Direct Visualization of Near-Fields in Nanoplasmonics and

May 17, 2012 - Success has been made in developing methods to probe the effect of their presence, but it remains difficult to directly image optically...
0 downloads 0 Views 340KB Size
Letter pubs.acs.org/NanoLett

Direct Visualization of Near-Fields in Nanoplasmonics and Nanophotonics Aycan Yurtsever and Ahmed H. Zewail* Physical Biology Center for Ultrafast Science and Technology Arthur Amos Noyes Laboratory of Chemical Physics California Institute of Technology, Pasadena, California 91125, United States ABSTRACT: Electric fields of nanoscale particles are fundamental to our understanding of nanoplasmonics and nanophotonics. Success has been made in developing methods to probe the effect of their presence, but it remains difficult to directly image optically induced electric fields at the nanoscale and especially when ensembles of particles are involved. Here, using ultrafast electron microscopy, we report the space-time visualization of photon-induced electric fields for ensembles of silver nanoparticles having different sizes, shapes, and separations. The high-field-of-view measurements enable parallel processing of many particles in the ensemble with high throughput of information. Directly in the image, the evanescent fields are observed and visualized when the particles are polarized with the optical excitation. Because the particle size is smaller than the wavelength of light, the near-fields are those of nanoplasmonics and are precursors of far-field nanophotonics. The reported results pave the way for quantitative studies of fields in ensembles of complex morphologies with the nanoparticles being embedded or interfacial. KEYWORDS: plasmonics, near-fields, photonics, electron imaging, PINEM, EELS Localized nanoscale electric fields that are induced optically exhibit unique phenomena of fundamental importance to nanoplasmonics and are the precursors for electromagnetic light in nanophotonics.1 In recent years, they have been considered for efficient photovoltaic2 and light-harvesting devices,3 single molecule detection,4 biomolecular labeling and manipulation, and surface-enhanced Raman scattering.5 Several techniques have been utilized for probing the fields, and these include optical methods,6−10 cathodoluminesce,11 and transmission electron microscopy.12−14 The latter probes the fields with nanometer and subnanometer resolution but the measured quantity is the dielectric response of the system to an impinging swift electron and not to a photon excitation. Moreover, the energy resolution is limited by the near 1 eV range characteristic of conventional electron microscopes; it is inferior to the resolution of optical methods even when the resolution can be improved to hundred(s) of millielectronvolts in some cases.15 Near-field optical techniques, on the other hand, can measure electric fields that are excited by a well-defined photon energy, but the spatial resolution of optical microscopes are diffraction limited, typically in the (sub)micrometer range, although recent advances have enabled detection with (sub)hundred nanometer resolution under special circumstances.16 This optical-based resolution is still far below the atomic resolution capability of electron microscopy. More relevant here is the fact that optical near-field techniques must be made at the interface to a vacuum and thus do not probe “buried fields” such as those produced by embedded particles and used in, for example, solar-cell energy conversion designs. © 2012 American Chemical Society

In this Letter, electron pulses of femtosecond (fs) duration synchronized with fs optical pulses are utilized to image nearfields of nanoparticle ensembles, thus reaching the spatial and energy resolution characteristics of electron and optical microscopy, respectively, and with the temporal resolution being on the scale of the fields generated, the femtosecond domain. These fields are evanescent, oscillating in time, and have nanoscale decay length, but they are the source of the electromagnetic fields (photonics) in the far-field (Figure 1). The results presented display the fields for the entire ensemble visualized at once and with different polarizations of the initiating photon field. The fields change depending on the particle size, shape, and separation. We discuss the theoretical basis for the behavior in space and for different polarizations. Electron Imaging of Fields. In free space, photons and electrons do not interact effectively because of momentum mismatch. Following an inelastic scattering between a photon and an electron, the change in the electron’s longitudinal momentum can be expressed as Δp = pf − pi = (2m(Ei ± nℏω))1/2 − (2mEi)1/2, where ℏω is the energy of the laser photon, Ei is the initial energy of the electrons, pi and pf are the initial and final momentum, respectively, and m is the mass of the electron; n = 0, 1, 2,... is an integer reflecting the number of photons, and ± denotes the energy gain (“absorption”) or loss (“emission”) of photons by the electron. We ignore the relativistic effects of the swift electron for the sake of simplicity; these effects do not alter the picture, as was recently shown in Received: May 1, 2012 Published: May 17, 2012 3334

dx.doi.org/10.1021/nl301643k | Nano Lett. 2012, 12, 3334−3338

Nano Letters

Letter

IEELS(x , y) =

e ℏω

−∞

∫+∞

Ezelectron(x , y , z)dz

(1)

Here, e is the unit charge and Eelectron is the z-component z (parallel to electron’s trajectory) of the electron-induced field. We note that IEELS(x,y) in this expression is an integral over z and varies in value in the x−y plane of the image; in energyfiltered images, it is these x−y changes that are monitored. The field is a function of the material property and its form can be deduced from the solution of Maxwell’s (or Poisson’s) equations. However, a simplified classical picture considering the motion of a relativistic electron (its field becomes a “pancake-looking” instead of a spherical field for nonrelativistic moving electron) near the material gives the simple (uncertainty) relationship b ≅ vΔt ≅ v/ω, where b is the impact parameter, ℏω is the energy filtered to record images, and Δt is the transit time of the electron in the induced field of the material; b is “resolution parameter” and the larger the ℏω (e.g., core-loss energy) the smaller b and better the resolution,17 as in this picture the field simply scales with e−(ω/v)x, and similarly for the y-direction. If the swift electron is 10 nm away from the point of interest, for instance, the field has a transit time that is much less than a femtosecond. This implies that many frequencies and energies are inherently present in these fields and hence the inelastic scattering at a wide range of energies. In photon-induced near-field electron microscopy (PINEM),18 a nanoparticle (having a length d) is excited with a photon (d ≪ λ) which creates the evanescent near-field that turns into electromagnetic waves in the far-field. In our experiments, an ultrashort optical pulse induces the near-field and simultaneously an ultrashort electron pulse images the field by utilizing the inelastic scattering of the ultrafast electron with the field’s longitudinal component, that is, the component parallel to the trajectory of the electron acceleration. A quantized energy exchange between the ultrafast electron and induced field takes place and gives rise to a well-resolved gain peaks in the energy spectrum.18 This is entirely different from the case of EELS, where the field is induced by the swift electron and the loss spectrum is determined by internal excitations in the material. In this three-body (electron, photon, and nanostructure) interaction picture, the energy of the ultrafast electron is quantized because of the presence of a time-oscillating electric field, the evanescent field in the vicinity of the nanoparticle. The electron’s initial energy, Ei = pi2/2m, now takes the form of Ei ± nℏω after the scattering. The extend of the needed spatial localization in the longitudinal direction for momentum conservation can be estimated by considering the uncertainty relationship between Δz and Δp . For the condition mentioned above of Δp = ℏω/v to hold, a confinement with a full-widthat-half-maximum of Δz = 2π/ω in real space is needed; the distribution of both Δz and Δp is taken to be Gaussian. This gives Δz to be less than 350 nm for photons with ℏω = 2.4 eV and electrons with v = 0.7c . It follows that a nanoparticle having a dimension less than this value will produce a scattering potential that has a significant component at ω/v and will make the ultrafast electrons inelastically couple to the field. A rigorous theoretical formulation was reported recently19,20 using a semiclassical approach in which the Schrödinger equation of the ultrafast electron was solved in the presence of the generated field. The Hamiltonian is constructed from the classical vector potential of the electromagnetic field in the

Figure 1. Schematic for the three-body interactions between an electron, photon, and a nanostructure. Shown is the oscillating dipole excited by the photon and the near-field produced, which is probed by the electron; both the electron and photon (E⃗ y) propagate in the zdirection and the near-field yielding EM radiation in the x̂ and ẑ directions. We note that the green clouds represent z-component of the field, which is probed in PINEM; the x−y components are not shown. The lines of the oscillating near-field between the + and − charges (Figure 4) give rise to the far-field indicated with the red-blue shapes (decreasing energy per unit area). Although a dipolar near-field falls as 1/r3, for the nanoparticle dimensions considered here, it can be approximated with an exponential decay (see ref 19).

our laboratory. In electron microscopes, when the electron is accelerated to 200 keV, the photon energy is five orders of magnitude smaller than the energy of the electron, that is, ℏω ≪Ei . Hence, the above momentum expression can be safely be approximated to yield Δp = ±nℏω/v, where v is the electron’s speed; for n = 1, Δp = ±ℏω/v. On the other hand, the momentum of an absorbed or emitted single photon momentum is ±ℏω/c, where c is the speed of light. Since v = 0.7c for 200 keV, electrons Δp is always larger than the momentum a photon can provide. This important result implies that, in free space, electrons and photons do not couple because the momentum cannot be conserved. Spatially confined electric fields along the electron’s propagation direction can provide the necessary momentum condition for the inelastic scattering to occur. Such localized electric potentials are intrinsically available in the vicinity of nano-objects as evanescent surface plasmon fields. The spatial confinement by the nanostructure leads to a large momentum spread, as dictated by the uncertainty principle (ΔzΔp ∼ h); with various components, some will facilitate the coupling as discussed below. In the classical picture for inelastic scattering of a moving electron in a field, that is, the (E⃗ ·v)⃗ interaction, only the electric field component that is parallel to the electron trajectory is of significance. Such electric fields in conventional electron energy loss spectroscopy (EELS) are induced in the material by the moving electron. At every point along its trajectory, the swift electron polarizes the particle, which in turn creates the induced electric field. The total energy loss can be written as the work done by these fields on the electron, W ≡ ΔE = vΔP ≡ v ∫ F dt ≡ ∫ F dz = ∫ eEz dz. The energy loss probability can be expressed as the work divided by ℏω, where I is given by 3335

dx.doi.org/10.1021/nl301643k | Nano Lett. 2012, 12, 3334−3338

Nano Letters

Letter

Coulomb gauge, hence the “semiclassical” nature of the approach.19−21 Here, we simply give the essential expression,19 which is the quantity that can be related to experimental observables −∞

IPINEM(x , y ; td) = |

∫+∞

were acquired and computationally summed up after correcting for any sample drift. To account for the chromaticity of the energy spectrometer at low magnifications, the slit position was adjusted at the edges of the detector and the resulting images were digitally analyzed. The silver nanoparticle sample was prepared by pipetting a droplet of the particles solution on the graphene/graphite substrate and was let to dry in an open air. Different shaped (triangular plates and spherical) particles were prepared and transferred to the microscope. Results and Discussion. Shown in Figure 2 is the brightfield TEM image of the nanoparticle ensemble, and the PINEM

Ezphoton(x , y , z ; td)e−i(ω / v)zdz|2 (2)

We note that the PINEM intensity directly maps the zcomponent of the field which varies spatially in the x−y plane of observation and as a function of the time delay (td) between the photon and electron. This expression, which differs from that of IEELS (eq 1), states that the experimental intensities are the Fourier transform of the z -component of the photoninduced electric field, Ephoton at a single reciprocal value of ω/v. z This apparent difference from (eq 1) is because in PINEM there is only one ω, which is of the photon used, whereas in EELS a spectrum of ω induced by the electron is invoked. In general, the expression involves Bessel functions,19 but the important point is that the field can be obtained from the experimentally measured intensities using ultrafast nanoscale electron probes and Fourier analysis, as demonstrated recently for subparticle imaging in 4D ultrafast electron microscopy (UEM).22,23 The PINEM images are obtained by using the energy filtering methodology in UEM. By filtering only the energy-gained electrons to form an image, we can visualize fields resulting from the interaction between electrons and photons through the nanostructure and on the femtosecond time scale. Experimental Methodhology. Experiments were conducted in Caltech’s UEM-2, which is a modified transmission electron microscope equipped with a high power femtosecond laser system.24,25 The synchronous excitation and probing of the nanoparticle ensembles was achieved by overlapping the electron and photon pulses on the sample, both in space and time. In these studies, the wavelength of the optical excitation pulse was 520 nm (ℏω = 2.4 eV), and its pulse duration was 220 fs. For every excitation pulse the specimen was probed with an electron bunch of similar duration18,19,22 and an average kinetic energy of 200 keV. Temporal synchronization of these two pulses was achieved with an optical delay line in an interferometry arrangement. The zero-of-time was determined when the maximum of the inelastic scattering peaks in PINEM was obtained, and data was then acquired. The average power of the optical laser pulse was typically 20 mW at 400 kHz repetition rate and it was focused to ∼40 μm diameter area on the sample. The linear polarization was controlled using a halfwave plate. PINEM images were formed by analyzing the scattered electrons with a postcolumn energy spectrometer. The energy selecting slit had a width of 10 eV (to cover the +nℏω region) and was centered at +6 eV at the energy gain side of the zeroloss-peak (ZLP). Accordingly, only the electrons that had gained energy between 1 and 11 eV are included in image formation. Individual image acquisition time was 20 s and drifts during this period were less than one pixel. To minimize the contribution from the ZLP tail and any possible diffraction effects (or, in general, any effect that is not a result of the excitation pulse), an off-resonance image taken far from timezero, at −10 ps, was acquired right after each image acquisition at time-zero. This off-resonance image was used as a reference frame and was subtracted from the time-zero image. To increase the signal-to-noise ratio, between 20 and 40 frames

Figure 2. PINEM images of a silver nanoparticle ensemble for two linear polarizations, Ep1 and Ep2. The sum image of Ep1 and Ep2 is also shown. The bottom-right depicts the bright field transmission electron microscope image of the same area. Using such images, structural shapes of the nanoparticles can be correlated with their near-field evanescent electric fields, both inside and outside the particles. Arrows at the upper right corner denote the linear polarization direction of the excitation laser pulse. The scale bar is 500 nm and the images are shown in colors for enhancement of contrast. The white circle is to guide the eye for the polarization effect of a single particle.

frames for two orthogonal polarizations of the excitation pulse. In addition, the sum of both polarizations is displayed for mimicking an excitation with a circularly polarized light; indeed the highly anisotropic field distributions observed for the two polarization directions become isotropic in the summed images. With such high-field-of-view images one can visualize the plasmonic fields of the whole ensemble and the effect of particle shape and separation on these fields. In Figure 3, we show images for two particles at close approach and for spherical and triangular particle array. For comparison, we also display TEM images obtained by energy filtering in the loss, not gain, region. The dominant feature in the PINEM images is the dipolar response of the particles. Since the particle length scale is much less than that of the wavelength of the excitation pulse, the dipolar response and dipole−dipole interactions are dominant factors in PINEM images, as experimentally observed. 3336

dx.doi.org/10.1021/nl301643k | Nano Lett. 2012, 12, 3334−3338

Nano Letters

Letter

Qualitatively, the results can be understood by examining the charge distributions and symmetries of particles dipoles, as shown in Figure 4. Recalling that PINEM probes the z-

Figure 3. PINEM images of two close-by particles and triangular particle assembly for two polarizations. The dipole-shaped fields and nodal planes are evident (see text). For comparison, we display in the lower panel images obtained using electron energy loss, instead of the electron energy gain of PINEM, and no dipolar near-field is present; the weak fringe around the bright field in the lower right panel is due to a slight defocus of the objective lens. The lower left (middle) panels are energy-filtered images using surface (bulk) plasmon energy selection (the spot in the middle image is due to a crystalline grain which diffracts and causes darker contrast due to the selected objective aperture. The particles in the top panel are separated by 70 nm (edgeto-edge), and the particle radius in the bottom panel is 45 nm..

A comparison of the two polarization images shows that the orientation of the induced dipole changes by 90° when the polarization is rotated by the same amount. This is a direct experimental evidence that the imaged PINEM contrast is induced by the excitation photon pulse. In contrast, EELS signals, where the observed energy filtered images are induced by the electric field of the impinging fast electron and bear the symmetry of the imaged structure (Figure 3), there is no distinct polarization to observe. For example the data in Figure 3 for EELS was obtained by filtering the surface and bulk plasmons at the energy of 5 and 25 eV (using 5 eV energy selecting slit), respectively. As discussed above following eq 1, b should then be 25 and 5 nm, respectively, and for surface plasmon filtering only the spherical field at the particle surface would be present in the image, as shown Figure 3.26 Similarly, for a triangular particle, the energy loss images show the spatial extend of different modes depending on the energy window used.13 In PINEM, this symmetry is broken because of the linear polarization of the incident photons, especially when the particle size is smaller than the wavelength of the incident photons. As shown in Figure 3, indeed PINEM images display the dipolar nature of the field, the selective polarization by the exciting photons and the effect of particle shape on the induced fields. The polarization dependence is reproduced quantitatively using the theoretical approach of ref 19; the PINEM field F = F0K1(Δkb̃)cos(ϕ), where ϕ is the laser polarization angle K1 is the modified Bessel function, and in this case Δk = ω/v.

Figure 4. Theoretical contour pictures and semiclassical calculations of the observed PINEM fields. The top panel qualitatively accounts for the presence of nodal planes and change in their directions with polarization change. The lower panel gives the PINEM field calculated using the theory presented in ref 19 by Park et al.; the calculations are made for touching particles and in this case the PINEM fields are nearly orbital-type pictures. Note the difference in field contributions in the z and x−y directions and the resulting nodal planes.

component of the electric field (Figure 4) it is understood that some nodal planes will exist. At special symmetry planes, the Ez value may vanish even though the lateral Exy components are nonzero. For the Ep1 polarization, the plane passing through the center of the particles and perpendicular to the polarization vector is the vanishing Ez plane in the PINEM images. These planes are indicated by dashed lines in the figure and, remarkably, are visualized in our experimental images. When the polarization is rotated by a 90° in the plane of the sample, the z-component of the field between the particles diminishes even though the lateral x−y field gets enhanced, again as observed experimentally. In conclusion, the ability of visualizing near-fields of nanoparticles in space and time promises a range of applications. Here, we demonstrated such imaging of fields using particles of different shapes and separations and using different polarizations. Unlike in electron energy loss spectroscopy, PINEM imaging, using electron energy gain, provides 3337

dx.doi.org/10.1021/nl301643k | Nano Lett. 2012, 12, 3334−3338

Nano Letters

Letter

maps of fields (induced by the photon) of the whole ensemble at once and enables examination of particles at close-by separations of nanometers. With the power spectrum of the zcomponent of the field observed at every point in the lateral x− y plane, it should be possible to reconstruct all components of the electric field in three dimensions with nanometer spatial and femtosecond temporal resolution. Knowledge of such fields is critical to the fundamentals and applications of nanoscale plasmonics and photonics.



(13) Nelayah, J.; Kociak, M.; Stephan, O.; Garcia de Abajo, F. J.; Tence, M.; Henrard, L.; Taverna, D.; Pastoriza-Santos, I.; Liz-Marzan, L. M.; Colliex, C. Mapping surface plasmons on a single metallic nanoparticle. Nat. Phys. 2007, 3, 348−353. (14) Kociak, M.; García de Abajo, J. Nanoscale mapping of plasmons, photons, and excitons. MRS Bull. 2012, 37, 39−46. (15) Schaffer, B.; Riegler, K.; Kothleitner, G.; Grogger, W.; Hofer, F. Monochromated, spatially resolved electron energy-loss spectroscopic measurements of gold nanoparticles in the plasmon range. Micron 2009, 40, 269−273. (16) Rang, M.; Jones, A. C.; Zhou, F.; Li, Z.-Y.; Wiley, B. J.; Xia, Y.; Raschke, M. B. Optical Near-Field Mapping of Plasmonic Nanoprisms. Nano Lett. 2008, 8, 3357−3363. (17) Muller, D. A.; Kourkoutis, L. F.; Murfitt, M.; Song, J. H.; Hwang, H. Y.; Silcox, J.; Dellby, N.; Krivanek, O. L. Atomic-Scale Chemical Imaging of Composition and Bonding by AberrationCorrected Microscopy. Science 2008, 319, 1073−1076. (18) Barwick, B.; Flannigan, D. J.; Zewail, A. H. Photon-induced near-field electron microscopy. Nature 2009, 462, 902−906. (19) Park, S. T.; Lin, M. M.; Zewail, A. H. Photon-induced near-field electron microscopy (PINEM): theoretical and experimental. New J. Phys. 2010, 12, 123028. (20) Howie, A. Photon interactions for electron microscopy aplications. Eur. Phys. J. - Appl. Phys. 2011, 54, 33502. (21) García de Abajo, F. J. Optical excitations in electron microscopy. Rev. Modern Phys. 2010, 82, 209−275. (22) Yurtsever, A.; van der Veen, R. M.; Zewail, A. H. Subparticle Ultrafast Spectrum Imaging in 4D Electron Microscopy. Science 2012, 335, 59−64. (23) Batson, P. E. Plasmonic Modes Revealed. Science 2012, 335, 47−48. (24) Zewail, A. H.; Thomas, J. M. 4D Electron Microscopy; Imperial College Press: London, 2009. (25) Zewail, A. H. Four-Dimensional Electron Microscopy. Science 2010, 328, 187−193. (26) Ugarte, D.; Colliex, C.; Trebbia, P. Surface- and interfaceplasmon modes on small semiconducting spheres. Phys. Rev. B 1992, 45, 4332−4343.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the National Science Foundation (DMR-0964886) and the Air Force Office of Scientific Research (FA9550-11-1-0055) in the Gordon and Betty Moore Center for Physical Biology at the California Institute of Technology. We wish to thank Dr. Sang Tae Park for the helpful discussion on the theory of PINEM and for the calculation provided in Figure 4 (bottom panel).We are especially grateful to Dr. Spencer Baskin who contributed to the excellent performance of the laser system and optical arrangement inside the microscope.



REFERENCES

(1) Prasad, P. N. Nanophotonics; Wiley-Interscience: Hoboken, NJ, 2004. (2) Atwater, H. A.; Polman, A. Plasmonics for improved photovoltaic devices. Nat. Mater. 2010, 9, 205−213. (3) Aubry, A.; Lei, D. Y.; Fernández-Domínguez, A. I.; Sonnefraud, Y.; Maier, S. A.; Pendry, J. B. Plasmonic Light-Harvesting Devices over the Whole Visible Spectrum. Nano Lett. 2010, 10, 2574−2579. (4) Liu, N.; Tang, M. L.; Hentschel, M.; Giessen, H.; Alivisatos, A. P. Nanoantenna-enhanced gas sensing in a single tailored nanofocus. Nat. Mater. 2011, 10, 631−636. (5) Kneipp, K. Surface-Enhanced Raman Scattering. Phys. Today 2007, 60, 40−46. (6) Krenn, J. R.; Dereux, A.; Weeber, J. C.; Bourillot, E.; Lacroute, Y.; Goudonnet, J. P.; Schider, G.; Gotschy, W.; Leitner, A.; Aussenegg, F. R.; Girard, C. Squeezing the Optical Near-Field Zone by Plasmon Coupling of Metallic Nanoparticles. Phys. Rev. Lett. 1999, 82, 2590− 2593. (7) Maier, S. A.; Brongersma, M. L.; Kik, P. G.; Atwater, H. A. Observation of near-field coupling in metal nanoparticle chains using far-field polarization spectroscopy. Phys. Rev. B 2002, 65, 193408. (8) Su, K. H.; Wei, Q. H.; Zhang, X.; Mock, J. J.; Smith, D. R.; Schultz, S. Interparticle Coupling Effects on Plasmon Resonances of Nanogold Particles. Nano Lett. 2003, 3, 1087−1090. (9) Sundaramurthy, A.; Crozier, K. B.; Kino, G. S.; Fromm, D. P.; Schuck, P. J.; Moerner, W. E. Field enhancement and gap-dependent resonance in a system of two opposing tip-to-tip Au nanotriangles. Phys. Rev. B 2005, 72, 165409. (10) Giannini, V.; Fernández-Domínguez, A. I.; Heck, S. C.; Maier, S. A. Plasmonic Nanoantennas: Fundamentals and Their Use in Controlling the Radiative Properties of Nanoemitters. Chem. Rev. 2011, 111, 3888−3912. (11) Gómez-Medina, R.; Yamamoto, N.; Nakano, M.; García de Abajo, F. J. Mapping plasmons in nanoantennas via cathodoluminescence. New J. Phys. 2008, 10, 105009. (12) Batson, P. E. Surface Plasmon Coupling in Clusters of Small Spheres. Phys. Rev. Lett. 1982, 49, 936−940. 3338

dx.doi.org/10.1021/nl301643k | Nano Lett. 2012, 12, 3334−3338