A plasmon-tuned ‘gold sandwich’ for metal enhanced fluorescence in silica coated NaYF₄:Yb,Er upconversion nanoparticles

Abstract

The low quantum yield of luminescence from lanthanide-doped up-conversion nanoparticles (UCNPs) has been enhanced by using an optimized ‘gold sandwich’ with a transparent top layer and a reflecting bottom layer at 980 nm excitation. Erbium (Er)-doped UCNPs, with a NaYF4:Yb,Er core, were synthesized by a thermal decomposition process and coated with silica to assist in metal-enhanced fluorescence (MEF). A bottom layer of thick coalesced gold island film, acting as a mirror, increases the optical path length of the 980 nm radiation through the UCNP layer dispersed on it. This layer enhances the UCNPs' 540 nm green emission by a factor of 5–8 compared to that in the absence of the gold reflector. A thin nanoparticle-like gold layer on top of the UCNPs, with a surface plasmon absorption around ∼550 nm, completes the sandwich, which augments the luminescence enhancement by another factor of ∼2.5, thus taking the net enhancement factor to ∼13–19 when compared to the luminescence in the absence of the gold-sandwich. The surface plasmon absorption in the top gold layer enhances the local electric field at the UCNPs to promote their radiative decay. Compared to previous reports, mostly for the solution state, the current case study is a solid state measurement.

---

Introduction

Bioimaging by photo luminescence (PL) generally employs exogenous contrast agents, such as organic dyes,¹ fluorescent proteins,² metal complexes,³ and semiconductor quantum dots (QDs).⁴ Most of these conventional contrast agents utilize Stokes-shifted emission, using excitation in the range of the ultraviolet (UV) or blue-green visible spectrum. As a process, this suffers from unwanted autofluorescence, strong light scattering from biological tissues, low penetration depth of high energy photons, and possible DNA damage and cell death. Moreover, there is also serious concern about the toxicity of heavy metal-based QDs for bioimaging, as they contain toxic elements. Lanthanide-doped up-conversion nanoparticles (UCNPs) are a promising class of new-generation bioimaging agents that uses excitation in the biological transparency window (∼1000 nm) to ensure deep tissue penetration.^5,6 The up-conversion process utilizes sequential absorption of multiple low energy (IR) photons through the use of long-lifetime and ladder-like energy levels of trivalent lanthanide ions embedded in an appropriate inorganic host lattice to produce higher energy (visible) anti-Stokes luminescence.⁷ However, the quantum yield of UCNPs is relatively low (<1%) due to the small absorption cross-sections.⁸ The up-conversion PL intensity (I_UCPL) generally has a dependence on the power P of the excitation light given by⁹

I_UCPL = KPⁿ (1)

where ‘n’ is an integer that determines the number of photons utilized in the PL process, and K is a material-related coefficient. The up-conversion quantum yield (UCQY), defined as the ratio of the number of the emitted up-converted photons to the number of the absorbed NIR photons, is strongly dependent on P and is given by⁹

UCQY ∞ Pⁿ⁻¹ (2)

In addition, compared to their bulk counterparts, the dopant ion concentration on the surface of the UCNPs is relatively high and is expected to be quenched by surface quenchers.¹⁰ These factors resulted in a reduced I_UCPL, limiting their application in deep tissue imaging.

Five strategies have been employed to achieve a high UCQY in UCNPs: (i) selection of novel host materials, (ii) tailoring the local crystal field, (iii) engineering the energy transfers, (iv) suppression of surface-related deactivations, and (v) plasmonic enhancement, of which the latter will be the topic of this report. From the theory of metal-enhanced fluorescence (MEF), it is well known that localized surface plasmon resonance (LSPR) in metallic structures can enhance the fluorescence from adjacent fluorophores,^11,12 when they are separated by a spacer of suitable thickness. This strategy has been used to enhance UCPL from NaYF₄ doped with lanthanide ion pairs of Yb³⁺/Er³⁺ or Yb³⁺/Tm³⁺, with a suitable spacer, deposited on solid metallic films of dense gold NPs producing ∼5.2 fold enhancement, or on silver NPs for ∼45 fold enhancement,¹³ or on gold nanorods to produce 22.6 fold enhancement.¹⁴ The fluorescence enhancement factor has to be considered carefully as it depends on several aspects, such as the material, size, and morphology of the fluorophores, as well as the plasmonic nanostructures, type of excitation, and particular emission band frequency, among others.¹⁵ Extraordinarily high enhancement factors, exceeding 100, have been reported when using pillar or rod-like metallic nanostructures that act as an antenna in concentrating light to the fluorophores.¹⁶ One previous study showed ∼5 fold overall enhancement of emission in NaYF₄:Yb,Er UCNPs, without a shell or a spacer, when coupled with sputtered gold island films (GIF),¹⁷ which should normally cause PL quenching. These effects have been attributed to an increased concentration of excitation light from the metallic nanostructures, rather than conventional MEF, which requires a dielectric spacer.¹⁸ Er-doped nanoparticles, although known for green emission around 540 nm, have been found to emit in the red and in the infrared. These emissions in the red or infrared had enhancements in the range of 450×.¹⁹ In contrast, the conventional MEF of the 540 nm band has been mostly lower than 10×,¹⁵ rarely exceeding 50×.^16,20 In other reports, the mechanisms behind the enhanced UCPL from metal island films were inconsistent and unclear because of the large mismatch in the plasmon absorption bands of these films (400–600 nm) and the conventional excitation for the UCNPs (980 nm).

In this work, we propose a plasmon tunable ‘gold sandwich’, within which the MEF-active core–shell UCNPs are embedded for the enhancement of the 540 nm green emission band. To maximize the MEF from the UCNPs, we used a ‘gold sandwich’ whose component layers are optically absorbing in the visible (top layer), and reflecting in the infrared (bottom layer) to couple efficiently to the UCNPs. The thick bottom gold layer has been tuned to reflect the 980 nm excitation like a mirror, and increase the net absorption. The top gold layer, transparent to the excitation wavelength, has been tuned to increase the local electric field at the emission wavelength (520–540 nm) of the Er-doped UCNPs. Our results are critically analysed against the available literature for a better understanding of the proposed design.

Experimental

Materials

Yttrium trichloride (YCl₃, 99.99%), ytterbium(III) chloride (YbCl₃, 99.99%), erbium(III) chloride (ErCl₃, 99.99%), polyoxyethylene (5) nonylphenyl ether (CO-520, average molecular weight of 441), sodium hydroxide (98%), ammonium fluoride (98%), silver nitrate (98%), ethylene glycol (99%), and ethanolamine (99%) were purchased from Sigma-Aldrich (USA). Oleic acid (90%), octadecene (90%) and tetraethylorthosilicate (TEOS, >99%) were purchased from Acros (Belgium). All the chemicals were used as received without further purification.

Synthesis of NaYF₄:Yb,Er core nanoparticles

Up-converting lanthanide NaYF₄:18% Yb, 2%Er nanoparticles were synthesized as reported previously,²¹ as follows: YCl₃ (0.8 mmol), YbCl₃ (0.18 mmol) and ErCl₃ (0.02 mmol) were mixed with 6 mL oleic acid and 15 mL octadecene in a 50 mL flask. The solution was heated to 160 °C under vigorous stirring to form a homogeneous solution and then cooled to room temperature. Subsequently, 10 mL methanol solution of NaOH (2.5 mmol), and NH₄F (4 mmol) were slowly added into the flask and vigorously stirred for 30 minutes. The solution was then slowly heated to remove methanol, and degassed at 100 °C for 10 minutes, followed by heating at 300 °C for 1 h under nitrogen flow. After the solution was cooled naturally, the nanoparticles were precipitated from the solution with acetone and washed three times with ethanol/water (1 : 1 v/v).

Synthesis of the SiO₂ shell on the NaYF₄:Yb,Er nanoparticles

To coat the hydrophobic NaYF₄:Yb,Er nanocrystals, in cyclohexane, surfactants and ammonia were added to form a water-in-oil reverse microemulsion. A relatively high nanocrystal concentration was used, and the emulsion was sonicated to ensure that all the nanocrystals were encapsulated in a microemulsion pool. The procedure to coat the NaYF₄:Yb,Er nanoparticles with a SiO₂ layer, as reported previously,²¹ is as follows: 0.1 mL of CO-520, 6 mL of cyclohexane, and 4 mL of 10 mM NaYF₄:Yb,Er nanoparticles in solution in cyclohexane were mixed and stirred for 10 min; 0.4 mL of CO-520 and 0.08 mL of 30 wt% ammonia were then added and the container was sealed and sonicated for 20 minutes until a transparent emulsion was formed. Then, 0.04 mL of TEOS was added into the solution. The thickness of silica coated on the outside of the UCNPs could be adjusted by controlling the concentration of TEOS. The solution was stirred for 48 hours at a speed of 600 rpm. NaYF₄:Yb,Er/SiO₂ core/shell nanoparticles were precipitated by adding acetone, and washed twice with ethanol/water (1 : 1 v/v), and then re-dispersed in water.

Synthesis and preparation of single and sandwich layer of gold-coated Si and NaYF₄:Yb,Er/SiO₂ nanoparticles

The synthesis and preparation of GIFs on crystalline silicon (c-Si) substrates for MEF was done as follows: firstly, Si substrates (1 cm × 2 cm) were cleaned and sonicated in DI water and acetone, separately, for 15 and 30 minutes, respectively, followed by drying under a nitrogen stream. Gold was coated on these substrates with a sputter coater (Emitech K550X, United Kingdom) for 2, 4, and 8 minutes. Subsequently, 50 μL of 10 mM NaYF₄:Yb,Er/SiO₂ core/shell nanoparticles were drop-coated on each gold-coated c-Si substrate and dried in ambient air. Secondly, to complete the ‘gold sandwich’, a top coat of gold nano-island was also applied on the NaYF₄:Yb,Er/SiO₂ nanoparticles dispersed on the bottom gold layer. The thickness of the top gold coating was controlled by varying the sputtering time from 1–8 minutes.

Characterization of morphology, structure and emission

Transmission electron microscope (TEM) images were taken using a JEM-2000EX, JEOL, Japan. High-resolution TEM (HRTEM), and selected area electron diffraction (SAED) patterns were taken with a JEM-2010F, JEOL, Japan. Scanning electron microscopy (SEM) images for investigating the morphology and surface roughness of gold thin-film-coated silicon wafer were obtained using a field emission SEM (FESEM, 6700F, JEOL, Japan). UV-vis absorption measurement was performed using a UV-visible-near-infrared spectrophotometer (V-770, JASCO, Japan). Up-conversion and metal-enhanced fluorescence spectra were measured with a Fluorolog-3 fluorescence spectrofluorometer (Jobin Yvon, USA), using a 980 nm laser head (SDL-5000T, Shanghai – China) with a 60° angle of incidence and power up to 5 W as the excitation source. All the measurements were performed at room temperature. All the fluorescence spectra presented here are an average of 5 spectra obtained on different locations of the substrate.

Results and discussion

The morphology and size distribution of the wet-chemically prepared NaYF₄:Yb,Er nanoparticles, with and without the SiO₂ shell, are shown in Fig. S1 (see ESI†). The core UCNPs have a lattice spacing of 0.3 nm, and a diameter of 8 ± 2 nm. The use of a SiO₂ shell on the core UCNP is important for stabilizing the core and enabling dispersion in aqueous media instead of cyclohexane, which cannot be used for cellular or tissue imaging, and for using the core–shell structure in MEF. The most commonly used methods for developing a SiO₂ shell on the core UCNP is the Stöber method and the microemulsion method.¹⁷ The microemulsion method has been used for coating SiO₂ on hydrophobic nanocrystals, such as QDs and Fe₃O₄ nanoparticles.^18,19 However, coating individual nanoparticles with very thin shells is quite challenging.^18,19 The SiO₂ shell thickness could be tuned to achieve an optimum plasmon coupling to the core UCNP when dispersed on metal nanoparticles or thin films to be used for MEF.^20,22,23 In this case, a SiO₂ shell of 10 nm was used (Fig. S1, ESI†). The chemical composition and structure of the core and the core/shell-type UCNPs are shown in Fig. S2 (ESI†).

Fig. 1a shows optical photographs of green luminescence from the NaYF₄:Yb,Er core (in cyclohexane) and the NaYF₄:Yb,Er/SiO₂ core/shell nanoparticles (in water), respectively. Characteristic emission peaks of Er-doped UCNPs, without (dotted line) and with (solid line) the SiO₂ coating, were observed at 520, 540, and 655 nm due to transitions from ²H_11/2 to ⁴I_15/2, ⁴S_3/2 to ⁴I_15/2, and ⁴F_9/2 to ⁴I_15/2, respectively (Fig. 1b). The spectra indicate that there was about a 40% decrease in the green emission intensity at 520 and 540 nm in the core–shell structure when compared to the core-only (Fig. 1b). However, the decrease in the yellow and red emissions at 550 and 655 nm, respectively, was about 25% in the core–shell NPs. This is why the core–shell UCNPs appear more yellowish than the predominantly green-core UCNPs (Fig. 1a). Fig. S3 (ESI†) shows the power-dependent UCNP emission spectra of the core and the core/shell type NPs. Such a decrease in fluorescence intensity post-silica-coating has been observed before.²¹

Fig. 1 Optical photograph of quartz cells showing green luminescence from (a) core NaYF₄:Yb,Er nanocrystals (in hexane) (left), and core–shell-type NaYF₄:Yb,Er/SiO₂ (in water) (right) nanoparticles when excited with a 980 nm laser. (b) The up-conversion emission spectra of the core (dotted line), and core–shell (solid line)-type nanoparticles under 2 W excitation; (c) photostability, for ∼2 h, of the core and core–shell nanoparticles of NaYF₄:Yb,Er under 4 W continuous-wave 980 nm excitation. (d) Absorption spectra of the core (dashed line), and the core shell (solid line) nanoparticles of α-NaYF₄:Yb,Er.

Although the 540 nm emission from the NaYF₄:Yb,Er core was consistently higher than the NaYF₄:Yb,Er/SiO₂ core/shell NPs, both sets demonstrate remarkable photostability for a period of 2 hours under continuous exposure to 980 nm excitation (Fig. 1c). The core-only UCNPs displayed higher optical absorption around 980 nm than the core/shell UCNPs (Fig. 1d). A reduced absorption of the host and absorber matrix in the UCNPs is expected as the effective refractive index around the core is changed by the application of the shell.

Metallic nanostructures, including nanofilms and nanoparticles (NPs) have been widely used in MEF, including the UCNPs.²⁴ The optical properties of these metal films can be tailored by manipulating the organization of the constituent nanostructures, columns,^24,25 islands, or layers.^26,27 In particular, ultrathin gold or silver island films show frequency-tunable surface plasmon (SP) absorption bands in the visible and near infrared (400–1000 nm).^28,29 The LSPR of the metallic nanostructures then increases the radiative rate and/or the excitation intensity, producing luminescence enhancement. The luminescence quenching, which stems from resonance energy transfer from the fluorophore or UCNPs to the metallic structure, can be prevented by introducing a spacer layer to delineate the fluorophore from the metal.³⁰ MEF in UCNPs has been reported before, using a variety of gold or silver nano-morphologies, including film,¹⁷ nanoparticles,¹³ and nanorods.¹⁴ However, the enhancement factors were not impressive,¹⁷ and sometimes conflicted with the theoretical requirement of a spacer layer (SiO₂ used here). Intriguingly, MEF in UCNPs, with 980 nm excitation, was reported with the use of Ag/Au NPs and attributed to SP absorption, which is in the visible spectrum (400–600 nm).³¹

Here, we introduce our design of a ‘gold sandwich’ for the observation of MEF in the UCNPs (Fig. 2). The gold sandwich would have a top (T), and bottom (B) layer, both tunable in surface coverage, morphology, and thickness by the sputtering time used. With a sputtering machine at our disposal, this would mean low-density gold nano-islands, coalesced gold islands, and thick layers of gold island with gold nano-clusters, with increasing sputtering time. Fixed volume UCNPs dispersed on clean silicon served as the control (Fig. 2a). Gold-coated silicon (bottom layer) was prepared as a function of the gold sputtering time and optimized for the maximum luminescence from the same volume of UCNPs dispersed on them (Fig. 2b). The sputtering time for the bottom gold layer, which yields the best luminescence from the UCNPs, would be chosen for the application of the top gold layer, whose sputtering time would have to be optimized separately (Fig. 2c and d). This way, we would have a ‘gold sandwich’ that will result in the best possible MEF from these UCNPs. We have carefully prepared gold-coated c-Si substrates by sputtering and optimized for the best MEF.

Fig. 2 Schematic of the proposed ‘gold sandwich’ for metal-enhanced fluorescence in up-conversion nanoparticles (UCNPs): (a) conventional setup of UCNPs on bare silicon; (b) metal-enhanced fluorescence (MEF) of UCNPs on gold-coated silicon substrate, and completing a ‘gold sandwich’ by the use of a (c) thin, and (d) thick top gold layer on the UCNPs for MEF. The red arrow signifies the excitation light, and the green arrow signifies the up-conversion visible emission. The longer green arrow signifies stronger emission.

Fig. 3a–c shows SEM images of the bottom gold layer, on c-Si, morphology for different sputtering times of 2, 4, and 8 minutes, respectively. The SEM images clearly reveal discontinuous GIF morphology for 2 minutes coating time (Fig. 3a), which becomes a coalesced quasi-continuous film by 4 minutes (Fig. 3b) coating time. Further increase in the sputtering time yielded fully coalesced, continuous, and comparatively thicker gold films (Fig. 3c). Fig. 3d shows a representative SEM image of the gold sandwich structure with the UCNPs embedded therein. The thickness of the UCNP layer, shown in Fig. 3d, is much higher than those used for the fluorescence measurement and just used to take the SEM image. Coupling of the UCNP phosphors with these metallic surfaces should lead to enhanced fluorescence according to earlier reports.^30,32 Tuning the optical reflection and absorption of the gold sandwich may efficiently couple more light into, and concentrate a plasmonic electric field at the UCNP site to facilitate enhanced fluorescence.

Fig. 3 Top-view SEM images showing the morphology of the gold island films on Si substrate produced by sputtering for (a) 2, (b) 4, and (c) 8 min. (d) Cross-section view of a representative gold sandwich structure showing top Au layer-core–shell UCNPs – bottom Au layer. The inset (in d) shows a high-magnification SEM image of the gold sandwich structure.

To verify the hypothesis mentioned above, fixed-volume UCNPs were dispersed on these gold-coated c-Si and measured for their luminescence. To characterize the gold sandwich, we will use a (B + T) identification tag, where B and T represent the sputtering time (in minutes) of the bottom (B) and top (T) gold coating, respectively. So, (0 + 0) would represent luminescence from UCNPs on bare c-Si with no top or bottom gold layer, (2 + 0) would represent a sample with a bottom gold layer of 2 minutes and no top gold layer, and (8 + 1) would imply a gold sandwich with a bottom gold coating of 8 minutes, and a top gold coating for 1 minute. All samples were prepared and measured under identical conditions.

Fig. 4a exhibits MEF from the core–shell UCNPs for all the samples (2 + 0), (4 + 0), and (8 + 0), compared to (0 + 0), where (8 + 0) showed the best fluorescence signal, which we would identify as the optimized bottom gold layer coating time. Maximum enhancements of ∼4, ∼6, and ∼5 fold were observed for the 520, 540, and 655 nm emissions, respectively, for the (8 + 0) compared to the (0 + 0) configuration. The spectral emissions were similar to those in solution (Fig. 1b) with signature peaks at 540 and 655 nm and a comparatively weaker emission peak at 520 nm.²¹ The intensity variation of the main spectral features of the UCNP emissions at 520, 540, and 655 nm, as a function of the excitation power density, is shown in Fig. 4b–d, respectively, for the (0 + 0), (2 + 0), (4 + 0), and (8 + 0) configurations. The slopes of the linear fits of the intensity variation (Fig. 4b–d) are between 1.8–2.1, indicating approximately 2 participating photons (‘n’) in the luminescence process, according to eqn (1), which is consistent with earlier reports.^9,33–35 The spectral information is consistent for all excitation powers used (Fig. S4, ESI†).

Fig. 4 (a) Up-conversion luminescence spectra, and power dependence of the (b) 520, (c) 540, and (d) 655 nm peak emission of NaYF₄:Yb,Er/SiO₂ on Si only (0 + 0), and Si coated with 2 (2 + 0), 4 (4 + 0), and 8 (8 + 0) minutes of gold. A laser power density of 4 W cm⁻² was used in (a). The line joining the data points in (b–d) represents linear fits to the data points (symbols) with the slopes mentioned against each label. The spectra in (a), and the symbols in (b–d) represent average measurements done on five different spots on the samples. The legends in each plot are written in the (B + T) format, which represents sputtering time (in minutes) of the bottom and top gold coating, respectively.

These very samples were then coated with a top gold layer, completing the sandwich. The sputtering time of the top layer was varied from 1–8 minutes. However, the best emission enhancement was observed for the 1 minute top coating, and further thickening of the top layer resulted in a decrease in the emission intensity. Fig. 5a exhibits MEF from the core–shell UCNPs for the samples (8 + 0) and (8 + 1), compared to (0 + 0). This presentation would reflect the efficacy of the bottom and top gold layer, separately, in the emission process, with respect to those without any use of gold. The ‘gold sandwich’ (8 + 1) yields enhancement factors of ∼13, 17, and 19 for the emissions at 520, 540, and 655 nm, respectively, against the no-gold (0 + 0) configuration, and enhancement factors of ∼2.5, 2.3, and 2.7, respectively, against the (8 + 0) configuration. The ‘gold sandwich’ proved to be the most efficient enhancer of luminescence compared to other configurations. Spectral data for thicker top-coating configurations of the ‘gold sandwich’, such as (8 + 2) to (8 + 8), and other bottom layer variations, are shown in Fig. S5 (ESI†) for a complete understanding.

Fig. 5 (a) Up-conversion luminescence spectra, and power dependence of the (b) 520, (c) 540, and (d) 655 nm peak emission of NaYF₄:Yb,Er/SiO₂ on Si only (0 + 0), on Si with 8 min Au bottom-coated (8 + 0), and on Si with 8 min bottom and 1 min top Au-coated (8 + 1). Inset in (a) shows the comparison of the maximum enhancement factor of 520, 540, and 655 nm peak emission from (8 + 0) and (8 + 1) sample configurations, normalized with respect to the control (0 + 0), assumed to be 1. The insets in (b–d) show the variation of the intensities of 520, 540 and 655 nm emission lines at lower than 1 W cm⁻² power density. A laser power density of 4 W cm⁻² was used in (a). The line joining the data points in (b–d) represents linear fits to the data points (symbols) with the slopes mentioned against each legend. The legends in each plot are written in the (B + T) format, which represents sputtering time (in minutes) of the bottom and top gold coating, respectively.

The intensity variation of the main spectral features of the UCNP emission at 520, 540, and 655 nm, as a function of excitation power density is shown in Fig. 5b–d, respectively, for the (0 + 0), (8 + 0), and (8 + 1) configurations. The slopes of the linear fits of such variation indicate the number of participating photons in the luminescence process. In the power density range of 1–5 W cm⁻², all of the peak intensity variations reveal a slope between 1.75–2.03 for all of the (0 + 0), (8 + 0), and (8 + 1) configurations, indicating a two-photon involvement in the luminescence process. However, in the <1 W cm⁻² range of power densities, the (8 + 1) configuration yielded much higher slopes in excess of 4, whereas the other configurations exceeded 3 (inset, Fig. 5b–d). This indicates the possibility of a much more efficient multiphoton (>3) process demonstrated in the ‘gold sandwich’, compared to the two-photon processes demonstrated in other substrates here. The insets in Fig. 5b–d indicate a higher slope, and therefore higher ‘n’ values exceeding 3. Multi or three-photon processes are indicated by the presence of the signature emission peaks at 384 and 410 nm, appearing due to the transitions from ⁴G_11/2 to ⁴I_15/2 and ²H_9/2 to ⁴I_15/2, respectively,^36,37 as observed in our case also (Fig. S6 (ESI†)). The blue peaks are also enhanced in the presence of the plasmonic sandwich (8 + 1) case, compared to the (8 + 0) and (0 + 0) cases (Fig. S6 (ESI†)). The general spectral features other than peak intensity, such as peak position, may not change with the increase in ‘n’⁹ but the probability of high-energy transitions appearing in the blue region of the spectrum is increased as observed here when using the ‘gold sandwich’.

To understand the effect of the bottom and top GIFs constituting the sandwich, we have to study their optical properties. First, let us consider the bottom GIF (on silicon) of the sandwich, as shown in Fig. 6a. The bare Si substrate, as control, shows a typical silicon reflectance spectrum (black line) with a decreased absorption below the band gap at 1100 nm energy. The data shows that the reflectance near 980 nm increases significantly (absorption decreasing) with increasing thickness (2–8 minutes of sputtering) of the gold layer. The result indicates a mirror-like gold coating (8 minute gold sputtering on Si), which reflects 90% of the 980 nm excitation, which is also the absorption band of the UCNPs (shown by the dashed band around 1000 nm in Fig. 6a). The elevated reflection of the 980 nm radiation by the bottom gold layer would ensure increased optical path length and superior absorption in the UCNP layer. Such enhancement was also observed on the 2 and 4 minute gold-coated Si substrates, but in lesser proportions (Fig. 4a).

Fig. 6 (a) Reflectance spectra (left axis) of silicon only (star), and silicon substrates coated with 2 (triangle), 4 (circle), and 8 (square) minutes of gold by sputtering. Short dashed line (green) shows NaYF₄:Yb,Er/SiO₂ absorption band (right axis) at 980 nm. (b) Reflectance (hollow symbols), and transmittance (solid symbols) spectra (left axis) of gold island on quartz with 1 (star), 2 (triangle), 4 (circle), and 8 (square) minutes of sputtering. Short dashed line (green) shows NaYF₄:Yb,Er/SiO₂ luminescence spectra (right axis).

The efficiency of MEF will be weaker on the discontinuous island films, say the (2 + 0) or (4 + 0) cases, because a significant number of the UCNPs would be sitting on bare Si, which would absorb, instead of reflecting, most of the available 980 nm radiation. The observed enhancement for the (2 + 0) and (4 + 0) cases is similar to that reported by Zhang et al.,¹⁷ who obtained a 5 fold enhancement just by a single coating on the core of UCNPs, without a SiO₂ shell on them, with gold NPs. A possible explanation of this emission enhancement from UCNPs, without a dielectric spacer, is the concentration of excitation light on the UCNPs by the gold NPs,¹⁸ unlike what we observed in our case for the core–shell UCNPs.

Next, we discuss the optical properties of the top layer of the gold sandwich. The transmittance (T/solid symbol), and reflectance (R/hollow symbol) of the different gold top layers are shown in Fig. 6b. The spectra indicate a decrease in the 980 nm transmittance as the gold thickness increases, as expected for a metallic layer. The result presented in Fig. 5a can be explained as follows: the top gold layer produced by 1 minute of sputtering, having a nanoparticle morphology, showed a surface plasmon (SP) absorption (decreased transmission) band around ∼560 nm, which nearly resonated at the 540 nm emission of the UCNPs (shown by the dashed-line spectrum in Fig. 6b),³⁸ and will transmit the 980 nm excitation to irradiate the UCNPs inside. However, as the sputtering time increased, the plasmon band of the top gold layer, say for the (8 + 2), (8 + 4), and (8 + 8) samples, shifted to longer wavelengths, decreasing the transmittance, and therefore the 980 nm photon flux on the UCNPs, and thereby decreasing the net luminescence (Fig. S5, ESI†) compared to the (8 + 1) sample. Photon flux is critical for UCNPs that depend on multiphoton processes for luminescence.

However, this does not explain why the luminescence in the (8 + 1) sandwich configuration is higher than (8 + 0). This is interesting, because, intuitively, the 520–540 nm emission from the UCNPs should be absorbed by the SP of the top gold layer, which is in the visible spectrum for the 1 minute coating case, before it reaches the detector, which should measure a lower photon count. In fact, there is a previous report of such a plasmon-induced quenching of luminescence.³⁹ However, experimentally we have observed an enhancement of nearly 2.5 times in the (8 + 1) configuration with respect to the (8 + 0) configuration. Firstly, it is possible that the absorption in the top gold layer produces SP that alters the local electric field near the UCNPs. The increase in the local electric field can then alter (i) the excitation rate, and/or (ii) the radiative decay rates in the UCNPs, resulting in the enhanced luminescence. It has been reported previously that the presence of plasmonic fields near the UCNPs, causing fluorescence enhancement, would also result in a decrease in the fluorescence lifetime.¹³ Whether the plasmons at the top influence any one or both of the above factors cannot be independently clarified presently. Next, the optical cross-section of the UCNP, normally a fraction of its physical size, which controls the absorption, can be greatly enhanced in the presence of the gold layers.⁴⁰ This may result in a higher incident field at the excitation wavelength on the UCNP. Rare-earth-doped NPs in the presence of gold clusters have demonstrated 250× enhanced fluorescence attributed to a concentration of the incident light.¹⁸

It may be tempting to treat this ‘gold sandwich’ design as a metallic shell-encased UCNP. But fundamentally, it is not. This is because the metallic shell could be tuned, by thickness, to either resonate with the excitation or the emission wavelength to have a ‘radiating plasmon’.⁴⁰

A uniform gold shell-encased UCNP may be technically similar to the case of a single UCNP placed between two gold NPs that has demonstrated an enhancement factor of 4.8, and has clearly shown the gold plasmonic influence on the excitation and emission processes in the UCNPs.³⁴

In contrast, our result indicates the possible requirement of non-uniform shell thickness for efficient up-conversion luminescence. We have shown by detailed tuning of the gold layer thickness that the top-layer SP should be resonant with the emission (transparent to the excitation), and that the bottom layer should act as a reflector at the excitation wavelength to concentrate the energy for efficient MEF.

As mentioned in the Introduction, the net enhancement factors are often difficult to compare directly because of variations in the design geometry (core, or core–shell) of the UCNP crystal, shell thickness, use of specific metallic nanostructures, emission bands being considered, and so on. Table 1 in the ESI† shows plasmonic enhancement of erbium-doped UCNPs for the 540 nm peak, only in a different measurement configuration. This requires careful consideration before a meaningful case-by-case comparison of the enhancement factor can be made.

Conclusions

In summary, we have proposed and successfully demonstrated the use of a ‘gold sandwich’ for metal-enhanced fluorescence from NaYF₄:Yb,Er/SiO₂ core–shell nanoparticles (in the solid state) with core sizes 8 nm in diameter prepared by thermal decomposition, and ∼10 nm thick SiO₂ shell prepared by the microemulsion technique. The nanocrystals were characterized by TEM, EDS, and SAED. We have optimized a thick gold bottom layer, with higher reflectance in the infrared region, to work best for the fluorescence enhancement by increasing the optical path length, and therefore efficient multiphoton absorption of the excitation in the UCNP layer. On the other hand, a thin island-like top coating, with surface plasmon absorption in the visible, completes the ‘gold sandwich’. The non-agglomerated gold top-coating is transparent to the 980 nm excitation, but absorbs the 520–655 nm emission from the nanoparticles, and in turn enhances the local electric field, assisting in efficient two- or higher multiphoton excitation, and/or radiative decay rates within the ladder-like energy states of the UCNPs. A maximum enhancement factor of 19 was observed when using the ‘gold sandwich’, compared to the bare nanocrystals without the use of any gold surface.

Acknowledgements

This study was supported by the Ministry of Science and Technology, Taiwan, under Grant no. MOST-101-2112-M-010-003-MY3 and 104-2112-M-010-002-MY3. The authors acknowledge support from the ATU plan of the Ministry of Education, Taiwan. R. V. M acknowledges the support from National Yang-Ming University, Taiwan, for his doctoral fellowship. Additional support from the Biophotonics and Molecular Imaging Research Center (BMIRC), National Yang Ming University is also gratefully acknowledged.

Notes and references

1. J. Zhang, R. E. Campbell, A. Y. Ting and R. Y. Tsien, Nat. Rev. Mol. Cell Biol., 2002, 3, 906–918.
2. E. Betzig, G. H. Patterson, R. Sougrat, O. W. Lindwasser, S. Olenych, J. S. Bonifacino, M. W. Davidson, J. Lippincott-Schwartz and H. F. Hess, Science, 2006, 313, 1642–1645.
3. W. Jiang, B. Y. Kim, J. T. Rutka and C. W. Chan, Nat. Nanotechnol., 2008, 3, 145–150.
4. D. R. Larson, W. R. Zipfe, R. M. Williams, S. W. Clark, M. P. Bruchez, F. W. Wise and W. W. Webb, Science, 2003, 300, 1434–1436.
5. R. Kumar, M. Nyk, T. Y. Ohulchanskyy, C. A. Flask and P. N. Prasad, Adv. Funct. Mater., 2009, 19, 853–859.
6. M. Haase and H. Schafer, Angew. Chem., Int. Ed., 2011, 50, 5808–5829.
7. F. Auzel, Chem. Rev., 2004, 104, 139–173.
8. R. Kolesov, K. Xia, R. Reuter, R. Stöhr, A. Zappe, J. Meijer, P. R. Hemmer and J. Wrachtrup, Nat. Commun., 2012, 3, 1029.
9. M. Pollnau, D. R. Gamelin, S. R. Luthi, H. U. Gudel and M. P. Hehlen, Phys. Rev. B: Condens. Matter, 2000, 61, 3337–3346.
10. Y. Liu, D. Tu, H. Zhu, E. Ma and X. Chen, Nanoscale, 2013, 5, 1369–1384.
11. M. A. Noginov, G. Zhu, A. M. Belgrave, R. Bakker, V. M. Shalaev, E. E. Narimanov, S. Stout, E. Herz, T. Suteewong and U. Wiesner, Nature, 2009, 460, 1110–1112.
12. J. M. Luther, P. K. Jain, T. Ewers and A. P. Alivisatos, Nat. Mater., 2011, 10, 361–366.
13. M. Saboktakin, X. C. Ye, S. J. Oh, S. H. Hong, A. T. Fafarman, U. K. Chettiar, N. Engheta, C. B. Murray and C. R. Kagan, ACS Nano, 2012, 6, 8758–8766.
14. A. L. Feng, M. Li You, L. Tian, S. Singamaneni, M. Liu, Z. Duan, T. J. Lu, F. Xu and M. Lin, Sci. Rep., 2015, 5, 7779.
15. D. M. Wu, A. Garcia-Etxarri, A. Salleo and J. A. Dionne, J. Phys. Chem. Lett., 2014, 5, 4020–4031.
16. W. Zhang, F. Ding and S. Y. Chou, Adv. Mater., 2012, 24, OP236.
17. H. Zhang, D. Xu, Y. Huang and X. F. Duan, Chem. Commun., 2011, 47, 979–981.
18. M. Eichailbum and K. Rademann, Adv. Funct. Mater., 2009, 19, 2045–2052.
19. E. Verhagen, L. Kuipers and A. Polman, Opt. Express, 2009, 17, 14586–14598.
20. M. Saboktakin, X. Ye, U. K. Chettiar, N. Engheta, C. B. Murray and C. R. Kagan, ACS Nano, 2013, 7, 7186–7192.
21. Z. Li, Y. Zhang and S. Jiang, Adv. Mater., 2008, 20, 4765–4769.
22. C. Graf, D. L. J. Vossen, A. Imhof and A. van Blaaderen, Langmuir, 2003, 19, 6693–6700.
23. F. Zhang, G. B. Braun, Y. F. Shi, Y. C. Zhang, X. H. Sun, N. O. Reich, D. Y. Zhao and G. J. Stucky, J. Am. Chem. Soc., 2010, 132, 2850–2851.
24. T. R. Jensen, G. C. Schatz and R. P. Van Duyne, J. Phys. Chem. B, 1999, 103, 2394–2401.
25. M. D. Malinsky, K. L. Kelly, G. C. Schatz and R. P. Van Duyne, J. Am. Chem. Soc., 2001, 123, 1471–1482.
26. G. Kalyuzhny, A. Vaskevich, G. Ashkenasy, A. Shanzer and I. Rubinstein, J. Phys. Chem. B, 2000, 104, 8238–8244.
27. G. Kalyuzhny, A. Vaskevich, M. A. Schneeweiss and I. Rubinstein, Chem.–Eur. J., 2002, 8, 3850–3857.
28. J. Shumaker-Parry, H. S. Rochholz and M. Kreiter, Adv. Mater., 2005, 17, 2131–2134.
29. V. Myroshnychenko, J. Rodriguez-Fernandez, I. Pastoriza-Santos, A. M. Funston, C. Novo, P. Mulvaney, L. M. Liz-Marzan and F. J. Garcia de Abajo, Chem. Soc. Rev., 2008, 37, 1792–1805.
30. J. R. Lakowicz, Anal. Biochem., 2001, 298, 1–24.
31. P. Yuan, Y. H. Lee, M. K. Gnanasammandhan, Z. Guan, Y. Zhang and Q. H. Xu, Nanoscale, 2012, 4, 5132–5137.
32. S. Schietinger, T. Aichele, H. Q. Wang, T. Nann and O. Benson, Nano Lett., 2010, 10, 134–138.
33. G. Y. Chen, G. Somesfalean, Y. Liu, Z. G. Zhang, Q. Sun and F. P. Wang, Phys. Rev. B: Condens. Matter Mater. Phys., 2007, 75, 195204.
34. J. F. Suyver, A. Aebischer, S. Garcia-Revilla, P. Gerner and H. U. Gudel, Phys. Rev. B: Condens. Matter Mater. Phys., 2005, 71, 125123.
35. G. Chen, H. Qiu, P. N. Prasad and X. Chen, Chem. Rev., 2014, 114, 5161–5214.
36. W. Xu, Z. G. Zhang and W. W. Cao, Opt. Lett., 2012, 37, 4865–4867.
37. H. Dong, L. D. Sun and C. H. Yan, Nanoscale, 2013, 5, 5703–5714.
38. H. Sun, M. Yu, G. Wang, X. Sun and J. Lian, J. Phys. Chem., 2012, 116, 9000–9008.
39. E. Dulkeith, A. C. Morteani, T. Niedereichholz, T. A. Klar, J. Feldmann, S. A. Levi, F. C. J. M. van Veggel, D. N. Reinhoudt, M. Möller and D. I. Gittins, Phys. Rev. Lett., 2002, 89, 203002.
40. J. R. Lakowicz, M. H. Chowdhury, K. Ray, J. Zhang, Y. Fu, R. Badugu, C. R. Sabanayagam, K. Nowaczyk, H. Szmacinski, K. Aslan and C. D. Geddes, Proc. SPIE-Int. Soc. Opt. Eng., 2006, 6099, 609909.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra20273j

---