1. Introduction

Optical tweezers were first demonstrated in 1986 by Ashkin et al. [1] and they are now routinely used to trap and manipulate micron-sized particles using a single laser beam. The trapping mechanism relies upon the extremely high electric field gradient produced near the beam waist of a tightly focused laser beam which draws a dielectric object towards the region of highest light intensity i.e. the beam focus. For micron sized particles, a few milliwatts of focused laser light creates a gradient force sufficient to overcome both the scattering and gravitational forces, thus creating a stable 3D optical trap. Optical tweezers are a powerful tool in biosciences since biological objects are easily trapped and the trapping mechanisms for a transparent, dielectric particle apply [2, 3]. The trapping prevents the particles from drifting out of the laser beam due to Brownian motion and also permits one particle to be studied at a time rather than an average signal from numerous species.

Previously, standard Raman spectroscopy and optical tweezers were combined with a near infrared trapping laser as the excitation source to study living biological cells [4]. However, only about one in a million photons is Raman scattered so the resulting spectral emission is weak, and consequently long acquisition times are required, potentially resulting in radiation damage to the sample. Optical tweezers have also been incorporated within a confocal microscope to study resonance Raman (RR) of biological cells [5]. In this example an infrared (830 nm) tweezing laser was combined with a confocal microscope incorporating an Argon/Krypton laser to produce the excitation wavelengths necessary to excite the resonance Raman response. However, this required a dual system comprising an optical tweezers and a confocal microscope, which separated the optical paths of the trapping and excitation lasers below and above the sample.

The technique of surface enhanced resonance Raman scattering (SERRS) uses a roughened surface of a suitable metal to enhance the signal due to an interaction between the adsorbed molecular species and the surface plasmon [6]. By choosing an excitation wavelength that is resonant both with an absorption maxima of the chromophore and the surface plasmon of the metal, the SERRS signal can be up to 10¹⁴ times greater than the Raman one. The sensitivity of SERRS can be even greater than fluorescence [7], allowing detection of single molecules [8, 9]. SERRS has a key advantage over fluorescence in that the resulting spectral peaks are clearly resolved from each other. This makes it easy to distinguish simultaneously between different chromophores in a mixture, with the resulting wavelength shifted spectra containing information on the molecular structure.

Trapping metallic beads to create a combined SERRS and optical tweezers system provides an attractive platform for microsystems and microfluidic technologies in which sensory particles can be independently manipulated, both with respect to each other and/or other entities. The trapping of metallic particles is more difficult than dielectric particles due to the increased scattering forces arising from their highly reflective surface [10, 11, 12, 13]. Our silica particles have discrete silver deposits on their surface, which provide the SERRS active site. However, the partial coating means that the particles are sufficiently transparent to allow light to refract through the sphere, enabling the partially silvered micro-particles to trap in a similar way as dielectric particls i.e. in the intensity maximum of the laser beam [14]. In our earlier work, the intensity of the 532nm laser beam, required for optical trapping, rapidly photobleached the chromophore within 100 ms making it unsuitable for monitoring/sensing applications [15].

We now report an elegant solution to this problem, demonstrating that a sustainable SERRS spectrum can be achieved by trapping a partially silvered, dye coated particle in a TEM₀₀ optical trap. The optical trap is formed using a 1064 nm laser, typically operating with an output power between 500–1000 mW. Prior to coupling the laser into the microscope objective, a frequency doubling KTiOPO₄ (KTP) crystal is inserted at an intermediate beam waist to produce 4–16 µW of 532 nm co-linear light. Both the IR and the green wavelengths suffer a loss within the coupling optics and objective lens of the order of 50% before reaching the trapping plane. This second harmonic generated light, at 532 nm, excites a surface plasmon in the silver, which interacts with the chromophore, eliminating the need for separate trapping and excitation lasers and the associated optical complexity.

2. Experimental set-up

The experimental set-up is shown in Fig. 1. The optical tweezers trapping laser was a continuous-wave, variable power Nd:YAG laser emitting at 1064 nm. The beam was expanded, collimated and passed through an afocal telescope, into the intermediate focal point of which was placed a frequency doubling KTP crystal that generated microwatts of power at 532 nm. This combined beam was passed through a high numerical aperture objective lens, NA 1.3, bringing both the 1064 nm and 532 nm light to a common focus. The sample cell was mounted on a piezoelectric stage with 100 µm of travel in the x, y and z directions, in an inverted microscope configuration allowing easy access to the sample plane. Particles were viewed using a CCD camera, with unwanted wavelengths filtered out. The SERRS spectrum was recorded using an Ocean Optics R2001-532 Raman spectrometer fiber coupled to collect light from the image plane of the microscope. The fiber diameter of 600 microns corresponds to a diameter in the sample plane of 6 microns, therefore spectra are from single particles. The beam steering mirror, was reimaged onto the back aperture of the objective lens, allowing the the optical trap to be positioned such that the SERRS emission was imaged onto the spectrometer collection fiber.

 

Fig. 1. Schematic diagram of experimental set-up.

Download Full Size | PDF

The particles were 1.5 µm diameter silica spheres (Lancaster, UK) partially coated in silver using a Tollen’s reagent method [16]. Transmission electron microscopy images of the partially silver coated particles showed that the Tollen’s method resulted in discrete silver deposits up to 130nm in diameter. The amount of silver coating on the particles varied, with some spheres carrying virtually no silver while others were compeletely, although unevenly, coated. We have observed that trapping of these particles with a TEM00 beam was possible provided that the particle had less than 25% of it’s surface coated with silver [14] In such cases we postulate that the spheres are sufficiently transparent that they trap due to the same mechanism as a dielectric particle. The azo dye (3, 5-dimethoxy-4-(6’-azobenztriazoyl)-phenylamine) at 10^-4M concentration was strongly adsorbed onto the partially silvered microparticles. This chromophore has an absorption maxima close to the excitation wavelength of 532 nm and the surface plasmon resonance of the partially coated sphere, giving us suitable conditions to observe a SERRS signal.

3. Results and discussion

Individual partially silvered, dye coated particles were trapped using a 1064 nm laser with a trapping power of 436 mW at the trapping plane. The frequency doubling crystal introduced 10.9 µW of co-linear green excitation light and the resulting SERRS spectra was recorded. The time decay of the SERRS intensity for several such particles was measured using the fiber coupled spectrometer with an integration time of 2s. Figure 2 shows the intensity fall-off of the 1352 cm^-1 Raman shifted peak for a selection of trapped particles, with different particles giving SERRS emission for between 1 and 4 minutes. SERRS emission with similar time durations were found for trapping laser powers of 365 mW (producing 9.3 µW of green) and 510 mW (with 15.3 µW of green) respectively.

Possible factor affecting the SERRS response, resulting in a range of photodegradation rates shown in Fig. 2, include how much silver and dye are on the surface of the silica sphere and the orientation of the metal and dye with respect to the small amount of incident excitation light. Studies of molecules interacting with small metal spheres also suggest that the decay rate of the emitting molecule is dependent upon the proximity of the adsorbed molecule to the metal surface, with the lifetime decreasing when the molecule-sphere separation is less than the wavelength of the emitted light [17, 18].

 

Fig. 2. Photo-degradation of several different individually trapped, silver and dye-coated particles using 436 mW of 1064 nm laser light which, after transmission through the frequency doubling crystal, introduced 10.9 µW of 532 nm excitation light in the trapping plane.

Download Full Size | PDF

 

Fig. 3. Comparison of the spectra recorded from the silver and dye coated micro-particle in the optical tweezers to the same dye adsorbed onto a silver colloid and recorded using a dedicated Raman system.

Download Full Size | PDF

To confirm that the emission did correspond to SERRS, the spectrum recorded was compared to SERRS spectra recorded from the same dye adsorbed onto a silver colloid taken in a dedicated Raman system i.e. no optical tweezers. The Raman system comprised of a Renishaw 2000 Raman Microprobe with a CCD spectrometer where the excitation was provided by an argon-ion laser at 514.5 nm and 3 mW. As shown in Fig. 3, both spectra are in good agreement. The two principle features of the SERRS spectra are the 1352 cm^-1 peak, due to the azo (-N=N-) stretch of the dye, and the 1601 cm^-1 peak, resulting from the symmetric C-C stretch in the aromatic rings. As expected, no signal was observed when dye was adsorbed onto plain silica particles and trapped in the optical tweezers, indicating that the SERRS active particles do have a molecular species adsorbed onto a partial metallic coating and is not ablated before trapping.

 

Fig. 4. The spectra of a SERRS active bead being trapped, released and re-trapped. The particle is only active when trapped; when it is not trapped the intensity falls down to noise levels.

Download Full Size | PDF

To demonstrate that the SERRS emission is associated with the particle and not its specific orientation within the trap, we trapped, released and re-trapped a single active particle. The intensity SERRS emission at the 1352 cm^-1 Raman shifted peak from a single particle is shown in Fig. 4. Obviously, when the bead is not in the optical trap, the intensity falls back down to noise levels. Upon re-trapping, the signal returns to a similar intensity value prior to its release and continues to photobleach.

4. Conclusion

We have shown that, by using a frequency doubling KTP crystal to introduce small quantities of co-linear 532 nm excitation light into an infrared optical trap, it is possible to record a SERRS signal from a trapped particle that is sustainable over several minutes. To the best of our knowledge, we believe that this is the first time that a prolonged SERRS signal has been recorded from a particle that is trapped in optical tweezers. By combining the spectroscopic nature of SERRS with our ability to trap and manipulate these beads, we will be able to develop new platforms, such as biological probes, that can readily compete with those being developed based on fluorescence.