Biological sensing and control of emission dynamics of quantum dot bioconjugates using arrays of long metallic nanorods

Abstract

We study biological sensing using plasmonic and photonic-plasmonic resonances of arrays of ultralong metallic nanorods and analyze the impact of these resonances on emission dynamics of quantum dot bioconjugates. We demonstrate that the LSPRs and plasmonic lattice modes of such array can be used to detect a single self-assembled monolayer of alkanethiol at the visible (550 nm) and near infrared (770 nm) range with well resolved shifts. We study adsorption of streptavidin-quantum dot conjugates to this monolayer, demonstrating that formation of nearly two dimensional arrays of quantum dots with limited emission blinking can lead to extra well-defined wavelength shifts in these modes. Using spectrally-resolved lifetime measurements we study the emission dynamics of such quantum dot bioconjugates within their monodispersed size distribution. We show that, despite their close vicinity to the nanorods, the rate of energy transfer from these quantum dots to nanorods is rather weak, while the plasmon field enhancement can be strong. Our results reveal that the nanorods present a strongly wavelength or size-dependent non-radiative decay channel to the quantum dot bioconjugates.

I. INTRODUCTION

Optical coupling of localized surface plasmon resonances (LSPRs) in arrays of metallic nanoantennas (mANTs) can form plasmonic lattice modes (PLMs).^1–6 From application point of view these modes are particularly appealing since spectrally they can be much narrower than the LSPRs of the individual m-ANTs. Particularly, extensive research is devoted to investigate how such modes can be used for biological and chemical sensing.^7–13 A prime advantage of sensors based on arrays of m-ANTs is that the targets can not only cause wavelength shift, but also can induce significant changes in the optical scattering between LSPRs.⁸ To utilize this feature, particularly for biological applications, one requires direct contact between the m-ANTs and the targets. In this regards the recent investigations regarding generation of PLMs in inhomogeneous media, wherein the refractive indices of the substrate and the ambient environments of m-ANTs are different, are particularly useful.⁹ In such media PLM can be generated by either increasing the polarizablity or sizes of m-ANTs,^3,6,14–17 or by making hybrid structures via capping of m-ANTs with a thin layer of silicon.¹⁸

In this paper we study biological sensing using plasmonic and photonic-plasmonics resonances of arrays of ultralong (~ 1300 nm long) metallic nanorods (NRs) and investigate their impact on emission dynamics of quantum dot bioconjugates. These arrays can support distinct multipolar LSPRs and PLMs in the visible and near infrared range, respectively, when the incident light is polarized along their short axes. We show that at both of these resonances, conjugation of a self-assembled monolayer (SAM) of alkanethiol can lead to distinct wavelength shifts. After activation we use this SAM as receptor for semiconductor QDs conjugated with streptavidin. In the presence of characteristic emission blinking, which indicates formation of a nearly two dimensional monolayer of QDs on the SAM, we demonstrate further wavelength shifts in both LSPR and PLMs with good amount of resolution. These indicate high sensitivity of the arrays to variations of the refractive index of the environment, which can clearly distinguish the SAM and QD bioconjugates from each other. They also show the possibility of dual-wavelength sensing of a biological or chemical target.

To study the impact of NRs on the radiative and non-radiative decay of QD bioconjugates, particularly energy transfer and plasmonic field enhancement, we used spectrally-resolved lifetime measurements. This technique allows us to study these features while considering the monodispersed size distribution of the QDs. We scanned the decay of such QDs from blue to red side of their emission bands, i.e. small to large QDs, demonstrating how the long gold NRs influence the interdot FRET and the FRET from QDs to NRs. Our results show that these NRs can substantially reduce the radiative lifetime of the QDs, as one expects from their near plasmon fields.¹⁹ The non-radiative decay rate, however, remains nearly unchanged. This highlights that although QDs are in very close proximity of the NRs, their FRET rate to the NRs is not substantial. We also show that the impact of the NRs are most visible in the shorter wavelength range associated with QDs with smaller core sizes.

Note that NRs have significant applications for ultra-sensitive detection of biological molecules,^20–22 subwave-length imaging,²³ investigation of quantum plasmonics.²⁴ Additionally, since QDs can be very efficient and stable emitters, they can offer unique opportunities for detection of biological molecules based on FRET process.^25,26 Moreover, hybrid systems consisting of QDs and NRs can offer new physics for sensing based on quantum effects,^27,28 and creation of novel quantum devices.²⁹

II. METHODOLOGY

We used e-beam lithography to fabricate long Au NR arrays on glass substrates. As the top-view of scanning electron microscopy (SEM) image of such arrays in Fig. 1a shows, the width (W) and length (L) of each NR were ~219 nm and ~1290. The lattice constants of the array were a_x=1.4 μm and a_y=0.5 μm. The nominal thickness of NRs were 40 nm. To measure extinction spectra of these samples we used a transmission setup consisting of a halogen lamp, polarizer, microscopic objective, and collective lens. Samples were placed between the objective and the collective lens which directed part of the transmitted light towards a sensitive spectrometer (Ocean Optics TE-cooled QE-Pro) using an optical fiber. The collective lens was placed at sufficient distance right beneath the microscope objective to ensure a well-defined and small-angle spread of illumination.

FIG. 1.

(a) Top view SEM images of ultralong NR array. (b) Schematic of the energy transfer from a QD to a NR. In (a) the lattice constants are a_x=1.4 and a_y=0.5 μm and the length (L) and width (W) of each NR are, respectively, 1290 and 219 nm.

To demonstrate the biological sensing of such arrays, as schematically shown in Fig. 2 (upper panel) we used carboxylic terminated self-assembled monolayer (SAM) followed by its conjugation with streptavidin-conjugated QDs (strep-QDs).^30–32 For this we used a thiol mixture consisting of 3-mercaptopropanol (3-MPOH) and 12-mercaptododecanoic acid (12-MDA), acquired from Sigma Aldrich. Using a special chamber we flowed this mixture over the sample for 18 hours, creating an alkanethiol monolayer (SAM) on the surface of gold m-ANTs (Fig. 2, upper panel).^30–32 After activating the terminal carboxylic groups of the monolayer into amine reactive succinimide esters with 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) and N-hydroxysuccinimide (NHS) solutions, they were immediately incubated in 5 μl of strep-QD conjugates mixed in water for 30 mins. We used low concentration of strep-QD conjugates to avoid aggregation. We recorded the extinction spectra of the long NR array after the SAM and after the strep-QD conjugation. The strep-QDs were acquired from ThermoFisher Scientific. They had Cd-SeTe cores with ZnS shells and their emission peaks were centered at about 705 nm. Each QD had approximately 5 to 10 biotin-binding streptavidins that were covalently attached to it.

FIG. 2.

Upper panel: Schematic illustration of immobilization of biological SAM and strep-QDs on NRs. Lower panel: Schematics of the setup components used for detection of emission spectra of strep-QDs (A), spectrally-resolved lifetime measurements (B), and optical imaging (C). Here S-PMT, mono-λ, and EMCCD stand for single photon PMT detector, monochromators, and Electron Multiplying Charge Coupled Device camera, respectively. CW 514 and 30-ps 450 refer, respectively, to continuous wave 514 nm laser diode and 30-ps pulsed laser with 450 nm wavelength.

To spectrally analyze the emission of strep-QDs we used a continuous wave laser diode with 514 nm wavelength. A combination of dichroic mirrors and filters were used to separate the emission of the QDs from the laser beam, before sending it to spectrometer (Fig. 2, compartment A). To study dynamics of emission of QDs we carried out spectrally-resolved lifetime measurements using a time correlated single photon counter (Picoquant PicoTime harp 260) with a single-photon photomultiplier tube (PMT). For this, before the emission of the QDs reaches the PMT, we used a monochromator to disperse it spectrally (Fig. 2, compartment B). To illustrate the monolayer nature of conjugation of strep-QDs to SAM we used a sensitive ANDOR iXon Ultra 897 Electron Multiplying Charge Coupled Device (EMCCD) camera (Fig. 2, compartment C).

III. BIOSENSING USING ULTRALONG NANOROD ARRAYS

We start our investigation with the measurement of extinction spectra of the sample when the NRs are exposed to air from the top, i.e., no bio-materials. The results presented in Fig. 3a show that when the incident light is polarized along the y-axis, we can see the spectrum contains a small peak at about 533 nm (peak A) and a sharp peak at about 755 nm (peak B). The sharp peak is an indication that the sizes of NRs considered in this paper are large enough such that the array can support PLM even though the structure is asymmetric.^9,16 Peak B is in close vicinity of Rayleigh Anomaly (RA) wavelength (λ_±1,0) at about 756 nm. Here λ_±1,0 = n_subsa_y, wherein n_subs=1.512 is the refractive index of the substrate. As shown in Fig. 3b (line 2), when the incident light is polarized along the x-axis, we see the spectrum contains a small peak at about 707 and a broad peak at about 954 nm. These peaks are probably associated, respectively, with the seventh and fifth order multipolar LSPRs.^33,34

FIG. 3.

Extinction spectra of the NR arrays in the absence (lines 1 and 2) and in the presence of the biological SAM (dashed lines) and strep-QDs (thin solid lines). In (a) the incident light was polarized along the y-axis and in (b) along the x-axis.

To study variation of the optical responses of the NR arrays under very small changes in the refractive index, we added the receptor (the biological SAM) and then the target (strep-QDs). The spectra after the SAM and then strep-QD show clear red shifts. Fig. 3a shows that for the case of peak A, SAM leads to ~5 nm shift and strep-QDs to an extra 7 nm (Fig. 3a, left inset). For the case of peak B the corresponding shifts are 8 nm and 7 nm, respectively (right inset). When the incident light polarization is along x-axis, we do not see significant changes in the shape of broad peak. Its wavelength, however, is shifted by about 21 and 29 nm after adding SAM and strep-QDs, respectively. Despite such larger shifts, the optical response associated with the x-axis is prone to large amount of errors because of its low wavelength resolution.

A key feature of using strep-QDs is that their emission can be used to detect the uniformity of the biological conjugation process. We found that variations of the QD emission intensity on the regions with NRs was about 25% (not shown). As shown in the Fig. 4a inset, on the glass regions, however, we observe nonuniform residual of strep-QDs with large fluctuation of about 65%. These results show a fairly uniform bonding via the strong affinity of the SAM with gold and strep-QDs with the SAM. Fig. 4a shows the averaged emission spectra of these QDs on the NR arrays. The results show that the emission intensity with polarization along the x-axis (blue solid line) is less than that along the y-axis (red dashed line). This suggests that the emission of the QD bioconjugates are partially polarized. Fig. 4b shows the optical image of the strep-QDs on the NR arrays taken by the EMCCD camera. Note that this image shows the shadows of the NRs, which are formed via back scattering of the emission of the QDs. We observed limited amount of emission blinking, suggesting that the conjugation process formed a monolayer of QDs close to the surfaces of the NRs.

FIG. 4.

(a) The relative emission spectra of strep-QDs on NRs arrays after passing a polarization analyzer. The solid line represents the case when the analyzer axis was along the x-axis and the dashed line the case when it was along the y-axis. (b) Image of this region taken by EMCCD camera. The inset in (a) shows variations of emission of strep-QDs in different spots on glass (off the NR region).

IV. EMISSION DYNAMICS OF QD BIOCONJUGATES IN THE PRESENCE OF LONG NRS

The monolayer of strep-QDs close to the NRs provide us a testbed for investigation of the impacts of plasmonic effects on the interdot FRET and energy transfer between QDs and NRs. To see this note that the width of emission spectra of the QDs presented in Fig. 4a is an indication of their core size distribution. The shorter wavelength sides of these spectra are due to QDs with smaller cores while the larger QDs decay at longer wavelengths. Fig. 5 shows the results for spectrally-resolved lifetime measurements of these QDs on glass (a) and on the NR arrays (b). These results are captured at 650 (line 1), 675 (line 2), 705 (line 3), 725 (line 4), and 750 nm (line 5). They show that at all wavelengths the initial decay is faster. This is particularly more pronounced in the presence of NRs (Fig. 5b). Additionally, as expected, we found that the decay rates at shorter wavelengths are faster than those at longer wavelengths. This is partially due to the transfer of energy from the smaller QDs to larger ones.³⁵ To study the decay seen in Fig. 5 we carried out a bi-exponential fitting using:^36–38

FIG. 5.

Average spectrally-resolved emission decay of strep-QDs in the absence (a) and presence (b) of the NR arrays. Here lines 1, 2, 3, 4, and 5 refer to decay at 650, 675, 705, 725, and 750 nm, respectively.

$$I(t)=A_f{e}^{\frac{t}{{\tau }_f}}+A_s{e}^{\frac{t}{{\tau }_s}}.$$ (1)

Here A_f and τ_f refer to the amplitude and decay time of the fast process and A_s and τ_s to those of the slow process. Such a distribution is related to the fact that the surfaces of colloidal QDs play important roles in carrier relaxation and recombination processes.³⁹ The fast process can be related to non-radiative decay, including FRET, and the slow process to radiative decay of the QDs.⁴⁰

Fig. 6 shows the outcomes of the fitting to the results shown in Fig. 5a (strep-QDs in the absence of the NRs). These results highlight energy transfer from smaller QDs to larger QDs within the monodispersed size distribution of the QDs. This process is seen here as the decline of A_f (circles) with wavelength, indicating the number of QDs available as donors decreases at longer wavelengths (Fig. 6a). This is confirmed by the fact A_s increases with wavelength, suggesting the radiative emission becomes more dominated in the larger QDs. On the other hand, the results presented in Fig. 6c show that τ_f is relatively very short. In fact, for short wavelength (around 650 nm), τ_f is about 1.5 ns, indicating fast decay via transfer of energy to larger QDs and entrapment of electrons in the surface defects. As wavelength increases, τ_f starts to increase (Fig. 6c inset), reaching about 2.5 ns. Such a rise can be due to the fact that for larger QDs, interdot energy transfer is less probable. The results in Fig. 6c also show that at long wavelengths (around 750 nm) τ_s is about 100 ns, about two times of that at 650 nm. Combination of this and the reduction of A_f /A_s by about 2.5 times across the 650–750 nm wavelength range (Fig. 6b) favor more efficient emission for the larger QDs.

FIG. 6.

Average variations of (a) A_s (squares) and A_f (circles), and (b) their ratio in the absence of long NRs as a function of wavelength. (c) shows variation of τ_s (squares) and τ_f (circles) with wavelength. The inset in (c) shows the detailed variations of τ_f with wavelength.

The results shown in Fig. 7 demonstrate the impact of the NRs on the emission dynamics of the strep-QDs. Fig. 7a shows that a prime effect of the NR arrays is enhancement of A_f. In fact for wavelengths less than about 720 nm, in contrast to that seen in Fig. 6a, A_f is higher than A_s. This suggests that under this condition, in addition to the defect sites and interdot FRET, energy transfer from QDs to the NRs can play as a major role as a non-radiative decay channel for excitons. This process, which reduces the chance of the radiative decay of QDs, seems wavelength dependent way. In fact, for wavelength longer than 720 nm A_f becomes smaller than A_s, indicating the rise of radiative decay probability of QDs as their core sizes are increased. This can be seen further in Fig. 7b that shows A_f /A_s undergoes a nearly linear decline with wavelength. The overall value of A_f /A_s, however, remains significantly higher than that seen in Fig. 6b, indicating the dominant contribution of NRs as non-radiative channel for the QDs. The other profound effect of NRs is significant shortening of τ_s. As seen in Fig. 7c, compared to the case of Fig. 6c, here τ_s is reduces by about one half. This can be due to the near plasmon fields of the NRs, which are known to increase radiative decay rates of the QDs.^41,42 The detailed variations of τ_f with wavelength in the presence of the NR arrays is shown in the inset of Fig. 7c. These results are very similar to those in the absence of the NRs (Fig. 6c). This suggests that the NRs mostly contributed to increase of the non-radiative sites available to the QDs (Fig. 7a). The rate of this process, however, is so small such that it does not change the overall values of τ_f, compared to those obtained in the absence of NRs (Fig. 6c). This can be due to the nature of the plasmon fields of the long NRs and the distance of the strep-QDs from them.

FIG. 7.

Average variations of (a) A_s (squares) and A_f (circles), and (b) their ratio in the presence of the long NRs as a function of wavelength. (c) shows variation of τ_s (squares) and τ_f (circles) with wavelength. The inset in (c) shows the detailed variations of τ_f with wavelength.

Note that the emission dynamics of QDs in the presence of m-ANTs strongly depend on wavelengths of both excitons and LSPRs. Usually FRET from QDs to mANTs occurs most efficiently when the exciton wavelength is close to the plasmonic absorption peak of mANTs. On the other, plasmonic field enhancement can happen at a different wavelength than that of the LSPR absorption peak. The emission wavelengths of the QDs used in this paper were centered at about 705 nm. Therefore, they were closed to the longitudinal multipolar LSPR that occurred at ~707 nm (Fig. 3b) and the PLM (Fig. 3a). Such a configuration can be correlated with the fact that why, based on Fig. 7, the QDs did not efficiency give up their energies to NRs, while they were influenced by their plasmon fields. Additionally, one of the main advantages of PLMs for sensor applications is their narrow spectral widths.^8,43 Depending on their structural parameters, however, they may become optically saturated, i.e., show small refractive index sensitivity, when they are fully developed.

Note also that the emission dynamics of QDs shown in Figs. 6 and 7 depend on the spacings between the QDs and variations of their core sizes. In general narrow spectral linewidth is a sign of small size fluctuations and, therefore, less variations in the QD transition energies. This makes the absorption (or emission) wavelengths of the QD donors and acceptors close to each other, leading to less efficient interdot FRET. Other the other hand, in a wider distribution the chance of spectral matching of emission of donors and absorption of acceptors is higher. Such a spectral match can make the interdot FRET more efficient.³⁵ Additionally, the results in Figs. 6 and 7 show that although the QDs were able to blink, they were close enough to each other to establish efficient interdot FRET.

V. CONCLUSIONS

We studied variations of the optical responses of arrays of long Au NRs that could support LSPRs and PLMs in the presence of biochemical agents. We applied various experimental techniques to study the way streptavidin-QD conjugates were anchored to such NRs via a biological SAM. Our imaging and time-resolved measurements showed QDs were formed in a monolayer such that they could exhibit some limited amount of blinking behavior. We showed emission dynamics of different sizes of the QDs in their monodispersed distribution, highlighting large plasmon field enhancement but small rate of FRET from QDs to NRs. We discovered that when the incident light was polarized along the long axes of the NRs, these arrays can support unique secondary peaks capable of detection of minuscules changes in refractive index of the environment. We also showed the PLMs of such arrays do not present a high refractive sensitivity while the high order multipolar LSPRs can.