Optical trapping through the localized surface-plasmon resonance of engineered gold nanoblock pairs

: We have investigated the plasmonic trapping of dielectric nanoparticles by using engineered gold nanoblock pairs with ~5-nm gaps. Pairs with surface-plasmon resonance peaks at the incident wavelength allow the trapping of 350-nm-diameter nanoparticles with 200 W/cm 2 laser intensities, and their plasmon resonance properties and trapping performance are drastically modified by varying the nanoblock size of ~20%. In addition, plasmon resonance properties of nanoblock pairs strongly depend on the direction of the linear polarization of the incident laser, which determines the trapping performance.

The optical trapping of nanometer-sized objects by using plasmonic nanostructures has attracted considerable attention in recent years because it significantly improves the trapping performance compared with conventional optical tweezers that use a tightly focused laser beam [1][2][3][4][5][6][7][8][9]. Localized surface plasmons, i.e., resonant charge-density oscillations confined to metal nanostructures, can efficiently convert propagating light into nanoscale confined and strongly enhanced optical fields, which generate high-intensity gradients responsible for the trapping mechanism. Among the huge variety of plasmonic nanostructures, a pair of metal nanoparticles separated by a nanometric gap produces an intense optical spot that is approximately two orders of magnitude smaller than the wavelength of the incident light [10][11][12], and it enables the optical trapping and confinement of dielectric nanoparticles such as polystyrene and living biological specimens with reduced laser intensity compared with conventional optical tweezers [6,7]. Furthermore, the nanogap structure was shown to trap metal and semiconductor nanoparticles with 10-nm dimensions [8,9]. The parallel trapping of micro-and nanoparticles has also demonstrated using patterned plasmonic nanostructures [7,13]. This plasmonic trapping by nanostructures deposited on a substrate is a promising technique for future biological and chemical lab-on-a-chip devices operated with light [14].
The localized surface plasmon also gives rise to spectrally selective light absorption and scattering. This plasmon resonance property strongly depends on the geometry and size of nanostructures. For example, anisotropic nanostructures such as nanorod and nanoparticle pairs have multiple resonance peaks [15][16][17][18][19][20]. Moreover, by varying the size of the gold and silver nanostructures, their color covers an extended wavelength range that includes most of the visible and infrared regions [15,21]. Thus, tailoring the structural geometry and size allows a direct control of the plasmon resonance property.
In this study, we demonstrate that the size and orientation of pairs of gold nanoblocks with ~5-nm determine their surface-plasmon resonance properties and trapping performance in conjunction with the wavelength and the polarization direction of the incident laser.  Figure 1 illustrates the optical setup. A continuous-wave Ti:sapphire laser (800-nm wavelength, 100-kHz linewidth) was used as the plasmonic-trapping light source. The linearly polarized beam was introduced into an inverted optical microscope and was focused by an oilimmersion objective [100 × , numerical aperture (NA) = 1.35]. A section of the focused beam with the incident angle smaller than the critical angle was masked by a knife edge at the pupil plane of the imaging system, and thus, samples were illuminated under focused total internal reflection (TIR, spot area = 0.9 µm × 2.7 µm). The polarization angle of the incident beam was controlled by a half-wave plate. A 50-µm-deep liquid chamber containing 350-nmdiameter spherical polystyrene particles (refractive index of 1.57) in water (refractive index of 1.33) was prepared on a glass substrate with nanostructures. The particle concentration of 2.7 x 10 9 particles/ml results in average numbers of 1.6 x 10 −3 nanoparticles in the observation volume obtained from the spot area and decay distance (0.3 µm) of evanescent wave. The plasmonic trapping was viewed using a CCD camera to monitor scattered light from trapped particles by disturbing the evanescent field. Note that nanoparticles were not trapped using only the focused TIR irradiation without plasmonic nanostructures.
Surface-plasmon resonance properties of nanostructures were characterized by their extinction spectra. Collimated white light from a halogen lamp illuminated sample through a condenser lens. Transmitted white light through the sample was collected by the same microscope objective, passed through an optical fiber, and directed to an optical spectrometer equipped with a CCD detector cooled to −120 °C by liquid N 2 . A ~0.5-µm-diameter collection area was used for the objective-fiber combination used here. Extinction spectra of the nanostructures were obtained by calculating (I b (λ)-I m (λ))/I b (λ), where I m (λ) and I b (λ) are the spectra of the transmitted white light through the sample with and without the nanostructures, respectively. An absorptive sheet polarizer was placed in front of the detector to analyze the polarization of transmitted white light.  A nanostructured sample is shown in the scanning electron microscope (SEM) image of Fig. 2(a). The sample was fabricated on a glass substrate by using electron beam lithography (EBL) and lift-off techniques [16,22]. Briefly put, the EBL technique was used to define masks of planar patterns of nanostructures on substrates and thin gold films were deposited over the mask by sputtering. Gold nanostructures were subsequently obtained by a lift-off of protected areas. This method yielded regular 3 µm × 3 µm two-dimensional arrays of pairs of diagonally aligned gold nanoblocks with nanometer-scale gaps. Within the array, all nanoblock pairs were oriented in the same direction.
Data discussed here were obtained from three samples with the same gap distance d ≈5 nm, height h = 40 nm, and pair-pair separation s = 320 nm. However, in-plane sizes of nanoblocks differed. With such a small gap distance, the electromagnetic interaction between two nanoblocks splits the localized plasmon resonance of an individual nanoblock into two resonances of longitudinal and transverse modes for the nanoblock pair [15]. The excitation of the longitudinal mode leads to a large field enhancement around the gap between nanoblocks [10,22]. Note that this field enhancement extends not only within the nanogap but also outside it (see Fig. 3), which is important for the plasmonic trapping of nanoparticles larger than the gap size. Although it was difficult to observe an isolated pair of nanoblocks in the optical micrograph, the nanoblock array was clearly imaged, as shown in Fig. 2(b). In addition, because nanoparticles experience an optical force near the nanostructure surface, they could be trapped with a higher probability for a pair array, whose surface area is relatively large compared with a single pair. These features were convenient for our experiments.   Fig. 4, bottom). The pairs with an 80-nm block width exhibit a surface-plasmon resonance peak at an 800-nm incident wavelength, and the 70-and 100-nm pairs exhibit surface-plasmon resonance peaks below and above 800 nm, respectively. The extinction intensity at 800-nm for the 80-nm pairs is 2.4 and 1.7 times higher than those for the 70-and 100-nm pairs, respectively. The calculated field enhancement factor at 800-nm (M 800 = |E max /E 0 | 2 ) is about 6300 for the 80-nm pairs (see Fig.  3), and the field enhancement factors (M 800 ) for the 70-and 100-nm pairs are 2.3 and 2.0 times lower, respectively. Such nanoblock pairs of different sizes are used to optically trap nanoparticles. The incident laser power was fixed at 15 µW, corresponding to intensity in the spot area of 750 W/cm 2 . For conventional optical tweezers, which were obtained by removing the knife edge and beam block (see Fig. 1), an intensity of 2 MW/cm 2 was required to trap 350-nm-diameter polystyrene particles for a period of a few tens of seconds. This intensity is more than three orders of magnitude larger than that required for the plasmonic trapping of the same particles. Other groups report the plasmonic trapping of 200-nm-diameter polystyrene particles with the incident intensity > kW/cm 2 [7]. In addition to optical gradients, coupling to the surface plasmon resonances can produce local heating in the metal and heat dissipation within the chamber, which results in convection. However, a local temperature elevation in the gold nanostructure for low incident intensity of 750 W/cm 2 is estimated to be less than 0.1 °C from some previous reports which experimentally and theoretically evaluate it [9,23]. Thus, the trapping contribution from plasmon-thermal fluid convection can be negligible.  In the first frame of these movies (see Fig. 5 (a)-(c)), the particles are not trapped and bright scattered light is observed around nanoblock pairs. When nanoparticles pass sufficiently close to the pairs, they get trapped and appear as bright scattering spots, as shown in Fig. 5 (d). The position of this bright spots fluctuates slightly, suggesting that particle Brownian motion is still present. This phenomenon distinguishes between light scattering from particles and that from nanoblock pairs. The escape of particles from the pairs is known from the disappearance of the fluctuating bright spots. On such process, the increase and decrease of the light scattered from particles are done in one step. On the other hand, the scattered light must gradually increase and decrease for optical trapping of multiple particles, as proved in the previous study [24,25]. In addition, as aforementioned, the average number of nanoparticles in the observation volume is estimated to be 1.6 x 10 −3 particles. These strongly support single-particle trapping in the present experiments. For the 70-and 100-nm nanoblock widths, nanoparticles become trapped and then escape from pairs within a few seconds, as shown in Figs. 5(a) (Media 1) and 5(c) (Media 3), respectively. However, the 80-nm nanoblock pairs optically traps nanoparticles for at least ten times longer, as shown in Fig.  5(b) (Media 2). Thus, we conclude that nanoblock pairs with the plasmon resonance peak at the incident wavelength allow the stable trapping of nanoparticles. Note that a variation in the block size of only 20% drastically modifies the plasmon resonance property and the trapping performance of nanoblock pairs. For 80-nm nanoblock pairs, the peak of the transverse plasmon resonance is away from the wavelength of the incident laser [ Fig. 6(a)]. Moreover, nanogap-mode plasmons are not excited with transverse polarization, as shown in Refs [15,22]. Therefore, the trapping performance of nanoblock pairs changes when the incident polarization is converted from longitudinal to transverse. As shown in Fig. 6(b) (Media 4), this polarization conversion leads to a decrease in the light-scattering intensity from pairs and the subsequent release of trapped nanoparticles. The incident intensity of 200 W/cm 2 was four orders of magnitude less than that for conventional optical tweezers. With transverse polarization, we were unable to trap nanoparticles on pairs despite attempting to do so for five hours. Thus, for example, for two types of nanoblock pairs with longitudinal axes orthogonal to each other, we can select the trapping pair via the polarization of the incident laser.
In summary, we have investigated the trapping performance and the surface-plasmon resonance property of engineered pairs of gold nanoblocks with ~5-nm gaps. Pairs with surface-plasmon resonance peaks at the incident wavelength allow the plasmonic trapping of 350-nm-sized nanoparticles with a laser intensity four orders of magnitude smaller than conventional optical tweezers. The trapping performance and plasmon resonance property are drastically modified by varying the block size. In addition, the plasmon resonance property of nanoblock pairs strongly depends on the direction of the incident linear polarization, which determines the trapping performance of pairs. Thus, an arrangement of nanoblock pairs with various sizes and orientations can allow selective nanoparticle trapping and patterning that can be optically controlled by the incident optical wavelength and polarization. This technique offers a new potential for plasmonic trapping by metal nanostructures.