1. Introduction

Laser tweezers Raman spectroscopy (LTRS) [1–4], a technique combining optical tweezers with confocal Raman spectroscopy, has proven to be a powerful technique for the label-free chemical analysis of single particles [5,6], and in particular of single living cells suspended in an aqueous medium for biomedical and biological applications [7,8]. An individual cell or micro-particle is continuously trapped within the laser focus, and at the same time, its Raman spectrum is acquired. A simple LTRS system typically uses only one laser beam for both optical trapping and Raman excitation, and uses one objective to focus the laser beam to generate far-field optical trapping forces and to collect Raman scattering light [7–10]. LTRS is an emerging, new analytical method for particle identification and characterization, studying sample heterogeneity, and monitoring cellular dynamics at the single cell level [5–10].

Objective based LTRS systems are normally built on a large microscope platform. In addition to its bulky footprint, a significant disadvantage of conventional LTRS systems is the difficulty in trapping sub-micron particles because of the diffraction limit [11]. In particular, the analysis of metallic nanoparticles is challenging because of the large scattering forces that are generated. The recent development of near-field optical technologies allows for the energy of the laser field to be highly concentrated in a localized volume that is beyond the diffraction limit. Based on this advantage, near-field optical tweezers are able to provide strong localized optical gradient forces for the stable manipulation of individual nanoparticles or biomolecules with diameters smaller than 100 nm [12]. Many near-field optical trapping techniques have been developed, including plasmon-assisted nanotweezers [13–16], waveguide-based optical transport [17,18], and photonic crystal resonators [19,20]. At the same time, several nanophotonic techniques have also been exploited for Raman spectroscopy, such as tip-enhanced Raman spectroscopy [21], evanescent wave excitation of surface enhanced Raman scattering (SERS) by using an optical fiber taper on a SERS substrate [22] or using a Ag NP-coated sapphire optical fiber [23], and evanescent wave excitation of spontaneous Raman scattering by using a silicon nitride nanophotonic waveguide [24]. With the development of nanophotonics for both near-field optical tweezers and near-field Raman spectroscopy, a logical next step in this field is to realize a novel near-field nanotweezers Raman spectroscopy approach for label-free chemical analysis of single trapped nanoparticles that offers unique capabilities beyond what conventional LTRS can currently provide. Such a system for Raman analysis of single nanoparticles using the evanescent wave for both trapping and spectral interrogation has not previously been demonstrated. Løvhaugen et al. recently reported the use of Raman spectroscopy to interrogate micron-sized particles trapped by an optical waveguide [25], in which a high power laser was coupled into an optical waveguide to generate an evanescent field to trap particles and a second laser was focused through an objective for far-field Raman excitation. This work demonstrated a simple combination of near-field optical trapping with conventional far-field Raman spectroscopy. Kumar’s group [26] reported plasmofluidic SERS from assembled plasmonic NPs by using a single evanescent wave at a metal–fluid interface to assemble plasmonic NPs and also excite Raman scattering signals from the aggregation of a large number of assembled NPs.

In this work, we show a direct integration of near field light scattering (NLS) and trapping with evanescent wave excited (or waveguide excited) SERS of individual NPs using a single laser coupled waveguide, a technique we refer to as NLS-Raman. An evanescent wave generated from the waveguide was used for both trapping of metallic NPs and excitation of SERS spectra. Poly (allylamine) coated silver NPs (a type of shell-isolated NPs) with 4-aminothiophenol (4-ATP) adsorbed on the Ag surface were used in this study to demonstrate the importance of a shell coating around the bare metallic NP for protecting the waveguide from particle-induced laser damage (i.e. ‘burning’). We report evanescent wave excited anti-Stokes and Stokes SERS spectra of 4-ATP from evanescent wave trapped single Ag NPs. As a comparison, far-field SERS spectra from waveguide trapped single Ag NPs were also collected. Important advantages of NLS-Raman have been demonstrated that are expected to significantly expand the use of LTRS for applications in Raman-based chemical analysis at the nanoscale.

2. Experimental

2.1 Near-field light scattering Raman (NLS-Raman) system

Figure 1(a) shows a schematic illustration of the integration of an optical waveguide system with a confocal Raman spectroscopy setup. A laser was coupled into a silicon nitride waveguide (Optofluidics) to generate a near-field evanescent wave capable of trapping metallic NPs in solution on the waveguide surface. Intense scattered light from the evanescent wave and evanescent wave excited SERS signals from trapped metallic NPs were collected by an objective (Leica, 50 × , NA = 0.55). A second laser can be coupled into the objective for far-field Raman spectroscopy of single NPs trapped on the waveguide. A detailed diagram of the experimental setup is shown in Fig. 1(b). Two laser wavelengths (785 nm, 1064 nm) were tested in this study. A 785 nm laser (Crystalaser, RCL-080-785-S) was coupled into the microscope objective for acquiring far-field Raman spectra. A 1064 nm laser (NanoTweezerTM instrument, Optofluidics, maximum 350 mW) or 785 nm laser (Sacher Lasertechnik, maximum 1 W) was used to generate evanescent waves by coupling the laser beams into the waveguide with an optical fiber. Before the fiber coupling, a laser line filter at 1064 nm (Chroma, ZET1064) or 785 nm (Semrock, LD01-785) was used to clean the spectral profile of the evanescent Raman excitation light. A short-pass filter (Edmund Optics, cut-off wavelength at 1000 nm) or a long-pass filter (Semrock, LP02-785) was placed before the entrance slit of the spectrometer (Princeton Instruments, LS785) for measuring evanescent wave excited anti-Stokes SERS spectra (λ_exc = 1064 nm) or evanescent wave excited or far-field Stokes SERS spectra (λ_exc = 785 nm), respectively. A LED white light source was filtered by a band pass filter (Edmund Optics, center wavelength at 507 nm) for bright-field illumination of the waveguide. A hot mirror (Thorlabs) was used to separate the bright-field imaging light from the Raman scattering light or the far-field Raman excitation laser light. In addition, the hot mirror allows for the transmission of a small amount (< 3%) of intense scattered evanescent wave light at 1064 nm or 785 nm from waveguide trapped NPs, which are detected by an imaging camera (IDS imaging, μEye), to locate the trapped NPs. This optical design enables us to simultaneously image the waveguide with bright-field, observe NPs being trapped by their scattered NIR light, and measure SERS signals without any interference. The use of 785 nm and 1064 nm for Stokes and anti-Stokes signal acquisitions, respectively, allowed us to conveniently use a single spectrometer and detector for all measurements.

 

Fig. 1 (a) Illustration of the nanotweezer SERS system. (b) Schematic diagram of the experimental setup. F: Filter; BS: Beam splitter; FC: Fiber coupler.

Download Full Size | PDF

2.2 Waveguide and microfluidic chip

The photonic chips (Fig. 2) used in these experiments were supplied by Optofluidics, Inc. Each chip is a 1 cm² silicon substrate with silicon nitride (Si₃N₄) waveguides patterned by standard microfabrication techniques. Waveguides have rectangular cross-sections with a height of 250 nm and a width of 600 nm. The waveguide is cladded by SiO₂ (3 μm above and 5 μm below) across the entire chip with the exception of a 200 μm × 200 μm experimental window in the center region of the chip (Fig. 2(c)). In the experimental window, the top cladding layer is removed, allowing for interaction of the sample with the exposed waveguide. Each chip contains three waveguides for experimental redundancy. The waveguides are separated by 1 mm at the input edge of the chip and converge after the experiment window to a corner of the chip on the output edge where the transmitted light is collected by a photodector (Fig. 2(b)).

 

Fig. 2 (a) Chip assembly consisting of waveguide chip, cassette, and three bonded optical fibers. (b) Schematic illustration of waveguide chip. (c) Microscope view of chip at experimental window.

Download Full Size | PDF

Three polarization maintaining fibers (L = 10 cm) supported by a silicon v-groove array are optically aligned to the three waveguides and permanently bonded to the chip. The chip assembly is loaded into an aluminum cassette (Fig. 2(a)), which provides structural support and durability while handling. Laser light is coupled into one waveguide at a time during experimental use by mating one of the three FC/APC terminated input fibers to an FC/APC terminated laser delivery fiber.

Each chip has two through-holes for fluidic access as shown in Fig. 2(b). A microfluidic channel is formed on the chip by a 25 μm thick silicone gasket capped with an 8 mm glass coverslip, allowing for viewing of the experimental window with a high magnification objective. For macro-microfluidic coupling, the chip cassette is seated on a custom-built holder with two fluidic ports. Fluid lines are coupled to the chip via o-rings that seal around the through-holes on the backside of the chip. Flow control is achieved via a pressure-regulated vacuum pump. The microfluidic chips can be reused by flushing repeatedly with washing buffer of 1% pluronic F-68 surfactant in 0.22 μm filtered deionized water.

2.3 Chemicals for nanoparticle synthesis

AgNO₃, 4-aminothiophenol (4-ATP), sodium citrate, and sodium hydroxide were purchased from Sigma-Aldrich (St. Louis, MO), and poly (allylamine) (Mw = 15,000) was purchased from Polysciences (Warrington, PA). Reagents and solvents were obtained commercially and used without further purification. All glassware were cleaned with laboratory grade detergent and rinsed with DI water thoroughly prior to use.

2.4 Synthesis of citrate stabilized silver nanoparticles

Citrate stabilized silver nanoparticles were synthesized according to the established Turkevich method [27]. In short, 50 mL of 0.4 mM AgNO₃ solution was placed in a 125 mL Erlenmeyer flask and boiled on a hot plate. Then, 450 µL of 38.8 mM sodium citrate solution was added into the boiling solution. The mixture then changed to a turbid yellow color in a few minutes and remained boiling for 10 more minutes to stabilize the synthesized nanoparticles. Then the solution was gradually cooled down to room temperature. The solution was purified by centrifugation with a bench-top centrifuge at 8000 rpm for 10 min and resuspended with DI water until further usage. This citrate stabilized silver nanoparticle solution is denoted as “1”.

2.5 Synthesis of 4-ATP conjugated silver nanoparticles

1 mM solution of 4-ATP was prepared in ethanol and then further diluted to 100 µM in DI water. 50 µL of this solution was added drop-wise into 5 mL of solution 1 with mild stirring on a magnetic stirring plate. The mixture was stirred for 30 min and purified with a bench-top centrifuge at 8000 rpm for 10 min and resuspended with DI water. This 4-ATP conjugated silver nanoparticle solution is denoted as “2”.

2.6 Synthesis of poly (allylamine) covered silver nanoparticles

The polyelectrolyte layer covering the silver nanoparticles was synthesized using a procedure that was slightly modified from a previous report [28]. In short, a 4 g/L solution of the Poly (allylamine) solution was prepared in 2 mL of DI water. Then 200 µL of solution “2” was added drop-wise into the stirring polymer solution and continuously stirred for 1 hour before purification by centrifugation at 8000 rpm for 10 min. The purified nanoparticles were resuspended in DI water and stored until further usage. This poly (allylamine) covered 4-ATP silver nanoparticle is denoted as “3”.

3. Results

3.1 Waveguide trapping of SERS Ag NPs

Nanostructured waveguides can generate strong localized evanescent waves with strong gradient forces for NP trapping [12,17,29]. Our initial experiments focused on trapping bare Ag NPs. A water solution containing the bare NPs (~10⁸ particles/ml) was introduced into the waveguide chip through the microfluidic channel. Particles traversing over the waveguide surface were attracted to the evanescent wave emanating from the waveguide and became trapped. However, trapping of bare NPs resulted in the generation of bubbles around the trapped NP and even burning of the waveguide in some instances (Fig. 3(a)), presumably due to the high temperatures around the hot spot between the trapped metallic NPs and the waveguide surface. Applying lower laser powers could help reduce this effect somewhat, but the tradeoff is a lower trapping efficiency of the single NPs. It has been shown previously that shell-isolated metallic NPs can prevent NPs from having direct contact with surrounding material [30]. Using this shell-isolated strategy for the purpose of providing a buffer region between the waveguide and the metallic NP to minimize the bubbling and burning effects, we synthesized poly (allylamine) coated Ag NPs (~60-80 nm) with the Raman reporter molecule 4-aminothiophenol (4-ATP) adsorbed on the Ag surface (Fig. 3(b)). The ultrathin coated poly film (1-2 nm thickness) prevents direct contact of the trapped Ag NPs with the waveguide. With this additional poly coating on the metal NPs, we observed a significant decrease in the bubbling or burning effects that we saw when trapping bare particles, which allowed us to observe the trapping behavior of the poly-coated metallic NPs on the waveguide. For this waveguide, most of the trapped NPs were transported by scattering forces along the waveguide in the laser propagation direction, while some NPs were stably trapped and immobilized at some points on the waveguide. Intrinsic waveguide defects, which create higher local intensity and thus stronger optical gradient forces, or variations in thickness of poly film coatings of NPs, can result in immobility of trapped NPs. The trapped NPs strongly scatter the evanescent light and show up as bright spots along the waveguide (Fig. 3(b)) within the effective trapping area (square area, 200 × 200 μm²), which makes it easier to visualize and locate the trapped Ag NPs on the waveguide surface.

 

Fig. 3 Images of waveguides during the trapping of (a) single bare Ag NPs and (b) single poly (allylamine) coated (beige shading in inset figure) Ag NPs with adsorbed 4-aminothiophenol (4-ATP) (green icons in inset figure) molecules using a 1064 nm laser. Trapping of bare Ag NPs leads to bubble formation and damage to the waveguide. Trapping of poly-coated Ag NPs are visualized by the light scattering on the waveguide surface.

Download Full Size | PDF

3.2 Evanescent wave excited SERS from trapped single Ag NPs

Poly-coated Ag NPs with adsorbed 4-ATP molecules can emit strong SERS signals. Figure 4(a) (curve I) shows a far-field SERS spectrum of an aggregate of poly-coated Ag NPs adsorbed with 4-ATP that was deposited on a quartz coverslip. No Raman peaks were observed for poly-coated Ag NPs without 4-ATP (curve II, Fig. 4(a)), which exhibited the same background base line as bare Ag NPs (curve III, Fig. 4(a)). These spectra indicate that the polymer coating itself does not contribute a Raman background of its own. The immobilization of some trapped single Ag NPs on the waveguide allowed for the collection of their SERS spectra using a microscope objective. Figure 4b (curve I) shows the evanescent wave excited SERS spectrum of a single trapped poly-coated Ag NP with 4-ATP that was excited with the 785 nm evanescent wave. This is the first realization of a waveguide-based approach for near-field trapping and simultaneous SERS detection of single NPs using a single laser source. The evanescent wave excited spectrum of waveguide trapped poly-coated Ag NPs without 4-ATP is shown in curve II of Fig. 4(b). The background is attributed primarily to the excitation of Raman signals from the glass optical fiber that was connected to the waveguide. When the flow rate of the Ag NP solution inside the microfluidic channel was adjusted to zero, some of the trapped NPs would remain immobilized on the waveguide surface even when the evanescent wave was removed. Far-field SERS spectra of these single Ag NPs were acquired using the 785 nm laser coupled into the microscope objective. Figure 4(c) shows both evanescent wave excited (curve I) and far-field (curve II) SERS spectra of single trapped Ag NPs excited by the evanescent wave and by the objective focused laser light, respectively. For the waveguide excited SERS spectrum in curve I, 24 mW laser power was measured at the output of the laser delivery fiber, prior to coupling to the chip assembly fiber. The optical power measured by a calibrated silicon photodetector positioned at the chip output edge was 0.29 mW. The experimental window is separated from the outputs by 11 mm of total waveguide length. Using an experimentally measured attenuation coefficient of 0.3 dB/mm for 785 nm light in the silicon nitride waveguide, the estimated optical power in the waveguide at the experimental window is calculated to be 0.63 mW, neglecting additional bending losses due to the curvature of the waveguide between the window and output. For the far-field SERS spectrum in curve II, the measured laser power after the 50 × objective was 1.3 mW. The integration time for both SERS measurements was 1 s. The Raman spectra in Fig. 4(c) were original data, as measured, without any background subtraction or intensity normalization. A comparison of the spectra shows that both the waveguide excited and far-field spectra exhibit a similar SERS spectral profile of the 4-ATP molecules. However, the waveguide excited SERS spectrum has a higher background than that of the far-field spectrum, which is due to the spectral background from the optical fiber. Meanwhile, the far-field SERS spectrum contains a sharp Raman peak at 525 cm⁻¹ and a broad Raman peak between 925-1010 cm⁻¹, which are associated with the spontaneous Raman spectrum of the SiO₂ substrate of the waveguide chip. This was confirmed by using the far-field laser beam to acquire a spectrum of a region of the pure SiO₂ substrate away from the waveguide, which showed the same two Raman peaks with almost the same intensities. Although the evanescent wave excited SERS spectrum shows stronger Raman peak intensities than the far-field SERS spectrum, we observed that for some other NPs that were analyzed under the same experimental conditions, those NPs showed stronger far-field SERS signals than waveguide excited SERS signals. This interesting observation suggests that the localized evanescent wave and the objective focused far-field laser light may excite different SERS hot-spots for the same single NP that is being analyzed.

 

Fig. 4 (a) Far-field 785 nm laser excited SERS spectra of 4-ATP (curve I) from an aggregation of poly-coated Ag NPs deposited on a quartz coverslip, (curve II) from an aggregation of poly-coated Ag NPs without 4-ATP, and (curve III) from an aggregation of bare Ag NPs. (b) 785 nm evanescent wave excited SERS spectra of 4-ATP (curve I) from waveguide trapped single poly-coated Ag NPs and (curve II) from waveguide trapped poly-coated Ag NPs without 4-ATP. (c) Evanescent wave excited (curve I) and far-field (curve II) SERS spectra of 4-ATP for the same single poly-coated Ag NP on the waveguide surface.

Download Full Size | PDF

3.3 Evanescent wave excited anti-Stokes SERS from trapped single Ag NPs

Another important result of this work is the observation of strong waveguide excited anti-Stokes SERS of trapped single Ag NPs. The evanescent wave generates anti-Stokes SERS of 4-ATP from trapped poly-coated Ag NPs when the 1064 nm laser was coupled into the waveguide, as shown in Fig. 5(a) (curve I). A flat background baseline (curve II, Fig. 5(a)) was observed for the measured anti-Stokes SERS spectrum of poly-coated Ag NPs without 4-ATP. During the trapping of the single poly-coated Ag NPs with 4-ATP by the 1064 nm evanescent wave and the collection of its anti-Stokes SERS spectrum, the far-field Stokes SERS spectrum was also collected simultaneously by using the 785 nm excitation from the objective. Figures 5(b) and 5(c) show the far-field Stokes SERS and waveguide excited anti-Stokes SERS for the same waveguide trapped single Ag NPs, respectively. The power of the far-field 785 nm laser was 1.3 mW measured after the objective. The power of the 1064 nm was 30 mW before being coupled to the chip assembly fiber. The measured optical power at the chip output was 1.3 mW. Using an experimentally measured attenuation coefficient of 0.17 dB/mm for 1064 nm light in the silicon nitride waveguide, the calculated optical power in the waveguide at the experimental window is estimated to be 2.0 mW, neglecting additional bending losses as before. 1 s integration time was used to measure the far-field Stokes and evanescent wave excited anti-Stokes SERS signals. For the same Ag NP, the far-field SERS spectrum excited by the 785 nm laser was roughly 5-20 times stronger than the waveguide excited anti-Stokes SERS spectrum excited by the 1064 nm evanescent wave.

 

Fig. 5 (a) Waveguide excited anti-Stokes SERS spectra of 4-ATP (curve I) excited by 1064 nm evanescent wave from waveguide trapped poly-coated Ag NPs and (curve II) from waveguide trapped poly-coated Ag NPs without 4-ATP. (b) Far-field SERS spectra excited by 785 nm laser and (c) waveguide excited anti-Stokes SERS spectra excited by 1064 nm evanescent wave of 4-ATP for the same single poly-coated Ag NPs trapped by waveguide with 1064 nm evanescent wave.

Download Full Size | PDF

4. Discussion

We have shown that a silicon nitride (Si₃N₄) waveguide coupled with a single laser source can be used to realize near-field optical trapping of single poly-coated Ag NPs and SERS spectroscopy of 4-ATP adsorbed on the NP surface. There are several important advantages of NLS-Raman as presented in this study. First, the use of an evanescent wave for both optical trapping and SERS excitation on a waveguide microchip demonstrates a new LTRS approach that utilizes advanced nanophotonic technology, which has important implications for eventually developing more miniaturized, compact LTRS systems for near-field single particle chemical analysis compared to the large microscope-based platforms that are currently used. Second, the near-field optical trapping allows us to trap and analyze single, metallic NPs with diameters < 100 nm, which for the first time extends LTRS down to the nanoscale for single nanoparticle analysis applications. Demonstration of the acquisition of evanescent wave excited Stokes and anti-Stokes SERS spectra of single trapped NPs with this NLS-Raman system suggests that NLS-Raman could potentially be a powerful, new analytical technique for efficiently characterizing the SERS properties of metallic nanoparticles at the single particle level in solution. The addition of a shell coating on the NP surface was an important feature of the NPs that we used in our experiments because it helped reduce damage to the NP and the waveguide surface upon evanescent field excitation that we observed in the case of bare NPs. Although we chose a polymer coating, other coatings around the metal NP, such as a glass shell, were shown to work as well (data not shown). Another advantage of performing SERS spectroscopy with near-field illumination is that it produces spectra that avoids background interference signals, such as from the waveguide substrate, because of the highly localized evanescent wave on the waveguide surface that primarily excites the SERS signal of the NP. In contrast, the far field excitation beam used for far field SERS spectroscopy has a much longer Rayleigh range that also probes a significant fraction of the surface of the waveguide chip. A qualitative comparison shows that the intensity of the waveguide excited SERS spectra is comparable to that of the far field spectra (on the same NP), even when using low laser powers coupled into the waveguide (e.g. the 785 nm laser power we used was around 10-30 mW before coupled to the chip assembly fiber). While the waveguide excited Stokes SERS spectra at 785 nm excitation did exhibit some background signal interference from the optical fiber, the background of the waveguide excited anti-Stokes spectra at 1064 nm excitation was almost flat, thus showing the potential advantage of anti-Stokes SERS spectroscopy for avoiding signals from the glass fiber, waveguide, and other optical components in the system. A low power of 10-30 mW of 1064 nm laser coupled into the waveguide was sufficient to generate trapping forces and anti-Stokes SERS signals. Anti-Stokes SERS spectroscopy with a 1064 nm laser may be a viable spectroscopy approach for future NLS-Raman applications.

For the waveguide geometry used in this work, the immobilization of some trapped NPs at fixed points on the waveguide allowed us to measure their SERS spectra. This can be improved upon in the future by exploring the use of waveguides with different designs [31,32], including waveguide resonators [19,20], that can realize stable trapping of NPs at fixed locations on the waveguide for SERS analysis. While a bulky, microscope objective was used in this study to collect the evanescent wave excited SERS signals, it is also possible, in principle, to measure these SERS signals from one of the ends of the waveguide since the SERS light will also propagate along the waveguide, as demonstrated in other recently published studies [22–24]. This configuration would eliminate the need for a large, bulky detection optic, which would help improve the compactness of the system for developing LTRS on a lab-on-a-chip platform in the future.

5. Conclusions

In summary, we have demonstrated an NLS-Raman system that can realize near-field optical trapping and SERS of single Ag NPs by using only one laser coupled waveguide. Although the system is designed to trap single nanoparticles, it is possible that in some of our measurements, SERS spectra from a few nanoparticles that cluster together could have been acquired. A polymer coating on the surface of the Ag NP helps to protect the particle and waveguide from damage to allow for the acquisition of high-quality SERS signals of 4-ATP. In the future, we have plans to explore the addition of coatings on the waveguide instead, which would allow for the stable trapping of bare metallic nanoparticles as well. We report the observation of waveguide excited Stokes SERS spectra excited by a 785 nm evanescent wave and anti-Stokes SERS spectra excited by a 1064 nm evanescent wave from single Ag NPs trapped by the same evanescent wave. Furthermore, far-field SERS spectra from the same trapped single Ag NPs have been measured for comparison purposes. This technique is a significant advancement in LTRS by exploiting the advantages of near-field nanophotonics to enable Raman analysis at the nanoscale. Given the recent and growing interest in the development of SERS labels (i.e. metallic nanoparticles with Raman reporter molecules, often encapsulated in an outer shell) as an alternative to fluorescent dyes for bioassay and cell labeling applications, we envision that this NLS-Raman system will be a useful new tool for single SERS particle characterization. We also envision many other applications of NLS-Raman for the chemical analysis and sensing of nanoscale particles, cells, and biomolecules.