3.1. Photoluminescence in Bulk Glass
The absorption spectrum in the range 250–800 nm obtained for the PZnAg-Eu glass is shown in
Figure 1. The UV edge is around 270 nm mainly due to silver ions absorption as previously observed for silver ions containing zinc phosphate glass [
28]. Additional characteristic absorption bands of Eu
3+ ions are observed at 318, 362, 380, 393, 464 and 526 nm assigned to the transition from the ground state
7F
0 to excited states
5H
3,
5D
4,
5G
2,
5L
6,
5D
2 and
5D
1, respectively. There is no evidence of Eu
2+ formation in the glasses and no absorption band above 300 nm, which could be attributed to surface plasmon resonance or silver cluster species such as (Ag
2)
+ or (Ag
3)
2+, for instance.
Fluorescence spectroscopy have been measured under different UV excitation wavelengths to investigate the emission properties of silver and Eu
3+ ions, and the results are shown in
Figure 2a. Under 235 nm excitation, two broad emission bands are observed due to two different intrinsic environments of silver with peak emission at 290 (site A) and 380 nm (site B), as well as the characteristics emissions of the 4f–4f transitions of Eu
3+:
5D
0-
7F
j (J = 0–6). The site A emission is assigned to Ag
+ ions with the
3D
J→
1S
0 transition, while site B emission is assigned to (Ag
+)
2 pairs with the S
1, T
1→S
0 transition [
29]. Note that the considered silver ion concentration makes site B largely preponderant compared to site A [
2]. Note also that emission spectra of
Figure 2a emitted with the present PZnAg:Eu glass are highly similar to that reported for a very close composition of PZnAg (without Europium insertion) in Figure 2 from Petit et al. [
2]: this makes us confident that there is no significant existence of broadband Eu
2+ emission in the visible range in the present paper in the pristine glass (nor in the laser-induced modifications). With the increase in the excitation wavelength to 260 nm, almost exclusively site B emission as well as the characteristic emission of Eu
3+ are observed. Excitation at UV light about 235 nm or 260 nm, promotes the electron to the excited state of both sites A and B, or site B, respectively. Thus, the excited electrons can relax to the ground state radiatively, but also non-radiatively, including relaxation paths involving energy transfers to the neighboring Eu
3+ ions. The occurrence of energy transfer from silver sites to Eu
3+ ions is favored by the energy level accordance between the excited states of the silver sites with the
5L
6 level of Eu
3+ ions. Under excitation at 365 nm, only the line-shape characteristics emissions of the Eu
3+ transitions at 578, 590, 612, 652, and 702 nm, corresponding to the Eu
3+ ions emission
5D
0→
7F
J (J = 0–4), are observed.
Based on the photoluminescence spectra of the glasses presented in
Figure 2, the colors of samples under the three distinct excitations at 235, 260, and 365 nm were characterized by International Commission on Illumination (CIE) chromaticity diagram (
Figure 2b). The colors shift from purplish pink to red with the increase in excitation wavelength from 235 to 365 nm, which demonstrates that the luminescence color can be systematically tuned by changing the excitation wavelength.
The photoluminescence excitation spectra are shown in
Figure 3. By monitoring the emission at 290 nm (site A), a broad excitation band centered at 230 nm is observed, attributed to Ag
+ ion. At 340 nm (site B), two overlapping broad excitation bands are observed, centered at 230 and 260 nm, from the Ag
+ ion and (Ag
+)
2 pairs, respectively. When monitoring the Eu
3+ emissions at 612 or 700 nm, besides the two broad excitation bands of silver originating from the energy transfer of Ag
+ ions and (Ag
+)
2 pairs energy transfer to the rare earth ions, several characteristic f–f transitions’ sharp lines are observed. Under UV light, Ag
+ ions and (Ag
+)
2 pairs can be excited to the
3D
j and S
1 energy levels, respectively, and then the non-radiative S
1→S
0 and
3D
J→S
0 transitions may transfer the excitation energy non-radiatively to Eu
3+ ions, with such populated Eu
3+ excited levels leading then to the observed typical Eu
3+ fluorescent emission. Such non-radiative energy transfer is understood as a non-radiative resonant relaxation from the silver species (donor) to the Eu
3+ ions (acceptor), without any electron transfer. Indeed, the overlapping of excitation bands evidences the potential ability of energy transfer from silver ionic species to the Eu
3+ ions. Furthermore, no fluorescence emission signature of Eu
2+ ions was observed, which corroborates the absence of electron transfer in the present energy transfer.
3.2. Direct Laser Writing of Fluorescence Structures
Figure 4a shows a wide-field fluorescence microscopy image of coil patterns obtained by DLW for PZnAg:Eu glass sample motion with controlled speed from 10 to 100 µm/s (corresponding to C, B and A lines), and for different irradiances from 4.0 to 6.7 TW cm
−2 (labeled columns 3, 4, 5 and 6). These patterns allow for investigating the influence of the laser parameters in the silver clusters growth and their interaction with the Eu
3+ ions. A high-resolution confocal image for excitation 435 nm (obj. 100× NA1.4) is given in
Figure 4b, showing the detailed distribution of the fluorescent double-line structure produced by several laser passes. These double lines correspond to the formation of parallel planes containing the silver clusters at the border of the voxel of interaction. This phenomenon has been described in detail elsewhere [
4].
Micro-absorptions measurements were performed in transmission mode both in a pristine area of the glass and in structures zones over a 50 × 50 μm
2 surface giving a direct experimental access to the local differential micro-absorption due to the laser-induced structures. They are presented in
Figure 5a,b for speeds of 10 and 100 µm/s, respectively. These spectra were decomposed by Gaussian contributions. Two contributions are observed around 335 nm (3.7 eV) and around 285 nm (4.35 eV). These two contributions increase in intensity with the irradiance and decrease in intensity with the speed. As can be seen, the relative intensity of these two bands are modulated by the dose deposition indicating that several centers are formed during the DLW process. The strongest dose tends to favor the absorption band around 335 nm (3.7 eV).
The micro-emission spectroscopy of laser-induced silver clusters are presented in
Figure 6 for two different excitations wavelengths at 325 and 405 nm as a function of the sample motion velocity for 6.75 TW.cm
−2 irradiance (
Figure 6a,c) and as a function of the irradiance for a fixed sample motion velocity of 10 µm/s (
Figure 6b,d).
At 325 nm (
Figure 6a,b), the emission spectra consist of two major contributions: one broad emission ranging from 400 to 700 nm attributed to silver nanoclusters and sharp emission peaks within the 575 to 715 nm region assigned to Europium ions. The decrease in sample velocity (
Figure 6a) and the increase in the irradiance (
Figure 6d) corresponding to a rise in the deposited dose promote the increase in the broad emission band intensity assigned to the formation of Ag
mn+ species and also promote the emission of the europium ions. The most intense emission of Eu
3+ around 612 nm is attributed to the hypersensitive electric dipole transition (
5D
0→
7F
2). This emission is strongly influenced by the chemical environment of the Eu
3+ ions in the host network. On the other hand, the
5D
0→
7F
1 transition occurs via the magnetic dipole mechanism and its intensity is not sensitive to the chemical environment of Eu
3+ ions. Therefore, the emission intensity ratio of these two transitions can be used to highlight a site evolution of the rare earth ion. In the present case, one can notice that the spectral shape of the 4f–4f transition remains constant and the intensity ratio between the
5D
0→
7F
2 and
5D
0→
7F
1 also remains at a constant value of 4.1 ± 0.1. Thus, an absence of major modification of the chemical environment at the vicinity of the Eu
3+ after DLW can be assumed.
Micro-fluorescence emission spectra demonstrate interesting behavior in terms of display and colorimetry. Indeed, the fluorescence emission analysis under UV excitation at 325 nm and associated projection in the XYZ tri-chromatic coordinates in the CIE chromaticity diagram shows that the chromatic coordinates can be continuously controlled in 3D by means of laser irradiation, namely from red (almost pristine glass with quasi-null action of the laser inscription) to white light emission, as shown in
Figure 7. In the case of the largest irradiances (I = 6.7 TW/cm
2 for patterns A6, B6 and C6 depicted in
Figure 7a), the trichromatic coordinates get very localized in the white emission domain whatever the three used velocities suggesting that the broadband silver cluster emission dominates the reddish contribution of Eu
3+ ions in the resulting tri-chromatic behavior. Similarly, in the case of the lowest velocity v = 10 µm/s (namely, the largest number of cumulated pulses with patterns C3 to C6 depicted in
Figure 7c), the deposited dose is also high enough to produce enough efficient fluorescent silver clusters to overcome the Eu
3+ emission and dominate the resulting tri-chromatic behavior. However, a larger velocity of 50 µm/s tends to limit the amount of silver clusters and its fluorescence emission amplitude, locally allowing for controlling the tri-chromatic coordinates between the initial red coordinates of the pristine glass and the white light emission of the silver clusters, as shown between points B2 and B3 in
Figure 7b. Such measurements result from micro-luminescence in confocal microscopy, so that they are measured at the silver cluster positions. By adjusting the probed volume of the material on the one hand, and by fixing the filling rate of silver cluster (typically with the inter-distance between successive laser passes in a given plane, or by producing thick multi-plane patterns) on the other, one can fully control the continuous evolution of the measured tri-chromatic coordinate between red to white emission. This thus suggests the ability to perform the localized 3D inscription of emitting areas with tunable tri-chromatic coordinates, which goes beyond the spatially homogeneous control of the tri-chromatic coordinates means of selecting different fluorescence excitation wavelengths, as shown in
Figure 2b (from the pristine Eu
3+-based red emission to a pinkish emission resulting from the red Eu
3+ ion emission and the blue Ag+ ion emission in the pristine glass). Laser inscription thus allows an additional experimental parameter to tune the 3D localized tri-chromatic coordinates.
To further investigate the energy transfer from silver clusters to Eu
3+ ions, time-resolved spectroscopy of silver nanocluster emission was obtained for the PZnAg and PZnAg:Eu glasses, for a collection over the 470–550 nm spectral range and using a pulsed excitation (6 ps, 4.75 MHz) at 435 nm. The collected fluorescence predominantly corresponds to that of silver clusters, because Eu
3+ ions show no emission in this spectral range and because Ag
+ ions of the glass matrix show negligible fluorescence excitation. In these conditions,
Figure 8a,b show the FAST-FLIM images for the laser inscribed patterns, where the color code and the brightness depict the pixel-to-pixel FAST-FLIM average lifetime and the associated fluorescence intensity, respectively. For both the PZnAg and PZnAg:Eu glasses, the FAST-FLIM images corroborate the double-line spatial distribution of laser-induced fluorescent silver clusters, as shown in
Figure 4. The intensity weighted FAST-FLIM lifetime histograms were extracted from
Figure 8a,b, and are reported in
Figure 8c. These histograms clearly show the mean lifetime shortening of the silver cluster excited states in the presence of Eu
3+ ions, with a typical peak value of the histograms occurring at a shorter lifetime of 1.94 ns for the PZnAg:Eu glass, compared to the 2.44 ns lifetime for the PZnAg glass. This supports the hypothesis of a quenching by energy transfer from these clusters to the neighboring Eu
3+ ions which directly affects the emission rate of the cluster emission.
Figure 8c also reveals a lower emission intensity of silver clusters in the presence of Eu
3+ ions, such partial emission quenching being also consistent with nonradiative energy transfer by resonant energy transfer from silver clusters to Eu
3+ ions. By properly taking into account the normalization with the incident pump intensity at 435 nm for both histograms of
Figure 8c, the corresponding time-integrated intensity of both PZnAg and PZnAg:Eu glasses led to the intensity quenching
of the silver cluster emission. Note that such global intensity quenching includes any sources of quenching, namely the energy transfers at play towards the Eu
3+ ions but also other possible phenomena whose contributions will be discussed below.
As the fluorescence emission of the silver clusters corresponds to a broadband multi-band emission that covers the whole visible range and depends on laser irradiation parameters, the fluorescence lifetime was also evaluated with a multiple exponential decay model in the sub-ns and ns range. As detailed in the experimental section, curve fitting with a multi-exponential decay model was performed on decays with better statistics, corresponding to the binning of pixels of larger regions of interest. As shown in
Figure 8d, optimal time decay fitting was achieved with a four-exponential model with the adjusted amplitudes and time constants being reported in the
Table 1. Such non-usual fitting appeared to be necessary, and sufficient, to correctly fit the experimental decay with respect to standard acceptable statistical parameters. However, it is not attempted thereby to attribute each component to a precise population of emitters. This fitting is exclusively used to determine the average decay time, taking into account the statistical characteristics of single photon counting. The addition of components would not be meaningful, leading to covariance of the fitting parameters. Other models were not taken into consideration, since the estimation of an average value was sufficient in the context of this study, in order to characterize energy transfer efficiency. Still, the used model demonstrates that the complexity of lifetime decay of the system under investigation requires at least four typical decay times, but reality may be even more complex with additional, much shorter decay times.
Based on the fitting results from the table, the intensity-weighted average emission lifetime
is of
= 2.44 ns for the PZnAg glass, and it decreases down to
1.94 ns in the Eu
3+-containing glass, in agreement with the lifetimes denoted by the peak positions of the histograms of
Figure 8c. This result corroborates the occurrence of energy transfer from Ag
mn+ species levels to Eu
3+ ions. According to the expected lifetime of europium Eu
3+ in such glass, which is expected to be around a millisecond for the most intense emission around 612 nm, the collected decay measurement strictly corresponds to the silver cluster contribution.
To go further in the interpretation of energy transfers, based on the fitting results from
Figure 8d and summarized in the table, the amplitude-weighted average emission lifetime
was also estimated, leading to
= 1.34 ns and to
= 0.98 ns for the PZ:Ag and PZnAg:Eu glasses, respectively. The amplitude-weighted quenching rate
of the emitting silver clusters can be estimated as
. The comparison of the amplitude-weighted quenching rate
with the intensity quenching rate
helps in sorting the contributions of the emission quenching of the silver clusters. Indeed, the ratio
indicates that 66% of the emission quenching of the silver clusters in presence of Eu
3+ ions is actually due to energy transfer. In addition, the estimation
indicates that 33% of the emission quenching of the silver clusters in the presence of Eu
3+ ions stands for a lower concentrations of silver emitters, affecting the fluorescence intensities but not the lifetimes. This can reasonably be interpreted by a slightly lower efficiency of silver cluster production during the femtosecond laser writing process. Indeed, in the laser inscription stage, we observed that Eu
3+ ions undergo multi-photon absorption and subsequent reddish fluorescence emission, which partially reduces the effective femtosecond laser intensity and thus completes the activation of the silver photochemistry and the associated creation and growth of silver clusters.
Regarding the europium emission intensity, the 612 nm Eu
3+ emission increases up to 5-fold and to 3.5-fold for excitations at 325 and 405 nm, respectively (
Figure 9a,b), indicating an energy transfer from silver species to Eu
3+ ions [
19,
25,
27]. The Eu
3+ emission enhancement is clearly dependent on the excitation wavelength, as a larger efficiency is observed for an excitation at 325 nm. Such wavelength excitation dependence may be directly interpreted from
Figure 5, with a more efficient excitation of silver clusters at 325 nm compared to 405 nm, leading thus to a more intense fluorescence emission and consequently to a larger amount of excited silver clusters compatible with resonant non-radiative energy transfers to Eu
3+ ions. Still, the existence of various populations of distinct silver clusters created during DLW has been reported [
13], showing a partially selective excitation with radiations at 325 or 405 nm which may also play a selective role in the observed energy transfer. Indeed, these distinct types of silver clusters may show distinct energy transfer efficiencies to the Eu
3+ ions. The present spectroscopy cannot directly identify the nuclearities and associated stabilized redox charges of these silver species, nor their respective populations [
13].
According to the singlet-triplet energy level model of Ag
mn+ species proposed by Velazquez et al. [
29] and to the results obtained from the photoluminescent excitation and emission spectra of the Eu
3+ and Ag
mn+ species in the co-doped glass before and after laser–matter interaction, the energy transfer pathway was tentatively described in
Figure 10. Under UV-visible light, Ag
+ ions and Ag
mn+ species can be excited to the S
1 and T
2 energy levels, and then the S
1→S
0 and T
2→S
0 transitions may transfer the excitation energy to Eu
3+ by relaxations, thus populating the
5H
3 level and by non-radiatively relaxing from the
5D
4 and
5L
6 levels down to the
5D
0 energy level. The subsequent radiative transitions from the
5D
0 energy upper level leads to the visible Eu
3+ emissions observed in
Figure 2 and
Figure 6. In addition, it is worth mentioning that the spin-forbidden transition of T
2→S
0 with relatively longer lifetime may mainly contribute to the above energy transfer processes. Still, at this level of description, the present spectroscopy cannot directly identify the mechanisms at play during the non-radiative resonant energy transfers; namely, one cannot directly discriminate between Dexter resonant energy transfers (DET) related to electron exchange operating at a very short scale (requirement of the wave function overlap) and Förster resonant energy transfers (FRET) related to dipole–dipole interactions spanning longer scales up to several nanometers typically (but highly sensitive to the solicited oscillation strength, which is known to be rather weak for the lanthanide ions) [
30]. A balance between DET and FRET, both occurring simultaneously, may even exist. Such discrimination in the origin of the involved resonant energy transfer could further be addressed thanks to a set of glasses with various rare earth doping concentrations, which is beyond the scope of the present article.
Following the laser–matter interaction, a subsequent heat treatment performed at 400 °C (T
g +20 °C) for 30 min in the glass revealed the crystallization of plasmonic silver nanoparticles. Such plasmonic structure precipitation will not be described in detail in the present paper, as they were investigated earlier by Marquestaut et al. [
13]. These plasmonic structures led to a quenching of the energy transfer between Ag species and Eu
3+ ions, so that the enhancement of the Eu
3+ emission was mostly cancelled after thermal treatment, as shown in
Figure 11 for an excitation at 325 nm. Indeed, the Eu
3+ emission enhancement ratio drops from 2.5- to 5-fold down to 1.2- to 1.8-fold (with an enhancement drop from 5-fold down to 1.2-fold for the highest laser-deposited dose from pattern C6). The precipitation of silver nanoparticles decreases the active population of silver clusters that can be excited and further transfer energy to Eu
3+ ions, thus reducing the energy transfer ability. Additionally, such silver nanoparticles can also contribute to quenching the fluorescence emission of the remaining silver clusters, further reducing the probability of energy transfer to Eu
3+ ions.