Generating Ultrabroadband Deep-UV Radiation and Sub-10 nm Gap

Jun 17, 2019 - We experimentally investigate the interaction between hybrid-morphology gold optical antennas and a few-cycle Ti:sapphire laser up to ...
0 downloads 0 Views 5MB Size
Download PDF
Letter Cite This: Nano Lett. 2019, 19, 4779−4786

pubs.acs.org/NanoLett

Generating Ultrabroadband Deep-UV Radiation and Sub-10 nm Gap by Hybrid-Morphology Gold Antennas Liping Shi,*,†,‡,@ Jose ́ R. C. Andrade,*,†,‡,@ Ayhan Tajalli,†,‡ Jiao Geng,*,† Juemin Yi,§ Torsten Heidenblut,‡,∥ Frans B. Segerink,⊥ Ihar Babushkin,†,‡ Maria Kholodtsova,# Hamed Merdji,# Bert Bastiaens,∇ Uwe Morgner,†,‡ and Milutin Kovacev*,†,‡ †

Institute of Quantum Optics, Leibniz University Hannover, Welfengarten 1, 30167, Hannover, Germany Cluster of Excellence PhoenixD (Photonics, Optics, and Engineering-Innovation Across Disciplines), 30167, Hannover, Germany § Institute of Physics and Center of Interface Science, Carl von Ossietzky University Oldenburg, 26129, Oldenburg, Germany ∥ Institute of Materials Science, Leibniz University Hannover, An der University 2, 30823, Garbsen, Hannover Germany ⊥ Optical Sciences, MESA+ Institute for Nanotechnology, University of Twente, P.O. Box 217, 7500AE Enschede, The Netherlands # LIDYL, CEA, CNRS, Universite Paris-Saclay, CEA Saclay 91191, Gif-sur-Yvette, France ∇ Laser Physics and Nonlinear Optics, MESA+ Institute for Nanotechnology, University of Twente, 7500AE Enschede, The Netherlands

Downloaded via UNIV OF SOUTHERN INDIANA on July 17, 2019 at 10:12:05 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

‡

S Supporting Information *

ABSTRACT: We experimentally investigate the interaction between hybrid-morphology gold optical antennas and a fewcycle Ti:sapphire laser up to ablative intensities, demonstrating rich nonlinear plasmonic effects and promising applications in coherent frequency upconversion and nanofabrication technology. The two-dimensional array of hybrid antennas consists of elliptical apertures combined with bowties in its minor axis. The plasmonic resonance frequency of the bowties is red-shifted with respect to the laser central frequency and thus mainly enhances the third harmonic spectrum at long wavelengths. The gold film between two neighboring elliptical apertures forms an hourglass-shaped structure, which acts as a “plasmonic lens” and thus strongly reinforces surface currents into a small area. This enhanced surface current produces a rotating magnetic field that deeply penetrates into the substrate. At resonant frequency, the magnetic field is further intensified by the bowties. The resonant frequency of the hourglass is blueshifted with respect to the laser central frequency. Consequently, it spectacularly extends the third harmonic spectrum toward short wavelengths. The resultant third harmonic signal ranges from 230 to 300 nm, much broader than the emission from a sapphire crystal. In addition, the concentration of surface current within the neck of the hourglass antenna results in a structural modification through laser ablation, producing sub-10 nm sharp metallic gaps. Moreover, after laser illumination the optical field hotspots are imprinted around the antennas, allowing us to confirm the subwavelength enhancement of the electric near-field intensity. KEYWORDS: Nonlinear plasmonics, broadband deep-ultraviolet, laser ablation, near-field mapping, nanoscale gaps

S

So far, a myriad of techniques have been shown for the realization of coherent deep-ultraviolet light where the usual method requires the use of a large bandwidth infrared laser and subsequent nonlinear interaction to upconvert the light to the 200−300 nm range. Sum-frequency of the fundamental and second harmonic fields using nonlinear optical crystals,8,9 direct third-harmonic generation (THG),10,11 four-wavemixing,12,13 gas-filled photonic crystal fiber,14 or even direct lasing in the deep-ultraviolet15 have all been presented as mechanisms to access this range, but few have shown capabilities of achieving ultrabroadband radiation supporting

pectroscopy and microscopy tools enable us to peer into the rich structure of matter and allow us to better understand our world. The more readily available these tools are, the more research and novel knowledge is created. Laserbased light sources for spectroscopy and microscopy can be found in virtually every region of the electromagnetic spectrum. However, some frequency bands are harder to reach than others, limiting the potential for new discoveries in those regions. Deep-ultraviolet (200−300 nm) coherent sources would find themselves used in probing electronic transitions of molecules and solids, high-temporal resolution pump−probe photoelectron spectroscopy, photoelectron emission microscopy, steering of ultrafast electron dynamics in the valence shell of molecules, solids, nanoparticles, clusters, and so forth.1−7 © 2019 American Chemical Society

Received: May 23, 2019 Revised: June 11, 2019 Published: June 17, 2019 4779

DOI: 10.1021/acs.nanolett.9b02100 Nano Lett. 2019, 19, 4779−4786

Letter

Nano Letters

Figure 1. (a) Schematic illustration of third-harmonic generation from a plasmonic metasurface. Inset: Laser spectrum (red curve) and retrieved phase (black curve) from the dispersion-scan measurement. (b) SEM image of several unit cells of the metasurface. Scale bar: 400 nm. Dashed box covers an hourglass-shaped structure formed between neighboring ellipses. (c) Artistic rendering of a unit cell, where E⃗ ,j,⃗ and H⃗ denote vectorial electric field, surface current, and magnetic field, respectively. (d) Polarization analysis of the far-field THG enhanced by the antennas, here 0° denotes y-polarization. Measured THG spectra of the laser source in dependence of glass insertion for the sapphire substrate (e) and the plasmonic metasurface (f). The optimal insertions of glass wedges are indicated by solid lines. (g) Spectrally resolved enhancement factor of THG emission from the nanoantennas with respect to the bare substrate.

generation in the deep-ultraviolet region, where the conventional nonlinear crystals reach their limits both in transparency and in phase-matching conditions.30 Over the past decade, various nanostructures, such as commonly used rod- or bowtie-shaped plasmonic antennas or apertures,21,31−33 three-dimensional plasmonic nanocup,34 alldielectric nanodisks with higher thermal damage threshold compared to plasmonic ones,35 and metal−dielectric hybrid nanoantennas36 have been widely employed to boost the generation efficiency of harmonics in visible and deepultraviolet spectral regime. Recently, Semmlinger et al. have reported the first use of zinc oxide nanoresonators to generate coherent vacuum-ultraviolet radiation.37 Nevertheless, metasurface-enhanced harmonic generation of spectra supporting few-cycle pulses is largely unexplored, as the bandwidth of these nanoresonators is usually not sufficient to cover the entire spectrum of few-cycle pulses.23,31,32,38,39 Therefore, nanoantennas exhibiting high-order resonant modes are of interest to broaden the spectrum of harmonics toward shortwavelength.40 However, the high-order plasmonic resonances decay very fast, limiting the near-field electric enhancement as well as far-field spectral investigations. Another path to a wider response bandwidth would be by the simultaneous excitation of two orthogonal dipole modes, as recently demonstrated.33 However, the free-standing rectangular plasmonic nanoholes for this purpose produce elliptical electromagnetic radiation. In this Letter, we report on a hybrid plasmonic nanostructure for the generation of linearly polarized ultrabroadband deep-ultraviolet coherent light. Our work relies on direct THG, which is driven by an electric dipole resonance allied to an enhanced magnetic field. This way we are able to engineer the response of our generation medium to drastically broaden the THG spectrum of a few-cycle Ti:sapphire

few-cycle pulses in the deep-ultraviolet, especially if the source has low fluence. This is mainly due to phase-mismatches and the fact that although the third harmonic process typically broadens the spectrum, the ratio between bandwidth and central frequency is smaller when compared to that one of the driving field. Ideally one would want a process which would directly convert each individual frequency component to the new frequency region, that is, broadening the spectrum by a factor of three as well. Below we show how a tailored antenna response approaches this scenario. The nonlinear response of natural materials is intrinsically low. Therefore, usually one needs to employ thick samples and complex phase-matching conditions to enhance the efficiency of nonlinear frequency conversion. Recently, thanks to the development of nanotechnology, it has been demonstrated that some quasi-two-dimensional artificial materials such as metallic and all-dielectric metasurfaces can significantly boost nonlinear optical effects in a subwavelength volume, using low fluence lasers at high-repetition rates.16 Metallic nanoantennas being resonantly driven experience a collective oscillation of the conduction electrons on the surface, that is, localized surface plasmons. This behavior results in an enhancement of the absorption, scattering, and electric near-field strength of the impinging light in the vicinity of the antennas. 17−19 Consequently, by fabricating a densely packed array of antennas, a plasmonic metasurface, one can significantly increase the nonlinear frequency conversion efficiency without the limitation of phase-matching.20−29 This is rather attractive as one can extend the wavelength range of compact laser sources that are used in applications where the light flux needs to be spread over several pulses, for example, photoelectronbased imaging and time-of-flight spectroscopy. Especially, the plasmonic nanoantennas are promising for coherent light 4780

DOI: 10.1021/acs.nanolett.9b02100 Nano Lett. 2019, 19, 4779−4786

Letter

Nano Letters

Figure 2. (a) FDTD-based simulation (blue curve) and measurement (red curve) of linear transmission spectrum of the metasurface. Insets: vectorial electric field map along the tip-to-tip of a bowtie in the y−z plane, excited by the monochromatic plane wave at 600 nm (left inset) and at 1170 nm (right inset). (b) FEM-based simulation of intensity enhancement factor of electric field in the center of the bowtie gap and the magnetic field surrounding the hourglass-antenna. λ0 = 840 nm denotes the laser central wavelength. The surface current distribution (c) and electric intensity map (d) at the air−gold interface as well as in the x−z plane (e) along the waist of hourglass-antenna. The excitation source is a monochromatic plane wave at 1170 nm, which is polarized along the y-direction and illuminates the sample from substrate to gold-film.

oscillator. The challenge is to perfectly match the plasmonic resonance with the laser spectrum so that the THG bandwidth can be efficiently broadened. We show that the generated 15 nm wide (fwhm) deep-ultraviolet radiation from the sapphire substrate is effectively broadened to 40 nm by the plasmonic structures, hence supporting 3.1 fs laser pulses. In addition, we introduce an alternative experimental approach for an intuitive and qualitative observation of the plasmonic phase response and its effects. Moreover, we demonstrate here that irradiation of our prestructured antennas around critical intensity enables the production of sub-10 nm metallic gaps, without lateral thermal damage to the gold film. Finally, using the few-cycle light source to irradiate the plasmonic metasurface, we find a high-resolution method to visualize the nanoscale distribution of the near-field enhancement not only around the gold surface but also inside the gold film. This direct imprinting of the nearfield enhancement is beneficial to better understand the nonlinear plasmonic effects, such as field-driven photoemission, harmonic generation, and self-organized reshaping of nanoantennas in strong femtosecond fields. Figure 1a schematically illustrates the experimental setup. Pulses with a width of 7.6 fs from a Ti:sapphire oscillator with a repetition rate of 100 MHz and pulse energy of 1 nJ are tightly focused onto either a plasmonic metasurface or its sapphire substrate. The laser spectrum spans from 650 to 1000 nm with a central wavelength of 840 nm (see inset of Figure 1a). Figure 1b shows the scanning electron microscopy (SEM) image of a representive area of the metasurface, which includes an array of elliptical nanoapertures combined with a pair of tipto-tip triangles (bowtie) on its minor axis. The semimajor (xaxis) and semiminor (y-axis) axis of the ellipse are 190 and 130 nm, respectively. The radius of curvature at the bowtie tips, and the gap size are both 20 nm. The gold film between neighboring elliptical apertures forms an hourglass-shaped nanostructure with a waist of ∼30 nm, as indicated by the dashed box in Figure 1b. Figure 1c schematically illustrates a unit cell of our metasurface. Broadband double-chirped mirrors and a pair of fused silica wedges are employed to control the

dispersion and to measure the fundamental pulse characteristics based on the dispersion scan technique (d-scan).41−44 In d-scan, one measures the nonlinear signal of the test pulse as a function of introduced (known) dispersion, for example, insertion of glass wedges. Feeding both the measured nonlinear signal trace and an independently measured fundamental pulse spectrum to an optimization algorithm, the phase of the fundamental pulse can be retrieved. Figure 1e,f shows the measured THG spectra of the laser source independent of glass insertion for the sapphire substrate and the plasmonic metasurface, respectively. In order to avoid sample degradation, the incident peak intensity is set to ∼0.05 TW/cm2 (see Supporting Information). The upper left inset in Figure 1a depicts the measured spectrum of the laser source together with the retrieved spectral phase at the optimum glass insertion (marked by the solid line in Figure 1e) using the THG d-scan technique. The corresponding retrieved pulse duration of the fundamental pulse is ∼7.6 fs. We find that the optimal wedge insertion to achieve the strongest THG emission through the plasmonic structure (marked by the solid line in Figure 1f) differs from that for the bare substrate. The phase variation is equivalent to the introduction of a 200 μm of fused silica. This is attributed to an additional phase shift induced by the plasmons.45−48 Therefore, an additional negative dispersion compensation is required for the nanostructures. Compared to the THG emission from the bare sapphire substrate, which is about 15 nm wide (fwhm), the plasmonic structures obviously generate a nearly three times broader spectrum of about 40 nm (fwhm), which supports few cycle pulses down to only 3.1 fs. For a better visualization, we plot the spectrally resolved THG enhancement factor at the wedge insertion points for two traces where the integrated harmonic photon counts across the entire spectral range (230−300 nm) are the highest. As shown in Figure 1g, the THG enhancement factor is at its minimum in the vicinity of the laser central wavelength. The harmonic enhancement factor peaks at around 240 nm. In addition, a deep-ultraviolet polarizer is employed to analyze the polar4781

DOI: 10.1021/acs.nanolett.9b02100 Nano Lett. 2019, 19, 4779−4786

Letter

Nano Letters ization state of the far-field THG (Figure 1d). The measurement demonstrates that the broadened THG is linearly polarized along the y-axis. We further investigate the linear response of the nanostructures. Using a FDTD-based open source MEEP (see Methods), we numerically simulate the linear transmission spectrum ranging from 500 to 2000 nm (blue curve in Figure 2a). Two transmission peaks are observed in the visible (∼600 nm) and near-infrared (∼1170 nm) regime. In order to gain physical insight into these resonances, the vectorial map of the electric field in the y−z plane (cross-sectional view along the tip-to-tip of a bowtie) is calculated using finite-elementmethod (JCMwave) and shown in the insets of Figure 2a. For the resonance at 600 nm, the field distribution (left inset in Figure 2a) shows that charges are accumulated at the air−gold and substrate−gold interfaces. The electric field in the z-axis exhibits the profile of a standing wave with antinodes at each interface. This field distribution is usually ascribed to Fabry− Perot resonance.49 Regarding the resonance at ∼1170 nm, the vectorial electric field map diplays that the induced current mainly flows through the bowties (see Figure 2c), indicating that collective charges accumulate at the bowtie tips. The plasmons emit back into free space as photons at the tips, leading to optical transmission enhancement. This dipole-like plasmonic resonance gives rise to the strong field enhancement in the air gap region [see right inset in Figure 2a,d,e]. The dashed curve in Figure 2b shows the finite element method (FEM)-based calculation (see Methods) electric field enhancement factor in the middle of the bowtie gap at various wavelengths. Indeed, we find that the electric field peaks at around the plasmonic resonance wavelegnth. Using the near-infrared laser as an in situ light source, we measure the linear transmission of the metasurface when the laser’s electric field is polarized along the bowtie’s direction (yaxis), as shown in Figure 2a (red curve). In the laser spectral range, the measured transmission spectrum is in agreement with the simulated one. The deep valley at around 870 nm is located in the gap between the plasmonic and Fabry−Perot resonance, where the field amplification is relatively weak. Therefore, the enhancement of THG corresponding to the laser central wavelength (λ0 = 840 nm, indicated by the arrow in Figure 2b) is rather low, whereas the enhancement factor of THG above 290 nm [see Figure 1g] starts to increase due to the plasmonic resonance centered at ∼1170 nm. This redshift of plasmonic resonance with respect to the laser central frequency broadens the THG spectrum toward long wavelengths. In addition, we note a small transmission dip appearing at ∼720 nm; meanwhile, the magnetic field is strongly enhanced in this spectral area [red curve, Figure 2b]. The surface current density distribution shows that at this incident light frequency the induced current is tightly reinforced into the hourglass structure (Figure 3a). This current generates a significantly localized rotating magnetic field (Figure 3b) and thus the electric field (Figure 3c) that deeply penetrates into the substrate, shown as cross-sectional profiles in x − z plane.50−52 Near the resonant frequency, the magnetic field is further intensified by the bowties (see Supporting Information). The penetration of electromagnetic fields into the substrate results in the enhancement of THG emission from the substrate crystal. In addition, instead of field confinement in the air gap, now the electric field is evidently enhanced surrounding the arms of the hourglass-antenna, as well as on the surface of its waist position (Figure 3d). Because

Figure 3. Surface current distribution (a) and electric intensity map (d) at air−gold interface. The excitation source is a monochromatic plane wave at 720 nm, which is polarized along the y-direction and illuminates the sample from substrate to gold-film. Vectorial magnetic field (b) and electric field maps (c) in the x−z plane along the waist of hourglass-antenna. The dashed box denotes the boundary of gold. SEM images of pristine nanostructure (e) after long-term exposure by intense laser pulses with a peak intensity of 0.08 TW/cm2 (f) and 0.15 TW/cm2. (g) The white arrows in (g) indicate the deposition of material also surrounding the hourglass-antenna. Scale bars: 20 nm. (h) Two-dimensional carbon map of the exposed antenna in (f).

of the very high intrinsic nonlinearty of gold,53 this strong field amplification on the antenna’s surface further explains the significant broadening of the THG spectrum toward shortwavelength. In order to qualitatively verify the subwavelength electric field enhancement in experiments, we carry out a long-term (30 min) laser exposure to modify the antennas’ morphology via strong-field effects [see Figure S1b in Supporting Information]. The incident peak intensity is 0.6 TW/cm2, however the Gaussian distribution of the laser spatial intensity allows us to investigate the exposed antennas at different intensities. A high-resolution SEM image of the illuminated antenna (0.08 TW/cm2) is displayed in Figure 3f. Compared to the pristine nanoantenna (Figure 3e), we clearly observe a thin film deposit in the bowtie gap. Using energy-dispersive Xray (EDX) spectroscopy, we confirm that the dominant element of this deposit is carbon (Figure 3h). Carbon may stem from the dissociation of hydrocarbon molecules that are inevitably adsorbed on the Au surface during the sample growing and milling processes or from the ambient environment. When interacting with an ionizing radiation source such as the intense laser and the generated deep-ultraviolet light, these molecules dissociate and produce carbon atoms. This dissociation rate depends on the photon flux; therefore, most of the carbon atoms will stay in the field-enhanced regions, that is, optical hotspots.54 When increasing the laser intensity from 0.08 to 0.15 TW/cm2, carbon deposits start to appear also around the arms of the hourglass-antenna (Figure 3g). This 4782

DOI: 10.1021/acs.nanolett.9b02100 Nano Lett. 2019, 19, 4779−4786

Letter

Nano Letters

Figure 4. (a) FDTD-based simulation of the electric intensity enhancement at the air−gold interface of the metasurface. The excitation source is a Fourier-limited 7.6 fs pulse centered at 840 nm, which is polarized along the y-axis and propagates from substrate to gold-film. SEM images of two antennas after exposure in a peak intensity of 0.6 TW/cm2 (b,c), and four antennas irradiated by the intensity of 0.4 TW/cm2 (d,j−l). The arrows in the SEM images are guides for the eye, indicating where the materials are deposited near the laser-produced small gap. (e) Simulation of electric near-field strength of antennas with a small laser-produced gap (6 nm) in the left-side arm. The illumination source is the same as that in (a). The material deposition seems to follow the electromagnetic field isolines which provides more contrast to the SEM images. There exists a qualitative correspondence between the simulated near-field maps and the contrast patterns in the SEM images. (f) Spectra of THG emission from the pristine metasurface (red curve) and from the ablated metasurface (blue area). (g−i) Two-dimensional EDX maps of Au, Al, and C of the illuminated antenna shown in (d).

our FIB (20 nm). It should be pointed out that this ablationinduced gap formation is reproducible. Figure 4 (j−l) shows three examples, in which sub-10 nm gaps are observed. On plasmonic metasurfaces, the collective electron oscillations are transferred to hot carriers via electron−electron scattering, which results in heating of the metallic lattice by means of ohmic losses. Given that the plasmons usually exhibit an inhomogeneous distribution on the metal surface, localized thermal hotspots will appear in the metal where the surface current density is maximal.55−59 Analogous to the optical lens, the hourglass-like structure acts as a nanoscale plasmonic lens,

indicates that the surface current has oscillated through the hourglass-antenna. In order to confirm the field enhancement inside gold, we image two representative antennas illuminated by a peak intensity of 0.6 TW/cm2 (Figure 4b,c) which show that the hourglass structure is strongly damaged. When controlling the laser intensity to be 0.4 TW/cm2, the damage becomes gentle. A self-organized, smooth, and narrow metallic gap is produced without lateral damage to the rest of the hourglass-film. Using high-resolution SEM, the gap size is measured to be 6 nm (Figure 4d), surpassing the gap limit that can be obtained by 4783

DOI: 10.1021/acs.nanolett.9b02100 Nano Lett. 2019, 19, 4779−4786

Letter

Nano Letters

estimated to be ∼10 V/nm,68,69 which indicates that the incident field strength (∼1 V/nm) at the bowtie tips is enhanced by at least 1 order of magnitude. This enhancement factor is confirmed by the numerical calculation (Figure 4e). We also investigate the impact of the laser-induced gap in the hourglass-shaped gold film on THG spectrum. With the disruption of the gold film at the waist position of the hourglass-antenna, the conditions for magnetic field enhancement vanish. The hourglass structure is modified to another bowtie-like antenna. As mentioned, the laser simultaneously illuminates 200 unit cells of the plasmonic metasurface. A portion of the hourglass-shaped nanostructures morph to small gaps. Consequently, as shown in Figure 4f, THG enhanced by the hourglass is drastically weakened; conversely, THG enhanced by the bowtie is further boosted by the laserproduced gaps. In conclusion, we have demonstrated the use of a hybrid plasmonic metasurface to efficiently broaden the THG spectrum of a broadband Ti:sapphire oscillator, obtaining a nanoscale ultrabroad source in the deep-ultraviolet spectral region. We have also shown a strong-field-based method to directly visualize the shape of the electric near-field both in and surrounding the gold nanoantennas. Moreover, we have reproducibly fabricated sub-10 nm plasmonic gaps via plasmon-enhanced near-field ablation mechanisms. Some effects such as the Purcell mechanism70 and single-molecule strong coupling71 only occur in sub-10 nm gaps, as the field enhancement factor in the gap region exponentially increases with the decrease of the gap size. Therefore, reproducible, reliable, and precise fabrication of a sharp and small space between metallic structures is essential. The general fabrication technique such as electron beam lithography and Ga+-FIB can easily obtain lateral feature sizes between 10 and 20 nm. However, it is challenging to fabricate sub-10 nm metallic gaps by these standard processes. The method in this Letter to fabricate ultrasmall gaps in gold film can be applied in nonlinear plasmonics, electrically driven optical antennas, and molecular fluoresence enhancement. Methods. Numerical Simulations. We perform a set of simulations based on both finite difference time domain (FDTD, open source MEEP) and finite element method (JCMsuite) to investigate the linear spectral response, electromagnetic field enhancement factor, electric near-field maps, vectorial magnetic field maps, and surface current density. To model the structures, we adopt gold dielectric permittivity values obtained empirically by Johnson and Christy,72 and refractive index of sapphire substrate obtained by Malitson.73 The geometry of the antenna for numerical simulations is very close to the realistic conditions. A 50 nm thick gold film is first set on a sapphire substrate. Elliptical apertures combined with a pair of tip-to-tip equilateral triangles (bowtie) on its minor axis are defined in the gold film. The semimajor (x-axis) and semiminor (y-axis) axis of the ellipse are set to 190 and 130 nm, respectively. The radius of curvature at the bowtie tips, and the gap size are both 20 nm. The finite element method is more adapted to accurately describe complex geometries and to represent the near-field maps. Therefore, the enhancement of electromagnetic fields at resonant frequencies [inset of Figure 1a, Figure 2b,d,e, Figure 3b−d] and surface current density [Figure 2c and Figure 3a] are calculated by finite element method. The incident beams are monochromatic plane waves, which illuminate substrateside first and polarize along the bowtie orientation. The

leading to a very strong current, and thus a thermal hotspot in its waist position. When the incident laser is intense enough, gold in the thermal hotspot experiences a phase change above the critical point. Because of the variation of surface tension at high temperatures, melting and subsequent boiling cause thermal damage. Besides thermal effects, nonthermal processes may also contribute to the formation of such a small gap. As shown in Figure 4a, in the neck of hourglass-antenna, the peak intensity of incident pulses is enhanced by a factor of β ≈ 40. Therefore, we can esimate the mean kinetic energy of hot carriers:60−62 16 πβI t Te ≈ 3 λ n0 = 2.6 eV, where I0 = 0.4 TW/cm2 and t = 7.6 fs 0 e

denote the laser intensity and pulse duration, respectively, and ne = 5.9 × 1028 m−3 is the free electron density of gold. This energy approaches the work function of gold in air.63,64 This leads to a possible ablation mechanism where a portion of hot electrons escape from the surface before scattering and depositing their energy to the lattice. Subsequently, electrostatic interaction between the freed electrons and the positively charged surface can result in removal of ions from the gold film.60,61 In addition, the sudden injection of a very dense plasmonic current into a small area may kick some atoms out of the surface via collisions, a similar process as electromigration that is usually induced by strong direct current.65 Thermal damage is a rather volatile and strong ablation process, which leads to the unwanted lateral damage of the gold film; in contrast, these nonthermal processes are more gentle and reproducible, which in principle have the potential capability of moving atoms one by one when properly controlled.62,65 Therefore, we consider that the formation of a small gap in the neck of hourglass-antenna is more likely induced by the nonthermal ablation mechanisms. Nevertheless, the full description of the underlying physics is still an open question and further testing of the dynamics of the formation of the nanogap is required. In addition, we carry out energy dispersive X-ray spectroscopy (EDS) to verify the formation of a nanometer-scale gap. The two-dimensional EDS maps of the exposed structure are shown in Figure 4g−i, which display the elemental distribution of Au, Al, and C, respectively. Indeed, we find that Au is ablated from the laser-produced gap (Figure 4g). The strong signal of Al from the substrate is thus observed (Figure 4h). Carbon atoms also accumulate in the opening gap, an identical process to that of the bowtie gap as mentioned above (Figure 4i). The bright rims of the material deposit appear so due to the SEM edge effect. It is rather impressive to see that the concave- and convex-shaped outlines are clearly observable in the bowtie- and laser-produced-gap [see Figure 4j−l], respectively. These features seem to follow the outlines of the simulated field map (Figure 4e). The outlines of material deposit may reflect isolines of electric near-field strength. Moreover, the rims of the material deposited in the laserablated gap appear brighter than those of the FIB-milled bowtie gap in the high-resolution SEM image. This confirms that the former exhibits a higher amount of material deposit, confirming a stronger field-enhancement factor due to its smaller gap distance. We also observe that the gap size of the FIB-milled bowtie increases to 30 nm after long-term illumination. The radius of curvature at the tips also becomes larger. These structural changes are due to evaporation of gold ions through the plasmonically enhanced localized field at the surface layer.66,67 The threshold for this phenomenon is 4784

DOI: 10.1021/acs.nanolett.9b02100 Nano Lett. 2019, 19, 4779−4786

Nano Letters

■

periodic boundary condition in the x−y plane and the transparent boundary in the z-axis are adopted, respectively. The surface current density is extracted from the electric field maps on the gold surface via the following equation: j=⃗ −iω Im(ϵAu)E⃗ , where ω, ϵAu, and E⃗ denote the light frequency, gold dielectric constant, and electric field strength, respectively. FDTD is more adapted to spectroscopic studies when a large number of wavelengths is of interest. Therefore, the linear transmission spectrum [Figure 2a], evolution of intensity enhancement on the surface of pristine antenna [Figure 4a], and electric field enhancement of laser ablated antenna [Figure 4f] are calculated by FDTD. We model the nanostructure with cubic elements of 2 × 2 × 2 nm dimensions. The incident laser pulse is assumed to be transform-limited, having an 840 nm central wavelength with a 7.6 fs pulse duration. The timedomain computation was implemented with an incremental step of 0.0066 fs. Periodic boundary conditions in x − y plane and perfectly matched layer in z-axis are adopted, respectively. During the spectral transmission simulations, after the source has turned off, it keeps running for an additional 50 time units until the square amplitude has decayed by 10−15 from its peak to ensure that the Fourier transforms have converged. For the laser ablated hourglass structure, the air gap is set to 6 nm. Sample Fabrication, SEM Images, and EDS Analysis. Using focused ion beam (at Optical Sciences, MESA+Institute for Nanotechnology, University of Twente in The Netherlands), nanoantennas are fabricated by patterning a 50 nm Au film deposited on a sapphire crystal. The thickness of the sapphire substrate is 500 μm. The nanostructures are arranged in square arrays of A 10 μm × 10 μm. The SEM images were performed by focused ion beam-scanning electron microscopes (FIB-SEM, Zeiss Auriga). The EDS element maps were performed on the same device, equipped with an Oxford XMax80-detector. Setups for Linear Transmission and Nonlinear Harmonic Generation Measurements. To efficiently detect the generated deep-ultraviolet radiation, a confocal monochromator setup (modified McPherson 234/302, 1200 lines/mm concave grating for 110−310 nm) is used to collect the generated THG in transmission geometry, where the generation volume acts as an entrance slit. The concave grating in combination with a photomultiplier tube (Hamamatsu H8259-09) and photon counter (Scientific Research SR400) capable of single-photon counting is used to collect and measure the transmitted third harmonic signal. The incident peak intensity on the metasurface is estimated to be 0.6 TW/cm2 with a focal beam diameter of 7 μm.The focal diameter is evaluated by THG z-scan method. The variations of laser intensity during experiments are also based on z-scan. A deep-ultraviolet polarizer [Thorlabs, GLB10-UV - Glan-Laser alpha-BBO Polarizer, 10 mm CA, UV Coating (220−370 nm)] is installed directly after the sample to analyze the polarization state of generated THG. A broadband Ti:sapphire femtosecond oscillator is utilized as an in situ light source to enable linear transmission measurement in the applied spectral region. Two neutral filters are used to attenuate the power by 5 orders of magnitude. Two fused silica lenses are used to collect and focus the transmitted laser beam on a fiber spectrometer (Ocean Optics Maya 2000). The transmission of the laser beam from a large gold-free area is defined as a reference signal.

Letter

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.9b02100.

■

Additional figures and information (PDF)

AUTHOR INFORMATION

Corresponding Authors

*E-mail: *E-mail: *E-mail: *E-mail:

[email protected]. [email protected]. [email protected]. [email protected].

ORCID

Liping Shi: 0000-0002-1998-6805 Juemin Yi: 0000-0003-3239-676X Author Contributions @

L.S. and J.R.C.A. contributed equally.

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors thank funding supports from Deutsche Forschungsgemeinschaft (DFG) (KO 3798/4-1) and from German Research Foundation under Germany’s Excellence Strategy EXC-2123 and Germany’s Excellence Strategy within the Cluster of Excellence PhoenixD (EXC 2122, Project ID 390833453), Lower Saxony through “Quanten und Nanometrologie” (QUANOMET, Project Nanophotonik). H.M. acknowledges support from the PETACom FET Open H2020, support from the French ministry of research through the ANR grants 2014 “IPEX”, 2017 “PACHA”, the DGA RAPID grant “SWIM”, the LABEX “PALM” (ANR-10-LABX-0039-PALM) through he grants “Plasmon-X”, “STAMPS”, and “HILAC”. We acknowledge financial support from the French ASTRE program through the “NanoLight” grant.

■

REFERENCES

(1) Altmann, R.; Galtier, S.; Dreissen, L.; Eikema, K. Phys. Rev. Lett. 2016, 117, 173201. (2) Altmann, R.; Dreissen, L.; Salumbides, E.; Ubachs, W.; Eikema, K. Phys. Rev. Lett. 2018, 120, 043204. (3) McFerran, J.; Yi, L.; Mejri, S.; Di Manno, S.; Zhang, W.; Guéna, J.; Le Coq, Y.; Bize, S. Phys. Rev. Lett. 2012, 108, 183004. (4) Ozaki, Y.; Kawata, S. Far-and Deep-ultraviolet spectroscopy; Springer, 2015. (5) Kubota, Y.; Watanabe, K.; Tsuda, O.; Taniguchi, T. Science 2007, 317, 932−934. (6) Popmintchev, D.; Hernández-García, C.; Dollar, F.; Mancuso, C.; Pérez-Hernández, J. A.; Chen, M.-C.; Hankla, A.; Gao, X.; Shim, B.; Gaeta, A. L.; et al. Science 2015, 350, 1225−1231. (7) Cingöz, A.; Yost, D. C.; Allison, T. K.; Ruehl, A.; Fermann, M. E.; Hartl, I.; Ye, J. Nature 2012, 482, 68. (8) Baum, P.; Lochbrunner, S.; Riedle, E. Opt. Lett. 2004, 29, 1686− 1688. (9) Zhao, S.; Gong, P.; Bai, L.; Xu, X.; Zhang, S.; Sun, Z.; Lin, Z.; Hong, M.; Chen, C.; Luo, J. Nat. Commun. 2014, 5, 4019. (10) Trushin, S. A.; Kosma, K.; Fuß, W.; Schmid, W. E. Opt. Lett. 2007, 32, 2432−2434. (11) Seres, J.; Seres, E.; Serrat, C.; Schumm, T. Opt. Express 2018, 26, 21900−21909. (12) Fuji, T.; Horio, T.; Suzuki, T. Opt. Lett. 2007, 32, 2481−2483. 4785

DOI: 10.1021/acs.nanolett.9b02100 Nano Lett. 2019, 19, 4779−4786

Letter

Nano Letters

Crespo, H. M.; Martens, R. L.; Nagy, T. IEEE J. Sel. Top. Quantum Electron. 2019, 25, 1−7. (44) Escoto, E.; Tajalli, A.; Nagy, T.; Steinmeyer, G. J. Opt. Soc. Am. B 2018, 35, 8−19. (45) Yi, J.-M.; Smirnov, V.; Piao, X.; Hong, J.; Kollmann, H.; Silies, M.; Wang, W.; Groß, P.; Vogelgesang, R.; Park, e. a.; Namkyoo. ACS Nano 2016, 10, 475−483. (46) Yi, J.-M.; Hou, D.; Kollmann, H.; Smirnov, V.; Pápa, Z.; Dombi, P.; Silies, M.; Lienau, C. ACS Photonics 2017, 4, 347−354. (47) Brinks, D.; Castro-Lopez, M.; Hildner, R.; van Hulst, N. F. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 18386−18390. (48) Accanto, N.; Piatkowski, L.; Renger, J.; van Hulst, N. F. Nano Lett. 2014, 14, 4078−4082. (49) Guo, H.; Meyrath, T. P.; Zentgraf, T.; Liu, N.; Fu, L.; Schweizer, H.; Giessen, H. Opt. Express 2008, 16, 7756−7766. (50) Grosjean, T.; Mivelle, M.; Baida, F.; Burr, G.; Fischer, U. Nano Lett. 2011, 11, 1009−1013. (51) Zhou, N.; Kinzel, E. C.; Xu, X. Opt. Lett. 2011, 36, 2764−2766. (52) Wang, D.; Yang, T.; Crozier, K. B. Opt. Express 2010, 18, 10388−10394. (53) Wolf, D.; Schumacher, T.; Lippitz, M. Nat. Commun. 2016, 7, 10361. (54) Shi, L.; Nicolas, R.; Andrade, J. R.; Boutu, W.; Franz, D.; Heidenblut, T.; Reinhardt, C.; Morgner, U.; Merdji, H.; Kovacev, M. ACS Photonics 2018, 5, 1208−1214. (55) Valev, V. K.; Denkova, D.; Zheng, X.; Kuznetsov, A.; et al. Adv. Mater. 2012, 24, OP29−OP35. (56) Denkova, D.; Verellen, N.; Silhanek, A. V.; Valev, V. K.; Dorpe, P. V.; Moshchalkov, V. V. ACS Nano 2013, 7, 3168−3176. (57) Valev, V.; Silhanek, A.; Jeyaram, Y.; Denkova, D.; De Clercq, B.; Petkov, V.; Zheng, X.; Volskiy, V.; Gillijns, W.; Vandenbosch, e. a.; et al. Phys. Rev. Lett. 2011, 106, 226803. (58) Coppens, Z. J.; Li, W.; Walker, D. G.; Valentine, J. G. Nano Lett. 2013, 13, 1023−1028. (59) Baffou, G.; Girard, C.; Quidant, R. Phys. Rev. Lett. 2010, 104, 136805. (60) Gamaly, E. G.; Rode, A. V.; Luther-Davies, B.; Tikhonchuk, V. T. Phys. Plasmas 2002, 9, 949−957. (61) Gamaly, E. G.; Rode, A. V.; Tikhonchuk, V. T.; Luther-Davies, B. Appl. Surf. Sci. 2002, 197, 699−704. (62) Gamaly, E. G. Phys. Rep. 2011, 508, 91−243. (63) Iwaszczuk, K.; Zalkovskij, M.; Strikwerda, A. C.; Jepsen, P. U. Optica 2015, 2, 116−123. (64) Kern, J.; Kullock, R.; Prangsma, J.; Emmerling, M.; Kamp, M.; Hecht, B. Nat. Photonics 2015, 9, 582. (65) Baumans, X. D.; Lombardo, J.; Brisbois, J.; Shaw, G.; Zharinov, V. S.; He, G.; Yu, H.; Yuan, J.; Zhu, B.; Jin, K.; Kramer, R. B. G.; Vondel, J. V. d.; Silhanek, A. V. Small 2017, 13, 1700384. (66) Shi, L.; Iwan, B.; Ripault, Q.; Andrade, J. R.; Han, S.; Kim, H.; Boutu, W.; Franz, D.; Nicolas, R.; Heidenblut, T. e. a.; et al. Phys. Rev. Appl. 2018, 9, 024001. (67) Herrmann, L. O.; Valev, V. K.; Tserkezis, C.; Barnard, J. S.; Kasera, S.; Scherman, O. A.; Aizpurua, J.; Baumberg, J. J. Nat. Commun. 2014, 5, 4568. (68) Plech, A.; Kotaidis, V.; Lorenc, M.; Boneberg, J. Nat. Phys. 2006, 2, 44. (69) Plech, A.; Leiderer, P.; Boneberg, J. Laser Photonics Rev. 2009, 3, 435−451. (70) Chikkaraddy, R.; de Nijs, B.; Benz, F.; Barrow, S. J.; Scherman, O. A.; Rosta, E.; Demetriadou, A.; Fox, P.; Hess, O.; Baumberg, J. J. Nature 2016, 535, 127. (71) Kinkhabwala, A.; Yu, Z.; Fan, S.; Avlasevich, Y.; Müllen, K.; Moerner, W. Nat. Photonics 2009, 3, 654. (72) Johnson, P. B.; Christy, R.-W. Phys. Rev. B 1972, 6, 4370. (73) Malitson, I. H. J. Opt. Soc. Am. 1962, 52, 1377−1379.

(13) Peters, E.; Diddams, S. A.; Fendel, P.; Reinhardt, S.; Hänsch, T.; Udem, T. Opt. Express 2009, 17, 9183−9190. (14) Köttig, F.; Tani, F.; Biersach, C. M.; Travers, J. C.; Russell, P. S. J. Optica 2017, 4, 1272−1276. (15) Ewing, J. IEEE J. Sel. Top. Quantum Electron. 2000, 6, 1061− 1071. (16) Krasnok, A.; Tymchenko, M.; Alu, A. Mater. Today 2018, 21, 8−21. (17) Muehlschlegel, P.; Eisler, H.-J.; Martin, O. J.; Hecht, B.; Pohl, D. Science 2005, 308, 1607−1609. (18) Kauranen, M.; Zayats, A. V. Nat. Photonics 2012, 6, 737. (19) Novotny, L.; Van Hulst, N. Nat. Photonics 2011, 5, 83. (20) Lippitz, M.; van Dijk, M. A.; Orrit, M. Nano Lett. 2005, 5, 799− 802. (21) Aouani, H.; Rahmani, M.; Navarro-Cía, M.; Maier, S. A. Nat. Nanotechnol. 2014, 9, 290. (22) Hajisalem, G.; Nezami, M. S.; Gordon, R. Nano Lett. 2014, 14, 6651−6654. (23) Hentschel, M.; Utikal, T.; Giessen, H.; Lippitz, M. Nano Lett. 2012, 12, 3778−3782. (24) Metzger, B.; Hentschel, M.; Schumacher, T.; Lippitz, M.; Ye, X.; Murray, C. B.; Knabe, B.; Buse, K.; Giessen, H. Nano Lett. 2014, 14, 2867−2872. (25) Lassiter, J. B.; Chen, X.; Liu, X.; Ciracì, C.; Hoang, T. B.; Larouche, S.; Oh, S.-H.; Mikkelsen, M. H.; Smith, D. R. ACS Photonics 2014, 1, 1212−1217. (26) Metzger, B.; Schumacher, T.; Hentschel, M.; Lippitz, M.; Giessen, H. ACS Photonics 2014, 1, 471−476. (27) Schwartz, O.; Oron, D. Nano Lett. 2009, 9, 4093−4097. (28) Sartorello, G.; Olivier, N.; Zhang, J.; Yue, W.; Gosztola, D. J.; Wiederrecht, G. P.; Wurtz, G.; Zayats, A. V. ACS Photonics 2016, 3, 1517−1522. (29) Mesch, M.; Metzger, B.; Hentschel, M.; Giessen, H. Nano Lett. 2016, 16, 3155−3159. (30) Ahmadivand, A.; Semmlinger, M.; Dong, L.; Gerislioglu, B.; Nordlander, P.; Halas, N. J. Nano Lett. 2019, 19, 605−611. (31) Hanke, T.; Krauss, G.; Träutlein, D.; Wild, B.; Bratschitsch, R.; Leitenstorfer, A. Phys. Rev. Lett. 2009, 103, 257404. (32) Hanke, T.; Cesar, J.; Knittel, V.; Trügler, A.; Hohenester, U.; Leitenstorfer, A.; Bratschitsch, R. Nano Lett. 2012, 12, 992−996. (33) Shi, L.; Andrade, J. R.; Yi, J.-M.; Marinskas, M.; Reinhardt, C.; Almeida, E.; Morgner, U.; Kovacev, M. ACS Photonics 2019, 6, 858− 863. (34) Zhang, Y.; Grady, N. K.; Ayala-Orozco, C.; Halas, N. J. Nano Lett. 2011, 11, 5519−5523. (35) Shcherbakov, M. R.; Neshev, D. N.; Hopkins, B.; Shorokhov, A. S.; Staude, I.; Melik-Gaykazyan, E. V.; Decker, M.; Ezhov, A. A.; Miroshnichenko, A. E.; Brener, I.; Fdeyanin, A. A.; Kivshar, Y. S. Nano Lett. 2014, 14, 6488−6492. (36) Shibanuma, T.; Grinblat, G.; Albella, P.; Maier, S. A. Nano Lett. 2017, 17, 2647−2651. (37) Semmlinger, M.; Tseng, M. L.; Yang, J.; Zhang, M.; Zhang, C.; Tsai, W.-Y.; Tsai, D. P.; Nordlander, P.; Halas, N. J. Nano Lett. 2018, 18, 5738−5743. (38) Schraml, K.; Regler, A.; Bartl, J.; Glashagen, G.; Wierzbowski, J.; Finley, J. J.; Kaniber, M. Optica 2016, 3, 1453−1459. (39) Shi, L.; Iwan, B.; Nicolas, R.; Ripault, Q.; Andrade, J. R.; Han, S.; Kim, H.; Boutu, W.; Franz, D.; Heidenblut, e. a.; Torsten. Optica 2017, 4, 1038−1043. (40) Melentiev, P. N.; Afanasiev, A. E.; Kuzin, A. A.; Gusev, V. M.; Kompanets, O. N.; Esenaliev, R. O.; Balykin, V. I. Nano Lett. 2016, 16, 1138−1142. (41) Miranda, M.; Fordell, T.; Arnold, C.; LHuillier, A.; Crespo, H. Opt. Express 2012, 20, 688−697. (42) Tajalli, A.; Chanteau, B.; Kretschmar, M.; Kurz, H. G.; Zuber, D.; Kovačev, M.; Morgner, U.; Nagy, T. Opt. Lett. 2016, 41, 5246− 5249. (43) Tajalli, A.; Ouille, M.; Vernier, A.; Bohle, F.; Escoto, E.; Kleinert, S.; Romero, R.; Csontos, J.; Morgner, U.; Steinmeyer, G.; 4786

DOI: 10.1021/acs.nanolett.9b02100 Nano Lett. 2019, 19, 4779−4786