Mixed metal zero-mode guides (ZMWs) for tunable fluorescence enhancement

Zero-mode waveguides (ZMWs) are capable of modifying fluorescence emission through interactions with surface plasmon modes leading to either plasmon-enhanced fluorescence or quenching. Enhancement requires spectral overlap of the plasmon modes with the absorption or emission of the fluorophore. Thus, enhancement is limited to fluorophores in resonance with metals (e.g. Al, Au, Ag) used for ZMWs. The ability to tune interactions to match a wider range of fluorophores across the visible spectra would significantly extend the utility of ZMWs. We fabricated ZMWs composed of aluminum and gold individually and also in mixtures of three different ratios, (Al : Au; 75 : 25, 50 : 50, 25 : 75). We characterized the effect of mixed-metal ZMWs on single-molecule emission for a range fluorophores across the visible spectrum. Mixed metal ZMWs exhibited a shift in the spectral range where they exhibited the maximum fluorescence enhancement allowing us to match the emission of fluorophores that were nonresonant with single metal ZMWs. We also compared the effect of mixed-metal ZMWs on the photophysical properties of fluorescent molecules due to metal–molecule interactions. We quantified changes in fluorescence lifetimes and photostability that were dependent on the ratio of Au and Al. Tuning the enhancement properties of ZMWs by changing the ratio of Au and Al allowed us to match the fluorescence of fluorophores that emit in different regions of the visible spectrum.


Surface Functionalization of ZMWs
All five types of ZMWs i.e 100Al, 75Al, 50Al, 25Al and 100Au were cleaned by thorough rinsing with GenPure 18 MΩ pure DI H 2 O, 100% ethanol, followed by plasma cleaning with Harrick plasma cleaner for 5 minutes. To avoid nonspecific binding of dye molecules on the wall of ZMWs, a protective coating of PVPA was employed by adding 2% V/V aqueous solution of PVPA (Poly (vinylphosphonic acid)) at 110°C for 2 minutes to the ZMW, followed by a rinsing with DI water and let dry for 10 minutes at 80°C on a hot plate. As PVPA has higher binding affinity for metals than glass, it preferentially binds with the metal wall of ZMW's 3 . PVPA coated ZMWs were then functionalized with biotin-neutrAvidin linker chemistry. 2 mg/ml Biotin-PEG-Silane (Laysan Bio) in 95% ethanol was added to it for 30 minutes, rinsed with 1X PBS buffer 7.0 and 100 µM neutrAvidin (Sigma Aldrich) in 1X PBS buffer 7.0, left idle for 2 hours to promote binding, and finally rinsed with 1X PBS buffer 7.0 to remove any residual unbound neutrAvidin.
Single fluorophore binding. Fluorophore molecules were bound to the surface neutrAvidin by biotin -neutrAvidin linkage (Fig.  S2). Four different types of biotin-conjugated fluorophores were used: ATTO 550-biotin, ATTO 590-biotin, ATTO 610-biotin and ATTO 647N-biotin whose fluorescence spectra cover almost the entire visible range. All the fluorophores were purchased from ATTO-tech, Germany. Biotin bound fluorophore molecules were diluted stepwise in 1X PBS buffer 7.0 to achieve single molecule level concentrations. As NeutrAvidin has 4 binding sites for biotin, non-fluorescent biotin was added with the fluorophore solution to avoid multiple fluorophore binding with a single neutrAvidin molecule.
For Glass, 1-10 pM of biotin-bound fluorophore in 1X PBS buffer pH 7.0: 100 nM of biotin in 1X PBS buffer pH 7.0 was added to for 5 minutes and rinsed thoroughly with 1X PBS buffer pH 7.0. For ZMW's, 1-10 nM of biotin bound fluorophore in 1X PBS buffer pH 7.0: 100 nM of biotin in 1X PBS buffer pH 7.0 was added to for 10 minutes and rinsed thoroughly with 1X PBS buffer pH 7.0.
Single molecule data acquisition. Time tagged data acquisition was performed using a custom built confocal microscope setup on an inverted, dual stack Olympus IX-83 microscope frame. For the excitation source, a SuperK Extreme Supercontinuum Free Space Pulsed Laser was used. Substrates were placed on a piezo electric stage (Mad City lab) and raster scanned at 30x30µm 2 -50x50µm 2 area monodirectionally with excitation light with 4 ms dwell time in each pixel. Excitation laser light was filtered by passing it through narrow band pass filter optics to remove stray light and focused on the substrate surface using an Olympus 60X x 1.45 NA oil objective. Emitted light along with some reflected excitation light travel back through objective. A dichroic mirror, however, filters out the reflected excitation light but permits emission light to pass through it. Emission light then passed through a 100 µm pin hole, an emission filter, and finally collected by an avalanche photo diode (APD). The APD converts single photons into an electrical signal, which is time tagged using a photon counter (picoharp 300). The Picoharp 300 communicates with the PC and Symphotime 64 software assigns the photon to each pixel generating an image of the area scanned such that fluorophores appear as bright spots in the image. The Piezo electric nano positioning stage was then used to position the fluorophore in the location of the confocal beam and the excitation beam was unshuttered to expose the fluorophore for at least 10 seconds beyond the time period the molecule photobleached. Fluorescence emission data was analyzed using Symphotime 64 software as a plot of fluorescence intensity time trace and fluorescence lifetime histogram. ATTO 550 molecules were excited with 532 nm laser light and filtered through double excitation filters (ZET 532/10X, chroma), and emission was passed through ET 542 LP (Chroma) and ET 575/40M (Chroma). Excitation power was 1.03 µW. Both ATTO 550 and ATTO 610 molecules were excited with 594 nm laser light filtered through a single excitation filter (ZET 594/10X, chroma), and emission was passed through HQ 650/75M. However, the laser power for ATTO 590 and ATTO 610 was 0.18 µW and 2.19 µW respectively. ATTO 647N molecules were excited with 640 nm laser light filtered through double excitation filters (ZET 640/10X, chroma), and emission was passed through double ET 673/44M (chroma). Laser power was 1.25 µW.
Reflectance. Reflectance spectra were recorded using a 2 inch diameter integrating sphere (Thorlabs) coupled with an Ocean Optics QEPro spectrometer equipped with a thermoelectric cooled CCD detector and an Ocean Optics Deuterium-Tungsten Halogen light source. The substrates were placed against the exit port of the integrating sphere and angled slightly to reduce specular reflection directly back to the entrance port.

Data analysis
A custom MATLAB script was used to extract fluorescence intensity time trace data and plot the fluorescence intensity vs time for each molecule. The average fluorescence intensity of each molecule was calculated by subtracting the average intensity of the time points beyond photobleaching from the average intensity of the time points before bleaching. Survival time of molecules in each substrate was calculated by fitting the photobleaching times of single molecules with single exponential decay. The fluorescence lifetime of each molecule was calculated by fitting the fluorescence lifetime histogram using n-exponential deconvolution. The average fluorescence intensity was reported as mean±SEM, however, both average survival time and fluorescence lifetime were reported as mean±SD.