1. Introduction

Solid immersion lenses (SILs) can be used in optical microscopy to improve spatial resolution [1]. In this technique, the gap between the sample and the objective lens is filled by solid material with a high refractive index n, which increases the numerical aperture (NA). For the combination of a traditional hemispherical SIL (HSIL), i.e. a hemispherical lens with a thickness equal to its radius, and a conventional objective lens, the NA can be increased by a factor of n [1] compared to that of the conventional objective lens alone; while, for the combination of an aplanatic SIL (ASIL), i.e. a truncated ball lens of thickness r + r/n where r is the radius, and a conventional objective lens, it can be increased by a factor of n² up to a maximum value of n [2]. Thus generally speaking the application of ASILs in an optical microscope can achieve higher NA and resolution than that of HSILs. Besides, in contrast to HSILs, ASILs can provide a larger effective field of view (FOV) taking into account the tolerance of geometrical aberrations induced by the SILs [3]; therefore in our opinion ASILs are more preferable than HSILs in the development of wide-field high-resolution optical microscopy.

Previous studies demonstrated experimentally that wide-field high-resolution fluorescence microscopy with an effective NA of 1.85 could be achieved by applying an ASIL made of high-index optical glass [4]. Fluorescence microscopy using an ASIL not only represents a low-cost technique for achieving high resolution on a real-time and wide-field basis, but it also works as a universal platform for other techniques in need of a large NA. Based on this idea, the combination of solid immersion fluorescence (SIF) microscopy and structured illumination microscopy (SIM) [5–7] was investigated [8]. Theoretically, the combination could increase the NA by a factor of 2n² subject to a maximum value of 2n. In [8], it was demonstrated that the combination can produce a wide-field high-resolution microscopic system with a bandwidth corresponding to an NA of 3.

Another promising case that may take advantage of high NA values in SIF microscopy is objective-based total internal reflection fluorescence (TIRF) microscopy [9]. TIRF microscopy is a wide-field high-resolution fluorescence microscopy used to image the surface of cells by utilizing a thin field of illumination. This illumination is generated by shining a tilted beam, commonly from a laser source, to the interface between the cells and the substrate, which creates a shallow evanescent field that excites fluorophores on the cell surface. TIRF microscopy is widely used in the study of cell-substrate contacts [10], membrane dynamics [11], single molecule imaging [12] and endocytic and exocytotic events [13]. TIRF is often favored as a way of imaging a very shallow depth from the cell-substrate surface in order to gain high sensitivity or/and high selectivity. In objective-based TIRF microscopy the imaging depth is regulated by the NA of the objective lens. As reported in [4], SIF microscopy is able to achieve an NA much higher than those of conventional oil immersion TIRF microscopy objective lenses, so it is promising to explore TIRF microscopy based on a SIF microscope platform.

In this paper, we report a new approach combining high NA SIF microscopy and TIRF microscopy, which we refer to as solid immersion total internal reflection fluorescence (SITIRF) microscopy, and the highly confined surface imaging capability in this new microscopic technique.

2. Theory

TIRF microscopy employs total internal reflection (TIR) illumination, as shown in Fig. 1 , to selectively illuminate and excite fluorophores in a thin optical plane close to the interface between the two materials. TIR occurs at an interface formed by two media with different indices of refraction. According to Snell’s law, for light propagating from an optically denser (refractive index: n₁) to a less-dense (refractive index: n₂) medium, i.e., n₁> n₂, if the incident angle of the beam is larger than the critical angle θ_c, which is defined by:

 

Fig. 1 TIR illumination scheme. When the incident angle of the incoming wave is larger than the critical angle θ_c, the wave is totally internally reflected. The evanescent wave propagates in the x direction and decays exponentially in the low-index (n₂) material in the z direction.

Download Full Size | PDF

From this equation we can see that, in order to confine penetration depth to a small amount at the interface, it is essential to have an illumination NA that is significantly greater than the refractive index of the medium surrounding the sample, i.e. NA_ill >> n₂.

In a practical TIRF microscope used in cell imaging, the most common medium around cell samples is water, with an index of 1.33. In theory any objective lens providing a NA higher than this value is capable of generating TIR; however, using objective lenses with NAs just above 1.33 by a small margin will produce TIR at an angle insufficiently larger than the critical angle, and the evanescent field in this case will be deep. Typical TIRF microscopy objective lenses made by the major manufacturers have the NAs in the range of 1.45-1.49, and the highest NA available in market is 1.65 [15] but unfortunately that objective requires the use of expensive n = 1.78 coverslips and special n = 1.78 coupling oil which is volatile and leaves a crystalline residue. Recently it was demonstrated that SIF microscopy employing an ASIL can provide an illumination NA as high as 2 [4], and ASIL-based fluorescence microscopy is generally adaptable to other types of wide-field high-resolution fluorescence microscopy. If the TIR illumination scheme is used in ASIL-based fluorescence microscopy, highly confined surface imaging is potentially achievable owing to the high NA performance in SIF microscopy.

As reported in [4] the NA in SIF microscopy employing an ASIL is effectively determined by the index of the ASIL material, and the illumination NA of 2 was demonstrated by applying an ASIL made of high-index optical glass (S-LAH79) that has a refractive index of n_d = 2.003 and an Abbe number of ν_d = 28.27. The major issue in the application of this ASIL is the appearance of severe chromatic aberration (CA) due to the strong dispersion in the optical glass, as the low Abbe number indicated. In practice a very narrow band emission filter, with a 3 nm spectral bandwidth (full width at half maximum, FWHM), was used to reduce CA to a negligible level and maintain high-resolution performance. However the narrow bandwidth of this filter hindered the fluorescence imaging because of the lack of detectable emitted photons. It is desirable to find optical materials with less dispersion, i.e. a higher Abbe number, to fabricate ASILs. In our current research, zirconium dioxide (ZrO₂), a type of crystalline material with n_d = 2.215 and ν_d = 51.11, was used to fabricate ASILs. In contrast to S-LAH79 optical glass, ZrO₂ features both higher index and lower dispersion, which could offer many benefits, including (a) the elimination of CA, which allows the use of wide band emission filters, with a consequent improvement in the efficiency of fluorescence detection; (b) a higher index, leading to a higher NA and gains in imaging resolution, as the illumination and imaging beams share the same objective lens in an objective-based TIRF microscope; (c) a higher illumination NA, which creates a shallower evanescent field.

In a SITIRF microscope employing an ASIL made of ZrO₂ the highest achievable illumination NA equals the index of ZrO₂, which is 2.230 at a wavelength of 473 nm. The solid blue curve depicted in Fig. 2 shows how the intensity decays along the z direction from the ASIL-air interface, according to Eq. (2). The dashed blue curve in the same figure displays the intensity decay from an illumination NA of 1.45, a characteristic NA value in conventional TIRF microscopes. The decay of the evanescent wave in the SITIRF microscope is more pronounced than in a conventional TIRF microscope, showing that a highly confined excitation depth is theoretically obtainable in SITIRF microscopy. Quantitatively from Eq. (4) we know that the penetration depths are 19 nm and 36 nm in the SITIRF and conventional TIRF microscopes, respectively, which are marked by the circles in Fig. 2. The intensity decays happening at an ASIL-water interface with different illumination NAs are depicted by red curves in Fig. 2, and in this case the penetration depths are 21 nm and 65 nm in the SITIRF and conventional TIRF microscopes, respectively. From these curves we know SITIRF microscopy can perform highly confined evanescent field illumination at similar level with either air medium or water medium, but conventional TIRF microscopy shows an obviously different penetration depth with different media. Therefore we can conclude SITIRF microscopy has a more general use when implementing surface imaging, and for this reason the following experiments were carried out in air.

 

Fig. 2 Simulated intensity decays along the z direction from the ASIL-air (blue curves) and ASIL-water (red curves) interfaces with illumination NA values of 2.2 and 1.45. Theoretical penetration depths are also marked by circles in each case. An illumination wavelength of 473 nm is considered.

Download Full Size | PDF

To investigate the CA performance of ZrO₂ ASILs and to guide the choices of emission filters that can be used in practice, we used the optical design software ZEMAX (ZEMAX Development Corporation) to simulate the CA of 3 mm diameter ASILs made of ZrO₂ at a central wavelength of 473 nm . The on-axis spot diagrams with an Airy disk reference from the simulations are shown in Fig. 3 . The spot diagrams in Fig. 3(a) show that with a 10 nm imaging spectral width, i.e. from 468 nm to 478 nm, the geometric point spread function (PSF) at all wavelengths is well within the Airy disk when the central wavelength is focused at the paraxial focal point, indicating diffraction-limited resolution. If the imaging spectral width is extended to 20 nm, i.e. from 463 nm to 483 nm, the resolution is still close to diffraction-limited performance (Fig. 3(b)). Our simulations also show that the nearly diffraction-limited imaging performance can be maintained within a 30 μm diameter FOV. According to these simulation results, we chose to use emission filters with spectral bandwidths no wider than 20 nm in our experiments in order to achieve high-resolution imaging.

 

Fig. 3 Spot diagrams showing PSF with Airy disk reference at the central wavelength (black circle) of 473 nm for an isotropic point source at the aplanatic point of a 3 mm ZrO₂ ASIL. (a) The PSF for ASIL imaging for three wavelengths: 468 (blue crosses), 478 nm (red squares) and 473 nm (green crosses) which is focused at the paraxial focal point and therefore forms a diffraction-limited image corresponding to the Airy disk. (b) The PSF for ASIL imaging for three wavelengths: 463 (blue crosses), 483 nm (red squares) and 473 nm (green dots) which is focused at the paraxial focal point and therefore forms a nearly diffraction-limited image corresponding to the Airy disk.

Download Full Size | PDF

3. Experimental methods

The optical arrangement of SITIRF microscopy is shown in Fig. 4 . The fundamental construction of this system is similar to the SIF microscope system described in [4], in which an epi-fluorescence configuration was applied, while to manipulate the incident illumination angle a mirror and a steering lens, as enclosed by the dashed box in Fig. 4, were mounted on a motorized translation stage (PT1/M-Z8, Thorlabs) to allow radial positioning of the illumination focusing spot on the back focal plane (BFP). When the focusing spot was on-axis, a conventional Köhler illumination was formed; once the focusing spot was off-axis to produce a tilted but collimated illumination above the critical angle, a TIR illumination was generated. This configuration allowed us not only to conveniently transform the system between an epi-fluorescence microscope and a SITIRF microscope, but also to exert fine control over the illumination NA because the radial position of the focused beam on the BFP has a linear relationship with the illumination NA [4]:

 

Fig. 4 Schematic of the SITIRF microscope system. A mirror and a steering lens were mounted on a motorized translation stage to change the incident illumination angle. The dark blue/red shading illustrates the Köhler illumination beam from either of the two lasers, the light blue shading illustrates LED illumination beam, the green shading indicates the fluorescent light forming the specimen images, and the red dashed lines represent the fluorescent light forming the BFP images.

Download Full Size | PDF

Here, f is the focal length of the conventional objective lens and n is the index of ASIL material, so we can conclude the effective focal length in our case was about 0.8 mm.

In this system, a 0.55 NA long working distance objective lens (50 × Plan Apo Infinity-Corrected Long WD Objective, Mitutoyo, focal length 4 mm) and a 3 mm diameter ASIL made of ZrO₂ were used to form an ASIL-objective. The ASIL was firstly purchased from Knight Optical Ltd as custom made ZrO₂ ball lenses, and then grinded and polished to the aplanatic configuration. The distance between the convex surface of the ASIL and the front surface of the Mitutoyo objective was 8.2 mm and the medium between them was air. The illumination NA of the ASIL-objective was determined by the index of the ASIL material. In our case, a 473 nm diode pumped solid state (DPSS) laser (MBL-III-473, Laser2000) and a 632.8 nm Helium-Neon (He-Ne) laser (31-2140-000, Coherent) were available for excitation; the indices of ZrO₂ at these wavelengths are 2.230 and 2.208, respectively, so we conclude that the maximum available illumination NA was about 2.2. For bright-field imaging, an LED (LXHL-LE5C, Lumileds) with a central wavelength of 505 nm was also used as an additional light source. A multiband dichroic filter (Z470/635rpc, Chroma) was employed to reflect either of the laser beams into the objective, while it was also used as a 50/50 beam splitter for the bright-field LED illumination. A group of emission filters was mounted on a filter wheel for easy selection of imaging wavelength. In order to maintain nearly diffraction-limited imaging and to conform to the simulation results (Fig. 3) all the emission filters had a spectral bandwidth of less than 20 nm. The imaging path was divided by a beam splitter to enable the simultaneous capture of sample images and monitoring of BFP distribution, recorded by a CCD camera (ORCA-ERG, Hamamatsu) and a CMOS camera (DCC1545M, Thorlabs), respectively. The FOVs for sample and BFP imaging were approximately 29 × 22 µm and 10 × 8 mm.

The ASIL was firstly cleaned by immersing it in Acetone and sonicating it for 15 minutes, and then cleaned by immersing it in Piranha solution for 10 minutes. To enhance cell adhesion, the flat surface of the ASIL was coated by Poly-L-Lysine. After cleaning the ASIL was glued on a brass holder and then mounted on a kinematic mount in the system. Once the ASIL was finely aligned the samples could be directly drop cast on its flat surface for imaging.

4. Experiments and results

As mentioned previously, the SITIRF system could be conveniently switched between epi-fluorescence mode and TIRF modes, and the illumination NA in TIRF mode could be finely controlled by the motorized translation stage, so samples could be imaged in three different modes: (a) epi-fluorescence, (b) TIRF with an illumination NA of 1.45 typical of commercial TIRF microscopes, designated conventional NA TIRF mode, and (c) TIRF with an illumination NA of 2.2, designated high NA TIRF mode. The first two modes were used to mimic the performance of an epi-fluorescence microscope and a conventional TIRF microscope, respectively. Comparison of images captured in these three different modes revealed the effective imaging depths in epi-fluorescence, conventional TIRF and SITIRF microscopes. The goal of this work is to demonstrate that the SITIRF system working in high NA TIRF mode is able to provide an imaging depth shallower than that in a conventional TIRF microscope.

It is worth noting that the penetration depth does not represent an absolute limit on the depth of features in a sample that will be observed [16]. In other words, the imaging depth may be deeper than the penetration depth. The penetration depth is artificially defined as the distance at which the intensity of the evanescent field decays to 1/e (corresponding to 37%) of its value at the interface. In addition, the exponential decay of the evanescent field is an oversimplification; precise calibration shows that approximately 90% of the evanescent field is represented by an exponential with a decay consistent with theory whereas the remaining 10% of the field is represented by an exponential with a much longer decay constant and is identified as scattering [17]. The source of the scattered light is mainly from multiple reflections at various interfaces and scattering in elements in the objective lens. Moreover fluorescence emitted as a result of excitation by the evanescent field propagates beyond the theoretical penetration depth and elicits further signals from fluorophores, giving an increased imaging depth. In terms of genuine cell samples the local refractive indices of intracellular components are more complicated and heterogeneous; therefore the actual evanescent field may be significantly affected and stronger scattering may occur. For the SITIRF system working in high NA TIRF mode, the higher emission collection efficiency for proximal fluorophores [18] increase the steepness of the decay from pure single exponential function, which is beneficial for the surface confinement in SITIRF microscopy.

1 µm Nile red (excitation/emission peaks: 535/575 nm) fluorescent beads (FluoSphere F-8819, Invitrogen) were used to examine the highly confined surface imaging of the SITIRF microscope. The excitation source was a DPSS laser at 473 nm, and the emission filter was centered at 570 nm with 10 nm spectral bandwidth (570/10 nm, FB570-10, Thorlabs). The images captured in different modes are displayed in Fig. 5 , and the 3D surface plots of a single bead, as marked by the dashed box, are also shown on the right-hand panel. In epi-fluorescence mode the images of these beads, as displayed in Fig. 5(a), 5(d), show a gradually declining intensity distribution from the center to the edge in a 1 µm diameter area. However, in conventional NA TIRF mode (Fig. 5(b), 5(e)) there is an intensity peak on one side where the TIR illumination beam is incident on the fluorescent bead. In this case the 1 µm beads converted evanescent light into propagating comet-shaped light due to relatively deep penetration depth of the evanescent wave [19] and morphology-dependent scattering [20]. The similarity of the asymmetric fluorescent emission in Fig. 5(b) and 5(e) with images obtained using an Olympus 60 × 1.45 NA oil immersion objective lens [17] confirm that the SITIRF microscope in conventional NA TIRF mode faithfully imitates a conventional TIRF microscope. Figure 5(e), 5(f) shows that in high NA TIRF mode the penetration depth was significantly reduced, theoretically to 19 nm at an ASIL-air interface, so that only a small area where the bead contacted the substrate was illuminated, and then the disruption on the evanescent field was eliminated. As a consequence of this confined illumination, a relatively intensely fluorescent area can be seen at the center of the bead image (Fig. 5(e)) which we regard as the effective imaging area in high NA TIRF mode. Thus the shallower imaging depth in high NA TIRF mode results in a smaller effective imaging area than for epi-fluorescence or conventional NA TIRF modes. From the images shown in Fig. 5 it is clear that the SITIRF microscope is able to provide a stronger confinement of imaging depth than that in either the epi-fluorescence microscope or the conventional TIRF microscope, even on a sample with densely concentrated fluorophores. In Fig. 5(c), 5(f) the residual fluorescence emission in regions outside the effective imaging area arises mainly from excitation by the scattering light at subcritical angles. The amount of this scattering-induced emission remained approximately constant for different illumination NAs as illustrated in Fig. 5, which again is in agreement with the report in [17]. The effective imaging depth in high NA TIRF mode can be quantified, based on the geometric model of a fluorescent bead shown in Fig. 6 . The imaging depth $\overline{AB}$can be calculated by:

 

Fig. 5 Images of 1 µm diameter fluorescent beads in (a) epi-fluorescence mode, (b) conventional NA TIRF mode with an illumination NA of 1.45, and (c) high NA TIRF mode with an illumination NA of 2.2. 3D surface plots (d, e, f) of a single bead, as indicated by the dashed box, in different modes are shown on the right-hand panel.

Download Full Size | PDF

 

Fig. 6 Geometric model of a fluorescent bead.

Download Full Size | PDF

In addition to fluorescent bead images, BFP images, as shown in Fig. 7 , were also captured using a 490 nm longpass filter (NT62-983, Edmund Optics) as an emission filter. The illumination NA can be measured from the BFP images as the location of each illumination focusing spot is clearly shown. The radial positions of the focused beam on the BFP in each case are indicated by the yellow leader lines in Fig. 7. According to Eq. (5) we can quantify the illumination NA in conventional NA TIRF mode and high NA TIRF mode as 1.45 and 2.2, respectively, which are consistent with the theoretical values.

 

Fig. 7 BFP images captured with 1 µm fluorescent bead samples in (a) epi-fluorescence mode, (b) conventional NA TIRF mode with illumination NA of 1.45, and (c) high NA TIRF mode with illumination NA of 2.2. The yellow dashed circles indicate the full NA of the system which is 2.2; the radial positions of the focused beam in each case are indicated by the yellow leader lines.

Download Full Size | PDF

Next, we imaged a mixture of 40 nm yellow-green (excitation/emission peak: 505/515 nm) fluorescent beads (FluoSphere F-8780, Invitrogen) and 1 µm Nile red fluorescent beads. In this case the excitation was 473 nm and a 510/10 nm emission filter (FB510-10, Thorlabs) was used. This filter efficiently transmitted the fluorescence wavelengths from the yellow-green beads while eliminating those from the Nile red beads, although a small amount of the Nile red fluorescence did leak through the filter due to the high fluorescence intensity. In Fig. 8 we show the images of the mixed samples in (a,e) bright-field mode, (b,f) epi-fluorescence mode, (c,g) conventional NA TIRF mode with illumination NA of 1.45, and (d,h) high NA TIRF mode with illumination NA of 2.2. The 40 nm fluorescent beads maintained a similar level of visibility in these images (except in the bright-field mode due to the weak scattering) because in each case they were very close to the ASIL surface and kept within different illumination depths. However only in high NA TIRF mode did the 1 µm bead became virtually undetectable (Fig. 8(d), 8(h)) because most of the bead was above the penetration depth. Theoretically the volume of the fluorophore in each 1 µm bead that can be detected in high NA mode was less than 28% of that in conventional NA mode, which made the beads nearly undetectable in high NA mode when the similar illumination power density was maintained.

 

Fig. 8 Images of a mixture of 40 nm and 1 µm fluorescent beads in (a, e) bright-field mode (b, f) epi-fluorescence mode, (c, g) conventional NA TIRF mode with an illumination NA of 1.45, and (d, h) high NA TIRF mode with an illumination NA of 2.2.

Download Full Size | PDF

In [4] it was demonstrated SIF microscopy is a type of wide-field high-resolution microscopy; accordingly, SITIRF microscopy, as a modification of SIF microscopy, should also perform high-resolution on a wide-field basis. In Fig. 8(e)–8(h) the fluorescent beads were located at the periphery of the FOV but there is no noticeable degradation of resolution, which proves the SITIRF is able to perform wide-field high-resolution imaging. Quantitatively the resolution performance of the high NA SITIRF system can be estimated from the FWHM of a two-dimensional Gaussian fit to the image of each 40 nm bead as these beads effectively acted as point emitters. The FWHM of a single 40 nm bead image from the average over 10 separate beads was measured as 174 ± 11.5 nm. Considering the actual size of these beads we may use a simple deconvolution method to estimate the PSF [21]:

In order to evaluate the ability of the SITIRF microscope to record images of biological samples confined to a few tens of nanometers from the lens surface we examined cells of a mutant of the photosynthetic cyanobacterium Synechocystis PCC6803. These bacteria not only have a cell envelope consisting of an outer membrane, a periplasmic space and a plasma membrane, but also have a true internal membrane system comprised of multiple layers of thylakoid membranes. In Synechocystis the thylakoid membranes, to which native photosynthetic phycocyanin pigments are located, are physically discontinuous from the plasma membrane [22, 23] and therefore the thylakoid membranes represent an independent intracellular compartment. In the mutant studied here the PetC3 Rieske FeS protein of the cytochrome b₆f complex was tagged with the green fluorescent protein (GFP). Previous work on this mutant using scanning confocal microscopy had shown that PetC3–GFP protein is mainly localized in the peripheral plasma membrane rather than to the internal thylakoid membrane [24]. Thus, a halo of fluorescence was observed in single cell images, emitted from the PetC3–GFP protein, and the internal parts of the cell showed emission from the phycocyanin pigments localized to the thylakoid membranes [24]. Images of PetC3–GFP Synechocystis cells, were recorded using the SITIRF microscope in different modes (Fig. 9 ), using 473 nm and 633 nm lasers to excite GFP and phycocyanin, respectively, and 512/18 nm and 655/15 nm emission filters (FF01-512/18-25, FF01-655/15-25, Semrock) to extract GFP and phycocyanin fluorescent emission signals. The epi-fluorescence images (Fig. 9(b)–9(d)) show that although signals from both GFP and phycocyanin can be separately recorded, there is little spatial information available. The GFP channel in conventional NA TIRF mode with an illumination NA of 1.45 shows that the PetC3–GFP protein is on the peripheral plasma membrane of the cells; the signal in the central regions of the cell (Fig. 9(e)) arises from the TIRF confinement which selects the signal from the peripheral plasma membrane closest to the ASIL surface. The extended evanescent field in conventional TIRF mode results in additional detection of a halo of GFP fluorescence from rising parts of the plasma membrane as the cell curves up and away from the surface (Fig. 9(e)). Furthermore, phycocyanin fluorescence from internal membranes is also seen, again because of the larger penetration depth in conventional TIRF mode. In contrast, the phycocyanin image obtained using the high NA SITIRF (Fig. 9(i)) has much lower amplitude and there is no halo of GFP fluorescence either (Fig. 9(h)). As a result, GFP signal dominates the merged image in Fig. 9(j) because the imaging depth was confined no deeper than the plasma membrane. The cell wall and outer membrane in Synechocystis cells are only 20 nm thick [25], so that quantitatively we can estimate the imaging depth in the high NA SITIRF microscope was confined to this level in this case.

 

Fig. 9 Images of Synechocystis cells expressing PetC3-GFP in (a) bright-field mode, epi-fluorescence mode in (b) GFP channel, (c) phycocyanin channel, and (d) by merging two channels; in conventional NA TIRF mode with illumination NA of 1.45 in (e) GFP channel, (f) phycocyanin channel, and (g) by merging two channels; and high NA TIRF mode with illumination NA of 2.2 in (h) GFP channel, (i)phycocyanin channel, and (j) by merging two channels. Scale bar: 2 µm.

Download Full Size | PDF

5. Conclusions and discussion

In this paper, we have investigated the application of total internal reflection illumination in wide-field high-resolution solid immersion fluorescence microscopy employing an aplanatic solid immersion lens. This microscopy, referred to as solid immersion total internal reflection fluorescence microscopy, is capable of performing highly confined surface imaging. The ASIL employed in our system was made of a modern crystalline material, ZrO₂, which has a higher refractive index and lower dispersion than that of conventional optical glasses. The application of high-index ZrO₂ ASIL allowed SITIRF microscopy to achieve higher resolution and shallower imaging depth. In addition, the low dispersion of the material removed the requirement for the narrow band emission filters normally needed to compensate for chromatic aberration, and enabled the use of emission filters with a wider spectral band. The novel design of the experimental SITIRF system enabled us to image the same sample in different illumination NAs, and to evaluate and compare the imaging depths in epi-fluorescence mode, conventional NA TIRF mode and high NA TIRF mode. We demonstrated that highly confined surface imaging can be performed by the SITIRF system in high NA TIRF mode by imaging 1 µm fluorescent beads, a mixture of 40 nm and 1 µm fluorescent beads and also by imaging a GFP-tagged mutant of the cyanobacterium Synechocystis PCC6803. We also demonstrated the SITIRF system can provide high lateral resolution imaging on a wide-field basis.

In contrast to conventional TIRF microscopy, the distinctive advantage of high NA SITIRF microscopy is that highly confined surface imaging enables highly selective imaging on cellular features closely adjacent to the substrate, which makes it more efficient to investigate the structures and dynamics of cell membranes. In addition, high NA SITIRF microscopy considerably eliminates background fluorescence by confining imaging within a very shallow depth, and as a result the resolution and contrast performance are enhanced compared to conventional TIRF microscopy, features that are particularly beneficial for single molecular imaging.

Although we studied highly confined surface imaging using SITIRF microscopy in this work, this technique has great potential for superior performance in many other situations in which conventional TIRF microscopy is applied. SITIRF microscopy can achieve a higher illumination NA, i.e. a larger incident angle of illumination, than that in conventional TIRF microscopy, a feature that leads to the possibility of observing very densely packed cellular components with evanescent wave illumination. For example, lysosomes in EMT6 cells have an index of 1.6 [26] and melanin in cuttlefish has an even higher index of 1.6-1.7 [27]. High NA SITIRF microcopy clearly has the potential to become a powerful tool for studies of the structural dynamics of these components. Another aspect in which SITIRF microscopy may have superior performance exists in axial resolution. The strong z-axis dependence of the excitation intensity in TIRF microscopy has been used to obtain sub-wavelength resolution in the axial direction; for example, TIRF microscopy was employed to observe the axial motion of single secretary granules with an axial resolution of 30 nm [28] and the complexes on the plasma membrane of living cells to 10 nm axial resolution [29]. SITIRF microscopy could deliver an even higher axial resolution owing to its higher restraint capability and a finer adjustability of the evanescent field, so these characteristics could be of value for enhancing the performance of variable-angle total internal reflection fluorescence microscopy (VA-TIRFM) [30] as well. The idea of combining high NA SILs with TIR illumination may also be applied in non-fluorescence microscopy for real-time imaging of live cellular events [31].

The major practical problems with the SITIRF microscope as it presently stands include the difficulty of mounting real biological samples on the ASIL, the complexity of ASIL cleaning and realignment procedure following each experiment, and the fixed position of effective FOV. We proposed several methods to solve these problems in [4], and future improvements of the SITIRF microscope include all the aspects discussed in this article in order to make it suitable for routine use.