Supporting Information for. Electrical control of Förster energy transfer.

Transcription

1 1 Supporting Information for Electrical control of Förster energy transfer. Klaus Becker 1, John M. Lupton 1*, Josef Müller 1, Andrey. L. Rogach 1, Dmitri V. Talapin, Horst Weller & Jochen Feldmann 1 1 Photonics and Optoelectronics Group, Physics Department and CeNS, Ludwig- Maximilians-Universität, 8799 Munich, Germany Institute of Physical Chemistry, University of Hamburg, Grindelallee 117, 146 Hamburg, Germany 1. Energy transfer in the ensemble Figure S1: Chemical structure of the cyanine dye employed. Fig. S demonstrates that energy transfer can occur from nanocrystals to the dye in the ensemble at room temperature by considering the PL excitation spectrum of the blend in polystyrene. Upon detection in the spectral tail of the dye emission at 71 nm, the PL excitation spectrum closely follows both the dye absorption and the nanocrystal absorption, which are overlaid in Fig. S. A nanocrystal-only control sample showed no detectable emission at 71 nm.

2 1. 1. Absorption (a.u.) Nanorods Dye λdetect PL intensity (a.u.) Wavelength (nm). Figure S: PL excitation spectra of a sample containing nanocrystals and dye molecules dispersed in a polystyrene matrix at 3 K (emission detected at 71 nm). The occurrence of energy transfer is confirmed by the presence of both the nanocrystal and the dye absorption features in the PL excitation curve.. Electrical control of energy transfer Although energy transfer is readily verified in the ensemble, it will not occur between every adjacent nanocrystal and dye molecule. The reason for this is that, especially as the temperature is reduced, the absorption of the individual molecules narrows as does the emission of the individual nanocrystal donors. Explicit examples of this are given in Figs. C and F, which show that the single particle or single molecule spectra are substantially narrower than the ensemble spectra. This narrowing of the spectra on the single molecule level by overcoming energetic disorder and inhomogeneous broadening makes electrical control of Förster energy transfer possible. Fig. S3 shows calculations of the energy transfer rate in dependence of the mean Stark shift of the nanocrystal emission spectra. To do this, we approximated the absorption and emission spectra using Gaussians. The single dye molecule absorption spectra could

3 3 not be measured directly. However, we can use the single dye emission spectra to provide an upper limit for the linewidth of the dye absorption of 7 nm at 5 K. Fig. S3A shows the fit to the ensemble emission of nanocrystals and absorption of the dye. The dye spectrum is readily described by two Gaussian curves, which have their physical origin in the inhomogeneously broadened zero-phonon transition band and a strong vibronic replica offset by ~16 cm -1 due to characteristic carbon-carbon modes. We can model the transition of the absorption from the ensemble to the single molecule by simply reducing the width of these two Gaussians describing the absorption. Likewise, the nanocrystal emission linewidth is set to the experimental value of 1.6 nm at 5 K. This is shown in panels C and E.

4 4 A B 1 Norm. PL ; Norm. absorption C E D F Energy transfer rate (a.u.) Wavelength (nm) Stark shift (nm) Figure S3: Absorption and emission of nanocrystals (blue) and dye molecules (red) in the ensemble (A and dotted lines) and on the single particle level (C, E) approximated by Gaussians. The single particle spectra are much narrower than the ensemble spectra, which are inhomogeneously broadened because every individual particle or molecule has a different transition energy. The arrows indicate the typical single particle Stark shift of 1 nm. The energy transfer rate is calculated from the spectral overlap of donor (nanocrystal) and acceptor (dye). Whereas only a small modulation is calculated for the ensemble (B), depending on the relative position of donor and acceptor with respect to each other at zero field, energy transfer can be fully switched on (panel D) or off (panel F) within a spectral shift of the donor of 1 nm. Note that the calculation for the ensemble is purely

5 5 hypothetical, as it is not possible to induce such a large Stark shift in the ensemble due to the different random orientations of the individual nanocrystals. The energy transfer rate is proportional to the spectral overlap between nanocrystal emission (donor) and dye absorption (acceptor) according to K 1 R D A FD ( ω) σ A ( ω) 6 dω where K D A is the transfer rate from donor to acceptor, R is the separation between donor and acceptor, F D is the donor fluorescence spectrum and σ A is the absorption spectrum of the acceptor in dependence of photon frequency ω. As we are only interested in how a mean change in donor fluorescence spectrum will affect the energy transfer rate, we need not consider R. Fig. S3B shows the calculated energy transfer rate in dependence of a mean Stark shift of the nanocrystal emission spectrum. Note that this calculation is purely hypothetical: in the ensemble, all nanocrystals have different random orientations so that the Stark effect of the individual nanocrystals averages out and no spectral shift is observed. Assuming a Stark shift of such a broad spectrum were observed, the change in energy transfer efficiency would be marginal with at most 65 % for an exceptionally large spectral shift of 5 nm. The situation is very different once the spectral width of donor and acceptor becomes substantially smaller than the achievable Stark shifts. Typical spectral shifts (e.g. Fig. 3A) are of order 1 nm (indicated by the black arrows in Fig. S3). Fig. S3C shows the Gaussian

6 6 describing the nanocrystal emission spectrum and the proposed single dye molecule absorption spectrum. The calculated energy transfer rate depends sensitively on the Stark shift and thus on the magnitude of the electric field applied: within a shift of 1 nm energy transfer can be completely switched on and off. This case, in which zero energy transfer occurs at zero shift and a ~1 nm Stark shift leads to complete energy transfer corresponds to the experimental situation observed in Fig. 3B. The dotted line shows the hypothetical case for the ensemble for comparison. The scenario observed in the experiment and shown in Fig. 3C is reconstructed in the calculations in panels E and F. Here, the nanocrystal emission is in resonance with the vibronic replica of the dye absorption at zero field: the energy transfer efficiency is highest for zero Stark shift. Within a 1 nm spectral shift of the donor, however, energy transfer can be fully suppressed as shown in panel F in complete agreement with experiment. The transition linewidth of donor and acceptor therefore crucially influences the electrical controllability of Förster energy transfer. In addition, these simple calculations illustrate that spectral overlap determined from ensemble measurements only provide a statistical average over events occurring on the single molecule level. Good spectral overlap in the ensemble may be generated artificially by increasing energetic disorder of the sample, e.g. by increasing the size distribution of the nanocrystals. On average, however, this will actually decrease the overall energy transfer efficiency from donor to acceptor as the number of microscopic donor-acceptor pairs with spectral overlap will decrease when energetic disorder (the inhomogeneous distribution) increases with respect to the single particle/molecule linewidth.

7 7 Both the molecules and the nanocrystals display spectral diffusion, a random spectral jitter of the emission energy with time. This jitter is superimposed on the spectral shift induced by the electric field. The narrower the electronic transitions are, the greater the accuracy required from the Stark effect to achieve resonance. The random jitter of the transitions can mean that dye and nanocrystal randomly drift in and out of resonance. As the temperature of the sample is increased, the transition linewidths of donor and acceptor increase. The random spectral jitter then becomes less crucial. The experiment therefore requires a tradeoff between FRET modulation depth achievable and insensitivity to spectral diffusion. Fig. S4 shows calculations of the energy transfer rate at three different temperatures, performed using Gaussians describing emission and absorption as above. The transition linewidths were set to the experimental values observed in emission (nanocrystals:.5 nm (5 K), 1.6 nm (5 K), 4.5 nm (15 K) ; dye molecules: 4.3 nm (5 K), 7 nm (5 K), and ~1 nm (15 K)). Clearly, for the temperature of 5 K used in the experiment complete modulation of energy transfer is still achievable for a Stark shift of 1 nm. Energy transfer rate (a.u.) 5 Ensemble T=5 K T=5 K T=15 K Stark shift (nm) Figure S4: Calculated energy transfer rates between a single nanocrystal and a single dye molecule in dependence of the nanocrystal Stark shift for different temperatures. The hypothetical case of the ensemble is also shown.

8 8 3. Evidence for energy transfer from single nanocrystals to single molecules Having provided a discussion of the mechanisms of the effect, in the following we present further experimental evidence for energy transfer from single nanocrystals to single molecules. Energy transfer excitation of single dye molecules by individual nanocrystal nanoantennae is evidenced by the fact that the number of single emitting dye molecules per unit area observed within the spatially continuous dye film scales with nanocrystal concentration. The PL transient of a single dye molecule excited by energy transfer from a nanocrystal is shown in Fig. S5 and exhibits strong intensity fluctuations. This is a consequence of three effects: the intrinsic dye blinking, the nanocrystal blinking, and the nanocrystal spectral diffusion, which randomly modifies the dye-nanocrystal coupling and thus the energy transfer efficiency and the resulting dye brightness. The single-step irreversible photobleaching which occurs at the end of the trace provides evidence for the presence of only one emitting dye molecule. 5 PL Intensity (a.u.) Time (s) Figure S5: Temporal evolution of PL intensity of a single dye molecule excited by energy transfer from a single nanocrystal. Strong fluctuations of the dye s PL intensity are observed. The single-step photobleaching at t=6 s is strongly suggestive of emission from a single dye molecule.

9 9 Fig. S6 shows a spectral fluorescence trace of a single molecule excited by energy transfer from a nanocrystal absorber. Besides exhibiting fluorescence intermittency (blinking), the molecular emission also shows spectral diffusion, i.e. shifts to higher and lower wavelengths. Whereas the blinking may be a characteristic of the nanocrystal superimposed on the dye emission due to time variable energy transfer, the spectral diffusion is a clear signature of single molecule emission. Wavelength (nm) Time (s) Figure S6: Spectral trace of a single dye molecule photopumped by an adjacent nanocrystal. The trace exhibits blinking and spectral diffusion, which are both signatures of a single emitter. Fig. S7 illustrates the evolution of PL intensity of dye molecules and single nanocrystals blended into a film with a comparatively high concentration of dye molecules corresponding to a typical intermolecular spacing of ~6 nm. Stepwise photobleaching of multiple dye acceptors excited by a single nanocrystal donor is seen in Fig. S7A. Three steps are identified in the histogram in panel B, suggestive of three molecules coupling to the nanorod in this particular case. Fig. S7C demonstrates the recovery of a single nanocrystal donor surrounded by several dye molecules. As the acceptors photobleach with

10 1 time, the PL of the more photostable nanocrystal donor increases. The strong fluctuations in PL with time, which are untypical for these rather stable emitters, are a consequence of varying coupling strength to the adjacent dye acceptors. Three maxima are observed in the histogram in panel D, which suggest that in this case at least two acceptors couple to the single donor. # of events A B Dye emission PL Intensity (a.u.) 4 C NC emission D Time (s) 1 3 # of events Figure S7: (A) Temporal evolution of dye emission excited by a single nanocrystal embedded in a dye film with comparatively small (~6 nm) average intermolecular spacing. Stepwise photobleaching arises due to sequential bleaching of acceptors. The steps in PL intensity are visualized in the histogram in (B). (C) shows the typical recovery of single nanocrystal emission in the blend. As the surrounding dye molecules bleach, the nanocrystal emission increases in intensity due to the lack of acceptors. The strong fluctuation in intensity demonstrates the temporal changes in energy transfer efficiency and thus PL quenching of the nanocrystal by the acceptors. The histogram in (D) illustrates the steps in nanocrystal intensity.

11 11 In order to provide further evidence for single molecule emission, we considered the polarization anisotropy of the nanocrystal pumped dye emission as shown in Fig. S8. The emission is evidently fully linearly polarized. PL Intensity Polarizer angle ( ) Figure S8: Polarisation anisotropy of the dye emission photopumped by a single nanocrystal. The linear polarisation is strongly suggestive of emission occurring from a single dye molecule. A further question relates to the orientation of the molecular dipole with respect to the transition dipole of the nanocrystal. Fig. S9 illustrates that the average polarization of single dye molecules excited by nanocrystal nanoantennae preferentially lies in the polarization plane of the exciting laser (referenced to a polarizer angle of in this case). As the elongated nanocrystals absorb and emit linearly polarized light, excitation energy is apparently preferentially passed on to molecules whose dipoles lie parallel to those of the nanorods. Fig. S9 shows an average over 88 single dye molecules. In contrast, direct excitation of the dye background using linearly polarized laser light leads to only weakly polarized PL, as shown in Fig. S1. This residual polarization is due to the polarization memory of the individual dye molecules and is not dominated by the instrument response.

12 1 PL Intensity (a.u.) Polarizer angle ( ) Figure S9: Polarization dependent PL intensity averaged over 88 single dye molecules excited via energy transfer from single nanocrystals. The strong anisotropy of the emission which preferentially lies parallel to the plane of polarization of the exciting laser ( ) demonstrates the polarization dependence of direct excitation of nanocrystals and consecutive energy transfer towards dye molecules. The preference in polarization due to the anisotropic nanocrystal transition dipole is maintained after the energy transfer step to the dye. PL Intensity (a.u.) Polarizer angle ( ) Figure S1: Polarization dependent PL intensity of an ensemble of dye molecules dispersed in polystyrene and excited directly with linearly polarized light at 514 nm. Compared to Fig. S9, only a weak polarization dependence is observed, demonstrating an isotropic dipole configuration. The individual dye molecule spectra are distinct from the ensemble and exhibit a scatter in energy due the inhomogeneous broadening visible in the absorption. Fig. S11 shows the

13 13 scatter of peak positions extracted from single dye molecules at 5 K together with the ensemble fluorescence spectrum. Three typical single molecule spectra are also shown. 5 Counts 15 1 Intensity Wavelength (nm) Figure S11: Scatter of peak positions of individual dye molecule spectra. The ensemble spectrum is superimposed (black line), as are three typical single molecule spectra at 5 K. The dye molecules do not exhibit any detectable response to an electric field. Fig. S1 shows the PL spectrum of molecules dispersed in a polystyrene matrix under application of a lateral electric field. No change in shape or intensity of the PL spectrum is observed, demonstrating that the strong PL modulation with field in the case of the dye nanocrystal couple is a direct consequence of the nanocrystal photophysical properties being imparted on the dye system.

14 14 Integrated PL Intensity V= V V=3 V PL Intensity Wavelength (nm) Time (s) V 3 V Figure S1: Influence of the electric field on the polystyrene:dye blend ensemble emission. The spectrally integrated PL intensity is not affected by an applied electric field of approx. 3 kv/cm. The inset shows the emission spectra, which do not change upon application of a field.

15 15 Figure S13: Fluorescence microscope images of the spectrally resolved PL of pure nanocrystals (panels A to C) and a blend of nanocrystals and dye molecules (panel D) at 5 K for high nanocrystal concentrations (interparticle separation approx. 5 nm, i.e. the same concentration as employed in Figs. 3B and C) in a polystyrene matrix. The y-axis corresponds to the spatial position on the sample.

16 16 Finally, Fig. S13 confirms that the concentration of nanocrystals used has no direct effect on the spectral properties or the nanocrystal QCSE. Panels A-C show a nanocrystal-only sample, in which the nanocrystals are dispersed at a concentration identical to that of the nanocrystals in the sample for electrically controllable energy transfer (cf. Figs. 3B&C). Due to the high nanocrystal concentration many single particle spectra are superimposed in this visualization. Panels A to C demonstrate the influence of an applied electric field (bias -3 V, V, and 3 V) on the emission. The QCSE is visible for some nanocrystals in the emission (marked by the white circles given as a guide to the eye). Importantly, no distinct emission from the nanocrystals is observed above 67 nm for any applied bias. This allows a definite exclusion of the possibility of far-red emissive nanocrystal aggregates appearing at elevated particle densities. For comparison, the emission from a single dye molecule excited via energy transfer from a nanocrystal is seen in the raw data in panel D. In this case the emission of the dye molecule is centred at 675 nm.