Time-resolved optical pump/x-ray probe spectroscopy

Transcription

1 4.3 Project part P03 Time-resolved optical pump/x-ray probe spectroscopy Principal investigators: Christian Spielmann Physikalisches Institut EP1 Universität Würzburg Am Hubland D Würzburg Phone: +49 (0) Fax: +49 (0) spielmann@physik.uni-wuerzburg.de Allocation to technical disciplines (according to code of OeStat*) 1 % 2 % 3 %

2 Report on Project Parts Summary The development of reliable femtosecond solid-state laser brought new possibilities into time-resolved spectroscopy (Zewail 2001). For the first time it became possible, in principle to monitor the nuclear motion of molecules, crystal lattices and other out-ofequilibrium structures. However, usually it is very difficult to map the experimental observations to the structural dynamics. Therefore, experimental approaches are needed that can overcome the limitation of optical studies for structural determination, while the high temporal resolution of femtosecond lasers is maintained. Techniques such as X-ray diffraction (XRD) (Rousse 2001), X-ray absorption spectroscopy (XAS) (Nakano 1999), or X- ray photoelectron spectroscopy (XPS) (Nuggent Glandorf 2001) deliver much more direct information about the structure. The key to the successful realization was the development of laser driven x-ray sources. Various schemes have been demonstrated but all of them rely on availability of state of the art femtosecond solid-state laser system. In this project part we concentrated on the optimized generation of soft x-ray pulses via high harmonic generation (HHG) (Brabec 2000) and their application for time resolved spectroscopy. In a first set of experiments we demonstrated qualitative amplitude shaping of the soft x-ray spectrum produced in the process of high-harmonic generation. This is accomplished by applying adaptive femtosecond pulse shaping methods. We performed the basic operations of complete spectral control by selective enhancement and suppression as well as the shift of spectral components (Pfeifer 2005a, Pfeifer 2005b). Our ability to qualitatively engineer the coherent spectral properties by application of temporal and spatial laser-pulse-shaping methods has immediate consequences for the developing field of time-resolved x-ray spectroscopy. On one hand the possibility to select single harmonic line without impairing photon numbers or time structure will be indispensable for application in time resolved XPS. Moreover control over the spectrum is directly related to the control over the temporal structure. HHG sources have been limited so far to the 100eV range, but recently we were able to demonstrate an extension of the cut-off to nearly 1 kev (Seres E. 2004, Seres J 2005). Using very short driving laser pulses, the spectrum becomes continuous near the cut off and will be ideally suited for absorption spectroscopy. Due to the coherent generation process, the XUV pulses are always shorter than the driving laser pulses. The XUV signal has been intense enough in an energy range up to 500eV opening the way to EXAFS (extended x-ray absorption fine structure). In a first proof-of principle experiment we followed structural changes in Silicon after excitation with an intense laser pulse with EXAFS. We measured phonon spectra of amorphous-si for the first time in the time domain (Seres E. 2005).

3 P03 - Time-resolved optical pump/x-ray probe spectroscopy (Spielmann) Results and discussion According to the proposal for the second period, the main objectives were the set up of a soft x-ray source and the development of the necessary instrumentation to use the x-ray pulses for time resolved spectroscopy. To reach these goals the following tasks have been defined: (A) Set up of a high-brightness high-harmonic source (HH-source) (B) Setup of x-ray microscopes and interferometer illuminated with a HH source (C) Time resolved x-ray spectroscopy with HH radiation (A) High brightness high harmonic source: Based on our current understanding of HH generation, it is necessary to shorten the driving laser pulses to increase the conversion efficiency and to extend the cut-off towards shorter wavelengths. In the current funding of this project major efforts have been made to establish an optimized laser system (Brabec 2000). The pulses of the existing laser systems have to be shortened and the pulse energy has to be boosted up. In collaboration with P02 we developed a three stage amplifier system, capable of generating 10fs pulses with energy up to 4mJ at a repetition rate of 1 khz (Seres J. 2003). A key element to obtain the reported parameters has been a broadband acousto-optical pulses shaper (AOPDF). At begin of our work no broadband AOPDF was available (Seres E. 2003). In cooperation with the manufacturer (FASTLITE) we tested the first prototype broadband AOPDF. Our measurements allowed then an optimization of the device meeting our requirements. The development of a laser driven pulsed x-ray source in the several 100eV range will have a great impact on biochemistry and biophysics for mapping the molecular structure, or photoelectron spectroscopy of magnetic metals, requiring photon energies around the L- edge (0.5-2 kev) (Bressler 2004). Despite more and more powerful laser systems the x-ray photon yield of high harmonic sources was very limited. In the low energy range reabsorption limits a growth of the signal over a longer propagation length. However, reabsorption plays only a minor role in the high energy range where the phase mismatch between the driving laser pulse and the generated radiation limits a further growth of the signal. To solve this problem several quasi-phase matching schemas have been suggested. Recently quasi phase-matching in a modulated waveguides lead to enhanced conversion efficiency in the water window at 4.4nm (Gibson 2003). In our experiments we employed a different approach to generate coherent soft-x-ray radiation covering the full water window with a high efficiency. The output beam of the three stage amplifier system (Seres J. 2003) has been focused into a He gas jet with an estimated peak intensity of 8x10 15 W/cm 2. Such high peak intensity opens the way to produce high

4 Report on Project Parts harmonic radiation with photon energy up to 1.5 kev using the approximation for single-atom emission. The generated soft x-ray radiation has been characterized with a 1-m grazing incidence scanning spectrograph. To calibrate the monochromator we put Ti and Al filter into the beam path. We were able to clearly resolve the Aluminum L-edge at 73 ev, the carbon K- edge and Titanium L-edge. At an increased sensitivity we were able to safely detect a signal for photon energies up to 700eV (Seres E. 2004). To compare the yield of our source with other sources we estimated the number of photons after optimization of the laser and target parameters. Taking into account the sensitivity of the channeltron, the diffraction efficiency of the grating and the effect of the slits in the monochromator, we estimated the following photon numbers in the 5 % band concerning to the spectral resolution: 9x10 7 photons/s at 100 ev, 5x10 7 photons/s at 200 ev; 6.6x10 6 photons/s below Carbon K-edge; 1.3x10 6 photons/s at the Nitrogen K-edge; 8.6x10 5 photons/s below the Titanium L-edge; 3.3x10 5 photons/s at Oxygen K-edge and 1x10 5 photons/s at 700 ev. Using an adiabatic approximation we have estimated a coherence length of only a few microns. Such a short coherence length will not support the observed high photon numbers. A possible explanation is non-adiabatic self-phase matching (NSPM), which requires that ionization is confined to a few optical cycles (Tempea 2000). Applying ADK theory we found for our experimental parameters, that the ionization takes mainly place during three and half optical cycles at the leading edge of the pulse. According to Tempea et. al, we can expect under these experimental conditions an enhancement of the conversion efficiency due to NSPM. According to this theory a further shortening of the pulses and/or higher intensity should result at shorter wavelengths. We repeated this experiment with an increased intensity well above W/cm 2 and replaced the monochromator by an energy dispersive x- ray spectrometer. Under these experimental conditions we could demonstrate spatially coherent soft x-ray radiation up to photon energies of 1keV. Further improvements are predicted and should allow an extension up to several kev (Seres J. 2005). Beside the work on improving our laser system we followed another route to optimize our x-ray source. Under standard experimental conditions, where usually near-transform limited pulses are used for HHG, the spectrum exhibits a well-known shape. The highharmonic intensity stays roughly constant for many orders up to ~300 (Brabec 2000) and finally vanishes abruptly at a particular photon energy. In our efforts to control the process of

5 P03 - Time-resolved optical pump/x-ray probe spectroscopy (Spielmann) HHG, It has been shown that with an optimization algorithm in conjunction with a deformablemirror pulse shaper (Pfeifer 2005a) it is possible to find femtosecond laser pulse shapes which enhance the total harmonic yield (Bartels 2000). With this technique we are also able to achieve quasi-monochromatic emission of HHG by suppressing neighboring harmonic orders. This possibility provides a solution to the experimental obstacle of selecting single harmonics for pump-probe experiments. In addition, we have shown that the high-order harmonic spectra can be amplitude shaped in a much more comprehensive way (Pfeifer 2005a). The magnitude of control ranges from generation of single harmonic peaks via selective generation of certain parts of the spectrum. The first experiments have been performed inside a gas-filled hollow fiber. During these experiments it became evident that not only the gas pressure but also the spatial fiber modes are important parameters to control the process HHG. With the control of the spatial beam profile, the coupling efficiency into different modes can be optimized and the level of control of the generated XUV radiation further enhanced. This is accomplished using an electronically addressable phase-only spatial light modulator (SLM). In a test setup the shaping capabilities of the SLM were demonstrated by excitation of a variety of different fiber modes and complex combinations thereof at the exit. The generation of high harmonics reveals that the excitation of different fiber modes results in different harmonic spectra for each of these modes. With a feedback loop we optimized the overall harmonic signal (Pfeifer 2005b). For the optimized laser beam profiles we measured HHG profiles and it became evident, that nonlinear coupling between the fiber modes is very important. Therefore is very hard to calculate the spatial intensity and phase profile of the laser pulse that would transform into a single fiber mode after some propagation distance in the fiber. Even in case this was possible we would still need to exactly reproduce the calculated optimal spatial amplitude and phase structure of the beam experimentally. Using an evolutionary algorithm this problem can be circumvented and easily allows finding the optimal initial spatial laser pulse shape and thus enhancing the brightness of the harmonics. (B) X-ray microscopy and interferometry Coherent soft x-ray sources open the way to new capabilities in high-resolution imaging, site- and element-specific spectroscopy and bio-microscopy. In this funding period we were able to demonstrate imaging with a table-top soft x-ray microscope (Wieland 2002). By combining a laser driven high harmonic light source, optimized for having a maximum brightness at around 100eV, a pair of multilayer mirrors to select a narrow spectral band and acting simultaneously as a condenser and a Fresnel zone plate as microscope objective, we

6 Report on Project Parts were able to resolve 200 nm structures of a diatome sample (Wieland 2005). Further, the pulsed nature of our x-ray source offers the possibility of time-resolved spectromicroscopy with a temporal resolution in the order of a few femtoseconds. The presented microscope is suited for investigations in the spectral range between the Al- and Si-L-edge below 100 ev photon energy. Interferometry is one of the most powerful tools known from the visible (VIS) spectral range used for a variety of applications and measuring techniques based on the sensitivity to the phase of the electromagnetic wave. The transfer of common interferometer setups from the VIS to shorter wavelengths, in particular to extreme ultra-violet (EUV) or soft x-ray wavelength suffers from hardly available high-performance optical components. One way to overcome this difficulty is the use of diffractive optics as beam splitting and recombining elements, as it is known from shearing or grating interferometer setups (Goulielmakis 2002). Here, the low diffraction efficiency of the order of a few per cent of the used gratings often is a severe disadvantage. Diffractive optical elements like Fresnel zone plates may be used for interferometer setups, as well. A first zone plate acts as beam splitter, a second one recombines the separated beams and creates the interferogram. Zone plate interferometry in the visible spectral range has been reported decades ago, recently common path interferometry based on zone plates was demonstrated at 4 kev photon energy (Wilhein 2001). Interferometry using a two zone plate common path interferometer designed for operation at 13 nm wavelength has been demonstrated in the current funding period. The interferometer was operated at a high-harmonic source providing radiation with a high degree of spatial and temporal coherence required by the interferometer. A precise Fig. 2: Schematic of a common-path two-zone-plate interferometer. Interference of different diffraction orders can be observed; the most promising combinations are (+1, 1) and (+1,+1). In the experiments reported, the distance between the zone plates equals the difference of the focal lengths.

7 P03 - Time-resolved optical pump/x-ray probe spectroscopy (Spielmann) alignment of the zone plates with respect to each other was achieved for optimal control of the interference pattern. As a possible application of the interferometer we demonstrated the simultaneous determination of the real and complex part of number of refraction of Zr in the 100eV-regime (Wieland 2003). A promising extension would be the combination of interferometry with spectroscopy for detailed analysis of the number of refraction near absorption edges. (C) Time resolved x-ray spectroscopy with HH radiation In this chapter we report on the generation of soft x-ray pulses via high harmonic generation and their first use for time resolved XAS to investigate the structural dynamic of amorphous silicon with a temporal resolution of about 20fs. To our knowledge this is the highest temporal resolution ever demonstrated in XAS. To tackle time-resolved XAS in the soft x-ray regime the light source must meet the following requirements: a) it must provide continuum radiation, b) it must provide ultrafast pulses, and c) it should have a sufficient photon flux (Bressler 2004). We realized it via high harmonic generation. HHG is a line radiation, and therefore of limited use for XAS. However, using very short driving laser pulses the line spectrum becomes continuous near the cut off (Brabec 2000). Due the generation process the XUV pulses are also always shorter than the driving laser pulses. The short pulse duration and the excellent spatial and spectral characteristic make HHG based sources well suited for time-resolved XAS. Our pump-probe experimental setup based on the Ti:sapphire CPA amplifier system developed in cooperation with P02. For the described experiment it was extremely important, that our parameters have been stable over an extended time (Seres J. 2003). Most of the energy of the output beam was tightly focused with a broadband mirror with a focusing length of 150 mm into a Ne gas jet at an intensity of about W/cm 2. The XUV radiation hits our sample, which is a 100nm thick silicon film, consisting of randomly oriented micro-crystallites (amorphous silicon a-si). The transmitted beam is launched into a scanning grazing incidence monochromator the output of which was connected to lock in amplifier. The laser and gas jet parameters have been optimized to maximize the signal at around 100eV, where we want to study dynamical structure modifications of silicon via changes of absorption near the L-edge. The signal has been safely above the noise level up to energies of about 500eV open the way to EXAFS (extended x-ray absorption fine structure) (Rehr 2000). A small fraction of the output beam energy is delayed and focused onto the sample obtaining a pump fluence nearly two orders of magnitude below the damage threshold. Due to chopping of the pump beam and using a lock in amplifier it is now easily possible to detect changes of the transmitted spectra as small as a 10-4.

8 Report on Project Parts The pump pulse modifies the conduction and valence band density of states (DOS) in Si, via single and two-photon absorption. These modifications have their signature also in the fine-structure of the soft-x-ray absorption spectrum (Brown 1972, Nakano 1999). To investigate the carrier dynamic we recorded the difference spectra in the vicinity of the L- edge as a function of the delay. From these spectra we have identified a fast and a slow time constant (Seres E. 2005). Comparing our findings with experiments based on conventional optical spectroscopy of silicon the fast time constant of about 200fs is in reasonable agreement with the previously observed electron-phonon relaxation time in a-si. The longer time constant of about 80ps corresponds to the carrier recombination time of electron and holes across the Si band gap (Sundram 2002). Further detailed calculations are necessary to fully understand the observed dynamics. To follow structural changes it is much more convenient to measure modification of the absorption spectrum in a range far above the edge, which is known in the literature as EXAFS. From a static absorption measurement we estimated the atomic distance with an uncertainty of less than 5% compared to synchrotron measurements (Glover 2003). In a second set of measurements we recorded difference spectra in a range up to 500eV as a function of the delay. At a fixed energy we Fourier-transformed the time series and obtained a spectrum having maxima at about 4 and 16THz (Seres E. 2005). These frequencies agree very well with the predicted numbers for coherent phonons in Si after laser pulse excitation (Stampfli 1992, Stampfli 1994). Summing up, our experimental success depended critically on the parameters and long-term reliability of our Ti:sapphire CPA system developed within the SFB. We demonstrated time resolved x-ray absorption spectroscopy with a resolution in the sub-20 fs range around the L- edge (100 ev) of amorphous silicon (XANES) and gathered information beyond the Si L-edge about the atomic structure with EXAFS. It allowed to gain insight in the dynamics of optical phonons and a direct way and to observe changes of the interatomic distances with a resolution of less than 20fs. Our setup can be easily adapted for other materials such as Carbon, and open the way s to gather information about the fast dynamical processes in molecules. References Bartels, R. S. Backus, E. Zeek, L. Misoguti, G. Vdovin, I. Christov, M. M. Murnane, and H. C. Kapteyn, 2000, Nature, 406, 164 Brabec T., F. Krausz, 2000, Rev. Mod. Phys. 72, 545 Bressler C., M. Chergui, 2004, Chem. Rev. 104, 1781 Brown F. C., O. P. Rustgi, 1972, Phys. Rev. Lett. 28, 497

9 P03 - Time-resolved optical pump/x-ray probe spectroscopy (Spielmann) Gibson EA, Paul A, Wagner N, Tobey R, Gaudiosi D, Backus S, Christov IP, Aquila A, Gullikson EM, Attwood DT, Murnane MM, Kapteyn HC, 2003, Science Glover C.J., F. J. Foran, M. C. Ridgway, 2003, Nucl. Meth. Instr. B 199, 195 Goulielmakis, E., G. Nersisyan, N. A. Papadogiannis, D. Charalambidis, G. D. Tsakiris, and K. Witte, 2002, Appl. Phys. B 74, 197 Nakano H., Y. Goto, P. Lu, T. Nishikawa, and N. Usegui, 1999,, Appl. Phys. Lett 75, 2350 Nugent-Glandorf,-L.; M. Scheer, D. A. Samuels, A. M. Mulhisen, E. R. Grant, Y. Xueming, V. M. Bierbaum, S. R. Leone, 2001, Phys. Rev. Lett. 87, 33 Pfeifer T., D. Walter, C. Winterfeldt, Ch. Spielmann, G. Gerber, 2005a, Appl. Phys B Pfeifer T., R. Kemmer, R. Spitzenpfeil, D. Walter, C. Winterfeldt, G. Gerber, Ch. Spielmann, 2005b, Opt. Lett. 30, 1497 Rousse A, Rischel C, Gauthier JC, 2001, Rev Mod. Phys.73, 17 Rehr J.J., Albers R.C., 2000, Rev Mod. Phys. 72, 621 Seres E., R. Herzog, J. Seres, D. Kaplan, C. Spielmann, 2003, Opt. Express 11, 240 Seres E., J. Seres, F. Krausz, Ch. Spielmann, 2004, Phys Rev Lett. 92, Seres E., C. Spielmann 2005 Time resolved x-ray absorption spectroscopy, in preparation Seres J., A. Müller, E. Seres, K. O Keeffe, M. Lenner, R. F. Herzog, D. Kaplan, Ch. Spielmann, F. Krausz, 2003, Opt. Lett. 28, 1832 Seres J., E. Seres, A. J. Verhoef, G. Tempea, Ch. Streli, P. Wobrauschek, V. Yakovlev, A. Scrinzi, Ch. Spielmann, F. Krausz, 2005 Nature 433, 596 (2005) Stampfli P., K. H. Bennemann, 1992, Phys. Rev. B 46, Stampfli P., K. H. Bennemann, 1994, Phys. Rev. B 49, 7299 Sundram S.K., E. Mazur, 2002, Nature Materials 1, 217 Tempea, G., M. Geissler, M. Schnürer, and T. Brabec, 2000, Phys. Rev. Lett. 84, 4329 Wieland, M., R. Frueke, T. Wilhein, Ch. Spielmann, M. Pohl, U.Kleineberg, 2002, Appl. Phys. Lett. 81, 2520 Wieland M., T. Wilhein, C. Spielmannn, U. Kleineberg, 2003, Appl. Phys. B. 76, 885 Wieland M., Ch. Spielmann, U. Kleineberg, U. Heinzmann, T. Wilhein, 2005, Ultramicroscopy 102, 93 Wilhein, T., B. Kaulich, J. Susini, 2001 Opt. Comm. 193, 19. Zewail A.H. J., 2000, J. Phys Chem A 104, Collaboration within and beyond the SFB Collaboration with P02: In the second funding period we extended the successful collaboration with P02. A major goal of the SFB was the development of the advanced femtosecond source (AFS). At begin we characterized a newly developed broadband pulse shaper (DAZZLER). It is a key component for reaching the design goals. In a next step we completed the AFS and demonstrated sub-10fs pulses with an energy of several mj. This source has been later on used to extend the cut-off of high harmonic generation up to 1keV

10 Report on Project Parts Collaboration with P11: To understand the origin of the extension of the HH cut-off and to obtain guidelines for a further optimization of the source numerical simulations are indispensable. These simulations have been provided by P11. With help of these results it was possible to confirm the observed efficient generation of high harmonics form neutrals and to support the hypothesis of generating harmonics from ions. Collaborations with G. Gerber University of Würzburg: Another route to improve the HH source is adaptive shaping the laser parameters. These experiments have been conducted in close collaboration with Gerber s group in Würzburg. They provided the necessary know-how about evolutionary algorithms and how to design an optimum control experiment. In this cooperation we have employed adaptive control into high harmonic generation, enabling the generation of engineered XUV spectra and opening the way to adaptive control with XUV pulses. The results have been published in several publications. Collaboration with Prof. T. Wilhein University for Applied Science Remagen and PD U. Kleineberg University of Bielefeld: The major aim of this collaboration was the development of the XUV microscope and interferometer. Both setups have been illuminated with the HHG source developed within the SFB. Our partner provided the necessary optical components and the some of the vacuum equipment. In a series of joint experiments we could reach the goals described in the proposal. The results have been published in several publications and presented at international conferences.