1. Introduction

Conventional optical tweezers focus laser beams to achieve trapping of micron and sub-micron particles [1]. It is extremely difficult to trap particles less than 100 nm in size with reasonable powers due to diffraction limitations. Even optical fiber probes, which remove the need for a cumbersome microscope setup, are still subject to diffraction and can only trap relatively large particles [2–8]. To overcome this diffraction limitation researchers have exploited subwavelength nanoapertures that have the ability to locally confine a high field intensity in a subwavelength scaled volume [9–12].

Nanoaperture tweezers allow for trapping < 100 nm particles [13–18], down to single digit nm range (including proteins) [19–21]. They still suffer from requiring a microscope and precise optical alignment. That is why people have worked on integrating nanoapertures (and other metal nanostructures) on the ends of optical fibers [22–33]. Most of these past works did nanofabrication on the tip directly, which is complicated and not scalable. (One exception is in [32] where there was no nanofabrication on the tip, but the fiber was randomly metal-coated to achieve random nanostructures). For the broad adoption, however, it is desirable to use instead a standard planar nanofabrication approach, so the challenge is to use a standard planar fabrication approach along with fiber tips.

To address this challenge, we consider transferring a planar nanostructure to the tip post fabrication. Several approaches aiming to transfer metal nanostructures onto different substrates have been demonstrated. These include, template-stripping of arrays of gold nano-tip wedges evaporated on a sharply shaped by etching silicon substrates [34], and a template-stripping of different shaped-array nanostructures (like holes, wires, pyramids, etc.) from a planar silicon substrate onto a stretchable polydimethylsiloxane [35]. A near-field imaging probe (sharp gold apex on a tungsten wire) has been demonstrated using template stripping [36]. We have also used template stripping to make double nanohole structures for optical trapping [17].

Here, we use a carefully aligned stripping approach to get a nanoaperture in a metal film aligned with the core of a standard single mode fiber cleaved end. The integrated nanoaperture fiber tweezer (NAFT) in our approach was directly nanofabricated on a metallic film and then transferred by a stripping procedure. The standard planar nanofabrication that allows for broad adoption, as opposed to nanofabrication on the fiber tip which is cumbersome because it requires mounting fiber tips in a focused ion beam (FIB) milling instrument.

2. Fabrication

Figure 1 shows a scanning electron microscope (SEM) image and a schematic (not to scale) diagram of the NAFT. We used a bow-tie nanohole that allows for two transmission peaks, one less than 1 micron and one at 1550 nm, which is a standard telecom wavelength. We estimate the thickness of the epoxy layer from side-view imaging on SEM (not shown).

 

Fig. 1 (a) Scanning electron microscope (SEM) image of the integrated NAFT. The inset shows the plasmonic aperture milled at the center of the NAFT; the scale-bar is 0.5 µm. (b) Schematic diagram of the integrated NAFT (not to scale). The epoxy used is the Norland optical adhesive 61 (NOA 61); a photopolymer liquid that cures when exposed to ultraviolet light.

Download Full Size | PDF

Figure 2(a) shows the various fabrication and integration steps for the NAFT. In step 1, we evaporate 100 nm thick gold films on a clean glass substrate without an adhesion layer. Figures 2(b) and 2(c) show step 2, where we used the focused ion beam (FIB) to mill rings of 125 µm diameter and the nanoaperture. The 125 µm diameter was chosen to match the diameter of the fiber. The gradually tapered gap nanoaperture was designed by finite difference time domain (FDTD) simulations and placed at the center of the ring to a few micron accuracy. The ring was fabricated at 1 k × magnification, whereas the aperture at 35 k × . The gold appendages appear adjacent to the circular gold film in Fig. 2(b) are for discharging the tip when later imaging in SEM. They also serve as alignment markers to know the orientation of the nanoaperture. Step 3 shows that multiple rings with apertures were milled in a single sample, allowing for production tens of tips from a single nanofabrication run.

 

Fig. 2 (a) Schematics for the different NAFT integration steps. 1: evaporated 100 nm gold (Au) film on an optically transparent, glass, substrate sample (no adhesion material). 2: single structure (a 125 µm ring plus the central nanoaperture) milled on the Au sample. 3: multiple structures on one Au sample. 4: an ultra-violet curable epoxy immersed single mode fiber aligned with the nanohole in the Au film. 5: strip-off the Au film, and 6: shows the integrated NAFT fixed within a protection plastic cone. (b) SEM image of a 100 nm Au film after being milled using the FIB, the ring inner radius is 125 µm and the outer radius is 132.5 µm. (c) SEM image for the plasmonic aperture. The aperture gap is ~70 nm along the y-axis. The black arrow shows the optimum polarization for maximum field confinement and transmission. (d) Schematic for the setup used to integrate the NAFT. LD is laser source (980 nm), EAC is azimuthal and elevation angle controller, LDM is long distance microscope, UV is ultra-violet light source, CL is collimator, MMF is multimode fiber, and OSA is optical spectrum analyzer.

Download Full Size | PDF

The critical alignment comes in step 4, using the setup in Fig. 2(d). First the fiber is aligned to be perpendicular to the gold film. We retracted it using the upper 3D-stage and inserted a glass-slide with 5 µl drop of UV curable epoxy (Norland NOA 61). The cleaved-fiber end was immersed in the epoxy drop for a few seconds. We removed the glass-slide containing the epoxy and moved the fiber to within ~300-500 µm of the gold film. The top and bottom 3D-stages were used to align the fiber with a ring in the gold film. A laser diode was used to send light through the fiber, which was then transmitted through the aperture and monitored by an optical spectrometer. Further alignment was performed by maximizing the signal intensity from the laser diode and bringing the epoxy in contact with the gold film. The fiber polarizer was also used to maximize the field intensity detected which can be observed when the electric field vector is polarized along the y-axis of the antenna aperture. At the point of contact, the epoxy was with a UV light for about 5 minutes. It is important not to apply too much pressure at contact, otherwise the epoxy goes into the aperture.

As shown in step 5, we retracted the fiber with the gold film attached to the tip. A plastic cone was used to protect the fiber (step 6). The gap between the fiber-head and the plane of the cone-base can be purposely adjusted. The fiber we used had an 8.2 µm core-diameter and 125 µm cladding-diameter (Corning SMF-28). We recognize that SMF-28 is not a single mode fiber for 980 nm. This choice is motivated by the desire to perform measurements on fluorescent nanoparticles that emit in the 1550 nm telecom window so that the trapping laser can also serve as an excitation source and the fiber can serve as a detection channel.

3. NAFT test for trapping

Figure 3(a) shows the setup used to trap with the NAFT. The setup comprises a fiber-coupled 980-nm laser, fiber polarization control paddles, the integrated NAFT, a solution glass bottle, a neutral density filter and a femtowatt photodetector (Thorlabs PDF10A). The neutral density filter was used to prevent saturation of the photodetector. The setup does not use any microscope imaging and is compact. In the future, the setup may be revised with translation stages for translocation [24].

 

Fig. 3 (a) Schematic of the experimental setup used for trapping 20 nm and 30 nm polystyrene nanospheres, where I(λ) (red pulse) represents laser signal and v(t) (purple waveform) represents output voltage. (b) 20 nm particle trapping signal record. Trapping event in this time-window occurred after 13.585 seconds from turning on the laser. The inset is just an enlarged view of the trapping jump. (c) 30 nm particle trapping signal record. Trapping event in this time-window occurred after 44.175 seconds from turning on the laser. The inset is an enlarged view of the trapping jump. (d) 30 nm particle release signature in response to turning off the laser power. More about power dependence of trapping can be found in [37].

Download Full Size | PDF

We used the same integrated NAFT for trapping two different sizes polystyrene nanospheres (refractive index 1.5731 [38]) of diameters 20 and 30 nm suspended in a 0.1 (v/v %) deionized-water solution. The optical power incident on the nanoaperture was estimated to be 0.5 mW/µm² for 50 mW of laser power. The polarization was controlled to be along the y-axis (see Fig. 2(c)). Figures 3(b) and 3(c) show time-windows of the two nanoparticles trapping signals detected at the photodetector and recorded through a data-acquisition-unit. Both insets show small jumps of around 0.7% and 0.9%. A particle trapping release signature is also illustrated in Fig. 3(d) for the 30 nm polystyrene bead when the laser power is turned off and on. With deionized water alone, no jumps were observed. The experiment was repeated to achieve multiple trapping events for each particle size and on multiple fibers.

Although, in the past, our group and others have identified multiple trapping events by a double step [15,39], we have not noticed such multiple trapping events throughout the test. We think that due to charge repulsion of the nanoparticles, multiple particle trapping is rare. The transition step in trapping has a similar duration as found in our past works (of the order of 1-10 ms) [16,37]. This suggests that the trapping stiffness is similar. Further studies are required to quantify fully the power dependence of trapping and the autocorrelation.

It is worth noting that we have fabricated 23 integrated structures with the current method with a working (trapping) yield of 13. The structures that did not work had alignment issues (nanoaperture with core) or damage to the gold surface.

4. FDTD simulation results

To understand the positive jumps observed on the photodetector output voltage, we used the FDTD numerical software, Lumerical Solutions Inc., release 2017b, version 8.18.1365. The change in far field electric field intensity can be investigated by positioning a 25 nm polystyrene nanosphere in the aperture gap. As it is known, the far field of an antenna can be measured by probing the near fields over an enclosing surface [40–42]. Numerically, however, we can record the near field amplitude and phase at very close points on a virtual surface that can be located very close to the near field source.

Here, we used a 5 µm × 5 µm × 2.3 µm 3D FDTD simulation region divided, along the z-axis, into three main material regions: the epoxy region, 1.2 µm, the gold film region, 0.1 µm, and the water region, 1.0 µm, with the nanoaperture centers the simulation region in the xy-plane. The aperture size extends 225 nm in the x-direction, 175 nm in the y-direction, and 100 nm in the z-direction. To excite the structure, we used a y-polarized 980 nm planewave source located in the epoxy region 1.0 µm away from the gold film. To calculate the near field, however, we used a virtual planar surface (monitor) that is only 2.5 nm away from the nanoaperture surface in the + z-direction. The total number of the 3D cells generated is 52,032,946: 252,000 cells of them are uniform of total size 225 nm × 175 nm × 100 nm and we defined these cells to fill the aperture region for the sake of improving simulation accuracy; the size of each individual cell of these cells is 2.5 nm × 2.5 nm × 2.5 nm. In addition, we set the simulation accuracy to level 5, time stability factor to 0.95, auto shutoff to 1 × 10⁻⁵.

We have approximated the taper angle in the FIB process by noting the top and bottom borders found in the SEM image as shown in Fig. 2(c). Several designs were attempted (rectangular, double-nanohole, rectangular with cusps) and the bow-tie gave the best results. The taper can also impact trapping efficiency, as has been studied elsewhere [43].

Figure 4 shows the normalized far field power emerging from the aperture, calculated within a cone specified by the photodetector aperture, without and with the nanoparticle (25 nm). The observed step in the power pattern is around 0.1% when the particle enters the aperture. This is less than the average observed power jump of around 0.8%, but reasonable considering that we did not place the particle at the highest intensity point (in order to exploit symmetry), and that there are differences between an idealized structure and the one actually fabricated.

 

Fig. 4 x-plane (blue lines) and y-plane (red lines) far-field normalized power for the no-particle (dash-dot lines) and with-particle (solid lines) cases. Patterns are FDTD calculated 50 mm away from the nanoaperture for a half-cone-angle of 0.63° dictated by a photodetector aperture’s diameter of 1.1 mm. Inset (a) shows an enlarged view of a central part of the yz-plane (x = 0 plane) of the 3D simulation region with the 25 nm particle located in the tapered aperture. Insets (b), (c), and (d) show the no particle (NP), with particle (WP), and the difference (Diff.) in electric near field intensity at 2.5 nm away from the aperture water-side surface.

Download Full Size | PDF

5. Conclusion

In conclusion, we have demonstrated a simple template stripping approach to integrate a nanoaperture fiber trap without the need for fabrication directly on the tip of the optical fiber. The NAFT replaces bulky microscope optics typically used in trapping with a fiber and a near-field aperture. We demonstrated trapping of 20 nm and 30 nm polystyrene particles, and the experimental results are supported by FDTD simulations. Moving forward, we expect that this approach will be useful in the trapping and translocation of nanoparticles [24], as well for trapping fluorescent nanoparticles and monitoring their emission directly within the fiber.