1. Introduction

Optical fiber tapers with a submicron-diameter are a versatile tool for sensing applications. A significant fraction of the light guided through the tapered region is contained in the evanescent field and thus interacts with the medium surrounding the taper. Low loss tapers can be produced [1] that are the basis for highly sensitive absorbance measurements of single atomic layers [2]. Even the fluorescence of a small number of atoms in a magneto-optical trap can be detected [3]. A different sensing concept is to functionalize the surface of a fiber taper in order to make it optically sensitive to changes of the environment. This has been shown for humidity [4] and hydrogen [5] sensors. Optimum control of functionalization is obtained if only a single active particle is placed on a fiber taper. This has been demonstrated, e.g., by mechanically picking-up single diamond nanocrystals with fiber probes [6, 7]. In this paper, however, we present a non-contact method of assembling such a sensor element by landing a preselected micro particle on the tapered region of a fiber with the help of a segmented linear Paul trap. Such a technique is very flexible and particularly appropriate if an extremely clean environment is required.

A Paul trap uses an oscillating electric field in order to trap charged particles [8]. Linear Paul traps are widely-used tools to confine single ions for high resolution spectroscopy [9] or as quantum registers [10] for quantum information processing [11]. Another application concerns controlled implanting of trapped ions as shown by W. Schnitzler et al. [12]. However, also larger particles, ranging from a few nanometers [13] to micrometers [14] have been trapped and investigated. In the following we will demonstrate how a linear Paul trap can be utilized to functionalize a fiber taper sensor in a controlled way.

2. Trapping, storing, and manipulating single particles in a segmented linear Paul trap

To prepare and select single particles for subsequent landing, we use a segmented linear Paul trap. Such a trap confines charged particles in a two-dimensional oscillating electric field created by a geometry of four main electrodes. Confinement in the remaining direction along the trap’s symmetry axis is typically obtained with the help of additional auxiliary electrodes, e. g., rods [15], rings [16] or segments of the main electrodes [17]. In the latter case segmentation is used to define well-separated trapping and spectroscopy regions, respectively. Transport between these regions is obtained by changing the voltage applied to the individual electrodes [18, 19].

A schematic of our trap can be seen in Fig. 1. The trap has a total length of 70mmand consists of four cylindrical electrodes made of brass tubes mounted on glass fiber rods, which act also as a cable guide. The opposing electrodes are electrically connected. A radio-frequency (rf) voltage is applied to one pair of electrodes while the other pair is subdivided into twelve segments with lengths of 2, 5, and 7mm. A varnish painted on the end facets of the segments provide electrical insulation. The electrodes with an outer diameter of 2R=4mm have a minimum distance to the center of the trap of r ₀=3mm. A sinusoidally varying rf-voltage is produced by a function generator and amplified to a voltage amplitude of typically V_{r f}=1kV with a frequency of Ω=2π·5kHz. The additional direct current (dc) voltages are provided by laboratory power supplies and are applied to the segments by a switch box. Typical values are V _2,4-12=20V and V _1,3=0-40V. The electrodes are mounted on two insulating plates inside an air-tight box allowing sufficient optical access. The trap is operated under ambient conditions. Air-friction results in damping of the particle motion [20] establishing a wider stability-region, i.e. the possibility to trap particles with a wider mass-range compared to vacuum. Based on the parameters quoted above and an estimated charge of 10¹⁵C per particle it is theoretically possible to trap polystyrene spheres within the range from 25nm up to 4µm in size. This range can be extended by changing the trapping frequency Ω. With our trap particle storing times on the order of several hours can easily be achieved.

The segmented geometry provides different regions in the trap for different purposes. The particles are inserted at one end of the trap (segments 10–12, see Fig. 1) whereas at the other end, the spectroscopy region (segments 1–3), fluorescence measurements and the deposition on the fiber taper are performed. This subdivision protects the fiber taper from unwanted deposition and, in addition, avoids contamination of the electrodes in the spectroscopy region by charged particles, which would modify the trapping potential in an uncontrolled way.

Charging and insertion of particles occurs via electrospray injection (ESI) [21]. Any kind of particles, which can be brought in suspension (e.g. water, ethanol), can be injected with this method. A high voltage of about 2kV is applied to the metallic capillary (inner diameter of 110µm) of a syringe containing the suspension. Charged droplets leave the metal capillary along a potential gradient oriented towards the trap. The droplets reduce in size due to evaporation. When the Coulomb repulsion exceeds the surface tension, fission of the droplets (Coulomb explosion) occurs [22]. Finally, when the remaining solvent is completely evaporated, the charged particles are captured by the linear trap.

 

Fig. 1. (Color online) Schematic of the Paul trap: The trap consists of two rf-electrodes and two dc-electrodes. The dc-electrodes are subdivided into twelve segments with lengths of 2, 5 and 7mm. The total length of the trap is 70mm. Segments 1–3 constitute the spectroscopy region, where the fluorescence measurements and the deposition on the optical fiber taper are performed. Loading of the trap occurs from the righthand side (segments 10–12.) The segmentation allows confinement of several particles at once, isolation of single particles, and their transfer within the trap.

Download Full Size | PDF

3. Optical setup

In our experiment (setup shown in Fig. 2) we use dye-doped polystyrene beads (Invitrogen, USA, FluoSpheres F8801), which can efficiently be excited via a fiber coupled argon-ion laser operating at 514nm or a frequency doubled Nd:YVO₄ laser at 532nm. In order to preselect a desired particle before placing it onto the taper it is necessary to investigate its optical properties while it is still hovering inside the trap. By flipping mirror FM2 and keeping mirror FM1 upright it is possible to excite a trapped particle through the microscope objective. During first fluorescence studies the fiber taper is removed from the trap and covered up in order to avoid unwanted contamination. A low power microscope objective with a magnification of 4 and NA=0.1 is optimum to find a particle within the trapping region due to its large field of view combined with a large field of depth. The position of the trap can be controlled by motorized translation stages in order to bring particles in the trap center into focus. The fluorescence of a particle is collected by the same microscope objective. A dichroic mirror and additional long pass filters are used to suppress the pump light before the fluorescence is imaged on a sensitive CCD camera or dispersed by a spectrometer. After deposition of a suitable particle, the taper also provides optical access. This can easily be achieved by flipping mirror FM1 and keeping mirror FM2 in the upright position. The flexible setup allows rapid switching between both detection modi.

A photodiode was added to the setup in order to measure the change of fiber transmission while a particle is placed on the taper. In this case mirror FM1 has to be flipped and mirror FM3 has to be upright. As indicated in the figure, the Paul trap and the taper are contained in an airtight box, in order to reduce air turbulence.

Optical fiber tapers are produced by heating a section of a standard optical fiber while pulling the unheated ends apart. In this process the elongated region is thinned down to the desired diameter. There exist various approaches to heat the fiber, e.g., using a CO₂-laser [23, 24] or flame-heating [25, 1]. We use a homemade pulling setup based on a small ceramic heater similar to [26]. A variety of fiber types can be used and tapers with a diameter down to a few hundred nanometers can be produced.

 

Fig. 2. (Color online) Schematic of optical setup: solid (green) and dashed (red) lines indicate the optical paths for the excitation laser and fluorescence detection, respectively. By flipping either mirror FM1 or FM2 it is possible to excite and detect particles hovering in the trap through the microscope objective or to measure deposited particles directly through the optical fiber taper. FM3 allows measurement of the change in transmission via a photodiode as a particle is placed onto the taper.

Download Full Size | PDF

4. Deposition of preselected polystyrene particles on fiber taper

In the following we will present the deposition of a single dye-doped polystyrene particle. The entire procedure is illustrated in Fig. 3 and by a series of short movie clips (Media 1–4)[27]. Initially, a suspension of 100nm-sized polystyrene spheres doped with a fluorescent dye and dissolved in ethanol is prepared by sonication and loaded into the ESI-syringe for injection. The concentration of the suspension is chosen to be 2×10⁷ particles/ml. The particles were negatively charged by a voltage of 1.7kV applied by the ESI and injected in the trap.

Several particles can be loaded into the trap simultaneously and can then be transferred between different trapping regions in the segmented trap as described in section 2. An additional He-Ne laser (not shown in Fig. 2) beam is directed along the axial center of the trap in order to observe the particles via light scattering. In this way trapped particles can be imaged to the CDD camera. This greatly facilitates the manipulation procedure.

A single trapped particle can be seen in Fig. 3(a) (Media 1) hovering in the middle of the trap. In the next step the particle is moved into the focus of the microscope objective as shown in Fig. 3(b) (Media 2). A photoluminescence spectrum of the particle is recorded and plotted in Fig. 4(a). This procedure allows to preselect a particle based on its spectral properties. If a particle is not considered to be suitable, it can be easily ejected by switching off the axial confinement potential at the end of the trap. Then, another particle from the storage region at the other side of the trap can be transported to the spectroscopy region to repeat the measurement.

Once a suitable particle is found, all other particles are removed from the trap and the selected particle is moved into the storage region, before the fiber taper is uncovered and moved slowly into the trap (see Fig. 3(c) (Media 3)).

The fiber taper can be observed via scattered light from an argon-ion laser beam coupled into the taper. In Fig. 3(c) the fiber taper can be identified as a thin line glued into a U-shaped holder. The taper used for this experiment has a diameter of 700nm. Such a thin diameter was chosen in order to increase the evanescent field [2] as well as to ensure a high coupling efficiency of fluorescence light into the fundamental mode of the taper.

When the taper is placed within the trap, the selected particle is transported back to the spectroscopy region. By adjusting the trap position with respect to the taper and changing the electrode voltages it is possible to move the particle close to the taper, where it is eventually attracted and landed on the taper. The success probability is nearly 100%. Once the particle is placed it can be observed as a bright scattering center on the taper as shown in Fig. 3(d) (Media 4).

 

Fig. 3. (a) Particle after electrospray injection trapped between electrodes (Media 1), (b) particle moved into spectroscopy region of the trap (Media 2), (c) optical fiber taper is uncovered and moved into trap (Media 3), (d) particle has landed onto fiber taper (Media 4). Annotation: i. trapped particle, ii. microscope objective, iii. cover for optical taper, iv. segmented electrode, v. high-voltage electrode, vi. holder for electrodes, vii. fiber optical taper (seen due to light scattered out of a guided fiber mode), viii. U-shaped holder for optical fiber taper. [27]

Download Full Size | PDF

We found that only negatively charged particles can be attracted by the taper. This is due to positive charges on the glass surface since a repulsive effect could be observed to positively charged particles.

At this point the taper provides optical access to the particle. By coupling the argon-ion laser into the taper it is possible to excite the deposited particles on the surface via the evanescent field of the taper. At the same time, the taper collects the fluorescence of the dye-doped particle. Figure 4(b) shows the fluorescence spectrum measured via the fiber taper. Obviously, both spectra, taken by the microscope of a freely hovering particle in the trap and the spectrum collect via the fiber of a particle landed on the fiber (Fig. 4(a) and (b)) coincide and show the typical emission characteristics of the dye.

The reliable procedure of selection and placement of particles can be repeated also for one and the same taper as shown in Fig. 5. The four particles are evenly spaced along the taper, where the minimal distance between the particles is given by Coulomb repulsion due to their similar initial charge from the ESI process. The lateral precision of the placed particle is limited by the motion of the particle in the trap. For our trap the observed amplitude of motion is of the order of 1–2µm. However, as seen in Fig. 5, the surface charge distribution has a major influence when landing the particles. In future experiments, electrostatic guiding of particles can be envisioned by modification of the charge distribution on the taper, e.g., by irradiating with focused UV light. This would also allow for a guided and more precise deposition on more complicated structures than optical fiber tapers.

Alternative methods to manipulate and deposit nano-sized objects using scanning probe techniques have been demonstrated recently [28, 29, 30]. For example, single nanodiamond crystals where picked-up and deposited using fiber tapers [6, 7]. However, it is important to point out that our method is non-contact. Thus no mechanical stress is applied which may cause damage to the fiber taper. Our method is of particular relevance for particles which require an extremely clean environment. For example, particles sensitive to oxygen (or other abundant gases) can directly be spray-injected from a buffer solution in a Paul trap in vacuum. With also the taper in vacuum the assembled system would have never been exposed to air. Also, further contamination of the taper due to parasitic particles transferred in case of the mechanical contact techniques can be avoided, since only trapped and spectroscopically preselected particles are deposited. Finally, having a scanning probe in a clean environment, not to mention vacuum, is much more cumbersome. A trade-off is precision of positioning which is expected to be better with scanning probes.

 

Fig. 4. (a) Fluorescence spectrum of a dye-doped particle trapped in the spectroscopy region of the Paul trap collected by the microscope objective, (b) spectrum of the same particle after deposition onto a fiber taper of 700nm in diameter collected via the same fiber. The excitation power was P_ext=30µW at 514nm.

Download Full Size | PDF

 

Fig. 5. Four particles placed evenly spaced onto a fiber taper. The distance between the particles on the taper is roughly 2mm.

Download Full Size | PDF

We investigated the landing process of a single particle in more detail. In order to do this we measured the transmission of a frequency doubled Nd:YVO₄ laser at 532nm through a fiber taper of 850nm in diameter during the landing. Figure 6(a) shows a sudden decrease in transmission corresponding to additional scattering by the landed particle. The transmission changes to 46% which indicates that in this experiment the single trapped and landed particle corresponds in fact to a cluster of several 100nm-sized polystyrene beads. For smaller objects of a few hundred nanometer in diameter (corresponding to a single bead) a much smaller change in transmission would have been expected [31]. The size of the landed particle cluster in this experiment was measured subsequently using a microscope. Figure 6(b) shows the micrograph indicating a particle size of 1.5µm.

 

Fig. 6. (a) Decrease in transmission at 532nm while landing a single 1.5µm-sized particle consisting of a cluster of polystyrene beads on a 850nm diameter taper. The transmission is normalized to the transmission through the taper before the particle is placed (corresponds to 1.0). The value of 0 corresponds to the transmitted signal when the laser is turned off. (b) Microscope image of the landed particle.

Download Full Size | PDF

In order to confirm the reduction of transmission due to landing of a particle we performed FDTD simulations (software by Lumerical inc., Canada). We modeled the problem as single mode transmission through a glass cylinder piercing a larger polystyrene sphere which corresponds well to the experimental system as seen in fig. 6(b).

Figure 7 shows the results for a glass cylinder of diameter d_t=850nm and length 100µm and different sizes of the polystyrene cluster. The correspondence is convincing when taking into account deviations of the bead cluster from a perfect sphere as well as possible guidance of light via higher order modes which are subject to higher scattering losses.

In addition to the robustness and flexibility of a single active particle coupled to a fiber taper, there is also an improved fluorescence yield compared to collection of fluorescence from a freely hovering particle via a microscope objective. This is clearly seen by comparing the different count rates of the spectra in Fig. 4 for collection of fluorescence from a particle trapped in the spectroscopy region of the trap via the microscope objective (a) and collection of fluorescence from a landed particle on the tapered fiber via the fiber itself (b). In both cases the same pump power of 30µW at 514nm was used, but the fluorescence signal collected through the taper from the landed is 11 times larger than the signal collected through the microscope objective from the trapped particle. An order of magnitude estimation can be performed to explain this difference.

One aspect concerns excitation efficiency. The focal spot of a free beam after our microscope objective has a theoretical minimum diameter of 17µm. For a 1.5µm sized particle the areal overlap is only 0.8%. A similar argument can be made for the particle attached onto the fiber by considering the evanescent field surrounding the taper. For a 700nm taper in diameter at a wavelength of 515nm about 4% [32] of the power is guided outside the taper and is available for exciting the particle. In our case the overall transmission through the taper including coupling and fiber losses were measured to be 6.3%. Multiplying this by the fractional evanescent field we found a number of 0.25% to be compared to the areal overlap in the free beam excitation. Thus, excitation via the tapered fiber should be approximately 3 times less efficient.

 

Fig. 7. (Color online) Results of FDTD simulation for the transmission for the HE11 mode at 532nm in a glass cylinder d_t=850nm of length 100µm piercing a polystyrene sphere of diameter d_s. The transmission through the glass cylinder is given by the time averaged Poynting vector integrated across the end facet of the cylinder and normalized by the power of the seeded mode. The unit cell size of the simulation is 33nm and perfectly matched layers are used as boundary condition.

Download Full Size | PDF

The other aspect concerns the collection efficiency of the fluorescence emission. The microscope lens of NA=0.1 collects only 0.25% of the emission of an arbitrarily oriented dipole. When comparing to collection via the fiber one has to take into account that only a thin layer around the taper will couple evanescently to the guided mode in the taper. In a cluster of dyedoped beads those ones closest to the surface will couple most efficiently while the coupling efficiency exponentially decreases with the distance from the fiber surface. Based on our FDTD simulations for a 700nm taper the collection efficiency of emission from an arbitrarily oriented dipole on the surface and at a distance of 100nm from the surface is 21% and 8%, respectively. Thus, as an estimation, the emission from a 100nm layer around the taper with an average coupling efficiency is considered to be 15%. For such a taper and a spherical cluster of 1.5µm in diameter enclosing the taper approximately 5% of the total fluorescence of the cluster is collected. This is about 20 times larger than collection via the microscope objective.

By combining the estimation for the excitation and the enhanced fluorescence collection a factor of 7 improvement for fluorescence signal in case of the particle on the fiber taper compared to detection via the microscope objective is expected. This is slightly underestimated, but a convincing order-of-magnitude agreement with the observed factor of 11. The estimation confirms the experimental observation that a particle attached on a fiber taper establishes a more efficient configuration for a sensor application than using simply a single particle alone and observing it with a long-distance microscope objective as used in our experiment. This is another benefit in addition to the inherent robustness of a fiber coupled configuration.

5. Conclusion

We demonstrated a novel technique that allows transfer of preselected particles from a linear Paul trap onto an optical fiber taper. The measured collection efficiency of light from fluorescence particles landed on the taper exceeded the detection of a freely hovering particle via a microscope objective by an order of magnitude. FDTD simulations confirmed this result.

In principle, a great variety of particles that can be brought into a suspension can be trapped, preselected, and landed on fiber tapers or on other fragile micro- and nanophotonic components. The procedure can also be repeated in a reliable way with different particles. Therefore, our method allows a versatile non-contact functionalization of fiber-coupled photonic components with active particles, e.g., for sensing applications or as fundamental light sources, such as single photon sources.