Letter pubs.acs.org/NanoLett
Nanostructured Potential of Optical Trapping Using a Plasmonic Nanoblock Pair Yoshito Tanaka, Shogo Kaneda, and Keiji Sasaki* Research Institute for Electronic Science, Hokkaido University, Sapporo 001-0020, Japan S Supporting Information *
ABSTRACT: We performed two-dimensional mapping of optical trapping potentials experienced by a 100 nm dielectric particle above a plasmonresonant gold nanoblock pair with a gap of several nanometers. Our results demonstrate that the potentials have nanoscale spatial structures that reflect the near-field landscape of the nanoblock pair. When an incident polarization parallel to the pair axis is rotated by 90°, a single potential well turns into multiple potential wells separated by a distance smaller than the diffraction limit; this is associated with super-resolution optical trapping. In addition, we show that the trap stiffness can be enhanced by approximately 3 orders of magnitude compared to that with conventional far-field trapping. KEYWORDS: Nanogap, localized surface plasmon, optical trapping, nanoparticle pair, trap stiffness
O
However, there has been little experimental work related to quantitative analysis of LSP trapping potentials experienced by a nanoparticle. In this Letter, we measure the plasmonic trapping potentials for a 100 nm dielectric particle above a gold nanoblock pair. We demonstrate that the potential has a nanoscale spatial structure that does not have a simple harmonic shape but rather reflects the plasmonic field landscape consisting of multiple localized optical spots. The potential shape as well as the plasmonic landscape is dramatically changed by rotating the linear polarization of the incident laser. Double potential wells separated by a distance of approximately 230 nm smaller than the diffraction limit are observed with the polarization direction perpendicular to the pair axis, respectively. In addition, we show that with the polarization direction along the pair axis, the trap stiffness can be amplified by approximately 3 orders of magnitude compared with the stiffness achieved with far-field trapping. The scanning electron microscope image of the plasmonic nanostructure comprised of a pair of diagonally aligned gold nanoblocks separated by a single-nanometer-sized gap is shown in Figure 1a. The 30 nm thick gold structure was fabricated on a glass coverslip by electron beam lithography combined with a lift-off process. In the experiment, a 1064 nm near-infrared (NIR) laser beam was directed on the gold nanostructure through an oil immersion objective [100×, numerical aperture (NA) = 1.35] to excite the surface plasmon of the nanostructure for optical trapping, as shown in Figure 1b (also shown in Supporting Information Figure S1). The gold nanostructure was immersed in a diluted aqueous solution of 100 nm fluorescent polystyrene nanoparticles. The excitation of
ptical trapping has attracted significant attention because of its successful implementation in the fields of biology and physics with applications ranging from single-molecule force measurements to optical particle sorting.1−4 The wellestablished optical trapping phenomenon relies on the field gradients of a focused laser beam, which induce a restoring force toward the laser focus where the intensity is maximum. In this far-field optical trapping, the diffraction-limited focus of the laser beam makes it difficult to trap and position nanometersized specimens. The limitation of the focused spot size sets a maximum for the achievable intensity and field gradients. Consequently, extremely high incident laser power is required to produce a trapping potential that is sufficiently deep to overcome the Brownian motion of nanometer-sized objects. This is not appropriate for many systems, particularly for interesting biological samples where intense optical irradiation is a concern. To circumvent the limitations imposed by light diffraction, the use of the localized surface plasmon (LSP) in metal nanostructures has recently been explored.5−16 A plasmonic nanostructure functions as a nanolens, and it can efficiently focus and confine propagating light to nanoscale volumes, which can be orders of magnitude smaller than the wavelength of the light. In particular, a pair of metal nanoparticles separated by a small gap exhibits superior light nanoconcentration properties.17−21 Metal nanoparticle pairs have been experimentally demonstrated to provide trapping and confinement of dielectric nanoparticles, such as those of polystyrene and living biological specimens, with reduced laser power compared to that required for far-field trapping.5,6,9−11 The plasmonic trapping of metal nanoparticles with 10 nm dimensions has also been discussed through monitoring of the Rayleigh scattering spectra of the nanoparticle pair.7 The implementation of plasmonic trapping in the life and physical sciences requires deep understanding of its properties. © XXXX American Chemical Society
Received: February 14, 2013 Revised: March 29, 2013
A
dx.doi.org/10.1021/nl4005892 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
differentiate the motion of the trapped 100 nm nanoparticle from the total scattering. Thus, plasmonic trapping was observed by monitoring the nanoparticle fluorescence. For fluorescence excitation, a 532 nm line of a green laser beam was passed through the same optical path as the NIR laser beam. These laser beams were defocused on the nanostructure to obtain uniform illumination. The intensity of the green laser at the nanostructure (15 W/cm2) was significantly weaker than that of the NIR laser (on the order of 10 kW/cm2) and therefore the plasmonic trapping by the green laser was not apparent. The fluorescence of the particles around the nanostructure was efficiently collected with the same high-NA objective, passed thorough a tube lens, and projected onto a charge coupled device (CCD) camera. A three-dimensional finite-difference time-domain simulation was performed to examine the near-field distribution of the nanoblock pair.23 We obtained the trapping potential created by the plasmonic nanostructure using Boltzmann statistics of the position fluctuations of the trapped particle. This method utilizes the random Brownian motion of the particles and has been shown to be highly effective for probing arbitrary force fields.24−27 Righini et al. measured the trapping potentials for a 3.55 μm polystyrene bead at a 4.8 μm gold pad, which is based on the propagating surface plasmon.25 To capture the instantaneous positions of the moving trapped particle, we must use a position detector with better temporal resolution than the cutoff frequency of the particle motion in the optical trap.28 The cutoff frequency is calculated according to fc = k/2πγ, where k is the trap stiffness, γ = 6πηr is the Stokes friction coefficient (η is the water viscosity and r is the particle radius), and fc ≈ 200 Hz in our experiment. A quadrant photodiode (QPD) with a high-
Figure 1. (a) Scanning electron micrograph of a gold nanoblock pair with an interparticle edge-to-edge separation of ∼6 nm. The scale bar is 100 nm. (b) Schematic illustration of the experimental setup. The green laser for fluorescence excitation was modulated at 30 Hz with a pulse of 300 μs and was synchronized with the CCD camera frame rate to capture the instantaneous positions of the moving trapped nanoparticle.
the surface plasmon resonances can produce local heating in the metal and heat dissipation, which results in convection. A thin sample solution of ∼5 μm was used to damp plasmonthermal fluid convection.22 The presence of gold nanostructures with strong light-scattering properties makes it difficult to
Figure 2. (a) Calculated field intensity distribution in the x−y plane at a height 5 nm above a model nanoblock pair with the incident polarization direction along the pair axis. (b) Experimentally measured potential map of plasmonic trapping of the nanoparticle using the nanoblock pair with the near-field landscape of (a). The incident laser power and the illumination spot diameter were 500 μW and 1 μm, respectively, corresponding to the incident intensity of 60 kW/cm2. The potential was obtained from the lateral position distribution of the trapped nanoparticle. (c,d) The potential profiles and the particle position histograms along the x- and y-axes of (b). The continuous red and blue lines in (c) and (d) represent the parabolic fits of the potentials around the minima and the corresponding Gaussian functions, respectively. While the potential profile of (d) is well approximated by a parabolic function (red line), the profile of (c) shows deviations from the parabolic function on both sides. The solid gray lines are instrumental response functions observed by the measurement of a particle adhered to a glass substrate. The scale bar is 150 nm. B
dx.doi.org/10.1021/nl4005892 | Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
be one of the indicator of the statistical error including measurement noises caused by vibration of the stage, instability in electronics, fluctuation of the laser power, and so forth. In other words, they represent the uncertainty of the measured trapping potentials.26−28 The solid red lines represent the parabolic fits of the potentials around the minima. Although the potential profile along the transverse direction is well approximated by a parabolic function, we observed some deviations from the parabolic form on both sides of the potential profile along the longitudinal direction. For the transverse direction, a single optical near-field spot is generated around the nanogap, and on the other hand, for the longitudinal direction, multiple near-field spots not only around the nanogap but also at the outer ends of the blocks (see Figure 2a and Supporting Information Figure S3). As the trapped particle moves closer to the outer ends, the gradient force toward the outer side against the nanogap field force is increased, reducing the trap strength on both sides of the potential along the longitudinal direction. Spring constants of the harmonic oscillators were calculated with the deconvolution of the instrumental response function.26 The resulting trap stiffnesses around the potential minima were kx = 2.3 fN nm−1 in the longitudinal direction and ky = 2.0 fN nm−1 in the transverse direction. This stiffness anisotropy occurs because the near field around the nanogap is more strongly confined along the longitudinal direction than along the transverse direction, as shown in Figure 2a. The near-field trapping is supported by numerical simulations of the optical force exerted on a 100 nm polystyrene particle, which is obtained through integration of the Maxwell stress tensor for the electric field distribution over the particle surface. When an incident light intensity is taken to be the same as that used in the experiment, the trap stiffnesses for small displacements from the gap center are of the same order as those measured experimentally. Reasonable agreement for the anisotropy of the trap stiffnesses is also observed. For far-field optical trapping of the nanoparticle using a laser beam focused to a diffractionlimited spot with the same incident intensity, its trap stiffness is estimated to be 2.1 × 10−3 fN nm−1.33 Thus, the nanoblock pair could amplify the trap stiffness by approximately 3 orders of magnitude compared to the stiffness generated by far-field optical trapping. Figure 3 shows the dependences of the trap stiffnesses, kx and ky, on the incident laser intensity. The trap stiffnesses increased
temporal resolution has been used to detect the position of the light-scattering particle for measuring the potential of far-field trapping.26−29 Unfortunately, it was difficult for the present experiment to track 100-nm-sized emitting particle with the QPD owing to the very weak fluorescent intensity. Therefore, we employed the CCD camera with chirping illumination. The highly sensitive CCD camera was operated at 30 fps with an exposure time of approximately 30 ms, which is longer than 1/fc = 5 ms. Consequently, the trapping potential could not be accurately measured owing to the motion blur resulting from time-averaging a signal over a long exposure time (Supporting Information Movie 1 and Figure S2).12,16,30 Thus, the green laser for fluorescence excitation was modulated at a frequency of 30 Hz with a pulse duration of 300 μs (
