2023 Volume 20 Issue 1 Article ID: e200011
Near-field scanning optical microscopy (NSOM) is a super-resolution optical microscopy based on nanometrically small near-field light at a metallic tip. It can be combined with various types of optical measurement techniques, including Raman spectroscopy, infrared absorption spectroscopy, and photoluminescence measurements, which provides unique analytical capabilities to a variety of scientific fields. In particular, to understand nanoscale details of advance materials and physical phenomena, NSOM has been often adopted in the fields of material science and physical chemistry. However, owing to the recent critical developments showing the great potential for biological studies, NSOM has also recently gained much attention in the biological field. In this article, we introduce recent developments made in NSOM, aiming at biological applications. The drastic improvement in the imaging speed has shown a promising application of NSOM for super-resolution optical observation of biological dynamics. Furthermore, stable imaging and broadband imaging were made possible owing to the advanced technologies, which provide a unique imaging method to the biological field. As NSOM has not been well exploited in biological studies to date, several rooms need to be explored to determine its distinct advantages. We discuss the possibility and perspective of NSOM for biological applications. This review article is an extended version of the Japanese article, Development of Near-field Scanning Optical Microscopy toward Its Application for Biological Studies, published in SEIBUTSU BUTSURI Vol. 62, p. 128–130 (2022).
Near-field scanning optical microscopy (NSOM) has a great potential to contribute to biological fields owing to its label-free super-resolution capability. Nevertheless, it has not been well exploited in biological fields since its invention a few decades ago. Recent crucial developments made in NSOM for high-speed imaging, stable imaging, and broadband imaging would open a door for NSOM to truly contribute to biological studies as a novel super-resolution bio-imaging tool.
Near-field scanning optical microscopy (NSOM) (aperture-less type) is one of the super-resolution optical microscopies, which is based on near-field light generated at a metallic tip via the localized plasmon resonance (collective oscillation of free electrons). The near-field light is confined as small as the size of the nanometrically sharp tip apex. Therefore, NSOM enables super-resolution optical imaging by raster-scanning the near-field light with a typical spatial resolution of 10–20 nm, as shown in Figure 1 [1,2]. Unlike super-resolution fluorescence microscopies [3,4], NSOM can be combined with various optical techniques, including Raman spectroscopy [5,6], infrared absorption spectroscopy [7], and/or photoluminescence measurements [8], which is a strong advantage compared with other super-resolution optical techniques. Since its development a few decades ago [9], it has contributed to various scientific fields from material science to physical chemistry [10–15]. Furthermore, it has gained much attention in the biological fields and has recently been utilized for some biological samples [16–21]. Several technical developments have also been made in NSOM, aiming at biological applications. In this review, we introduce recent advances in NSOM techniques, particularly for biological applications.
Schematic of near-field scanning optical microscopy.
High-speed imaging is one of the long-standing challenges to applying NSOM for biological studies. It typically takes a few tens of minutes to obtain an image because it physically raster-scans a metallic tip. A scanning probe microscope, such as atomic force microscopy (AFM) or scanning tunneling microscopy, is often used for tip scanning, which has restricted the imaging speed. NSOM has been utilized for static samples, such as dried or fixed biological samples [16–21]; however, capturing the motions of live samples has been impossible because of the slow imaging speed.
We have recently developed high-speed NSOM (HS-NSOM) to overcome this issue [22]. To improve the imaging speed of NSOM, we have adopted high-speed AFM (HS-AFM). HS-AFM is a well-recognized technique that allows the video recording of dynamic motions of bio-molecules with a high spatiotemporal resolution [23]. AFM topographic images were obtained at as fast as 10 frames/s by rapidly scanning a micro-cantilever tip or small sample stage. HS-AFM has contributed to numerous significant biological discoveries owing to such a great imaging capability [24–32]. As aforementioned, NSOM often utilizes AFM for scanning and position control of a metallic tip, and HS-AFM can be easily combined in NSOM. High-speed optical imaging with nanoscale spatial resolution would provide new insight into the biological fields.
Figure 2 shows the experimental setup of HS-NSOM. HS-AFM was installed on an inverted optical microscope. The incident laser was focused on the tip apex through the oil-immersion objective lens to excite the near-field light. We put the spatial mask for evanescent illumination of the metallic tip. The signals were collected through the same objective with the back-scattering configuration. The pinhole was inserted as a confocal element to remove unwanted noises. The signals were finally detected by the highly sensitive avalanche photodiode. The avalanche photodiode was synchronized with the tip scanning motion to construct near-field optical images.
Experimental setup of HS-NSOM. Reprinted from Ref. 22 with permission from Elsevier.
The metallic tip is one of the most important parts in NSOM. Therefore, we developed a fabrication method of a metallic tip on a micro-cantilever. The micro-cantilevers are much shorter and smaller than conventional cantilevers; therefore, they have a high resonance frequency, which is essential for high-speed imaging in HS-AFM. The fabrication procedure of the metallic tip and the scanning electron microscopy (SEM) images of each fabrication step are shown in Figures. 3 (A) and 3 (B), respectively. An amorphous carbon tip was fabricated via electron beam deposition [28]. Subsequently, it was smoothly coated with a silver layer using the sputtering technique. The sputtering direction was parallel to the cantilever so that both sides of the cantilever were evenly coated with silver, which avoided the thermal bending of the cantilever. Note that the silver layer made the cantilever slightly thicker and stiffer so that its resonance frequency increased from ~400 kHz to ~600 kHz. The spring constant should be also slightly increased due to the same reason. The rod-like silver structure was formed at the end of the cantilever by smoothly coating the carbon tip with silver, which acts as an efficient optical antenna that facilitates the localized plasmon resonance. The localized plasmon resonance and the resulting strong near-field light were confirmed in numerical simulation (Fig. 3(C)). Figure 3 (D) shows the near-field light intensity dependence on the rod length with different silver coating thicknesses. We confirmed that strong near-field intensity was obtained at appropriate rod lengths.
Metallic tip fabricated on a micro-cantilever for HS-NSOM. (A) Fabrication procedure of the metallic tip. (B) SEM images of each fabrication step. (C) Simulated electric field distribution map around the rod-like silver structure. (D) Near-field light intensity dependence on the rod length with different coating thicknesses. Reprinted from Ref. 22 with permission from Elsevier.
Finally, we performed HS-NSOM imaging with the constructed system. We used DNA labeled with fluorescent YOYO-1 as a sample. We rapidly scanned the metallic tip on the DNA sample in a liquid environment and detected fluorescent signals excited by the near-field light. The super-resolution near-field fluorescence image of DNA was obtained, as shown in Figure 4 (A). The imaging time was 10 s, which was much faster than that of the conventional NSOM. For comparison, the confocal fluorescence image of DNA was also shown in Figure 4 (B). Figure 4 (C) shows the line profiles of fluorescent intensities along with white dotted lines in Figures. 4 (A, B) for both near-field and confocal images. The spatial resolution was approximately 39 nm in the near-field imaging, which was far beyond the diffraction limit. We further demonstrated the time-lapse NSOM imaging as shown in Figure 4 (D), wherein we obtained near-field fluorescence images every 8 s. The dynamic process of photobleaching was observed. Moreover, we tested the high-speed near-field fluorescence imaging with a fluorescent polystyrene bead and achieved an imaging time of approximately 3.5 s, which was approximately 100 times faster than the imaging speed of ordinary NSOM. To further improve the imaging speed, we have currently been working on the development of HS-NSOM. The imaging speed has currently been limited not by the scanning speed of a tip, but by low efficiency of near-field optical measurements so that the photon number is not high enough to image faster. Our optical setup still has many rooms to be improved, such as a galvano mirror scanner, in the near future. We believe that this significant improvement in the imaging speed is an important step forward toward the label-free near-field optical observation of biological dynamics.
HS-NSOM imaging of DNA. (A) Near-field fluorescence image of DNA obtained in 8 s. (B) Confocal fluorescence image of DNA. (C) Fluorescence intensity line profiles along with the white dotted lines in (A) and (B). (D) Time-lapse near-field fluorescence imaging of a DNA fraction. Reprinted from Ref. 22 with permission from Elsevier.
Improving the imaging speed in NSOM leads to the optical observation of nanoscale biological dynamics. Meanwhile, imaging of samples for long-term duration is also effective for biological samples in some cases. For example, near-field Raman imaging of biological samples requires an extremely long duration as we have to detect extremely weak Raman signals from multiple points. Note that in this case, we suppose not dynamic but static biological samples to be imaged for a long duration. However, stable NSOM imaging for a long-term duration is challenging owing to the mechanical drift of the system. In NSOM, a metallic tip must be placed exactly at the center of the incident focal spot with nanoscale precision for efficient excitation of the near-field light and efficient detection of near-field signals. However, the tip and/or laser focus can easily drift. The tip can drift laterally, and the laser focus, that is, an objective lens, can drift vertically. It deteriorates NSOM measurements even if the amount of drift is the nanoscale. Therefore, the imaging time in NSOM has typically been limited to several tens of minutes. Stable and robust NSOM enables super-resolution imaging of bio-samples that could not be imaged owing to weak optical signals.
Therefore, to hold the tip in the exact center of the incident focus, we developed the drift compensation system [33]. We employed a galvano mirror scanner for the lateral drift compensation and used an objective lens piezo positioner for the vertical drift. By rapidly scanning the incident laser around the tip apex using the galvano mirror scanner, the lateral drift was detected. By detecting the scattered signal from the tip apex, the tip position was determined. When the tip position drifted with respect to the laser focus, it was automatically compensated by the galvano mirror scanner. As for the vertical drift, a guide laser was introduced to the objective lens with a certain angle, which was monitored using a position sensor. When the laser focus drifted vertically, which means the incident laser was defocused owing to the drift of the objective lens, the vertical drift was detected as the lateral drift of the guide laser on the position sensor. It was compensated by the objective lens piezo positioner through the PID feedback. More details were described in the previous report [33].
Figure 5 (A) shows the time variation of the Raman intensity from the silicon cantilever tip with and without the compensation. The silicon cantilever tip was placed at the focal center, and the Raman signal from the silicon tip at 520 cm–1 was monitored. In the case of no compensation, the Raman signal gradually decreased and completely disappeared after 50 min. In contrast, the Raman intensity was constant for 140 min with the compensation. We confirmed that our compensation system can hold the tip at the exact center of the focal spot for a long duration. We performed long-term near-field Raman imaging of tungsten disulfide (WS2), an atomically thin two-dimensional material, as shown in Figure 5 (B). Here, the imaging time was 7 h. We obtained a super-resolution near-field Raman image with a much larger field of view than that of ordinary NSOM techniques owing to such stable and long-term near-field optical imaging. We have overcome the limitation of the imaging time owing to the mechanical drift, which in principle allows imaging forever unless it meets another factor that limits the imaging time. Another limiting factor, for example, can be tip degradation. As we used a silver tip, which is oxidized within several hours, it limits the imaging time. However, this can be simply solved by replacing silver to gold or coating silver with a thin protection layer.
Stable NSOM imaging with the drift compensation system (A) The time variation of the Raman intensity of the silicon cantilever tip at 520 cm–1 with and without compensation. (B) Stable long-term near-field Raman imaging of tungsten disulfide (WS2), an atomically thin two-dimensional material, with an imaging time of 7 hours. The image was constructed by the Raman intensity of A1g mode from WS2 at 422 cm–1. Reprinted from Ref. 33 in accordance with the CC BY license.
As the Raman signals of biological samples are usually very weak, obtaining a Raman signal with a reasonably high signal-to-noise ratio can require several seconds. Therefore, performing near-field Raman imaging for such a sample in a limited imaging time is not feasible, because it requires at least several hours for taking an image with approximately 100×100 pixel numbers. There have been only a few studies reporting the near-field Raman imaging of biological samples [17,20,21]. We believe that our NSOM apparatus equipped with the novel drift compensation system will be a useful tool for super-resolution near-field bio-imaging.
The near-field light plays a central role in NSOM, which is generated through the localized plasmon resonance at the metallic tip. Recently, plasmon nanofocusing has merged as another method for generating the near-field light [34–36]. In plasmon nanofocusing, plasmons are excited at a plasmon coupler on the metallic tip shaft and propagated toward the tip apex, compressing their energies and eventually creating a strong near-field light at the tip apex. Because plasmon nanofocusing generates the near-field light on a different mechanism, it has different characteristics that provide unique functions to NSOM.
Suppression of background noises caused by incident light is one of the distinctive characteristics of plasmon nanofocusing. In an ordinary NSOM, the tip apex is directly irradiated with the incident light to excite the localized plasmon resonance, which means that the near-field and incident lights are spatially overlapped. The strong scattered noises caused by the incident light often overwhelm the near-field signals, which significantly degrade the sensitivity in NSOM measurements. In contrast, in plasmon nanofocusing, a plasmon coupler, such as a grating structure, on the tip shaft is irradiated with the incident light, which is located far from the apex. Therefore, the incident light is spatially separated from the near-field light at the tip apex, which drastically improves the sensitivity [36,37]. We indeed demonstrated NSOM measurements based on plasmon nanofocusing and confirmed high contrast in near-field optical images owing to background suppression [38]. This characteristic is important, particularly for biological applications, because it significantly suppresses not only the background noises but also the photo damage of biological samples.
Another unique characteristic of plasmon nanofocusing is broadband. The localized plasmon resonance is literally a resonance phenomenon that occurs only at a resonance wavelength. Therefore, it works only in a narrow wavelength range. The typical bandwidth of the plasmon resonance is 100–200 nm only. In contrast, plasmon nanofocusing is a phenomenon based on the propagation of plasmons; therefore, it works over a broad wavelength range [39–41]. Therefore, we do not have to adjust the wavelength to plasmons, and we can readily select a suitable wavelength for biological samples. Moreover, we can exploit optical techniques that require multiple wavelengths, including non-linear optical techniques, which widely extend the capability of NSOM for biological applications [42,43].
We fabricated a metallic tip suitable for plasmon nanofocusing to demonstrate broadband plasmon nanofocusing and its application for NSOM. Figure 6 shows the fabrication procedure and the fabricated metallic tip. We used a commercially available silicon cantilever tip. It was oxidized in an electric furnace with water vapor to convert silicon to oxidized silicon, which is optically transparent. As the tip originally has a pyramidal shape, one surface of the pyramidal tip was coated with a silver thin layer via thermal evaporation. The coating thickness was 40 nm. The silver layer automatically formed a tapered structure owing to the triangle shape of the evaporated surface. The evaporation direction was perpendicular to the evaporated surface, and the evaporation speed was as fast as a few nm/s, which made the silver coating smooth enough for plasmon propagation. The surface roughness was <1 nm. Finally, a slit structure was fabricated as a plasmon coupler by focused ion beam lithography at a distance of approximately 6 μm from the apex, which is far enough to spatially separate the near-field and incident lights. The slit structure is more suitable for broadband coupling than the grating structure.
Fabrication of the metallic tip for plasmon nanofocusing (A) Fabrication procedure of the metallic tip for plasmon nanofocusing. (B) SEM image of the fabricated tip. The inset shows the side view of the tip. Scale bars, 2 μm (inset, 200 nm). Reprinted from Ref. 39 in accordance with the CC BY license.
To evaluate the broadband property of plasmon nanofocusing on the fabricated metallic tip, we performed numerical simulations. A similar tip structure was designed for simulations based on the finite-difference time-domain method. Figure 7 (A) shows simulated electric field distribution maps in the vicinity of the tip apex under the plasmon nanofocusing process at different excitation wavelengths. The confined near-field light was observed at the apex for wavelengths with a range of 460–1,200 nm. We confirmed that plasmon nanofocusing was excited over the wide wavelength range from the entire visible region to the near-infrared region. We irradiated the slit on the tip using white light and observed it through several band-pass filters to experimentally verify the broadband property, as shown in Figure 7 (B). The dashed circles indicate the position of the tip apex. A bright spot, which is a scattered signal generated from the near-field light, was observed at the tip apex for wavelengths with a range of 500–800 nm over the entire visible range, which again confirms the broadband characteristic of plasmon nanofocusing. Notably, the detection range of the observation camera is narrower than the wavelength range of plasmon nanofocusing; therefore, experiments were performed only in the visible range. However, we believe that plasmon nanofocusing can be also excited in the near-infrared region at a wavelength of more than 800 nm.
Investigation of the broadband property of plasmon nanofocusing. (A) Simulated electric field distribution maps in the vicinity of the tip apex under the plasmon nanofocusing process at different excitation wavelengths. (B) Optical images of the metallic tip with white light irradiation to the slit, as observed through several band-pass filters. The dashed circles indicate the position of the tip apex, where bright spots were observed as scattered optical signals from the near-field light. Scale bars, 100 nm (A) and 2 μm (B). Reprinted from Ref. 39 in accordance with the CC BY license.
As we confirmed that plasmon nanofocusing was excited in a wide wavelength range, which enables the creation of broadband near-field light, we applied this novel tiny light source for NSOM measurements. By illuminating samples using the broadband near-field light, a scattering spectrum of samples is obtained with the nanoscale spatial resolution. Figure 8 (A) shows the AFM image of carbon nanotubes (CNTs) used as a sample. The left one is the metallic-type CNTs (m-CNTs), and the right one is the semiconducting-type CNTs (s-NCTs). To obtain scattering spectra, we located the broadband near-field light on the red and blue crosses in Figure 8 (A), as shown in Figure 8 (B). The scattering spectra were normalized using a reference spectrum, which was obtained without samples. We observed that the semiconducting-type CNT showed three peaks at 620, 680, and 730 nm, whereas the metallic-type CNT showed only a single peak at 620 nm. The peaks are attributed to the resonant scattering due to energy bandgaps of CNTs. Therefore, we can investigate energy bandgaps at the nanoscale using the broadband near-field light. We obtained the scattering spectra along with the yellow dotted line shown in Figure 8 (A). As presented in Figure 8 (C), the scattering peaks appeared only at the location of CNTs, which confirmed that the scattering peaks originated from CNTs. Furthermore, the quick spectral change indicates a high spatial resolution far beyond the diffraction limit of light. We further performed super-resolution scattering spectral imaging on the same sample, as shown in Figure 8 (D). For the 620-nm wavelength, both the semiconducting- and metallic-type CNTs were imaged, whereas only the semiconducting-type CNT was imaged at 680- and 730-nm wavelengths, which corresponds to the results shown in Figure 8 (B). Although we have applied this technique for CNTs, this advanced NSOM technique will surely contribute to biological studies in the near future.
Broadband NSOM imaging using plasmon nanofocusing. (A) AFM image of CNTs. (B) Near-field scattering spectra obtained at the red and blue crosses in (A). (C) Near-field scattering spectra obtained along the yellow dotted line in (A). (D) Near-field scattering spectral imaging of CNTs constructed at 620, 680, and 730 nm. Scale bars, 100 nm (A and D). Reprinted from Ref. 39 in accordance with the CC BY license.
In this review, we introduced newly developed NSOM techniques, focusing on high-speed imaging, stable imaging, and broadband imaging. Although NSOM has still not been well adopted in the biological fields, its imaging and analytical capabilities have been drastically improved in recent years.
Considering the biological applications of NSOM, as we have to physically touch a sample with a tip, we think that a suitable biological target is cell membranes rather than cell cytoplasm. Using our HS-NSOM technique, it would be possible to observe dynamics of lipid rafts and/or membrane proteins. For example, diffusion speed of lipid rafts is typically several hundreds of nanometers per second, and it is slower for membrane proteins. If the imaging speed is improved up to 10 frames per second or more, which is same as the imaging speed of HS-AFM, we can capture their motions at the nanoscale without labelling, which should give a significant impact to the biological fields. Regarding the stable NSOM imaging technique, we can image much wider areas of cell membranes or lipid bilayers, although samples should be fixed in this case. It is now possible to image for more than several hours, which means that we can image a several times wider area at once. Therefore, we can study these biological samples in detail, or have a higher chance to discover rare biological events or signals hidden in cell membranes.
A concern regarding the biological applications of NSOM is the contamination of metallic tips during measurements in physiological conditions. Biological substances and contaminations can be easily adsorbed to the surface of metallic tips. The adsorbed substances provide noises to the near-field signal and even degrade the metallic tip itself. Due to this issue, NSOM measurements have been mostly performed in air or vacuum and are very rare in physiological conditions. However, some important developments are already made to solve this issue, including the chemical synthesis of an extremely thin silica coating on metallic tips, which works as a protection layer against contaminations [44]. A more reliable method to deposit thin coatings of silica or alumina using atomic layer deposition was also recently reported [45,46]. A multilayer metal coating with chemical modification of a thin zirconia layer has also shown great stability in liquid environment [47,48].
The essential and unique advantage of NSOM is that we can combine various optical techniques with the nanoscale spatial resolution in a label-free manner. As a novel tool that provides a new insight to optically visualize nanoscale details of biological samples, we expect that NSOM will widely and deeply contribute to biological fields in the near future.
The author declares no conflicts of interest.
T.U. wrote the manuscript.
The evidence data generated and/or analyzed during the current study are available from the corresponding author on reasonable request.
This work was partly supported by Grant-in-Aid for Scientific Research (B) 20H02658, and JST PRESTO Grant Number JPMJPR19G2.
