Set-up for ultrafast time-resolved x-ray diffraction using a femtosecond laser-plasma kev x-ray-source

Transcription

1 Set-up for ultrafast time-resolved x-ray diffraction using a femtosecond laser-plasma kev x-ray-source C. Blome, K. Sokolowski-Tinten *, C. Dietrich, A. Tarasevitch, D. von der Linde Inst. for Laser- and Plasmaphysics, University of Essen, Essen, Germany, Phone: , Fax: , kst@ilp.physik.uni-essen.de * corresponding author

2 C. Blome, K. Sokolowski-Tinten, C. Dietrich, A. Tarasevitch, D. von der Linde Abstract. A short-pulse 4.51keV Ti K α femtosecond laser-plasma driven hard x-ray source has been built and characterized. This source is incorporated into a set-up for ultrafast time-resolved x-ray diffraction experiments, to study ultrafast lattice dynamics in laser-excited materials. The present knowledge of the atomic structure of matter is to a large extent due to x-ray spectroscopies. However, direct observation of atomic motion, for example during chemical reactions or phase transitions, is very difficult because they occur transiently on time-scales comparable with the natural oscillation periods of atoms and molecules, that is femtoseconds to picoseconds. Although ultrafast optical techniques are well established they probe only the electronic properties and provide only indirect information on structure. With the recent progress in ultrafast laser technology the situation is rapidly changing. Chirped pulse amplification has enabled efficient amplification of femtosecond laser pulses and lead to the development of table-top laser systems up to the multi-tw power level. When focused onto solid targets, such intense pulses generate high-density plasmas emitting short bursts of hard x-rays up to the MeV region [1]. This new kind of high brightness ultrashort pulse x-ray sources allows to extend well-established experimental techniques (pump-probe-scheme) of time-resolved optical spectroscopy into the x-ray range, thus enabling direct quantitative studies of ultrafast atomic motion deep inside the bulk of matter [2]. In this contribution we describe a set-up for ultrafast time-resolved x-ray diffraction experiments, which is based on a femtosecond laser-plasma driven x-ray source. We have chosen the K α -emission from laser-excited titanium, because the 4.51keV photon energy of the Ti-K α -line is high enough to observe Bragg diffraction from crystalline materials without any restrictions, but can be produced with sufficient efficiency even with a laser of moderate energy/power (in our case: 120fs, up to 150mJ). In addition, due to the generation process (fast, laser-accelerated electrons penetrate the cold material beneath the hot plasma and create characteristic line radiation [3]) it is expected that the K α -radiation can be produced with sub-ps pulse duration. Fig. 1 displays spectra of our source in the kev-range. The overview-spectra shown in the left viewgraph have been obtained from flat thin-film targets (500nm Ti on glass) using a x- ray CCD (thinned, back-illuminated) in the single photon counting mode with pulse height analysis (to assure single-photon detection we had to reduce the laser energy). At 4.51keV and 4.93keV, the K α - and K β -lines of Ti are visible, as well as a weak line at 8.6keV which we attribute to K α -emission of Znimpurities of the plasma-deposited films. The right viewgraph shows a high-resolution spectrum of the

3 spin-orbit-split Ti-K α -emission obtained from a bulk Ti-target (wire, see below) with a Si-(311) crystal spectrometer. Figure 1. Spectra of the Ti-K α -source. Left: overview obtained on thin-film targets by photon counting/pulse height analysis with x-ray CCD. Right: High-resolution spectrum of the spin-orbit split K α -lines obtained with a Si-(311)-crystal spectrometer. With a single 20mJ laser pulse and optimum focusing (see below) we obtain 3x10 8 K α -photons per pulse from the thin film targets and up to 10 9 photons/pulse from a bulk target. Similar efficiency ratios between thin-film and bulk targets have been also observed for other materials (Cu). For the timeresolved diffraction set-up we do not use flat targets, but have constructed a wire-target. A 250µm thick Ti-wire is moved with high precision and at constant velocity through the focus of the laser. This target design is very compact which simplifies shielding issues and allows virtually infinite measurement times. At 10Hz repetition rate and a wire velocity of 200µm/pulse a 500m wire role will stand nearly 70h! In the following we discuss experiments aimed to optimize our source for time-resolved diffraction experiments. As has been described above, characteristic K α -radiation is produced by fast, laseraccelerated electrons. The efficiency of K α -production depends on the energy of the electrons and is therefore, strongly influenced by the parameters of the laser-generated plasma. In this context it is common experience in many laboratories [4] and also supported by theoretical calculations [5] that for given laser and material parameters the highest possible laser intensities not necessarily lead to the highest K α -yield. This is demonstrated by Fig. 2, which shows the K α -yield at 4.51 kev (dots) as a function of the position of the focusing lens (f=150mm) relative to the Ti-target. The K α -signal has been normalized to the yield with the Ti-target exactly in the focal plane of the lens. The highest yield is obtained 0.5mm away from focus, which corresponds to approximately two times the Rayleigh length. At the same time the background signal detected by the CCD (squares; arbitrarily normalized to fit into the plot window of Fig. 2) due to hard x-rays is significantly reduced. Therefore, optimizing the focusing conditions does not only increase the K α -flux, but allows at the same time improvement of the signal-to-noise ratio. Another way to tailor the plasma properties is the use of a double-pulse excitation scheme which separates the steps of plasma- and X-ray generation. A first, relatively weak pulse is used solely for plasma generation. After a certain delay the main pulse interacts then with the pre-formed plasma and generates the fast electrons required for K α -production.

4 Figure 2. Normalized x-ray signal as a function of lens position (relative to the best focus). Dots: Ti-K α -emission; Squares: hard background. Figure 3. Ti-K α -yield in a two-pulse excitation scheme as a function of delay time between the plasma generating pulse and the main pulse. Fig. 3 shows the yield of the Ti-K α -source as a function of the time delay between the plasma generating pulse (I W/cm 2 ) and the main pulse (I 5x10 16 W/cm 2 ), which produces the K α -radiation. In nearly perfect agreement with results obtained at a Si-K α -source at 1.8keV [6], we achieve an enhancement of almost an order of magnitude at a delay time of a few ps. It should be kept in mind that the K α -radiation from such laser-generated plasmas is emitted incoherently into 4π solid angle. Therefore, efficient use of the produced x-rays in an optical pump, x-ray probe experiment requires focusing of the x-rays onto the surface of the sample under investigation with a spot size comparable or smaller than the area excited by the optical pump. In our set-up we use a toroidally bent Si-crystal with 311-surface orientation as a focusing x-ray mirror. As has been discussed in detail in reference 7, monochromatic point-to-point imaging of the plasma-source can be achieved in this way. The left part of Fig. 4 displays the experimental geometry. Figure 4. Focusing of the K α -radiation with a toroidally bent crystal. Left: experimental geometry; Upper right: focal distribution containing 10 4 K α -photons; Lower right: rocking curve of 390nm (111)-Ge-film on (111)-Si.

5 For a given crystallographic orientation of the mirror the horizontal and vertical bending radii are determined by the imaging geometry on the Rowland circle and the requirement that the Bragg-condition has to be fulfilled for the chosen wavelength. In the upper right part of Fig. 4 the result of such a focusing experiment with our Ti-K α -source is shown. It displays the distribution of the focused K α -radiation as detected with our x-ray CCD (pixel size 27µm). The spot is nearly circular and exhibits a FWHM of approximately 85µm, as demonstrated by the horizontal cross section shown in the upper right. Typically the focus contains detected K α -photons per pulse. In the diffraction experiments the sample under investigation will be placed exactly at the image point (under the appropriate Bragg-angle). An example is shown in the lower left of Fig. 4. It represents the diffraction pattern of a 390nm (111)-oriented Ge-film grown on a (111)-Si- substrate by surfactantmediated heteroepitaxy [8]. The K α -radiation is diffracted at two distinct Bragg-angles (Ge: o, Si: 26 o ) due to the different lattice constants of the two diamond-like materials (Ge: 5.65Å, Si: 5.43Å). While the rocking curve width of the bulk substrate is mainly determined by the properties of the focused radiation (source size, spectral bandwidth), the large linewidth of the overlayer results from the finite film thickness. Because the Si x-ray mirror captures a large solid angle (in our case 2 o horizontally and 7 o vertically), such diffraction patterns are directly obtained from a single CCD-image without any angle scanning of the sample. Therefore, even with small exposure times (in this case 1min) complete rocking curves can be measured with high signal-to-noise ratio. In conclusion, we have built and characterized a set-up for ultrafast time-resolved x-ray diffraction experiments which is based on a short-pulse 4.51keV Ti K α femtosecond laser-plasma driven hard x-ray source. Time resolved experiment to study ultrafast lattice dynamics in laser-excited materials are on the way. Acknowledgements C.B. acknowledges travel support by the EU. K.S.T. gratefully acknowledges support by the Deutsche Forschungsgemeinschaft and by the Stiftung Flughafen Frankfurt/Main. The authors are indebted to I. Uschmann and E. Förster for preparation of the x-ray mirror. References [1] Kmetec J.D. et al., Phys. Rev. Lett. 68 (1992) [2] Rischel C. et al., Nature 390 (1997) ; Larsson J. et al., Appl. Phys. A 66 (1998) ; Rose-Petruck C. et al., Nature 398 (1999) ; Chin A.H. et al., Phys. Rev. Lett. 83 (1999) ; Siders C.W. et al., Science 286 (1999) ; Lindenberg A.M. et al., Phys. Rev. Lett. 84 (2000) ; Cavalleri A. et al., Phys. Rev. Lett. 85 (2000) [3] Rousse A. et al., Phys. Rev. E 50 (1994) [4] Eder D.C. et al., Appl. Phys. A 70 (2000) [5] Reich Ch. et al., Phys. Rev. Lett. 84 (2000) [6] Bastiani S., Phys. Rev. E 56 (1997) [7] Missalla T. et al., Rev. Sc. Instr. 70 (1999) [8] Horn-von-Hoegen M., Appl. Phys. A 59 (1994)