Photoluminescence Enhancement, Blinking Suppression, and Improved Biexciton Quantum Yield of Single Quantum Dots in Zero Mode Waveguides

Abstract

The capability of quantum dots to generate both single and multiexcitons can be harnessed for a wide variety of applications, including those that require high optical gain. Here, we use time-correlated photoluminescence (PL) spectroscopy to demonstrate that the isolation of single CdSeTe/ZnS core-shell, nanocrystal quantum dots (QDs) in Zero Mode Waveguides (ZMWs) leads to a significant modification in PL intensity, blinking dynamics, and biexciton behavior. QDs in aluminum ZMWs (AlZMWs) exhibited a 15-fold increase in biexciton emission, indicating a preferential enhancement of the biexciton radiative decay rate as compared to the single exciton rate. The increase in biexciton behavior was accompanied by a decrease in blinking events due to a shortening in the dark state residence time. These results indicate that plasmon mediated enhanced decay rates of QDs in AlZMWs lead to substantial changes in the photophysical properties of single quantum dots, including an increase in biexciton behavior.

Graphical Abstract

Quantum dots (QDs) have been extensively studied due to their potential application in a variety of fields such as optoelectronics^1–4, bio labeling, and bioimaging^5–6. One of the exceptional properties of QDs is their ability to populate both single excitonic (X) and multiexcitonic (MX) energy levels and subsequently undergo either single-photon or multiple photon emission⁷. The ability to control the extent of single and multiphoton emission is of interest for a number of applications. For example, in the field of quantum communication where single-photon emission is desirable, QDs with low MX emission efficiency can be utilized as a single-photon source^8–10. Multiphoton emission in QDs has shown promise for applications in areas where materials of high optical gain are desired, such as lasing^10–11. Biexciton-exciton (BX-X) cascade which leads to the emission of two photons at a time from a single QD can potentially be applied for quantum computing¹².

Despite the potential of QDs to be utilized for a wide range of applications, they suffer from two significant drawbacks. (1) They exhibit fluorescence intermittency (blinking) characterized by frequent transitions between photoluminescence (PL) intensity on and off states¹³. PL blinking poses severe challenges to the utility of QDs as a single-photon source due to the resulting interruption in photon supply^14–15. (2) The MX quantum yield of QDs is also limited due to nonradiative auger recombination¹⁶. This energy loss is due to the recombination energy of an e-h pair being used to excite an electron or hole. Auger rates in MX systems can be 10¹-10² times faster than radiative recombination rates, which leads to low emission quantum yields¹⁷. PL blinking is potentially due to QD charging¹⁸. The excitation of a neutral QD generates an e-h pair leading to photon emission and the observed on-state in the fluorescence intensity. However, the electron or hole can be trapped in the surrounding matrix leading to a charged QD core. Further excitation of the charged QD leads to either a negatively or positively charged exciton. This triggers nonradiative auger recombination given rise to the observed non-emissive off-states¹⁹. When the charged QD is neutralized radiative recombination is restored leading to a return to the emissive state. QDs can also exhibit multiple emission states in their PL intensity, instead of just two-level PL blinking¹⁸. Several models have also been reported to describe this phenomenon. According to the multiple recombination center (MRC) model, carriers potentially recombine non-radiatively with centers on the QD surface^{18, 20}. This process then competes with electron-hole radiative recombination. The activation and deactivation process of the recombination centers drives the PL intensity over time^{18, 20}. Other models suggest that hot carriers (electron or holes) can be trapped by surface states losing their energy through nonradiative recombination and the loss of PL intensity. It is also possible that some carriers can avoid these processes and cool down to the band edge and radiatively recombine leading to a contribution to the observed PL intensity^{18, 21}. Passivating surface traps or moderating auger recombination rates should then reduce PL blinking and improve quantum yields¹⁸. A reduction in blinking has been observed when the QD surface was passivated with organic ligands^22–23. Auger recombination efficiency can be reduced by growing a thick shell layer on a small QD core^24,25. Core/shell based QDs exhibit reduced auger rates with almost non-blinking PL intensity²⁶ and improved optical gain²⁴.

An alternative approach to limit blinking utilizes the coupling of QDs to plasmonic nanostructures²⁵. In this approach, the inherent QD auger rates are not affected. Rather, radiative recombination pathways become more efficient due to interactions between plasmons and QDs²⁵. Enhanced radiative decay rates compete with auger rates or any other associated nonradiative rates that are responsible for fluorescence intermittency and that normally reduce MX emission quantum yield. Plasmonic nanostructures are capable of enhancing the local electric field due to plasmon resonances, which can ultimately both increase the excitation intensity and alter the radiative decay rate of the QD²⁷. While both $\text{X}$ emission and $\text{BX}$ emission can be modified due to coupling with plasmons, the extent of the enhancement is different. $\text{BX}$ emission is enhanced at a higher rate than $\text{X}$ emission because $\text{BX}$ excitation intensity is proportional to the fourth power of the local electric field while $\text{X}$ excitation intensity is modified as the square of the field^{10, 28–29}. Several previous studies have utilized nanostructures for PL intensity enhancement^30–34, blinking suppression^35–36 and improving biexciton quantum yield (${\text{ϕ}}_{\text{BX}}$)^{14, 17, 37–39}. The magnitude of the changes in the optical and photophysical properties of QDs depends on a wide range of factors including on the size and shape^35–36, metal composition^{17, 37} and orientation of the nanostructures⁴⁰. Zero mode waveguides (ZMWs), which are characterized by nanoscale holes (typically <250 nm in diameter) in thin metal films, are an excellent tool for single-molecule imaging^41–42. Like other nanostructures, ZMWs can be used as a nanoantenna, creating an enhanced local electromagnetic field within the apertures⁴³. ZMWs have the potential to modify the optical and photophysical properties of QDs. Like other nanostructures, the plasmonic effect depends on the size, shape, and metallic composition of the ZMWs as well as the spatial position of the QD with regard to the metal walls⁴⁴. Fluorophores in close proximity to the wall of the ZMW experience maximum excitation intensity enhancement⁴³. Here, we investigated plasmon assisted modification of the optical and photophysical properties of single QDs immobilized in ZMWs.

A schematic diagram of the experimental setup for these studies is shown in Figure 1a. AlZMWs with 200 nm in diameter (Figure S1), were fabricated according to published protocols in 100 nm metal films^43–45. Electron beam lithography (EBL) and lift-off techniques were utilized to fabricate the ZMWs as detailed in the Supporting Information. Figure 1b shows the SEM image of an array of AlZMWs and a single AlZMW at higher magnification. Commercially available core/shell QDs (Qdot 705 ITKTM carboxy quantum dots, Thermofisher) were used for these studies. The core of the QDs is composed of CdSeTe and the shell contains ZnS. The pyramidal shaped QDs has an average size of ~ 8 × 13 nm. The absorption (Figure S2a) and PL spectrum (Figure S2b) of bulk QDs are shown in Supporting Information. The PL maximum was found to be at 708 nm with a FWHM of 74 nm. Single QDs were immobilized on glass and in AlZMWs by using EDC (1-ethyl-3-(3-dimethyl aminopropyl) carbodiimide) as a linker (Figure S3).

Figure 1.

Schematic of the setup used for photophysical studies of single QDs in ZMW. (a) Single QDs immobilized in ZMWs are excited with pulsed laser (532 nm) and the filtered excitation laser is focused to the single QDs by using an oil immersion objective. PL emission from the single QDs passes through the pinhole which is then sent to APDs and finally to TCSPC by using suitable emission filters and the beam splitter. (b) SEM image of an array of ZMWs and a single ZMW.

A custom-built confocal microscope was used to image single QDs. To understand the influence of ZMWs on QD exciton behavior, we studied the PL intensity, decay rates, blinking dynamics, and the photon correlation studies of QDs isolated in AlZMWs. QDs immobilized on a glass substrate was used as a reference. To determine the distribution of QDs, we raster scanned an area of 50μm x 50μm on a glass surface and on AlZMWs. Figure 2a and figure 2c show the raster scanned image of the QDs on glass and in AlZMW, respectively. Only a fraction of the AlZMW wells is occupied with QDs. Among the total of 196 possible wells with a 3.5μm pitch between the wells in the 50μm x 50μm area, only 35 wells are occupied with QDs which is ~20% of the total wells. Based on a Poisson distribution of QD occupancy, having less than 37% of the wells containing a QD is an indication of a low probability of dual occupancy and high probability of single particle level isolation^46–47.

Figure 2.

PL intensity enhancement and blinking suppression of single QDs in ZMW. (a) Raster scanned image of the QDs immobilized on a 50μm x 50μm area of glass substrate. (b)PL intensity time trace of a single QD immobilized on glass, with the bimodal distribution of the PL intensities representing on/off blinking states (top panel), the blue straight line is used to distinguish the on/off blinking events. (c) Raster scanned image of the QDs immobilized in a 50μm x 50μm area of AlZMW. (d) PL intensity time trace of a single QD immobilized in AlZMW, with almost unimodal distribution of the PL intensities signifying blinking suppression (bottom panel), the blue straight line is used to distinguish the on/off blinking events. (e) Average PL intensity of individual QDs immobilized on glass (black bars) and in AlZMW (red bars), single QDs in AlZMWs exhibit enhancement in PL intensity.

We acquired the PL intensity of individual QDs on glass and in AlZMWs using the same optical setup. Figure 2b and figure 2d show PL intensity time traces of single QDs on glass and in AlZMWs. The PL intensity of QDs in both cases exhibits significant intermittency characterized as a transition between states of high emissivity and the non-emissive background level. Discrete blinking of PL intensity between on/off state signifies single QD behavior^{10, 28}. However, it is also notable in Figure 2d that the PL intensity of the QD in AlZMW is larger, and blinking is suppressed which can be observed as almost unimodal distribution of the photon count histogram in comparison to the bimodal distribution of the photon count histogram of the QD isolated on glass substrate. On average, the PL intensity of QDs in AlZMWs was ~2.5 times higher than QDs on glass (Figure 2e). The PL intensity enhancement observed for QDs in AlZMWs might be due to an increased local electric field through plasmon resonance, which leads to an enhancement in the excitation intensity. However, plasmon assisted improvement in the radiative decay rates of the single QDs and the collection efficiency of the emission might also contribute to their PL intensity enhancement in ZMWs. To gain insight into plasmon mediated alterations in the decay rate of single QDs, we acquired the PL lifetime of isolated QDs on glass and in AlZMW (Figures 3a and 3b). PL decay of QDs on glass best fits with a bi-exponential decay with a longer component that averaged ${\text{τ}}_1=60\pm 20\text{ns}$ (Figure S4a) and a faster component that averaged ${\text{τ}}_2=5\pm 4\text{ns}$ (Figure S4b). The longer and shorter PL lifetime components presumably represent the neutral and the charged (trion) $\text{X}$ state of single QDs, respectively. The average amplitude for ${\text{τ}}_1(60\pm 20\text{ns})$ and ${\text{τ}}_2(5\pm 4\text{ns})$ was found to be 50% for each, which resulted in an overall weighted average PL lifetime of $30\pm 10\text{ns}$ (Figure 3c). On the other hand, PL decay of QDs in AlZMWs best fit with a triexponential decay of ${\text{τ}}_1=35\pm 9\text{ns}$ (Figure S5a), ${\text{τ}}_2=8\pm 2\text{ns}$ (Figure S5b), and ${\text{τ}}_3=0.9\pm 0.2\text{ns}$ (Figure S5c) with an average amplitude of 10%, 30% and 60% respectively. In addition to the 2 components associated with $\text{X}$ decay, the sub-nanosecond decay component is attributed to $\text{BX}$ decay. Taking the weighted average based on the amplitude of all the decay components, the average PL lifetime of QDs in AlZMWs was 6 ± 3 ns (Figure 3d). Thus, the decay rate of QDs in AlZMWs is enhanced by ~5 times. However, the enhanced decay rate might be due to radiative or nonradiative pathways. If the enhancement in the decay rate is due to a radiative pathway, then the photoluminescence quantum yield (PLQY) should be improved. Similarly, the PLQY of QD should be quenched if the increase in the decay rate is from a nonradiative decay pathway. At low excitation intensities, enhancement in PL intensity (${\text{η}}_{\text{PL}}$) is the product of three contributing factors- excitation intensity enhancement (${\text{η}}_{\text{exc}}$), an improvement in quantum yield (${\text{η}}_{\text{ϕ}}$), and an increase in collection efficiency of the emission (${\text{η}}_{\text{k}}$)⁴⁸. Thus, we can write ${\text{η}}_{\text{PL}}={\text{η}}_{\text{exc}}{\text{η}}_{\text{ϕ}}{\text{η}}_{\text{k}}$. Considering that the experimentaly determined ${\text{η}}_{\text{PL}}=2.5$, is already known, we can determine ${\text{η}}_{\text{ϕ}}$ by quantifying ${\text{η}}_{\text{exc}}$ and ${\text{η}}_{\text{k}}$. Excitation enhancement (${\text{η}}_{\text{exc}}$) can be estimated by calculating the distribution of the near field around a dipole in the presence or absence of a ZMW, at a given excitation wavelength. We utilized the finite difference time domain (FDTD) simulations to calculate the near field distribution in an AlZMW at 532 nm (Figure 3e). Averaging the near field enhancement over the entire ZMW hole gives approximately a 2-fold enhancement in the local electric field in AlZMWs. Alternatively, plasmon mediated enhancement in excitation intensity (${\text{η}}_{\text{exc}}$) of QDs can be computed from the amplitude of $\text{X}\left(a_x\right)$ and $\text{BX}\left({\text{a}}_{\text{bx}}\right)$ decay lifetimes where the amplitude ratio of these two decays represents excitation intensity enhancement, regardless of the emission process⁴⁹. By calculating the amplitude ratios (${\text{a}}_{\text{b}x}/{\text{a}}_{\text{x}1}+{\text{a}}_{\text{x}2}$) of single QDs, which reveal an average value of 2.0 ± 0.7 representing the excitation field enhancement in AlZMW. ${\text{a}}_{\text{x}1}$ and ${\text{a}}_{\text{x}2}$ are denoted by the amplitude of neutral and charged $\text{X}$ decay. These results are almost similar to those obtained from FDTD simulations (Figure 3f). Plasmonic nanostructures can modify the radiation pattern or angular distribution of the emission intensity of the emitters, in addition to their ability to alter the excitation and decay processes of the emitters^50–52. We anticipate that the enhancement due to an increased collection efficiency will be minimal⁵³. Thus, we adopted a value for the enhancement in collection efficiency (${\text{η}}_{\text{k}}$) of 1.5 as has been used for similarly sized ZMWs⁴⁸. Nanostructures hold both localized and also propagative surface plasmon modes⁵⁴. Radiation coupled to propagative modes is lost, as they do not reradiate to the far-field. The propagative mode character of a single nanopore can be determined by calculating the relative magnitude of the real and imaginary part of the propagation constant ($\text{γ}$)^{50, 55}. A larger value of the real part or smaller value of imaginary part of $\text{γ}$ corresponds to greater propagation^{50, 55}. In general, the real part of $\text{γ}$ increases with an increasing diameter of the nanoaperture^{50, 55}. The 200 nm diameter of our ZMWs is considerably smaller in comparison to the NIR emission of the QDs used here indicating insignificant interaction with the surface plasmon polaritons of AlZMWs. Thus, a 2-fold enhancement in excitation intensity and 1.5-fold enhancement in collection efficiency indicates that ${\text{η}}_{\text{ϕ}}$ is decreased in AlZMWs. Quenching of PLQY of QDs in AlZMWs is not surprising because the inherent PLQY of QDs is high (~70%). We can verify the improvement in radiative/non-radiative decay rates using the following expression (adopted from Masuo et.al 2016) where the enhancement in radiative and nonradiative decay rates of QDs in AlZMWs can be expressed as⁷ (detailed equations can be found in Supporting Information);

$${\text{η}}_{\mathrm{r}}=\frac{{\mathrm{I}}_{\mathrm{z}}}{{\mathrm{I}}_{\mathrm{g}}}\frac{{\text{τ}}_{\mathrm{g}}}{{\text{τ}}_{\mathrm{z}}}\frac{1}{{\text{η}}_{\mathrm{e}\mathrm{x}\mathrm{c}}}\frac{1}{{\text{η}}_{\kappa }}$$ (1)

$${\text{η}}_{\mathrm{n}\mathrm{r}}=\frac{\left(\frac{{\text{τ}}_{\mathrm{g}}}{{\text{τ}}_{\mathrm{z}}}\right)-{\text{ϕ}}_{\mathrm{g}}{\text{η}}_{\mathrm{r}}}{\left(1-{\text{ϕ}}_{\mathrm{g}}\right)}$$ (2)

Figure 3.

Shortening in PL lifetime and enhancement in excitation intensity of single QDs in ZMW. (a) PL lifetime decay of a single QD immobilized on glass, best fitted with a biexponential decay. (b) PL lifetime decay curve of a single QD immobilized in AlZMW and the curve best fit with a triexponential decay giving rise to an additional sub-nanosecond PL decay component attributed to biexciton decay. (c) PL lifetime (weighted average) histogram of the single QDs immobilized on glass. (d) PL lifetime (weighted average) histogram of the single QDs immobilized in AlZMW exhibiting a clear reduction in the PL lifetime in compare to the QDs on glass as shown in Figure 3c. (e) Near electric field distribution in an AlZMW, estimated by FDTD. The color bar represents the magnitude of enhancement in near electric field inside the AlZMW which is calculated by dividing the near electric field in AlZMWs with that of the near electric field in glass. The color bar is drawn in logarithmic scale (f) Average enhancement in excitation intensity by using FDTD and from the ratio of the amplitudes of biexciton and exciton lifetimes (${\text{ab}}_{\text{x}}/{\text{a}}_{\text{x}}$).

Where, ${\text{η}}_{\text{exe }}$, ${\text{η}}_{\text{r}}$, ${\text{η}}_{\text{nr}}$ and ${\text{η}}_{\text{κ}}$ represent enhancement in the excitation intensity, radiative decay rate nonradiative decay rate and collection efficiency respectively; $\text{I}$, $\text{τ}$ and $\text{ϕ}$ refer to the PL intensity, the PL lifetime and PLQY of QD, respectively. The subscript $\text{z}$ and $\text{g}$ represent the QD in AlZMW and on glass, respectively.

Using the obtained experimental value of the PL intensity enhancement factor, ${\text{η}}_{\text{PL}}=2.5$, the PL lifetime enhancement factor, $\frac{{\text{τ}}_{\text{g}}}{{\text{τ}}_{\text{z}}}\sim 5$, the excitation intensity enhancement factor ${\text{η}}_{\text{exc}}\sim 2$ and enhancement in collection efficiency ${\text{η}}_{\text{k}}=1.5$ the enhancement factor for the radiative decay rate is found to be ${\text{η}}_{\text{r}}\sim 4.3$. By considering the PLQY of QDs in aqueous solution as ${\text{ϕ}}_{\text{g}}\sim 0.7$, the enhancement factor for the nonradiative decay rate is found to be ${\text{η}}_{\text{nr}}\sim 7.0$.

The enhancement in ${\text{η}}_{\text{nr}}$ is most likely due to non-radiative energy transfer from the QDs to the metal surface of the AlZMWs. However, an overall enhancement in total decay rate has the potential to modify blinking dynamics by competing with and lowering the probability of transitions into states responsible for off-state blinking. To investigate if enhancement in total decay rates depressed the frequency of blinking of single QDs in AlZMW, we studied the distribution of the on and off characteristics of individual QDs immobilized on glass and in AlZMW. On/off-state events in the PL intensity trace of single QDs were identified by setting up a threshold PL intensity value, which is shown as the straight blue line in Figures 2a and 2b. PL intensity values lying above this threshold value are designated as on-state events, whereas PL intensity values lower than this threshold are considered as off-state events.

Figures 4a and 4b show the distributions of on/off state blinking events of a representative single QD on glass and in AlZMW, respectively. Distribution of on/off state events fit well (${\text{r}}^2\ge 0.99$) with a power equation; $\text{N}(\text{t})={\text{a}}^{\ast }{\text{t}}^{-\text{α}}$, where $\text{N}(\text{t})$ is the number of blinking events at time t and α is the power coefficient. The power coefficient values for on/off-state blinking events (${\text{α}}_{\text{on}}/{\text{α}}_{\text{off}}$), are found to be within the range of 1.0 – 2.0, which signifies the discrete blinking behavior of single QDs⁵⁶ both on glass and in AlZMW. However, the relative value of the (${\text{α}}_{\text{on}}/{\text{α}}_{\text{off}}$), of a single QD is also the determinant of the extent of on/off-state blinking events in that QD. Thus, the larger ${\text{α}}_{\text{off}}(1.95)$ value in comparison to the ${\text{α}}_{\text{on}}(1.20)$ of a QD in an AlZMW indicates the faster off-state blinking events correspond to lower off-state residence in AlZMWs. Likewise, similar values for the coefficients (${\text{α}}_{\text{on}}/{\text{α}}_{\text{off}}$), (1.44/1.36) for QDs on glass indicate a nearly equal share of on/off-state duration in the PL intensity time trace. To further elucidate the effect of ZMWs on the blinking dynamics of single QDs, we computed the average duration of off-state events on glass and in AlZMWs. The average length of off-state events was calculated by dividing the total time a QD spent in an off-state by the total number of off-state events. Figures 2c and 2d show the off-state blinking duration of QDs immobilized on glass and in AlZMW. The average off-state blinking duration of QDs is reduced by a factor of 3 from 140.0 ± 5.0 ms on the glass to 45.0 ± 2.0 ms in AlZMWs. The results provide further evidence that plasmon-QD interactions shorten the off-state events in AlZMW.

Figure 4.

Blinking suppression of single QDs in ZMWs. (a) Distribution of the on/off state blinking events of a representative QD immobilized on glass, the distribution fits well, with the power equation, $\text{N}(\text{t})={\text{a}}^{\ast }{t}^{-a}$; the power co-efficient for the on-state blinking distributions (green triangles), $a_{on}=1.44$ and the power co-efficient for the off-state blinking distributions (orange triangles), $a_{off}=1.36$. (b) Distribution of the on/off state blinking events of a representative QD immobilized in AlZMW, which also fits well with power equation, $\text{N}(\text{t})={\text{a}}^{\ast }{t}^{-a}$; the power co-efficient for on-state blinking distributions (green circles), $a_{on}=1.20$ and the power co-efficient for the off-state blinking distributions (orange circles), $a_{off}=1.95$. (c) Average duration of the off-state blinking events of individual QDs on glass. (d) Average duration of the off-state blinking events of individual QDs in AlZMW.

The surface plasmon spectrum of nanostructures is relatively broad allowing for modification of both the $\text{X}$ and $\text{BX}$ emission processes of QDs³⁴. Thus, a QD immobilized in an AlZMW is expected to experience a plasmonic effect on both $\text{X}$ and $\text{BX}$ processes. To understand the plasmonic impact of AlZMWs on $\text{X}$ and $\text{BX}$ processes of single QDs, we performed photon correlation spectroscopy (PCS) of the emitted photons from single QDs immobilized on glass and in AlZMW using a Hanbury Brown - Twiss (HBT) optical setup⁵⁷. In a single-photon regime, cross-correlation of the photons arrived at two APDs results in an insignificant correlation at zero delay time ($\text{τ}=0$)¹⁷. However, a significant correlation value at zero delay time corresponds to multiphoton emission events¹⁷. Figures 5a and 5b show the second-order correlation coefficient ${\text{g}}^2(\text{τ})$ values of emitted photons from a QD on glass and in AlZMWs, respectively. It is evident that the second-order correlation at zero delay is very low for the QD on the glass (Figure 5a) which is a signature of antibunching character indicating single-photon emission during each excitation pulse²⁵. Thus, $\text{X}$ emission is the major contributor in the PL intensity of QDs on the glass. The relative contribution of $\text{X}$ and $\text{BX}$ emission can be quantified using low laser power by taking the area ratio of the central peak ($\text{τ}=0$) to the side peaks ($\text{τ}=\text{time interval between pulses}$)⁵⁸.

$$\frac{{\text{ϕ}}_{\mathrm{B}\mathrm{X}}}{{\text{ϕ}}_{\mathrm{X}}}=\frac{\text{Area of central peak}}{\text{Area of side peak}}$$

where, ${\text{ϕ}}_{\text{X}}$ and ${\text{ϕ}}_{\text{BX}}$ represent $\text{X}$ and $\text{BX}$ quantum yield, respectively. We computed the area ratio of the central peak to the average area of the side peaks of the individual QDs on glass and AlZMW and the area ratio is denoted as ${\text{g}}^2(0)$. We found the average ${\text{g}}^2(0)$ value of 0.1 ± 0.1 (Figure 5c) for the QDs on glass. Thus, the contribution of $\text{BX}$ emission is approximately 10% of $\text{X}$ emission for QDs on the glass. The small ${\text{ϕ}}_{\text{BX}}$ of QDs on glass is likely due to the nonradiative recombination of biexcitons through an efficient auger recombination process²⁸. In contrast to glass, QDs in AlZMWs do not exhibit antibunching behavior as evidenced by a rise in the central peak (Figure 5b). The area of the central peak is similar to that of the side peaks, which indicates the enhancement of $\text{BX}$ emission²⁵. The average ${\text{g}}^2(0)$ value of single QDs was 1.5 ± 0.3 (Figure 5c) for the QDs in AlZMW. Thus, the ${\text{ϕ}}_{\text{BX}}$ of single QDs is enhanced by approximately 15 times in AlZMWs making it larger than ${\text{ϕ}}_{\text{X}}\left({\text{ϕ}}_{\text{BX}}>{\text{ϕ}}_{\text{X}}\right)$. This larger enhancement of the ${\text{ϕ}}_{\text{BX}}$ in AlZMWs could be the result of a decrease in auger rates for $\text{BXs}$ or alternatively in a more substantial increase of $\text{BX}$ radiative rate $\left({\text{K}}_{\text{r},\text{BX}}\right)$ as compared to the $\text{X}$ radiative rate $\left({\text{K}}_{\text{r},\text{x}}\right)$ of QDs as a result of coupling with AlZMW surface plasmons¹⁷. The MX radiative decay scales as the square of the number of excitons indicating that the radiative rate for biexcitons must be enhanced in excess of four times the exciton radiative rate in order to overcome the higher $\text{BX}$ auger rates of biexcitons.¹⁷ It has been proposed that QDs could generate multipole components leading to coupling with the multipole plasmon modes of a gold film¹⁷. This is unlikely to occur in an exciton where dipoles dominate¹⁷. In the absence of a plasmonic structure biexcitons can align such that it leads to rapid non-radiative auger recombination¹⁷. However, in the proximity of a plasmonic nanostructure, a biexciton which could be considered as a combination of two dipoles or in other words, a quadrupole, can couple radiatively with the relatively more abundant quadrupole plasmon modes of nanostructures¹⁷. Thus, plasmon-QD interaction can preferentially enhance the ${\text{K}}_{\text{r},\text{BX}}$ in comparison to the ${\text{K}}_{\text{r},\text{x}}$ and also can potentially lower the $\text{BX}$ auger rates¹⁷. We believe a similar concept is applicable to the significant enhancement in ${\text{ϕ}}_{\text{BX}}$ of QDs in AlZMWs where the $\text{BXs}$ couple strongly with the quadrupole plasmon modes of AlZMWs in comparison to their $\text{X}$ counterpart. This would enhance the ${\text{K}}_{\text{r},\text{BX}}$ more significantly than ${\text{K}}_{\text{r},\text{x}}$. On the excitation side, significant enhancement in $\text{BXs}$ can be explained by the more substantial effect of the plasmon-enhanced excitation intensity on $\text{BXs}$ as compared to $\text{Xs}$. Plasmonic nanostructures have been shown to enhance the local electric field due to the excitation of surface plasmons increasing the excitation intensity of nearby QD²⁸. The population of $\text{Xs}$ and MXs are both increased but to a different extent. The absorption of QDs follows a Poisson distribution and the possibility of creating the m^th exciton is described by equation 3 (adopted from Dey et.al 2018)²⁸,

$${\mathrm{P}}_{\mathrm{m}-\mathrm{t}\mathrm{h}} = \frac{{({\text{η}}_{\mathrm{e}\mathrm{x}\mathrm{c}}<\mathrm{N}>)}^{\mathrm{m}}{\mathrm{e}}^{-{\text{η}}_{\mathrm{e}\mathrm{x}\mathrm{c}}<\mathrm{N}>}}{\mathrm{m}!}$$ (3)

Figure 5.

Bunching character of emitted photons from single QDs in ZMW, representing enhanced biexciton yield. (a) Second order cross correlation (${\text{g}}^2(\text{τ})$) of the photons emitted from a single QD immobilized on glass, the insignificant correlation value at zero delay time ($\text{τ}=0$) corresponds to the antibunching character of the photons. (b) Second order cross correlation (${\text{g}}^2(\text{τ})$) of the photons emitted from a single QD immobilized in AlZMW, a significant rise of the correlation value at zero delay time ($\text{τ}=0$) correspond to the bunching character of the photons. (c) An average ${\text{g}}^2(0)$ value (area ratio of the central peak ($\text{τ}=0$) to the average area of the side peaks ($\text{τ}$ = time interval between pulses)) of the QDs immobilized on glass and in AlZMW, exhibiting an increased ${\text{g}}^2(0)$ value corresponding to the enhanced ${\text{ϕ}}_{\text{BX}}$ of the QDs in AlZMW. (d) A Scatter plot of the ${\text{g}}^2(0)$ value VS PL intensity of single QDs immobilized on glass (black circles) and in AlZMW (red circles). (e) PL spectrum of a single QD on glass and in AlZMW.

Where, $\text{m}=1$ for $\text{X}$, $\text{m}=2$ for $\text{BX}$, and so on. $<N>$ is the average number of photons absorbed per pulse, ${\text{η}}_{\text{exc}}$ = enhancement in excitation intensity of QDs in AlZMW.

Thus, with the 2-fold enhancement in excitation intensity of QDs in AlZMW (${\text{η}}_{\text{exc}}=2$), the probability of $\text{X}$ generation is increased by 2-fold. However, the probability of $\text{BX}$ generation is enhanced by 4 times. The $\text{BX}$ excitation rate of QDs is then increased to a more significant extent than the $\text{X}$ excitation rate. Thus, AlZMWs likely increase the probability of generating $\text{BXs}$ to a greater extent than $\text{Xs}$ in QDs, and $\text{BX}$ recombination channels become more efficient due to their resonant coupling with the plasmon quadrupole modes of AlZMWs¹⁷. Increased ${\text{ϕ}}_{\text{BX}}$ has been observed in the case of other QD-nanostructure systems, including QDs coupled with a rough Au surface^{17, 59}, Au nanoparticle^{10, 28}, nanorods²⁵, or a nanogap between Au nanorods and an Au film³⁴. Park et al. also observed a significant enhancement in ${\text{ϕ}}_{\text{BX}}$ of single QDs on a rough Ag surface³⁷. However, they postulate that instead of an improvement, both the ${\text{K}}_{\text{r},\text{BX}}$ and ${\text{K}}_{\text{r},\text{BX}}$ are quenched and ${\text{K}}_{\text{r},\text{BX}}$ is quenched to a lesser extent than the ${\text{K}}_{\text{r},\text{BX}}$ which makes ${\text{ϕ}}_{\text{BX}}/{\text{ϕ}}_{\text{X}}$ >1³⁷. In this case quenching of the radiative rates also results in a significant decrease in PL intensity. We did not observe a decrease in the PL intensity, suggesting that this is not the mechanism for QDs in AlZMWs. The differences in our observations could be due to the PVPA protective layer used to prevent non-specific binding of QDs on the metal walls of the ZMWs in contrast to studies using metal films where the QD is directly in contact with the surface. A scatter plot of the PL intensity of single QDs with their individual ${\text{ϕ}}_{\text{BX}}/{\text{ϕ}}_{\text{X}}$ ratio further demonstrates the enhancement of ${\text{ϕ}}_{\text{BX}}$ in AlZMW (Figure 5d). To investigate if plasmon mediated alteration in photon statistics has any effect on the PL spectrum, we compared the PL spectrum of a single QD in AlZMW with the PL spectrum of single QD on glass (Figure 5e).

There was no obser difference between the spectra. There was no indication of an extra peak at the high energy side, which indicates that most of the higher order excitons emitted from the QD are BX²⁸. The absence of peaks for triexcitons or other higher order excitons is presumably due to the lower intensity excitation source, which might not be sufficient to excite that higher energy excitons²⁸. The $\text{BX}$ emission spectra often exhibit a redshift of 15–20 meV, which is equivalent to 4–6nm from the $\text{X}$ emission spectrum^{7, 11, 28, 60}. This shift is difficult to distinguish at room temperature with a large FWHM (~ 42 nm) and was not observed in our data. One important point to mention that the width of the ${\text{g}}^2(\text{τ})$ peaks became narrower for QDs in AlZMW compared to QDs on glass. The average FWHM value of the ${\text{g}}^2(\text{τ})$ on glass is 66 ± 2 ns, whereas it reduces to 36 ± 3 ns (Figure S6) in AlZMWs. This narrowing is an indication of a shortened excited state lifetime which is supported by the measured ~5-fold decrease in PL lifetime of the QDs in AlZMWs. Shortening in the PL lifetime of QDs in AlZMWs likely results from non-radiative energy transfer from the excited state to the ZMW metal and the recombination of excitons to the plasmon modes of AlZMWs.¹⁷ Key results from the modification in the photophysical behavior of QDs due to AlZMWs are summarized in Table 1.

Table 1.

Key photophysical changes of QDs in ZMWs

Enhancement in PL intensity (${\text{η}}_{\text{PL}}$)	2.5x
Enhancement in excitation intensity (${\text{η}}_{\text{exc}}$)	2.0x
Enhancement in radiative decay rate (${\text{η}}_{\text{r}}$)	4.3x
Enhancement in nonradiative decay rate (${\text{η}}_{\text{nr}}$)	7.0x
Reduction in an off-state event duration	3.0x
Enhancement in biexciton quantum yield	15.0x

In conclusion, we studied the effect of AlZMWs on the photophysical properties of QDs. We found a ∼2.5-fold enhancement in PL intensity resulting from a 2-fold enhancement in excitation intensity due to the increased local electric field, 1.5-fold enhancement in collection efficiency and slight quenching in the PLQY of QDs. We also calculated changes in radiative and nonradiative decay rates of QDs isolated in AlZMWs. We found a ∼ 4.3-fold enhancement in the radiative decay rate and a 7.0-fold increase in nonradiative decay, which resulted in the slight quenching of PLQY. We also investigated the influence of plasmon-QD interactions on $\text{X}$ and $\text{BX}$ emission processes and found that the ${\text{ϕ}}_{\text{BX}}$ of QDs is increased by ∼ 15 times. The large enhancement in ${\text{ϕ}}_{\text{BX}}$ indicates that plasmon-QD interactions lead to more efficient decay of $\text{BX}$ radiative pathways. We also observed less blinking for QDs in AlZMWs, which presumably resulted from an enhanced decay rate that accelerated e-h recombination process in QDs.

ABBREVIATIONS

ZMWs

Zero mode waveguides

QDs

Quantum dots

PLQY

Photoluminescence quantum yield

X

Single exciton

BX

Biexciton