Journal of the Korean Physical Society, Vol. 73, No. 2, July 2018, pp. 227∼234 Overview Articles

Extreme Metrology for Ultrafast Electron Dynamics at the Atomic Scale

Dong Eon Kim∗

Physics Depart. & Center for Atto. Sci. and Tech,POSTECH and Max Planck POSTECH/KOREA Research Initiative, Pohang 37673, Korea

(Received 23 May 2018)

The past two decades have witnessed the remarkable advance in the new metrology for ultra-fast electron dynamics, which allows one to control material processes at electron level and studydynamics far away from equilibrium. This excitement is presented with a brief history of the de-velopment and exemplary experiments performed with the new metrology. Applications reviewedinclude atomic and molecular, condensed-phase science, and advances in X-ray Free electron lasers(XFELs), with an emphasis on attosecond XFEL studies. The future perspectives are envisioned.

PACS numbers: 03.65.Xp, 32.30.−r, 33.20.−t, 78.47.J−, 82.53.KpKeywords: Ultrafast science, Attosecond science, Few-cycle pulse, Isolated attosecond pulseDOI: 10.3938/jkps.73.227

I. INTRODUCTION

At the dawn of the new 21st century, there have beenlots of discussions about the scientific achievements inthe 20th century and the perspectives in the 21st centuryat individual scientists’ level as well as at organizationallevel all over the world. The Basic Energy Sciences Ad-visory Committee (BESAC) of Department of Energy ofthe United States of America has identified five grandchallenges (Fig. 1), which are “to be scientifically deepand demanding, and be relevant to the broad portfolioof basic science, and promise real dividends in devices ormethods that can significantly improve the quality of life[1]”; these are

(1) to control material processes at the level of elec-trons

(2) to design and perfect atom-efficient and energy-efficient synthesis of new forms of matter with tai-lored properties,

(3) to understand and control the remarkable proper-ties of matter that emerge from complex correla-tions of atomic and electronic constituents,

(4) to master energy and information on the nanoscaleto create new technologies with capabilities rivalingthose of living things,

(5) to characterize and control matter away - especiallyfar away - from equilibrium.

∗E-mail: kimd@postech.ac.kr

Items (1) and (5) concern the dynamics of a system ofinterest. Quantum theory, which was developed duringthe 1st half of the 20th century, has increased our un-derstanding of atomic, chemical, and condensed phasesof materials. Overall, the 20th century science was pri-marily concerned with how quantum systems are orga-nized and manifest their properties and functionalities,and therefore considered phenomena that are in equilib-rium.

In contrast, in the 21st century, interest is growingabout the study of how quantum systems evolve, andeventually, how to induce such quantum systems to be-have as desired. In this sense, we are entering a newscientific paradigm, “Control Age.” For example, scien-tists would like to move electrons around during chemi-cal reaction processes that are far from equilibrium. Thenew era of science calls for new tools to control elec-tron behavior in matters at the utmost time scale (fem-tosecond to attosecond) with atomic spatial resolution.The last decade has observed exciting advances in thedevelopment of attosecond visible and extreme ultravi-olet (XUV) pulse, and femtosecond hard x-ray pulse,along with appropriate metrologies for proper utilizationof such sources.

In this overview article, I try to share the excitementin the development of new metrologies, and in the newinsights that these tools have provided in aspects of bothfundamental science and future technology. Although Ihave tried to highlight all important work, this article isnot meant to cover all progress. Apologies go to manyexcellent scientists whose work is not cited here.

This article has the following structure. Section IIdiscusses the brief history in relation with the develop-ment of few-cycle lasers and their applications to gener-

pISSN:0374-4884/eISSN:1976-8524 -227- c©2018 The Korean Physical Society

-228- Journal of the Korean Physical Society, Vol. 73, No. 2, July 2018

Fig. 1. (Color online) Feature article about the BESACreport published in Physics Today 2007. Reproduced from[1] with permission.

ation of isolated attosecond pulses. The following sec-tions are dedicated to the discussion of several topicsthat I consider to be important. Section III considersatomic, molecular phase science, and Sec. IV considerscondensed-phase science. Section V covers advances inX-ray Free electron lasers (XFELs) with an emphasison attosecond XFEL studies. Section VI concludes withperspectives for the next decade.

II. BRIEF HISTORY OF ULTRAFASTSOURCE DEVELOPMENT

Efforts to understand the dynamics of natural phe-nomena began almost 100 years before the invention ofthe laser in 1960. In 1864, Toepler used a short spark togenerate sound waves and subsequently obtained theirimages by using an electronically-delayed second spark[2]. This technique was improved by Abraham andLemoine in 1899 to the limit of the temporal resolutionimposed by the duration of a light flash [3]. This tem-poral resolution of nanoseconds was the limit until thearrival of the laser in 1960. The pulse duration of lasersis being decreased with the advance of laser technolo-gies such as mode-locking, so the temporal resolution ofpump-probe experiments had greatly been improved toa few tens of femtosecond by the 1990s. Utilizing thistechnology, Zewail performed a series of pioneering stud-ies that opened a new field called femto-chemistry [4,5].Using visible or near-infrared (NIR) lasers with pulse du-rations of a few tens of femtoseconds, femtochemistry, inprinciple, follows the dynamics of atoms in various chem-ical reactions, most of which occur on a picosecond timescale.

An electron’s response time is much shorter than thepicosecond time scale of atoms’ motion, so the electronsare assumed to be in equilibrium with a given atomicarrangement at each moment during a reaction process.This is the Born-Oppenheimer approximation; it doesnot apply to the study of electron dynamics, which takesplace on a sub-femtosecond time scale. Such investiga-tion requires new tools, namely attosecond metrology.

Fig. 2. (Color online) Change of carrier-envelop phase(CEP). Within a few cycle pulse, the electric field oscillationis significantly different for different CEPs.

The advance to the attosecond time domain of tempo-ral resolution took one and a half decades. This is thedomain that is required when tracking electron dynamicsin atoms, molecules and condensed matter. The advancetoward the attosecond time domain was made possibleby the development of three key technologies: discoveryof Ti:S crystal [6], Kerr-lens mode-locking (KLM) [7] andchirped pulse amplification (CPA) [8]. These three keytechnologies enable production of high-energy femtosec-ond pulses that enable execution of various nonlinearprocesses such as generation of an octave-spanning su-percontinuum (OSSC) and of high-order harmonics, bothof which are the essential ingredients for the generationof an isolated attosecond pulse. As supercontinuum gen-eration has become successful and used in various areas,the escalating demand for the control of dispersion, espe-cially the group velocity dispersion (GVD) has led to thedevelopment of chirp mirrors (CMs) KLM lasers basedon them [9]. The combination of OSSC and CM has ledto the generation of few-cycle femtosecond lasers. Asthe pulse duration is reduced to a few cycles, the phaseof the electric field oscillation within the envelope of apulse becomes important. The variation of the electricfield within the envelope of a few-cycle pulse is large andsignificantly dependent on the phase of the field with therespect to the peak of the envelope (Fig. 2), which hasbeen called the carrier-envelop phase (CEP). The forcethat acts on an electron is governed by the electric fielditself, not by the envelop; as a result, different field os-cillations within the same envelop shape exert differenteffects on electron dynamics. Control of electrons on anattosecond time scale, and consequent generation of anisolated attosecond pulse requires complete control of theCEP of a few-cycle pulse.

The solution to the control of CEP or the controlledlight waveform came from adoption of a frequency-combtechnique, for the invention of which T. Hansch wasawarded Nobel prize in 2005 [10]. The collaborationbetween time-domain and frequency-domain metrologyprovided a great synergy and opened up the avenue tocontrol electron dynamics in microcosms on attosecondtimescales. Such a CEP-stabilized few cycle laser was

Extreme Metrology for Ultrafast Electron Dynamics at the Atomic Scale – Dong Eon Kim -229-

Fig. 3. (Color online) Strong-field ionization and pump-probe setting for its real-time observation. Reproduced from[18] with permission.

applied to high-order harmonics generation (HHG), andsuccessfully generated an isolated attosecond pulse forthe first time [11] in 2001, with improved precision later[12]. Since then, additional techniques for isolated at-tosecond pulses have been developed, adding new char-acteristics and capabilities: polarization gating [13], gen-eralized double optical gating [14], two-color gating [15],and field synthesizer [16].

III. ULTRAFAST ATOMIC ANDMOLECULAR PHYSICS

1. Real time observation of electron tunneling

Tunneling of an electron through a potential barrier isa bizarre phenomenon that occurs in quantum mechani-cal world, but does not have a counterpart in the classicalworld. Attosecond time resolution has enabled observa-tion of a tunneling event in real time, and attempts havealso been made to measure the duration of the tunnelingevent.

Light-field tunneling ionization traces back to a semi-nal paper by Keldysh [17] in 1965, shortly after the in-vention of the laser. Keldysh suggested that a valenceelectron may escape by tunneling if its potential is sup-pressed by the light field (Fig. 3(a)). If the dimension-less Keldysh parameter is less than unity, the ioniza-tion is predicted to occur during a fraction of the half-oscillation cycle of the light field (Fig. 3(c)) under theassumption that the photon energy is much smaller thanionization energy. The following scenario for real-timeobservation of tunneling event using attosecond pulsewas envisioned: an attosecond XUV pulse generates ionsin excited states, from which electrons exit by tunnel-ing under an NIR few-cycle probe pulse with a delay;

Fig. 4. (Color online) left: Energy levels and transitions inNe1+ and Ne2+ ions. right: (a) Ne2+ ion yield versus delaybetween the sub-femtosecond XUV pump and the few-cyclenear-infrared probe: experiment. Thick red line: average offive adjacent data points of the six scanned data; thin greyline: the same but recorded without CEP stabilization of NIRprobe laser. Inset: ionization step with a rise time of 380 as(full-width at half-maximum of the Gaussian function derivedfrom the error function). (b) Simulation results. Reproducedfrom [18] with permission.

as a result, doubly-charged ions are produced. The pro-duction of doubly-charged ions is expected to show sub-half-cycle steps in the ionization profile(Figs. 3(b) and(d)). Krausz’ group at Max Planck Inst. for QuantumOptics has performed a series of experiments using Negas, and has confirmed the sub-cycle step in ion pro-duction, and achieved the first real-time observation ofelectron tunneling [18]. The left figure in Fig. 4 showsthe relevant energy levels in neutral Ne and ionized Ne(Fig. 4, left). Absorbed XUV photons of ∼ 90 eV pro-duce singly-charged or doubly-charged Ne with higherprobability for Ne1+ than for Ne2+. The absorption andshake-up processes produce populations in 2p−2nl con-figuration. The subsequent few-cycle laser field at a delaywith respect to the attosecond XUV pulse removes elec-trons from these shake-up states, and yields additionaldouble ionization. The experimental data is shown onthe right panel in Fig. 4 obtained with a peak intensityof ∼ 7 × 1013 Wcm−2 of the NIR probe (the thick redline: the average of five adjacent data points of the 6scanned data; the thin grey line: the same but recordedwithout the CEP stabilization of NIR probe laser). Theinset depicts an ionization step with a rise time of 380 asbased on the simulation shown below (Fig. 4(b)) usingKeldysh theory of tunneling ionization for the experi-mental conditions.

-230- Journal of the Korean Physical Society, Vol. 73, No. 2, July 2018

Fig. 5. (Color online) Attoclock experiment with angularstreaking. Reproduced from [19] with permission.

2. Electron tunneling time measurement

The experimental capabilities associated with attosec-ond metrology have allowed us to enter situations inwhich genuine understanding of tunneling time becomesimportant, and have also provided the tools to addressan unresolved question of tunneling time.

The tunneling time has been highly debated and isalmost as old as quantum mechanics. Many conflict-ing theories have been developed; they advocated dif-ferent tunneling times such as the Larmor time [20],the Buttiker-Landauer time [21], the Eisenbud-Wignertime [22] and the Pollack-Miller time [23]. Until now, nodefinitive experiment has been conducted to judge which,if any, of these times is correct.

Eckle et al. devised the idea of projecting the tunnel-ing time to the momentum distribution of a tunnelingelectron, by using ‘angular streaking’ [24]. The idea ofangular streaking tells us that the final momentum of theelectron can provide us with highly accurate informationabout the time of its first appearance at the tunnel exit.

The experimental setup (Fig. 5, upper) is equipped witha velocity-mapping imaging (VMI) device. The linearly-polarized pulse enters a quarter-wave plate, and therebybecomes elliptically polarized with an ellipticity of 84%.The 2nd half-wave plate rotates the major axis of thepolarization to a convenient angle. This light interactswith a gas jet in a vacuum chamber to produce pho-toelectrons, which are collected by the VMI; the resultis an image of the electron momentum distributions inthe plane of polarization (Fig. 5, inset in bottom). Theradially- integrated distribution is fitted by a Gaussianfunction (Fig. 5, red line in bottom panel) to extractthe center of the momenta distribution. Experimentaltunneling delay time is proportional to this angle and isextracted from the experimentally-measured angle.

The first of these “attoclock experiments” [25] set anupper limit of 40 as to electron tunneling time, and more-thorough experiments were performed later [26]. Exper-imental results indicate that only Larmor time agreeswell with the attoclock measurements. The Larmor timecan be expressed as an average quantity over the proba-bility amplitudes of paths in the Feynman path integralapproach; i.e., the experimental observations imply thatthe time taken by an electron to tunnel is probabilisticrather than deterministic.

3. Counter Rotating Wave effect in atoms

Attosecond metrology has renewed long-standing in-terest in the understanding and manipulation of electrondynamics in atoms, molecules, and solids under strongfields. For the study of light-atom interaction, the rotat-ing wave approximation has been widely used; it holds inthe weak-field regime in which the Rabi frequency, whichis proportional to strength of the light field, is muchsmaller than the light frequency. For atoms illuminatedby a strong light field, this approximation breaks down,because the counter-rotating wave (CRW) terms in theHamiltonian and their effect can no longer be neglected.These CRW effects on the dynamics are themselves in-teresting and can lead to applications in circuit quantumelectrodynamics [27], quantum computation (QC) usingSQUIDS [28] and in the creation of entanglement be-tween two atoms [29].

The CRW effects have recently been theoretically stud-ied [30], but no clear experimental observation has beenmade until recently. Attosecond transient absorptionspectroscopy has been used to achieve the first obser-vation of ultrafast sub-cycle CRW features with opticalfrequencies on a sub-femtosecond time scale [31]. TheCRW effect was studied in highly excited 4d−1np statesin Xe that was illuminated by an intense NIR pulse. All4d−1np states are autoionizing, with a lifetime of 6.6 fs,due to Auger decay of the unstable 4d hole. Conse-quently, the CRW should be faster than the lifetime tobe observed. For an 800-nm pulse, the optical period is

Extreme Metrology for Ultrafast Electron Dynamics at the Atomic Scale – Dong Eon Kim -231-

Fig. 6. (Color online) Schematic layout of attosecond tran-sient absorption spectroscopy (upper panel) Experimentaland simulation data (bottom panel). Adopted from [31].

2.6 fs; hence, the CRW effect occurs within 1.3 fs andbeats several times within the core lifetime.

Therefore, attosecond time-resolution is required. Toobserve such ultrafast electron dynamics in real time,an attosecond transient absorption spectroscopy (ATAS)has been adopted (Fig. 6, upper panel). AttosecondXUV pulses were generated from Ne, using few-cycle(∼ 4 fs) 800-nm pulses. A concentric two-segment curvedmirror along with a special metal filter separated theattosecond XUV and the NIR pulses, and provided aprecise delay between them. The two beams were thenfocused co-linearly onto a quasi-static gas cell containingXe as a target gas. The absorption spectrum was ob-tained by measuring the transmitted spectrum Igas fromthe gas target and also the spectrum I0 taken in the ab-sence of the gas target as a function of delay betweenNIR and XUV pulses. Experimental absorption spec-tra at different delays between XUV and NIR pulse, andTDCIS calculations (Fig. 6, bottom) agree well with the2ω oscillation, which is the manifestation of the CRWeffect.

4. Ultrafast charge migration

Theoretical studies of sudden ionization of a complexmolecule by attosecond pulses suggest the possibility of

ultrafast charge dynamics (“charge migration”) along themolecular structure, on a temporal scale ranging from afew femtoseconds down to tens of attoseconds [32]. Theinvestigation of such electron dynamics in molecule mayallow development of attochemistry [33], which will en-able us to better understand the effects of electronic cor-relations in molecular chemistry.

The first try with an isolated attosecond pulse wasmade in 2014 [34]. Isolated attosecond pulses of photonenergy between 15 and 35 eV with a pulse duration of300 as initiated charge migration in an amino acid. De-layed waveform-controlled NIR pulses with a pulse dura-tion of 4 fs probed the subsequent development by mea-suring ion production with a time-of-flight mass spec-trometer (TOFMS). Charge migration was manifestedas an oscillation with a period of 4.3 fs in production ofdoubly-charged immonium ions; it is due to the loss ofthe carboxylic group. This oscillation period is shorterthan the vibrational response of the molecule. Numericalsimulations of the temporal evolution indicated that thecoherent superposition of electronic states created by anattosecond XUV pulse is the origin of charge migration,and that the measured oscillation period is the same asthat of the charge density variation around the aminefunctional group.

IV. ULTRAFAST CONDENSED MATTERPHYSICS

1. Time delay of photoemission

Photoemission spectroscopy has been widely usedto study electronic properties of condensed matter.Quantum-mechanical calculation of electronic ground-state configurations and delocalized conduction bandstates have provided good understanding of the momen-tum and energy distribution of photoelectrons. How-ever, the dynamics of the photoemission process is notcorrectly considered in conventional models of photoe-mission in condensed matter.

The first attosecond time-resolved study [35] was con-ducted in tungsten by using the attosecond streakingmethod. It revealed that electrons coming from delocal-ized conduction-band states arrives ∼ 100 as earlier thanthose that originate from localized core states. In con-trast, in magnesium, photoelectrons that originate fromthe core and valence-band states arrived simultaneouslyat the surface, within the experimental uncertainty of±20 as [36]. Theoretical investigations suggest that theobserved delays are a result of electron propagation.

Recent attosecond time-resolved photoemission spec-troscopy [37] using the van der Waals crystal WSe2 showsthat the photoemission events are time-ordered with in-creasing initial-state angular momentum. During the ini-tial stage of the photoemission process, the excited-statedynamics is affected by the local environment (internal

-232- Journal of the Korean Physical Society, Vol. 73, No. 2, July 2018

Fig. 7. (Color online) Elementary steps in photoemissionfrom the van der Waals crystal WSe2. A 300-as long XUVpulse at a photon energy of 91 eV excite photoelectrons thatare then streaked in the IR field. The initial stage of photoe-mission is dominated by intra-atomic processes: the photo-electron wave (green) created by XUV excitation from the W4f state (blue) is governed by the effective radial potential(red) composed of the HS potential UHS and the centrifugalterm. Right panel: schematic of wave packet propagationin the later stage; it is dominated by a 1D potential thataccounts for the inner potential UIP of WSe2 and the inter-action with the remaining photohole. Reproduced from [37]with permission.

structure of the atom). Hence, the dependence of theemission on the angular moment due to intra-atomic in-teractions naturally arises (Fig. 7, top left). These effectshave been neglected in conventional models of photoe-mission from condensed matter. New modeling requiresthe inclusion of both intra-atomic delays and propaga-tion effects. (Fig. 7, top). The relative delays amongthe four photoemission channels measured could onlybe explained by considering both propagation and intra-atomic delays. Attosecond time-resolved photoemissionfrom condensed matter provides important benchmarks,which will lead to improved understanding of photoe-mission dynamics in ways that have never been exploredbefore.

2. Ultrafast tunneling and metallization

Isolated attosecond pulses have also been used to ex-plore tunneling processes in condensed matter. Such

(a) (b)

(c) (d)

Fig. 8. (Color online) (a) Normalized transferred chargeQP per pulse as a function of change in propagation lengthor the change in CEP (b) Measured maximum transferredcharge QP (CE = 0) as a function of laser intensity. (c) and(d) are the same as (a) and (b), respectively, but for quartz(blue) and calcium fluoride (green). Adopted from [40].

an experiment has been conducted in silicon [38]. Un-der laser intensities > 1010 W·cm−2 at a wavelength of800 nm, electrons from the valence band make transi-tions by the multiphoton process or tunnel ionizationto the conduction band through a band gap of 3.2 eV.The transmission of an attosecond XUV pulse througha 250 nm thick single-crystalline free-standing Si mem-brane has been measured as a function of the time delaywith respect to the 800 nm pump pulse. The tempo-ral evolution of the transmission measured at ∼ 100 eVshows a step-like behavior with a step of about 450 as,synchronized with the half-cycle period of the NIR elec-tric field. Each step also gives an upper limit for thecarrier-carrier scattering time. The observed evolution ofthe ionization process agrees with the expectation fromtunnel ionization. The experiment demonstrates thatelectron tunneling becomes dominant in silicon at laserintensities >∼ 1012 W·cm−2.

The ultrafast switching of conduction properties of amaterial by using optical pulses is very attractive fromthe technological point of view, because switching speedsof extant devices have stagnated due to the limits of cur-rent technology. A CEP-stabilized 800-nm laser lightwas used to demonstrate that the ac conductivity offused silica can be increased by > 18 orders of magni-tude within 1 fs, and that this change can be controlled[39]. Wider spread of this technology demands the bet-ter understanding of whether the strong field behavioris universally similar for different dielectrics. To addressthis question, a series of the experiments has been con-ducted for sapphire, calcium fluoride and quartz. The ex-

Extreme Metrology for Ultrafast Electron Dynamics at the Atomic Scale – Dong Eon Kim -233-

Fig. 9. (Color online) Schematic layout for the generation of synchronized TW as XFEL with E-SASE section and undulatortapering. Adopted from [56].

perimental results showed remarkably similar responses(Fig. 8) [40]. This similarity despite the distinct differ-ences in their physical properties, suggests the universal-ity of the physical picture explained by the localizationof Wannier-Stark states. These results may provide away to develop petahertz-rate optoelectronics.

V. ATTOSECOND XFEL

XFEL [41–43] sources have been developed in the lastdecade. They provide a new frontier in X-ray science dueto their remarkable peak power of 10 - 50 GW and pulseduration of a few to 100 fs.

Attosecond metrology based on HHG by few-cyclefemtosecond lasers [11] will be further enriched by the de-velopment of isolated attosecond XFEL pulses. XFELspresently achieve higher power and shorter wavelengthsthan HHG sources, and even a higher powers and shorterpulses are expected; these properties suggest that XFELpulses are considered as a future light source. The devel-opment of high-intensity attosecond X-ray sources willgreatly extend the variety of ultrafast processes that canbe explored. For example, single-molecule imaging to de-termine the structure of a molecule is only possible withan attosecond hard X-ray pulse [44]. Another immediateapplication of such a pulse is the observation of real-timechanges in the probability distribution of an electron’sposition [45–47]; such an observation would constitute agreat advance in time-resolved diffraction experiments,and enable four-dimensional imaging with picometer spa-tial resolution and attosecond temporal resolution [45].Such attosecond X-ray pulses would allow the investiga-tion of phenomena in ultrafast science and X-ray nonlin-ear science [48].

Therefore, research has been conducted to learn howto generate isolated attosecond XFEL pulses . Most ofthese works predict XFEL pulse radiations with powersat the gigawatt level (∼ 108 photons/pulse) or durationslonger than 100 as. Scientists desire to use an isolatedas XFEL with a pulse duration < 100 as and number ofphotons > 1010 (∼ terawatt power).

A suggested method [49] to generate an isolated ter-awatt attosecond XFEL pulse includes a slotted foil andE-SASE [50] section to generate many current spikes,with optical and electron beam delay units to delay X-ray pulses and electron beams. Variants of this methodhave also been proposed [51–55].

Although all of these proposals present innovative ap-proaches, they share a common disadvantage of usingoptical and electron delay units. Use of such units re-quires numerous elements, such as optical mirrors, fourbending magnets in a chicane, control systems for eachdelay unit. This complexity may cause instability anddifficulty in control of X-ray pulses with attosecond reso-lution. A question, “without any delay unit, can a singlecurrent spike produce terawatt as XFEL pulse?” thencomes naturally. All relevant complications involved forsolving this problem have been addressed [56], and theauthors proposed a simple scheme (Fig. 9) to producean isolated terawatt XFEL pulse without any optical orelectron delay units between undulator modules. Thisdesign represents a realistic solution to realize a ter-awatt attosecond XFEL, and also provides flexibility inthe change of XFEL operation mode from normal SASEmode to as XFEL mode. This scheme will open terawattattosecond XFEL as a new regime of operation.

VI. FUTURE PERSPECTIVE

The two decades, since the first demonstration of at-tosecond pulses produced by using HHG in gases, havewitnessed impressive progress in technological point ofview (attosecond metrology) as well as in scientific pointsof view (attosecond science). Routine generation of iso-lated attosecond pulses allows design of investigation ofreal-time observation and control of ultrafast electrondynamics in various forms of matter. We should utilizethis new ability to control and manipulate electrons inatoms, molecules and condensed matter, to address prob-lems that are of direct relevance to life and medicine, asset out by the BESAC report.

The current capability of attosecond metrology is lim-ited to the XUV spectral region (<∼ 100 eV). Consider-

-234- Journal of the Korean Physical Society, Vol. 73, No. 2, July 2018

ing current tremendous efforts toward development ofhigh-power mid-infrared (up to ∼ 10 µm) lasers, thegeneration of isolated attosecond pulses with photon en-ergy of few kiloelectronvolts (soft X-ray spectral region)at a repetition rate of kilohertz or less, is imminent.Along with this advance, attosecond XFEL will becomea unique and powerful tool for the hard x-ray (∼ 10 keV)spectral region. These resources will fling open a door toall the electron dynamics taking place in nature.

Novel tools with attosecond resolution and atomic res-olution called for by the BESAC report will then beavailable. Control of processes at the level of electronsand understanding of states far away from equilibriumwill also enable great increase in our understanding ofthe inner working-principles in all the matter, includingthe basic components of life, and their complexes. Suchknowledge will certainly help us to reveal complicatedelectron processes in highly complex systems. The nextdecade is expected to observe this development.

REFERENCES

[1] G. R. Fleming and M. A. Ratner, Physics Today Julyissue 61, 28 (2008).

[2] P. Krehl and S. Engemann, Shock Waves 5, 1 (1995).[3] H. Abraham and T. Lemoine, Compt. Rend. 129, 206

(1899).[4] A. H. Zewail, Femtochemistry vol I and II (Singapore:

World Scientific, 1994).[5] A. Zewail, J. Phys. Chem. A 104, 5660 (2000).[6] P. F. Moulton, J. Opt. Soc. Am. B 3, 125 (1986).[7] D. E. Spence, P. N. Kean and W. Sibbett, Opt. Lett. 16,

42 (1991).[8] P. Maine, D. Strickland, P. Bado, M. Pessot and G.

Mourou, IEEE J. Quantum Electron. 24, 398 (1988).[9] R. Szipocs, K. Ferencz, C. Spielmann and F. Krausz,

Opt. Lett. 19, 201 (1994).[10] T. W. Hansch, Rev. Mod. Phys. 78, 1297 (2006).[11] M. Hentschel et al., Nature 414, 509 (2001).[12] R. Kienberger et al., Nature 427, 817 (2004).[13] G. Sansone et al., Science 314, 443 (2006).[14] X. Feng, S. Gilbertson, H. Mashiko, H. Wang, S. D.

Khan, M. Chini, Y. Wu, K. Zhao and Z. Chang Z, Phys.Rev. Lett. 103, 183901 (2009).

[15] E. J. Takahashi, P. Lan, O. D. Mucke, Y. Nabekawa andK. Midorikawa, Nat. Commun. 4, 2691 (2013).

[16] A. Wirth et al., Science 334, 195 (2011).[17] L. V. Keldysh, Sov. Phys. JETP 20, 1307 (1965).[18] M. Uiberacker et al., Nature 446, 627 (2007).[19] A. S. Landsman, U. Keller, Physics Reports 547, 1

(2015).[20] A. I. Baz’, Sov. J. Nucl. Phys. 4, 182 (1967).[21] M. Buttiker and R. Landauer, Phys. Rev. Lett. 49, 1739

(1982).[22] E. P. Wigner, Phys. Rev. 98, 145 (1955).[23] E. Pollak and W. H. Miller, Phys. Rev. Lett. 53, 115

(1984).

[24] L. Gallmann, C. Cirelli and U. Keller, Ann., Rev. Phys.Chem. 63, 447 (2012).

[25] P. Eckle et al., Science 322, 1525 (2008).[26] P. Eckle et al., Nat. Phys. 4, 565 (2008).[27] T. Niemczyk et al., Nat. Phys. 6, 772 (2010).[28] P. Forn-Dıaz, J. Lisenfeld, D. Marcos, J. J. Carcıa-Ripoll,

E. Solano, C. J. P. M. Harmans and J. E. Mooij, Phys.Rev. Lett. 105, 237001 (2010).

[29] Z. Ficek, J. Jing and Z. G. Lu, Phys. Scr. T 140, 014005(2010).

[30] A. N. Pfeiffer and S. R. Leone, Phys. Rev. A 85, 053422(2012).

[31] M. Anand, S. Pabst, O. Kwon and D. E. Kim, Phys.Rev. A 95, 053420 (2017).

[32] L. S. Cederbaum and J. Zobeley, Chem Phys. Lett. 307,205 (1999).

[33] Lepine F, M. Y. Ivanov and M. J. J. Vrakking, Nat.Photon. 8, 195 (2015).

[34] F. Calegari et al., Science 346, 336 (2014).[35] A. L. Cavalieri et al., Nature 449, 1029 (2007).[36] S. Neppl et al., Phys. Rev. Lett. 109, 087401 (2012).[37] F. Siek, S. Neb, P. Bartz, M. Hensen, C. Struber,

S. Fiechter, M. Torrent-Sucarrat, V-M. Silkin, E-E.Krasovskii, N-M. Kabachnik, S. Fritzsche, R-D. Muino,P-M. Echenique, A-K. Kazansky, N. Muller, W. Pfeifferand U. Heinzmann, Science 357, 1274 (2017).

[38] M. Schultze et al., Science 346, 1348 (2014).[39] A. Schiffrin et al., Nature 493, 70 (2013).[40] O. Kwon, T. Paasch-Colberg, V. Apalkov, B-K. Kim, J-

J. Kim, M. I. Stockman and D. Kim, Sci. Reports. 6,21272 (2016).

[41] P. Emma et al., Nat. Photon. 4, 641 (2010).[42] T. Ishikawa et al., Nat. Photon. 6, 540 (2012).[43] H-S. Kang et al., Nat. Photon. 11, 708 (2017).[44] A. Fratalocchi and G. Ruocco, Phys. Rev. Lett. 106,

105504 (2011).[45] F. Krausz and M. Ivanov, Attosecond, physics Rev. Mod.

Phys. 81, 163 (2009).[46] G. Dixit, O. Vendrell and R. Santra, Proc. Natl Acad.

Sci. USA 109, 11636 (2012).[47] D. Cho, J. R. Rouxel, M. Kowalewski, J. Y. Lee and S.

Mukamel, J. Chem. Theory Comput. 14, 329 (2018).[48] M. Fuchs, M. Trigo, J. Chen and S. Ghimire, Nat. Phys.

11, 964 (2015).[49] T. Tanaka, Phys. Rev. Lett. 110, 084801 (2013).[50] A. A. Zholents, Phys. Rev. ST Accel. Beams 8, 040701

(2005).[51] E. Prat and S. Reiche, Phys. Rev. Lett. 114, 244801

(2015).[52] E. Prat, F. Lohl and S. Reiche, Phys. Rev. ST Accel.

Beams 18, 100701 (2015).[53] T. Tanaka, Y. W. Parc, Y. Kida, R. Kinjo, C. H. Shim,

I. S. Ko, B. Kim, D. E. Kim and E. Prat, J. SynchrotronRad. 23, 1273 (2016).

[54] Z. Wang, C. Feng and Z. Zhao, Phys. Rev. Accel. Beams20, 040701 (2017).

[55] S. Kumar, Y. W. Parc, A. S. Landsman and D. E. Kim,Sci. Rep. 6, 37700 (2016).

[56] C. H. Shim, Y. W. Parc, S. Kumar, I. S. Ko and D. E.Kim, Sci. Reports 8, 7463 (2018).