Attosecond Control and Measurement: Lightwave Electronics

1 S C I E N T I F I C R E P O R T S 

1 . 3 A T T O S E C O N D A N D H I G H - F I E L D 

P H Y S I C S D I V I S I O N 

Ferenc Krausz 

Max-Planck-Institut für Quantenoptik • Progress Report 2007/2008 131

GROUP MEMBERS 

ASSOCIATED SCIENTIFIC STAFF 

- Prof. U. Kleineberg (LMU 1 ) - S. Becker 

- Dr. Jingquan Lin - A. Henig 

- M. Hofstetter - D. Kiefer 

- Prof. D. Habs - K. Golab 

(Max-Planck Fellow, LMU) - R. Weingartner 

- Dr. U. Schramm - M. Fuchs 

(until 9/2006) 

- Dr. J. Schreiber (Sabbatical from 

2/2008 till 1/2009) 

POSTDOCTORAl STAFF & GUEST RESEARChERS 

- MPQ: 

- M. Centurion - X. Gu - M. Schultze 

- Y. Deng - S. Karsch - M. Siebold 

- R. Ernstorfer - Zs. Major - G. Tsakiris 

- E. Fill - Y. Mikhailova - S. Trushin 

- M. Geissler - A. Marcinkevicius - L. Veisz 

(until 8/2008) (until 12/2006) - T. J. Wang 

- E. Goulielmakis - J. Meyer-ter-Vehn - T. Wittmann 

- F. Grüner - Y. Nomura - H.-C. Wu 

- LMU: 

- A. Apolonskiy - M. Uiberacker 

- J. Fülöp (until 1/2007) (until 4/2008) 

- S. Naumov (until 12/2006) - V. Yakovlev 

- V. Pervak - A. Sugita 

- J. Rauschenberger 

Ph.D CANDIDATES 

- MPQ: 

- I. Ahmad - S. Klingebiel - P. Reckenthäler 

- A. Buck - K. Kosma - S. Rykovanov 

- A. Fernandez* - J. Li* - M. Schultze 

- J. Gagnon - C. Lin* - F. Tavella* 

- T. Ganz - Y. Nomura - A. Tronnier* 

- D. Herrmann - J. Osterhoff - A. Verhoef* 

- B. Horvath - A. Popp - C. Wandt 

- N. Ishii* - I. Pupeza - M. Wen 

- LMU: 

- R. Hörlein - K. Schmid - C. Teisset 

- T. Metzger 

DIPlOMA CANDIDATES 

- MPQ: 

- S. Benavides* - I. Grguras* - W. Schweinberger 

- R. Graf* - B. Reiter* - R. Tautz 

- D. Grupe* - J. Schmidt* 

- LMU: 

- M. Kraus - B. Marx* - A. Wytrykus* 

- A. S. Kruber* 

1 LMU= Ludwig Maximilians Universität 

1 S C I E N T I F I C R E P O R T S 

TEChNICAl 

- MPQ: 

- A. Böswald - H. Haas - A. Raufer 

- G. Brandl - A. Horn - H.-P. Schönauer 

- M. Fischer 

- LMU: 

- S. Herbst 

MANAGEMENT, OUTREACh 

- MPQ: - LMU: - MAP 2 : 

- Th. Naeser - B. Ferus - K. Adler 

- B. Schütz - Ch. Hacken- - C. Reichelt 

- M. Wild berger 

KEY COllABORATORS 

- D. Attwood - K. Kompa - H. Schröder 

(Berkeley) (MPQ) (MPQ) 

- J. Barth - N. Kroo - A. Scrinzi 

(TUM) (Budapest) (TU Wien) 

- J. Burgdörfer - W. Leemans - D. Sutter 

(Vienna) (Berkeley) (Trumpf GmbH) 

- D. Charalambidis - R. Levine - M. Stockman 

(Crete) (Jerusalem) (Atlanta) 

- P. Feulner - M. Lezius - L. Strüder 

(TUM) (MPQ) (HLL) 

- I. Földes - D. Menzel - T. Taira 

(Budapest) (TUM) (IMS, Japan) 

- M. Geissler - R. Moshammer - V. Tarnetsky 

(Belfast) (Heidelberg) (Novosibirsk) 

- T. W. Hänsch - C. Nicolaides - A. V. Tikhonravov 

(MPQ, LMU) (Athens) & M. K. Trubetskov 

- D. Hoffmann - M. Nisoli (Moscow) 

(Aachen) (Milan) - A. Tünnermann 

- M. Hegelich - L. O´Silva (Jena) 

(Los Alamos) (Lissabon) - T. Udem 

- J. Hein - G. G. Paulus (MPQ) 

(Jena) (Jena) - J. Ullrich 

- S. Hooker - G. Pretzler (Heidelberg)) 

(Oxford) (Düsseldorf) - M. Vrakking 

- M. Yu. Ivanov - F. Remacle (Amsterdam 

(Ottawa) (Liège) - M. Zepf 

(Belfast) 

JUNIOR RESEARCh GROUPS 

- Attosecond Dynamics - Dr. R. Kienberger: 

- Dr. A. Cavalieri - M. Fieß - W. Helml 

- Dr. G. Marcus - U. Graf - E. Magerl 

- Attosecond Imaging - Dr. M. Kling: 

- Dr. Th. Uphues - O. Herrwerth - I. Znakovskaya 

- Dr. S. Zherebtsov - A. Wirth 

* received degree during report period; 

2 MAP= Munich Centre for Advanced Photonics 

132 Max-Planck-Institut für Quantenoptik • Progress Report 2007/2008

INTRODUCTION 

The motion of electrons and light wave oscillations 

(being mutually the cause of each other) occur at an 

inconceivable pace, measured in billionths of a billionth 

of a second, in attoseconds (asec) [1]. By drawing on the 

interaction of electrons and light, attosecond science 

[2] aims at measuring, controlling, and exploiting 

these processes: electron motion and light waves. The 

symbiosis of light and electrons is now being extended 

from nature to technology. 

Controlled few-cycle light waves [3] exert a force 

on electrons that is controlled and variable on the 

attosecond scale. Electron motion driven by this force has 

allowed the reproducible generation and measurement 

of isolated attosecond pulses [4,5]. These tools and 

techniques gave birth to a new technology, which we 

dub lightwave electronics [6]. It allows control and 

measurement of the atomic-scale motion of electrons 

and light field oscillations, just as microwave electronics 

permits control and measurement of nanometer-scale 

current and microwave fields. Lightwave electronics has 

spawned experimental attosecond science and advanced 

metrology to the electronic time scale. Steering electron 

motion in molecules [7], real-time observation of the 

electrons’ quantum transitions deep inside atoms [8], 

their escape via tunneling [9], and their atomic-scale 

transport in solids [10] demonstrate the power of the 

new technology. 

Confinement of light energy to within a few wave cycles 

is opening another field of physics, related to ultrastrong 

fields and forces. The power of laser light and the 

strength of its electromagnetic fields grow inversely with 

pulse duration and may access unprecedented levels. 

The resultant light forces ionize matter, drive electrons to 

near the velocity of light – producing relativistic electrons 

– and can even completely separate them from ions, thus 

generating giant electric fields in plasma waves. These 

light and plasma fields can accelerate electrons within 

micro- and milli-meters to energies acquirable only over 

tens to hundreds of meters in conventional accelerators 

[11-14]. Control of the emerging multi-million-volt 

electrons with ultra-strong fields of controlled light 

waveforms will extend attosecond control to relativistic 

electron motion and open a new branch of attosecond 

science: high-field attosecond physics. It may give rise to 

intense attosecond pulses [15] and ultra-bright sources 

of hard X-rays and other energetic particles. 

The mission of the Joint LMU-MPQ Laboratory for 

Attosecond Physics (LAP) affiliated with the MPQ and 

the Department of Physics of the LMU is to develop 

and advance attosecond science, including lightwave 

electronics, for steering and probing the pico- to 

nanometre-scale motion of electrons bound to atomic 

structures, and high-field attosecond physics, for 

1 . 3 . 1 S U M M A R Y O F S C I E N T I F I C A C T I V I T I E S 

controlling with nano- to micrometre precision the 

trajectories of relativistic electrons, and demonstrate 

applications of the new technologies in interdisciplinary 

collaborations. The “critical mass” required for successful 

pursuit of these demanding long-term objectives is now 

emerging thanks to very substantial resources being 

devoted to the research activities in these areas by the 

Max-Planck Gesellschaft (MPG) and by the Ludwig- 

Maximilians-Universität (LMU). 

Whilst the MPG provides invaluable regular funding 

for running LAP as well as for several major projects 

including the Petawatt Field Synthesizer, the LMU is the 

host university of the Cluster of Excellence: Munich- 

Centre for Advanced Photonics – MAP (directors: D. 

Habs and F. Krausz) [16], which sponsors a number of 

LAP projects and supplies LAP with optical components 

of key importance from its newly-created MAP Service 

Center. Moreover, MAP has permitted the establishment 

of several new professorships: W3 (full, tenured): theory 

of relativistic light-matter interactions, W2 (associate, 

tenured track): high-intensity laser technology; W2: 

brilliant X-ray sources; W2: ultrafast spectroscopy on 

surfaces; W2: theory of microscopic electron dynamics. 

Furthermore the successor of Prof. Habs is now being 

appointed at the level of W3, several years before 

the retirement of Prof. Habs, in the area of laserbased 

particle acceleration and high-field science. We 

shall pursue LAP’s long-term research goals in close 

collaboration with these colleagues. Once all these 

groups are established, some 150-200 researchers at 

LAP and the associated MAP-groups will be active in 

the fields of research outlined above, forming a critical 

mass for efficient pursuit of LAP’s many ambitious and 

demanding long-term goals. 

1.3.1 SUMMARY OF RESEARCh ACTIVITIES 

Our research is organized in three main areas: 

• advancing femtosecond technology towards 

higher average or peak powers, towards 

bandwidths spanning an octave and beyond 

and towards stabilizing the carrier-envelope 

phase and thereby the waveform of laser 

pulses 

and with the improved femtosecond tools developing 

and advancing 

• lightwave electronics: attosecond control and 

spectroscopy, for gaining insight into and 

control over the motion of electrons in atomic 

structures 

and 

• high-field attosecond science: relativistic 

electron control, for developing intense 

attosecond sources and ultra-brilliant sources 

of high-energy particles (X-ray photons, 


1 . 3 AT T O S E C O N D A N D H I G H - F I E L D P H Y S I C S D I V I S I O N 

electrons and ions) for a range of applications 

in physics and life sciences. 

Each of these areas builds upon strong synergies 

between theoretical and experimental research and is 

based on a number of collaborations with groups from 

all over the world. The facilities of our LMU-MPQ joint 

laboratory are situated at MPQ and the LMU Physics 

Building at Coulombwall 1. LAP members working at 

both locations closely co-operate, forming a single 

coherent group with weekly meetings, discussions and 

intense communications on a daily basis. 

1.3.1.1 PUShING ThE FRONTIERS OF FEMTOSECOND 

TEChNOlOGY 

Attosecond technology is based upon and has grown 

out of femtosecond technology. Improving femtosecond 

technology towards 

(i) bandwidths approaching, reaching, and 

exceeding an octave, 

(ii) intense pulses at higher (MHz) repetition 

rates, i.e. higher average powers, 

(iii) higher peak powers both at kilohertz as 

well as 10 Hertz repetition rates 

and endowing it with 

(iv) carrier-envelope phase stabilization 

constitute basic requirements for advancing attosecond 

technology. 

Advances (i) and (iv) enable researches to generate 

intense laser pulses with stabilized waveform. Strongfield 

(multi-photon-assisted) interaction of these 

waveform-stabilized pulses with matter has provided 

access to the value of their carrier-envelope phase 

[3], allowing the determination and thereby full 

control of their waveform. We dub the sources of 

these waveform-controlled few-cycle light pulses light 

waveform synthesizers and consider them as the major 

technological pillars of attosecond science. Increasing 

their bandwidth (i) improves attosecond metrology 

by generating shorter isolated attosecond pulses with 

higher efficiency (see next chapter). Increasing their 

pulse repetition rate into the MHz regime (ii) opens 

the door for endowing coincidence spectroscopy with 

attosecond temporal resolution. Last but not least, 

increasing their peak power to the multi-terawatt 

regime (iii) is the prerequisite for extending attosecond 

control to relativistic electrons. 

BROADBAND ChIRPED DIElECTRIC MUlTIlAYER 

MIRRORS 

constitute an enabling technology for all these 

developments: they are the only optical components 

capable of providing dispersion control 

(i) over bandwidths required for shaping optical 

waveforms within the wave cycle, 

(ii) with extremely low loss required for high-averagepower 

MHz-rate femtosecond pulse generation 

in high-Q laser resonators and passive build-up 

cavities, 

(iii) with low amount of material withstanding high 

intensities, making them ideal for tailoring, 

steering and focusing ultrabroad-band, multiterawatt 

pulses. 

Therefore we continue – even some 15 years after its 

invention [18] – our efforts to advance this technology. 

In collaboration with A. V. Tikhonravov and M. K. 

Trubestkov from Moscow State University, Vladimir 

Pervak and Alexander Apolonskiy have developed a 

family of dispersive mirrors for precision dispersion 

control over an octave or more in the visible (VIS), nearinfrared 

(NIR) spectral range for widespread use in LAP’s 

ultrafast laser “workhorses” (see next section). 

With MAP Service Center’s newly-installed electron-beam 

evaporation system, the team has recently succeeded 

in the first realization of a chirped multilayer dielectric 

mirror providing dispersion control over the spectral 

range of 300-900 nm and the first use of hafnium oxide 

in a chirped mirror [19]. Simultaneously, the chirped 

mirror technology is being extended to the mid IR 

(MIR) up to wavelengths of 3 μm. Precision dispersion 

metrology via white light interferometry is also being 

extended to the same spectral range (from MIR to UV). 

These developments allow extension of broadband 

dispersion control to a spectral range spanning several 

octaves, paving the way towards the generation 

of mono-cycle to sub-cycle optical waveforms and 

arbitrary optical waveform synthesis, as well as towards 

frequency combs over the UV-VIS-NIR-MIR spectral 

range. Moreover, hafnium-oxide-based chirped mirrors 

will lead to high-damage-threshold broadband optics 

for high-power applications. 

lIGhT WAVEFORM SYNThESIZERS 

constitute key tools for attosecond science. Thanks to 

ongoing advances in chirped multilayer technology, we 

can now reliably generate intense (sub-terawatt) sub- 

1.5-cycle laser pulses in the VIS/NIR spectral range at 

kHz repetition rates in several workhorse laser systems 

on a daily basis in our laboratory. Figure 1 shows the 

spectrum and interferometric autocorrelation recorded 

by Adrian Cavalieri and coworkers, featuring clean, ~3.5fs-duration 

(full width at half maximum) laser pulses 

carried at a wavelength of ~ 750 nm (field oscillation 

period ~ 2.5 fs) [6,20]. This performance has recently 

been reproduced by Sergei Trushin, Izhar Ahmad and 

co-workers, and by Eleftherios Goulielmakis and Martin 

Schultze using other LAP laser systems, demonstrating 


A 

B 

C 

D 

E 

Figure 1: (A) Photo of the chirped dielectric multilayer 

mirrors providing high reflectivity and dispersion 

control over the spectral range of 450 – 1000 nm. 

(B) Schematic of the hollow-fibre/chirped mirror 

compressor seeded with carrier-envelope-phasestabilized 

25-fs, 1-mJ, near-infrared laser pulses 

from a Ti:sapphire laser system. Spectrum (C) and 

interferometric autocorrelation (D) of sub-3.5-fs laser 

pulses carried at ~ 750 nm. (E) Possible electric field 

waveforms of the generated waveform-controlled 

sub-1.5-cycle laser pulses. 


the maturity of the technology of near-single-cycle 

optical waveform generation. This progress had direct 

implication to attosecond technology as reviewed in the 

next chapter. 

The intense near-mono-cycle NIR pulses offer the 

potential for few-femtosecond pulse generation in the 

deep UV and VUV by direct harmonic conversion, or – 

with their different spectral components independently 

adjustable, in terms of both amplitude and phase – for 

the synthesis of a wide range of optical waveforms. 

Ulrich Graf and Ivanka Grguras, under the supervision 

of Eleftherios Goulielmakis, pursued these goals. Ulrich 

Graf has succeeded in generating sub-4-fs, microjoule 

pulses in the deep UV, at a carrier wavelength of about 

250 nm by third-harmonic generation in a high-pressure 

neon gas cell and is now working on the extension of 

the technique to higher (5 th and 7 th ) harmonics and 

shorter pulse durations. Ivanka Grguras has set up a 

proto-typical 3-channel optical waveform synthesizer 

for the 400-1200nm spectral range using chirped, 

dichroic beam splitters designed, manufactured, 

and characterized by Vladimir Pervak and Alexander 

Apolonskiy. Preliminary frequency-resolved optical gating 

measurements of the signals transmitted through the 

individual channels have yielded bandwidth-limited fewcycle 

pulses, characterization of the overall throughput 

is under way. The programmable optical waveforms, 

once available, are expected to allow versatile steering 

of the atomic-scale motion of electrons in a wide range 

of systems. For example, they provide a large degree 

of freedom for controlling the electron trajectories in 

attosecond pulse production by high-order harmonics, 

offering manipulation of the generation process for 

different purposes (e.g. fine tuning the photon energy 

or optimizing the efficiency of generation). As another 

example, in Figure 2C(iii) a high-frequency wavepacket 

preceds a more slowly oscillating intense waveform, 

A 

Figure 2: (A) Chirped-mirror-based optical waveform 

synthesizer. Generic scheme of a multi-channel 

optical waveform synthesizer based on wide-band 

chirped dichroic beamsplitters (DBS). 


B 

C 

D 


Figure 2: (B) Reflectivity curves of DBSs designed 

and manufactured for a prototypical three-channel 

synthesizer covering the spectral range of 400- 

1200 nm, which was recently generated from a 

neon-gas-filled capillary, see Fig. 1. Channel #1: 

400-550 nm, Channel #2: 550-800 nm, Channel 

#3: 800-1200 nm) designed to support band pass 

reflectance with minimized additional spectral 

phase. (C) Representative waveforms that can be 

synthesized with this prototype optical waveform 

synthesizer, by adjusting amplitude and phase (delay) 

of the individual channels (6 adjustable “knobs”). (D) 

Top-view of the synthesizer illuminated with whitelight 

supercontinuum from the system shown in 

Figure 1. The scheme can be readily scaled to more 

independently-addressable channels (as indicated 

in panel A) and extended towards longer as well as 

shorter wavelength, where chirped multilayer optics 

are now becoming available. 

which may open the door for a new type of reaction 

control: the high-frequency (UV or VUV) prepulse 

selectively creates a localized electronic excitation at 

the selected site of the molecule and the subsequent 

synthesized control wave steers the wavepacket across 

the molecule towards a target bond which is to be 

destroyed to trigger a reaction. 

lIGhT WAVEFORM SYNThESIS AT lONGER 

WAVElENGThS 

affords promise for scaling high-order harmonic 

generation to shorter wavelengths, owing to the 

quadratic scaling of (maximum) harmonic photon 

energy with the wavelength of the driving light pulse. 

This prospect motivated us to extend waveformcontrolled 

light pulse generation to the MIR. To this end 

we have produced broadband MIR light by differencefrequency 

generation from a few-cycle Ti:sapphire laser 

and amplified it by optical parametric amplification 

(OPA). 

136 Max-Planck-Institut für Quantenoptik • Progress Report 2007/2008 

A 

B 

Figure 3: (A) Third-harmonic frequency-resolved 

optical gating spectrogram of the millijoule-scale, 

waveform-controlled mid-IR few-cycle pulses 

generated by a broadband optical parametric 

amplifier. (B) Temporal intensity profile and phase 

of the mid-IR pulse retrieved from the TH-FROG 

measurement, revealing a pulse duration of 15 fs. 

Inset: Measured spectrum of the pulse.

Building upon the preliminary work of Andrius Baltuska, 

Takao Fuji and Nobuhisa Ishii [21], persistent efforts 

of Xun Gu, Yunpei Deng and coworkers have recently 

led to the first intense waveform-controlled sub-twocycle 

light pulses in the MIR, at a carrier wavelength of 

2.1 μm (Figure 3). 

With the high-power diode-pumped disc-laser being 

developed by Thomas Metzger and Wolfram Helml 

as a new pump source for the MIR-OPA, we pursue 

the development of the first terawatt infrared light 

wave synthesizer, which we dubbed LWS-1 (for target 

parameters, see survey of projects and goals). We expect 

LWS-1 to allow us to significantly advance attosecond 

metrology and spectroscopy, both in terms of temporal 

resolution and wavelength coverage. 

lIGhT WAVEFORM SYNThESIS AT RElATIVISTIC 

INTENSITIES 

may allow attosecond control of the trajectories of 

relativistic electrons, which, in turn, holds promise for 

attosecond pulse generation with high intensity and/or at 

high (X-ray) photon energies and for the development of 

compact sources of brilliant X-rays. These prospects and 

the expected applications, which we shall briefly discuss 

in Chapter 1.3.1.3, have motivated us to pursue scaling 

the technology of light waveform synthesis to relativistic 

intensity levels. In 2003, Andrius Marcinkevicius and 

Franz Tavella started developing a 10-Hz near-infrared 

noncollinear optical parametric chirped-pulse amplifier 

(OPCPA), which we dubbed LWS-10, for producing 

few-cycle pulses with a peak power of about 10 TW. 

After the leave of A. M. in 2006, Laszlo Veisz took over 

the project and – in collaboration with Franz Tavella – 

recently demonstrated reliable sub-10-fs, 10-TW pulse 

generation from this system, constituting the world’s 

first multi-terawatt few-cycle light source [22], see 

Figure 4. Simultaneously, Tibor Wittmann and coworkers 

have managed to develop a single-shot carrier-envelope 

phase meter, which will allow to measure and control 

the phase and thereby the waveform of the intense fewcycle 

pulses delivered at a low repetition rate [23]. 

At the same time, Stefan Karsch, together with Zsuzsanna 

Major, Mathias Siebold, Sergei Trushin and several PhD 

students work on the development of a petawattscale 

few-cycle light source, dubbed the Petawatt Field 

Synthesizer (PFS), in the framework of a project funded 

by the Max Planck Society. Once commissioned, PFS 

will constitute the world’s first high-power light source 

delivering 

• few-cycle, waveform-controlled light with 

petawatt-scale peak power 

• petawatt-scale pulses at a 10 Hz repetition 

rate from compact (laboratory-scale) system 


A 

B 

Figure 4: Light wave synthesizer - 10 (LWS-10): 

the world’s first 10-TW sub-10-fs light source. (A) 

Overview of the system occupying approximately 

10 m 2 on an optical table. For details see Ref. 

[22]. (B) Single-shot second-order autocorrelation 

trace routinely recorded during operation shows a 

FWHM correlation width of 10.7 fs, implying a pulse 

duration (FWHM) of 7.6 fs under the assumption of 

a Gaussian pulse shape. 

and expected to open a new chapter in attosecond and 

high-field science. The concept PFS is considered as a 

key technology for the realization of the Extreme Light 

Infrastructure (ELI), a pan-European project aiming at 

the development of an exawatt-scale light source (1 

Exawatt = 10 18 W)*. Thanks to the recent successful 

demonstration of the two main technological pillars 

of the enterprise: (i) the feasibility of ultrashort-pulsedriven 

broadband OPCPA by József Fülöp, Zsuzsanna 

Major, Stefan Karsch and coworkers, see Figure 5 [24], 

and (ii) the feasibility of scaling diode-pumped Yb:YAG 

amplifiers to the Joule energy regime and beyond, by 

Mathias Siebold, Marco Klingebiel, Christian Wandt and 

collaborators from the Univ. Jena [25,26], the second 

phase of the project was now approved and we hope 

that the system can be commissioned before the end 

of 2010. 

* ELI was recently included in the ESFRI roadmap for future large-scale infrastructures in 

Europe and its Preparatory Phase has recently been funded by the EU. 



Figure 5: Ultrashort-pulse-driven OPCPA test system using ATLAS. (A) Pulse-front tilt matching of pump and 

signal pulses. Maximum gain bandwidth (B) and corresponding transform limited pulse duration (C). Spectrum 

transmitted by our chirped-mirror compressor (D) along with pulse reconstructed from SPIDER measurements (E). 

lIGhT WAVEFORM SYNThESIS AT MEGAhERTZ 

REPETITION RATES 

affords promise to produce attosecond pulses at MHz 

repetition rate, which, in turn, would open the door for 

combining attosecond spectroscopy with coincidence 

detection, a long-standing dream of researcher in the 

field of attosecond science. By inventing the concept of 

chirped-pulse oscillators (CPO) our group has previously 

shown how to scale the energy and peak power of 

femtosecond pulses from mode-locked oscillators by 

lowering their repetition rate from 50-100 MHz by an 

order of magnitude [27]. Recently, a collaborative effort 

of Akira Ozawa, Thomas Udem (from the group of Prof. 

T. Hänsch) and Jens Rauschenberger, Alma Fernandez 

and Alexander Apolonskiy from our group coupled 

sub-microjoule energy pulses from our Ti:sapphire 

chirped-pulse oscillator into a passive build-up cavity to 

generate high-order harmonics up to photon energies of 

~ 30 eV (wavelength ~ 40 nm) with average powers of 

about one microwatt [28]. Implementing the concept 

with few-cycle light, which we currently pursue, will 

allow extension of frequency combs to the soft-X-ray 

regime and attosecond pulses at MHz repetition rates 

for attosecond coincidence spectroscopy. 

1.3.1.2 lIGhTWAVE ElECTRONICS: ATTOSECOND 

CONTROl & SPECTROSCOPY 

At the time of the previous progress report, extreme 

ultraviolet (XUV) pulses of a duration of 250 as at a 

photon energy of about 100 eV, containing some 10 7 

photons/pulse and delivered at a 1 kHz repetition rate 

have represented the state of the art of attosecond 

technology [4]. In our newly-developed AS-1 

attosecond beamline [6], these sub-femtosecond pulses 

along with their few-cycle NIR drivers have served as 

a standard tool for attosecond real-time interrogation 

of atomic-scale electron dynamics over the past two 

years. Simultaneously, the improved NIR waveforms 

reported above have been used to push the frontiers of 

attosecond pulse generation and metrology. 

ATTOSECOND SPECTROSCOPY 

has – in its first implementation [8] – drawn on 

electrons leaving the atom after their excitation. Intraatomic 

motion has been explored by probing outgoing 

electrons. Probing the transient population of shakeup 

states by means of strong-field-induced tunnelling 

ionization offers another means of interrogating 

intra-atomic electron dynamics (Figure 6). If lightfield-induced 

electron tunnelling occurs as predicted 

by Keldysh some four decades ago, the probability of 

setting the electron free from these shake-up states 

increases in sub-femtosecond steps near the oscillation 

peaks of the NIR laser field, where the binding potential 

is suppressed. Matthias Uiberacker, Martin Schultze and 

Thorsten Uphues, with collaborators from the group of 

Karl Kompa, from Bielefeld and Amsterdam, have now 

experimentally verified this long-standing theoretical 

prediction, the cornerstone of strong-field theories, 

by ionizing neon atoms with a 250-as X-ray pulse and 

removing the shake-up electron from neon ions by a 


Figure 6: Electronic excitation and relaxation in 

atoms, molecules and solids. Motion is initiated by 

an attosecond XUV pulse and measured by sampling 

the temporal evolution of outgoing electrons via 

attosecond streaking spectroscopy or by probing the 

transient population of bound states via attosecond 

tunnelling spectroscopy. Both methods are 

implemented with a waveform-controlled few-cycle 

NIR pulse synchronised with attosecond accuracy 

to the XUV trigger pulse. The labels +1 and +2 

indicated single and double ionization. 

time-delayed few-cycle NIR field [9]. The measurement 

has yielded a temporal ionization profile revealing 

sharp steps approximately half a laser cycle apart, just 

as predicted by Keldysh 40 years ago. Each sub-cycle 

ionization step was measured to last less than 380 

attoseconds, setting a corresponding upper limit on 

shake up and tunnelling. 

This sub-femtosecond ionization step offers an 

attosecond probe for interrogating the transient 

population dynamics of excited (bound) states in atoms, 

molecules or solids following a sudden excitation. 

We have termed this novel technique attosecond 

tunnelling spectroscopy and used it for probing multielectron 

dynamics such as shake up and single as well 

as cascaded Auger processes for the first time with 

attosecond resolution. In these experiments [9] we not 

only confirmed several results previously obtained by 

frequency-domain measurements but also shed light 

on dynamics that could not be accessed by frequencydomain 

techniques so far. 


Until recently, attosecond spectroscopy was restricted 

to isolated atoms in the gas phase. Can it possibly 

be extended to condensed-matter systems: to solids 

and surfaces? In an international collaborative effort, 

led by Adrian Cavalieri and coordinated by Reinhard 

Kienberger, with important contributions from Bielefeld, 

San Sebastian, and Vienna, we have been able to answer 

this question affirmatively, by attosecond streaking of 

photoelectrons liberated from a tungsten crystal by a 

300-as XUV pulse [10]. We have studied two different 

types of electrons, liberated from localized core states 

of the 4f band and the delocalized states of the 

conduction band near the Fermi energy. The recorded 

streaking spectrograms, the artistic representations of 

Figure 7: Artistic representation of streaking 

spectrograms of photoelectrons, released by a 300attosecond 

95-eV XUV pulse from a tungsten crystal 

and recorded by a 5-fs, 750-nm NIR laser pulse. 

“Electrons move in solids at very high speed – 

traversing atomic layers and interfaces within tens 

to hundreds of attoseconds. These time intervals 

will ultimately limit the speed of the electronics 

of the future. Physicists have now experimentally 

probed such electron dynamics in real time. The 

cover illustrates the first attosecond spectroscopic 

measurement in a solid, revealing a 110-attsecond 

difference in the travel time of two different types 

of electrons following photoexcitation in a tungsten 

crystal. The ability to time electrons moving in solids 

over merely a few interatomic distances makes it 

possible to probe the solid-state electronic processes 

occurring at the ultimate speed limit and thus helps 

to advance technologies such as computation, data 

storage and photovoltaics,which all rely on exquisite 

control of electron transport in ever smaller structures 

of solid matter. [Cover art: Barbara Ferus, MPQ/LAP]” 

Description of cover, from the 25 October 2007 issue 

of Nature, page ix. 



which are shown in Figure 7, indicate sub-femtosecond 

emission time for both types of electrons. Closer 

inspection has yielded a discernible temporal shift 

between the two streaking spectrograms: careful data 

analysis reveals a delay in emission of the core electrons 

with respect to the conduction-band electrons of about 

110 attoseconds. This means that the former reach the 

surface – from the several-Angstrom depth which they 

can escape from – some 110 as later than the latter. The 

analysis of Pedro Echenique has revealed that only half 

of this delay can be accounted for by the lower kinetic 

energy of core electrons, the other half is attributed to 

a larger effective mass of these electrons. This proof-ofconcept 

study opens the door for direct time-domain 

access to a wide range of hyperfast electron processes 

in solids and surfaces, such as e.g. charge transfer in 

host-guest systems or within molecular assemblies on 

surfaces, charge screening and e-e scattering in metals, 

or collective motion in surface plasmons. 

After being successfully demonstrated in atoms and 

solids, attosecond spectroscopy now remains to be 

extended to molecular systems. As a matter of fact, 

the valence electronic states are typically separated by 

one or more electronvolts, indicative of a hyperfast 

motion of electron wavepackets composed of such 

states. The few-femtosecond-duration broadband 

deep UV and VUV pulses reported in Chapter 1.3.1.1 

lend themselves to launching electron wavepackets 

on molecular orbitals with substantial probability, 

whilst synchronized attosecond XUV pulses would 

ideally suit for probing the unfolding electronic and 

structural dynamics in molecules. Both of these tools 

are derived from waveform-controlled few-cycle NIR 

pulses, our basic tool, which ensures attosecond 

synchronism between the deep-UV/VUV and XUV 

pulses. In order that these tools become available for 

real-time observation of hyperfast electronic motion 

as well as complex, intertwined electronic and nuclear 

dynamics in molecules, Marcus Fieß, Wolfram Helml, 

and Eleftherios Goulielmakis – under the guidance of 

Reinhard Kienberger – have developed a next-generation 

attosecond beamline: AS-2, which accommodates two 

frequency-conversion assemblies, for attosecond XUV 

and few-femtosecond UV/VUV pulse generation. Both 

beams can be recombined on target with attosecond 

timing accuracy. 

Encouraged by the successful extension of attosecond 

spectroscopy to solids [10], a team led by Reinhard 

Kienberger: Elisabeth Magerl, Adrian Cavalieri, and Ralph 

Ernstorfer – in close cooperation with our MAP partners: 

Peter Feulner, Dietrich Menzel and Johannes Barth from 

the Technische Universität München, are developing 

the world’s first ultrahigh vacuum beamline for 

attosecond surface and solid-state spectroscopy: AS-3. 

We are hopeful that measurement in AS-2 and AS-3 

will make several contributions to the highlights of our 

forthcoming progress report. 

ADVANCING TIME-RESOlVED METROlOGY INTO 

ThE SUB-100-AS DOMAIN 

is indispensable for direct time-domain insight into 

electron correlations on the atomic scale. These govern 

or affect such fundamental processes as the intraatomic 

energy transfer between electrons, the response 

of atomic electron systems to external influence, or its 

rearrangement following the sudden loss of one or more 

electrons. Detailed insight into the electronic response 

of atomic-scale solid-state structures to strong external 

fields of infrared or visible light, e.g. via non-adiabatic 

tunneling is required for the development of solid-state 

lightwave electronics and will also rely on sub-100-as 

temporal resolution. 

By confining optical field ionization to a single, wellcontrolled 

light oscillation period, the sub-1.5-cycle NIR 

laser pulses described in Chapter 1.3.1.1 along with 

novel broadband XUV multilayer mirrors developed 

by Michael Hofstetter and Ulf Kleineberg have recently 

permitted our team, Eleftherios Goulielmakis, Martin 

Schultze and coworkers, to generate isolated sub-100as 

pulses of XUV light [29], the measured XUV pulse 

profile and the NIR electric field waveform are shown 

in Figure 8. Detailed evaluation of the measurement 

has been performed with a new algorithm developed 

by Justin Gagnon and Vladislav Yakovlev [30]. The 

quasi-monocycle driving field benefits attosecond pulse 

generation in several ways. Abrupt onset of ionization 

within a single half cycle minimizes the backgroundfree 

electron density at the instant of harmonic pulse 

emission, minimizing thereby distortion of the driving 

wave and its dephasing with the generated harmonics 

during propagation. Hence coherent build-up of the 

harmonic emission over extended propagation is 

maximized. In addition, order-of-magnitude variation of 

the ionization probability between adjacent half-cycles 

creates unique conditions for single sub-100-as pulse 

emission without the need for sophisticated gating 

techniques. 

On the measurement side, improved resolution results 

from three (almost equally) important advances: (i) 

shorter XUV pulse duration, (ii) higher signal-to-noise 

ratio (S/N) and/or shorter overall measurement time due 

to the increased XUV photon flux, and (iii) the feasibility 

of stronger streaking before the onset of the NIR-fieldinduced 

ionization in attosecond streaking or enhanced 

S/N due to reduced number of tunnelling steps in 

attosecond tunnelling spectroscopy. 

Consequences of these favourable conditions include the 

routine generation of isolated sub-100-as XUV pulses 

(at a carrier photon energy of about 80 eV), for details 



Figure 8: High-fidelity measurement of an isolated sub-100-attosecond pulse and its quasi-mono-cycle 3.3-fsduration 

(FWHM) driver laser waveform allows precision tests of basic models of strong-field interactions for the 

first time. Isolation of a single attosecond pulse means isolation of a single tunnelling and recollision event and 

wavepacket motion between these the instants of ionization and recollision within an oscillation period of the 

driving laser field (shown in red in panel A) that is precisely known from an attosecond streaking measurement. 

A simple analysis of classical electron trajectories between ionization and recollision (black lines in panel A) in the 

framework of the semiclassical model of Paul Corkum implies a positive and negative chirp of the emitted XUV 

pulse for trajectories shorter and longer than the one resulting in highest-energy emission (green line in panel B). 

As an example of the power of high-fidelity attosecond metrology, we have calculated this intrinsic spectral chirp, 

i.e. the variation of group delay versus photon energy, carried by the attosecond pulse during its emergence, from 

the chirp measured by attosecond streaking and the known dispersion of the metal filter traversed by the pulse on 

its way from the source to the measurement. The degree of agreement of the measured chirp with the prediction 

of the semiclassical model is stunning, indicating that the emission from long trajectories is completely suppressed 

in the far field by spatial filtering, and secondly classical electron trajectory analysis provides a powerful tool for 

predicting the outcome of strong-field interactions. 

see Selected Reprint 11) “Single-cycle nonlinear optics” 

on pages 185 - 188, with a flux of greater than 10 11 

photons/s, exceeding by several orders of magnitude 

the flux of previously-reported sources of sub-fs light; 

a sub-10% measurement accuracy of the XUV pulse 

parameters and the NIR field evolution, suggesting that 

the temporal resolution of this pump-probe sampling 

system may approach or even reach the atomic unit of 

time (~ 24 as). 

OUR lONG-TERM OBJECTIVES 

Attosecond metrology includes the scaling of attosecond 

pulse generation towards higher photon energies 

and combination of high temporal with high spatial 

resolution in spectroscopy. 

Vladislav Yakovlev has analytically as well as numerically 

studied the generation of soft-X-ray harmonics with 

few-cycle waveforms of wavelength, λ L , varied from the 

near to the mid IR [31]. His studies predict that selfphase-matching 

becomes more efficient for longer 

wavelengths and under specific circumstances can 

provide ideal conditions for coherent growth of soft- 

X-ray harmonics over extended propagation lengths. 

Enhanced phase matching may thereby overcompensate 

the decrease in single-atom emission intensity for longer 

Figure 9: Spectral intensity distribution of soft-X-ray 

harmonics generated with a two-cycle laser pulse of 

a carrier wavelength ranging from 0.75 μm to 3 μm. 

λ L , and may result in orders of magnitude enhancement 

of the harmonic yield (see e.g. in Figure 9 harmonics in 

the 400 eV and 700 eV range for λ L = 2.1 μm and 

λ L = 1.5 μm, respectively), as well as efficient filtering of 

isolated attosecond pulses by restricting phase matching 

to a limited spectral band (in the example shown in 

Figure 8: 350-450nm and 600-700eV for λ L = 2.1 μm 

and λ L = 1.5 μm , respectively). 



In another study, Vladislav Yakovlev has investigated the 

possibility of coherent superposition of laser-driven SXR 

harmonics in successive sources by one and the same 

laser pulse. His prediction of the feasibility of quasi- 

A 

B 

Figure 10: High-order harmonic generation from 

successive sources. (A) Schematic illustration of a 

sequence of harmonic sources formed by gas jets. 

The sources are arranged along the optical axis of 

a focused laser beam and pumped by one and the 

same laser pulse. In our first proof-of-concept study 

we have employed only two sources (depicted in 

black), further ones can be added for scaling the 

SXR harmonic yield in future experiments (depicted 

in grey). (B) Increasing the density of atomic dipole 

emitters tends to increase the coherent harmonic 

yield. However, increasing density causes the dipole 

oscillators to increasingly dephase along the laser 

propagation direction. In a single source, this leads to 

saturation of the harmonic yield at an atomic density 

corresponding to a backing pressure of ~ 40 mbar, 

red curve (calculated) and squares (measured). 

Splitting the generation medium into two equal 

sections and moving them apart so that the phase 

of the atomic dipole oscillations gets shifted by in 

the focused laser beam allows the atomic density to 

be increased by a factor of two (at backing pressure 

of ~ 80 mbar), leading to saturation at a factor of 

four higher harmonic intensities (blue curve and 

diamonds). 

phase-matched SXR harmonic generation by a focused 

laser beam in a gas medium of modulated density 

has meanwhile been confirmed by our experimental 

observation of constructive and destructive interference 

between the harmonic signals originating from two 

successive sources, see Figure 10 [32]. Our proof-ofconcept 

study opens the prospect of enhancing the 

photon flux of SXR harmonic sources to levels enabling 

researchers to tackle a range of applications in physical 

as well as life sciences. 

Should attosecond pulse generation be scalable to 

photon energies of several kiloelectronvolts, attosecond 

X-ray diffraction might allow, one day, 4-dimensional 

microscopy with attosecond temporal and picometre 

spatial resolution. Drawing on the pioneering work of 

Ahmed Zewail, we also pursue this long-term goal by 

advancing ultrafast electron diffraction. To this end, 

Ernst Fill, Laszlo Veisz, and Sasha Apolonskiy have 

presented a novel concept of an electron gun for 

generating few-femtosecond- to sub-femtosecondduration 

electron bunches [33]. The basic idea is 

to utilize a DC acceleration stage combined with a 

microwave cavity, the time-dependent field of which 

generates an electron energy chirp for bunching at the 

target. To reduce space charge broadening, the number 

of electrons in the bunch is reduced and the gun is 

operated at a MHz repetition rate. In the absence of 

space charge (one electron per bunch), the duration 

of the generated electron wavepacket may potentially 

be shortened below 1 fs. The team around Ernst Fill, 

Martin Centurion and Peter Reckenthäler, have also 

devised a technique for measuring the bunch duration 

with potentially attosecond resolution [34] and have 

also performed preliminary time-resolved experiments 

with (multi-)electron bunches, which allowed them to 

image dynamics changes in the density distribution of a 

laser-induced plasma [35]. 

1.3.1. 3 hIGh-FIElD ATTOSECOND SCIENCE: RElA- 

TIVISTIC ElECTRON CONTROl 

As described in Chapter 1.3.1.1, we pursue scaling of 

waveform-controlled few-cycle laser pulse generation to 

relativistic intensities in the expectation that controlling 

the motion of relativistic electrons with attosecond 

precision in time and nano- to micrometre precision in 

space paves the way towards scaling attosecond pulse 

generation to higher flux and/or photon energies and 

may allow the development of compact, laboratory 

sources of brilliant X-rays. Simultaneously with our 

efforts aiming at laser development, in collaboration 

with Dieter Habs and his group, we are also conducting 

relevant proof-of-concept studies drawing on our 

laser sources currently available. Two routes are being 

pursued at MPQ: the development of a compact X-ray 

free electron laser seeded by laser-accelerated electron 

bunches and coherent harmonic XUV/X-ray pulse 

generated from relativistically-driven surfaces. In this 


chapter we summarize first promising results of these 

proof-of-concept experiments. 

lASER-DRIVEN ElECTRON ACCElERATION 

affords promise for the development of compact, 

laboratory-sized X-ray free electron lasers. Simulations 

of our collaborators, Florian Grüner, Stefan Becker, 

Figure 11: Simulation results of Michael Geissler and 

Jürgen Meyer-ter-Vehn showing the physical state 

of MPQ’s few-cycle-laser-driven plasma accelerator 

after the laser pulse has travelled a distance of 

130 μm inside the plasma. (A) Electron density 

(grey-scale false-colour plot) and the instantaneous 

laser intensity (rainbow-scale false-colour diagram). 

It is evident that the laser pulse fits within the 

bubble radius. The red line shows the longitudinal 

accelerating field caused by charge separation in the 

generated plasma wave (forward-directed field has 

positive value). The field strength is about 0.45 TV/m 

at the location of the bunch. (B) Lineouts of electron 

density and instantaneous laser intensity along the 

optical axis of the laser beam. The injected electron 

bunch (red) contains 4.5 pC charge and has an order 

of magnitude higher density than the background 

plasma. There is no direct interaction between the 

laser pulse and the electron bunch during its forward 

acceleration in the plasma “bubble”. 


Dieter Habs and coworkers, indicate that seeding lasergenerated, 

ultrahigh-peak-current, GeV-energy electron 

bunches into a compact, several-metre-long undulator 

may allow – one day – the realization of X-ray laser 

operation, producing ultra-brilliant, few-femtosecond 

X-ray pulses of several-Angström wavelength [36]. 

Comprehensive 3-dimensional PIC simulations of Michael 

Geissler, Jürgen Meyer-ter-Vehn (MPQ) and Alexander 

Pukhov (Düsseldorf) have predicted the feasibility of 

making electron bunches with the required parameters 

available with laser-driven plasma accelerators. 

A promising implementation relies on “broken” plasma 

waves driven by a laser pulse shorter than the half 

plasma oscillation period (Figure 11). 

In the absence of intense laser pulses of the required 

duration, first experiments drew on longer (several- 

10-fs) driver pulses. Under these circumstances, 

mono-energetic electron acceleration is preceded 

by a nonlinear interaction of the laser pulse with the 

Figure 12: Typical spectra of sub-10-fs-laser-driven 

mono-energetic electron beams free from thermal 

background. (A) Spectra with mean energies of 13.4, 

17.8 and 23 MeV, and relative energy spreads of 

11%, 4.3% and 5.7% (FWHM) respectively. (B) Results 

obtained with a spectrometer capable of accessing 

electron energies down to the non-relativistic 

regime, confirming the absence of an exponential 

background down to energies in the keV range. Here 

the mean energies and the relative energy spread 

are 4.1 and 9.7 MeV, and 14% and 9.5% (FWHM), 

respectively. 



Figure 13: Gigaelectronvolt-scale laser wake field accelerator at MPQ. G: GRENOUILLE pulse diagnostics, C1-C5: 

cameras; D1, D2: diodes; L1, L2: scintillating screens; S: spectrometer; W: wedge; OAP: off-axis parabola; CAP: 

capillary waveguide. The inset shows the incoming laser focus at the capillary entrance. 

relativistic plasma, which shortens its duration into the 

required domain. Longer-than-optimal driver pulses 

compromise efficiency as well as reproducibility, and 

result in copious amounts of low-energy electrons 

accompanying the mono-energetic emission with an 

exponentially-decaying spectrum, forming a “thermal” 

background. By using the multi-terawatt, sub-10-fs 

pulses from LWS-10, Karl Schmid, Laszlo Veisz and 

coworkers, in close cooperation with Ulrich Schramm, 

Dieter Habs and collaborators from the Univ. Düsseldorf 

have recently demonstrated the first electron accelerator 

based on high-density plasma waves driven with 

laser pulses that fit in one half of the plasma period 

(Figure 11) [14]. Direct “impulsive” excitation of the 

plasma wave permits mono-energetic electron 

acceleration virtually free from a thermal background 

for the first time. In our experiments, 5-terawatt, 8femtosecond 

laser pulses yield electron bunches up to 

energies of 25 MeV (Figure 12). The flux of low-energy 

electrons dramatically reduced as compared to earlier 

experiments also manifests itself in a strongly-reduced 

secondary radiation emerging from the accelerator and 

offers the potential for enhancing efficiency and stability 

with more intense driver pulses. 

The electron energy can be further boosted by laser-driven 

plasma-wake-field acceleration (LWFA) in a discharge 

capillary to the gigaelectronvolt range, as demonstrated 

by Wim Leemans and coworkers at Berkeley in 2006. 

Jens Osterhoff, Antonia Popp, Zsuzsanna Major, Stefan 

Karsch and coworkers, in close co-operation with Dieter 

Habs and his group, have recently also demonstrated 

LWFA to the GeV frontier from a discharge capillary 

waveguide provided by Simon Hooker from Oxford 

University by driving the accelerator with 0.75 J, 40fs, 

800-nm pulses from ATLAS, MPQ’s multiterawatt 

Ti:Sa laser, see Figure 13 [12]. 

Quasi-monoenergetic electron beams with energies as 

high as 500 MeV have been detected, with features 

reaching up to 1 GeV, albeit with large shot-to-shot 

fluctuations. The beam divergence and the pointing 

stability in this case are around or below 1 mrad and 

8 mrad, respectively. These shot-to-shot electron-bunch 

parameter variations are greatly suppressed by utilizing 

the capillary as a gas-vessel without pre-ionization by an 

external electrical discharge. In this regime of operation 

quasi-monoenergetic electron beams of up to ~200 

MeV in energy have been observed (Figure 14). These 

beams emitted within a low-divergence cone of mrad 

FWHM display unprecedented shot-to-shot stability 

in energy (2.5% RMS), pointing (1.4 mrad RMS) and 

charge (16% RMS) owing to a highly reproducible gasdensity 

profile within the interaction volume (Figure15) 

[13]. The excellent stability of bunch parameters affords 

promise for the potential ability of laser-accelerated 

electron bunches to meet the stringent criteria for 

seeding free electron lasers. 

INTENSE ATTOSECOND XUV/SXR PUlSES 

might allow triggering as well as probing ultrafast 

electron dynamics with attosecond pulses, i.e. 

attosecond XUV-pump/XUV-probe spectroscopy, 

which has not been possible so far. Relativistic laserplasma 

interactions have been identified as a promising 

approach to achieving this goal. Recent experiments 

confirmed that relativistically-driven overdense plasmas 

are able to convert infrared laser light into harmonic 

XUV radiation with unparalleled efficiency and that the 

generation technique is scalable towards hard X-rays. 

In a recent experiment (Figure 15) Yutaka Nomura, 

Rainer Hörlein, Sergey Rykovanov, George Tsakiris and 

coworkers, in collaboration with the teams of Matt Zepf 

from Belfast and Dimitris Charalambidis from Heraklion, 

have recently succeeded in obtaining conclusive 

experimental evidence for the phases of the XUV 

harmonics emanating from the interaction that they are 



Figure 14: (A) False-color images of 40 consecutive, spatially-dispersed electron beams on the fluorescence screen 

L2 behind the permanent dipole magnet (see Figure 13) recorded at a plasma density of n e ≈ 7.3 x 10 18 cm -3 . 

Each shot is normalized to its maximum value. (B) Five sample electron spectra from the same data set. (C) Five 

exemplary spectra for an electron density of n e ≈ 6.8 x 10 18 cm -3 . Ten consecutive fluorescence images obtained 

under those conditions are presented in the inset. 

synchronized to allow attosecond temporal bunching 

(Figure 16) [15]. Along with the previous findings about 

energy conversion and recent advances in high-power 

laser technology, our experiment demonstrates the 

feasibility of confining unprecedented amounts of light 

energy to within less than a femtosecond, initiating 

thus the symbiosis of relativistic high-field science and 

attosecond science. 

1.3.1.4 VISIONS: SPACE-TIME IMAGING & BIO- 

MEDICAl APPlICATIONS 

Attosecond spectroscopy has enabled us to observe in 

real time how electrons undergo quantum transitions. 

However, in these experiments, we could not obtain 

information about how the electrons move in space. 

The atomic-scale density distribution of electrons can 

Figure 15: Fluorescence screen images of the radial intensity profile of the accelerated electron beam at the 

entrance of the dipole spectrometer (L1 in Figure 13). (A) Summed signal of 74 electron beams and their individual 

centers at n e ≈ 7.8 x 10 18 cm -3 . RMS shot-to-shot pointing fluctuations under these conditions are 1.4 mrad (yaxis) 

and 2.2 mrad (x-axis), respectively. (B) The signal of a single electron beam with a FWHM divergence of 1.6 

mrad. The crosses in the background originate from markings on the screens. 



Figure 16: Surface harmonic generation from relativistic 

plasmas and its temporal characterization. 

The p-polarized laser beam is focused onto a 

rotating disk-shaped target. The light reflected in 

the specular direction is re-collimated by a parabolic 

mirror and deflected towards a glass plate. The 

reflection off the glass plate close to Brewster’s 

angle (~57°) suppresses the p-polarized IR laser light 

while reflecting a substantial fraction (~5% in p-polarization) 

of the XUV harmonic emission. Passage 

through a 150-nm In filter selects harmonics greater 

than H8 and suppresses low-order harmonics as 

well as the residual laser light. The beam of selected 

harmonics (from H8 to H14) is focused by a spherical 

mirror split into two halves, serving as a focusing 

wave-front divider. The insets show data of a typical 

mass (upper) and harmonic (lower) spectrum. 

be resolved in space by conventional microscopic and 

X-ray or electron diffraction techniques, but they can do 

so only if the system is at rest. Our long-term research 

goal is to combine sub-atomic resolution in space and 

in time by developing sources of sub-femtosecond, 

Angström-wavelength electron and photon pulses. By 

recording a series of freeze-frame sub-femtosecond 

diffraction images of electron distributions by means 

of attosecond electron or X-ray diffraction, we will be 

able to observe atomic-scale electronic motion in matter 

with sub-atomic (picometre and attosecond) resolution 

in space and time. Once this dream comes true, we shall 

be able to make movies of all dynamics of the microcosm 

outside the atomic core. 

Our pursued laser-based high-energy particle and 

photon sources also open up exciting prospects for 

applications in life sciences. Ultra-brilliant, femtosecond 

X-ray pulses from a future laser-driven XFEL may permit 

determination of the atomic structure of biomolecules 

without the need for crystallization by recording 

flash X-ray diffraction images of single molecules. 

Furthermore, collimated, laser-driven beams of mono- 

Figure 17. Nonlinear volume autocorrelation of the 

coherent XUV beam comprising harmonics H8 to 

H14. The red dots represent He + signal produced by 

two-photon ionization while the blue dots depict 

H 2 O + signal resulting from single-photon ionization. 

The H 2 O signal level relative to the He + signal is 

scaled arbitrarily. (A) Coarse scan over a delay interval 

corresponding to the laser pulse duration. A Gaussian 

pulse fit (green dashed line), yields the overall 

duration of the XUV emission of T XUV = 44±20 fs. 

(B) A fine scan taken near zero delay with a delay 

step size of 133 as, corresponding to about 20 data 

points per laser cycle. The dashed green line is a fit 

to the raw data of a sequence of Gaussian pulses 

of τ XUV = 0.9 fs in duration to the second-order XUV 

autocorrelation signal. 

energetic protons/ions and X-rays affords promise for 

opening a new chapter in cancer therapy and diagnosis, 

respectively. One of the main long-term objectives of 

the Munich-Centre for Advanced Photonics (MAP) [16] 

initiated and led by Dieter Habs and Ferenc Krausz is to 

bring together the requisite brain power and expertise 

for developing these novel photon-based ultra-brilliant 

electron/proton/ion/X-ray sources and demonstrating 

their suitability for atomic-resolution bioimaging, 


high-resolution, low-dose cancer diagnostics and costeffective 

cancer therapy. 

REFERENCES 

[1] M. Hentschel et al.; Attosecond metrology. Nature 

414, 509 (2001). 

[2] P. B. Corkum and F. Krausz; Attosecond science. 

Nature Phys. 3, 381 (2007). 

[3] A. Baltuska et al.; Attosecond control of electronic 

processes by intense light fields. Nature 421, 611 

(2003). 

[4] R. Kienberger et al.; Atomic transient recorder. 

Nature 427, 817 (2004). 

[5] E. Goulielmakis et al.; Direct measurement of light 

waves. Science 305, 1267 (2004). 

[6] E. Goulielmakis et al.; Attosecond control and 

measurement: lightwave electronics. Science 317, 769 

(2007). 

[7] M. F. Kling et al.; Control of electron localization in 

molecular dissociation. Science 312, 246 (2006). 

[8] M. Drescher et al.; Time-resolved atomic inner-shell 

spectroscopy. Nature 419, 803 (2002). 

[9] M. Uiberacker et al.; Attosecond real-time 

observation of electron tunnelling in atoms. 

Nature 446, 627 (2007). 

[10] A. L. Cavalieri et al.; Attosecond spectroscopy in 

condensed matter. Nature 449, 1029 (2007). 

[11] A. Pukhov and J. Meyer-ter-Vehn; Laser wake field 

acceleration: the highly non-linear broken-wave regime. 

Appl. Phys. B 74, 355 (2002). 

[12] S. Karsch et al.; GeV-scale electron acceleration in 

a gas-filled capillary discharge waveguide. New J. Phys. 

9 415 (2007). 

[13] J. Osterhoff et al.; Generation of Stable, Low- 

Divergence Electron Beams by Laser Wakefield 

Acceleration in a Steady-State-Flow Gas Cell. Submitted 

to Physical Review Letters. (2008). 

[14] K. Schmid et al.; Few-Cycle-Laser-Driven Electron 

Accelerator. Submitted to Nature Physics (2008). 

[15] Y. Nomura et al.; Attosecond phase locking of 

harmonics emitted from laser-produced plasmas. 

Submitted to Nature Physics (2008). 

[16] www.munich-photonics.de 

[17] G. Paulus et al.; Measurement of the phase of fewcycle 

laser pulses. Phys. Rev. Lett. 91, 253004 (2003). 

[18] R. Szipöcs, K. Ferencz, Ch. Spielmann, F. Krausz; 

Chirped multilayer coatings for broadband dispersion 

control in femtosecond lasers. Opt. Lett. 19, 201 

(1994). 

[19] V. Pervak, F. Krausz, A. Apolonski; Dispersion 

control over the UV-VIS-NIR spectral range with HfO 2 / 

SiO 2 chirped dielectric multilayers. Opt. Lett. 32, 1183 

(2007). 

[20] A. L. Cavalieri et al.; Intense 1.5-cycle near infrared 


laser waveforms and their use for the generation of 

ultra-broadband soft-x-ray harmonic continua. New J. 

Phys. 9, 242 (2007). 

[21] T. Fuji et al.; Parametric amplification of few-cycle 

carrier-envelope phase-stable pulses at 2.1 μm. Opt. 

Lett. 31, 1103 (2006). 

[22] F. Tavella et al.; Dispersion management for a sub- 

10-fs, 10 TW optical parametric chirped-pulse amplifier. 

Opt. Lett. 32, 2227 (2007). 

[23] T. Wittmann et al.; Single-shot measurement of 

the carrier-envelope phase. Submitted to Nature Phys. 

(2008) 

[24] J. A. Fülöp et al.; Short-pulse optical parametric 

chirped-pulse amplification for the generation of highpower 

few-cycle pulses, New J. Phys. 9, 438 (2007). 

[25] M. Siebold et al.; High-energy, diode-pumped, 

nanosecond Yb:YAG MOPA system. Opt. Exp. 16, 3675 

(2008). 

[26] Ch. Wandt et al.; Generation of 220 mJ nanosecond 

pulses at a 10 Hz repetition rate with excellent beam 

quality in a diode-pumped Yb:YAG MOPA system. Opt. 

Lett. 33, 1111 (2008). 

[27] S. Naumov et al.; Approaching the microjoule 

frontier with femtosecond laser oscillators. New J. Phys. 

7, 216 (2005). 

[28] A. Ozawa et al.; High harmonic frequency combs 

for high resolution spectroscopy. Phy. Rev. Let. 100, 

253901 (2008). 

[29] E. Goulielmakis et al.; Single-cycle nonlinear optics. 

Science 320, 1614 (2008). 

[30] J. Gagnon, E. Goulielmakis, V. S. Yakovlev; The 

accurate FROG characterization of attosecond pulses 

from streaking measurements., Accepted for Publication 

in Appl. Phys. B (2008). 

[31] V. S. Yakovlev, M. Y. Ivanov, F. Krausz; Enhanced 

phase-matching for generation of soft X-ray harmonics 

and attosecond pulses in atomic gases. Opt. Exp. 15, 

No. 23, p. 153511-15364 (12 November 2007). 

[32] J. Seres et al.; Coherent superposition of laserdriven 

soft-X-ray harmonics from successive sources. 

Nature Phys. 3, 878 (2007). 

[33] E. Fill, L. Veisz, A. Apolonskiy, F. Krausz; Sub-fs 

electron pulses for ultrafast electron diffraction. New J. 

Phys. 8, 272 (2006); L.Veisz et al. Hybrid DC-AC electron 

gun for fs-electron pulse generation. New J. Phys. 9, 

451 (2007). 

[34] P. Reckenthaeler et al.; Proposed method for 

measuring the duration of electron pulses by attosecond 

streaking. Phys. Rev. A 77, 042902 (2008). 

[35] M. Centurion et al.; Picosecond electron 

deflectometry of optical-field ionized plasmas. Nature 

Phot. 2, 315 (2008). 

[36] F. Grüner et al.; Design considerations for table-top, 

laser-based VUV and X-ray free electron lasers, Appl. 

Phys. B 86, 431-435 



1.3.1.5 ATTOSECOND DYNAMICS 

Leader: Dr. R. Kienberger 

The main goal of the Junior Research Group is to 

investigate electronic processes in molecules and on 

solid surfaces on an attosecond timescale. Efforts have 

been undertaken to improve the necessary tools, i.e. 

phase-stabilized few-cycle laser pulses, ultrashort pulses 

in the ultraviolet (UV) and vacuum ultraviolet (VUV), 

and to develop attosecond beamlines for the envisaged 

experiments. 

IMPROVEMENT OF ThE lASER SYSTEM TO 1.5 CYClE 

lASER PUlSES 

The natural limit nature sets to the duration of a light 

pulse is one oscillation cycle of the carrier wavelength. 

The pulse duration is inversely proportional to the 

spectral width of the pulse, i.e. the broader the spectrum 

the shorter the pulse. In a chirped pulse laser amplifier, 

pulses are temporally stretched for the amplification 

and have to be compressed subsequently. During the 

compression, e.g. in prisms, the pulses normally lose 

spectral width due to parasitic effects like self-phase 

modulation (SPM). To overcome this problem, we 

developed a “hybrid” compressor consisting of prisms 

and additional chirped mirrors which take over the last 

part of the compression without the problem of SPM, 

leading to amplified laser pulses lasting only 18 fs. 

Sending theses pulses into a gas filled hollow core fiber 

and final compression resulted in 0.4 mJ laser pulses as 

short as 3.8 fs – only 1.5 cycles of the carrier wave [5, see 

reprint]. This is a new world record and made possible 

the generation of intense, ultrashort UV pulses. 

GENERATION OF UlTRAShORT UV AND VUV PUlSES 

Since the separation of quantum states which are 

relevant for electronic dynamics in molecules is on the 

order of 5 to 10 eV, ultrashort light pulses at this photon 

energy have to be generated. The UV/VUV generation 

was performed in a gas jet and a special “self diffraction 

FROG” (Frequency Resolved Optical Gating) device 

has been set up for the full characterization of the UV 

pulses. 

The spectra we were able to generate in the UV are 

extraordinarily broad (Figure 1) and could be compressed 

to 3.7 fs. 

ATTOSECOND TEChNOlOGY ON A SOlID SURFACE 

In a collaboration with the University of Bielefeld, a first 

attosecond experiment on a solid surface was performed 

1 . 3 AT T O S E C O N D A N D H I G H - F I E L D D I V I S I O N 

[2, see reprint]. We were able to measure a time delay 

between conduction band and core level electrons in 

single-crystal tungsten as short as 110 as. 

Figure 1: Ultrabroad UV spectrum supporting 3 fs 

pulses 

REFERENCES 

[1] Remacle F., Kienberger R., Krausz F., Levine, R.D.; 

“On the feasibility of an ultrafast purely electronic 

reorganization in lithium hydride”, Chemical Physics 

338 (2-3): 342-347 ( 2007). 

[2] A. L. Cavalieri, N. Müller, Th. Uphues, V. Yakovlev, 

A. Baltuska, B. Horvath, B. Schmidt, L. Blümel, S. 

Hendel, P. M. Echenique, M. Drescher, U. Kleineberg, 

R. Kienberger, F. Krausz, U. Heinzmann; „Attosecond 

time-resolved photoemission from solids”, Nature 449, 

1029 (2007). 

[3] Kienberger R, Uiberacker M, Kling MF, Krausz F.; 

„Attosecond physics comes of age: from tracing to 

steering electrons at sub-atomic scales” (invited); J. 

Mod. Opt. 54 (13-15), 1985-1998 (2007). 

[4] E. Goulielmakis, V. Yakovlev, A. L. Cavalieri, M. 

Uiberacker, V. Pervak, A. Apolonsky, R. Kienberger, 

U. Kleineberg, F. Krausz; “Attosecond Control and 

Measurement: Lightwave Electronics”, Science 317, 

769 (2007). 

[5] A. L. Cavalieri, E. Goulielmakis, B. Horvath, W. Helml, 

M. Schultze, M. Fieß, V. Pervak, L. Veisz, V. Yakovlev, 

M. Uiberacker, A. Apolonski, F. Krausz, R. Kienberger; 

“Intense 1.5-cycle near infrared laser waveforms and 

their use for the generation of ultrabroad-band soft-Xray 

harmonic continua”, NJP 9, 242 (2007). 

[6] MB. Gaarde, M. Murakami, R. Kienberger; „Spatial 

separation of large dynamical blueshift and harmonic 

generation”, Phys. Rev. A 74 (5), 053401 (2006) 

[7] A. Scrinzi, M.Yu. Ivanov, R. Kienberger and D. M. 

Villeneuve; “Attosecond Physics” (invited); J. Phys. B: At. 

Mol. Opt. Phys. 39, R1 – R37 (2006). 


1.3.1.6 ATTOSECOND IMAGING 

Leader: Dr. Matthias Kling 

With waveform-controlled laser fields to control 

electronic motion and (single) attosecond XUV pulses 

to probe electronic motion in real-time two powerful 

tools are available to explore electron dynamics in 

atoms and molecules. We have proposed to extend 

the application of these tools to studies of correlated 

and collective electron motion in nanosystems. One of 

our aims was to utilize imaging technologies as e.g. 

velocity-map imaging (VMI) to monitor the ultrafast 

electron dynamics and its control. As a demonstration 

that VMI can indeed be used to monitor the waveformcontrol 

of electron emission we have investigated 

atomic systems, where the obtained results have been 

compared to theoretical predictions (by e.g. solving 

the time-dependent Schrödinger equation (TDSE)). 

This control is applied to molecular systems, where the 

light-steering of binding electrons is used to coherently 

control a reaction. As a highlight, the waveform control 

of electron localization in molecular dissociation was 

time-resolved by an attosecond pump-probe experiment 

for the first time. 

The full spatio-temporal characterization of nanometerscale 

collective electron dynamics with attosecond 

temporal resolution was studied theoretically within 

the collaboration with Prof. Mark Stockman (GSU, 

Atlanta, USA). The original idea for studies on isolated 

nanoparticles was extended to photoelectron emission 

microscopy (PEEM) of nanosystems on surfaces, leading 

to a new technique dubbed “attosecond nanoplasmonic 

microscope”. Prof. Ulf Kleineberg (LMU) and his 

coworkers are currently building a UHV experiment 

for time-resolved Photoelectron Emission Microscopy, 

which will be capable for the first time to track electron 

dynamics in nanostructures on surfaces on a ~100 

fsec time scale and with 10 nm spacial resolution. This 

effort complements the development of the attosecond 

beamline AS 3. 

IMAGING OF SUB-FEMTOSECOND CONTROl OF 

ElECTRON DYNAMICS 

Control of the electric field waveform of laser light 

pulses has opened a new way to study and control 

electron dynamics in strong-field processes. We have 

utilized velocity-map imaging to study the angleresolved 

electron emission from Ar, Kr and Xe atoms 

with waveform-controlled few-cycle pulses. The data 

for Ar is compared to simulations using the strongfield 

approximation (SFA) and full time-dependent 

Schrödinger equation (TDSE) calculations. We find a 

pronounced asymmetry in both the energy and angular 

distributions of the electron emission that critically 


depends on the carrier-envelope phase of the laser field. 

The generally weak asymmetry of low-energy electrons 

can be explained by their deflection in the oscillating 

laser field that cancels the initial non-linear dependence 

of the ionization yield on the field strength. In the 

plateau region of the argon data, the asymmetry is 

larger and the phase of its oscillation with CEP shifts 

not only with the electron kinetic energy but also with 

the emission angle. A triangular pattern suggests that 

the observation of the asymmetry at a constant energy 

but two different emission angles can serve as a way to 

measure the phase CEP of the laser pulse without ±π 

ambiguity. In the case of xenon, significant asymmetry 

was observed even below 2 Up in particular for high 

emission angles around 60 degrees. This unique 

dependence indicates that VMI can also be used instead 

of Stereo-ATI detection to determine the CEP with high 

accuracy. 

A B 

C 

Figure 1: A) and B) show raw two-dimensional 

photoelectron images for above-threshold ionization 

of argon with 5fs laser pulses for phases of ϕ = 0 

and ϕ = π, measured by velocity map imaging. As 

seen in C), where the asymmetry of the emission is 

plotted (red = upwards, blue = downwards), the 

emission can be controlled with the laser phase. 

A pronounced shift in the phase of the asymmetry 

oscillation that is seen in the experimental data is 

reproduced well beyond 5 Up by both SFA and TDSE 

calculations. While a phase jump is visible for the SFA 

calculated asymmetries at around 5 Up, the difference 

between the TDSE calculated values and the experimental 


data stays consistently below 0.1 rad and thus shows 

excellent agreement all the way down to ca. 2 U p (below 

this value, noise is dominating the curves due to the low 

amplitude of the asymmetry). We have demonstrated 

that velocity-map imaging is not only a powerful tool 

for measuring the full momentum distributions of ATI 

in rare gases but also provides a new way to determine 

the CEP of few-cycle laser pulses from the angular 

distribution of the emitted electrons. 

WAVEFORM CONTROl OF ElECTRON lOCAlIZATION 

IN MOlECUlES 

A previously explored single-color control scheme for 

attosecond electron localization in molecules is difficult 

to extend to systems with dissociation times much longer 

than the duration of the waveform-controlled fewcycle 

laser field (in previous experiments, 5 fs pulses at 

760 nm were used) as the field strength of the laser will 

be to low to coherently couple two electronic states 

close to the break-up of the molecule. 

As a demonstration for a suitable way to extend the 

control scheme, in collaboration with the groups of Anne 

L’Huillier (Lund), Marc Vrakking (Amolf), Franck Lepine 

(Lyon) and Mauro Nisoli (Milan), we have performed 

a two-color experiment, where an isolated attosecond 

laser pulse ionizes D and part of the ionized molecules 

2 

dissociate. The single attosecond pulse is generated by 

means of high-harmonic generation in Krypton using the 

polarization gating technique. In the photo-ionization 

+ by the attosecond pulse the repulsive 2pσ state is 

u 

populated when the photon-energy of the pulse is high 

enough (above ca. 29 eV). Interaction of this dissociating 

wave packet with a time delayed moderately strong IR 

field localizes the electron on the upper or lower D + ion. 

By varying the delay between the XUV pulse and the 

few-cycle IR pulse the asymmetry in the ejection of D + 

ions can be controlled with attosecond time resolution. 

This experiment may be viewed as a first example of 

the observation of attosecond time-resolved electron 

dynamics in molecular physics. As seen in Figure 2 the 

asymmetry shows a strong dependence on the kinetic 

energy of the D + fragment. 

A model solving the 1D time-dependent Schrödinger 

equation (TDSE) was used to confirm the interpretation 

and to give insight into the origin of the observed 

energy dependence. In the model a nuclear wave 

+ packet is projected on the repulsive 2pσ state and 

u 

+ propagated on the bound 1sσ and the repulsive 

g 

+ 2pσ state of the molecular ion in the presence of a 

u 

few-cycle IR field. The IR field couples the two states 

which generates a coherent superposition where the 

electron can be localized on one or the other ion. Just as 

in the experiment the timing of the IR field determines 

the electron localization. The kinetic energy dependence 


Figure 2: A) D + ion kinetic energy spectra from the 

dissociative ionization of D 2 versus the delay between 

the 300 as XUV pulse at 30 eV and a phase-stable 

6-fs IR pulse; B) Asymmetry of the D + ion emission 

along the laser polarization as a function of the D + 

kinetic energy and the delay between the XUV and 

IR pulses. 

of the electron localization, which is observed in the 

experiment and the calculations, likely, originates from 

the initial spread of the wave packet. 

Electron transfer processes are ubiquitous in chemistry. 

The present combination of one- and two-color 

experiments, where electron localization is first 

controlled on attosecond timescales using the controlled 

waveform of a few-cycle IR laser pulse and then using 

the controlled delay between an isolated attosecond 

pulse and a few-cycle IR laser pulse, are first examples of 

strong-field control of chemical processes on attosecond 

timescales and of the direct observation of attosecond 

electron dynamics in molecules, paving the way towards 

attempts to observe and control electron dynamics in 

more complicated molecules and nanostructures. The 

results of these studies are currently written up. 

ATTOSECOND NANOPlASMONIC MICROSCOPY 

Nanoplasmonics deals with collective electronic 

dynamics on the surface of metal nanostructures, which 

arises as a result of excitations called surface plasmons. 


This field, which has recently undergone rapid growth, 

could benefit applications such as computing and 

information storage on the nanoscale, the ultrasensitive 

detection and spectroscopy of physical, chemical and 

biological nanosized objects, and the development of 

optoelectronic devices. Because of their broad spectral 

bandwidth, surface plasmons undergo ultrafast 

dynamics with timescales as short as a few hundred 

attoseconds. So far, the spatiotemporal dynamics of 

optical fields localized on the nanoscale has been hidden 

from direct access in the real space and time domain. 

In collaboration with Prof. Mark Stockman we have 

proposed and developed a theory for an attosecond 

nanoplasmonic-field microscope. The main advantage 

of such a microscope is that it is non-invasive with 

respect to the nanoplasmonic fields. The principle of this 

microscope is based on the photoemission of electrons 

by an XUV attosecond pulse that is synchronized with 

a waveform-stabilized driving optical field. Information 

about the nanoplasmonic fields is imprinted in the 

energy of the XUV-emitted electrons owing to their 

acceleration in the instantaneous electrostatic potential 

of the surface plasmon oscillations excited by the optical 

field. 

The attosecond nanoplasmonic microscope will open up 

unique possibilities to directly study and control ultrafast 

photoprocesses in surface plasmonic nanosystems and 

circuits. It images the local nanoplasmonic field in real 

space with nanometre-scale spatial resolution and 

in real time with approx. 100 as temporal resolution. 

This approach will be especially important in ultrafast 

nanoplasmonic systems where very tight localization 

of optical fields occurs, for example, when a shaped 

pulse of radiation induces nanolocalized fields at a 

desired nanosite. The microscope could also be used 

to study various plasmon enhanced photoprocesses 

such as femtosecond photochemistry, light detection, 

and solar energy conversion where the processes of 

the energy transfer can be ultrafast. For example, 

with nanoplasmonic antennas, which couple together 

molecular and semiconductor systems with external 

fields, the ultrafast kinetics of the energy exchange 

can be studied through the corresponding kinetics of 

the nanoplasmonic fields. One of the most important 

potential uses of the attosecond nanoplasmonicfield 

microscope will be in the design and study of 

elements and devices for ultrafast and ultradense (at 

the nanoscale) optical and optoelectronic information 

processing and storage. To bring practical advantages 

over existing electronic and optoelectronic technology, 

such nanoscale devices must necessarily be ultrafast 

(with a subpicosecond or femtosecond response time). 

Examples of such devices include optical nanotransistors, 

memory cells, nanoplasmonic media for the mass storage 

of information, and nanoplasmonic interconnects 


where both the nanolocalization of energy and its 

ultrafast temporal kinetics are important. Currently, in 

collaboration with Prof. Ulf Kleineberg (LMU) we are 

preparing the setup and samples for first PEEM studies. 

Figure 3: After excitation of the collective electron 

motions (plasmons) in a nanoparticle assembly by an 

optical light pulse, the resulting plasmon dynamics 

can be probed with high spatial and temporal 

resolution by the generation of photoelectrons und 

their acceleration in the plasmonic fields. This figure 

shows calculated snapshots of the plasmonic fields 

at different time delays after excitation. The spatial 

resolution is given by the limit of the optics of the 

photoemission electron microscope (PEEM) that is 

used for the detection of the electrons (approx. 20 nm). 

The time resolution (approx. 100 attoseconds) is 

given by the duration of the XUV probe pulse. These 

theoretical calculations show for the first time, that 

the direct and non-invasive observation of rapidly 

evolving nanoplasmonic fields becomes possible (see 

the two snap shots in the lower part of the figure, 

which show the fields within a time interval of 1 fs 

(= 10 -15 s)). 

REFERENCES 

[1] M.F. Kling, and M.J.J. Vrakking; Attosecond Electron 

Dynamics, Annu. Rev. Phys. Chem. 59, 463-492 

(2008). 

[2] M.F. Kling, Ch. Siedschlag, I. Znakovskaya, A.J. 

Verhoef, S. Zherebtsov, F. Krausz, M. Lezius, and M.J.J. 

Vrakking; Strong-field control of electron localization 

during molecular dissociation, Mol.Phys. 106, 455-465 

(2008). 


[3] M.F. Kling, J. Rauschenberger, A.J. Verhoef, E. 

Hasovic, T. Uphues, D.B. Milosevic, H.G. Muller, and 

M.J.J. Vrakking; Imaging of carrier-envelope phase 

effects in above-threshold ionization with intense fewcycle 

laser fields, New J. Phys. 10, 025024 (2008). 

[4] Th. Uphues, M. Schultze, M.F. Kling, M. Uiberacker, 

S. Hendel, U. Heinzmann, N.M. Kabaschnik, and M. 

Drescher; Ion-charge-state chronoscopy of cascaded 

atomic Auger decay, New J. Phys. 10, 025009 (2008). 

[5] A. Gijsbertsen, W. Siu, M.F. Kling, P. Johnsson, 

P. Jansen, S. Stolte, and M.J.J. Vrakking; Direct 

Determination of the Sign of the NO Dipole Moment, 

Phys.Rev.Lett. 99, 213003 (2007) 

[6] R. Kienberger, M. Uiberacker, M.F. Kling, and F. 

Krausz; Attosecond physics comes of age: from tracing 

to steering electrons at sub-atomic scales, J.Mod.Opt. 

54, 1985-1998 (2007) 

[7] M.I. Stockman, M.F. Kling, U. Kleineberg, and F. 

Krausz; Attosecond nanoplasmonic field microscope, 

Nature Photonics 1, 539 (2007). 

[8] F. Lepine, M.F. Kling, Y. Ni, J. Khan, O. Ghafur, T. 

Martchenko, E. Gustafsson, P. Johnsson, K. Varju, T. 

Remetter, A. L’Huillier, M.J.J. Vrakking; Short XUV pulses 

to characterize field-free molecular alignment, J. Mod. 

Opt. 54, 953 (2007). 

[9] M. Uiberacker, Th. Uphues, M. Schultze, A.J. 

Verhoef, V. Yakovlev, M.F. Kling, J. Rauschenberger, 

N.M. Kabaschnik, H. Schröder, M. Lezius, K.L. Kompa, 

H.-G. Muller, M.J.J. Vrakking, S. Hendel, U. Kleineberg, 

U. Heinzmann, M. Drescher, F. Krausz; Attosecond realtime 

observation of electron tunneling in atoms, Nature 

446, 627-632 (2007) 



1.3.2 SURVEY OF ThE RESEARCh ACTIVITIES 

1 . 3 . 2 S U R V E Y O F T H E R E S E A R C H A C T I V I T I E S 

Pushing the frontiers of femtosecond technology 

Project coordinators: A. Apolonskiy, R. Kienberger, S. Karsch, L. Veisz 

Project Objectives Team 

Chirped multilayer 

dielectric mirrors 

Generation of terawattscale 

near-monocycle laser 

pulses with controlled 

waveform at several kHz 

repetition rate in the near 

infrared (LWS-N1) and in 

the mid-infrared (LWS-M1) 

Development of a highfield 

waveform synthesizer 

producing few-cycle nearinfrared 

laser pulses with a 

peak power of several 10 

terawatt and controlled 

waveform at a 10 Hz 

repetition rate (LWS-10) 

Development of the 

Petawatt Field Synthesizer 

(PFS) 

Design, manufacturing & characterization of broadband 

mirrors for optical waveform synthesis in the 

visible, infrared and ultraviolet spectral range 

LWS-N1: 4fs/2mJ/0.5TW@0.75μm&4kHz 

with Ti:sapphire MOPA + hollow-fibre/chirped-mirror 

compressor 

LWS-M1: 

12fs/10mJ/1TW@2.1μm&10kHz 

by parametric amplification pumped of by a sub-2ps, 

50-mJ, diode-pumped Yb:YAG disk laser 

Current status: 

8fs/70mJ/10TW@0.8μm&10Hz 

Target: 

5fs/600mJ/100TW@0.8μm&10Hz 

using an optical chirped-pulse amplifier chain 

pumped by a commercial picosecond Nd:YAG laser 

Target parameters: 

3J/~1PW@1.2μm&10Hz 

V. Pervak (PL) 

I. Grguras 

A. Apolonskiy 

External collaborators: 

A. V. Tikhonravov (Moscow) 

M. K. Trubetskov (Moscow) 

A. Cavalieri (PL: N1) 

X. Gu (PL: M1) 

Y. Deng 

W. Helml 

G. Marcus 

T. Metzger 

B. Horváth 

J. Li 

E. Magerl, 

R. Kienberger 


H. Giesen (Stuttgart) 

D. Sutter (Trumpf GmbH) 

L. Veisz 

A. Buck 

D. Herrmann 

K. Schmid 

F. Tavella 

T. Wittman 

A. Marcinkevicius (until 

12/2006) 

Zs. Major (PL) 

M. Siebold (PL) 

S. Trushin (PL) 

I. Ahmad 

S. Klingebiel 

Ch. Wandt 

T.-J. Wang 

S. Karsch 


J. Hein (Jena) 

A. Tünnermann (Jena) 

PL = Project leader 


1 . 3 AT T O S E C O N D A N D H I G H - F I E L D O H Y S I C S D I V I S I O N 

Pushing the frontiers of femtosecond technology (continued) 

Project coordinators: A. Apolonskiy, R. Kienberger, S. Karsch, L. Veisz 


Scaling femtosecond 

tech-nology towards 

multi-kW intra-cavity 

average power levels and 

development of MHzrate 

XUV sources with 

milliwatt-scale average 

power 

Generation, measurement 

and applications of few-fs 

sub-relativistic electron 

bunches 

Millijoule-energy femtosecond laser pulses inside 

solid-state femtosecond laser oscillators and passive 

build-up cavities for intracavity 

production of UV/VUV/XUV/SXR light 

Several-10-keV, few-electron-bunches produced at 

MHz repetition rates in synchrony with MHz-rate, 

microjoule-energy few-cycle laser pulses for scaling 

time-resolved electron diffraction towards the 1femtosecond 

frontier 

lightwave electronics: attosecond control and metrology 

Project coordinators: R. Kienberger, U. Kleineberg, M. Kling 

Projects Objectives Team 

Chirped multilayer metallic 

mirrors 

Attosecond pulse 

generation from atomic 

harmonics 

Design, manufacturing & characterization of 

broadband mirrors for attosecond XUV/SXR pulse 

technology (10 eV – 1000 eV) 

Scaling towards microjoule pulse energies or higher 

(several hundred to thousand electronvolt) photon 

energies by using LWS-10 and LWS-N1/LWS-M1 

as a driver, respectively, and exploiting quasi-phase 

matching schemes 


J. Rauschenberger (PL) 

R. Graf 

J. Pupeza 

A. Sugita 

C. Teisset 

A. Apolonskiy 


T. Udem & T. Hänsch (MPQ) 

D. Hoffmann (Aachen) 

A. Tünnermann (Jena) 

E. Fill (PL: Technology) 

P. Baum (PL: Applications) 

M. Centurion 

P. Reckenthäler 

L. Veisz 

A. Apolonskiy 


V. Tarnetsky (Novosibirsk) 

A. Zewail (Pasadena) 

U. Kleineberg (PL) 

M. Hofstetter 

J. Lin 


A. L. Aquila 

E. M. Gullikson & 

D. T. Attwood (Berkeley) 

G. Marcus (PL: LWS-10, M1) 

A. Cavalieri (PL: LWS-N1) 

D. Herrmann 

M. Hofstetter 

U. Kleineberg 

V. Yakovlev (theory) 

L. Veisz 

R. Kienberger 


D. Charalambidis (Heraklion)


lightwave electronics: attosecond control and metrology (continued) 

Project coordinators: R. Kienberger, U. Kleineberg, M. Kling 


Towards ultrawide-band 

waveform synthesis and 

mono-cycle UV/VUV pulse 

generation 

Attosecond spectroscopy 

in isolated atoms and 

nanoparticles 

Attosecond spectroscopy 

in molecules and solids 

Towards solid-state 

lightwave electronics 

Sub-cycle shaping of the electromagnetic field of 

light by Fourier-synthesis of intense (sub-terawatt) 

arbitrary waveforms in the NIR/VIS/UV spectral range 

& few-fs UV/VUV pulses 

Real-time observation of electron dynamics and 

correlations in atoms and collective motion in 

nanoparticles by using isolated attosecond XUV 

pulses and few-cycle NIR waveforms for pump/ 

probe spectroscopy and sophisticated electron and 

ion detection techniques (including coincidence 

detection) 

Real-time observation of electron motion (individual 

as well as collective) in molecules (in the gas phase 

and on surfaces) and in solid-state structures by 

using isolated attosecond XUV pulses and few-cycle 

NIR/UV waveforms for pump/probe spectroscopy, 

also in combination with photo-electron emission 

microscopy (PEEM) 

Demonstration of controlling atomic-scale electron 

motion in solids with light fields 

E. Goulielmakis (PL) 

R. Ernstorfer 

U. Graf 

I. Grguras 

V. Pervak, 

A. Apolonskiy 

R. Kienberger 

M. Kling (PL) 

E. Goulielmakis 

O. Herrwerth 

U. Kleineberg 

A. Wirth 

S. Zherebtsov 

I. Znakovskaya 


K. Kompa 

M. Lezius & 

H. Schröder (MPQ) 

M. Vrakking (Amsterdam) 

R. Moshammer & 

J. Ullrich (Heidelberg) 

A. Cavalieri (PL: solids) 

M. Schultze (PL: molecules) 

J. Lin (PEEM) 

R. Ernstorfer 

M. Fieß 

E. Magerl 

R. Kienberger 

U. Kleineberg 


J. Barth 

P. Feulner & 

D. Menzel (TUM) 

K. Kompa &. 

M. Lezius (MPQ) 

R. Levine (Jerusalem) 

F. Remacle (Liège) 

R. Ernstorfer (PL) 

U. Graf 

E. Goulielmakis 

R. Kienberger 


J. Barth 

P. Feulner & 

D. Menzel (TUM) 


1 . 3 AT T O S E C O N D A N D H I G H - F I E L D O H Y S I C S D I V I S I O N 

high-field attosecond science 

Project coordinators: F. Grüner, D. Habs, S. Karsch, G. Tsakiris, L. Veisz 


Few-fs to sub-fs relativistic 

electron bunches 

Laser-plasma acceleration 

of electrons to superrelativistic 

(>>100 MeV) 

energies 

Reproducible generation of mono-energetic relativistic 

(1-100MeV) few-fs electron bunches in underdense 

plasmas driven by few-cycle laser pulses from LWS-10, 

sub-femtosecond bunch slicing by the inverse freeelectron 

laser mechanism, temporal characterization 

and applications for time-resolved experiments 

Reproducible generation of collimated, high-current, 

mono-energetic electron bunches of energies 

ranging from 100 MeV to several GeV by means of 

laser wake field acceleration (LWFA) for producing 

undulator radiation and free-electron lasing, and 

other applications 


K. Schmid (few-fs) 

C. Sears (sub-fs) 

S. Becker 

A. Buck 

F. Grüner 

S. Karsch 

Y. Mikhailova (theory) 

D. Habs 

D. Herrmann 

R. Tautz 

L. Veisz 


W. Leemans (Berkeley) 

S. Karsch 

S. Becker 

M. Fuchs 

F. Grüner 

D. Habs 

Zs. Major 

B. Marx 

J. Osterhoff 

A. Popp, 

R. Weingartner 

M. Fuchs 

L. Veisz 


M. Geissler (Belfast) 

S. Hooker (Oxford) 

W. Leemans (Berkeley)


high-field attosecond science (continued) 

Project coordinators: F. Grüner, D. Habs, S. Karsch, G. Tsakiris, L. Veisz 


Towards a laser-based, 

compact X-ray freeelectron 

laser (XFEL) 

Towards intense 

attosecond XUV pulses 

Ion and X-ray beams from 

thin foils 

Optimization of LWFA and preparation of the 

accelerated electron beam for undulator seeding, 

demonstration of undulator radiation and FEL 

operation at XUV wavelengths and subsequently at 

successively shorter wavelengths 

Generation of intense single attosecond XUV pulses 

from few-cycle-driven relativistic plasmas at solid 

surfaces, their characterization and applications for 

XUV nonlinear optics and XUV-pump/XUV-probe 

spectroscopy 

Generation of mono-energetic ion beams from ultrathin 

diamond foils driven by ~ 100 TW laser pulses 

(from ATLAS) and of coherent X-rays by Thomson 

scattering from electron sheaths driven out of the foil 

by multi-TW laser pulses (from LWS-10) 

F. Grüner 

S. Becker 

M. Fuchs 

R. Weingartner 

D. Habs 

S. Karsch 

Zs. Major 

B. Marx 

J. Osterhoff 

A. Popp 

L. Veisz 


M. Geissler (Belfast) 

S. Hooker (Oxford) 

W. Leemans (Berkeley) 

G. Tsakiris 

A. Buck 

R. Hörlein 

S. Karsch 

Z. Major 

Y. Mikhailova 

Y. Nomura 

S. Rykovanov 

L. Veisz 


D. Charalambidis (Heraklion) 

M. Zepf (QU Belfast) 

M. Geissler (Qu Belfast) 

I. Foeldes (KFKI-Hungary) 

D. Habs 

A. Buck 

L. Veisz 


M. Hegelich (Los Alamos) 


1.3.3 SElECTED REPRINTS 


1) Design considerations for table-top, laser-based VUV 

and X-ray free electron lasers 

F. Grüner, S. Becker, U. Schramm, T. Eichner, M. Fuchs, 

R. Weingartner, D. Habs, J. Meyer-ter-Vehn, M. Geissler, 

M. Ferrario, L. Serafini, B. van der Geer, H. Backe, W. 

Lauth, S. Reiche, 

Applied Physics B 86, 431 (2007). 

MPQ Progress Report: page 160 

2) Attosecond real-time observation of electron tunnelling 

in atoms 

M. Uiberacker, Th. Uphues, M. Schultze, A. J. Verhoef, 

V. Yakovlev, M. F. Kling, J. Rauschenberger, N. M. Kabachnik, 

H. Schröder, M. Lezius, K. L. Kompa, H.-G. 

Muller, M. J. J. Vrakking, S. Hendel, U. Kleineberg, U. 

Heinzmann, M. Drescher, F. Krausz, 

Nature 446, 627 (2007). 


3) Dispersion-control over the ultraviolet-visible-near-infrared 

spectral range with HfO 2 /SiO 2- chirped dielectric 

multilayers 

V. Pervak, F. Krausz, A. Apolonski, 

Optics Letters 32, 1183 (2007). 

MPQ Progress Report: page 171* 

4) Dispersion management for a sub-10-fs, 10-TW optical 

parametric chirped-pulse amplifier 

F. Tavella, Y. Nomura, L. Veisz, V. Pervak, A. Marcinkevicius, 

F. Krausz, 

Optics Letters 32, 2227 (2007). 


5) Attosecond control and measurement: lightwave 

electronics 

E. Goulielmakis, V. S. Yakovlev, A. L. Cavalieri, M. Uiberacker, 

V. Pervak, A. Apolonski, R. Kienberger, U. Kleineberg, 

F. Krausz, 

Science 317, 769 (2007). 


6) Coherent superposition of laser-driven soft-X-ray harmonics 

from successive sources 

J. Seres, V. S. Yakovlev, E. Seres, Ch. Streli, P. Wobrauschek, 

Ch. Spielmann, F. Krausz, 

Nature Physics 3, 878 (2007). 


7) Enhanced phase-matching for generation of soft Xray 

harmonics and attosecond pulses in atomic gases 

V. S. Yakovlev, M. Ivanov, F. Krausz, 

Optics Express 15, 15351 (2007). 

Downloadable from http://www.opticsexpress.org/issue.cfm?volume=15&issue=23 


8) GeV-scale electron acceleration in a gas-filled capillary 

discharge waveguide 

S. Karsch, J. Osterhoff, A. Popp, T. P. Rowlands-Rees, Zs. 

Major, M. Fuchs, B. Marx, R. Hörlein, K. Schmid, L. Veisz, 

S. Becker, U. Schramm, B. Hidding, G. Pretzler, D. Habs, 

F. Grüner, F. Krausz, S. M. Hooker, 

New Journal of Physics 9, 415 (2007). 

Downloadable from http://www.iop.org/EJ/abstract/1367-2630/9/11/415 


9) Hybrid dc–ac electron gun for fs-electron pulse generation 

L. Veisz, G. Kurkin, K. Chernov, V. Tarnetsky, A. Apolonski, 

F Krausz, E. Fill, 




10) Picosecond electron deflectometry of optical-field 

ionized plasmas 

M. Centurion, P. Reckenthaeler, S. A. Trushin, F. Krausz, 

E. E. Fill, 



11) Single-cycle nonlinear optics 

E. Goulielmakis, M. Schultze, M. Hofstetter, V. S. Yakovlev, 

J. Gagnon, M. Uiberacker, A. L. Aquila, E. M. 

Gullikson, D. T. Attwood, R. Kienberger, F. Krausz, U. 

Kleineberg, 

Science 320, 1614 (2008) 


12) Intense single attosecond pulses from surface harmonics 

using the polarization gating technique 

S G Rykovanov, M Geissler, J Meyer-ter-Vehn and 

G D Tsakiris 





*) 1st page


Dr. R. Kienberger 

1) Intense 1.5-cycle near infrared laser waveforms and 

their use for the generation of ultra-broadband soft-xray 

harmonic continua 

A. L. Cavalieri, E. Goulielmakis, B. Horvath, W. Helml, M. 

Schultze, M. Fieß, V. Pervak, L. Veisz, V. S. Yakovlev, M. 

Uiberacker, A. Apolonski, F. Krausz, R. Kienberger, 


Downloadable from http://www.iop.org/Select/abstract/1367-2630/9/7/242 


2) Attosecond spectroscopy in condensed matter 

A. L. Cavalieri, N. Müller, Th. Uphues, V. S. Yakovlev, A. 

Baltuska, B. Horvath, B. Schmidt, L. Blümel, R. Holzwarth, 

S. Hendel, M. Drescher, U. Kleineberg, P. M. Echenique, 

R. Kienberger, F. Krausz, U. Heinzmann, 

Nature 449, 1029 (2007). 


Dr. M. Kling 

1) Attosecond nanoplasmonic-field microscope 

M. I. Stockman, M. F. Kling, U. Kleineberg, F. Krausz, 



1 . 3 . 3 S E L E C T E D R E P R I N T S 

*) 1st page 


Appl. Phys. B 86, 431–435 (2007) 

DOI: 10.1007/s00340-006-2565-7 

f. grüner 1,� 

s. becker 1 

u. schramm 1 

t. eichner 1 

m. fuchs 1 

r. weingartner 1 

d. habs 1 

j. meyer-ter-vehn 2 

m. geissler 2 

m. ferrario 3 

l. serafini 3 

b. van der geer 4 

h. backe 5 

w. lauth 5 

s. reiche 6 


Received: 12 December 2006 

Published online: 27 January 2007 • © Springer-Verlag 2007 

Applied Physics B 

Lasers and Optics 

Design considerations for table-top, 

laser-based VUV and X-ray free 

electron lasers 

ABSTRACT A recent breakthrough in laser-plasma accelerators, based upon ultrashort 

high-intensity lasers, demonstrated the generation of quasi-monoenergetic GeVelectrons. 

With future Petawatt lasers ultra-high beam currents of ∼ 100 kA in ∼ 10 fs 

can be expected, allowing for drastic reduction in the undulator length of free-electronlasers 

(FELs). We present a discussion of the key aspects of a table-top FEL design, 

including energy loss and chirps induced by space-charge and wakefields. These 

effects become important for an optimized table-top FEL operation. A first proof-ofprinciple 

VUV case is considered as well as a table-top X-ray-FEL which may also 

open a brilliant light source for new methods in clinical diagnostics. 

PACS 41.60.Cr; 52.38.Kd 

1 Introduction 

The pursuit of table-top FELs 

combines two rapidly developing fields: 

laser-plasma accelerators, where the 

generation of intense quasi-monoenergetic 

electron bunches up to the GeV 

range has been achieved [1–4], and 

large-scale X-ray free-electron lasers 

(XFELs) that are expected to deliver 

photon beams with unprecedented peak 

brilliance. A prominent application of 

such FEL pulses is single molecule 

imaging [5]. The proposed laser-plasma 

accelerator-based FELs would not only 

allow a greater availability due to their 

smaller size and costs, but also offer new 

features, such as pulses synchronized 

to the phase-controlled few-cycle driver 

laser [6] for pump–probe experiments. 

Moreover, the X-ray energy of a tabletop 

XFEL can be as large as required for 

medical diagnostics (above 20 keV [7]). 

1 Department of Physics, Ludwig-Maximilians-Universität München, 85748 Garching, Germany 

2 Max-Planck Institute of Quantum Optics, 85748 Garching, Germany 

3 INFN, 00044 Frascati, Italy 

4 Pulsar Physics, 3762 Soest, Netherlands 

5 Institute of Nuclear Physics, Mainz, Germany 

6 University of California Los Angeles, UCLA, Los Angeles, CA, USA 

The mechanism for the generation 

of intense (nC charge) quasi-monoenergetic 

electron pulses by laser-plasma 

accelerators requires an ultrashort, highintensity 

laser-pulse with a length 

shorter than the plasma wavelength 

(on the µm-scale corresponding to gas 

densities of 10 19 cm −3 ). Due to the 

ponderomotive force, plasma electrons 

are blown out transversely, leaving an 

electron-free zone – the bubble – behind 

the laser pulse [8]. These electrons 

return to the axis after half a plasma oscillation, 

thus determining the size of 

the bubble in the order of the plasma 

wavelength. Typically about 10 9 ...10 10 

electrons are captured into the bubble, 

as found both experimentally and from 

scaling laws of relevant bubble parameters 

within the framework of similarity 

arguments [9]. Due to the inertial positive 

ion background, these electrons 

experience a strong electrical field gra- 

� Fax: +49 89 2891 4072, E-mail: florian.gruener@physik.uni-muenchen.de 

dient of up to TV/m. Particle-in-cell 

(PIC) simulations [10] show that the 

bubble electrons form a stem that is geometrically 

considerably smaller than the 

bubble as seen in Fig. 1. 

For these dimensions, beam currents 

of the order of ∼ 100 kA (i.e. ∼ 1 nC 

charge within ∼ 10 fs) can be reached 

without the need for any bunch compressor. 

For preparation of the required 

gas densities, either supersonic gas-jets 

or capillaries [11] are used, the latter 

consisting of a ∼ 300 µm-thin gas-filled 

channel with a parabolic radial ion density 

profile generated by a second laser 

pulse or discharge. The advantage of 

the capillary is that the laser can be 

guided beyond the Rayleigh length allowing 

longer acceleration distances 

and hence higher electron energies. The 

very recent experiments by Leemans et 

al. [4], utilizing a 40-TW laser pulse 

of 38 fs duration and a gas density 

of 4.3 × 10 18 cm −3 , clearly show that 

1 GeV electron beams can now be produced 

with capillaries. However, the 

measured charge of 30 pC is significantly 

below our design goal of 1 nC. 

In order to further improve the resulting 

current, we propose the scheme described 

below. The number of bubble 

electrons can be increased when using 

plasmas with higher density such as in 

Fig. 1, because the feeding process is 

more efficient if more plasma electrons 

are present. An increased gas density 

requires shorter laser-pulses (sub-10fs), 

such that the entire laser pulse fits 

into one plasma period. In this case, the 

entire laser pulse energy can be used, 


Rapid communication

432 Applied Physics B – Lasers and Optics 


FIGURE 1 (Color online) Snapshot of electron density (in units of 10 20 cm −3 ) from PIC simulation. 

The geometrical size of the electron-free cavity (“bubble”) behind the laser spot corresponds to 

the plasma period, which is about 8 µm in this case (gas density 1.8 × 10 19 cm −3 ). The stem of the 

high-energy electrons is much shorter than the plasma period 

while longer pulses lose energy during 

self-shortening for entering the bubble 

regime. Therefore, in contrast to all previous 

laser-acceleration experiments, 

we propose to use a different driver 

laser, namely a ∼ 5 fs-Petawatt laser 

(like the Petawatt-field-synthesizer PFS 

at MPQ [12]). Own PIC simulations 

show that such ultrashort lasers can capture 

more than 1 nC charge inside the 

bubble. The smaller plasma period leads 

to smaller bubbles, hence shorter bubble 

stems and thus higher currents (1 nC 

in 10 fs). However, the expected final 

energy in the bubble regime is, according 

to the scaling laws, lower for shorter 

laser pulses. To overcome this limit, we 

also propose to use a capillary and suggest 

the following scenario therein: with 

a longitudinal plasma density gradient, 

the laser pulse can be forced to gradually 

increase its diameter, thus adiabatically 

turning from the pure bubble regime 

into the pure blow-out regime long 

before the laser is depleted. Figure 1 

shows a snapshot of a bubble around 

the transition from the bubble to the 

blow-out regime, where the electrons 

blown away from the laser and flowing 

around the bubble are so strongly 

deflected by the electric field of the captured 

bubble electrons, that electrons 

are no longer scattered into the bubble 

(which is not charge-neutralized yet). 

The number and absolute energy spread 

of the bubble electrons is thus frozen, 

but their energy is still increased due to 

the present bubble fields. The remaining 

energy of the laser allows maintaining 

of the bubble structure for the remaining 

distance inside the capillary, where 

the laser is then guided along to overcome 

the Rayleigh limit. It is only for 

this stage with increased laser beam 

size, that the capillary-induced guiding 

is relevant. Note that this scheme 

is a two-stage approach, but within one 

and the same capillary, where the laser 

turns adiabatically from one stage into 

the other. It can be expected that the energy 

spread for 100 MeV is about 1% 

(as confirmed by measurements [2]), 

but 0.1% for 1 GeV 1 . Normalized emittances 

are found both from simulations 

and experiments to range between 

0.1–1 mm mrad. 

1 Note that the detector resolution in the cited 

GeV-experiment did not suffice to resolve energy 

spreads below one percent. 

2 Table-top 

free-electron-lasers 

An FEL requires an undulator 

which is an arrangement of magnets 

with an alternating transverse magnetic 

field. Electrons in an undulator are 

forced on a sinusoidal trajectory and 

can thus couple with a co-propagating 

radiation field. The induced energy 

modulation yields a current modulation 

from the dispersion of the undulator 

field. This modulation is called 

micro-bunching expressing the fact that 

the electrons are grouped into small 

bunches separated by a fixed distance. 

Therefore, electrons emit coherent radiation 

with a wavelength equal to 

the periodic length between the microbunches. 

In a self-amplification of spontaneous 

emission (SASE) FEL, there is 

no initial radiation field and the seed 

has to be built up by the spontaneous 

(incoherent) emission [13]. We present 

quantitative arguments, which are complemented 

by SASE FEL (GENESIS 

1.3 [14]) simulations. The gain length, 

which is the e-folding length of the exponential 

amplification of the radiation 

power, is 

Lgain,ideal = λu 

4π √ , (1) 

3� 

with undulator period λu and the basic 

scaling parameter �. This Pierce or FEL 

parameter describes the conversion efficiency 

from the electron beam power 

into the FEL radiation power and reads 

for the one-dimensional and ideal case 

(neglecting energy spread, emittance, 

diffraction, and time-dependence) 

[13, 15] 

� = 1 

� � � � 

2 

1/3 

I λu Au 

. (2) 

2γ IA 2πσx 

Here γ = Ebeam/mc2 is the electron 

beam energy, I the beam current, IA = 

17 kA the Alfven-current, σx the beam 

size and Au = au[J0(ζ) − J1(ζ)] (planar 

undulator), whereby a2 u = K 2 /2, 

K is the undulator parameter (K = 

0.93λu[cm]B0[T] and B0 the magnetic 

field strength on the undulator axis), 

ζ = a 2 u /(2(1 + a2 u 

)), and J are Bessel 

functions. In the presence of energy 

spread, emittance, and diffraction, a correction 

factor Λ is introduced for the 

gain length [15]: Lgain = Lgain,ideal(1 + 



GRÜNER et al. Design considerations for table-top, laser-based VUV and X-ray free electron lasers 433 

Λ). The saturation length Lsat is the 

length along the undulator at which 

maximum micro-bunching is reached. 

The power of the FEL radiation at this 

point is the saturation power Psat, while 

the shot noise power Pn is the power 

of the spontaneous undulator radiation 

from which the SASE FEL starts. The 

coupling factor α = 1/9 describes the 

efficiency at which the noise power couples 

to the FEL gain process. The saturation 

length reads 

� � 

Psat 

Lsat = Lgain ln , (3) 

αPn 

where the saturation power scales as 

� �2 1 

Psat ∼ (Iλu) 

1 + Λ 

4/3 . (4) 

The FEL-wavelength is, in case of a planar 

undulator, 

λ = λu 

2γ 2 

� 

1 + 

2 � 

K 

. (5) 

2 

Thus, for reaching a certain wavelength 

λ, a shorter undulator period λu allows 

the use of less energetic electrons. In the 

following Table 1 and Fig. 2, we compare 

the FLASH VUV FEL (DESY) in 

the so-called femtosecond mode [16] 

with our corresponding table-top VUV 

FEL as well as a table-top XFEL operating 

with 1.2 GeV electrons, as discussed 

below. 

The importance of the ultra-high 

current in the table-top case is evident. 

The smaller undulator period allows 

a smaller beam energy, hence decreasing 

the gain and saturation lengths. The 

Pierce parameter gives the upper limit 

of the acceptable energy spread σγ /γ . 

Thus, for compensating the relatively 

large energy spread of laser-plasma accelerators 

and for maintaining a large 

output power, a beam current significantly 

above ∼ 17 kA is mandatory for 

keeping � large and Λ small. 

2.1 Space charge effects 

Such ultra-high beam currents 

are subject to strong space-charge 

forces. After release into vacuum, the 

electron bunch starts expanding, for 

which there are two sources: (i) its 

own space-charge, driving a Coulombexplosion, 

i.e. (transverse) space-charge 

expansion and (longitudinal) debunching, 

and (ii) its initial divergence, which 

drives a linear transverse expansion. 

Such expansions transform potential energy 

into kinetic, hence changing the 

initial energy distribution and bunch 

form. First, we want to discuss an extreme 

case, that is, an upper limit of the 

charge that can be expected from a laserplasma 

accelerator at lower energies. 

For this study we take a beam energy of 

γ0 = 260, 1.25 nC charge, and a Gaussian 

bunch with sizes σx,y,z = 1 µm, corresponding 

to I = 150 kA, and an initial 

divergence of θ0 = 1 mrad. In the rest 

frame of the bunch, its length amounts 

Parameter FLASH (fs) TT-VUV-FEL TT-XFEL 

current 1.3 kA 50 kA 160 kA 

norm. emitt. 6 mm mrad 1 mm mrad 1 mm mrad 

beam size 170 µm 30 µm 30 µm 

energy 461.5 MeV 150 MeV 1.74 GeV 

energy spread 0.04% 0.5% 0.1% 

und. period 27.3 mm 5 mm 5 mm 

wavelength 30 nm 32 nm 0.25 nm 

Pierce par. 0.002 0.01 0.0015 

sat. length 19 m 0.8 m 5 m 

pulse length 55 fs 4 fs 4 fs 

sat. power 0.8 GW 2.0 GW 58 GW 

TABLE 1 Parameters for the comparison between the DESY femtosecond-mode Flash-VUV case 

and the corresponding table-top VUV FEL as well as a table-top X-ray FEL (rms values) 


FIGURE 2 Saturation lengths (3) 

and (inset) degradation factor Λ as 

a function of electron beam current 

I: DESY’s FLASH (dashed curve, 

triangle) and the table-top VUV scenario 

(solid line). The circles denote 

specific GENESIS runs 

to σ � z = σzγ0 = 260 µm. For such an 

aspect-ratio, the transverse Coulombexplosion 

dominates over the longitudinal 

one (GPT [17] simulations reveal 

that after a time period of 1 ps 

σ � x = 73 µm, while σ � z = 268 µm). Figure 

3 shows the electron distribution 

in the bunch rest frame from the GPT 

simulations. 

It can be seen that the transverse 

Coulomb explosion dominates the longitudinal 

one. We want to emphasize 

here that studies of space-charge effects 

of such ultra-dense relativistic bunches 

can be simulated most accurately only 

with codes (such as GPT and CSRtrack 

[18]) which utilize point-to-point 

interactions (a result also found in [19]). 

In contrast, other codes merely based 

upon Poisson-solvers cannot cope with 

large relative particle motions within the 

bunch rest frame in case of large initial 

energy spreads and Coulomb explosioninduced 

motion. In comparison with the 

Coulomb-driven explosion, the linear 

divergence-driven transverse expansion 

is even larger: after a distance of 1 cm 

in the laboratory frame, the beam size 

is increased to 10 µm, while the transverse 

Coulomb-explosion yields 6 µm 

only. Hence, even in the extreme case 

the transverse bunch expansion is dominated 

by the initial divergence and for 

lower charges this dominance increases. 

If γ0 is the bunch energy just before 

release into vacuum, γ � and β� the energy 

and velocity of a test electron in 

the bunch rest frame, γ and β ≈ 1 for 

its laboratory frame energy and velocity, 

then one can clearly distinguish 

between a transverse-dominated expansion 

and a longitudinal-dominated 

one: in the first case, where the electron 

moves in the rest frame with velocity 

±β� along transverse direction, 

γ = γ0γ � , while in the latter case, where

434 Applied Physics B – Lasers and Optics 


FIGURE 3 (Color online) Projected spatial electron distribution within the bunch rest frame in the 

extreme case, at τ = 0 s (left) and τ = 4 ps (right), from a GPT run. The color code indicates the electron 

energy (blue is initial kinetic energy, red is increased energy). The transverse expansion is clearly 

dominating 

the electron moves with velocity ±β � 

in the beam direction, γ = γ0γ � (1 ± β � ). 

Hence, in a purely transverse expansion 

all electrons gain kinetic energy, 

while in a purely longitudinal expansion, 

some electrons lose energy. The 

inset of Fig. 4 shows the absolute kinetic 

electron energy distribution along 

the bunch at a distance of z = 3 cm after 

its release into vacuum (in case of 0.6 

nC). This spectrum shows that the bunch 

under study mostly undergoes a transverse 

expansion. The maximum energy 

gain lies in the middle of the bunch, 

where the density is highest, i.e., where 

initially the highest potential energy was 

stored. That the space-charge driven 

transverse expansion is weaker than the 

divergence-driven one is also confirmed 

by the fact that (even in the extreme 

case) the divergence is increased only 

to 1.3 mrad. After passing a focusing 

system (a triplet of quadrupoles, positioned 

at z = 4 cm, having a length of 

about 10 cm), the Gaussian-like energy 

distribution is transformed into a quasilinear 

one, as depicted in Fig. 4. The 

reason is that in the resulting quasicollimated 

low-divergence beam faster 

electrons can catch up the slower ones. 

Note that after z = 3 cm (inset) the 

fastest electrons are symmetrically distributed 

around the middle of the bunch. 

These electrons start overtaking the 

slower ones, that is, they move forward 

within the bunch. If initially longitudinal 

debunching dominates, the 

fastest electrons are found in the head 

and the slowest ones in the tail of the 

bunch. 

A key parameter for SASE to occur 

is the energy spread σγ /γ , which must 

always be smaller than the Pierce parameter 

�. The relevant energy spread 

is taken over a bunch slice with the size 

of one cooperation length, i.e. the slippage 

length of the radiation along the 

bunch over one gain length. As seen in 

Fig. 4, a large fraction of the bunch fulfills 

σγ /γ < �. Due to the slippage of the 

radiation relative to the bunch, the linear 

energy chirp must be compensated, 

because the energy factor γ in (5) increases 

along the bunch. However, in 

case of short-period undulators the gap 

is correspondingly small (typically one 

third of the undulator period) and such 

small gaps lead to strong wakefields 

also causing an energy variation along 

the bunch [23, 24]. In case of ultrashort 

FIGURE 4 Energy vs. position 

(in co-moving frame) along the 

bunch before entering the undulator 

(GPT simulation): the 

Gaussian-like distribution after 

a drift of z = 3 cm behind the gasjet 

(inset), indicating a transversedominated 

bunch expansion, is 

stretched into a quasi-linear one 

behind the focusing system. The 

vertical lines show the slice energy 

spread σγ and the horizontal lines 

mark the initial energy γ0 = 260 

bunches, the characteristic length of the 

resistive wake potential is longer than 

the bunch length and, hence, the energy 

change is found to be negative 

with a linear few-percent variation along 

the bunch. This in turn implies that the 

bunch as a whole is decelerated during 

the passage of the undulator, whereby 

the head loses less energy than the tail. 

It turns out that the linear energy chirp 

induced by space-charge corresponds 

well with the wakefield-induced slowing 

down of the entire electron bunch. 

Therefore, the radiation slipping in the 

forward direction interacts with electrons 

of effectively constant energy, if 

by varying the gap the bunch energy 

loss is tuned with respect to the slippage. 

In other words, the space-charge 

effects and the wakefield effects cancel 

each other. We have found with simulations 

that for a first proof-of-principle of 

a table-top FEL the comparably low energy 

of 130–150 MeV is well suited to 

cope with the wakefields in the undulator. 

In this sense the table-top VUV case 

given in Table 1 can be regarded as an 

optimal test case. 

Coherent synchrotron radiation 

(CSR) [20] as another source for increasing 

the (slice) energy spread within 

the undulator was found to have negligible 

impact. This was shown with CSRtrack 

simulations which take into account 

that in our cases the bunches have 

a much larger transverse size (σx,y ∼ 

30 µm) than length (σz ∼ 1 µm). 

2.2 Table-top XFEL 

So far we have discussed 

a proof-of-principle scenario at relatively 

low electron energies, where 

space-charge effects play a dominant 

role leading to a linear energy 

chirp. The situation of the proposed 

table-top X-FEL (TT-XFEL) is different. 

For reaching a wavelength of 

λ = 0.25 nm, an electron energy of 

1.74 GeV is needed in case of a period 

of λu = 5 mm. Space charge effects are 

much weaker here. For this less demanding 

situation we have confirmed 

with four different simulation codes 

(GPT, ASTRA [21], CSRtrack, HOM- 

DYN [22]) that above 1 GeV, with the 

same parameters as in the extreme case 

given above, Coulomb-explosion leads 

to a projected energy chirp of below 

0.3% and a bunch elongation with a fac- 



GRÜNER et al. Design considerations for table-top, laser-based VUV and X-ray free electron lasers 435 

tor below 1.1. Furthermore, CSR can 

be neglected in agreement with the 1D 

theory. Experimentally the most demanding 

constraint is that the Pierce 

parameter is one order of magnitude 

smaller than in the test case at 130 MeV, 

hence the electron (slice) energy spread 

should be as small as 0.1%. This goal 

seems to be within reach considering the 

bubble-transformation scenario mentioned 

above. 

Without the effect of wakefields 

GENESIS simulations have shown that 

this TT-XFEL scenario with an undulator 

length of only 5 m yields 8 × 10 11 

photons/bunch within ∼ 4 fs and 0.2% 

bandwidth, a divergence of 10 µrad, and 

a beam size of 20 µm. However, the 

wakefields become the dominant degrading 

effect as the required undulator 

length is larger and the Pierce parameter 

reduced, hence also the tolerance with 

respect to variations in the electron energy. 

But since there is no initial spacecharge-induced 

energy chirp, one must 

find another method for compensating 

the wakefield-induced energy variation. 

A suitable method for compensating 

the wakefields for the TT-XFEL would 

be tapering, i.e., varying the undulator 

period along the undulator. Due to the 

fact that the undulator parameter K is 

smaller than unity, tapering via K, i.e., 

by gap variation, could only be used as 

fine-tuning. Depending on the specific 

material properties of the undulator surface, 

our first wake calculations show 

that the bunch center loses over the entire 

5 m undulator length about 10% 

of its initial energy. Due to the linear 

wake field variation along the bunch 

only a fraction of it will undergo SASE. 

A detailed study will be further investigated 

in a future paper. 

2.3 Experimental status 

In order to demonstrate practical 

feasibility, we have built and tested 

a miniature focusing system and undulator 

consisting of permanent mag- 

nets. The focusing triplet consists of 

mini-quadrupoles with an aperture of 

just 5 mm, hence allowing for measured 

gradients of 530 T/m. Their focusing 

strength was measured with a 600 MeV 

electron beam (at Mainz Microtron facility 

MAMI, Mainz, Germany). A first 

test hybrid undulator with a period of 

only 5 mm and a peak magnetic field 

strength of ∼ 1 T has been built, and 

produced, with a 855 MeV beam (also 

at MAMI), an undulator radiation spectrum 

as expected. These results will be 

published elsewhere. 

3 Conclusion 

We have shown by means 

of analytical estimates and SASE FEL 

simulations that laser-plasma accelerator-based 

FELs can only be operated 

with meter-scale undulators. The key 

parameter is the ultra-high electron peak 

current, which significantly reduces the 

gain length and increases the tolerance 

with respect to the energy spread and 

emittance. The latter would also allow 

increasing of the TT-XFEL photon 

energy into the medically relevant 

range of 20–50 keV, because the limiting 

quantum fluctuations in the spontaneous 

undulator radiation, scale with 

γ 4 [25] and the required electron energy 

of the TT-XFEL is comparatively 

small. 

ACKNOWLEDGEMENTS We thank 

W. Decking, M. Dohlus, K. Flöttmann, T. Limberg, 

and J. Rossbach (all DESY) for fruitful 

discussions. Supported by Deutsche Forschungsgemeinschaft 

through the DFG-Cluster of Excellence 

Munich-Centre for Advanced Photonics. 

This work was also supported by DFG under Contract 

No. TR18. 

REFERENCES 

1 S.P.D. Mangles, C. Murphy, Z. Najmudin, 

A. Thomas, J. Collier, A. Dangor, E. Divall, 

P. Foster, J. Gallacher, C. Hooker, 

D. Jaroszynski, A. Langley, W. Mori, P. Norreys, 

F. Tsung, R. Viskup, B. Walton, 

K. Krushelnick, Nature 431, 535 (2004) 


2 C.G.R. Geddes, C. Toth, J. van Tilborg, 

E. Esarey, C. Schroeder, D. Bruhwiller, 

C. Nieter, J. Cary, W. Leemans, Nature 431, 

538 (2004) 

3 J. Faure, Y. Glinec, A. Pukhov, S. Kiselev, 

S. Gordienko, E. Lefebvre, J.-P. Rousseau, 

F. Burgy, V. Malka, Nature 431, 541 

(2004) 

4 W.P. Leemans, B. Nagler, A. Gonsalves, 

C. Toth, K. Nakamura, C. Geddes, E. Esarey, 

C. Schroeder, S. Hooker, Nature Phys. 2, 696 

(2006) 

5 R. Neutze, R. Wouts, D. van der Spoel, 

E. Weckert, J. Hajdu, Nature 406, 752 

(2000) 

6 A. Baltus, T. Udem, M. Uiberacker, 

M. Hentschel, E. Goulielmakis, C. Gohle, 

R. Holzwarth, V.S. Yakovlev, A. Scrinzi, 

T.W. Hänsch, F. Krausz, Nature 421, 611 

(2003) 

7 W. Thomlinson, P. Suortti, D. Chapman, 

Nucl. Instrum. Methods Phys. Res. A 543, 

288 (2005) 

8 A. Pukhov, J. Meyer-ter-Vehn, Appl. Phys. B 

74, 355 (2002) 

9 A. Pukhov, S. Gordienko, Philos. Trans. R. 

Soc. London A 364, 623 (2006) 

10 M. Geissler, J. Schreiber, J. Meyer-ter-Vehn, 

New J. Phys. 8, 186 (2006) 

11 D.J. Spence, A. Butler, S. Hooker, J. Opt. 

Soc. Am. B 20, 138 (2003) 

12 http://www.attoworld.de/research/PFS.html 

13 R. Bonifacio, C. Pellegrini, L.M. Narduci, 

Opt. Commun. 50, 373 (1984) 

14 S. Reiche, Nucl. Instrum. Methods Phys. 

Res. A 429, 243 (1999) 

15 M. Xie, Nucl. Instrum. Methods Phys. Res. A 

445, 59 (2000) 

16 Technical Design Report: 

http://www.hasylab.desy.de/facility/fel/vuv/ 

ttf2_update/TESLA-FEL2002-01.pdf 

17 M.J. de Loos, S.B. van der Geer, Proc. 5th 

Eur. Part. Acc. Conf., Sitges, 1241 (1996) 

18 M. Dohlus, T. Limberg, Proc. FEL 2004 

Conf., 18-21 

19 G. Fubiana, J. Qiang, E. Esarey, W.P. Leemans, 

Phys. Rev. ST Accel. Beams 9, 064 402 

(2006) 

20 E. Saldin, E.A. Schneidmiller, M.V. Yurkov, 

Nucl. Instrum. Methods Phys. Res. A 417, 

158 (1998) 

21 K. Floettmann, Astra User’s Manual, 2000, 

http://www.desy.de/mpyflo/ 

22 M. Ferrario, J.E. Clendenin, D.T. Palmer, 

J.B. Rosenzweig, L. Serafini, Proc. of the 2nd 

ICFA Adv. Acc. Workshop on “The Physics 

of High Brightness Beams”, UCLA, Nov., 

1999; see also SLAC-PUB-8400 

23 A.W. Chao, Physics of Collective Beam Instabilities 

in High Energy Accelerators (Wiley, 

New York, 1993) 

24 K. Bane, SLAC-PUB-11829 (2006) 

25 M. Sands, Phys. Rev. 97, 470 (1955)

Vol 446 | 5 April 2007 |doi:10.1038/nature05648 


Attosecond real-time observation of 

electron tunnelling in atoms 

ARTICLES 

M. Uiberacker 1,2 , Th. Uphues 3 , M. Schultze 2 , A. J. Verhoef 2,4 , V. Yakovlev 1 , M. F. Kling 5 , J. Rauschenberger 1,2 , 

N. M. Kabachnik 3,6 , H. Schröder 2 , M. Lezius 2 , K. L. Kompa 2 , H.-G. Muller 5 , M. J. J. Vrakking 5 , S. Hendel 3 , 

U. Kleineberg 1 , U. Heinzmann 3 , M. Drescher 7 & F. Krausz 1,2,4 

Atoms exposed to intense light lose one or more electrons and become ions. In strong fields, the process is predicted to occur 

via tunnelling through the binding potential that is suppressed by the light field near the peaks of its oscillations. Here we 

report the real-time observation of this most elementary step in strong-field interactions: light-induced electron tunnelling. 

The process is found to deplete atomic bound states in sharp steps lasting several hundred attoseconds. This suggests a new 

technique, attosecond tunnelling, for probing short-lived, transient states of atoms or molecules with high temporal 

resolution. The utility of attosecond tunnelling is demonstrated by capturing multi-electron excitation (shake-up) and 

relaxation (cascaded Auger decay) processes with subfemtosecond resolution. 

With the invention of the laser in 1960 and subsequent advances in 

ultrashort light pulse generation, light fields with a strength of several 

volts per angstrom have become routinely available. They rival the 

fields acting on valence electrons in atomic systems, allowing their 

release from atoms, molecules and solids. These advances sparked a 

revolution in studying the interaction of electrons with light. The 

primary step in strong-field interactions is the liberation of electrons 

from their atomic bound state. The revolutionary theory of Keldysh1 and subsequent work2–6 suggested that a valence electron may escape 

by tunnelling through its atomic binding potential suppressed by the 

light field (Fig. 1a). If the dimensionless parameter: 

c ~ vL 

pffiffiffiffiffiffiffiffiffiffiffiffi 

2mWb 

ð1Þ 

jejE0 is less than one, under the assumption of "vL = Wb ionization is 

predicted to be confined to short intervals lasting a fraction of the half 

oscillation cycle of the light field (Fig. 1b). Here E0 and vL stand for 

the amplitude and angular frequency of the oscillations of the laser 

electric field EL(t) 5 E0e(t)cos(vLt 1 Q), with e(t) being the amplitude 

envelope function, and e, m and Wb the charge, mass and binding 

energy of the electron. Recent studies6 suggest that tunnelling 

remains the dominant ionization mechanism even for c substantially 

exceeding one, that is, under conditions when the potential barrier 

formed by the atomic binding potential and the ionizing light field 

varies during tunnelling (non-adiabatic regime). 

In this work we report what we believe is the first real-time observation 

of light-induced electron tunnelling. The observation of ionization 

occurring in subfemtosecond steps spaced by the half laser 

cycle up to values of the Keldysh parameter as high as three is in good 

agreement with analytic and numerical calculations. Our approach 

provides experimental access to all forms of optical field ionization— 

both adiabatic and nonadiabatic tunnelling, as well as barriersuppression 

ionization—and allows us to test models of these processes 

for the first time. 

Once the process of field ionization is fully understood, the technique 

of attosecond tunnelling will provide direct time-domain 

insight into a wide range of multi-electron dynamics and electron– 

electron interactions, ultimately with a resolution approaching 

the atomic unit of time (,24 as). We demonstrate this potential by 

probing shake-up and Auger cascade processes with subfemtosecond 

resolution. 

Attosecond probing of electron dynamics 

Figure 2 illustrates different options for attosecond sampling of electronic 

motion in atoms or molecules. A subfemtosecond extreme 

ultraviolet (XUV) pulse triggers the motion by exciting a valence 

or core electron (Fig. 2a, b). The unfolding excitation and relaxation 

processes (Fig. 2) could, in principle, be probed by a delayed replica 

of the pulse. However, the low flux of currently available subfemtosecond 

XUV pulses and the low two-photon transition probabilities 

in the XUV and X-ray regimes have thwarted this straightforward 

extension of conventional pump–probe techniques into the XUV– 

X-ray spectral range. 

A few-cycle wave of visible or near-infrared (NIR) light with controlled 

waveform 7 in combination with a highly nonlinear process 

may replace the subfemtosecond pulse either in probing or starting 

electron dynamics. This was first demonstrated by attosecond streaking 

8,9 : the strong-field interaction of a few-cycle light wave with free 

electrons released by a subfemtosecond XUV excitation pulse results 

in broadening and shifting of their final momentum distribution. 

Recording the streaked spectra of the emitted photo- and Auger 

electrons versus delay between the XUV pump and the few-cycle laser 

probe allowed us to retrieve the XUV pulse 10,11 and the laser field 12 as 

well as the inner-atomic relaxation dynamics 13 with subfemtosecond 

resolution. 

Here, we demonstrate that nonlinear interaction of the same light 

wave with bound electrons ionizes in subfemtosecond steps and 

hence offers a means of probing intra-atomic and intra-molecular 

electron dynamics—including when no free electrons are released— 

by means of attosecond tunnelling. This approach relies on the fact 

that energetic photo-excitation as well as subsequent rearrangement 

1 Department für Physik, Ludwig-Maximilians-Universität, Am Coulombwall 1, 2 Max-Planck-Institut für Quantenoptik, Hans-Kopfermann-Strasse 1, D-85748 Garching, Germany. 

3 Fakultät für Physik, Universität Bielefeld, Universitätsstrasse 25, D-33615 Bielefeld, Germany. 4 Technische Universität Wien, Gusshausstrasse 27, A-1040 Vienna, Austria. 5 FOM- 

Instituut voor Atoom- en Molecuulfysica (AMOLF), Kruislaan 407, 1098 SJ, Amsterdam, The Netherlands. 6 Institute of Nuclear Physics, Moscow State University, Moscow 119992, 

Russia. 7 Institut für Experimentalphysik, Universität Hamburg, Luruper Chaussee 149, D-22671 Hamburg, Germany. 

©2007 Nature Publishing Group 

Max-Planck-Institut für Quantenoptik • Progress Report 2007/2008 165 

627


ARTICLES NATURE |Vol 446 | 5 April 2007 

via Auger decay is accompanied by transitions to unoccupied orbitals 

via ‘shake-up’ (Fig. 2d–f) 14–16 . Here, we use the term ‘shake-up’ in a 

broad sense, standing for all possible processes populating excited 

ionic states (including instantaneous and non-instantaneous ones), 

henceforth referred to as shake-up states (represented by levels 1, 2 

and 3 in the example of Fig. 2). These populations can be probed via 

optical field ionization by a strong, few-cycle NIR pulse of variable 

delay Dt, with respect to the subfemtosecond XUV excitation (Fig. 1c) 

by measuring the number of ions resulting from the XUV pump–NIR 

probe exposure as a function of Dt (Fig. 1d). 

Shake-up usually populates several quantum states in the valence 

band, from which electrons can be freed by the NIR probe. Hence, the 

ion yield will constitute an integral signal, with contributions from all 

shake-up states up to a certain binding energy from which ionization 

is feasible for the intensity chosen. Population dynamics of individual 

states of significantly differing binding energy can be retrieved by 

pump–probe scans repeated at different NIR probe intensities and/ 

or from the temporal separation of the depletion of the states in the 

same delay scan, as explained in Fig. 1d and demonstrated in Fig. 4. 

The ion yield constitutes an integral signal in a temporal sense, too. 

The shake-up states are exposed to the ionizing NIR field from the 

moment they have been populated until the end of the NIR pulse. The 

a 

W 

b 

0 

–W b 

Ionization 

rate 

e – 

0 x 

Core 

Time 

c 

XUV 

pump 

∆t (> 0) 

E L (t)=E 0 ε(t)cos(ω L t+ϕ) 

d 

Ionization 

yield 

T L /2 

Depletion of states 

with binding energy 

W 1 and W 2 

Delay time, ∆t 

beginning of this time interval within the NIR pulse is adjusted with 

the delay Dt between the XUV pump and the NIR probe. As the delay 

is scanned from large negative values (NIR probe first) to large positive 

values (XUV pump first) the measured ion yield (Fig. 1d) starts 

increasing at Dt , 0 owing to ionization on the trailing edge of the 

NIR pulse and continues to increase with increasing Dt because the 

XUV pump shifts towards the peak of the ionizing NIR probe and the 

shake-up states are exposed to ever higher NIR probe intensities. 

In the absence of Auger decay, shake-up excitation results from 

photoionization only (Fig. 2d). The time-dependent ionization 

dynamics sketched in Fig. 1d can then be traced by measuring the 

yield of doubly charged ions as a function of the delay between the 

XUV pump and the sampling NIR light field. The temporal ionization 

gradients sketched in Fig. 1d incorporate the finite duration of 

the XUV excitation, a possible delayed response of shake-up and the 

tunnelling dynamics. With a sufficiently rapid (=1 fs) excitation, 

these measurements can thus provide direct insight into the temporal 

evolution of shake-up (in the presence of a strong optical field) and 

light-induced tunnelling. In what follows, first we prove this concept 

by exposing neon atoms to our subfemtosecond XUV pump and fewcycle 

NIR probe pulses (Fig. 3) and measuring the yield of the product 

of the pump–probe exposure, Ne 21 , versus Dt (Fig. 4). Then we 

extend the approach to probing electrons shaken up in xenon atoms 

during Auger decay. Our primary observables in this case are higher 

charged states, Xe N1 (for N . 2). Measuring their yield versus Dt 

displays the course of an Auger cascade. 


NIR 

probe 

Figure 1 | Strong-field ionization and pump-probe setting for its real-time 

observation. a, Exposing an atom to a strong NIR laser field will result in a 

modified potential (solid curve) composed of the Coulomb potential 

(dashed curve) and the time-dependent effective potential of the laser pulse. 

The laser is polarized along the x direction and W b is the binding energy of 

the electron. At sufficiently high laser field strengths the atomic binding 

potential is suppressed to a small barrier in the x or –x direction for the 

maxima and minima of the laser electric field, respectively, allowing optical 

tunnelling to become the dominant ionization mechanism. b, The highly 

nonlinear dependence of the tunnelling rate on the width of the potential 

barrier confines ionization to time intervals of very short duration near the 

field oscillation maxima. In the electric field of a few-cycle NIR laser pulse 

(thin line) ionization is predicted to be restricted to several subfemtosecond 

intervals (thick line). c, Concept for tracing optical-field ionization: a 

subfemtosecond XUV pulse generates ions in excited (shake-up) states, from 

which a time-delayed NIR few-cycle probe pulse liberates electrons to 

produce doubly charged ions. A delay Dt 5 0 is defined as the coincidence of 

the peaks of the envelopes of the NIR and XUV pulse. Dt . 0 implies that the 

peak of the XUV pulse envelope precedes that of the NIR pulse. d, Yield of 

doubly charged ions versus delay Dt between the XUV pump and the NIR 

probe, as predicted qualitatively for the case of electrons being prepared in 

two shake-up states of significantly differing binding energy (W 1 and 

W 2 ? W 1) for liberation via strong-field ionization (see text for discussion). 

The state of low binding energy is depleted at low intensities, where multiphoton 

ionization dominates and hence sub-cycle structure in the ionization 

dynamics is absent. By contrast, the state of high binding energy is emptied 

by tunnelling, resulting in pronounced sub-half-cycle steps in the ionization 

profile. 

628 

Kinetic energy 

0 

Binding energy 


+1 +1 

Valence Core-level 

photophotoemissionemission Probing by attosecond 

Streaking 

+2 

Auger 

decay 

1 

2 

+1 +1 

Tunnelling 

Photo-emission 

and shake-up 

a b c d e f 

3 

Unoccupied 

valence 

Occupied 

valence 

Core 

orbital 

+2 Final 

Auger charge 

decay and state 

shake-up 

Figure 2 | Probing electron dynamics in atoms, molecules or solids with 

attosecond sampling techniques. A subfemtosecond XUV pulse triggers the 

motion by inducing valence (process a) or core photoelectron emission 

(process b). The temporal evolution of photo- and Auger electron emission 

(process c) can be probed via attosecond streaking to retrieve the XUV pump 

pulse or the sampling NIR field and trace inner-shell relaxation dynamics. 

XUV photoexcitation as well as subsequent Auger decay processes are 

usually accompanied by shake-up of another electron to a previously 

unoccupied level (processes d, e and f). In this case the liberated electrons 

will be ejected at a reduced kinetic energy compared to the cases without 

shake-up processes a, b and c. The difference in energy is used for shaking up 

bound electrons (represented by curved black arrows to levels 1, 2 and 3). 

For sufficiently strong probing laser fields, the shake-up electrons can be 

liberated by tunnelling ionization. The temporal evolution of the tunnelling 

current will provide information about the inner-atomic electron dynamics 

that populates and/or depopulates the interrogated shake-up states and the 

duration of the process that have populated the levels on atto- and 

femtosecond timescales. Note that the final charge state given in the figure is 

increased by attosecond tunnelling, while it remains unchanged in the case 

of attosecond streaking. The observable for streaking is the momentum 

distribution of liberated electrons, whereas in tunnelling it is the number of 

ions in different charge states.

Experimental set-up 

For a detailed description of the attosecond pump–probe apparatus, 

see the Supplementary Information. Briefly, the XUV pump originates 

from high-harmonic generation in a neon gas jet exposed to 

300-mJ, 750-nm waveform-controlled laser pulses with a duration of 

tL < (5.5 6 0.5) fs, at a repetition rate of 3 kHz. The collinear, linearly 

polarized XUV and NIR light beams are passed through several 

filters and reflected by a concentric double-mirror arrangement 

(Mo/Si multilayer: , 9 eV bandwidth at photon energy of 91 eV). 

The mirror introduces a variable delay between the XUV and the NIR 

fields, isolates a single or twin XUV pulse (depending on the carrierenvelope 

phase of the laser pulse) of tX < 250 as duration 11 by filtering 

the cut-off part of the harmonic emission spectrum and focuses 

both beams into a jet of atoms under scrutiny. The absolute delay 

between the XUV and the NIR signals is determined with an accuracy 

of better than 60.5 fs; for details see Supplementary Information. 

Ions created in the common focus of the two beams are detected 

by a time-of-flight ion spectrometer (reflectron) that combines 

high mass resolution (Dm/m < 10 23 ) with the capability of analysing 

Ne 

3s 

9.7% 

3p 

0.2% 

4p 

3d 4s 

4d 

1.0% 

12.3% 

2s –1 

2p –1 


NATURE | Vol 446 |5 April 2007 ARTICLES 

Energy (eV) 

70 

60 

50 

40 

30 

20 

10 

0 

62.53 eV 

48.47 eV 

21.56 eV 

Shake-up satellites 

51.3% 

2p –2 25.0% nl 

XUV + IR (∆t > 0): 93.3% 

XUV: 95.2% 

Ne + 

1.9% 

( 1 S) 

( 1 D) 

( 3P) 2p –2 

6.7% 

4.8% 

Ne 2+ 

Figure 3 | Energy levels and transitions in Ne 11 and Ne 21 ions relevant to 

this study. The plotted levels represent energies required to ionize and 

possibly excite a neutral atom from its ground state. Absorption of photons 

of about 91-eV energy can produce singly or doubly charged ions with 

probabilities of 95.2% and 4.8%, respectively. Owing to shake-up a small 

fraction of the singly-charged ions is produced in 2p 22 nl configurations, 

where the probabilities of different channels are known from electron 

spectroscopy (see Supplementary Information). Each of these 

configurations, represented by a green box, consists of 2p 22 ( 3 P)nl, 

2p 22 ( 1 D)nl, and 2p 22 ( 1 S)nl states (the atomic terms in parentheses describe 

electrons not involved in the interaction). Our few-cycle laser field following 

the XUV pulse (Dt . 0) can remove electrons from these shake-up states, 

thus increasing the probability of double ionization to 6.7%. Depending on 

the initial 2p 22 nl state, the doubly ionized ion is left in one of the 3 P, 1 D or 1 S 

states. On its own, the laser pulse produces only singly charged ions (in 

configuration 2p 21 ) above the detection limit. 

particles within a micrometre-scale detection volume 17–19 . The 

target and background pressures were ,10 22 and ,10 28 mbar, 

respectively. 

Shake-up and tunnelling 

To probe shake-up and light-induced electron tunneling, we ionized 

neon atoms with our subfemtosecond XUV pulse. Figure 3 shows the 

level structure and transitions relevant to our experiments. The core 

shell was not accessed by our XUV photons, and hence Auger decay is 

absent 20 . The threshold energies for single and double ionization 

from the outer shell of Ne are 21.56 and 62.53 eV, respectively 21 . 

The XUV photons produced Ne 11 and Ne 21 ions with a ratio of 

a 1,000 

Ion counts 

b 

900 

800 

700 

Ionization yield (arbitrary units) 


Ionization yield (arbitrary units) 


–1.0 

2p –2 4p 

380 as 

–0.5 0.0 0.5 1.0 

Delay time, ∆t (fs) 

2p –2 3d 

T L /2 

2p –2 3p 

2p –2 3s 

|E L | 

–14 –12 –10 –8 –6 –4 –2 0 2 4 


Figure 4 | Ne 21 ion yield versus delay between the subfemtoscond XUV 

pump and the few-cycle NIR probe: experiment and modelling. The peak 

intensity of the NIR probe was (7 6 1) 3 10 13 W cm 22 . a, The experimental 

data were acquired from six delay scans repeated under the same 

experimental conditions. The signal was accumulated for 3 s at each delay 

setting. The squares and the error bars show the average and the standard 

error (s.e.m.) of the results of the six measurements. The thick red line shows 

the average of five adjacent data points; the thin grey line shows the same as 

the thick line but recorded with NIR probe pulses of random carrierenvelope 

phase. In the inset, squares, triangles, and circles depict an 

ionization step extracted from three different measurements normalized 

to give the same change in the ionization yield. The solid line shows an 

error-function fit to the data yielding a rise time of 380 as (full-width at 

half-maximum, FWHM, of the gaussian function derived from the error 

function). The zero of delay is set arbitrarily to the centre of the tunnelling 

process. b, Simulation of the pump–probe experiment based on the 

nonadiabatic theory of tunnel ionization 6 . The thin coloured lines show the 

calculated fractional ionization yields contributed by electrons liberated 

from different shake-up states. The thick blue line depicts the overall 

ionization rate obtained by totalling the fractional rates (and by adding the 

result to the background measured at large negative delays). The simulations 

were carried out for a gaussian 250-as XUV pulse and a gaussian 5.5-fs laser 

pulse with a peak intensity of 7 3 10 13 W cm 22 . The black solid curve 

represents the absolute value of the laser field. For discussion of the results, 

see text. 

629



(19.7 6 0.5):1 (with ,2,500 Ne 11 ions created per second), in good 

agreement with the results of synchrotron measurements 22 . A few per 

cent of the Ne 11 ions were promoted into 2p 22 nl (principal quantum 

number n: 3 or 4; quantum orbit l: s, p or d) configurations 23 (Fig. 3). 

These satellite states can only decay radiatively on a picosecond timescale. 

Double ionization by the NIR field was not observed at the intensity 

level chosen. The XUV-generated Ne 21 yield was therefore not 

affected by the NIR probe for Dt = 2t L but was significantly 

enhanced by the laser field for Dt approaching zero and becoming 

positive. The NIR-induced Ne 21 yield enhancement amounted to 

(40 6 4)% of the XUV-produced Ne 21 yield at a NIR peak intensity 

of (7 6 1) 3 10 13 W cm 22 . The absence of this enhancement for 

Dt = 2tL clearly indicates that the laser sets electrons free from states 

excited by the XUV pulse as sketched in Fig. 2d. The laser-induced 

change in the Ne 21 yield amounts to some 2% of the XUV-produced 

Ne 11 ions. This implies that a substantial fraction of the population 

of the 2p 22 nl shake-up satellites must have been depleted by field 

ionization. 

Figure 4a shows the number of Ne 21 ions detected as a function of 

delay Dt between the XUV pump and NIR probe. Figure 4b compares 

the prediction of the Yudin–Ivanov theory 6 (lines) with the experimental 

data (squares). In our modelling the shake-up states were 

populated instantly during XUV photoionization (for more details, 

see Supplementary Information). The calculations are in reasonable 

agreement with our measurements and reveal how the different 

shake-up states are depleted sequentially by laser-field ionization. 

The signal starts increasing at large negative delays owing to depletion 

of the 2p 22 4p state (relative population ,12%) and the 2p 22 3d state 

(,10%) at NIR intensity levels reached some 10 and 6 fs after the 

peak of the NIR probe (Dt , 210 and 26 fs), respectively. A more 

dramatic increase in the Ne 21 yield is observed as the delay 

approaches zero, as a consequence of the depletion of the most highly 

populated 2p 22 3p (,50%) and 2p 22 3s (,25%) states. In spite of 

their relatively high binding energy (,10 and 13 eV, respectively), 

these states are also depleted before zero delay, that is, by the field 

oscillation cycles comprised in the trailing edge of the pulse, leaving 

no room for increasing the Ne 21 yield with increasing Dt beyond 0. 

This main contribution to the Ne 21 yield emerges within approximately 

one and a half wave cycles of the NIR field, ,(3/2)TL 5 3p/vL, 

in several sharp steps that are spaced by ,TL/2; this clearly shows that 

field-induced tunnelling is the main cause of the observed increase in 

the Ne 21 yield. This conclusion is also supported by the disappearance 

of the steps in a pump–probe scan performed with a randomly 

varying carrier-envelope phase of the NIR probe pulses (grey line in 

Fig. 4a). 

The main shake-up population (residing in the 2p 22 3p and 2p 22 3s 

states) is depleted at NIR intensities corresponding to a Keldysh 

parameter c of the order of three. Hence, our experiment verifies 

not only the existence of light-field-induced tunnelling, as predicted 

by Keldysh some four decades ago 1 , but also confirms the dominant 

role of this ionization mechanism up to c values substantially exceeding 

1, as predicted recently by Yudin and Ivanov 6 . This conclusion 

is also backed by numerical solutions of the time-dependent 

Schrödinger equation (see Supplementary Information). 

The steepness of the ionization steps and the dips preceding them 

in the measured data are not well reproduced by our model, which 

neglects the influence of electron–electron interactions and that of 

the strong NIR field on the XUV-induced transitions populating the 

shake-up states. Recent work 24 and our TDSE simulations (see Supplementary 

Information) indicate that the influence of the strong 

laser field on the XUV excitation process may (at least partially) be 

responsible for this discrepancy. 

We feel that these experiments afford profound insight into fundamental 

electronic processes such as tunnelling and shake-up by 

contrasting theoretical models with time-domain data. To exploit 

this potential both (1) accurate models of shake-up in the presence 

630 

of a strong laser field need to be developed and (2) the temporal 

resolution needs to be improved by using shorter XUV pulses 25 

and improving the signal-to-noise ratio as well as the accuracy of 

determining the zero of delay. These advances will allow determination 

of the attosecond temporal evolution of the light-field-induced 

tunnelling current and they will provide deep insight into the nature 

of the electron–electron interactions responsible for shake-up. 

Once models for shake-up and tunnelling have been tested and 

verified with attosecond precision, the technique of attosecond tunnelling 

will provide direct time-domain insight into a wide range of 

multi-electron dynamics inside atoms and molecules by probing the 

transient population of excited valence states while these dynamics 

are unfolding. With improved signal-to-noise and XUV pulse duration, 

the temporal resolution may potentially approach the atomic 

unit of time (,24 as). At present, the observed rise time of the Ne 21 

yield of less than 400 as (which sets a corresponding upper limit on 

the time it takes the excited electronic states to become populated 

during XUV ionization and on tunneling; see inset in Fig. 4a) dictates 

the temporal resolution of our pump–probe approach. In the next 

section we demonstrate its applicability to probing intra-atomic 

multi-electron dynamics in real time. 

Photon 

energy 

5s –1 

3.3% 

5p –1 

9.7% 

Xe Xe + Xe 2+ Xe 3+ Xe 4+ 


Energy (eV) 

120 

100 

80 

60 

40 

20 

} 

XUV -I 


0 

8.9% 

Shake-up 

satellites 

A1 

4d –15p –1 5p 

nl 

–4 nl n′l′ 

4d –1 

78.0% 

A1 

A1 

A1 

} 

β α 

5p –2 

γ 

A2 

A2 

NIR-I 

Energy (eV) 

NIR-DI 

XUV mirror reflectivity 

100 

90 

80 

5p –3 

5p –4 

Shake-up 

satellites 

8s,8p 

4d –15p –1 7s,7p 

5d,6p 

6p 

5d 

6s,5d 

nl 

Figure 5 | Energy levels and transitions in xenon ions relevant to the 

current study. The relative population of states in Xe 11 is given for an XUV 

excitation energy of 90 eV from ref. 30. The XUV light preferably ionizes 

from the 4d 21 shell, creating an inner-shell vacancy. A subsequent Auger 

process (green arrows labelled A1) will follow with 99% probability 26 and is 

predominantly decaying to Xe 21 in configuration 5p 22 . Some of the 

populated states, a and b (presumably 5s 21 5p 22 7p states 28,31 ) can further 

decay to Xe 31 via a second Auger process (green arrow labelled A2), leading 

to triply charged ions. The A1 process also populates states below the 

threshold for Xe 31 . These are denoted by c and represent 5s 21 5p 22 6p as well 

as 5p 23 nl configurations. The laser pulse can ionize these states (red arrow 

labelled NIR-I). Furthermore, a series of 4d 21 shake-up satellites in Xe 11 is 

also populated—with a small probability—by the XUV pulse. The inset 

shows the possible configurations 29 together with the XUV mirror 

reflectivity determining the XUV excitation spectrum. The satellites mainly 

decay to the a, b, c states, with a small fraction of the Xe 21 population 

ending up in 5p 24 nln9l9 configurations. These states are short-lived and 

decay via the A2 process. Before this occurs, the nln9l9 electrons can be 

liberated by the laser field to yield Xe 41 in the 5p 24 state (red arrow labelled 

NIR-DI).

Multi-electron relaxation 

As a first application of the technique of attosecond tunnelling, we 

captured Auger cascade xenon atoms following excitation by a subfemtosecond 

XUV pulse. Figure 5 sketches the relevant energy levels 

and transitions. Energy-resolved synchrotron measurements have 

revealed that (1) the 91-eV XUV pulse will preferably liberate electrons 

from the 4d orbital 22 , (2) the vacancy decays by subsequent 

single (A1 in Fig. 5) and cascaded (A1 and A2 in Fig. 5) Auger 

processes, leading to Xe 21 and Xe 31 , respectively 26 , and (3) the lifetimes 

of the 4d3/2 and 4d5/2 holes are 6.3 6 0.2 and 5.9 6 0.2 fs, 

respectively 27 . These time-integral measurements have hitherto been 

able to set only a lower limit of 23 fs for the time constant of A2 (ref. 

27). 

To trace this dynamics in real time, we simultaneously recorded 

the number of Xe ions emerging in different charged states as a 

function of Dt. At the laser intensity of (7 6 1) 3 10 13 W cm 22 used 

in this experiment, the XUV-induced Xe 11 yield is buried in lasergenerated 

background, preventing delay-dependent effects from 

coming to light in the Xe 11 signal. This is not the case for higher 

charged states. With increasing delay, rapid exponential increase was 

observed in the Xe 31 signal near Dt 5 0 (Fig. 6b), concurrent with a 

significant decrease of the Xe 21 yield. The background in the Xe 31 

signal arises from the A1–A2 Auger cascade discussed above. From 


NATURE | Vol 446 |5 April 2007 ARTICLES 

a 

Ion counts 

b 

Ion counts (10 3 ) 

40 

30 

20 

10 

0 

12 

10 

8 

τ A1 = 6.0 ± 0.7 fs 

τ A1 

τ A2 = 30.8 ± 1.4 fs 

0 50 100 


150 200 

Xe 4+ 

Xe 3+ 

Figure 6 | Xe 41 and Xe 31 ion yields versus delay between the 

subfemtosecond XUV pump and the few-cycle NIR probe. The data have 

been compiled from the results of five delay scans repeated under the same 

experimental conditions. The signals were accumulated for 20 s at each delay 

setting. The squares and the error bars show the average and the s.e.m. of the 

results of these measurements. a, Assuming a sampling function identical to 

the ionization profile measured in the neon experiment (see solid line in 

Fig. 4a), a double exponential fit (see Supplementary Information) to the 

measured Xe 41 yield versus Dt (solid line) yields the Auger decay times 

tA1 5 6.0 6 0.7 fs and tA2 5 30.8 6 1.4 fs. b, With the temporal evolution of 

the A1 process acquired from the Xe 41 data and using the fact that the rise of 

the Xe 31 yield with t comprises both A1 and the laser-induced ionization 

process, we can obtain the temporal evolution of the laser-induced transition 

from that of the Xe 31 yield. The exponential rise from about 7,000 to 11,500 

counts is consistent with a laser-induced ionization time of 5.8 6 2.5 fs 

(FWHM). This is comparable to tL, indicating that low-order, one-photon 

and/or two-photon transitions may promote electrons from the c states of 

Xe 21 to the 5p 23 states of Xe 31 (see Fig. 5). Note that the fluctuations in the 

Xe 31 signal can be largely accounted for by XUV intensity variations, which 

can be efficiently eliminated by normalization to, for example, the Xe 21 ion 

yield (see Supplementary Information). 


Fig. 5 we infer that the laser-induced increase in the Xe 31 yield is due 

to ionization from the c-states in Xe 21 , which cannot spontaneously 

decay into Xe 31 . This NIR-probe-induced transition (denoted by 

NIR-I in Fig. 5) yields Xe 31 in configuration 5p 23 . Because these 

are populated by the first Auger process, A1, the exponential increase 

in the Xe 31 signal is the convolution of the A1 decay and the NIRinduced 

ionization process. No decrease of the enhanced Xe 31 yield 

was observed up to our maximum delay of 300 fs, indicating that the 

lifetime of the c states was longer than 1 ps. 

Charge-states higher than Xe 31 cannot be created with the XUV 

pulse alone, because the XUV photon energy is below the threshold 

for Xe 41 production (,105 eV). However, with the probe switched 

on, we did observe Xe 41 ions for Dt . 2tL (Fig. 6a). The Xe 41 signal 

first grows within a few femtoseconds, followed by a longer decay. 

The Xe 41 ions are created by NIR-induced double ionization 

(denoted by NIR-DI in Fig. 5) from the intermediate doubly excited 

5p 24 nln9l9 (Fig. 5) and/or singly excited 5s 22 5p 21 nl (not shown in 

Fig. 5) states of Xe 21 . These states are populated by A1 from the 

satellites 4d 21 5p 21 nl of the 4d 21 state upon emission of low-energy 

electrons 28,29 and emptied by A2 to states of larger binding energy in 

Xe 31 , which cannot be reached by the NIR probe. 

From these results we conclude that the exponential rise and decay 

of the Xe 41 signal in Fig. 6a reveal the evolution of the A1 and A2 

Auger decays, respectively. The laser-induced double ionization may 

be either sequential or non-sequential (with the second step induced 

by recollision of the first electron). In either case, in the first step a 

binding energy similar to that of the most highly populated shake-up 

states in the neon experiment must be overcome by tunnel ionization. 

Hence, the probing (ionization) process may be assumed to be 

the same as measured in the neon experiment, see Fig. 4a. With this 

sampling function, fitting the result of a simple rate-equation analysis 

including the transitions A1 and A2 to the Xe 41 data shown in Fig. 6 

yields decay times of t A1 5 6.0 6 0.7 fs and t A2 5 30.8 6 1.4 fs for the 

A1 and A2 Auger processes, respectively. Both time constants are in 

accordance with the results of energy-resolved measurements 27 . 

Conclusions and outlook 

We have reported the observation of light-induced electron tunnelling 

from atoms in real time. Electrons are found to escape from their 

atomic binding potential within several subfemtosecond time intervals 

near the oscillation peaks of the ionizing few-cycle near-infrared 

laser field. Our results are in good agreement with the predictions of 

the theory Keldysh put forward four decades ago. The observed subfemtosecond 

ionization steps provide a powerful means of probing 

the transient population of short-lived valence electronic states in 

excited atoms or molecules, offering direct, time-domain access to a 

wide range of multi-electron dynamics unfolding on an attosecond 

to femtosecond timescale. Proof-of-principle attosecond tunnelling 

experiments in neon and xenon demonstrate this potential. 

Simultaneous implementation of attosecond tunnelling and attosecond 

streaking spectroscopy along with scaling of the techniques 

to higher photon energies and shorter X-ray pulse durations will 

provide unprecedented insight into the transient electronic states 

of matter. 

Received 2 November 2006; accepted 26 January 2007. 

1. Keldysh, L. V. Ionization in the field of a strong electromagnetic wave. Sov. Phys. 

JETP 20, 1307–1314 (1965). 

2. Faisal, F. H. M. Multiple absorption of laser photons by atoms. J. Phys. B 6, 

L89–L92 (1973). 

3. Reiss, H. R. Effect of an intense electromagnetic field on a weakly bound system. 

Phys. Rev. A 22, 1786–1813 (1980). 

4. Brabec, T. & Krausz, F. Intense few-cycle laser fields: frontiers of nonlinear optics. 

Rev. Mod. Phys. 72, 545–591 (2000). 

5. Scrinzi, A., Geissler, M. & Brabec, T. Ionization above the coulomb barrier. Phys. 

Rev. Lett. 83, 706–709 (1999). 

6. Yudin, G. L. & Ivanov, M. Yu. Nonadiabatic tunnel ionization: looking inside a laser 

cycle. Phys. Rev. A 64, 013409 (2001). 


631



7. Baltuska, A. et al. Attosecond control of electronic processes by intense light 

fields. Nature 421, 611–615 (2003). 

8. Itatani, J. et al. Attosecond streak camera. Phys. Rev. Lett. 88, 173903 (2002). 

9. Kitzler, M., Milosevic, N., Scrinzi, A., Krausz, F. & Brabec, T. Quantum theory of 

attosecond XUV pulse measurement by laser-dressed photoionization. Phys. Rev. 

Lett. 88, 173904 (2002). 

10. Hentschel, M. et al. Attosecond metrology. Nature 414, 509–513 (2001). 

11. Kienberger, R. et al. Atomic transient recorder. Nature 427, 817–821 (2004). 

12. Goulielmakis, E. et al. Direct measurement of light waves. Science 305, 1267–1269 

(2004). 

13. Drescher, M. et al. Time-resolved atomic inner-shell spectroscopy. Nature 419, 

783–787 (2002). 

14. Svensson, S., Eriksson, B., Martensson, N., Wendin, G. & Gelius, U. Electron shakeup 

and correlation satellites and continuum shake-off distributions in x-ray 

photoelectron spectra of the rare gas atoms. J. Electron Spectrosc. Related 

Phenomena 47, 327–384 (1988). 

15. Aksela, H., Aksela, S. & Kabachnik, N. Resonant and nonresonant Auger 

recombination. In VUV and Soft X-Ray Photoionization (eds Becker, U. & Shirley, D. 

A.) 401–440 (Plenum, New York, 1996). 

16. Istomin, A. Y., Manakov, N. L. & Starace, A. F. Perturbative analysis of the triply 

differential cross section and circular dichroism in photo-double-ionization of He. 

Phys. Rev. A 69, 032713 (2004). 

17. Schröder, H., Wagner, M., Kaesdorf, S. & Kompa, K. L. Surface-analysis by laser 

ionization. Ber. Bunsenges. Phys. Chem. 97, 1688–1692 (1993). 

18. Wagner, M. & Schröder, H. A novel 4 grid ion reflector for saturation of laser 

multiphoton ionization yields in a time-of-flight mass-spectrometer. Int. J. Mass 

Spectrom. 128, 31–45 (1993). 

19. Witzel, B., Schröder, H., Kaesdorf, S. & Kompa, K. L. Exact determination of 

spatially resolved ion concentrations in focused laser beams. Int. J. Mass Spectrom. 

172, 229–238 (1998). 

20. Larkins, F. P. Charge state dependence of x-ray and Auger electron emission 

spectra for rare-gas atoms—II. The neon atom. J. Phys. B 4, 14–19 (1971). 

21. National Institute of Standards and Technology Physical Reference Data Æhttp:// 

physics.nist.gov/PhysRefData/æ (1994). 

22. Holland, D. M. P., Codling, K., West, J. B. & Marr, G. V. Multiple photoionization in 

the rare gases from threshold to 280 eV. J. Phys. B 12, 2465–2484 (1979). 

23. Becker, U. & Shirley, D. A. Partial Cross Sections and Angular Distributions. In 

VUV and Soft X-Ray Photoionization (eds Becker, U. & Shirley, D. A.) 135–173 

(Plenum, New York, 1996). 

632 


24. Smirnova, O., Spanner, M. & Ivanov, M. Y. Coulomb and polarization effects in 

laser-assisted XUV ionization. J. Phys. B 39, 323–339 (2006). 

25. Sansone, G. et al. Isolated single-cycle attosecond pulses. Science 314, 443–446 

(2006). 

26. Kämmerling, B., Krässig, B. & Schmidt, V. Direct measurement for the decay 

probabilities of 4dj hole states in xenon by means of photoelectron-ion 

coincidences. J. Phys. B 25, 3621–3629 (1992). 

27. Penent, F., Palaudoux, J., Lablanquie, P. & Andric, L. Multielectron 

spectroscopy: the xenon 4d hole double Auger decay. Phys. Rev. Lett. 95, 083002 

(2005). 

28. Lablanquie, P. et al. Photoemission of threshold electrons in the vicinity of the 

xenon 4d hole: dynamics of Auger decay. J. Phys. B 35, 3265–3295 (2002). 

29. Hayaishi, T. et al. Manifestation of Kr 3d and Xe 4d conjugate 

shake-up satellites in threshold-electron spectra. Phys. Rev. A. 44, R2771–R2774 

(1991). 

30. Becker, U. et al. Subshell photoionization of Xe between 40 and 1000 eV. Phys. 

Rev. A 39, 3902–3911 (1989). 

31. Viefhaus, J. et al. Auger cascades versus direct double Auger: relaxation 

processes following photoionization of the Kr 3d and Xe 4d, 3d inner shells. J. 

Phys. B 38, 3885–3903 (2005). 

Supplementary Information is linked to the online version of the paper at 

www.nature.com/nature. 

Acknowledgements We thank A. F. Starace for discussions. We are grateful for 

financial support from the Volkswagenstiftung (Germany), the Marie Curie 

Research Training Network XTRA, Laserlab Europe, and a Marie Curie 

Intra-European Fellowship. F.K. acknowledges support from the FWF (Austria). 

The research of M.F.K. and M.J.J.V. is part of the research programme of the 

Stichting voor Fundamenteel Onderzoek der Materie, which is financially 

supported by the Nederlandse Organisatie voor Wetenschappelijk Onderzoek. 

This research was supported by the cluster of excellence Munich Centre for 

Advanced Photonics (www.munich-photonics.de). 

Author Contributions M.U., Th.U., M.S. and A.J.V. contributed equally to this work. 

Author Information Reprints and permissions information is available at 

www.nature.com/reprints. The authors declare no competing financial interests. 

Correspondence and requests for materials should be addressed to M.U. 

(matthias.uiberacker@mpq.mpg.de) or F.K. (ferenc.krausz@mpq.mpg.de). 



Dispersion control over the 

ultraviolet–visible–near-infrared spectral range 

with HfO 2/SiO 2-chirped dielectric multilayers 

V. Pervak 

Max-Planck-Institute of Quantum Optics, Hans-Kopfermann-Strasse 1, D-85748 Garching, Germany 

F. Krausz 

Max-Planck-Institute of Quantum Optics, Hans-Kopfermann-Strasse 1, D-85748 Garching, Germany 

and Ludwig-Maximilians-Universitaet Muenchen, Am Coulombwall 1, D-85748 Garching, Germany 

A. Apolonski 

Ludwig-Maximilians-Universitaet Muenchen, Am Coulombwall 1, D-85748 Garching, Germany 

and Institute of Automation and Electrometry, Russian Academy of Sciences, 630090 Novosibirsk, Russia 

Received December 21, 2006; accepted February 1, 2007; 

posted February 13, 2007 (Doc. ID 78368); published April 3, 2007 

We report the first realization, to the best of our knowledge, of a chirped multilayer dielectric mirror providing 

dispersion control over the spectral range of 300–900 nm and the first use of hafnium oxide in a 

chirped mirror. The technology opens the door to the reliable and reproducible generation of monocycle laser 

pulses in the blue–violet spectral range, will benefit the development of optical waveform and frequencycomb 

synthesizers over the ultraviolet–visible–near-infrared spectral range, and permits the development of 

ultrabroadband-chirped multilayers for high-power applications. © 2007 Optical Society of America 

OCIS codes: 320.5520, 320.7160, 310.1620. 

Advancing ultrashort laser pulse generation to the 

limit set by the oscillation cycle of light has been pursued 

ever since the discovery of lasers. Pulses comprising 

an ever-decreasing number of wave cycles allow 

more efficient exploitation of nonlinear optical 

effects 1 with implications as striking as the generation 

of single subfemtosecond light pulses. 2 Moreover, 

the controlled superposition of light frequencies extending 

over more than one octave opens, along with 

carrier-envelope phase control, 3–5 the way to shaping 

the subcycle (i.e., subfemtosecond) evolution of light 

fields in laser pulses and thereby to a new way of 

quantum control based on the light-force-directed 

charge in atoms, molecules, or solids. 6 In this Letter, 

we present a chirped multilayer mirror offering high 

reflectivity and controlled group-delay dispersion 

(GDD) over some 1.5 octaves spanning from ultraviolet 

(UV) to near-infrared (NIR) frequencies. This 

technology may become instrumental for the development 

of future ultrawideband optical waveform 

synthesizers. 

There have been several approaches to generating 

monocycle optical pulses: (i) phase-coherent superposition 

of pulses from different laser sources 7,8 ; (ii) 

phase-coherent synthesis of Raman sidebands by exploiting 

vibrational or rotational transitions in 

molecules 9,10 ; (iii) multicolor optical parametric 

generation 11 ; and last but not least, (iv) by means of 

supercontinuum generation based on self- or crossphase 

modulation. 12–14 What all these techniques 

have in common is that they rely on some optical device 

that can be used to adjust the phase (and amplitude) 

of individual groups of frequencies independently. 

This has, so far, been implemented by 

May 1, 2007 / Vol. 32, No. 9 / OPTICS LETTERS 1183 

separating the spectral components of a broadband 

signal in space 15 and addressing the spectral channels 

in the Fourier plane by a spatial light modulator 

(SLM) based on liquid crystals 15 or acoustic waves. 16 

The concept can, in principle, be extended to bandwidths 

exceeding one octave, 17 however, the UV–IR 

absorption and the low damage threshold of SLMs restricts 

the technology from being scaled to bandwidths 

of several octaves and ever higher pulse energies. 

In this Letter, we demonstrate that chirped 

multilayer mirrors may provide a promising alternative 

for specific applications, such as monocycle pulse 

generation or wideband frequency-comb generation. 

Chirped mirrors (CM) have been continuously improved 

ever since their invention in 1994. 18 Progress 

has been made in terms of bandwidth, losses, GDD, 

and the ability to compensate higher-order spectral 

phase errors introduced by optical components. 19–27 

As a result of the efforts of several groups, by the 

turn of the millennium, CM-based optical systems 

have been capable of controlling broadband radiation 

over spectral ranges approaching an octave in the 

visible–NIR domain. 12,13,28–30 

Recently, we demonstrated a chirped multilayer 

mirror supporting sub-3 fs pulses carried at a wavelength 

of 600 nm. 31 This technological progress paves 

the way toward the generation of monocycle light 

pulses, wideband optical waveform synthesis, and 

frequency-comb generation from compact, userfriendly 

laser systems. Many applications motivate 

the extension of these capabilities into the UV spectral 

range. The dramatically increasing dispersion of 

optical materials toward the UV absorption bands 

prevented chirped dielectric multilayer technology 

0146-9592/07/091183-3/$15.00 © 2007 Optical Society of America 



Dispersion management for a sub-10-fs, 10 TW 

optical parametric chirped-pulse amplifier 

Franz Tavella, 1,4 Yutaka Nomura, 1 Laszlo Veisz, 1 Vladimir Pervak, 2 Andrius Marcinkevi~ius, 1,3,5 and 

Ferenc Krausz 1,2 

1 Max-Planck-Institut für Quantenoptik, Hans-Kopfermann-Strasse 1, 85748 Garching, Germany 

2 Department für Physik, Ludwig-Maximilians-Universität München, am Coulombwall 1, 85748 Garching, Germany 

3 Present affiliation, IMRA America Inc., 1044 Woodridge Avenue, Ann Arbor, Michigan 48105, USA 

4 franz.tavella@mpq.mpg.de 

5 amarcink@imra.com 

Received March 19, 2007; revised June 6, 2007; accepted June 11, 2007; 

posted June 11, 2007 (Doc. ID 81025); published July 24, 2007 

We report the amplification of three-cycle, 8.5 fs optical pulses in a near-infrared noncollinear optical parametric 

chirped-pulse amplifier (OPCPA) up to energies of 80 mJ. Improved dispersion management in the 

amplifier by means of a combination of reflection grisms and a chirped-mirror stretcher allowed us to recompress 

the amplified pulses to within 6% of their Fourier limit. The novel ultrabroad, ultraprecise dispersion 

control technology presented in this work opens the way to scaling multiterawatt technology to even 

shorter pulses by optimizing the OPCPA bandwidth. © 2007 Optical Society of America 

OCIS codes: 190.4970, 190.4410, 320.5520. 

High-peak-power few-cycle sources are of interest for 

a number of applications in nonlinear optics, highfield, 

and ultrafast science [1]. Few-cycle pulses not 

only offer high peak powers from compact systems 

(relative value) but also enable the emergence of entirely 

new technologies such as the generation and 

measurement and spectroscopic applications of isolated 

attosecond pulses and steering the atomic-scale 

motion of electrons with controlled light fields [2–7] 

(absolute value). Noncollinear optical parametric amplification 

offers amplification over spectral ranges 

sufficiently broad for few-cycle pulse synthesis, but 

dispersion control during stretching and recompression 

has remained a major challenge because of the 

large bandwidth over which the dispersion needs to 

be compensated with high accuracy. This is one of the 

reasons why many terawatt-scale optical chirpedpulse 

amplifiers [8,9] are designed for higher energy 

amplification and for pulse durations longer than 

those potentially allowed by the amplification bandwidth. 

Only recently amplification and adaptive 

pulse compression of more than 100-THz bandwidths 

to the few-cycle regime were demonstrated reaching 

the terawatt regime [10–12]. 

In this Letter we report the implementation of an 

ultrabroadband grism-pair stretcher capable of controlling 

group delay over a dynamic range of tens of 

picoseconds and a bandwidth exceeding 100 THz. 

This improvement led to what we believe to be the 

first sub-10-fs, 10 TW light source ever reported and 

holds promise for further shortening of multiterawatt 

light pulses by means of optical parametric amplifier 

(OPA) bandwidth engineering. 

The schematic layout of the our novel stretchercompressor 

system is depicted in Fig. 1. The OPA 

chain is described in previous work [12]. Stretching is 

implemented by the negative dispersion of a pair of 

grisms [13]. These hybrid elements combining the 

dispersive effects of diffraction gratings and prisms 

were recently demonstrated to be capable of introduc- 

August 1, 2007 / Vol. 32, No. 15 / OPTICS LETTERS 2227 

ing near-linear dispersion with high throughput over 

a bandwidth of 60 nm in the near infrared [13–15]. 

Here we demonstrate that they can also control optical 

delay efficiently over a bandwidth of 300 nm in 

the same spectral range. Our positive-dispersion 

compressor is composed of glass blocks and chirped 

mirrors (CM2). The latter are used for final compression 

to prevent excessive nonlinear effects in the 

glass compressors. Residual higher-order dispersion 

of the system is compensated by another pair of 

chirped mirrors (CM1) and a programmable acoustooptic 

dispersive filter (AODF), dubbed Dazzler [16]. 

The grism stretcher is designed to compensate for 

the dispersion of (1) the glass compressor (160 mm of 

SF57 and 100 mm of fused silica), (2) the BBO crystals 

in the optical parametric chirped-pulse amplifier 

(OPCPA) chain (a total of 15-mm path length), and 

(3) the acousto-optical filter (43 mm of TeO 2) at the 

central wavelengths of 850 nm of our amplifier chain. 

Fig. 1. Optical layout of the OPCPA experimental setup. 

0146-9592/07/152227-3/$15.00 © 2007 Optical Society of America 


that is responsible for such quantum-mechanical 

phenomena as electron diffraction through crystals. 

Likewise, the bound electron state possesses a 

quantum phase, which can change sign from 

place to place in the molecule. The quantum 

phases of the returning electron wave interfere 

with the different parts of the molecular bound 

state wave function, creating attosecond modulation 

of the electronic structure that is recorded 

in the shape of the HHG spectrum (Fig. 2). 

The physics of HHG is still under active 

discussion, and this coherent scattering picture 

has many subtleties that are still unresolved. 

Nonetheless, the effect of molecular shape on 

the HHG spectrum is genuine and dramatic. For 

example, two simple molecules O2 and N2, are 

similar enough in their gross structure, each 

consisting of two atoms separated by about an 

angstrom, and a difference of only two electrons 

out of more than a dozen; but their highest 

occupied orbitals, the ones that will field-ionize 

in an intense laser field, have very different symmetries. 

HHG spectra reveal this difference, and 

recent experiments have used the spectra to perform 

tomographic reconstructions of the wave 

function responsible for ionization. 

Conclusion 

Only a few years ago, the direct measurement of 

transient phenomena lasting less than an optical 

cycle seemed a supreme challenge. Now scien- 

REVIEW 


tists have brought together techniques from atomic 

and laser physics to make these measurements 

possible, if not yet routine. Attoscience is a new 

field, with the promise to reveal some of the 

fastest processes of chemistry and atomic physics, 

or to freeze motion that will allow us to view 

the structure of matter under extreme conditions. 

The subject is still dominated by attempts to 

understand and improve the sources, and to interpret 

the data, and there are still frontiers in 

source development: For example, attosecond 

pulses at subangstrom wavelengths have not been 

demonstrated. There are even more challenges 

in the theory of attosecond molecular dynamics: 

One of the most difficult issues yet to be resolved 

in attoscience is how we should describe 

the structure and motion of electrons and nuclei 

in physical chemistry and molecular physics to 

accommodate processes that are so fast that cherished 

concepts such as potential energy surfaces 

or the Born Oppenheimer approximation are no 

longer valid. Despite these difficulties, the rapid 

progress in attoscience over the past few years is 

not abating, and we may anticipate new physical 

insights in the coming decade. 

References and Notes 

1. W.-M. Yao et al., J. Phys. G: Nucl. Part. Phys. 33, 

1 (2006); http://meetings.aps.org/link/BAPS.2007.APR. 

B2.1. 

2. D. E. Spence, P. N. Kean, W. Sibbett, Opt. Lett. 16, 42 

(1991). 

Attosecond Control and Measurement: 

Lightwave Electronics 

E. Goulielmakis, 1 V. S. Yakovlev, 2 A. L. Cavalieri, 1 M. Uiberacker, 2 V. Pervak, 2 A. Apolonski, 2 

R. Kienberger, 1 U. Kleineberg, 2 F. Krausz 1,2 * 

Electrons emit light, carry electric current, and bind atoms together to form molecules. Insight into 

and control of their atomic-scale motion are the key to understanding the functioning of biological 

systems, developing efficient sources of x-ray light, and speeding up electronics. Capturing and 

steering this electron motion require attosecond resolution and control, respectively (1 attosecond = 

10 −18 seconds). A recent revolution in technology has afforded these capabilities: Controlled light waves 

can steer electrons inside and around atoms, marking the birth of lightwave electronics. Isolated 

attosecond pulses, well reproduced and fully characterized, demonstrate the power of the new 

technology. Controlled few-cycle light waves and synchronized attosecond pulses constitute its key tools. 

We review the current state of lightwave electronics and highlight some future directions. 

Quantum mechanics predicts the characteristic 

time scale for the rapidity of 

microscopic dynamics as Dt ~ ħ/DW, 

where DW is the spacing between the 

relevant energy levels of the microscopic system 

and ħ is Planck’s constant. The milli–electron 

volt and multi–electron volt energy spacing of 

vibrational and electronic energy levels, respectively, 

imply that structural dynamics of mole- 

cules and solids as well as related chemical 

reactions and phase transitions evolve on a femtosecond 

time scale, whereas electronic motion on 

the atomic scale is to be clocked in attoseconds. 

Before the invention of the laser, the resolution 

of time-resolved spectroscopy was limited 

by the nanosecond duration of pulses of incoherent 

light. The laser and the successive technological 

developments for the generation and 

3. M. Nisoli, S. De Silvestri, O. Svelto, Appl. Phys. Lett. 68, 

2793 (1996). 

4. C. P. Hauri et al., Appl. Phys. B 79, 673 (2004). 

5. R. L. Fork, C. H. Brito Cruz, P. C. Becker, C. V. Shank, 

Opt. Lett. 12, 483 (1987). 

6. E. Matsubara et al., J. Opt. Soc. Am. B 24, 985 

(2007). 

7. M. Hentschel et al., Nature 414, 509 (2001). 

8. Ahmed Zewail received the 1999 Nobel Prize in 

Chemistry for his studies of the transition states of 

chemical reactions using femtosecond spectroscopy. 

9. S. Augst et al., Phys. Rev. Lett. 63, 2212 (1989). 

10. P. B. Corkum, Phys. Rev. Lett. 71, 1994 (1993). 

11 X. F. Li et al., Phys. Rev. A. 39, 5751 (1989). 

12. J. L. Krause et al., Phys. Rev. A. 45, 4998 (1992). 

13. K. J. Schafer, K. C. Kulander, Phys. Rev. Lett. 78, 638 

(1997). 

14. R. López-Martens et al., Phys. Rev. Lett. 94, 033001 

(2005). 

15. N. M. Naumova, J. A. Nees, G. A. Mourou, Phys. Plasmas 

12, 056707 (2005). 

16. A. A. Zholents, G. Penn, Phys. Rev. ST Accel. Beams 8, 

050704 (2005). 

17. M. Lein, N. Hay, R. Velotta, J. P. Marangos, P. L. Knight, 

Phys. Rev. A. 66, 023805 (2002). 

18. J. Itatani et al., Nature 432, 867 (2004). 

19. T. Kanai, S. Minemoto, H. Sakai, Nature 435, 470 

(2005). 

20. This paper was written at the Stanford PULSE Center, 

with support from the NSF and from the Stanford Linear 

Accelerator Center, a national laboratory operated by 

Stanford University on behalf of the U.S. Department of 

Energy, Office of Basic Energy Sciences. We acknowledge 

useful discussions with M. Guehr, who also provided Fig. 2 

for this article. 

10.1126/science.1142135 

SPECIALSECTION 

measurement of ultrashort laser pulses improved 

the resolving power of pump-probe 

spectroscopy from several nanoseconds to several 

femtoseconds (1). The birth of femtosecond 

technology permitted real-time observation of the 

breakage and formation of chemical bonds (2). 

We review the recent developments in the 

optical technology that have led to the breaking 

of the femtosecond barrier and provided real-time 

access to intra-atomic electron dynamics. Consequences 

include the observation of electronic 

motion deep inside (i.e., in inner shells of ) atoms 

(3) and its control in real time (4). We address the 

underlying physical concepts and highlight the 

current status as well as future prospects of attosecond 

technology (5). 

Femtosecond Technology: Control and 

Measurement with the Amplitude and 

Frequency of Light 

Control and measurement of dynamics are 

intertwined. Time-resolved measurement relies 

on a physical quantity varying in a controlled, 

reproducible fashion on the relevant time scale. 

1 Max-Planck-Institut für Quantenoptik (MPQ), Hans- 

Kopfermann-Straße 1, D-85748 Garching, Germany. 

2 Department für Physik, Ludwig-Maximilians-Universität, 

Am Coulombwall 1, D-85748 Garching, Germany. 

*To whom correspondence should be addressed. E-mail: 

krausz@lmu.de 

www.sciencemag.org SCIENCE VOL 317 10 AUGUST 2007 769 


on November 12, 2007 

www.sciencemag.org 

Downloaded from

770 

Attosecond Spectroscopy 

Femtosecond technology is the result of controlling 

the nonlinear polarization of matter with the 

amplitude of light. Controlling the absorption 

and/or refractive index in this way has yielded— 

along with group delay dispersion control— 

femtosecond pulses from laser oscillators by means 

of passive mode locking (1). Measurement of 

the pulses relies on the same physical effect: 

control of the nonlinear polarization of matter 

by a replica of the pulse in the presence of the 

pulse to be characterized (6, 7). 

The controlled, well-characterized evolution 

of the amplitude envelope and carrier-frequency 

sweep (chirp) of ultrashort laser pulses permits 

measurement (2) and control (8) of quantum 

transitions on a femtosecond time scale. Reliance 

on these cycle-averaged quantities implies that 

measurement resolution and control speed are 

ultimately limited by the carrier wave cycle. The 

carrier wave cycle period is about 3 fs in the near 

infrared, where low dispersion favors the generation 

of the shortest laser pulses. 

Toward Attosecond Technology: 

X-ray–Pump/X-ray–Probe Spectroscopy? 

Femtosecond measurement and control techniques 

utilizing nonlinear material response could, 

in principle, be extended into the attosecond time 

domain by using intense attosecond pulses of extreme 

ultraviolet (XUV) or x-ray light. Unfortunately, 

in these regions of the optical spectrum, the 

probability of two-photon absorption is prohibitively 

low. X-ray–pump/x-ray–probe spectroscopy 

and x-ray quantum control therefore rely on x-ray 

intensities that can be attained only with large-scale 

free-electron lasers (9). Even though these sources 

are expected to eventually deliver their radiation 

in subfemtosecond pulses (10) and XUV 

sources pumped by large-scale, high energy 

lasers have already pushed the frontiers of 

nonlinear optics to the range of several tens of 

electron volts (11, 12), proliferation of attosecond 

technology and its widespread applications 

call for another approach—one that relies on light 

sources suitable for small laboratories. 

An Alternative Route: Light-Induced 

Electronic Motion Within the Wave 

Cycle of Light 

The electric field of visible, near-infrared or 

infrared (henceforth, referred to collectively as 

NIR) laser light, E L(t), exerts a force on electrons 

that varies on a subfemtosecond scale. The use of 

this gradient for initiating and probing the 

subsequent dynamics with attosecond timing 

precision and resolution has led to the emergence 

of an attosecond technology that does not rely on 

the existence of intense x-ray pulses. 

According to theory, strong-field–induced 

ionization of atoms is confined to subfemtosecond 

intervals near the peaks of the oscillating 

NIR field (Fig. 1A), setting a subfemtosecond 

electron wave packet yfree(t) free; this prediction 


was recently confirmed by real-time observation 

(13). The subfemtosecond ionization process, 

locked to the peak of NIR field oscillation with 

corresponding precision, may serve as a subfemtosecond 

“starter gun” for a wide range of 

dynamics (including electronic and subsequent 

nuclear motion and related dynamics of molecular 

structure). 

The energetic radiation concurrent with opticalfield 

ionization in a linearly polarized laser field 

(14) provides a means of triggering motion in a 

more gentle way, without applying an intense 

field. The broadband XUV light building up 

during the propagation of the driving pulse 

through an ensemble of atoms was predicted to 

emerge in a subfemtosecond burst within every 

half optical cycle that is sufficiently intense for 

ionization (15). In this Review, we focus on these 

photonic tools of attosecond technology, noting 

that the recolliding electron wave packet itself 

can also be used to explore dynamics (16–18). 

The electronic (and concurrent) phenomena 

following the attosecond trigger can be tracked 

with comparable resolution with the use of the 

generating NIR field, the oscillations of which are 

synchronized with the trigger event (tunnel 

ionization, recollision, or XUV burst from recollision). 

Let us consider an electron ejected during 

the unfolding evolution of the system under 

scrutiny. In the presence of a linearly polarized 

NIR field, outgoing electrons collected along 

the direction of the laser electric field EL(t) 

(Fig. 1B) suffer a momentum change DpðtÞ ¼ 

e∫ ∞ t ELðt′Þdt′ ¼ eALðtÞ. Here, e is the charge of 

the electron, t is the instant of its emission, and A L 

(t) is the vector potential of the laser field. This 

change varies monotonically within a half wave 

cycle of the laser field, mapping the temporal 

evolution of the emitted subfemtosecond electron 

wave packet to a corresponding final momentum 

and energy distribution of the emitted electrons. 

We recognize here a new embodiment of 

the basic concept of streak imaging, with the 

streaking controlled by the NIR field in this case. 

The electrons can gain or lose an energy of 

several electron volts depending on whether they 

have been emitted some hundred attoseconds 

sooner or later than the peak of the field cycle 

(Fig. 1B), endowing this light-field–driven streak 

camera with attosecond resolution (19, 20). 

The physical concepts and mechanisms outlined 

above have opened a realistic prospect of 

measuring and controlling electron dynamics on 

an attosecond time scale. However, full exploitation 

of the potential offered by this conceptual 

framework requires precise control over the applied 

laser field. 

Steering Electrons with the Electric 

Field of Controlled Light Waves 

In a laser pulse comprising many field cycles, the 

subfemtosecond electron and photon bursts born 

in the ionizing field emerge every half cycle, 

10 AUGUST 2007 VOL 317 SCIENCE www.sciencemag.org 

resulting in a train of subfemtosecond bursts (21). 

Its spectrum consists of equidistant lines that 

represent high-order odd harmonics of the 

driving field (22). The subfemtosecond pulse 

train and its multicycle driver constitute powerful 

tools for controlling and tracking electron 

dynamics that terminate within the NIR field 

oscillation cycle (23–25). The characteristics 

of the individual bursts in the train can be difficult 

to control and measure. Extension of attosecond 

control and metrology to a wide range of 

electronic phenomena (including all those extending 

in time over more than half an optical cycle) 

calls for the isolation of one to several attosecond 

pulses and precise control of their properties (26). 

Isolation was predicted to be achievable by 

manipulating the polarization state (polarization 

gating) of the driving field (27) or shortening its 

duration to nearly a single cycle of the carrier 

wave (28). In fact, intense few-cycle laser pulses 

lasting several femtoseconds (1) led to the first 

observation of light lasting for less than 1 fs (29). 

Full control of the number and properties of bursts 

isolated, however, requires precise control of the 

driving electric-field waveform E L(t). 

Intense few-cycle laser pulses with controlled 

waveform (4)—along with careful bandpass filtering 

of the highest-energy (cutoff) harmonics— 

have indeed permitted the reproducible generation 

of single and twin subfemtosecond pulses with 

cosine- and the sine-shaped electric field waveforms, 

respectively, as well as their reliable measurement 

by means of attosecond streaking (30, 31) 

(fig. S1). The series of streaked electron spectra 

shown in Fig. 1C yields complete information 

about the driving laser field and allows determination 

of the temporal shape, duration, and a 

possible chirp of the emitted subfemtosecond 

XUV pulse (Fig. 1D) as well as the degree of 

its synchronism with its driver (31–33). These 

experiments reveal that few-cycle light with 

controlled waveform permits reproducible generation 

and complete characterization of subfemtosecond 

light. Owing to the attosecond 

synchronism between them, these tools allow 

attosecond metrology without the need for highintensity 

x-rays. 

Lightwave Electronics: Versatile 

Technology for Control and Chronoscopy 

on the Electronic Time Scale 

Controlled light fields permit control of microscopic 

electric currents at the atomic scale just as 

synthesized microwave fields permit control of 

currents at the mesoscopic scale in semiconductor 

chips. By analogy to microwave electronics, we 

propose to name this new technology lightwave 

electronics. In marked contrast with previous 

quantum control through controlling transitions 

between quantum states with the amplitude and 

frequency of light (8), lightwave electronics gives 

way to controlling dynamics directly by the force 

that the electric field of intense light exerts on 




Downloaded from

electrons. This new approach is powerful in that 

(i) it provides a direct way of affecting the position 

and momentum of electrons and (ii) control is 

matched in speed to the electronic time scale. 

The first notable manifestations of the 

power of lightwave electronics include controlling 

subfemtosecond XUV emission (4, 25, 31–35); 

molecular dissociation (36); measuring subfemtosecond 

XUV and electron pulses (21, 31–33, 35); 

and imaging dynamic changes in molecular 

structure (16–18) by means of steering electron 

wave packets with light fields. Sub–wave cycle 

(i.e., subfemtosecond) control of electron wave 

packet motion can be accomplished by mixing 

multicycle fields (25, 35). However, full control 

over the entire system evolution from attosecond 

to femtosecond time scales—and, consequently, 

precision attosecond measurements—relies on 

Fig. 1. The birth and measurement of a subfemtosecond (XUV) light 

pulse. (A) The field of a femtosecond NIR laser pulse, EL(t), is able to 

suppress the Coulomb potential in atoms sufficiently to allow a valence 

electron to tunnel through the narrow barrier and release a subfemtosecond 

wave packet, y free(t), near the peaks of its most intense oscillations. 

The wave packet is subsequently removed from the vicinity of 

the atomic core and less than a period later pushed back by the reversed 

field (green trajectory). Upon recollision, it interferes with the boundstate 

portion of the electron wave function. This interference results in 

high-frequency oscillations, emitting broadband XUV light. The highestfrequency 

portion, with intensity IX(t), is temporally confined to a small 

fraction of the optical period. (B) Concept of optical-field–driven streak 

imaging of electron emission from atoms. Electrons released by an XUV 

pulse parallel to the direction of electric field (red line) suffer a change in 

their initial momenta that is proportional to the vector potential of the 

field (black line) at the instant of release, mapping the intensity profile of 


optical waveforms that can be accurately controlled 

in shape and polarization state (4, 37) 

and confined to several cycles. Exploitation of 

the full potential of lightwave electronics for 

attosecond control and metrology requires 

waveform-controlled broadband light. 

The attosecond streaking spectrogram shown 

in Fig. 1C yields the complete history of photoelectron 

emission induced by a subfemtosecond 

XUV pulse as well as of the streaking optical field 

(38). Similar spectrograms can also be recorded 

with secondary (Auger) electrons, which uncover 

the history of a quantum transition of an electron 

deep inside an atom in real time (3). Attosecond 

streaking is not the only way of observing the 

temporal evolution of atomic-scale electron motion. 

Energetic excitation of atoms, molecules, or 

solids is often accompanied by the promotion of 


valence electrons to excited states (referred to as a 

shake up). The transient population of these states 

provides information about the instantaneous 

(electronic) state of the system. This population 

can be probed by means of attosecond tunneling 

induced by the time-delayed waveform-controlled 

few-cycle laser field (13). 

In attosecond streaking and tunneling spectroscopy, 

the subfemtosecond XUV pulse serves 

as a pump and the controlled optical wave as the 

probe. Their roles can be interchanged: The optical 

field may—by means of tunnel ionization 

(13)—initiate an electronic process and control 

the unfolding dynamics (36), and the subfemtosecond 

XUV pulse may probe this process by, for 

example, photoelectron spectroscopy (39). 

Attosecond technology based on a controlled 

optical wave and a synchronized subfemtosecond 

the emitted electron and hence of the ionizing subfemtosecond XUV 

pulse into a corresponding final momentum and energy distribution of 

electrons. (C) Streaked spectra of photoelectrons released from neon 

atoms by a single subfemtosecond XUV pulse (ħw X ≈ 95 eV) recorded for 

a series of delays between the XUV pulse and NIR field (streaking 

spectrogram). The laser field causes only a moderate broadening 

(streaking) of the electron spectra; its main effect is the shift of the 

center of mass of the electron spectrum. In the limit of large initial 

kinetic energy of the electrons, this shift is linearly proportional to the 

vector potential of the field at the instant of impact of the XUV pulse (see 

white line). (D) Electric field of the NIR wave and intensity of the XUV 

pulse as retrieved from the streaking spectrogram shown in (C). The 

measurement confirms that the few-cycle waveform must be near cosineshaped 

to warrant the production of a single subfemtosecond pulse and 

reveals a near-Fourier–limited XUV pulse of a duration of 250 attoseconds. 

arb. u., arbitrary units. 





Downloaded from

772 


pulse is both versatile and sufficiently compact 

and affordable to allow proliferation in small 

laboratories. The core infrastructure required is 

sketched in Fig. 2. The scope of the technology 

largely depends on the characteristics and flexibility 

of its key tools: synthesized fields of laser 

light and attosecond pulses synchronized to them. 

From Controlled Light Waveforms Toward Optical 

Waveform Synthesis 

Lightwave electronics benefits from ever-broader 

optical bandwidth in several ways. Superposition 

of spectral components beyond an octave permits 

subcycle shaping of the light waveform 

and hence sculpting of the electric force on the 

electronic time scale for steering electrons in 

atomic systems. The increased bandwidth and 

improved dispersion control also lead to shorter 

optical pulses, which allow advancing subfemtosecond 

XUV pulse generation 

in terms of all relevant parameters: 

duration, intensity, and photon 

energy. 

Intense few-cycle optical 

pulses (~5 fs) were first generated 

(40) by spectral broadening of 

femtosecond pulses (initially ~25 

fs) in a hollow-core waveguide, 

and the resultant supercontinuum 

was compressed by reflection off 

of chirped multilayer dielectric 

mirrors (41) (Fig. 3A). Combined 

with waveform control (4), the 

technology matured to become a 

workhorse for attosecond science. 

Its current state is represented by 

the results summarized in Fig. 3, B 

and C. The spectrum of near–20-fs, 

800-nm pulses is first broadened 

by self-phase modulation in a gasfilled 

capillary to a supercontin- 

uum reaching from 400 to 1000 

nm, followed by compression of 

the ~ 400-μJ pulses to ~3.5 fs 

with octave-spanning chirped mirrors 

(42). These pulses constitute 

the shortest controlled optical 

waveforms demonstrated to date. 

Figure 3D shows how a 

single subfemtosecond XUV 

pulse can emerge from an atom 

ionized by a linearly polarized 

few-cycle light field. The green 

lines depict the kinetic energy, 

W kin, with which different portions 

of the freed electron wave packet (y free in 

Fig. 1B) return to the core as a function of return 

time, determining the energy of the emitted 

harmonic photon, ħw = Wkin + Wb, where Wb is 

the binding energy of the electron. For the 

optimum (near-cosine) waveform (Fig. 3D), we 

can estimate—in the limit of ħw X >> W b—the 

bandwidth over which photons of the highest 


energy (near the cutoff of the emitted spectrum 

depicted in violet) are emitted from one recollision 

only as 

DWsingle�pulse 


ℏwX 

≈ 0:6 

Tosc 

TFWHM 

2 

ð1Þ 

where T osc and T FWHM stand for the oscillation 

period (osc) of the carrier wave and the full width 

at half maximum (FWHM) duration of the 

laser pulse ionizing the atoms, respectively 

(43). Equation 1 underscores the importance of 

minimizing the number of cycles within the 

driver pulse. The intense sub–1.5 cycle (TFWHM 

< 1.5Tosc) laser pulses that can now be produced 

with commercially available ingredients of ultrafast 

laser technology (Fig. 3, A to C) constitute a 

promising tool for pushing the frontiers of 

attosecond-pulse technology. 

Advancing femtosecond technology to its 

ultimate limit is the key to extending the frontiers 

of both measuring and controlling dynamics with 

attosecond precision. Metrology benefits from 

broader spectral coverage and shorter duration of 

subfemtosecond pulses. The scope of control critically 

depends on the variety of waveforms, which 

is determined by the available bandwidth of intense 

coherent radiation. The shortest intense laser pulses 

also provide ideal conditions for generating the 

broadest supercontinua of coherent light, extending 

over more then three octaves (44). Advanced 

chirped multilayer mirror technology holds promise 

for versatile and scalable (multichannel) implementation 

of optical waveform synthesis. A 

prototypical three-channel synthesizer operational 

over the 1- to 3-eV band indicates the vast variety 

of possibilities for steering electrons in atomic 

systems with synthesized optical waves (fig. S2). 

From Subfemtosecond Toward 

Attosecond Pulses 

Equation 1 suggests that femtosecond technology 

must approach the monocycle limit to push the 

frontiers of XUV pulse generation toward and 

below 100 as. However, there are other options. 

Modulation of the polarization state of the ioniz- 

Fig. 2. Schematic of an experimental setup for attosecond-pulse generation and attosecond metrology as well as 

spectroscopy: the AS-1 attosecond beamline at MPQ. The intense, waveform-controlled few-cycle NIR laser pulse 

generates a subfemtosecond XUV pulse in the first interaction medium (jet of noble gas). The collinear XUV and NIR 

beams then propagate into a second vacuum chamber, where they are focused by a two-component XUV multilayer 

mirror into a gas target. The inner and outer part of the two-component mirror reflects and focuses the XUV and the 

(more divergent) NIR beam, respectively. By positioning the internal mirror with a nanometer-precision piezotranslator, 

the XUV pulse can be delayed with respect to the NIR pulse with attosecond accuracy. Analysis of the generated 

electrons and/or ions as a function of delay between the subfemtosecond XUV and the waveform-controlled NIR pulse 

permits characterization of the attosecond tools (Fig. 1) as well as real-time observation of atomic-scale electron 

dynamics in all forms of matter by means of attosecond streaking spectroscopy (3) and attosecond tunneling 

spectroscopy (13) with the same apparatus. In the former case, the final energy distribution of the outgoing electron is 

analyzed with a time-of-flight spectrometer; in the latter case, the observables measured as a function of delay are 

multiple-charged ions. ACF, auto-correlation function; CCD, charged-coupled device. 


ing NIR pulse (33, 37, 45) or of its amplitude by 

coherent addition of second harmonic radiation 

(46, 47) can efficiently increase DWsingle-pulse 

even by using several-cycle NIR driver pulses. 

Implementation of polarization gating with 

waveform-controlled, approximately two-cycle 

NIR pulses (37) has recently led to a spectacular 

achievement: the generation of isolated XUV 



Downloaded from

pulses at ħwcarrier ≈ 36 eV with a bandwidth of 

DW single-pulse ≈ 15 eV. Along with dispersion 

control and trajectory selection (48), these isolated 

pulses resulted in near–single-cycle XUV pulses 

that had a duration of 130 as (33) (see the lowenergy 

streaking spectrogram in Fig. 4A). 

The increased DWsingle-pulse with several-cycle 

pulses comes, however, at the expense of efficiency 

because (i) only a small temporal fraction of the 

driver pulse participates in the generation process 


and (ii) the cycles preceding the generation moment 

pre-ionize the atoms, causing substantial 

depletion of the ground state before the XUV pulse 

can be emitted. The unprecedented confinement of 

electromagnetic energy into a single wellcontrolled 

oscillation of light in sub–1.5-cycle 

NIR pulses (Fig. 3) avoids these shortcomings and 

offers several more benefits. Indeed, the high 

contrast between the wave crests at t = –Tosc/2 and 

earlier ones in Fig. 3D permits—thanks to the 


exponential scaling of ionization probability with 

the field strength—adjustment of the pulse intensity 

so as to allow the electron to survive with a 

high probability in its ground state until t = –T osc/2 

(in Fig. 3D) and then to be set free with a high 

probability near this instant through tunneling. The 

resulting wave packet, yfree (Fig. 1A), is launched 

with unprecedented amplitude. Upon its return to 

the core at about t ≈ Tosc/4, it interferes with the 

ground state portion of the wave function causing 

Fig. 3. Toward intense, monocycle optical pulses. 

(A) Hollow-fiber/chirped-mirror optical pulse compressor 

for the generation of intense, few-cycle 

laser pulses. (B) The spectrum of millijoule-energy, 

near–20-fs, phase-controlled near-infrared laser 

pulses delivered at a repetition rate of 3 kHz from 

a Ti:sapphire laser (Femtopower, Femtolasers 

GmbH) is first broadened to more than an octave 

(~450 to 950 nm) in a neon-filled capillary 

(pressure ~ 2 bar, length ~ 1 m, bore diameter 

250 mm). (C) The spectrally broadened pulses are 

compressed by octave-spanning chirped multilayer 

mirrors. The interferometric autocorrelation trace 

indicates a near–bandwidth-limited ~3.5-fs pulse 

carried at lL ≈ 720 nm. (D) Energies of XUV 

photons emitted from an atom exposed to a 

linearly polarized 3.5-fs, 720-nm Gaussian laser 

pulse that is sufficiently intense for efficient tunnel 

ionization. The photon energies (green lines) are 

depicted as a function of the return time of the 

electron to the atomic core and have been 

obtained by analyzing classical electron trajectories 

(43). The width of the energy band within 

which only one recollision contributes to XUV 

emission determines the bandwidth DW single-pulse 

available for the generation of an isolated attosecond 

pulse. Based on the classical trajectory analysis, 

DWsingle-pulse is found to be maximum for a near– 

cosine-shaped waveform, E L(t) = E 0e(t)cos(w Lt + 

ϕ), with the carrier-envelope phase ϕ varying 

between p/12 and p/6 depending on the pulse 

shape and intensity. Here, E 0 and w L stand for the 

amplitude and angular frequency of the oscillations 

of the laser electric field, with e(t) being 

the amplitude envelope function. The table summarizes 

bandwidths over which the sub–1.5-cycle 

pulses are predicted to support the emergence of 

an isolated attosecond pulse at different XUV 

photon energies. The estimated pulse durations 

derive from the conservative assumption that no 

spectral components are available for the pulse 

synthesis outside the given spectral window. 





Downloaded from

774 


high-frequency (soft–x-ray) dipole oscillations with 

unprecedented amplitude. Hence, the intense near– 

single-cycle waveforms appear to constitute ideal 

tools for the pursuit of powerful soft–x-ray pulses 

with pulse durations approaching the atomic unit 

of time (see table in Fig. 3D) and photon energies 

approaching the kilo–electron volt frontier. 

A first indication of the potential that near– 

single-cycle NIR pulses offer for attosecond-pulse 

generation is provided by the high-energy streaking 

spectrogram in Fig. 4A. It has been recorded 

with sub–two-cycle (near–4-fs) NIR (750-nm) 

pulses and XUV pulses filtered by a molybdensilicon 

chirped multilayer mirror with a highreflectivity 

band of ~16 eV (FWHM) centered at 

~93 eV (Fig. 4B). The bandwidth (FWHM) of the 

bandpass filtered XUV light amounts to ~13 eV. 

The streaking spectrogram indicates isolated sub– 

Fig. 4. Current frontiers of attosecond technology. 

(A) Attosecond streaking spectrograms recorded 

with few-cycle NIR pulses (l L ≈ 750 nm). The lowenergy 

spectrogram (bottom) shows the carrier 

photon energy of an XUV pulse, ħwX ≈ 36 eV, with 

argon target atoms (courtesy of M. Nisoli). The 

high-energy spectrogram (top) shows ħw X ≈ 93 eV, 

with neon target atoms (49). (B) Computed 

reflectivity and group delay of the chirped Mo/Si 

multilayer mirror used for filtering and focusing the 

isolated sub–170-as, 93-eV pulses. (C) Computed 

reflectivity and group delay of an ultrabroadband 

chirped multilayer Mo/Si mirror designed for compensating 

the chirp carried by an attosecond pulse 

filtered near cutoff from short-trajectory emission 

(group delay dispersion ~ –0.007 fs 2 ) and for reflecting 

a band sufficient for sub–100-as pulse 

generation (bandwidth > 35 eV). (D) Spectrum of 

high-order harmonics emitted from neon ionized 

by ~4-fs–duration NIR (~720-nm) pulses near 

cutoff as transmitted through a high-pass filter 

(150-nm palladium). The transmission of the Pd 

foil is shown by the red curve. The spectrum has 

been recorded with the carrier-envelope phase ϕ 

of the waveform-controlled NIR pulse set to 

provide the broadest continuum. Variation of ϕ 

reshapes the overall distribution of the continuum 

(42) rather than introducing a pronounced 

harmonic-like structure, as observed with severalcycle 

driver pulses for appropriate phase setting 

(dashed line). The smooth continuum with a 

bandwidth of >30 eV suggests the feasibility of 

sub–100-as XUV pulse generation without manipulation 

of the driving pulse (e.g., superimposing 

its second harmonic or gating its polarization). 


170-as pulses that are near bandwidth-limited 

(49). The pulses contain more than 10 6 photons, 

which are delivered at a repetition rate of 3 kHz, 

implying a photon flux of >3 × 10 9 photons per 

second transported in a near–diffraction-limited 

beam. With the Mo/Si mirror, which has a 10-cm 

focal length, used in the MPQ attosecond beamline 

AS-1 (Fig. 2), this beam can be focused to a 

spot diameter of several micrometers. 

The recently demonstrated, waveformcontrolled 

sub–1.5-cycle NIR pulses are capable 

of producing DWsingle-pulse > 30 eV at ħwX ≈ 120 

eV (Fig. 4D), offering the potential for generating 

sub–100-as pulses at a wavelength of ~10 nm (42). 

To exploit this potential, researchers must develop 

ultrabroadband soft–x-ray multilayer mirrors with 

precisely controlled group-delay (GD) dispersion. 

Figure 4C plots the reflectivity and GD of the first 


chirped multilayer mirror design that supports 

sub–100-as pulses. Chirped multilayer mirror 

technology (both metallic and dielectric) is likely 

to play as important a role in advancing attosecond 

technology as it has played in advancing femtosecond 

technology to its ultimate limits. 

Prospects 

New research tools allow scientists to seek answers 

to questions that, in the absence of a means 

of addressing them, have never been posed 

before. Grand questions serve as a compass 

providing directions for research that promises 

the most extensive return. We pose a few 

questions across disparate areas of science, connected 

by the fundamental role that atomic-scale 

electronic motion plays in physical, chemical, and 

biological processes. 




Downloaded from

Atomic physics and x-ray science. How is the 

energy of an x-ray photon distributed between 

electrons upon its absorption by an atom? Can 

electronic transitions deep inside atoms be 

affected by controlled ultrastrong external fields 

rivaling in strength the internal Coulomb fields, 

e.g., for opening up novel routes to efficient, compact 

x-ray lasers? 

Physical chemistry, molecular biology, bioinformatics, 

and photovoltaics. Can controlled 

light fields offer a fundamentally new way of 

modifying the structure and/or composition of 

molecules by driving electron wave packets across 

molecules with synthesized optical fields? What 

are the microscopic mechanisms underlying biological 

information transport? Can charge-transfer 

in host-guest systems (e.g., dye-semiconductor 

assemblies) be exploited for developing solar cells 

with unprecedented efficiency? 

Information technology. Can electron-based 

information processing and storage be downscaled 

to atomic dimensions and sped up to the 

atomic time scale (i.e., to optical frequencies)? 

Can these ultimate limits be reached by exploiting 

electric interactions (electronics) or magnetic 

interactions (spintronics) or collective electron 

motion (plasmonics)? Which incarnation of lightwave 

electronics will be the ultimate electronbased 

information technology? 

The answers to these questions will rely on 

exploring and controlling the microscopic motion 

of electrons, on atomic scales in space and time. 

Attosecond technology now offers the tools for 

tackling these and many other exciting questions. 

The importance of the answers being sought will 

drive its proliferation. 


1. T. Brabec, F. Krausz, Rev. Mod. Phys. 72, 545 (2000). 

2. A. H. Zewail, J. Phys. Chem. A 104, 5660 (2000). 


3. M. Drescher et al., Nature 419, 803 (2002). 

4. A. Baltuska et al., Nature 421, 611 (2003). 

5 For a comprehensive historical review, see P. Agostini, 

L. F. DiMauro, Rep. Prog. Phys. 67, 813 (2004). 

6. R. Trebino et al., Rev. Sci. Instrum. 68, 3277 (1997). 

7. C. Iaconis, I. A. Walmsley, IEEE J. Quantum Electron. 35, 

501 (1999). 

8. P. W. Brumer, M. Shapiro, Principles of the Quantum 

Control of Molecular Processes (Wiley, New York, 2003). 

9. W. Ackermann et al., Nature Phot. 1, 336 (2007). 

10. A. A. Zholents, W. M. Fawley, Phys. Rev. Lett. 92, 224801 

(2004). 

11. P. Tzallas, D. Charalambidis, N. A. Papadogiannis, K. Witte, 

G. Tsakiris, Nature 426, 267 (2003). 

12. T. Sekikawa, A. Kosuge, T. Kanai, S. Watanabe, Nature 

432, 605 (2004). 

13. M. Uiberacker et al., Nature 446, 627 (2007). 


15. P. Antoine, A. L’Huillier, M. Lewenstein, Phys. Rev. Lett. 

77, 1234 (1996). 

16. H. Niikura et al., Nature 417, 917 (2002). 

17. T. Kanai, S. Minemoto, H. Sakai, Nature 435, 470 

(2005). 

18. S. Baker et al., Science 312, 424 (2006). 

19. The first implementation of the basic concept of a 

light-field–driven streak camera has drawn on an 

orthogonal detection geometry (electrons collected 

along a direction orthogonal to the electric field vector 

of the streaking NIR field (29). 

20. The laser field needed to induce this change in electron 

energy is orders of magnitude less intense than that 

required for strong-field ionization. Hence, this 

streaking field has negligible influence on the atomic 

or molecular processes under study, unless its oscillations 

happen to be in resonance with a transition from an 

occupied to an unoccupied quantum state of the system. 

21. P. M. Paul et al., Science 292, 1689 (2001). 

22. M. Ferray et al., J. Phys. B 21, L31 (1988). 

23. K. J. Schafer, M. B. Gaarde, A. Heinrich, J. Biegert, 

U. Keller, Phys. Rev. Lett. 92, 023003 (2004). 

24. T. Remetter et al., Nature Phys. 2, 323 (2006). 

25. J. Mauritsson et al., Phys. Rev. Lett. 97, 013001 (2006). 

26. Experiments that can be performed with subfemtosecond 

pulse trains [including those described in (23–25)] 

would benefit from using a small number of wellcontrolled 

and characterized subfemtosecond pulses. 

27. P. B. Corkum, N. H. Burnett, M. Yu. Ivanov, Opt. Lett. 19, 

1870 (1994). 

REVIEW 

Harnessing Attosecond Science in the 

Quest for Coherent X-rays 

Henry Kapteyn, Oren Cohen, Ivan Christov, Margaret Murnane* 

Modern laser technology has revolutionized the sensitivity and precision of spectroscopy by 

providing coherent light in a spectrum spanning the infrared, visible, and ultraviolet wavelength 

regimes. However, the generation of shorter-wavelength coherent pulses in the x-ray region 

has proven much more challenging. The recent emergence of high harmonic generation techniques 

opens the door to this possibility. Here we review the new science that is enabled by an ability to 

manipulate and control electrons on attosecond time scales, ranging from new tabletop sources of 

coherent x-rays to an ability to follow complex electron dynamics in molecules and materials. We 

also explore the implications of these advances for the future of molecular structural 

characterization schemes that currently rely so heavily on scattering from incoherent x-ray sources. 

Next year, 2008, will mark the 50th anniversary 

of the revolutionary paper by 

Schawlow and Townes that proposed the 

laser (1). This paper extended concepts first used to 

demonstrate the maser in the microwave region of 

the spectrum into the visible spectrum. Soon after 

the laser was demonstrated, scientists discovered 

how to control laser light to generate extremely 

28. I. P. Christov, M. M. Murnane, H. C. Kapteyn, Phys. Rev. 

Lett. 78, 1251 (1997). 

29. M. Hentschel et al., Nature 414, 509 (2001). 

30. J. Itatani et al., Phys. Rev. Lett. 88, 173903 (2002). 

31. R. Kienberger et al., Nature 427, 817 (2004). 

32. E. Goulielmakis et al., Science 305, 1267 (2004). 

33. G. Sansone et al., Science 314, 443 (2006). 

34. I. P. Christov, R. Bartels, H. C. Kapteyn, M. M. Murnane, 

Phys. Rev. Lett. 86, 5458 (2001). 

35. N. Dudovich et al., Nature Phys. 2, 781 (2006). 

36. M. F. Kling et al., Science 312, 246 (2006). 

37. I. J. Sola et al., Nature Phys. 2, 319 (2006). 

38. F. Quéré, Y. Mairesse, J. Itatani, J. Mod. Opt. 52, 339 (2005). 

39. A. D. Bandrauk, S. Chelkowski, H. S. Nguyen, Int. J. 

Quantum Chem. 100, 834 (2004). 

40. M. Nisoli et al., Opt. Lett. 22, 522 (1997). 

41. R. Szipöcs, K. Ferencz, C. Spielmann, F. Krausz, Opt. Lett. 

19, 201 (1994). 

42. A. L. Cavalieri et al., New J. Phys. 9, 242 (2007). 

43. The electron kinetic energies have been calculated in 

terms of the classical description of the center-of-mass 

motion of the freed electron wave packet (Fig. 2), under 

the assumption of a Gaussian pulse shape and by using 

the strong-field approximation. The factor of ~0.6 in 

Eq. 1 depends on the pulse shape and intensity but varies 

less than 15% in the relevant parameter range. 

44. N. Aközbek et al., New J. Phys. 8, 177 (2006). 

45. Z. Chang, Phys. Rev. A. 70, 043802 (2004). 

46. T. Pfeifer et al., Phys. Rev. Lett. 97, 163901 (2006). 

47. Y. Oishi, M. Kaku, A. Suda, F. Kannari, K. Midorikawa, 

Opt. Exp. 14, 7230 (2006). 

48. R. Lopez-Martens et al., Phys. Rev. Lett. 94, 033001 (2005). 

49. M. Schultze et al., New J. Phys. 9, 243 (2007). 

50. We apologize that many original research papers could not 

be cited because of space limitations. This work is 

supported by the Max-Planck-Society and the Deutsche 

Forschungsgemeinschaft through the DFG Cluster of 

Excellence Munich-Centre for Advanced Photonics 

(www.munich-photonics.de). E.G. acknowledges support 

from a Marie Curie Intra-European Fellowship. We thank B. 

Ferus, M. Hofstätter, B. Horvath, and M. Schultze for their 

support in the preparation of this manuscript. 

Supporting Online Material 

www.sciencemag.org/cgi/content/full/317/5839/769/DC1 

Figs. S1 and S2 

References 

10.1126/science.1142855 


short nanosecond, picosecond, and even femtosecond 

pulses. Given the origin of the laser, it was 

also natural to attempt to generate coherent light at 

shorter and shorter wavelengths. However, this 

effort proved very challenging because of the punishing 

power scaling inherent in lasers. Basic physics 

dictates that the energy required to implement a 

laser scales roughly as 1/l 5 ; that is, a laser at a 10 

times shorter wavelength (l) requires ~100,000 

times the input power. Thus, the first x-ray lasers 

implemented in the 1980s used the building-sized 

Nova fusion laser at Lawrence Livermore National 

Laboratory as a power source to generate soft (relatively 

long-wavelength) x-rays. Since that initial 

x-ray laser, considerable progress has been made in 

downscaling the laser needed as the power source 

JILA and the National Science Foundation Center for 

Extreme Ultraviolet Science and Technology, University of 

Colorado at Boulder, Boulder, CO 80309–0440, USA. 


murnane@jila.colorado.edu 





Downloaded from

ARTICLES 


Coherent superposition of laser-driven 

soft-X-ray harmonics from successive 

sources 

J. SERES 1,2 , V. S. YAKOVLEV 3 , E. SERES 1,2 , CH. STRELI 4 , P. WOBRAUSCHEK 4 , CH. SPIELMANN 2 

AND F. KRAUSZ 3,5 * 

1 Institut für Photonik, Technische Universität Wien, A-1040 Wien, Austria 

2 Physikalisches Institut EP1, Universität Würzburg, D-97074 Würzburg, Germany 

3 Department für Physik, Ludwig-Maximilians-Universität München, D-85748 Garching, Germany 

4 Atominstitut der Österreichischen Universtitäten, Technische Universität Wien, A-1020 Wien, Austria 

5 Max-Planck-Institut für Quantenoptik, D-85748 Garching, Germany 

*e-mail: krausz@lmu.de 

Published online: 11 November 2007; doi:10.1038/nphys775 

High-order harmonic generation from atoms ionized by femtosecond laser pulses has been a promising approach for the development 

of coherent short-wavelength sources. However, the realization of a powerful harmonic X-ray source has been hampered by a phase 

velocity mismatch between the driving wave and its harmonics, limiting their coherent build-up to a short propagation length and 

thereby compromising the efficiency of a single source. Here, we report coherent superposition of laser-driven soft-X-ray (SXR) 

harmonics, at wavelengths of 2–5 nm, generated in two successive sources by one and the same laser pulse. Observation of constructive 

and destructive interference suggests the feasibility of quasi-phase-matched SXR harmonic generation by a focused laser beam in a 

gas medium of modulated density. Our proof-of-concept study opens the prospect of enhancing the photon flux of SXR harmonic 

sources to levels enabling researchers to tackle a range of applications in physical as well as life sciences. 

The quest for powerful laboratory sources of coherent soft-X-ray 

(SXR) light has been ongoing since the discovery of the laser. 

So far, high-order harmonic generation from atoms ionized by 

ultrashort laser pulses 1,2 constitutes the only technique providing 

coherent short-wavelength radiation at any photon energy up 

to the kiloelectronvolt regime 3–11 . Laser-driven high-harmonic 

sources (henceforth briefly referred to as harmonic sources) up 

to photon energies of 100 eV are now in widespread use 12 and 

have played a key role in pushing the frontiers of nonlinear 

optics into the extreme-ultraviolet region 13–17 and ultrafast science 

into the attosecond domain 18–23 . Above 100 eV, rapidly decreasing 

efficiency prevented harmonic sources from becoming useful for 

applications. The low harmonic conversion efficiency implies that 

most of the incident laser photons are transmitted through the 

harmonic source, offering the possibility of being reused for 

creating harmonics in successive sources and—by exploiting their 

coherence—adding them to increase the overall harmonic flux. 

Here, we demonstrate the feasibility of this approach. 

Efficient generation of high-order harmonics of intense laser 

light relies on a large number of ionizing atoms emitting 

high-frequency radiation with proper phase that allows coherent 

build-up of light on propagation through the generation medium. 

Free electrons, unavoidable concomitants of the generation 

process, tend to severely limit the propagation length over which 

phase-matching can be achieved (henceforth briefly referred to as 

the coherence length). Femtosecond laser pulses have permitted 

the generation of harmonics in the extreme-ultraviolet regime 

(10–100 eV) at sufficiently low levels of ionization, so that the 

coherence length could be extended beyond the absorption 

length 4,5 , leading to the production of microjoule-energy coherent 

extreme-ultraviolet pulses 6,7 . 

Generation of SXR harmonics (>100 eV) is favoured by 

lowered absorption but suffers from increased free-electron density 

implied by the higher laser intensities required. Hence, the 

efficiency of SXR harmonic sources is limited by dephasing of 

the atomic dipole oscillators driven at different positions in the 

generation medium 24 . The SXR harmonic yield has recently been 

improved by implementing quasi-phase-matching (QPM) in a 

gas-filled hollow-core fibre with a modulated inner diameter 8 or 

counter-propagating laser pulses 25,26 . At higher photon energies, 

subcycle modification of few-cycle driving fields has been identified 

as being beneficial for enhancing phase-matching 27 (referred to 

as non-adiabatic self-phase-matching), which allowed extension 

of harmonic generation beyond the kiloelectronvolt frontier 10,11 . 

Unfortunately the fluxes demonstrated so far are still insufficient 

for most applications. 

Here, we provide the first experimental evidence of the 

feasibility of enhancing the efficiency of SXR harmonic generation 

by coherent superposition of harmonics generated in successive 

sources traversed by the same pump laser beam (Fig. 1). SXR 

harmonic emission from a single source is maximized by using 

few-cycle driver pulses, which maximize both the single-atom 

emission intensity and the coherence length for SXR harmonics 24 , 

and is limited by dephasing. The phase of the atomic dipole 

878 nature physics VOL 3 DECEMBER 2007 www.nature.com/naturephysics 

© 2007 Nature Publishing Group 



Enhanced phase-matching for 

generation of soft X-ray harmonics and 

attosecond pulses in atomic gases 

Vladislav S. Yakovlev 1∗ , Misha Ivanov 2 , Ferenc Krausz 1,3 

1 Department of Physics, Ludwig-Maximilians-Universität München, Am Coulombwall 1, 

D-85748 Garching, Germany 

2 National Research Council of Canada, M-23A, Ottawa, Ontario, Canada K1A OR6 

3 Max-Planck-Institut für Quantenoptik, Hans-Kopfermann-Strasse 1, D-85748 Garching, 

Germany 

∗ Corresponding author: Vladislav.Yakovlev@physik.uni-muenchen.de 

Abstract: We theoretically investigate the generation of high harmonics 

and attosecond pulses by mid-infrared (IR) driving fields. Conditions for 

coherent build-up of high harmonics are revisited. We show that the coherence 

length dictated by ionization-induced dephasing does not constitute 

an ultimate limitation to the coherent growth of soft X-ray (> 100 eV) harmonics 

driven by few-cycle mid-IR driving pulses: perfect phase-matching, 

similar to non-adiabatic self-phase-matching, can be achieved even without 

non-linear deformation of the driving pulse. Our trajectory-based analysis 

of phase-matching reveals several important advantages of using longer 

laser wavelengths: conversion efficiency can be improved by orders of 

magnitude, phase-matched build-up of harmonics can be achieved in a jet 

with a high gas pressure, and isolated attosecond pulses can be extracted 

from plateau harmonics. 

© 2007 Optical Society of America 

OCIS codes: (190.7110) Ultrafast nonlinear optics; (270.6620) Strong-field processes; 

999.9999 Attosecond science. 

References and links 

1. P. B. Corkum, F. Krausz, “Attosecond science,” Nat. Phys. 3, 381–387 (2007). 

2. M. Hentschel, R. Kienberger, Ch. Spielmann, G. A. Reider, N. Milosevic, T. Brabec, P. Corkum, U. Heinzmann, 

M. Drescher, F. Krausz, “Attosecond metrology,” Nature 414, 509–513 (2001). 

3. R. Kienberger, E. Goulielmakis, M. Uiberacker, A. Baltuˇska, V. Yakovlev , F. Bammer, A. Scrinzi, Th. Westerwalbesloh, 

U. Kleineberg, U. Heinzmann, M. Drescher, F. Krausz, “Atomic transient recorder,” Nature 427, 

817–821 (2004). 

4. G. Sansone, E. Benedetti, F. Calegari, C. Vozzi, L. Avaldi, R. Flammini, L. Poletto, P. Villoresi, C. Altucci, R. 

Velotta, S. Stagira, S. De Silvestri, M. Nisoli, “Isolated single-cycle attosecond pulses,” Science 314, 443–446 

(2006). 

5. J. Seres, E. Seres, A. J. Verhoef, G. Tempea, C. Streli, P. Wobrauschek, V. Yakovlev, A. Scrinzi, C. Spielmann, 

F. Krausz, “Source of coherent kiloelectronvolt X-rays,” Nature 433, 596 (2005). 

6. A. Gordon, F. X. Kärtner, “Scaling of keV HHG photon yield with drive wavelength,” Opt. Express 13, 2941– 

2947 (2005). 

7. J. Tate, T. Auguste, H. G. Muller, P. Salieres, P. Agostini, L. F. DiMauro, “Scaling of wave-packet dynamics in 

an intense midinfrared field,” Phys. Rev. Lett. 98, 013901 (2007). 

8. S. Gordienko, A. Pukhov, O. Shorokhov, T. Baeva, “Relativistic Doppler Effect: Universal Spectra and Zeptosecond 

Pulses”, Phys. Rev. Lett. 93, 115002 (2004). 

9. G. D. Tsakiris, K. Eidmann, J. Meyer-ter-Vehn, Ferenc Krausz, “Route to intense single attosecond pulses,” New 

J. Phys. 8, 1–20 (2006). 

#84689 - $15.00 USD Received 29 Jun 2007; revised 21 Sep 2007; accepted 25 Sep 2007; published 5 Nov 2007 

(C) 2007 OSA 12 November 2007 / Vol. 15, No. 23 / OPTICS EXPRESS 15351 



New Journal of Physics 

T h e o p e n – a c c e s s j o u r n a l f o r p h y s i c s 

GeV-scale electron acceleration in a gas-filled 

capillary discharge waveguide 

S Karsch 1,6 , J Osterhoff 1 , A Popp 1 , T P Rowlands-Rees 2 , 

Zs Major 1 , M Fuchs 1,3 , B Marx 1,3 , R Hörlein 1,3 , K Schmid 1,3 , 

L Veisz 1 , S Becker 3 , U Schramm 4 , B Hidding 5 , G Pretzler 5 , 

D Habs 3 , F Grüner 1 , F Krausz 1,3 and S M Hooker 2 

1 Max-Planck-Institut für Quantenoptik, Hans-Kopfermann-Str. 1, 


2 Clarendon Laboratory, University of Oxford, Parks Road, 

Oxford OX1 3PU, UK 

3 Sektion Physik der Ludwig-Maximilians-Universität München, 

Am Coulombwall 1, D-85748 Garching, Germany 

4 Forschungszentrum Dresden-Rossendorf, Bautzner Landstr. 128, 

D-01328 Dresden, Germany 

5 Institut für Laser- und Plasmaphysik, Heinrich-Heine-Universität, 

Universitätsstr. 1, D-40225 Düsseldorf, Germany 

E-mail: stefan.karsch@mpq.mpg.de 

New Journal of Physics 9 (2007) 415 

Received 14 September 2007 

Published 23 November 2007 

Online at http://www.njp.org/ 

doi:10.1088/1367-2630/9/11/415 

Abstract. We report experimental results on laser-driven electron acceleration 

with low divergence. The electron beam was generated by focussing 750 mJ, 

42 fs laser pulses into a gas-filled capillary discharge waveguide at electron 

densities in the range between 10 18 and 10 19 cm −3 . Quasi-monoenergetic electron 

bunches with energies as high as 500 MeV have been detected, with features 

reaching up to 1 GeV, albeit with large shot-to-shot fluctuations. A more stable 

regime with higher bunch charge (20–45 pC) and less energy (200–300 MeV) 

could also be observed. The beam divergence and the pointing stability are 

around or below 1 mrad and 8 mrad, respectively. These findings are consistent 

with self-injection of electrons into a breaking plasma wave. 

6 Author to whom any correspondence should be addressed. 

New Journal of Physics 9 (2007) 415 PII: S1367-2630(07)60033-0 

1367-2630/07/010415+11$30.00 © IOP Publishing Ltd and Deutsche Physikalische Gesellschaft 





Hybrid dc–ac electron gun for fs-electron 

pulse generation 

L Veisz 1,4 , G Kurkin 2 , K Chernov 2 , V Tarnetsky 2 , A Apolonski 3 , 

F Krausz 1,3 and E Fill 1 

1 Max-Planck-Institut für Quantenoptik, Hans-Kopfermann-Strasse 1, 


2 Budker Institute for Nuclear Physics, 630090 Novosibirsk, Russia 

3 Department für Physik der Ludwig-Maximilians-Universität München, 

Am Coulombwall 1, D-85748 Garching, Germany 

E-mail: laszlo.veisz@mpq.mpg.de 


Received 5 August 2007 

Published 20 December 2007 


doi:10.1088/1367-2630/9/12/451 

Abstract. We present a new concept of an electron gun for generating 

subrelativistic few-femtosecond (fs) electron pulses. The basic idea is to utilize 

a dc acceleration stage combined with an RF cavity, the ac field of which 

generates an electron energy chirp for bunching at the target. To reduce space 

charge (SC) broadening the number of electrons in the bunch is reduced and 

the gun is operated at a megahertz (MHz) repetition rate for providing a high 

average number of electrons at the target. Simulations of the electron gun were 

carried out under the condition of no SC and with SC assuming various numbers 

of electrons in the bunch. Transversal effects such as defocusing after the dc 

extraction hole were also taken into account. A detailed analysis of the sensitivity 

of the pulse duration to various parameters was performed to test the realizability 

of the concept. Such electron pulses will allow significant advances in the 

field of ultrafast electron diffraction. 






Picosecond electron deflectometry 

of optical-field ionized plasmas 

MARTIN CENTURION 1 * † , PETER RECKENTHAELER 1,2† , SERGEI A. TRUSHIN 1 , FERENC KRAUSZ 1,2 

AND ERNST E. FILL 1 

1 Max-Planck-Institut für Quantenoptik, Hans-Kopfermann-Strasse 1, D-85748 Garching, Germany 

2 Ludwig-Maximilians-Universität München, Am Coulombwall 1, D-85748 Garching, Germany 

† These authors contributed equally to this work. 

*e-mail: martin.centurion@mpq.mpg.de 

Published online: 20 April 2008; doi:10.1038/nphoton.2008.77 

Optical-field ionized plasmas are of great interest owing to their 

unique properties and the fact that they suit many applications, 

such as the study of nuclear fusion 1 , generation of energetic 

electrons 2–5 and ions 6,7 , X-ray emission 8,9 , X-ray lasers 10–12 

and extreme-UV attosecond pulse generation 13 . A detailed 

knowledge of the plasma dynamics can be critical for 

optimizing a given application. Here we demonstrate a method 

for real-time imaging of the electric-field distribution in 

optical-field ionized plasmas with ultrahigh temporal 

resolution, yielding information that is not accessible by other 

methods. The technique, based on electron deflectometry, 

yields images that reveal a positively charged core and a cloud 

of electrons expanding far beyond the Debye length. 

The parameters of an optical-field ionized (OFI) plasma can be 

varied over a wide range. The maximum ionization stage can be set 

by the intensity of the laser pulse generating the plasma 14 , the 

electron temperature can be controlled by applying linear or 

circular polarization 15 , and, by appropriately choosing the 

backing pressure of the nozzle, cluster plasmas 16 can be 

generated. Various methods have been applied to investigate the 

parameters of OFI plasmas. The electron temperature has been 

measured by Thomson scattering 17 . The ionization stage has been 

determined by ion spectrometry 18 and by recording the emission 

of X-rays from the plasma 11 . The time-resolved plasma density 

profile has been measured by optical interferometry and 

holography 19,20 . Moiré deflectometry has been used to determine 

the density profile in the plasma channel and its lateral 

expansion 21 . Spectrometry of ions emitted from the plasma has 

yielded information on ion velocity and temperature 22 . 

Radiography using energetic (MeV) protons has been used to 

diagnose density perturbations and transient fields in highdensity 

plasmas with a temporal resolution of 100 ps (refs 23–25). 

Here, we demonstrate a new diagnostics technique for OFI 

plasmas with a temporal resolution of 2.7 ps and very high 

sensitivity. The technique is deflectometry using monoenergetic 

20-keV electron pulses, which are directed onto an OFI nitrogen 

plasma generated by a 50-fs Ti:sapphire laser pulse. The electrons 

are deflected by the fields resulting from charge separation, and 

the resulting distortion of the electron beam yields time-resolved 

images of the plasma. Pump–probe experiments on the plasma 

evolution capture changes within a few picoseconds, with a 

spatial resolution of 30 mm. This direct time-resolved imaging of 

plasma fields reveals features not accessible to other methods. 

Such knowledge can help to improve the setting of parameters 

for optimizing particular applications. As an example, X-ray 

lasers and soft X-ray sources may greatly benefit from a 

better understanding of the dynamics of laser-generated 

plasmas. The new technique has the potential to lead to better 

control of plasma electron and ion accelerators and improve 

their features. The high sensitivity of electron deflectometry is 

based on the fact that even small charge imbalances within the 

plasma are observable as distortions in the spatial profile of 

the electron beam. The narrow energy spread and low emittance 

(1 ¼ 1.5 mrad mm) of the electron beam allows plasma fields 

below 1 � 10 6 V m 21 to be detected. 

An application of the method is illustrated in Fig. 1, which 

shows the evolution of a nitrogen OFI plasma. Initially, a small 

depleted region appears in the electron beam in the area of the 

laser focus. This hole expands for approximately 80 ps, after which 

a spot develops in the centre (Fig. 1e,f). The spot becomes brighter 

than the initial electron beam for a time delay T . 100 ps, and its 

intensity increases up to 200 ps. Beyond 200 ps (not shown in the 

figure) the brightness of the spot slowly decreases. 

Simultaneously with the depleted region, two bright lobes 

appear on each side of the plasma region, along the line of laser 

propagation (vertical in Figs 1 and 2). These lobes then move 

away from the focal region in opposite directions. The sidelobes 

reach the boundary of the electron beam after approximately 

100 ps (Fig. 1e) and then remain static. The pattern remained 

qualitatively unchanged until the maximum delay time in the 

experiment (T ¼ 300 ps), with only a slight decrease in the 

brightness of the main features. 

Figure 2 focuses on the initial stages of the plasma expansion. 

The duration of the electron pulses was decreased by reducing 

the number of electrons per pulse. Figure 2a–d shows the plasma 

evolution with a step size of �2.7 ps. The formation of an 

expanding feature is clearly visible after the first step. After only 

5.3 ps (Fig. 2c) there is already a well-defined boundary 

and deflected electrons appear in the shadow of the nozzle. For 

T . 5.3 ps (Fig. 2d–i) the hole expands and the number of 

electrons deflected out of the focal region increases. The sidelobes 

at the boundary of the plasma also appear at 5.3 ps (Fig. 2c), 

then increase in brightness and move away from the centre. At 

T ¼ 24 ps (Fig. 2g) the fraction of electrons ejected reaches its 


LETTERS 

nature photonics | VOL 2 | MAY 2008 | www.nature.com/naturephotonics 315 

© 2008 Nature Publishing Group

REPORTS 

1614 


Single-Cycle Nonlinear Optics 

E. Goulielmakis, 1 * M. Schultze, 1 M. Hofstetter, 2 V. S. Yakovlev, 2 J. Gagnon, 1 

M. Uiberacker, 2 A. L. Aquila, 3 E. M. Gullikson, 3 D. T. Attwood, 3 R. Kienberger, 1 

F. Krausz, 1,2 * U. Kleineberg 2 * 

Nonlinear optics plays a central role in the advancement of optical science and laser-based 

technologies. We report on the confinement of the nonlinear interaction of light with matter to a single 

wave cycle and demonstrate its utility for time-resolved and strong-field science. The electric field of 

3.3-femtosecond, 0.72-micron laser pulses with a controlled and measured waveform ionizes atoms 

near the crests of the central wave cycle, with ionization being virtually switched off outside this 

interval. Isolated sub-100-attosecond pulses of extreme ultraviolet light (photon energy ~ 80 electron 

volts), containing ~0.5 nanojoule of energy, emerge from the interaction with a conversion efficiency 

of ~10 –6 . These tools enable the study of the precision control of electron motion with light fields and 

electron-electron interactions with a resolution approaching the atomic unit of time (~24 attoseconds). 

N 

onlinear electron-light interactions driven 

by strong light fields of controlled wave- 

form (1) have allowed for the control of 

electronic motion at light frequencies and the realtime 

observation of electron dynamics inside and 

between atoms with ~100-as resolution (2–7). 

However, time-domain access to a number of 

fundamental processes, such as the intra-atomic 

energy transfer between electrons (resulting, for 

example, in shake-up) (8), the response of an 

atomic electron system to external influence (e.g., 

to ionizing radiation) (9) and its rearrangement 

after the sudden loss of one or more electrons 

Fig. 1. Simulation of sub-femtosecond XUV emission from 

neon atoms ionized by a linearly polarized, sub-1.5-cycle, 720nm 

laser field. E0 and aL(t) are inferred from best agreement 

between the modeled (17) and measured (Fig. 2) spectra and 

the streaking spectrogram (Fig. 3), respectively. The laser field 

liberates electrons near its most intense wave crests. (A) 

Classical trajectories of maximum return energy (left panels) 

and spectra of emerging XUV emission (right panels) are 

shown for waveforms consistent with a L(t) inferred from Fig. 3 

(correspondence established by colors and line style). The 

numbers in (i) to (iii) quantify, in units of 10 –4 , the ionization 

probability and, hence, the squared modulus of the amplitude of 

the electron wave packets launched. This amplitude substantially 

dictates the intensity of XUV emission upon recollision: 

Contrast ionization probability in (i) to (iii) with the corresponding 

emission intensities in (iv) to (vi). The pink “dottedline” 

emission is not visible in (v) because of the lower 

ionization probability (by two orders of magnitude) with 

respect to that resulting in the purple “solid-line” emission 

[see (ii)]. The gray dashed-and-dotted lines denote the bandpass 

used in our experiments (fig. S2). ϕ = 70° and ϕ = 135° 

yield highest contrast [see (B)] and highest XUV cutoff energy, 

respectively. arb.u., arbitrary units. (B) Contrast versus CE 

phase. Here, contrast is defined as the ratio of the energy of 

the main attosecond XUV pulse to the overall XUV emission 

energy transmitted through the bandpass [gray dashed-anddotted 

lines in (A)]. 

(10), the charge transfer in biologically relevant 

molecules (11) and related changes in chemical 

reactivity (12) or because of nonadiabatic tunneling 

(13, 14), would require (or benefit from) 

an improved temporal resolution. 

We used waveform-controlled sub-1.5-cycle 

near-infrared (NIR) light to demonstrate the generation 

of robust, energetic, isolated sub-100-as 

pulses of extreme ultraviolet (XUV) radiation and 

their precise temporal characterization. Photoionization 

confined to a single wave cycle results 

in observables (such as high-harmonic photons 

and electrons emitted by above-threshold ioniza- 

A 

Distance of electron from the core (Å) 

B 

Contrast 

20 

10 

0 

-10 

-20 

20 

10 

0 

-10 

-20 

20 

10 

0 

-10 

-20 

1.0 

0.9 

0.8 

0.7 

0.6 

(i) 

(ii) 

(iii) 

E L (t) 

E L (t) 

E L (t) 

0.49 

5.9 

36 

14 

48 

26 

45 

8.6 

0.3 

-3.75 0 

Time (fs) 

3.75 

tion) that can now be related to several distinguishable 

subcycle ionization events and subsequent 

electron trajectories with a known timing with 

respect to the driving field, whose strength and 

temporal evolution is accurately known (3). These 

circumstances provide ideal conditions for testing 

models of strong-field control of electron motion 

and electron-electron interactions. 

The generation of attosecond pulses benefits 

from the abrupt onset of ionization within a single 

half-cycle, which minimizes the density of free 

electrons and, hence, the distortion of the driving 

wave and its dephasing with the generated harmonic 

wave. As a result, the coherent build-up of the 

harmonic emission over an extended propagation 

is maximized. In addition, the order-of-magnitude 

variation of the ionization probability between adjacent 

half-cycles creates unique conditions for 

single sub-100-as pulse emission without the need 

for sophisticated gating techniques (5, 15, 16). 

On the measurement side, improved resolution 

results from three provisions: (i) shorter 

1 Max-Planck-Institut für Quantenoptik, Hans-Kopfermann- 

Strasse 1, D-85748 Garching, Germany. 2 Department für 

Physik, Ludwig-Maximilians-Universität, Am Coulombwall 

1, D-85748 Garching. 3 Center for X-Ray Optics, Lawrence 

Berkeley National Laboratory, Berkeley, CA 94720, USA. 

*To whom correspondence should be addressed. E-mail: elgo@ 

mpq.mpg.de (E.G.); krausz@lmu.de (F.K.); ulf.kleineberg@ 

physik.uni-muenchen.de (U.K.) 

0 

40 60 80 100 120 

Energy (eV) 


XUV spectral intensity (arb.u.) 

1 

0 

1 

0 

1 

(iv) 

(v) 

(vi) 

ϕ = 0° 

ϕ = 70° 

ϕ = 135° 

0 20 40 60 80 100 120 140 160 180 

Carrier envelope phase, ϕ (deg) 

20 JUNE 2008 VOL 320 SCIENCE www.sciencemag.org 

on June 24, 2008 


Downloaded from

Photoelectron energy (eV) 

70 

60 

50 

40 

A 

B 

70 

60 

50 

40 


0 1 

20 40 60 80 100 120 140 160 

Carrier-envelope phase (deg) 

Electron counts (arb. u.) 

1 

C 

0 

1 

D 

0 

1 

E 

XUV pulse duration, (ii) improved signal-to-noise 

(S/N) ratio due to the increased XUV photon 

flux, and (iii) stronger streaking before the onset 

of the NIR field–induced ionization in attosecond 

streaking (2) or enhanced S/N ratio due to a 

reduced number of tunneling steps in attosecond 

tunneling spectroscopy (14). 

Figure 1 summarizes results of the modeling 

of the single-cycle interaction of ionizing NIR 

radiation with an ensemble of neon atoms (17). In 

Fig. 1A, the left panels plot possible NIR electric 

waveforms,ELðtÞ ¼ E0aLðtÞe −iðwLtþϕÞ þ cc 

(where cc stands for complex conjugate) derived 

from our streaking measurements (as presented 

in the next sections) for different settings of the 

carrier-envelope (CE) phase, ϕ. Here, E0 is the 

peak electric-field strength, aL(t) is the normalized 

complex amplitude envelope, and wL is the 

carrier frequency. The probability of ionization 

outside the central cycle is more than two orders 

of magnitude lower than that at the field maximum 

and hence is negligible. 

The spectra of XUVemissions originating from 

the individual recollisions (18) are predicted to 

differ by tens of electron volts in cut-off energy and 

by up to orders of magnitude in intensity as a consequence 

of the single-cycle nature of the driving 

field. The strong variation of emission energies and 

intensities within a single wave cycle creates ideal 

conditions for isolated sub-100-as pulse generation. 

Indeed, filtering radiation with the bandpass 

depicted by the dashed-and-dotted line is predicted 

to isolate XUVradiation with more than 90% of its 

energy delivered in a single attosecond pulse for a 

range of CE phases as broad as 30° ≤ ϕ ≤ 90° (Fig. 

1B). In contrast, with few-cycle-driven harmonic 

generation resulting in isolated subfemtosecond 

pulses over only a relatively narrow range of the 

CE phase near ϕ ≈ 0° (3), single-cycle excitation 

appears to permit robust isolated attosecond pulses 

for a variety of driver waveforms, ranging from 

near-cosine– to sine-shaped ones, owing to the 

order-of-magnitude variation of the ionization 

probability within a single wave cycle. 

We used phase-controlled sub-1.5-cycle laser 

pulses carried at a wavelength of lL = 2pc/wL = 

720 nm (19) to generate XUV harmonics in a 

neon gas jet up to photon energies of ~110 eV 

(fig. S1). The emerging XUV pulse—following 

a spectral filtering through a bandpass (dashedand-dotted 

line in Fig. 1A) introduced by metal 

foils and a Mo/Si multilayer mirror (fig. S2)— 

subsequently propagates, along with its NIR driver 

wave, through a second jet of neon atoms in 

which the XUV pulse ionizes the atoms in the 

presence of the NIR field. The freed electrons 

with initial momenta directed along the electricfield 

vector of the linearly polarized NIR field are 

collected and analyzed with time-of-flight spectrometry 

(17). 

The variation of the measured photoelectron 

spectra versus CE phase shows good agreement 

with the predictions of our simulations (Fig. 2, 

A and B). Figure 2, C to E, shows plots of 

electron spectra corresponding to the CE phase 


ϕ = 70° 

ϕ = 130° 

ϕ = 170° 

0 

30 40 50 60 70 80 

Photoelectron energy (eV) 

Fig. 2. Control of bandpass-filtered XUV emission with the waveform of monocycle light. Measured 

(A) and simulated (B) (17) photoelectron spectra versus CE phase, with the delay increased in steps of 

~11° ( p /16 rad). (C to E) Spectra measured at the CE phase setting closest to the values selected in 

Fig. 1A. The zero of the CE phase scale in (A) was set to yield the best agreement with the modeled 

spectra in (B). 

Photoelectron energy ( eV) 

XUV intensity (arb.u.) 

90 

80 

70 

60 

50 

40 

30 

A B 

−4 −2 0 

Delay (fs) 

2 4 

1.0 C 1.0 D 

0.8 

0.6 

0.4 

0.2 

τ x= 80 ±5 as 

-300 -200 -100 0 100 200 300 

Time (as) 

4 

3 

2 

1 

phase (rad) 

Photoelectron energy ( eV) 

90 

80 

70 

60 

50 

40 

30 

XUV spectral intensity (arb.u.) 

0.8 

0.6 

0.4 

0.2 

−4 −2 0 

Delay (fs) 

2 4 

φ″=(1.5 ± 0.2)×10 3 as 2 

40 50 60 70 80 90 100 110 

Photon energy (eV) 

Fig. 3. Sub-100-as XUV pulse retrieval. (A) Measured ATR spectrogram compiled from 126 energy 

spectra of photoelectrons launched by an XUV pulse with a bandwidth of ~28 eV (FWHM) and recorded 

at delay settings increased in steps of 80 as. Here, a positive delay corresponds to the XUV pulse 

arriving before the NIR pulse. The high flux of the XUV source allows this spectrogram to be recorded 

within ~30 min. (B) ATR spectrogram reconstructed after ~10 3 iterations of the FROG algorithm (17). 

(C) Retrieved temporal intensity profile and spectral phase of the XUV pulse. The intrinsic chirp of the 

XUV emission (Fig. 4B) is almost fully compensated by a 300-nm-thick Zr foil introduced into the XUV 

beam between the attosecond source and the ATR measurement. Arrows indicate the temporal FWHM of 

the XUV pulse. (D) XUV spectra evaluated from the measurement of the XUV-generated photoelectron 

spectrum in the absence of the NIR streaking field (blue dashed line) and from the ATR retrieval (blue 

solid line). The black dotted line indicates the retrieved spectral phase. 

0 

-3 

0.6 

0.5 

0.4 

0.3 

0.2 

0.1 

phase (rad) 

REPORTS 

www.sciencemag.org SCIENCE VOL 320 20 JUNE 2008 1615 

on June 24, 2008 


Downloaded from

REPORTS 

1616 

settings selected in Fig. 1A. Apart from a downshift 

by the ionization potential of neon (21.5 eV), 

they reveal close resemblance to the XUV spectra 

transmitted through the bandpass in Fig. 1A(v), 

(vi), and (iv), respectively. Figure 2C depicts the 

broadest filtered spectrum produced by a single 

recollision [full-line spectrum in Fig. 1A(v)]. 

Emission from the same recollision dominates 

also in the spectrum shown in Fig. 2E, with this 

spectrum red-shifted and (upon transmission through 

the bandpass) correspondingly narrowed, as predicted 

in Fig. 1A(iv). The two humps of the 

spectrum plotted in Fig. 2D are indicative of contributions 

from two recollisions, in accordance 

with the “solid-line” and “dotted-line” contributions 

to the emission spectrum in Fig. 1A(vi). 

Before measuring the XUV pulse, we optimized 

the generation process by “fine-tuning” 

the NIR laser peak intensity to achieve the 

broadest possible XUV spectrum transmitted 

through the bandpass in the range of CE phase 

settings where the contrast is maximized (50° – 80°, 

according to Fig. 1B) to generate a clean single 

pulse with the shortest possible duration. For the 

temporal characterization of the generated XUV 

supercontinuum, we shined the NIR field into the 

neon atoms ionized by the XUV pulse to implement 

the atomic transient recorder (ATR) 

technique introduced in (2, 3). The NIR field 

boosts or decreases the momenta of the photoelectrons, 

depending on their instant of release 

within the 2.4-fs period of the laser field, resulting 

in broadened and shifted (streaked) 

spectra of the electrons’ final energy distribution. 

Figure 3A is a plot of a series of streaked spectra 

recorded versus delay between the XUVand NIR 

pulse, which we refer to as an ATR or streaking 

spectrogram. It is practically equivalent to the 

spectrogram obtained by frequency-resolved 

optical gating (FROG) (20), with the oscillating 

NIR field constituting an attosecond phase gate 

in the present case (21). As a consequence, a 

FROG retrieval algorithm (22) allows complete 

determination of both the (gated) XUV pulse and 

the (gating) NIR laser field (17). The reconstructed 

ATR spectrogram is plotted in Fig. 3B 

and reveals excellent agreement with the measured 

one. 

The retrieved temporal intensity profile and 

phase of the XUV pulse are shown in Fig. 3C. 

The pulse duration of t x = 80 ± 5 as is close to 

its transform limit of 75 as, with a small positive 

chirp of f′′ = (1.5 ± 0.2) × 10 3 as 2 being responsible 

for the deviation. As a further consistency 

check of the attosecond pulse retrieval, we 

compared the measured NIR field–free electron 

spectrum (dashed blue line in Fig. 3D) with the 

electron spectrum calculated from the retrieved 

attosecond pulse (solid line in Fig. 3D). Given 

that the pulse retrieval draws on streaked spectra 

that are strongly distorted with respect to the 

field-free one, the degree of agreement between 

the retrieved and directly measured spectrum 

provides yet another conclusive testimony of the 

reliability of the retrieved data. 


The ATR retrieval algorithm indicates the 

presence of a satellite pulse accompanying the 

main attosecond pulse, containing ~1% of the energy 

of the main pulse. This amount of satellite 

is consistent with the depth of the experimentally 

observed modulation in the XUV spectrum 

(Fig. 3D). However, this result is inconsistent 

with our numerical modeling, which predicts a 

satellite energy content of some 6 to 7% for the 

optimum range of CE phase settings (Fig. 1B). 

From analysis of the streaked spectra recorded 

at the maximum of the NIR electric field (fig. 

S3), where the momentum of the electrons released 

by the main attosecond pulse and its 

satellite is shifted in opposite directions, we 

inferred a relative satellite energy of ~8%, which 

is in good agreement with the prediction of our 

modeling. As a consequence, the fringe visibility 

in the XUV spectrum is lower than was 

implied by the relative amplitude of the satellite 

pulse. The discrepancy may originate from a 

temporal jitter between the main pulse and the 

satellite pulse and/or from a different spatial 

amplitude distribution of the beams transporting 

the emission from the adjacent recollision 

events. The important lesson from these findings 

is that the fringe visibility in the XUV spectrum 

does not allow a reliable determination of the 

A 

80 as 

XUV pulse envelope 

B 

Photon energy (eV) 

120 

100 

80 

60 

78 

52 

26 

Electric field (GV/m) 

energy carried by the satellite(s) accompanying 

the main attosecond pulse. 

The laser waveform evaluated from the ATR 

measurement is preserved between the location 

of the generation and measurement. Figure 4A 

illustrates the evaluated NIR waveform with 

electric-field amplitude corresponding to a peak 

intensity of I 0 ~ (5.8 ± 0.5) × 10 14 W/cm 2 , as 

evaluated from the cut-off of our XUV spectra. 

The pulse duration [full width at half maximum 

(FWHM)] of 3.3 fs is in good agreement with 

the results of previous interferometric autocorrelation 

measurements (19), and the evaluated 

CE phase of ~50° is consistent with the optimum 

contrast according to our modeling (Fig. 1B). 

Accurate knowledge of the attosecond XUV 

pulse parameters, the temporal evolution of the 

generating NIR field, and the emergence of the 

former from a single recollision permit one to 

perform precision tests of models of light-electron 

interactions underlying the ionization and subsequent 

attosecond pulse generation processes. As 

an example, we calculated the intrinsic spectral 

chirp (i.e., the variation of the group delay versus 

frequency) carried by the attosecond XUV pulse 

during its emergence from the chirp measured by 

the ATR and the known dispersion of the 

components traversed by the pulse on its way 

Laser electric field E L(t) 

Recolliding electron 

trajectories 

0 

-2 0 2 

4 

Time (fs) 

Short Long 

trajectories 

0.5 1.0 1.5 2.0 2.5 3.0 

Recollision time (fs) 

Fig. 4. (A) Retrieved electric field of the NIR laser pulse used for generating and measuring the 

attosecond XUV pulse shown in Fig. 3. The duration of the pulse (FWHM of the cycle-averaged intensity 

profile) is tL = 3.3 fs, and the CE phase is evaluated as ϕ ~ 50°. The classical electron trajectories 

responsible for the emission of the filtered attosecond pulse are calculated with the plotted electric field 

and shown in the same panel. The color coding indicates the final return energy of the electrons. (B) 

Energy of the recolliding electron plus ionization potential (which is equal to the emitted XUV photon 

energy) versus recollision time evaluated from the classical trajectory analysis (solid green line), and 

emitted photon energy versus emission time (dashed purple line) inferred from the chirp of the measured 

attosecond pulse and the dispersion of the metal filter that the attosecond pulse passes through before the 

measurement. The basic idea for the graphical representation is borrowed from (29). Error bars indicate 

the uncertainty in the retrieved group delay. 

20 JUNE 2008 VOL 320 SCIENCE www.sciencemag.org 


60 

30 

Electron excursion (Å) 

on June 24, 2008 


Downloaded from

from the source to the measurement (fig. S2). The 

result (purple dashed line in Fig. 4B) is compared 

with the intrinsic attosecond chirp (green solid 

line in Fig. 4B) calculated from a classical 

trajectory analysis (23, 24). There is a notable 

discrepancy at the high-energy components of 

the wave packet, possibly because of quantum 

effects near cutoff. Nevertheless, the agreement 

with the attochirp resulting from short trajectories 

is stunning in the main part of the spectrum, 

where the S/N ratio is excellent. This agreement 

indicates the powerfulness of semiclassical modeling 

of strong-field interactions (25, 26) and the 

negligible role of long trajectories in contributing 

to the XUV radiation in the far field (27). 

In a similar way, the confinement of interaction 

between the ionizing field and the atom to a 

single wave cycle will permit accurate quantitative 

tests of theories of strong-field ionization. The 

sub-100-as XUV pulses emerging from the interaction 

with a flux greater than 10 11 photons/s— 

along with their monocycle NIR driver wave—will 

push the resolution limit of attosecond spectroscopy 

to the atomic unit of time (~24 as) and allow 

for the real-time observation of electron correla- 


tions, by means of streaking (6), tunneling (14), 

or absorption (28) spectroscopy. 


1. A. Baltuska et al., Nature 421, 611 (2003). 

2. R. Kienberger et al., Nature 427, 817 (2004). 


4. M. F. Kling et al., Science 312, 246 (2006). 

5. G. Sansone et al., Science 314, 443 (2006). 

6. A. L. Cavalieri et al., Nature 449, 1029 (2007). 


8. S. Svensson, B. Eriksson, N. Martensson, G. Wendin, 

U. J. Gelius, J. Electron Spectrosc. Relat. Phenom. 47, 

327 (1988). 

9. S. X. Hu, L. A. Collins, Phys. Rev. Lett. 96, 073004 (2006). 

10. J. Breidbach, L. S. Cederbaum, Phys. Rev. Lett. 94, 

033901 (2005). 

11. A. I. Kuleff, J. Breidbach, L. S. Cederbaum, J. Chem. Phys. 

123, 044111 (2005). 

12. F. Remacle, R. D. Levine, Proc. Natl. Acad. Sci. U.S.A. 

103, 6793 (2006). 

13. G. Yudin, M. Y. Ivanov, Phys. Rev. A 64, 035401 (2001). 

14. M. Uiberacker et al., Nature 446, 627 (2007). 

15. T. Pfeifer et al., Phys. Rev. Lett. 97, 163901 (2006). 

16. Y. Oishi et al., Opt. Express 14, 7230 (2006). 

17. See supporting material on Science Online. 

18. C. A. Haworth et al., Nature Phys. 3, 52 (2007). 

19. A. L. Cavalieri et al., New J. Phys. 9, 242 (2007). 

20. R. Trebino, D. J. Kane, J. Opt. Soc. Am. A 10, 1101 (1993). 

21. Y. Mairesse, F. Quéré, Phys. Rev. A 71, 011401(R) (2005). 

The Formation Conditions 

of Chondrules and Chondrites 

C. M. O’D. Alexander, 1 * J. N. Grossman, 2 D. S. Ebel, 3 F. J. Ciesla 1 

Chondrules, which are roughly millimeter-sized silicate-rich spherules, dominate the most 

primitive meteorites, the chondrites. They formed as molten droplets and, judging from their 

abundances in chondrites, are the products of one of the most energetic processes that operated in 

the early inner solar system. The conditions and mechanism of chondrule formation remain 

poorly understood. Here we show that the abundance of the volatile element sodium remained 

relatively constant during chondrule formation. Prevention of the evaporation of sodium requires 

that chondrules formed in regions with much higher solid densities than predicted by known 

nebular concentration mechanisms. These regions would probably have been self-gravitating. Our 

model explains many other chemical characteristics of chondrules and also implies that chondrule 

and planetesimal formation were linked. 

C 

hondrules make up ~20 to 80 volume % of 

most chondrites and formed at peak temper- 

atures of ~1700 to 2100 K (1). Chondrules 

in the different chondrite groups have distinct physical 

and chemical properties (2), as well as distinct 

age ranges (3), indicating that they formed in relatively 

local environments via a process that operated 

at least periodically between ~1 and 4 million years 

after the formation of the solar system. 

It was long thought that individual chondrules 

behaved as chemically closed systems during their 

formation, inheriting their compositions from their 

precursors (4, 5). However, for likely cooling rates 

of 10 to 1000 K/hour (1) and at the low pressures 

1 Department of Terrestrial Magnetism, Carnegie Institution of 

Washington, Washington, DC 20015, USA. 2 U.S. Geological 

Survey, Reston, VA 20192, USA. 3 American Museum of 

Natural History, New York, NY 10024, USA. 


alexande@dtm.ciw.edu 

(total pressure ≈ 10 −6 to 10 −3 bars) of the solar 

protoplanetary disk (nebula), experiments (6–8), 

natural analogs (9, 10), and theoretical calculations 

(11, 12) all show that there should be extensive 

evaporation of major and minor elements, in 

the order S > Na, K > Fe > Si > Mg. 

Elemental fractionations in chondrules are 

generally a function of volatility (4, 5). If evaporation 

in the nebula produced the alkali metal and 

Fe fractionations, the more volatile elements (such 

as S) should be entirely absent, which they are not. 

In addition, the fractionated elements should exhibit 

large and systematic isotopic fractionations, 

which they do not (13). 

Here we demonstrate that chondrules did indeed 

behave as essentially closed systems during 

melting, at least for elements with volatilities less 

than or equal to that of Na. We also propose a 

means of resolving the apparent conflict between 

this result and experimental and theoretical expec- 

22. D. J. Kane, G. Rodriguez, A. J. Taylor, T. S. Clement, 

J. Opt. Soc. Am. B 14, 935 (1997). 

23. P. Salières et al., Science 292, 902 (2001). 

24. V. S. Yakovlev, A. Scrinzi, Phys. Rev. Lett. 91, 153901 

(2003). 


26. K. J. Schafer, B. Yang, L. F. DiMauro, K. C. Kulander, 

Phys. Rev. Lett. 70, 1599 (1993). 

27. R. López-Martens et al., Phys. Rev. Lett. 94, 033001 

(2005). 

28. Z.-H. Loh et al., Phys. Rev. Lett. 98, 143601 (2007). 

29. K. Varjú et al., Laser Phys. 15, 888 (2005). 

30. Supported by the Max Planck Society and the 

Deutsche Forschungsgemeinschaft Cluster of Excellence: 

Munich Centre for Advanced Photonics (www.munichphotonics.de). 

E.G. acknowledges a Marie-Curie 

fellowship (MEIF-CT-2005-02440) and a Marie-Curie 

Reintegration grant (MERG-CT-2007-208643). A.L.A. 

is supported by the NSF Extreme Ultraviolet Engineering 

Research Center. R.K. acknowledges support from 

the Sofia Kovalevskaya award of the Alexander von 

Humboldt Foundation. 

Supporting Online Material 

www.sciencemag.org/cgi/content/320/5883/1614/DC1 

SOM Text 

Figs. S1 to S4 

References 

17 March 2008; accepted 27 May 2008 

10.1126/science.1157846 

tations that chondrules should have suffered considerable 

evaporation during formation. Our 

conclusions have implications for mechanisms of 

dust concentration in the solar nebula, for chondrule 

formation, and for planetesimal formation. 

Chondrules are dominated by olivine 

[(Mg,Fe)2SiO4], pyroxene [(Mg,Fe,Ca)SiO3], 

Fe-Ni metal, and quenched silicate melt (glass). 

Many of the more volatile elements (such as Na) 

can diffuse rapidly, particularly in melts and 

glasses. Therefore, it is possible that volatiles were 

completely lost when chondrules melted, and 

reentered the chondrules during cooling or even 

after solidification. However, Na clinopyroxene/ 

glass ratios show that the Na contents of the final 

chondrule melts (now glass) had approximately 

their observed, relatively high abundances at 

temperatures of ~1600 to 1200 K (14–16). 

Calculations suggest that chondrule melts 

could have been stabilized in the nebula by substantially 

enriching solids (chondrule precursors 

or other dust) relative to gas (11, 12). This also 

substantially increases the condensation temperatures 

of even highly volatile elements such as S 

(11, 12). Even in solid-enriched systems, there is an 

initial phase of evaporation when a chondrule melts, 

but subsequent chondrule/gas re-equilibration would 

erase any isotopic fractionations (12). If the enrichment 

of solids is high, little evaporation may be 

needed to reach chondrule/gas equilibrium, and 

the behavior of volatile elements during cooling 

would resemble closed-system behavior. However, 

even at a high total pressure of 10 −3 bars, with a 

solids enrichment of 1000 relative to the solar composition, 

all the Na would evaporate at near-liquidus 

temperatures, and substantial recondensation only 

begins well below 1600 K (11). Locally enriching 

chondrule-sized or smaller solids by 1000 times 

www.sciencemag.org SCIENCE VOL 320 20 JUNE 2008 1617 


REPORTS 

on June 24, 2008 


Downloaded from






Intense 1.5-cycle near infrared laser waveforms and 

their use for the generation of ultra-broadband 

soft-x-ray harmonic continua 

A L Cavalieri 1,3 , E Goulielmakis 1 , B Horvath 1 , W Helml 1 , 

M Schultze 1 , M Fieß 1 , V Pervak 2 , L Veisz 1 , V S Yakovlev 2 , 

M Uiberacker 2 , A Apolonski 2 , F Krausz 1 and R Kienberger 1,3 

1 Max-Planck-Institut für Quantenoptik, Hans-Kopfermann-Str. 1, 


2 Department für Physik, Ludwig Maximilians Universität, Am Coulombwall 1, 


E-mail: adrian.cavalieri@mpq.mpg.de or reinhard.kienberger@mpq.mpg.de 


Received 20 June 2007 

Published 31 July 2007 


doi:10.1088/1367-2630/9/7/242 

Abstract. We demonstrate sub-millijoule-energy, sub-4 fs-duration nearinfrared 

laser pulses with a controlled waveform comprised of approximately 1.5 

optical cycles within the full-width at half-maximum (FWHM) of their temporal 

intensity profile. We further demonstrate the utility of these pulses for producing 

high-order harmonic continua of unprecedented bandwidth at photon energies 

around 100 eV. Ultra-broadband coherent continua extending from 90 eV to more 

than 130 eV with smooth spectral intensity distributions that exhibit dramatic, 

never-before-observed sensitivity to the carrier-envelope offset (CEO) phase of 

the driver laser pulse were generated. These results suggest the feasibility of 

sub-100-attosecond XUV pulse generation for attosecond spectroscopy in the 

100 eV range, and of a simple yet highly sensitive on-line CEO phase detector 

with sub-50-ms response time. 





Vol 449 | 25 October 2007 | doi:10.1038/nature06229 


LETTERS 

Attosecond spectroscopy in condensed matter 

A. L. Cavalieri 1 , N. Müller 2 , Th. Uphues 1,2 , V. S. Yakovlev 3 , A. Baltusˇka 1,4 , B. Horvath 1 , B. Schmidt 5 , L. Blümel 5 , 

R. Holzwarth 5 , S. Hendel 2 , M. Drescher 6 , U. Kleineberg 3 , P. M. Echenique 7 , R. Kienberger 1 , F. Krausz 1,3 

& U. Heinzmann 2 

Comprehensive knowledge of the dynamic behaviour of electrons 

in condensed-matter systems is pertinent to the development of 

many modern technologies, such as semiconductor and molecular 

electronics, optoelectronics, information processing and photovoltaics. 

Yet it remains challenging to probe electronic processes, 

many of which take place in the attosecond (1 as 5 10 218 s) regime. 

In contrast, atomic motion occurs on the femtosecond (1 fs 5 

10 215 s) timescale and has been mapped in solids in real time 1,2 

using femtosecond X-ray sources 3 . Here we extend the attosecond 

techniques 4,5 previously used to study isolated atoms in the gas 

phase to observe electron motion in condensed-matter systems 

and on surfaces in real time. We demonstrate our ability to obtain 

direct time-domain access to charge dynamics with attosecond 

resolution by probing photoelectron emission from single-crystal 

tungsten. Our data reveal a delay of approximately 100 attoseconds 

between the emission of photoelectrons that originate 

from localized core states of the metal, and those that are freed 

from delocalized conduction-band states. These results illustrate 

that attosecond metrology constitutes a powerful tool for exploring 

not only gas-phase systems, but also fundamental electronic 

processes occurring on the attosecond timescale in condensedmatter 

systems and on surfaces. 

Photoemission spectroscopy is based on the photoelectric effect, 

first explained by Einstein more than 100 years ago 6 . According to 

Einstein’s law, photoelectrons ejected from a metal surface by light 

will have a kinetic energy that depends on the incident photon energy 

and the electron’s original bound state energy. Photoelectron spectra 

will thus provide information about the electronic structure of the 

metal if well-characterized light sources are used 7 . Indeed, experiments 

using a broad range of photon energies have now been performed 

to determine the steady-state electronic properties of many 

bulk materials, thin films, and surfaces. The photoemission process 

itself involves three steps: excitation, transport, and ultimately escape 

of the photoelectron through the surface 8 . Here, in a proof-ofprinciple 

experiment, we combine this spectroscopy with attosecond 

temporal resolution to obtain time-domain insight into the electron 

transport stage of the photoemission process. The measurements 

represent (to our knowledge) the first direct attosecond timeresolved 

observation of electron transport in a condensed-matter 

system, and we expect that they will trigger other experimental 

research into the dynamics of processes that have attracted interest 

in solid-state and surface science. Such processes include charge 

transfer 9,10 , charge screening 11 , image charge creation and decay 12 , 

electron–electron scattering 13 , and collective electronic motion 14 . 

Time-resolved photoemission spectroscopy was originally implemented 

in the picosecond (1 ps 5 10 212 s) to femtosecond regime, 

using first visible 15–17 and then extreme ultraviolet (XUV) 18,19 radiation. 

These experiments utilize one light pulse to trigger the 

dynamics, followed by a second light pulse to induce photoemission 

and thereby probe the transient state. Experiments using the laserassisted 

photoelectric effect have been carried out 20–22 ; but the XUV 

photoemission lasted over several wave cycles of the coincident nearinfrared 

(NIR) light, limiting the time resolution to .10 fs. To overcome 

this limitation, we use single sub-femtosecond XUV pulses 4,5 

for pumping, and coincident NIR waveform-controlled few-cycle 

laser pulses 23 as a probe 5 . The XUV pulse triggers the photoemission 

process, with only those photoelectron wave packets initiated from 

the uppermost atomic layers escaping without inelastic collision. 

Delayable 

XUV mirror 

To preparation 

chamber 

Tungsten 

crystal 

Zirconium foil,150 nm, 

on pellicle 

1 Max-Planck-Institut für Quantenoptik, Hans-Kopfermann-Str. 1, D-85748 Garching, Germany. 2 Fakultät für Physik, Universität Bielefeld, D-33615 Bielefeld, Germany. 3 Department 

für Physik, Ludwig-Maximilians-Universität, Am Coulombwall 1, D-85748 Garching, Germany. 4 Institut für Photonik, Technische Universität Wien, Gußhausstr. 27, A-1040 Wien, 

Austria. 5 Menlo Systems GmbH, Am Klopferspitz 19, D-82152 Martinsried, Germany. 6 Institut für Experimentalphysik, Universität Hamburg, Luruper Chaussee 149, D-22761 

Hamburg, Germany. 7 Dpto. Fisica de Materiales UPV/EHU, Centro Mixto CSIC-UPV/EHU and Donostia International Physics Center (DPIC), Paseo Manual de Lardizabal 4, 20018 

San Sebastian, Spain. 

Silver 

mirror 



e – 

TOF 

Neon-filled 

tube 

Figure 1 | Experimental set-up. Waveform-controlled, ,5-fs, 750-nm, 400-mJ 

laser pulses are focused with a mirror of 500-mm focal length into a ,2-mmdiameter 

tube filled with neon to generate XUV radiation by high-harmonic 

generation. The collinear XUV and NIR beams co-propagate towards a cored 

two-part mirror in the measurement chamber that is maintained under 

ultrahigh vacuum in a 1-m-length differential vacuum pumping stage, until 

the beams are separated by a pellicle/zirconium foil assembly. The zirconium 

foil transmits the XUV but blocks the NIR. The XUV radiation, indicated by 

the blue beam, is incident on a 6 eV (full-width at half-maximum FWHM) 

broad multilayer band-pass reflector centred at ,91 eV, which is mounted on 

a piezo-electric delay stage. With a proper XUV spectrum, the multilayer 

mirror reflects and focuses ,300-as (FWHM) XUV pulses. The NIR pulse, 

indicated by the violet beam, is reflected by a stationary (silver) outer annular 

mirror confocal with the inner mirror (f 5 12.5 cm). Both pulses are focused 

onto the (110) surface of a tungsten single crystal that is mounted on a 

manipulator to control the angle of incidence. The manipulator is also used to 

retract the crystal into a preparatory chamber for cleaning. Resultant XUVinduced 

photoemission, which is detected by the time-of-flight spectrometer 

(TOF), is streaked by the coincident NIR laser pulse. 

1029


LETTERS NATURE |Vol 449 | 25 October 2007 

Photoexcited electron wave packets propagate through the material 

in upper conduction bands, ultimately leaving the surface with an 

average kinetic energy determined by the XUV photon energy, the 

initial binding energy, and the material work function. An attosecond 

transient recorder (ATR), previously developed and used in gasphase 

experiments 5 , is used to observe the emitted photoelectron 

wave packet. In this scheme, the photoelectron momentum is further 

influenced by the electric field of a coincident few-cycle NIR laser 

pulse, giving rise to a ‘streaked’ final momentum distribution 24,25 . 

An ATR spectrogram is compiled by measuring a series of streaked 

photoelectron spectra with a time-of-flight detector, recorded as a 

function of time delay between the XUV pump and NIR streaking 

field. Important characteristics of the emitted electron wave packets, 

including their duration and frequency sweep, or ‘chirp’, can be determined 

from the spectra 24,25 . If measured for two or more different types 

of electrons, the complete ATR spectrograms can also yield relative 

timing information about the arrival of the wave packets on the surface, 

because the streaking effect is negligible until the electrons emerge 

from the surface (see Methods). The resolution of the ATR depends on 

the duration of the XUV excitation, the gradient of the streaking NIR 

field, and the signal-to-noise ratio in the photoelectron spectra. 

In comparison to experiments performed on isolated atoms, ATR 

measurements in condensed-matter systems are more complicated 

because the photoelectron wave packets can be released from energy 

a 

Kinetic energy (eV) Kinetic energy (eV) 

100 

90 

80 

70 

60 

50 

40 

30 

8 7 6 5 4 3 2 1 0 

100 

90 

80 

70 

60 

50 

40 

Raw spectrum (1) 

ATI subtracted and smoothed (1) 

Streaked spectrum (2) 

Conduction-band 

photoemission 

4f-states 

photoemission 

c d 

Measured streaked spectrum 

Reconstructed spectrum 

30 

8 7 6 5 4 3 2 1 0 

Intensity (a.u.) 

Figure 2 | Attosecond time-resolved photoemission spectra. 

a, Photoelectron spectra collected at two different relative delays. The 

spectrum recorded at the delay indicated by the white dashed line labelled 

‘(1)’ in b is far from the zero of delay as defined by the overlap of the 

maximums of the NIR and XUV pulse envelopes. This spectrum shows 

pronounced peaks corresponding to the 4f-state and conduction-band 

photoemission. The blue line shows the raw spectrum as recorded by the 

time-of-flight detector, and the red line shows the corresponding spectrum 

after subtraction of NIR-induced ATI background and numerical 

smoothing. The 4f photoemission is peaked near 56 eV. The conductionband 

photoemission is peaked near 83 eV, owing to a high density of d-band 

conducting states just below the Fermi energy. Ef denotes the kinetic energy 

of a photoelectron excited from the Fermi energy level. The other spectrum 

in a (displayed only after ATI subtraction and smoothing) was recorded near 

zero delay, as indicated by the white dashed line labelled ‘(2)’. At this delay, 

1030 

E f 

bands containing many distinct states rather than from a single, 

isolated energy level. Unoccupied conduction-band states just above 

the Fermi energy (defined by the highest occupied energy level in the 

absence of thermal excitation) might become populated by singlephoton 

absorption of the leading edge of the NIR probe field prior to 

XUV photoemission, which is an unwanted complication. In contrast 

to conduction-band states, the localized 4f core states of tungsten 

are deeply bound and fully populated. Therefore, these states are 

unsusceptible to this potential influence of the streaking field, and 

constitute an ideal test case for proof of the extension of attosecond 

metrology to solids. 

As a further challenge, above-threshold ionization (ATI) by the 

streaking field can, in condensed matter, generate energetic photoelectrons, 

obscuring detection of XUV-induced photoelectrons. ATI 

is favoured by the low work function of metals (as compared to the 

relatively high ionization potential of isolated atoms), which limits 

the intensity of the applied streaking field to levels far below those 

that can be used in gas-phase experiments. 

As indicated by the diagram of the experimental set-up in Fig. 1, 

streaked photoemission spectra from a tungsten (110) crystal surface 

were recorded by collecting electrons within a narrow cone aligned 

perpendicularly to the surface. The relative delay between the XUV 

pulse and the NIR waveform-controlled streaking field was varied in 

300-as steps in a sequence chosen to minimize systematic error, with 

–6 –4 –2 0 


b 

100 

90 

80 

70 

60 

50 

40 

30 

100 

90 

80 

70 

60 

50 

40 

30 

(1) 

–6 


(2) 

(2) 

–4 –2 0 

Relative delay (fs) 

2 4 

2 4 

the XUV pulse peak coincides with the NIR field maximum on our target. 

Consequently, the NIR vector potential crosses zero, giving rise to the 

strongest streaking and a clear broadening of the photoemission peaks is 

observed. b, The full ATI subtracted spectrogram for both the 4f-states and 

conduction-band photoemission. The streaking waveform (vector potential) 

is evident in both spectrograms, proving the extension of attosecond 

metrology to condensed-matter systems. We can expect that the energy 

modulation of the conduction-band peak should be a little larger than that of 

the 4f peak owing to their initial photoelectron kinetic energy. The 

amplitude of the spectral shift has a square-root dependence on initial 

kinetic energy (KE) 24 , yielding an expected ratio in spectral shift of 

p ffiffiffiffiffiffiffiffiffiffiffiffiffi. 

p ffiffiffiffiffiffiffiffiffi 

KEcond KE4f

spectra integrated for 60 s at each delay. (For a brief description of the 

experiment, see Methods Summary; full details regarding set-up, 

measurements and data analysis are provided as Supplementary 

Information.) Characteristic spectra obtained with our system are 

shown in Fig. 2a, indicating that emission from the conduction band 

occurs at a kinetic energy of ,83 eV while emission from the localized 

4f states occurs at ,56 eV. At kinetic energies significantly 

below the 4f peak, the measured spectrum is due to NIR-induced 

ATI photoelectrons and XUV-generated photoelectrons that have 

undergone inelastic scattering. 

The two distinct background components were distinguished by 

recording an additional photoelectron spectrum without the NIR 

streaking field. The ATI component was subsequently subtracted 

from the measured data (see Supplementary Information). This subtraction 

is illustrated for a fixed delay in Fig. 2a, and was performed at 

each of the delay steps, resulting in the full spectrogram presented in 

Fig. 2b. Here, a positive relative delay corresponds to the XUV pulse 

arriving earlier with respect to the streaking field at the surface. Both 

the 4f and conduction-band photoemission exhibit a pronounced 

periodic upshift and downshift in energy as a function of relative 

delay and, as in previous gas-phase experiments, the spectrogram 

reveals the waveform (vector potential) of the streaking field 5,26,27 . 

Our ability to resolve the field oscillation indicates that the photoemission 

from the 4f core states and from the conduction band is subfemtosecond 

in duration, and proves that attosecond metrology has 

been successfully extended to condensed-matter systems. 

Further examination reveals that the 4f spectrogram is shifted 

along the delay coordinate with respect to the conduction-band 

spectrogram. This effect is readily apparent upon inspection of the 

smoothed spectrograms that are obtained by interpolation of the 

measured data and shown in Fig. 3a. We quantify the temporal shift 

in the measured data by evaluating, for each delay step, the centre-ofmass 

(COM) of the spectral regions spanning the 4f and conductionband 

peaks that cover the energy intervals 47–66 eV and 66–110 eV, 

respectively. Characterizing the periodic motion of the peaks through 

their COM requires no assumptions or fitting parameters, yet yields 

timing information that is invariant to fluctuations in the instantaneous 

laser parameters. The approach is also relatively insensitive to 

inelastic scattered background photoelectrons, which could not be 

subtracted from our measurements. As a result, the COM accurately 

describes the streaking-induced time-dependence of the energy shift 

of the 4f and conduction-band peaks, as shown in Fig. 3b. 

By comparing the COM trajectories of the 4f and conduction band 

at the seven zero-crossings of the vector potential, we obtain seven 

independent measurements of their relative timing. This yields a 

temporal shift of Dt 5 110 6 70 as between the ATR spectrograms 

of the conduction-band and 4f photoelectrons. (The error estimate 

results from a straightforward extrapolation of the error in calculating 

the COM; see Supplementary Information.) This shift or delay 

was observed in different independent measurements made at different 

locations on the tungsten sample, with the results corroborating 

the above value of Dt to within the measurement error. We note 

that the rather large error associated with our Dt value could be most 

effectively reduced in future measurements by using higher XUV 

photon energies and fluxes. 

The shift between the two spectrograms indicates that, on average, 

photoelectrons originating from the localized 4f states emerge from 

the tungsten surface approximately 100 as later than those originating 

from the delocalized conduction band—even though the 

photoemission process for both types of electrons is initiated simultaneously 

by the same XUV pulse. The delay effect thus occurs during 

transport of the excited photoelectrons to the surface, illustrating 

that our technique provides a means to directly observe features of 

electron wave packet propagation towards the surface with attosecond 

precision. 

By adapting a quantum mechanical model used in previous gasphase 

streaking experiments 28 , we are able to reconstruct the measured 


NATURE | Vol 449 |25 October 2007 LETTERS 

spectra and spectrograms. The modelling of the streaking experiment 

requires some assumptions, leaving several parameters (such as duration 

of the electron wave packets, their chirp, and their emission time) 

for optimization. Figure 2c and d shows the reconstructions that best 

agree with experiment. These were obtained for wave packets with a 

duration of ,300 as (full-width at half-maximum, FWHM) and 

assuming a delay of ,100 as between the emission times of the electron 

wave packets, which supports the conclusions drawn from the COM 

analysis. 

Our measurements also indicate that electron wave packets 

launched from both the localized 4f and delocalized conduction-band 

states are nearly undistorted on propagation to the surface. To explain 

the observed delay, we consider the group velocities for the two different 

photoelectron wave packets travelling in the solid. The crucial point 

is that after absorption of an XUV photon, the electron is excited 

into an upper conduction band region that depends on the electron’s 

Energy shift (eV) 4f-states kinetic energy (eV) 

65 

63 

61 

59 

57 

55 

1 

0.5 

0 

–0.5 

–6 

–1 –6 


a 

b 

4f states 

Cond. band 

∆t = 110 ± 70 as 

–4 –2 0 2 4 



∆t 

–4 –2 0 2 4 


91 

89 

87 

85 

83 

Cond.-band kinetic 

energy (eV) 

Figure 3 | Evidence of delayed photoemission. a, The 4f and conductionband 

spectrograms, following cubic-spline interpolation of the measured data 

(but without background subtraction). The spectral region between ,65 eV 

and ,83 eV has been omitted to more easily compare the edges of the 4f and 

conduction-band peaks. A small shift in the relative delay is evident, as 

indicated by the white dashed lines through the fringes, and can be seen at 

each fringe maximum and minimum. Quantification of the shift of the 4f with 

respect to the conduction-band spectrogram is made by COM analysis, and is 

summarized in b. The energy intervals, within which the COMs were 

calculated, are 47–66 eV for the 4f photoemission peak and 66–110 eV for the 

conduction-band photoemission peak. Vertical error bars (61 s.d.) are 

calculated from noise in the measured spectra (see Supplementary 

Information for details). For ease of visual comparison, the COM energy shift 

of the 4f spectral region was scaled by a factor of 2.5, to offset the stabilizing 

effect of the background plateau underneath the 4f peak (see Supplementary 

Information), in order to illuminate the ,100 as delay in emission. Rescaling 

these COM data points along the energy axis cannot influence the measured 

delay. The COM data points were fitted with a damped sinusoid, which 

corresponds to the NIR streaking field, to guide the eye. 

1031


LETTERS NATURE |Vol 449 | 25 October 2007 

initial binding energy and the XUV photon energy. If the relationship 

between momentum and energy of such an electron were that of a free 

electron, the ratio of velocities of the conduction-band and 4f electrons 

travelling with energies of 85 eV and 58 eV, respectively, would be a 

factor of ,1.2 (energy is defined with respect to the Fermi energy inside 

the material). However, elastic interactions with atoms in the crystal 

lattice, which give rise to electronic band structure, modify the 

momentum–energy relationship. This implies that in tungsten for 

XUV photon energies of ,91 eV, the mean velocity of the conductionband 

photoelectrons is approximately twice that of the 4f photoelectrons 

(see Supplementary Information). We also note that due to 

their longer inelastic mean free paths 29 , the slow 4f photoelectrons 

originate, on average, from ,1 A ˚ deeper in the tungsten crystal than 

the fast conduction-band photoelectrons. On the basis of these 

considerations, we estimated the absolute delay between the initial 

excitation of a photoelectron and its escape through the surface to 

be ,60 as and ,150 as for the conduction and 4f photoelectrons, 

respectively. These values suggest a relative delay between photoelectron 

emissions from the surface of ,90 as, which is in good 

agreement with the observed delay in the emission of the 4f and 

conduction-band photoelectrons. 

In summary, our observations demonstrate the successful 

extension of attosecond metrology to condensed-matter systems. 

Although the current experimental apparatus provides access to only 

the relative group delay in electron wave packet propagation, future 

measurements of absolute emission delays may be feasible using the 

same methods, for example by direct comparison with gas-phase 

ATR data. At this point, attosecond photoemission spectroscopy 

presents a clear path toward ultimately uncovering the intermediate 

processes leading to ejection of a photoelectron. This information 

might shed new light on data previously obtained with conventional 

time-integral photoemission spectroscopy and allow for accurate 

description of charge dynamics on the electronic timescale in both 

condensed matter and on surfaces. 

METHODS SUMMARY 

A 1-kHz-repetition-rate, waveform-controlled, few-cycle, ,5-fs, 400-mJ, 750-nm 

Ti:sapphire laser system is the front-end of our apparatus and is used in combination 

with proper spectral filtering to efficiently generate isolated attosecond 

pulses of XUV radiation by high-harmonic generation. As shown in Fig. 1, the 

XUV and NIR pulses co-propagate towards a two-part focusing mirror at near 

normal incidence. The inner component is a Mo/Si multilayer mirror and reflects 

the XUV radiation over a bandwidth of ,6 eV (FWHM) centred at ,91 eV, 

supporting 300-as transform-limited pulses 30 . The XUV mirror is mounted on a 

translation stage, providing a precise delay between the XUV pump and the NIR 

streaking pulse. The temporal resolution that can be achieved in pump-probe 

experiments using these pulses and this apparatus is expected to be a small 

fraction of the pulse half-width because XUV pulses generated with similar 

spectra and multilayer optics have previously been fully characterized and 

observed to be gaussian 24 . 

The tungsten surface must be sufficiently free of contamination to minimize 

photoelectron scattering on emission. Therefore, the measurement chamber is 

maintained under ultrahigh vacuum conditions with typical background pressures 

of , 10 29 mbar, which suppresses the accumulation of contaminates on 

the crystal surface to a level permitting a full ATR spectrogram to be recorded 

without interruption. 

In our application of the ATR, detection of electrons occurs in the direction 

normal to the tungsten crystal surface, and the NIR streaking field is incident on 

the tungsten (110) crystal surface near Brewster’s angle (,75u). For this angle of 

incidence, owing to the refractive index of tungsten, photoelectron wave packets 

are not efficiently accelerated in the direction of observation until they emerge 

from the surface. Even though the streaking field penetrates the tungsten crystal, 

inside the material the electric field component along the surface normal is 

weaker by a factor of approximately 16, allowing us to neglect streaking effects 

until the electron wave packets emerge from the surface. The absence of effective 

streaking until the photoelectron emerges from the surface is generally the case 

for solids, allowing us to time processes occurring within the material. 

Additional, detailed description of the experimental apparatus, measurement 

technique, and data analysis is provided in Supplementary Information. 

1032 


Received 20 June; accepted 3 September 2007. 

1. Reis, D. A. & Lindenberg, A. M. in Light Scattering in Solids IX (eds Cardona, M. & 

Merlin, R.) 371–422 (Topics in Applied Physics 108, Springer, Berlin, 2007). 

2. Fritz, D. M. et al. Ultrafast bond softening in bismuth: Mapping a solid’s 

interatomic potential with X-rays. Science 315, 633–636 (2007). 

3. Pfeifer, T., Spielmann, C. & Gerber, G. Femto-second X-ray science. Rep. Prog. 

Phys. 69, 443–505 (2006). 



6. Einstein, A. Über einen die Erzeugung und Verwandlung des Lichts betreffenden 

heuristischen Gesichtspunkt. Ann. Phys. 17, 132–148 (1905). 

7. Siegbahn, K. Electron-spectroscopy — outlook. J. Electron Spectrosc. Relat. Phenom. 

5, 3–97 (1974). 

8. Berglund, C. N. & Spicer, W. E. Photoemission studies of copper and silver: theory. 

Phys. Rev. 136, A1030–A1044 (1964). 

9. Brühwiler, P. A., Karis, O. & Martensson, N. Charge-transfer dynamics studied 

using resonant core spectroscopies. Rev. Mod. Phys. 74, 703–740 (2002). 

10. Föhlisch, A. et al. Direct observation of electron dynamics in the attosecond 

domain. Nature 436, 373–376 (2005). 

11. Borisov, A., Sánchez-Portal, D., Díez-Muino, R. & Echenique, P. M. Dimensionality 

effects in time-dependent screening. Chem. Phys. Lett. 387, 132–137 (2004). 

12. Huber, R. et al. How many-particle interactions develop after ultrafast excitation 

of an electron-hole plasma. Nature 414, 286–289 (2001). 

13. Haight, R. Electron dynamics at surfaces. Surf. Sci. Rep. 21, 277–325 (1995). 

14. Cavalleri, A. et al. Tracking the motion of charges in a terahertz light field by 

femtosecond X-ray diffraction. Nature 442, 664–666 (2006). 

15. Yen, R. et al. Picosecond laser interaction with metallic zirconium. Appl. Phys. Lett. 

40, 185–187 (1982). 

16. Höfer, U. et al. Time-resolved coherent photoelectron spectroscopy of quantized 

electronic states on metal surfaces. Science 277, 1480–1482 (1997). 

17. Petek, H. & Ogawa, S. Femtosecond time-resolved two-photon photoemission 

studies of electron dynamics in metals. Prog. Surf. Sci. 56, 239–310 (1997). 

18. Haight, R. & Peale, D. R. Tunable photoemission with harmonics of subpicosecond 

lasers. Rev. Sci. Instrum. 65, 1853–1857 (1994). 

19. Siffalovic, P. et al. Laser-based apparatus for extended ultraviolet femtosecond 

time-resolved photoemission spectroscopy. Rev. Sci. Instrum. 72, 30–35 (2001). 

20. Schins, J. M. et al. Observation of laser-assisted Auger decay in argon. Phys. Rev. 

Lett. 73, 2180–2183 (1994). 

21. Glover, T. E., Schoenlein, R. W., Chin, A. H. & Shank, C. V. Observation of laser 

assisted photoelectric effect and femtosecond high order harmonic radiation. 

Phys. Rev. Lett. 73, 2180–2183 (1994). 

22. Miaja-Avila, L. et al. Laser-assisted photoelectric effect from surfaces. Phys. Rev. 

Lett. 97, 113604 (2006). 

23. Baltuska, A. et al. Attosecond control of electronic processes by intense light 

fields. Nature 422, 611–615 (2003). 

24. Quéré, F., Mairesse, Y. & Itatani, J. Temporal characterization of attosecond XUV 

fields. J. Mod. Opt. 52, 339–360 (2005). 

25. Yakovlev, V., Bammer, F. & Scrinzi, A. Attosecond streaking measurements. 

J. Mod. Opt. 52, 395–410 (2005). 

26. Goulielmakis, E. et al. Direct measurement of light waves. Science 305, 1267–1269 

(2004). 

27. Sansone, G. et al. Isolated single-cycle attosecond pulses. Science 314, 443–446 

(2006). 

28. Kitzler, M., Milosevic, N., Scrinzi, A., Krausz, F. & Brabec, T. Quantum theory of 

attosecond XUV pulse measurement by laser dressed photoionization. Phys. Rev. 

Lett. 88, 173904 (2002). 

29. Tanuma, S., Powell, C. J. & Penn, D. R. Calculations of electron inelastic mean free 

paths. 2. Data for 27 elements over the 50–2000-eV range. Surf. Interface Anal. 17, 

911–926 (1991). 

30. Wonisch, A. et al. Design, fabrication, and analysis of chirped multilayer mirrors 

for reflection of extreme-ultraviolet attosecond pulses. Appl. Opt. 45, 4147–4156 

(2006). 

Supplementary Information is linked to the online version of the paper at 

www.nature.com/nature. 

Acknowledgements We thank W. Hachmann for expeditious preparation of the 

XUV multilayer optical substrate. We acknowledge partial financial support by the 

Deutsche Forschungsgemeinschaft through the DFG Cluster of Excellence Munich 

Centre for Advanced Photonics, and through the SFB 613, and by the Volkswagen 

Stiftung Germany, and by the EURYI scheme award. P.M.E. acknowledges support 

from the Basque and Spanish Governments. R.K. acknowledges a fellowship from 

the Austrian Academy of Sciences and additional support from the Sofja 

Kovalevskaja Award of the Alexander von Humboldt Foundation. The apparatus to 

generate attosecond pulses was constructed at Technische Universität Wien, 

thanks to the support of the FWF. 

Author Information Reprints and permissions information is available at 

www.nature.com/reprints. Correspondence and requests for materials should be 

addressed to A.L.C. (adrian.cavalieri@mpq.mpg.de) or F.K. (krausz@lmu.de) or 

U.H. (uheinzm@physik.uni-bielefeld.de). 



Attosecond nanoplasmonic-field 

microscope 

MARK I. STOCKMAN1,2 *, MATTHIAS F. KLING2 , ULF KLEINEBERG3 AND FERENC KRAUSZ2,3 * 

1Department of Physics and Astronomy, Georgia State University, Atlanta, Georgia 30303, USA 

2Max-Planck-Institut für Quantenoptik, Hans-Kopfermann-Straße 1, D-85748 Garching, Germany 

3Ludwig-Maximilians-Universität München, Department für Physik, Am Coulombwall 1, D-85748 Garching, Germany 

*e-mail: mstockman@gsu.edu; ferenc.krausz@mpq.mpg.de 

Published online: 3 September 2007; doi:10.1038/nphoton.2007.169 

Nanoplasmonics deals with collective electronic dynamics on the surface of metal nanostructures, which arises as a result of 

excitations called surface plasmons. This field, which has recently undergone rapid growth, could benefit applications such as 

computing and information storage on the nanoscale, the ultrasensitive detection and spectroscopy of physical, chemical and 

biological nanosized objects, and the development of optoelectronic devices. Because of their broad spectral bandwidth, surface 

plasmons undergo ultrafast dynamics with timescales as short as a few hundred attoseconds. So far, the spatiotemporal dynamics 

of optical fields localized on the nanoscale has been hidden from direct access in the real space and time domain. Here, we 

propose an approach that will, for the first time, provide direct, non-invasive access to the nanoplasmonic collective dynamics, 

with nanometre-scale spatial resolution and temporal resolution on the order of 100 attoseconds. The method, which combines 

photoelectron emission microscopy and attosecond streaking spectroscopy, offers a valuable way of probing nanolocalized optical 

fields that will be interesting both from a fundamental point of view and in light of the existing and potential applications 

of nanoplasmonics. 

Recently, significant attention has been devoted to the development 

of attosecond science and technology, in particular regarding pulse 

generation, physical-system excitation, detection, spectroscopy and 

electron-motion control on the attosecond timescale 1–13 . In 

nanoscience, which is another rapidly evolving field, an important 

issue is the study and exploitation of effects that are both ultrafast 

and localized on the nanoscale. The localization length of surface 

plasmon eigenmodes in nanoplasmonic systems is determined by 

the size of the constituent nanoparticles and can be on the order 

of several nanometres 14 . Furthermore, the relaxation rate of the 

surface plasmon polarization is in the 10–100 fs range across the 

plasmonic spectrum, allowing coherent control of nanoscale 

energy localization with femtosecond laser light 15–25 . Importantly, 

collective motion in nanoplasmonic systems unfolds on much 

shorter attosecond timescales, as defined by the inverse spectral 

bandwidth of the plasmonic resonant region. In this paper we 

propose an approach that enables direct measurement of the 

spatiotemporal dynamics of nanolocalized optical fields with 

100-as temporal resolution and nanometre-scale spatial resolution. 

Nanoplasmonic fields could be used in place of electrons in 

ultrafast computation and information storage on the nanoscale, 

where the fields act as the medium that carries, processes and 

stores information. 

The proposed attosecond nanoplasmonic-field microscope 

combines two modern techniques: photoelectron emission 

microscopy and attosecond streaking spectroscopy 26 . A plasmonic 

nanostructure is excited by an intense, waveform-controlled field 

in the optical spectral range that drives collective electron 

oscillations (the quantum of which is called a surface plasmon). 

This generates optical fields localized on the nanometre scale, 

which we will refer to as nanolocalized optical fields. An 

ARTICLES 

attosecond extreme ultraviolet (XUV) pulse, which is produced 

from and synchronized with the driving optical pulse, is then 

sent to the system. This XUV pulse produces photoelectrons that, 

owing to their large energy and short emission time (determined 

by the attosecond-pulse duration), escape from the nanometresized 

regions of local electric fields enhanced by plasmon 

resonances within a fraction of the oscillation period of the 

driven plasmonic field. The ultrashort escape time implies a final 

energy change of the emitted photoelectrons that is proportional 

to the local electric potential at the surface at the instant of 

electron release. This is in sharp contrast to previous attosecond 

streaking experiments performed in macroscopic volumes of gasphase 

media, where the electron escape times are longer than the 

optical period; consequently, the change in electron energy 

probes the vector potential of the optical field 26 . For 

nanoplasmonic systems, the imaging of the emitted XUVinduced 

photoelectrons by an energy-resolving photoelectron 

emission microscope (PEEM) probes the electric-field potential 

at the surface as a function of the XUV-pulse incidence time and 

the position at the surface with an attosecond temporal and 

nanometre-scale spatial resolution. 

Schematically our approach is illustrated in Fig. 1. Similarly to 

ref. 27, it is based on the use of attosecond XUV pulses that are 

synchronized with the waveform-controlled optical fields 5 , which 

are used here to drive the nanolocalized excitations of a 

plasmonic nanostructure. A metal (plasmonic) nanosystem is 

driven by an optical ultrashort laser pulse with frequency within 

the range of the plasmonic resonances (from near UV to near 

infrared, NIR). The wavelength of the optical radiation is orders 

of magnitude greater than the characteristic size of the 

nanosystem, which we assume to be on the order of tens of 

nature photonics | VOL 1 | SEPTEMBER 2007 | www.nature.com/naturephotonics 539 



ARTICLES 

XUV pulse 

Optical pulse 


Local plasmonic field 

PEEM 

Photoelectrons 

Time 

Figure 1 Schematic of the system and photoprocesses. The nanosystem is 

shown in a plane denoted in light blue. Instantaneous local fields, which are 

excited by the optical pulse, are shown as a three-dimensional plot. The local 

optical field at a point of the maximum field (‘hottest spot’) is shown as a 

function of time by a red waveform, enhanced with respect to the excitation 

field. The application of an XUV pulse is shown by a violet waveform that is 

temporally delayed with respect to the excitation field. The XUV excitation 

causes the emission of photoelectrons shown by the blue arrows, which are 

accelerated by the local plasmonic potential. They are detected with spatial and 

energy resolution by a PEEM. 

nanometres. The plasmonic nanosystem responds to the almost 

uniform field of the optical excitation with ‘hot spots’, where the 

local fields are enhanced with respect to the excitation field by a 

factor that depends on the quality of the localized surface 

plasmon resonances and can be as high as a few hundred for 

silver nanosystems. These hot spots are indicated as the peaks of 

the local field amplitude in Fig. 1. For the highest of these 

peaks (‘hottest spot’), we show (as the red waveform) the 

temporal dynamics of the local electric field. Similar to direct 

measurements of the optical field’s instantaneous magnitude 27 , 

an attosecond XUV pulse is incident on the system 

synchronously with the optical excitation waveform. It causes the 

photoemission of electrons whose energy is high enough to select 

them from the background of above-threshold ionization (ATI) 

and multiphoton emission created by the optical field. 

These XUV electrons are accelerated by the local plasmonic 

electric fields, which define their energies. Imaging such a system 

in a PEEM with energy resolution will allow one to see the hot 

spots and find the instantaneous magnitude of the plasmonic 

field at the site and time of the emission. As the emission instant 

is defined by the XUV pulse, this would allow one to measure 

the complete spatiotemporal dynamics of the local fields. The 

corresponding spatial resolution is defined solely by the electron 

optics of the PEEM (the de Broglie wavelength of the XUV 

540 


electrons is much smaller and is not a limiting factor) and can 

realistically be on the order of nanometres. The temporal 

resolution is determined by the duration of the attosecond pulse 

and the time of flight of the photoelectrons through the localfield 

region, which can be on the order of (a few) hundred 

attoseconds. Very importantly, this proposed approach to the 

visualization of the attosecond–nanometre dynamics is noninvasive 

with respect to the plasmonic fields because of the low 

power of the XUV pulses. 

ELECTRON ACCELERATION BY NANOPLASMONIC FIELDS 

Different regimes of the electron emission by an XUV pulse will 

now be discussed to identify those realistic for our conditions 

and conducive to our goals. The escape velocity for an electron 

can be estimated as ve ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 

p 

2ðhvXUV 2 W f Þ=m, 

where m is the 

electron mass, vXUV is the frequency of the XUV pulse, Wf is the 

metal workfunction and h is the reduced Planck’s constant. 

The electron escape time from a region of local field of size b is 

te ¼ b/ve. The most important case for our purposes is the one 

to be called a regime of instantaneous acceleration, when an 

electron leaves the local-field region much faster than this field 

oscillates in time, that is, 


t e � T ð1Þ 

where T ¼ 2p/v is a characteristic period of the optical excitation, 

and v is the excitation-pulse carrier frequency. 

Under this condition, an electron is driven by a nearly frozen, 

instantaneous local electrostatic potential f(r, t XUV), where t XUV 

is the emission time defined with a sufficient precision by the 

incidence time of the XUV pulse. The final kinetic energy of an 

electron, which is measured by the PEEM, is found from energy 

conservation, 

E XUVðr; t XUVÞ ¼ hv XUV � W f þ efðr; t XUVÞ ð2Þ 

where e is the electron charge and r the emission point. In sharp 

contrast to attosecond streaking in the gas phase 26 , this energy 

does not depend on the initial momentum of the photoelectron 

or its subsequent flight trajectory. 

Estimates for the currently available XUV pulse parameters 

are chosen as follows 28 : pulse duration t p ¼ 170 as and photon 

energy hv XUV ¼ 91 eV. In such a case, assuming the workfunction 

W f ¼ 5 eV, we obtain the escape velocity v e ¼ 5 � 10 8 cm s 21 and 

the escape time for the localization distance b ¼ 1 nm as t e ¼ 180 

as. Considering an NIR driving pulse with a characteristic period 

T � 3 fs, we see that the instantaneous-regime condition (1) is 

well satisfied for an optical field localization length up to several 

nanometres. Temporal resolution t r of the attosecond plasmonicfield 

microscope is determined by both flight time t e through 

the region of nanolocalized optical fields and the duration t p 

of the XUV pulse itself, t r � t e þ t p. For an NIR driving field and 

b . 3 nm, this resolution is sufficient, t r � T. For visible excitation 

and plasmonic frequencies, sufficient temporal resolution can be 

achieved in the future with attosecond pulses delivered with 

photon energy hv XUV of several hundred eV to a keV and duration 

below 100 as. 

In the instantaneous regime (1) and (2), we can collect a large 

number of electrons leaving the surface without sacrificing the 

temporal resolution provided by the attosecond streaking, owing 

to the fact that the electron energy shift is independent of the 

emission angle. This is in sharp contrast to the regime of 

the multiple electron oscillations induced by the optical field 

nature photonics | VOL 1 | SEPTEMBER 2007 | www.nature.com/naturephotonics

(which, for brevity, will be referred to as the ‘oscillatory regime’). 

The oscillatory regime was dominant in the earlier experiments 

with the gas phase 27,29 . In this regime, the dwelling (escape) time 

of an emitted electron in the region of the streaking optical field 

is large enough, t e & T. In that case, the motion is governed not 

by the scalar potential but by the vector potential A(t), and the 

electron velocity v undergoes many oscillations 26 . As a result, the 

field-dependent part of the final electron energy E XUV is 

dominated by a term DE XUV ¼ (e/mc)v 0A(t XUV) that depends on 

the direction of emission velocity v 0. In energy-resolved 

microscopy, this would lead to a drastic decrease in the intensity 

of the spectral features and their smearing-out. 

The oscillatory regime of the electron acceleration also took 

place in the experiments 30,31 where electrons were emitted from a 

plasmonic metal surface in a multiphoton process excited by an 

NIR field. After the emission, these electrons are accelerated by 

local fields of the surface plasmon polaritons (SPPs). The major 

difference between these experiments and our proposed 

attosecond microscope arises from the fact that the electron 

emission energy of refs 30 and 31 is much lower, and the field 

localization radius characteristic of SPPs is much larger than in 

the present case, b � 1 mm. Thus, in those experiments, the 

Fermi edge of the emitted electron distribution was significantly 

smeared out. 

In contrast, in our proposed approach, E XUV of equation (2) 

does not depend on the direction of emission or a specific 

trajectory of the electron motion. This will allow electrons to be 

collected in a wide solid angle of emission, which is of utmost 

importance in providing a good signal-to-noise ratio, without 

sacrificing the energy resolution (and, consequently, the accuracy 

of the plasmonic potential measurement). It will be shown in the 

following that for the moderate optical intensities assumed, the 

shift of the electron energy due to the acceleration in the local 

field DE XUV ¼ ef is on the order of 10 eV and should be easily 

measurable. Thus, this instantaneous regime is ideal for the 

direct measurement of the attosecond–nanometre dynamics of 

the local plasmonic potential and is the most desirable. However, 

at present it can only be achieved for strongly localized 

plasmonic fields whose extension should not exceed a few 

nanometres. This restriction applies only until attosecond 

pulses become available at much higher photon energies (�1 keV 

and beyond). 

CALCULATIONS AND RESULTS 

PLANAR METAL NANOSTRUCTURES IN THE ATTOSECOND PLASMONIC-FIELD MICROSCOPE 

Concentrating on this most important case of the instantaneous 

acceleration described by equations (1) and (2), we start with a 

model of a silver 32 random planar composite (whose geometry 

will be described in the next paragraph). We apply an s-polarized 

(along the z-direction), waveform-controlled 5 , 5.5-fs optical pulse 

as shown in Fig. 2a. We have computed the electric potential 

f(r, t) using the quasi-static spectral-expansion Green’s function 

method 14,15,33 . The excitation intensity is kept at a moderate level 

of I ¼ 10 GW cm 22 to ensure non-damaging conditions of the 

excitation. The carrier frequency of the optical (NIR) pulse is 

chosen to be 1.55 eV (corresponding to an 800-nm vacuum 

wavelength). The temporal kinetics of the local field at a site 

where it reaches its global maximum is displayed in Fig. 2b. It 

shows the initial period of the driven oscillations for t , 20 fs, 

where the response closely reproduces, albeit with a delay, the 

5-fs excitation pulse, which reflects the bandwidth of this 

plasmonic system. At longer times, after the end of the excitation 

pulse, the free-induction evolution shows the interference 


Excitation field (a.u.) 

Hot spot field (a.u.) 

1 

0 

–1 

20 

0 

–20 

100 200 

t (fs) 

100 200 

t (fs) 

ARTICLES 

Figure 2 Excitation field and kinetics of the local field at a hot spot. 

a, Excitation field as a function of time. b, Local field at the position of the 

maximum (‘hottest spot’) in Fig. 3b as a function of time. The field magnitude is 

shown relative to the amplitude of the excitation pulse, which is set to 1. 

The vertical arrow denotes the oscillation period within which an XUV pulse is 

applied to probe the local field. 

beatings of several plasmonic eigenmodes. The maximum 

enhancement in this hot spot is Q � 30. 

The random planar composite is generated as a collection of 

uncorrelated silver cubes (monomers) positioned on a plane in 

vacuum, which is illustrated in Fig. 3a for a monomer size of 4 nm. 

As is characteristic of plasmonic nanosystems, there are hot spots 

of local fields induced by the optical excitation. The local field 

dynamics is shown in Fig. 2b for the ‘hottest spot’, where the local 

field reaches its global maximum in space and time. We assume 

that an XUV attosecond pulse is incident at this system delayed 

with respect to the waveform-controlled 5 driving field in such a 

way that it probes the system close to the instance of the local field 

maximum (indicated by an arrow in Fig. 2b). According to 

equation (2), we compute the energy shift of an electron emitted by 

such an XUV pulse, which is due to the acceleration in the local 

plasmonic fields, as DE XUV(r, t XUV) ¼ ef(r, t XUV). We assume that 

these photoelectrons are spatially resolved by a PEEM and show in 

Fig. 3b–f a series of the electron energy distributions with an 

interval of Dt XUV � 200–400 as during a half period of the driving 

field. The distributions are shown as a three-dimensional map in 

panel 3b and as topographic maps in panels c–f. Even for the 

moderate excitation intensities used, the energy shift jDE XUVj in 

hot spots of the plasmonic potential is rather large (�10 eV) and, 

consequently, relatively easily measurable. There is a pronounced 

nanometre–attosecond kinetics of the electron energy observed in 

these distributions, with sharp hot spots indicative of those of the 

local fields. These hot spots are concentrated around the edges and 

voids of the metal nanostructure as is a general trend in 

nanoplasmonics. (See Supplementary Information for more details, 




ARTICLES 


z (nm) 

z (nm) 

z (nm) 

60 

40 

20 

0 

0 

60 

40 

20 

60 

40 

20 

20 40 60 

x (nm) 

t XUV = 16.87 fs 

20 40 60 

x (nm) 

t XUV = 17.64 fs 

20 40 60 

x (nm) 

in the movie Attosecond_Electron_Energy_tm5.5_130_180_.2mov 

of the XUV electron energy distribution for a longer, 30-fs interval 

of the evolution.) 

Note that after leaving the region of the enhanced local fields, 

the electron’s motion is oscillatory and defined by the 

vector potential of the excitation field in the empty 

space. The maximum change of the electron energy due to 

this motion is given by a leading term (hv XUV 2 W f)j, where 

j � p {8pe 2 I(t XUV)/[mcv 2 (hv XUV 2 W f)]} is a dimensionless 

parameter (c is the speed of light in a vacuum); it is assumed 

that j � 1. In fact, for I ¼ 10 GW cm 22 , j � 5 � 10 23 ; thus, this 

energy change in the free space is small enough and can be 

safely neglected. 

XUV ELECTRON ENERGIES FOR NANOSHELLS 

It may also be useful to carry out the first experiments without 

spatial resolution by studying the energy distribution of the 

XUV-emitted electrons. As an example, we consider in such a 

case a much simpler nanoplasmonic system such as metal 

nanoshells 34 . Such nanoshells are frequency tunable by adjusting 

their aspect ratio A (the ratio of the inner to outer shell radius). 

Assuming small shell radius (R . 10 nm), we can use the quasielectrostatic 

approximation where the corresponding solutions 

are obtained in a simple analytical form. 

Consider for instance a silver nanoshell of R ¼ 2.5 nm on a core 

with a permittivity of 1 ¼ 10, which is resonant to 800 nm 

radiation for A ¼ 0.831. For such a case, the energy shift of the 


0 

12 

–12 

ΔE XUV (eV) 

60 

z (nm) 

40 

z (nm) 

z (nm) 

20 

60 

40 

20 

60 

40 

20 

t XUV = 16.68 fs 

20 

40 

x (nm) 

t XUV = 17.25 fs 

60 

0 

5 

–5 

–10 

10 

ΔE XUV (eV) 

20 40 60 

x (nm) 

t XUV = 18.03 fs 

20 40 60 

x (nm) 

Figure 3 Topography of a nanosystem and spatiotemporal kinetics of the local field potential as detected by the attosecond plasmonic-field microscope. 

Intensity of the excitation optical pulse I ¼ 10 GW cm 22 , and photon energy hv ¼ 1.55 eV. a, Topography of the nanosystem: random planar composite consisting of 

4 nm� 4 nm � 4 nm silver cubes arranged on a plane with fill factor 0.5. This composite is smoothed to within 2 nm to improve numerical precision. 

b–f, Distributions of the energy shift DE XUV of electrons emitted by the XUV pulse in the plane of this nanostructure shown for different moments t XUV (as indicated in 

the panels) of the XUV pulse incidence within the half cycle of local field oscillations. b, A three-dimensional map. c–f, Topographic colour maps showing details of 

the nanometre–attosecond spatiotemporal kinetics. 

542 


nature photonics | VOL 1 | SEPTEMBER 2007 | www.nature.com/naturephotonics

electrons is shown in Fig. 4, as a function of both the phase delay w 

between the driving NIR field and the incident attosecond pulse, 

and the azimuthal angle u of the electron-emission point (for 

I ¼ 10 GW cm 22 and hv ¼ 1.55 eV). To avoid any confusion, we 

emphasize that the electron energy does not depend on the 

direction of the electron velocity but only on the azimuth of a 

point at the nanoshell surface from which the emission took place. 

In the exact resonance for Fig. 4a, the maximum shift is 

jDE XUVj � 10 eV, on the same order as for the random planar 

composite above. The maximum is reached for w ¼ p/2, as 

expected for resonant excitation. The nanoshell resonance is very 

sharp; therefore, a small change of A to a value of 0.829 detunes 

the resonance and leads to a significant reduction in the 

magnitude of the effect and a shift in its phase dependence 

(compare with Fig. 4b). 

DISCUSSION AND CONCLUSIONS 

ΔE XUV (eV) 

10 

0 

–10 

0 

2 

The state-of-the-art of attosecond technology for the attosecond 

plasmonic-field microscope proposed in this paper is represented 

by t p ¼ 170 as, hv XUV ¼ 93 eV XUV pulses containing &�10 6 

photons and delivered at a 3-kHz repetition rate 28,35 . As has been 

demonstrated earlier, such XUV pulses can be focused to a spot 

diameter of 2 mm (ref. 36). 

The signal-to-noise ratio of the proposed attosecond 

nanoplasmonic-field microscope is limited by the electron 

currents that can be obtained from nanoscopic surface areas for 

the presently available intensities of the XUV sources 28,35 . In 

Materials and Methods the XUV photoelectron current emitted 

per 1-nm 2 area of the nanosystem surface is estimated. The 

number of photoelectrons per unit area of the nanosystem 

surface and unit time of observation (photoelectron fluence) is 

j � 15 nm 22 s 21 for gold and j � 10 nm 22 s 21 for silver. For the 

spatial resolution of PEEM available at present, r � 10 nm, this 

yields the electron flux from the area of resolution as 

N 0 ¼ pr 2 j � 4,500 s 21 for gold and N 0 � 3,000 s 21 for silver. 

Such electron counts should not pose any problem for detection 

with PEEM. 

Future steps towards the experimental implementation of the 

attosecond nanoplasmonic field microscope will use ultrashort 

pulsed XUV radiation with photon energies exceeding 

hv XUV ¼ 100 eV. At present, single-attosecond XUV pulses are 

obtained by high-harmonic generation driven by a few-cycle 

φ 

4 


A = 0.831 

6 

0 

1 

2 

θ 

3 

ΔE XUV (eV) 

5 

0 

–5 

0 

2 

waveform-stabilized Ti:sapphire laser pulse (800 nm, 5 fs) in a Ne 

gas jet target. Selecting a broadband (�9 eV full-width at-halfmaximum 

(FWHM), centre frequency 91 eV) spectrum from the 

harmonic plateau range close to its cut-off by means of a Mo/Si 

multilayer-coated XUV mirror, single-attosecond XUV pulses have 

been extracted and pulse duration down to 250 as FWHM has been 

experimentally verified 7 . Improvements (broader-band mirrors) 

have yielded pulses shorter than 200 as. This offers a temporal 

resolution of at least 100 as. The excitation optical pulse and the 

probing XUV pulse are refocused to the sample by means of an 

interferometric double-mirror configuration allowing for controlled, 

variable time delays with 10-as steps. The spatially and energyresolved 

detection of photoelectrons is achieved using a PEEM that 

is equipped with a flight drift tube followed by a two-dimensional 

delay-line detector. The spatial resolution of this time-resolved 

PEEM is limited by the electron optics to 20 nm. Silver and gold 

samples have been grown (see Supplementary Information for 

details, Experimental_Steps.pdf). 

To conclude, we have proposed and developed a theory for an 

attosecond nanoplasmonic-field microscope. The main advantage 

of such a microscope is that it is non-invasive with respect to the 

nanoplasmonic fields. The principle of this microscope is based 

on the photoemission of electrons by an XUV attosecond pulse 

that is synchronized with a waveform-stabilized driving optical 

field. Information about the nanoplasmonic fields is imprinted in 

the energy of the XUV-emitted electrons owing to their 

acceleration in the instantaneous electrostatic potential of the 

surface plasmon oscillations excited by the optical field. 

Our proposed microscope will open up unique possibilities to 

directly study and control ultrafast photoprocesses in surface 

plasmonic nanosystems and circuits. It images the local 

nanoplasmonic field in real space with nanometre-scale spatial 

resolution and in real time with �100 as temporal resolution. 

This approach will be especially important in ultrafast 

nanoplasmonic systems where very tight localization of optical 

fields occurs, for example, when a shaped pulse of radiation 

induces nanolocalized fields at a desired nanosite. The 

microscope could also be used to study various plasmonenhanced 

photoprocesses such as femtosecond photochemistry, 

light detection, and solar energy conversion where the processes 

of the energy transfer can be ultrafast. For example, with 

nanoplasmonic antennas, which couple together molecular and 

semiconductor systems with external fields, the ultrafast kinetics 


φ 

4 

A = 0.829 

6 

0 

1 

2 

θ 

ARTICLES 

Figure 4 Energy shift of electrons emitted from the surface of silver nanoshells as a function of the azimuthal angle u of the emission point and the phase 

w of the delay between the driving optical radiation and the probing attosecond XUV pulse. The intensity of the excitation optical pulse is I ¼ 10 GW cm 22 , 

the photon energy is hv ¼ 1.55 eV and the outer radius of the nanoshells is R ¼ 2.5 nm. a,b, Data for the nanoshells with aspect ratio A ¼ 0.831 (a) and for 

A ¼ 0.829 (b). 



3

ARTICLES 


of the energy exchange can be studied through the corresponding 

kinetics of the nanoplasmonic fields. 

One of the most important potential uses of the attosecond 

nanoplasmonic-field microscope will be in the design and study 

of elements and devices for ultrafast and ultradense (at the 

nanoscale) optical and optoelectronic information processing and 

storage. To bring practical advantages over existing electronic and 

optoelectronic technology, such nanoscale devices must 

necessarily be ultrafast (with a subpicosecond or femtosecond 

response time). Examples of such devices include optical 

nanotransistors, memory cells, nanoplasmonic media for the 

mass storage of information, and nanoplasmonic interconnects 

where both the nanolocalization of energy and its ultrafast 

temporal kinetics are important. 

MATERIALS AND METHODS 

Here we estimate the XUV electron photocurrent that can be obtained per 1-nm 2 

area of the nanosystem metal surface. We adopt the values of the XUV photon flux 

for state-of-the-art XUV pulses 28,35 : J XUV � 10 9 s 21 (the flux averaged over time). 

The XUV pulses can be focused 36 to a spot with a radius R � 1 mm. This yields an 

XUV photon fluence I X � 8 � 10 16 cm 22 s 21 . For hv XUV ¼ 91 eV photon energy, 

the photoionization cross-section per atom for silver is s a ¼ 7.1 Mb (ref. 37) and 

for gold is s a ¼ 10.9 Mb (refs 37, 38). 

Consider emission from an area a ¼ 1 nm 2 . For electron energies �100 eV, 

the elastic escape depth (from which electrons come without losing their energy 

to collisions) is h ¼ 0.5 nm (ref. 39). The total photoemission cross-section from 

atoms within this elastic depth and area a is s tot ¼ rahs a, where r is the number 

density of atoms. From this we obtain the expected values of XUV photoelectron 

fluence j � 15 nm 22 s 21 for gold and j � 10 nm 22 s 21 for silver. 

Received 30 March 2007; accepted 23 July 2007; published 3 September 2007. 

References 

1. Paul, P. M. et al. Observation of a train of attosecond pulses from high harmonic generation. Science 

292, 1689–1692 (2001). 


3. Niikura, H. et al. Sub-laser-cycle electron pulses for probing molecular dynamics. Nature 417, 

917–922 (2002). 

4. Drescher, M. et al. Time-resolved atomic inner-shell spectroscopy. Nature 419, 803–807 (2002). 

5. Baltuska, A. et al. Attosecond control of electronic processes by intense light fields. Nature 421, 

611–615 (2003). 

6. Niikura, H. et al. Probing molecular dynamics with attosecond resolution using correlated wave 

packet pairs. Nature 421, 826–829 (2003). 


8. Sekikawa, T., Kosuge, A., Kanai, T. & Watanabe, S. Nonlinear optics in the extreme ultraviolet. Nature 

432, 605–608 (2004). 

9. Lopez-Martens, R. et al. Amplitude and phase control of attosecond light pulses. Phys. Rev. Lett. 94, 

033001 (2005). 

10. Baker, S. et al. Probing proton dynamics in molecules on an attosecond time scale. Science 312, 

424–427 (2006). 

11. Dudovich, N. et al. Measuring and controlling the birth of attosecond XUV pulses. Nature Phys. 2, 

781–786 (2006). 

12. Sansone, G. et al. Isolated single-cycle attosecond pulses. Science 314, 443–446 (2006). 

13. Kling, M. F. et al. Control of electron localization in molecular dissociation. Science 312, 

246–248 (2006). 

544 


14. Stockman, M. I., Faleev, S. V. & Bergman, D. J. Localization versus delocalization of surface plasmons 

in nanosystems: Can one state have both characteristics? Phys. Rev. Lett. 87, 167401 (2001). 

15. Stockman, M. I., Faleev, S. V. & Bergman, D. J. Coherent control of femtosecond energy localization 

in nanosystems. Phys. Rev. Lett. 88, 067402 (2002). 

16. Bergman, D. J. & Stockman, M. I. Surface plasmon amplification by stimulated emission of 

radiation: Quantum generation of coherent surface plasmons in nanosystems. Phys. Rev. Lett. 90, 

027402 (2003). 

17. Lehmann, J. et al. Surface plasmon dynamics in silver nanoparticles studied by femtosecond timeresolved 

photoemission. Phys. Rev. Lett. 85, 2921–2924 (2000). 

18. Zentgraf, T., Christ, A., Kuhl, J. & Giessen, H. Tailoring the ultrafast dephasing of quasiparticles in 

metallic photonic crystals. Phys. Rev. Lett. 93, 243901 (2004). 

19. Kubo, A. et al. Femtosecond imaging of surface plasmon dynamics in a nanostructured silver film. 

Nano Lett. 5, 1123–1127 (2005). 

20. Stockman, M. I. & Hewageegana, P. Nanolocalized nonlinear electron photoemission under coherent 

control. Nano Lett. 5, 2325–2329 (2005). 

21. Brixner, T. et al. Quantum control by ultrafast polarization shaping. Phys. Rev. Lett. 92, 

208301 (2004). 

22. Brixner, T., d. Abajo, F. J. G., Schneider, J. & Pfeiffer, W. Nanoscopic ultrafast space–time-resolved 

spectroscopy. Phys. Rev. Lett. 95, 093901 (2005). 

23. Sukharev, M. & Seideman, T. Phase and polarization control as a route to plasmonic nanodevices. 

Nano Lett. 6, 715–719 (2006). 

24. Aeschlimann, M. et al. Adaptive subwavelength control of nano-optical fields. Nature 446, 

301–304 (2007). 

25. Pelton, M., Liu, M. Z., Park, S., Scherer, N. F. & Guyot-Sionnest, P. Ultrafast resonant optical 

scattering from single gold nanorods: Large nonlinearities and plasmon saturation. Phys. Rev. B 73, 

155419 (2006). 

26. Corkum, P. B. & Krausz, F. Attosecond science. Nature Phys. 3, 381–387 (2007). 

27. Goulielmakis, E. et al. Direct measurement of light waves. Science 305, 1267–1269 (2004). 

28. Schultze, M. et al. Powerful 170-attosecond XUV pulses generated with few-cycle laser pulses and 

broadband multilayer optics. New J. Phys. 9, 243 (2007). 

29. Drescher, M. et al. X-ray pulses approaching the attosecond frontier. Science 291, 1923–1927 (2001). 

30. Kupersztych, J., Monchicourt, P. & Raynaud, M. Ponderomotive acceleration of photoelectrons in 

surface-plasmon-assisted multiphoton photoelectric emission. Phys. Rev. Lett. 86, 

5180–5183 (2001). 

31. Kupersztych, J. & Raynaud, M. Anomalous multiphoton photoelectric effect in ultrashort time 

scales. Phys. Rev. Lett. 95, 147401 (2005). 

32. Johnson, P. B. & Christy, R. W. Optical constants of noble metals. Phys. Rev. B 6, 4370–4379 (1972). 

33. Stockman, M. I., Bergman, D. J. & Kobayashi, T. Coherent control of nanoscale localization of 

ultrafast optical excitation in nanosystems. Phys. Rev. B 69, 054202–10 (2004). 

34. Nehl, C. L. et al. Scattering spectra of single gold nanoshells. Nano Lett. 4, 2355–2359 (2004). 

35. Goulielmakis, E. et al. Attosecond control and measurement: Lightwave electronics. Science 

(in the press). 

36. Schnurer, M. et al. Guiding and high-harmonic generation of sub-10-fs pulses in hollow-core fibers 

at 10(15) w/cm 2 . Appl. Phys. B 67, 263–266 (1998). 

37. Henke, B. L., Lee, P., Tanaka, T. J., Shimabukuro, R. L. & Fujikawa, B. K. Low-energy X-ray 

interaction coefficients: Photoabsorption, scattering, and reflection. Atomic Data and Nuclear Data 

Tables 27, 1–131 (1982). 

38. Manson, S. T. in Photoemission in Solids Vol. 1, (eds Cardona, M. & Ley, L.) 135–163 (Springer, 

Berlin, New York, 1978). 

39. Tanuma, S., Powell, C. J. & Penn, D. R. Calculations of electron inelastic mean free paths. ii. Data for 

27 elements over the 50–2000 eV range. Surf. Interface Anal. 17, 911–926 (1991). 

Acknowledgements 

The work of M.I.S. is supported by grants from the Chemical Sciences, Biosciences and Geosciences 

Division of the Office of Basic Energy Sciences, Office of Science, US Department of Energy, a grant CHE- 

0507147 from NSF, and a grant from the US-Israel BSF. M.I.S.’s work at the Max-Planck-Institute for 

Quantum Optics (Garching, Germany) was supported by a Research Stipend of the Max Planck Society. 

The work of M.F.K., U.K., and F.K. was partially supported by the German Science Foundation (DFG) 

through the Cluster of Excellence Munich Center for Advanced Photonics. M.F.K. acknowledges support 

by an EU reintegration grant and the DFG Emmy–Noether program. MIS acknowledges helpful 

discussions with S. Manson regarding photoelectron cross-sections and with P. Corkum regarding 

charging of the surfaces. 

Correspondence and requests for materials should be addressed to M.I.S. or F.K. 

Supplementary information accompanies this paper on www.nature.com/naturephotonics. 

Competing financial interests 

The authors declare no competing financial interests. 

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/ 

nature photonics | VOL 1 | SEPTEMBER 2007 | www.nature.com/naturephotonics

Attosecond Control and Measurement: Lightwave Electronics

Create successful ePaper yourself

Delete template?

Save as template?