LETTERS plasmon waveguides. 0. Energy attenuation lengths of 6 dB per 30 nm were predicted for plasmon waveguides consisting of closely spaced pherical Au nanoparticles on an indium tin oxide(ITo)substrate O Control B so, it was shown that a geometry change to spheroidal Ag particles 3+23 should allow foranattenuationlengthincrease toabout 6dB per 250nm 329+14mm ref 10). This kind of attenuation length is sufficient to allow detection of energy transport by localexcitation. Plasmon waveguides were fabricated by using electron-beam 3 lithography with lift-off on ITO-coated quartz slides, which allowed for 0.6 a good control over particle size and spacing. The waveguide structures consist of rod-shaped Ag nanoparticles with dimensions of 90 nm x 30 nm x 30 nm and a surface-to-surface spacing of 50 nm between adjacent particles. The long axes of the individual nanoparticles were oriented perpendicular to the waveguide chain axis to allow for an increased near-field coupling between the particles. The inset of Fig. I shows a scanning electron micrograph of one of these plasmon waveguides. To allow for the determination of the plasmon 0o00o resonances of the fabricated structures with a high signal-to-noiseratio using far-field spectroscopy, many plasmon waveguides were arranged in a 100 umx 100 um grid with a grating constant of 1 um as depictedin Distance (um) Fig. 2b. It has previously been shown that cross talk between different waveguides is negligible for this grating constant".. Thus, far-field extinction spectra on these arrays reflect the properties of individual Figure 3 Evidence for energy transportin plasmon waveguides by the widthof the plasmon waveguides and provide a probe of the near-field coupling intensity of fluorescent nanospheres. Individual data sets represent averagesoffive between the nanoparticles composing each waveguide. parallel cuts along the plasmon waveguide direction through the fiuorescent spots Figure I shows the far-field extinction spectrum of the fabricated highlighted in Fig. 2c, d for isolated nanospheres(controlAand B, blackand redd plasmon waveguides taken under normal-incidence white-light es located on top of plasmon waveguides (WG 1 andwG2, green illumination witha spot sizeof 100 amand a polarizationalong thelong and blue data points) as depicted in the inset Gaussian line-shape fits to the data an of nanoparticles and thus perpendicular tothe waveguide-chain increased width for nanospheres located on plasmon waveguides kis(red triangles). The extinction spectrum of a grid of single Ag nanoparticles of the same geometry with an inter-particle spacing of I um is also shown, for which the inter-particle coupling is negligible (black squares). The single-particle extinction spectrum peaks at filtered out by using a band-pass filter, and the 610 nmdyeemission was 2. 18 eV for a polarization along the long particle axis, corresponding to detected in the far-field with an avalanche photo diode. This scheme a resonance wavelength of 570 nm. The extinction spectrum of the enables the observation of energy transport in the following way: first, plasmon waveguide shows a resonance shift of about 100 meV to higher energy is transferred from the illuminating tip to the plasmon agreement with theoretical and numerical studies of aggregated noble- nanoparticle structure and excites a luoresenlaepagates along the merges because of near-field coupling between the particles, in iveguide. The excitation subsequentl osphere placed on metal nanoparticles and plasmon waveguides. 2. According to top of a waveguide at a sufficient distance from the excitation source numerical simulations, this resonance shift of 100 meV translates into a Energy transport would result in dye emission even when the maximumenergy attenuation length of the order of 6dB per 200nm for microscope tip is located away from the dye, and thus would manifest anexcitation at the single-particle resonance(2.18 eV in these samples), itself in an increased spatial width of the fluorescence spot of a at which the energy transfer in a plasmon waveguide is predicted to be nanosphere attached to a plasmon waveguide compared with a single mostefficient, independent of inter-particle spacing. 2. free nanosphere To probe energy transport directly in the fabricated plasmor Figure 2c,d shows simultaneously obtained topography and waveguides, local excitation is necessary as opposed to far-field fluorescent NSOM scans of thecontrol area outside the grid( Fig 2b,red excitation of all particles in the arrays. To accomplish this, the tip of an box)with single nanospheres only and of the plasmon waveguide grid illumination mode near-field scanning optical microscope(NSOM)(Fig 2b, green box) together with fluorescent nanospheres, respectively (Nanonics NSOM-100)(ref. 17)is used as a local excitation source for The samples were illuminated at the single-particle resonance nanoparticles in plasmon waveguides. To excite the mode of least wavelength of 570 nm by usingamicroscopetip witha 100nmaperture damping, laser light from a dye laser at a wavelength of 570 nm, The scans were done in constant gap mode, ensuring a fixed vertical responding to the single-particle resonance, was coupled into a distance between the exciting tip and the fluorescent nanospheres both multi-mode optical fibre attached to the Al-coated NSOM tip used for for isolated spheres and for spheres located on plasmon waveguides. excitation Figure 2a shows a schematic of our approach to excitation The scan direction was perpendicular to the plasmon waveguides, and and energy transport detection Power transport away from the directly images were built up from left to right. Figure 2c shows a scan of a excited nanoparticles in the plasmon waveguide is probed by the control area consisting of single nanospheres only, which was taken placement of carboxyl-coated polystyrene nanospheres(Molecular immediately before scanning the plasmon waveguide area. Probes Fluospheres F-8801, diameter 110+8 nm)filled with fluorescent Single nanospheres are clearly resolved in both the topography and the electron-beam-fabricated plasmon e waveguide structure. For this, the fluorescent NSOM image. Intensity variations are observed between a thin polylysine layer, and the nanospheres were subsequently variationsin the nanosphere diameter; theelongation of the fluorescent randomly deposited from an aqueous solution. The fluorescent dyes spots is due to a tip artefact Figure 2d shows the subsequent scan of a used show a strong absorption, peaking at 580-590 nm near the sample area comprising four plasmon waveguides in the left part of the plasmon resonance wavelength of a single fabricated Ag particle and image and fluorescent nanospheres. Two nanosphe heres(highlighted by having their emission maximum at 610 nm. The excitation light was red circles) are located on top of pla naturematerialsVol2lApriL2003iwww.nature.com/naturematerials @2003 Nature Publishing GroupLETTERS nature materials | VOL 2 | APRIL 2003 | www.nature.com/naturematerials 231 plasmon waveguides9,10. Energy attenuation lengths of 6 dB per 30 nm were predicted for plasmon waveguides consisting of closely spaced spherical Au nanoparticles on an indium tin oxide (ITO) substrate. Also, it was shown that a geometry change to spheroidal Ag particles should allow for an attenuation length increase to about 6dB per 250nm (ref. 10). This kind of attenuation length is sufficient to allow detection of energy transport by local excitation. Plasmon waveguides were fabricated by using electron-beam lithography with lift-off on ITO-coated quartz slides,which allowed for a good control over particle size and spacing. The waveguide structures consist of rod-shaped Ag nanoparticles with dimensions of 90 nm × 30 nm × 30 nm and a surface-to-surface spacing of 50 nm between adjacent particles. The long axes of the individual nanoparticles were oriented perpendicular to the waveguide chain axis to allow for an increased near-field coupling between the particles10. The inset of Fig.1 shows a scanning electron micrograph of one of these plasmon waveguides. To allow for the determination of the plasmon resonances of the fabricated structures with a high signal-to-noise ratio using far-field spectroscopy, many plasmon waveguides were arranged in a 100µm×100µm grid with a grating constant of 1µm as depicted in Fig. 2b. It has previously been shown that cross talk between different waveguides is negligible for this grating constant9,15. Thus, far-field extinction spectra on these arrays reflect the properties of individual plasmon waveguides and provide a probe of the near-field coupling between the nanoparticles composing each waveguide. Figure 1 shows the far-field extinction spectrum of the fabricated plasmon waveguides taken under normal-incidence white-light illumination with a spot size of 100µm and a polarization along the long axis of the nanoparticles and thus perpendicular to the waveguide-chain axis (red triangles). The extinction spectrum of a grid of single Ag nanoparticles of the same geometry with an inter-particle spacing of 1 µm is also shown, for which the inter-particle coupling is negligible (black squares). The single-particle extinction spectrum peaks at 2.18 eV for a polarization along the long particle axis, corresponding to a resonance wavelength of 570 nm. The extinction spectrum of the plasmon waveguide shows a resonance shift of about 100meV to higher energies because of near-field coupling between the particles, in agreement with theoretical and numerical studies of aggregated noble￾metal nanoparticles16 and plasmon waveguides4,12. According to numerical simulations,this resonance shift of 100 meV translates into a maximum energy attenuation length of the order of 6dB per 200nm for an excitation at the single-particle resonance (2.18 eV in these samples), at which the energy transfer in a plasmon waveguide is predicted to be most efficient,independent of inter-particle spacing4,12. To probe energy transport directly in the fabricated plasmon waveguides, local excitation is necessary as opposed to far-field excitation of all particles in the arrays. To accomplish this, the tip of an illumination mode near-field scanning optical microscope (NSOM) (Nanonics NSOM-100) (ref. 17) is used as a local excitation source for nanoparticles in plasmon waveguides. To excite the mode of least damping, laser light from a dye laser at a wavelength of 570 nm, corresponding to the single-particle resonance, was coupled into a multi-mode optical fibre attached to the Al-coated NSOM tip used for excitation. Figure 2a shows a schematic of our approach to excitation and energy transport detection.Power transport away from the directly excited nanoparticles in the plasmon waveguide is probed by the placement of carboxyl-coated polystyrene nanospheres (Molecular Probes Fluospheres F-8801,diameter 110±8nm) filled with fluorescent molecules18 in close proximity to the waveguide structure. For this, the electron-beam-fabricated plasmon waveguide sample was coated with a thin polylysine layer, and the nanospheres were subsequently randomly deposited from an aqueous solution. The fluorescent dyes used show a strong absorption, peaking at 580–590 nm near the plasmon resonance wavelength of a single fabricated Ag particle and having their emission maximum at 610 nm. The excitation light was filtered out by using a band-pass filter,and the 610 nm dye emission was detected in the far-field with an avalanche photo diode. This scheme enables the observation of energy transport in the following way: first, energy is transferred from the illuminating tip to the plasmon waveguide. The excitation subsequently propagates along the nanoparticle structure and excites a fluorescent nanosphere placed on top of a waveguide at a sufficient distance from the excitation source. Energy transport would result in dye emission even when the microscope tip is located away from the dye, and thus would manifest itself in an increased spatial width of the fluorescence spot of a nanosphere attached to a plasmon waveguide compared with a single free nanosphere. Figure 2c,d shows simultaneously obtained topography and fluorescent NSOM scans of the control area outside the grid (Fig.2b,red box) with single nanospheres only and of the plasmon waveguide grid (Fig.2b,green box) together with fluorescent nanospheres,respectively. The samples were illuminated at the single-particle resonance wavelength of 570nm by using a microscope tip with a 100nm aperture. The scans were done in constant gap mode, ensuring a fixed vertical distance between the exciting tip and the fluorescent nanospheres both for isolated spheres and for spheres located on plasmon waveguides. The scan direction was perpendicular to the plasmon waveguides, and images were built up from left to right. Figure 2c shows a scan of a control area consisting of single nanospheres only, which was taken immediately before scanning the plasmon waveguide area. Single nanospheres are clearly resolved in both the topography and the fluorescent NSOM image. Intensity variations are observed between different fluorescent nanospheres, which are attributed to slight variations in the nanosphere diameter;the elongation of the fluorescent spots is due to a tip artefact. Figure 2d shows the subsequent scan of a sample area comprising four plasmon waveguides in the left part of the image and fluorescent nanospheres. Two nanospheres (highlighted by red circles) are located on top of plasmon waveguides, and one Figure 3 Evidence for energy transport in plasmon waveguides by the width of the intensity of fluorescent nanospheres. Individual data sets represent averages of five parallel cuts along the plasmon waveguide direction through the fluorescent spots highlighted in Fig. 2c,d for isolated nanospheres (control A and B,black and red data points) and nanospheres located on top of plasmon waveguides (WG 1 and WG2,green and blue data points) as depicted in the inset.Gaussian line-shape fits to the data show an increased width for nanospheres located on plasmon waveguides. –0.2 0.2 0.6 1.0 1.4 174±17 nm 193±23 nm 329±14 nm 343±27 nm Fluorescence intensity (a.u.) Fit widths: 0.0 0.4 0.8 1.2 1.6 Distance (µm) Control A Control B WG1 WG2 © 2003 Nature Publishing Group