1. Introduction

There has been an intense scientific and technological interest in the development of new materials as near-infrared (NIR) super-broadband and intense luminescence sources for the advancement of optical amplifiers technology, tunable lasers and amplified signals transmission [1–3]. Rare-earth (RE) ion -doped glasses are an alternative solution since the demand for optical devices that operate in the NIR region has been increasing.

Bulk glasses with high solubilities of RE ions represent a large optical confinement (medium gain [2,4,5]), which offers an attractive way of producing transparent, small and high-emission quantum yield devices. In contrast, the glasses fabrication with low contaminants content is still limited by the traditional technique of melting the powders precursor in crucible, despite the recent progresses observed in this area [6]. Specific properties of the host matrix, such as high refractive index, large optical transmission (visible and NIR regions) and low phonon energies are required for the achievement of high emission yield. In this sense, tellurite glasses (whose major component is TeO₂) have appeared as optimal candidates among the oxide glasses because of their wide transparency range (from 400 to 5500 nm), high refractive index (~1.8-2.0), relatively low phonon energy (~800 cm⁻¹) in comparison to silicate glasses [7–9] and good chemical and mechanical resistance [10,11]. Nevertheless, there exists a technological limitation to the use of RE ions for optical emission and amplification, since each RE ion covers the O-, E-, S-, C-, L- and U- bands of the optical communications separately (in the 1000 - 1680 nm region) [1,12]. The emission bands of the trivalent RE ions are difficult to surpass since the f-f transitions of the 4f orbitals are confined in the inner-shell and their nature is usually mediated by the electric dipole transitions which are sensitive to local environmental [13,14]. Several approaches are currently under development for a full use of the optical bandwidth through glassy systems, such as (i) doping with metal ions, e.g., bismuth [15–18], nickel [19–21], bismuth-nickel codoped [22], and, chromium [23], (ii) generation of supercontinuum light in highly non-linear optical fibers [24–26] and (iii) co-doping with RE ions [27–29]. Therefore, the expansion in the gain bandwidth in glassy materials for the manufacture of more efficient optical devices in the transmission network is a key factor for the present and future development of optical communications.

Noble metal nanoparticles (NPs) embedded in RE-doped tellurite glasses can either improve or quench the RE spectroscopic properties for optical applications in telecommunication bands [30–34] and light emission by upconversion processes [35–37]. The full understanding of the interactions involved in NPs and RE ions is still under debate. Some researchers have ascribed such an emission enhancement to the energy transfer (ET) between NP:RE, while others have attributed it to an improvement in the local field near the NPs (which exponentially decays from the NP surface). In fact, both interaction effects are produced from a coherent collective oscillation of the surface electrons of the metal NP (nonpropagating excitations), also known as localized surface plamon resonance (LSPR) [38,39], which enables the confinement of the electromagnetic energy ($E_p=\hslash {\left(n{e}^2/\left(m{\epsilon }_0\right)\right)}^{0.5}$ [39]) into volumes smaller than the diffraction limit (λ₀/n)³, where ${\epsilon }_0$is the permittivity of the free space, n is the electron density, e and m are the charge and mass of the electron, respectively, n is the refractive index of the surrounding medium and λ₀ is the incident excitation wavelength. Moreover the polarization of the NP can be either longitudinal (electric dipole parallel to the NP:RE axis) or transverse (electric dipole perpendicular to the NP:RE axis), depending on the NP shape, e.g. non-spherical NP, nanorods and others [40–43]. Such electric dipole coupling is formed when both NP:RE have the same potential, but opposite polarity [38]. The controlled production of anisotropic NPs embedded in glassy materials [44], like tellurite glasses [30–37,43] and the comprehension of the basic concepts involved constitute a current challenge that may bring us new technological applications.

The present research is focused on the emission enhancement and widening of broadbands in the NIR region in Er³⁺-Tm³⁺-doped tellurite glass by means of gold or silver NPs embedded in the glass for gain amplification. Such an emission bandwidth is obtained under excitation at 976 nm laser diode through an integrated emission between the Er³⁺ and Tm³⁺ ion levels via efficient ET processes: (Er³⁺)⁴I_11/2→(Tm³⁺)³H₅, (Er³⁺)⁴I_13/2→(Tm³⁺)³H₄ and (Er³⁺)⁴I_13/2→(Tm³⁺)³F₄, which cover all the S-, C + L- and U-bands of the optical telecommunications. These ETs are mediated by either silver or gold NPs (SNPs or GNPs). The NIR emission in the 1450-2010 nm region is ascribed to the following electronic transitions of Er³⁺ ions: ⁴I_13/2→⁴I_15/2 (1530 nm), ²H_11/2→⁴I_9/2 (1680 nm) and ⁴S_3/2→⁴I_9/2 (1720 nm), and the electronic transitions of Tm³⁺ ions: ³H₄→³F₄ (1490 nm) and ³F₄→³H₆ (1810 nm). Meanwhile, the NPs produce the following effects: (i) increment in the ET quantum yield, resulting in an emission improvement, and (ii) variations in the oscillator strength (local crystal field), resulting in a blue/red band shift. The latter may change the line shape and broaden the NIR emission band. Furthermore, the most outstanding characteristic of glassy materials and more particularly tellurite glasses has arisen from their potential applications as optical fiber amplifiers or fiber laser covering the entire telecommunication window.

2. Experimental section

Glasses of 74TeO₂-5ZnO-15Na₂O-5GeO₂-1Er₂O₃:xTm₂O₃ (mol%) nominal composition with x = 0.05 or 0.06 mol% and containing 0.25 mol% of Ag or Au added in the form of chloride (AgCl or AuCl₃) were prepared by the traditional melt-casting technique. More details about the preparation are provided in a previous paper [29]. For clarity, the studied glass samples are labeled as TxMy, where M represents the noble metal (S or G for silver and gold, respectively), x = 5 or 6 for the concentration of Tm₂O₃ (0.05 or 0.06 mol%, respectively), and y is the annealing time at 300 °C (y = 2, 4, 6, 8, 10 and 12 hours) – for the samples without NPs, i.e. 5 hours. All the glass samples were prepared from high purity starting materials (3N and above). Silver or gold NPs were generated in the host matrix by thermal mobility according to the annealing time after the decomposition of the metal chloride during the glass melting [31–33,43,45]. The final samples were cut into 10 × 30 × 1 mm³ square pieces and polished until obtaining adequate transparency for the optical characterizations.

Dark-field microscopic images were obtained by an optical microscope (Olympus BX53 with a Darkfield dry condenser 10 × NA 0.8) coupled to a digital monochrome camera (Olympus XM10) and using a 75 W Xe lamp as a light source. A transmission dark-field objective (Olympus UPlanSApo 60x/1.2W NA 0.9 - NPLan) was used to focus the image of NPs embedded in the glasses.

The thermal properties of the glass samples were examined by differential scanning calorimetry (Netzsch DSC 404F3 Pegasus) in sealed Al pans at 10 °C/min heating rate. The density of the samples was measured by an MD-300S (Alfa Mirage) electronic densimeter and the Archimedes method using distilled water as the immersion liquid. Multiple density measurements were performed for each sample composition so as to obtain accurate values ( ± 5 mg/cm3 precision).

The transmission spectra were recorded from 350 to 2000 nm on a Varian Cary 500Scan UV-VI-NIR double beam spectrophotometer of ± 0.3 nm resolution. Steady-state luminescence spectra were obtained from a Nanolog spectrofluorimeter from Horiba Jobin Yvon equipped with a liquid-nitrogen-cooled Symphony InGaAs near-infrared detector. A pig-tailed diode laser at 976 nm (power 40 mW) was used as an external excitation source. The spectral slit width was 10 nm for the emission signal and the data acquired were corrected by instrumental factors. The samples were excited by a Pico Quant pulsed diode laser at 972 nm (model LDH-P-C) with pulse train of 60 ps and dead time of 20 ms for the measurements of the ⁴I_13/2 excited state lifetime. The time resolved emission signal centered at 1535 ± 4 nm was measured by a Hamamatsu NIR-PMT module detector coupled to the Nanolog system by employing the method of Time Correlated Single-Photon Counting (TCSPC). Decay analysis software (DAS6) was used for obtaining the lifetime values by fitting a single exponential function. Measurements of both steady-state emission and lifetime were taken at room temperature. The laser beam was always collimated and focused at the center of the glass surface with a lens of 50 mm focal length.

3. Results and discussion

The glass transition temperatures Tg, determined for T5 and T6 samples are Tg = 311 ± 2°C and 309 ± 2°C, respectively. The annealing temperature used to generate and grow the NPs in both glasses was below their glass transition temperature Tg, at 300 °C. Such an experimental procedure was chosen so as to prevent a possible interaction between the glass network and the NPs mobile precursors [46], and a slow diffusion of the metallic Ag⁰ or Au⁰ particles [33]. Besides, the glass transition temperatures of the samples containing NPs were also determined: Tg = 303 ± 2°C and Tg = 301 ± 2°C for T5Sy and T6Sy samples, respectively, and Tg = 305 ± 2°C and Tg = 304 ± 2°C for T5Gy and T6Gy samples, respectively. The measured density of the samples, the calculated number density N for the Er³⁺, Tm³⁺, Ag⁰ and Au⁰ species, and the average distance r_a-b between them [33,43] are summarized in Table 1. The values of glass densities, N_Er³⁺, N_Tm³⁺, N_Ag⁰, N_Au⁰, r_Ag⁰-_Ag⁰, r_Au⁰_-Au⁰, r_Er³⁺-_Ag⁰ and r_Er³⁺_-Au⁰ are almost invariant for all the samples studied. The average distance r_Er³⁺_-Tm³⁺ in both glasses without NP is the same. Distance r_Er³⁺_-Tm³⁺ increases similarly upon the addition of metallic NPs (for both T5My and T6My samples). Likewise, from the values reported in Table 1, we can deduce the following trends:

Table 1. Measured and calculated physical properties of TxMy glass samples: glass density, number (ionic or atomic) density N and average distance between rare-earth-ions and metallic-atoms (M⁰).

View Table

Almost the same r_Er³⁺_-M⁰ distance for the T5Gy and T6Gy, and T5Sy and T6Sy samples and a clear increase in the r_Tm³⁺_-Ag⁰ distance from the T5Sy and T6Sy samples are observed. The Ag⁰ and Au⁰ atoms are closer to the Er³⁺ ions than to the Tm³⁺ ions, and the Tm³⁺ ions are closer to the Au⁰ atoms than to the Ag⁰ atoms. Besides, an increased r_Tm³⁺_-M⁰ distance with the increases of the N_Tm³⁺ was observed. We have assumed that each Ag⁰ or Au⁰ atom has become a nucleation center and formed one NP during the annealing process (the growth is governed by the glass viscosity and the Ag⁰ or Au⁰ atoms mobility [45]) of specific size and shape. To maintain a thermodynamic and electronic charge equilibrium we can assume that the separation distance between RE and NP remains similar to the distance between RE ions and Ag⁰/Au⁰ atoms.

In summary, we have observed that: (i) the introduction of the metallic NPs increases the separation distance between the RE ions; (ii) the separation distances between Er³⁺ ions and SNP are lower than those of Er³⁺ ions and GNP, independently of the Tm³⁺ ions density N_Tm³⁺, suggesting a larger interaction (coupling) of SNP:Er³⁺ dipoles than GNP:Er³⁺, and in contrast; (iii) the separation distances between Tm³⁺ ions and SNP are larger than those of Tm³⁺ ions and GNP, suggesting a larger interaction of GNP:Tm³⁺ dipoles in comparison to SNP:Tm³⁺. Dark-field illumination is a microscopy technique largely used by biologists to observe unstained samples by making them appear brightly lit against a dark background [47]. The light scattered (transmitted) from the NP is directed to a camera image plane through an aperture of 0.1 mm diameter so as to reduce scattered light from other regions of the glass. The dark-field microscopic images obtained from T5S12 and T5G12 samples are shown in Fig. 1. From this, we can be observed that the size/shape and distribution those NPs are non-homogeneous inside the glasses, in agree with the literature [30–37]. Although the GNPs or SNPs embedded in the glass do not show any fluorescence, the surface plasmon scattering they produce can be detected as a white light. The absence of fluorescence can be explained by the absorption of both Er³⁺ and Tm³⁺ ions present in the glass, which results in the colors emission turn-off of those NPs.

 

Fig. 1 Optical dark-field microscopic images of the NPs embedded in T5S12 (a) and T5G12 (b) samples.

Download Full Size | PDF

The RE ions absorption bands can indeed overlap the NPs resonance wavelength, as verified in the glasses absorption spectra shown in Fig. 2. Nonetheless, from these dark-field images the different shapes of the NPs (spherical, elliptic and more complex) can be distinguished. Such an unorganized growth was predicted in [45], which demonstrates the optical properties of the NPs embedded in the glass can be explored.

 

Fig. 2 Absorption spectra of (a) T5Sy, (b) T6Sy, (c) T5Gy, and (d) T6Gy glass samples. The inset shows a zoom of the absorption bands of transitions (a) ³H₆→³F₄, (b) ³H₆→³H₅, (c) ³H₆→³H₄ and (d) ³H₆→³F_2-3 in Tm³⁺ ions. All spectra show a high absorption of the Er³⁺ ions without NPs in comparison with the other samples, since the NPs increase the absorption these glasses.

Download Full Size | PDF

The room temperature absorption spectra of the TxMy samples are shown in Fig. 2. Eight intense transitions from the ground state to excited states ⁴I_15/2→(⁴I_13/2, ⁴I_11/2, ⁴I_9/2, ⁴F_9/2, ⁴S_3/2, ²H_11/2, ⁴F_7/2 and ⁴F_3/2) are attributed to the presence of Er³⁺ ions and correspond to the bands centered at 1538, 978, 806, 649, 544, 520, 488 and 444 nm, respectively. Five subtle transitions from the ground state to excited states ³H₆→(³F₄, ³H₅, ³H₄, ³F₃ and ³F₂) are also observed for the Tm³⁺ ions and correspond to the bands centered at 1750, 1190, 785, 660 and 651 nm, respectively. Due to the small concentration of silver and gold (small atomic density N_Ag⁰ or N_Au⁰) and the overlap of the LSPR band with the Er³⁺ or Tm³⁺ ions absorption bands, no plasmon band is observed in the absorption spectra recorded in the TxMy glasses. Nevertheless, the activity of the plasmon band associated with the presence of GNPs or SNPs in tellurite glasses was demonstrated in our previous investigations into the growth and geometry dependence on the optical properties of LSPR modes [31,32,45].

The luminescence spectra of the glasses in the 1400-2120 nm range under 976 nm laser diode excitation (with a diode power of 40 mW) are shown in Fig. 3. Note that the measurement reproducibility was carefully verified for all samples studied.

 

Fig. 3 Near infrared luminescence spectra under excitation at 976 nm of (a) T5Sy, (b) T6Sy, (c) T5Gy, and (d) T6Gy glass samples showing the ⁴I_13/2→⁴I_15/2 and ³F₄→³H₆ radiative transitions of the Er³⁺ and Tm³⁺ ions, respectively, and their splitting due to the Stark effect.

Download Full Size | PDF

The oscillator strength (f_fi) of the trivalent RE ion is a fundamental physical quantity in analytical spectroscopy which determines the sensitivity of the 4f-4f transitions due to the chemical nature and symmetry of the ligand environment. This phenomenon is referred to as hypersensitivity transition and has been correlated with the coordination numbers and symmetries of the RE ions in the host matrix. On the other hand, LSPR creates electromagnetic resonance fields that can improve the local field [38]. Such fields are formed through oscillating dipoles (from the NPs) around the RE ion, hence an additional oscillating electric field produced near the RE ion, which may induce strong changes in 4f–4f transitions [31] (e.g. a blue/red band shift or an enhanced/quenched luminescence). The induced oscillating dipoles depend on the size, shape and isotropic dipolar polarizabilities of the NPs environment. Therefore, evident variations occur in the luminescence spectra due to the NPs embedded in the glass, Fig. 3. In the first approximation, we can use the dynamic coupling mechanism [48–50] and the interaction energy between the NP:RE is given by: ${\text{H}}_{\text{DC}}=e{\displaystyle \sum _{i,j}\overrightarrow{{\mu }_j}{\text{n}}_j\frac{\overrightarrow{r_i}-\overrightarrow{R_j}}{{\left|\overrightarrow{r_i}-\overrightarrow{R_j}\right|}^3}}$, where$\overrightarrow{{\mu }_j}\left(={\alpha }_j{\overrightarrow{E}}_{inc}\right)$ is the electric dipole moment induced in the NP by an incident electric field ${\overrightarrow{E}}_{inc}$, $\left|\overrightarrow{r_i}-\overrightarrow{R_j}\right|$is the separation distance between NP:RE, n_j is the density of the conduction electrons inside a plasma characterized by carriers with charge N’e.

According to our experimental conditions, we must consider that: (i) as the growth of the NPs is governed by the thermal treatment, NPs can have different geometrical ratios (e.g. ellipsoidal NPs), therefore the polarizability is [51] ${\alpha }_j=\frac{4\pi }{3}{a}_1{a}_2{a}_3\frac{{\epsilon }_{NP}-\epsilon \left(\omega \right)}{\epsilon \left(\omega \right)-L_i\left({\epsilon }_{NP}-\epsilon \left(\omega \right)\right)}$, where principal axes a₁, a₂ and a₃ introduce geometrical depolarization factors L_i (for spherical particles, L₁ = L₂ = L₃ = 1/3, or spheroidal particles L₁ = L₂), $\epsilon \left(\omega \right)$ and ${\epsilon }_{NP}$ are, respectively, the dielectric and NP permittivity as a function of the excitation frequency; (ii) we can assume a couple of particles (NP and RE ions) forming an electric dipole of −V/2 and + V/2 potentials, respectively [33]; (iii) the ET between Er³⁺:Tm³⁺ is mediated by the NPs [38,43]; (iv) the ET from the RE ion to NP is more probable than the ET from NP to the RE ion, since the excited state lifetime of NPs is extremely shorter (ns order) in comparison with the higher energy excited states of Er³⁺ and Tm³⁺ ions (μs or ms order) [29,52]; (v) the luminescence quenching can be described by the ET between NP:RE and re-absorption by the NPs which are in resonance with the emissions of RE ions [33,38,43]. Such considerations are summarized in Fig. 4. About the point (ii), in the case with multiple interactions at different distances between NPs:REs, the charge from the RE ion is the same, because the ${\overrightarrow{E}}_{inc}$employed to excite these particles will be absorbed for the RE ion due to the cross-section absorption those is more than the NPs [Fig. 2], case contrary the light incident (λ₀≠λ_spp – wavelength surface plasmon resonance) is scattered for the NPs, and this scattered light is absorbed for the RE ion which emits and this coupled/resonates with the NP.

 

Fig. 4 Schematic representation of the Er³⁺-Tm³⁺:NP interactions in two different scenarios: (a) the RE ions do not interact with each other, but with the NP and; (b) the interaction occurs between NP:RE and RE₁:RE₂. Equipotential surface (electric multipole coupling) with electric potentials −V/2 and + V/2 between NP:REs,$\text{d}{}_{\text{1,2}}=\left|\overrightarrow{r_i}-\overrightarrow{R_j}\right|$.

Download Full Size | PDF

Several interesting aspects about the oscillator strength obtained by dynamic coupling mechanisms between NP:RE can be discussed. We can write the transition probability from any excited state transition decaying to background (2→1) as $A_{21}=\frac{8{\pi }^2{e}^2}{{\lambda }^{2}mc}{f}_{21}$, where c is the light velocity. The intensity emitted from a glass can be expressed by $I_{em}=h\upsilon N_2{A}_{21}$, where N₂ is the population in the excited level, h is the Planck constant and $\upsilon$ is the line frequency, and we have the following relation $I_{em}\propto f_{21}$. In this frame, the NPs contribute only to f₂₁ when ${\overrightarrow{E}}_{inc}$is present and the states of the REs have opposite parity [31,38]. Figure 3 shows the magnitude of the H_DC effects on the crystalline potential as changes in the spectral lines, their widths, and shapes. In glasses, since different sites can be occupied by the RE ion, the Stark splitting is not well determined. However, here the manufactured process was strictly the same for all the samples, therefore the Stark splitting must be invariant in our samples. In this manner, the presence of NPs in the host matrix can modify the bandwidth of the Stark energy levels, which results in a blue/red shift and/or a broadening/increase in the RE ions emission as well as the emission intensity.

A subtle increase in N_Tm³⁺ induces a decrease in the emission intensity of the Er³⁺ and Tm³⁺ ions due to the ET from Er³⁺ to Tm³⁺ ions (donor-acceptor) and a cross-relaxation process in which the quenching reduces the luminescence quantum efficiency of the Er³⁺-Tm³⁺ ions [Fig. 4]. A similar behavior was observed by V.A.G. Rivera [29]. Besides, such ET process between the Er³⁺ to Tm³⁺ ions still remains in the TxMy samples since the separation distance between the RE ions does not change with decreasing Tm³⁺ ions concentration, Table 1.

On the one hand, in comparison to the T5 glass sample, the T5S08 sample shows 282% and 238% improvements in the emission intensity near 1535 and 1795 nm for the Er³⁺:(⁴I_13/2 →⁴I_15/2) and Tm³⁺:(³F₄→³H₆) transitions, respectively. On the other hand, in comparison with the T6 sample, the T6S10 sample shows 233% and 248% improvements in the emission intensity for the same Er³⁺ and Tm³⁺ ions transitions. Such values of NIR emission enhancement induced by NPs have been reported for the first time in tellurite glasses. In order to analyze the oscillator strength f_fi, the values of the intensity peaks (I_λ-peak) were taken from the emission spectra [Fig. 3], with and without NPs at 1535 ± 6 and 1795 ± 8 nm wavelengths [Fig. 5(a)]. The ± 6 or 8 nm deviations are due to the blue/red shift produced by the longitudinal/transverse polarization of the NPs [38,43]. Figure 5(a) provides an overview of the amplitude of the luminescence enhancement (↑) or quenching (↓) observed in the glass samples. The emission peak intensity strongly depends on the annealing time, and NPs nature and shape, therefore: (i) the peak intensity ratio I_λ-1535/I_λ-1795 varies from the glass without NPs (Tx) to the samples containing NPs (TxMy), which confirms the H_DC effects; (ii) the T5S08 sample exhibits a more intense emission than the other TxMy samples, and the T5G10 sample shows a more intense emission than the other TxGy samples; (iii) I_λ-1535 > I_λ-1795 for the TxSy samples (except the T6S06 and T6S10 samples), indicating that the SNP:Er³⁺ coupling is more intense than SNP:Tm³⁺, which is in agreement with the average separation distance for these ions (r_Er³⁺-_Ag⁰ < r_Er³⁺-_Au⁰); (iv) in contrast, for TxGy samples, I_λ-1535 < I_λ-1795 (except for the T6G10 sample), indicating that the coupling of GNP is more intense with the Tm³⁺ ions than with the Er³⁺ ions, which is also in agreement with the average separation distance for these ions (r_Tm³⁺-_Au⁰ < r_Tm³⁺-_Ag⁰); (v) although most of the TxSy samples show higher emission intensities, some of them also show a serious quenching in comparison to the TxGy samples, see doted/dashed lines in Fig. 5(a). We can conclude that the multidipole SNP:(Er³⁺-Tm³⁺) displays a strong coupling due to a short separation distance, which can improve luminescence, or quench, due to the ET between SNP:RE or re-absorption by the SNPs, which are in resonance with the emissions of RE ions [43].

 

Fig. 5 (a) Intensity of the emission peaks at 1535 ± 6 nm (⁴I_13/2 →⁴I_15/2 transition of Er³⁺ ion) and 1795 ± 8 nm (³F₄→³H₆ transition of Tm³⁺ ion) for the TxMy samples studied. (b) Bandwidth of the emission band centered at 1535 ± 6 nm for the TxMy samples studied. (c) Emission spectra of the T5S08 and T6S10 samples which exhibit the most intense emission, and T5G08, T5G10, T6G02 and T6G12 samples which exhibit the broadest band.

Download Full Size | PDF

In a previous study on the expansion of broadband emissions in NIR [29], we achieved a covering of the optical telecommunications S-, C + L- and U-bands by optical pump at 976 nm (for the Tm0.075 sample in [29]) through an integrated emission by means of appropriate Er³⁺-Tm³⁺ co-doping ions and an ET among levels (Er³⁺)⁴I_11/2→(Tm³⁺)³H₅, (Er³⁺)⁴I_13/2→(Tm³⁺)³H₄ and (Er³⁺)⁴I_13/2→(Tm³⁺)³F₄. Whereas the sample labeled Tm0.050 in [29] is similar to the T5 sample of this study, the emission bandwidth observed here is larger than that of the Tm0.050 sample (112 nm vs. 79 nm, respectively). This difference may be attributed to the host glass composition (with different sodium and zinc oxide concentrations) and the annealing time. Therefore, the T6 sample shows a 118 nm bandwidth. The bandwidths of T5 and T6 samples [Fig. 5(c)], are less important than that of the Tm0.075 sample (134 nm bandwidth [29]).

The TxGy samples exhibit larger bandwidths in comparison to the other samples, of which the most promising are T5G08, T6G02, T5G10 and T6G12 with expanded broadbands of 411, 455, 464 and 467 nm, respectively, Fig. 5(b). To the best of our knowledge, this is the first time such large values of FWHM have been reported for an NIR emission band in tellurite glasses. The FWHM values were determined at 1535 ± 6 nm, covering the second part of the S-band (1460 ± 10 to 1530 nm), the entire C + L-band (1530-1625 nm) and the U-band (1625-1675 nm), and extended beyond 285 nm (up 1995 ± 10 nm). This broadband emission results from the combined emission of the Er³⁺ and Tm³⁺ ions through efficient ET processes mediated by the metallic NPs via H_DC [Fig. 4], according to the following transitions: (Er³⁺)⁴I_11/2→(Tm³⁺)³H₅, (Er³⁺)⁴I_13/2→(Tm³⁺)³H₄ and (Er³⁺)⁴I_13/2→(Tm³⁺)³F₄, as shown in Fig. 6. The broadband emission results from the energy splitting of the Stark levels of each Er³⁺ ion (with 7 ( = (2J + 1)/2)) and Tm³⁺ ion (with 8 ( = 2J + 1)), the multipole interactions between NP:RE, and also the multipole interactions between the host matrix ions [29,38]. The NP:RE interactions were described through H_DC, i.e. the perturbation theory is applicable here. We can indeed add a “perturbation series”, i.e. the “k” terms, since we have different separation distances d_i,j, shape and size of the NPs [Fig. 4], which are responsible for shifting and splitting the luminescence spectra, as shown in Fig. 5(c), due to dynamic coupling mechanism. Therefore, the states have different crystal-field representations, and more than one type of absorption/emission dipole can be simultaneously excited or emitted.

 

Fig. 6 Schematic energy level diagrams of (a) TxSy samples and (b) TxGy samples. Ground state absorption (GSA) of the Er³⁺ under 976 nm excitation resulting in a population inversion of the high (via ESA) and lower levels through nonradiative (NR) decays. Radiative ET among Er³⁺-Tm³⁺ ions (ET_iE-iT), nonradiative energy transfer ET_NR, and ET between the NP:RE (ET_SNP or ET_GNP). τ_NP is the finite plasmon lifetime (ns order) and the red curved lines represent the displacement of a small volume charge ${\text{n}}_{\text{j}}=q_{p,j}\left(r\right)e^{i{\omega }_{p}t}$ to the quantum system.

Download Full Size | PDF

In this scenario, we can use a modal approach similar to that developed in recent quantum theories that describes the propagation of surface plasmon created from the LSPR [53,54]. We consider displacements varying in time as harmonic (damped) oscillators, i.e. ${\text{n}}_{\text{j}}={\text{n}}_{\text{j}}\left(r\right)e^{i\omega t}$. The optical properties of an NP are ruled by the collective excitation of its free electron cloud and create an electric dipole in the glass with polarizability α_j. In this way, as described in [43], the ET between NP:RE could have either a positive or a negative impact for a specific electronic transition and depends on the efficiency of the coupling. Let us consider a set of orthonormal vector functions $q_p\left(r\right)$ (eigenfunctions) representing the amplitude of a harmonic displacement [54] that could be coupled with the RE electronic transitions. We can write ${\text{n}}_{\text{j}}=q_{j,p}\left(r\right)e^{i{\omega }_{p}t}$ and finally ${\text{H}}_{\text{DC}}=e{\displaystyle \sum _{i,j}{\overrightarrow{\mu }}_{j,p}{q}_{p,j}\left(r\right)e^{i{\omega }_{p}t}\frac{\overrightarrow{r_i}-\overrightarrow{R_j}}{{\left|\overrightarrow{r_i}-\overrightarrow{R_j}\right|}^3}}$.

The Hamiltonian H_DC is expressed as a function of the set $q_{j,p}$ eigenmodes associated with the conduction electron density generated by different NPs (j) through its collective free oscillations at each resonance frequency ω_p of the NP and their geometric dependence by means of ${\alpha }_{j,p}$. The quantum behavior of RE ions and plasmons (in LSPR mode) interacts through H_DC through $q_{j,p}$, resulting in a plasmon-photon coupling [Fig. 6], which can be verified as an enhancement/quenching of luminescence, a widening of the broadband emission and a change in its line shape, as observed in the luminescence spectra in Figs. 3 and 5.

Because of this process, the plasmon states on the RE ions are coupled, and the respective energies are modified, nevertheless, the occupation of a given plasmon mode on a given electronic level of the RE ion (Er³⁺ or Tm³⁺, Fig. 6) will depend of the efficiency of the ET between NP and RE. Furthermore, when the higher energy levels of Er³⁺ and Tm³⁺ ions are coupled with the NP (the collective free oscillations with each resonance frequencyω_p), they can generate a white light, as initially observed in the NP in Fig. 1.

Figure 6 shows the schematic energy level diagrams in the vicinity of either an SNP [Fig. 6(a)] or GNP [Fig. 6(b)] of the Er³⁺-Tm³⁺ ions doped tellurite glass. The excitation at 976 nm stimulates the Er³⁺ ion from the ground level ⁴I_15/2 to the ⁴I_11/2 level and then to the ⁴F_7/2 level through an excited state absorption process (ESA), in which the nonradiative decays populate the levels with lower energy of the Er³⁺ ion. Likewise, the ET process from Er³⁺ to Tm³⁺ ions can populate the other energy levels of Tm³⁺ ion and the levels of lower energy by nonradiative decays. Er³⁺:(⁴S_3/2) and Tm³⁺:(³H₄) excited states can also be populated by ET between two neighboring Er³⁺-Tm³⁺ ions in the present system. Besides, the plasmon-photon coupling has been drawn as a red curved line (ET_NP) representing the displacement of a small volume charge ${\text{n}}_{\text{j,p}}=q_{j,p}\left(r\right)e^{i{\omega }_{p}t}$ to the quantum system. Accordingly, the NIR emission in the 1450-2010 nm region can be unambiguously ascribed to the following transitions of the Er³⁺ ions: ⁴I_13/2→⁴I_15/2 (1530 nm), ²H_11/2→⁴I_9/2 (1680 nm) and ⁴S_3/2→⁴I_9/2 (1720 nm) and of the Tm³⁺ ions: ³H₄→³F₄ (1490 nm) and ³F₄→³H₆ (1810 nm), Fig. 6.

Another effect produced by the presence of NPs is the variation in the ⁴I_13/2 lifetime, shown in Fig. 7, due to radiative/nonradiative ET after the incidence of the pump energy to excite the NPs. This means there occur an increment in the light absorption due to the presence of NPs [Fig. 2] and results in a lifetime increase in the T5My and T6Sy samples in comparison with the T5 and T6 samples, respectively. A decrease in the lifetime is observed for the T6Gy samples and more particularly for T5G08, T5G10, T6G02 and T6G12, which also exhibit an expanded broadband [↔, Fig. 7] and higher emission intensity [↑, Fig. 7]. In this manner, the increase or decrease of the lifetime is due to the presence of the NPs with a little or large damping ($\gamma =1/{\tau }^{NP}$ [33]), respectively, which favors or not the ET processes, Fig. 6. The T5G10 sample has shown improvements in both luminescence intensity and bandwidth broadening, which makes the sample an excellent candidate as an optical gain medium for potential applications in amplifying waveguides for plasmonic devices and other photonic nano-devices [4,55,56].

 

Fig. 7 Lifetime of the Er³⁺:⁴I_13/2 level for all samples studied. Dashed and dotted lines indicate, respectively, the lifetime of the sample without SNP and GNP for comparison. The T5G08, T5G10, T6G02 and T6G12 samples exhibit an expanded broadband (↔). The T5S08, T6S10 and T5G10 samples exhibit an improved luminescence (↑).

Download Full Size | PDF

Conclusions

Er³⁺-Tm³⁺-doped tellurite glasses are good host matrices with broadband NIR emissions of 112 and 118 nm under excitation at 976 nm for glasses co-doped with 1 mol.% of Er³⁺ and 0.05 and 0.06 mol% of Tm³⁺, respectively. For the first time the tuning in the spectral properties of Er³⁺-Tm³⁺ ions in tellurite glasses in the near-infrared region has been demonstrated through the incorporation of either SNP or GNP. A decrease in both emission intensity and lifetime of the ⁴I_13/2 level of Er³⁺ ion was observed by increasing the Tm³⁺ concentration, due to energy transfers. These tellurite glasses have enabled the formation of SNP and GNP embedded in the host matrix of different shape and size as a function of the annealing time, as verified through dark-field microscopy. The presence of NPs induces the formation of electric multiple dipoles activated by the ET between the Er³⁺-Tm³⁺ ions and the NPs, forming LSPR. The separation distance between Er³⁺-Tm³⁺ ions increases with the presence of the NPs, the SNP is closer to the Er³⁺ ions and GNP is closer to Tm³⁺. In this framework, the dynamic coupling mechanism that describes the interaction energy between the NP:RE was used through the Hamiltonian H_DC expressed as a function of the eigenmodes set. Such eigenmodes were associated with the conduction electron density generated by different NPs through its collective free oscillations at each resonance frequency of the NP and their geometric dependence. Therefore, changes in the Stark splitting, blue/red shift, and/or an increase in the bandwidth of the Er³⁺-Tm³⁺ NIR emissions can be described through the H_DC. The T5S08 and T6S10 glass samples showed an increase in the intensity of the Er³⁺:(⁴I_13/2→⁴I_15/2) and Tm³⁺:(³F₄→³H₆) emissions in comparison with that of the other samples. Likewise, the T5G08, T5G10, T6G02 and T6G12 glass samples exhibited larger bandwidths in comparison to those of the other samples. Both emission enhancement and broadband widening are in agreement with the Er³⁺:⁴I_13/2 lifetimes obtained under diode laser at 976 nm as the excitation source. To the best of our knowledge, such values of NIR emission from RE ions controlled via photon-plasmon interaction have been reported for the first time in tellurite glasses in this study. From a technological point of view, all the NIR luminescence features reported here, especially for the T5G10 sample and its NIR emission (473 nm bandwidth and enhancement of luminescence of approximately 44%) make these glasses promising candidates for broadband optical fiber amplifiers in the optical communication window and supercontinuum laser source. From a theoretical point of view, this study also brings fresh insight, since the Er³⁺-Tm³⁺-doped tellurite glasses with embedded metallic NPs have enabled us experimental study the photon-plasmon interactions, which have resulted in an improvement in the emission quantum yield of the RE ions.