Compact Hybrid (Gold Nanodendrite-Quantum Dots) Assembly

Oct 15, 2014 - increased by 4 times during the formation of the compact hybrid. (AuND-QDs) assembly. Both experiment and finite-difference time...
2 downloads 0 Views 4MB Size
Article pubs.acs.org/ac

Compact Hybrid (Gold Nanodendrite-Quantum Dots) Assembly: Plasmon Enhanced Fluorescence-Based Platform for Small Molecule Sensing in Solution Huide Chen and Yunsheng Xia* Key Laboratory of Functional Molecular Solids, Ministry of Education; College of Chemistry and Materials Science, Anhui Normal University, Wuhu 241000, China S Supporting Information *

ABSTRACT: In this study, we have presented a novel plasmon enhanced fluorescence (PEF) system for label-free sensing of small molecules in bulk solution. The amine-terminated gold nanodendrite (AuND) and carboxyl-terminated QDs directly assemble each other by amine−carboxyl attraction. Without any spacer layers, PEF can be increased by 4 times during the formation of the compact hybrid (AuND-QDs) assembly. Both experiment and finite-difference time domain calculation results indicate that the distinct solution-PEF effect is ascribed to two reasons: (1) The used AuNDs simultaneously possess four features in morphology and topology, well-defined superstructure, sharp tips and edges, moderately elongated subunits, and smaller size. (2) The hybrid (AuND-QDs) assembly has a very compact structure. So, the fluorescence is well enhanced by the effective increase of excitation and radiative decay rates with the decrease of scattering effect. The (AuND-QDs) assembly is then employed for sensing of trinitrotoluene (TNT), one of the highly explosive and environmentally detrimental substances, in bulk solution. The sensing principle is that the analytes can react with primary amines on the AuND surface and form Meisenheimer complexes, which break the preformed assemblies and result in the fluorescence recovery of the QDs. The linear range is 0−8.8 nM with 0.05 nM detection limit. The present quasipicomole level sensitivity is one of the best results for fluorescent TNT sensing. The developed method is successfully applied to TNT sensing in real environmental samples, indicating the practical potential.

T

In addition to FRET, there are two additional effects of metal NPs on fluorophores. One mechanism is metal increased rate of excitation, and another is metal increased rate of radiative decay.6 These two photophysical effects can cause fluorescence enhancement of the fluorophores, which are often called plasmon enhanced fluorescence (PEF) or metal enhanced fluorescence. PEF behaviors have been reliably observed in both organic dyes and inorganic QDs,7 and they can also be used for sensing applications by several groups. However, to date, most PEF sensing systems are focused on substrates,8 and analyte targets are often limited to DNA and proteins assisted by labeling technique.9 In fact, the existing substrate-based PEF sensing platform has difficulty in sensing small molecules or ions, because they are not suitable to be labeled. Instead, the solution−PEF system is more appropriate for these substances. Unfortunately, PEF is rarely observed in bulk solution.10 The reason is that large metal particles (subwavelength sized metal core plus an additional layer with tens of nanometers) are often used for effective PEF effect.8c,11 Such bulky particles cannot work in solution because (1) they are not stable in solution due to gravitational effect and (2) they strongly scatter the enhanced emission and lead to fluorescence quenching.

he hybrid assembly consisting of gold nanoparticles (AuNPs) and quantum dots (QDs) has attracted considerable attention ranging from fundamental science study to technical applications because of the coupling effects of plasmon−exciton interactions.1 Among these, (AuNP-QDs) assembly-based fluorescence resonance energy transfer (FRET) systems have been well used for sensing and biosensing by versatile surface modification.2 In terms of these systems, two types of sensing strategies are often employed. The first one is the “turn off” mode. Dispersed AuNPs and QDs form a hybrid assembly in the presence of analyte targets; FRET from QDs to AuNPs is then opened and results in fluorescent quenching.3 The second one is the “turn on” mode. Preformed hybrid (AuNP-QDs) assembly is broken by analyte targets, which switches off the FRET and recovers the quenched fluorescence.4 In both cases, fluorescence quenching process is essential. As we know, in addition to energy transfer, there are lots of factors that can cause fluorescence quenching, which would impair the signal reliability.5 Therefore, a conceptual demonstration of a (AuNP-QDs) platform, in which signal readout is independent of any quenching effects, will be significant for both fundamental research and applications. Such a platform, on the one hand, means a better reliability in signal output; on the other hand, it needs a fundamentally different photophysical principle beyond FRET. © 2014 American Chemical Society

Received: April 24, 2014 Accepted: October 15, 2014 Published: October 15, 2014 11062

dx.doi.org/10.1021/ac5031804 | Anal. Chem. 2014, 86, 11062−11069

Analytical Chemistry

Article

Instruments and Characterizations. Steady-state fluorescence measurements were performed using a Hitachi F-4600 spectrofluorometer equipped with a 1 cm quartz cuvette. Fluorescence lifetimes were measured by the time correlated single photo counting technique on the combined steady state and lifetime spectrometer (Edinburgh Analytical Instruments, FLS920). Characterizations of transmission electron microscopy (TEM) were carried out on Tecnai G2 20 ST (FEI) under the accelerating voltage of 200 kV. Characterizations of scanning electron microscopy (SEM) were carried out on Hitachi S-4800 under the accelerating voltage of 5 kV. A Hitachi-U-3010 spectrometer was used to record the UV− visible spectra. Fourier transform infrared (FT-IR) spectra were recorded from a KBr window on a PerkinElmer PE-983 FT-IR spectrophotometer. ξ-Potential values were tested by dynamic light scattering (Zetasizer Nano ZS series, Malvern Instruments) with 633 nm laser wavelength and a measurement angle of 173° (backscatter detection) at 25 °C. Synthesis and Modification of AuNDs. AuNDs were prepared in aqueous solutions by the modified Peng’s method.12 The Au seeds were first synthesized. A 20 mL sample of aqueous HAuCl4 (0.25 mM) was put in a 25 mL flask. After the solution was brought to a boil under stirring, 200 μL of 5% sodium citrate solution was added. The reaction was allowed to run until the color of solution became wine red. To grow AuNDs, 100 mL of HAuCl4 (0.25 mM) aqueous solution was added to a round-bottom flask, followed by adjusting the pH to 11.4 by 1 M NaOH solution. The mixture of 800 μL of NH2OH·HCl solution (40 mM) and 8 mL of the Au seeds then was added into the above solution at 30 °C. With slight stirring, the color of the solution turned blue green, indicating the formation of AuNDs. For obtaining cysteaminemodified AuNDs, AuNDs were first centrifuged to remove other ions in solution. The precipitate was redispersed in water, and subsequently a small amount of cysteamine was added under vigorous stirring. The mixture was stirred at room temperature for 3 h and then was allowed to age without agitation at 4 °C overnight. The resultant AuNDs were collected by centrifugation to remove excess cysteamine, and the purified cysteamine-modified AuNDs were dispersed in water directly. The resultant AuNDs solution could be well kept for at least 1 week in 4 °C. Because there is no extinction coefficient data for AuNDs, their concentrations were estimated according to spherical gold NPs with identical size. The calculation method and related data are based on a previous report.13 Synthesis and Purification of CdTe and CdTe@CdS Core@Shell QDs. TGA-capped CdTe QDs were prepared in aqueous solution using the method described previously.4b Briefly, 1.25 × 10−3 mol of CdC12·2.5H2O was dissolved in 100 mL of water, and 3.0 × 10−3 mol of TGA was added under stirring, followed by adjusting the pH to 11.2 by adding dropwise 1.0 M NaOH. The solution was deaerated by N2 bubbling for 30 min. Under vigorous stirring, 1.6 × 10−4 mol of freshly prepared oxygen-free NaHTe was injected into the above solution. Afterward, the solution mixture was heated to 100 °C and refluxed for different times to obtain different sized CdTe QDs; for example, QDs-545 can be obtained after reflux for 2 h, and the reflux time for QDs-600 is 8 h. We then used previous reports for CdTe@CdS core@shell QDs.14 50 mL of crude CdTe (λEm= 570 nm) solution was cooled to room temperature and deaerated by N2 bubbling for 30 min. Under vigorous stirring, 8.0 × 10−5 mol of TAA, which acted as a

Therefore, to sense small molecules/ions in solution by PEF platform, the above challenges in both science and technology should be faced and conquered. Herein, by a combination of experiments and finitedifference time domain (FDTD) calculations, we rationally design a novel hybrid assembly made of gold nanodendrite (AuND) and QDs. Without any spacer layers, PEF can increase by 4 times in bulk solution. The key points for the distinct PEF effect can be ascribed to two reasons. First, the used AuNDs simultaneously possess four features in morphology and topological structure: (1) The AuNDs have a unique superstructure; (2) the AuNDs possess many sharp tips and edges; (3) each AuND contains lots of moderately elongated subunits; and (4) the AuNDs are rather small (62 nm in diameter). Different from the existing PEF systems, the present design needs not any spacer layers and possesses a very compact structure (the amine-terminated gold nanodendrite (AuND) and carboxyl-terminated QDs directly assemble each other by amine-carboxyl attraction). So, in the (AuND-QDs) assembly, the fluorescence can be well enhanced by the effective increase of excitation and radiative decay rates with the decrease of scattering effect. The (AuND-QDs) assembly can be employed for sensing of trinitrotoluene (TNT), one of the highly explosive and environmentally detrimental substances, in bulk solution. The sensing principle is that the analytes can react with primary amines on the AuND surface and form Meisenheimer complexes, which breaks the preformed assemblies and results in the fluorescence recovery of the QD (Scheme 1). The detection limit is as low as 0.05 nM; such Scheme 1. Schematic Illustration of Plasmon Enhanced Fluorescence-Based Platform for TNT Sensing Enabled by the (AuND-QDs) Assembly

quasi-picomole level sensitivity is one of the best results for fluorescent TNT sensing. This plasmon enhanced fluorescencebased platform is successfully applied to TNT sensing in real environmental samples, indicating the promise in practicability.



EXPERIMENTAL SECTION Materials. Thioglycollic acid (TGA), cysteamine, and Te powder were purchased from Alfa Aesar. Trinitrotoluene (TNT), 2,4-dinitrotoluene (DNT), and nitrobenzene (NB) were obtained from Sigma-Aldrich. Trihydrate chloroauric acid (HAuCl 4 ·3H 2 O), sodium citrate, sodium borohydride (NaBH4), cadmium chloride (CdC12·2.5H2O), hydroxylamine hydrochloride (NH2OH·HCl), sodium hydroxide (NaOH), thioacetamide (TAA), and other routine chemicals were acquired from Shanghai Reagent Co. All solutions were prepared with double deionized water. 11063

dx.doi.org/10.1021/ac5031804 | Anal. Chem. 2014, 86, 11062−11069

Analytical Chemistry

Article

Figure 1. (A) TEM image of the cystamine-modified AuNDs. (B) TEM image of the carboxyl-terminated CdTe QDs-600. (C) Normalized SPR absorption spectra of the AuNDs and the three emissive CdTe QDs. The inset of (C) is the photo of the AuND solution in room light.



sulfur source, was added to the QD solution. Next, the mixture solution was heated and maintained at 100 °C in an oil bath for 3 h to form the CdTe@CdS core@shell structure (λEm = 600 nm). For purification of the CdTe and CdTe@CdS QDs, 5 mL of ethanol was added to 5 mL of QD solution, and then the mixture was centrifuged, and the purified CdTe@CdS QDs were redispersed in 5 mL of water as stock solution, which was stable at least 8 months at 4 °C. The QD stock solution was further diluted to 2.28 × 10−7 M by water for experiments. The concentration of CdTe@CdS QDs was estimated according to the concentration of CdTe core using the method reported previously.15 Evolutions of Fluorescence of QDs with Increasing AuND Concentrations. 50 μL of purified CdTe@CdS core@ shell QD solution (2.28 × 10−7 M) and various amounts of cysteamine-modified AuNDs (2.3 × 10−10 M) were placed in a series of 5 mL colorimetric tubes. The mixture was diluted to 2 mL with water and mixed thoroughly. Five minutes later, the mixed solutions were transferred separately into a 1 cm quartz cuvette. The fluorescence spectra were recorded by the fluorescence spectrophotometer. Procedures for TNT Sensing. 50 μL of purified CdTe@ CdS core@shell QD solution (2.28 × 10−7 M) and 100 μL of AuNDs (2.3 × 10−10 M) were placed in a series of 5 mL colorimetric tubes. Five minutes later, various amounts of analytes were added. The mixtures were diluted to 2 mL with water and mixed thoroughly. Five minutes later, the mixed solutions were transferred separately into a 1 cm quartz cuvette. The fluorescence spectra were recorded by the fluorescence spectrophotometer. Procedures for TNT Sensing in Real Samples. Tap water (from Lab) and lake water (Jinghu Lake, Wuhu) samples were used after quiescence for 24 h, and then the supernatant was used for experiment. For sensing of TNT in the water samples, pure and TNT-spiked water samples (20 μL) were added to an aqueous dispersion of (AuND-QDs) assembly; the concentrations of AuNDs and QDs were 11.5 pM and 5.7 nM, respectively. Five minutes later, the mixed solutions were transferred separately into a 1 cm quartz cuvette. The fluorescence spectra were recorded by the fluorescence spectrophotometer.



RESULTS AND DISCUSSION

PEF in Solution Enabled by the Hybrid (AuND-QDs) Assembly. As shown in Figure 1A, the monodisperse AuNDs are 62 nm in diameter with only 7% size distribution. Because their size is much smaller than the wavelength of visible light, the AuND containing solution is completely transparent (inset of Figure 1C). Furthermore, each AuND possesses tens of elongated and sharp subunits. These features in size and structure are critical to the PEF in bulk solution, as discussed below. To study their interaction with negatively charged carboxylterminated QDs (Supporting Information Figure S1), the AuNDs were then modified by cysteamine. As shown in Figure S2 (Supporting Information), the ξ-potential values turn from −24.3 to +11.0 mV; at the same time, the characteristic absorption bands of cysteamine are observed in the FTIR spectrum (Supporting Information Figure S3). These data clearly demonstrate that cysteamine molecules are successfully modified on the AuND surface by the Au−S bond. The AuNDs can well keep their morphology and optical behaviors after the modification, as demonstrated by large-scale SEM images and extinction spectra (Supporting Information Figures S4 and S5), respectively. Figure 1B is the TEM image of TGA-capped CdTe QDs (λEm = 600 nm); they possess clear lattice fringe with 3.5−4.0 nm diameter. For better understanding of the interaction of AuNDs and QDs, we synthesized two other QDs with different emission wavelengths (Figure 1C). Because of their emission peak values, the three QDs are referred to as QDs-545, QDs-600, and QDs-625, respectively. As amine-terminated AuNDs were introduced into the solution containing carboxyl-terminated QDs-600, they formed hybrid (AuND-QDs) assembly by electrostatic and/or hydrogen-bonding attraction (Figure 2A and the left part of Scheme 1). The amine−carboxyl attraction effect has been employed for constructing various hybrid assemblies, such as (Au nanosphere-QDs),16 (Au nanorod-QDs),4b etc. Meanwhile, an obvious fluorescence enhancement process is observed (Figure 3A), and the enhancement can increase up to 3 times by 11.5 pM AuNDs. As shown in Supportng Information Figure S6, the profile of the AuND extinction spectra remains almost invariable in the assembly process, which indicates the present PEF results from the AuND-QDs interactions instead of the AuND aggregation. Therefore, the observed PEF mechanism is different from that of the previous Ag@SiO2 core@shell NP aggregation system.17 In addition, the present 3 times is a rather conservative datum. In the experiments, even more than 8 times enhancement has been observed. We then studied the interactions of the AuNDs with two other sized QDs. As shown in Figure 3B and Supporting Information Figure S8, the

FDTD CALCULATIONS

The FDTD calculations were performed using a soft package EastFDTD Solutions v4.0, developed by Dongjun Technology Co. The details can be found in the Supporting Information. 11064

dx.doi.org/10.1021/ac5031804 | Anal. Chem. 2014, 86, 11062−11069

Analytical Chemistry

Article

with 10−20 nm thick is essential for PEF. Otherwise, the fluorophores would be quenched because of damping of the dipole oscillators by the nearby metal NPs, just as reported previously.3,4 So, it is interesting to ask why the fluorescence enhancement can be observed in solution enabled by such a compact hybrid assembly? To answer this question, it is necessary to retrospect the three kinds of effects of metallic surfaces on fluorophores. One is energy transfer quenching to the metals. A second mechanism is an increase in intensity by the metal amplifying the incident field. The last one is an increased radiative decay rate by the metals, which also results in fluorescence enhancement. For better understanding, the three effects are described in a Jablonski diagram (Figure 4A), as highlighted by km, Em, and Γm, respectively. The quantum yield (Q) and lifetime (τ) in the absence of metal are given by Q=

Γ Γ + k nr

(1)

τ = (Γ + k nr)−1

(2)

where knr and Γ are nonradiative and radiative decay, respectively.20 In the presence of metal NPs, the quantum yield and lifetime can be described as

Figure 2. TEM images of the hybrid (AuND-QDs) assembly before (A) and after (B) addition of 8.8 nM TNT. The concentrations of AuNDs and QDs are 11.5 pM and 5.7 nM, respectively. The AuND and parts of the QDs are marked by white and red dashed curves, respectively.

Qm =

Γ + Γm Γ + Γm + k nr + k m

τm = (Γ + Γm + k nr + k m)−1

(3) (4)

where km is the nonradiative decay rate of FRET effect, and Γm is the metal increased radiative decay rate.9a On the basis of the formulas 3 and 4, the first mechanism (km) causes the decrease in both quantum yield and lifetime. The third effect (Γm) can lead to the increase of quantum yield with lifetime decrease. In terms of the second effect, the increased rate of excitation (from E to E+Em) can result in a fluorescence enhancement without changing intrinsic Q and τ. However, such enhancement can be considered as an “apparent” enhancement of the quantum yield, because the increased rate of the excitation results from the metal plasmon instead of increasing excitation power of the light source. According to Figure 4A, to obtain fluorescence enhancement, the contribution of the second and third effects should exceed that of the first one. To understand the effects of AuND topological structure, we first prepared spherical AuNPs with similar size, and studied their interaction with the QDs. As shown in Supporting Information Figure S9, an obvious fluorescence quenching process is observed at identical conditions. This control experiment demonstrates that the structure feature of the AuNDs is critical to the QD emission enhancement. We then employed FDTD calculations to understand the significance of the AuND structure. Figure 4C and D shows the surface electric field and near field images of the spherical and dendritic Au NPs with identical size, respectively. In Figure 4C and D, we can find that the electric field intensity produced by the AuNDs is 5−8 higher than that of the spherical AuNPs. The reason is that the AuNDs possess many subunits, and their plasmon coupling can lead to very high electric field intensity (“hot spot” effect). Such an effect can effectively increase the rate of excitation, which results in the QD fluorescence enhancement. In fact, the actual electric

Figure 3. (A) Evolution of fluorescence spectra of QDs-600 (5.7 nM) with increasing AuNDs concentrations. The AuNDs concentrations from bottom to top are 0, 1.2, 2.3, 3.5, 4.6, 5.8, 6.9, 9.2, and 11.5 pM, respectively. (B) PEF effects of QDs-545, QDs-600, and QDs-625. F and F0 are the fluorescence intensities of the QDs in the presence and absence of the AuNDs, respectively.

emission of both QDs-545 and QDs-625 can also be enhanced, but their enhanced extents are smaller than that of QDs-600. According to Figures 1C and 3B, the better is the optical band overlap, the higher is the PEF effect, which is in agreement with that of some substrate-based PEF systems.8b,18 As compared to previous (AuNP-QDs) assemblies with fluorescence enhancement behaviors, this system possesses a distinct difference in assembly structure.7a,19 Herein, no any additional spacers between the AuNDs and QDs (SiO2 or polymer layers) are used. At present, a popular view believes that an insulating layer 11065

dx.doi.org/10.1021/ac5031804 | Anal. Chem. 2014, 86, 11062−11069

Analytical Chemistry

Article

Figure 4. (A) Jablonski diagram for the free-space condition and the modified form in the presence of metallic particles. (B) Fluorescence decay curves of the QDs in the absence and presence of the AuNDs. The concentrations of AuNDs and QDs are 11.5 pM and 5.7 nM, respectively. (C)The distribution of surface electric field simulated with single Au nanosphere and AuND. (D) Near-field images simulated with single Au nanosphere and AuND.

Figure 5. (A) UV−vis absorption spectra TNT (1 mM) in the absence and presence of cysteamine (2.5 mg). Inset: The corresponding photographs of TNT solutions without (left) and with (right) cysteamine. After reaction, an obvious absorption peak appears at about 500 nm, and the solution turns from colorless to orange, indicating the formation of Meisenheimer complexes. (B) Evolution of fluorescence spectra of CdTe@CdS core@ shell QDs (5.7 nM) with increasing AuNDs concentrations. The AuNDs concentrations from bottom to top are 0, 1.2, 2.3, 3.5, 4.6, 5.8, 6.9, 8.0, 9.2, and 11.5 pM, respectively. (C) Evolutions of emission spectra of the hybrid (AuND-QDs) assemblies with increasing TNT, whose concentrations from top to bottom are 0, 0.88, 2.2, 4.0, 5.5, 6.6, 7.7, and 8.8 nM, respectively. The dashed line represents the emission of the QDs alone. (D) Plots of the (AuND-QDs) assembly fluorescence intensities versus TNT concentrations.

field intensity of the AuNDs is higher than that of the calculation result. The reason is that the AuNDs possess many sharp edges and tips, which can well enhance electric field intensity by lightning rod effect.21 Because the edges and tips are too irregular to accurately estimate, it is not considered in the calculation. Each subunit of the AuNDs is then an elongated branch with an aspect ratio about 2. On the basis of the calculation of Lakowicz,22 the quasi one-dimensional structure, which has a 1.75 aspect ratio, possesses the strongest capacity for the increase of radiative decay rate. In this regard, the present moderately elongated branches are the ideal structure for a more effective PEF effect. Third, the size of the AuNDs is much smaller (62 nm) than that of visible light

wavelength, which results in a weak scattering effect. Finally, the present hybrid assembly possesses a compact structure because no any SiO2/polymer spacer layers are used. Generally, too close between AuNPs and QDs can lead to the enhancement of nonradiative quenching (km), which is adverse for PEF. However, on the one hand, in the present design, the remarkably increased Em and Γm can well compensate fluorescence quenching; on the other hand, the compact structure can further decrease the light scattering in bulk solution. As a result, the enhanced emission light can smoothly shine out from the solution because of the weak scattering effect (Supporting Information Figure S10). On the basis of the analysis above, the present solution-based PEF can be ascribed 11066

dx.doi.org/10.1021/ac5031804 | Anal. Chem. 2014, 86, 11062−11069

Analytical Chemistry

Article

intensity and the concentrations of TNT in the range from 0 to 8.8 nM (Figure 5D). The detection limit (signal-to-noise ratio of 3) is as low as 0.05 nM. As far as we know, it is one of the most sensitive fluorescent systems for TNT sensing (Supporting Information Table S1). Conceivably, the present higher sensitivity results from two reasons: (1) favorable PEF effect of the assembly system, and (2) higher binding effect of the AuND surface and the analyte targets. It should be noted that the present sensing principle is completely independent of any quenching processes, which is fundamentally different from both the conventional electron transfer quenching mechanism26 and our previous turn-on sensing mode.4b Such “from fluorescent enhancement to recovery” sensing motif is promising to provide more reliable signal readout. According to Figure 5C, almost one TNT molecule replaces one QD in metrology (8.8 vs 5.7 nM). In fact, it is hardly possible, because the TNT molecule is too small as compared to the QD. The reason might result from two aspects. One is the error of particle concentration estimation. For example, because there is no extinction coefficient data for CdTe@CdS QDs, the core@ shell QD concentration is obtained according to the CdTe core. During the formation of CdTe@CdS core@shell structure, the original CdTe core can partially dissolve by Ostwald ripping effect. As a result, the true concentration of the QDs is lower than that of the estimated value. Another reason is only partial QDs bind to the AuND surface, because electrostatic attraction is weaker than that of covalent bond (dozens of kJ/mol vs hundreds of kJ/mol). In terms of the latter, the present AuND-QDs system has the potential for a better PEF effect if the two-particle binding efficiency can be well enhanced (such as covalent binding,2c,29 DNA hybridization,4d,30 antigen−antibody reactions,31 etc.), which would enhance the signal to background ratio and result in a higher sensitivity. Selectivity is a critical parameter to evaluate the performances of a fluorescent chemosensor. For real applications in environmental samples, there can be impurities due to some anions and metal ions. As shown in Figure 6A and B, the common eight ions in natural water and some familiar transition metal ions do not interfere with TNT sensing as their concentrations are 100−1000 times higher than that of the analyte. In comparison, the present sensing platform exhibits lower tolerance toward Cu2+, Hg2+, and Pb2+ ions. Generally, the concentrations of these heavy metal ions in environmental samples are rather lower. In this point, the present (AuND-QDs) assembly can probably act as one of the feasible sensing platforms for commonly environmental water samples. We then studied the effects of common interfering agents, including DNT and NB, on TNT sensing. As described in Figure 6C, the (AuND-QDs) assemblies have almost no responses to NB and DNT in the whole concentration ranges of TNT sensing. The excellent selectivity over NB and DNT probably results from that the electron deficiency of these two molecules is not high enough for competing with the aminecarboxyl binding, because there are only one and two electronwithdrawing nitro groups in each aromatic ring, respectively. These results are similar to those of previous (gold nanorodQDs) system.4b In addition to sensitivity and selectivity, the response times for PEF and subsequent TNT sensing were also studied. As shown in Supporting Information Figure S14, the PEF and recovery processes occur rapidly in the first 4 min, both of which can reach a steady state within 6 min. In contrast, the

to rational design of (1) gold NP morphology and topology, and (2) hybrid assembly structure. The remaining question is how to quantify the three effects in the hybrid (AuND-QDs) assembly. Figure 4B shows an obvious decrease of lifetime decay (from 20.68 to 8.40 ns) in the presence of AuNDs. Because the distance from the QDs to the AuNDs is less than 10 nm,4b km will surely play roles in this system. At the same time, Γm can also decrease the lifetime. On the other hand, the above three processes can modify the steady-state fluorescence simultaneously (km, fluorescence quenching; Em and Γm, fluorescence enhancement). As a result, it is difficult to quantificationally calculate the contribution of the three mechanisms, respectively. Because of the complexity of metal−fluorophore interactions (at least three kinds of mechanisms, and that lots of factors can affect their interactions), such a fundamental photophysical issue needs further study in both experimental and theoretical aspects. Sensing Applications of the Hybrid (AuND-QDs) Assembly. The formation of the (AuND-QDs) assembly is based on electrostatic attraction instead of stronger covalent binding.23 Such an interaction means that competing reactionbased sensing mode is promising to be carried out by the present system.24 To demonstrate the proof-of-concept, TNT sensing was studied because of four reasons. First, TNT is a small molecule that is difficult to detect by the existing substrate-based PEF platform. TNT sensing is important because it is a highly explosive and environmentally detrimental substance.25 Third, most existing TNT fluorescent sensing systems are based on the quenching effect of its electron deficiency property.26 In this regard, exploration of PEF effect for TNT sensing is fundamentally significant. Finally, the present hybrid assembly possesses the convenience for TNT sensing because the AuND surface has TNT binding sites (primary amine groups).27 It is known that Te atoms at CdTe particle surface are sensitive to oxygen and some transition/heavy metal ions, which can lead to the emission instability of the QDs. To obtain better analytical performance, especially a higher selectivity, preformed CdTe QDs were coated with a CdS layer (Supporting Information Figure S11). As shown in Figure 5B, the fluorescence of CdTe@CdS QDs can also be enhanced by the AuNDs, and the PEF effect is even better than that of CdTe (4 times vs 3 times). As TNT molecules were introduced into the (AuND-QDs) system, the fluorescence decreased correspondingly (Figure 5C). In contrast, in QD-only solution, the fluorescence remained constant at the same condition (Supporting Information Figure S12). This control experiment demonstrated that the decrease of fluorescence resulted from the QD fluorescence recovery instead of the TNT quenching effect. The reason for the fluorescence recovery is that the (AuND-QDs) assemblies are broken by the added TNT molecules, as described in Figure 2B. As we know, the benzene ring of the TNT molecule is highly electron deficient because of the synergistic electron-withdrawing effect of its 2,4,6-trinitro substituent groups. As a result, it can react with electron-rich primary amine and form Meisenheimer complex (Figure 5A);28 such a covalence-like interaction is stronger than that of electrostatic binding of amine-carboxyl groups, and the QDs around the AuNDs are replaced by TNT molecules (Scheme 1).4b,24 As described in Figure 5C, the more TNT is added, the more recovery is obtained, and a more than 80% fluorescence recovery is observed in the presence of 8.8 nM. There is a good linear relationship (R = 0.995) between the fluorescence 11067

dx.doi.org/10.1021/ac5031804 | Anal. Chem. 2014, 86, 11062−11069

Analytical Chemistry

Article

Figure 7. Sensing of TNT in tap (A) and lake (B) water samples by the hybrid (AuND−QDs) assemblies.

platform for quenching- and labeling-free sensing applications. Of course, for more effective applications and better sensing performances, it is urgent to more deeply and quantitatively investigate the fluorescence modulation of hybrid (AuNP-QDs) assemblies.

Figure 6. Fluorescence responses of the (AuND-QDs) assemblies to eight common ions in environmental water samples (A), some familiar transition/heavy metal ions (B), and TNT analogues (C). In (A), the concentration of Ca2+ is 1.0 μM, and other ion concentrations are 10 μM; TNT concentration is 8.8 nM. In (B), the concentrations of Cu2+, Hg2+, and Pb2+ are 0.05 μM, and other ion concentrations are 1.0 μM; TNT concentration is 8.8 nM. F0 and F are the fluorescence intensities of the (AuND-QDs) assemblies in the absence and presence of the corresponding substances, respectively.



ASSOCIATED CONTENT

S Supporting Information *

Some additional experimental data and the details of FDTD calculations. This material is available free of charge via the Internet at http://pubs.acs.org.



(AuND-QDs) assemblies themselves can remain stable for a rather longer time (60 min, Supporting Information Figure S15). These results demonstrate that the present sensing platform is feasible for rapid TNT sensing. To further investigate the potential practical applications of this method, the sensing of TNT in tap and lake water samples was carried out. As shown in Figure 7, almost no responses to the (AuND-QDs) assemblies are observed until the samples are spiked with some TNT. Furthermore, the fluorescence recovery degrees are in agreement with the concentrations of added TNT (relative errors