Letter pubs.acs.org/NanoLett
Application of Plasmonic Bowtie Nanoantenna Arrays for Optical Trapping, Stacking, and Sorting Brian J. Roxworthy,†,⊥ Kaspar D. Ko,‡,⊥ Anil Kumar,† Kin Hung Fung,§ Edmond K. C. Chow,∥ Gang Logan Liu,† Nicholas X. Fang,§ and Kimani C. Toussaint, Jr.*,‡,¶ †
Department of Electrical and Computer Engineering and ‡Department of Mechanical Science and Engineering, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801, United States § Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States ∥ Micro and Nanotechnology Laboratory, University of Illinois at Urbana-Champiagn, Urbana, Illinois 61801, United States S Supporting Information *
ABSTRACT: We present the use of Au bowtie nanoantenna arrays (BNAs) for highly efficient, multipurpose particle manipulation with unprecedented low input power and low-numerical aperture (NA) focusing. Optical trapping efficiencies measured are up to 20× the efficiencies of conventional high-NA optical traps and are among the highest reported to date. Empirically obtained plasmonic optical trapping “phase diagrams″ are introduced to detail the trapping response of the BNAs as a function of input power, wavelength, polarization, particle diameter, and BNA array spacing (number density). Using these diagrams, parameters are chosen, employing strictly the degrees-of-freedom of the input light, to engineer specific trapping tasks including (1) dexterous, singleparticle trapping and manipulation, (2) trapping and manipulation of two- and threedimensional particle clusters, and (3) particle sorting. The use of low input power densities (power and NA) suggests that this bowtie nanoantenna trapping system will be particularly attractive for lab-on-a-chip technology or biological applications aimed at reducing specimen photodamage. KEYWORDS: Plasmonics, optical nanoantennas, optical trapping, particle sorting, optical trapping phase diagrams
C
and convective forces.15 The nonoptical forces are generally regarded as deleterious to the plasmonic optical trap, and previous efforts have aimed at minimizing thermo-fluidic forces.10,17 In addition, current implementations of plasmonic trapping involve confining particles at the surface of a single nanostructure,12,13 which precludes manipulation (e.g., smooth translation) of the trapped objects over an extended plasmonically generated potential energy landscape. In this work, we demonstrate for the first time the use of arrays of Au bowtie nanoantennas (BNAs) for multipurpose optical trapping and manipulation of submicrometer to micrometer-sized objects. In contrast to other plasmonic tweezers, we find that BNAs permit particle trapping, manipulation and sorting utilizing only the optical parameter space, namely, low input power densities, wavelength and polarization. The BNAs are excited using low-NA (0.6 NA) illumination, and we show that counterintuitively, a central requirement for stable, single-particle trapping is to excite the BNAs off resonance. Specifically, we tune the delicate interplay of optical and thermally induced forces to generate the following three distinct states of trapping: (1) lateral
onventional optical tweezers make use of high-numerical aperture (1.2−1.4 NA), oil/water immersion lenses to tightly focus an incident laser exhibiting high spatial-mode purity.1,2 This establishes strong optical gradient forces that confine objects in three dimensions, leading to a host of applications in biology,3−5 colloidal dynamics,6 particle sorting,7 and lab-on-a-chip technology.8 In these applications, the diffraction limit sets an upper bound on the achievable intensity gradient, and in turn limits the maximum optical trapping force for a given input power. In an effort to circumvent this problem, recent efforts have turned toward the near-field confinement properties of plasmonic nanostructures.9 There has been several exciting developments in the field of plasmonic optical tweezers.10−15 Compared to conventional tweezers, the plasmonic counterpart utilizes nanostructures or nanoantennas to create high local-field enhancements and to provide flexibility in shaping the optical potential energy landscape,13 attributes that have been shown both theoretically11 and experimentally to be more suited for trapping nanoparticles,12,16 investigation of colloidal dynamics,14 and carrying out basic studies on the interplay between near-field optical and thermal forces.10 Plasmonic tweezers typically rely on resonant excitation of the nanoantennas, which produces large localized intensities and enhanced optical gradient forces as well as significant heat generation, leading to thermophoretic © 2011 American Chemical Society
Received: October 28, 2011 Revised: December 21, 2011 Published: December 30, 2011 796
dx.doi.org/10.1021/nl203811q | Nano Lett. 2012, 12, 796−801
Nano Letters
Letter
delocalization, (2) single-particle trapping, and (3) multiple trapping of two-dimensional (2D) and three-dimensional (3D) hexagonally packed clusters. We refer to these states as phases and develop novel optical trapping “phase diagrams” to characterize the BNA trapping behavior over the optical parameter space and as a function of nanostructure geometry. The phase diagrams are produced empirically to classify the phase behavior over the full parameter range; they are used to engineer specific trapping tasks including dexterous, highly efficient particle manipulation, and particle sorting for a fixed nanostructure design (e.g., array spacing). In doing so, we show that BNAs extend the capabilities of current plasmonic sorting techniques. Moreover, BNAs are well suited to biological applications since they provide efficient particle trapping and manipulation performance with low power densities at the trapped particle location (108-109 W/m2), thereby minimizing specimen damage.13,18 The BNAs consist of two equilateral triangles, each with a 120 nm tip-to-base height, placed tip-to-tip and separated by a 20-nm gap as described elsewhere.19 Large 80 × 80 μm2 of 50nm thick Au BNAs are fabricated onto a 25-nm indium tin oxide (ITO)-coated glass substrate using e-beam lithography. A 100-nm polymethyl methacrylate (PMMA) (950 K, 2 % in anisole) layer is spin-coated on the substrate and baked at 200 C. Electron beam exposure is performed at 50 kV accelerating voltage with 100 pA beam current. After e-beam exposure and subsequent development, a 3-nm Ti adhesion layer and 50 nm of gold are deposited with an e-beam evaporator. Metal lift-off is performed by soaking the sample in acetone for 30 min. The BNAs are placed with center-to-center spacings of 425 × 425, 475 × 475, 525 × 525, and 575 × 575 nm with corresponding bowtie number densities of 10.3, 8.8, 7.6, and 6.7 bowties/μm2, respectively. Using a 0.6 NA objective (Olympus LUCPlanFLN), one of two interchangeable trapping sources (660 and 685 nm, spatially filtered diode lasers) is focused onto the BNAs from the substrate side. Trapping chambers are made with a 13-mm diameter gasket (Ivitrogen CoverWell) sandwiched between the BNA sample and a 24 × 60 mm microscope coverslip (Corning). The BNAs are immersed in an aqueous solution containing 0.5, 1.0, or 1.5-μm diameter polystyrene spheres (or combinations thereof) and both a galvanometer-based scanner and the microscope stage are used for precise control of the beam and trapped-particle trajectories. Stable optical trapping is attained due to the local field enhancement that establishes a sufficient axial intensity gradient to overcome the optical scattering force as well as destabilizing axial thermal and convective forces.11,15 In order to assess the strength of the single-particle trap, we measure the optical trapping efficiency given by
ε(a , h) =
8 ⎛⎜ h − a ⎞⎟ − 0.9588 ln 15 ⎝ a ⎠
(2)
is the lubrication value of the fluid-force correction, which accounts for the particle radius and its height (h) above the substrate (valid when h < 100 nm).21 A value of 25 nm is assumed for the calculation of trapping efficiencies. At heights >25 nm, the local intensity rapidly approaches the incident intensity (Supporting Information Figure S1), indicating that the particle does not experience the influence of the near field enhancement and will scatter axially due to the low-NA illumination. Figure 1 shows the experimentally measured trapping efficiencies for all parameters considered in this study. Each
Figure 1. Experimental trapping efficiencies. (a) Trapping efficiencies for a 685-nm, horizontally polarized input beam. Blue, green, and gray bars correspond to 0.5, 1.0, and 1.5-μm beads, respectively. The inset shows efficiencies for a vertically polarized input with red, tan, and gray bars corresponding to 0.5, 1.0 and 1.5-μm beads, respectively. (b) Trapping efficiencies for a 660-nm, horizontally polarized beam. The inset indicates vertically polarized results.
efficiency value results from the average of 15 force measurements taken over 3 power levels. The measured optical trapping efficiencies for the horizontally polarized, 685-nm laser and 0.6-NA illumination are given in Figure 1a, and the 660-nm efficiencies are given in Figure 1b; the inset figures indicate the respective vertically polarized results. Here, horizontally (vertically) polarized refers to an input polarization along (perpendicular to) the bowtie long axis. The largest trapping efficiencies are 1.72, 0.65, and 0.27 for the 1.5, 1.0, and 0.5-μm beads, respectively. The measurement error is 1 implies the existence of auxiliary trapping forces. In addition to establishing an extended optical potential energy landscape, the BNAs act as localized heat sources under incident illumination, generating heat energy according to Qheat = σabsI0, where σabs is the effective absorption cross section of the BNAs and I0 is the incident intensity.24 The tight spatial confinement of the electric fields by the BNAs gives rise to large temperature gradients near the illumination point,22 and as a result generates toroidal Rayleigh-Bénard convection currents consisting of a radially symmetric lateral fluid velocity component parallel to the BNA surface and pointing toward the point of illumination as well as a net axial fluid velocity at the point of illumination.25−27 The lateral fluid force associated with Rayleigh-Bénard tends to balance the applied Stokes’ drag force, thereby serving to strengthen the trap along the lateral dimension; the thermally induced forces are likely contributing to the overall trapping force such that Fmax = Foptical + Fthermo−fluidic, where Foptical is the optical gradient force and Fthermo−fluidic is the lateral component of the RayleighBénard drag force plus thermophoretic forces. The efficiency results indicate the extraordinary trapping ability of BNA-based plasmonic optical traps; such high efficiencies permit (large field-of-view) single-particle trapping with as little as 15, 25, and 50 μW at the focal plane for 1.5, 1.0, and 0.5-μm beads, respectively. The corresponding nominal focal power densities are 4 × 107, 6.8 × 107, 1.4 × 108 W/m2, respectively. While BNAs have been shown to produce intensity enhancements on the order of 103 in the gap region,19 an intensity enhancement of ∼1.5-5× is expected at the axial location of the particle above the BNAs for the off-resonance illumination. This leads to a maximum power density on the order of ∼109 W/m2 incident on the trapped particles. In general, the BNA trapping efficiency decreases with lower bowtie number density and is typically larger for horizontal excitation compared to vertical. In addition, the optical trapping efficiency increases with particle diameter, which is consistent with conventional trapping results.20 However, it is difficult to pinpoint an exact trend in trapping efficiency over the parameter space since it is dictated by the interaction of the particles with the complicated optical potential energy landscape established by the BNAs as well as thermal and fluid forces.11,14,15 The exact nature of the trapping efficiency 798
dx.doi.org/10.1021/nl203811q | Nano Lett. 2012, 12, 796−801
Nano Letters
Letter
Figure 2. Trapping phase diagrams. (a) Phase diagram for 0.5-μm beads with a horizontally polarized, 685-nm input; the inset image displays the single-particle trapping/manipulation capability. (b) Phase diagram for 1.5-μm beads; the inset image shows multiple trapping of a 3D cluster of beads. The indicated y-axis values correspond to the various BNA spacings used. (c) Combination of Figure 2a,b showing the combinations of power and bowtie number density suitable for sorting the two particle species, indicated by the orange regions.
video 1 (si_002.avi)]. This “billiard ball” type stacking is observed for all three particle diameters, and the behavior implies that the near field intensity gradients formed by the BNAs are coupling at least 2 μm into solution. For distances >∼50 nm above the BNAs, the local field enhancement drops to 1 suggesting there is some auxiliary optical effect extending much beyond the range of the BNAs. This effect is consistent with longitudinal optical binding in which the near-field intensity is refocused by sublayer dielectric spheres.28 Another possibility is a variant of the Talbot effect, where the BNAs act as a diffractive element that causes constructive interference peaks along the optical axis.29,30 Figure 2a,b shows the trapping phases for 0.5 and 1.5-μm diameter beads using horizontally polarized, 685-nm illumination (the full set of phase diagrams is available in Supporting Information Figures S3 and S4). The vertical coordinate axis indicates bowtie number density and the horizontal dotted lines correspond to experimentally available number densities (array spacings). Single-particle trapping, focal-power threshold values range from 66-115 μW (4.4-7.7 × 107 W/m2) for 0.5-μm beads and 38-95 μW (2.5-6.3× 107 W/m2) for 1.5-μm diameter beads and the corresponding multiple-trapping-only power thresholds are 598-1064 μW (4-7.1 × 108 W/m2) and 361-390 μW (2.42.6× 108 W/m2), respectively. Note that the reported power densities are at the output of the objective lens and the error in the power values at the phase transitions is 1 W) input lasers to ensure stable trapping at all sites in the array.34,35 Therefore, a combined BNA-HOT system could take advantage of simpler laser sources (e.g., laser diodes) to study in-plane colloidal dynamics, or to form an optical sieve for particle sorting by both the selective trapping and rejection mechanisms detailed above. In this manner, the BNAs could be dynamically reconfigured for particle sorting by simply adjusting the input power or polarization, thus removing the need for structures specifically designed to trap only one type of particle.15 Overall, BNAs offer an advanced plasmonic trapping scheme that combines the high trapping efficiency and particle sorting capabilities of other plasmonic trapping systems into a single, multipurpose trapping platform with the additional, unique ability to form and manipulate three-dimensional, selfassembled clusters of particles. This phenomenon involves the coupling of optical near fields (in the BNA gap regions) a distance >2 μm into solution and may indeed be the first demonstration of longitudinal optical binding in a plasmonic trapping scenario. We have introduced plasmonic optical trapping phase diagrams to condense a wide range of trapping parameters into a set of useful diagrams that allow one to engineer multipurpose optical trapping tasks using BNAs. The trapping phases are governed by the interplay of thermal and optical forces and thus the trapping phase diagram concept may find use in other trapping scenarios involving heat generation, such as trapping plasmonic nanoparticles.31 Furthermore, a generalization of the phase diagrams could be useful for plasmonic nanoantenna systems that are coupled, for example, to a molecule of interest or a solar cell in order to take advantage of a preferred reaction pathway (e.g., frequency conversion)19 in the presence of deleterious heating effects.
■
ASSOCIATED CONTENT
S Supporting Information *
FDTD simulation results, full set of optical trapping phase diagrams, and tabulated trapping efficiency results. This material is available free of charge via the Internet at http:// pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Author Contributions ⊥
These authors contributed equally to this work.
Notes ¶
Affiliate in the Department of Electrical and Computer Engineering, and Bioengineering
■
ACKNOWLEDGMENTS The authors acknowledge support from the National Science Foundation (NSF ECCS 10-25868), and the University of Illinois at Urbana-Champaign research start-up funds.
■
REFERENCES
(1) Ashkin, A.; Dziedzic, J. M.; Bjorkholm, J. E.; Chu, P. Opt. Lett. 1986, 11, 288−290. (2) Neuman, K. C.; Block, S. M. Rev. Sci. Instrum. 2004, 75, 2787− 2809. (3) Fazal, F. M.; Block, S. M. Nat. Photonics 2011, 5, 318−321. 801
dx.doi.org/10.1021/nl203811q | Nano Lett. 2012, 12, 796−801
