plasma at point A, and the seed at point B. The spatial overlap of the pump and the seed were guaranteed by imaging the two beams above the nozzle with a CCD camera. The temporal overlap was achieved by scanning of the optical delay between the pump and seed with precision better than ps. The output seed passed through a long wavelength (> 850 nm) band pass filter so that any reflected pump beam light was filtered out. A small part of the amplified seed pulse was directed to a spectrometer for spectral analysis. The rest of the beam was split by a 50% beam splitter. One part was directed to a power meter and the second part was directed to an autocorrelator. When the amplification entered the nonlinear regime, the increased bandwidth made the growth of the seed pulse more robust than thermal Raman in the nonuniform plasma. Pulse compression in the nonlinear regime is accompanied by bandwidth broadening. The pump (for pumping of both the Barium Nitrate crystal and the plasma) and the input seed had a bandwidth of about 12 nm. The bandwidth of the amplified seed was ~ 19 nm FWHM, which is an independent indication of reaching the nonlinear regime. With the high amplification of the seed pulse, we were able to perform a pulse duration measurement of the amplified seed using a standard autocorrelator (Positive Light Model SSA) at various plasma length. The pulse

220

S. Suckewer et al.

Fig.8. Schematic of the RBS amplifier & compressor experimental setup

was observed to lengthen to a maximum of ~ 800 fs when the amplification was below 10. The pulse duration began to decrease when the ampli fication approached 30. he trend that the pulse became shorter as the amplification became higher is clearly shown in Fig. 9. The shortest pulse we have observed so far had a FWHM of ~ 150 fs when the amplification was ~ 60. At this point, the power of the seed pulse increased more than 200 times, and its intensity increased ~ 1000 times when spatial beam narrowing was taken into account. The intensity of the amplified seed pulse at point A in Fig. 8 was 1.7 x 1015 W/cm2. The pump pulse diameter was about 50 μm FWHM at point A and had an intensity of ~ 1 x 1014 W/cm2, hence the intensity of the amplified seed exceeded the pump intensity by more than order of magnitude. Simulation results are also shown in Fig.9. Simulations were performed using the average 1D particle-in-cell code (aPIC). The seed and pump pulses were assumed to be initially Gaussian with FWHM given by the experimental values of 500 fs and 10 ps, respectively. The initial seed intensity was 1.6 x 1012 W/cm2 and pump intensity was 1 x 1014 W/cm2. The plasma density was chosen so that the plasma wave (including a thermal shift) was resonant with the beat frequency of the lasers. The electrons were assumed to be initially homogeneous with a Maxwellian velocity distribution of Te=50eV and the ions were assumed to be motionless over the short timescale of the simulation. The seed energy amplification (the final seed energy divided by the initial seed energy) and the FWHM of the seed pulse duration are plotted as a function of the amplification of the seed (function of interaction length ). The simulation clearly shows pulse

Feasibility of 3.4 Nm Laser Pumped by Ultraintense RBS Laser

221

Fig. 9. Pulse duration vs. output energy (amplification); amplified pulses: modeling (solid lines) and experiments (circles)

stretching of the seed during the initial amplification (until roughly 1mm, corresponding to an energy amplification of ~20 ), followed by pulse shortening to ~240 fs at an energy amplification of ~180 at the end of the 2 mm plasma. While the pulse duration evolution agreed quite well with the experimental measurements, we did not observe an energy gain factor as high as 180. This may be because the amplified seed pulse experienced energy loss due to heating and ionization induced diffraction of the plasma. The simulation results shown here are one-dimensional and are, in detail, sensitive to the initial value chosen for the temperature (which is not yet well known experimentally). Nevertheless, the qualitative features of the experiment are well-reproduced including pulse expansion followed by noticeable compression and amplification of the seed pulse after its intensity becomes larger than that of the pump.

5. Two-passes RBS amplification In Fig. 10 are presented recent results for amplification in two-passes set up. The system employed additional mirrors, and interferometeric filters in order for pump pulse, after amplifying seed pulse in plasma in the 1st pass, to be directed back to approximately the same plasma region and interact again with reflected back into the plasma amplified seed. This way in the second pass the seed was increased farther amplified by factor ~2.

222

S. Suckewer et al.

Amplified Seed Energy

This was just demonstration of possibility to amplification in second pass due to quite significant “left over” pump energy from the 1st pass. Such 2passes or even multi-pass systems could significantly improve efficiency of RBS amplifier/compressor. Single pass

(a) Pump delay (psec)

Double

(b) Pump delay (psec)

Fig. 10. Amplified seed energy in single pass (a) and in double pass (b) for input seed of 5 μJ and pumping energy ~ 62 mJ in the 1st pass and ~ 40 mJ in second pass *) Present address: The University of Texas at Austin

References

1. Matthews, D, Hagelstein, P, Rosen, M., et al.: ‘Demonstration of soft X-Ray Amplifier’, Phys. Rev. Lett., 54, 110, 1985. 2. Suckewer, S., Skinner, C.H., Milchberg, H., Keane, C., and Voorhees, D.: ‘Amplification of stimulated soft-x-ray emission in a confined plasma column’, Phys. Rev. Lett., 55, 1753, 1985; also Suckewer, S., Skinner, C.H., Kim, D., Valeo, E., Voorhees, D., and Wouters, Al, ‘Divergence measurements of soft-x-ray laser beam’, Phys. Rev. Lett., 57, 1004, 1986. 3. Dunn, J., Osterheld, A.L., Shepherd, R., White, W., Shlyaptsev, V.N., and Stewart, R.E., ‘Demonstration of x-ray amplification in transient gain nickellike palladium’, Phys. Rev. Lett., 80, 2825-2828, 1998; also Dunn, J., Li, Y., Osterheld, A.L., Nilsen, J., Hunter, J.R., and Shlyaptwev, V.N. ‘Gain satura-

Feasibility of 3.4 Nm Laser Pumped by Ultraintense RBS Laser

223

tion regime for laser driven tabletop transient Ni-like ion x-ray lasers’, Phys. Rev. Lett., 84, 4834-4837, 2000. 4. Nickles, P.V., Shlyaptsev, V.N., Kalachnikov, M., Schnurer, M., Will, I. Sandner, W., ‘Short pulse x-ray laser 32.6 nm based on transient gain in Ne-like titanium’, Phys. Rev. Lett., 78, 2748-2751, 1997. 5. Nilsen, J., MacGowan, B., DaSilva, L.B., and Moreno, J.C., ‘Prepulse technique for producing low-Z Ne-like x-ray lasers’, Phys. Rev. A, 48, 4682-4685, 1993. 6. Keenan, R., Dunn, J., Patel, P.K., Price, D.F., Smith, R.F., and Shlyaptsev, V.N., ‘High- repetition-rate grazing-incidence pumped x-ray laser operating at 18.9 nm’, Phys. Rev. Lett., 94, 103901, 2005. 7. Luther, B.M., Wang, Y. M., Larotonda, A., Alessi, D., Marconi, M.C., Rocca, J.J., and Shlyaptsev, V.N., ‘Saturated high repetition-rate 18.9-nm tabletop laser in nickel-like molybdenum’, Optics Lett., 30, 165-167, 2005. 8. Nagata, Y., Midorikowa, K., Obara, M., Tashiro, H., and Toyoda, K., Phys. Rev. Lett., 71, 3774, 1993. 9. Korobkin, D., Nam, C.H., Goltsov, A., and Suckewer, S., ‘Demonstration of soft x- ray lasing to ground state in Li III’, Phys. Rev. Lett,. 77, 5206-5209, 1996. 10. Goltsov, A., Morozov, A., Suckewer, S., Elton, R., Feldman, U., Krushelnick, K., Jones, T., Moore, C., Seely, J., Sprangle, P., Ting, A., and Zigler. A., ‘Is efficiency of gain generation in LiIII 13.5-nm laser with 0.25μm subpicosecond pulses the same as with 1 μm?’, IEEE J. Sel. Top. Quantum Electron., 5(6):1453–1459, 1999. 11. Rocca, J.J., Shlyaptsev, V.N., Tomasel, F.G., Cortazar, O.D., Harshorn, D., and Chilla, J.L., Phys. Rev. Lett., 73, 2192, 1994 ; also Benware, B.R., Macchietto, C.D., Moreno, C.H., and Rocca, J.J., ‘Demonstration of a high average power tabletop soft x-ray laser’, Phys. Rev. Lett., 81, 5804-5806, 1998 ; also Macchietto, C.D., Benware, B.R., Rocca, J.J., ‘Generation of millijoulelevel soft-x-ray laser pulses at a 4-Hz repetition rate in a highly saturated tabletop capillary discharge amplifier’, Opt. Lett., 24, 1115-1117, 1999. 12. Lemoff, B.E., Yin, G.Y., Gordon III, C.L., Barty, C.P., and Harris, S.E., ‘Demonstration of a 10-Hz Femtosecond-Pulse-Driven XUV laser at 41.8 nm in Xe IX’, Phys. Rev. Lett ,74, 1574-1577, 1995. 13. Attwood, D.T., Soft X-rays and extreme ultraviolet radiation, Cambridge University Press, 1999. 14. Keldysh, L.V., ‘Ionization in the field of strong electromagnetic wave’, Soviet Physics JETP, 20, 1307, 1965. 15. Perelomov, A.M., Popov, V.S., and Terent’ev, M.V., Soviet Physics JETP, 23, 924, 1965. 16. Burnett N.H., and Corkum, P.B., ‘Cold-plasma production for recombination extreme-ultraviolet lasers by optical-field-induced ionization’, JOSA B, 6, 1995, 1989.

224

S. Suckewer et al.

17. Ping, Y. Ph.D. Thesis ‘Soft X-ray Lasers and Raman Amplification in Plasmas’, Princeton University, 2002. 18. Avitzour, Y., Suckewer, S., and Valeo, E., ‘Numerical investigation of recombination gain in the LiIII transition to ground state’. Phys. Rev. E., 69, 046409, 2004. 19. Y. Avitzour Ph.D. Thesis “Numerical Modeling of Recombination X-Ray Lasers in Transition to Ground State”, Princeton University (2006). 20. Avitzour, Y., Suckewer, S., ‘The feasibility of achieving gain in transition to ground state of CVI at 3.4nm’, submitted for publication (August.2006) 21. Shvets, G., Fisch, N.J., Pukhov A., and Meyer-ter-Vehn, J., ‘Superradiant Amplification of an Ultrashort Laser Pulse in a Plasma by a Counterpropagating Pump’, Phys. Rev. Lett., 81, 4879, 1998; also Fisch, N.J., and Malkin, V.M., ‘Generation of Ultra-high Intensity Laser Pulses’, Physics of Plasma, 10, 2056-2063, 2003. 22. Malkin V. M., Shvets G. and Fisch N. J., ‘Fast compression of laser beams to highly overcritical powers’, Phys. Rev. Lett., 82, 4448, 1999;also Malkin V. M., Shvets G. and Fisch N. J., ‘Detuned Raman amplification of short laser pulses in plasma’, Phys. Rev. Lett,. 84, 1208, 2000. 23. Cheng, W., Avitzour, Y., Ping, Y., Suckewer, S., Fisch, N.J., Hur, M.S., and Wurtele, J.S., ‘Reaching Nonlinear Regime in Large Raman Amplification of Ultrashort Laser Pulses’, Physical Review Letters, 94, 2005.

Status and Prospects on Soft X-Ray Lasers Seeded by a High Harmonic Beam at LOA

S. Sebban1 , Ph. Zeitoun1, G. Faivre1, S. Hallou1, A. S. Morlens1, J.P. Goddet1 B. Cros2, G. Vieux2 and G Maynard2,T. Mocek4 and M. Kozlová4, J.P. Causmes5 and H. Merdji5 1

Laboratoire d’Optique Appliquée, chemin de la hunière, 91128 Palaiseau LPGP, Université Paris-Sud, 91405 Orsay, France 3 Institute of Physics, Department of X-Ray Lasers, Prague, Czech Republic 4 Service des Photons, Atomes et Molecules, CEA, 91191 Gif-surYvette,France 2

Summary. Thanks to the most recent works on x-ray laser and on high order harmonics (HHG), it is now possible to produce an energetic beam having at the same time the required optical properties. The solution consists in seeding the XRL amplifier medium with another beam (HHG). This experiment was successfully realized in LOA. We studied seeding of two x-ray laser transitions, 4d-4p at 32.8 nm in Kr8+ and 5d-5p 41.8 nm in Xe8+. The amplifying medium is generated by focussing a high energy circularly polarized, 35 fs 10 Hz Ti: sapphire laser system in a few mm cell filled with gas (xenon or krypton). We succeeded to increase from a factor 10 to 200 the HHG energy, without deteriorating their optical qualities. The resulting beam was polarized, coherent and we estimate the output energy to be about 0.5 µJ.

1 Introduction For decades, synchrotrons have been the acme of XUV sources, providing X-rays for biology, chemistry, physics applications and industrial developments, resulting in major scientific progress. But tomorrow’s most attractive applications (e.g. structural Biology) require pulses with much shorter duration (fs) and much higher energy (mJ) than delivered by synchrotrons. X-ray Free Electron lasers should fulfil these requirements by providing ultra-intense beams, but their limited number will stir tremendous beamtime pressure, slowing down the spreading of applications. Laser-driven XUV sources are comparatively inexpensive and widely avail-

226

S. Sebban et al.

able, but they have reached a bottleneck in the race for ultra-high intensities. In this pioneer work, we have set up and characterised the first true XUV laser chain, overcoming these bottlenecks for the first time. By combining the high optical quality of high harmonics as a seed [1] , with a highly energetic x-ray laser plasma amplifier [2-3], we produced the first highly saturated, energetic, femtosecond, fully coherent and fully polarised tabletop x-ray laser operating at 10 Hz [4]. This technique, easily applicable on all existing laser-driven XUV facilities, opens the way to ultra-high intensities studies worldwide. Since some years x-ray laser community try to promote this source for applications. In this aim, it is necessary to offer an optically good beam to the users. Nevertheless most of x-ray laser (XRL) are using amplified spontaneous emission, i. e. laser has no cavity to select a spatial mode. Consequently the beam can not be polarized or spatially coherent and the wave front is not regular. Thanks to the most recent works on x-ray laser and on high order harmonics (HHG), it is now possible to produce an energetic beam (from 0.1 to 1mJ) having at the same time the required optical properties. The solution consists in seeding the XRL amplifier medium with another beam (HHG). This experiment was successfully realized in LOA. We studied seeding of two x-ray laser transitions, 32.8 nm (Kr8+) and 41.8 nm (Xe8+). We succeeded to increase from a factor 15 to 600 the HHG energy, without damaging their optical qualities. This structure seems to be the beginning of a new generation of XRL, with considerable improvements compare to the actual XRL. Furthermore this technique can help for the measurement of many plasma parameters.

2 Experimental setup The XUV amplifying laser chain was made of three parts: a high harmonics generation (HHG) seed, a focusing system, and an optical field ionized x-ray laser (XRL) amplifier medium. The seed beam was obtained by focusing a 20 mJ, 30 fs, infrared laser in a gas cell filled with argon or xenon .This seed was image relayed onto the entrance of the x-ray laser (XRL) amplifier (Kr or Xe gas longitudinally pumped by a 1 J, 30 fs laser) by means of a toroidal XUV mirror. This X-ray laser operates at 32.8 nm and 42.8 nm using Kr and Xe respectively. After amplification, the output beam was either analyzed by an XUV spectrometer or XUV CCD camera. The experiment has been performed in LOA with the multiterawatt, 30 fs, 10 Hz, Ti:sapphire laser emitting at 815 nm. The laser delivers two chirped beams having separate pulse compressor enabling to optimise in-

Status and Prospects on Soft X-Ray Lasers Seeded by a High Harmonic

227

dependently the XRL or the HHG drivers. One beam, lately called “pump”, contains most of the energy (about 1 J on target) and is used for creating the amplifying XRL medium. The second beam, called “probe”, has energy around 20 mJ on target and generates the HHG (Figure 1) Harmonics were produced by focusing the probe beam into a gas cell of variable pressure and length, filled with argon. The linearly polarised beam was focused with a f=1.5 m, aspherical lens. The focal spot has been 15 -2 measured to be about 150 µm leading to an intensity of about 10 W.cm . The harmonic flux was optimised by changing the conditions of laser interaction with the active medium, such as moving the focal spot position, changing the gas pressure, and the cell length. The highest HHG flux was obtained for a pressure of 25 mbar, 10 mm long cell, and focal plane situated 4 cm behind the gas cell. The main optimisation was the fine adjustment of the wavelength of one harmonic (25th) so as to coincide with the 8+ wavelength of Kr XRL (32.8 nm). This was done on a day-to-day basisby adjusting the chirp of the laser generating the HHG [5-6].

Fig. 1. Schematic description of the experimental set-up

The amplifying plasma was created by focusing the pump beam into another gas cell of variable length (from 0 to 6 mm) and pressure. The cell was filled with krypton or xenon held at uniform pressure. Fundamental radiation from the Ti:Sapphire laser was circularly polarised by a quarterwave plate and focused with an f=0.85 m, spherical on-axis mirror into the gas cell. Circular polarisation ensures the creation of energetic free electrons that collisionally pump the 4d-4p or 5d-5p lasing lasing transitions. The measured focal spot was around 50 µm in diameter leading to a net in17 -2 tensity of about 5×10 Wcm . The output of the harmonic source was re-imaged on the entrance of the amplifier plasma, using a 1:1 grazing incidence (4°) gold coated toroidal

228

S. Sebban et al.

mirror. The theoretical reflectivity is around 80%. The focal spot of HHG has been measured to be around 150 µm. Special attention has been paid to the temporal and spatial overlapping between the harmonic and the X-ray laser beams. As the OFI XRL is longitudinally pumped, the position of the gain region coincides with the position of the IR driving laser. Since also the HHG are collinear with the IR beam, the superposition of the HHG and the XRL gain region has been done in IR. Temporal synchronisation of the HHG with the period of XRL amplification could not be done in such a simple way since the plasma starts to amplify after the interaction of the driving laser with the gas. Based on previous experiment, we may assume that the gain will appear around 5 ps after laser interaction and will last about 8 ps [7]. Since HHG are emitted only during the interaction of the IR laser with the gas, the synchronisation of the seed and the XRL period of amplification was done by adjusting the delay between the IR pump and probe beams. The XUV beams generated in this experiment were analysed by an onaxis XUV spectrometer and by making monochromatic images of the beam cross-section. These images were obtained by placing a removable 45°, XUV mirror in the beam path, redirecting the beam towards a cooled, thin, back-illuminated charge-coupled device (CCD). A 300 nm thick aluminium filter has been used to remove the IR beams. The interferential multilayer mirror has been designed to achieve a high reflectivity (estimated at 50%) and a strong polarisation at 45°. The XUV mirror is made of a stack of 30 Mo/B4C/Si tri-layers.

3 Results and discussion The delay between XRL plasma creation and HHG injection was variable to reach optimum amplification conditions, synchronizing HHG seeding with maximum gain. Four emission spectra are displayed in figure 2, corresponding to (a) high harmonics alone, (b) x-ray laser alone, (c) XRL and HHG seeded long after gain extinction, and (d) HHG synchronised with maximum XRL gain. The amplification period starts 5 ps after the interaction of the IR laser with the gas and lasts about 8 ps. Figure 2 (d) clearly shows that a strong amplification of the seed has been achieved. The HHG were seeded at about twice the intensity of the ASE x-ray laser, leading to an enhancement of the output signal by a factor of 13. Since the level of seeding may influence many laser parameters (intensity, pulse duration, ratio of the energies in the seeded beam to this in the ASE beam), it is crucial to understand how is the seeding level related to the saturation intensity.

Status and Prospects on Soft X-Ray Lasers Seeded by a High Harmonic

229

Fig. 2. Experimental spectra under conditions : (a) HHG only, (b) ASE XRL only, (c) both HHG and XRL in a non-amplifying timing, and (d) the amplified seeded XRL.

The amplification law (output intensity versus amplifier length) of the amplified spontaneous emission (ASE) XRL has been measured to clarify the level of seeding. The gas cell length ranged from 0 to 4 mm and the intensities have been fitted taking into account saturation effects [3]. A -1 small-signal gain coefficient of 80 cm has been inferred, reaching saturation above 1.7 mm plasma length. Clearly, the spectrum in Fig. 2(d) was obtained for an HHG seeding above saturation intensity (about 4.5 times higher). In the second part of the experiment, the seeding intensity was reduced down to 1/50th of the saturation intensity. The output signal was 38 times higher than the saturation intensity; the amplification factor became as large as 200 i.e. 15 times higher than for the amplification achieved with above-saturation-intensity seeding. The delay line has also allowed to measure the gain duration for several pressure. This measurement was done for both gas, krypton end xenon. For the Xe8+ at 41,8 nm (fig3a), the maximum gain was obtained for a pressure of 15 Torr and the duration of the gain was about 8ps. For Kr8+ at 32.8 nm (fig3b) the best pressure was 15 Torr, and for this pressure the gain duration was 4 ps. Using this method the temporal profile of the gain of the collisionally excited Xe8+ laser at 41.8 nm has been measured for various gas densities and compared to simulations of the atomic processes

230



S. Sebban et al.

a)

b)

c) Fig. 3. (a) Measured amplification, (b) extracted average gain and (c) calculated gain coefficient as a function of function of the injection delay for Xe, and for 5 gas pressures

performed with a time-dependent collisional radiative code as shown in figure 3 [8]. The seeded XRL beam was also characterised in terms of optical parameters such as divergence, spatial coherence and polarisation. These parameters were directly inferred from images of the beam cross-section (fig. 5 and 6). Theoretically, the ratio of the mirror reflectivity for the 25th harmonic (32.8 nm) to the 23rd (35.65 nm) or to the 27th (30.37 nm) is about 10. Therefore only the 25th harmonic and/or the XRL might be reflected towards the detector. The seed divergence is imposed by both the geometry of the toroidal mirror and the HHG intrinsic divergence. The divergence of the seeded XRL was slightly lower than those of the seed. The brightest part of the image in Fig. 4 corresponds to the seeded XRL while the weak surrounding, circular signal is the ASE x-ray laser. The diver-

Status and Prospects on Soft X-Ray Lasers Seeded by a High Harmonic

231

gence of the seeded XRL was slightly lower than those of the seed. The brightest part of the image in Fig. 4 corresponds to the seeded XRL while the weak surrounding, circular signal is the ASE x-ray laser. The divergence of the ASE x-ray laser was found to be much larger (>12 mrad) than that of the seeded beam (1 mrad).

Fig. 4. 3D images (false color) of the seeded XRL cross-section as recorded by the XUV CCD detector after reflection on the monochromatic XUV mirror. The central strong signal (gray) belongs to the seeded XRL while the ASE XRL illuminating essentially the whole filter generated a circular signal (purple). The insert shows the seeding HHG signal recorded with the same set-up and displayed on the same scale as the main image.

The multilayer mirror has been designed to be highly polarising. The extinction ratio (intensity peak/lowest intensity) has been investigated by measuring the reflected signal versus the angle of polarisation of the HHG. Figure 5 displays the variation of both the HHG and the seeded XRL intensity versus the angle of polarisation The extinction ratio of the mirror is as high as 20. From these curves, it appears that the seeded XRL is polarised and follows well the initial polarisation of HHG. No sign of depolarisation has been observed. Inserts in Fig. 6 show the beam cross-section when the HHG were s- polarised (peak reflectivity) and p- polarised (minimum reflectivity). As expected, the ASE XRL alone is unpolarized. Also we have observed that the coherence of the seeded XRL beam can be dramatically improved, with the respect to seed HHG beam. In this aim, we have characterized the spatial profile and conducted a series of Young’s two-slit experiments to measure and systematically compare the

232

S. Sebban et al.

Fig. 5. Variation of the HHG (open diamond) and seeded XRL (black square) intensities versus the angle of polarization.

Fig. 6. Fringe visibility of the HHG beam, the ASE emission and the amplified 32.8 nm pulse as a function of the slit separation.

spatial coherence of the seed HHG beam, the ASE amplifier emission and the seeded amplifier emission at 32.8 nm. Figure 6 shows the normalized complex degree of coherence ⎟µ12⎟ plotted for a gaussian profile. The coherent radius of the ASE emission is about 98 µm which is about 150 times smaller that the measured beam diameter. As the OFI XRL emission can be considered as spatial incoherent but temporally coherent, the equivalent incoherent source size which is here about 76 µm should closely correspond to transverse dimension of the plasma amplifier. For the seed HHG beam we estimate a coherent radius of 102 µm which corresponds to about a quarter of full beam diameter. When amplified by the population inverted plasma, the coherent radius build upto 232 µm which corresponds to about 60% of the central disc of the beam profile, showing the benefic action of spatial filtering on the improvement of the transverse coherence of the generated pulse.

Status and Prospects on Soft X-Ray Lasers Seeded by a High Harmonic

233

4 Conclusion Future work will be in a near future to seed in an OFI capillary (ASE x30 compare to OFI in a gas cell). With this kind of amplifier, the expected output energy is about 30 µJ). Later we plan to seed an XRL plasma created from a solid target such as to increase the output energy by at least three order of magnitudes (~ 1 mJ) thanks to the much higher density of XRL amplifying medium [9]. This rise in density and temperature will also lead to a spectral width broadening and consequently to even shorter pulse duration. As this technique matures, future compact XUV laser chains should achieve beams performances close to those of VUV Free-electron lasers [10].

References [1] Kazamias, S. et al, Global optimization of High Harmonic generation. Phys. Rev. Lett., 90, 193901 (2003) [2] Sebban, S., Demonstration of a Ni-Like Kr Optical-Field-Ionization at 32.8 nm. Phys. Rev. Lett., 89, 253901 (2002). [3] Sebban, S. et al, Saturated Amplification of a Collisionally Pumped OpticalField-Ionization Soft X-Ray Laser at 41.8 nm. Phys. Rev. Lett. 86, 3004 (2001). [4] Ph. Zeitoun. et al. , A high-intensity highly coherent soft X-ray femtosecond laser seeded by a high harmonic beam, Nature vol 431, 426-429 (2004) [5] Lee, D. G., Kim, J. H., Hong, K. H., Nam, C. H., Coherent Control of HighOrder Harmonics with Chirped Femtosecond Laser Pulses. Phys. Rev. Lett, 87, 24, 243902 (2001) [6] Reitze, D. H., Enhancement of high-order harmonic generation at tuned wavelengths through adaptative control. Opt. Lett., 29 1, p. 86 (2004) [7] T. Mocek et al, Characterization of collisionally pumped optical-fieldionization soft x-ray lasers. Appl. Phys. B, in press [8] T. Mocek, S. Sebban, G. Maynard, Ph. Zeitoun, G. Faivre, A. Hallou, M. Fajardo, S. Kazamias, B. Cros, D. Aubert, G. de Lachèze-Murel, J. P. Rousseau, and J. Dubau “Absolute Time-Resolved X-Ray Laser Gain Measurement”, Phys. Rev. Lett. 95, 173902 (2005) [9] Y. Wang et al. Phys. Rev. Lett 97 (2006) 12901 [10] Ayvazyan, A. et al, A new powerful source for coherent VUV radiation : Demonstration of exponential growth and saturation at the TTF free-electron laser. Eur. Phys. J. D, 20, 149 (2002)

Generation of Efficient Optical-Field-Ionization X-Ray Lasers in Cluster Jets M.-C. Chou, T.-S. Hung and J.-Y. Lin* Department of Physics, National Chung Cheng University, Chia-Yi 621, Taiwan P.-H. Lin, S.-Y. Chen and J. Wang Institute of Atomic and Molecular Sciences, Academia Sinica, Taipei 106, Taiwan

Summary: We experimentally demonstrate the amplification of optical-fieldionization soft x-ray lasers in krypton cluster jets at pump energy of only 200 mJ. With the employment of small f-number optics, nature diffraction and ionizationinduced refraction in plasmas can be suppressed to enable the pump pulse producing a longer column of gains provided the focal position is optimized. A simple wave propagation code is developed to study and optimize the laser propagation in optical-field driven plasmas. Good agreements between experimental observations and simulations are found. The further enhancement of Kr x-ray laser output at 32.8 nm is achieved by guiding the pump beam in the optically preformed plasma waveguide. More than 4 times of x-ray photons are generated from pure Kr plasma waveguide at the same pump energy. Significant reductions of the pump energy make these pumping configurations favorable for practical high-repetitionrate operations.

1 Introduction Optical-field-ionization (OFI) has been considered as an efficient way for pumping tabletop, high-repetition-rate soft x-ray lasers through collisional excitation or recombination processes, because the electron energy distribution and the ionization balance in plasmas can be precisely controlled by the pump laser intensity and polarization [1-5]. A linearly polarized laser pulse can minimize the mean electron temperature in plasmas and promote strong recombination, whereas the hot electrons produced by a circularly polarized pulse can be used for collisional-excitation pumping. Moreover,

236

M.-C. Chou et al.

the appearance intensity for various ion species in the OFI plasma is discrete and highly selective. These properties can be applied to enhance ion utilization, thereby achieve efficient lasing. Saturated lasing of collisional excitation OFI x-ray lasers was first demonstrated in Xe8+ ions for the 5d-5p transition at 41.8 nm by focusing an intense, circularly polarized laser pulse into a gas cell filled with pure xenon [6]. A similar result was later obtained in a xenon cluster jet operated at high backing pressures [7]. Strong amplification at 32.8 nm for the 4d– 4p transition of Ni-like krypton has also been demonstrated with a 760-mJ, 30-fs laser pulse [8]. Recently, enhancement of OFI x-ray lasing in preformed plasma waveguides driven by discharge was reported. The gain media were a capillary tube filled with xenon and buffer gas [9] or a multimode capillary waveguide [10]. By seeding an OFI x-ray laser amplifier with a pulse produced from high harmonic generation, a 20-fold increase in the x-ray output was achieved and sub-picosecond x-ray pulse was produced [11]. In this paper, we demonstrate low-threshold lasing of the 32.8-nm Kr8+ 4d–4p line by using tight-focusing pumping configurations. In the meantime, we use gas jets as the gain media to avoid accumulated optical damage. The combination of low pumping threshold and gas-jet media facilitates long-time high-repetition-rate operation, which is an important aspect for practical applications. We show that low-threshold OFI x-ray lasing can be achieved by tight focusing the pump laser at an optimal focal position. Both computer simulation and experimental diagnosis show that the balance between the converging power of the focusing optics, diffraction, and ionization defocusing can increase the gain length significantly as the spatial profile of the pump beam is well controlled at the optimal size. Near saturated lasing was observed at a pump energy of only 200 mJ, and the lasing threshold is only 70 mJ. In order to further improve the efficiency and beam quality of OFI x-ray lasers, a preformed plasma waveguide is developed to guide the intense pump pulse in OFI plasmas. The guided laser pulse can delivered a high pump intensity in a smaller diameter at a considerably low pulse energy. The interaction length and thus the gain-length product of x-ray lasers can also be extended significantly. We show that the plasma waveguide can be efficiently produced in a Kr/H2 cluster jet operated at a high backing pressure with axicon ignitor-heater scheme [12]. More than 70% of the pump energy in the vacuum focal spot is observed to transmit through the all optically preformed waveguide. The waveguide with buffer gas H2 can be used to develop recombination OFI x-ray lasers as electrons generated from high-density hydrogen are relatively cool [13]. On the other hand, strong amplifications of collisional excitation OFI Kr x-ray laser at 32.8

Generation of Efficient Optical-Field-Ionization X-Ray Lasers in Cluster Jets 237

nm is observed in pure Kr waveguide. The spectral brightness of Kr lasers amplified in waveguides is enhanced by a factor of > 100 compared to that obtained with tight focusing configuration.

2 Experimental Setup The experimental setup was similar to that described in Ref. 7, 12. A 10TW, 45-fs, 810-nm, and 10-Hz Ti:sapphire laser system (upgraded from the laser system in Ref. 14) based on the chirped-pulse amplification technique was used in this experiment. The pump pulse was focused by an offaxis parabolic mirror of 30-cm focal length onto a cluster jet. The focal spot size of the pump pulse was 7.5-µm diameter in full width at half maximum (FWHM) with 80% energy enclosed in a Gaussian-fit profile. A quarter-wave plate was used to change the pump polarization. The cluster jet was produced from a slit nozzle with a 5-mm×500-µm outlet and a round throat of 1.2-mm diameter. In the major axis of the outlet, the inner opening angle of the nozzle is 5.4°, and in the minor axis the inner width shrunk from 1.2 mm at the throat to 0.5 mm at the outlet. To produce a gas target with sharp boundaries, a rectangular aperture was mounted on the output of the cluster jet. So, the gas density profile had a flat-top region of 3.5 mm in length and a sharp boundary of 450 µm at both edges along the major axis. The average atom density increased linearly with the backing pressure and was measured to be 2.8×1017–3.2×1019 cm-3 for a backing pressure of 0.041–4.8 MPa. The preformed plasma waveguide was produced in pure krypton gases or the mixture of krypton and hydrogen gases with the axicon ignitor-heater scheme. The ignitor was a compressed 45 fs pulse, and the heater was the uncompressed beam with a pulse duration of 80 ps. They were focused by an axicon of 30° base angle to a line focus of ~8-mm length. The longitudinal intensity distribution of the line focus can be approximately controlled by a beam expander installed before the axicon. A relay-imaging system was used to measure the injection beam profile at the end of the plasma waveguide to verify the guiding of the injection beam. The on-axis diagnostic was a flat-field spectrometer (FFS) made of an aperiodic grazing-incidence grating with an average groove density of 1200/mm and a back-illuminated 16-bit soft x-ray CCD camera. A 0.5 µm-thick Al filter is used to block the pump beam and to attenuate x-ray emission. A dipole magnet was put in front of the spectrometer to deflect the electrons emitted from the interaction region, guarding against false signals from electrons in the x-ray CCD camera. Mach-Zehnder interferometry with a probe pulse passing transversely through the cluster jet

238

M.-C. Chou et al.

Fig. 1. X-ray laser output as a function of the focal position at atom density of 1×1018 cm-3. The pump energy was 200 mJ and the pump polarization ellipticity was 0.55. The effective f-number of the focusing optics is 7. Inset: Kr x-ray laser spectrum at 32.8 nm.

was used to measure plasma density distribution 5 ps after the pump pulse. The propagation of the pump pulse in the cluster jet can be analyzed from the interferograms. By calibrating the grating reflectivity, the filter transmittance, and the CCD response, the absolute emission yield was obtained.

3 Experimental results and discussions 3.1 Low-threshold OFI Kr lasers with tight focusing pumping To create a uniform and elongated plasma column for x-ray amplification, we scan the focal position of the pump laser in vacuum relative to the entrance of the gas target along the laser propagation direction. At each position, the diameter and the wave front curvature of the pump beam at the front boundary of the gas target is different, so is the nonlinear evolution of the beam profile in the plasma. Fig. 1 shows the strong dependence of the lasing signal on the focal position. The inset shows the on-axis time integrated Kr x-ray laser spectrum for a pump polarization ellipticity of 0.55 at an atom density of 1×1018 cm-3 and a focal position of 2.2 mm behind the front edge of the gas target. When the laser pulse is focused at the front edge, the intensity is about 5.5×1018 W/cm2, which is much higher than needed for producing the correct ionization balance. Over-ionization depletes Kr8+ ions and shortens the gain length. In addition, overproduction

Generation of Efficient Optical-Field-Ionization X-Ray Lasers in Cluster Jets 239

Fig. 2. Dependence of the calculated gain length on the focal position for different focusing optics. The parameters used in the simulation are the same as in Fig. 1, except that the focal length of the pump laser is varied.

of free electrons promotes ionization defocusing which makes the laser beam diverge quickly. As the laser intensity drops below OFI threshold of Kr8+ ions, the lasing gain is also terminated at the rear portion of the gas jet. On the contrary, if the focal spot is moved to the rear edge of the gas jet, the beam diameter at the front edge becomes relatively large. The pump intensity becomes too low to generate lasing ions at the front portion of the gas jet. Consequently there exists an optimal focal position to achieve a good balance among beam convergence, diffraction, and ionization defocusing for the laser beam to maintain its intensity over a long distance. To study and optimize the pump beam propagation in OFI plasmas, a source dependent expansion method [15] is used to numerically solve three-dimensional nonlinear wave equations for the pump pulse. The computation takes account of photoionization, ionization energy depletions, and laser propagations in a plasma with spatially and temporally varying density. Since the lasing plasma is predominately generated by the laser pulse via optical-field ionization, the ion density, electron density, and photonionization rate in specific stages of ionization can be calculated using the Ammosov-Delone-Krainov formulas for a given field [16]. In turn, the spatiotemporal electron and ion density profile determine the propagation of the pump pulse. These two processes are described by a system of nonlinearly coupled equations which can be numerically integrated. Fig. 2 shows that the dependence of the calculated effective length of the gain re

240

M.-C. Chou et al.

Fig. 3. X-ray lasing signal and calculated gain length as a function of pump energy. Solid square: lasing signal at 32.8 nm produced by a pump pulse with a polarization ellipticity of 0.55 at an atom density of 1×1018 cm-3. The focal position is at 2.2 mm behind the front edge of the gas target. Solid line: calculated gain length under the same condition.

gion on the focal position for various focusing configuration. All the parameters used in the simulation are the same as in Fig. 1, except that the focal length of the pump laser is varied. The result shows that the x-ray amplification length is increasingly more sensitive to the focal position as the f-number decreases. The length of the gain region is clearly enhanced by tight focusing at an optimal position. The dependence of the lasing output on the pump energy was measured to explore how OFI lasers work under low-energy pumping. As shown in Fig. 3, it was observed that the x-ray laser output was enhanced dramatically (up to factor of 1000) from pump energy of 70 mJ to 200 mJ. In the small signal region, the laser output is expected to increase exponentially with the length of the gain region. Simulations reveal that the effective amplification length increases monotonically with the pump energy when the pump beam is focused at the optimal position. The departure of simulation results from experimental data at large pump energy indicates that the laser gain may have started to saturate, as the saturation effect is not included in the simulation model. A small signal gain coefficient of 61±9 cm-1 is derived from the experimental and simulation data at the pump energy up to 130 mJ in Fig.3. A gain-length product of 15.9±1.3 is expected when the pump energy is 200 mJ. The absolute yield under 200-mJ pumping is estimated to be ~1×109 photons/pulse. This value is close to that of

Generation of Efficient Optical-Field-Ionization X-Ray Lasers in Cluster Jets 241

saturated Kr x-ray lasers reported in Ref. 8 with a 760-mJ, 30-fs pump pulse. It is interesting to note that in our experiments the optimal polarization ellipticity for maximum x-ray lasing was found to be in the range of 0.5– 0.6 regardless of the atom density. The result can be explained by the preionization of the clusters at the ionization front. If collisional heating and ionization at the ionization front can ionize Kr atoms to Kr3+ or Kr4+, these pre-ionized electrons may lose their energy via radiative cooling during the nanoplasma expansion and through collision with other colder electrons or surrounding neutral atoms during the uniform plasma expansion. As a result, upon the arrival of the main peak of the pump pulse, the energies of these pre-ionized electrons may fall below the threshold of collisional excitation and thus they cannot contribute to collisional excitation. This will greatly reduce the gain. A reduction of the polarization ellipticity toward linear polarization can shift the electron energy spectrum toward the lowenergy end for subsequent ionization stages from Kr3+ or Kr4+, thereby increases the collisional excitation rate. The observation suggests that by using time-domain polarization shaping [17], it may be possible to further optimize the electron energy spectrum in accordance with the ionization stages. This will further increase the gain and efficiency. 3.2 Generation of OFI Kr lasers in preformed plasma waveguide Figure 4(a) shows the interferogram at an ignitor-heater separation of 200 ps and a probe delay of 2.5 ns with respect to the heater in the 5-mm Kr/H cluster jet operated at 630-psi backing pressure. The ratio of Kr to H2 in the gas reservoir of the cluster jet was 1:11. The pulse energy for ignitor and heater were 43 mJ and 290 mJ, respectively. It was found that plasma waveguide cannot be generated if only the ignitor or the heater was used. Instead, a uniform plasma waveguide extended to ~4.4 mm was observed when both ignitor and heater were deployed. These results show that the plasma waveguide can be produced much more efficiently when the axicon scheme is used in conjunction with the ignitor-heater scheme. When only a short laser pulse is used, even though it has sufficient energy to ionize the gas, the pulse duration is too short to heat the plasma via inverse bremsstrahlung absorption. When only a long pulse is used, the laser intensity is too low to produce seed electrons so that there is no inverse bremsstrahlung absorption to drive the formation of plasma waveguide. In the ignitor-heater scheme, the short low-energy pulse first ionizes the gas to provide seed electrons by multi-photon ionization, so that the following long high-energy pulse can be absorbed efficiently to drive the expansion

242

M.-C. Chou et al.

Fig. 4. Interferogram (a) and beam profile of the injected pump beam at ~ 5 mm after its focus (b) for various cases in the mixture gas of 1:11 Kr/H at a backing pressure of 630 psi. The ignitor-heater separation is 200 ps.

and the subsequent collisional ionization which makes the plasma density higher in the outer region than in the center of the beam. Guiding of the pump beam was verified from the beam profile imaging at the end of the plasma waveguide for an injection delay of 2.5 ns. As shown in Fig. 4(b), the pump pulse can not be guided at a small beam diameter without the presence of the plasma waveguide. The guided beam size was measured to be ~ 14 µm (FWHM) with more than 60% of the energy in the vacuum focal spot transmitting through the preformed waveguide. It is also noted that the guiding efficiency of the Kr/H plasma waveguide was enhanced to ~75% when the energy of the pump pulse increased from 15 mJ to 210 mJ. The pedestal of the high-energy pump pulse may induce pre-ionization and heating of plasmas to enlarge the density dip for a better guiding. Although high quality plasma waveguide can be generated in the mixture of Kr and H2 gases, it is believed that cold electrons produced during the ionization of hydrogen atoms can promote strong recombination in Kr8+ ions and thus reduce the lasing gain significantly. In our experiments, the energy of the pump pulse was varied systematically to obtain the optimal pump intensity for producing the correct ionization balance in the plasma waveguide. No lasing signal of Ni-like Kr 4d-4p line was observed. The effective gain-length product in plasma waveguide may become too small to produce a strong lasing output as the small signal gain coefficient is reduced dramatically.

Generation of Efficient Optical-Field-Ionization X-Ray Lasers in Cluster Jets 243

Fig. 5. Interferogram and beam profile of the injected pump beam at ~ 5 mm after its focus for various cases in pure Kr at a backing pressure of 200 psi. The ignitor-heater separation is 200 ps.

Without doping hydrogen as the buffer gas, the high efficiency plasma waveguide can not be generated in the pure Kr as shown in Fig.5. Only ~10% of the energy in the vacuum focal spot is observed to transmit through the preformed waveguide. For pump energy of 210 mJ. The laser intensity at the exit of the waveguide was estimated to be 1.5 × 1017 W/cm2 which is high enough to ionize Kr atoms to Ni-like ion state. Strong lasing signals from Kr8+ 4d–4p transition at 32.8 nm was detected by the on-axis x-ray flat-field spectrometer. Compared to the optimal lasing output with tight focusing pumping, more than 4 times of x-ray photons were generated in the Kr plasma waveguide with a small divergence angle of ~ 6 mrad. This indicates the brightness of Kr lasers can be enhanced by a factor of > 100 when the preformed plasma waveguide is presented.

4 Summary In summary, a low-threshold configuration of OFI x-ray laser in krypton cluster jets was demonstrated. The reduction of the pump energy is achieved by carefully balancing the focusing, diffraction, and ionization defocusing to maintain a high-intensity pump profile over an extended length. In contrary, an efficient method for generating extended plasma waveguides is developed by using the axicon lens in conjunction with the ignitor-heater scheme. High efficiency plasma waveguides in Kr/H mixed gas jets are generated with more than 70% of pump energy observed to transmit through a ~5-mm long waveguide channel. This high efficiency plasma waveguide is particularly suitable for producing recombination OFI x-ray lasers as the electron generated from hydrogen can be very cool to promote recombination. The generation of collisonal excitation Ni-like Kr x-ray lasers at 32.8 nm in a 5-mm pure Kr waveguide at a pump energy

This page intentionally blank

244

M.-C. Chou et al.

of only 210 mJ is also demonstrated. The brightness of Kr lasers amplified in waveguides is enhanced by more than 2 orders of magnitude compared to that obtained with free propagation, tight focusing pumping. In order to further increase the x-ray laser efficiency, the ratio between Kr and H2 has to be optimized to obtain a reasonable gain coefficient and guiding efficiency in the preformed plasma waveguide. Such that the gain-length product of x-ray lasers can be maximized while the pump energy can be kept at low value. With time-domain polarization shaping, it is possible to further reduce the threshold and increase the efficiency. We believe this is a promising direction toward the development of low-cost, high-repetitionrate soft x-ray lasers for practical applications.

References 1. Nagata, Y. et al., Phys. Rev. Lett. 71, 3774 (1993). 2. Lemoff, B.E. et al., Opt. Lett. 19, 569 (1994). 3. B.E. Lemoff B. E et a., Phys. Rev. Lett. 74, 1574 (1995). 4. Korobkin D.V. et al.,Phys. Rev. Lett. 77, 5206 (1996). 5. Hooker S.M., Epp P.T., and Yin G.Y., J. Opt. Soc. Am. B 10, 2735 (1997). 6. Sebban, S. et al., Phys. Rev. Lett. 86, 3004 (2001). 7. Chu H.-H. et al., Phys. Rev. A 71, 061804(R) (2005). 8. Sebban S. et al., Phys. Rev. Lett. 89, 253901 (2002). 9. Butler A. et al., Phys. Rev. Lett. 91, 205001 (2003). 10. Mocek T. et al., Phys. Rev. A 71, 013804 (2005). 11. Zeitoun Ph. et al., Nature 431, 426 (2004). 12. Xiao Y.-F. et al., Phys. Plasmas 11, L21 (2004). 13. Grout M.J. et al., Opt. Commun. 141, 213 (1997). 14. Chu H.-H. et al., Appl. Phys. B 79, 193 (2004). 15. Sprangle P., Penano J.R., and Hafizi. B., Phys. Rev. E 66, 046418 (2002). 16. Ammosov, M.V., Delone N.B., and Krainov V.B., Soviet Phys. JETP 64, 1191 (1986). 17. Brixner T. and Gerber G., Opt. Lett. 26, 557-559 (2001).

Experimental Investigation of the Parameter Space of Optical-Field-Ionization Collisional-Excitation X-Ray Lasers in a Cluster Jet M.-C. Chou1,2, P.-H. Lin1,3, T.-S. Hung1,2, J.-Y. Lin2, J. Wang1,3,4 and S.-Y. Chen1,4 1

Institute of Atomic and Molecular Sciences, Academia Sinica, Taipei 106, Taiwan

2

Department of Physics, National Chung Cheng University, Chia-Yi 621, Taiwan

3

Department of Physics, National Taiwan University, Taipei 106, Taiwan

4

Department of Physics, National Central University, Chun-Li 320, Taiwan

Summary. Optical-field-ionization collisional-excitation soft x-ray lasers in clustered gas jets were experimentally investigated in detail. A tomographic measurement based on laser machining technique was used to resolve the growth of x-ray lasing intensity as a function of position in a cluster jet, exploring the origins of the dependences of x-ray lasing intensity on atom density and laser polarization. The presence of optimal atom density was found to be a result of ionizationinduced refraction. An unexpected observation is that circular polarization appears to be not the optimal polarization ellipticity, which may be a manifestation of effects of pre-ionization at the laser-cluster ionization front.

1 Introduction The first saturated operation of collisional-excitation optical-fieldionization (OFI) x-ray laser was demonstrated with 5d-5p transition at 41.8 nm of Pd-like Xe in a xenon gas cell [1]. A similar result was later achieved in a Xe clustered gas jet [2]. Strong amplification at 32.8 nm for 4d-4p transition of Ni-like Kr has also been demonstrated with a 760 mJ, 30 fs laser pulse [3]. Recently, enhancement of OFI x-ray lasing in a preformed plasma waveguide driven by discharge in a capillary tube filled with xenon and buffer gas [4] and that in a multi-mode capillary waveguide [5] were reported. In addition, by seeding an OFI x-ray laser

246

M.-C. Chou et al.

amplifier with a pulse produced from high harmonic generation, enhancement of the x-ray laser intensity and quality was achieved [6]. Here we report the investigation of the parameter space of OFI collisional-excitation x-ray lasers in a clustered gas jet for Pd-like Xe at 41.8 nm and Ni-like Kr at 32.8 nm. A tight focusing configuration was used to explore how the lasers work under low-energy pumping. A tomographic measurement technique based on laser machining was used to resolve the growth and decay of the x-ray lasing intensity as a function of position in the cluster jet. Maximum x-ray lasing was found to appear at a pump polarization ellipticity other than circular polarization. This may be ascribed to the effect of pre-ionization during the laser-cluster interaction at the ionization front.

2 Experimental setup A 10-TW, 45-fs, 810-nm, and 10-Hz Ti:sapphire laser system based on chirped-pulse amplification technique (upgraded from the laser system in Ref. 7) was used in this experiment. A 210-mJ, 45-fs pump pulse was used for preparation of the lasing ionization stages and above-thresholdionization (ATI) heating of electrons, and a 30-mJ, 45-fs pulse set to be 6 ns earlier than the pump pulse was used as the machining pulse for the tomographic measurement to be described below. The pump pulse was focused with an off-axis parabolic mirror of 30-cm focal length onto a cluster jet and the focal spot size was 10-μm in diameter. The cluster slit-jet has a 3.5 mm flat-top region with 750-μm slopes at both edges. Propagating perpendicularly to the pump pulse, the machining pulse was imaged from the location of a knife edge onto the interaction region by a spherical lens of 20-cm focal length with a demagnification factor of 3. In the meantime it was focused in the vertical direction to a width of 20 μm by this spherical lens in combination with a cylindrical concave lens of 75-cm focal length. Based on the laser machining technique as that in Ref. 8, by scanning the knife-edge position the end of the region in which the pump pulse interacts with clusters was varied and the growth of x-ray lasing intensity with pump-pulse propagation in the cluster jet was resolved tomographically. The primary diagnostics for x-ray was a flat-field grazingincidence x-ray spectrometer consisting of an aperiodically ruled grating and a 16-bit x-ray charge-coupled device (CCD) camera. The spectrometer was used to measure the x-ray lasing spectrum and the lasing divergence angle in the direction of pump laser propagation. Aluminum filters of 0.5 μm thickness were used to block transmitted pump laser pulses and attenu-

Experimental Investigation of the Parameter Space of Optical-Field

247

ate x-ray emission. By calibrating the grating reflectivity, the filter transmittance, and the CCD response, the absolute emission yield was obtained.

3 Dependence on atom density The inset in Fig. 1(a) shows the Xe x-ray spectrum. The photon number of the 41.8 nm lasing line reached about 109 photons and the divergence angle was 13 mrad. The dependence of number of photons for Pd-like Xe on atom density is shown in Fig. 1(a). It was observed that an optimal atom density for maximum lasing is present at 7.4×1017 cm-3, which agrees with the number found in previous experiments using a gas cell [1]. It is expected that increase of atom density should result in a larger gain and thus a larger lasing photon number. However, if the atom density is increased beyond an optimal value, the lasing photon number is expected to drop as a result of ionization-induced refraction [1]. The ionization-induced refraction refers to the defocusing of a laser beam caused by the plasma with a higher on-axis density than that at the radially outer region produced by optical-field ionization. Similar results for 32.8 nm x-ray lasing were observed. The inset in Fig. 1(b) shows the Kr x-ray spectrum. The photon number of the 32.8-nm lasing line reached about 109 photons and the divergence angle was 14 mrad. The dependence of number of photons for Ni-like Kr lasing at 32.8-nm on atom density is shown in Fig. 1(b). The optimal atom density was 1.2×1018 cm-3.

Fig. 1. Number of photons of Pd-like Xe lasing at 41.8 nm, (a), and Ni-like Kr lasing at 32.8 nm, (b), as functions of atom density. Insets in (a) and (b) show the x-ray lasing spectrum at atom densities of 7.4×1017 cm-3 for Xe and 1.2×1018 cm-3 for Kr. The pump polarization ellipticity were 0.65 for Xe and 0.55 for Kr. The energy of the pump pulse was 210 mJ and the focal position of the pump pulse was 2500 μm after the entrance of the cluster jet both for Xe and Kr.

248

M.-C. Chou et al.

To gain further insight on the decrease of lasing photon number at higher atom densities, we used the tomographic technique to measure the lasing photon number of Pd-like Xe at 41.8 nm as a function of position in the cluster jet at various atom densities, as shown in Fig.2. The results reveal that the decreased lasing photon number at an atom density higher than the optimal value comes not only from the decrease of gain length and increase of length of absorption as a result of ionization-induced refraction, but also from a reduction of gain. Aside from ionization-induced refraction, line-width broadening due to increased ion momentum obtained during dissociation of larger clusters and/or increased rate of collisional deexcitation of the upper level may also contribute to the reduction of gain.

Fig. 2. Number of photons of Pd-like Xe lasing at 41.8 nm as a function of position in the cluster jet for various atom densities. The energies of the pump pulse and the machining pulse were 210 mJ and 30 mJ, respectively. The pump polarization ellipticity was 0.65 and the focal position of the pump pulse was 2500 μm after the entrance of the cluster jet. The temporal separation between the machining pulse and the pump pulse was 6 ns.

4 Dependence on polarization ellipticity Figure 3 shows the number of photons of Pd-like Xe lasing at 41.8 nm and Ni-like Kr lasing at 32.8 nm as a function of pump polarization for various atom densities. The optimal polarization ellipticity for maximum x-ray lasing was found to be in the range of 0.6-0.7 for Xe and 0.55 for Kr regardless of the atom density. This shows that the effects of atom density

This page intentionally blank

Experimental Investigation of the Parameter Space of Optical-Field

249

and polarization ellipticity are decoupled. We observed unexpectedly that the optimal pump polarization ellipticity for maximum x-ray lasing is not the circular polarization.

Fig. 3. Number of photons of Pd-like Xe lasing at 41.8 nm, (a), and Ni-like Kr lasing at 32.8 nm, (b), as a function of pump polarization for various atom densities. The energy of the pump pulse was 210 mJ and its focal position was 2500 μm after the entrance of the cluster jet.

To further examine the x-ray amplification process under different polarization ellipticities of the pump pulse, the photon number of Pd-like Xe lasing at 41.8 nm as functions of position in the cluster jet were measured using the tomographic method, as shown in Fig. 4. This result indicates that a mechanism that changes the electron energy spectrum produced by ATI heating was present and resulted in a decreased collisional-excitation rate when the polarization ellipticity was larger than the optimal value. One possible explanation is the effects from pre-ionization of the clusters at the ionization front. If the collisional heating and ionization in the lasercluster interaction at the ionization front can ionize Xe atoms to Xe3+ or Xe4+, these pre-ionized electrons may lose their energy via radiative cooling during the nanoplasma expansion and through collision with other colder electrons or surrounding neutral atoms during the uniform plasma expansion. As a result, upon the arrival of the main peak of the pump pulse at roughly 2 ns later [2] the energies of these pre-ionized electrons may fall below the threshold of collisional excitation and thus they cannot contribute to collisional excitation. This will greatly reduce the lasing photon number. However, under this circumstance, if the ATI energies of the electrons produced from the subsequent ionization stages by the main peak of the pump pulse are lowered to fall in the region above and near the excitation threshold, in which the collisional-excitation cross section is larger, the gain can be raised. This can be achieved by shifting the polarization ellipticity of the pump pulse away from the circular polarization, resulting in

250

M.-C. Chou et al.

Fig. 4. Number of photons of Pd-like Xe lasing at 41.8 nm as functions of position in the cluster jet for 7.4×1017 cm-3 atom density and various pump polarization ellipticities. The energies of the pump pulse and the machining pulse were 210 mJ and 30 mJ, respectively. The focal position of the pump pulse was 2500 μm after the entrance of the cluster jet. The temporal separation between the machining pulse and the pump pulse was 6 ns.

the observed optimal polarization ellipticity at less than 1. This also explains the decoupling of optimal atom density and optimal polarization ellipticity, because by the time the main peak of the pump pulse arrives to heat up the plasma again, the cluster structure has already disintegrated. Also note that in Fig. 4 the position of maximum lasing photon number does not shift with variation of laser polarization ellipticity. This is reasonable since the ionization fraction is not very different for varying polarization ellipticity when the laser intensity is well above the ionization intensity threshold.

References 1. 2. 3. 4. 5. 6. 7. 8.

S. Sebban et al., Phys. Rev. Lett. 86, 3004 (2001) H.-H. Chu et al., Phys. Rev. A 71, 061804(R) (2005) S. Sebban et al., Phys. Rev. Lett. 89, 253901 (2002) A. Bulter et al., Phys. Rev. Lett. 91, 205001 (2003) T. Mocek et al., Phys. Rev. A 71, 013804 (2005) Ph. Zeitoun et al., Nature 431, 426 (2004) H.-H. Chu et al., Appl. Phys. B 79, 193 (2004) C.-H. Pai et al., Phys. Plasmas 12, 070707 (2005)

Collisionally Pumped X-Ray Lasers for Shorter Wavelengths G.J. Pert Department of Physics, University of York, York, YO10
5DD, U.K.

Summary. The scaling of the pumping requirements for X-ray lasers in the range 50-100Å is investigated using analytic models for the interaction of the laser with solid targets. It is found that due to the rapid increase of the electron temperature required to establish Ni-like ionisation that the irradiance needed in the pre-pulse to generate the plasma increases rapidly with atomic number. In particular the mass of the heated plasma is dominated by upstream thermal conduction. In contrast due to the high temperatures achieved in the pre-pulse relatively little energy is required to further heat the plasma to generate strong gain. The large mass of upstream plasma at wavelengths ~50Å is comparable with that in freestanding foils, and suggests the use of the latter for targets at shorter wavelengths. The validity of these models is confirmed by simulation.

1 Introduction Soft X-ray lasers using grazing incidence pumping and nickel-like ions are now established as efficient sources of radiation in the super-100Å waveband [1]. Underlying this approach is the separation of the ionisation and excitation phases of the laser, and matching the operating density to the optimum for pumping the population inversion [2]. The first ionisation pulse (or pre-pulse) tends to be long and the second excitation (or main pulse) short. In this paper we examine whether these ideas can be applied to sub-100Å systems. We find two problems arise: firstly the optimum pumping density is at or above the critical density of the fundamental of the pumping laser, and secondly that the scaling of the electron temperature required to generate the Ni-like ionisation stage scales adversely with atomic number. The laser irradiation conditions required to achieve the necessary background plasma are identified by using appropriate analytic models of laser-plasma interaction developed many years ago. We find that, at the temperatures necessary for Ni-like ionisation, thermal conduction plays an important role, dominating the energy transfer upstream rather than

252





G.J. Pert

downstream into the expanding plasma plume. This has the advantage that the plasma at densities in excess of that of the absorption layer will be hotter than predicted directly from the standard expansion models, and will allow more rapid ionisation and a higher pumping density. Using simulation applied to a variety of representative atomic systems suitable for generating laser action at these wavelengths, we identify a strategy which offers a possible route to achieving these aims reasonably efficiently. This is based on a pre-pulse comprising both fundamental and harmonic radiation generated from the pump laser, and a normally incident fundamental main pulse. However we find that the pump laser energy required scales very adversely with the atomic number and therefore wavelength to the extent that typically about 100J/cm for a target of width 100µm is required in the pre-pulse to ionise the shortest wavelength systems such as tantalum or ytterbium lasing at 45Å or 50Å respectively. In practice these values which neglect any transverse heat losses are probably an underestimate of those needed in an experiment. At these high pre-pulse energies a relatively small main pulse of only a few 10sJ/cm is needed to generate the gain. Energies typically of a few 10s of µJ are generated from these designs.

2 Analytic Models Simple one dimensional analytic models for the interaction of laser on solid targets were developed many years ago. These differ only in the nature of the absorption process and the role of thermal conduction. 2.1 Deflagration Model In this regime [3] the plasma flow is assumed steady state with localised absorption at the critical density,
c, by unspecified mechanisms. The upstream flow, and downstream isothermal rarefaction are both supported by thermal conduction from the absorption layer. The flow is therefore a classical Chapman-Jouget deflagration flow, with heat input given by the irradiance. The electron temperature is given by

Collisionally Pumped X-Ray Lasers for Shorter Wavelengths





253

Te
0.397

1 2 3  c 2 3 Z R g

where Z is the ionisation,

(1)

R g the gas constant per unit mass and


= 1 + Ti Z Te . The above equation neglects the energy required for ionisation and that held in the conduction zone, hence the effective irradiance
 is significantly less than that actually required. 2.2 Self-regulating Model In this case absorption is global by inverse bremsstrahlung absorption, and thermal conduction is neglected [4]. Since the plasma grows in time, the model is not steady state [5].

Te
0.423

1 b 1 4   1 2 1 4 Z R g

(2)

where
is the pulse length and b the appropriate inverse bremsstrahlung constant. The characteristic absorption density at the isothermal sonic point

 a
0.784 b  3 8   1 4  3 8

(3)

This model is closely related to that above when the time dependency and the different absorption density are taken into account 2.3 Thermal Conduction Zone Upstream of the absorption layer in either model, thermal conduction will be important if the temperature is high. This gives rise to layers of hot plasma, which over long times will pass through the absorption zone. However if the pulse is short (as is the case here) the upstream plasma will have a substantial heat content at the end of the pulse which must be included in the overall energy balance. The length, mass and energy of this zone may be expressed in terms of the characteristic length [6]

This page intentionally blank

254





G.J. Pert

2ac 4 L= a

c2 =


Z kTe M

(4)

determined by the values at the absorption layer. For a flux limiting factor of 0.1, the length, mass and energy of the layer take the values: Length 0.0629L : Mass 0.1767a L : Energy 0.1089ac2L

(5)

where a is the characteristic constant for electronic thermal conduction.

3 Ionisation We may estimate the minimum temperature needed to achieve the Ni-like ionisation stage in the plasma at the conclusion of the laser pulse from that needed for equilibrium at the appropriate ionisation stage. Using the screened hydrogenic model for the ionisation and excited state energy levels, we find that the ionisation temperature scales only weakly with electron density over the density range of interest, but strongly with atomic number Z:

Te
Aion Z 5.5

(6)

where the constant Aion varies with ionisation stage and density. For the Nilike ion and electron density 1021 cm-3 we find Aion
5.8710-8 eV. Using this temperature (6) we may use equations(1), (2) and (5) to calculate the minimum irradiance needed to achieve the Cu-like ionisation , since this has a sufficient population of Ni-like ions to generate gain. As a set of scaling relations we find for the irradiance

 Ade Z 8.25  Ace Z 13.4 
 e 10 +   Asr Z 

(7)

and for the mass ablation rate

 A m Z 3  A m Z 9 µ&
 dm 4.5  + c   Asr Z 

(8)

Collisionally Pumped X-Ray Lasers for Shorter Wavelengths





255

where Ade , Asre and Ace , and Adm , Asrm and Acm refer to the empirical coefficients from the deflagration, self-regulating and thermal conduction models respectively. The self-regulating/deflagration transition is determined by the absorption density, namely if
a <
c the flow is selfregulating and vica versa. 3.1 Ionisation Time Thus far we have assumed that the ionisation takes place over times short compared to the pulse duration. To check this assumption is valid we show in fig. 1 the time taken to reach the Cu-like ionisation stage at the Ni-like equilibrium temperature. It can be seen that this time is nearly linearly proportional to the electron density. Furthermore we note that at an electron density of 1021 cm-3 the time is about 200ps. for Z=70 and increases rapidly thereafter. We therefore conclude that we require prepulse durations of at least 100ps. 1.5 20

-3

Ionization time (ns.)

5 × 10 cm 21 -3 1 × 10 cm 21 -3 4 × 10 cm 22 -3 1 × 10 cm 1

0.5

0 40

50

70 60 Atomic number

80

90

Fig. 1. Plot of the time required to achieve the Cu-like ionisation at the equilibrium temperature of the Ni-like ion as a function of the atomic number

4. Optimum Pumping Density It can be shown that there are electron densities at which the population inversion density and the gain are maximised. Fig.2 shows the variation of the fractional inversion density and a parameter measuring gain defined below in dysprosium over a range of electron densities and temperatures.

256





G.J. Pert

  Population inversion fraction = 1  q 2
q1  q0  g 2 g1 

Gain parameter = Population inversion fraction
1020 ne

(9)

It is clear from Fig.1 and Fig.2 that typical conditions after the final heating require an electron density of about 1021 cm-3 to achieve both ionisation with a pulse length of about 100ps, and high gain. This is the critical density for the fundamental of Nd-glass laser radiation at 1.06µm. However since the absorption in these systems is dominated by inverse bremsstrahlung absorption, this suggests that we need to use the harmonic radiation at 0.53µm as the pre-pulse to generate the ionisation and create a suitable density profile. 10 500eV 1000eV 2000eV 3000eV

1

Gain

er

met

para

Population fraction

-1

10

-2

10

Inversion fraction -3

10

-4

10

-5

10

19

10

20

10

21

10 -3 Electron density (cm )

22

10

23

10

Fig. 2. Plots of the inversion density and gain parameter in dysprosium as functions of the electron density for different temperatures.

Unfortunately, harmonic generation is typically only about 50% efficient, the remaining 50% remaining in the fundamental. To obtain maximum efficiency therefore it is advantageous to focus both harmonic and fundamental on to the plasma, if a suitable focusing arrangement can be designed. We will assume this can be done. From equation (6) we see that for dysprosium, the temperature required for ionisation is about 500eV. From fig.2 it is clear that significant gain can be developed by the pre-pulse alone. Furthermore it can be seen that

Collisionally Pumped X-Ray Lasers for Shorter Wavelengths





257

an increase in temperature of a further 500eV, which is achieved without additional ionisation or hydrodynamic effects is all that is required from the main pulse. The energy needed by the main pulse is therefore substantially less than that in the pre-pulse.

4 Examples from Simulation The analytic modelling presented above has been entirely one dimensional, appropriate for a plasma width of 100µm. We will use the 1_d code EHYBRID to simulate such plasmas, where the hydrodynamics are essentially 1 dimensional. The pre-pulse will be assumed to have a constant irradiance over the full pulse length of 100 or 200ps and the main pulse a Gaussian temporal profile of 2ps half-width. The ionisation is calculated using screened hydrogenic energy levels for all ions except the lasing Ni-like stage. 4.1 Dysprosium (Z=66) Fig.3 shows the gain and electron density spatial profiles along the plasma axis at increasing times measured from the onset of the main pulse. It can be seen that the maximum gain increases with time until about 10ps and decreases slowly thereafter. However we note that increasing gain is associated with a shift to higher density. This is due to thermal conduction heating plasma at densities greater than the absorption density, with consequent higher gain as reflected in fig.2. However we also note that this move to higher density is accompanied by an increase in the gradient of the electron density, leading to progressive deterioration of the output due to refraction. We note that at the electron density of 1021 cm-3, the gain decreases slowly as the plasma cools. Fig.4 shows output from plasma similar to that shown in fig.3. It will be noted that although the magnitude of the outputs differ by a factor of 10, their temporal profiles are similar, indicating that the temporal profile is controlled by the spatial profile of the expansion resulting from the prepulse. The peak power occurring at 7.5ps is significantly earlier than gain maximum, reflecting the limitation imposed by the density gradient as the gain moves to higher density. The relatively large change in output from a small change in main pulse energy should be noted, cosequent to increased temperature and stronger pumping, consistent with fig.2.

258





G.J. Pert

4.2 Ytterbium (Z=70) Ni-like ytterbium lases at 50Å. Fig.5 shows the gain and electron density profile in this case. It is immediately apparent that the pre-pulse energy is greatly increased over that needed for dysprosium (by a factor 4) despite the relatively small increase in the atomic number. This is larger than our

1e+23

80 70

−1

Gain (cm )

50

1e+22 −3

1e+21

40

Electron density (cm )

2 ps 6 ps 10ps 14ps 18ps

60

30 1e+20

20 10 0 −0.001

0

0.001 0.002 Distance (cm)

0.003

1e+19 0.004

Fig. 3. Plots of the gain at 59 Å and electron density for dysprosium pumped by a pre-pulse of mixed harmonic and fundamental of 23.75 J/cm in 200 ps and a fundamental main pulse of 8.8 J/cm in 2 ps all at normal incidence.

model predictions (factor 2), even when account is taken of the reduction in inverse bremsstrahlung absorption due to the higher temperature. The additional energy is required to meet the increasing demands of radiation escaping from the hot plasma. It can be seen that reasonable gain is generated over quite a large plasma body with a small density gradient lasting quite long times ~25ps. We therefore expect to obtain a relatively long output pulse at quite low power, but with significant total energy. Fig.6 confirms this expectation.

5 Foils at Shorter Wavelengths Referring to equation (8) we see that the mass scales rapidly with atomic number. Fig.7 illustrates this behaviour and shows that for Z > 67 the mass ablated exceeds 100µg. This suggests that we may limit the mass of ionised plasma by using thin foils, which are uniformly heated by thermal

Collisionally Pumped X-Ray Lasers for Shorter Wavelengths





259

conduction – exploding foils. The analysis of this system can be carried out reasonably easily and shows that the ratio of kinetic to thermal energies is about unity for a foil expansion. This is significantly larger than for slabs, where much of the mass is in the slowly moving conduction zone. Consequently the energy balance is in favour of slabs up to atomic numbers of about 73. 250 Power

Power (MW)

8.8J/cm 5.4J/cm (×10)

10

200

150

100

Energy

Energy (µJ)

Energy

20

Power 50

0

0

5

10 Time (ps)

0 20

15

Fig. 4. Plots of the output power and energy at 59 Å generated from a 1 cm. length of dysprosium plasma by pre-pulse irradiation with a pure harmonic beam of 23.75 J/cm in 200ps and a main pulse irradiation by normally incident fundamental beams of 5.4 J/cm and 8.8J/cm Note that in the 5.4 J/cm case the power and energy are both multiplied by a factor of 10.

23

60

Gain(cm

-1)

40

22

10

30

21

20

10

-3

2.5ps 5.0ps 7.5ps 10.0ps 12.5ps 15.0ps 17.5ps

50

Electron density (cm )

10

10

0

20

0

0.001

0.002 Distance (cm)

0.003

10 0.004

Fig. 5. Plots of the gain and electron density from ytterbium at 50 Å pumped by a combination harmonic/fundamental pre-pulse of energy 97 J/cm in 200 ps and a fundamental main pulse of 23 J/cm

260





G.J. Pert

10

80 70

8

50

Power Energy

6

40 4

Energy (µJ)

Power (MW)

60

30 20

2 10 0

0

10

5

15 Time (ps)

20

0 30

25

Fig. 6. Plot of the output power and energy at 50 Å from a 1cm plasma length of ytterbium heated by a pre-pulse of 97 J/cm in 200ps and main pulse 23 J/cm in 2 ps. 20000

Foil Slab

-2

-2

Mass (µg/cm )

Energy

10000 200 Mass

0 60

65

70 Atomic number

Total energy (J/cm )

400

75

0 80

Fig. 7. Comparison of the mass and energy required to achieve Cu-like ionisation in slabs and 100 µ g foils with a laser pulse of 100 ps duration at 0.5 µ m wavelength.

However foils have two further advantages. Firstly the density gradient at peak density is zero due to their symmetry. Secondly it is possible to use

Collisionally Pumped X-Ray Lasers for Shorter Wavelengths





261

them at high density by adjusting the pulse length so that the main pulse is applied before much expansion has occurred. This requires a short pulse and one must take care to ensure that thermal conduction has burnt through the plasma and that the ionisation is completed. With care operation at electron densities ~1022 cm-3 can be achieved.

6. Conclusions Simple analytic models have been used to identify the conditions necessary to pump Ni-like ions to generate laser action in the range 5070Å. It is found that ionisation is a limiting factor, both in respect of the pre-pulse input energy and the time required. Since the densities required both for ionisation and inversion are high > 1021 cm-3, harmonic radiation at 0.53 µm is needed, and grazing incidence pumping does not offer any advantage. At the high temperatures needed thermal conduction, even when flux limited, plays a key role. Although greatly increasing the energy demands, it does allow a mixture of the fundamental and harmonic wavelengths to be used, provided that the practical difficulties can be overcome. Since the temperature is high at the conclusion of the main pulse, a relatively small additional energy is required from the main pulse, which being short is not affected by hydrodynamic or conduction losses. The main pulse energy required is therefore only about _ of that in the prepulse. For shorter wavelengths it is suggested that foils with limited mass may allow a reduction in the pump energy, and improve the density and gradients in the plasma.

References 1. Y.Wang et.al. Phys. Rev. A 72, 053807, 2005 2. G.J.Pert Phys. Rev. A, 73, 033809, 2005 3 C.Fauquignon and F.Floux Phys. Fluids 13, 386, 1970 4 I.V.Afanas’ev et.al. Appl. Maths. Mech. 30, 1218, 1966 5 G.J.Pert J Plasma Phys. 36, 415, 1986 6 G.J.Pert J Plasma Phys. 29, 415, 1983

Inner Shell Lasing In Titanium via Electron Collisional Ionization E. F. Spracklen and G. J. Pert Department of Physics, University of York, York, YO10 5DD, UK

Summary. The possibility of generating gain in the x-ray region of the spectrum through the technique of electron collisional ionization is studied in titanium . As in previous work transient inversions of moderate size are seen to be produced on the L2 – M1 and L3 – M1 transitions provided that the collisional effects of the secondary and Auger electrons are ignored. Inclusion of these effects is seen to significantly decrease both the magnitude and duration of the gain obtained.

1 Introduction Inner-shell x-ray laser schemes present the possibility of generating gain at wavelengths of several angstroms in small scale devices. Conventionally such techniques have involved using x-ray photo-pumping to generate the inner-shell transitions as originally discussed by Duguay [1]. Widespread implementation of such devices has however been curtailed due to the difficulty of generating x-ray pulses with durations of the order of the inversion lifetimes. An alternative approach, outlined by Barty et al [2], is to use fs electron pulses to generate the inversion. However, although such pulses can be relatively easily produced via laser-plasma interactions, atomic electron impact cross-sections favor outer shell over inner shell ionization. Such techniques therefore cannot produce gain directly and instead rely upon Auger transitions to generate transient inversions. Initial studies into the possibility of generating gain in such a manner identified Ti as the most favorable candidate for lasing and predicted gains of up to 30 cm-1 [3]. The work in question however assumed mono-energetic electron pulses and ignored the collisional effects of the secondary and Auger electrons produced. Further simulations carried out using more realistic electron

264

E. F. Spracklen and G. J. Pert

fluxes predicted gains of only around 10-3 cm-1 [4]. Yet again however this work ignored the secondary electron effects. This paper investigates the possibility of generating gain on the L2 - M1 and L3 - M1 transitions in Ti using physically realistic electron pulses and taking full account of the collisional effects of secondary and Auger electrons.

2 Method The production of a hot electron flux via a laser-plasma interaction is simulated using a 2-D PIC code. The superthermal electrons produced are followed using Monte-Carlo methods with the titanium level populations obtained by numerical solutions of the relevant rate equations. Within this atomic code the secondary electron energies are calculated using the symmetrized binary encounter theory of Burgess and Percival [5]. In all the simulations performed the titanium lasing media is assumed to be initially un-ionised and at solid density.

3 Results Simulations performed using a variety of laser and plasma parameters predict larger gains upon the L3 – M1 transition in contradiction of the work of Barty et al [3] – figure 1. However, in Barty’s work the L2 and L3 levels are treated as having identical transition rates and electron impact cross-sections. The larger degeneracy of the L3 level therefore leads to smaller gains being generated upon the L3 – M1 transition. Treating the L2 and L3 levels separately it can be shown that L2 states possess faster filling rates and smaller impact cross-sections [6,7] explaining the larger L3 – M1 gains observed. Asides from this discrepancy it can be seen that the gains obtained are in the region of those predicted by Barty et al rather than those forecast by Upcraft [4]. Examining the spectra in more detail it is seen that the gain in the initial titanium regions is not consistent with that obtained at larger depths. This feature arises as the initial gain is produced by thermal electrons obtained in the PIC simulation before the hot electrons are generated. The gain in the deeper regions meanwhile is produced by the hot electron flux with the gain fall-off obtained with increasing depth a consequence of the small number of hot electrons produced in typical laser plasma interactions.

This page intentionally blank

Inner Shell Lasing In Titanium via Electron Collisional Ionization





265

The thermal electron effects can be negated by allowing the plasma electron flux to pass through a thin atomic filter – figure 2. It can be seen that, provided the filters are thick enough, the anomalous behaviour of the gain at small penetration depths disappears. However as the filter thickness is increased the gain produced is seen to decrease as a result of the increasing numbers of electrons being removed from the flux. 12

8

Gain (cm**-1)

6

9

4

6

2

3

0

0

0.2

0.4

0.8 1 0.6 Penetration Distance (microns)

1.2

0

0.2

0.4

0.8 1 0.6 Penetration Distance (microns)

1.2

Gain Duration (fs)

L3-M1 L2-M1

L3-M1 L2-M1

0

Fig. 1. Titanium gain characteristics obtained from electron fluxes produced in the interaction of a 10 fs, I
2 = 1018 W cm-2 µm2 laser with a 1 keV plasma. 12

8

9

4

6

2

3

0

0

0.2

0.4

0.8 1 0.6 Penetration Distance (microns)

1.2

0

0.2

0.4

0.8 1 0.6 Penetration Distance (microns)

1.2

Gain Duration (fs)

6

Gain (cm**-1)

X=0 X = 0.1 X = 0.3 X = 0.5

X=0 X = 0.1 X = 0.3 X = 0.5

0

Fig. 2. L3-M1 gain characteristics for electron fluxes produced in the interaction of a 10 fs, I
2 = 1018 W cm-2 µm2 laser with a 1 keV plasma filtered by aluminum filters of thickness X µm.

Allowing for different irradiance beams it is seen that all the gain spectra exhibit the same behaviour at shallow penetration depths due to the thermal plasma electrons – figure 3. For deeper regions however larger but

266





E. F. Spracklen and G. J. Pert

shorter lived gains are obtained for the more intense beams. This feature can be attributed to the increasing number of hot electrons generated by the higher intensity pulses. These electrons are capable of penetrating further into the titanium and thus produce a more significant inversion in the deeper segments. The increasing electron number however causes more rapid over-ionisation leading to the afore-mentioned decrease in gain duration. 30

X = 16 X = 17 X = 18 X = 19

6 Gain (cm**-1)

X = 16 X = 17 X = 18 X = 19

25

20

15

4

Gain Duration (fs)

8

10

2 5

0

0

0.2

0.4

0.8 1 0.6 Penetration Distance (microns)

1.2

0

0.2

0.4

0.8 1 0.6 Penetration Distance (microns)

0

1.2

Fig. 3. L3-M1 gain characteristics for electron fluxes produced by the interaction of 10 fs, I
2 = 10X W cm-2 µm2 lasers with 1 keV plasmas.

Allowing for the collisional effects of the secondary and Auger electrons generated during the interaction is seen to significantly decrease both the gain magnitude and duration – figure 4. This decrease arises as a result of the fact that the secondary electrons are overwhelmingly produced at low energies [5] and thus act to destroy any inversion. 4

2

1.5

2

1

1

0.5

-1

Gain (cm )

3

0 0.2

0.4

0.6

0.8 1 Penetration Distance (µm)

1.2

0.2

0.4

0.6

0.8 1 Penetration Distance (µm)

1.2

Gain Duration (fs)

X = 16 X = 17 X = 18 X = 19

X = 19 X = 18 X = 17 X = 16

0

Fig. 4. L3-M1 gain characteristics for electron fluxes produced by the interaction of 10 fs, I
2 = 10X W cm-2 µm2 lasers with 1 keV plasmas allowing for secondary and Auger electrons.

Inner Shell Lasing In Titanium via Electron Collisional Ionization

267

Similar effects are also seen in simulations of photo-ionised inner-shell devices. In such schemes it is possible to somewhat alleviate the effects of the secondary electrons by doping the lasing material with hydrogen [8]. Such an approach cannot however be adopted in electron collisional devices due to the similar electron impact cross-sections of hydrogen and titanium at intermediate to high energies.

4 Conclusions The gain and population dynamics of the L23 – M1 transitions in titanium have been investigated in the case of electron collisional ionization. Moderate gains of fs duration are predicted when the collisional effects of the Auger and secondary electrons are ignored. Inclusion of these effects is seen to significantly decrease both the magnitude and duration of the gains obtained.

References 1. Duguay, M. A. and Rentzepis, P. M., Appl. Phys. Lett., 10, 350 (1967) 2. Barty, C. P. J., Guo, T., Le Blanc, C., Raksi, F., Rose-Petruck, C., Squier, J., Walker, B. C., Wilson, K. R., Yakovlev, V. V. and Yamakawa, K., X-ray Lasers IOP Conference Series, 151, 282 (1996) 3. Kim, D., Toth, C. and Barty, C. P. J., Phys. Rev. A., 59, 4129 (1999) 4. Upcraft, L. M., X-ray Lasers 2002: 8th International Conference on X-Ray Lasers, 349 (2002) 5 Burgess, A. and Percival, I. C., Adv. At. Mol. Phys., 34, 109 (1968) 6. Deutsch, H., Margreiter, D. and Mark, T. D., Z. Phys. D., 29, 31 (1994) 7 Chen, M. H. and Crasemann, B., At. Data Nucl. Data Tables, 24, 13 (1979) 8. Kapteyn, H. C., Appl. Opt., 31 4931 (1992)

Gain Generation in the Critical Density Region of a TCE XRL D. Ursescu1, D. Zimmer1,2, T. Kühl1,2, B. Zielbauer1,2,3 and G.J. Pert4 1

Gesellschaft für Schwerionenforschung mbH, Darmstadt, Germany; Johannes Gutenberg Universität, Mainz, Germany; 3 Max Born Institute, Berlin, Germany; 4 University of York, United Kingdom

2

Summary. Significant reduction of the total pumping energy for a transient collisionally excited (TCE) x-ray laser (XRL) was made possible by using nonnormal main pulse pumping. In the attempt to scale to shorter XRL wavelengths, it is described here a theoretical investigation of the influence of the pumping pulse parameters on the gain generation in a Ni-like Ag x-ray laser (XRL). The possibility of gain generation closer to the critical density is shown by the EHYBRID code for a set of optimized parameters, corresponding to a possible setup at the PHELIX laser facility.

1 Introduction The advance of the TCE XRL in the last few years was marked by the introduction of a non-normal incidence focusing system (see e.g. [1]) and of the grazing incidence angle pumping (GRIP) technique for the main pumping pulse (MP) [2]. At the introduction of GRIP technique it was largely believed that the optimal angle is determined by the density where the gain appears, in the range of 1020 cm-3. A number of experiments were then investigating the effect of the MP non-normal incidence angle on the XRL output [3,4,5], with the results that slightly smaller incidence angles on target are beneficial to the output of the XRL. Also on the theoretical side, a study concerning Ag and Sm [6] has shown that the incidence angle for Sm of 45° at 1054 nm would reduce the pumping needs of the XRL a factor of three. In the present paper the same modeling program is used as in [6] and the influence of the incidence angle is investigated for Ni-like Ag XRL. It is found that the incidence angle of the main pulse has

270

D. Ursescu et al.

different effects depending on the pumping conditions. With parameters typical for the PHELIX preamplifier [7], the optimal MP incidence angle is found to be 45° (see section 2). Moreover a simple plasma shaping technique is implemented to control the density profile of the pre-plasma, by a variation of the incidence angle of the pre-pulse (PP). It is found that for large PP incidence angles on target, the XRL output improves, due to a better interplay of the electron temperature and the electron density in the gain region.

2 Influence of the MP incidence angle The parameters used for the modelling of the Ag Ni-like XRL were selected to fit a possible experiment with the PHELIX preamplifier. Two pulses are used, a long one to create the pre-plasma, and the short pulse which produces the strong excitation needed for lasing. The PP duration is 800 ps with a total energy of 4 J in a line focus of 10 mm length and 50 μm width and it is followed at it's conclusion by a MP with 2 ps duration and 2 J energy, with the same line focus length. In this evaluation the PP incidence angle on target is 0° while the MP incidence angle is varied from 0° to 75°. The EHYBRID code was used employing detailed level information of the Ni-like Ag ions, in order to obtain good description of the main lasing line, at 13.9 nm wavelength. The local XRL signal is defined as the product of the density normalized to critical density with the exponential of the gain g multiplied with the length of the target l, for a small target length (l=0.1 mm).

sl =

ne ⋅ (e gl − 1) nc

The results for the local XRL signal are shown in fig.1 for MP incidence angles from 0° to 75° on a logarithmic scale at the moment when the highest local signal is obtained, relative to the arrival of the main pulse. For the 75° incidence angle the MP energy is deposited in a quite diluted plasma region (0.5 1020 cm-3) and the local signal is less than unity. The highest local signal obtained is for 45° incidence angle, and corresponds to a density close to critical density which is 1.1 1021 cm-3 in this case. The gain appears after the MP is gone, when the heated plasma region reaches the Ni-like state. However, due to ionization and plasma expansion, the electron density at that given distance from the target increases so that the local signal is enhanced.

Gain Generation in the Critical Density Region of a TCE XRL

271

When the MP incidence angle is 0°, the energy is deposited close to the critical density so the heated mass is much larger and the heat conductivity is large, so the temperature reached at the end of the MP is about 500 eV.

Fig. 1. Local signal as a function of the position relative to the target for several main pulse incidence angles

Increasing the angle, the maximum temperature increases to a value of 2150 eV in the case of 45° and the cooling of the plasma takes longer time. This time is needed to build up the Ni-like charge state which, as a consequence, will see higher temperatures of the electrons. For larger MP angles, the maximum electron temperature further increases but the ionization process is becoming slower and the Ni-like charge state appears at a much later time when the plasma cooled again; in this case also the electron density does not increase any more in a significant way. In conclusion, the balance of the ionization dynamics and temperature dynamics determines the optimal angle for MP. In the optimal case, the local gain is enhanced by the increase of the electron density in the gain region.

3 Influence of the PP incidence angle Further optimization was performed by variation of the PP incidence angle. This is a simple plasma shaping technique which allows controlling the density profile together with the temperature and the charge state of the plasma generated by the PP. With this technique, the plasma is heated at a given density region determined by the chosen incidence angle in the same way as in the MP case.

This page intentionally blank

272

D. Ursescu et al.

The highest local signal is obtained in the case of a large incidence angle of 75° for a 30° MP incidence angle as shown in fig. 2. For comparison the maximum local signal for PP incidence angles from 0° to 75° are also represented.

Fig. 2. Local signal as a function of the position relative to the target for several prepulse incidence angles

The local XRL signal builds up in a region close to the target surface and in the simulation is a factor of 6 higher than the one obtained in the best case of the 0° PP angle (shown in fig. 1). In the cases presented in fig. 2, the electron density region has densities above the critical density of the pumping laser, being placed at 1.5 to 1.9 1021 cm-3 electron density. In conclusion, here we presented a systematic modelling study of the influence of the incidence angles for the PP and MP for a Ag XRL. It is found that strong local signal appears for MP angles around 45° in the case of normal PP incidence. When varying the PP angle, the optimal angle is found 75° for PP using an MP incidence angle of 30°. In this last case a factor of 6 in the local XRL signal is expected. The results correspond to a possible setup at the PHELIX laser facility. Furthermore they open new perspective for scaling TCE XRL towards sub-10 nm wavelengths [7].

References 1. Neumayer, P.; Alvarez, J.; Mos, B.B.; Borneis, S.; Brück, K.; Gaul, E.W.; Häfner, C.; Janulewicz, K.A.; Kuehl, T.; Marx, D.; Reinhard, I.; Tomaselli, M.; Nickles, P.V.; Sandner, W.; Seelig, W.: 'X-ray laser spectroscopy on

Gain Generation in the Critical Density Region of a TCE XRL

273

lithium-like ions', Soft X-Ray Lasers and Applications IV, SPIE, 4505, 236242, 2001 2. Keenan, R.; Dunn, J.; Patel, P.K.; Price, D.F.; Smith, R.F. & Shlyaptsev, V.N.: 'High-Repetition-Rate Grazing-Incidence Pumped X-Ray Laser Operating at 18.9 nm', Phys. Rev. Lett., 94, 103901, 2005 3. Neumayer, P.; Seelig, W.; Cassou, K.; Klisnick, A.; Ros, D.; Ursescu, D.; Kuehl, T.; Borneis, S.; Gaul, E.; Geithner, W.; Haefner, C. & Wiewior, P. 'Transient collisionally excited X-ray laser in nickel-like zirconium pumped with the PHELIX laser facility', Appl. Phys. B, 78, 957-959, 2004 4. Kazamias, S.; Cassou, K.; Ros, D.; Plé, F.; Jamelot, G.; Klisnick, A.; Lundh, O.; Lindau, F.; Persson, A.; Wahlström, C. G.; de Rossi, S.; Joyeux, D.; Zielbauer, B.; Ursescu, D.; Kühl, Th.: 'A 10 Hz, 3 µJ transient X-ray laser pumped in grazing incidence geometry', these proceedings 5. Luther, B.M.; Wang, Y.; Larotonda, M.A.; Aless, D.; Berrill, M.; Marconi, M.C.; Rocca, J.J. & Shlyaptsev, V.N.: 'Saturated high-repetition-rate 18.9-nm tabletop laser in nickellike molybdenum', Opt. Lett., 30, 165, 2005 6. Pert, G.J.: 'Optimizing the performance of nickel-like collisionally pumped xray lasers', Phys. Rev. A, 73, 033809, 2006 7. Kühl, Th.; Ursescu, D.; Bagnoud, V.; Javorkova, D.; Rosmej, O.; Zimmer, D.; Ros, D.; Cassou, K.; Kazamias, S.; Klisnick, A.; Zielbauer, B.; Janulewicz, K.; Nickles, P.; Pert, G.; Neumayer, P.; Dunn, J.: 'A Non-Normal Incidence Pumped Ni-like Zr XRL for Spectroscopy of Li-like Heavy Ions at GSI/FAIR', these proceedings

2-D Hydrodynamic Simulation of Laser Plasma Generation For Transiently Pumped Soft X-Ray Amplifier K. Cassou, S. Kazamias, A. Klisnick and D.Ros LIXAM– Université Paris XI 91400 Orsay – France

Ph. Zeitoun Laboratoire d’optique Appliquée – ENSTA, Ecole Polytechnique - 91620 Palaiseau France

E. Oliva, P. Velarde, C. Garcia and F. Ogando Institute de Fusion Nuclear (DENIM), Universidad Politecnica de Madrid Madrid Spain

Summary. A soft x-ray laser amplifier based on solid target plasma is numerically investigated using a 2-D hydrodynamic code ARWEN with radiative transport solved by multigroup method based on adaptive mesh refinement. The code has been used to describe the spatial and temporal plasma evolution and, ultimately, to understand how to generate an ideal preformed plasma in the transient collisional pumping scheme. Firstly, we examine the influence of the laser driver spatial profile on the characteristics of the preformed plasma. We show that using a super Gaussian, instead of gaussian, spatial transverse profile leads to a substantial reduction of the transverse refraction by two orders of magnitude and to an enlargement of the gain zone surface by about a factor of 2. Secondly, we perform a study on the pre-pulse significance in the transient collisional scheme, as it was done several years ago for the J=0-1 Ne-like line in the quasi-steady-state pumping. All above studies were carried out for an iron target with gain on the J = 0- 1 neon like transition at λ= 25.5nm

276

K. Cassou et al.

1 Introduction These In recent years, worldwide soft-x-ray laser research has progressed to the point where many laboratories have routinely been able to produce stimulated emission over a wide range of wavelengths. Demonstration of multi-millijoule soft-x-ray laser has been achieved in the quasisteady-state (QSS) collisional pumping scheme [1]. Significant progress has also been made in the past few years to push the soft-x-ray lasers toward higher intensity, shorter pulse duration, shorter wavelength, and particularly much higher repetition rate with much less pumping laser energy. These accomplishments were made by the use of high-repetition-rate laser system, and the implementation of the transient-collisional-excitation (TCE) x-ray laser scheme [2,3] TheTCE scheme is often poorly homogeneous exhibiting many large- and small-scale structures. Since the beam homogeneity is a key parameter for focusing applications or imaging experiments, the problem has to be overcome. These structures originate from two separate phenomena. First, speckle-like pattern is generated from the interferences of several incoherent sub-pupils [4]. This effect may be easily alleviated by seeding the amplifier with a nearly fully coherent source such as high harmonic generation [5]. This has been achieved recently [6] and opens the route to a new generation of soft-x-ray lasers [7]. The second phenomenon is the beam degradation because of its propagation along the plasma amplifier that is non-homogeneous. Control and improvement of the plasma homogeneity remains one of the key studies for the increase of soft-x-ray laser performance. More precisely, the non-uniformity of the plasma density induces a spatial fluctuation of the index of refraction that in turn refracts the soft-x-ray laser beam. Although soft-x-ray laser beam degradation was observed on other pumping schemes, it was not as crucial as now with the transient scheme. Deleterious effects of refraction associated to the transient scheme consist in beam degradation but also in the reduction of the output soft-x-ray laser energy by refracting the beam out of the gain region before reaching the end of the amplifier [8]. All the previous works related to the refraction underwent by soft-x-ray laser has shown that the refraction impacts the beam propagation in not only horizontal but also in vertical directions (Fig. 1). It has been demonstrated that the use of bent target and pre- or double-pulse pumping compensate for horizontal refraction of the soft-x-ray beams in the plasma [9, 10]. The hydrodynamic of soft-x-ray plasma amplifier is a bidimensional problem. Up to now, vertical refraction has been treated only with phenomenological modeling using 1.5D hydrocode post-

2-D Hydrodynamic Simulation of Laser Plasma Generation

277

processed by ray-tracing code but assuming ab initio plasma vertical profiles [11]. This approach prevents one from a realistic description of plasma structure along the vertical direction. As it will be pointed out in this paper, the two dimensional (2D) effects are strongly impacting the plasma shape, requiring a full 2D hydrocode for approaching a realistic description. Furthermore, the bidimensional modeling pointed out a large overestimation of the vertical size of the gain zone calculated with 1.5D hydrocode. In this conference, we report numerical modeling of the impact of the spatial shaping of the driving laser beam in order to improve the soft-x-ray laser amplifier quality by reducing the vertical refraction combined with a broadening of the amplification surface, in the transient collisional schme. We demonstrate by numerical modeling that refraction might be dramatically reduced by using (n=10) super-Gaussian laser (SGL) spatial profile instead of the classical Gaussian laser (GL) spatial profile [12]. We also observed that for (n=10) super-Gaussian, the gain region size might be increased by a factor of 2 as compared to the Gaussian spatial profile case, doubling the soft-x-ray laser (SXRL) output energy. This additional unexpected result enhances the interest of SGL spatial profile. We show that even with low prepulse energy before the long pulse, the effect reminds for optimal pre-plasma formation.

2 ARWEN code and modelling To model the laser deposition influence and the following hydrodynamic plasma evolution we used the bidimensional (2D) hydrocode ARWEN. The ARWEN [13] code is based on adaptive mesh refinement (AMR) fluid dynamic and radiation transport calculations. The radiation intensity is calculated with a discrete energy multigroup scheme [14] coupled to the adaptive algorithm. To our knowledge this is the first time that an AMR code is used to model SXRL plasmas. As compared to older 2D hydrocodes, AMR technique is much faster enabling to run a full, highly resolved hydrodynamic case in a reasonable CPU time (8h) as required to achieve a complete case. Simulations are realized with a base cell grid of 128×64 cells and two refinement levels. The radiation transport is initialized with 16 angles. The simulation window is defined by a 240 µm×120µm square side with a slab target of 10 µm. The vacuum in front of the target is set at a baking pressure of 10−3 mbar. The goal of hydrodynamic modeling of the

278

K. Cassou et al.

Fig. 1. Schematic view of modeling geometry

laser-produced plasma is to find optimum laser drive spatial conditions for the long laser pulse to obtain the largest possible gain region associated with negligible vertical density gradients (Fig. 1).For the sake of clarity we define the axis along which the soft-x-ray laser propagates as z, the vertical direction y as the axis perpendicular to the driving laser incidence, x direction. For all the simulations of interest, the target is illuminated by a λ=800 nm laser having a Gaussian pulse duration of 400ps full width at half maximum (FWHM) and 1 J energy corresponding to the typically

Fig. 2. Transverse vertical spatial laser profiles for different n-value

This page intentionally blank

2-D Hydrodynamic Simulation of Laser Plasma Generation

279

uncompressed chirped pulse of Ti:Sa laser system. 100ps after the peak of the long pulse, a second short 5ps (FWHM) duration pulse of 2 J heats preformed plasma with temporal and spatial shape identical to the long laser pulse. The spatial deposition profile of the laser pulse energy on the target, modeled by: 2n

⎛ y ⎞ ⎟ −⎜ ⎜ 2σ y ⎟ ⎝ ⎠

I ( y) = I0e

is Gaussian for n=1 or super-Gaussian for n greater than 1. Our study has been performed considering the cases of n=2, 5, and n=10. For all Gaussian or super-Gaussian lasers, the widths at half maximum are set at 100 µm (Fig. 2). The target length is set at 5 mm. We assume a homogeneous deposition along the z direction. The SGL spatial profile can be realized experimentally by using diffractive optics or by coupling a standard combination of a cylindrical and spherical lens to an adaptive mirror [15]. With this scheme the line focus is achieved thanks to the use of the lenses while the adaptive optic [16] serves to shape the transverse (vertical) profile of the line focus. The longitudinal line shape is normally sufficiently homogeneous to prevent any correction. We consider that both the long and short pulses are focused with the same optics meaning that the gain region properties of interest (ions density, electron density gradient, sizes) are given by the energy deposition of the long pulse while electronic temperature profile is given by the short laser pulse.

3 Vertical shaping of the pre-plasma It is fairly justified to expect that the preforming process (i. e. long laser pulse or combination of prepulse and long pulse) is very important for SXRL operating in transient collisional scheme and more especially for SXRL pumped in the GRIP geometry [17,18]. As not only the state of the plasma column but also the pump pulse propagation are affected by the plasma preforming process. We concentrated our interest on the plasma density shape and temperature. The ionization is not illustrated because in the LTE assumption it follows the electron temperature. On the other hand the parameters of the heating pulse decide about the final plasma temperature.

280

K. Cassou et al.

3.1 Influence of transverse spatial profile (n) Figures Fig. 3 and Fig. 4 displayed respectively 2D maps of the electron densities and temperatures for n=1 (left side) and n=10 (right side) for a delay of 100ps after the peak of the long laser pulse. During the early times of the laser-target interaction, the laser pulse shapes strongly imprint Gaussian or super-Gaussian density profiles. However, later during the plasma evolution, at delay Δt=100ps, when we choose to fire the short pulse to heat up the plasma, the density and temperature profiles do not reproduce so precisely the laser shapes (Figs. 3a,b). On both plasmas created by Gaussian and super-Gaussian lasers the edges looks perturbed. For n =10 SGL, at 100 ps after the peak of the long pulse, plasma jets appear clearly around y=40 µm and 190 µm while the central part of the plasma is homogeneous in terms of electron density and temperature. For both GL and SGL generated plasmas cases, the plasma centers are expanding with the highest velocities because most of the laser energy is laid down there, while the sides are cooler and less dense leading to lower lateral velocities.

a)

b)

Fig. 3. Electron density map,

These results in the plasma inhomogeneities observed on the edges. For plasma created by SGL profile (Fig. 3b) there is no cold dense plasma on each side formed by the feet of the Gaussian laser pulse that increase the cooling of the plasma by thermal conduction, as compared with plasmas created by GL profile pulse. This is even reinforced by the lateral expansion for the case of plasmas created by super-Gaussian beam which creates jets containing the expansion in the x-axis. Besides, the density profiles follow somehow the laser deposition profile, creating a kind of

2-D Hydrodynamic Simulation of Laser Plasma Generation

281

plateau on the electron density with sizes increasing with n. For n=1, the plateau is about 25% of the laser full width while it rises 40% for n=5 to 60 % for n=10 SGL profile. On other and, we can notice in Fig. 4a,b the electron temperature is more homogeneous in the dense part of the plasma for the SGL spatial profile. The homogeneous region, in terms of temperature follows the size of the density plateau. During the plasma expansion, the edge of the plasma in the case of SGL profile, the temperature decrease rapidly but the center part remind homogeneous. We can observe cold plasma region on each side of the main plasma in the case of SGL spatial profile, where there is no laser energy lay down. These plasmas are created by the x-ray coming from the main plasma which produced dense and cold plasma. The direct consequence of this enlargement of the hottest and densest parts of the plasmas, is that neonlike ions, i.e., lasing ions, are found in regions with boundaries at ±25 µm from the plasma centre for GL pulse, increasing to ±40 µm for n=5 SGL pulse up to ±50 µm for n=10. Note that for all GL or SGL cases, lasing ions are found from x=20→40 µm. This means that the potentially amplifying surface, and then the maximum output energy of the SXRL, is multiplied by a factor of 2 for n=10 SGL vertical profile as compared to GL.

Fig. 4. Electron temperature distribution for a) GL profile, n=1 b) SGL profile, n10 for a delay Δt=100ps after the peak of the long pulse.

As previously discussed, the electron density gradients, particularly along the y direction, may dramatically reduce the net SXRL amplification and destroy the beam homogeneity. Comparison of figures (Fig. 3a) and

282

K. Cassou et al.

(Fig. 3b) shows an impressive reduction by three orders of magnitude of the vertical density gradient in the region of peak gain localization for n=10 SGL ∇yNe~1.1×1021 cm−4 as compared to n=1 case ∇yNe~6.0×1024 cm−4. Following the calculation in the paraxial approximation propagation of soft-x-ray beam, this shows that the refraction is negligible for n=10 SGL profile [19]. Note that electron density gradients along the x direction are equivalent for any GL- or SGL-produced plasmas (n=2, 5 and 10). 3.2 Prepulse influence on the vertical plasma shape As we show, the electron density and temperature mainly follow the laser energy deposition. We investigate now the effect on the imprint of the laser energy deposition when the long pulse irradiated the plasma formed by low energy prepulse.

Fig. 5. Pre-plasma formed by combination of prepulse and long pulse for a delay of 100 ps after the peak of the long pulse: a) GL laser spatial profile b) SGL spatial profile

We add a prepulse 2.3 ns before the peak of the long pulse containing 0.1% of its energy. This leads to laser intensity on the target of 2 1010 W.cm-2 and 5 1011 W.cm-2. The prepulse spatial profile width and shape are the same as the long pulse, considering they are often focused by the same optical system. To decrease the computation time we cut the simulation window in half in the y direction using the x-axial symmetry. We Show in Fig. 5 that the density distribution (logarithm scale) is still more homogeneous in the case of SGL spatial profile in the vicinity of the critical surface. If we compare with the Fig.3 where there is no pre-pulse,

2-D Hydrodynamic Simulation of Laser Plasma Generation

283

the electron density gradient fall by 2 orders of magnitude on the x-axis [20, 21]. In the case of SGL profile, the density is higher for a same x coordinate compared to GL profile case. This could be explained by the difference in the velocity distribution and particularly, the difference of the ratio of (vx/vy) which is closer to one in the case of SGL profile. Finally, we can notice, the beginning of a jet formation in y=60µm generated by the long pulse and the second one expanding to the top of the figure from y=75µm to y=120µm. These structures are not generate in the case of a perfect GL spatial profile.

4 Plasma heating and gain estimation Thus at high electron density Ne~1020–1021 cm−3, where the second pulse is absorbed, the GL generated plasma has a temperature of about Te~470 eV at its center y=120 µm dropping to Te~ 200 eV at y=75 µm and y=175 µm. For SGL profile pulse, the peak temperature is nearly the same, however on the edges y=75 µm and y=175 µm a much higher temperature of Te ~320 eV is achieved. Besides, the density profiles follow somehow the laser deposition profile, creating a kind of plateau on the electron density with sizes increasing with n. For n=1, the plateau is about 25% of the laser full width while it rises 40% for n=5 to 60 % for n =10 SGL profile. This means that the potentially amplifying surface, and then the maximum output energy of the SXRL, is multiplied by a factor of 2 for n=10 SGL vertical profile as compared to GL.

Fig. 6. Gain map a) Super Gaussian profile, n=10 b) Gaussian profile, n=1.

284

K. Cassou et al.

To achieve 2D gain maps, we postprocessed the outputs of ARWEN with a simplified three-level atomic model. Such a calculation is known to slightly overestimate the gain [20]. We assumed that the fundamental level 1s22s22p6 is much more populated than the other levels. The upper lasing level (1s22s22p53p,J=0) is linked to the fundamental level by a forbidden transition and following the conclusions of Goldstein et al. [22], we assume that is mainly populated by direct collisional excitations. The lower lasing level (1s22s22p53s,J=1) is populated and depopulated through both collisional and radiative transitions. Collisional and radiative rates have been obtained from previous works on neon-like iron [23]. Spectral width is calculated assuming both homogeneous and inhomogeneous broadening based on ARWEN output. The electron density is given by ARWEN at the instant of firing the second pulse. From Figs.6a and 6b we can observe in the 2D gain maps that the shapes of the gain regions follow the energy deposition profiles. Close maximum gain, around 136 cm−1, is found for all profiles. Yet, it is interesting to note that the peak gain zone is two times larger over the y direction for the n=10 SGL 80 µm as compared to GL case 40 µm.

5 Conclusion This study underlines the need of complete bidimensional hydrodynamic code to achieve realistic soft-x-ray plasma amplifier description. Indeed, we show that lateral thermal conduction reduces the gain zone to a third of laser width when Gaussian laser shape is considered. By using a 2D hydrocode, we have numerically demonstrated that n=10 SGL spatial profile reduces the vertical refraction to an insignificant level. Consequently, an amplified soft-x-ray beam of high homogeneity would be achievable. Another beneficial phenomenon observed is an enlargement of the gain region surface by a factor of 2 while using super- Gaussian spatial profile as compared to classical Gaussian laser profile. Broadening of effective amplification surface would lead to doubling of the output energy while keeping driving laser energy constant. These two points allow a better output soft-x-ray laser energy extraction ensuring to double the output energy without increasing the driving laser energy.

2-D Hydrodynamic Simulation of Laser Plasma Generation

References 1. B. Rus et al., Phys. Rev. A 66, 063806 (2002). 2. P. V. Nickles et al., Phys. Rev. Lett. 78, 2748 (1997). 3. J. Dunn et al., Phys. Rev. Lett. 84, 4834 (2000). 4. O. Guilbaud et al., Europhys. Lett. 74, 823, (2006). 5. R. Bartels, Nature 406, 164 (2000). 6. Ph. Zeitoun et al., Nature 431, 427 (2004). 7. Y. Wang et al. Phys. Rev. Lett. 97, 123901 (2006) 8. S. Le Pape and Ph. Zeitoun, Opt. Commun. 219, 323 (2003). 9. J. G. Lunney, Appl. Phys. Lett. 48, 891 (1986). 10. R. Kodama et al., Phys. Rev. Lett. 73, 3215 (1994). 11. J. A. Plowes, Opt. Commun. 116, 260 (1995). 12. K. Cassou et al. Phys. Rev. A, 76, 1 (2006) 13. F. Ogando and P. Velarde, J. Quant. Spectrosc. Radiat. Transf. 71, 541 (2001). 14. L. H. Howell et al., Numer. Heat Transfer, Part B 35, 47 (1999). 15. G.-Y. Yoon et al., Appl. Opt. 36, 847 (1997). 16. T. A. Planchon et al., Opt. Commun. 216, 25 (2003). 17. K. Cassou et al. Optics Letters, to be published. 18. S. Kazamias et al, in this proceedings 19. E. Fill, J. Opt. Soc. Am. B 14, 1505 (1997). 20. J. Nilsen et al., Phys. Rev. A, 48, 4682 (1993) 21. Y. Li, G. P. Preztler et al., Phys. Plasmas 4, 164 (1997). 22. W. H. Goldstein et al., Phys. Rev. A 36, 3607 (1987). 23. M. Cornille et al., At. Data Nucl. Data Tables 58, 1 (1994).

285

2D Hydrodynamic Simulation of Focus-Line Plasma in Ni-Like Ag X-Ray Laser Research W. Zheng and G. Zhang (Institute of Applied Physics and Computational Mathematics, P. O. Box 8009, Beijing 100088,China)

Summary. In the research of X-ray laser (XRL) driven by line focused laser, because focus-line had very large aspect ratio with narrow width (100μm), the jet of plasma showed clear 2D structure, and all sorts of peculiar 2D distributions of near-field and far-field XRL intensity were already observed in experiments, such as “bow-like” field image of XRL on ShengGuang-II facility. In order to understand and simulate this phenomenon, we developed a 2D non-equilibrium radiation hydrodynamics code named XRL2D, which was introduced in this paper briefly. The hydrodynamics behavior of Ni-like Ag XRL had been simulated by XRL2D code, simulation results were also presented in this paper. Similar to the QSS XRL experiments on ShengGuang-II, a slab Ag target was irradiated by a 1ω laser beam (85J) with 100μm×1.8cm focus-line, drive laser included a pre-pulse (2.55J) and a main pulse(82.45J), both with width of 100ps, and the interval of two pulses was 3ns. The abundance distribution of Ni-like ion appeared bow-like structure. The 2D gain calculation was our future work.

1. Introduction In the study of X-ray laser (XRL) driven by laser, line-focus beam were used popularly. Because focus line had a long length and a narrow width (100μm), the plasma expanded two dimensionally, in addition to the nonuniform of drive beam, hydrodynamic behavior of plasma and gain distribution of XRL can’t been described by 1D code. For this 2D characteristic of plasma expansion, many peculiar distributions of near and far field XRL intensity have been observed in many experiments [1-4]. In the experiments conducted on ShengGuang-II facility, the near- and farfield intensity distribution of Quasi-Steady-State (QSS) Ni-like Ag 13.9nm XRL presented a bow-like shape, which were showed in fig.1. This XRL were produced by irradiating Ag slab target (1.6cm long) with 100μm×1.8cm line focused beam (1ω/100ps/100J) at an intensity of 4~6×1013W/cm2, drive laser beam included two pulses, energy rate of pre-

288

W. Zheng and G. Zhang

pulse and main-pulse was 3~5%, main-pulse delayed by ~3ns. There were two reasons to explain this bow-like shape. First, the gradient of electron density of plasma made XRL refracted out of the center area. Second, the gain of XRL maybe have this bow-like profile. In order to understand physical processes in it exactly and to study 2D effects of interaction between laser and plasma, we developed a non-equilibrium radiation hydrodynamics 2D code (XRL2D). The brief introduction of this code was presented in this paper, as well as the hydrodynamic simulation result of QSS Ni-like Ag XRL.

Fig. 1. The intensity distribution of Ni-like Ag 13.9nm XRL (experiments on ShengGuang-II facility). (a) Near-field image of two opposite coupling targets, and (b) far-field image of single target. x was the normal direction of target surface, y was the tangent direction of target surface. θ was the refraction angle

2. Introduction of XRL2D code Inverse bremsstrahlung absorption and resonant absorption of drive laser were considered, and geometric optic approximation was used. Flux-limit heat-conduct approximation and flux-limit multi-group diffusion approximation (~100 groups) were adopted. Average atomic model only considering principal quantum number was used in code. Considered micro-processes included bremsstrahlung and its inverse process, electron collisional ionization and three-body recombination, photo ionization and photo recombination, electron collisional excitation and de-excitation, line emission and line absorption, di-electron recombination and self ionization. In the view of Lagrange, equations of irradiation hydrodynamics employed by code were listed below, v du 1 ≈ − ∇( pe + pi + pr + q ) dt ρ

(1)

2-D Hydrodynamic Simulation of Focus-Line Plasma

289

v d (CveTe + V ) 1 d ⎛1⎞ ≈ − ∇ ⋅ Fe − pe ⎜⎜ ⎟⎟ + Wie + Wl − Wr ρ dt dt ⎝ ρ ⎠

(2)

v d (CviTi ) 1 d ⎛1⎞ ≈ − ∇ ⋅ Fi − pi ⎜⎜ ⎟⎟ − Wie ρ dt dt ⎝ ρ ⎠

(3)

v ρν ∂fν ∂ ⎛ 1 ⎞ ∂fν = −∇ ⋅ Fr + ⎜ ⎟ + Dν − Cν fν ∂t 3 ∂ν ∂t ⎜⎝ ρ ⎟⎠

(4)

v

In which, u was the hydrodynamic velocity, ρ was the matter density, pe, pi and pr were the pressures of electron respectively, ion and irradiation, q was the Von-Neumann viscosity, CveTe and CviTi were the electron and ion internal energy per unit mass, fv was the photo number of unit quantum state, v was the frequency of photo. V was combine energy of ion, Wie was the term of energy exchange between electron and ion, Wl was the term of laser energy deposit, Wr was the term of energy exchange between matter and irradiation. Dv, Cv were the emission and absorption coefficients of photo. Fe, Fi and Fr were flux of electron, ion and irradiation respectively. r Fe = FDe ⋅ f e FLe

and

F

( Fr

De

were

Le v FLe = N e kTe kTe / me

+ f e FLe

), where, f

diffusion

e

flux

v

was electron flux-limitv factor, FDe

and

direct

flux,

FDe = − K e ∇Te

,

, so as well as flux of ion and irradiation. ALE method adopted by code included three steps. First was Lagrange step. Differential scheme of momentum equation was IGA scheme. In order to decrease calculation amount, split scheme was used, diffusion processes of electron, ion and irradiation were separated from local processes such as atomic processes, energy exchange between electron and ion, temperature increase induced by energy deposit of laser, and local photo-absorption and irradiation et al. using implicit scheme, local processes were all coupled. Second step was rezone, mesh were rezoned every time-step. Last step was remap, 2-order precision remap scheme were adopted.

3. Hydrodynamic simulation of QSS XRL The calculation model was according to the experiments of Ni-like Ag XRL on ShengGuang-II described above. Drive laser (1ω) irradiated Ag slab target with 100 μm×1.8 cm line focused at a intensity of 4.7×1013 W/cm2, laser beam included pre-pulse (2.55 J) and a main-pulse (82.45 J)

290

W. Zheng and G. Zhang

delayed 3ns. Both were Gaussian pulse in time with a width of 100ps, and flat-top pulse in space. Electronic flux-limit factor (fe) of 0.03 was applied. The incidence direction of drive beam was reverse to x axis,

Fig. 2. Electron temperature and density distribution before the main-pulse. Meshes were also plotted.

Fig. 3. Electron temperature and density distribution 20ps after main-pulse peak time.

initial target surface was located at x=0, the target thickness was 30μm. The tota l mesh number was 150×100. At the beginning, area out of target surface was filled with rareness gas (Ag) with a density of 10-6g/cm3. In calculation, meshes were kept perpendicularly always by rezoning, see in fig.2-3. For the symmetric of problem, only half physical area was calculated. Fig.2 showed the simulation results before main-pulse. Pre-pulse only prepared a pre-plasma. After expanding for 3ns, Pre-plasma cooled down

2-D Hydrodynamic Simulation of Focus-Line Plasma

291

no more than 20 eV, and while approaching target surface, electron temperature decreased more and more. Simulation results 20 ps after main pulse peak time showed in fig.3, when gain of XRL would appear (the result from 1D simulation of JB19 code). According to our simulation, in the

Fig. 4. At center of focus line, electron temperature Te, radiation temperature Tr, electron density Ne, abundance of Ni-like ion PNi_like and laser energy deposit WL distribution along normal direction of target surface, 20ps after main-pulse peak time.

period of incidence of main pulse, plasma moved a little because the width of main pulse was short enough. With the arrival of main-pulse, plasma at laser energy deposit area was heated and ionized dramatically. At this area, for rapid electron collisional ionization, Ni-like ion dominated at lower electron temperature (1 because the preplasma is not transparent for the pedestal pulse. The solutions are

Similarity Model of Longitudinally Pumped X-Ray Laser

329

4/3 8/9

T = a1 t 3/ 4

3/ 4

(1 − t m t 17 / 9

−17 / 9

Tm t m

+

t

L = b1 t

3/2

)

3/ 4

, here a

1

a1

(1 − t m t 17 / 9

−17 / 9

Lm t

+

−1

= 28.45 I m 1 L A

2/3

−17 / 9

2

1/ 2

−17 / 9

10 / 9

)

1/ 2

, here b

1

b1 t m

= 17.81 × 10

−1

−4

I m1 L

(3) 1/ 4

c1 t

n0 = (1 − t m t 17 / 9

−17 / 9

−5 / 4

, here c

1

c1

+

28 / 9

4

t

−17 / 9

)

20

= (5.69 × 10 )

4

−1

I

−2

5

m1 L A

1/ 4

nm t m

b) During the time of t2trans ≤ t ≤ t2L

The plasma changes to be transparent for the pedestal pulse again with the laser heating. Considering the physics meaning of the t2trans we get

t 2 trans = 0.51 I

−7 /15

23 / 15

m1 L A

−2 / 3

Λ

8 /15

λ

16 /15

(4)

Also 5/ 2

T = a2 (1 − t 2 trans t 2/5

5/3

−5 / 3

L = b2 t [1 − t 2 trans t 4 / 15

5/3

+

−5 / 3

5/3

T2 trans t 2 trans t +

−5 / 3

a2 15 / 4 −5 / 3 L2 trans t 25 / 12

2/5

]

b2 t 2 trans

4 / 15

t[1 − t 2 trans t 5/3

−5 / 3

+

(8.08 × 10 )

20 15 / 2

c2

c2 t 15 / 2

, here c2 =

−5 / 3 35 / 6

]

2 / 15

2

, here b2 = (1.29 × 10 −2 )15 / 4 a2 A

2 / 15

n0 =

−1

) , here a2 = 19.46 I m1 L λ Λ

a2 A

−5 / 4

15 / 2

m1 L

5/2

n2 trans t 2 trans (5)

c) During the time of t2L≤ t ≤ t3L

The main pulse is so short that the plasma is not transparent for the laser and there is not enough time to change the state of ion charge. At time t2L

330

Y. J. Li et al.

the solutions can be obtained by considering the initial conditions before t2L: 2/3

5 / 3 −5 / 3

T = a 3 t (1 − t 2 L t

+

T2 L t 2 L

t

−5 / 3

−1

), here a 3 = 23.12 keV I 2 m AZ

−1

a3 2

1/ 2 3 / 2

L = b3 t

n0 = c3 t 1/ 2

5 / 3 −5 / 3

(1 − t 2 L t

−3 / 2

L2 L

+

t

4/3

−5 / 3 1 / 2

)

b3 t 2 L −5 / 3 −5 / 3

(1 − t 2 L t

2

−1

3

c3

+

, here b3 = (5.16 × 10 −2 ) 2 I 2 m

4/3

t

−5 / 3

n2 L t 2 L

)

−1 / 2

, here c3 = (3.16 × 10 ) 20

2

m Z 1

2 2

I2 A (6)

d) During the time of t3L≤ t

After the time t3L the main pulse is turned off and the plasma continues to expand adiabatically. The solutions can be obtained using the condition before t3L: 2/3

T = T3 L t 3 L t

−2 / 3

−5 / 9

L = L3 L t3 L t

5/9

7/9

n0 = n3 L t 3 L t

−7 / 9

(7)

here T3L L3L and n3L are the temperature scale length and density at the t3L respectively.

3 Results and Discussion We firstly calculate the experiment results of T.Ozaki et al. under the same condition to demonstrate the credibility of the model. Then we investigate the hydrodynamic of longitudinally pumped x-ray laser plasma using the model. 3.1 Credibility of the model

We calculate the longitudinally pumped nickel-like molybdenum slab xray lasers under the same experimental conditions of T.Ozaki’s. The calculation results indicate that the average ionization state only reach to 10 not 14 required by nickel-like molybdenum at the pre-pulse laser intensity of 1.5×1011 W/cm2 and the duration of 300ps which is a bit bigger than 9 from T.Ozaki et al. But the pedestal of main pulse with the intensity

Similarity Model of Longitudinally Pumped X-Ray Laser

331

3.0×1011 W/cm2 and the duration 300ps heat the electron to 88.1 eV which is quite similar to Ozaki’s results. The ionization is 14.04 and the electron density reach to 1.01×1020 cm-3 according with Ozaki’s results. The comparison shows successfully that the results from this simplified model well match the calculation by Ozaki. 3.2 Affection of the pedestal of main pulse on the ionization

In order to realize the affection of the pedestal to x-ray laser we calculate the affection on the ionization of pulse pedestal with different duration and intensity respectively as shown in Fig 1. 14.15

Z (Average ionization)

(a) 14.10 14.05 14.00 13.95 13.90

0.1

0.2

0.3

0.4

0.5

0.6

7

8

t2L (ns)

(b)

Z (Average ionization)

16.0 15.5 15.0 14.5 14.0 3

4

5

6 11

2

Intensity (10 W/cm )

Fig. 1. The average ionization vs the duration (a) and the intensity (b) of the pedestal of the main pulse.

332

Y. J. Li et al.

Fig.1 shows that the ionization just decreases from 14.12 to 13.93 (stay Ni-like) with the increase of the duration of main pulse pedestal from 100 ps to 600 ps under the condition that the intensity is invariable (3.0×1011 W/cm2). While the ionization increases rapidly from 14.03 to 15.96 (over Ni-like state) with the increase of intensity of pulse pedestal from 3.0×1011 W/cm2 to 8.0×1011 W/cm2 when the duration of main pulse pedestal (300 ps) is invariable. It is clear that the variety of the ionization is sensitive to the intensity of pulse pedestal. 3.3 The hydrodynamic of the longitudinally pumped x-ray laser

We know that people always pay more attention to the variety of temperature scale and density of plasma with different value of z along propagation of pumped laser (z axis) in longitudinally pumped scheme due to its direct affection if the plasma can form the uniform population inversion and if x-ray laser can be amplified effectively instead of being absorbed. In order to investigate the above problems particularly we calculate the electron temperature the ionization state and the variety of density respectively with the different value of z under the same experimental conditions of T.Ozaki’s as shown in Fig.2. We firstly calculate different value of z dependent electron temperature before the ending of the main pulse pedestal as shown in Fig.2 (a). It should be noted that the temperature rapidly decreases from 90 eV to 5.9 eV before 600 μm and then approaches to 5.6 eV after 600 μm which is the same to the electron

temperature without main pulse pedestal indicating that the pumped laser intensity has approached to 0 after 600 μm. The different value of z dependent electron temperature in Fig.2 (b) has also demonstrated the above conclusion: the ionization decreases from 14.03 to 5.69 with the variety of z from 0 to 600 μm and then approaches to 5.6. At last we calculate the variety of the electron density with the different value of z (shown in Fig.2 (c)). The electron density varies from 1.01×1020 cm-3 to 1.84×1020 cm-3 and then approaches to 1.95×1020 cm-3 after 600 μm. Therefore we can conclude that the intensity of pumping laser decreases rapidly along propagation in the plasma due to the absorption and refraction. When the value of z exceeds a definite value the proportion of Ni-like ion is much little because of the low temperature. Here the x-ray laser will be absorbed instead of being enhanced by plasma. So it is important to choose the appropriate slab length. We also calculate the different value of z dependent electron temperature at the end of pulse pedestal with different electron density as shown in Fig.2(d) due to the condition that we can vary different incident

Similarity Model of Longitudinally Pumped X-Ray Laser

333

100

14

Z (Average ionization)

(a)

80

T (eV)

60 40 20 0 0.00

0.05

0.10

0.15

(b)

12 10 8 6 4

0.20

0.00

0.05

0.10

0.15

0.20

z (cm)

z(cm)

100

2.0

(c) 80

(d)

1.4

T (eV)

-3

1.6

20

n0 (10 cm )

1.8 60 40 20

1.2

0

1.0 0.00

0.05

0.10

z (cm)

0.15

0.20

0

500

1000

1500

2000

-3

z (10 mm)

Fig. 2. The curve of the temperature (a) ionization (b) and electron density (c) with z axis along the longitudinal direction before the main pulse arrive. The delay time is 4ns other conditions are same as fig.1. (d) The curve of the temperature with different density along z axis at the end time of the main pulse. Here the solid line is for n0=1.0×1019 cm-3 the dash line is for n0=4.0×1019 cm-3 the dot line is for n0=8.0×1019 cm-3 the dash-dot line is for n0=1.2×1020 cm-3 the dot-dot line is for n0=4.0×1020 cm-3 the short dash line is for n0=8.0×1020 cm-3.

density through choosing different longitudinally incident position in the experiment. It is noted that the electron temperature changes obviously with z in different electron density. It also shows that the electron temperature varies only from 88.11 eV to 83.62 eV when the density is lower 1.0×1019 cm-3 for instance while the electron temperature decreases rapidly to the initial value of 5.37 eV in the 8.0×1020 cm-3. Therefore it is more effective method to choose the proper element of which the requirement of the density in the gain region is lower.

This page intentionally blank

334

Y. J. Li et al.

4 Conclusions A similarity model is developed for the longitudinally pumping x-ray lasers. The results are in agreement with the experimental results. The hydrodynamics of the scheme are investigated using the model. The results show that (1) The pedestal of the longitudinal pumping pulse play an important role in the Ni-like ionization state (2) The absorption of longitudinally pumping laser in the plasma result in the asymmetry of the temperature and the density in the laser propagation direction so that the gain length is limited. (3) It is an effective method to choose the proper element of which the required density is low in the gain region.

References 1. P. V. Nickles et al. Phys. Rev. Lett 78 2748 1997. 2. R. Li et al. Phys. Rev. E 57 7093 1998. 3. T. Ozaki et al. Phys. Rev. Lett 89 253902 2002. 4. Y. J. Li et al. 9th International conference of x-ray lasers page 333 Beijing May 24-28 2004. 5. Y. J. Li et.al. Phys. Rev. E 63 036410 2001.

Numerical Analysis of Plasma Medium of a Fully Coherent X-Ray Laser N. Ohnishi1, M. Nishikino2 and A. Sasaki2 1

Department of Aerospace Engineering Tohoku University 6-6-01 Aramaki-Aza-Aoba Aoba-ku Sendai 980-8579 Japan 2 Kansai Photon Science Institute Japan Atomic Energy Agency 8-1 Umemidai Kizu Souraku Kyoto 619-0215 Japan

Summary. Two-dimensional (2D) radiation hydrodynamics simulations have been performed to investigate the generation and the refraction influence in the plasma medium of a fully coherent x-ray laser at 13.9 nm by the double-target configuration. The local energy deposition of the main laser pulse generates a blast wave near the critical density surface and the density dip structure is gradually formed behind the blast wave. The size of the density dip structure is about 10 μm after 50 ps of the main pulse. The three-dimension (3D) ray-trace calculation using the result of the 2D simulation shows the x-rays pass through the density dip with less refraction. The size and the position of the density dip area are similar to the light source of the fully coherent x-ray laser.

1 Introduction The development of the brilliant x-ray source has been progressed for the material science and the biological science and the soft x-ray free-electron lasers (XFELs) [1 2] will open up possibilities of the new scientific area. The laser-driven x-ray lasers (XRLs) [3–9] with highly spatial and temporal coherence are applicable light sources as a scientific tool for the purpose of the practical application of the XFEL. Since the first demonstration of the laser-driven soft XRL [10] the improvements of the beam properties have been demonstrated by a number of the groups [11– 15]. The significantly improvement of the beam quality such as spatial and temporal coherence is the important issue for the practical applications. Recently soft-XRLs with small divergence and highly spatial coherence have been demonstrated by the double-target amplification [16–18] and the longitudinal pumping [19]. In contrast to the transient-collisionalexcitation (TCE) XRL [20 21] the highly spatial coherent XRL is

336

N. Ohnishi, M. Nishikino and A. Sasaki

developed with discharged capillary plasma [22] and the amplification of the high-order harmonic generation [23]. A fully coherent XRL has firstly demonstrated at a wavelength of 13.9 nm by the double-target amplification [16–18]. In the double-target amplification the XRL from the first target is used as a seeding light and a coherent portion of the seeding XRL is amplified in the gain medium of the second target. Since the spatial coherence of the TCE XRL is limited by the short length of the gain medium the amplifier of the second target was separated from the first target by a sufficient distance with the condition of the Fresnel number N < 1. When the delay time between two pumping pulse is optimum condition the highly directional XRL is generated. The total gain of the amplification of the second target is about 1000 with the gain coefficient of about 8 cm−1 and the output energy is achieved about 1 μJ [24 25]. In this configuration the gain medium of the second target works as an active spatial filter. From the first experimental result and the simple analysis the spatial position of the gain medium of the second target is generated in a lower density area for avoiding the influence of the x-ray refraction. Then the position of the gain medium is about 100 μm from the target surface. However the gain region is about 20 μm distance from the target surface and the source size of the gain medium is about 20 μm in the 2D imaging experiment [26]. It turns out that there is a difference between the simple estimation and the experimental result. Therefore we have performed the 2D radiation hydrodynamics simulation to investigate the generation and the refraction influence of the gain medium plasma. In this paper we present the numerical analysis of the gain medium plasma of the fully coherent XRL using a 2D radiation hydrodynamics code and the 3D ray-tracing of x-ray laser has been conducted with arranging flow-fields of the 2D simulation result. We have carried out the comparison of the gain medium between the experimental result and the simulation one. The effects of hydrodynamic properties of the gain medium plasma have been investigated for finding the optimal pumping laser condition.

2 Numerical Methods and Conditions The computational code for plasma medium of x-ray laser should be able to properly reproduce high-temperature dense plasma. In the preset paper we have developed a 2D radiation hydrodynamics code based on the RAICHO code [27] which was originally developed for inertial fusion

Numerical Analysis of Plasma Medium of a Fully Coherent X-Ray Laser

337

plasmas with some modifications for including the effects of thermal nonequilibrium realistic equation of state and laser refraction. The governing equations consist of inviscid compressible hydrodynamic equations with the additional terms due to electron/ion thermal conduction x-ray radiation and laser absorption in energy equation. Also radiative transfer equation is simultaneously solved to estimate the radiation source term. Numerical fluxes due to inviscid compressible flow are estimated by AUSM-DV approximate Riemann solver [28] with second-order spatial accuracy. The radiative transfer equation is approximated by the multigroup diffusion with variable Eddington factor. Non-LTE (CRE) emissivity and opacity is tabulated before the simulation based on the averaged ion model [29]. The electron and ion thermal conductions are modeled with the assumption of the flux-limited diffusion with classical conductivities. The flux-limiter was set to be 0.1 in the present simulation. In our code two-temperatures for electrons and ions are adopted with the relaxation term because they must be in thermally non-equilibrium in a corona. The laser absorption of inverse-bremsstrahlung is calculated by the ray-tracing manner including the refraction due to the electron density gradient [30].

Fig. 1. Simulation conditions for plasma medium of the second target.

The computations are conducted to investigate the density effect of the plasma medium of the second target in the oscillator-amplifier configuration. Figure 1-(a) shows the simulation area of the second target. The simulation domain with 50 _m long and 150 _m wide is calculated on the center of the target with 301 x 51 computational grids. This number of grids is sufficient for capturing the plasma features. The target material is silver. The laser wavelength is 1.06 _m. Input laser pulses consist of the Gaussian shaped pulses shown in Fig. 1-(b) which are pico-sec prepulse 100-ps prepulse and pico-sec main pulse. The pulse widths are 12 ps 100 ps and 12 ps at FWHM respectively. The pulse energies are 1J, 3 J and 12 J respectively. Since the pedestal of the laser pulse affects the plasma condition [7] we performed the calculation including the pedestal with 1012

338

N. Ohnishi, M. Nishikino and A. Sasaki

W/cm2. The timing of t = 0 ps is the peak of the main pulse. The silver target is irradiated from right hand side of Fig. 1-(b) by the line focused laser whose spatial width is 20 μm (10 μm in the computational domain) at the focal position. In the present study the spatial distribution of the laser intensity is uniform in that width. The laser is divided into 100 computational rays and the rays are parallel to each other without divergence.

3 Results and Discussion Electron temperature is rapidly increased by the main pulse near the critical densisy and exceeds 2 keV in the preformed plasma. Then pressure becomes very high while density gradient is relatively gradual. This leads the formation of a blast wave. The blast wave formation depends on the geometrical condition. Figure 2 shows the typical profiles of electron number density and electron temperature with different dimensions at 50 ps after the main pulse. In 1D (planar) geometry we cannot find the formation of the blast wave. However the propagation of the shock wave towards the corona region and the low density dip (circled region in Fig. 2(b)) can be found with 2D simulation. This density dip may affect on the refraction of the incident x-ray laser from the first target.

Fig. 2. Profiles of electron number density and electron temperature by (a) 1D planar simulation and (b) 2D simulation (in axis of symmetry) at 50 ps after the main pulse.

In order to confirm this density dip effect we have performed 3D raytracing of injected x-ray laser in the variable flowfield along the incident direction. The flowfield felt by the injected rays was constructed with arranging the time series of 2D simulation. Figure 3-(c) shows the contours of electron number density at 50 ps after the main pulse. The rays in the

Numerical Analysis of Plasma Medium of a Fully Coherent X-Ray Laser

339

density dip around X = 20–30 μm and Y = 0–10 μm meander though the dip while the other rays are refracted to the expansion direction. Figure 3-(b) shows the ray positions at the cross section of the second target (6 mm from the incidence). In the present study a gain length for each ray is defined by the light path integrated over the assumed gain region where 1020 < ne < 1021 cm−3 and 0.25 < Te < 1.0 keV. The length for the rays in the density dip can be longer than that for the other rays because that dip is in the condition of the gain region. In this timing the length is about 5 mm. The size of the ray passed through the density dip is about 10 and 20 μm for the horizontal and vertical direction respectively. The non-refracted rays are located at X = 20–30 μm. Figure 3-(a) shows the near-field pattern (NFP) of the second target of the double-target amplified XRL in the case of the gain length of about 6 mm. The NFP was measured using a near-field imaging system with magnification of 8.5[26]. The size and the position of transmitted rays where the non-refracted rays are found are similar to those of intense area of the NFP.

Fig. 3. (a) The NFP of the 6-mm length second target of the double-target amplified XRL. (b) The ray positions in the cross section of the second target at 6 mm from the incidence. (c) The contours of electron number density at 50 ps after the main pulse.

4 Conclusion We have performed 2D radiation hydrodynamics simulations and 3D raytracing for investigating the plasma medium of the fully coherent x-ray

340

N. Ohnishi, M. Nishikino and A. Sasaki

laser. The results suggest that the density dip behind the blast wave near the critical density suppresses the refraction of the injected x-ray laser from the first taget. We are planning the coupling computation with unsteady atomic process for detailed analysis of the fully coherent x-ray laser.

References [1] J. Arthur Rev. Sci. Instrum. 73 1393 (2002). [2] V. Ayvazyan et al. Phys. Rev. Lett. 88 104802 (2002). [3] M. Kalachnikov et al. Phys. Rev. A 57 4778 (1998). [4] A. Klisnick et al. J. Opt. Soc. Am. B 17 1093 (2000). [5] J. Dunn et al. Phys. Rev. Lett. 84 4834 (2000). [6] R. Tommasini J. Nilsen and E. Fill Proc. SPIE 4505 85 (2001). [7] T. Kawachi et al. Phys. Rev. A 66 033815 (2002). [8] S. Sebban et al. Phys. Rev. Lett. 89 253901 (2002). [9] B. M. Luther et al. Opt. Lett. 30 165 (2005). [10]D. L. Matthews et al. Phys. Rev. Lett. 54 110 (1985). [11]C. L. S. Lewis et al. Opt. Commun. 91 71 (1992). [12]R. E. Burge et al. J. Opt. Soc. Am. B 14 2742 (1997). [13]H. Daido et al. J. Opt. Soc. Am. B 16 2295 (1999). [14]G. Cairns et al. Appl. Phys. B 58 51 (1994). [15]R. Kodama et al. Phys. Rev. Lett. 73 3215 (1994). [16]M. Tanaka et al. Opt. Lett. 28 1680 (2003). [17]M. Nishikino et al. Phys. Rev. A 68 061802(R) (2003). [18]M. Nishikino et al. IEEE Journal of Selected topics in quantum electronics 10 1382 (2004). [19]T. Ozaki et al. Phys. Rev. Lett. 89 253902 (2002). [20]P. V. Nickles et al. Phys. Rev. Lett. 78 2748 (1997). [21]J. Dunn et al. Phys. Rev. Lett. 80 2825 (1998). [22]Y. Liu et al. Phys. Rev. A 63 033802 (2001). [23]Ph. Zeitoun et al. Nature 431 426 (2004). [24]T. Kawachi et al. Proc. SPIE 5919 1 (2005). [25]M. Nishikino et al. AIP Conf. Proc. 827 499 (2006). [26]M. Tanaka et al. Surface Rev. Lett. 9 641 (2002). [27]N. Ohnishi et al. J. Quant. Spectrosc. Radiat. Transf. 71 551 (2001). [28]Y. Wada and M. S. Liou AIAA Paper 94-0083 (1994). [30]H. Takabe and T. Nishikawa J. Quant. Spectrosc. Radiat. Transf. 51 379 (1994). [31] A. L. Edwards and J. A. Fleck Jr. J. Appl. Phys. 50 4307 (1979).

Longitudinally Pumped Ne-Like Titanium X-Ray Laser Simulation with a Post-Processor Code Coupled to EHYBRID A. Demir, E. Akman, S. Bilikmen*, P. Demir*, S. İnce*, E. Kacar, E. Yurdanur* and S. Yaltkaya** University of Kocaeli Laser Technologies Research and Application Center Aslanbey Campus Kocaeli / Turkiye * Middle East Technical University Department of Physics Ankara / Turkiye ** Akdeniz University Department of Physics Antalya / Turkiye

Summary. Longitudinal pumping with a grazing incidence scheme of Ne-like Ti has been modeled by using the EHYBRID and a post-processor code. The atomic data that are required in the simulation are obtained by using the Cowan code. The variation of the Ne-like Ti x-ray laser gain at 32.6 nm is calculated for a fixed delay time with a variation in the incidence angle and a fixed incidence angle with a variation in the delay time. The post processor code has been used to simulate the x-ray resonance lines between 17 and 29 Å.

1 Introduction The first x-ray laser was put into practice in the LLNL Novette laser facility in 1984 [1]. In this first demonstration kilo joule scale optical pump energy with a transverse scheme was used to ionize Ne-like Se thin foils and the output wavelength was seen at 20 nm. Since then numerous amount of experiments have been held throughout by using Li-like Ne-like Ni-like targets [2-5] with different pumping schemes [6-9]. Besides these the progress in the area has also been due to the technological improvements achieved in the laser technology. The general aim in these researches has always been to increase the gain efficiency and the repetition rate of the x-ray lasers with a shorter wavelength output and make them compact for their further usage in science and technology. In the grazing incidence pumping scheme a double irradiation is used for lasing output. The first long duration and low intensity transversely

342

A. Demir et al.

pumped pulse is used to ionize the target and form a plasma column. After a delay a second higher intensity shorter duration pulse irradiates the plasma column at grazing incidence. The angle of incidence is chosen so that at a preferential electron density the pulse is refracted into the gain region. In this way the beam will pass through the same density region and the absorption and therefore the efficiency will be increased. 2 Method of Calculation

During this study the modelling of Ne-like Ti has been done by using a modified version of the original EHYBRID code [10]. This modification allowed the spontaneous transition rates to be calculated from the absorption oscillator strengths which are used in the evaluation of the ionisation balance. Besides this the line intensities are calculated with a post-processor code that is coupled to EHYBRID by simulating the excited level population densities. The required atomic data are obtained by using the Cowan code [11]. 112 Ne-like 1s22s22p6 - 1s22s22p5 nl and 214 F-like 1s22s22p5 - 1s22s22p4 nl resonance line intensities have been calculated for the simulation of the spectral emission from the titanium plasma. 3 Results and Discussion

In the simulations firstly a 4 mm slab target is irradiated by a pre-pulse of 70 mJ energy and 200 ps pulse duration to form a plasma column. Afterwards a main pulse of 80 mJ energy and 1.2 ps duration has been sent with a grazing incidence for observing x-ray lasing (Fig.1.). The pulses have a 800 nm wavelength. The optimum delay time between the pulses and the incidence angle are searched for.

Fig. 1. Grazing Incidence Pumping Scheme

Longitudinally Pumped Ne-Like Titanium X-Ray Laser Simulation

343

The variation of the gain of the 32.6 nm Ne-like Ti laser line as a function of the incidence angle of the main pulse to the normal of the target surface for a 500 ps delay time is shown in Fig.2. Maximum gain is obtained around 10o of incidence angle. 32.6 nm Ne-like Ti Lasing Line

50

-1

Gain( cm )

40

30

20

10

0 0

10

20

30

40

50

60

70

80

90

Incidence Angle

Fig. 2. Variation of the gain of the 32.6 nm Ne-like Ti laser line as a function of the incidence angle of the main pulse to the normal of the target surface. Data was obtained using a 70 mJ 200 ps duration pre-pulse followed by an 80 mJ 1.2 ps duration grazing incidence short pulse separated by a 500 ps time delay.

Simulation is performed to determine optimum delay time between the two pulses. Pre heating pulse has 70 mJ energy and 200 ps pulse duration in the simulation. The main pulse has 1.2 ps pulse duration and 80 mJ of energy. Fig.3. shows the variation of the gain as a function of the time delay with a grazing incidence angle of 10o. 70 32.6 nm Ne-like Ti Lasing Line 60

Gain (cm-1)

50

40

30

20

10

0 200

250

300

350

400

450

500

550

600

650

700

Delay Time (ps)

Fig. 3. Variation of the gain of the 32.6 nm Ne-like Ti laser beam as a function of the delay time between the two pulses. Data was obtained using a 70 mJ 200 ps duration pre-pulse followed by a 80 mJ 1.2 ps duration grazing incidence shot at 10o to the normal.

344

A. Demir et al.

Fig.4. shows the simulated time integrated Ne-like and F-like Ti resonance line spectrum for total pulse energy of 150 mJ. The Ne-like resonance lines are more intense than the F-like resonance lines. Ne-like 3d-2p

3.0E+15

Ne-like 3s-2p

F-like 3s-2p

F-like 3s-2p

1.0E+15

F-like 4s-2p

Ne-like 4s-2p

2.0E+15

F-like 3d-2p

Intensity (photons/Å)

4.0E+15

0.0E+00 17

19

21

23 Wavelength Å

25

27

29

Fig. 4. The time integrated Ne-like and F-like resonance line spectrum emitted from Ti plasma.

Fig.5. shows the Ne-like plus F-like resonance line intensity and gain of the Ne-like Ti lasing lines at 32.6 nm as a function of time. The FWHM of the gain is ~64 ps for 32.6 nm. 45

1.4E+22

40

1.2E+22

30 8.0E+21

25

6.0E+21

20

Gain (cm-1)

Intensity (photons/Å)

35 1.0E+22

15 4.0E+21 10 2.0E+21

0.0E+00 710

Ne-like and F-like Resonance Line 32.6 nm Gain Line

720

730

740

750

760 770 Time (ps)

780

790

800

5 0 810

Fig. 5. Variation of the gain and the intensity of the 32.6 nm Ne-like Ti laser beam as a function of time. Data was obtained using a 70 mJ 200 ps duration prepulse followed by a 80 mJ 1.2 ps duration grazing incidence shot at 10o to the normal. The time delay between the two pulses is 500 ps.

Longitudinally Pumped Ne-Like Titanium X-Ray Laser Simulation

345

Ti(I) 3d 4s4p-3d 4s5s

3

2

430

480

530 580 Wavelength (nm)

Ti(III) 3p 3d5d-3p 3d5f

6

6

630

6

6

3

750

0 380

Ti(III) 3p 3d5s-3p 3d5p

3

Ti(I) 3d 4p-3d 4d

3

Ti(I) 3d 4s4p-3d 4s5s

2

3

3

Ti(I) 3d 4s-3d 4p

2 2 2 3

3

1500

Ti(I) 3d 4s -3d 4s4p

6 6

Ti(II) 3d -3d 4p

6

2250

6

Intensity (Au)

3000

Ti(III) 3p 3d4f-3p 3d5g 3 3 + Ti(I) 3d 4s-3d 4p

6

3750

Ti(III) 3p 3d5p-3p 3d5d 3 3 + Ti(I) 3d 4s-3d 4p

6

4500

Ti(II) + Ti (III) 3p 3d4f-3p 3d5g 3 3 + Ti(I) 3d 4s-3d 4p

Fig.6. shows the visible line spectrum emitted from Titanium preplasma created using pre-pulse laser conditions. The energy of the laser pulse is 200 mJ and the pulse duration is approximately 6 ns. The pre-pulse has created low ionized plasma and the main pulse interacts with the preformed plasma to create lasing action. The focus width is 100 μm.

680

730

Fig. 6. Experimentally observed visible spectrum of Titanium under pre-pulse condition.

4 Conclusions

During the simulations firstly the optimum incidence angle corresponding to the irradiation of a Ti target with a 70 mJ 200 ps pre-pulse which is followed by a 80 mJ 1.2 ps main pulse after 500 ps of time delay has been investigated and the optimum incidence angle is found at 10o. The pulses have an 800 nm wavelength. After this at 10o the optimum time delay is searched between 200-700 ps for the data given above and it is seen that 500 ps is near to the optimum case. The FWHM of the gain and the resonance lines between 17 and 29 Å for these optimum cases have been investigated.

346

A. Demir et al.

Acknowledgements This project has been supported by DPT under the contract number K120710.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

Matthews D.L. et al.: Phys. Rev. Let. 54 110 1985 Nickles P.V. et al.: Phys. Rev. Let. 78 2748 1997 Zhang J. et al.: Phys. Rev. A 53 3640 1996 Zhang J. et al.: Phys. Rev. Let. 78 3857 1997 Zhizhan Xu et al.: App. Phys. Let. 63 1023 1993 Kolacek K.: Proc. National Laser Symposium Indore India 2001 Sebban S. et al.: Phys. Rev. Let. 86 3004 2001 Ruxin Li et al.: Phys. Rev. E 57 7093 1998 Keenan R. et al. : Phys. Rev. Let. 94 103901-1 2005 Pert G.: J. Fluid. Mech. 131 401 1983 Cowan R.D.: J. Opt. Soc. Am. 58 808 1968

Simulation of Longitudinally Pumped Ni-Like Molybdenum X-Ray Laser Medium Using PostProcessor Code Coupled to EHYBRID E. Kacar, P. Demir, P. Demir*, A. Demir and S. Yaltkaya** Laser Technologies Research and Application Center University of Kocaeli Aslanbey Campus Kocaeli / TURKEY * Department of Physics METU Ankara/TURKEY ** Department of Physics University of Akdeniz Antalya/TURKEY

Summary. Longitudinally pumped Ni-like Mo x-ray laser media is modeled using EHYBRID and a post-processor code. The required atomic data are obtained using the Cowan code. In this study the pre-formed plasma is pumped on longitudinal direction with a grazing angle. Variation of the x-ray laser gain of the 18.9 nm and 18.2 nm Ni-like Mo lines according to the incidence angle of the main heating pulse normal to the target surface is calculated for different main pulse energies. X-ray resonance lines between 26 and 37 Å emitted from molybdenum plasma have been simulated using post-processor coupled with EHYBRID.

1 Introduction A longitudinal Ni-like Mo x-ray laser at 18.9 nm was proposed in 1998 [1] and it was demonstrated by using a short pulse which pumps the inversion along a pre-formed plasma column [2-5]. The pump energy required to produce a saturated Ni-like Mo x-ray laser can be reduced by pumping preplasma column from a longitudinal direction. An 18.9 nm Ni-like Mo x-ray laser operating close to saturation at 10 Hz repetition rate have been reported to be produced by pumping with 14° grazing incidence angle and 150 mJ of total pumping energy [3]. A plasma column is preformed by irradiating a slab target transversely with a long pulse. Then the short pulse is sent with a grazing angle to the target surface through the laser-produced plasma after a delay and refracted back into the gain region from a predetermined electron density. Refraction of the short pulse in the plasma depends on the incidence angle of the short pulse relative to the target surface. Path length and absorption of the short pulse is increased by

348

E. Kacar et al.

refraction [3]. In experimental investigations dependencies of the longitudinally pumped Ni-like Mo x-ray laser intensity on parameters such as delay time heating pulses’ energy ratio intensity of the main pulse have been observed and the optimal conditions have been obtained [6]. In this paper longitudinally pumped Ni-like Mo x-ray laser is modelled by using EHYBRID code [7]. The modelling was performed for different parameters such as delay time between pulses heating pulses’ energy incidence angle of the main pulse.

2 Method of Grazing Incidence Pumping Simulations Longitudinally pumped Ni-like Mo x-ray laser is modelled using EHYBRID code [7]. Line intensities are calculated in a post-processor coupled with the EHYBRID using the simulated excited level population densities. The orginal EHYBRID code was modified to calculate the spontaneous transition rates from the absorption oscillator strengths used in the evaluation of the ionisation balance. The required atomic data are obtained using the Cowan code [8]. 128 Ni-like 1s22s22p63s23p63d10 1s22s22p63s23p63d9 nl and 230 Co-like 1s22s22p63s23p63d9 1s22s22p63s23p63d8 nl resonance line intensities have been calculated for the simulation of the spectral emission from the molibdenium plasma. In our simulations laser pulses of 0.8 µm wavelength focused on 4 mm target have been simulated. Molibdenium has been pumped with a prepulse and a short pulse. The pre-pulse has 70 mJ energy and 200 ps pulse duration and the following short pulse has 80 mJ energy and 1.2 ps pulse duration. The pre-pulse and the main pulse are separetad by 300 ps.

3 Results and Discussion

A plasma column is preformed by irradiating a slab target transversely with a long pulse. Then the short pulse is sent with a grazing angle to the target surface through the laser-produced plasma after a delay and refracted back into the gain region from a predetermined electron density. Refraction of the short pulse in the plasma depends on the incidence angle of the short pulse relative to the target surface. Variation of the gain of the 18.9 nm and 18.2 nm Ni-like Mo laser lines as a function of the incidence angle of the main heating pulse normal to the target surface is shown in Fig. 1. Maximum gain is obtained around 20° of incidence angle. And also

Simulation of Longitudinally Pumped Ni-Like Molybdenum X-Ray Laser

349

Fig. 1 shows the plasma gain widths for each incidence angle of the main pulse . Gain for 18.9 nm Gain for 18.2 nm

400

Gain width

10

350

-1

Gain (cm )

250 6 200 4

150

Gain width (µm)

8

300

100 2 50 0

0 0

10

20

30

40

50

60

70

80

Incidence Angle

Fig. 1. Variation of the gain of the 18.9 nm and 18.2 nm Ni-like Mo laser lines as a function of the incidence angle of the main heating pulse normal to the target surface. Data was obtained using a 70 mJ 200 ps duration prepulse followed by a 80 mJ 1.2 ps duration grazing incidence short pulse separated by a 300 ps.

90 -1

18.2 nm 18.9 nm

350

80

300 70 250

60

200

50 40

150

30 100 20 50

10

0

Gain of 18.9 mn Ni-like Mo line (cm )

-1

Gain of 18.2 mn Ni-like Mo line (cm )

Simulation is performed to determine optimum energy level of pumping in the range of 18.9 nm and 18.2 nm. Pre heating pulse has 70 mJ energy and 200 ps pulse duration in the simulation. Delay time between the pre-pulse and the short pulse is 300 ps and the main pulse duration is 1.2ps. Fig. 2. shows the variation of gain as a function of the main pulse energy at 20° of incidence angle.

0 0

30

60

90

120

150

180

210

240

Main pulse energy (mJ)

Fig. 2. Variation of the gain of the 18.9 nm and 18.2 nm Ni-like Mo laser beam as a function of the main heating pulse energy at 20o of incidence angle. Data was obtained using a 70 mJ 200 ps duration prepulse followed by a 1.2 ps duration grazing incidence short pulse separated by a 300 ps.

350

E. Kacar et al.

Fig. 3. shows the simulated time-integrated Ni-like and Co-like Mo resonance line spectrum for total pulse energy 150 mJ. The Co-like resonance lines are more intense than the Ni-like resonance lines. Co-like 4f - 3d

5.0E+15

Ni-like 4p - 3d

Ni-like 4f - 3d

Ni-like 5f - 3d

1.0E+16

Co-like 4p - 3d

1.5E+16 Co-like 5f - 3d

Intensity (photons/Å)

2.0E+16

0.0E+00 26

28

30

32

34

36

38

Wavelength (Å)

Fig. 3. The time integrated Ni-like and Co-like resonance line spectrum emitted from Mo plasma.

Fig.4. shows the Ni-like plus Co-like resonance line intensity and gain coefficient of the Ni-like Mo lasing lines at 18.9 nm and 18.2 nm as a function of time from the start of the short pulse interaction time. The FWHM of the gain coefficient duration is ∼ 23 ps for 18.9 nm and for 18.2 nm. Ni-like and Co like resonance line

4.0E+19

300

18.9 nm Gain line 250

200 -1

Gain (cm )

Intensity (photons/Å)

18.2 nm Gain line 3.0E+19

150

2.0E+19

100 1.0E+19 50

0.0E+00 495

0 515

535

555

575

595

Time (ps)

Fig. 4. The time variation of the resonance lines for Ni-like plus Co-like lasing lines and gain coefficient for 18.9 nm and 18.2 nm Ni-like Mo lasing lines.

Simulation of Longitudinally Pumped Ni-Like Molybdenum X-Ray Laser

351

4 Conclusion Longidutinally pumped Ni-like Mo x-ray laser medium is modelled to determine optimum grazing incidence angle of the main heating pulse normal to the target surface and optimum main heating pulse energy. Optimum grazing incidence angle of the main pulse is found around ∼ 20° under the explained simulation conditions in the text. X-ray laser output at 18.2 nm and 18.9 nm and intensity of resonance lines between 26 Å and 37 Å emitted from molibdenium plasma has been simulated using the EHYBRID. Acknowledgements

This project is supported by DPT under the contract number K120710.

References 1. 2. 3. 4. 5.

Li R. et al. Phys.Rev.A 66 047402 2002. Ozaki T. et al. Phys.Rev.A 66 047402 2002. Keenan R.et al. Phys.Rev.Lett. 94 103901 2005. Tümmler J. et al. Phys.Rev.E 72 037401 2005. Larotondo M.A. et al. IEEE J. of Selected Topics in Quant. Electronics 10 6 1363 2006. 6. Kuroda H. et al. 29th EPS Conf. On Plasma Phys. And Contr.Fusion Monterux ECA 26B D-5.005 2002. 7. Pert G.J. Fluid. Mech. 131 401 1983. 8. Cowan R.D. J.Opt.Soc.Am. 58 808 1968

X-Ray Laser at 10–15 nm in Pd-like Ions Er XXIII – Re XXX. E.P. Ivanova, A.L. Ivanov and T.E. Pakhomova Institute of Spectroscopy of Russian Academy of Sciences

Summary. The intensive output radiation at 10-15 nm is possible by the use of the Pd-like scheme of X-ray laser. We have performed the calculations of spectroscopic constants kinetics of level populations in plasmas and gains for eight Pd-like ions ErXXIII – ReXXX. The laser action proved to be efficient at the optically self-pumping transition 4d95f 1P1 – 4d95d 1D1 and at the conventional transition 4d95d 1S0 – 4d95p 3P1. The optimum plasma parameters for lasing and time evolution of gain are calculated. The experimental scheme is implied which use an ultra-short laser pulse pumping.

1 Introduction. The first observations of the Pd-like X-ray laser [1-2] have proved that it is most efficient in compare with the Ne-like either Ni-like schemes. The first reason: the ionization of the electron shells with n=6 occurs at relatively smaller electron energy; thus smaller energy of pumping source is necessary. The second: the optimal electron density (ne=neopt) for X-ray lasing is relatively smaller (ne ~3·1018 - 4·1020 cm-3 for Pd-like ions with Z=54 - 75); hence the recombination and bremstraulung radiation plasma losses are negligible. In [3] we have interpreted the results of experiments [1-2] and suggested the improvements for the pumping and target parameters in order to increase the X-ray radiation yield. During last decade the spectra of Pd-like ions with Z = 52-60 were carefully analyzed (see [4] and references herein) however the spectroscopic data for the heavy ions (Z > 65) are known from the only work [5]. Recently in several independent experiments investigating the interaction between the optical field of intense laser pulse and a xenon cluster beam an anomalously high quantum yield in the plasma radiation in the range 10-15 nm was recorded [6-7]. In our work [8] the nature of this phenomenon was interpreted high conversion efficiency is shown to be possible when producing plasma with optimum parameters for the

354

E.P. Ivanova, A.L. Ivanov and T.E. Pakhomova

amplification of spontaneous radiation on the definite transitions in Xe XXVII. At present nanowires grow by electro-deposition using nonporous template have attracted much attention due to the simplicity of this technique for preparing nano-scaled materials. For instance a standard way to fabricate lead or tin nanowires arrays is suggested in [9]. The production of nanowires arrays of other heavy elements is quite possible. In the present work we imply the experimental scheme similar to that described in [6-7] but nanowires (or dust) array of heavy element (Z = 68-75) is used instead of Xe clusters. Recently the method of relativistic perturbation theory with model potential of zero approximation (RPTMP) was used for the calculation of energy levels of Ag- Rh- and Pd- like ions with Z≤86 [10]. The results of [10] are used here for the calculation of the spectroscopic constants and gains of the Pd-like ions Er XXIII – Re XXX. Optimal plasma parameters for the X-ray lasing at 10-15 nm are determined. The method of gain calculation may be found elsewhere ([8] see the references herein).

2. X-ray laser in Pd-like ions. There are two principal X-ray laser transitions in Pd-like ions; i) 4d3/25d3/2 [J=0]–4d3/25p1/2 [J=1] (5d-5p 0-1); the amplification of this line emission was observed in [1-2]; ii) 4d3/25f5/2 [J=1] – 4d3/25d3/2 [J=0] (5f–5d 1-1). The amplification on the 5f-5d 1-1 optically self-pumping transition is weak in the light Pd-like ions and was not observed experimentally yet. However the amplification grows significantly along the sequence. The kinetics of level population is calculated here with the following assumptions: i) Plasma is produced in the shape of cylinder with diameter d = 100 μm and length L. ii) The pumping pulse parameters are such that plasma with initial temperature Te (at the moment τ = 0) in which the Pdlike ions constitute 90% of the plasma and are in the ground state is produced immediately after the interaction of the optical field laser with array of nanowires or dust). iii) Plasma parameters: electron density (ne) temperature (Te) and diameter (d) are not varying with time. iv) The electron and ion energy distribution are Maxwellian and the shape of the distribution plays no significant role in calculating the rates of the transitions induced by electron-ion collisions. v) The ion temperature Ti = Te/10. For each ion Er XXIII – Re XXX we calculate the gain g(neTed|τ) with unchanged d = 100 μm. For each transition at given Te the optimal value

This page intentionally blank

X-Ray Laser at 10–15 nm in Pd-like Ions Er XXIII – Re XXX.

355

ne= neopt is determined by the condition that the time averaged ĝ = g(neoptTed|τ) is maximum value. Figure 1 shows the time dependences of the gains for the 5d-5p 0-1 (a λ = 107.8 Å) and 5f-5d 1-1 (b λ = 139.5Å) transitions in W XXIX. The decay of the gain of the 5d-5p 0-1 transition is conditioned predominantly by ionization of the W XXIX into higher stages while the decay for 5f-5d 1-1 transition is due to collisional mixing of level populations ⎯ as a result the gain disappears on a shorter time interval than it does on the 5d-5p 0-1 transition. The gain is presented for three values of Te and ne = 1020 cm-1 for the 5d-5p 0-1 transition (Fig.1a) ne = 3·1020 cm-3 for the 5f-5d 1-1 transition (Fig.1b). 120

100

a) g (cm-

80

60 40

20 0 0.01

0.1

1

τ

10

100

1

τ

10

100

300

b) g (cm250 200

150 100

50 0 0,01

0,1

Fig. 1. Time evolution of the gain (g) in W XXIX for two transitions at Te = 0.5 (open circles) 1 (black circles) 1.5 (triangles) keV; d = 100 μm; a) 5d-5p 0-1 transition (λ = 10.78 nm); ne = 1020 cm-3; b) 5f-5d 1-1 transition (λ = 13.95 nm) ne = 3·1020 cm-3.

356

E.P. Ivanova, A.L. Ivanov and T.E. Pakhomova

a)

b)

λ [Å] Fig. 2. Model W XXIX spectra calculated with accounting for theamplification at Te=1.5 keV d=100 μm λ = 100 Å - 160 Å a) ne = 1020 cm-3 L=0.4 cm b) ne = 3·1020 cm-3 L=0.15 cm.

The model spectra in the region 10 – 16 nm with accounting for the amplification are shown in Fig.2 ab. One can see also the second strong 5d-5p 0-1 laser transition at 12.08 nm and the second strong 5f-5d 1-1 transition at 14.48 nm. We used the time-averaged values of g(τ)=ĝ to compute the spectra. In Fig.2 the plasma length L corresponds to the saturation of the amplification. At relatively low Te≤ 700 eV amplification saturation is conditioned by the time duration of lasing. At high Te ≥ 1000 eV time-averaged ĝ is more than 100 cm-1 in this case the saturation is

X-Ray Laser at 10–15 nm in Pd-like Ions Er XXIII – Re XXX.

357

caused by too high inversion value. In this case Lsat determination should be based on the model that accounts for the interaction between X-ray laser field and Pd-like ions. Table 1. Characteristics of X-ray laser on the transition 5d-5p 0-1 in the Pd-like ions.

Z λ [nm] Aul· 1010 Al0· 1011 neopt 1019 R· 10-10 Δν· 1012 I0· 1029 g0 cm-1

68

69

70

71

72

73

74

75

14.61

13.84

13.15

12.49

11.88

11.32

10.78

10.29

4.70

5.52

6.44

7.45

8.57

9.80

11.2

12.7

2.69

3.30

3.94

4.65

5.44

6.31

7.26

8.29

4.0

4.0

5.0

5.0

6.0

6.0

7.0

7.0

9.18

8.75

8.00

7.09

6.56

6.06

5.65

5.02

1.4

1.4

1.4

2.2

1.7

2.0

2.1

2.3

0.8

0.9

1.0

1.2

1.7

2.0

2.6

2.1

32.0

30.4

40.1

17.4

42.9

35.9

44.4

24.6

To make an estimate of Lsat the condition ĝ·Lsat ≈ 15-16 might be used. The results of spectroscopic constants and gain calculations for Er XXIII – Re XXX are summarized in the Tables 1 and Tables 2. For all Pd-like ions the calculations have been performed with the same values Te = 1000 eV d = 100 μm. In the table 1 time-averaging for g(τ) I0(τ) is performed on the interval 0 - 21 ps in the table 2 - on the interval 0 – 5.2 ps. One can see that radiative and collisional transition probabilities as well as time averaged emissive power I0 change gradually along this piece of the sequence. However ĝ value exhibits abrupt jumps along Z what is meant by the inversion sensitivity to a mixing of level populations due to the multitude of transitions induced by electron collision.

358

E.P. Ivanova, A.L. Ivanov and T.E. Pakhomova

Table 2. Characteristics of X-ray laser on the transition with a short-lived inversion 5f-5d 1-1 in Pd-like ions.

Z λ [nm] Aul · 1010 Au0· 1012 neopt 1020 R· 10-11 Δν· 1012 I0· 1030 Ĝ cm-1

68

69

70

71

72

73

74

75

17.64

16.92

16.24

15.61

15.02

14.47

13.95

13.47

7.41

7.92

8.43

8.96

9.50

1.00

1.06

1.12

1.06

1.33

1.64

2.00

2.39

2.83

3.31

3.85

1.4

1.8

2.2

2.6

3.0

3.2

3.6

3.9

2.74

2.71

2.68

2.63

2.57

2.56

2.57

2.28

2.2

2.8

3.1

3.7

3.4

4.1

3.8

5.2

0.3

0.4

0.5

0.5

0.7

0.9

1.0

1.1

186.0

120.8

115.9

76.9

116.3

104.9

125.3

68.5

Note: Wavelengths λ (nm) radiative transition probabilities (RTP) from upper to low active level Aul(s-1). RTP from the low active level 4d3/25p1/2 [J=1] to the ground level Al0(s-1) RTP from the upper active level 4d3/25f5/2 to the ground level Au0(s-1). The rates of level excitation by electron collision from the ground state per unit volume R(cm--3s-1) the Voight line width Δν(s-1) the time-averaged transition emissive power (without accounting for an amplification) I0(eV/cm3s) and the time-averaged gain ĝ (cm--1) were calculated at ne=neopt Te=1000 eV d=100μm.

3. Conclusion This investigation allows drawing the following principal conclusions: a. In all ions ĝL ≥ 14 is possible at Te ≥ 800 eV. b. The time it takes for the gain decay (τlas) reduces at larger ne.τlas = 30 – 60 ps for the 5d-5p 0-1 transition; τlas = 6 – 10 ps for the 5f – 5d 1-1 transition. In each ion the value neopt for the transition 5d-5p 0-1 is few times smaller than for the optically self-pumping transition 5f-5d 1-1. c. Amplifications on the 5f-5d 1-1 transitions with a short-lived inversion are possible only with the use of an high intensity ultra

X-Ray Laser at 10–15 nm in Pd-like Ions Er XXIII – Re XXX.

359

short laser pulse pumping. Amplifications on the 5d-5p0-1 transitions are possible with either ultra short or long lasting laser pulse pumping. This investigation confirmed beyond any reasonable doubt that the Pd-like X-ray laser scheme is promising to attain the most efficiency as compared to the Ne-like either Ni-like scheme. The other models of an X-ray laser derived from the Pd-like scheme are under development.

References 1. Lemoff B.E. Barty C.P.J. Harris S.E.: Opt. Let. 19 569-572 1994. 2. Sebban S. et.al. : Phys. Rev. Lett. 86 3004-3007 2001. 3. Ivanova E.P. Ivanov A.L.: ‘Theoretical search for optimal pump parameters for observing spontaneous radiation amplification on the λ=41.8 nm transition of Xe IX in plasma’ Quantum Electronics 34 1013-1017 2004. 4. Churilov S.S. et al: ‘Analysis of the spectra of Pd-like Praseodymium and Neodymium’ Physica Scripta 71 589-598 2005. 5. Sugar J. Kaufman V. Rowan W.: ‘Observation of Pd-like resonance lines through Pt32+ and Zn-like resonance lines of Er38+ and Hf42+’ J. Opt. Soc. Am. 10 799-801 1993. 6. Ter-Avetisyan S. Schnürer M. Stiel H. et.al. ‘Absolute extreme ultraviolet yield from femtosecond-laser excited Xe clusters’ Phys. Rev. E 64 0364041-8 2001. 7. Mori M. et. al. ‘Extreme ultraviolet emission from Xe clusters excited by high- intensity lasers’ J. Applied Physics 60 3595-3601 2001. 8. Ivanova E.P. Ivanov A.L. ‘A Superpower Source of Far-Ultraviolet Monochromatic Radiation’ J. Exp. Theor. Phys. 100 844-856 2005. 9. Michotte S. ‘Standard manufacture and properties of arrays of Pb and Sn nanowires’ Int. J. of Modern Physics B 17 4601-4618 2003. 10. Ivanova E.P. ‘Energy levels of the ions of the isoelectronic sequences Silver and Rhodium’ Optics and Spectroscopy (Russian)

Temporal Characterization of Attosecond Harmonic Pulses C. H. Nam and K. T. Kim Department of Physics and Coherent X-ray Research Center, Korea Advanced Institute of Science and Technology (KAIST), Daejeon 305-701, Korea

Summary. Techniques to characterize attosecond harmonic pulses based on the cross-correlation between harmonic and IR laser pulses are presented. The selfcompression of an attosecond harmonic pulse in the harmonic generation itself has been demonstrated, obtaining near transform-limited attosecond harmonic pulses.

1. Introduction High harmonic x-ray sources possess unique features suitable for attosecond physics and science [1,2]. High harmonics from atoms driven by intense femtosecond laser pulse can form an attosecond pulse train or a single attosecond pulse when properly controlled and selected. For proper understanding of the interactions between attosecond pulses with matter, the temporal structure of attosecond pulses should first be well characterized. Since efficient nonlinear material for two photon processes is difficult to realize in the xuv wavelength region, autocorrelation techniques, normally used for the characterization of femtosecond pulses, can be applied only to limited cases of low-order harmonic pulses. Cross correlation techniques based on the photoionization of atoms by high harmonic and femtosecond laser pulses, acting simultaneously, are thus useful for the characterization of attosecond pulses. As a realization of the cross correlation technique, the reconstruction of attosecond beating by interference of two-photon transition (RABITT) method was demonstrated by Paul et al. [3]. An attosecond pulse train with 250-as duration was measured, and attosecond intrinsic chirp structure was revealed as theoretically predicted [4,5]. For complete characterization of attosecond pulses, the frequency resolved optical gating method for the complete reconstruction of attosecond bursts (FROG CRAB) was proposed by Mairesse et al. [6]. The demon-

362

C. H. Nam and K. T. Kim

stration of FROG-CRAB would provide the detailed temporal structure of attosecond harmonic pulses. Here we present techniques to characterize attosecond harmonic pulses, especially the FROG-CRAB method, and the self-compression of an attosecond harmonic pulse to generate near transform-limited attosecond harmonic pulses.

2. Temporal characterization of attosecond harmonic pulses Attosecond pulses obtained from high harmonic generation contain a complex chirp structure. In single atom calculations, each single pulse in an attosecond pulse train is positively chirped [5], while each harmonic is negatively chirped [7]. For rigorous temporal characterization of such complex attosecond harmonic pulses, one needs to precisely determine the spectral phase and amplitude of the attosecond pulses. In the RABITT method, photoionization of atoms by attosecond harmonic pulse and IR laser pulse is used for the temporal reconstruction of the attosecond pulse. The photoelectron spectrum shows sidebands due to the interference between photoelectrons with same energy but generated through different two-photon transitions. The sideband is modulated as the time delay between the harmonic and IR probe pulses changes. The phase information of the attosecond pulses can then be found in the side band modulation as A cos ( 2ω0τ + Δϕq ) .

Here, A is the amplitude of the modulation, ω0 is the laser frequency, τ is the time delay between harmonic and laser pulses, and Δϕ0 is the phase difference between (q+1)th and (q-1)th harmonics. The spectral amplitude can be obtained from the photoelectron spectrum obtained only with the harmonic pulse. The reconstruction of the attosecond harmonic pulse is possible from those phase and amplitude information. However, this technique has some drawbacks because the attosecond harmonic pulses are assumed to be the sum of the discrete harmonics, separated by 2ω0. It cannot be used for the reconstruction of a single attosecond pulse having continuum spectrum. Furthermore, the reconstruction result is always an attosecond train with identical individual pulses which is the average of the real pulse train. The chirp information on individual pulses is not available. The FROG CRAB method can provide the complete information on attosecond pulses. In this technique, the photoelectron spectra obtained with harmonic and laser pulses with time delay τ can be represented by

Temporal Characterization of Attosecond Harmonic Pulses S ( ω ,τ ) =

∫

+∞

−∞

2

dt G ( t ) E X ( t − τ ) eiωt .

363

(1)

Here E X ( t ) is the harmonic electric field to be measured and G(t) is the phase gate function defined by +∞ G ( t ) = exp ⎡i ∫ dt ′ { v ⋅ A ( t ′ ) + A 2 ( t ′ ) / 2}⎤ , ⎣⎢ t ⎦⎥

(2)

where v is electron velocity and A(t) is the vector potential of IR laser pulse. Since Eq. (1) is the spectrogram expressed in frequency and time delay, a conventional FROG inversion algorithm, such as the principal component generalized projection algorithm (PCGPA), may be used to reconstruct the attosecond harmonic pulse [6]. However, slight modification is necessary to overcome the problem that the gate function does not have same values at both ends. Consequently, one can retrieve the harmonic electric field containing the detailed information of the attosecond harmonic pulse.

3. Experiment setup The schematic of the experimental setup is shown in Fig. 1. A 1-kHz Ti:sapphire laser, generating pulses of 30-fs duration, was used to obtain high harmonics. The laser beam was split into two parts by a beam splitter; the first beam was focused into a gas cell for high harmonic generation. The second beam was used as a probe laser beam. After harmonic generation, the transmitted laser beam was blocked by a 200-nm aluminum filter

Fig. 1. Schematics of the experiment setup.

364

C. H. Nam and K. T. Kim

to completely eliminate the laser light. The harmonic and the probe beams were combined using a mirror having a hole in the center, and both beams were then focused together, using a gold-coated toroidal mirror, into an effusive atomic beam and generated photoelectrons were measured with a time-of-flight photoelectron spectrometer.

4. Results

Intensity (arb. units)

First, applying the RABITT technique, we achieved the generation of selfcompressed attosecond harmonic pulses. A 30-fs laser pulse with intensity of 2.5×1014 W/cm2 was applied to an argon gas cell with length of 12 mm. Photoelectron spectra were obtained for harmonics from 17th to 41st orders. As the argon pressure was increased from 15 torr to 40 torr, the harmonics below 25th order were severely absorbed, and the intrinsic positive chirp was compensated by the negative group delay dispersion of argon itself. The reconstructed temporal profile of the attosecond harmonic pulse is shown in Fig. 2. The measured pulse width of 206 as is slightly longer than the transform-limited width of 200 as. 1.0 0.8 0.6 0.4 0.2 0.0 -600

-400

-200

0

200

Time (as)

400

600

Fig. 2. Self-compressed attosecond harmonic pulse measured using RABITT.

Second, FROG CRAB measurements were carried out using the same experiment setup, but with different harmonic conditions due to the poor energy resolution of the spectrometer at high photoelectron energy. Attosecond harmonic pulses, consisting of harmonics up to 31st order, were generated with a 6-mm-long 25-torr argon gas cell. Photoelectron spectra obtained by changing the time delay between the harmonic and IR laser pulses is shown in Fig. 3 (a). The temporal reconstruction of the harmonic pulses was carried out using a modified principal component generalized

Temporal Characterization of Attosecond Harmonic Pulses

365

a)

b) Fig. 3. (a) Photoelectron spectra obtained by applying harmonic pulse with IR laser pulse with time delay. (b) Temporal profile of an attosecond harmonic pulse reconstructed using the FROG CRAB technique.

projections algorithm (PCGPA). Figure 3(b) shows the temporal reconstruction of the attosecond harmonic pulse. The envelope width of the pulse train is 11 fs and the width of the single attosecond pulse at the center of the train is 230 as. In this case, the intrinsic attosecond chirp at the center of the train and the harmonic chirp of the 17th order are estimated to be 1.4×10-32 s2 and -2.3×10-28 s2, respectively. Since the modified PCGPA

This page intentionally blank

366

C. H. Nam and K. T. Kim

is a blind FROG technique, the FROG CRAB measurement provides the information on both harmonic pulse and laser pulse.

5. Conclusion We have demonstrated the temporal characterization of attosecond harmonic pulses using both RABITT and FROG CRAB techniques. The 206as self-compressed attosecond pulses was reconstructed using RABITT. And the FROG CRAB technique has been applied for the complete reconstruction of attosecond harmonic pulses, obtaining an attosecond pulse train of 11-fs envelope width and 230-as width at the pulse peak.

References [1] M. Hentschel, R. Kienberger, Ch. Spielmann, G. A. Reider, N. Milosevic, T. Brabec, P. Corkum, U. Heinzmann, M. Drescher, F. Krausz, Nature 414, 509 (2001). [2] M. Drescher, M. Hentschel, R. Kienberger, M. Uiberacker, V. Yakovlev, A. Scrinzi, Th. Westerwalbesloh, U. Kleineberg, U. Heinzmann, F. Krausz, Nature 419, 803 (2002). [3] P. M. Paul, E. S. Toma, P. Breger, G. Mullot, F. Augé, Ph. Balcou, H. G. Muller, and P. Agostini, Science 292, 1689 (2001). [4] Y. Mairesse, A. de Bohan, L. J. Frasinski, H. Merdji, L. C. Dinu, P. Monchicourt, P. Breger, M. Kovacev, R. Taïeb, B. Carré, H. G. Muller, P. Agostini, and P. Salières, Science 302, 1540 (2003). [5] K. T. Kim, C. M. Kim, M.–G. Baik, G. Umesh, and C. H. Nam, Phys. Rev. A 69, 051805(R) (2004). [6] Y. Mairesse and F. Quéré, Phys. Rev. A 71, 011401(R) (2005) [7] J.-H. Kim and C. H. Nam, Phys. Rev. A 65, 033801 (2002).

Nonlinear Interaction of Intense Attosecond XUV Pulses with Atoms and Molecules K. Midorikawa, T. Shimizu and Y. Nabekawa Laser Technology Laboratory, RIKEN

Summary. We have observed nonlinear optical processes such as two-photon double ionization and above threshold ionization of rare gases in the xuv region with intense high-order harmonics. Using two-photon double ionization in He, the pulse width of the 27th (42 eV) harmonic was measured by an autocorrelation technique, and found it to be 8 ns. A train of attosecond pulses was also characterized directly by the energy-resolved autocorrelation of the above threshold ionized electrons.

1 Introduction The progress of chirped pulse amplification and femtosecond Ti:sapphire laser enables the study of the interaction of strong optical fields with atoms and molecules. A variety of interesting phenomena, including highharmonic generation, high energy radiation/particles emission, and Coulomb explosion of molecules, have been investigated intensively. The underlying physics of these extremely nonlinear optical phenomena have been explored through the interaction with low-frequency radiations such as infrared or visible light; however, little has been understood concerning the interaction of intense high-frequency radiation. Great interest has been aroused recently in the interaction of intense high-frequency radiation with matters. Yet no one could observe it because of the lack of an intense coherent light source in this spectral region. The induced phenomena are expected to be much different from those caused by low frequency radiation because the electron quiver energy—the cycle averaged electron energy in the optical field—is proportional to a square of the wavelength. In this paper, we report on the generation of intense soft X-ray pulses and its application to nonlinear multiphoton processes. Using such nonlinear processes, the temporal width of the 42-eV soft x-ray pulse was meas-

368

K. Midorikawa, T. Shimizu and Y. Nabekawa

ured directly by an autocorrelation technique. A train of attosecond pulses was also characterized by frequency-resolved autocorrelation. Intense high harmonics produced by the phase-matching technique enable the observation of these nonlinear optical processes.

2 High-Power High Harmonic Generation For application of high harmonics to nonlinear optics in the XUV region, high focused intensity is crucial because the cross section of nonlinear processes in atoms or molecules tends to decrease rapidly with decrease of the pump wavelength. To increase high harmonic energy by improving conversion efficiency from the pump energy, phase matching is essential. It is, however, not easy to satisfy the phase matching condition along the interaction length, because in contrast to low-order harmonic generation in the perturbative regime, the atomic dipole phase is dependent on the driving laser intensity [1]. This means that the atomic phase varies rapidly around the focus when the pump pulse is tightly focused in the nonlinear medium. Furthermore, in addition to the medium’s dispersion, nonlinear phenomena such as self-focusing and plasma defocusing accompanied by a high-intensity pump pulse also make the experimental achievement of the phase matching quite difficult. Overcoming such difficulties, the energy scaling of high-order harmonics under the phase-matched condition has been achieved using a long interaction length and a loosely focused pumping geometry [2]. This method shows a linear increase in harmonic energy with respect to the geometrical focusing area of the pump pulse, while keeping an almost perfect spatial profile of the harmonic output. Peak powers of 130 MW at 62.3 nm in Xe [3], 10 MW at 29.6 nm in Ar [2], and 1 MW at 13.5 nm in Ne [4] are obtained with femtosecond, high-power Ti:sapphire laser pulses.

3 Focusing Property of High Harmonics Although the advent of high harmonics having excellent beam quality and spatial coherence, together with high output energy, opens up the possibility of focusing XUV beams as intensity sufficient for inducing nonlinear optical processes, a technique for focusing an intense, coherent soft x-ray beam and its characterization should be established before the observation of nonlinear phenomena. Figure 1(a) shows the 1/e2 beam spot size of the 27th harmonic beam as a function of distance from the off-axis parabolic

Nonlinear Interaction of Intense Attosecond XUV Pulses with Atoms

369

mirror. In this measurement, the 27th harmonic wave was selected and focused to a Ce:YAG scintillator by an off-axis parabolic mirror with a SiC/Mg multilayer coating having 40% reflectivity. A set of image relay optics transferred the visible light image onto an image-intensified chargecoupled device (CCD) camera. The red circles and blue squares correspond to the horizontal and vertical spot sizes, respectively. There is little difference between the horizontal and vertical diameters, showing that circularity is maintained during focusing. The best-fit curve provided a figure of M2 = 1.4, indicating nearly perfect spatial beam quality [5]. The focusing property was also investigated by an ablation pattern produced on a gold-coated mirror placed at the focal position. Figure 1(b) shows the atomic force microscope image of the ablation pattern. Note that the pattern was produced by a single shot irradiation of the focused 27th harmonic pulse. The diameter of the circular hole is approximately 2 μm, which agrees well with the scintillator experiment. Because the high-order harmonics have excellent spatial coherence and beam quality, the focusability might be dominated by the performance of the mirror. The focused intensity of the 27th harmonic wave was estimated from those experimental results of the energy, spot size, and pulse width. Assuming that the pulse width of the 27th harmonic wave was the same as the fundamental pulse, the maximum focused intensities of 1.0 x 1014 W/cm2 were obtained. 20 Horiz. Vert. 15 μm

p o

0 .5

10

0 - 0 .5 - 10

5

-5

0

5 μm 60

60.5

0

61

61.5

62

5 1 0 μm

Distance from focusing mirror [mm]

(a)

(b)

Fig. 1. Focusing property of the 27th harmonic pulse: (a) spot size as a function of middor distance and (b) ablation pattern produced on a gold coated mirror surface by a single-shot irradiation.

370

K. Midorikawa, T. Shimizu and Y. Nabekawa

4 Two-Photon Double Ionization in He at 42 eV As the first step in investigating the interaction of intense high-frequency radiation with matter, the He atom has been selected as a target. At the very high frequency and low intensity limit, double ionization by single photon absorption has been investigated in detail using synchrotron radiation; this process has been understood as “shake off.” On the other hand, in the low-frequency and high-intensity regime realized by femtosecond high intensity Ti:sapphire lasers, the occurrence of double ionization has also been reported and the process was explained by a rescattering model. Between these two extreme experimental conditions, the occurrence of twophoton double ionization by intense XUV lights was theoretically predicted about 20 years ago. Particularly, two-photon double ionization in He has attracted much interest and has been studied theoretically in a number of papers because it provides insight into the electron-electron interaction (that is, electron correlation) and paves the way for unexplored aspects of the three-body problem [6, 7]. Figure 2(a) shows the ion spectrum produced by the interaction between the focused 27th harmonic pulses and 3 He. The production of 3He2+ results dominantly from the one-photon absorption of 3He. The strongest peak of 1.8 μs can be assigned to 3He2+. The signal of 3He2+ clearly appears between the H+ signal and H2+ signal. The generation of doubly charged He2+ confirms the observation of a nonlinear optical process (two-photon absorption) in the soft x-ray region. To further confirm that the He2+ ions are produced via the two-photon process, the 27th harmonic intensity dependence of the He2+ yield was investigated. As shown in Fig. 2(b), the result clearly showed the slope of two, giving more evidence of the occurrence of the two-photon double ionization. This is the first observation of a nonlinear optical process in the soft x-ray region [8, 9]. The XUV nonlinear optical process is not only interesting in the field of atomic or molecular physics, it is also of importance for the direct measurement of the pulse width by autocorrelation method in the ultrafast optics field. With the two-photon double ionization in He, the pulse duration of the 42-eV soft x-ray pulse was measured by means of autocorrelation. For the autocorrelation measurement, an autocorrelator splits a measured pulse equally into two pulses and spatially overlaps the separated delayed pulses at the focus point. The construction of the autocorrelator for the XUV light is, however, not straightforward because no beam splitter or high reflectance-/transmittance optics are available. Therefore, a novel autocorrelator using a split beam separator was designed [10]. The delayed pair of harmonic pulses is produced by spatially dividing the harmonic

Nonlinear Interaction of Intense Attosecond XUV Pulses with Atoms

371

8 6

3

He2+

(a)

3

H2

H+

He+

+

He2+ yield

4

(b)

2

0.01

8 6

Slope = 2

4 2

0.001 1.0

1.2

1.4

1.6

1.8

2.0

Time of flight ( μs)

2.2

2.4

2

3

4

5 6 7 8 9

1

2

+ He yield Harmonic Intensity

Fig. 2. (a) The time of flight ion spectrum and (b) doubly charged helium ion yield as a function of the 27th harmonic intensity.

beam with two beam separators of SiC substrates. One of the divided pulses is delayed or advanced to another pulse by moving one of the separators placed on a translation stage with a piezo-actuator. The measurement principle using this split beam separator is the same as an allreflective interferometric autocorrelator. With this autocorrelator and twophoton double ionization in He, a pulse width of the 42-eV soft X-ray radiation was measured for the first time and it was determined to be 8 fs [8].

5 Temporal Characterization of XUV Attosecond Pulse Train by Autocorrelation Although the temporal width of the soft X-ray pulse was measured directly by an intensity autocorrelation technique, the pulse width is limited by the bandwidth of the 27th harmonic wave. The narrow bandwidth of a SiC/Mg multilayer mirror has high reflectivity only for the 27th harmonic wavelength. If several harmonics are simultaneously focused with a broadband mirror, a train of extremely short pulses would be observed. As is well known, Fourier synthesis of harmonic wave fields exhibits a train of ultrashort pulses. The pulse width decreases in inverse proportion to the number of harmonics. The optical field in the XUV region generated as high order harmonics of a femtosecond laser pulse is one of the most significant examples of the Fourier synthesis because it can serve an attosecond pulse train. For characterization of an attosecond pulse train, several harmonics should be simultaneously measured and the information of individual harmonic phase should be determined. But the phase information can hardly

372

K. Midorikawa, T. Shimizu and Y. Nabekawa

be obtained from ion signals. Therefore, a new method to reconstruct an attosecond pulse train was proposed [11]. This method uses electrons produced by the above threshold ionization (ATI) process, another type of nonlinear mutlitphoton process. In the ATI process, an ejected electron absorbs the photons in excess of the minimum required for ionization. The ejected electrons carry the information of the phase of the XUV pulse used for ionization.

Fig. 3. Mode-resolved autocorrelation trace of two-photon above threshold ionization in Ar and the reconstrcuted train of attosecond pulses.

In the experiment, the electron energy spectra produced by the two-photon ATI process in Ar are utilized as mode-resolved signals of the autocorrelation measurement for the pulse train and its envelope [11, 12]. To simplify the analysis of the ATI electron spectra, only three harmonics from 11th to 15th were selected by passing through a Sn thin-foil filter.

Nonlinear Interaction of Intense Attosecond XUV Pulses with Atoms

373

When these three harmonics are simultaneously focused to Ar gas, five peaks form the lowest-order mode at 22nd (11th + 11th) to the highest order mode at 30th (15th + 15th) are expected to appear in the photo-electron spectrum. Among these five processes, the 24th (11th + 13th) and the 28th (13th + 15th) processes show beating due to interference of two harmonics, while only the 26th (13th + 13th, 11th + 15th) process contains the phase information between the three harmonics. The three-dimensional image of the mode-resolved autocorrelation with two-photon ionization electrons is shown in Fig. 3. From the spectral analysis, the chirp among the three harmonic fields was specified and it was found that a pulse duration should be shorter than 450 as, which is the first determination of the chirp in the attosecond pulse train with an autocorrelation technique [11]. Conforming to the custom that a method characterizing ultrashort pulses is named after creatures like FROG (frequency resolved optical gating) or RABITT (resolution of attosecond beating by interference of two-photon transitions) in the ultrafast optics field, this method is called PANTHER (photoelectron analysis of nonresonant two-photon ionization for harmonic electric-field reconstruction). PANTHER will open the way to full characterization of an attosecond pulse train.

Acknowledgment The authors wish to thank Drs. K. Yamanouchi, H. Hasegawa, K. Furusawa, K. Ishikawa, E. Takahashi, T. Okino, A. Suda, and H. Mashiko for their experimental contribution and helpful discussions.

References 1. P. Saliéres, A. L'Huillier, M. Lewenstein: Phys. Rev. Lett. 74, 3776–3779, 1995. 2. E. Takahashi, Y. Nabekawa, T. Otsuka, M. Obara, K. Midorikawa: Phys. Rev. A 66, 021802(R), 2002. 3. E. Takahashi, Y. Nabekawa, K. Midorikawa: Opt. Lett. 27, 1920 –1922, 2002. 4. E. J. Takahashi, Y.NabekawA, K. Midorikawa: Appl. Phys. Lett. 84, 4-6, 2002. 5. H. Mashiko, A. Suda. K. Midorikawa: Opt. Lett. 29, 1927-1929, 2004. 6. H. Bachau, P. Lambropoulus: Phys. Rev. A 44, R9-R12, 1991. 7. P. Lambropoulos, L. A. A. Nikolopoulos, M. G. Makris: Phys. Rev. A 72, 013410, 2005.

374

K. Midorikawa, T. Shimizu and Y. Nabekawa

8. Y. Nabekawa, H. Hasegawa, E. J. Takahashi, K. Midorikawa: Phys. Rev. Lett. 94, 043001, 2005. 9. H. Hasegawa, E. J. Takahashi, Y. Nabekawa, K. L. Ishikawa, K. Midorikawa: Phys. Rev. A 71, 023407, 2005. 10. H. Hasegawa, E. J. Takahashi, Y. Nabekawa, K. Midorikawa: Laser Phys. 15, 812-820, 2005. 11. Y. Nabekawa, T. Shimizu, T. Okino, K. Furusawa, H. Hasegawa, K. Yamnouchi, K. Midorikawa: Phys. Rev. Lett. 96, 083901, 2006. 12. K. Furusawa, T. Okino, T. Shimizu, H. Hasegawa, Y. Nabekawa, K. Yamanouchi, and K. Midorikawa: Appl. Phys. B 83, 203-211, 2006.

Anomalous Enhancement of Single High-Order Harmonic Generation at 61 nm and 47 nm by Indium and Tin Due to Strong Resonance M. Suzuki, M. Baba, R. A. Ganeev and H. Kuroda Institute for Solid State Physics, University of Tokyo T. Ozaki Institut national de la recherché, Universite du Quebec

Summary. We have successfully demonstrated the intensity enhancement of single high-order harmonic at a wavelength of 61.26 and 46.76 nm using low ionized indium and tin ions in laser ablation plume as the nonlinear medium. The ablation plume was produced by irradiating of solid tin target with 10 mJ energy picosecond laser pulse. Using the indium target, the 13th harmonic at the wavelength of 61.26 nm was obtained with conversion efficiency of 8x10-5. Using the tin target, the 17th harmonic at a wavelength of 46.76 nm was observed with conversion efficiency of about 10-4. We attribute the strong enhancement of single high-order harmonic to multiphoton resonance with a strong radiative transition of the In and Sn II ions.

1 Introduction High-order harmonic generation (HHG) is a very attractive radiation source in the extreme ultraviolet (XUV) and soft x-ray region, with good spatial quality and femtosecond resolution. One of the most important subjects of HHG research is the enhancement of conversion efficiency. In the past, such conversion efficiency enhancement has been pursued by controlling the phase matching condition in the gas filled capillary or gas cell. The strongest output energy of harmonics were 7 μJ for the 11th harmonic at a wavelength of 72.7 nm, 4.7μJ for 13th harmonic at a wavelength of 62.3 nm, and 1μJ for 15th harmonic at a wavelength of 54 nm using xenon gas-cell.1 Very recent results have reported an enhancement of high-order harmonic conversion efficiency when a two-color (fundamental and sec-

376

M. Suzuki et al.

ond harmonic field) orthogonally polarized driving field is used.2 Using the helium from the gas-jet, the conversion efficiency of the 38th harmonic at a wavelength of 21.6 nm was 5x10-5. An alternative approach is the resonance enhancement of HHG, which has been proposed by theoretical calculations.3 Experimentally, the resonance enhancement of the 13th and the 15th order harmonics generation using the argon gas medium has been observed.4 However the enhancement in these works were a factor of two to three, and the harmonic spectrum consisted of multiple harmonics. In this proceeding, we report the first observation of the single highorder harmonic enhancement in the XUV region. In this experiments, the nonlinear medium was indium and tin ablation plumes, produced by a lowenergy laser pulse, instead of the conventional gas medium. Using the tin target, the intensity of the 17th order harmonic at the wavelength of 46.76 nm was 20 times higher than its neighbor harmonics. The output energy of only this harmonic is measured to be about 1.1 μJ. Such output energy that was obtained the maximum output energy in the XUV region. Using the indium target, the conversion efficiency of the 13th harmonic at a wavelength of 61 nm was about 10-4, which was two orders of magnitude higher than its neighboring harmonics. The output energy of the 13th harmonic was measured to be 0.8 μJ.5 The origin of these enhancements is attributed to resonance with strong radiative transitions, produced within the laserablated plume.

2 Experiments The schematic of experimental setup was described in another proceeding.6 The pump laser was a commercial, chirped pulse amplification laser system (Spectra Physics: TAS-10F), whose output was further amplified using a homemade three-pass amplifier operating at a 10 Hz repetition rate. A pre-pulse was split from the amplified laser beam by a beam splitter before a pulse compressor. The pre-pulse energy is 10 mJ with pulse duration of 210 ps. A main pump pulse output at a center wavelength of 795 nm has energy of 10 mJ with pulse duration of 150 fs. A cylindrical lens focuses the pre-pulse onto a solid tin target placed within a vacuum chamber, which generates a laser ablation plume that contains low-charged ions. The size of the line focus on the target surface was 600 μm width and 1mm long, and the intensity of the pre-pulse was varied between 1010-1011Wcm2. The main pulse is focused onto the ablation plume by a spherical lens (focus length of 200 mm), 100 ns after pre-pulse irradiation. The intensity of the main pulse at the plasma plume is 1014Wcm-2. The spectrum of the

Anomalous Enhancement of Single High-Order Harmonic Generation

377

generated high-order harmonics was measured by grazing incidence spectrometer with a gold-coated Hitachi 1200 grooves/mm flat-field grating. A gold-coated grazing incidence cylindrical mirror was used to image the target harmonics at the ablation plume onto detector plane. The XUV spectrum was detected using a micro-channel plate (MCP) with a phosphor screen read-out (Hamamatsu, model F2813-22P), and the optical output from the phosphor screen was recorded using a CCD camera (Hamamatsu model C4880). The detail of an absolute calibration of the spectrometer was described in elsewhere6.

3 Results and discussions Figure 1 shows the spectra of HHG from tin and vanadium laser-ablated plume pumped by femtosecond laser pulse. The plateau and cut-off of the harmonics have observed in these experiments. The second ionization potential of tin and vanadium are 14.63 eV and 14.65 eV, respectively. Therefore, the cut-off order for both targets should be the same, since the cut-off order has been shown to depend on the ionization potential. Though the signal is weak, high-order harmonics up to the 23rd (wavelength: 34.56 nm) order was observed in these experiments with both tin and vanadium. However, a strong 17th order harmonic at a wavelength of 46.76 nm was observed using tin laser ablation plume, as shown in Fig. 1(a). The intensity of the 17th harmonics was 20 times higher than those of its neighbors. The conversion efficiency of the 17th harmonic was measured to be about 1.1x10-4, and this output energy of 1.1 μJ was obtained from the pump laser energy of 10 mJ. For vanadium ablation plume, the strong single HHG was not observed, as can be seen in Fig. 1(b). To confirm the nature of this strong emission at the wavelength of 46.76 nm, we investigated the effect of pump laser polarization on HHG. Figure 2 shows the intensity of the 46.76 nm line as a function of the laser polarization ellipticity. A quarter-wave plate was installed after the focusing lens to change the pump laser polarization from the linear to circular. The line intensities shown in Fig. 2 are normalized to that generated using linearly polarized pump laser, corresponding to the zero position. The intensity of emission at the wavelength of 46.76 nm decreased after 10-degree rotation of the quarter-wave plate, which completely disappeared after 50-degree rotation of the quarter-wave plate. This tendency is consistent with that of HHG, which leads us to conclude that the strong emission at 46.76 nm should be generated by HHG.

M. Suzuki et al.

Intensity (a.u.)

500

17th

(a) Sn

Intensity (a. u. )

378

1000

19th 21 200 30

15th

st

40

50

13th 60

11th 70

(b) V 400 x1000 300

200 30

80

40

50

60

70

80

Wavelength (nm)

Wavelength (nm)

17th order harmonic intensity (a. u.)

Fig. 1. HHG spectra from (a) tin and (b) vanadium laser ablation irradiated by femetoseconf laser pulse. The ordinate of spectrum (b) is 1000 times increased. The 17th order harmonic at wavelength of 46.76 nm was obtained using tin laser ablation.

1.0

Linear polarization

0.8 Circular polarization

0.6 0.4 0.2 0.0 0

20

40

60

Angle (degree)

Fig. 2.. The 17th order harmonic at the wavelength of 46.76 nm as a function of the quarter wave-plate rotation. The position of linear polarization is zero. By changing the rotation, the laser polarization changes the linear to circular.

To investigate the mechanism of enhancement for the 17th harmonic, the central wavelength of the pump laser pulse was changed from 795 nm to 778 nm. Figure 3 shows the HHG spectra from the laser-ablated tin laser with three different laser wavelengths of 795 nm, 782nm, and 775 nm. In Fig. 3(a), one sees that the intensity of the 17th harmonic using 795 nm wavelength pump dominates the harmonic spectrum. The intensity of the 17th harmonic is 20 times higher than that of other harmonics. However, in Fig 3(b), the intensity of the 17th harmonic using 782 nm wavelength pump is decreased, and is almost same as that of other harmonics. In Fig. 3(c),

Anomalous Enhancement of Single High-Order Harmonic Generation

379

the intensity of the 17th harmonic with 775 nm wavelength pump is further decreased. In the case, the 17th harmonic intensity is weaker than that of the 13th and 11th harmonics. The above results show that the intensity of the 17th harmonic gradually decreased as the wavelength of the pump laser become shorter. In the past work, the strong Sn II ion has been shown to posses a strong transition of the 4d105s 25p2P3/2-4d95s25p2 (1D) 2D5/2 at the wavelength of 47.256 nm.7 The gf-value of this transition has been calculated to be 1.52 and this value is 5 times larger than other transition from ground state of Sn II. Therefore, the enhancement of the 17th harmonic with 795 nm wavelength laser pulse can be explained be due to resonance with this transition driven by AC-Stark shift. By changing the pumping laser wavelength from 795 nm to 778 nm, the wavelength of the 17th harmonic is changed from 46.76 nm to 45.76 nm. Therefore, the wavelength of 17th harmonic pumped by laser wavelength of 775 nm is farther away from the 4d105s 25p2P3/2-4d95s25p2 (1D) 2D5/2 transition, at the wavelength of 47.256 nm. As a result, the resonance condition of the 17th order harmonic is weaker when pumped by a 778 nm, compared with the case for 795 nm pump.

7000 (a) 795 nm 17H

Intensity (a.u.)

6000 5000

19H

15H 13H

11H

4000 (b) 782 nm X6

3000 2000 (c) 778 nm

X6

1000 0 30

40

50

60

70

80

Wavelength (nm)

Fig. 3. HHG spectra from tin laser ablation for pump laser with central wavelength of (a) 795 nm, (b) 782 nm, and (c) 778 nm. The ordinate of spectra (b) and (c) are 6 times increased. The intensity of 17th order harmonic gradually decreased as the central wavelength of laser become shorter.

380

M. Suzuki et al.

(a) 795nm 25

th

23

rd

21

st

19

th

17

th

15

th

13

th

11

th

(a) 782nm

30

40

50

60

Wavelength (nm)

Fig. 4. HHG spectra from indium laser ablation for pump laser with central wavelength of (a) 795 nm, (b) 782 nm. The intensity of the 13th harmonic is two orders of magnitude higher than its neighboring harmonics.

Figure 4 shows the typical spectra of HHG from the laser ablation indium plume. For indium, the 4d105s2 1S0-4d95s25p (2D) 1P1 transition of In II, which have an absorption oscillator strength (gf–value) of 1.11,8 can be driven into resonance with the 13th order harmonic by AC-Shark shift. The intensity of 13th harmonic for indium is attributed to such resonance of the harmonic wavelength with that of a strong radiative transition. By changing the laser wavelength from 796 nm to 782 nm, the 15th harmonic at the wavelength of 52.13 nm increased, and the intensity of 13th harmonic decreased at the same time. The reason of 15th harmonic enhancement is due to resonance with the 4d105s5p 3P2-4d95s5p2 (2P) 3F3 transition of In II, which has a gf-value of 0.30. The enhancement of 15th order harmonic intensity is lower than that of 13th harmonic because the gf-value of 4d105s5p 3 P2-4d95s5p2 (2P) 3F3 transition is lower than that of the 4d105s2 1S04d95s25p (2D) 1P1 transition. Furthermore the central wavelength of the 13th harmonic was driven away from resonance with the 4d105s2 1S0-4d95s25p (2D) 1P1 transition when using 782 nm wavelength laser, thereby decreasing the 13th order harmonics.

4 Conclusions We have observed the strong resonance enhancement of the single highorder harmonic generation at the wavelengths of 46.76 nm and 61.26 nm using tin and indium laser ablation plumes irradiated by femtosecond laser pulse. Conversion efficiencies of these harmonics were about 10-4 in the

Anomalous Enhancement of Single High-Order Harmonic Generation

381

XUV region. We attribute the strong harmonic intensity to resonance with a strong oscillator strength transitions of Sn and In.

References 1. 2.

3. 4. 5. 6. 7. 8.

Takahashi, E., Nabekawa, Y., Midorikawa, K.: 'Generation of 10-μJ coherent extreme-ultraviolet light by use of high-order harmonics', Opt. Lett., 27, 1920-1922, 2002. Kim, I.J., Kim, C. M., Kim. H. T., Lee, G. H., Lee, Y. S., Park, J. Y., Cho, D. J.,Nam, C. H.:'Highly Efficient High-Harmonic Generation in an Orthogonally Polarized Two-Color Laser Field ' Phys. Rev. Lett. 94, 243901 2005. Gaarde, M B., Schafer. K. J.: 'Enhancement of many harmonics via a single multiphoton resonance ' Phys.Rev. A 64, 013820 2003. Toma, E. S., Antoine, Ph., Bohan, A. de., Huller, H. G.: 'Resonanceenhancement high-harmonic generation ' J. Phys. B 32, 5843-5852 1999. Ganeev. R. A., Suzuki. M., Baba, M., Kuroda, H., Ozaki, T.: 'Strong resonance enhancement of a single harmonic generated in the extreme ultraviolet range' Opt. Lett. 31, 1699-1701 2006. Ganeev, R. A., Baba, M., Suzuki, M., Kuroda: 'High-order harmonic generation from silver plasma', Phys. Lett. A, 339, 103-109, 2005. Duffy, G., Kampen, P. van.,Dunne, P: ' 4d-5p transitions in the extreme ultraviolet photoabsorption spectra of Sn II and Sn II' J. Phys. B 34, 3171-3178 2001. Duffy,G.,Dunne, P.: ' The photoabsorption spectrum of an indium laser produced plasma ' J. Phys. B34, L173-L178 2001.

Enhanced High Harmonic Generation from Ions Using a Capillary Discharge Plasma

M. Grishama, D. M. Gaudiosib, B. Reagana, T. Popmintchevb, M. Berrilla, O. Cohenb, B. C. Walkerc, M. M. Murnaneb, H. C. Kapteynb and J. J. Roccaa

NSF ERC for Extreme Ultraviolet Science and Technology a Electrical and Computer Engineering Department, Colorado State University, Fort Collins, CO b JILA and Physics Department, University of Colorado at Boulder c Department Physics and Astronomy, University of Delaware.

Summary. We demonstrate a significant extension of the high harmonic cutoff observed in xenon, up to 150 eV, by generating harmonics from ions in a capillary discharge plasma. The pre-ionized plasma generated by the capillary discharge dramatically reduces ionization-induced defocusing and energy loss of the driving laser due to ionization, allowing higher photon energies to be generated from xenon ions. We also demonstrate enhancement of the harmonic flux of nearly two orders of magnitude at photon energies around 90 eV when the capillary discharge is used to ionize xenon, compared with harmonic generation in a hollow waveguide. The use of a capillary discharge plasma as a medium for high harmonic generation shows great promise for extending efficient harmonic generation to much shorter wavelengths.

The extension of high harmonic generation (HHG) to higher photon energies from neutral atoms is not limited by the laser intensity, but is rather limited by the intensity at which all the neutral atoms are ionized. Ionization of the gas by the driving laser pulse leads to ionization-induced defocusing and beam attenuation, therefore limiting the peak intensity and cutoff photon energy. Generation from ions removes this limitation by reducing ionization-induced defocusing and beam attenuation and allows for extension of the cutoff harmonic [1]. A preformed plasma waveguide has been suggested as an advantageous media in which to generate efficient high harmonics [2,3]. Capillary discharges can conveniently produced elongated plasma waveguides suitable for guiding high intensity

384

M. Grisham et al.

light pulses [3,4]. In this work, we demonstrate a significant extension of the high harmonic cutoff observed in xenon using a capillary discharge plasma waveguide. We have extended the cutoff by 60 eV, and have enhanced the harmonic flux at 90 eV photon energy by an order of magnitude.

Fig. 1. Schematic of the experimental setup for generating high-order harmonics from a capillary discharge created plasma

The experiment (fig. 1) made use of a 10-Hz, two stage Ti:sapphire laser system capable of generating 12 mJ pulses with a 28 fs duration. The laser pulses were focused into the entrance of a 175 um diameter, 5 cm long capillary discharge plasma. The capillary was filled with 2-4 Torr of xenon and simmered with a small DC current of ~10 mA. A main current pulse with a peak amplitude of ~5A and duration of ~5 µs generates a plasma with a density profile that is minimum on-axis, constituting in an index waveguide. The discharge was operated at 10 Hz matching the repetition rate of the laser used in this experiment. The high harmonic spectrum was measured using a flat field EUV spectrometer with an X-ray CCD camera.

Enhanced High Harmonic Generation from Ions

385

HHG Signal (Arbitrary Units)

Thin aluminum or zirconium foil filters were used to block the fundamental laser light as well as calibrate the spectrum.

10

10

10

3

2

1

80

100

120

140

160

Photon Energy (eV)

Fig. 2. Harmonic spectrum observed through thin Al-coated Zr filters. The solid line shows harmonics when the discharge is on, while the dotted line corresponds to generation in a hollow core fiber with the discharge on.

A comparison of typical spectra obtained in the capillary dicharge and in a hollow-core fiber is shown in figure 2. In both situations 5 mJ, 28 fs laser pulses were focused into the waveguides filled with 2.8 Torr of xenon. The laser pulses were injected into the discharge 2 μs after the intiation of the current pulse. The use of the discharge clearly extends the highest observable harmonic from 90 eV to ~ 150 eV. Above photon energies of 75 eV the discharge enhances the HHG signal. Harmonics around 85 eV are enhanced by an order of magnitude. Both the flux enhancement and cutoff extension are due to a reduction in the losses in the waveguide by the discharge plasma. The reduction of laser induced-ionization by the discharge also results in spectra with more distinct harmonic peaks as shown in Fig. 3. This is due to a reduction of the severe SPM broadening of the driving laser resulting from the rapid ionization of neutral Xenon. Time-resolved visible spectroscopy of the discharge plasma shows that the plasma is completely ionized at the time delay that corresponds to the observation of the highest harmonics. At this time, the spectrum is dominated by Xe II and Xe III lines, with Xe I lines (dominant before the onset

386

M. Grisham et al.

of the discharge pulse) practically unobservable. The temporal evolution of typical lines of each species is shown in Fig. 4. These spectra confirm 4

HHG Signal (Arbitrary Units)

10

3

10

Aluminum L-edge 2

10

45

50

55

60

65

70

75

Photon Energy (eV) Fig. 3. Harmonic spectra obtained with (solid line) and without (dotted line) the discharge. The harmonics peaks are more distinctly resolved when the discharge is used.

the results of hydrodynamic / atomic physics simulations of the capillary discharge plasma which predict that the plasma is fully ionized. This is also in agreement with ADK ionization rate calculations that show harmonics greater than 80 eV could only be generated at an intensity where the plasma is fully ionized. For a particular photon energy of 120 ev, figure 4 shows delays between the rise of the current pulse and injection of the laser as well as a complete current profile. Before the current pulse there is no signal at this photon energy. The flux then rises to a peak 2 μs after the initiation of the current pulse and then slowly returns to zero. This clearly demonstrates the advantages of HHG in a capillary discharge created plasma. Spectroscopy measurements of the plasma show that at the time of maximum high harmonic emission the plasma is completely ionized. It is possible that the delay of the maximum HHG signal is due to the presence of plasma nonuniformities caused by the fast rise time of the current pulse.

This page intentionally blank

Enhanced High Harmonic Generation from Ions

387

(a)

(b)

Fig. 4. (a) Time evolution of the harmonic signal at 120 eV in Xe (diamonds). The current pulse is shown for reference (solid). (b) time evolution of the visible emission lines from Xe I (823.16 nm) (dotted), Xe II (418.01 nm) (dashed), and Xe III (410.92 nm) (solid).

In conclusion, we have demonstrated that a capillary discharge plasma waveguide can be used to extend high harmonic generation to shorter wavelengths. We have obtained high harmonic generation from Xenon to a photon energy of 150 eV. This is an increase of more than 70 eV over the highest previously published cutoff observed in xenon. Preliminary experiments in krypton and argon show similar enhancement. The capillary discharge overcomes the limitations imposed by ionization-induced defocusing, allowing for much higher laser intensities and harmonic generation at shorter wavelengths. This technique combined with quasi-phase matching methods promises to achieve efficient high harmonic generation at high photon energies.

Acknowledgement Work supported by the U.S. Department of Energy, Chemical Sciences, Geosciences, Biosciences Division of the Office of Basic Energy Sciences and by the NSF ERC for Extreme Ultraviolet Science and Technology under NSF Award No. 0310717. BCW acknowledges support from the JILA Visiting Fellows Program.

388

M. Grisham et al.

References 1. EA Gibson et al., “High-Order Harmonic Generation up to 250 eV from Highly Ionized Argon”, Phys. Rev. Lett. 92, 033001 (2004) 2. H. M. Milchberg, C. G. Durfee III, and T. J. McIlrath, “High-Order Frequency Conversion in the Plasma Waveguide”, Phys. Rev. Lett. 75, 2494–2497 (1995) 3. A Butler, DJ Spence, and SM Hooker, “Guiding of High-Intensity Laser Pulses with a Hydrogen-Filled Capillary Discharge Waveguide”, Phys. Rev. Lett. 89, 185003 (2002) 4. Y. Wang, et al., “Capillary discharge-driven metal vapor plasma waveguides”, Phys Rev. E 72, 026413 (2005)

Generation of High-Order Harmonic Continuum Supporting Single Attosecond Pulse in Argon Driven by Intense 7 fs Laser Pulse Y. H. Zheng 1, H. Xiong 1, Y. Peng 2, H. Xu 2, X. Yang 2, Z. N. Zeng 1, X. W. Chen 1, R. X. Li 1, H. P. Zeng 2 and Z. Z. Xu 1 1

State Key Laboratory of High Field Laser Physics, Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Shanghai 201800, China

2

Key Laboratory of Optical and Magnetic Resonance Spectroscopy of Ministry of Education, Department of Physics, East China Normal University, Shanghai 200062, China

Summary. High-order harmonic continuum in the cutoff is demonstrated with an argon gas cell driven by 0.4 mJ/7 fs (FWHM) ultrashort intense laser pulse. We find that the spectral structure, the modulation depth and the continuum bandwidth of the high-order harmonic spectra vary when the carrier-envelope phase (CEP) of driving laser pulse is stabilized at different values. At some CEP values, a continuous spectrum of 0.08 J/cm2, and the exothermic reaction of silicide formation starts. Main degradation mechanisms of MLs are discussed. The results of this study can be used for development of advanced multilayer mirrors capable handling the intense radiation conditions of new generation coherent X-ray sources.

1 Introduction Multilayer (ML) X-ray mirrors are widely used as optical components in the regions of UV and soft X-rays (1-50 nm) due to their high efficiency and flexibility in the parameters and forms. Under aggressive environment (heating, ion irradiation and exposure to high-power laser radiation) they can degrade as a result of the structural nonequilibrium conditions [1-4]. Growing power of new generation coherent X-ray sources (tabletop laser [5], FEL [6] and others) makes the stability of ML optical properties even more critical. The Sc/Si MLs developed for the wavelength range 35-50 nm have been successfully used in many applications with the capillary-discharge laser generating at the wavelength λ=46.9 nm [7]. Knowledge of radiationinduced degradation mechanisms in the Sc/Si structures will help finding high-end edges of the application and the ways to enhance their stability.

564

Yu. P. Pershyn et al.

On the other hand, the very laser radiation represents a specific interest for the surface processing [8]. Diffusion mixing of the layers and formation of the chemical compounds at moderate temperatures indicate the excessive free energy of the MLs. Moreover, the energy released in the mixing process can trigger self-sustained reaction observed in multilayer nanofoils [9]. In this work we discuss features of a single-shot laser influence onto Sc/Si ML degradation process related to the nonequilibrium of the layered structure.

2 Experimental Sc/Si MLs with the period of ~27 nm were deposited onto silicon and float glass substrates by the method of DC magnetron sputtering. Part of Sccontaining layer in the period was ~0.5. Number of the periods was 10 (Sc/Si/10) on glass substrate and 33 (Sc/Si/33) on Si. MLs deposited onto both type of substrates were exposed to the focused laser radiation operating at λ=46.9 nm. The energy of the pulsed (1.2 ns) laser beam was ~ 0.13 mJ. To vary the fluence value (F) we translated ML sample with respect to the beam focus. Each laser shot stroke the sample surface in a new ML region at normal incidence. Scanning electron microscope JSM-820 was used to get the information on surface morphology and chemical composition of Sc/Si MLs. Crosssectional images were produced with the help of transmission electron microscope PEM-U (SELMI, Ukraine) at accelerating voltage of 100 kV.

3 Results and discussions

3.1 Scanning electron microscopy study We investigated different laser imprints (LIs) obtained with the change of irradiated area on the Sc/Si ML surface, which corresponded to a set of the fluences in the range of 0.04-5.00 J/cm2. We observed traces of a ML melting at the fluence values starting from F~0.08 J/cm2. This fluence corresponds to the melting threshold predicted using the thermal diffusion model [10] for the MLs with average thermal characteristics being between Sc and Si.

Analysis of 46.9-nm Pulsed Laser Radiation Aftereffects

565

An incidental change in the morphology of the molten surface was observed at the fluences up to F~0.3 J/cm2 (fig. 1a) in the irradiated areas of Sc/Si/33 ML deposited on Si substrate. Note that the rectangular undamaged area in the center of LI is a shadow of the ML sample inherited from the irradiation scheme [11]. Further growth of the fluence resulted in the melt cracking structures. Cracks advent is an evidence of deep ML melting; better contrast SEM images obtained with the reflected electrons (penetration depth up to ~1 μm) have denoted this fact. First signs of evaporation were revealed at F~0.9 J/cm2 (fig. 1b) in the form of small micron-sized pits (visible in fig. 1b as black dots inside the LI center) indicating a melt boiling. Simultaneously a “crown” was formed as a result of pressing out the melt to the LI rim by excessive vapor pressure. The thermal diffusion model gives the lower threshold fluence for evaporation to be F~0.2 J/cm2. Perhaps, the evaporation process would start before the cracks appearing at the surface, but to define the threshold value more precisely one need to use more accurate and sensitive instruments (for example, probe beam deflection technique [12]). At F~2.2 J/cm2 the active evaporation or ablation of Sc/Si/33 ML becomes visible. The specific features of this stage are a crater formation in the LI center (fig. 1c) and appearing solidified drops around the LI. According to electron microanalysis there is no Sc within the crater of evaporated region, i.e. ML is completely absent in this region. The crater as a rule occupies less than a half the LI area. Presence of the drops means the ML being removed from the center in liquid phase as well. So, the mechanism of a combined ablation is characteristic for the Sc/Si ML: an ejection of the melt and vaporization; it is similar to that for metals [13].

0.3 J/cm2

0.9 J/cm2

10 μm

10 μm

a

b

2.2 J/cm2

10 μm

c

Fig. 1. SEM images of Sc/Si/33 ML on Si substrate irradiated with 46.9-nm laser pulses of different fluences.

Despite the fact that ~97% of absorbed energy is concentrated in top 10 periods, melting and ablation in Sc/Si/10 ML deposited on the glass goes

566

Yu. P. Pershyn et al.

in completely different way as compared to Sc/Si/33 ML. Melting as a rule is accompanied by cracking and flaking the ML practically up to F~0.5 J/cm2 (fig. 2a). An ablation processes in some LI areas begins at F~0.6 J/cm2 (fig. 2b) that is considerably less than that for Sc/Si/33 (2.2 J/cm2). Fig. 2c shows SEM picture for LI at 1.4 J/cm2 with ML removed from the most LIs. Judging from the presence of the drops the ablation mechanism here didn’t change. The rest of laser energy (~3%) is concentrated within the substrate surface because of high absorption [14] and low heat conductivity of the glass [15] compared to Si and Sc [16]. However this energy is insufficient to reduce the observed threshold of the ablation for Sc/Si/10 ML at least to one-third. Taking into account the strong absorption of practically all incident energy in the ML material we expected the similar behavior of MLs with 10 and 33 periods under the irradiation. We believe the difference of heat conductivity in glass and the ML material is the reason of observed distinction.

0.3-0.5 J/cm2

0.6 J/cm2

10 μm

10 μm

a

b

1.4 J/cm2

10 μm

c

Fig. 2. SEM images of Sc/Si/10 ML on float glass substrate irradiated with 46.9nm laser pulses of different fluences.

According to our calculation, at F~0.4 J/cm2 the laser radiation can ablate the Sc/Si/10 ML material completely. However, it looks like it doesn’t take place (see fig. 2a). Our estimations show that even at F~0.6 J/cm2 (fig. 2b) the bottom two periods cannot be melted by the absorbed laser energy directly. Complete ablation can occur at F≥1.4 J/cm2, with all the layers being melted under the laser irradiation. 3.2 Transmission electron microscopy study Fig. 2 shows cross-sectional image of the Sc/Si/33 ML after irradiation by the laser beam at F~0.13 J/cm2 (laser source is on the left). The most of

This page intentionally blank

Analysis of 46.9-nm Pulsed Laser Radiation Aftereffects

567

ML has been molten, and according to electron diffraction analysis the alloy composition corresponds to Sc3Si5 silicide. Only 7 periods of 33 survived at the substrate (on the right). This ratio of molten and survived volumes indicates the fluence approaching the full structure melting threshold for the Sc/Si/33 ML. The estimations, however, show that only 4 periods can be molten at that fluency, i.e. only one-sixth of the value observed in the experiment. Such a discrepancy can be explained by the suggestion that the formation of the Sc3Si5 silicide results in a release of up to 570 kJ/mol [17]. This energy is enough to heat and melt the material. For the short laser pulse duration (~1.2 ns) no efficient diffusion process can occur, therefore that melting is believed to be the main mechanism of ML degradation.

F~0.13 J/cm2

Sc3Si5

ML

Si

Fig. 3. Cross-sectional TEM image of Sc/Si/33 ML after irradiation at fluence of 0.13 J/cm2. Laser beam falls from the left. Substrate (S) is on the right.

So, we see that melting of the Sc/Si ML by the laser beam initiates the exothermic reaction, which, from one hand, facilitates the ML ablation (for F≥1.4 J/cm2) and, from the other hand, enlarges the molten region at low fluences (F10-4.Thus, it is very attractive to use relativistic harmonics for the attosecond pulse generation. The theory of the attosecond pulse generation [3,27] in the framework of the sliding mirror model

Intense Femto-Second Laser-Driven X-Ray Source

601

[24] has been developed. As was demonstrated by the PIC simulations [23,24], when a few-cycle relativistic-irradiance laser pulse interacts with a thin foil target as shown in Fig. 6, the ions remain at rest, while the electrons move as a whole. We consider the case when the electron motion in the direction perpendicular to the foil is negligible, so they move ("slide") mainly along the foil. Perpendicular electron motion can be neglected in two cases. First, when it is suppressed by the strong electric charge separation field Ech = 2πenl, which is of the order of the laser field amplitude E0 or more in this case; here e is the electron charge, n and l are the foil density and thickness. Second, when the thin foil target is irradiated by two identical laser pulses from both sides. We consider a p-polarized incident pulse obliquely incident on the thin foil. The problem was solved using the plane wave approximation in the boosted reference frame moving along the foil with the velocity V = c sinθ, where c is the velocity of light and θ is the incidence angle; in this reference frame, the incidence is normal [26]. Under the above assumptions, calculation has been performed. Both reflected and transmitted pulses contain harmonics of the incident laser radiation as shown in Fig. 7; furthermore, the spectral phase varies little at high harmonic orders. After spectral filtering, these harmonics form isolated attosecond pulses, like in the case of harmonics generated in gases for few-cycle driver pulses. Using spectral filtering, it is possible to obtain extreme ultraviolet pulses with the duration of few hundred attoseconds and the conversion efficiency of 10-8 – 10-5 [3, 27]. In order to obtain high conversion efficiency, we should not use spectral filtering technique. We found new regime of attosecond pulse generation in transmission, where the conversion efficiency can be as high as several percent. Example of the transmitted attosecond pulse is shown in Fig. 9 (a). Driver laser pulse has the duration of 6.3 fs and the dimensionless amplitude a0 = eE0/(mcω0) = 16 (I0 = 5.5×1020 W/cm2 for λ0 = 800 nm). Normalized aerial foil density is ε0 = 13.6, which corresponds, for example, to a fully-ionized 10-nm carbon foil. Duration of the transmitted attosecond pulse is τFWHM = 0.44/ω0 (190 as for λ0 = 800 nm), the energy conversion efficiency into the main pulse is 0.033. For the 5-µm focal spot this corresponds to 12 mJ, 60 TW attosecond pulse, which is at least six orders of magnitude larger than can be generated in gases. We take advantage of the analytic theory to establish the optimum parameters for the attosecond pulse generation in this regime. We obtain that the dimensionless pulse amplitude should be approximately equal to the normalized aerial foil density: a0 ≈ ε0, which can be cast into the form I0 ≈ 1.1×1019 W/cm2× (n/1024cm-3)2(l/nm)2. If a0 ε0, the pulse is transmitted through the foil almost without change; thus, only small part of the laser energy is transformed to higher frequencies, so the interaction is not efficient, and the transmitted pulse remains femtosecond. Dependencies of attosecond pulse duration and conversion efficiency on a0 under the optimum conditions are shown in Fig. 8 (b). The established relation a0 ≈ ε0 is rather general. It represents the condition of strongest coupling between a laser and a thin foil, so it holds not only for the attosecond pulse generation, but also for harmonic generation [3], ion acceleration [28-30] and so on. Experimental implementation of the sliding mirror technique requires clean femtosecond pulses with extremely low ASE(amplified spontaneous emission) level so the thin foil is not destroyed before the arrival of the main pulse. We will employ techniques to dramatically reduce the ASE level.

n E r2 oscillations

Et2

0

, E0

l

Fig. 6. Schematic view of the laser driven sliding mirror interaction geometry.

Intense Femto-Second Laser-Driven X-Ray Source S 7 10

Φ, rad 6

7

10

S

S 104

S

1

10

4

10

4 10

ω/ω0

Φ

1

10

603

2 0

-2

10

0

20

40

ω/ω0

Fig. 7. Spectrum of reflected pulse and the spectral phase after the subtraction of a linear term 1.76ω; squares on the phase curve correspond to peaks on the spectrum curve. Inset shows the same spectrum in the log-log scale. a0 = 15 (I0 = 4.8×1020 W/cm2), τ0 = 6.3 fs, ε0 = 13.5. 2

Et

(a)

E0

2

Et

100 25

50 E0

0

0 -10 -2.5 -5 ω0τFWHM

0.17

η

(b)

0.6

τFWHM, fs (λ0 = 800 nm)

0.25

0 10 ω0t t, fs ( λ = 800 nm) 0 0

η

0.04 0.03

τFWHM

0.4 3

4 5 6 7 8 910

0.02 0.01 a0 20

Fig. 8. (a): Transmitted attosecond pulse generation without spectral filtering. Shown is the intensity itself, not the intensity envelope. a0 = 16 (I0 = 5.5×1020 W/cm2), τ0 = 6.3 fs, ε0 = 13.6. Attosecond pulse: τFWHM = 190 as, the energy conversion efficiency into the main pulse is 0.033. (b): Dependences of the main attosecond pulse duration and energy conversion efficiency into it on a0 under the optimum conditions.

604

H. Daido et al.

6 Summary The high intensity femtosecond laser driven quantum beams such as x-ray, electron, proton, THz radiation and their combinations for fruitful applications such as various pump-probe technique are described. A well defined thin plasma layer in a thin foil target created by a clean femtosecond pulse opens collective relativistic phenomena which may bring a lot of fruitful applications.

References [1] M. Borghesi et al., Fusion Science and Technol. 49 April (2006). [2] K. W. Ledingham et al., Phys. Rev. Lett.84, 899(2000). [3] A. Pirozhkov et al., Phys. Plasmas 13, 13107(2006) [4] Hamster et al., Phys. Rev. Lett.71, 2725(1993). [5] T. Tajima and J. M. Dawson, Phys. Rev. Lett.43, 267(1979). [6] L. Willingale et al., Phys. Rev. Lett.96, 245002(2006). [7] T. Detmire et al., Phys. Rev. Lett. 78, 3121(1997). [8] M. Mori et al., Laser Physics 16, 1092(2006). [9] T. Fujii et al., Appl. Phys. Lett. 83, 1524(2003). [10] D-K Ko and J. Lee, Abstracts of ICUIL 2006 Ref:ICUIL067-Session1 “Recent progress on APRI PW Project and related high field physics researches”, pp. 49-50. [11] H. Shwoerer et al., Phys. Rev. Lett. 86, 2317(2001). [12] A. Rousse et al., Rev. Mod. Phys. 73, 17(2001). [13] C. Reich et al., Phys. Rev. Lett. 84, 4846 (2000). [14] K. Ta Phuoc et al., Phys. Rev. Lett. 91, 195001(2003). [15] A. Rousse et al., Phys. Rev. Lett.93, 135005(2004). [16] A. Sagisaka et al., Appl. Phys. B84, 415(2006) [17] M. Nishiuchi et al., Phys. Lett. A357, 339(2006). [18] Borghesi et al., Phys. Rev. Lett. 92,55003(2004). [19] M. Nishiuchi et al., Repetitive highly collimated intense proton beam with sub-MeV energy range driven by a compact few tera-watt femto-second laser, submitted for publication. [20] S. Nashima et al., IEEE Proc. 2004 Conf. Optelectronics and Microelectronics materials and Devices(COMMAD 2004), Brisbane, Australia, 8-10 December 2004, (2004) p.303 [21] M. Mori et al., Phys. Lett. A356, 146(2006) [22] Kando, Proc. APRC symposium 2006 ; H. Schwoerer et al., Phys. Rev. Lett. 96, 014802 (2006). [23] S. V. Bulanov et al., Phys. Plasmas 1, 745 (1994). [24] V. A. Vshivkov et al., Phys. Plasmas 5, 2727 (1998).

Intense Femto-Second Laser-Driven X-Ray Source

605

[25] M. Zepf et al., Phys. Rev. E 58, R5253 (1998); A. Tarasevitch et al., Phys. Rev. A 62, 023816 (2000); I. Watts et al., Phys. Rev. Lett. 88, 155001 (2002); I. Watts et al., Phys. Rev. E 66, 036409 (2002); U. Teubner et al., Phys. Rev. A 67, 013816 (2003). [26] U. Teubner et al., Phys. Rev. Lett. 92, 185001 (2004). [27] A. Pirozhkov et al., Phys. Lett. A.349, 256(2006) [28] K. Matsukado et al., Phys. Rev. Lett. 91, 215001 (2003). [29] T. Esirkepov et al. Phys. Rev. Lett. 89,175003(2002) [30] T. Esirkepov et al. Phys. Rev. Lett. 92, 175003(2004)

Radiative Properties and Hydrodynamics of LaserProduced Tin Plasma for Efficient Extreme Ultraviolet Light Source S. Fujioka1, H. Nishimura1, K. Nishihara1, Y. Tao1, T. Aota1, T. Ando1, K. Nagai1, T. Norimatsu1, N. Miyanaga1, Y. Izawa1, K. Mima1, H. Tanuma2, H. Ohnishi2, A. Sunahara3, Y. Shimada3 and A. Sasaki4 1) Institute of Laser Engineering, Osaka University, 2) Department of Physics Tokyo Metropolitan University, 3) Institute for Laser Technology, 4) Japan Atomic Energy Agency.

Summary. A laser-produced tin (Sn) plasma is an attractive extreme-ultraviolet (EUV) light source for the next generation lithography in terms of its brightness and compactness. Radiative properties and hydrodynamics of the laser-produced Sn plasmas are quite important for investigating the optimum conditions for EUV generation. Several experiments have been performed to clarify the above issues; (i) EUV spectra emitted from isolated Snq+ (5 < q < 15) ions and opacity spectrum of a 30-eV Sn plasma have been measured as fundamental data for accurate modeling radiation energy transport in plasmas, (ii) electron density profile in 13.5 nm emission dominant region of a laser-produced Sn plasma was measured with a laser interferometer for understanding hydrodynamics of an EUV source plasma. (iii) benchmarking one-dimensional (1D) radiation-hydrodynamic simulation code with multiple laser beam irradiated spherical 1D Sn plasma. Based on the experimental results and calculations, it was found that optically thinner plasma emits 13.5 nm light more efficiently. Optical depth of Sn plasma is actively controlled with changing laser pulse duration and the use of low-density porous target.

1. Introduction Laser-produced high-Z plasma is a typical example of radiationhydrodynamics, in which the radiative properties of a plasma are tightly coupled with hydrodynamic motion via energy transport. A clear understanding of radiation-hydrodynamics is important for the fields of inertial fusion energy, astro- and planetary physics, and x-ray source applications.Extreme ultraviolet (EUV) light sources for microlithography are receiving much attention as an application of laser-produced high-Z plasma. EUV lithography (EUVL) is a promising technology for volume produc-

608

S. Fujioka et al.

tion of next-generation microprocessors whose node size is less than 40 nm [1]. A commercial EUVL system requires > 300 W of output EUV power into a solid angle of 2π sr. from a plasma source of 13.5 nm wavelength within a 2% bandwidth (BW). Laser produced tin (Sn) plasma has a highly intense emission peak at 13.5 nm of wavelength, thus much effort is devoted to the development of the Sn-based EUV light source [2].

2. Radiation properties of laser-produced Sn plasma

2.1 Absolute opacity of 30-eV Sn plasma The opacity, as well as the emissivity, of laser-produced Sn plasma is so high for 13.5-nm light that the light emitted from deep within the Sn plasma is absorbed strongly during propagation through surrounding plasma as it expands. To obtain high conversion efficiency (CE) from incident laser energy to output EUV energy, the plasma size should be controlled to attain an appropriate optical depth, i.e., the product of the mass absorption coefficient and area density of the plasma for 13.5-nm light. The opacity of Sn plasma in the dominant EUV emission region is a critical parameter for investigating the optimum conditions for EUV generation, however no reliable experimental data has been available. The electron temperature and ion density of the dominant EUV emission region are in the ranges from 20 to 80 eV and from 1017 to 1020 cm-3, respectively [3]. Opacity measurements of Sn plasma were performed on the Gekko-XII laser facility [4]. Figure1 (a) shows the experimental set up for opacity measurement. A radiation-confining gold cavity, called the "dog-bone (DB)" [5], was used to uniformly heat an opacity sample. The opacity sample consisted of a thin Sn layer sandwiched between two 100-nm-thick CH tampers mounted on an observation window (200 μm x 200 μm) of the DB. The CH tamper was used to minimize the density gradient of the Sn plasma. The area density of the Sn layer in the sample was 2.04 +/- 0.18 x 10-5 g/cm2, which was analyzed with an inductively coupled plasma (ICP) device. Six beams of the Gekko-XII laser [6] (1.053 μm wavelength, 500 ps pulse duration) were focused through two inlet holes of 500 μm diameter onto the inner surface of x-ray generation sections set at both end of the DB. X-ray radiation from the sections propagated diffusively toward the central cavity for thermal radiation confinement.

Radiative Properties and Hydrodynamics of Laser-Produced Tin Plasma

609

The radiatively heated sample was backlit with broadband EUV light from another Sn plasma generated with a laser pulse of 1.053 mm wavelength and 500 ps duration at 1 x 1013 W/cm2. The backlight source was set 2 mm away from the DB and backlight x-rays were shone at 1.0 ns after the peak of the heating x-ray. The transmitted EUV spectrum through the sample was measured by a grazing incidence spectrograph (GIS) coupled with an x-ray streak camera (XSC). The Sn sample was heated by a TR = 50 eV thermal radiation pulse with a Gaussian shape (500-ps FWHM). Radiation hydrodynamic simulation (ILESTA-1D) [7] predicts an averaged temperature and density of the heated Sn sample of 30 eV and 0.01 g/cm3 at 1 ns after the peak of the heating x-ray pulse. Since the spectrum inevitably includes self-emission from the Sn and CH plasmas, we separately measured the self-emission spectra from the sandwiched Sn sample and from the CH tamper alone.

Fig.1 (a) Schematic of opacity measurement, (b) transmission of 30 eV Sn plasma for EUV light.

The dots in Fig. 1 (b) represent the raw measured spectrum of the transmission, while the solid line in Fig. 1 (b) is the smoothed measured spectrum in consideration of the spectral resolution of the GIS-XSC. The dash-dotted, dashed, and dotted lines represent the spectra calculated by an atomic code HULLAC [8] with electron temperatures of 20.9, 31.0, and 40.3 eV, respectively. The configuration interaction between the 4dn, 4dn1 4f, 4dn-15p, 4dn-15f, and 4p54dn configurations were taken into account, since the configuration interaction changes the wavelength and strength of emission lines considerably [9]. The population of the ionization state was calculated under the collisional radiative equilibrium condition. Strong absorption is seen around 13.5 nm, arising mainly due to the 4p-4d and 4d-4f transitions of Sn8+ to Sn13+

610

S. Fujioka et al.

The transmission at 13.5 nm, the most interesting wavelength, shows quite good agreement between the measurement and calculation, demonstrating that the HULLAC calculation is useful for calculating 13.5 nm light generation. The measured mass absorption coefficient of the Sn plasma at 13.5 nm is 0.96 +/- 0.18 x 105 cm2/g. 2.2 Identification of transition energies of multiply charged ions For accurate modeling of atomic process in plasma, spectroscopic data of Sn ions is necessary. However, the available spectroscopic information on multiply charged Sn ions is fairly limited at present. Even though much spectroscopic data has been reported, the energy levels of multiply charged Sn ions have not been established yet, because of the complexity of their electric structures attributed to the strong electron correlation among the large number of active electron. Photon-emission spectroscopy, in which the line intensities of the photon emissions are measured following the charge-transfer reaction in collisions of multiply charged ions with neutral target gases, is a very powerful experimental technique to investigate the transition energies and energy levels of multiply charged ions [10]. The multiply charged Sn ions were produced in a 14.25 GHz ECR (electron cyclotron resonance) ion source at Tokyo Metropolitan University. The Snq+ (5 < q < 15) ions were extracted with an electric potential of 20 kV and select by a double-focusing dipole magnet according to their massto-charge ratio. The ion beam was directed into a collision chamber, where the ion beam interacts a target gas jet ejected from a capillary plate. The background pressure in the collision chamber was 6 x 10-6 Pa and the target gas pressure in the chamber was held at about 1 x 10-3 Pa during the measurement. The target gas pressure was low enough to maintain the single-collision conditions. The EUV emission from the collision center was observed at 90 deg. to the ion beam direction with a GIS spectrometer equipped with a toroidaltype conversing mirror. As a photon detector, a liquid nitrogen cooled CCD camera was attached to the EUV spectrometer and an emission spectrum in the wavelength range of 6 – 24 nm was accumulated simultaneously. Spectral resolution of the spectrometer is about 0.1 nm. Wavelength calibration was performed using the established EUV emission lines of oxygen ions. From comparison between the measured spectrum and calculation, photon wavelength calculated by the HULLAC code is systematically different by 0.5 nm from the measured one for 4d-4f transitions. Therefore calculated wavelength of 4d-4f transition is manually adjusted to match the

Radiative Properties and Hydrodynamics of Laser-Produced Tin Plasma

611

experimental results for coupling with radiation-hydrodynamic simulation. Furthermore, the measurement reveals transitions corresponding to the dip observed in the opacity spectrum, i.e. the characteristic absorption peaks at 11, 12, 16 and 18 nm are caused by 4d-5f of Sn9+, 4d-5f of Sn8+ 4d-5p of Sn10+ and 4d-5p of Sn9+, respectively.

3. Hydrodynamics of 13.5 nm emission dominant region in laser-produced Sn plasma

3.1 Dynamic imaging of 13.5 nm emission from laser-produced Sn plasmas EUV emission from a laser-produced Sn plasma was imaged on an XSC using a monochromatic EUV Schwarzschild microscope. The microscope employed two concentric spherical mirrors with Mo/Si multilayer coating, operating at 13.5 nm in 4% BW [11]. A 0.4 µm Zr filter over coated on a 0.5 µm CH foil was placed in front of the mirrors to block visible light and particles from the plasma. In the present experiment, spatial and temporal resolutions of the microscope were checked to be better than 15 m and 1.5 ns, respectively. The microscope was installed in the plane of laser incidence and parallel with the target. A 10 ns laser pulse at 1064 nm with energy up to 2 J was illuminated onto the target. The laser beam was focused by an F/4 lens onto the target surface at normal incidence. The focal spot size was measured with both an optical imaging system and a monochromatic EUV pinhole camera. The diameter defined by FWHM was 220 μm and the width defined at 10% of the peak intensity was 450 μm. Targets used in the experiment consists of a strip Sn foil with a width of 200 μm and a thickness of 15 μm. It was placed on a 1 μm-thick CH film. Since CH plasma has much less average mass, CH plasma expands faster than Sn plasma, and CH plasma acts as a tamper to prevent lateral expansion of Sn plasma. Figure 2 shows temporal evolution of the EUV emission profiles at laser intensity of 1 x 1011 W/cm2, which is appropriate for efficient EUV generation [12]. Zero time represents the peak of the laser pulse. The EUV emission region expands with time and the peak of the EUV emission moves away from the target surface. At the peak of the laser pulse, the peak of the EUV emission locates 200 μm away from the target surface.

612

S. Fujioka et al.

Fig. 2 Temporal evolution of EUV emission observed from the side direction.

3.2 Density profile of 13.5-nm dominant emission region in laserproduced Sn plasma The electron density profile of laser-produced Sn plasma was investigated using two interferometers: a Mach-Zehnder interferometer using a probing beam of 266 nm UV light and a Michelson interferometer using 532 nm green light [13]. A small amount of laser energy, split from the heating beam, was used as a probe beam. The wavelength of the probe beam was converted to 532 and 266 nm. High temporal resolution was achieved by using a visible framing camera (HAMAMATSU C7400), which was synchronized with the laser pulse with a jitter less than 1 ns. The gate interval was set to 1 ns. The temporal resolution was less than 1.5 ns. The spatial resolution limited by the camera was 15 µm. For the both the green and UV interferometers, an interferometer data evaluation algorithm was used to calculate the Abel inversion to extract the density map from the phase shift map. Density profiles along the center of the plasma are plotted in Fig. 3. The circles an triangles represent the data from the green and UV interferometers, respectively. The error bars arise from the Abel inversion process, due to the irregular phase profiles. The solid line is an exponential decay fit to the experimental points with two typical scale length of 70 and 120 μm. Electron density of the EUV emission dominant region (EDR) is in range from 1019 to 1020 cm-3, those are much lower than the critical density for 1.064 μm light.

Radiative Properties and Hydrodynamics of Laser-Produced Tin Plasma

613

The measured density profile was compared with that predicted by onedimensional (1D) code (STAR-1D) [15]. It can be seen in Fig. 3 that there is a considerable discrepancy in their absolute values. A possible reason for this discrepancy arises from three dimensional plasma expansion. Because of the finite size of the focal spot, a lateral plasma expansion occurs as well as a longitude expansion. It is noted that the plasma scale length and the diameter of the focal spot are comparable, so multi-dimensional expansion becomes to have significant effects on plasma structure. As is expected, two-dimensional radiation hydrodynamic code reproduces the density reduction caused by the multi-dimensional expansion.

Fig. 3 Comparison of electron density profile between measurement and simulation. Dashed and solid lines are calculated by 1D and 2D code.

4. Benchmarking radiation-hydrodynamic simulation code with one-dimensional spherical plasmas Radiation-hydrodynamic simulation code is a powerful tool to investigate the optimum conditions for the EUV generation. Its reliability is proven only by experimental data, however a plasma produced from a planar target irradiated by single laser beam shows the multi-dimensional effects. The multi-dimensional plasma expansion make it difficult to compare the experimental results with 1D simulation. For eliminating these problems, spherical targets were irradiated uniformly with a multiple laser beam [14] on Gekko-XII laser facility [6]. Spherical plastic targets coated with a 1 μm-thick tin layer, were used in the experiment. The target diameter was varied from 300 to 750 μm. A 1.2 ns Gaussian pulse of 1.05 μm wavelength and 0.5 – 15 J in total energy

614

S. Fujioka et al.

was focused with an F/3 lens and the focal point was displaced based on the target diameter such that the target was illuminated uniformly. The laser intensity on the target was varied from 2 x 1010 to 1 x 1012 W/cm2 by adjusting the laser energy and the target diameter. The radiation-hydrodynamic code (STAR-1D) [15] describes radiation transport with a multi-group diffusion approximation, in which photon from 0.5 to 1500 eV is divided into 1500 groups. Electron thermal conduction is treated with a flux-limited Spitzer-Harm model, and laser absorption is calculated using a ray-tracing method. Tabulated emissivities and opacities calculated by the HULLAC code under the collisional radiative equilibrium condition are coupled with the radiation-hydrodynamic code.

Fig. 4 EUV spectra emitted from multiple laser beams irradiated spherical target. (a) experiment, (b) calculated by 1D code.

Figure 4 shows the comparison of spectra between experiment [14] and calculation at the laser intensity of 9 x 1011 W/cm2. The EUV spectrum was measured with a transmission-grating spectrometer coupled with a CCD camera. Its spectral resolution is better than 0.42 nm. The simulation spectrum is smoothed with taking into account of the instrument spectral resolution. The calculated spectrum shows quite good agreement with the experimental results. The agreement is observed in the wide range of the laser intensity from 1010 to 1012 W/cm2. The atomic process and energy transport included in the simulation code are proven to be appropriate for the EUV source modeling. The conversion efficiencies (CEs) from incident laser energy to output 13.5 nm 2 % BW energy were measured with a calibrated energy-meter called E-mon. Figure shows the CE dependence as a function of laser intensity. The highest CE of 3% was attained at an intensity of 0.5 – 1 x 1011 W/cm2. The calculated CEs are almost consistent with, but somehow higher than, the experimental values, because 20 % of intensity non-

Radiative Properties and Hydrodynamics of Laser-Produced Tin Plasma

615

uniformity still remains on a spherical target irradiated by finite (twelve) laser beams.

5. Active control of the laser produced Sn plasma for efficient EUV generation

5.1

Dependence of EUV-CEs on laser pulse duration

Optical depth of a produced Sn plasma is a key parameter to obtain efficient EUV light [4]. The optical depth can be controlled by changing the laser pulse duration. We investigated dependence of optical depth on the laser pulse duration. Pulse duration, which makes the optical depth unity, is calculated to be 3.7 ns at 1 x 1011 W/cm2 of 1.064 µm laser [16]. Experiments were carried out using a Q-switched Nd:YAG laser of 1.064 μm in wavelength. Pulse duration was changed from 2 to 9 ns using a pockels cell pulse slicer, and from 1 to 2 ns using a stimulated-Brillouinscattering pulse compressor. The laser was focused onto Sn plates with an F/30 lens from the target normal direction. Spot size was changed from 300 to 900 μm to obtain laser intensity in the range from 1 x 1010 to 1 x 1012 W/cm2. A 100-nm thick Sn layer coated on a plastic plate was used as a target to minimize influence of oxidation layer formed at the target surface on EUV conversion. Absolute 13.5 nm light energy was measured with the E-mon installed at 45 deg. with respect to the target normal. The spectral response of the E-MON, the measured spectral shape and measured angular distribution of the EUV emission were taken into account for the CE evaluation. Figure 5 shows dependence of EUV CEs on the incident laser intensities for various pulse durations. The optimum laser intensities are in the range from 5.0 x 1010 W/cm2 to 1.0 x 1011 W/cm2 for all the pulse durations. The maximum CE of 2.2 % was attained with the pulse duration of 2.3 ns at 5.0 x 1010 W/cm2. The CEs increase with shortening laser pulse duration for longer than 2.3 ns, however, the peak CE which obtained with 1.2 ns pulse is lower than that with 2.3 ns pulse. This result implies that the optimum pulse duration is determined not only by the optical depth but also by a fraction of laser energy absorbed in the EUV-EDR. The fraction of laser energy absorbed in the EDR were calculated to be, respectively, 25, 42, 73 and 85 % in 1.2, 2.3, 5.6 and 8.5 ns pulse produced plasmas. Therefore , for shorter pulse produced plasma, incident laser energy is not directly ab-

616

S. Fujioka et al.

sorbed in the EDR, but in higher dense region. Optimum laser pulse should have enough duration to absorb sufficient laser energy in the EDR.

Fig. 5 Dependence of EUV-CEs on laser pulse duration.

5.2

Low-density porous target for efficient EUV generation

The other method to change plasma properties is the use of low density porous target [17]. Experiments were carried out using a Q-switched Nd:YAG laser (1.064 m in wavelength and 10 ns in pulse duration) focused with an F/30 lens at the target normal. The focal spot size was optically measured to be 500 µm, and was unchanged within this experiment. Three kinds of planar targets were used. The first target is a solid-density Sn foil, whose initial density is 7.28 g/cm3. The second and third ones are low-density SnO2 foils, whose initial densities are respectively 23 % (1.62 g/cm3) and 7 % (0.49 g/cm3) of the solid SnO2 density. The low-density targets were fabricated using monodispersed polystyrene nanoparticles and liquid Sn chloride [18]. After calcination process of the chloride, the lowdensity SnO2 targets have porous structures whose cell size is about 1 µm. Dependence of the EUV-CEs on the laser intensity is shown in Fig. 6. In the case of the solid-density targets, the EUV-CEs depend weakly on the laser intensity, and almost constant CE of 1.2 % is obtained. On the other hand, the EUV-CEs from the low-density targets show relatively strong dependence on the laser intensity, this trend is similar to those for the shorter laser pulse (1.2 ns and 2.2 ns) produced -- optically thinner -- Sn plasmas. The highest CE of 2.2 % was obtained at the optimum intensity of 5 x 1010 W/cm2 for the 7% low-density SnO2 targets. The peak CE for

This page intentionally blank

Radiative Properties and Hydrodynamics of Laser-Produced Tin Plasma

617

the 7 % low-density SnO2 targets is 1.7 times higher than that for the soliddensity Sn targets.

Fig. 6 Dependence of EUV-CEs on initial target density.

Acknowledgement This work was performed under the auspices of a Leading Project promoted by MEXT (Japanese Ministry of Education, Culture, Sports, Science and Technology).

References [1] Silfvast W. T. et al., Appl. Opt. 32, 6895 (1993). [2] Jin F. et al., Appl. Opt. 34, 5750 1994); Spitzer R. C. et al., J. Appl. Phys. 79, 2251 (1996); Shimoura A. et al., Appl. Phys. Lett. 75, 2026 (1999); Choi I. W. et al., J. Opt. Soc. Am. B 17, 1616 (2000); Aota T. et al., Phys. Rev. Lett. 94, 015004 (2005). [3] Nishihara K. et al., in “EUV sources for Lithography” (SPIE, Bellingham, WA, 2006), Vol. PM149, Chap. 11, p. 339. [4] Fujioka S. et al., Phys. Rev. Lett. 95 235004 (2005). [5] Eidmann K. et al., Phys. Rev. E 52, 6703 (1995). [6] Yamanaka C. et al., IEEE J. Quantum Electron. 17, 1639 (1981). [7] Takabe H. et al., Phys. Fluids 31, 2884 (1988).

618

S. Fujioka et al.

[8] Bar-Shalom A. et al., Phys. Rev. E 56, R70 (1997). [9] Svendsen W. et al., Phys. Rev. A 50, 3710 (1994). [10] Tanuma H. et al., Nucl. Instrum. Meth. Phys. Res. B 235, 331 (2005). [11] Tao Y. et al., Rev. Sci. Instrum. 75, 5173 (2004). [12] Tao Y. et al., Appl. Phys. Lett. 87, 241502 (2005). [13] Tao Y. et al., Appl. Phys. Lett. 86, 201501 (2005). [14] Shimada Y. et al., Appl. Phys. Lett. 86, 051501 (2005). [15] Sunaharra A. et al., “Proceedings of the 3rd International EUVL Symposium”, Miyazaki, Japan, 2004. [16] Ando T. et al., to be published in Appl. Phys. Lett. [17] Okuno T. et al., Appl. Phys. Lett. 88, 161501 (2006). [18] Gu Q. et al., Chem. Mater. 17, 1115 (2005).

Laser Physics Research Relevant to Laser-Electron X-Ray Generator A.V. Vinogradov, M.V. Gorbunkov, Yu.Ya. Maslova and Yu.V. Shabalin P.N. Lebedev Physical Institute, Moscow, Russia Summary. A prototype of laser unit for Laser Electron X-Ray Generator is constructed on the basis of the optoelectronic control. The laser system in which an optoelectronic negative feedback is realized by means of a signal reflected from an intracavity Pockels cell polarizer is proposed and tested. The design provides flexible control over pulse train time structure.

1 Introduction Laser produced plasma proved to be a convenient and flexible source of incoherent and coherent soft X-rays probably including water window in the near future [1]. However the scaling of laboratory X-ray lasers and laser plasma X-ray sources to photon energies higher than 5 keV is questionable. Laser plasma based hard X-rays sources are still uncompetitive with conventional X-ray tubes, synchrotrons and free electron lasers. Meanwhile a dramatic (several orders of magnitude) gap exists between Xray tubes and accelerator based X-rays sources in respect of average power, brightness, sizes, cost, monochromaticity, tunability etc. The filling in of this gap would be beneficial for various scientific and commercial applications. In other words a new X-ray source that could bring together the compactness of X-ray tubes and X-ray beam manipulation ability of synchrotron radiation (SR) beamlines without the substantial rise of cost and loss of average power is highly desirable. The applications are biological and medical imaging, material structure and chemical analysis, protein crystallography, microscopy and microtomography for life sciences, medical diagnostics, industrial online X-ray inspection, security and custom control in ports and border terminals. A promising candidate to fill in the gap between conventional and SR sources is laser-electron X-ray generator (LEXG) based on Thomson scattering [2]. The idea to use Thomson scattering of laser radiation by relativistic electrons to extend the output energy spectrum down to hard X-ray of several tens keV range was n [3] and attracted last decade several research groups [4-12]

620

A. V. Vinogradov et al.

Pulse picker

. Fig. 1. The near analogue of the LEXG laser system [13].

A project of LEXG developed jointly by Moscow State University and proposed in Institute of Quantum Radiophysics of Lebedev Physical Institute is aimed at the system containing pulsed synchrotron and a repetitive picosecond laser [11, 12]. Electron bunch is circulating in a storage ring and appears in the interaction chamber (IC) with the period 10-20 ns. Original laser beam time structure is a millisecond train of pulses separated by microsecond-scale interval. Thanks to multiplication in optical circulator the period of laser signal is transformed from microsecond-scale interval into 10-20 ns inside the IC that provides the most efficient coupling of laser and electron beams. Since the microsecond interval is far above a master oscillator’s resonator round-trip (generally about 10 ns), one has to use pulse picking by means of an electrooptical modulator (see Fig. 1). A well-known method to obtain highly stable trains of picosecond pulses is to apply a system of feedback loops, which proved to be efficient in hundred microsecond range [14, 15]. Furthermore, feedbacks allow not only to stabilize the pulse amplitude, but also to obtain regular pulsations with controlled period far exceeding a resonator round trip time Tr [16]. In this case a laser radiation looks like microgroups of picosecond pulses separated by microsecond interval. It is important to note that such mode has an attractive advantage. In fact, after single pulses at 0.5 MHz are picked from a train of pulses at 100 MHz, only 0.5% of average power is utilized. The mode of regular pulsations with 2-microsecond (0.5 MHz) period is

Laser Physics Research Relevant to Laser-Electron X-Ray Generator

621

more favorable, since it provides increase in intensity for the pulses picked in pulsation peaks.

2 Two Feedback Loops-Controlled Laser To investigate the dynamics of a picosecond laser controlled by feedbacks we use the approach describing a mode-locked laser as an object with discontinuous control [16, 17]. In the simplest case of system with one negative feedback (NFB) loop delayed by a resonator round trip the control can be described by means of the so-called logistic mapping

xn +1 = rxn (1 − xn )

(1)

where xn stands for a normalized energy at the n-th pass (a pass corresponds to a laser cavity round trip), r is an overall gain including active medium gain and passive losses, and the term in brackets represents the one-pass delayed feedback loop action. The maximum acceptable gain for steady operation is rmax = 3 [16]. When r exceeds the threshold value rmax, the logistic mapping demonstrates a well-known nonlinear dynamics [18].

Fig. 2. Calculated log of pulsation period T(rmax) over relative feedback sensitivity: curves are in close agreement in the region of negative argument. The upper curve is an approximation.

Another dynamics can be observed in a system controlled by two feedback loops (delayed by one and two round trips) described by the equation

xn +1 = rxn (1 − α xn − xn −1 )

(2)

where α is relative feedback sensitivity, its negative value denotes that feedback loop delayed by a resonator round trip is positive (PFB). The

622

A. V. Vinogradov et al.

analysis shows that if −1 < α < 1 a nonlinear dynamics displayed by (2) differs fundamentally from that of the logistic mapping (1): regular pulsation with period above three round-trips appears. The pulsation period T (rmax) at r = rmax (α) calculated by means of the mapping stationary point stability analysis and the approach based on differential equations (approximation) is presented in Fig. 2. For negative α (i.e. positive and negative feedback combination control) at r = rmax (α) the pulsation is harmonic and the period can be expressed as

T ( rmax (α ) ) =

2π α +1

(3)

Formula (3) implies that regular dynamics with large periods (tens and hundreds of round-trips) can be observed at α close to - 1 (see Fig. 2).

3 Optoelectronic Feedback Designed for Pulsation Mode The above discussion showed that pulsations with period far exceeding a resonator round trip time are expected in a laser controlled by a combination of feedbacks where a negative feedback loop is delayed by one resonator round trip with regard to the positive one. With this aim in mind we designed a laser system in which an optoelectronic negative feedback is realized by means of a signal reflected from an intracavity Pockels cell polarizer. Self-mode-locking [19] in the negative feedback-controlled laser is a known technique for stable ultrashort pulses generation. The laser modelocked by a fast time-shifted negative optoelectronic feedback loop [20] is a very simple and reliable source of light pulses with about hundred picosecond duration. In such a laser the losses caused by an intracavity Pockels cell look like a periodic asymmetric "saw" with a long front and short tail. The tail is forming due to fast charge of the Pockels cell capacity by photocurrent generated in optoelectronic element under ultrashort light pulse. The long front is formed by slow discharge of intracavity cell capacity through the control system resistor. Stable self-mode-locking occurs with temporal delay in feedback control system corresponding to light pulse passage through the Pockels cell at the moment of low intracavity losses. Discharge time should be [20, 21] about several cavity round trip time Tr. Optimal delay for short pulse generation is approximately Tr. Usually the optical control signal is taken at the moment after passage of intracavity Pockels cell polarizer. The control proved to be efficient in laser output stabilization [16, 17].

Laser Physics Research Relevant to Laser-Electron X-Ray Generator

623

The laser dynamics changes dramatically if an optoelectronic feedback is realized by means of a signal reflected from an intracavity Pockels cell polarizer. The diagram of discontinuous control in the laser is shown in Fig. 3. When the laser pulse length varies slightly from pass to pass, the (n+1)-th pulse energy in the cavity En+1 is related to En as En +1 = En G 2 RPn , where G is a gain in a round-trip and R is an output mirror reflectivity (see Fig. 3). Similarly to [5], we use the Pockels cell transmission Pn = P0 (1 − Bn ) where Bn is control signal, and P0 is the initial Pockels cell transmission when feedback is off. In the proposed laser design Bn is the difference between incident light flux on the polarizer and transmitted flux. G2REn-1 - En NFB

Output mirror, R

→G2REn-1

→En

Polarizer, P

Gain medium, G

Fig. 3. The diagram of discontinuous control in a laser with NFB: the control signal is getting from Pockels cell polarizer.

As a result for the recurrence relation we have

(

En +1 = EnG 2 RP0 1 − ( En −1G 2 R − En ) Using r = G 2 RP0 and En = xn

)

(4)

P0 leads to r

P ⎞ ⎛ xn +1 = xn r ⎜1 − xn −1 + 0 xn ⎟ r ⎠ ⎝

(5)

Applying (3) and taking into account the Pockels cell transmission

⎛ U π⎞ P (U ) = cos 2 ⎜ ⎜ U λ 2 2 ⎟⎟ ⎝ ⎠

(6)

where U is a static bias voltage and U λ 2 is a cell half-wave voltage, we estimate the nonlinear dynamics development threshold gain rmax(U) and oscillation period T(U) in two limits.

624

A. V. Vinogradov et al.

When U is far less than U λ 2

T (U ) = 2

2π U λ 2 U

, rmax (U ) = 1 +

π2 ⎛ U ⎞

2

⎜ ⎟ 4 ⎜⎝ U λ 2 ⎟⎠

When U → U λ 2 , T (U ) → 2π , rmax (U ) → 2 . Thus, we conclude that in the laser system controlled by optoelectronic NFB with the control signal getting from the Pockels cell a large period regular pulsation can be obtained. The period increases and the nonlinear dynamics development threshold decreases when the Pockels cell bias voltage U decrease.

4 Laser Dynamics Simulation Results To complete our demonstration of the laser, where the optoelectronic control signal is taken from an intracavity Pockels cell polarizer, however, we still have to show that the regime of pulsations can be reproduced qualitatively by extending the mapping (4) to take into account the laser output radiation fine time structure evolution as well as Pockels cell voltage variation at the time scale of Tr depending on the feedback delay time, Pockels cell capacity discharge time, active medium gain, and the cell bias voltage Ust following the same approach as in our previous works [16,21] The goal is to investigate self mode locking in pulsation regime with corresponding significant control voltage and time-dependent Pockels cell transmission variation. The main question was about the possibility to obtain a single short pulse circulating in a laser cavity. In the numerical simulation spontaneous emission noise was placed in the laser cavity and subsequently transformed by time-dependent transmission P(t) of the Pockels cell. On the other hand P(t) is calculated from the feedback signal proportional to the laser intensity I(t) reflected from the intracavity polarizer. The simulation proved that the discussed laser system allows not only to obtain stable mode-locking in quasi CW mode, but also to generate short pulses microgroups with controlled period far exceeding a resonator round trip time. To show gain sensitivity of microgroup envelope, the trains at gain r = 1.03 and 1.09 are shown in Fig. 4 (in both cases Ust = 0.05 U λ 2 ).

Laser Physics Research Relevant to Laser-Electron X-Ray Generator

625

a

b

Fig. 4. Simulated trains: r = 1.03 (a) and 1.09 (b).

Experiment The experiments were performed using a flash-lamp pumped Nd:YAG laser (∅6.3×60 mm rod was used). A PC-controlled laser pumping based on the incomplete discharge of large capacity allowed us to vary pump duration

Fig. 5. Millisecond lamp pumped YAG-Nd laser designed for mcs scale pulsations. AM active laser medium; M1, M2 cavity mirrors; P polarizer; IA iris aperture; MT – mirror telescope; Pockels cell DKDP electrooptical crystal ; CC feedback control circuits.

626

A. V. Vinogradov et al.

up to 3.9 milliseconds. Laser cavity was made by flat wedge-shaped reflectors of 0.98 and 0.35 reflection coefficient respectively. We introduced an intracavity mirror telescope 3:1 to enlarge a laser mode volume in the active media and thus to raise the laser output. Total cavity length was 150 cm. Iris aperture was used for mode selection. Two-pass Pockels cell was based on a multilayer polarizer (Brewster angle) and 8×8×11 mm3 DKDP a

b

Fig. 6. Traces for a lasing power (a) and a pump lamp running (b), discharge time is 3,9 ms obtained with C8-14 scope; time scale 0.5 ms/div.

(Uλ/2= 3.9 kV) crystal with antireflection faces placed close to the laser mirror. Crystal was installed directly on the control circuit plate. Static bias voltage Ust applied to the Pockels cell was varied in a range 0÷2 kV. A high voltage silicon mesa-structure was used as a control element of an optoelectronic system. Discharge time of intracavity Pockels cell ca pacity was set to 20 ns (2Tr). The control signal was taken from an intracavity Pockels cell polarizer (Fig. 5). Pin-diode and C8-14 storage (low resolution) and digital TDS-3052 (5 GS/s, 500 MHz) oscilloscopes were used for laser emission registration. Control voltage vs time was registered by using of specially made HF voltage divider and fast oscilloscope C7-19 (5 GHz frequency bond). Fine time structure of the laser light was investigated using streak-camera AGAT SF-3M (time resolution < 2 ps) synchronized with laser pulses with specially-made electronic delay system. Pumping pulse shape is shown in Fig. 6, b. Setting an optimal diameter of intracavity iris aperture we obtained a picosecond pulse train of stable amplitude with total number of pulses up to 350000 (Fig. 6, а). By varying Ust and fine angular DKPD crystal tuning we obtained modes of regular pulsations having periods 0.5, 1, 1.3, and 2 microseconds. The experiments showed that despite of significant control

Laser Physics Research Relevant to Laser-Electron X-Ray Generator Fine time structure, 50 ns/div.

gain increase

Pulsations, 2 mcs/div.

627

Overview trace, 0.5 ms/div.

Fig. 7. The development of regular microsecond pulsations and a microgroup fine time structure in millisecond laser. Pulsation period is 200 round trips (2 mcs). The dynamics is shown in three time scales: millisecond, microsecond and nanosecond.

voltage and corresponding time-dependent Pockels cell transmission variation a single pulse is circulating in the laser cavity for all modes. An example of 2 microsecond period pulsation development is shown in Fig. 7. Millisecond oscilloscope traces (see Fig. 8) show that the peak pulsation power (feedback is on) is several times higher than a steady-state power level in free-running lasing (feedback is off, all other conditions being equal).

a

b

Fig. 8. Overview oscilloscope traces: a – feedback is switched off, b – feedback is switched on (time scale 0.5 ms/div). Vertical scales are equal.

628

A. V. Vinogradov et al.

We are grateful to N. A. Borisevich, V. G. Tunkin and V. A. Petukhov for fruitful discussions and to A. M. Chekmarev for help in work. The work was partially supported by the Program of Fundamental Research of RAS, subprogram “Laser systems” and RFBR grant 05-02-17448a.

References 1. R. Toth, J.C. Kieffer, A. Krol, et al.: 'Phase contrast micro-CT with an ultrafast laser-based hard x-ray source', Proc. SPIE, p. 591813-1 – 591813-8, 2005. 2. P. Sprangle, A. Ting, E. Esarey, and A. Fisher: 'Tunable, short pulse hard xrays from a compact laser synchrotron source', J. Appl. Phys. 72, p. 5032, 1992. 3. A. Luccio and A.B. Brik: 'Methods and Apparatus for Producing X-rays', US Patent 4,598,415 July 1 1986, European Patent 0,05032 24.08.1988. 4. Z. Huang and R.D. Ruth: 'Laser-Electron Storage Ring', Rhys. Rev. Letters, Vol. 80, No. 5, p. 976, 1998. 5. R.J. Loewen: SLAC-R-632, June 2003, Ph.D. thesis, Stanford University, Stanford CA. 6. A. Agafonov, V. Androsov, J.I.M. Botman et al.: 'Status of Kharkov x-ray generator NESTOR', Proc. SPIE, 5917, p. 97, 2005. 7. F.E. Carroll: 'Tunable Monochromatic X Rays: A New Paradigm in Medicine', AJR 179, p. 583, 2002. 8. W.J. Brown, S.G. Anderson, S.P.J. Barty et al.: 'Experimental characterization of an ultrafast Thomson scattering x-ray source with three-dimensional time and frequency-domain analysis', Physical Review Special Topics – Accelerators and Beams, 7, 060702, p. 1, 2004. 9. K. Dobashi, A. Fukasawa, M. Uesaka et al.: 'Design of Compact Monochromatic Tunable Hard X-Ray Source Based on X-band Linac', Japanese Journal of Applied Physics, Vol. 44, No.4A, p. 1999, 2005. 10. T. Yanagida, T. Nakajyo, S. Ito et al.: 'Development of high-brightness hard x-ray source by Laser-Compton scattering', Proc. SPIE, Vol. 5918, p. 231, 2005. 11. M.V. Gorbunkov, V.G. Tunkin, E.G. Bessonov et al.: 'Proposal of a Compact Repetitive Dichromatic X-ray Generator with Millisecond Duty Cycle for Medical Applications', Proc. SPIE, 5919, OU1-OU6, 2005. 12. I.A. Artyukov, E.G. Bessonov, A.V. Vinogradov et al.: 'Laser-electron X-ray Generator', Preprint INP MSU 2006-7/806. 13. S. Schreiber, I. Will, D. Sertore et al.: 'Running experience with the laser system for the RF gun based injector at the TESLA Test Facility linac', Nuclear Instruments and Methods in Physics Research A 445, 427-431, 2000. 14. M.V. Gorbunkov, Yu.Ya. Maslova, A.M. Chekmarev et al.: 'Pulse generation of a large number of picosecond pulses microtrains in Nd:YAG laser controlled by negative feedback loop', Proc. of MIPT, Moscow, p. 50, 2005.

Laser Physics Research Relevant to Laser-Electron X-Ray Generator

629

15. M.V. Gorbunkov, A.V. Konyashkin, P.V. Kostryukov et al. 'Pulsed-diodepumped, all-solid-state, electro-optically controlled picosecond Nd:YAG lasers', RUS Quantum Electronics 35 (1), p.2, 2005. 16. I.M. Bayanov, V.M. Gordienko, M.G. Zvereva, S.A. Magnitski, A.P. Tarasevich. A High-Stable Negative-Feedback Picosecond YAG:Nd3+ Laser. RUS Quantum Electronics, 16 (8), p. 1545, 1989. 17. M.V. Gorbunkov, Yu.V. Shabalin: 'Two-Loop Feedback Controlled Laser: New Possibilities For Ultrashort Pulses Generation And High-Level Stabilization', Proc. SPIE, Vol. 4751, p. 463, 2002. 18. H.G. Schuster: Deterministic Chaos. An Introduction, Physik-Verlag, Weinheim, 1984. 19. V.K. Makukha, V.M. Semibalamut, V.S. Smirnov: 'Simulated Emission of Ultrashort Pulses from a Negative Feedback Laser', Sov. Quantum Electronics 4, No.5, p.1023-1027, 1977. 20. D.B. Vorchik, M.V. Gorbunkov: 'Self Mode-Locked Nd-YAG Laser Under Fast Delayed Negative Feedback by Means of High-Voltage Assemblies of Inversely-Shifted Silicon p-n Junctions', Physical Foundations of Electronic Laser Engineering, Proceedings of MIPT, Moscow, p. 4 - 11, 1995. 21. M.V. Gorbunkov, Yu.V. Shabalin: 'Picosecond YAG:Nd3+ Laser with SelfExcitation of HF Oscillations in Optoelectronic Negative Feedback System', Bulletin of the LPI 8, p.38-50, 1998

Laser Electron Generator of the X-Ray Radiation I. A. Artyukov, E. G. Bessonov, A. V. Vinogradov, M. V. Gorbunkov, Yu. Ya. Maslova, N. L. Popov, A. A. Postnov, Yu. A. Uspenski, R. M. Feshchenko and Yu.V. Shabalin P.N. Lebedev Physical Institute of RAS 19991 53 Leninski Pr, Moscow, Russia

Yu. L. Slovokhotov and Ya.V. Zubavichus Institute of elemental organic compounds of RAS, 119991 28 Vavilov st, Moscow, Russia

B.S. Ishanov, A.V. Poseryaev and V.I. Shvedunov Institute of Nuclear Physics of MSU

P. V. Kostrukov and V. G. Tunkin International Laser Center of MSU

Summary. The possibility of the creation and the application prospects of the laser-electron X-ray generator based on the Thompson scattering of the laser radiation on a bunch of relativistic electrons are considered. Such a generator fills the existing gap between X-ray tubes and synchrotron sources, which is several orders of magnitude in terms of the brightness, average intensity, size and also in the construction and exploitation costs. The layout of beam-lines and experimental stations intended for the applications of the X-ray laser-electron generator to the investigation of the elemental composition and material structure and biological objects is discussed.

632

I. A. Artyukov et al.

1 Introduction The development of the technologies connected to the generation and utilization of the X-ray radiation is an important part of the scientific and technical progress in lots of fields of the industry including metallurgy, chemical and pharmaceutical industry and also in such areas of the human experience as medicine, whole agricultural and food trade, ecological monitoring, social security, work of the custom and boarder terminals etc. For the introduction of many scientific developments in areas the investigators need constant access to X-ray sources. The modern generators of X-radiation can be divided into two main classes. X-ray tubes (with a fixed or rotating anode) and boosters of electrons: synchrotrons and storage rings. The X-ray tubes are used in overwhelming majority of production devices and apparatuses. The tubes with a fixed anode are reliable enough, compact, and simple to service and rather inexpensive - from hundreds up to several thousand dollars. The modern tubes with a rotated anode give in 10-100 times higher emission power, however they cost tens and hundred thousand dollars and is much more complex in exploitation. Common faults of this class of X-ray sources are absence of directivity of X-ray radiation, broad and, at a given material of the anode practically not variable spectrum, rather a small emission power and impossibility of generating bright monochromatic xray radiation. As compared to X-ray tubes, synchrotron accelerators and the storage rings are large power-intensive research installations with a closed path of the electron beam of tens and hundreds meters length. All over the world there are about hundred accelerators intended for deriving of X-ray synchrotron radiation (SR). This radiation has high luminosity, directedness and wide spectrum, with possibility of obtaining tunable monochromatic radiation. However size and cost - tens and hundred millions USD – of the modern synchrotron sources seriously restrict their applications, which do not satisfy the needs of the science and practice. Thus, now there is objective need for a new source of X-rays, which would fill in the gap existing between X-ray tubes and synchrotron centers. An X-ray source adequate to the formulated requirements, can be built on the basis of complex systems, which combine a compact high-current electron accelerator and laser emitting intensive light pulses. X-rays are in this case generated at head on collision of electronic and laser bunches; in other words, the photons of high energies bear as a result of diversion of the electron beam from the linear trajectory in the field of an intensive light wave. The relevant elementary process is well investigated and wears a title Thomson or Compton scattering (depending on quantity of parameter

Laser Electron Generator of the X-Ray Radiation

633

EħωL/(mc2)2, determining the size of the quantum contributions, E - energy of an electron, ħωL - energy of the laser photon). From the end of 70th years of XX century the Compton scattering on bunches of relativistic electrons serves as an effective method of deriving γ - photons (down to energies ~ 2 GeV), used in photonuclear reactions. However in generation of photons of lower energies (~ 10 - 100 keV), which have the greatest interest for the applications, the X-ray tubes and synchrotron radiation until recently remained beyond the competition. Now new active solid-state mediums using pumping by laser diodes, diode bars and matrixes, allow to generate and enhance trains of picosecond light pulses in compact devices and with a high efficiency. On the other hand, the modern electron accelerators allow to generate bunches with high luminosity, which can be focused in a spot with the size about 10 microns, and the modern accelerating structures can supply rate of acceleration up to 50 MeV/m, that allows to build installations of the small sizes. The integration of lasers and accelerators in one device enables to create rather a cheap compact source of intensive X-rays for the scientific and applied purposes.

2 Thompson scattering by relativistic electrons The research on deriving gamma and X-rays by the scattering of laser pulses on relativistic electrons has been carried on for more than 40 years [1-7]. In this section we shall give an estimate of intensity and luminosity of the laser-electronic generator and compare it by the last parameter to Xray tubes and sources of SR. For the electrons with energies Ee = γmc2 ~ 25 – 50 MeV and laser photons ħωL ≤ 2 eV, which provide X-ray generation in the range 5-50 kev interesting for us, the inequality holds:

2γ hω L 99.9% Pockels cell 1

P 1 R>99.9% Pockels cell 2

T c = 10 ns

1 ms

40eV

Wp

N 0 g0

)3

N 2 g2

A21 N 1 g1 Wg

Fig. 2. Simplified 4 level scheme

37.2 nm -

e -collisions

5

2p 3s

2

3

P

3

P1 Na +

1

2p6 1 S0

0

2p 6 3s

Na

Novel Attempts to Realize an Innershell X-Ray Laser in Na

651

Rate equations dN 0 dt

= − N 0W p (t ) − N 0W g (t ) − N 0 N el C

dN el dt

= N 0W p (t ) + N 0Wg (t ) + N 0 N el C

dN1 dt

= N 0Wg (t ) + N 2 A21 + N 0 N el C

dN 2 dt

= 14 N 0W p (t ) − N 2 A21

dN 3 dt

= 34 N 0W p (t )

In these 5 equations, N0 is the density of sodium atoms in the ground state (2p63s), N1 and N2 are the population densities of the lower (2p6 1S0) and upper (2p53s 1P1) laser levels and N3 is the population density of the 2p53s 3 P level manifold, which however does not contribute to the laser process itself. Wp describes the total pump rate to the excited 2p53s states of the sodium ion and Wg the corresponding pump rate for the lower laser level (sodium ion ground state). According to the level degeneracies, only ¼ of the pump rate Wp is used for the upper laser level. The pump rates Wp,g are given by Wp,g = Ip/hνp* σp,g, where Ip is the intensity of the x-ray pump laser (hνp: photon energy) and σp.g are the photoionization cross sections for extraction of an inner p-electron, leading to the population of the upper laser level, and of an outer s-electron, leading to the population of the Na+ ground state. According to Fig. 1 and the data of Ref. 3, σp is about 8 Mb at the pump wavelength of 21.2 nm and σg is about a factor of 200 smaller. With the ionization of sodium atoms electrons are produced, which may then populate ionic levels by collisional excitation of sodium atoms. Of special importance thereby is the population of the lower laser level, leading to a reduction of the population inversion. This population of the lower laser level is described by the rate NelC, where Nel is the electron density. The collisional coupling rate C is estimated as 10-7 cm3/s (Ref. 7). From the equations, the inversion density ΔN = N2 – g2/g1*N1 can be calculated, where g2 and g1 are the degeneracy factors of the levels, with g2= 3 and g1 =1. With the inversion ΔN the gain coefficient g is then given as g = σstim*ΔN, with the stimulated emission cross section σstim. Assuming a Doppler broadened linewidth at a temperature of about 400 oC and a lifetime τ of the upper laser level of 400 ps (A21 = 2.5x109 s-1), σstim amounts to 5x10-14 cm-2. At a temperature of 400 oC the sodium density is about 1016 cm-3. For numerical calculations of the gain, the intensity, duration (pulse shape) and spatial change of the pump pulse (depending on the focussing) have to be chosen. Also the starting density N0 may change in pump beam

652

B. Wellegehausen et al.

direction and perpendicular to it, which depends on the used sodium oven. In the following, calculations of the inversion density and gain coefficient for specific scenarios (assuming constant intensity and density) will be presented and discussed. Assuming pulses with a sech2 pulse form and different peak intensities, a pulse duration (FWHM) of 100 ps and a starting sodium density of 1016 cm-3, the time development of the relative inversion density ΔN/N0 shown in Fig.3 is obtained. Population inversion is achieved in a certain time window, first at an intensity of about 109 Wcm-2 just around the maximum of the pulse. With increasing intensity this window shifts to earlier times and the inversion density finally saturates due to the depletion of the sodium density N0. The inversion density at the maximum intensity of 1013 Wcm-2 corresponds to a gain coefficient of about 28 cm-1.

Fig. 3. Time behaviour of the relative inversion density ΔN/N0 in units of τ (1/A21 = 400 ps)

To achieve a larger gain for the given density of 1016 cm-3, pulses with shorter pulse duration have to be used, as can be seen from Fig.4. For a pulse duration of 50 ps a gain coefficient of 65 cm-1 can already be obtained at a pump intensity of 5x1011 Wcm-2. To reduce the saturation at higher pump intensities, which is due to the depletion of the sodium density, the initial particle density has to be increased further, as can be seen from Fig. 5. For a given pump intensity and pulse duration an optimum density exists, which for the 50 ps pulse duration and intensity of 5x1011 Wcm-2 is around 2.5x1016 cm-3, leading to a gain coefficient of more than 100 cm-1

Novel Attempts to Realize an Innershell X-Ray Laser in Na

653

Fig. 4. Gain versus pump intensity for different pulse durations (sech2 pulse) at a sodium density of 1016cm-3

Fig. 5. Gain versus sodium density for different pulse durations (sech2 pulse) at an intensity of 5x1011Wcm-2

654

B. Wellegehausen et al.

3 Planned experiment In the planned experiment, the x-ray laser radiation will be focussed with an off axis paraboloid into the sodium vapour which is generated in an

Fig. 6: Time dependent gain (in unit of τ = 400 ps) for the measured Zn-x-ray laser pulses (pulses superimposed at maximum intensity)

oven. The vapour streams out of a nozzle head, which has 800 µm holes on both sides and a length which can be varied between 5-15 mm. The oven is located in a vacuum chamber, which can be attached to the target chamber of the x-ray laser. The x-ray laser delivers pulses with energies between 1 mJ - 10 mJ and pulse durations of 150ps (single pass) and 250 ps (double pass) (Fig.6), with the larger energies for the double pass geometry. Assuming a pulse energy at the sodium target of 0.25 mJ/1 mJ for the 150 ps/250 ps pulses and a focus radius of 50 μm, the 150 ps/250 ps pulses deliver peak intensities of 0.75x1011cm-2 and 2.2x1011Wcm-2. For a sodium density of 1016 cm-3 these pulses then yield maximum gain coefficients of 52 cm-1 and 62 cm-1 (Fig.6). The large gain is due to the steep raising of the pulses and comparable to the gain for a 75 ps sech2 pulse (Fig.4). For interaction lengths of about 5mm-10mm and assuming constant pump intensity, saturated amplification should be possible.

Novel Attempts to Realize an Innershell X-Ray Laser in Na

655

4 Conclusions The presented calculations yield high gain in a wide parameter range. Planned more detailed calculations will include changes of the pump intensity within the vapour due to the focussing and absorption, and also changes of the vapour density especially at the nozzle exit. These effects may slightly reduce the gain. But as the x-ray laser data are assumed conservative and further improvements expected, good chances for realization of the innershell x-ray laser in sodium exist.

References 1. 2. 3. 4. 5. 6. 7.

M.A. Duguay and P.M. Rentzipis; Appl. Phys. Lett. 10,350 (1967) W.T. Silfvast and O.R. Wood II; J. Opt. Soc. Am. B4, 609 (1987) J.J. Yeh and I. Lindau; Atomic Data and Nuclear Data Tables 32, 1 (1985) S. Meyer et al.; Inst. Phys. Conf. Ser. No 151, 173 (1996) S. Meyer et al.; Inst. Phys. Conf. Ser. No 159, 313 (1998) B. Rus et al.; Phys. Rev. A 66, 063806 – 12 (2002) W. Lotz; Z. Physik 216, 241 (1968)

310 Angstroms X-Ray Lasing Under Beta-Decay of Tritium M. Yu. Romanovsky1 and V. K. Bityukov2 1 2

A.M.Prokhorov General Physics Institute of RAS Moscow State Institute for Radiotechnics, Electronics, and Automatics

Summary. The beta-decay of atomic tritium produces the ion 3He+ at the first exited state 2s with the probability of 25%. This state can irradiate two photons (two-photon lifetime is about 0.01 ms) or one photon (the lifetime is 173 s). The stimulated emission of one-photon may depress the two-photon decay. This is a soft X-ray laser with the wavelength 310 angstroms (40.8 eV). The first problem of two-photon decay depression is solved by largereflectivity mirrors for 310 A. The periodic array of non-dispersive, periodically positioned, quasi-dielectric cylinders in vacuum oriented along the axis of X-ray emission has narrow-band reflection of about 100% for wavelength closed to the distance between dielectric cylinders. Such array with quasi-dielectric cylinders periodically located at about 300 angstroms chess table can be realized by forthcoming lithography methods. Two such mirrors shape the proper resonator for generated emission and form good divergency of (coherent) X-ray beam. The second problem of 310 A radiation absorption by tritium atoms in bounded-free electron transitions may be solved by application of (strong) permanent magnetic field to the tritium sample. The magnetic field produces the spectrum of permitted free electron energy in one direction (of X-ray propagation). While the energy of possible free electron 27.2 eV is out of this spectrum, the bounded-free transition is forbidden, and the absorption of stimulated emission is depressed by several orders and becomes negligible small. The theoretical limit of X-ray continuous laser power is 0.15 mW per cubic cm of liquid atomic tritium. The required magnetic field is about 300 kGs.

1 Introduction: irradiation of tritium under the beta-decay. The beta-decay of any nucleus provides the nucleus with the additive charge +e. After the small time of the beta-decay (it is about the flight time of decayed electron through the atom), the electron system of the atom becomes in a non-stationary state. All electrons have some additional energy with respect to new stationary states. The deepest K- and L- electrons have the energy

658

M. Yu. Romanovsky and V. K. Bityukov

E

− K ,L

Z 2e2 = 2rB (n − δ − ) 2

before the beta-decay, and the energy

E

+ K ,L

( Z + 1) 2 e 2 = 2rB ( n − δ + ) 2

after the beta-decay, where Ze is the charge of initial parent nucleus, n is the principal quantum number (n = 1 for K-electrons and n = 2 for Lelectrons), rB is the Bohr radius, δ- and δ+ are quantum defects of atomic states of initial and product nuclei. To obtain a population inversion, the L states of the product nuclei should have less energy that K state of the parent one. It means that

( Z + 1) 2 e 2 Z 2e2 ΔE = E − E = − + ≤0 2rB (2 − δ + ) 2 2rB (1 − δ − ) 2 + L

− K

Since δ- ~ δ+ ~ δ < 1, this inequality does not fulfill for any pair of nuclei, taken place in beta-decay, except one. Auger-processes start in any nuclei (except one) immediately after the beta-decay [1], and normally do not accompany by any X-rays since the effective “width” of state in an atom with product nucleus (of “product” ion) is very large. There is one atom where radiation processes permitted not only by energy, but due to the fact the electron in a product atom appears in some stationary state. This is atomic tritium. The product ion is 3He+ is produced by beta-decay of tritium. Indeed, the energy of the ground state of tritium (Z = 1, δ- = δ+ = 0) E1T = - e2/2rB while the energy of the first excited state of + 2 2 2 3He (Z = 2, n = 2) ion is E2He = - 2 ·e /2rB·2 = E1T. The probability to appear at this state is about 25% [2]. The process is fast due to the absence of large transformation of a wave-function. At the same time, there are no mechanisms of an energy transfer for transitions to the ground state (the probability is about 70%, see, for example, [2]) as well as to other excited states (total probability 5%). It means, looks, that these lasts transitions occur while some body (nearest atom) collides with the decayed tritium (3He+ ion) and rates of these transitions are (very) slow. Thus, there is a long-life temporal population inversion on the transition 2s – 1s in 3He+ ion. Indeed, the beta-decay of atomic tritium leads to an appearance of the part (denote this part as α) of ions 3He+ at the meta-stable state 2s. There

310 Angstroms X-Ray Lasing Under Beta-Decay of Tritium

659

are 3 possible decays of this state: two-photon decay to 1s state, and the cascade transition to 1s state via Lamb transition to 2p state. The last process is very slow and can be neglected. The third possible decay is the forbidden one-photon M1 transition to 1s state (the electro-dipole and electroquadropole transitions are forbidden) [3]. Thus the radiation of tritium can be calculated easily. The two-photon component is very wide: from zeros for 0 eV and 40.8 eV quanta energy to the maximum of quanta with the 20.4 eV energy. The one-photon component with quanta energy 40.8 eV (wavelength λ = 310 A) is narrow (its natural width is about 19 orders less than the width of the two-photon component), but the maximum is 12 orders larger (the total ratio of 7 orders is the ratio of lifetimes of one-photon transition τ1 ~ 175 s and twophoton transition τ2 ~ 0.92·10-5 s). The spectrum of tritium radiation (it is, in fact, the radiation spectrum of an ion 3He+) is presented on Fig.1. The total rate of tritium emission under the beta-decay (the radiation of ions 3He+) is

v = αΛ -9

where Λ = ln2/T= 1.78·10 s, T is the period of tritium half-decay. The rate of two-component decay is

v2 = αΛ

τ1 ≈v τ1 + τ 2

while the rate of one-photon decay is τ2/τ1 ~ 0.5·10-7 times less.

Fig. 1. The radiation spectrum of tritium under the beta-decay in arbitrary units: the radiation spectrum of an ion 3He+. The width of one-photon radiation is 19 orders less than the width of two-photon radiation (the scale along X-axis is not kept).

This page intentionally blank

660

M. Yu. Romanovsky and V. K. Bityukov

The total power of quantum radiation of ℏ ω1 = 40.8 eV is

P1 = αΛhω1 N 0

τ2 τ1

where N0 is the total number of (tritium) atoms. The one cubic cm of liquid atomic tritium (N0 = 5·1022) irradiates P1 ~ 3α·10-11 W spontaneously, the total power of this two-photon is about one mW. Note that the total power of beta-decay of atomic tritium is

Pβ = ΛN 0

Emax

∫ f ( E )dE 0

where f(E) is the energy spectrum of decayed electrons. The value Pβ ~ 0.15 W for one cubic cm of liquid atomic tritium, i.e. Pβ is two orders larger than the total radiation power. The radiation of tritium is isotropic. To create X-ray laser with the beta-decayed tritium as active medium, several problems have to be solved. The first problem is to get atomic tritium with high enough density. It is in most technical. The second problem is to depress two-photon radiation. It is solved through a creation of conditions for one-photon stimulated radiation with the wavelength λ = 310 A. Such conditions achieve due to good reflectivity resonator (high reflectivity mirrors) for λ = 310 A. This problem is solved [4]. At least, losses of stimulated one-photon radiation (per one pass of resonator) should be less than the gain (per one pass of resonator), i.e. these losses should be depressed specially.

2 High-reflectivity mirrors for λ = 310 A. The first problem of two-photon decay depression is solved by largereflectivity mirrors for 310 A. The periodic array of non-dispersive, periodically positioned, dielectric cylinders in vacuum oriented along the axis of X-ray emission has narrow-band reflection of 100% for wavelength closed to the distance between dielectric cylinders [4]. Such array with dielectric cylinders periodically located at about 300 angstroms chess table can be realized by usual lithography methods. Two such mirrors shape the proper resonator for generated emission and form good divergency of (coherent) X-ray beam.

310 Angstroms X-Ray Lasing Under Beta-Decay of Tritium

661

In Fig. 2, the reflection coefficient is shown as a function of the wavelength expressed in units of array step Dg. The solid-blue and dashed-red curves correspond, respectively, to dielectric constants ε = 2 and ε = 4. As one can see the array becomes a perfect reflector within a fairly narrow wavelength range centered at the resonant wavelength that is slightly larger than the period Dg. Similar results have been obtained for dielectric grating structures. The resonant pattern is associated with the so-called Wood anomalies [5], and can be explained by the existence of trapped modes or guided wave resonances [6].

Fig. 2. Calculated zero-order reflection coefficient for a periodic array of dielectric cylinders in vacuum described in the text. Results are presented as a function of the wavelength of the incident radiation measured in units of the period Dg. The solid blue and dashed red curves correspond, respectively, to the array of cylinders with dielectric constants ε = 2 and ε = 4.

Of course, all medium (of cylinders) can not be considered as dielectrics for X-ray radiation: such radiation excites currents in cylinders (heads). Nevertheless, the volume of such currents is very small due to the cylinders radius, and losses may be considered as negligibly small. Indeed, the cylinders mirror can be compared with the metal grating mirror [4]. The reflection coefficient from “dense” silver grating is about 85% with zero transmission (losses are about 15%) while the grating holes square is about 3%. The square of cylinders head is about 0.7%, therefore total losses are much less than 1%, which is quite enough for the proposed X-ray laser. The good (theoretically, very closed to one) reflectivity for λ = 310 A plays the role in a solutions of damp losses. As well, this perfect reflectiv-

662

M. Yu. Romanovsky and V. K. Bityukov

ity leads to the stimulated emission of λ = 310 A with a sharp space diagram and to the depression of two-photon radiation. Thus, the periodic array of non-dispersive, periodically positioned, dielectric cylinders in vacuum oriented along the axis of X-ray emission solves two problem of Xray laser by tritium: the depression of two-photon radiation (all energy will be forwarded to the one-photon stimulated emission) and increase of final power output of X-ray laser. Moreover, the sharp space diagram of stimulated emission permits to solve the problem of radiation losses.

3 Damping of radiation losses inside an active medium. The third problem of 310 A stimulated emission is the (large) radiation absorption. There are two types of absorption: by bounded electron transitions in 3He+ ions which transited to ground state, and by bounded-free electron transitions in by tritium atoms. Since the concentration of 3He+ ions is small, this type ob absorption can be neglected. The absorption of 40.8 eV quanta radiation in tritium is much more larger. The cross-section of this process can be estimated by well-known Kramers formula 3

σ vn = 7.9 × 10 −18

n ⎛ ωn ⎞ ⎜ ⎟ cm 2 2 Z ⎝ω ⎠

where ωn is the minimal frequency needed for an ionization of hydrogen atom from the n state. For tritium Z = 1, n = 1, and σ1n = 2.9·10-19 cm2. The bound-free absorption coefficient in tritium is proportional to the density, and for liquid tritium (density n0 = 5·1022 cm-3) is equal 14600 cm-1. Even for moderate (gaseous) tritium n ~ 1019 cm-3, this absorption coefficient remains large enough to prevent any stimulated emission (the gain is not so large). How to solve this problem? It is necessary to transform the spectrum of free electrons, appeared due to an absorption of stimulated radiation with λ = 310 A. This spectrum should have the forbidden energy band near the 27.2 eV. It is impossible to realize uniformly in space, but possible for free electrons propagated in the definite direction. Indeed, the magnetic field produces the discrete spectrum of particles, moving at the direction perpendicular to the field direction. Classical motion of charged particles in this case goes along circles, radiuses of these circles depend on particles energy continuously. Quantum effect is the change of the continuous radi-

310 Angstroms X-Ray Lasing Under Beta-Decay of Tritium

663

uses spectrum (in one direction!) to a linear spectrum. The differences between two nearest values of possible energy ΔE is

ΔE =

ehH = hωL mc

where H is an amplitude of (permanent) magnetic field, e is an electron charge, m is an electron mass, ωL is the Larmor frequency (see, for example, [7]. This energy difference should be larger than the energy difference of stimulated emission arises by line broadening ℏ Δω, i.e.

ωL > Δω If the permanent magnetic field provides this inequality, the energy value of 27.2 eV becomes forbidden for free electrons, moving along the axis of stimulated radiation. Since only these electrons can absorb quanta of stimulated radiation due to the momentum conservation law, the absorption coefficient along the X-ray axis becomes depressed strongly. Experimentally, this fine depression can be achieved by tuning of the magnetic field amplitude.

4 Estimates. Since the total radiation of tritium under the beta-decay in good resonator goes at the wavelength of 310 A, the output power of stimulated emission may reach the value

P1s = αΛhω1 N 0 Thus the theoretical limit of X-ray continuous laser power is 0.15 mW per cubic cm of liquid atomic tritium, α = 0.25. Of course, an experimental preparation of liquid atomic tritium is doubt: atomic tritium with gas densities is much more achievable. The gas atomic tritium has to be prepared specially. The gas with room temperature and normal pressure, the expected output laser power is about 0.1 μW per cubic cm. The broadening of 310 A X-ray laser radiation in above normal gas conditions has the non-uniform character (collision broadening is small). The recoil linewidth ΔωR ≈ 4·1012 s-1, the Doppler linewidth ΔωD ≈ 0.4·1012

664

M. Yu. Romanovsky and V. K. Bityukov

s-1, thus the total non-uniform linewidth Δω = ΔωR + ΔωD ≈ 4.4·1012 sThis limits the magnetic field above the value

1

H > mcΔω / e

or, for normal gas atomic tritium, H > 2.2·105 CGS. Such magnetic field can support the existence of tritium in form of atomic gas.

Fig. 3. The possible experimental scheme of atomic tritium X-ray laser. The gaseous atomic tritium is contained in the tube. Cylinder array mirrors are located in vacuum near the side windows of the tube.

The possible experimental scheme of 310 A X-ray laser based on the beta-decay of atomic tritium can be the following (Fig.3). The tritium tube can have the volume up to 102 – 103 cm3, therefore the expected X-ray laser power output may be up to 1 mW while α = 0.25. Thus, the combination of three devices: the tube with gaseous atomic tritium, high reflectivity mirrors (cylinders array), and strong magnetic field may provide X-ray lasing of λ = 310 A with output power up to 1 mW.

References 1. K.Siegbahn. Beta-and Gamma-rays Spectroscopy. Amsterdam (1955). 2. P.V.Elyutin, V.D.Krivchenkov. Quantum mechanics. Moscow: Science (1976) [in Russian]. 3. Berestetskii V. B., Landau L. D., Lifshitz E.M., Pitaevskii L. P. Quantum Electrodynamics. Oxford, UK: Pergamon Press (1982). 4. A.G.Borisov, S.V.Shabanov. J. Comp.Phys., 209, 643 (2005). 5. R.W. Wood, Phys. Rev. 48, 928 (1935) 6. R. Magnusson and S.S. Wang, Appl. Phys. Lett. 61, 1022 (1992);

310 Angstroms X-Ray Lasing Under Beta-Decay of Tritium 7. 8. 9. 10. 11.

665

S. Peng, G.M. Morris, Opt. Lett. 21, 549 (1996); T. Peter, R. Br¨auer, O. Bryngdahl, Optics Communications 139, 177 (1997); K. Koshino, Phys. Rev. B 67, 165213 (2003); L. Pilozzi, A. D’Andera, R. Del Sole, Phys. Rev. B 54, 10763 (1996). L.D.Landau, E.M.Lifshitz. Quantum mechanics: non-relativistic theory. 3rd ed., Oxford, UK: Pergamon Press (1991).

Capillary Discharge Soft X-Ray Laser at University of L’Aquila: Laboratory Survey A. Ritucci, G. Tomassetti, A. Reale and P. Zuppella Physics Department of University of L’Aquila, gc LNGS-INFN, via Vetoio 67010 Coppito L’Aquila, Italy

Summary. We report on different experimental investigations regarding the use of the 0.3 mJ, 0.3 Hz, 1.7 ns, 46.9 nm Ar laser developed at the University of L’Aquila. Firstly, we have investigated the potentiality of the EUV laser in the ablation of hard dielectrics such as CaF2 and LiF crystals, which are largely transparent to the visible and to the ultraviolet laser light. We determined the ablation thresholds and the ablation rates for these two crystals and compared the results with the values obtained with other kinds of visible-ultraviolet laser systems. Secondly, we investigated a method to manipulate the intense soft X-ray laser beam based on the use of hollow waveguides. We used monocapillary tubes in a multireflection regime to achieve a substantial modification of the intensity distribution of the beam. These results can be of great interest for several applications of these soft X-ray laser sources.

1 Introduction Among the variety of soft X-ray lasers, which are recently under fast development in several laboratories worldwide, the capillary discharge soft X-ray Ar laser operating at 46.9 nm remains, to now, the most reliable one [1-2]. This laser is the most suitable for a large number of laboratory investigations for its simplicity of use, its compactness and the relatively large output energy per pulse. Many of such investigations may be of relevance in the field of applied science and the basic physics [3-4]. Moreover they can represent a valuable starting point in view of the applications of the new, high rep-rate, tabletop EUV lasers in laser-produced-plasmas having shorter wavelengths and smaller pulse durations [5]. The aim of this paper is to report some experimental investigations, which have been performed by the use of the 46.9 nm Ar laser, recently developed at the University of L’Aquila [2]. The laser under consideration operates at the reprate of 0.3 Hz, with 0.3 mJ/pulse and 1.7 ns pulse duration.

This page intentionally blank

668

A. Ritucci et al.

2 EUV Laser Ablation Today, laser ablation has a significant role in the field of applied science for the processing of different kinds of materials and/or for their characterization. From the general point of view, laser ablation depends on different experimental parameters such as the pulse laser energy and duration, the energy fluence and strongly on the penetration depth of the laser light through the material. In this work we have focused our attention to hard dielectrics, such as CaF2, LiF, BaF2 or SiO2, which have relevant interest for several applications such as for miniaturized optical devices. These crystals are characterized by very large band gap (e.g. Eg ≈12.0, 13.6-14.5 or 9.1 eV for CaF2, LiF and BaF2 respectively [6]), so that they are largely transparent from the infrared up to the vacuum ultraviolet (where hν 1 J/cm2, the fluence is sufficiently high to produce evaporation of material over the whole beam cross section. The ablated crater has, in this case, a conical shape with regular vertical profile and deepness >1-1.5 µm. The analysis performed at the AFM (Fig. 1) for the large fluence confirms the regularity and the deepness of the ablated region. This behavior in the vertical profile of the craters is found both in the case of LiF and CaF2. Figures 4 (a) – (b) show the SEM images of LiF irradiated with the fluence of 0.8 J/cm2 (a) and 3 J/ cm2 ((b). Irregular reliefs on the ablated region can be reasonably attributed to the irregular distribution of the laser fluence due to the diffraction of the beam and not to liquid waves formed by the melting of the surface. This analysis shows also that the ablation processes at 46.9 nm is accompanied by the formation of micro-sized cracks inside the irradiated area.

Capillary Discharge Soft X-Ray Laser at University of L’Aquila

671

These cracks are due to the strong thermoelastic stress on the surface and to the brittleness of the materials. An interesting behavior is that the cracks are observed already very close to

Fig. 4 SEM image of an ablated region of LiF with fluence of (a) 0.8 J/cm2 and (b) 3 J/cm2.

6 µm

QuickTime™ and a decompressor are needed to see this picture.

QuickTime™ and a decompressor are needed to see this picture.

QuickTime™ and a decompressor are needed to see this picture.

Fig. 5 Ablated spot on LiF with fluence > 3 J/cm2 and using an improved optical configuration.

the ablation threshold at very low irradiation fluences and increase with the number of pulses. These cracks stand along preferential directions, which should correspond to the cleavage planes of the crystal. The higher density of short microcracks at the periphery (see Fig. 4 (a)-(b)) of the craters can be attributed to the different thermoelastic forces acting inside and at the edge of the laser spot. As the fluence is increased from the threshold, the evaporation of material is more efficient, cracks become less evident and a cleaner condition of ablation is found. Fractures on CaF2 samples are typically less evident and of smaller dimensions. They present irregular shape homogeneously distributed through the irradiated area. Approaching 3

672

A. Ritucci et al.

J/cm2, the mechanical stress of the surface becomes so large (see fig. 5) to detach away from the surface macroscopic pieces of material and the quality of ablation is lost. The presence of cracks even close to the ablation threshold is in contrast to what is generally expected by the shortening of the laser wavelength and suggests the necessity for the modeling and a better understanding of the ablation processes induced by EUV and soft Xrays. Concerning the resolution fig. 5 show an ablated spot obtained at a large fluence (>3 J/cm2) using an improved focusing system. The central hole has diameter of only 6 µm. The fluence in this case is so large to destroy substantially the crystal in proximity of the hole.

3 Capillary waveguides in the EUV Soft X-ray lasers are generally obtained without any optical cavity by the single pass amplification of a lasing line through elongated hot plasma columns. Unfortunately, the amplifying plasmas are often characterized by large density gradients where the laser emission can experience large refraction effects. As a consequence, the intensity distribution of a soft X-ray laser can have unusual and not uniform profiles, which are quite difficult to control. As an example, our laser is characterized by an annular shape with the divergence of 4.5 mrad. This beam profile may be inconvenient for several kinds of applications. The need for the manipulation of the beam of such a kind of coherent soft X-ray radiation pushes the need for testing different kinds of optical elements. In this section we have investigated the possibility to use hollow glass capillary waveguides to modify the intensity distribution of the soft X-ray laser beam. The soft X-ray propagation in different kinds of capillary optics (both in the form of mono- and poli-capillary structures) has been well studied for several years and it has been successfully employed on different kind of sources: synchrotrons, incoherent plasma sources (laser plasmas and z-pinches) and x-ray tubes (see for example refs. [12-13]) at wavelengths typically