Intrinsic luminescence blinking from plasmonic nanojunctions

Abstract

Plasmonic nanojunctions, consisting of adjacent metal structures with nanometre gaps, can support localised plasmon resonances that boost light matter interactions and concentrate electromagnetic fields at the nanoscale. In this regime, the optical response of the system is governed by poorly understood dynamical phenomena at the frontier between the bulk, molecular and atomic scales. Here, we report ubiquitous spectral fluctuations in the intrinsic light emission from photo-excited gold nanojunctions, which we attribute to the light-induced formation of domain boundaries and quantum-confined emitters inside the noble metal. Our data suggest that photoexcited carriers and gold adatom - molecule interactions play key roles in triggering luminescence blinking. Surprisingly, this internal restructuring of the metal has no measurable impact on the Raman signal and scattering spectrum of the plasmonic cavity. Our findings demonstrate that metal luminescence offers a valuable proxy to investigate atomic fluctuations in plasmonic cavities, complementary to other optical and electrical techniques.

Fig. 1. Blinking of metal photoluminescence (PL) in a single nanojunction.

a Schematic representation of a nanojunction, made of a gold mirror, a self-assembled biphenyl-4-thiol (BPhT) monolayer (g ~ 1 nm), and a faceted gold nanoparticle (d ~ 80 nm). Middle inset: simulation of electric field distribution in the nanojunction region. Lower inset: illustration of the luminescent nano-clusters or nano-domains forming under laser irradiation, which we invoke as the cause of PL fluctuations shown in d–f. b Schematic of the optical setup enabling three simultaneous types of measurement: PL under 532 nm (VIS) excitation, Raman scattering (R) under tunable near-infrared (NIR) excitation, and dark-field scattering (DF) under p- or s-polarized grazing angle white light (WL) illumination. BS beam-splitter, LP long-pass filter, C camera, S spectrometer. c PL spectrum of a single nanojunction averaged over the entire duration of panel d (orange curve) and DF scattering spectrum of the same nanojunction (gray curve, calibrated by the illumination spectrum). Labeling of the modes is detailed in Fig. 2g, h. d Time series of plasmon-enhanced PL (the color scale is saturated for better visibility of weak emission periods). Power density of the 532 nm laser: ~70 μW μm⁻², camera exposure time = 0.1 s, numerical aperture: 0.85, room temperature. e Individual examples of anomalous emission (referring to d) deviating from the typical baseline PL emission (shaded blue). f Time trace of the maximum PL intensity around the L01 mode, as marked by orange trace in d. The shaded blue area corresponds to the instrument-limited level of fluctuations, where I_PDF is the peak in the intensity probability density function (PDF) and σ is the standard deviation of the measurement noise at this signal intensity (see Supplementary Fig. 7).

Fig. 2. Multi-mode blinking: evidence for spatially localized fluctuating sources of emission.

a Fluctuating PL time trace from a nanojunction emitting from the L01 and S02 gap modes. Excitation power density at 532 nm: ~45 μW μm⁻², numerical aperture 0.85, exposure time: 0.1 s, room temperature. b Time series of peak PL intensities at the L01 (blue curve) and S02 (orange curve) resonances along with the Q factor of the L01 mode (gray-shaded area). c Examples of PL spectra with different Q factors (as fitted by Lorentzian functions), along with the typical baseline PL spectrum (blue-shaded area). d, e Distribution of the PL peak wavelength (d) and intensity (e) (relative to their time average denoted by brackets) around one resonance vs. the other, showing no correlations. f Distribution of relative PL intensity vs. Q factor for the L01 emission, showing a positive correlation. Individual spectra from c are highlighted with black, red, and blue circles. g, h Full-wave simulation (finite-element modeling) of the optical response of a faceted nanojunction. g Based on the surface charge distributions taken at resonance, the modes are identified as the lowest frequency Fabry–Perot-like transverse cavity mode S11, the dipolar bonding antenna mode L01, and the higher-order cavity mode S02, respectively. These modes feature distinct spatial distributions of their local photonic density of states (PDOS), which result in different far-field emission spectra (solid lines in h, offset for clarity) when a radiating point dipole is placed at the different locations shown by color-coded full circles in g. The yellow-shaded area in h corresponds to the calculated DF spectrum.

Fig. 3. Blinking PL with stable plasmon-enhanced Raman spectra.

a Time series of emission spectra acquired under dual excitation with 532 and 750 nm laser beams (first 120 s) with respective power densities ~200  and ~10 μW μm⁻², and then with 750 nm excitation alone (after 120 s). Camera exposure time = 1 s; numerical aperture: 0.95; room temperature. The color scale is saturated for better visibility of the fluctuations. b Time series of the PL intensity (cf. I_PL in c) and PL-subtracted Raman intensity (cf. I_R in c), normalized by their respective time-averages (〈I_PL〉 and 〈I_R〉). c Example of individual Raman+PL spectra (blue and orange curves) and time-averaged Raman spectrum under 750 nm excitation alone (120–180 s, gray area). The green and red-substituted points on the blue curve highlight how I_PL and I_R are defined.

Fig. 4. Blinking PL with stable dark field.

a Spectral time series from an individual nanojunction under sequential illuminations with 532 nm laser alone (PL), both 532 nm laser and white light (PL+DF, normalized by illumination spectrum) and white light alone (DF, normalized by illumination spectrum). The positions of maximum emission close to the L01 and S11 modes are shown as orange, blue, and purple traces for the PL, PL+DF, and DF regions, respectively. b, c Probability density functions (PDFs) for the relative L01 peak intensity (b) and wavelength shift (c) extracted from the PL (orange), DF (purple), and DF+PL (blue) regions in a. The DF+PL (blue) area in b are slightly offset away from the value of 1 to show it clearly. Experimental parameters: objective numerical aperture = 0.8; laser power density ~ 45 μW μm⁻²; white light is p-polarized; exposure time = 1 s, room temperature.

Fig. 5. PL blinking as a function of laser power and sample temperature.

a Temperature-dependent time series of PL intensity from a nanojunction, with the time-averaged PL spectrum shown in the inset. The PL signal was measured by a single photon counting module after spectral filtering (range shown in inset). Binning time, 1 ms. Excitation power density: ~50 μW μm⁻², b enlarged view of a revealing shorter blinking events. More data presented in Supplementary Fig. 16 suggest that no clear relationship exists between temperature and blinking statistics. c Probability density functions (PDFs) of peak PL intensity as a function of excitation intensity (room temperature) plotted against the re-scaled intensity δI_PL/σ to enable comparison. Here δI_PL represents the PL intensity deviation from the peak of the PDF; σ is the standard deviation of the measurement noise at this signal level. The PDFs are all centered around the averaged PL of the “OFF” state and normalized to the respective measurement noise (blue area). The corresponding time series can be found in Supplementary Fig. 17. d Autocorrelation function of the PL intensity trace at different excitation intensities, evidencing increased fluctuations at higher laser powers.