Several series of preform and fiber samples of Yb2O3-Al2O3-SiO2 glass with finely-graded composition steps were prepared by MCVD and solution doping under well-defined conditions using both oxidizing and reducing atmospheres during preform collapse. Their optical properties, including absorption and emission behavior in the NIR/VIS/UV region, have been characterized and correlated with the detailed glass composition. The results present an overview of the property spectrum, which should contribute to the further development of laser fibers and the discussion and control of disturbing effects such as photodarkening.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
The performance of rare-earth (RE) doped high power silica fiber lasers has been dramatically increased with output powers in the multi-kW range, high efficiency, reliability, and beam quality [1–5]. This progress is due to new design concepts, such as non-symmetrical double clads and large mode area (LMA) cores , and to the microstructuring of the fibers using “air clads”, inner “holey” clads, or solid “multifilament” cores [7–9]. Extreme power densities beyond the kW level and complicated fiber structures, however, place high demands on fiber material properties and preparation technology concerning both efficient laser operation and high power stability. The increased fiber attenuation induced by pump radiation, called photodarkening (PD), turned out to be a critical factor in high power laser action [10–12]. These pump-induced losses can impair the long-term power stability and efficiency of fiber lasers and amplifiers at high power loading. The thermal load caused by the absorption of PD color centers  at the pump and laser wavelength was identified as an additional issue, which can lead to refractive index changes in the laser core, mode instabilities, and degradation of the beam quality in nominal single mode working laser or amplifier systems during high power operation .
The ytterbium ion (Yb3+) is the favored lasing ion because of its long lifetime in the excited state, its simple energy-level scheme, and its small quantum defect between the pump and laser wavelength . Quartz glass and high silica glasses, made by applying gas phase deposition processes, are extremely important as host glasses because of their high glass stability, high purity, and low optical losses. However, the option of incorporating these RE ions in silica is limited, and it is difficult to achieve a high doping level of such ions. By adding co-dopants such as aluminium (Al) and/or phosphorus (P), as well as germanium (Ge), the solubility of RE ions in silica can be improved and the RE content increased without phase separation and crystallization. These co-dopants also influence further properties of glass, such as e.g. the refractive index of the fiber core, the absorption and emission properties of Yb ions, several aspects of the preparation technology of the laser fiber, and the PD kinetics on a large scale both favorably and unfavorably [16–19].
The optimization of the core composition and preparation conditions for different fiber laser applications is still a task for research and technology. Al is a very important solubilizer for RE ions in silica , and the most commercially available Yb fibers are co-doped with Al. The ambivalent role of this co-dopant was discussed . On the one hand, atomic precursors are inserted that promote the formation of color centers; on the other hand, a high Al content should be favored to prevent Yb clustering and to reduce PD effects . Therefore, in this paper we report on developments aimed at increasing the efficiency and power stability of Yb doped laser glass fibers with a composition of Yb2O3-Al2O3-SiO2. The absorption and emission properties of preforms made by applying the conventional modified chemical vapor deposition (MCVD) process in combination with solution doping technology [23,24] and of the drawn fibers were investigated for a wide range of Yb and Al concentrations. Moreover, the influence of the atmosphere during the collapsing step was studied. The latter is especially important because preparation according to non-MCVD routes (flame hydrolysis deposition [25,26] and powder sintering technology (REPUSIL) ) is mostly connected to a strongly reducing gas atmosphere. Moreover, it was assumed that PD is connected to a reduction in the Yb ions to a bivalent state . In detail, we investigated the absorption, fluorescent behavior, and fluorescent lifetime of the Yb3+ ion at a wavelength of around 1 µm, the UV absorption and important features of the spectrum such as the NIR excited visible absorption, and the UV excited visible emission of the material.
We report on the preparation and optical investigation of samples, in which the Yb and Al concentration was changed in a wide range, together with well-defined modifications of the preparation conditions, leading to correlations in composition and preparation conditions as the basis for a subsequent characterization of their laser properties. This work includes some results already reported in [19,22,29–32]. The specific aim is to obtain an overall view of the entire implementable concentration range in the Yb2O3-Al2O3-SiO2 system, unlike many reports in which only singular samples are investigated. Even though it is not the actual object of the investigations, supplementary samples of Yb2O3-SiO2 (without Al co-doping) were made and measured for comparison.
2.1 Preparation of the specimen
The fiber preform samples of 8 series (AO … DO, AR … DR) were prepared by MCVD in combination with solution doping technique according to the following route. At first, a flocculent deposit was formed from a gaseous mixture of SiCl4/O2 (and a small amount of POCl3 for AO and AR series) on the inner surface of the quartz glass carrier tube with an outer and inner diameter of 14 mm and 11.5 mm, respectively. This deposit was pre-sintered in a pure oxygen atmosphere at a temperature of 1335°C (P doped) and 1520°C (un-doped) in order to generate a defined relative density of 20% compared with a fully densified layer. This layer was impregnated with an aqueous solution of YbCl3 (0…2.6·10−4 mol·cm−3) and AlCl3 (0…5·10−4 mol·cm−3), dried in subsequent preparation steps at room temperature and about 1000°C, again in a pure oxygen atmosphere, and at 1400°C under Cl2/O2 atmosphere (11% Cl2). After a sintering step at 2000°C with pure oxygen, a glassy layer with a thickness of about 12 µm was built up on the inner tube surface. In order to complete the fiber preform, the tube was collapsed by two burner passes at about 2250°C to a solid rod (preform) where the doped layers have formed the preform core with a diameter of about 0.7 mm (outer preform diameter: about 8 mm). The collapsing of samples of the AO, BO, CO, and DO series was accomplished in a Cl2/O2 atmosphere (11% Cl2). The samples of the AR, BR, CR, and DR series were prepared under the same conditions, but during collapsing a pure helium (He) atmosphere was maintained in the tube’s interior. As already investigated and reported in , He treatment leads to a certain release of oxygen from the deposited layer with the result that the core glass is (slightly) transferred to a chemically reduced state, with distinct consequences for its optical properties. (As usual in the MCVD process, all steps have been performed with a moving oxygen-hydrogen torch and a flowing gas atmosphere inside the tube, disregarding the last collapsing step).
The following variations in the core composition were carried out:
- AO, AR series: Nearly constant Al content / variation in the Yb content
- BO, BR series: Nearly constant Yb content / variation in the Al content
- CO, CR series: Nearly constant Yb/Al ratio / variation in the Yb and Al content
- DO, DR: Without Al
The samples of the AO and AR series are additionally doped with a low P content (about 0.5 mol% P2O5). Samples without Al co-doping (DO and DR) can be prepared only at low Yb concentrations (up to about 0.2 mol% Yb2O3). With higher Yb concentrations a strong opacity appears, accompanied with high preform and fiber scattering loss. The effect marks the limit of solubility of Yb in the silica matrix.
The preform samples were characterized non-destructively by a beam deflection method  for the refractive index profile. The radial concentration profiles of Yb, Al, and P were determined by wavelength dispersive electron probe microanalysis (WD-EPMA) on thin polished preform slices (spatial resolution of about 3 µm, detection limit of 0.01 mol% Al2O3, Yb2O3, and P2O5). Fibers with a core and outer diameter of about 10 µm and 125 µm, respectively, were directly drawn from the preforms at a drawing speed of 10 m·min−1 at a drawing tension of about 0.2 N. The refractive index change (Δn) of the preform and fiber core is determined by the Yb content and the concentration of the co-dopants Al and P. An additivity rule is valid with respect to the molar composition with constant increments acc. to Eq. (1) .
In Table 1, the Yb (cYb2O3) and Al (cAl2O3) content of the prepared preform samples is shown, together with the measured fiber background loss β in the NIR at a wavelength of 1200 nm and the NA of the preform samples. As a consequence of the very low dopant concentrations of a few samples (CO1, CR1, CO2, CR2, DO, DR) and hence the very low NA of the corresponding fibers, the measured fiber attenuation is determined by strong bending losses, which exceed material-related losses. In all other cases, the bending losses are negligible, and the measured losses can be considered to really depend on the material properties.
2.2 Optical characterization
For the UV/VIS/NIR absorption measurements on preforms, polished slices with a thickness d between 0.2 and 2 mm (depending on the expected absorption effect) were illuminated by the collimated beam of a combined halogen/deuterium lamp. Immediately at the back of the sample, the transmitted light was collected by a UV-suitable silica fiber probe with a NA of about 0.2 and a diameter of 0.2 mm, respectively, in order to measure the absorption in the central homogeneous area of the doped core. The light collected in this manner is guided to an Instrument Systems Spectro 320D spectrometer (190 nm to 2150 nm). The exponential absorption coefficients α were determined by comparison of the signal intensity behind the preform core, I, and the pure silica preform cladding, I0, as34]. The samples were pumped at 976 nm (CW or pulsed).
3.1 Preform and fiber absorption
All prepared Yb doped preform samples show the typical Yb3+ absorption in the wavelength range between 800 nm and 1100 nm with a broad peak at about 920 nm and a small higher peak at 976 nm. One example is shown in the insert of Fig. 1. It is well known that both the spectral shape and the absolute absorption cross sections of Yb3+ show a certain distinct dependence on the kind and concentration of the co-doping elements . Here however, the Yb concentration change and the modification of the preparation process with increasing Al co-doping do not result in a remarkable change in the absorption coefficient, which is about 11 cm−1 /mol% Yb2O3 in the maximum at 976 nm and 2.4 cm−1 /mol% Yb2O3 in the maximum at 920 nm, respectively (see Fig. 1). The absorption increases apparently linearly with the Yb content in the investigated concentration region.
The co-doping of P in the AO and AR series is too small to exert a measurable influence on the absorption. The exact analysis of the data, however, reveals a certain influence of Al on absorption, which is small but significantly outside the adjustment limits. Increasing Al content gives rise to a slight decrease in the Yb absorption coefficient, as shown in Table 2 together with the respective root-mean-square-deviation (RMSD) of the fitting procedure. The spectral shape, however, shows no change at all with varying Al content.
Only the samples without Al (DO and DR) do not fit well in this picture. Apart from small spectral changes (shift in the absorption maxima by a few nanometers to shorter wavelengths, compare also ), the absorption coefficients are lowered against the general tendency of the increase that comes with decreasing Al concentration. Their values are with 0.4 cm−1 (915 nm)/1.9 cm−1 (975 nm) for DO and 0.3 cm−1 (915 nm)/1.3 cm−1 (975 nm) for DR, respectively, still within the RMSD limits because of the low concentrations.
In contrast to NIR absorption, important changes are observed in the UV/VIS region. Figures 2 and 3 show the typical spectra of preform samples of the AO, BO, CO series and sample DO prepared under oxidizing conditions in the UV region at a wavelength as low as 195 nm. The UV edge increases and apparently shifts to longer wavelengths with increasing Yb concentration (samples of the AO series and sample DO). All Yb doped preform samples show an Yb related absorption band in the wavelength region 220 nm … 240 nm with a tail up to 300 nm, in samples of the BO, CO series and sample DO as a weakly emphasized shoulder. In contrast, the spectrum of sample AO1 (without Yb) does not show absorption in this region, which demonstrates that the absorption is due to the presence of Yb ions in the glass network. This absorption is caused by Yb3+ transitions to a charge transfer (CT) state in the glass network, the properties of which and their relevance for PD have been extensively discussed in [35,36].
The shape of the spectra of the AO, BO, and CO series was analyzed by fitting it with a Gaussian distribution using a standard procedure which made it possible to identify three bands, one of which is located out of our measuring range at a wavelength below 190 nm. The location of the band maxima (λmax) and the absorption in the maximum (αmax) together with the respective root-mean-square-deviation (RMSD) of the fitting procedure are summarized in Table 3.
One example of spectral deconvolution is shown in Fig. 4(a). The band position shows a certain dependence both on the Al and Yb concentration. Increasing Al shifts each band to shorter wavelengths, up to about 10 nm for the highest Al concentrations. The influence of Yb is different for band 1 (shift to shorter wavelengths) compared with band 2 (shift to longer wavelengths) and can likewise reach values of up to 10 nm for extreme concentrations. Obviously, band 1 and band 2 move closer together with increasing Yb concentration. These relatively complex relations are responsible for the appearance of a more-or-less pronounced shoulder or a partial maximum in the spectra shown in Figs. 2 and 3.
The absorption is apparently linearly increased with the Yb concentration, as shown in Fig. 4(b) for band 1. The detailed analysis reveals that the maximum absorption of the bands is additionally influenced by the Al content. In the case of band 1 and band 2, the absorption is reduced by increasing Al content. The effect is not strong but distinctly beyond the error limit of the deconvolution procedure. This behavior is in general similar to the NIR-range absorption, and the increments are comparable in magnitude (compare Table 2). The data of band 3, which are shown in Table 3 in brackets, are not reliable because the band is outside of our measuring range. Note that the completely Al-free sample DO is the only one that does not fit well with the values in Table 3. The wavelength location of band 1 and band 2 in particular with 222 nm and 200 nm, respectively, is remarkably lower than that predicted by Table 3. The maximum absorption with 26 cm−1 and 10 cm−1, respectively, is likewise smaller than that predicted by Table 3 and only for band 1 still within the statistical error limits. This deviation is reminiscent of the IR absorption.
Figures 5 and 6 show the typical spectra of preform samples of the AR, BR, CR series and sample DR prepared under reducing conditions in the UV region as low as 195 nm. The comparison between preform samples made by He and O2 collapsing shows that via He collapsing, an additional absorption band system appears between wavelengths of 200 nm and 400 nm, which superimposes the Yb3+ absorption edge in the UV region. The same absorption bands were observed under stronger reducing treatment  and are related to a (small) part of Yb in a reduced state, shortly designed as Yb2+ [37,38]. This assignment is based on a comparison with investigations of Yb2+ in several crystals [39–41] and on a report of Yb2+ formation in quartz glass samples co-doped with Al , which is in good agreement with our observations. The tail of this absorption extends in the visible region and is responsible for the yellow color of the preform/fiber core.
The preform sample AR1 (without Yb) is also shown in Fig. 5(a). The typical Yb2+ features are absent, but He collapsing leads to an absorption increase in the UV with a peak at about 240 nm caused by oxygen deficiency centers (ODC) . In sample DR (without Al), the formation of Yb2+ by He collapsing is greatly reduced compared to Al co-doped samples, and the ODC features are relatively more pronounced, see Fig. 5(b).
Through the absorption difference between reduced and oxidized Yb doped samples at comparable concentrations, the absorption bands of Yb2+ can be exposed. For example, the spectra of the samples of the CR series are depicted in Fig. 7(a). (To correct a small concentration difference between oxidized and reduced samples, the spectra were linearly adjusted (compare Table 1)). A detailed investigation of these spectra shows that the absorption spectrum can be composed of 7 Gaussian components (one of which is located out of our measuring range below 190 nm). One example is shown in Fig. 7(b). This is in qualitative agreement with investigations on CaF2 , in which several components were also found and interpreted as different 4f→5d intra-configurational transitions.
The most conspicuous absorption bands in these spectra are located at 215 nm and 326 nm. According to a detailed analysis of Yb2+ formation through reduction with hydrogen and detection of the hydroxyl groups formed, the relationship between Yb2+ concentration and absorption at 326 nm was able to be determined , leading to a value of the absorption coefficient of 5·102 cm−1/ mol% YbO. If we apply this relationship to our measurement results, we can conclude that only a small proportion of Yb is reduced in our current samples. In Fig. 8, the absorption difference between the reduced and oxidized state is displayed for all samples. From Fig. 8, it follows that the concentration of Yb2+ reaches at most 0.03 mol% YbO in the case of more than 1 mol% Yb2O3 and is about 0.01 mol% YbO in the majority of cases. In contrast to the absorption behavior of Yb3+ determined at 920 nm and 976 nm (see Fig. 1) and in the UV region (see Fig. 4(b)), there is by no means a linear relationship between Yb2+ and the total amount of Yb, cYb2O3. Rather, it is a more complex relationship. Moreover, there is obviously a significant promoting effect of Al on Yb2+ formation. Detailed investigations revealed that the Yb2+ concentration can be expressed in good approximation by a proportionality to the square root of the product of the Yb and Al concentrations acc. to Eq. (4).8(a). From an evaluation of the log10-log10 fit, shown in Fig. 8(b), the statistic accuracy of the exponent in Eq. (4) can be determined as 0.50 ± 0.03 for both wavelengths.
In Table 1 and Fig. 9 it is shown that the background loss β at 1200 nm of the drawn fibers in samples of the AO, BO, and CO series is moderately increased with the total Yb concentration, cYb2O3, in a linear manner. Obviously, the background absorption at 1200 nm is caused by the long wavelength tail of the absorption of Yb. The relationship between absorption at 1200 nm and at 976 nm, i.e. in the maximum, is roughly 10−5. Similarly, as in the absorption maximum shown in Fig. 1, the influence of Al remains small. (The samples CO1, CO2, DO are not considered here for the reasons explained above).
The fiber samples of the AR, BR, and CR series show additional background losses at 1200 nm (see Table 1), probably caused by the long wavelength tail of the Yb2+ ions. In this case, we have a rough correlation between the excess loss at 1200 nm and the Yb2+ related absorption at 326 nm (see Fig. 10).
In  the absorption and emission properties of Yb/Al doped preforms and fibers with a nearly constant composition (∼ 0.4 mol% Yb2O3/∼4 mol% Al2O3) as a function of the atmosphere (O2/ He/ CO/ H2) during preform collapsing were investigated. There, it was shown that the UV band system of Yb2+ extends its tail to the visible and NIR region (beyond the Yb3+ absorption) and that there is a fairly good correlation between the absorption coefficients in the UV/VIS at 326 nm and the excess absorption (in comparison to the sample prepared under oxidizing conditions) of a fiber in the NIR at 1200 nm according to
3.2 Yb3+ near infrared fluorescence
With excitation at 976 nm, all investigated Yb doped fiber samples show the typical emission spectrum between 900 nm and 1150 nm for the transition between the 2F5/2 multiplet manifold and the 2F7/2 ground state. An example is given in the insert of Fig. 11. The shape of the spectrum is hardly affected by the Yb concentration and modification during preparation. In contrast to the invariable spectral behavior, the fluorescence lifetime of about (805 ± 15) µs at low Yb concentrations tends to decrease at higher concentrations, beginning at about 0.8 mol% Yb2O3 for oxidized samples (AO10/12) and about 0.36 mol% Yb2O3 for reduced samples (AR7-12, BR1/3, CR5-8), respectively (see Fig. 11). For lifetimes < 700 µs, decay kinetics deviates from purely exponential behavior. In contrast to samples with Al co-doping at comparable concentrations, the lifetime of the reduced sample without Al (sample DR) is decreased.
The fluorescence intensities (measured at 1075 nm) at low Yb concentrations are likewise identical for both oxidized and reduced samples and increase linearly with the Yb concentration, cYb2O3, up to the concentration points of about 0.6 mol% Yb2O3 for the reduced and 0.8 mol% Yb2O3 for the oxidized samples, respectively, where the lifetimes start to decrease. With further increasing concentrations the intensities pass through a maximum and decrease strongly (see Fig. 12). This remarkable deviation from the linearity at higher concentrations is in contrast to the absorption behavior pointing to strong concentration quenching.
In detail, we again find a small dependence on the Al concentration, as expected from the absorption behavior (see above) but is barely noticeable in the general behavior.
3.3 NIR excited visible fluorescence
A typical feature of Yb doped materials is the visible emission around 500 nm, if the Yb3+ is excited in the absorption maximum at 976 nm. This emission is responsible for the intense green shining of a lasing fiber. It is explained as cooperative luminescence, in which a pair of excited Yb3+ ions emits one photon with the double energy of the absorption transition [43,44]. Examples of spectra for selected samples of the AO and AR series and sample DO and DR, respectively, are shown in Fig. 13. The observed peaks at 488 nm, 500 nm, and about 510 nm are caused by Yb. Traces of Tm impurities in the material (in consequence of a Tm content of about 1 ppm in the used raw material YbCl3) give rise to an additional band at 475 nm. (The existence of Tm could be proved by further absorption spectral bands). With pumping at 976 nm, Tm3+ is excited via a three-step energy transfer from the excited Yb3+. This explains the strong increase in the Tm band with increasing concentration and its sensitive reaction to the reduction state of Yb. Subtracting the Tm related effect, the spectrum of the visible fluorescence does not show a remarkable change in the concentration or reduction state.
The concentration dependence on the visible fluorescence intensity in the peak maximum at 500 nm is indicated in Fig. 14. As can be seen, the intensity increases close to the square of the total Yb concentration, cYb2O3, as expected for a cooperative process. In detail, we again find the marginal dependence on the Al concentration, as expected from the absorption and NIR fluorescence. Only the sample DO does not fit very well into the picture of the majority of samples, both in contrast to the NIR fluorescence and the VIS emission of the sample DR! The reduced samples in general agree with the oxidized sample at low concentration, but the quenching starts earlier, closely resembling the behavior of the Yb3+ NIR fluorescence. Concentration quenching is clearly recognizable in the visible emission as well. Note, however, that the visible fluorescence is about two orders of magnitude lower than the NIR fluorescence.
There is a correlation between the fluorescence intensities at 500 nm and 1075 nm, as shown in Fig. 15 in the form of
This evidently demonstrates that the cooperative fluorescence can by no means be held responsible for the impairment of the NIR fluorescence concerning time and intensity, i.e. concentration quenching . Really, we have to consider the cooperative fluorescence a very small side effect, which however sensitively reacts to the concentration of the undisturbed excited Yb3+ ions.
3.4 UV excited fluorescence
Through UV excitation with a deuterium lamp, different emission bands can be observed. The Yb3+-typical NIR emission that is observable on all samples is not discussed here. The UV/VIS emission spectra of the oxidized and reduced samples are exemplified by the AO and AR series in Fig. 16, supplemented by the samples DO and DR without Al co-doping. The remarkable feature of the reduced Yb/Al doped samples in Fig. 16(b) is a visible fluorescence with a large band around 520 nm, as already reported earlier . This fluorescence is related to the formation of Yb2+ ions (compare [37,38,41]). The sample AR1 (without Yb) shows a peak at 400 nm, obviously in consequence of the oxygen deficiency absorption at 240 nm (see Fig. 5(a)). This peak also seems to exist in the Yb doped samples but is increasingly superimposed by the strong emission around 520 nm. In the sample DR (Yb doped, but without Al), the formation of Yb2+ is greatly reduced and the ODC features are relatively more pronounced, as can be expected from the absorption behavior. Note that in the case of UV excited fluorescence, the intensity of the emission is strongly influenced by a pump intensity decrease with increasing Yb concentration because of the very high absorption coefficient of the samples in the region around 200 nm (see Figs. 5 and 6), in which the lamp spectrum reaches its intensity maximum. Therefore, there is no real basis for discussions concerning the emission intensity change with the concentration, in contrast to the NIR and cooperative fluorescence behavior.
In the oxidized samples, the 400 nm peak is absent both in the un-doped and doped samples, respectively. This is also the case with the 520 nm peak in the doped samples. Only at the highest Yb concentration of 1.7 mol% Yb2O3 (sample AO12) does the 520 nm emission peak seem to be formed, however, nearly two orders of magnitude lower than under reduced preparation conditions.
Besides the visible fluorescence, some effects in the UV region are observed as shown in Fig. 16(a). A small peak at about 350 nm appears in the un-doped sample AO1 and decreases with increasing Yb content. Even if it is also present in the reduced samples, it can not be resolved because of the dominance of the 400 nm emission. Its reason is unclear and difficult to discuss because of the strong absorption and reabsorption effects in this wavelength region. Pump light scattering effects cannot be excluded.
The peaks around 300 nm in the Al co-doped samples at higher Yb concentrations (> 0.8 mol% Yb2O3, samples AO10 and AO12) and around 260 nm in the Yb doped sample DO (without Al), respectively, also seem to exist in the reduced samples to a certain extent. These emission bands could be caused by CT transitions to the 2F7/2 ground state of the Yb3+ion, in qualitative agreement with investigations of YAG:Yb (15%) under UV excitation [46–48]. There, the luminescence spectra exhibit two broad emission bands with a maximum around 333 nm and 500 nm. The 333 nm emission band, which corresponds to the transition from the CT state to the 2F7/2 ground state, is significantly more intense than the 500 nm band corresponding to the transition from the CT state to 2F5/2 exited state of the Yb3+ion, respectively. Thus, the low band at 520 nm in sample AO12 could be assigned to the CT transition to the 2F5/2 excited state of the Yb3+ion.
Both high-doped samples AO10 and AO12 with noticeable UV emission exhibit concentration quenching in the IR fluorescence (see Fig. 14(a)). Interestingly, the sample DO (with low Yb content, comparable with AO5, but without Al) shows noticeable UV emission as well, comparable with the highest doped sample AO12. The shift of the emission band to smaller wavelengths is obviously assigned to the corresponding shift in the absorption band.
A notable feature of the IR absorption in the Yb2O3-Al2O3-SiO2 system is its high constancy of the shape and size of the Yb-related absorption bands independent of the strong composition variation between about 0.5 mol% Al2O3 and 6.5 mol% Al2O3. The maximum positions remain practically unchanged, and the Yb-related absorption coefficient shows only a slight but systematic decrease with a magnitude of about 4.5% (920 nm) and 2% (976 nm), respectively, per mol% Al2O3. In contrast, the complete omission of the co-dopant or a change in the type of the co-dopant (e.g, replacement of Al by P or Ge ) results in much greater changes in the absorptive properties. This applies both to the oxidized and reduced samples, i.e. the reduction is without effect on the basic glass structure of Yb. In a first approach, such behavior could be explained by the formation of a stable “solvation shell” built up by the co-dopant Al according to . The preferred formation of more stable linkages Yb-O-Al compared with Yb-O-Si and Yb-O-Yb would protect the Yb ions from a statistical distribution in the glass network. In general, this assumption could duly explain the specific role of Al as an advantageous solubilizer for RE ions in silica glass. Note, however, that several recent investigations of the local molecular environment of Yb ions by electron paramagnetic resonance point to the formation of a “solvation shell” for P rather than Al co-doping [49,50,51], i.e. the formation of Yb-O-Si and Yb-O-Yb linkages is not suppressed by Al but only by P co-doping. The existence of Yb-O-Yb linkages is interpreted as “clustering”, as an incomplete dissolution of Yb in silica glass under the influence of Al.
Similar statements also apply to Yb absorption in the UV, although here a variation of the Yb and Al content is accompanied by larger spectral variations. Besides a change in the Yb-related absorption coefficient, again in the order of a few percent per mol% Al2O3, the band maxima are clearly shifted depending on both Al and Yb.
Today, the most accepted hypothesis as to the origin of the UV bands is charge transfer (CT) transitions for the longer UV wavelengths and 4f14 to 4f135d transition for the shorter UV wavelengths [52,53]. The role of these transitions in the process of PD was discussed several times [35,36]. The shift of the band 1 maximum to smaller wavelengths with increasing Yb content was observed in . Overall, the observed effects are still significantly lower than when changing the type of co-dopant .
In , an interesting interpretation is given for the origin of the UV bands. There, the long-wavelength peak around 240 nm is assigned to Yb-O-Al links and the peak around 215 nm to Yb-O-Si links. In samples without Al, only the second peak was observed, and in Al co-dopant samples both peaks appear. This interpretation, however, seems to be in contrast to our detailed investigation on a large number of samples with Al varying in a wide concentration range. In our analysis, both absorption bands show an intensity decrease with increasing Al content, which is however very low. On the basis of the interpretation in , we would rather expect an increase in the 240-nm-band and a decrease in the 210 nm band with increasing Al (with a total change of Al from 0.5 mol% Al2O3 to 6.5 mol% Al2O3, this is by a factor of 13). Moreover, the influence of Al should be remarkably larger. Furthermore, our investigation on samples without Al points to a shift of both bands rather than a disappearance of the second band. The observations concerning samples without Al is discussed below in detail.
The emission after UV excitation is in general very low in the oxidized samples. Only in very few samples has a weak emission band been observed at 260…300 nm. Even if the interpretation as CT luminescence is true (see above), the question remains as to why this emission is quenched in most cases and only appears both at very high Yb doping (with Al co-doping) and at low Yb doping (without Al co-doping).
At first glance, the reduced samples show prominent spectral phenomena in the UV and VIS region. This applies both to the absorption and UV excited emission. The analysis demonstrated above shows that these conspicuous effects are caused by Yb2+ ions, which are present in the glass network in a relatively small proportion. The majority of the Yb ions remains in the trivalent state with properties that are the same as those of the oxidized samples.
In contrast to the absorption properties of Yb3+, which are only weakly influenced by a variation in the Al concentration, the apparent absorption of Yb2+ (see Fig. 8) is strongly influenced by the Al content. This, however, is not a direct influence of Al on the optical properties but an influence on the concentration of the formed Yb2+ ions. In samples without Al, the reductive treatment does not form Yb2+ but rather reduces to a certain extent the silica matrix, recognizable by the formation of oxygen deficiency centers. The interplay of ODC and Yb2+ development depending on the Al content is also clearly reflected in the UV-excited visible emission bands (see Fig. 16(b)). It was mentioned earlier that that the presence of aluminum is necessary for the formation of divalent Yb in quartz glass . Beyond that, here we were able to show that the concentration of Yb2+ in the course of a reductive treatment exhibits a systematically continual increase with Al concentration (see Eq. (4)). Obviously, Al is directly involved into the reduction process of Yb. A simple explanation of the mechanism is currently not possible. Note, however, that Yb2+ is often discussed as a participating species in the process of the PD of Yb doped laser fibers [35,50,54]. The relationship in Eq. (4) could therefore become of importance for the control of PD in specially designed laser fibers. Note further that PD is directly inhibited by a reductive treatment in the preparation process of the laser fiber .
The basic attenuation of doped fibers in a wavelength region without apparent absorption peaks at 1200 nm is systematically influenced primarily by the absorption tail of Yb3+ (peak absorption 920 nm / 976 nm) and Yb2+ (peak absorption 326 nm). Certain deviations at high Yb concentrations, in the range of concentration quenching, indicate additional contribution on the fiber attenuation. This is obviously related to specific glass defects in the quenching region, the type of which is currently not definable.
The close similarity of the IR absorption spectra of all Al doped samples and the constancy of the Yb absorption coefficient mean that the optical properties of the majority of Yb ions in the ground state are practically not influenced by concentration effects in the investigated concentration range and also not by the applied reductive treatment with He atmosphere. These results correspond well with investigations of absorption properties of preform samples prepared by gas phase doping of Yb and Al in the MCVD process . In contrast, the properties of Yb ions in the excited state vary depending on the concentration and reduction state. Assuming that for low concentrations we have the properties of a typical singly excited Yb3+ ion, at a certain concentration point the mean lifetime is decreased and – as shown by the NIR fluorescence intensities – the stationary number of the excited ions is likewise decreased. Obviously, we have here the initiation of a non-radiative process, which further depopulates the excited level. Similar effects are observed in general for REs and referred to as “concentration quenching”, e.g. in the well-investigated case of the erbium (Er) ion , where the quenching starts at remarkably lower concentrations. Because of the simple level scheme of Yb3+, the reason for concentration quenching is not yet understood, although it has been repeatedly observed [57,58]. The inset point of concentration quenching is distinctly influenced by the reduction state. Note, however, that the number of Yb2+ ions formed by reduction is small, much smaller than the number of the quenched Yb3+ ions; therefore, a direct influence can be excluded. As already discussed, the observed fluorescence around 500 nm (the cooperative fluorescence) cannot likewise be directly responsible for concentration quenching because of its quantitative insignificance. Visible fluorescence is already produced at very low Yb concentrations. In addition, the strong relationship between visible and NIR fluorescence intensity (shown by Eq. (6)) is valid into the region of quenching (i.e., the visible emission in the product of non-quenched ions and the inset point of concentration quenching is not accompanied by a conspicuous effect in the course of cooperative emission). The intensity of the cooperative emission is often presumed as an indicator for the so-called “clustering” of RE ions (see e.g. ), and it really depends distinctly on the kind of co-dopant (e.g. Al, P, or Ge [49,18]). On the other hand, concentration quenching is likewise attributed to the conjunction of RE ion clustering through RE-O-RE bonds (see e.g. ). Even if we cannot explain in detail the reason for both “concentration quenching” and “clustering” in Yb2O3-Al2O3-SiO2 glass, our investigations show that there is no direct connection between them.
One important general result of the comprehensive investigation presented is that the optical properties of the Yb2O3-SiO2 (without Al) system cannot be considered a simple smooth extrapolation of the Yb2O3-Al2O3-SiO2 system to an Al content of zero. This results from the absorption features both in IR and UV concerning band position, spectral shape, and absorption coefficients. In all cases, the properties of the system without Al show a certain but significant deviation from the extrapolation values of the ternary system to zero Al (in general, the position is shifted to shorter wavelengths and the absorption coefficient is reduced). Similar abrupt changes also appear in the emission properties in NIR, VIS, and UV (compare the results shown in Figs. 14, 15, and 16). This behavior of the optical properties shows a parallelism to transport properties, demonstrated with the diffusion in the Yb2O3-Al2O3-SiO2 system . There, a smooth dependence of the Yb diffusion coefficients on the concentration of Al was observed in the concentration range between 3 mol% Al2O3 and 0.5 mol% Al2O3, but an unregular change in the concentration range below 0.5 mol% Al2O3 down to 0 mol% Al2O3. The measured values of the diffusion coefficients and the activation energy of diffusion for 0 mol% Al2O3 differ significantly from the extrapolated ones. Note that all samples in the present investigation exhibit an Al concentration of about or more than 0.5 mol% Al2O3, apart from the samples completely without Al. Undoubtedly, both optical and transport properties can be attributed to the same atomic peculiarities in the low Al range.
Optical properties of Yb2O3-Al2O3-SiO2 glass preforms made via MCVD in combination with solution doping technique and of the drawn fibers were investigated. Furthermore, the influence of reducing atmosphere during the collapsing step was studied. A variety of samples of well-defined composition and finely graded concentrations of Yb and Al was prepared covering the range between 0 mol% Yb2O3 and 1.7 mol% Yb2O3 and 0 mol% Al2O3 and 6.5 mol% Al2O3. The concentration range implemented is only limited by the scope of the preparation method and its principal applicability as a laser fiber. The investigations include absorption and emission measurements from UV to NIR with specific emphasis on the quantification of emission intensities. The results present an overview of the property spectrum in detailed dependency on sample composition and collapse atmosphere. The UV and NIR absorption was analyzed depending on Yb and Al concerning absorption wavelengths and intensities. The influence of the dopants on the basic fiber loss was clarified. Relationships concerning Yb2+ ion formation were elucidated. Regularities and relationships concerning the NIR and VIS fluorescence were found and discussed including such effects as concentration quenching and clustering. The results provide comprehensive and, in many details, new and important information, which should contribute to the further development of laser fibers and the discussion and control of disturbing effects such as photodarkening.
Bundesministerium für Bildung und Forschung (13N11972 (TEHFAI), 13N13653 (TEHFA II)).
1. Y. Jeong, J. K. Sahu, D. N. Payne, and J. Nilsson, “Ytterbium-doped large-core fiber laser with 1.36 kW continuous-wave output power,” Opt. Express 12(25), 6088–6092 (2004). [CrossRef]
2. D. J. Richardson, J. Nilsson, and W. A. Clarkson, “High power fiber lasers: current status and future perspectives,” J. Opt. Soc. Am. B 27(11), B63–B92 (2010). [CrossRef]
3. A. Popp, A. Voss, T. Graf, S. Unger, J. Kirchhof, and H. Bartelt, “Thin-disk laser-pumping of ytterbium-doped fiber laser,” Laser Phys. Lett. 8, 887–894 (2011). [CrossRef]
4. C. Jauregui, J. Limpert, and A. Tünnermann, “High-power fibre lasers,” Nat. Photonics 7(11), 861–867 (2013). [CrossRef]
5. M. Zervas and C. Codemard, “High Power Fiber Lasers: A Review,” IEEE J. Sel. Top. Quantum Electron. 20(5), 219–241 (2014). [CrossRef]
6. A. Tünnermann, H. Zellmer, W. Schöne, A. Giesen, and K. Contag, “New concepts for diode-pumped solid-state lasers,” in High-power diode lasers, (Ed. R. Diehl, Springer-VerlagBerlin Heidelberg, 2000).
7. A. Tünnermann, S. Höfer, A. Liem, J. Limpert, M. Reich, F. Röser, T. Schreiber, and H. Zellmer, “Large mode area fibers for high power laser operation based on solid- and air-microstructured cores,” Proc. SPIE 5709, 301–309 (2005). [CrossRef]
8. C. Wirth, O. Schmidt, A. Kliner, T. Schreiber, R. Eberhardt, and A. Tünnermann, “High-power tandem pumped fiber amplifier with an output power of 2.9 kW,” Opt. Lett. 36(16), 3061–3063 (2011). [CrossRef]
9. G. Canat, S. Jetschke, S. Unger, L. Lombard, P. Bourdon, J. Kirchhof, V. Jolivet, A. Dolfi, and O. Vasseur, “Multifilament-core fibers for high energy pulse amplification at 1.5 µm with excellent beam quality,” Opt. Lett. 33(22), 2701–2703 (2008). [CrossRef]
10. J. J. Koponen, M. J. Söderlund, H. J. Hoffman, and S. K. T. Tammela, “Measuring photodarkening from single-mode ytterbium doped silica fibers,” Opt. Express 14(24), 11539–11544 (2006). [CrossRef]
11. I. Manek-Hönninger, J. Boullet, T. Cardinal, F. Guillen, S. Ermeneux, M. Podgorski, R. Bello Doua, and F. Salin, “Photodarkening and photobleaching of an ytterbium-doped silica double-clad LMA fiber,” Opt. Express 15(4), 1606–1611 (2007). [CrossRef]
12. K. E. Mattsson, “Photo darkening of rare earth doped silica,” Opt. Express 19(21), 19797–19812 (2011). [CrossRef]
13. J. J. Montiel, I. Ponsoda, M. Söderlund, J. Koplow, J. Koponen, and S. Honkanen, “Photodarkening-induced increase of fiber temperature,” Appl. Opt. 49(22), 4139–4143 (2010). [CrossRef]
14. C. Jauregui, H.-J. Otto, F. Stutzki, J. Limpert, and A. Tünnermann, “Simplified modelling the mode instability threshold of high power fiber amplifiers in the presence of photodarkening,” Opt. Express 23(16), 20203–20218 (2015). [CrossRef]
15. P. D. Dragic, M. Cavillon, and J. Ballato, “Materials for optical fiber lasers: A review,” Appl. Phys. Rev. 5(4), 041301 (2018). [CrossRef]
16. S. Unger, A. Schwuchow, S. Jetschke, V. Reichel, A. Scheffel, and J. Kirchhof, “Optical properties of Yb- doped laser fibers in dependence on co-dopants and preparation conditions,” Proc. SPIE 6890, 689016 (2008). [CrossRef]
17. S. Unger, A. Schwuchow, J. Dellith, and J. Kirchhof, “Co-doped materials for high power fiber lasers: diffusion behaviour and optical properties,” Proc. SPIE 6469, 646913 (2007). [CrossRef]
18. J. Kirchhof, S. Unger, S. Jetschke, A. Schwuchow, M. Leich, and V. Reichel, “Yb doped silica based laser fibers: Correlation of photodarkening kinetics and related optical properties with the glass composition,” Proc. SPIE 7195, 71950S (2009). [CrossRef]
19. J. Kirchhof, S. Unger, A. Schwuchow, S. Jetschke, and B. Knappe, “Dopant interactions in high-power laser fibers,” Proc. SPIE 5723, 261–272 (2005). [CrossRef]
20. K. Arai, H. Namikawa, K. Kumata, and T. Honda, “Aluminum or phosphorus co-doping effects on the fluorescence and structural properties of neodymium-doped silica glass,” J. Appl. Phys. 59(10), 3430–3436 (1986). [CrossRef]
21. T. Arai, K. Ichii, S. Tanigawa, and M. Fujimaki, “Gamma radiation-induced photodarkening in ytterbium-doped silica glasses,” Proc. SPIE 7914, 79140K (2011). [CrossRef]
22. S. Jetschke, S. Unger, A. Schwuchow, M. Leich, and J. Kirchhof, “Efficient Yb laser fibers with low photodarkening by optimization of the core composition,” Opt. Express 16(20), 15540–15545 (2008). [CrossRef]
23. J. E. Townsend, S. B. Poole, and D. N. Payne, “Solution-doping technique for fabrication of rare-earth- doped optical fibres,” Electron. Lett. 23(7), 329–331 (1987). [CrossRef]
24. J. Kirchhof, S. Unger, and A. Schwuchow, “Fiber lasers: materials, structures and technologies,” Proc. SPIE 4957, 1–12 (2003). [CrossRef]
25. S. Tammela, M. Söderlund, J. Koponen, V. Philippov, and P. Stenius, “The potential of direct nanoparticle deposition for the next generation of optical fibers,” Proc. SPIE 6116, 61160G (2006). [CrossRef]
26. J. Wang, S. Gray, D. T. Walton, M. Li, X. Chen, A. Liu, and L. A. Zenteno, “Advanced vapor-doping, all- glass double-clad fibers,” Proc. SPIE 6890, 689006 (2008). [CrossRef]
27. A. Langner, G. Schötz, M. Such, T. Kayser, V. Reichel, S. Grimm, J. Kirchhof, V. Krause, and G. Rehmann, “A new material for high power laser fibers,” Proc. SPIE 6873, 687311 (2008). [CrossRef]
28. J. Jasapara, M. Andrejco, D. DiGiovanni, and R. Windeler, “Effect of heat and H2 gas on the photodarkening of Yb3+ fibers,” Conf. on Laser and Electro-Optics, Proc. CTuQ5 (2006).
29. J. Kirchhof, S. Unger, A. Schwuchow, S. Grimm, and V. Reichel, “Materials for high-power fiber lasers,” J. Non-Cryst. Solids 352(23-25), 2399–2403 (2006). [CrossRef]
30. J. Kirchhof, S. Unger, A. Schwuchow, S. Jetschke, V. Reichel, M. Leich, and A. Scheffel, “The influence of Yb2+ ions on optical properties and power stability of ytterbium doped laser fibers,” Proc. SPIE 7598, 75980B (2010). [CrossRef]
31. S. Jetschke, S. Unger, M. Leich, and J. Kirchhof, “Photodarkening kinetics as a function of Yb concentration and the role of Al codoping,” Appl. Opt. 51(32), 7758–7764 (2012). [CrossRef]
32. S. Jetschke, S. Unger, A. Schwuchow, M. Leich, V. Reichel, and J. Kirchhof, “Photodarkening in Yb-doped silica fibers: influence of the atmosphere during preform collapsing,” Proc. SPIE 6873, 68731G (2008). [CrossRef]
33. H. R. Müller and U. Röpke, “Preform index profiling with high spatial resolution,” Phys. Status Solidi A 66, 199–205 (1981). [CrossRef]
34. A. Schwuchow, S. Unger, S. Jetschke, and J. Kirchhof, “Advanced attenuation and fluorescence measurement methods in the investigation of photodarkening and related properties of ytterbium-doped fibers,” Appl. Opt. 53(7), 1466–1473 (2014). [CrossRef]
35. M. Engholm and L. Norin, “Preventing photodarkening in ytterbium-doped high power fiber lasers; correlation to the UV-transparency of the core glass,” Opt. Express 16(2), 1260–1268 (2008). [CrossRef]
36. K. K. Bobkov, A. A. Rybaltovsky, V. V. Velmiskin, M. E. Likhachev, M. M. Bubnov, E. M. Dianov, A. A. Umnikov, A. N. Guryanov, N. N. Vechkanov, and I. A. Shestakova, “Charge-transfer state excitation as the main mechanism of the photodarkening process in ytterbium-doped aluminosilicate fibres,” Quantum Electron. 44(12), 1129–1135 (2014). [CrossRef]
37. S. Lizzo, E. P. Klein Nagelvoort, R. Erens, A. Meijerink, and G. Blasse, “On the quenching of the Yb2+ lumiescence in different host lattices,” J. Phys. Chem. Solids 58(6), 963–968 (1997). [CrossRef]
38. S. Lizzo, A. Meijerink, G. J. Dirksen, and G. Blasse, “On the luminescence of divalent ytterbium in KMgF3 and NaMgF3,” J. Phys. Chem. Solids 56(7), 959–964 (1995). [CrossRef]
39. E. Loh, “Ultraviolet-absorption spectra of europium and ytterbium in alkaline earth fluorides,” Phys. Rev. 184(2), 348–352 (1969). [CrossRef]
40. T. Tsuboi, H. Witzke, and D. S. McClure, “The 4f14→4f13 5d transition of Yb2+ ion in NaCl crystals,” J. Lumin. 24-25, 305–308 (1981). [CrossRef]
41. S. M. Kaczmarek, T. Tsuboi, M. Ito, G. Boulon, and G. Leniec, “Optical study of Yb3+/Yb2+ conversion in CaF2 crystals,” J. Phys.: Condens. Matter 17, 3771–3786 (2005).
42. I. V. Kovaleva, V. P. Kolobkov, and G. P. Starostina, “About some peculiarities of the luminescence of the ions Eu2+ an Yb2+ in quartz glass,” Fiz. Khim. Stekla 12, 222–229 (1986).
43. E. Nakazawa and S. Shionoya, “Cooperative luminescence in YbPO4,” Phys. Rev. Lett. 25(25), 1710–1712 (1970). [CrossRef]
44. Y. G. Choi, Y. B. Shin, H. S. Seo, and K. H. Kim, “Spectral evolution of cooperative luminescence in an Yb3+doped silica optical fiber,” Chem. Phys. Lett. 364(1-2), 200–205 (2002). [CrossRef]
45. A. V. Kiryanov, Y. O. Barmenkov, I. L. Martinez, A. S. Kurkov, and E. M. Dianov, “Cooperative luminescence and absorption in ytterbium-doped silica fiber and the fiber nonlinear transmission coefficient at λ=980 nm with regard to the ytterbium ion-pairs effect,” Opt. Express 14(9), 3981–3992 (2006). [CrossRef]
46. L. van Pieterson, M. Heeroma, E. de Heer, and A. Meijerink, “Charge transfer luminescence of Yb3+,” J,” Luminescence 91(3-4), 177–193 (2000). [CrossRef]
47. N. Guerassimova, N. Garnier, C. Dujardin, A. G. Petrosyan, and C. Pedrini, “X-ray excited charge transfer luminescence of ytterbium-containing aluminium garnets,” Chem. Phys. Lett. 339(3-4), 197–202 (2001). [CrossRef]
48. E. Nakazawa, “Charge-transfer type luminescence of Yb3+ ions in LuPO4 and YPO4,” Chem. Phys. Lett. 56(1), 161–163 (1978). [CrossRef]
49. T. Deschamps, N. Ollier, H. Vezin, and C. Gonnet, “Clusters dissolution of Yb3+ in codoped SiO2-Al2O3-P2O5 glass fiber and its relevance to photodarkening,” J. Chem. Phys. 136(1), 014503 (2012). [CrossRef]
50. T. Deschamps, H. Vezin, C. Gonnet, and N. Ollier, “Evidence of AlOHC responsible for the radiation-induced darkening in Yb doped fiber,” Opt. Express 21(7), 8382–8392 (2013). [CrossRef]
51. C. Shao, F. Xie, F. Wang, M. Guo, R. Zhu, H. Zhang, M. Wang, S. Wang, C. Yu, and L. Hu, “UV absorption bands and its relevance to local structures of ytterbium ions in Yb3+ / Al3+ / P5+ -doped silica glasses,” J. Non-Cryst. Solids 512, 53–59 (2019). [CrossRef]
52. M. Engholm and L. Norin, “The role of charge transfer processes for the induced optical losses in ytterbium doped fiber lasers,” Proc. SPIE 7195, 71950T (2009). [CrossRef]
53. A. A. Rybaltovsky, S. S. Aleshkina, M. E. Likhachev, M. M. Bubnov, A. A. Umnikov, M. V. Yashkov, A. N. Gur’yanov, and E. M. Dianov, “Luminescence and photoinduced absorption in ytterbium-doped optical fibres,” Quantum Electron. 41(12), 1073–1079 (2011). [CrossRef]
54. S. Rydberg and M. Engholm, “Experimental evidence for the formation of divalent ytterbium in the photodarkening process of Yb-doped fiber lasers,” Opt. Express 21(6), 6681–6688 (2013). [CrossRef]
55. F. Lindner, C. Aichele, A. Schwuchow, M. Leich, A. Scheffel, and S. Unger, “Optical properties of Yb- doped fibers prepared by gas phase doping,” Proc. SPIE 8982, 89820R (2014). [CrossRef]
56. P. C. Becker, N. A. Olsson, and J. R. Simpson, “Erbium Doped Fiber Amplifiers, Fundamentals and Technology,” Academic Press, San Diego (1999).
57. R. Paschotta, J. Nilsson, P. R. Barber, J. E. Caplen, A. C. Tropper, and D. C. Hanna, “Lifetime quenching in Yb-doped fibres,” Opt. Commun. 136(5-6), 375–378 (1997). [CrossRef]
58. Z. Burshtein, Y. Kalisky, S. Z. Levy, P. Le Boulanger, and S. Rotman, “Impurity local phonon nonradiative quenching of Yb3+ fluorescence in ytterbium-doped silicate glasses,” IEEE J. Quantum Electron. 36(8), 1000–1007 (2000). [CrossRef]
59. A. Monteil, S. Chaussedent, G. Alombert-Goget, N. Gaumer, J. Obriot, S. J. L. Ribeiro, Y. Messaddeq, A. Chiasera, and M. Ferrari, “Clustering of rare earth in glasses, aluminum effect: experiments and modeling,” J. Non-Cryst. Solids 348, 44–50 (2004). [CrossRef]
60. S. Unger, J. Dellith, A. Scheffel, and J. Kirchhof, “Diffusion in Yb2O3-Al2O3-SiO2 glass,” Phys. Chem. Glasses: Eur. J. Glass Sci. Technol. B 52, 41–46 (2011).