1. Introduction

Fluorescence microscopy has become an indispensable tool in biological imaging due to its selectivity and sensitivity. However, even with bright and photostable markers at hand, auto-fluorescent cellular structures as well as non-ideal labeling efficiency make suppressing out-of-focus fluorescence challenging. This results in a serious bottleneck for crisp biological imaging [1]. A common concept for spatial selectivity and optical sectioning in fluorescence is based on total internal reflection fluorescence microscopy (TIRF-M) [2–4]. Another concept utilizing evanescent fields is based on Surface plasmon coupled emission (SPCE) [5–7]. Here, evanescent modes at the metal SPCE supporting nanocoated surface are manipulated. As a result, fluorescence can only be emitted through the coating within the narrow supercritical angle. The efficiency of SPCE depends on the fluorophore–nanocoating distance as well as its relative orientation [8, 9]. SPCE is well known to be able to discriminate close-to-surface signals from signals originating from higher distances [10–12], and surface plasmon resonant (SPR) illumination of plasmonic nanocoating has been shown to enhance the evanescent TIRF field excitation [13–15]. As demonstrated by several research groups [16, 17], SPCE reduces the observation volume in fluorescence correlation spectroscopy to several attoliters, suppresses fluorescence background and favors a certain dipole emission direction [16, 17]. Nevertheless, applications of SPR and SPCE for surface sensitive fluorescence spectroscopy and imaging are still rare. One reason may be due to the metal-induced energy transfer, which reduces the quantum yield of standard bright molecules [18]. Another reason may be due to the setup geometry, which requires very high numerical aperture (NA ≥ 1.65) objectives [16, 19, 20] or prisms [21, 22] for to collect SPCE.

Here, we theoretically and experimentally dissect the optimal operational regimes and conditions for a successful application of SPCE. First, we compare different plasmonic gold film thicknesses to find the best compromise between coating transparency, SPR excitation and SPCE efficiency. Second, we show how fluoplasmonics (f-Pics) microscopy on gold-dielectric coated coverslips can outperform common imaging techniques by extreme background suppression. We introduce suitable instrumentation and useful operative regime for efficient SPR excitation and SPCE detection that particularly supports live-cell experiments when extremely high photon background is the dominant hurdle.

2. Materials and methods

2.1 Mathematical modeling and simulations

Analytical expression of surface plasmon resonances and plasmonic excitation enhancements on thin noble metal films have been found previously by various analytical approaches (e.g [23, 24].). The interaction of a point emitter in the vicinity of the metallic surface is given by the Chance, Prock, and Silbey (CPS) theory [25, 26] with its radiation directivity [9]. Here, we use finite element method (FEM) calculations to have access to all parameters describing our experimental situation. With FEM, we simulate an emitter on our metal-dielectric multilayer system. From this, derive excitation field enhancement and radiation rates, fluorescence quenching, transmission through the multilayer system and the detection efficiency.

We performed FEM calculations (Comsol Multiphysics 4.4) for the radiative and non-radiative dipole emission rates and far-field radiation distribution, considering different dipole to surface heights and orientations (parallel and perpendicular). We simulated a dipole emitter in water (RI = 1.33, ${\text{QY}}_0$ = 0.35, emission peak: 665 nm) that mimics the situation for Alexa 647 that we use later in the experiment. We placed the dipole at different distances above an uncoated (RI = 1.78, N-LAFN21) and coated glass substrate. The coating design includes a 2 nm chromium layer followed by a gold layer (different thicknesses selected between 10 to 50 nm) and a capping layer of 6 nm silicon-nitride. The wavelength dependent complex refractive indices are extracted from literature [27, 28], also accessible by (https://refractiveindex.info/). The radiation rates $k_{rad}$ in close vicinity to the surface are directly connected with the line integral of the power outflow $\text{P}(\text{φ})$ [29]. To determine the amount of light collected by the objective lens (NA = 1.7), we calculate the detection ratio of radiation reaching the objective to total radiation:

Since excitation of evanescent waves and absorption effects of the gold coating are connected with reduced radiation strength, we can express the non-radiative rates by the difference between total and radiative dipole emission rate $({\text{k}}_{\text{tot}}-{\text{k}}_{\text{rad}}$), where the total dipole emission rate ${\text{k}}_{\text{tot}}$ given by the 360° line integral of the power outflow at the dipole position.

The resulting height dependent quantum yield QY(z) [12, 30] can be expressed as follows:

here ${\text{k}}_{\text{rad},0}$ represents the radiation rates of the dipole not affected by an interface. Note that emission rate values are averaged over the dipole orientations. The excitation is considered as an evanescent field [2]:

Here the excitation enhancement factor is ${\left|\text{E}/{\text{E}}_0\right|}^2$. The field enhancement of our multilayer system is simulated by oblique angle illumination of a glass/gold/coating/medium interface originating from the glass side. We used periodic ports parallel to the multilayers and periodic boundary conditions to complete the model in the perpendicular direction. The penetration depth of the evanescent field is [6]:

Here ${\text{RI}}_{\text{cover}}$ is the refractive index of the coverslip and $\sin {\text{θ}}_{\text{c}}={\text{RI}}_{\text{medium}}/{\text{RI}}_{\text{cover}}$ the sine of the critical angle. By multiplication of excitation field ${\text{I}}_{\text{exc}.}$, quantum yield $\text{QY}\left(\text{z}\right)$ and detection ratio of the radiative rates ${\text{k}}_{\text{rad},\text{obj}}(\text{z})/{\text{k}}_{\text{rad}}(z)$ we derive the fluorescence at a certain distance $z$ to the coverslip. Finally, we can calculate the theoretical intensities for fluorescent spheres and homogenous fluorescent background:

Here ${\text{ρ}}_{\text{Bead}}\left(\text{z}\right)=\text{π }\left(\text{Dz}-{\text{z}}^2\right)$ describes a spherical dye distribution with diameter D, ${\text{ρ}}_{\text{BGR}}$ represents fluorescent background induced by a homogenous dye distribution. We performed those calculations for different gold thicknesses and normalized the results to the uncoated case (standard TIRF) to pinpoint the best compromise between most efficient SPR illumination, SPCE collection, and lowest signal losses. As a measure for the image contrast, we calculate Signal/(Signal + BGR) which provides a value for contrast that is independent of the illumination strength.

2.2 Metal coating and coverslip preparation

Fabrication of the metal-dielectric coating is performed in a HEX deposition system (Korvus Technology Ltd.) equipped with a radio frequency magnetron sputtering (for dielectrics deposition) and an e-Beam evaporation system (for metals). Coverslips with high refractive index RI were used (NLAFN-21, 150 µm thick, UQG Optics Ltd.). The metal coating consists of a 2 nm adhesion promoter (chromium, 99,99%, Goodfellow), followed by a gold layer (99,99%, EVOCHEM) and a 6 nm Si₃N₄ layer (hot-pressed Si₃N₄ sputtering target, Goodfellow).

2.3 Experimental setup

We used a home-built total internal reflection fluorescence microscope (TIRF-M) setup in combination with a commercial base (IX71, Olympus). A digital illumination angle control was used to perform wide-field, TIRF and f-Pics imaging with red laser illumination of 640 nm (OBIS 640nm LX 40 mW). The beam was expanded to 4 mm beam diameter and deflected by a digital scan mirror (62xxH series galvanometer XY, Cambridge Technologies). The laser is focused (f = 200 mm, AC254-200-A-ML Thorlabs) onto the back focal plane (BFP) of the TIRF objective (APON 100XHOTIRF, Olympus, NA = 1.7). A linear polarizer ensured efficient excitation of surface plasmons by p-polarized light (half-wave plate AHWP05M-600, Thorlabs). Note that an objective lens with NA$\ge$1.65 is strictly required to be able to induce SPR excitation and/or SPCE collection of fluorescence in aqueous or cellular samples [19]: For gold coatings on standard glass coverslips, the SPR/SPCE angle is between 80° and 90° (for RI = 1.33 to 1.4 in visible spectral regime, calculations see [23]). This angle is higher than the maximum possible excitation/collection angle limited by numerical aperture ($\text{NA}\le 1.49$,${\vartheta }_{\text{max}}=78°$) of the objective lens. For NA = 1.7 (${\vartheta }_{\text{max}}$ = $72°$) the use of coverslips with a high refractive index (RI = 1.78) reduces the plasmon resonance angles below 70°, which makes SPR/SPCE possible in our TIRF setup.

The fluorescence signal is split and filtered by a dichroic mirror and a band pass filter (F73-421PH and F73-697 (697/58), AHF Analysentechnik). The fluorescence signal is projected onto a camera (Neo 5.5 sCMOS, Andor). Camera detection and laser illumination are synchronized to minimize light exposure of the sample. Scan mirror, laser power, and camera acquisition are controlled by analog signals (PCI-6733, National Instruments) and the python script “Shadowless TIRF” from Ellefsen, Dynes, and Parker [31]. Minor modifications in this python script allow synchronized mirror deflection, laser excitation, and camera detection. Images and image stacks are acquired with the software of the camera manufacturer (Solis, Andor). Calibration curves of the illumination angle θ show a linear behavior over the mirror deflection voltage (slope = 78.9°/V). The penetration depth was determined on uncoated substrates with 6 µm fluorescently stained microspheres (FI4806, Invitrogen) on poly-D-lysine coated coverslips in an aqueous environment (for method see [32]). We determined penetration depths of (179 ± 18) nm at illumination angle ($\vartheta$ = 54° ± 1°) and (89 ± 11) nm at SPR angle (60° ± 1°).

2.4 Imaging of fluorescent beads and background characterization

For characterization of the illumination angle dependent fluorescence signal and background suppression, we used 100 nm multi-color fluorescent beads (T7279, TetraSpeck Microspheres 0.1 µm, Invitrogen) on coated and uncoated coverslips excited at 640 nm. We experimentally confirmed the random distribution of fluorescent molecules of the fluorescent bead by polarization and implemented a polarizer in the emission path. To avoid photobleaching by the strong SPR excitation enhancement during the acquisition series, we used a relatively low laser power of 20 µW, camera binning 2 and exposure time of (0.5 – 1.0) seconds. For easy comparison, signals are normalized to counts/s. Analysis of signal and background was supported by the python module “Trackpy” [33]. “Trackpy” localizes and analyzes fluorescence signals originated from the beads on uncoated and coated coverslips. For each event, “Trackpy” calculates the amount of total and signal fluorescence in an area connected to the event so that the background can be calculated by the difference of total “Raw” and signal “Signal” intensity.

We performed a fluorescent background series by adding different concentrations of Alexa Fluor 647 (A647) labeled oligonucleotides (IBA, for sequence [34]) in ultrapure water (mili-Q, Millipore Corporation). In contrast to pure dye solution, the oligonucleotides coupled to the dye prevented unspecific binding to the surface in our experiments.

2.5 Cell culture, immunostaining, image acquisition and analysis

We cultured wild-type Chinese Hamster Ovary (CHO) cells on un-/ coated coverslips at 37°C and 5% CO₂ in Dulbecco`s Modified Eagle Medium (DMEM/F12, Thermofischer), and performed experiments on uncoated and coated coverslips in the same manner for comparison. The plasma membrane was fluorescently stained (Cellmask Deep Red, Invitrogen, 1x = 1 µl/ml). After 10 minutes of incubation with cell mask, the cells were washed 3 times with DMEM (Gibco, Lifetechnology). For image acquisition, we transferred the cell dish to a home-built imaging chamber that allowed measurements under physiological conditions. On the microscope, we kept the cells in phenol-red free DMEM buffered by 15 mM HEPES for pH-stability. Cell mask in excess (0 - 6 µl/ml) induced a significant fluorescence background. Image acquisition was at 100 µW laser power (before the objective) and 200 ms camera exposure time without binning, For wide-field microscopy the illumination angle was set to 0°, for TIRF and SPR to $\cong$60° reflecting the surface plasmon resonance angle to induce maximum fluorescence on coated coverslips.

For characterization of cell membrane signal and step sharpness, we use Gwyddion [35] to extract profiles from the cell membrane to coverslip transition and fit the line scan with a smooth step function ${\text{f}}_{\text{step}}\left(\text{x}\right)$.

Here, $\text{Signal}$ from the cell membrane is given by the step height of an error function $\text{erf}\left(\frac{\text{x}-{\text{x}}_0}{\text{w}}\right)$ at the position ${\text{x}}_0$ with $\text{w}$ as step width. We started the experiment without induced background and then subsequently increased the fluorescence background.

3. Results

First, we simulated SPR field enhancement (excitation pathway), and SPCE fluorescence (emission pathway) independently, to determine the optimal thickness of the metal coating. We simulated the situation of a fluorescent emitter on a gold coated versus an uncoated glass surface. Figure 1(a) shows the logarithmic electric field intensity distribution of both scenarios (uncoated and coated) for different emitter distances and orientations to the surface. Figure 1(b) shows the emission directivity of our emitter oriented parallel or perpendicular to the interface and for the dipole close to (30 nm) or far from (400 nm) the interface. The gray area illustrates the collection angle of the objective lens. For an uncoated interface, fluorescence emission follows the classical dumbbell radiation pattern with small modification by the refractive index difference at the interfaces. The “butterfly” shaped radiation of an emitter perpendicular (p-pol) to the surface is known as supercritical angled fluorescence (see e.g. [5]) and disappears for increased z. On the uncoated interfaces and for a parallelly oriented emitter the fluorescence pattern (s-pol) is less affected by refractive index mismatch. In the case of our gold coatings, the situation is different: The higher reflectivity and lower transparency of our coatings restrict transmittance of fluorescence, and thus light collection by the objective lens. Fluorescence coupled to the gold surface plasmons (SPCE) is able to pass the gold layer and is most efficient for p-pol fluorescence from emitters few tens of nanometers above the interface [9]. Thus, fluorescence originating from such fluorophores is most pronounced. Since a randomly dipole orientation is the most meaningful scenario in our (biological) experiments, we assumed an averaged dipole orientation.

 

Fig. 1 Finite element method calculations for surface plasmon coupled emission. (a) Electric field distribution log(|E|²) of an electric dipole emitter next to an uncoated and coated (gold 30 nm) medium/glass interface (scale bar 30 nm). (b) The corresponding far-field radiation patterns for a dipole with parallel (dark green) and perpendicular (blue) orientation are depicted for dipole-to-surface distances of 30 nm (solid) and 400 nm (dashed). (c) Calculated intensities for a fluorescent sphere (black, sphere diameter: 100 nm), and a homogenously distributed background (gray) for different gold thicknesses. Intensities were calculated w/ (solid) and w/o (dashed) plasmonic excitation enhancement (red).

Download Full Size | PDF

Second, to determine the ideal gold layer thickness for our scenario we simulated the emission of a 100 nm fluorescent sphere a homogenous distributed background separately. Intensities were normalized to the uncoated case. As shown in Fig. 1(c), best imaging performance is expected around 30 nm gold thickness where the plot of fluorescence on different gold layers as well as the expected background peak. A gold thickness of 30 nm leads to the highest intensity and a clear signal contrast even if the strongest SPR is observable for ~48 nm thick gold layers. Note that without excitation enhancement; detected fluorescence is strongly reduced compared to the case without coating.

Next, we simulated the fluorophore behavior at a distance z above the interface. In Fig. 2(a) we plotted the effective distance depending quantum yield QY for a near-infrared dye emitter (emission: 665 nm) with intrinsic quantum yield QY₀: For the chosen metal-dielectric nanolayers, fluorophores with QY₀ > 0.4 are strongly quenched in the first nanometers than those with lower QY₀ [12]. In other words, the effective QY of fluorophores with low QY₀ increases in the plasmonic environment while QY (and brightness) decreases in the case of high QY₀. For the following experiments, we will use the commonly used fluorescent dye A647 (emission: 665 nm, QY₀ = 0.35) that benefits from such quantum yield modification in the SPCE efficient region. Next, we simulated the fluorophore signal intensity in dependence of distance z above the interface; see Fig. 2(b). Here the signal is the product of the excitation field, quantum yield and the detection efficiency (for details see Methods section). Wide-field (WF) and TIRF illuminations (for penetration depths of 169 nm (${\vartheta }_{\text{crit}}\le \vartheta =50°$) and 72 nm (${\vartheta }_{\text{SPR}}=$58°) are modeled for an uncoated interface. The SPCE is calculated on the coated interfaces at SPR excitation. The TIRF and WF signal distribution follow their well-known decay behaviors. The signal in the case of SPCE starts lower due to fluorescence quenching, followed by a steep rise when efficiently coupling to surface waves, and peaks at ≈50 nm distance before dropping rapidly at higher interface distances. The simulations shown in Fig. 2 are in good agreement with the experimental results reported in the literature [36]. Note that the silicon nitride spacer layer prevented stronger fluorescence quenching at z ≈0 nm. To pinpoint the benefit of SPCE compared to WF and TIRF, we calculated Signal/(Signal + BGR) as a measure of contrast for 100 nm fluorescent beads (Signal) attached to the nanolayered interface and a homogenous background (BGR). The concentration ratio of BGR and Signal is given by the factor c, while here BGR = c⋅BGR_{c = 1}. In Fig. 2(c) we plot the Signal/(Signal + BGR). For all illumination methods reasonable contrast is given until a BGR exceeds the signal by a factor of 100 and drops later to zero depending on the illumination method. We confirmed that changes of the signal contrast are negligible for small quantum yield variations, which may occur in heterogeneous chemical environments or different buffer solutions.

 

Fig. 2 FEM calculations of fluorophore behavior. (a) Distance z depending relative fluorescence quantum yield $\text{QY}/{\text{QY}}_0$ for different free space quantum yield ${\text{QY}}_0$. (b) Calculated signal collected by an objective lens (NA = 1.7) for different fluorophore distances z to the uncoated (WF and TIRF) and coated (SPCE, 30nm gold coating) coverslip surface. (c) Contrast of fluorescence signal (Signal) and homogenous background (BGR) depending on dye concentration ratios for different illumination scenarios. A step like Signal/(Signal + BGR) distribution illustrates the signal-contrast in high-background environments. The signal-contrast is independent from the excitation amplitudes. For TIRF two different penetration depths (solid: 169 nm; dashed: 72 nm) were considered. The penetration depths for SPCE under SPR illumination is 64.4 ± 0.3 nm (extracted by a mono exponential decay fit, z > 20nm).

Download Full Size | PDF

For experimental validation of our theoretical simulations, we used a home-built TIRF-M. For intensity characterization of the fluorescent beads on coated and uncoated coverslips, a high NA objective (NA = 1.7) provided efficient excitation of SPR and collection of SPCE. Our TIRF setup allows computer assisted fine 0.5°deg changes of the excitation angle $\vartheta$ to precisely hit the critical (${\vartheta }_{\text{crit}}$) and SPR (${\vartheta }_{\text{SPR}}$) excitation angles. Figure 3(a) shows an illustration of the TIRF-M setup. By sweeping the excitation angle, different modes of fluorescent excitation can be induced. For uncoated interfaces, the fluorescent beads are visible for all illumination angles. The fluorescence signals for undercritical WF illumination and critical to overcritical TIRF illumination is shown in Fig. 3(b). When approaching the critical angle, the fluorescence is increased due to resonance excitation of the surface modes and then drops rapidly when further increasing the angle. In contrast, for coated coverslips (gold layer thickness > 25 nm) the beads are only visible at SPR illumination. A doughnut-shaped point-spread function (PSF), a well-known characteristic of the SPCE [37–39], is clearly observable. Analysis of signal and background was supported by the python module “Trackpy” [33]. Figure 3(c) shows the resulting average signal of beads for different gold thicknesses with highest values for uncoated and 30 nm gold-coated coverslips. Thus, the latter was picked as the best compromise to efficiently analyze samples in high background environments.

 

Fig. 3 Experimental f-Pics setup and performance scheme of the setup and sample configuration. (a) Main optical components include scan mirror, laser (640 nm), linear polarizer (P), focus lens (FL), dichroic mirror (DM) and bandpass filter (BP). The TIRF objective back focal plane (BFP) and the image plane (IP) are indicated. (b) Measurements of fluorescent spheres on uncoated and coated coverslips for different excitation angle $\vartheta$: undercritical $\vartheta <{\vartheta }_{\text{crit}}$, critical $\vartheta ={\vartheta }_{\text{crit}}$ and surface plasmon resonant (SPR) $\vartheta ={\vartheta }_{\text{SPR}}$ excitation. The circles indicate analyzed objects. Scale bar: 2 µm. (c) Excitation angle $\vartheta$ dependent fluorescence of spheres on gold coatings (left), and the maximal fluorescence for different gold thickness (right) with best performance for 30 nm gold layers.

Download Full Size | PDF

Figure 4 shows fluorescence images of beads on uncoated and gold coated glass coverslips in the presence of different background levels induced by respective cell mask concentrations. In Fig. 4(b) the results for the signal to background contrast are summarized. The step-like function observed in Fig. 4(b) is in compliance with the theoretical calculations shown in Fig. 2(c). For low BGR, the fluorescent beads are clearly observable on uncoated (WF and TIRF) and coated glass (SPR excitation). For such low BGR conditions, classic TIRF-M is clearly preferable as it provides higher photon yield and thus outperforms f-Pics as expected. The situation is different for high (100 nM) BGR: Even if exhibiting sufficient contrast in TIRF mode, beads show the best contrast in f-Pics where SPR excitation and SPCE provides background-suppression far beyond TIRF-M. Finally, at extreme BGR of 500 nM, beads only remain detectable for f-Pics.

 

Fig. 4 (a) Images of fluorescent beads (emission bandpass 697-58) in high background A647-oligonucleotides environment on uncoated (WF and TIRF) and coated (SPR, 30 nm gold) coverslips. (b) Results of the signal-to-BGR analysis for different BGR concentrations. Scale bars: 10 µm.

Download Full Size | PDF

Next, live-cell imaging is performed to test the versatility of our operational regime. Figure 5(a) shows respective cell images (with additional background 0, 2 and 4 µl/ml) detected in the presence of a high fluorescent background. For TIRF and SPR illumination (both by $\vartheta \cong$60°) the intensity profiles in Fig. 5(b) derived from panel (a) exhibit a step profile at the cell membrane. For WF illumination, the intensity rises further towards the image center because of the unselective axial illumination. We analyzed intensity profiles to identify cell signal and background, respectively. For all illumination modes, the step height and base values are determined; in addition, for TIRF and SPR the step width was described by a slanted step model, so that step base and height determine the signal-to-background contrast see Fig. 5(b). The step widths are 0.65 ± 0.21 µm and 1.27 ± 0.27 µm for TIRF and SPR illumination (cell numbers N > 10). The reason for the increased step size is the doughnut-like PSF on the metallic films resulting in blurrier images. For comparison, we measured a full width half maximum (FWHM) of 100 nm beads of ~330 nm on uncoated and ~435 nm on coated coverslips.

 

Fig. 5 (a) Fluorescence images of living membrane stained CHO cells on uncoated (WF, TIRF) and gold coated (30 nm, SPR) coverslips. (b) Example of lateral intensity profiles (open symbols) with data fits (solid lines) representing step models to dissect image contrast. (c) Comparison of image contrast for depicted imaging modalities. Scale bars: 5 µm.

Download Full Size | PDF

The results of background suppression are comparable to the previously shown experiments in Fig. 3 and the theoretical results in Fig. 2(c). For WF and TIRF, the contrast drops immediately when the staining concentration is increased. For the SPR illumination on coated coverslips, the contrast remains high, and only decreases if background concentration is increased to 6 µl/ml. Taken together, the background suppressed operative regime for this live-cell experiments on coated coverslip ranges from ~2 to 4 µl/ml staining concentrations.

4. Discussion

SPCE for fluorescence imaging has not been widely successful far; most likely due to its limited photon yield compared to other techniques. Here, we showed that there is an attractive SPCE operational regime for extremely high background applications including live-cell experiments in a noisy environment. In our experiments, the determined image contrast was decoupled from the excitation enhancement so that dissection of the interesting operational regime for f-Pics became evident. Thus, we systematically explored an experimental regime where surface plasmon assisted microscopy in our f-Pics approach provides higher image contrast compared to TIRF-M. As our data shows, f-Pics allows for imaging under high-background conditions. While 20 nM to 500 nM background fluorophore concentration would prevent imaging by WF and TIRF-M, SPCE is more selective and enables fluorescence imaging under this extreme background conditions.

For the respective coating design, we identified 30 nm gold as optimal layer thickness and confirmed this by computational modeling and imaging experiments. Unfortunately, the optimal gold layer thickness always represents a compromise when an epi-fluorescence, objective-based illumination mode is used. Therefore, this approach is not ideal for very low-light or single molecule applications where a prism-based setup [13, 14, 40] with decoupled excitation and detection pathways would provide more detection efficiency but no SPCE fluorescence selectivity.

To make the method more versatile for future applications, f-Pics need to address the reduction of ohmic losses inside the metal layer. The reduced quantum yield and the doughnut-shaped PSF of the SPCE are equally unfavorable for fluorescence-based applications. Approaches to overcome these issues are under current development. It has been already demonstrated that the doughnut-shaped PSF can be transferred with polarizers into an Airy-shaped PSF to increase the resolution preferable with deconvolution later [38, 39]. As alternative methods, supercritical angle fluorescence (SAF) approaches blocking fluorescence signals in the Fourier plane [41] are useful to distinguish between surface-near and surface-far fluorescence. In SAF approaches, the PSF stays Gaussian-shaped and the signal losses are lower than for SPCE; however, modifications in the instrumentation are required to separate undercritical and supercritical fluorescence in the detection, followed by computational recombination. This additional effort allows for excellent 3D superresolution [42, 43]. In comparison, the main advantage of nanocoatings such as used in f-Pics is their straight-forward experimental implementation. This, taken together with their capability of extreme background suppression under biologically relevant conditions, makes them unique and highly valuable for applications that otherwise suffer from high background levels.

Another innovative approach for nanocoatings are all-dielectric multilayer coatings, which support evanescent modes by Bloch surface waves without ohmic and diffraction losses [44]. If this issue is solved, we believe that such evanescent interface mode-supported fluorescence may have exciting prospective for surface sensitive fluorescence-based imaging and spectroscopy. It could be beneficial for high throughput screening (on non-purified substances), cell growth studies and near-surface binding studies where a reservoir of fluorescent dye is present or added for long-time studies. Imaging cell membrane receptors of adherent cells to study receptor interactions or internalization would be straightforward even if a high concentration of markers inside the cell would outshine the specific membrane signal.