Introduction

Sensitive detection of chemical species using fluorescence has become massively popular in analytical and biological sciences, particularly in cellular and molecular imaging, flow cytometry, medical diagnostics, DNA sequencing, forensics, and genetic analysis [1–6]. Unfortunately, the intrinsically high sensitivity of fluorimetry may be compromised by the limited brightness, excitation cross-section, and photostability of many common molecular probes, leading to unsatisfactory detection limits or excessively long observation times. Several years ago, pioneering work by several groups sought to overcome these limitations by exploiting the interaction between the electrons responsible for surface plasmon resonance in metal nanoparticles and nearby fluorophores [7–13]. By combining the amplification of the incident electric field at or near the surface of nanoparticles and the coupling of excited fluorophores with the electrons in the particles, they succeeded in enhancing excitation and emission rates several-fold, while shortened excited state lifetimes improved robustness towards photobleaching [14]. This process was termed metal-enhanced fluorescence (MEF) to distinguish it from surface-enhanced fluorescence (SEF), which mainly concerns the enhancement of the incident field near metallic surfaces without necessarily involving dipole-dipole coupling between the metal surface and the fluorophore [15]. Other names for this phenomenon include plasmon-controlled fluorescence (PCF) [16] or shell-isolated nanoparticle-enhanced fluorescence (SHINEF) [17]. Today MEF is being investigated as a powerful means to enhance the optical properties of commercially available luminescent dyes and to increase the detection sensitivity of various biological assays and imaging techniques [18–21].

Numerous studies performed using, for example, molecular assemblies over silver island films (SIF) deposited on glass slides [22–24], nanostructured planar systems fabricated with lithographic or ion beam methodologies [25, 26] and single-molecule studies [13, 25, 27] have shown that maximizing MEF for a given nanostructure-fluorophore system hinges on the optimization of several factors, notably the size, shape and chemical composition of the nanoparticles [27–31], the distance between metal nanoparticles and the fluorescent molecule(s) or between adjacent nanoparticles [23, 24, 32, 33] and the degree of overlap between the plasmon band and the fluorescence excitation-emission spectrum [34]. Unfortunately, the complexity of such precisely engineered nanofabricated plasmonic substrates tends to limit their use in biosensing applications, in particular where one needs to probe the interior of cells or extended sample volumes and for which freely diffusing nanostructures are often preferred [20, 35].

In comparison, a core-shell geometry where the nanostructure is in the form of a discrete metal nanoparticle separated from an outer fluorescent layer by a silica spacer shell (or other suitable dielectric material) retains all the advantages provided by MEF, including enhanced brightness and photostability, while being highly dispersible in aqueous media and easily conjugated to target biomolecules [36, 37]. Several studies have been published in recent years on the influence on fluorescence enhancement or quenching of structural parameters such as core shape and size, spacer shell thickness, and spectral overlap [23, 31, 38–40] and have led to a better understanding of structure-property relationships in MEF-enabled core-shell systems. However, because the colloidal synthesis methods generally used to fabricate core-shell nanoparticles do not provide the same level of geometrical control as state-of-the-art lithographic techniques, most studies examining the effects of structural parameters have been fragmented across different nanostructure types/compositions or limited to certain sizes or particular fluorophores. Therefore, a generalized model or methodology was not available to help determine the extent of MEF in multilayer core-shell nanoparticles. Fortunately, significant progress was made recently towards the synthesis of various multilayer core-shell metallic colloids, not only with silver or gold [41–43], but also more reactive elements like indium [30], aluminum [44], and copper [45], with excellent control over nanoparticle geometry, crystallinity, size uniformity, and shell thickness.

In the work presented herein, we have performed a detailed structure-property study in MEF-capable core-shell nanoparticles, using a seed-growth synthesis technique and rigid silica shells as spacers to precisely control the diameter of silver cores and the distance between the core surface and a thin layer of fluorescent silica on the outer surface of the nanostructures. Nanoparticles with nominal core diameters of 30, 40, 60, and 75 nm and spacer shell thicknesses from ∼3 to >25 nm were prepared and coated with a ∼1-nm outer silica layer doped with different fluorophores (fluorescein, eosin, and rhodamine B isothiocyanates) chosen for their varying degree of spectral overlap with the metal core plasmon band. The core-shell nanostructures were examined by transmission electron microscopy (TEM) and their concentration in each sample was determined by elemental analysis, and this information was used to determine the concentration of fluorescent molecules immobilized in the outer shell of each nanoparticle formulation. Their spectral properties were characterized by UV-VIS spectrometry, steady-state and time-resolved fluorimetry. Finally, fluorescence enhancement factors were determined and compared as a function of core size, spacer shell thickness, and spectral overlap.

Experimental Details

Chemicals and Reagents

Sodium citrate tribasic dihydrate (≥99.0 %, ACS grade), fluorescein 5(6)-isothiocyanate (FiTC; 90 %), rhodamine B isothiocyanate (RBiTC; mixed isomers), triethylamine (≥99.5 %), (3-aminopropyl)triethoxysilane (APTES; ≥98 %), ammonium hydroxide solution (28–30 % NH3), dimethylamine (DMA; 40 %wt solution in water), sodium borohydride (99.99 %), and silver nitrate (≥99.0 %) were purchased from Sigma-Aldrich. High purity silver nitrate (99.9995 %) was obtained from Strem Chemicals, eosin-5-isothiocyanate (EiTC) from Marker Gene Technologies, tetraethoxysilane (TEOS; 99.9 %) from Alfa Aesar, and anhydrous N,N-dimethylformamide (DMF; 99.8 %) from EMD Chemicals. Unless otherwise stated, every chemical reagent was used without further purification. Nanopure water (18 MΩ) was used in all experiments and anhydrous ethanol (Commercial Alcohols, Brampton, ON, Canada) was used as a solvent. All glassware for nanoparticles synthesis was conditioned with concentrated nitric acid and rinsed thoroughly with water.

Preparation of Ag Nanoparticles by Seed-Growth

The silver cores were prepared by a seed-growth procedure using silver seeds as a starting material to grow larger nanoparticles. This methodology was adapted from various other sources [46–48] and provides precise control over core diameter from ∼20 to more than 100 nm, with good repeatability with regards to polydispersity and plasmonic properties. In the present study, nanoparticles with well-controlled and uniform sizes (nominal diameters of 30, 40, 60, and 75 nm) were prepared using a succession of growth steps. In a 250-mL round bottom flask, 75 mL of nanopure water was heated to 70 °C under vigorous stirring, 20 mL of 1 %wt aqueous sodium citrate (34 mM) was added and the temperature was allowed to stabilize for 10 min. 1.7 mL of 1 %wt aqueous AgNO3 (59 mM, purity 99.0 %) and 0.2 mL of freshly made 1 %wt aqueous NaBH4 (264 mM) were then quickly added to the sodium citrate solution and the mixture was left to stir for 30 min before letting it cool to ambient temperature. This stock seed solution, with an average diameter of 5 to 10 nm and a plasmon band typically centered at 389 nm, was diluted with nanopure water to 100 mL and stored overnight in the dark at 4 °C to allow the deactivation of unreacted borohydride. Fifteen milliliters of this seed solution was added to a mixture of 75 mL of nanopure water and 2.0 mL of fresh 1 %wt sodium citrate solution brought to a boil in a 250-mL single-neck round bottom flask under reflux and under vigorous stirring using an oil bath at 120 °C. This was followed 30 sec later by the addition of 1.7 mL of fresh 1 %wt AgNO3 and the solution was left to stir for 60 min. At this point in the process, the resulting Ag cores had an average diameter of 20 nm and a plasmon band near 390 nm. The 30-nm silver cores were obtained by adding 1.7 mL of 1 %wt AgNO3 and 2 mL of 1 %wt sodium citrate to this suspension, stirring for an additional 60 min, and repeating this sequence a second time. The plasmon resonance of the resulting 30-nm nanoparticle solution displayed a maximum between 402 and 406 nm. The 45-nm silver cores were prepared using the 20-nm core mixture (diluted to 100 mL beforehand) as starting material using the following procedure: 75 mL of nanopure water and 10 mL of the 20-nm cores were mixed in a clean 250-mL round bottom flask and heated to 88 °C under vigorous stirring. 2.0 mL of 1 %wt sodium citrate was added to the mixture, followed 30 s later by the addition of 1.7 mL of fresh 1 %wt AgNO3. The mixture was left to stir for 2 h, allowed to cool to ambient temperature under moderate stirring, and diluted to 100 mL. The plasmon band of this solution was typically centered between 407 and 410 nm. Finally, the 60- and 75-nm silver cores were synthesized using the 45-nm nanoparticles as starting material: the 60-nm cores were prepared by mixing 50 to 40 mL of nanopure water (30 mL colloid + 60 mL water to prepare the 75-nm core size) and bringing the mixture to a temperature of 85 °C under vigorous stirring, followed by the addition of 2.0 mL 1 %wt sodium citrate and, 30 s later, 1.7 mL 1 %wt AgNO3. All colloid dispersions were kept in the dark at 4 °C before condensation of the silica spacer shell, typically 24 h after their synthesis.

Preparation of Ag@SiO2 Core-Shell Nanoparticles

In order to minimize the partial dissolution of silver by the ammonia used in conventional Stöber sol-gel procedures [49], the condensation of silica shells on our metallic colloids was performed using an ammonia-free synthesis route adapted from the literature [50]. In a 50-mL polypropylene conical tube, 0.20 to 0.85 mL (depending on the desired shell thickness) of a 9-mM TEOS solution in ethanol was mixed with 20 mL of anhydrous ethanol. 2.5 mL of Ag NPs dispersion in water and 0.25 mL of 40 % dimethylamine were added to the TEOS solution before sealing and stirring during 20–24 h. The resulting Ag@SiO2 core-shell NPs were centrifuged at speeds and durations optimized for each colloid size: 15 min at 12,000 RCF for the 30-nm NPs, 15 min at 10,000 RCF (45 nm), 10 min at 10,000 RCF (60 nm) or 10 min at 8000 RCF (75 nm). For the present study, each Ag@SiO2 population with a given core size and shell thickness was prepared in triplicate, then pooled and diluted with ethanol to an optical density of 1.0 (see Figures S1 and Table S1, Supplementary Material†).

Fluorescent Labeling of Ag@SiO2 Core-Shell Nanoparticles

Condensation of an outer thin silica shell incorporating fluorescent silane moieties on the Ag@SiO2 NPs was done using a procedure adapted from our previous work [39]. The fluorescent dyes used in the present study were commercially available isothiocyanate-functionalized organic fluorophores reacted with aminopropyltriethoxysilane (APS) using the following protocol: 23 μmoles of fluorophore was weighted (8.8 mg for FiTC, 16.0 mg for EiTC, 12.2 mg for RBiTC) and dissolved in 450 μL of anhydrous DMF. 6.4 μL of APS and 6.4 μL of triethylamine were added and the resulting mixture was left to react in the dark at room temperature for 3 h. The APS-modified fluorophore solution was diluted to 50 mL with anhydrous ethanol and stored in the dark at -20 °C until use (within 1 week after their preparation). In preparation for the condensation procedure, the actual nanoparticle concentration of the different Ag@SiO2 samples was determined by dissolving the cores and determining the silver content by inductively coupled plasma atomic emission spectrometry (ICP-AES) combined with the estimation of the average spherical core volume by TEM (see Supplementary Material† for details). Using this information, the concentration of each core-shell aliquot was adjusted to provide a total available silica surface of 4 × 1013 nm2/mL for the deposition of the outer silica layer, assuming an identical reactivity of the surface siloxane groups towards the silane coupling agent (independently of core size or shell thickness). Stöber condensation of the outer fluorescent silica shells on the Ag@SiO2 NPs was achieved in triplicate in 2 mL polypropylene vials by addition, in sequence, of 1.50 mL of the normalized Ag@SiO2 NP suspension in ethanol, 5.0 μL of the APS-derivatized fluorophore solution (0.46 mM), 45 μL of the 9-mM TEOS solution in ethanol, and 18 μL of aqueous ammonia. The vials were inserted in a 24-slot thermomixer (Model Thermomixer C, Eppendorf) and the reaction was allowed to proceed for 16–20 h at 20 °C and stirring at 800 rpm. The resulting fluorescent nanoparticles were purified by three consecutive centrifugations and finally dispersed in 800 μL of ethanol. The thickness of the fluorescent outer shell was estimated at (1.0 ± 0.2) nm from TEM images and incorporation yields were calculated (see Supplementary Material† for details).

Characterization

Nanoparticles were characterized by UV-visible spectrophotometry (Model Cary 50, Agilent Technologies), inductively coupled plasma atomic emission spectrometry or ICP-AES (Model Optima 3000, Perkin Elmer), transmission electronic microscopy or TEM (Model Tecnai G2 Spirit, FEI), spectrofluorimetry (Model Fluorolog 3, Jobin-Yvon Horiba), and time-correlated single photon counting time-resolved fluorimetry or TCSPC (Model FluoTime 200, PicoQuant GmbH). Fluorescence decay curves were fitted with a two-exponential model with reconvolution of the instrumental response function (FluoFit data analysis software, PicoQuant GmbH).

Results and Discussion

Absorption Spectra of Ag@SiO2 Nanoarchitectures

Four distinct Ag nanoparticles populations with diameters of (27 ± 4) nm, (40 ± 5) nm, (60 ± 8) nm, and (74 ± 10) nm were synthesized for this study (Fig. 1). Because of changes in absorption and scattering efficiencies and in the local intensity of the plasmon-generated local electric field, these silver colloids are expected to interact with light differently. Mie theory predicts that particles smaller than ∼30 nm support plasmonic oscillations of reduced amplitude and show a characteristic extinction dominated by absorptive phenomena, while larger particles are more efficient at scattering light [51]. Extinction cross-section values measured for the four core sizes vary accordingly, from (2.62 ± 0.02) × 107 nm2 for 27-nm nanoparticles to (19.8 ± 0.3) × 107 nm2 for the 74-nm nanoparticles (see Table S1 in Supplementary Material†). The extent of the plasmon-enhanced electric field into the surrounding media was evaluated by comparing the wavelength shift of the plasmon band for silver cores of various sizes coated with silica shell of varying thickness and dispersed in ethanol (Fig. 2). The results show that the red-shift of the plasmon band increases with spacer thickness until the oscillating electric field is completely confined in the silica shell, at which point the maximal wavelength becomes constant. This study also shows, albeit indirectly, how the plasmon-enhanced electric field extends to a greater distance from larger silver cores, i.e., ∼10 nm for 27-nm cores and over 25 nm for 75-nm cores (Fig. 2C). These experimental values are of particular interest for the design of multilayer concentric core-shell architectures and for controlling the range of LSPR sensing by adjusting the size of the metallic domains and the thickness of the dielectric shells.

Fig. 1
figure 1

a Representative transmission electron micrographs. b UV-VIS extinction spectra in H2O. c Size distribution histograms (C, N > 150, 3-nm bins) of Ag cores after seed-growth synthesis

Full size image
Fig. 2
figure 2

a Transmission electronic micrographs of (27 ± 4) nm Ag cores coated with silica shells (2.7 ± 0.3), (7.1 ± 0.4), (13.1 ± 0.9), and (17.6 ± 0.6) nm in thickness, respectively. b Comparison of extinction spectra for 27-nm Ag cores and Ag@SiO2 NPs with 15-nm thick coating dispersed in ethanol. c Plasmonic shift as a function of shell thickness for different core sizes

Full size image

Influence of Metal Core Size and Spacer Shell Thickness

Ag@SiO2 core-shell nanoparticles with various core diameters and space shell thicknesses were coated with a thin layer of fluorescein—a high quantum yield (95 % at pH > 8) fluorescent dye frequently used as a probe for MEF studies [52]—and their fluorescence intensity was compared as a function of core diameter and spacer shell thickness. The concentration of nanoparticles in each sample was adjusted to contain the same amount of fluorescein molecules incorporated in the outer silica layer (see Supplementary Material†) and care was taken to keep the optical density below 0.05 at the plasmon and fluorescence excitation wavelengths in order to minimize scattering and inner-filter effects. The results, shown in Fig. 3, mirror the plasmonic field range values extrapolated from the extinction measurements shown in Fig. 2, i.e., smaller cores support a weaker plasmon and only enhance the emission of fluorophores located closer from the metal surface (less than 10 nm away from the 27-nm cores). Interestingly, we observed that signal intensity values reach the same maximum values (within experimental error) at the same distance from the core (∼5 nm) for all nanoparticle core diameters except for the smaller ones. This result confirms a previous study performed on silver island films [27] in which the MEF enhancement was reported to be relatively constant, at a distance of ∼8 nm from the core, for silver particle diameters in the 40- to 70-nm range. As a consequence, a majority of the fluorescence signal originates from a region 5-10 nm from the metal surface for the 45-nm cores (5–15 nm for the 60- and 75-nm cores).

Fig. 3
figure 3

Fluorescence intensity for Ag@SiO2@SiO2 + FiTC (λexc = 485 nm, λem = 520 nm) with varying core sizes and silica spacer thickness (N = 3)

Full size image

Since the enhancement of fluorescence close to a metallic surface is known to be accompanied by a decrease in the lifetime of excited states, fluorescence lifetime measurements are a useful indicator of MEF. Fluorescence decay curves were acquired for Ag@SiO2@SiO2 + FiTC core-shell NPs as a function of core size and distance from the core to the outer fluorescent layer, and the results (Fig. 4 and Table S2 in Supplementary Material† for complete lifetime data) show fluorescence lifetime values increasing with distance from the core for all core diameters. The shorter range of the plasmon from the 27-nm cores hypothesized from Fig. 3 can be observed in the fluorescence lifetime values which seem to plateau when the fluorescent layer is positioned beyond 10 nm from the core surface. The independence of fluorescence enhancement on core size for 60- and 75-nm cores observed on Fig. 3 is confirmed by very similar excited state lifetimes for all spacer shell thickness values of the two core-shell nanostructures. Interestingly, the fluorescence lifetime values measured for the thicker spacers remain inferior to the value of 3.03 ns measured for FiTC in solid silica nanoparticles (SiO2@SiO2 + FiTC), an indication that the influence of the core plasmon extends beyond the silica shells investigated in this work.

Fig. 4
figure 4

Average fluorescence lifetimes for Ag@SiO2@SiO2 + FiTC with varying core sizes and silica spacer thickness (two-exponential model with IRF deconvolution using non-fluorescent Ag@SiO2 NPs suspended in ethanol; N = 3)

Full size image

Quantification of Enhancement Factor (EF)

Formalisms were developed previously for the determination of fluorescence enhancement factors (EF) for various systems including single molecules in bowtie nanoantennas [25] and silica-coated silver island films [24]. In the case of freely diffusing MEF-capable core-shell nanostructures, the development of a generalized methodology for the determination of EFs is hindered by the propensity of colloidal suspensions to aggregate and change with time; consequently, various strategies were devised by research groups to determine EFs for their particular systems [30, 36, 39, 53, 54].

The EF in core-shell systems is generally defined as F NP /F 0, where F NP is the emission intensity measured from a certain number of fluorescent molecules in the vicinity of plasmonic cores, and F 0 is that measured from an equal number of emitters without enhancement. Most of the discrepancy in EF values in literature comes from the difficulty of measuring F 0 in conditions comparable to F NP [53]. For example, Tovmachenko et al. [55] and Lin et al. [56] used silica nanoparticles coated with fluorescent molecules as a reference; a disadvantage to this approach is the difficulty to accurately quantify the concentration of silica NPs used to estimate F 0. Another method involves dissolving the metal cores with chloride or cyanide ions, nitric acid, or hydrogen peroxide and measuring the intrinsic signal of the remaining nanoshells [30, 36, 39]. However, this process may take from a few hours to a few days to complete depending on the oxidant used and the porosity of the silica shell, during which time nanoparticle aggregation and sedimentation can occur. Purification of the nanoshells by centrifugation may also induce losses, which will result in overestimated EF values.

The method used in the present work assumes an identical number of molecular emitters generating F NP and F 0 in the probed volume and requires that the concentration of fluorophore and nanoparticles be chosen so that inner-filter effects are minimized. As such, we calculated EF as the ratio of the raw fluorescence measured in core-shell suspensions by the signal of the incorporated molecules (F incorporated). However, this value is impossible to quantify directly: we chose to subtract the signal of the supernatant after condensation of the fluorescent outer shell (F residual) from a reaction blank (F total, see Supplementary Material†) in otherwise similar conditions of solvent, pH, etc.:

$$ EF=\frac{F_{\mathrm{NP}}}{F_0}=\frac{F_{\mathrm{NP}}}{F_{\mathrm{incorporated}}}=\frac{F_{NP}}{F_{\mathrm{total}}-{F}_{\mathrm{residual}}} $$

Figure 5a shows the EF for the Ag@SiO2@SiO2 + FiTC system as a function of core size and spacer shell thickness calculated from the intensity readings of Fig. 3. The maximum EF values of ∼10 obtained for fluorescein ∼5 nm away from cores 40–75 nm in diameter are similar to values measured in similar core-shell nanostructures by the core dissolution method [57, 58]. The region within 5 nm from the core is presumably dominated by quenching effects, while fluorescent molecules 15 nm or more from the core are largely beyond the reach of the plasmon. These values could be useful as guidelines for the design of simpler, spacer-less core-shell nanoparticles, knowing that fluorescent coatings thicker than 10–15 nm will not improve signal enhancements significantly [35, 59].

Fig. 5
figure 5

a Fluorescence enhancement factors (EF) for Ag@SiO2@SiO2 + FiTC (λexc = 485 nm, λem = 520 nm) with varying core sizes and silica spacer thickness (N = 3). b Enhancement factors for 74-nm cores with different xanthene fluorophores and silica spacer thickness

Full size image

Enhancement factors were also determined by the same method for Ag@SiO2@SiO2 + dye core-shell nanoparticles prepared with two other fluorophores with different quantum yields and degrees of spectral overlap with the core plasmonic band. Fluorescein, eosin, and rhodamine B are xanthene dyes with similar quantum yields (0.95, 0.70, and 0.95, respectively) and excited state lifetimes (3.9, 2.8, and 2.5 ns, respectively) in ethanol. However, the degree of overlap of their excitation-emission spectra with the plasmon band varies considerably, i.e., excitation bands are centered at 490, 530, and 550 nm for FiTC, EiTC, and RBiTC, respectively (Figure S5, Supplementary Material†). Figure 5b shows the enhancement factors measured for FiTC, EiTC, and RBiTC as a function of spacer thickness for 75-nm cores. There is a clear dependence on the overlap of the plasmonic band (maximum at 460 nm) with the fluorophores’ excitation band: EF values for ∼5-nm spacers decrease from 10.9 (FiTC), to 4.9 (EiTC) to 3.5 (RBiTC). For smaller particles with a blue-shifted extinction, this relation becomes especially apparent with the loss of good overlap for RBiTC (Figure S6 in Supplementary Material†). Since Mie theory predicts that the increase of the plasmonic core size induces a higher efficiency and a notable red-shift of the scattering mode, a dipole-dipole interaction for 60- and 75-nm cores can be achieved with EiTC and RBiTC: in comparison, this coupling is lost for absorption-dominated 27- and 40-nm silver cores. Interestingly, Fig. 5b also demonstrates that the optimal distance for maximal EF is identical for all fluorophores with sufficient overlap. Therefore, in these conditions, MEF can be used to investigate plasmonic properties in core-shell nanostructures and improve their design to incorporate sensitive fluorescent molecules of interest.

Summary and Conclusion

In this paper, we have used Ag@SiO2@SiO2 + fluorophore NPs with well-controlled geometry as tunable systems to study metal-enhanced fluorescence with regard to the position of molecules in the plasmon-induced electric field, the degree of spectral overlap with different fluorophores, and the size of the metallic nano-domain. The combination of steady-state and time-resolved measurements has allowed to identify optimal values for key parameters in the design of colloidal MEF sensors: (1) Metallic particles with low absorptive and high scattering efficiencies—e.g., larger silver NPs up to 75 nm—offer a better plasmonic range to induce dipole-dipole interactions with neighboring fluorophores; (2) Maximum enhancement of fluorescence is observed between 10 and 15 nm from the metal surface, whereas quenching phenomena are dominant 5 nm or less from the core; (3) For silver cores, MEF is highly dependant on the degree of spectral overlap with fluorophores. Our experiments involved three available commercial fluorophores (fluorescein, eosin, rhodamine B), but we believe the methodology and conclusions of the study could be extended to other nanoparticle compositions. Because many labeling molecules in biology and biochemistry have fluorescence properties in this region of the visible spectra (450–600 nm), we believe our conclusions on the optimal distance in plasmonic core-shell architectures could be generalized and applied in future generation of MEF-enabled probes.