1. Introduction

The efficient use of broadband light sources (such as solar irradiation) is often limited, for example, in photochemistry, because a single absorption mechanism usually does not cover the light source's entire spectral bandwidth. E.g., photosynthesis uses only the blue and red wavelength regions of the visible spectrum. It has therefore been proposed to employ photoluminescent light converters in order to also exploit the non-resonant part of the spectrum so as to enhance the number of photons available for photosynthesis [1,2]. However, not only the number of photons but also their efficient distribution within a reactor and their delivery to the active photosystem (e.g., avoiding local saturation and strong gradients in illumination intensity) are key factors in such applications [3]. Side-emitting optical fibers were identified as a means to illuminate turbid reactor environments or biofilms [4–6]. Spatially separating the original light source from the location of light emission, such illumination designs enable not only volumetric light dilution but also spectrally structured illumination [1], modulated illumination [7], and reduced heat influx [8].

The desire to extend the narrow spectral activity range has led to research into PL light conversion systems that convert unused bandwidth into usable emission. Thereby, trivalent rare-earth ions are the best-studied activator species for inorganic phosphors. Their 4f orbitals are well-shielded by the larger 5s² and 5d⁶ subshells, and intraconfigurational transitions within the f-subshell are therefore only weakly influenced by the pertinent structural or chemical environment [9]. However, such rare-earth-activated phosphors can usually be excited only within a relatively narrow wavelength range. In addition, they may exhibit a low interaction cross-section [10], so they must be used in high concentrations or with long interaction paths in order to obtain a high conversion efficiency. To this end, noble metal nanostructures can improve the interaction cross-section and extend the usable bandwidth; they may boost the performance of photoluminescent emitters by two different mechanisms: by local field enhancement or through an energy transfer process.

Local field enhancement is caused by a noble metal nanoparticle's local surface plasmon resonance (LSPR). In this process, the metal's valence electrons respond to an external electromagnetic field with a collective oscillation that increases the electric field near the nanostructure. This effect may facilitate photoluminescent emission [11–15]; it has often been used to enhance lanthanide emission with gold (Au) and silver (Ag) nanoparticles [16–19]. However, LSPR occurs only under resonant conditions, and it only uses a limited frequency range.

The energy transfer process is produced by smaller agglomerates of noble metal species, often referred to as “clusters”. These clusters consist of only a few atoms [20] and do not exhibit LSPR but enhance photoluminescent performance even under non-resonant conditions [21–25]. They absorb high-energy photons, undergo singlet-triplet intersystem crossing, and transfer the energy to an excited multiplet state of nearby lanthanide ions [24]. This non-resonant excitation is particularly interesting because it opens up a broader frequency range for interaction with an emitter and reduces the low absorption behavior of lanthanide PL emitters.

Using such non-resonant amplification of lanthanide phosphors leads to new possibilities in solar energy harvesting: previously unused photon energy in the UV range can be converted into light directed at a photosystems’ absorption bands, such as in microalgae. However, exploitation of this effect underlies various technological constraints. Although glasses provide long-term stability for incorporated noble metal species [26], clusters are usually generated by ion exchange [23,27] or ion irradiation [24,28–31]. Compared to the simple coating and curing processes presented in this paper, both techniques are complex and potentially not applicable for the modification of the surface of side-emitting fibers.

We now aim to demonstrate enhanced PL side-emission performance of coatings deposited on the outer surface of optical fibers. For this, we implement a sol-gel process first to show that stable Ag clusters can be readily incorporated into glassy matrices. To investigate the effects on photoluminescent emission, we choose europium(III) (Eu³⁺) as an archetype activator species with established literature on emission and excitation characteristics in many host materials [32,33]. Moreover, PL enhancement of europium by silver-doping has been reported before [18,19]. Specifically, europium was chosen as the PL emitter to demonstrate the feasibility of our approach, regardless of potential applications, which might need emission characteristics beyond what is possible using Eu³⁺. For example, when targeting photosynthesis, the main emission bands of Eu³⁺ do not ideally overlap with thylakoid or chlorophyll A/B absorption in terms [34]. However, for this specific application, the technique can be readily adapted using alternative emission species such as trivalent samarium or erbium ions, given that the concept has been successfully established.

Using the sol-gel technique, we can co-dope Ag and Eu at specific concentrations, and realize coatings on various substrate materials, as schematically shown in Fig. 1. Moreover, the noble metal species can be trapped within the glassy phase by avoiding high temperatures. The obtained coatings show enhanced photoluminescence under resonance conditions but, more importantly, even stronger out-of-resonance emission when pumping with light below 400 nm.

 

Fig. 1. (a) Schematic of a photoluminescent coating on optical fibers: the Ag clusters extract energy from UV-irradiation and transfer it to nearby Eu³⁺ emission centers. This non-resonant excitation enhances the photoluminescence of Eu³⁺, which emits red light in all directions. The excitation light is guided inside the fiber. (b-c) Photographs of a section of such a coated fiber under external UV illumination with and without Ag-co-doping, respectively (in the photographs, we show coated borosilicate fiber with a Eu³⁺ molar ratio of 0.04 in the sol). (d) Schematic of the fiber coating process: the glass fiber is pulled upwards through a vessel sealed on the lower end by a rubber septum.

Download Full Size | PDF

The paper is organized as follows: Section 2 describes the preparation of the sol, the sample coating, and the measurement procedures. In Section 3, the results of the measurements are presented and discussed in three parts: First, the photoluminescent enhancement of Eu/Ag co-doped sol-gel coatings on flat reference glass samples is investigated. Second, the origin of the photoluminescent enhancement of Eu³⁺ by Ag clusters is discussed. Third, the results are transferred to side-emitting optical fibers to achieve enhanced photoluminescence-based emission with a coating. Finally, we summarize the results of the work in Section 4.

2. Materials and methods

2.1 Sol-gel synthesis

The sol-gel system was prepared from tetraethyl orthosilicate (TEOS), triethyl borate (TEB), Eu(NO₃)₃·6H₂O, and AgNO₃ precursors. Ethanol (EtOH) was used as a solvent, de-ionized water (H₂O) as the reactant, and nitric acid (HNO₃) as the catalyst for the hydrolysis reaction. The reference substrate was fused silica (microscope slides, GE124, Ted Pella, Inc., USA). The binary TEOS-TEB sol was coated on fused silica to study the effect of Ag concentration on the photoluminescence of Eu³⁺. The final molar compositions of the sols are given in Table 1.

Table 1. Molar ratios in the employed Eu/Ag co-doped sol

View Table

The sol was prepared as follows: In the first beaker, TEOS, TEB, and EtOH were mixed. In the second beaker, Eu(NO₃)₃·6H₂O and AgNO₃ were dissolved in H₂O and HNO₃ (1M). After complete dissolution, the clear solution was transferred to beaker one under vigorous stirring. The final mixture was continuously stirred for another 20 minutes and then directly used for coating. The sol-gel synthesis was conducted under ambient conditions in the air.

2.2 Coating on flat glass and optical fibers

The coating on fused silica flat glass samples was applied by dip-coating with a drawing speed of 50 mm·min^-1. Uncoated and uncladded fused silica optical fiber with a diameter of 440 µm was selected for side-emission studies. The sol was filled in a rubber-sealed vial. The fibers were subsequently pulled upwards through the sol-filled vial with a pulling speed of 500 mm·min^-1. A fiber segment of ∼20 cm in length was left uncoated for light-guiding purposes.

2.3 Thermal treatment

After coating, the samples were promptly transferred to an oven and left to dry at 100 °C for 20 minutes. Then, they were placed for 30 minutes into a pre-heated muffle furnace at 400 °C for curing. Next, they were transferred directly to a hot furnace to induce particle growth. In this final step, the samples were heat-treated at 600 °C for 0, 10, 60, and 120 minutes. In the following, as-received refers to samples that were cured at 400 °C but on which no further heat-treatment was conducted. After heat-treatment, the samples were immediately removed from the furnace to quench particle nucleation and growth.

2.4 Optical characterization

Photoluminescence spectroscopy: Photoluminescence excitation and emission spectra were recorded with a high-resolution spectrofluorometer (Fluorolog 3, Horiba). The light source was a continuous-wave 450 W Xe-lamp, and the detector was a Hamamatsu R2658P photomultiplier. The photoluminescence excitation spectra were corrected for Xe-lamp intensity. Both photoluminescent and photoluminescence excitation spectra were corrected for the spectral response of the photomultiplier tube.

UV VIS spectroscopy: UV VIS spectra of Eu/Ag co-doped samples were collected in direct transmission mode on a Cary 5000 (Agilent) double-beam spectrometer between 200 and 600 nm with a spectral resolution of 1 nm. The optical extinction was calculated using the uncoated substrate as reference material.

Side-emission of fibers: The side-emitting properties of coated fiber were measured using a custom set-up shown in Fig. 2 [35]. The fiber was threaded through an integrating sphere coupled to a CCD-spectrometer (Ocean Optics, Maya2000 Pro). A motorized stage was used to move the integrating sphere incrementally alongside the optical fiber, and the spectra of the side-emitted light were recorded for every position. A passively Q-switched pulsed Cr:YAG laser emitting at 266 nm with an average power of 100 mW was used as a light source (MPL-N-266, HJ Optronics, USA). In this way, point-by-point emission data were obtained alongside the fiber axis in segments of up to 120 mm.

 

Fig. 2. The motorized linear stage for analyzing side-emission from a surface-modified optical fiber. The coated fiber is threaded through an integrating sphere, which is moved alongside the fiber. The emitted spectral distribution of the fiber is measured with a spectrometer and recorded with a computer as a function of longitudinal position. As the light source, a laser emitting in the UV range is used. More information on this setup can be found in [35,36]. Adopted with permission from Ref. [35] under the terms of the Optica Open Access Publishing Agreement.

Download Full Size | PDF

3. Results and discussion

3.1 Influence of Ag concentration on Eu³⁺ photoluminescence

3.1.1 Excitation with resonant wavelength

Incorporating silver into the Eu-doped coating increased the luminescent emission in all bands compared to coatings with no silver, as shown in Fig. 3(a) for excitation at the resonance wavelength of 394 nm. We observed that the photoluminescence could be enhanced ∼ 16-fold when adding 5% Ag by annealing for 10 min and ∼ 12-fold for 2% Ag after 60 min annealing. Aside from the overall increase in emission intensity, other spectral properties remained unaffected. For reference, similar annealing of an Ag-free Eu-doped coating did not cause any changes in the PL intensity.

 

Fig. 3. PL emission spectra of planar samples with varying Ag concentration and annealing time for excitation at the resonant wavelength of 394 nm: (a) Emission spectra of ⁵D₀ → ⁷F_J (J = 0,1,2,3) transitions: all emission bands increased uniformly, shown by way of example for 2% Ag. (b) Relative photoluminescent intensities of the ⁵D₀ → ⁷F₂ transition normalized to that of the sample 0% Ag, 60 min. (c) Photograph of two selected samples under UV light irradiation (contrast and brightness were enhanced for visibility).

Download Full Size | PDF

We quantify the emission enhancement of Eu³⁺ due to silver incorporation using the relative luminous intensity. This quantity is calculated by normalizing the peak height of the ⁵D₀ → ⁷F₂ induced electronic dipole transition to the corresponding peak height of the 0% Ag sample after 60 min of heat treatment. The latter transition is affected by the local symmetry of Eu³⁺. Comparison to the structure-insensitive magnetic dipole transition of ⁵D₀ → ⁷F₁ did not reveal any significant differences in the observed trends (not shown), indicating that the addition of Ag indeed affects only the PL intensity but not the spectral emission characteristics.

Higher silver concentrations led to higher emissivity at the beginning of the annealing process, as shown in Fig. 3(b) for 600 °C. However, annealing led to different outcomes for emission intensity, depending on silver concentration and annealing time: all samples initially showed an increase in the photoluminescence intensity. Then, while the PL intensity for the weakly co-doped samples continued to increase, the two most highly silver-doped samples reached a PL maximum for annealing times of 10 min and 60 min, respectively. Consequently, the PL intensity of the 2% Ag sample surpassed that of the 5% Ag sample for treatment times exceeding 60 min.

3.1.2 Excitation with non-resonant wavelengths

Excitation with non-resonant wavelengths (here: λ_ex < 394 nm) of the Ag-containing coatings resulted in the same characteristic emission spectra as did resonant excitation, but with a significantly higher intensity as compared to the non-Ag-containing and the Ag-containing coatings (as shown in Fig. 4). This enhanced emission increased with decreasing excitation wavelength and was more pronounced at higher Ag concentrations and shorter annealing times (see Fig. 5). The observed broadband enhancement is so extensive that it also enhances the excitation at the resonance wavelength of 394 nm. This padding of the excitation at resonance with the off-resonance excitation band, as seen in Figs. 4 and 5(a), leads to the emission enhancement observed in the previous paragraph and in Fig. 3(b).

 

Fig. 4. Excitation and emission spectra of an Ag-free coating (top) and a coating with 2 mol% Ag (bottom). With Ag incorporation, the emission is stronger for non-resonant excitation at 350 nm; without Ag, it is strongest for resonant excitation at 394 nm. The emission spectra were recorded with 350 nm non-resonant excitation (blue) and with 394 nm resonant excitation (red), respectively. Excitation spectra were recorded monitoring emission at 612 nm.

Download Full Size | PDF

 

Fig. 5. (a): PL emission spectra (λ_em= 612 nm) of coatings with 5, 2, and 1 mol% Ag for different annealing times. A broad out-of-resonance excitation band < 400 nm is visible for the Ag-containing coatings. The shape and size of this excitation band depend on Ag concentration and annealing time. The blue arrow indicates the strength of the out-of-resonance excitation. The resonant excitation wavelengths are magnified in the inset. (b) Schematic of the energy transfer from Ag clusters to Eu³⁺: the proposed mechanism for out-of-resonance excitation.

Download Full Size | PDF

We measured the non-resonant excitation response by recording the intensity of the ⁵D₀ → ⁷F₂ transition at 612 nm as a function of excitation wavelength in the range from 330 nm to 500 nm. The resulting excitation spectra in Fig. 5(a) represent the response of the induced electric dipole transition to a varying excitation wavelength. The off-resonance response generally increased for excitation at a lower wavelength. A broader and more intense out-of-resonance excitation band was observed with increasing Ag concentration. The shape and size of this wide excitation band depend on annealing time: the highest out-of-resonance PLE intensities were observed in non-annealed samples with high Ag concentration at the lowest wavelength of 330 nm. Here, further annealing reduced the PLE intensities. At lower Ag concentration (1%), the distinct resonant excitation peaks of Eu³⁺ at 394 nm became visible, whereas the out-of-resonance band still overlapped these for higher Ag concentrations (2% and 5%).

The height and width of the broad out-of-resonance excitation band changed upon annealing: the onset of the non-resonant excitation of the annealed sample shifted to a longer wavelength. These bands intersect with the spectra of the as-received sample (without annealing), as shown in the insets of Fig. 5. This behavior is consistent with the increase and decrease in emission under resonant excitation observed for 5 and 2, 1 mol% Ag during annealing in Fig. 3(b). Consequently, the non-annealed samples always showed higher enhancement for excitation wavelengths smaller than the point of intersection (>360nm) and vice versa.

The broad non-resonant excitation band below 400 nm can be explained by an energy transfer process between molecular-like Ag species and Eu³⁺ [21–25,37,38]. Figure 5(b) shows this schematically: an Ag cluster absorbs high-frequency photons and transfers the energy to nearby Eu³⁺, acting as a sensitizer [23]. By intersystem crossing, the excited electrons relax to the ⁵D₀ level, leading to an increased population density. As a result, the rate of radiative decay to ⁷F_J increased uniformly, enhancing the intensity in all emission bands.

The variations in shape and size of the out-of-resonant excitation spectrum upon annealing are related to a change in Ag cluster concentration during annealing. Especially in the high Ag samples, metallic nanoparticles are grown at the expense of clusters. Simo et al. observed a similar phenomenon while investigating the growth of Ag⁺ to Ag clusters and further to metallic Ag nanoparticles in ion-exchanged soda-lime silicate glasses [27]. They found that upon heating below a threshold temperature of 410 °C, small Ag clusters (mostly dimers) formed, which did not grow into metallic nanoparticles. Above this temperature, larger nanoparticles were generated as a function of annealing time, and LSPR was observed [27]. Their threshold temperature of 410 °C was close to the temperature of 400 °C we used here for curing the coatings.

3.2 Ag cluster and nanoparticle formation

The enhanced off-resonance photoluminescence is caused by Ag clusters, not by nanoparticles. Two observations confirm this: first, the samples with the strongest photoluminescence lack the extinction band at around 400 nm [39,40] related to nanoparticle plasmon resonance. Second, the annealing process leads to the formation of nanoparticles at the expense of clusters. This formation of NPs becomes visible by increased plasmon resonance in Fig. 6 and decreased broadband UV-absorption. It is accompanied by diminished photoluminescence in Fig. 3. We conclude that the diminishing PL emission during annealing is due to the vanishing of Ag-clusters, which grow into Ag-NPs.

 

Fig. 6. (a): Optical extinction of samples containing 5 mol% Ag for different annealing times at 600 °C. After annealing (10, 60, 120 min), the Ag LSPR signal becomes apparent, indicating plasmonic Ag nanoparticles. Optical loss in the UV is attributed to the absorption of Ag⁺ and Ag clusters of different sizes. A spectrum of Ag-free samples after treatment is added for comparison. (b): Schematic illustrating the growth of Ag⁺ to Ag clusters and metallic NPs upon annealing, adopted from [27].

Download Full Size | PDF

Ag clusters do not exhibit LSPR at around 400 nm and can be distinguished from nanoparticles by optical transmission measurements. Thus, we investigated the samples by UV-Vis photo-spectrometry to determine whether nanoparticles were present. Selected results of the extinction of samples containing 0 and 5 mol% at different annealing times are presented in Fig. 6(a). The coating's extinction E was calculated by normalizing the transmission spectra to the uncoated substrate.

When we compare the annealed Ag-containing samples, we observe two distinct phenomena: the spectra of the annealed samples show an extinction peak around 420 nm. This peak becomes more pronounced with increasing annealing time. Additionally, Ag-doped samples possess a strong UV extinction for wavelengths smaller than 300 nm, decreasing with annealing time. The un-annealed sample differs from this observation: it has an extinction minimum at 320 nm and features a broad extinction band at 550 nm and a shoulder signal at ∼ 250 nm. In comparison, the Ag-free sample did not exhibit any notable extinction bands in this spectral range.

The peak at 420 nm indicates LSPR [39], which confirms the formation of Ag nanoparticles particles during annealing. The assumed process is Ostwald-ripening, as shown schematically in Fig. 6(b) [27], whereby larger Ag nanoparticles grow at the expense of smaller Ag clusters [41]. This interpretation agrees with the observed photoluminescence excitation spectra in Fig. 3: The decrease in out-of-resonance excitation PL is accompanied by the increase of plasmon resonance in the extinction spectra in Fig. 6(a) upon annealing; NP growth leads to a decrease in the number density of clusters participating in the energy transfer process.

3.3 UV extinction due to silver incorporation

The addition of silver causes an overall increase in UV extinction compared to the silver-free coating. We suggest that atomic silver, silver clusters, and silver NPs contribute to UV extinction. While the detailed absorption mechanisms of Ag agglomerates are still unknown for sol-gel layers, the broad absorption envelope is attributed to overlapping contributions from various cluster species with different sizes.

Comparing spectra of Ag-doped to Ag-free coatings suggests that extinction at wavelengths < 300 nm is also caused by Ag species. Different silver species are known to cause absorption in the UV: Ag⁺ in silicate glass absorbs at wavelengths < 250 nm [42]. Furthermore, several molecule-like clusters absorb at wavelengths < 350 nm in glasses or aqueous solutions [27,42,43]. However, other than in these reports, we could not identify distinct peaks due to specific Ag cluster species (see Fig. 6). We suggest that a mixture of different Ag cluster sizes is present in the coatings. Consequently, the observed absorption band contains overlapping contributions from several species with variable sizes.

The extinction spectra of the reference sample and the as-received one strongly differed from those of the other samples. For the Ag-free reference, even the Eu³⁺-related absorption bands were hardly visible due to the low layer thickness and dopant concentration. This shows that nearly all transmission loss is due to the presence of Ag in the coating. The broad maximum and the shoulder in the extinction spectrum of the untreated Ag-containing samples may be due to non-dissolved AgNO₃ or residual carbon from incomplete curing. Alternatively, this absorption band could be caused by the formation of Ag₂O [44]. The PL enhancement at 350 nm was still the largest for this sample, although the absorption showed a distinct minimum at this wavelength.

3.4 Influences due to structural changes

Influences of PL emission due to structural changes of the matrix surrounding the europium next to the Ag clusters can be excluded within the measurement uncertainty. Several observations confirm this: PL measurements of samples with and without doping of Ag in Fig. 4 unambiguously show that silver is necessary for PL enhancement for excitation < 394 nm. Furthermore, both the coating and the europium show no signs of crystallization after annealing, as confirmed by grazing incidence X-ray diffraction conducted on individual films before and after annealing. Additionally, The PL lifetime measurement in Fig. 7(a) and the ED to MD transition ratio in Fig. 7. (b) show no changes due to the annealing of the coating in addition to the change from as-receive to annealed: for the annealed samples, the emission characteristics do only weakly depend on annealing time.

 

Fig. 7. (a) Intensity average lifetime of the luminescence for the 5% Ag sample for different annealing times. Excitation wavelength 350 nm and 394 nm; measurement at the ⁵D₀ → ⁷F₂ induced electronic dipole (ED) transition at 612 nm. (b) Peak height ratio of the induced electronic dipole transition ⁵D₀ → ⁷F₂ at 612 nm and the magnetic dipole (MD) transition ⁵D₀ → ⁷F₁ at 591 nm; excitation at 394 nm. Both graphs only show strong changes for the transition from as-received to annealed, probably due to densification accompanied by the release of residual OH-groups from the coating, which quenches luminescence and reduces lifetime [45,46].

Download Full Size | PDF

First, we rule out an intrinsic effect of the coating on the observed PL characteristics for Ag-free coatings by considering the 0% Ag curve in Fig. 5, the 1% Ag curve in Fig. 3, and the 0.1% Ag curve in Fig. 2: All examples show that when there is no or only little Ag in the sample, there is no significant increase in PL for the annealed samples. Thus, PL changes occur only in the presence of Ag.

Second, X-ray diffraction did not reveal any crystallization beyond the growth of metallic silver following thermal treatment.

Third, the PL lifetime measurement of the 5% Ag sample shows almost no change with increasing annealing time, indicating no structural changes due to annealing, as shown in Fig. 7. Only the as-received samples exhibit significantly shorter lifetimes, likely due to the presence of residual OH groups from the sol-gel process. OH vibrations increase the non-radiative decay rate [45,46]. Annealing densifies the coating and removes OH groups, leading to an initial increase in lifetime. Subsequently, the PL remains constant as the structure of the matrix does not change further.

Fourth, the ratio between the ED and MD transitions (often referred to as the asymmetry ratio) shows no systematic changes in the europium environment due to Ag concentration or annealing. This can be seen in Fig. 7(b), as compared to Fig. 3(b) or Fig. 5(a). As with the PL lifetime, the initial increase in the PL ratio for the transition from as-received to annealed samples can also be explained by the removal of residual OH groups and densification due to the annealing of sol-gel derived coatings [47]. As mentioned earlier, the ⁵D₀ → ⁷F₂ transition is sensitive to the environment, whereas the MD transition is not [32]. A comparison of the two intensities provides an opportunity to observe structural changes in the local environment of Eu³⁺ [48]. The ED to MD ratio stays almost constant after the initial increase because there is no further change in the matrix.

3.5 Luminescence-enhanced coatings on optical fibers

We now applied the Ag/Eu-co-doped coatings on the surface of optical fibers, generating a line-shaped light source with a broadband UV light converter. Thereby, the fiber served as a guide for the excitation light to the coating, and the coating as the emitter of the converted radiation. The coating emits converted light in a characteristic red glow but also transmits some UV excitation light. The observed side-emission intensity (per unit length) of the converted light is 1.28 ± 0.08 times higher than that of the unconverted light. This conversion ratio is almost constant over the measured distance.

When the guided UV light interacts with the coating, it is partially converted and released into the surrounding. This effect is seen in Fig. 8 (top): the UV light becomes apparent as light-blue side-emission on the uncoated part of the fiber guide and as red emission on the Eu/Ag-coated part. For the illustrated experiment, an unclad fused-silica fiber was coated with the 5 mol% Ag sol and cured at 400 °C (see section 2.1) to obtain high out-of-resonance PL enhancement. In the lower part of Fig. 8, the coated fiber was clamped to the fiber measurement set-up, and a 100 mV UV laser was coupled to the left-side fiber end. The integrating sphere was moved alongside the fiber [35,49] to record position-resolved emission spectra using the attached CCD-spectrometer. The result of such a scanning procedure is shown in Fig. 9(a).

 

Fig. 8. Luminescent emission from side-emitting fiber: Top: partially coated fiber showing UV emission and Eu³⁺-photoluminescence. Bottom: Experimental set-up with the fiber threaded through an integrating sphere connected to a spectrometer. The closed sphere (in the photograph: opened for visibility) is moved alongside the fiber to obtain position-resolved emission spectra. The white ceramic tubes are used to guide and shield the fiber inside the sphere; they act as the aperture. The baffle shields the sensor from direct exposure.

Download Full Size | PDF

 

Fig. 9. Position resolved spectra of the fiber coatings spectral exitance: the photoluminescent emission and the transmitted UV excitation. (a) shows a heat map of spectral emittance as a function of position and wavelength. (b) shows the average spectral exitance from z = −20 mm to 96 mm, resulting in three distinctive emission peaks of Eu³⁺ at 590 nm, 620 nm, and 700 nm, and one peak for the UV excitation at 267 nm. (c) shows the evolution of the radiant exitance for the PL emission integrated from 550 nm to 750 nm as a function of z-position. This is compared to the detected UV excitation integrated from 260 to 270 nm. Both were fitted with an exponential decay yielding 0.063 ± 0.003 mm⁻¹ for the PL and 0.062 ± 0.002 mm⁻¹ for the UV. The green curve (associated with the right axis) shows the conversion ratio, which is the ratio of the PL curve to the UV curve. It has an average ratio of 1.28 ± 0.08 in the homogeneously coated range from 0 mm to 100 mm. This shows that more energy is converted by the layer than transmitted and that the conversion rate is almost constant. (d) principle of the light conversion: guided light leaks in the coating and is converted to PL or transmitted.

Download Full Size | PDF

Figure 9(b) shows the spectrum averaged over all z-positions; it has the same distinctive emission peaks of Eu³⁺, which we already discussed in Section 3.1. We conclude that the guided UV light was converted to visible light by the Eu coating. Figure 9(c) shows the emissivity integrated over the wavelength range of 550 nm to 750 nm. We see that the emissivity of the fiber decayed approximately exponentially with increasing distance to the start of the coating.

The observed exponential decline follows the Beer-Lambert law, stating that a constant fraction α of the guided light flux $\phi $ is lost per unit distance $\textrm{d}\mathrm{\Phi }/\textrm{dz} ={-} \mathrm{\alpha \Phi }$, which is subsequently emitted with the radiant exitance $M ={-} \textrm{d}\mathrm{\Phi }/\textrm{dz}$. This leads to the observed exponential decay $\mathrm{\Phi } = {\mathrm{\Phi }_0}\exp({ - \mathrm{\alpha }z} )$ in transmitted and emitted light flux [36]. The prerequisite is that the modes in the fiber are in a steady-state power distribution [49]. An exponential fit to the integrated PL data yielded a decay constant of α = 0.063 ± 0.003 mm⁻¹. All peaks decay with approximately the same constant (not shown), which shows that the coating is homogeneous.

In addition to the desired emission of converted light, guided UV light is also emitted through the coating. The extent of transmission is shown in Fig. 9(c) in comparison to PL emission. This UV transmission also decreases exponentially, with a constant of α = 0.062 ± 0.002 mm⁻¹, which is almost identical to the decrease in PL emission. This similarity suggests a common loss mechanism. The ratio of converted PL emission to transmitted unconverted UV light is nearly constant with a ratio of 1.28 ± 0.08 over the range of the homogeneously coated part from z = 0 mm to 96 mm.

The common mechanism responsible for the lateral emission is the change in the total internal reflection of the fiber due to the coating, as shown in Fig. 9(d): When the fiber is uncoated, the light is guided at the glass-air interface. When the light reaches the coating, the state changes: because the refractive index contrast is lowered, most of the light in the fiber can no longer be guided and leaks into the coating. There, most of it is converted into PL emission, but some is able to pass through the coating. These experiments show that it is possible to convert guided UV light into visible radiation with the aid of out-of-resonance excitation of Eu. Furthermore, because the coating has low absorbance in the visible spectral range, it can be applied additionally to any side-emitting fiber to convert the UV part of the guided radiation for solar energy harvesting. Increasing the emission of visible radiation, such coatings would benefit applications that require structured illumination, for example, in photochemistry and photosynthesis [6].

4. Conclusions

In summary, we demonstrated strongly enhanced photoluminescent side-emission from an optical fiber using an Ag/Eu-codoped coating for broadband UV-Vis spectral conversion. Energy transfer from small Ag clusters uniformly increased the emission intensity of all Eu³⁺-emission bands, enabling efficient spectral conversion even at low coating thickness and low Eu³⁺ concentration. We observed the strongest out-of-resonance PL excitation in samples coated with sols of 5 mol% Ag, for which the PL emission intensity was increased by a factor of sixteen. Prolonged annealing led to the formation of metallic Ag particles with LSPR, which reduced the efficiency of non-resonant PL excitation. Our approach provides a facile method for fabricating side-emitting fiber light sources with broadband spectral conversion. The observed conversion ratio of converted to unconverted light was 1.28 ± 0.08, practically constant alongside the observed length of the side-emitting fiber.