Matching Nanoantenna Field Confinement to FRET Distances

Aug 3, 2015 - As opposed to gold structures of similar dimensions, which have a resonance above 700 nm,(10) the use of aluminum conveniently shifts th...
2 downloads 11 Views 3MB Size
Letter pubs.acs.org/NanoLett

Matching Nanoantenna Field Confinement to FRET Distances Enhances Förster Energy Transfer Rates Petru Ghenuche,† Mathieu Mivelle,‡ Juan de Torres,† Satish Babu Moparthi,† Hervé Rigneault,† Niek F. Van Hulst,‡,§ María F. García-Parajó,‡,§ and Jérôme Wenger*,† †

CNRS, Aix-Marseille Université, Centrale Marseille, Institut Fresnel, UMR 7249, 13013 Marseille, France ICFO-Institut de Ciencies Fotoniques, Mediterranean Technology Park, 08860 Castelldefels, Spain § ICREA-Institució Catalana de Recerca i Estudis Avançats, Barcelona 08010, Spain ‡

S Supporting Information *

ABSTRACT: Förster resonance energy transfer (FRET) is widely applied in chemistry, biology, and nanosciences to assess distances on sub-10 nm scale. Extending the range and applicability of FRET requires enhancement of the fluorescence energy transfer at a spatial scale comparable to the donor−acceptor distances. Plasmonic nanoantennas are ideal to concentrate optical fields at a nanoscale fully matching the FRET distance range. Here, we present a resonant aluminum nanogap antenna tailored to enhance single molecule FRET. A 20 nm gap confines light into a nanoscale volume, providing a field gradient on the scale of the donor−acceptor distance, a large 10-fold increase in the local density of optical states, and strong intensity enhancement. With our dedicated design, we obtain 20-fold enhancement on the fluorescence emission of donor and acceptor dyes, and most importantly up to 5-fold enhancement of the FRET rate for donor−acceptor separations of 10 nm. We also provide a thorough framework of the fluorescence photophysics occurring in the nanoscale gap volume. The presented enhancement of energy transfer flow at the nanoscale opens a yet unexplored facet of the various advantages of optical nanoantennas and provides a new strategy toward biological applications of single molecule FRET at micromolar concentrations. KEYWORDS: FRET, plasmonics, LDOS, nanoantenna, fluorescence enhancement, aluminum

O

nanoparticles26,27 observed negligible changes of FRET rates, whereas reports using dense layers of emitters in microcavities,28−30 arrays of nanoparticles,31−33 and nanoapertures34,35 indicated moderate FRET rate increases. Reaching a substantial FRET enhancement, thus, has remained challenging. Indeed, theoretical studies on FRET around single nanoparticles36−41 have predicted specific stringent conditions that depend on the nanoscale spatial configuration and on the spectral overlap of the plasmon resonances with the donor/ acceptor spectra to produce a major change in the local density of optical states (LDOS). To achieve a significant FRET rate enhancement by means of an optical antenna, several conditions should be met:22,40 (i) the field intensity enhancement should be large, (ii) light should be confined on similar dimensions as the donor− acceptor distance, typically 0−20 nm, (iii) the operation wavelength should be set in the visible region to accommodate for most common fluorescent dyes, (iv) the antenna bandwidth should be sufficiently broad to overlap both the donor and the

ptical nanoantennas provide strong confinement of light into nanoscale volumes1−3 and giant enhancement of the luminescence from a single quantum emitter.4−6 These features make optical antennas ideal for ultrasensitive molecular detection down to the single molecule level even at high micromolar concentrations, corresponding to the physiological conditions for most biomolecular interactions.7−10 Among the numerous single molecule techniques that can benefit from optical antennas, Förster resonance energy transfer (FRET) stands out as a major technique based on the near-field energy transfer from an excited donor to a ground state acceptor emitter.11,12 FRET is used to accurately measure the distance between two fluorescent sources at the nanometer scale,13,14 enabling the study of molecular conformations 15 and interaction dynamics.16 Moreover, FRET is a core phenomenon in photosynthesis,17,18 organic photovoltaics,19 lighting sources,20 or biosensing.21 With the broad field of FRET applications at hand, the impressive progress of optical antennas offers appealing opportunities to enhance FRET and extend its versatility. However, combining nanophotonics with FRET requires proper engineering of the energy flow at the nanoscale. Earlier works using mirrors,22,23 microresonators,24,25 and dielectric © XXXX American Chemical Society

Received: June 26, 2015 Revised: July 30, 2015

A

DOI: 10.1021/acs.nanolett.5b02535 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 1. Aluminum nanogap antenna for single-molecule FRET. (a) Scanning electron microscope image of a gap antenna fabricated by focused ion beam milling in an aluminum film. (b) Sketch of the nanogap region. The enhanced excitation intensity (red line) is confined on the same nanoscale dimensions as the donor−acceptor distance. (c) Jablonski diagram describing the FRET process. All rates are modified by the antenna. (d) Dark-field scattering spectra of the aluminum antenna (filled blue curve) with the illumination polarization parallel to the antenna axis. The antenna’s response covers both the Atto550 donor and the Atto647N acceptor spectra (green and red curves respectively, dashed lines are for absorption spectra, solid lines correspond to fluorescence emission). (e,f) Intensity enhancement computed with the excitation polarization parallel to the gap at 550 nm and the 670 nm.

per unit volume and frequency at the position of the dipole emitter where the energy can be released during the spontaneous emission process. This definition implies that the LDOS is proportional to the isolated donor total decay rate ΓDo (inverse of the fluorescence lifetime), taking into account both radiative and nonradiative decays.42 The antenna is directly milled by focused ion beam in a 50 nm thick aluminum film on a glass coverslip to reach gap sizes around 20 nm (Figure 1a) The Supporting Information Figure S1 shows a vertical cross section of the experimental configuration, and Supporting Information Figure S2 provides more electron microscope images of the nanoantennas. The dark-field scattering spectrum in Figure 1d indicates a broad resonance centered at 620 nm. As opposed to gold structures of similar dimensions, which have a resonance above 700 nm,10 the use of aluminum conveniently shifts the resonance toward the visible part of the spectrum. Figure S3 in the Supporting Information confirms that the resonance is related to the dimer antenna and is sensitive to the polarization orientation. Importantly, the antenna’s resonance covers both the absorption and emission spectra of the donor and acceptor dye pairs used in these studies (Atto550 as donor and Atto647N as acceptor). Numerical simulations of the intensity distributions display similar profiles at the 550 nm donor excitation wavelength and the 670 nm acceptor emission wavelength (Figure 1e,f), as further indication of the broadband spectral response of the aluminum gap antenna. Ensemble-based measurements can lead to discussions about collective effects in FRET.23,28 Therefore, we conduct all our experiments at the single molecule level on well-defined donor−acceptor fluorophore pairs on double-stranded DNA linkers. The FRET samples are formed by single Atto550Atto647N donor−acceptor pairs covalently attached on double stranded DNA molecules, providing a linker that enables accurate positioning of the donor and acceptor with

acceptor absorption and emission bands, (v) the antenna should have limited ohmic losses to the metal, and (vi) the antenna should induce minimal parasitic emission such as metal photoluminescence. So far, there has been no report of FRET in a structure featuring all these requirements. In this work, we present a resonant gap antenna optimized for all conditions to enhance the FRET rate of individual donor−acceptor pairs (Figure 1a,b). The 20 nm gap confines light into a nanoscale volume, providing strong intensity enhancement, a field gradient on the scale of the donor− acceptor distance, and a large 10-fold LDOS increase. The aluminum antenna has a resonance in the visible range of the spectrum with a broad bandwidth covering both the donor and acceptor spectra and also has negligible autoluminescence as compared to gold.35 As additional advantage, the antenna design is fully compatible with the detection of single fluorescent molecules in water solution thanks to a nanoaperture surrounding the nanogap antenna so as to efficiently screen for the far-field background. This unique combination of features enables the conclusive observation of significant FRET rate enhancement up to 5-fold in a nanogap with single molecule resolution. The Jablonski diagram in Figure 1c introduces our notations to describe the different photokinetic rates. It also illustrates the challenge of assessing the LDOS influence on FRET, as the FRET rate ΓFRET for dipole−dipole energy transfer directly competes with the donor direct radiative emission rate ΓDrad and the donor nonradiative energy losses to the environment ΓDnr (these decay rates are sometimes referred to as rate constants).24,25 For a resonant nanoantenna, we have to assume that all rates are affected by the photonic structure. In the absence of the acceptor, the isolated donor decay rate is noted ΓDo = ΓDrad + ΓDnr and encompasses both radiative and nonradiative transitions. Following the common approach in nanophotonics,42 the LDOS is defined as the number of modes B

DOI: 10.1021/acs.nanolett.5b02535 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 2. Raw fluorescence intensities demonstrate the occurrence of FRET in the gap antenna. (a) Acceptor fluorescence time traces in the antenna for different donor−acceptor distances, and with the acceptor alone (binning time 100 ms). The acceptor intensity increase in the presence of the donor is a direct indication for the occurrence of FRET. (b) Donor fluorescence time traces corresponding to the same experiment as in (a) but for the donor detection channel. The donor intensity decrease (quenching) in the presence of the acceptor is another direct indication of FRET. (c) FCS correlation functions corresponding to the time traces displayed in (a,b). The concentration of donor−acceptor pairs is kept constant at 1 μM, as indicated by the similar amplitudes of the FCS fluorescence correlation functions. The results of the FCS analysis are detailed in the Supporting Information Table S1.

Figure 3. Antenna enhanced FRET rates probed by accelerated donor photodynamics. (a−c) Normalized donor fluorescence decay traces in confocal (a), box aperture (b) and antenna with polarization parallel to the antenna dimer (c). Green traces correspond to isolated donor, blue traces to donor−acceptor D−A separation of 10.2 nm and orange traces to D−A distance of 6.8 nm. Black lines are numerical fits convoluted by the instrument response function (IRF). The presence of the antenna reduces the donor lifetime, demonstrating an enhanced local density of optical states (LDOS). The presence of the acceptor further accelerates the donor decay dynamics, which demonstrates the occurrence of FRET. (d) Isolated donor decay rates ΓDo and donor decay rates in the presence of the acceptor ΓDA = ΓDo + ΓFRET extracted from the data in (a−c) and Supporting Information Figure S5 for the antenna with perpendicular excitation. The difference between ΓDA and ΓDo corresponds to the FRET rate and is highlighted in bright color. The rate enhancement (right scale) is defined respective to the isolated donor decay rate in confocal case. (e) FRET rate ΓFRET as a function of the isolated donor decay rate ΓDo, which is proportional to the LDOS. A linear relationship is found for both donor−acceptor separations. From left to right, the data points correspond to the cases of confocal, box aperture, and antenna with perpendicular and parallel excitation polarization.

subnanometer resolution.13 To explore different conditions of FRET rates, we perform our experiments for two sets of donor−acceptor separations of, respectively, 20 base pairs (∼6.8 nm) and 30 base pairs (∼10.2 nm). Bringing the acceptor closer to the donor increases the energy transfer, whereas the LDOS set by the antenna remains unaffected. The

nanoantennas are covered by a solution containing the DNA FRET pairs at 1 μM concentration so that single molecules are detected as they cross the antenna hot spot. The molecules constantly diffuse through the 20 nm gap of the nanoantenna, so that our observations are not affected by photobleaching. We C

DOI: 10.1021/acs.nanolett.5b02535 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 4. Spatial distributions of FRET rate enhancement in a 22 nm aluminum gap. (a) Close-up view of the FRET rate enhancement in the antenna gap computed as the ratio of the electric field intensity emitted by the donor dipole in the presence and absence of the nanoantenna. The donor dipole position is indicated by the white arrow, and is located either at the center, or at 4 or 6 nm lateral shift. In all cases, the dipole is oriented along the dimer main axis. The aluminum antenna corresponds to the purple regions limited by the dashed line. All distances are in nanometers. (b) Horizontal cut along the images in (a) showing the FRET rate enhancement along the dimer axis for three different positions of the donor dipole respective to the antenna. The shaded regions represent the metal locations.

also set the illumination power to the minimal value of 10 μW to avoid the fluorescence saturation of the dyes. A direct evidence for the occurrence of FRET in the aluminum antenna is given by the evolution of the donor and acceptor fluorescence intensities as the donor−acceptor separation is reduced (Figure 2a,b). The simultaneous increase of the acceptor intensity and quenching of the donor intensity indicate energy transfer from the donor to the acceptor. To ensure that these observations are not related to concentration variations between the experiments, we analyze the fluorescence time traces by fluorescence correlation spectroscopy (FCS) to quantify the number of detected molecules.10,43 The similar amplitudes of the correlation curves shown in Figure 2c confirm that the concentrations are nearly identical. Analysis of the FCS correlation curves yields an average number of molecules in the hot spot of N* = 0.13 ± 0.02, which demonstrates the single-molecule feature of our experiments (see section 4 of the Supporting Information and Table S1 for all parameters of the FCS analysis, the average residence time of the DNA molecules in the antenna hot spot is 24 μs). At 1 μM concentration, observing 0.13 molecules corresponds to a detection volume of 220 zL (10−21 L) which nicely agrees with the size of the antenna hot spot computed in Figure 1e,f. Additionally, the data corresponding to the surrounding box aperture (in the absence of the antenna) confirms that the signal observed with the antenna stems from the gap region (Supporting Information Figure S4 and Table S2). The FCS analysis also quantifies the fluorescence enhancement in the antenna gap region.10 For the isolated Atto550 donor and isolated Atto647N acceptor, we obtain fluorescence enhancement factors of 8.5× and 17.3×, respectively (Supporting Information Tab. S1). These values are lower that previous

enhancements reported in gold antennas,10 yet this is explained by the lower excitation and radiative rate enhancements found for aluminum compared to gold. Moreover, in the current experiments we use fluorescent dyes with relatively high quantum yields, and so the emission efficiency can hardly be further enhanced (the quantum yield for Atto550 and Atto647N are 80% and 65%, respectively). To measure the LDOS enhancement brought by the aluminum nanogap antenna, we first record the fluorescence photodynamics of the isolated donor in the absence and in the presence of the antenna. The comparison of the time-correlated decay traces for the isolated donor in Figure 3a−c reveals faster emission dynamics in the antenna than in confocal or in the box aperture, indicating higher LDOS in the antenna (see also Supporting Information Figure S5a). The isolated donor total decay rate ΓDo is quantified by analyzing the fluorescence decay traces (full details are provided in the Methods section). Strikingly, the isolated donor decay rate ΓDo increases by 9.4× in the antenna with parallel excitation, yielding an equivalent LDOS enhancement. All results are summarized in Figure 3d and Supporting Information Table S3. An equivalent 10.5× increase is also observed for the isolated acceptor total decay rate (see Supporting Information Figure S5), demonstrating that the antenna has a large and spectrally broad influence on the LDOS. Next, we focus on the energy transfer rate. In all cases the presence of the acceptor further accelerates the donor emission dynamics, as clearly seen in Figure 3a−c and Supporting Information Figure S5. This demonstrates the occurrence of FRET, as the acceptor opens a new decay pathway for the donor. In the presence of the acceptor, the donor decay rate becomes ΓDA = ΓDo + ΓFRET, and is higher than the isolated D

DOI: 10.1021/acs.nanolett.5b02535 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 5. Nanogap FRET enhancement probed by fluorescence burst analysis. (a) Typical fluorescence time traces for Atto550-Atto647N FRET pairs in an aluminum nanogap antenna. For each detected fluorescence burst, a FRET efficiency is calculated (bottom trace). The binning time is 0.2 ms which is similar to the diffusion time of DNA samples in the antenna. Longer traces are provided in the Supporting Information Figure S8. (b,c) FRET efficiency histograms extracted from fluorescence burst analysis for donor−acceptor separations of 6.8 nm (b) and 10.2 nm (c). The number of detected events exceed 11 000 for the antenna and 6000 for the confocal reference. (d,e) Evolution of the FRET efficiency with the isolated donor decay rate ΓDo for the two D−A distances. Filled markers correspond to the center FRET efficiency deduced from the Gaussian fits of the histograms in (b,c). Empty markers represent the FRET efficiency computed independently from the lifetime data in Figure 3c. From left to right, the data points correspond to the cases of confocal, box aperture, and antenna with perpendicular and parallel excitation polarization. The dashed lines indicate the evolution of the FRET efficiency if the FRET rate were independent of the LDOS. The trend observed in our data (solid lines) refutes this hypothesis and demonstrates a dependence of the FRET rate on the LDOS in the aluminum nanogap, resulting in an enhanced FRET efficiency.

donor decay rate ΓDo by the FRET rate ΓFRET. The acceleration of the donor decay dynamics therefore quantifies the FRET rate ΓFRET = ΓDA − ΓDo, which is displayed in bright colors on the Figure 3d bars. Figure 3e shows ΓFRET as a function of ΓDo and the LDOS. Remarkably, a linear relationship is found for the different donor−acceptor distances and the different configurations. We observe a significant increase in ΓFRET in the antenna as compared to the confocal reference, implying that the antenna indeed enhances the FRET rate. For the 6.8 nm donor−acceptor distances, ΓFRET is enhanced by 2.9-fold, whereas for the 10.2 nm separation a 4.6-fold FRET rate enhancement is achieved. The higher FRET rate enhancement observed for larger donor−acceptor separation is a consequence of the fact that at short donor−acceptor distances (on the order of the Förster radius or below), the direct dipole− dipole interaction dominates the energy transfer, leaving little influence for the LDOS effect on the FRET rate.22,34,35,40 Interestingly, for larger distances with weaker dipole−dipole interaction and lower FRET efficiency, the influence of the antenna becomes more prominent and the extra contribution to the energy transfer mediated by the nanoantenna does noticeably enhance the apparent FRET rate. These results are important because they indicate that optical antennas can extend the spatial range of FRET to distances where dipole− dipole interactions would otherwise be too weak to result in a detectable FRET signal. Numerical simulations provide more insight in the spatial distribution of the FRET rate enhancement inside the nanogap region. A description of FRET following a quasiclassical

approach shows that the FRET rate enhancement in the nanoantenna scales as the donor power transferred to the acceptor.42 The FRET rate enhancement is thus proportional to the square of the electric field emitted by the donor at the acceptor position projected on the acceptor’s dipole axis.39,42 Assuming for simplicity an emission at a single frequency, we compute the enhancement of the energy transfer rate by calculating the ratio of the field intensity distribution |ED(rA)|2 created by the donor at the acceptor’s position in the presence and absence of the nanoantenna (Supporting Information Figure S7 displays the dipole intensity distributions). Figure 4 summarizes our main results for three different donor positions: in the center of the 22 nm gap and shifted from the center by +4 and +6 nm. The FRET enhancement maps reveal zones of enhanced energy transfer within the nanogap, with typical enhancement factors in the range 2−8×, in good agreement with the experimental results in Figure 3e. Moreover, the enhancement factors are higher if the donor dipole is shifted from the antenna center, breaking the symmetry and increasing the coupling between the two aluminum nanoparticles. We also observe that the FRET enhancement is larger for longer D−A separations, again in excellent agreement with the experimental results (Figure 3e). To complete and confirm the results from fluorescence decays and numerical simulations, we record the fluorescence bursts intensities as the individual FRET constructs diffuse across the antenna hot spot, and compute the corresponding FRET efficiencies EFRET for every detected burst. The FRET efficiency is commonly defined as the probability of energy E

DOI: 10.1021/acs.nanolett.5b02535 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters transfer over all donor transition events.12,13 For the EFRET calculation from the fluorescence bursts, we carefully take into account the direct excitation of the acceptor by the laser light, the donor emission crosstalk into the acceptor channel, and the differences in the quantum yields and detection efficiencies for the donor and acceptor emission (see Methods section for details). Typical time traces are displayed is Figure 5a and Supporting Information Figure S8. Figure 5b,c summarize the FRET efficiency histograms for 6.8 and 10.2 nm donor− acceptor distances. A reduction of the average FRET efficiency is observed as the LDOS is increased from the confocal to the antenna case. Nonetheless, the efficiencies are always above zero, which substantiate the occurrence of FRET. Moreover, the histograms in Figure 5b,c clearly differ from the histogram obtained with the isolated donor (Supporting Information Figure S9), which is centered at EFRET = (0.4 ± 3.1)% in the absence of the acceptor and provides a reference for the zero FRET case. The evolution of the FRET efficiency is further evidenced by representing EFRET as a function of ΓDo (Figure 5d,e). The data points obtained from the burst intensity analysis are represented as filled gray markers. To confirm the trend, the FRET efficiency is computed independently from the lifetime data in Figure 3d as EFRET = ΓFRET/(ΓFRET + ΓDo) = 1 − (ΓDo/ ΓDA). These results are represented as color circles in Figure 5d,e and nicely overlap with the average FRET efficiency computed from the burst analysis. The higher LDOS in the antenna gap corresponds to an increase in ΓDo, which contributes to lower the apparent FRET efficiency EFRET = ΓFRET/(ΓFRET + ΓDo) by increasing its denominator ratio. This loss is partly compensated by the enhanced FRET rate in the nanogap, so the FRET efficiency reduction is less pronounced than would occur if the FRET rate was independent of the LDOS. Clearly, our data do not follow the evolution of EFRET with ΓDo expected if the FRET rate was independent of the LDOS (dashed lines in Figure 5d,e, computed using the expression EFRET = ΓFRET/(ΓFRET + ΓDo) and fixing ΓFRET to its value obtained for the confocal measurements). Thus, the observed FRET efficiency confirms the FRET rate enhancement in the gap antenna. As additional proof of the consistency of our results, we compute the FRET rate from the average FRET efficiency in the histograms and compare it to the data obtained from lifetime analysis (Supporting Information Figure S10). Both measurements converge toward similar results. Previous reports considered a planar mirror22,23 or a microresonator24,25 where the electric field does not vary notably below the spatial scale of a half-wavelength and the LDOS variations remain below 2-fold. In comparison, the 20 nm aluminum nanogap shows strong field gradients, confinement of light to nanoscale dimensions, and large 10-fold LDOS enhancement. These new conditions lead to FRET rate enhancement, as clearly demonstrated by our set of data obtained from two independent measurement methods and supported by numerical simulations. To push further the analysis, we quantify the antenna effect on all photokinetic rates for both the donor and the acceptor (see Table S4 and Figure S11 of the Supporting Information). Our approach fully takes into account the nonradiative decay losses to the metal for both dyes. All the photokinetic rates described in the Jablonski diagram Figure 1d are enhanced by the gap antenna, with slight variations between donor and acceptor radiative rates related to the different positions of the emission wavelengths relative to the antenna resonance. The analysis reveals that nonradiative

losses to the metal become quite significant in the case of the aluminum gap antenna, being the current limiting factor for further enhancement of the FRET efficiency. In the quest of photonic structures to enhance FRET, we have reported an important observation of enhanced energy transfer within a resonant nanogap antenna having a large LDOS. Our dedicated design of aluminum nanogap antenna has specific advantages. The dimer gap antenna isolates detection volumes of 220 zL (2000-fold below the diffractionlimited confocal volume) accompanied by a 10-fold LDOS enhancement. The surrounding box aperture blocks the background from the solution and enables single molecule detection at concentrations of 1 μM. The choice of an aluminum layer provides a resonance in the visible range together with a broadband response covering both the donor and the acceptor spectra. Importantly, all our experiments are conducted at the single molecule level on well-defined donor− acceptor pairs, while both the donor and acceptor emission are monitored. With all these specificities, we are able to provide a complete picture of FRET in a nanoantenna, conclusively observe up to 5-fold enhancement of the FRET rate and demonstrate experimentally that the Förster energy transfer rate is influenced by the LDOS in the 20 nm nanogap. The antenna compatibility with the detection of single molecules in water solution at physiological conditions is an additional advantage for biophotonic applications,8,9 extending single molecule FRET toward higher physiological concentrations and higher sensitivities. Methods. Antenna Fabrication. Nanoantennas are milled by focused ion beam (Zeiss Auriga 60 FIB-SEM, 1 nm resolution GEMINI SEM, equipped with Orsay Optics 2.5 nm resolution Cobra ion column) on 50 nm thick aluminum films deposited by thermal evaporation (Oerlikon Leybold Univex 350). The half-sphere diameter is 80 nm with gap size of 22 nm, whereas the surrounding aperture dimensions are 300 × 100 nm2. DNA Samples. Double-stranded DNA constructs of 51 base pairs length are designed with one Atto550 donor on the forward strand, and one Atto647N acceptor on the reverse strand. The distances between fluorescent labels are set such that the donor and acceptor are separated by 20 or 30 base pairs (corresponding to ∼6.8 and 10.2 nm separations respectively). As 10 base pairs make a complete turn on the DNA double strand, the choice of D−A separation as multiples of 10 base pairs avoids taking into account the complex threedimensional structure of DNA to estimate the D−A distance.13 The characteristic Förster radius computed for Atto550 and Atto647N in pure water is 6.5 nm. Labeled HPLC-purified DNA single strands are obtained from IBA (Göttingen, Germany), modified with the corresponding N-hydroxysuccinimidyl ester (NHS) donor and acceptor fluorophore derivatives of ATTO550 and ATTO647N. Fluorophores are covalently linked to an amino-C6-modified thymidine with NHS-chemistry via base labeling. The forward strand sequence is 5′-CCTGAGCGTACTGCAGGATAGCCTATCGCGTGTCATATGCTGTTDCAGTGCG-3′. The reverse strand sequence is 5′-CGCACTGAACAGCATATGACACGCGAT20AGGCTATCCT30GCAGTACGCTCAGG-3′. The reference sequences carrying only the isolated donor or acceptor are constructed with unlabeled complementary strand, respectively. The strands are annealed at 10 μM concentration in 20 mM Tris, 1 mM EDTA, 500 mM NaCl, 12 mM MgCl2 F

DOI: 10.1021/acs.nanolett.5b02535 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

fit the experimental data. The short lifetime contribution for the box aperture corresponds to less than 15% of the total counts and is found independent of the polarization orientation or the box size. Because this biexponential contribution was not observed for circular apertures,34,35 we relate this short lifetime contribution to LDOS increase occurring near the corners of the rectangular box aperture. For the antenna, the fluorescence signal is dominated by the contribution from the gap region, hence the minor effects near the box corners are no longer observable. Additionally, it is worth mentioning that the fluorescence decays found with gold structures are affected by a fast sub-5 ps decay originating from the gold photoluminescence. A positive element is that aluminum antennas are free of this effect;35 this further eases the decay analysis near the zero delay time. Finally, the FRET rate is obtained as ΓFRET = ΓDA − ΓDo = (1/τDA) − (1/τDo), where τDA and τDo are the donor lifetime in the presence and absence of the acceptor, respectively. The FRET efficiency is then deduced as EFRET = 1 − (ΓDo/ΓDA) = 1 − (τDA/τDo) . Numerical Simulations of the FRET Rate Enhancement. The enhancement of the energy transfer rate as a function of the acceptor’s position is obtained by calculating the ratio of the field intensity distribution |ED(rA)|2 created by the donor in the presence and absence of the nanoantenna. The relative increase of |ED(rA)|2 with the nanoantenna as compared to free space directly relates to a higher rate of energy transfer to the acceptor dipole. Computations are performed using finitedifference time-domain FDTD method (RSoft Fullwave software) with a mesh size of 1 nm. The antenna parameters are set to reproduce the fabricated devices, with a hemispherical shape of the 80 nm aluminum nanoparticle and 22 nm gap. The emission wavelength is 600 nm, and the aluminum permittivity is taken from ref 46. FRET Efficiency Analysis. For every fluorescence burst, the number of detected photons in the acceptor channel na and in the donor channel nd are recorded. The threshold for burst recognition is set to the sum of the mean plus one standard deviation of the summed trace of donor and acceptor channels. The FRET efficiency is then computed according to the formula

buffer, and by heating to 95 °C for 5 min followed by slow cooling to room temperature. Double stranded DNA stocks are diluted in a 10 mM HEPES-NaOH buffer, pH 7.5 (SigmaAldrich). Experimental Setup. To unambiguously quantify the nanophotonic influence on the FRET process, our experiments record simultaneously the donor and the acceptor emission photodynamics at the single molecule level, including reference cases without antenna or acceptor. This provides several independent measurements to characterize FRET by monitoring either the donor photodynamics acceleration in the presence of the acceptor or the relative intensities of the donor and acceptor fluorescence. Experiments are performed on a confocal inverted microscope with a Zeiss C-Apochromat 63x 1.2NA water-immersion objective. The excitation source is a iChrome-TVIS laser (Toptica GmbH) delivering 3 ps pulses at 40 MHz repetition rate and 550 nm wavelength. The laser beam has a waist of 300 nm at the focal spot of the 1.2NA objective (as determined by FCS experiments on free Alexa Fluor 647 dyes). The average excitation power is set to 10 μW, well within the linear regime for the excitation of the fluorescent dyes (see Supporting Information Figure S12 for power dependence). The excitation conditions correspond to 0.035 mW/μm2 or 3.5 kW/cm2 power density and 0.1 mJ/cm2 energy fluence per pulse. This weak excitation intensity sets a maximum temperature increase of 2 °C.44 Moreover, the metal layer acts as an efficient heat sink to shrink this temperature elevation.45 The laser excitation is filtered by a set of two bandpass filters (Chroma ET525/70 M and Semrock FF01550/88). Dichroic mirrors (Chroma ZT594RDC and ZT633RDC) separate the donor and acceptor fluorescence from the reflected laser light. The detection is performed by two avalanche photodiodes (Micro Photon Devices MPD5CTC with