Raman spectroscopic technologies have been widely used in material characterization in chemistry, physics, biology, aerosol science, and life science [1–16]. Combining particle trapping with Raman spectroscopy by holding a particle in place long enough for data acquisition enables us to study a single microsized particle, spore, or cell, and then monitor the temporal process of a particle in various environments. The trapping also helps reduce the scattering and stray light from the sample holding substrate [1,3–12]. Among various trapping-associated technologies, laser-trapping Raman spectroscopy (LTRS) has been extensively used in the study of molecular biology with the help of microscopy, such as confocal microscopic Raman spectroscopy [1,3,4]. However, a biological system—even a single cell—is a complex mixture containing numerous biomolecules in various concentrations and forms, including amino acids, proteins, nucleic acids, coenzymes, flavins, lipids, and more. Some cells or spores can grow, change, and reproduce in buffer liquid or air; LTRS enables the study of these cells as they undergo these processes. However, different molecules located in different areas within a particle have different functions and responses to their environment. To detect and monitor different biomolecules as they respond to changes in various environments, and then to gain a new fundamental understanding of these reactions in a particle or a cell, measuring the Raman spectra from different positions within a single cell or a particle could be a very powerful tool. The reported position-resolved Raman spectra are realized either by recording multiple Raman spectra by spatially scanning the sample [12–15], or by measuring multiple monochromatic images by spectrally scanning a filter to allow the scattering signal to be exposed to a 2D detector at different wavelengths [16]. However, these results are obtained either from samples on a substrate or from bulk samples. Although there are numerous studies using the LTRS technique, the majority of them are performed in a liquid solution, with studies conducted in air increasing only recently [1,3–12], and there is no study that records the position-resolved Raman spectra from laser-trapped single airborne aerosol particles, except one work by Zhang et al. [6]. They have shown a 2D Raman image from a droplet levitated by an electrodynamic balance device, but with only vertically binned Raman spectra measured. In this Letter, we demonstrate the measurement of the position-resolved Raman spectra within a laser-trapped single airborne chemical droplet.

Figure 1(a) shows the top view of the experimental schematic, and Fig. 1(b) shows the front view of the droplet generator and the particle trapping part of the setup. A VX nerve agent chemical simulant, diethyl phthalate (DEPh; Sigma Aldrich, 99.5% stated purity), was used as the test material. The diethyl phthalate liquid was held in a 25 mL glass bottle without further purification. It was then used to produce micro-droplets about 20–25 μm in diameter using a drop-on-demand piezoelectric dispensing device (Microfab, MJ-ATP-01). A laser beam from a continuous wave (CW) Ar-ion laser (Coherent, Innova 300C FreD) operating at 488 nm with an output power of 750 mW was used as the light source for both droplet trapping and Raman excitation. The original Gaussian beam was converted into a hollow beam, with a ring-shaped transverse cross section, by passing through two axicon lenses (Del Mar Photonics, cone angle 175°). The horizontally propagating hollow beam was reflected up into a vertical beam by an elliptical mirror set at 45°, propagated through an aspheric lens ($\mathrm{NA}=0.55$) and focused onto a point centered inside a glass cell ($25\mathrm{mm}\times 25\mathrm{mm}\times 15\mathrm{mm}$). Beyond the focal point, the conical beam expanded and was reflected by a concave spherical mirror ($f=19\mathrm{mm}$, $\text{diameter}=25.4\mathrm{mm}$) and refocused at a second focal point aligned with the same beam axis. This trapping technique is modified from our recently developed setup [17] with the additional concave spherical mirror, which can capture a droplet within about 10 ejections [18]. This technique can trap both transparent and absorbing aerosol particles using relatively low numerical aperture ($\mathrm{NA}=0.55$) optics in the air and hold the particle at least 5 mm away from any optical surface; it greatly reduces the possibility of contamination and the influence of scattering and stray light from any optical components, and can be easily integrated into a wide range of spectroscopic characterization techniques. Ideal trapping conditions occur when the two focal points are perfectly aligned with each other and separated vertically by few tens of micrometers. Two holes ($\text{diameter}=8\mathrm{mm}$) at the centers of the concave mirror and the top cover of the glass cell were used for introducing particles into the trapping volume. Scattered light from the single trapped particle was collected with a microscopic objective lens (Mitutoyo $20\times$, $\mathrm{NA}=0.42$) and focused by a second lens to an image-reserved spectrograph (Princeton Instruments IsoPlane SCT-320). A long-pass filter was used to block the elastically scattered 488 nm light. The dispersed imaging Raman spectra were recorded by an electron-multiplying charge-coupled device (EMCCD; Princeton Instruments ProEM) running in its image mode. In general, a CCD runs in spectral mode for recording spectra by binning the pixels vertically.

 

Fig. 1. (a) Top view of the experimental schematic for measuring position-resolved Raman spectra of a laser-trapped single airborne aerosol particle, (b) front view of the droplet generator and the particle trapping part of the setup.

Download Full Size | PDF

Meanwhile, a portion of the collected scattered light at 488 nm was reflected by a dichroic beam splitter (long-pass, cutoff wavelength 488.5 nm) and directed onto a photomultiplier tube (PMT). The PMT recorded the elastic scattering light within a scattering angle of $90°\pm 2°$, and it was compared with the simulated resonance scattering spectrum computed using the Lorenz–Mie theory for particle size determination [18].

In order to obtain the position-resolved Raman spectra from the laser-trapped DEPh droplet, we project its magnified Raman scattering image onto the slit of the spectrograph. When the width of the spectrograph slit is set at 1000 μm, it barely allows the whole image from a 20 μm diameter droplet to pass through the slit and form a circular image on the EMCCD with a 100 pixel diameter, so the image on the pixel is also from a $200\mathrm{nm}\times 200\mathrm{nm}$ area size spot of the droplet. As the diffraction limitation at 488 nm prevented us from resolving a spot with a diameter smaller than $\sim 500\mathrm{nm}$, there was no means to further magnify the image to increase the resolution. Figure 2(a) shows typical imaging Raman spectra from a laser-trapped DEPh droplet, where the vertical pixels of the EMCCD represent the vertical spatial position of the droplet and the horizontal pixels represent the shifted wavenumber of the superposed Raman spectra from each horizontal spatial position of the droplet.

 

Fig. 2. Imaging Raman spectra from (a) the whole droplet and (b) one 200 nm vertically oriented central slice of a laser-trapped diethyl phthalate microdroplet with a 20 μm diameter; (c) the Raman spectra from the five positions marked as A, B, C, D, and E (with an equal $200\mathrm{nm}\times 200\mathrm{nm}$ area size) along the central slice in (b) of the trapped droplet spaced vertically from the center at $-8\mathrm{\mu m}$, $-4\mathrm{\mu m}$, 0 μm, $+4\mathrm{\mu m}$, and $+8\mathrm{\mu m}$, respectively.

Download Full Size | PDF

However, the superposition of the spectra from different horizontal positions of the droplet makes it difficult to retrieve the actual spectra from each horizontal position of the droplet. To avoid such a problem, the slit was set to be as narrow as 10 μm, and then only a vertical slice of the droplet with a total width of 200 nm was imaged onto the EMCCD. In such a condition, the vertical pixels of the EMCCD still represent the vertical spatial positions of the droplet and the horizontal pixels correspond to the shifted wavenumber of the Raman spectra. Therefore, the count number from each pixel of a horizontal line of the EMCCD gives the intensity of the Raman spectra from the corresponding position of a $200\mathrm{nm}\times 200\mathrm{nm}$ area of the droplet. By horizontally scanning the lens (L3 in Fig. 1) to allow each slice of the droplet image to enter the slit step by step, the Raman spectra from each section in the 2D area of the whole droplet could be measured.

As reported [18], the Raman spectra from a DEPh droplet showed additional stimulated Raman scattering (SRS) features and whispering gallery mode (WGM) resonance peaks from bulk liquid that only shows spontaneous scattering peaks. While the spontaneous Raman signal is generally weak, the SRS peaks exhibit stronger and sharper profiles, such as peaks at $\mathrm{\Delta }\upsilon =231$, 363, 387, 540, 568, 588, and $1054{\mathrm{cm}}^{-1}$, compared to the peaks in bulk liquid. Such a phenomenon has been widely reported within microcavity media, such as in a droplet and microsphere, where spontaneous Stokes photons injected into the medium lead to the amplification of the Raman transitions by total internal reflection within the microcavity [18–23].

Figure 2 shows the imaging Raman spectra when the slit of the spectrograph was set at 1000 μm [Fig. 2(a)] and 10 μm [Fig. 2(b)]. Although Fig. 2(a) contains all the Raman information from the different positions of the whole droplet, it is very hard to retrieve all the position-resolved spectra from the different positions of the droplet due to the superposition problem where both the wavenumber of the spectra and the horizontal positions of the droplet are stretched in the horizontal pixels. The spectrum in Fig. 2(b) eliminated the superposition problem of the scattering from different horizontal positions of the droplet, with only the Raman scattering from one 200 nm vertical slice from the droplet. Figure 2(c) gives five Raman spectra from the $200\mathrm{nm}\times 200\mathrm{nm}$ spots along the central slice of the droplet at positions spaced vertically from the center at 0 μm, $\pm 4\mathrm{\mu m}$, and $\pm 8\mathrm{\mu m}$, respectively. All these spectra exhibit a high signal-to-noise ratio with the same peaks that characterize the DEPh’s vibrational modes, but they show strong, sharp SRS peaks (marked by short arrows in curve A) varying in intensity at different positions, especially at the top and bottom positions near the droplet surface (curves A and E).

The microcavity-enhanced scattering phenomena have been observed in Raman scattering, elastic scattering, absorption, and fluorescence. When the wavelength of traveling light in a microcavity matches a WGM resonance, or “morphology dependent resonance (MDR),” these waves are generally propagating near the concave interface and could strongly enhance the signal by the input resonance of the illuminating wavelength, or by the output resonance of the scattering wavelength [5,7,8,18–23]. Double resonances contribute to very strong, sharp Raman peaks, and have more opportunities to produce strong SRS and Raman laser emissions. Thus, the WGM-related waves escape from the microcavity in the evanescent wave format tangentially, and the relative intensities of these cavity-enhanced signals depend on the origin of their position in a microcavity and the angle of observation.

In our experiment, the droplet is vertically trapped and illuminated by the focused 488 nm hollow beam; the cavity-enhanced effects will be stronger along the vertically or near-vertically oriented equatorial planes than in the horizontally oriented planes. Therefore, the collecting microscopic objective that was set horizontally facing the droplet should observe more cavity-enhanced signal from the top and bottom parts than from the middle part of the droplet. The Raman spectra from the middle parts of the droplet [B, C, and D curves in Fig. 2(c)] mainly contain the normal spontaneous Raman scattering peaks with some SRS signal at the same level of intensity, while the Raman spectra near the surface of the droplet [A and E curves in Fig. 2(c)] display strong SRS peaks, which are $\sim 5\text{times}$ (E curve at top) and $\sim 30\text{times}$ (A curve at bottom) stronger than the Raman peaks from the middle positions (curves B, C, and D), although they have similar intensity to the spontaneous Raman scattering peaks (e.g., the peak at $1735{\mathrm{cm}}^{-1}$) as shown in all curves. Chan’s group observed a similar intensity distribution of Raman scattering within a levitated droplet and explained that the two areas of the droplet with maximum intensities are the results of the enhanced light scattering at the top and the bottom of the droplet [6]. In our case, the SRS peaks at the bottom area are about 5−10 times stronger than those at the top area; we believe that these molecules at the bottom also received higher-intensity laser illumination by the focused laser beam than those at the top, except for the cavity-enhanced effect.

In case this intensity difference between the top and bottom of a droplet is caused by other reasons, e.g., a nonuniform molecule distribution within the droplet (more dense at the bottom), possibly redistributed by the gravitational and trapping optical force, position-resolved Raman spectra of a trapped DEPh droplet were recorded using a horizontally illuminated 527 nm laser beam (200 mW, 1 mm in diameter).

Figure 3(a) shows the Raman spectra from four different spots (A, B, C, and D) of the droplet with horizontal illumination at 527 nm. All four spots are located on a circle with a radius of 8 μm and have an equal-sized area of $200\mathrm{nm}\times 200\mathrm{nm}$. In Fig. 3(a), portions of the spectra between 1300 and $1600{\mathrm{cm}}^{-1}$ ($\sim 566–576\mathrm{nm}$) are not shown for a better view quality. At this region, the 527 nm excitation Raman peaks were dominated by the strong Raman scattering corresponding to the C-H stretching mode ($\sim 2800–3100{\mathrm{cm}}^{-1}$ in Raman shift) excited by the 488 nm laser. In such an optical configuration, all the Raman spectra from the four positions near the droplet surface exhibit the same spectral peaks, but with different intensities; position A, which the 527 nm laser is illuminating, has the strongest peaks followed by the opposite side of the droplet at position B, while the Raman scattering from the top and bottom positions has relatively the same intensity but is weaker than that from positions A and B. This further supported our explanation of the result shown in Fig. 2. However, we were not able see the SRS by 527 nm illumination, although we tried to increase the illumination intensity to 300 mW before the droplet was pushed away from the trap. In comparison, Fig. 3(b) represents the position-resolved Raman spectra by 488 nm illumination from the same positions A, B, C, and D shown in Fig. 3(a). Position C, which the 488 nm laser is illuminating, also has the strongest signal, followed by the opposite side of the droplet at position D, while the Raman scattering from the sides near the surface at positions A and B has SRS peaks associated with the spontaneous scattering peaks, with a weaker intensity than that from positions C and D.

 

Fig. 3. Raman spectra from the four positions marked as A, B, C, and D (with an equal $200\mathrm{nm}\times 200\mathrm{nm}$ area size) along the horizontal and vertical central slice in a laser-trapped diethyl phthalate microdroplet with a 20 μm diameter. All four spots are located on a circle with a radius of 8 μm; the droplets were illuminated (a) horizontally by a collimated 527 nm laser beam and (b) vertically by a 488 nm counterpropagating hollow beam.

Download Full Size | PDF

In summary, a technique that can measure position-resolved Raman spectra from equal $200\mathrm{nm}\times 200\mathrm{nm}$ area spots within a laser-trapped single airborne chemical microdroplet was demonstrated. This droplet was trapped using two focused counterpropagating hollow beams. The very stable trapping enables us to measure the slice-to-slice position-resolved Raman spectra from the whole droplet. This spatially resolved single-particle Raman technique could also be used to monitor various chemical reactions in different parts of a trapped particle in various environments such as in life sciences and atmospheric sciences.