Ionization avalanching in clusters ignited by extreme-ultraviolet driven seed electrons
Bernd Schütte, Mathias Arbeiter, Alexandre Mermillod-Blondin, Marc J. J. Vrakking, Arnaud Rouzée, Thomas Fennel
aa r X i v : . [ phy s i c s . a t m - c l u s ] N ov Ionization avalanching in clusters ignited by extreme-ultraviolet driven seed electrons
Bernd Sch¨utte,
1, 2, ∗ Mathias Arbeiter, Alexandre Mermillod-Blondin, Marc J. J. Vrakking, Arnaud Rouz´ee, and Thomas Fennel † Max-Born-Institut, Max-Born-Strasse 2A, 12489 Berlin, Germany Department of Physics, Imperial College London,South Kensington Campus, SW7 2AZ London, United Kingdom Institute of Physics, University of Rostock, Albert-Einstein-Str. 23, 18059 Rostock, Germany (Dated: September 15, 2018)We study the ionization dynamics of Ar clusters exposed to ultrashort near-infrared (NIR) laserpulses for intensities well below the threshold at which tunnel ionization ignites nanoplasma for-mation. We find that the emission of highly charged ions up to Ar can be switched on withunit contrast by generating only a few seed electrons with an ultrashort extreme ultraviolet (XUV)pulse prior to the NIR field. Molecular dynamics simulations can explain the experimental obser-vations and predict a generic scenario where efficient heating via inverse bremsstrahlung and NIRavalanching is followed by resonant collective nanoplasma heating. The temporally and spatiallywell-controlled injection of the XUV seed electrons opens new routes for controlling avalanchingand heating phenomena in nanostructures and solids, with implications for both fundamental andapplied laser-matter science. PACS numbers: 32.80.Fb, 36.40.-c, 36.40.Gk, 52.50.Jm
The strong-field ionization dynamics of solid-densitymatter exposed to NIR laser fields with intensities nearthe ionization threshold is a fundamental and challeng-ing problem in laser-matter interactions. A process thatis generic for this scenario is the generation of nanoscaleplasmas via ionization avalanching, resulting from thestrongly non-linear intertwining of collisional absorptionand ionization. The quest for a quantitative micro-scopic understanding of this process is driven by its sig-nificance for a wealth of applications. These includelaser-based material processing [1], high harmonic gener-ation (HHG) [2–6], the realization of ultrafast lightwaveelectronics in dielectrics [7] and the generation of shockwaves [8, 9].In the last two decades, intense laser-cluster interac-tions have become an important platform for studyingnon-linear, collective, and correlated many-body pro-cesses in nanoplasmas for a wide spectral range fromthe NIR to the x-ray domain [10, 11]. A result thatreceived particular attention was the unexpected highlycharged ion emission following ionization of clusters byeither an NIR [12] or a vacuum-ultraviolet (VUV) laserpulse [13]. By now it is well known that at long wave-lengths, avalanching and transient resonant plasmon ex-citations are crucial to explain the observed highly ef-ficient light absorption [14, 15], the emission of highlycharged ions [12, 16–18] and the generation of fast elec-trons [19–22]. In clusters, the expansion of an initiallyoverdense nanoplasma enables particularly strong reso-nant laser energy absorption, once the frequency of thecollective electronic dipole mode ω mie ∝ √ ρ ion , the so-called Mie plasmon [23], equals the laser frequency. Here ρ ion is the ion charge density. While collision-mediatednanoplasma heating via inverse bremsstrahlung (IBS) was predicted to be important for excitation and ioniza-tion in the NIR [14] and VUV ranges [13, 24, 25], verticalsingle-photon excitation of bound electrons becomes thekey absorption mechanism in the XUV and in the x-rayregime [26, 27].An important concept for strong-field ionization dy-namics of rare-gas clusters is the so-called “ionizationignition”, proposed by Rose-Petruck et al. [28]. Therein,the intensity threshold for efficient nanoplasma genera-tion and heating was connected with the atomic ioniza-tion threshold, as the latter determines the seed electrongeneration required for avalanching. Direct evidence forthis picture was found in a z -scan experiment on Xe clus-ters in He nanodroplets [29], where the sudden appear-ance of highly charged Xe ions was observed near theintensity threshold for tunnel ionization (TI) of atomicXe. Evidence for the hypothesis that few seed electronsare sufficient for ignition was provided in an NIR few-cycle experiment on He nanodroplets, where weak dop-ing with less than 10 Xe atoms was found to saturate theHe emission at intensities well below the TI thresh-old of He [30, 31]. In all these experiments, however,strong-field ionization by the NIR field itself was used togenerated seed electrons.In this Letter, we demonstrate an alternative conceptthat allows one to completely decouple the seed elec-tron generation from the NIR driven ionization dynam-ics, both spatially and temporally. The main idea is toinject seed electrons via photoionization with a moder-ately intense ultrashort XUV pulse (2 × W/cm )produced by HHG. The low intensity of the XUV pulsedistinguishes our scenario from that of the theoreticalstudy in [32], where strong cluster ionization was consid-ered by a VUV pump pulse. We show that IBS heating,subsequent efficient NIR avalanching, and resonant exci-tation remain possible well below the TI threshold andcan be triggered by just a few seed electrons. This factis evidenced by the emission of highly charged ions upto Ar under conditions where no ion emission is ob-served without seeding. For NIR intensities as low as3 × W/cm we find ion emission up to Ar , thoughthe ponderomotive energy of U p = 170 meV is two ordersof magnitude below the ionization potential of atomic Ar.The observations are well reproduced by molecular dy-namics simulations. Our findings enable the study oflow-intensity IBS during the early phases of avalanchingand open a route to steer the spatial and temporal plasmaformation in solids, with implications for laser materialprocessing.For the experiments, we use a Ti:Sapphire laser ampli-fier delivering pulses at 790 nm and operating at 50 Hz.The maximum energy achievable is 35 mJ for a pulse du-ration of 32 fs [33]. The stretched laser beam is split bya beamsplitter into 2 parts that are individually com-pressed by two separate grating compressors. Up to32 mJ of the pulse energy are used for HHG by focusingone beam with a 5 m long focal length spherical mir-ror into a 15 cm long gas cell statically filled with Kr.The NIR light used in the HHG process is blocked bya 100 nm thick Al filter after a propagation distance of5 m. The second NIR laser pulse is collinearly overlappedwith the XUV beam using a mirror with a 6 mm centralhole. The maximum NIR pulse energy of 3 mJ can bereduced by the combination of a λ /2 waveplate and apolarizer. Both the XUV and NIR laser pulses are fo-cused onto a cluster beam at the center of a velocity mapimaging spectrometer [34] by using a spherical multilayermirror with a focal length of 75 mm. The cluster beamis generated by a piezoelectric valve operating at 10 Hzwith a 0.5 mm diameter nozzle, and is placed 7 cm awayfrom a 0.2 mm diameter skimmer. We estimate the aver-age cluster sizes from the Hagena scaling law [35]. Usingthe velocity map imaging spectrometer, we record the2D projections of the electron and ion momentum distri-butions resulting from ionization by the two-color fields,from which we obtain kinetic energy spectra by using astandard Abel inversion procedure [36].Ion time-of-flight (TOF) spectra resulting from the ion-ization of Ar N ( h N i = 3500) by an XUV pulse and bythe XUV-NIR pulse sequences are shown in Fig. 1(a).The XUV intensity of I = 2 × W/cm is two ordersof magnitude lower than in our previous studies [37–39].Ionization from the XUV pulse only (black curve) resultsin the observation of Ar + ions and larger ionic fragmentssuch as dimers and trimers. When adding an NIR pulsewith a peak intensity of 5 × W/cm ( U p = 2 . are generated. Remarkably, ion chargesup to Ar are still observed when the NIR peak inten-sity is reduced by more than one order of magnitude to (b) -2 -1 s i gna l ( a r b . un i t s ) kinetic energy (eV) XUV only XUV+NIR s i gna l ( a r b . un i t s ) kinetic energy (eV) atoms XUV only XUV+NIR (I
NIR =5x10 W/cm ) XUV+NIR (I NIR =3x10 W/cm ) Ar Ar Ar Ar Ar Ar Ar Ar + Ar Ar s i gna l ( a r b . un i t s ) time of flight (microseconds) (a) FIG. 1. (a) Ion TOF spectra from Ar clusters with an av-erage size of h N i = 3500 atoms ionized by an XUV pulseonly ( I XUV = 2 × W/cm ) and with an additional NIRpulse at a time delay of 5 ps ( I NIR = 5 × W/cm or I NIR = 3 × W/cm ). The different TOF spectra areplotted with a vertical offset. (b) The black curve shows theelectron spectrum from clusters obtained by the XUV pulseonly. An additional NIR pulse at a time delay of 600 fs and ata peak intensity of 5 × W/cm (orange curve) stronglyenhances the electron signal. The vertical axis has a loga-rithmic scale. The inset displays a photoelectron spectrumfor the ionization of atomic Ar, with the main contributionscoming from the 11th, 13th, 15th, 17th harmonics. × W/cm (blue curve), corresponding to a pon-deromotive potential of only 170 meV. Most importantly,no ion or electron signal from clusters was observed withthe NIR pulse only in both cases, underlining the highcontrast between seeded and unseeded excitation.Our observations can be explained by a modified ig-nition model (see Fig. 2(a)). As shown in the photo-electron spectrum of atomic Ar in the inset of Fig. 1(b),the XUV spectrum contains contributions from the 11th( ~ ω = 17 . I = 2 × W/cm , see step (1) in Fig. 2(a).In agreement with our earlier work [37] we find indica-tions for frustrated recombination [41] by observing veryslow meV electrons in Fig. 1(b). This suggests the on-set of nanoplasma formation even at the moderate XUVintensity applied, cf. step (2) in Fig. 2(a). Following ig-nition of the cluster ionization with the XUV pulse, theNIR laser pulse can interact with quasifree nanoplasmaelectrons and with electrons that are weakly bound byatoms, see step (3) in Fig. 2(a) and Ref. [42]. For the FIG. 2. (a) Scheme of the two-color cluster ionization pro-cesses, in which the cluster is ignited by a moderately in-tense XUV pulse in step (1), leading to the formation of ananoplasma (step (2)). Neutral atoms are shown in blue,ions in red and electrons in black. In step (3), a time-delayedNIR pulse initially interacts with quasifree electrons and elec-trons that are weakly bound by atoms. Due to IBS heating,avalanching and resonance effects, the cluster is strongly ion-ized (step (4)). (b) Schematic of the NIR and XUV spatialprofiles at the focus for Gaussian pulses. Since the XUV pulse(black) has a much smaller focus diameter than the NIR pulse(orange), focal volume averaging over different NIR intensi-ties is avoided in the experiment. It is furthermore possibleto restrict the ionization with NIR pulses to clusters where ananoplasma is formed (green area), i.e. to the region of thehighest XUV intensities. chosen NIR pulse duration of 1 ps, electrons trappedwithin the clusters can be efficiently heated, resulting inextensive avalanching and strong ionization, which trig-gers cluster expansion (step 4 in Fig. 2(a)). Strong NIRdriven ionization is also supported by the fact that theelectron signal from Ar clusters (Fig. 1(b)) is increasedby up to two orders of magnitude when adding the NIRpulse ( I = 5 × W/cm ).In recent x-ray laser-cluster experiments it was demon-strated that the effects of focal volume averaging and thecluster size distribution crucially affect the measured ionspectra and need to be eliminated to measure meaningfulion charge state distributions at high x-ray intensity [43].A major advantage of our scheme is that the volume av-eraging over the NIR laser focus can be avoided via thespatially selective seed electron generation. As the NIRbeam has a significantly larger focal spot size ( ≈ µ m)than the XUV beam ( ≈ µ m), seeded avalanching andnanoplasma formation is restricted to the center of theNIR beam where the NIR intensities are close to the peakvalue (see Fig. 2(b)). Furthermore, NIR-induced electronand ion emission is restricted to larger clusters that ex-perience the XUV peak intensity as the number of seedelectrons per cluster scales with the number of clusteratoms N and will drop below unity for small clusters /lower XUV intensity. The resulting selectivity makes thepresented scheme suitable for a detailed comparison ofexperiment and theory.In order to analyze the evolution of the ionizationand avalanching processes, we have performed semiclas-sical molecular dynamics simulations [44] for parameterssimilar to the measurements, see Fig. 3. In agreement [ e V ] i nne r i on i z a t i on pe r a t o m c l u s t e r r ad i u s [ A ]
12 2 a v g . c ha r ge pe r a t o m a)c) b) ab s o r bed ene r g y pe r a t o m [ k e V ] NIR pulseXUV pulse distribution of effective charge state W/cm ii c ha r ge pe r a t o m XUV seed impactionization
12 2
12 2
12 2 m i e eff r e c o m b i na t i on W/cm W/cm -1 -2 -3 FIG. 3. Molecular dynamics simulations on Ar . A 30 fsXUV seed pulse at ~ ω = 20 eV ( I XUV = 2 . × W / cm )is followed by a 1 ps NIR pulse (delayed by ≈ h q ii i ,and predicted final effective charge state h q eff i by taking intoaccount electron-ion recombination (gray area). Insets showcorresponding simulated final effective ion charge spectra. (b)Temporal evolution of inner ionization (solid curves) and root-mean-square cluster radius (dashed curves); the inset showsthe evolution of inner ionization during the seeding step andthe subsequent avalanching process for the highest intensityon a logarithmic scale. The black dashed curve correspondsto an exponential growth. (c) Dynamics of the predicted Miefrequency ~ ω mie of the nanoplasma (solid curves) and evolu-tion of the total energy absorption (dashed curves). The grayareas indicate the intensity envelopes of the XUV and NIRfields. The dashed-dotted horizontal line corresponds to theNIR photon energy. with our experimental results, the predicted final ioncharge spectra show high ionization stages up to Ar (Fig. 3(a)) for the seeded case. The average final ioncharge states increase rapidly up to NIR intensities ofabout 4 × W/cm before they begin to level out.The stages of the seeded avalanching process and thehighly charged ion generation can be understood from thetime-resolved analysis of the simulations in Figs. 3(b)-(c). After the generation of only a few seed electrons,an exponential increase of the inner ionization (electronsremoved from host atoms), h q ii i , in the leading edge ofthe NIR pulse documents the self-amplified character of Intensity (10 W/cm ) C ha r ge s t a t e
321 100110
FIG. 4. Ion charge-state distributions from Ar clusters with h N i = 18000 atoms at different NIR intensities. An XUVpulse ( I = 2 × W/cm ) preceeds the NIR pulse by5 ps. The average charge state increases as a function ofthe NIR intensity and is peaked at Ar for an intensity of3 × W/cm . The ion signal is shown in a logarithmicscale. the avalanche, resulting in the generation of (quasi)freeelectrons in each optical half cycle that contribute to thegeneration of even more electrons in the next cycles viaimpact ionization (see inset of Fig. 3(b)). Note that val-ues of h q ii i > ω NIR = ω mie ismet twice. While the ionization driven first resonance isunimportant for the total energy absorption, the expan-sion induced resonance largely governs the total energyabsorption [11]. The fact that the expansion driven reso-nance occurs late in the trailing edge of the NIR pulse forthe lowest intensity explains the stronger effect of recom-bination in this case (cf. Fig. 3(a)), since recombinationis more efficient for lower electron temperature [45].The simulations show a fast increase of the ionizationstates as function of the NIR laser intensity and indi-cate the onset of saturation for intensities well belowthe TI threshold for Ar. This behavior is also observedin the experimental ion charge spectra shown for differ-ent intensities in Fig. 4. We can furthermore concludefrom the simulations that, under the given experimen-tal conditions, single-pulse excitation, both NIR-only aswell as XUV-only, does not significantly ionize the sys-tem ( q XUVavg < .
01 and q NIRavg = 0), highlighting the syner-getic action of XUV induced seeding and long-wavelengthdriven avalanching and nanoplasma heating.As multiphoton ionization is avoided in our experi- ment, we can conclude that the efficient absorption oflaser energy in early stages, i.e. far away from the plas-mon resonance condition, can only be explained by in-verse bremsstrahlung (IBS), in agreement with our simu-lations. Therefore, our experimental results confirm thatIBS is efficient in rare-gas clusters at ponderomotive po-tentials on the order of 100 meV. This shows that IBSmay indeed play an important role in the ionization ofclusters with intense VUV pulses, where similar pondero-motive potentials were applied [13, 24, 25]. In the future,our ignition method is expected to enable the quantita-tive study of IBS at low ponderomotive potentials, atlow plasma densities, and as function of wavelength. Itshould be applicable even in the VUV range, as long asVUV single-photon ionization can be excluded by choos-ing sufficiently small VUV photon energies.Our concept of cluster ignition using an ultrashortXUV pulse is very versatile and is expected to be ap-plicable in a wide range of laser parameters to studyand control fundamental processes of light-matter inter-actions. Using attosecond XUV pulses would allow theinvestigation of the plasma dynamics on a sub-NIR-cycletimescale. In the context of HHG in solids [2–6], ourapproach could be used to temporally control the HHGprocess in order to better understand and possibly avoidthe undesired damaging effects of plasma formation. Ap-plied to the surface ablation of semiconductors and di-electrics, an irradiation sequence composed of an (X)UVlaser pulse followed by an NIR laser pulse offers a highermicromachining precision and a higher efficiency thanconventional irradiation schemes involving a single coloror existing dual beam methods [46–49]. First, the inter-action footprint is determined by the size of the (X)UVspot only, leading to a dramatic improvement of the spa-tial precision. Furthermore, the overall efficiency of theprocess (i.e. the ratio of the energy required to evaporatethe volume removed to the energy contained in the laserelectromagnetic field) may significantly increase becausethe use of a low-intensity (X)UV pulse allows one to ap-ply an NIR pulse with a much lower intensity than inconventional schemes.In summary, we have reported on a novel concept toignite NIR-driven ionization of solid-density matter bygenerating seed electrons using an ultrashort XUV pulse.High charges up to Ar were generated from clusterswith an NIR pulse at an intensity of only 3 × W/cm ,i.e. far below the tunnel ionization threshold. From acombination of experimental and theoretical studies, wecould conclude that IBS and avalanching play a key rolefor the initial charging of the clusters, which is stronglyenhanced at the plasmon resonance, leading to the ob-served high charge states. Our results demonstrate forthe first time that inverse bremsstrahlung plays an im-portant role in rare-gas clusters interacting with intenselaser pulses, which have a ponderomotive potential as lowas 100 meV.We would like to thank M. Yu. Ivanov for fruitful dis-cussions. B. S. is grateful for funding from the DFG viaa research fellowship. M.A. and T.F. gratefully acknowl-edge financial support from the DFG within SFB652 andcomputer time provided by the North-German Super-computing Alliance (HLRN, project mvp00010). ∗ [email protected] † [email protected][1] L. Englert, M. Wollenhaupt, L. Haag, C. Sarpe-Tudoran,B. Rethfeld, and T. Baumert, Appl. Phys. A , 749(2008).[2] S. Ghimire, A. D. DiChiara, E. Sistrunk, P. Agostini,L. F. DiMauro, and D. A. Reis, Nat. Phys. , 138 (2011).[3] O. Schubert et al. , Nat. Photon. , 119 (2014).[4] M. Hohenleutner, F. Langer, O. Schubert, M. Knorr,U. Huttner, S. W. Koch, M. Kira, and R. Huber, Nature(London) , 572 (2015).[5] T. T. Luu, M. Garg, S. Y. Kruchinin, A. Moulet, M. T.Hassan, and E. Goulielmakis, Nature (London) , 498(2015).[6] G. Vampa, T. J. Hammond, N. Thir´e, B. E. Schmidt,F. L´egar´e, C. R. McDonald, T. Brabec, and P. B.Corkum, Nature (London) , 462 (2015).[7] M. Schultze et al. , Nature (London) , 75 (2013).[8] T. Pezeril, G. Saini, D. Veysset, S. Kooi, P. Fidkowski,R. Radovitzky, and K. A. Nelson, Phys. Rev. Lett. ,214503 (2011).[9] R. Ecault, L. Berthe, M. Boustie, F. Touchard, E. Le-scoute, A. Sollier, P. Mercier, and J. Benier, J. Phys. D , 235501 (2013).[10] U. Saalmann, C. Siedschlag, and J. Rost, J. Phys. B ,R39 (2006).[11] T. Fennel, K.-H. Meiwes-Broer, J. Tiggesb¨aumker, P.-G.Reinhard, P. M. Dinh, and E. Suraud, Rev. Mod. Phys. , 1793 (2010).[12] E. M. Snyder, S. A. Buzza, and A. W. Castleman, Phys.Rev. Lett. , 3347 (1996).[13] H. Wabnitz et al. , Nature (London) , 482 (2002).[14] T. Ditmire, T. Donnelly, A. M. Rubenchik, R. W. Fal-cone, and M. D. Perry, Phys. Rev. A , 3379 (1996).[15] J. Zweiback, T. Ditmire, and M. D. Perry, Phys. Rev. A , R3166 (1999).[16] L. K¨oller, M. Schumacher, J. K¨ohn, S. Teuber,J. Tiggesb¨aumker, and K. H. Meiwes-Broer, Phys. Rev.Lett. , 3783 (1999).[17] S. Zamith, T. Martchenko, Y. Ni, S. A. Aseyev, H. G.Muller, and M. J. J. Vrakking, Phys. Rev. A ,011201(R) (2004).[18] T. D¨oppner, T. Fennel, T. Diederich, J. Tiggesb¨aumker,and K. H. Meiwes-Broer, Phys. Rev. Lett. , 013401(2005).[19] Y. L. Shao, T. Ditmire, J. W. G. Tisch, E. Springate,J. P. Marangos, and M. H. R. Hutchinson, Phys. Rev.Lett. , 3343 (1996).[20] E. Springate, S. A. Aseyev, S. Zamith, and M. J. J. Vrakking, Phys. Rev. A , 053201 (2003).[21] U. Saalmann and J. M. Rost, Phys. Rev. Lett. ,133006 (2008).[22] T. Fennel, T. D¨oppner, J. Passig, C. H. Schaal,J. Tiggesb¨aumker, and K.-H. Meiwes-Broer, Phys. Rev.Lett. , 143401 (2007).[23] G. Mie, Ann. Phys. (Leipzig) , 377 (1908).[24] R. Santra and C. H. Greene, Phys. Rev. Lett. , 233401(2003).[25] C. Siedschlag and J.-M. Rost, Phys. Rev. Lett. ,043402 (2004).[26] C. Bostedt et al. , Phys. Rev. Lett. , 133401 (2008).[27] M. Arbeiter and T. Fennel, Phys. Rev. A , 013201(2010).[28] C. Rose-Petruck, K. J. Schafer, K. R. Wilson, andC. P. J. Barty, Phys. Rev. A , 1182 (1997).[29] T. D¨oppner, J. P. M¨uller, A. Przystawik, S. G¨ode,J. Tiggesb¨aumker, K.-H. Meiwes-Broer, C. Varin, L. Ra-munno, T. Brabec, and T. Fennel, Phys. Rev. Lett. ,053401 (2010).[30] A. Mikaberidze, U. Saalmann, and J. M. Rost, Phys.Rev. Lett. , 128102 (2009).[31] S. R. Krishnan et al. , Phys. Rev. Lett. , 173402(2011).[32] C. Siedschlag and J. M. Rost, Phys. Rev. A ,031401(R) (2005).[33] G. Gademann, F. Ple, P.-M. Paul, and M. J. J. Vrakking,Opt. Express , 24922 (2011).[34] A. T. J. B. Eppink and D. H. Parker, Rev. Sci. Inst. ,3477 (1997).[35] O. F. Hagena and W. Obert, J. Chem. Phys. , 1793(1972).[36] M. Vrakking, Rev. Sci. Instrum. , 4084 (2001).[37] B. Sch¨utte, M. Arbeiter, T. Fennel, M. J. J. Vrakking,and A. Rouz´ee, Phys. Rev. Lett. , 073003 (2014).[38] B. Sch¨utte, F. Campi, M. Arbeiter, T. Fennel, M. J. J.Vrakking, and A. Rouz´ee, Phys. Rev. Lett. , 253401(2014).[39] B. Sch¨utte, T. Oelze, M. Krikunova, M. Arbeiter, T. Fen-nel, M. J. J. Vrakking, and A. Rouz´ee, New J. Phys. ,033043 (2015).[40] W. F. Chan, G. Cooper, X. Guo, G. R. Burton, andC. E. Brion, Phys. Rev. A , 149 (1992).[41] T. Fennel, L. Ramunno, and T. Brabec, Phys. Rev. Lett. , 233401 (2007).[42] B. Sch¨utte, T. Oelze, M. Krikunova, M. Arbeiter, T. Fen-nel, M. J. J. Vrakking, and A. Rouz´ee, J. Phys. B (2015).[43] T. Gorkhover et al. , Phys. Rev. Lett. , 245005 (2012).[44] M. Arbeiter, C. Peltz, and T. Fennel, Phys. Rev. A ,043428 (2014).[45] Y. Hahn, Phys. Lett. A , 82 (1997).[46] F. Th´eberge and S. Chin, Appl. Phys. A , 1505 (2005).[47] S. Zoppel, J. Zehetner, and G. A. Reider, Appl. Surf.Sci. , 7692 (2007).[48] C.-H. Lin, Z.-H. Rao, L. Jiang, W.-J. Tsai, P.-H. Wu,C.-W. Chien, S.-J. Chen, and H.-L. Tsai, Opt. Lett. ,2490 (2010).[49] D. Grojo, M. Gertsvolf, S. Lei, T. Barillot, D. M. Rayner,and P. B. Corkum, Phys. Rev. B81