We report on the generation of broadband longitudinal fields within a tightly focused spot by using a segmented wave plate combined with a phase tailored broadband laser pulse. Their field distribution is characterized by observing the scattered light from a gold-coated glass fiber tip as it is scanned across the focused beam spot. It is observed that efficient coupling to the tip-enhanced field can be achieved over a broad bandwidth of more than 100nm, resulting in a positive contrast at the centre of the focus in the spectrally resolved Rayleigh scattering image. Temporal characteristics of the nonlinear excitation at the tip apex observed by using the fringe resolved autocorrelation technique indicate the possibilities of ultrafast spectroscopy by utilizing the tip-enhanced longitudinal fields.
© 2011 OSA
Near-field scanning optical microscopy (NSOM) allows us to obtain fine images of surfaces beyond the diffraction limit of light . In particular, apertureless NSOM [2–5] takes an advantage of a tip-enhancement effect primarily due to the localized surface plasmons at the apex of a sharp metallic tip , which allows for amplification of weak spectroscopic signals. This not only leads to high spatial resolution in imaging but also provides high sensitivity in spectroscopy. These features have been successfully applied to the tip-enhanced (TE) Raman spectroscopy (TERS) and imaging , with which even single-molecule sensitivity has been reported [8–11] with the spatial resolution of less than 30nm. Thus, apertureless NSOM has proved to be useful for obtaining one of the greatest spatial resolutions beyond the diffraction limit of light. Then, a natural question arises as to how short excitation is possible with such extremely localized TE fields since one of the virtues of optical excitations is its high temporal resolution that is not attainable by any other techniques. Short pulses were previously employed in the context of TE coherent anti-Stokes Raman scattering (CARS) [12,13] and four-wave mixing (FWM)  spectroscopy. TE-CARS allows for a direct access to the vibrational information of the sample while TE-FWM generated at the tip-apex serves as a tuneable localized nano-scale light source. Although the latter was initially considered as a source of background in the TE-CARS experiment , it turns out that linear spectroscopy can be conveniently performed by using an appropriately structured tip . However, the temporal resolution was not fully utilized in these experiments. Anderson et al. have recently demonstrated that it is possible to retrieve the plasmonic response of a free-standing metallic tip  by observing second harmonic generation (SHG) under the sub-10fs pulsed excitation using the side illumination scheme. However, the plasmonic response of the tip can be affected by the presence of the substrate at the proximity of the tip apex in realistic experiments. In addition, the bottom illumination scheme [7,12–14] is generally a favored option as it is perfectly compatible with the conventional inverted optical microscopes using a high numerical aperture (NA) objective lens. In this case, it has been shown that efficient coupling to the TE-fields can be accomplished by using a radially polarized light [16–18] although its compatibility to the broadband radiation has not been demonstrated yet.
In this paper, we first show that it is indeed possible to generate broadband longitudinal fields within a tightly focused spot by employing a segmented wave retarder. This is verified by observing spectral images of Rayleigh scattering obtained by scanning a gold-coated glass fiber tip around the spot. It is shown that, thanks to the efficient coupling to the TE longitudinal fields, quantitative characterization of the focused spot can be performed. Furthermore, by comparing the spectral content of the Rayleigh scattering images at different lateral positions, it is shown that the spectral responses of the spatially localized plasmons at the tip apex can be extracted. Secondly, we characterize the fringe resolved autocorrelation (FRAC) in the near-field by observing blue-shifted nonlinear emission from the tip. This provides a simple means of characterizing the temporal characteristics of the near-field excitations. The result also indicates interesting opportunities for studying the dynamics by using nonlinear spectroscopic techniques in the near-field.
2. Experimental setup
Our experimental setup is shown in Fig. 1 . We use a sub-10 fs Ti:sapphire laser oscillator (Femtolasers Fusion, λ0~790 nm, repetition rate of 80 MHz). Pre-compensation optics consist of a double prism pair (DDL: dispersive delay line) and a pair of chirped mirrors so that a large amount of group velocity dispersion (GVD) within the system, including the high NA objective (>4000 fs2), can be compensated for up to 3rd order. After expanding the beam at the first beam expander (BE1), both spectral amplitude and phase of the pulse are controlled by using a pulse shaper based on a spatial light modulator (Jenoptik, SLM-S640d), where the spectral resolution is characterized to be about 5 cm−1 per pixel. The pulse is then split into two replica pulses with an adjustable temporal delay in a dispersion-balanced, delay-stabilized autocorrelator. The collinearly combined beams subsequently pass through the second beam expander (BE2), a polarizer (ColorPol VISIR CW02), a long wavelength pass filter (LWPF: Omega, 3RD740LP), and a segmented retarder (Zpol: Nanophoton) prior to launching into the high NA objective (Nikon, Apo TIRF 100x NA = 1.49) through either a dichroic mirror (DM: Semrock, FF-735-Di01-25x36) or a non-polarizing beam splitter (NPBS: Melles Griot, 03BSC027) cube within an inverted optical microscope (Nikon TE2000). The DM is used whenever the dispersion compensation is important (i.e. Sections 2 and 5) while NPBS is used for observing the Rayleigh back scattering in Sections 3 and 4. When observing the blue-shifted nonlinear emission (NE) signal from the tip, an additional short wavelength pass filter (SWPF: Omega, 3RD730SP) is used in order to eliminate the Rayleigh background.
Figure 2(a) displays the wavelength dependent retardation characteristics through one of the eight segments of the Z-pol that we used. We applied a narrow band source (Δλ~5 nm), shaped by using the amplitude modulation capability of our pulse shaper, to the Z-pol in the crossed Nicol polarizer arrangement. We found that the variation of the retardation angles (determined by the polarizer angles at which the transmission minimum is obtained) over the wavelength range of interest is almost negligible. However, the extinction quickly degrades below the wavelength of 780 nm while it is retained to be an order of 0.01 for longer wavelengths. This suggests that the depolarization effect may limit the performance of our Z-pol at short wavelengths rather than the dispersion of the retardation effects. The GVD due to the Z-pol was found to be negligibly small as it is made of a sub-wavelength structure fabricated on a thin glass substrate.
Our NSOM setup uses a commercially available gold-coated glass fiber tip (for atomic force microscopy: AFM) with a typical apex radius of less than 50 nm. We attach it on a quartz tuning fork sensor  with epoxy glue. The resultant quality factor typically amounts to 3000. The tip-surface distance control is accomplished in the shear-force mode. The frequency shift of the oscillatory motion of the tuning fork is used as a feedback signal, and which is detected using a phase lock loop (PLL) circuit . We observe a typical frequency shift of 3 Hz over a distance of 20 nm from a glass surface. The frequency noise density amounts to less than 5 mHz/√Hz within a response bandwidth of 500 Hz. Therefore, the instantaneous measurement accuracy of the frequency shift was less than 0.2 Hz. Stable operation as an AFM is achieved at the tip-surface distance of approximately 5 nm (set point of 0.6 Hz). Prior to optical characterization, we scanned carbon nanotubes deposited on a glass substrate as a test sample, and we confirmed that the spatial resolution is less than 50nm.
Since we used a variety of transmission optical components in our system, we corrected the high order phase distortion at the focus by using the MIIPs technique . We placed a 20 μm thick BBO crystal on a glass cover slip, to which we focused the beam through the high NA objective. We detected the SH spectra in the forward direction through an additional short wavelength pass filter (FF01-680/SP-25). After a few iterations, we were able to compensate for the residual dispersion, resulting in the phase deviation of less than 0.2 radians over the entire bandwidth as shown in Fig. 2(b). Note that the total phase imposed by dispersive components such as filters often contains irregular phase profiles that can only be compensated for by using the pulse shaper . The quality of the temporal compression at the focus was also examined by observing SHG-FRAC, as shown in Fig. 2(c), which agrees well with the calculated FRAC trace by assuming the spectral intensity profile and the phase shown in Fig. 2(b). This demonstrates that the incident pulse is well compressed down to 10 fs (8 fringes within FWHM) despite a substantial amount of GVD present in our system.
3. Spatial characterization of the broadband longitudinal fields
In order to characterize the broadband longitudinal fields, we focused the beam to a glass cover slip and scanned the tip over the focused spot at low powers (~1 nW). The Rayleigh back-scattering spectra was observed by using a spectrometer (Princeton SP-2300i) equipped with thermo-electronically cooled CCD (Andor Technology DU401-BR-DD), which was synchronized with the scanning motion of the tip. Note that we used the NPBS in this instance so that the polarization dependence (for both transmission and reflection) was almost negligible over the entire bandwidth. This is important for observing the Rayleigh back-scattering images since any polarization dependent transmission through the DM causes an artifact on the images although the generated longitudinal fields are not affected by the DM.
The spectral images of the focused spot are shown in Fig. 3(a) . All images are scaled so as to maximize the visual contrast. We observed a single bright spot at the centre, in good agreement with the calculated spatial distribution of the longitudinal fields (see Fig. 4(a) ). Note that the strength of the Rayleigh scattering is an indirect measure of the dipole strength of the plasmonic fields at the tip apex , thus representing the magnitude of the tip enhancement effect. Without any significant enhancement, a negative contrast could be observed. It was found that the image contrast sensitively depends on the focus position. In the first row of images in Fig. 3(a), better contrasts are obtained at the wavelengths between 740 nm and 820 nm while the central spot is heavily distorted at longer wavelengths (e.g. 860 and 880 nm) due to the aberrations. This distortion can be significantly reduced by translating the focus by 200 nm (see the second row of the images) at the expense of compromised image contrasts at shorter wavelengths. The comparison between these two sets of the images reveals that the aberrations of the laser beam currently limit the excitation bandwidth at the high NA focus. However, since the positive contrast is maintained over the entire wavelength range of the incident pulse, it also indicates that good coupling to the TE fields can be achieved over a bandwidth of more than 100 nm as long as the focus position is properly adjusted.
We analyzed the centre of gravity and the full width (1/e2) of the focused spot, and the diameter of the annular ring pattern, as shown in Fig. 3(b) and (c), respectively. It is seen in Fig. 3(b) that the spot position deviates significantly in the x direction, compared to that in the y direction. This is attributed to the imperfect alignment of the dispersive optical elements such as prisms and gratings in our system. However, the centre of gravity remains within the range of ± 30 nm for the majority of the wavelengths, and which is approximately one tenth of the average spot width (~330 nm, see Fig. 3(c)). This suggests that even when slight misalignment of the tip occurs with respect to the spot centre, due to a long term drift of the microscope for instance, the spectral content to which the TE field couples will not drastically change. This relaxed tip alignment tolerance is one of the advantages of the radial polarization  and is particularly useful when dealing with broadband radiation sources.
Note that this is significantly different from the linear polarization, where the longitudinal fields are split into two lobes due to the destructive interference at the centre of the focus (see Fig. 4(b)). In addition, our numerical calculation predicts that the peak intensity of the longitudinal field is more than 3 times greater in the radial polarization than in the linear polarization due to the constructive interference at the centre, as shown in Fig. 4(c). This ensures efficient coupling to the TE fields of the tip. Both the spot width and the ring diameter gradually increase with the wavelength, as expected from their wavelength dependence.
4. Spectral analyses of the image: the plasmonic response of the tip
We found that the plasmonic response of the tip can be deduced from the Rayleigh scattering images. The red curve in Fig. 5(a) represents the average spectra taken at the central spot whilst the green curve corresponds to the background spectra taken at the edge of the images, as encircled in the inset. This background originates from the total internal reflection of the high NA components of the incident beam. The ratio between these two curves is plotted in Fig. 5(b), which corresponds to the scattering spectrum of the tip. By this means the instrumental response function can be cancelled out. The spectral profile of the scattering spectrum shows obvious inhomogeneous broadening with its peak at 815 nm and width of 63 nm. We found that it can be reasonably well fit by using two Lorentzian functions (the curve is squared in Fig. 5(b)), implying that there are two sources of plasmon resonances located near the tip apex. The peak wavelengths are 832 nm and 786 nm, respectively. From the FWHM widths (dν) obtained from the fitting, the dephasing times ( = 1/πdν) are estimated to be 14.3 fs and 10.1 fs, respectively. Note that the dephasing time is proportional to the magnitude of the field enhancement. Although these values are less than the reported value for a pure metallic tip , it is substantially longer than those observed in the particle plasmons , and which also supports the strong field enhancement observed as positive contrasts of the Rayleigh backscattering images. Our result suggests that the broadband plasmonic field enhancement can be maintained by the presence of multiple narrowband resonances.
Although the tip is often modeled with a single nanoparticle, we note that a single isolated spherical gold nanoparticle would not give rise to any plasmon resonances at these wavelengths. Since our tip is coated with a 50 nm thick gold film, it is likely that a rough metallic surface is the origin of these plasmon resonances. Therefore, it is not surprising that the multiple plasmon resonances are simultaneously observed although this is not always the case. Unfortunately, we found that the establishing the correlation between the tip structure and the resonance characteristics is not straightforward. Especially, we found that the SEM (scanning electron microscope) characterization of the tip prior to the optical experiment degrades the tip enhancement, possibly due to the contaminations caused by the electron beam deposition process. However, we were able to observe strong resonances that lead to a positive contrast with a probability of 20~30% without any active control of the tip structure. This fact, combined with our observation in Fig. 5, implies that it is possible to obtain the broadband tip enhancement by using appropriately structured tips, and which should substantially improve the reproducibility of the tip enhancement.
5. Temporal characterization of the TE longitudinal fields
Owing to the temporally compressed excitation combined with efficient coupling to the TE fields of the tip, we were able to observe the back-scattered nonlinear emission (NE) signal from the tip with incident powers as low as a few microwatts. The NE almost completely disappears either when the tip is displaced from the centre of the focus or when the tip is retracted from the substrate, as shown in Fig. 6(a) . The two peaks in the far-field spectrum (blue curve) are the residual far-field Rayleigh components resulted from the filter edges (LWPF and SWPF in Fig. 1(c)), as can be deduced from the fact that the peak at 740nm exhibits the identical intensity when the tip is approached. The intensity dependence of the total NE signal measured by integrating the NE spectra exhibits the slope value of 2.5 (see Fig. 6(b)), indicating that the signal contains not only FWM , but also two-photon luminescence (TPL) . Note that the spectral distribution of the TPL is different from that of FWM and, in general, extends to shorter wavelengths. However, we hardly observed any noticeable changes on the spectral profile when we varied the incident powers. This can be attributed to the spectral response of the spectrometer, where we observed a substantial decrease in the detection efficiency below the wavelength of 650 nm. This may diminish any subtle changes of the spectral profile at shorter wavelength.
At the incident power of 20 μW, we performed the NE-FRAC measurement by synchronously detecting the entire NE signals with a combination of an analog photomuliplier tube (PMT: Hamamatsu H5783) and a lock-in amplifier, in order to obtain high signal to noise ratio. The intensity modulation was provided by a mechanical chopper at the frequency of 1.2 kHz and the integration time was set to 100 ms at each delay. The lateral positioning accuracy of the tip required to maintain the NE signal was about ± 20 nm. However, this can be readily accomplished in our NSOM setup during the near-field NE-FRAC measurement (~10 min.).
The resultant NE-FRAC trace is shown in Fig. 6(c), where the SHG-FRAC presented in Fig. 2(c) is also superimposed for comparison. Since the peak to background ratio was only slightly greater than that of the SHG-FRAC (~10), the role of the FWM seems minor and the TPL is likely the dominant component of the NE signal. Contrary to the spectrometer, our PMT exhibits greater sensitivity at shorter wavelengths (<700nm). For this reason, the TPL can be detected with a better efficiency relative to the FWM, and which may decrease the contrast of the NE-FRAC trace. We found that the NE signal slowly fluctuates even without changing the delay although we used the delay-stabilized interferometer in the NE-FRAC measurement. We attributed this to the long term stability of the tip-sample distance control, which significantly affects the yield of the NE signal than those for the lateral positioning. For this reason, the fringe structures are often distorted (see Fig. 6(a)). However, it is clearly seen that the FWHM of the NE-FRAC is comparable to that of the SHG-FRAC, and which indicates that the excitation bandwidth is well maintained. The amount of the pedestals is substantially increased in the NE-FRAC although it steadily decays as we increase the delay. This is also explained by the interplay between the incident fields and the plasmonic response of the tip, as described below.
The excitation fields for the NE signal results from the convolution of the incident fields and the temporal response of the tip [25,27]. In the previous studies on the planer nanostructures with a single plasmon resonance, significantly different FRAC traces were observed depending on the detuning between the laser wavelength and the plasmon resonance wavelength. In the case of off-resonant excitation, broadening of the FRAC trace was not significant while the onset of the side wings was observed . The former is because the non-resonant excitation dominates the NE-FRAC signal near the zero delay. The latter was caused by the temporally varying phase of the plasmons that leads to the destructive interference and beating with the incident field. In our case, the excitation bandwidth is broader than those of the constituent plasmonic resonances. Therefore, most of the spectral components serve as off-resonant excitations for a given resonance. Thus, the increased amount of the pedestals can be attributable to the above explanations. However, this does not rule out the spatial localization of the tip enhancement because of the lightening rod effect at the tip apex , which can also be efficiently induced by the longitudinal excitation. Although the magnitude of the tip enhancement obtained by this effect would be smaller than that due to the plasmonic effect, the nonlinearity of the emission process leads to a substantial amount of signal enhancement and this effect would allow us to observe ultrafast dynamics within a near-field volume.
We did not observe any significant pedestals outside the measurement delay range (< ± 80 fs) shown in Fig. 6(c). This suggests the possibility of observing the Fourier-transform CARS  signals from the near-field volume. By adequately tailoring the phase and polarization of the incident pulse, the individual plasmons can be selectively excited to manipulate the local excitation . In such a case, the plasmonic field enhancement can be conveniently used in addition to the lightening rod effect described above.
We have experimentally demonstrated the generation of broadband longitudinal fields within a tightly focused spot so that the tip-enhanced fields at the tip apex of a gold coated glass fiber tip are efficiently excited. The Rayleigh scattering images of the spot provides the quantitative information about the spot profile as well as the spectral response of the tip itself. Furthermore, we demonstrated the near-field FRAC measurement by using the nonlinear emission signal of the metallic tip while it is approached to the substrate. This offers a simple means of characterizing the ultrafast TE longitudinal fields. Our measurement has shown that the temporally confined TE excitation is possible despite the complicated spectral response of the tip mainly due to the lightning rod effect of the tip that can be efficiently induced by using the broadband longitudinal fields. Our results indicate not only the possibility of studying the temporal dynamics within the TE fields, but also indicate interesting opportunities for ultrafast nonlinear near-field spectroscopy and imaging by taking advantages of the broadband longitudinal fields.
K.F. acknowledges the financial support by Grant-in-Aid for Young Scientists (B) 23760055 and N.H. gratefully acknowledges the financial support by Grant-in-Aid for Young Scientists (A) 21686007, both from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan.
References and links
1. A. Zayats and D. Richards, eds., “Nano-optics and near-field optical microscopy” (Artech House, Boston, 2008)
4. J. Jersch, F. Demming, L. J. Hildenhagen, and K. Dickmann, “Field enhancement of optical radiation in the nearfield of scanning probe microscope tips,” Appl. Phys., A Mater. Sci. Process. 66(1), 29–34 (1998). [CrossRef]
6. S. Kawata, ed., “Near-field optics and surface plasmon polaritons,” (Springer-Verlag, Berlin, 2001)
7. S. Kawata and V. M. Shalaev, eds., “Tip Enhancement,” (Elsevier, Amsterdam, 2007)
9. W. Zhang, B. S. Yeo, T. Schmidt, and R. Zenobi, “Single molecule tip-enhanced Raman spectroscopy with silver tips,” J. Phys. Chem. C 111(4), 1733–1738 (2007). [CrossRef]
11. C. C. Neacsu, J. Dreyer, N. Behr, and M. B. Raschke, “Scanning-probe Raman spectroscopy with single-molecule sensitivity,” Phys. Rev. B 73(19), 193406 (2006). [CrossRef]
12. T. Ichimura, N. Hayazawa, M. Hashimoto, Y. Inouye, and S. Kawata, “Tip-enhanced coherent anti-stokes Raman scattering for vibrational nanoimaging,” Phys. Rev. Lett. 92(22), 220801 (2004). [CrossRef]
13. N. Hayazawa, T. Ichimura, M. Hashimoto, Y. Inouye, and S. Kawata, “Amplification of coherent anti-stokes Raman scattering by a metallic nano-structure for a high resolution vibrational microscopy,” J. Appl. Phys. 95(5), 2676–2681 (2004). [CrossRef]
15. A. Anderson, K. S. Deryckx, X. G. Xu, G. Steinmeyer, and M. B. Raschke, “Few-femtosecond plasmon dephasing of a single metallic nanostructure from optical response function reconstruction by interferometric frequency resolved optical gating,” Nano Lett. 10(7), 2519–2524 (2010). [CrossRef]
16. N. Hayazawa, Y. Saito, and S. Kawata, “Detection and characterization of longitudinal field for tip-enhanced Raman spectroscopy,” Appl. Phys. Lett. 85(25), 6239–6241 (2004). [CrossRef]
18. L. G. Cançado, A. Hartschuh, and L. Novotny, “Tip-enhanced Raman spectroscopy of carbon nanotubes,” J. Raman Spectrosc. 40(10), 1420–1426 (2009). [CrossRef]
19. K. Karrai and R. D. Grober, “Piezoelectric tip-sample distance control for near field optical microscopes,” Appl. Phys. Lett. 66(14), 1842–1844 (1995). [CrossRef]
20. D. Kobayashi, S. Kawai, and H. Kawakatsu, “New FM detection techniques for scanning probe microscopy,” Jpn. J. Appl. Phys. 43(7B), 4566–4570 (2004). [CrossRef]
21. V. V. Lozovoy, I. Pastirk, and M. Dantus, “Multiphoton intrapulse interference. IV. Ultrashort laser pulse spectral phase characterization and compensation,” Opt. Lett. 29(7), 775–777 (2004). [CrossRef]
23. A. Bouhelier, M. R. Beversluis, and L. Novotny, “Near-field scattering of longitudinal fields,” Appl. Phys. Lett. 82(25), 4596–4598 (2003). [CrossRef]
24. N. Hayazawa, K. Furusawa, A. Taguchi, and S. Kawata, “One-photon and two-photon excited fluorescnece microscopies based on polarization-control: applications to tip-enhanced microscopy,” J. Appl. Phys. 106(11), 113103 (2009). [CrossRef]
25. B. Lamprescht, A. Leitner, and F. R. Aussenegg, “SHG studies of plasmon dephasing in nanoparticles,” Appl. Phys. B 68(3), 419–423 (1999). [CrossRef]
26. M. R. Beversluis, A. Bouhelier, and L. Novotny, “Continuum generation from single gold nanostructure through near-field mediated intraband transitions,” Phys. Rev. B 68(11), 115433 (2003). [CrossRef]
27. B. Lamprescht, J. R. Krenn, A. Leitner, and F. R. Aussenegg, “Resonant and off-resonant light-driven plasmons in metal nanoparticles studied by femtosecnd-resolution third-harmonic generation,” Phys. Rev. Lett. 83(21), 4421–4424 (1999). [CrossRef]
28. N. Hayazawa, K. Furusawa, A. Taguchi, S. Kawata, and H. Abe, “Tip-enhanced two-photon excited fluorescence microscopy with a silicon tip,” Appl. Phys. Lett. 94(19), 193112 (2009). [CrossRef]
30. M. Aeschlimann, M. Bauer, D. Bayer, T. Brixner, F. J. García de Abajo, W. Pfeiffer, M. Rohmer, C. Spindler, and F. Steeb, “Adaptive subwavelength control of nano-optical fields,” Nature 446(7133), 301–304 (2007). [CrossRef]