Research ArticleOPTICS

Label-free surface-sensitive photonic microscopy with high spatial resolution using azimuthal rotation illumination

See allHide authors and affiliations

Science Advances  29 Mar 2019:
Vol. 5, no. 3, eaav5335
DOI: 10.1126/sciadv.aav5335


Surface plasmon resonance microscopy (SPRM) with single-direction illumination is a powerful platform for biomedical imaging because of its wide-field, label-free, and high-surface-sensitivity imaging capabilities. However, two disadvantages prevent wider use of SPRM. The first is its poor spatial resolution that can be as large as several micrometers. The second is that SPRM requires use of metal films as sample substrates; this introduces working wavelength limitations. In addition, cell culture growth on metal films is not as universally available as growth on dielectric substrates. Here we show that use of azimuthal rotation illumination allows SPRM spatial resolution to be enhanced by up to an order of magnitude. The metal film can also be replaced by a dielectric multilayer and then a different label-free surface-sensitive photonic microscopy is developed, which has more choices in terms of the working wavelength, polarization, and imaging section, and will bring opportunities for applications in biology.


Analysis of the localization and dynamics of molecules and events occurring near the surfaces of cells is very important in cell biology. For example, the plasma membrane represents the barrier that all molecules must cross to enter or exit a cell, and large numbers of biological processes occur at or near this plasma membrane. These processes are difficult to image using traditional epifluorescence or confocal microscopy techniques because image details near the cell surface can easily be obscured by the fluorescence that originates from the internal contents of the cell. In contrast, total internal reflection fluorescence microscopy (TIRFM) is particularly suitable for imaging of plasma membranes because of its use of evanescent waves for selective illumination and excitation of fluorophores within a restricted specimen region (14). The imaged section is thinner (at 100 to 200 nm) than the sections that are obtained with most optical sectioning techniques, such as confocal (5) or multiphoton microscopy (6), where temporal resolutions are also limited by the need for confocal scanning. Following the rapid development of plasmonic and nanophotonic technologies, a label-free surface-sensitive photonic microscopy technique, surface plasmon resonance microscopy (SPRM), has been developed to image the surfaces or interfaces of cells (79). Surface plasmons (SPs) can be seen as one kind of evanescent waves that exist at the metal-dielectric interface, such as a metal sheet in air. They are also localized at the near-field region (such as 100 to 200 nm or less) on the substrate. They are more sensitive to environmental changes on surfaces than the evanescent waves that are used in TIRFM (10, 11). The localized electric field of SPs is also stronger than that of the evanescent waves used in TIRFM. So, SPs can be used to develop highly sensitive label-free surface imaging methods such as SPRM.

When compared with the surface-sensitive microscopy techniques of single-molecule TIRFM, SPRM offers the advantageous capability to monitor individual binding events continuously without the need to account for photo bleaching or blinking of the fluorophores. Therefore, in recent years, SPRM has found numerous applications in the study of biological targets, including cells (12), bacteria (13), viruses (7), DNA molecules (14), and proteins (15); the local electrochemical reactions of heterogeneous surfaces (16, 17); and the catalytic reactions of nanomaterials (18, 19). However, SPRM has obvious disadvantages that prevent the general use of this interesting technique. The first disadvantage is its poor spatial resolution, which is caused by the SPs propagation length (which is much larger than the diffraction limit) along its propagating direction (1219). The second disadvantage is that only noble metal films can be used as substrates for the cells in SPRM. For these metal films, which are most commonly gold or silver films, the SPs act as the imaging source, but the surface plasmon resonance (SPR) angle will increase as the wavelength of the incident beam decreases (20). However, the numerical aperture (NA) of the objective lens that is used in SPRM is finite, so SPRM can only use long-wavelength excitation beams at wavelengths such as 633 or 671 nm, which will certainly limit the potential applications of this high-surface-sensitivity imaging technique if shorter light wavelengths [e.g., ultraviolet (UV) light] were needed in some of these applications. In addition, cell cultures are more commonly grown on a glass substrate than on a metal film, which is another reason why TIRFM is more popular in cell biology applications than SPRM. In this work, to remove these two disadvantages, a galvanometric scanning system is adopted for use in the SPRM to realize azimuthal rotation illumination (ARI) for incidence angle scanning and azimuthal averaging, which combine to greatly enhance the spatial resolution of SPRM. Furthermore, we demonstrate a label-free surface-sensitive photonic microscopy method called Bloch surface wave microscopy (BSWM) with ARI, which uses the Bloch surface waves (BSWs) as the illumination sources to image the objects on the surface. BSW can be seen as the dielectric analog of SP and has the same merits, such as surface sensitivity and electric field enhancement. BSW also has some unique features that make it different from the SP, such as polarization. BSWM uses an all-dielectric multilayer (similar to the glass) for its substrate and has no limitations in terms of working wavelength.


Configuration for label-free surface-sensitive photonic microscopy

In this study, we demonstrate and compare two surface-sensitive photonic microscopy methods. The first is SPRM based on SPs on a thin silver film, and the second is BSWM based on BSWs on a dielectric multilayer. Like TIRFM, SPRM and BSWM are only sensitive to objects within the near-field region of the substrate, but not to all the objects placed on the substrate; in other words, these methods focus on the objects (e.g., the membrane proteins) within the evanescent field (BSWs or SPs) rather than those within the entire cell, so they can be called surface-sensitive photonic microscopy methods. Figure 1A shows a schematic diagram of the surface-sensitive photonic microscopy setup, in which a polarizer and a half-wave plate are used to control the polarization orientation of an incident laser beam at 635 nm. Using a pair of scanning galvanometers and a focusing lens, the expanded and collimated laser beam can be focused on any point in the back focal plane (BFP) of the objective (100×, NA of 1.49). The expanded and collimated beam will then exit the objective and strike the substrate at a given angle of incidence (θ). The beam that is reflected from the substrate will then be collected by the same objective and imaged on a scientific complementary metal-oxide semiconductor (sCMOS) array detector using a tube lens. Another polarizer was inserted before the detector to block the light that has not excited the SPs but has just been reflected by the metal film. Figure 1B shows the illumination configuration. When the focused spot was fixed at a specific position on the BFP, a collimated beam propagated outward from the objective at a particular angle (θ no rotating beam). If this angle (θ) is equal to the SPR angle, then the setup operates as the conventional SPRM (1219) in which the excited SPs will propagate on the metal film in an azimuthal direction (at the angle φ). For comparison with the proposed technique, the conventional microscopy is named SPRM with single-direction illumination (SPRM-SDI) here. In the proposed technique, the motion of the scanning galvanometers from which the incident beam will be reflected is controlled, thus allowing the focused spot to orbit (rotate) on the BFP of the objective to form a circular orbit. The exit angle (θ) of this rotating beam can be controlled based on the radius (r) of the circular orbit. If the exit angle (θ) is equal to the SPR angle, then SPs can be excited and can propagate on the metal plane at all azimuthal angles (φ) during the orbit of the focused spot [note here that, at the azimuthal direction (φ = 0) that is parallel to the incident polarization, SPs cannot be excited because the incident beam has S polarization relative to the plane of incidence; in S polarization, the polarization orientation is perpendicular to the plane of incidence] (20). Here, we call this proposed optical configuration SPRM with ARI (SPRM-ARI). The use of ARI allows incidence angle scanning and azimuthal averaging to be conducted automatically by the detector (camera) to enable capture of the final image. In the following work, the differences in the surface imaging performances of SPRM-SDI (conventional SPRM) and SPRM-ARI will be investigated and discussed.

Fig. 1 Schematic of the experimental setup and the substrates.

(A) Experimental setup. (B) Illustration of SDI and ARI. BFP, back focal plane. When the laser beam is focused on the BFP of the objective and this focal spot is fixed, it is then the SDI. When this focal spot is rotating and forms a ring on the BFP, it is then the ARI. (C) Thin silver film–coated glass operating as the substrate for SPRM. (D) Dielectric multilayer PBG structure operating as the substrate for BSWM.

For SPRM, a 45-nm-thick silver thin film coated on a glass surface was used as shown in Fig. 1C, where SPs can be excited at the metal-air interface. A dielectric multilayer that supports BSWs will also be proposed to replace the silver thin film and enable the demonstration of another type of label-free surface-sensitive photonic microscopy, BSWM, which uses the same setup that is shown in Fig. 1 (A and B). The dielectric multilayer photonic band gap (PBG) structure, where BSWs can be excited on the top surface, is shown in Fig. 1D (21). The thicknesses of each dielectric layer and total number of layers (14 layers) are exactly shown in Fig. 1D. Details of the fabrication process of the dielectric multilayer are described in the Materials and Method section. BFP images of the silver thin film, the dielectric multilayer, and the bare glass (substrate for TIRFM) are shown in fig. S1, which verify the excitation of the SPs and BSWs on the first two substrates, respectively (22). The angle-dependent excitation spectrum and the field characteristics of the SP and the BSW were simulated as shown in fig. S2. The electromagnetic fields of both the SPs and the BSWs decay exponentially from the substrates and thus only penetrate to depths of approximately 100 to 200 nm into the sample medium; this determines the section that can be imaged using each microscopy method and plays an important role in studies toward the visualization of the spatiotemporal dynamics of molecules located at or near the cell surface. Similar to the TIRFM, the imaged section is also thinner (at 100 to 200 nm) when compared with that obtained by most optical sectioning techniques, such as confocal or multiphoton microscopy (5, 6).

For simplicity, in this investigation, polymer nanowires, microwires, and nanospheres were used as objects to test the surface-sensitive imaging performances of the proposed photonic microscopes, although they can certainly be replaced with real cells in practical applications. The nanowire is approximately 150 nm in diameter, while the diameter of the microwire is approximately 3 to 4 μm. The commercial nanospheres are approximately 50 nm in diameter. Scanning electron microscopy (SEM) images of a nanowire, a microwire, and a nanosphere are shown in fig. S3.

Enhancing the spatial resolution of SPRM by using ARI

Figure 2A shows a bright-field image of a nanosphere with a diameter of approximately 50 nm that is placed on the Ag film. This single nanosphere is consistent with all previously reported works in that it produces a wave-like pattern with parabolic tails on the image (Fig. 2B) acquired using conventional SPRM (SPRM-SDI) (7, 8, 13, 15, 18). Both theoretical (23, 24) and experimental (18, 25, 26) investigations have indicated that this wave-like pattern is due to the scattered SPs generated by the nanosphere. This pattern can be regarded as the point spread function (PSF) of the SPRM-SDI (27). The tails of these curved fringes are as long as several micrometers (Fig. 2D) in the direction of plasmon propagation (along dashed line 1), which is consistent with the results reported in (7), (18), and (25); the spatial resolution of the SPRM-SDI along the plasmon propagation direction is thus quite poor and is of the same order as the SP propagation length on the silver-air interface at an incident wavelength of 635 nm (of micrometer scale) (7, 25, 27).

Fig. 2 Images acquired using SPRM-SDI and SPRM-ARI.

The Ag film operating as the substrate. A nanosphere with diameter of about 50 nm (A to C) and a winding polymer nanowire (G to I) were placed on the Ag film. (A and G) Bright-field images. (B and H) Corresponding images acquired using the SPRM-SDI. (C and I) Corresponding images acquired using the SPRM-ARI. (D to F) Intensity profiles along the dashed lines 1, 2, and 3, respectively. On (E) and (F), the peak-to-peak distances are about 460 and 370 nm, respectively. The scale bars on (A) and (G) are applicable to (B) and (C) and (H) and (I), respectively. The incident angle is fixed at θ = 43° for excitation of the SPs.

There are three irregular bright spots at the nanosphere’s position (Fig. 2B), and the distance between any two adjacent spots is approximately 460 nm (Fig. 2E). This distance can thus be regarded as the spatial resolution of the SPRM-SDI in the direction along dashed line 2. Therefore, from the pattern shown in Fig. 2B, we can conclude that the spatial resolution of the SPRM-SDI is very poor and that it differs along the different spatial directions; this will certainly induce artefacts in the images, particularly when the object to be imaged has a complex shape. For example, as shown in Fig. 2G, when a winding nanowire is placed on the Ag film to be imaged, multiple fringes and scattering noise appear around the winding nanowire in the SPRM-SDI image (Fig. 2H); this makes it difficult to recognize the nanowire if its shape was not known before the measurement. This winding nanowire is similar to a DNA molecule, which is anisotropic. Consequently, the scattering of propagating SPs by a DNA molecule or a winding nanowire is dependent on the orientation of the DNA molecule or nanowire relative to the plasmonic wave propagation direction, and this then determines the contrast and scattering patterns of the resulting SPRM-SDI images (14). This orientation is random and is in several directions, similar to the winding nanowire shown in Fig. 2G, so many irregular fringes and noise are contained in the SPRM-SDI image (Fig. 2H).

However, in the image of the single nanosphere acquired using the proposed SPRM-ARI, this wave-like pattern disappears and only two very small sized, closed bright spots (Fig. 2C) appear. The reason for this phenomenon can be explained as follows and will be analyzed with a two-dimensional model. With ARI, SPs can be excited along multiple azimuthal directions on the silver surface, meaning that the fringes or tails induced by SPs scattering can then be azimuthally averaged or homogenized. As a result, the nanosphere’s location appears to be much brighter than the other areas, which leads to the improved spatial resolution. The reason for the appearance of two closed bright spots rather than a solid bright spot can be explained on the basis of the PSF for surface plasmon–coupled emission microscopy (SPCEM) (28, 29). As reported in (28, 29), a fluorescent nanoparticle on a thin Ag film will form as a doughnut-shaped spot on the imaging plane of the SPCEM because the coupled emission has the same polarization (P polarization, where the orientation of the polarization is within the plane of incidence) in all the azimuthal directions and occurs at a single axial angle. This doughnut-shaped spot can be regarded as the PSF of the SPCEM. The principle is the same in the case of the nanosphere on the Ag film. The azimuthally rotating illumination will cause SPs to be excited on the Ag films in most azimuthal directions when scattered by the nanosphere. Similar to the coupled emission from the fluorescent spots, the light scattered by this nanosphere will also form a doughnut-shaped spot on the SPRM-ARI imaging plane. In this work, a polarizer was inserted in front of the camera, so that the doughnut-shaped spot was split into two bright spots, as shown in Fig. 2C. This splitting is the same behavior that would be observed when a polarizer is used after the radially or azimuthally polarized beam, where the cross section of this doughnut-shaped beam would also split into two spots (30).

The two closed bright spots can be regarded as the PSF of the SPRM-ARI here (29). When the patterns shown in Fig. 2 (B and C) are compared, it is clear that the spatial resolution of the SPRM-ARI is much better than that of the SPRM-SDI. The two closed bright spots are approximately 370 nm in size (peak-to-peak distance; Fig. 2F, line 3), so we can estimate that the spatial resolution of the SPRM-ARI is higher by up to approximately an order of magnitude along the plasmon propagation direction when compared with that of the SPRM-SDI. The high spatial resolution of the SPRM-ARI is advantageous in recognizing complex samples, e.g., the winding nanowire that has been resolved as shown in the SPRM-ARI image (Fig. 2I).

It should be noted here that the rotation speed of the focused spot on the BFP can be as fast as approximately 1 ms per circle, meaning that an SPRM-ARI image can be acquired in only 1 ms (this image acquisition speed is also dependent on the sCMOS response speed, and the full frame rate of the sCMOS used here is 100 frames per second). In our experiment, the rotation speed is set at 10 ms, which is faster than the speed of the differential plasmonic imaging technique. In that technique, the substrate is moved laterally by 500 nm; one image is captured before the lateral movement, and another is then captured after the movement. The subtraction of one image from the other removes the background interference patterns, and the final differential SPRM image is thus obtained (18). These mechanical movements and image processing steps will reduce the speed with which the final SPRM images are obtained.

Mechanism of spatial resolution enhancement using ARI

The SPRM-SDI image of a nanosphere can be modeled and simulated via a three-dimensional numerical method such as the finite element method or the finite-difference time-domain method. However, as reported in (18), the wave-like pattern with the parabolic tails shown in the SPRM-SDI image can be described using a simplified two-dimensional model. In our work, using this simplified model, the mechanism for spatial resolution enhancement by ARI-induced incidence angle scanning and azimuthal averaging can be demonstrated. In this model, the excited SPs can be described as a planar wave propagating in one direction, which can then be written asU1(x,y,φ)=Aei(kcos(φ)x+ksin(φ)y)(1)where φ is the propagation direction of the planar wave, k is the wavenumber, and A is the amplitude of the planar wave.

Because the nanosphere is very small when compared with the wavelength of the SP, it can be approximated as a point scatter that creates a circular wave with an origin at the nanosphere and is given by (18)U2(x,y)=Beαreikrr=x2+y2(2)

Here, B is the amplitude of the scattered wave, depending on the scattering strength and the amplitude of the plan wave, and α is the damping constant of the scattered wave. The term e− αr represent the decay of the scatter waves, and in the following calculation, the damping constant is selected as 0.5. The SPRM-SDI image can then be described as the superposition of the planar and circular waves.U(x,y,φ)=U1(x,y,φ)+U2(x,y)=Aei(kcos(φ)x+ksin(φ)y)+Beαreikr(3)

The intensity of the image is then given by|U(x,y,φ)|2=|A|2+|B|2e2αr+2ABeαr[cos(kcos(φ)x+ksin(φ)ykr)](4)

The final terms in the brackets produce an interference pattern and a long tail, as shown in Fig. 3 (A to C), where the planar wave propagation direction (φ) values are defined as 0°, 45°, and 90°, respectively. The patterns show that the orientation of the parabolic tail rotates with the change in the angle (φ).

Fig. 3 Simulated images captured using SPRM-SDI and SPRM-ARI.

The superposition of a planar wave (the illumination source) and a circular wave caused by scattering from a nanosphere led to the image shape that was observed in SPRM. In (A) to (C), the planar wave is propagating along a single direction defined as φ = 0°, 45°, and 90°, respectively, which corresponds to the illumination source used in SPRM-SDI. In (D), the propagation direction of the planar wave rotates from 0° to 360°, similar to the illumination source used in SPRM-ARI. The simulated area had dimensions of 6 μm by 6 μm.

In contrast, in SPRM-ARI, the propagation directions of the planar wave (φ) range from 0° to 360° at high speed to obtain the final image because of the use of ARI, so the intensity of the images captured using SPRM-ARI can be described using Eq. 5I(x,y)=φ=0φ=2π|U(x,y,φ)|2dφ(5)

The calculated image is shown in Fig. 3D, in which the parabolic tails have been averaged or removed, and only a single bright spot, similar to the Airy spot for conventional microscopy, is obtained. These simulated images verify that the spatial resolution of SPRM can be greatly enhanced by the introduction of ARI. A more rigorous calculation of the SPRM-SDI and SPRM-ARI images based on the theory reported in (23, 24) was also performed as shown in the Supplementary Materials, which gives out the same conclusion. ARI will induce the incidence angle scanning and azimuthal averaging and then enhance the spatial resolution of the SPRM.

Replacing the silver film with a dielectric multilayer to develop BSWM

As described earlier, a thin metal film must be used as the substrate for the objects under study (such as cells) in SPRM-SDI and SPRM-ARI. It is well known in cell biology that the techniques used to culture cells on glass substrates are more mature and generally applicable than those for growth on metal films. On the basis of our experiences in surface wave–related experiments, a thin metal film is not as stable or robust as a dielectric film, it will be damaged a little after each use and cannot be used for many times. Therefore, in the following experiment, we will demonstrate that the metal film can be replaced with a dielectric multilayer (in which the top layer is SiO2, i.e., the same material as glass) to develop a new type of label-free surface-sensitive photonic microscopy: BSWM. BSWs can be excited using this dielectric multilayer structure. This dielectric multilayer is much stable and can be used for many times in experiments. Similar to SPs, BSWs are also electromagnetic surface waves and are excited at the interface between the dielectric multilayer and the surrounding medium (31). These BSWs can be regarded as the dielectric analog of SPs, which means that they are also highly sensitive to their environment and can be used to develop a label-free imaging method (32, 33).

In this work, we present a BSWM technique with both SDI (BSWM-SDI) and ARI (BSWM-ARI) for the first time. Different from the BSWM reported in (34), which was developed to image the propagation characters (amplitude and phase ) of the excited BSWs, our BSWM technique uses the BSWs as the optical source to illuminate the objects (such as the nanosphere or nanowires) placed on the dielectric multilayer. By collecting the scattering signals from these objects, we will realize the label-free surface-sensitive imaging of these objects, but not imaging of the BSWs themselves. In our experimental setup, only one objective was used for both the excitation of the BSWs and the following imaging of the objects. For comparison with the SPRM, we used the same incident beam for the BSWM experiments. The setup is the same as that used for SPRM (Fig. 1A). The only difference here is that the silver film has been replaced by the all-dielectric multilayer, as shown in Fig. 1D. This dielectric multilayer sustains S-polarized BSWs at a wavelength of 635 nm (fig. S1C). The objects to be imaged are a polymeric nanowire on a polymeric microwire (Fig. 4, A to C) and a winding nanowire (Fig. 4, D to F). There are many fringes on the images (Fig. 4, B and D) that were acquired using BSWM-SDI, and these fringes obscure the images of the wires. In contrast, the wires are seen very distinctly on the images (Fig. 4, C and F) that were acquired using BSWM-ARI. Different from the bright-field image shown in Fig. 4A, in Fig. 4C, the nanowire is invisible in the areas where it crosses the microwire because when the nanowire is on the microwire and is far away from the dielectric multilayer surface; this section of the nanowire cannot be imaged using BSWs. The phenomena verify that, in a manner same as the SPRM images, BSWM images also focus on the targets within the evanescent field (100 to 200 nm) only, rather than those contained in the entire sample, which made it ideally suited to analyze the localization and dynamics of molecules and events occurring near the plasma membrane. Our results also verify that the ARI can enhance the spatial resolution of the BSWM.

Fig. 4 Images acquired using BSWM-SDI and BSWM-ARI working with S-polarized BSWs.

The dielectric multilayer operating as the substrate (Fig. 1D) sustains S-polarized BSWs at a wavelength of 635 nm. (A and D) Bright-field images. (B and E) Corresponding images acquired using BSWM-SDI. (C and F) Corresponding images acquired using BSWM-ARI. In (A) to (C), a polymeric nanowire was located on a polymeric microwire. These two straight wires were on the dielectric multilayer. In (D) to (F), a winding polymeric nanowire was placed on the dielectric multilayer. The scale bars shown on (A) and (D) are applicable to (B) and (C) and (E) (F), respectively. The angle of incidence was fixed at θ = 47°, at which the BSWs can be excited. Different from the bright-field image (A), in the BSWM-ARI image (C), the nanowire is invisible in the areas where it crosses the microwire (labeled with the oval), showing the surface-sensitive imaging capability of the BSWM-ARI.

Differences of the BSWM-ARI from the SPRM-ARI

In this section, we provide detailed descriptions on the differences of the BSWM-ARI from the SPRM-ARI, which are both wide-field label-free surface-sensitive photonic microscopy methods of high spatial resolution. First, as shown in Fig. 1, the setups used for the SPRM and BSWM are the same, and the only real difference is the material used for the substrate. In SPRM, a thin metal film such as a gold or silver film was used. In BSWM, an all-dielectric multilayer structure was used, where the top layer is SiO2, that is, the same material as the commonly used glass. Glass substrates are commonly known to be used almost universally for applications in cell biology or biomedical optics, rather than as metal substrates. The technique used for surface functionalization of the glass (SiO2) is both generally applicable and mature. Metal films may sometimes alter the conformation of the objects to be imaged or affect the measurements because of the existence of metal atoms on the substrate. Metal films may also present corrosion and oxidation rapidly in certain liquid or gaseous environments, which will certainly affect the surface roughness of the thin metal film and thus affect the image quality of the SPRM. However, metal films are conductive, so SPRM can be used to image local electrochemical currents (16). This type of imaging can also be realized if the dielectric multilayer is coated using a conductive dielectric layer such as indium tin oxide.

The second difference is that SPs can only be sustained in P polarization. BSWs can be sustained in both P polarization (Fig. 5) and S polarization (Fig. 4), depending on the thickness and the refractive index of each of the dielectric layers (35). In our experiments, another dielectric multilayer (with structural parameters that are shown in Fig. 5A) was used as the substrate for BSWM-ARI, where P-polarized BSWs can be excited at the interface between the air and the top SiO2 layer (35). The resonant dip in the angle-dependent reflection curve (Fig. 5B) and the dark arcs in the reflected BFP image of this dielectric multilayer (Fig. 5C) verified the existence of the P-polarized BSWs (22). When a single polymeric nanosphere with a diameter of 50 nm was placed on this dielectric multilayer (Fig. 5D), it could be imaged and detected, as shown in the BSWM-ARI image in Fig. 5E, meaning that BSWM can also work in the P-polarized mode.

Fig. 5 BSWM-ARI operating with P-polarized BSWs.

(A) The dielectric multilayer operating as the substrate for BSWM. The thickness of each layer is as labeled in the illustrations. The refractive index of the Si3N4 is 2.65 + i0.005, which is higher than that of the Si3N4 used in Fig. 1D. (B) Angle-dependent excitation spectrum. ne is the refractive index of the environment. When ne changes from 1.0 to 1.1, the resonant dip in the P-polarized BSWs will shift, demonstrating the sensing ability of the BSWs. (C) Reflected BFP image of the dielectric multilayer, in which the pair of dark arcs verifies the excitation of the P-polarized BSWs. The double-headed arrow line on (C) indicates the orientation of the incident polarization. A nanosphere was placed on this dielectric multilayer, and (D) shows the bright-field images. (E) Corresponding image acquired using BSWM-ARI. The scale bar shown on (D) is also applicable to (E). The angle of incidence is fixed at approximately θ = 43° for excitation of the P-polarized BSWs, and the incident wavelength is 635 nm.

Third, the incident wavelength and angle requirements of SPRM and BSWM are different. In SPRM, SPs can be sustained at the metal-air (or metal-water) interface at wavelengths ranging from UV to the near-infrared light band. However, because of the limited width of the PBG for any given dielectric multilayer, BSWs can only be sustained within a limited range of wavelengths (21). As described earlier, both SPRM and BSWM are based on the use of an oil-immersed objective, which is used for the excitation of the SPs and BSWs, in a manner similar to the well-known Kretschmann configuration (20). The NA of the objective is limited, and the most commonly used value is NA = 1.49, which corresponds to a collection angle of up to 80° [arcsine (1.49/1.515) = 80°]. For SP excitation, a shorter incident wavelength will result in a larger SPR angle. For example, at a gold-water interface, the SPR angle at a wavelength of 633 nm is approximately 72°, but this SPR angle will increase to 81° when the incident wavelength is reduced to 561 nm. At shorter wavelengths, SPs cannot be excited using this objective, and thus, the SPRM will not work. Most reported SPRMs work at longer wavelengths, e.g., 633 or 671 nm (7, 8, 14, 16, 18). In contrast, the appropriate design of the PBG of the dielectric multilayer will allow BSWs to be excited on the dielectric multilayer at both short and long wavelengths, and the BSW resonance angle can be shifted to a small angle because there are many potential material choices for the all-dielectric multilayer and the thickness of each of these layers can be controlled precisely. For example, BSWs can even be excited in the UV band with small angles of incidence (36). BSWM therefore provides greater choice of working wavelengths, which can be used to enhance the sensitivity of the photonic microscopy process. For example, if the biological sample to be imaged has an absorption peak at a short wavelength, e.g., at 532 nm, then the surface-sensitive photonic microscopy operating at this wavelength will be more sensitive to changes in the sample than the microscopies working at other wavelengths. Furthermore, light beams at shorter wavelengths are favorable for fluorescence excitation applications because many of the fluorophores used in cell biology can only be excited at short wavelengths. The experimental setup used for BSWM is the same as that used for TIRFM, except for the substrate, so BSWM can easily be switched to TIRFM after inclusion of a fluorescence filter when an illumination beam at a shorter wavelength is used. This means that both label-free and label-based surface-sensitive measurements can be performed using the same setup, which is useful for on-site observations.

In our experiments, a third dielectric multilayer, as shown in Fig. 6A, was used as the substrate for BSWM and could sustain S-polarized BSWs in water when the incident wavelength was only 532 nm. To the best of our knowledge, this wavelength is shorter than all previously reported wavelengths used in SPRM and also verifies that BSWM can work at short wavelengths and in liquid environments. The black arcs shown on the reflected BFP image of this multilayer (Fig. 6B) demonstrate that BSWs can be excited at the interface between the water and the top SiO2 layer (22). When a nanosphere was placed on this dielectric multilayer, it could also be imaged or detected on both the BSWM-SDI and BSWM-ARI images (Fig. 6, D and E). The sizes of the bright spots on the images captured via BSWM-ARI were also investigated, as displayed in fig. S4. Figure S4 shows that the spot sizes in Fig. 5E and Fig. 6E are approximately 350 and 380 nm, respectively. These spot sizes are similar to that shown in Fig. 2C, thus demonstrating that the spatial resolution of BSWM-ARI is similar to that of SPRM-ARI. While the spatial resolution for both BSWM-ARI and SPRM-ARI is limited by diffraction, these methods offer the merits of wide-field and label-free imaging capabilities.

Fig. 6 BSWM operating in short wavelength and liquid environments.

(A) Dielectric multilayer PBG structure operating as the substrate for BSWM. The upper space is filled with water, and the incident wavelength is 532 nm. The thickness of each layer is as labeled in the illustrations. (B) Reflected BFP image of the dielectric multilayer, in which the pair of dark arcs verifies the excitation of BSWs at the interface between the water and the top SiO2 layer. The double-headed arrow line on (B) represents the orientation of the incident polarization. A nanosphere was placed on this dielectric multilayer. (C) Bright-field image. (D and E) Corresponding images acquired using BSWM-SDI and BSWM-ARI, respectively. The scale bar on (C) is applicable to (D) and (E). The angle of incidence was fixed at approximately θ = 66° for excitation of the BSWs in the water.

As reported in (21), the excitation angle (θ) for the BSWs can be tuned with the structure dimensions, especially the top dielectric layer. The penetration depth of the BSWs into the top space can be described using a function of angle (θ), λ0/4π*(n0*sin θ)2(ne)2), where λ0 is the wavelength of the incident light in vacuum, n0 is refractive index of the substrate, and ne is the refractive index of the medium on the upper space, e.g., air or water. Then, the imaged section of the surface-sensitive photonic microscopy is determined by the penetration depth of the BSWs. So the imaged section of BSWM-ARI can also be tuned, as demonstrated earlier. For the three dielectric multilayers used here, the excitation angles (θ) of BSWs are 47°, 43°, and 66°, corresponding to the penetration depths, 105, 194, and 110 nm, respectively, of the evanescent fields, which will be useful for observing the cells in different depths.


Both SPs and BSWs are well-known phenomena that have been exploited for sensitive detection of biomolecular kinetics on surfaces. Extending these schemes to microscopy has presented some resolution issues due to optical wave coupling to either the SPs or the BSWs, both of which, with point coupling, lead to sets of trailing interference patterns. Although it is plausible that some of those imaging artefacts can be removed by proper software computations, we have performed this task experimentally by using the ARI (3739), thus greatly improving the spatial resolution of the images. These high-spatial-resolution images are obtained without image processing and without mechanical movement of the samples for differential processing, as would be used in SPRM-SDI; this means that imaging speed can be increased, which is helpful during real-time observation. The mechanism of the spatial resolution enhancement by using ARI is analyzed with a simple two-dimensional model.

We have also used an all-dielectric multilayer to replace the commonly used thin metal film substrates and have developed a new type of surface-sensitive photonic microscopy called BSWM. BSWM is both versatile and generally applicable. As demonstrated in this study, in addition to long wavelength operation (e.g., at 635 nm, which is commonly used in SPRM), BSWM can also operate at shorter wavelengths (e.g., 532 nm here), which cannot be realized in SPRM. The optical properties of the BSWs, such as their polarization states and penetration depths (corresponding to the thickness of the imaged section of the microscopy), can be tuned precisely via the thicknesses of the materials used for the dielectric layers. For example, BSWM operation in both P- and S-polarized modes is demonstrated in our work, and the thickness of the imaged section for BSWM has also been tuned from 100 to 200 nm using the structural dimensions. In contrast, the polarization of SPs is fixed at the P-polarized mode, and the thickness of the imaged section of the SPRM cannot be tuned via the thickness of the metal film.

Because they are sensitive to environmental changes on surfaces, both SPs and BSWs have been used to develop the high-performance sensors by monitoring the angular shift of the reflected light from the metal film or dielectric multilayer (10, 11, 32, 33). As far as the sensing function is concerned, if an additional CMOS or a quadrant diode was included in our experimental setup to record the optical signals on the BFP of the objective, the angular shift of the reflected light from the metal film or dielectric multilayer can be also monitored (40), then our setup can also work as a sensing instrument. It is a sensor based on an objective but not a bulk prism. Thus, we anticipate that the surface-sensitive photonic microscopy method with ARI will have a broad range of applications in both surface imaging and sensing, particularly for the analysis of the localization and dynamics of molecules and events located at or near cells because of its wide-field, label-free, and high-sensitivity surface imaging capabilities.


Experimental setup

A schematic of the setup, which is based on a commercial inverted microscopy setup, is shown in Fig. 1A. The laser beam at the wavelength of 635 nm comes from a commercial laser diode. The scanning galvanometer was from Thorlabs Inc., and its scanning speed can be controlled via software. The objective (100×, NA of 1.49) was from Nikon, Japan. The Neo sCMOS detector was from Andor Oxford Instruments (UK).

Sample preparation

The first dielectric multilayers were fabricated by plasma-enhanced chemical vapor deposition (PlasmaPro System 100, Oxford) of SiO2 and Si3N4 on a standard microscope cover glass (0.17 mm thick) at a vacuum pressure of <0.1 mtorr and at a temperature of 300°C. In this case, SiO2 was the low (L) refractive index dielectric, and Si3N4 was the high (H) refractive index dielectric. These dielectric layers had thicknesses of 105 and 88 nm, respectively. A total of 14 dielectric layers were deposited on top of each other. The thickness of the top SiO2 layer was approximately 240 nm. Another two dielectric multilayer structures were fabricated in the same way.

The microwires and nanowires were fabricated using the electrospinning technique. The nanowires were electrospun from a 2.0-ml solution of formic acid that contained 0.6 g of Nylon 6. The microwires were electrospun from a 1.7-ml tetrahydrofuran solution with 0.5 g of hydrogel D4 (HydroMed D4, a commercially available polyurethane-based hydrogel). Both solutions were stocked in a 2-ml disposable syringe. A voltage of 10 kV was applied to a stainless steel needle with a flat tip from a high-voltage power supply. The feed rate was maintained using a syringe pump. The feed rate was 0.1 mm min−1, and the internal diameter of the needle used for fabrication of the nanowires was 0.19 mm. For the microwires, the feed rate was 0.2 mm min−1, and the internal diameter of the needle was 0.50 mm. These wires were collected at a distance of 10 cm from the tip of the needle and then transferred to the substrates. The microwire was transferred to the substrate first and was followed by the nanowire, so that the nanowire was on top of the microwire.

The nonfluorescent polystyrene nanosphere (3050A), which has a diameter of approximately 46 ± 2 nm, was bought from Thermo Fisher Scientific (, and the attachment of these spheres to the dielectric multilayer and the Ag film was realized using the MET One Aerosol Particle Generator 255 Gen.


Supplementary material for this article is available at

Fig. S1. Surface waves sustained by Ag film or dielectric multilayer.

Fig. S2. Angle-dependent excitation spectra and field characteristics for the SPs and BSWs.

Fig. S3. SEM images of the polymeric nanowire and microwire and the nanospheres.

Fig. S4. Sizes of the bright spots on images captured using BSWM-ARI.

Fig. S5. The model used for a more rigorous calculation.

Fig. S6. Simulated SPRM-SDI and SPRM-ARI images based on a rigorous calculation.

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 by NSFC (91544218, 61427818, 11774330, 21574120, and 21774115), MOST (2016YFA0200601 and 2013CBA01703), the Science and Technological Fund of Anhui Province for Outstanding Youth (1608085J02 and 1608085J01), the Longshan Academic Talent Research Supporting program of SWUST (17LZX626), and the Fundamental Research Funds for the Central Universities (WK2340000084 and WK2030380008), Anhui Provincial Science and Technology Major Projects (18030901005), Anhui Initiative in Quantum Information Technologies, the foundation of Key Laboratory of Environmental Optics and Technology of Chinese Academy of Sciences (grant no. 2005DP173065-2019-XX). This work was also supported by grants from the National Institute of Health (GM125976, GM129561, EB018959, and S10OD019975). The work was partially carried out at the University of Science and Technology of China’s Centre for Micro and Nanoscale Research and Fabrication. D.Z. acknowledges Q. Zhan of the University of Dayton, X. Liu and C. Qian of USTC for the helpful discussions. Author contributions: D.Z. conceived the ideas, supervised the work, and drafted the manuscript. Y.K. carried out the optical experiments. J.C. and X.T. performed the numerical calculations. All authors discussed the results and contributed to the writing and editing of 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