Research ArticleOPTICS

White nanolight source for optical nanoimaging

See allHide authors and affiliations

Science Advances  03 Jun 2020:
Vol. 6, no. 23, eaba4179
DOI: 10.1126/sciadv.aba4179


Nanolight sources, which are based on resonant excitation of plasmons near a sharp metallic nanostructure, have attracted tremendous interest in the vast research fields of optical nanoimaging. However, being a resonant phenomenon, this ideally works only for one wavelength that resonates with the plasmons. Multiple wavelengths of light in a broad range confined to one spot within a nanometric volume would be an interesting form of light, useful in numerous applications. Plasmon nanofocusing can generate a nanolight source through the propagation and adiabatic compressions of plasmons on a tapered metallic nanostructure, which is independent of wavelength, as it is based on the propagation, rather than resonance, of plasmons. Here, we report the generation of a white nanolight source spanning over the entire visible range through plasmon nanofocusing and demonstrate spectral bandgap nanoimaging of carbon nanotubes. Our experimental demonstration of the white nanolight source would stimulate diverse research fields toward next-generation nanophotonic technologies.


Multiple wavelengths of light coexisting in a confined nanometric volume can constitute an interesting optical effect. Such unique nanolight holds great potential as a useful platform for diverse research fields, providing opportunities to probe a sample over a wide wavelength range, apply arbitrary wavelengths of light to an object, or induce light-light interaction between different wavelengths, all on the nanoscale. Optical antennas have played substantial roles over recent decades in confining light on the nanoscale through localized plasmon resonances in metallic nanostructures (1, 2). Since nanoscale localization also leads to light-field enhancement, various fundamental and applied studies on nanolight have been reported (36). Plasmon resonance, however, being a resonant phenomenon, leads to enhance a certain wavelength corresponding to the plasmon resonance wavelength and hence does not facilitate broadband nanolight source generation.

Plasmon nanofocusing has recently been gaining attention as an alternative approach for creating nanolight sources. In this phenomenon, a nanoscale light source is generated by the propagation and superfocusing of surface plasmon polaritons (SPPs) toward the apex of a tapered metallic structure. Adiabatic compression of light energy into a nanoscale volume leads to enormous enhancement of the light field at the apex (7, 8). In addition, since the metallic structure is remotely illuminated at a location far from the apex in this configuration, it drastically reduces the incident light background in comparison with the direct excitation of plasmon resonances, where the incident light illumination coincides with the nanolight source generated at the location of the observed sample (912). Although this ability of background-free characteristic has been recognized as a strong advantage of plasmon nanofocusing, another feature of plasmon nanofocusing that can be of major interest is the broadness of the wavelength range in the nanolight source. Unlike resonance-based effects, plasmon nanofocusing can be induced at any wavelength using the same metallic taper, since it is based on the propagation of SPPs. The broadband property of plasmon nanofocusing has been exploited, for instance, for four-wave mixing with a high nonlinear conversion efficiency, in which multiple wavelengths were involved in a nanoscale interaction (13, 14). Plasmon-nanofocused broadband light sources could become powerful tools for a wide range of research fields, as any wavelength of light could be confined to a nanoscale volume by using the same tapered structure.

Here, we introduce a white nanolight source spanning over the entire visible wavelength range generated via plasmon nanofocusing and demonstrate broadband energy bandgap optical imaging of carbon nanotubes (CNTs) using the white nanolight source. Although plasmon nanofocusing can be excited in a broad wavelength range, it has principally been used thus far in the near-infrared range or on the longer wavelength side of the visible range because of limitations imposed by the taper materials. Gold has been used as a material for conical tapered structures since it can be easily fabricated by elongating and sharpening a gold wire in an electrochemical etching process (1518). Although this fabrication method is well established to obtain gold conical tips with smooth surfaces, which are crucial for efficient SPP propagation, it is not suitable for other plasmonic materials such as silver and aluminum. When sharpened by a similar process, the surfaces of silver and aluminum wires become roughened by the chemical reactions. Although these materials are expected to give better results for their usage in plasmon resonance (1920), their roughness deteriorates the efficiency of SPP propagation. Therefore, plasmon nanofocusing–based optical imaging has mainly been performed in the near-infrared range, where gold is a better plasmonic material owing to lower ohmic losses, but not in the visible or ultraviolet ranges where silver or aluminum would be more suitable (19, 21, 22).

We have recently reported an efficient fabrication method for tapered metallic structures based on a thermal evaporation technique (23). We used a commercially available silicon cantilever with a pyramid-shaped tip. One surface of the pyramid was used as a base to obtain a two-dimensional metallic taper. By thermally evaporating the metal onto one surface of the pyramid, with the evaporation direction perpendicular to the surface, we obtained an extremely smooth metallic coating in the two-dimensional tapered shape. One of the advantages of our fabrication method is that any types of metal can be applied, as the thermal evaporation technique is not limited to particular types of metals. We thus used silver as a taper material and achieved highly efficient plasmon nanofocusing with 100% reproducibility at 642 nm. Some reports also showed a fabrication method of tapered metallic structures by means of the metal evaporation on a conical tip that was fabricated by electron beam deposition (2426). However, our extremely smooth and optimized metallic coating can further facilitate broadband plasmon nanofocusing with longer propagation length over a broad wavelength range. In the present report, we therefore describe the utilization of this technique for white plasmon nanofocusing over a broad range of visible wavelengths (Fig. 1A).

Fig. 1 Plasmon nanofocusing of white light for full spectral nanoanalysis.

(A) Schematic of plasmon nanofocusing for white light and spectral bandgap nanoanalysis. (B) Schematic of tapered metallic structure used for simulation. (C) Superposition of waves with different wave vectors.

Plasmon nanofocusing has other important advantages for its application in near-field optical imaging, such as direct spectroscopic detection, no need of a lock-in amplification, and the absence of any chromatic aberration. Since the direct illumination of laser light on a metallic tip apex to excite plasmon resonance in traditional plasmonic nanoimaging techniques results in a tremendous amount of background from the incident light, lock-in amplifier detection is required to selectively extract the nonlinear response from the near-field signals (27). In contrast, we can directly detect a broadband spectrum through a spectroscope, owing to the background-free excitation scheme of plasmon nanofocusing (12, 13, 28, 29). In addition, all wavelengths of light are confined within the same nanometric volume near the apex of a tapered structure, which guarantees the generation of a white nanolight source at the apex without any chromatic aberration. All these advantages provide a perfect solution for broadband spectral nanoimaging in the visible range. Since visible light carries an energy that can be used to probe electronic transitions, which play important roles in semiconductor applications, we exploited a white nanolight source for bandgap imaging of CNTs. Different bandgap energies were clearly imaged with nanoscale spatial resolution. As absorption in most materials or biological molecules is in the visible range, our results can contribute to diverse research fields. In addition, as discussed later, the plasmon-nanofocused white light source is not limited to bandgap analysis but has great potential as a unique form of light to realize various nanophotonics applications.


Design and fabrication of tapered metallic structure for broadband plasmon nanofocusing

To obtain a broadband white nanolight source, we designed and fabricated a tapered metallic structure. Figure 1B shows our design for the tapered structure, composed of an oxidized silicon pyramidal tip and a thin silver layer on one surface of the pyramid. The thickness of the silver layer and the apex size are 40 and 20 nm, respectively. Gratings are often used as a coupler between light and plasmons, where grating periods are adjusted to satisfy the wave vector matching between the incident light and plasmons. To facilitate the coupling of light of a certain wavelength, we need to accordingly adjust the grating periods, and hence, a particular grating couples only one particular wavelength. On the other hand, since white light can be considered as a combination of multiple wavelengths, a superposition of which will be a pulse wave function. Since a sinusoidal wave couples with a grating, a wave with pulse wave function would couple with a single slit, because superposition of gratings with varying periods would result in a structure close to a single slit. Therefore, a single slit is expected to serve as a good coupler for a white incident light of broadband, as depicted in Fig. 1C. From our experimental results as well as numerical calculations, we found that a single slit of 200 nm in silver worked best for coupling light in visible range. Hence, a 200-nm-wide slit was used as a white light coupler in our tapered structure.

To investigate the capability of our design to produce a nanolight source with a wide range of wavelengths, we calculated the electric field distributions in the vicinity of the apex at multiple excitation wavelengths using the finite-difference time-domain (FDTD) method (Fig. 2A). We observed strong electric fields confined at the tip apex at excitation wavelengths ranging from 460 to 1200 nm. Although the propagation length of plasmons in the visible range is expected to be tens of micrometers, which is large enough for our purpose, it has a wavelength dependence and becomes shorter as wavelength shortens. Therefore, slightly weaker electric fields were observed at wavelengths of 460 and 550 nm, and hence, the image contrasts were increased by five- and twofold, respectively, to improve the visual clarity of the images. A possible cause for concern is whether sufficient coupling can be obtained with a single slit compared to that produced by the commonly used grating structure. We confirmed that a single slit indeed had a few times lower coupling efficiency than a grating designed to couple a particular wavelength. However, the slight decrease in coupling efficiency did not largely affect our measurements. Since a typical coupling efficiency of an optimal grating structure is around 10%, we estimate the coupling efficiency of the single slit to be about 2 to 3%. Although a recent report has shown for a fiber coupled silver nanowire that high coupling efficiency at wavelengths of 532 and 633 nm could be achieved (30), we have confirmed that the coupling efficiency in our structure is high enough for experimental observations, as shown in experimental demonstrations later. Figure 2B shows the near-field spectrum monitored 6 nm below the tip apex, demonstrating the broadband characteristic of our plasmon nanofocusing scheme. The tapered structure with a 200-nm-wide slit enabled generation of a broadband nanolight source over a wide wavelength range, from about 400 to at least 1200 nm, spanning over the entire visible region and even reaching in the near-infrared region.

Fig. 2 Broadband property of plasmon nanofocusing evaluated by FDTD simulations.

(A) Electric field distribution maps in the vicinity of the apex of the tapered silver structure produced by FDTD simulations. Scale bars, 100 nm. The plasmon coupler slit, where white light was illuminated, is not shown, as it is out of the frame. (B) Simulated near-field spectrum detected 6 nm below the apex.

Subsequently, we fabricated the modeled structure. Figure 3A illustrates the fabrication process for the designed tapered structure. We used a commercially available silicon cantilever tip with a pyramidal shape. The cantilever was first oxidized at 1000°C in the presence of water vapor in an electric furnace to produce oxidized silicon, which is preferable compared to silicon in terms of energy loss during SPP propagation (31). We then deposited a 40-nm-thick silver layer on one surface of the pyramidal structure of the oxidized silicon tip. Obtaining a smooth silver coating is crucial to reduce energy loss during SPP propagation. To fabricate a smooth silver coating, we evaporated silver at a high evaporation rate of 1.5 nm/s with the evaporation direction perpendicular to the target surface. The detailed mechanism for obtaining a smooth silver coating is described in section S1 and fig. S1. Under these conditions, we could obtain a surface roughness under 1 nm, which is sufficient for SPP propagation. We confirmed in our previous study that such a smooth coating leads to highly efficient and reliable plasmon nanofocusing (23). Last, we fabricated a single slit of 200 nm in width at a distance of 8 μm from the tip apex via focused ion beam (FIB) lithography. This separation of 8 μm from the tip apex is sufficient to avoid any possible overlap between the incident and the plasmon-nanofocused light. Figure 3B shows the scanning electron microscopy image of the fabricated tapered structure. The side view of the silver coating, shown in the inset, confirms that a smooth silver layer was deposited on the tip surface. Tiny silver particles deposited on the other surfaces of the pyramidal tip do not affect the plasmon nanofocusing process, as they are sufficiently separated from the silver layer.

Fig. 3 Fabrication of a tapered silver structure on a cantilever tip.

(A) Schematic of fabrication process of the tapered silver structure on a cantilever tip. (B) Scanning electron microscopy image of the fabricated tapered silver structure on the cantilever tip. The inset shows a side view of the silver layer. Scale bars, 2 μm (inset, 200 nm).

White nanolight source generated by plasmon nanofocusing

To investigate whether the fabricated tapered structure produces confined white light through plasmon nanofocusing, we illuminated the slit structure with a supercontinuum laser that spans over the visible–to–near-infrared range and has high coherence, by tightly focusing the incident light at the center of the slit. Figure 4A shows an optical image of a fabricated tip illuminated by the supercontinuum laser. The dashed lines indicate the outer shape of the tip and the position of the slit. The inset shows a zoomed image of the tip apex. The incident polarization was perpendicular to the slit, as indicated by the arrow. A white light spot was clearly observed at the tip apex, which was the far-field radiation from the nanolight source induced by plasmon nanofocusing. To verify that the light spot was associated with plasmon nanofocusing, we investigated the dependence of the intensity of the light spot on the polarization of the incident light. When the incident polarization was rotated by 30°, the intensity of the light spot at the apex was reduced (Fig. 4B), and it disappeared when the polarization was parallel to the slit (Fig. 4C). This change in the intensity of the optical spot is plotted in a polar graph in Fig. 4D, where 0° corresponds to the polarization parallel to the slit and 90° corresponds to the perpendicular polarization. We observed a dipole-like intensity variation, which indicates that the light spot was indeed linked with the light coupling at the slit and hence originated from the far-field emission by the plasmon-focused nanolight source (29). Note that the best coupling is expected when incident polarization is perpendicular to the slit. This observation is also in good agreement with the simulation results (green curve in Fig. 4D).

Fig. 4 Optical observation of a white nanolight source generated through plasmon nanofocusing.

(A) Optical image of a tapered silver structure under illumination by supercontinuum laser at its slit. The locations of the boundaries of the tip as well as the slit are indicated by dashed lines. The inset shows a zoomed image of the apex. Incident polarization was normal to the slit as indicated by the arrow. (B and C) Optical images of the same tapered silver structure with supercontinuum laser illumination at different incident polarizations, as indicated by the arrows. (D) Polar graph of the light spot intensity at the apex with respect to the incident polarization; 0° and 90° correspond to parallel and perpendicular polarizations, respectively. (E) Optical images of the tapered silver structure illuminated with a supercontinuum laser, observed through a series of band-pass filters indicated by their central wavelengths. (F) Scattering spectrum of the optical spot at the apex of the tapered silver structure. a.u., arbitrary units. (G) Simulated near-field spectrum calculated at the tip apex. Scale bars, 2 μm (A and E).

Although the nanolight source looks white in color, this does not guarantee that it is broadband. We therefore observed the tapered structure through a series of band-pass filters. In these experiments, we could measure only the visible range because our optical setup was designed for visible light, and it is not sensitive in the near-infrared range. Bright optical spots were observed over the wavelength range of 500 to 800 nm, displayed as the colorful optical images in Fig. 4E. These images visually confirm that the nanolight includes components from a wide wavelength range. For a more accurate observation, we have also spectrally investigated the optical spot at the apex, as shown in Fig. 4F. Optical signal was obtained over a broadband spectral range from 450 to 700 nm, which almost spans over the entire visible range, proving that a white nanolight source was generated at the tip apex. This observation is also in accord with the simulation result, shown in Fig. 4G. In the experimental observation, a slightly higher intensity was obtained in the shorter wavelength range, compared with that in the simulation, which is because of the far-field scattering in our experiment. Note that we can only indirectly measure the near-field light by observing the scattered far-field light in the experiment, whereas the near-field spectrum was directly monitored in simulation. Since the scattering efficiency increases as the wavelength shortens, a relatively higher intensity was observed in the shorter wavelength range in the experiment.

Spectral bandgap imaging via plasmon-focused white nanolight source

Using the prepared plasmon-nanofocused white light source, we performed spectral nanoanalysis of CNTs. Near-field spectroscopic study of CNTs has been so far widely demonstrated, mostly through near-field Raman spectroscopy using near-field light excited by a narrow-band single-mode laser, which can analyze nanoscale chemical properties and atomic vibrations of CNTs (3235). Having a unique near-field light covering a broad wavelength range, one can investigate nanoscale scattering properties of a sample, which means that it would enable scattering spectrum measurements with nanoscale spatial resolution. The CNT samples are shown in the atomic force microscopy (AFM) image in Fig. 5A. The structure observed on the right side of the image is a bundle of semiconductingtype CNTs (s-CNTs), and the structure observed on the left side is a bundle of metallic-type CNTs (m-CNTs), which are identified in advance through the preparation process described in fig. S2. Both bundles are estimated to be composed of around five CNTs, considering their heights. By illuminating these CNT bundles with the white nanolight source, we obtained near-field scattering spectra for the two types of CNTs (Fig. 5B). The exposure time was 0.1 s, and the incident power was 10 μW. To avoid any experimental fluctuation, we normalized these spectra using the reference scattering spectrum measured without the sample. The s-CNTs and the m-CNTs have different spectral features. The spectrum of the s-CNTs shows three peaks at around 630, 680, and 730 nm, whereas only one peak at 620 nm appears in the near-field spectrum of the m-CNTs. The scattering intensity is enhanced when excitation wavelength matches the absorption wavelength, i.e., the energy bandgap, of CNTs due to strong interactions between the incident light and the sample, which was also demonstrated with a Rayleigh scattering spectroscopy (36). In our case, the white nanolight source localized at the tip apex interacts with CNT bundles that have multiple bandgaps.

Fig. 5 Optical nanoimaging of CNTs using the white nanolight source.

(A) An AFM image of CNT bundles. The structures observed on the left and the right parts of the image are the metallic (m-CNTs) and semiconducting (s-CNTs) CNTs, respectively, as identified during the sample preparation process. Scale bar, 100 nm. (B) Near-field spectra of s-CNTs and m-CNTs, obtained from the locations indicated by the blue and red crosses, respectively, in (A). (C) Near-field spectra obtained pixel by pixel along the dotted line in (A). (D to F) Bandgap images constructed at 620, 680, and 730 nm, respectively. Scale bars, 100 nm.

Since the nanolight source contains photons of multiple energies, the scattering signal increases for all those photons that have the same energy as one of the bandgaps of the CNTs, resulting in several peaky shapes in a scattering spectrum. We thus interpreted that these scattering peaks are attributed to energy bandgaps of CNTs, as the scattering intensity is enhanced through resonant scattering processes at bandgaps. We then obtained near-field spectra by scanning the white nanolight source along the yellow dotted line in Fig. 5A. As shown in Fig. 5C, some peaks appear only at locations where the CNTs exist, confirming that the peaks are associated with the CNT samples. Moreover, pixel-by-pixel spectra were obtained with a step of 15 nm, and drastic spectral changes were seen from one pixel to the next. This clearly indicates the extremely high spatial resolution of this imaging modality, owing to the confinement of the light field via plasmon nanofocusing. As a comparison with our technique, we have measured extinction spectra of both samples in solution by an ordinary ultraviolet-visible (UV-Vis) spectrometer (fig. S3). Our sample solution contains CNTs with various chiralities, and individual CNTs may have different energy bandgaps, i.e., extinction peaks. In the extinction spectra, however, we did not observe any particular peaks because of a large detection volume of the UV-Vis spectroscopic technique, which gives averaged information from multiple CNTs. As extinction peaks of individual CNTs were overlapped, it is not possible to identify the energy bandgaps. This illustrates the advantage of our technique using plasmon-nanofocused white nanolight source to reveal the bandgap energies of CNTs at the nanoscale spatial resolution. Last, we performed spectral imaging with the plasmon-nanofocused white light. Figure 5 (D to F) shows the near-field intensity images simultaneously obtained from signals at 620, 680, and 730 nm, respectively, which correspond to the peaks observed in Fig. 5B. Strong intensities observed in the images, shown by yellowish color, indicate the presence of CNTs with bandgaps of the corresponding photon energies. Since both m-CNTs and s-CNTs have a bandgap peak near 620 nm, both types of CNT were observed in Fig. 5D. In contrast, at 680 and 730 nm, only s-CNTs could be observed, as they display resonance peaks at these wavelengths. Thus, different energy bandgaps among the CNTs are clearly visualized with nanoscale spatial resolution by using the plasmon-nanofocused white light. By taking intensity line profile across a nanotube observed in Fig. 5F, we have evaluated the spatial resolution to be around 31 nm, which is a typical high spatial resolution far beyond the diffraction limit of light expected in a near-field experiment (fig. S4). Moreover, by combining this approach with Raman spectroscopic analysis, it is possible to extend our technique to examine the chirality of the CNT sample, which is another important property of CNTs. The diameters of the CNTs at the locations indicated by the red and blue crosses in Fig. 5A were calculated from the radial breathing mode (RBM) in Raman spectra (fig. S5) (37). Since we know in advance that the bandgap energies of the CNTs and were also able to estimate their diameters, the chirality of the m-CNTs was estimated to be (16, 1) or (15, 3), and that of the s-CNTs was also estimated to be (19, 0), (21, 1), or (14, 1). Thus, our spectral nanoanalysis, in combination with other analytical techniques, provides insights to aid investigation of the chirality of individual CNTs.


The plasmon-nanofocused white light source is a fundamental but effective state of light that, as we demonstrated, can enable bandgap nanoimaging. Such a unique light source is not limited in terms of applications to our current demonstration, but we believe that this work paves the way to a variety of possible applications. For instance, since electron transitions are common physical phenomena among all materials, our technique can be applied widely, including to the biological molecules as a probe of their absorption properties at nanoscale spatial resolution. Moving to longer wavelengths, a mid-infrared broadband nanolight source should be also highly beneficial for material science as well as molecular biology, facilitating simultaneous probing of multiple molecular vibrations over a broad wavelength range, in contrast to the restriction to specific vibrational bands in the case of the plasmon-resonance optical antennas (38, 39). The white nanolight source should also boost the analytical capability of surface-enhanced Raman spectroscopy (3) and tip-enhanced Raman spectroscopy (6, 40) to probe molecular vibrations. With the white nanolight technique, there is no requirement to precisely design optical antennas to tune the resonance wavelength, as it can be effective for all wavelengths using a single tapered structure. Furthermore, one can analyze the same molecule with different wavelengths at the single-molecule level, which we expect will provide deeper insights as the Raman spectrum differs depending on the excitation wavelength. The white nanolight source should not be limited to applications only within analytical science and technology but could also be used for resonant optical nanotrapping and designing nanoplatforms for photochemical reactions or broadband nanosensors, thanks to the advantages of its broadband and nanoscale characteristics.

In conclusion, we demonstrated generation of a white nanolight source at the apex of a tapered silver structure through plasmon nanofocusing and performed nanoanalysis of CNTs via full-spectrum near-field imaging. We designed and fabricated a tapered structure on a commercially available cantilever that induced plasmon nanofocusing over a wide wavelength range. The spectral bandgap nanoimaging technique should contribute not only to materials science but also to biological research, probing electronic transitions at the nanoscale. Our demonstration of bandgap nanoimaging is only one application example, and the plasmon-nanofocused white-light source should stimulate research in diverse fields, taking advantage of a powerful, fundamental nanoscale optical tool with excellent wavelength flexibility.


FDTD simulation

FDTD simulation (FUJITSU, Poynting for Optics) was conducted in a three-dimensional space. A 40-nm-thick silver layer was placed on a pyramidal silicon dioxide structure. The diameter of the tip apex was 20 nm, and the cone angle was 28°, which correspond to the experimental parameters. The slit structure was located at a distance of 4 μm from the apex, which, to save the calculation time, was shorter than the distance used in the actual experimental structure. The slit was illuminated with white light. The mesh size was 2 nm around the tip apex and 10 to 100 nm in the other areas. The total calculation volume was 4 μm by 1.7 μm by 8 μm.

Fabrication of tapered metallic structure

A silicon cantilever (NT-MDT, CSG-01/bare) was placed in an electric furnace (ISUZU, UT-21H) for 20 min at 1000°C with a flow of water vapor to convert silicon into oxidized silicon, which can be distinguished by the color change of the cantilever. Silver (purity 99.99%; Nilaco) was evaporated onto one surface of the pyramidal structure of the oxidized silicon cantilever tip in a thermal evaporator (ULVAC KIKO, VPC-1100). The evaporation direction with respect to the tip surface was controlled by a handmade sample holder that tilted the sample stage. The chamber pressure was 5 × 10−5 Pa, and the evaporation rate was 1.5 nm/s. A slit structure was then fabricated at a distance of approximately 8 μm from the tip apex via FIB lithography (HITACHI, FB2200).

Observation of optical images

To observe the light spot at the apex, the fabricated cantilever tip was mounted on an inverted microscope. The slit was illuminated with a supercontinuum laser (Leukos, Leukos-SM series) through an objective lens [numerical aperture (NA) = 0.9, ×100], and the optical images were also observed through the same objective. To observe the optical image at different wavelengths, we inserted band-pass filters in front of the observation camera under the supercontinuum laser illumination to the slit. We used seven band-pass filters, with center wavelengths at 500, 550, 600, 650, 700, 750, and 800 nm, respectively (500FS10-25, 550FS10-25, 600FS10-25, 650FS10-25, 700FS10-25, 750FS10-25, and 800FS10-25, respectively; Andover Corporation). The bandwidth for each filter was 10 nm. Scattering spectrum of the light spot at the tip apex was obtained using a spectrometer (Princeton Instruments, HRS-300), where we used a pinhole (diameter = 200 μm) to extract the signal only from the vicinity of the apex. The spectrum was normalized using the spectrum of the incident supercontinuum laser.

Bandgap nanoimaging

For near-field imaging, the same supercontinuum laser was focused onto the slit of our fabricated tip through an objective lens [NA = 0.28, working distance (WD) = 30 mm, ×20]. The position of the tip was controlled by an AFM system. The incident polarization was controlled by a polarizer. The illumination position in the x, y, and z directions was controlled by an actuator. A white nanolight source was then generated via plasmon nanofocusing, and the signal scattered by the sample was collected from the bottom through another objective (NA = 1.4, ×60). We used a pinhole (diameter = 30 μm) and a spatial mask to block background light scattered from the slit. Working as a confocal element, the pinhole efficiently extracts the signal from the vicinity of the tip apex. The spatial mask allows only high NA components to be transmitted, whereas the background light originating from the slit, propagating with low-NA angles, is blocked by the mask. This configuration completely blocks background from the slit, and thus, near-field components from the apex of the tapered structure can be efficiently detected through the spectroscope even without using a lock-in amplifier. Last, the signal was detected by an electron-multiplying charge-coupled device camera (Princeton Instruments, PIXIS: 100B_eXcelon) through a spectrometer (Princeton Instruments, HRS-300). The exposure time and the incident power were around 0.1 s and 10 μW, respectively, for one spectrum. The schematic of the setup is described in detail in section S6 and fig. S6.

Sample preparation

Metallic-type (NanoIntegris Inc., IsoNanotubes-M) and semiconducting-type (NanoIntegris Inc., IsoNanotubes-S) CNT powders were separately sonicated for 30 min in a solution of 1,2-dichloroethane. We first dropped and spin-casted a solution of s-CNTs on a clean glass slip and acquired an AFM image of the sample. We then dropped and spin-casted a solution of m-CNTs on the same glass slip and observed the same area by AFM. We confirmed that the s-CNTs did not move during the subsequent processes of dropping and spin-casting the m-CNTs. At the same time, we observed the newly added CNTs, which were the m-CNTs. Therefore, in this manner, we were able to distinguish both types of CNTs in advance, verifying the sample as suitable for demonstrating and evaluating our spectral bandgap nanoanalysis technique to probe different bandgap properties of the two types of CNTs. The details are also described in fig. S2.

Calculation of CNT diameters

Since the sample that we have used is a bundle of several single-walled CNTs, it was necessary to take into account the van der Waals interactions between the tubes in this calculation. The diameters of the CNTs were determined from the Raman shift of RBM peak, and the relationship between these quantities can be expressed as d [nm] = 232/(ω − 6.5)[cm−1], where d is the CNT diameter and ω is the RBM Raman shift (37).


Supplementary material for this article is available at

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: Funding: This work was supported in part by the JSPS Core-to-Core Program, the Grant-in-Aid for Early-Career Scientists 18K14148, Grant-in-Aid for Scientific Research (A) 19H00870, the Shorai Foundation for Science and Technology, the Inamori Foundation, and the Murata Science Foundation. Author contributions: T.U. conceived and designed this project. T.U. and M.T. performed the experiments and analyzed results. P.V. and Y.S. supervised this research. T.U. and P.V. wrote the manuscript. All authors contributed to discussion of the results and commented on the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article