Introduction

The choice of the active laser medium is quite essential for the construction of high peak power laser systems. To achieve femtosecond laser pulses with peak pulse energies in the 100 J range broad-band amplification materials are required. Since high power InGaAs laser diodes are commercially available, Yb³⁺-doped media are of increasing interest, because wavelengths of the Yb³⁺ absorption band agree well with the emission wavelengths of these laser diodes. Furthermore, Yb³⁺ doped materials have comparatively long luminescence lifetimes and a low quantum defect [1–3]. For the amplification of laser pulses up to the petawatt level, materials of very high homogeneity, good thermo-mechanical properties and a large optical aperture are needed.

Up to now, the material of choice was Yb³⁺ doped CaF₂. As a crystalline material, it possesses a high thermal conductivity; it offers a suitable amplification bandwidth and a high quantum efficiency. However, the high coefficient of thermal expansion (CTE) might be disadvantageous at high repetition rates, due to thermo-mechanical stresses induced during laser operation. Furthermore, the material is very expensive and its emission spectrum is not as smooth as the spectra of Yb³⁺ doped glasses [4–6]. Hence the latter are an interesting alternative.

The currently used laser glasses for high power applications are predominantly phosphate or fluorophosphate glasses [7, 8]. Although they provide long luminescence lifetimes if doped with Yb³⁺, their CTE are usually as high as 15 to 17·10⁻⁶ K⁻¹ and their thermal conductivity is low [9]. At high thermo-mechanical stress due to the induced temperature gradients, micro cracks are formed, which finally lead to a complete destruction of the laser material [10]. Furthermore these glasses are difficult to produce in large scale. For that reason in the past few years, low thermal expansion glasses have been proposed as laser materials for high peak power amplification [11]. Fused silica as a glass with a CTE as low as 0.5·10⁻⁶ K⁻¹ of course is highly thermal shock resistant, however, the solubility for rare-earth ions is very limited, at least if the rare earth ions are introduced via a melting route. If concentrations higher than 0.5 mol% are incorporated, clustering occurs and the luminescence lifetimes and quantum efficiency decrease strongly [12]. Recently, the luminescence and laser properties of Yb³⁺ doped alumino silicate (AS) glasses were reported [6]. Among these glasses, many systems are well studied with respect to their glass forming abilities and already produced in large scale. Here, lithium alumosilicate as well as magnesium alumosilicate glasses are to be mentioned. Most of these glasses exhibit excellent mechanical and thermo-mechanical properties. Due to their high mechanical strengths, high Young's moduli and their low CTE in particular, glass-fibers for the reinforcement of polymer matrix composites are predominantly produced from alumino silicate glasses [13–15]. Although, most commercial glass-ceramics are also based on alumino silicate glasses, the crystallization tendency is usually low, if no nucleation agents such as TiO₂ or ZrO₂ are added to the glass composition. Furthermore, AS are highly chemically durable and show low water and platinum solubilities, which scarcely dissolve from the ambient air or the crucible material, respectively. However, AS glasses generally have a high solubility for rare earth ions [16]. A drawback of these materials is the required high melting temperatures and the high viscosities even at high temperatures.

In this paper, a multi component Yb³⁺ doped alumino silicate glass in the system Li₂O/MgO/La₂O₃/Al₂O₃/SiO₂ is modified by additions of B₂O₃ and fluoride which should decrease the viscosity of the glass melt and enable lower melting temperatures, higher homogeneity of the glass and a lower degree of contamination with impurities such as platinum. The effect of these additives on the physical properties of the glasses and the luminescence properties of Yb³⁺ are reported.

Experimental procedure

Glasses in the system Li₂O/MgO/La₂O₃/Al₂O₃/B₂O₃/Yb₂O₃/SiO₂ were melted from optic grade raw materials (Fe < 10 ppm, other contaminating metals < 0.5 ppm) SiO₂ (SipurA1, Bremthaler Quarzitwerk, Germany), Li₂CO₃ (Centrum Odczynnikow Chemicznych, Poland) LiF (Chemiewerk Nünchritz, Germany), MgO (Merck, Germany), MgF₂ (Chemiewerk Nünchritz, Germany), Al₂O₃ (PM-5N, Pengda Munich, Germany), H₃BO₃ (Merck, Germany), La₂O₃ (Jenapharm, Germany) and Yb₂O₃ (Han Kyung, South Korea). The doping concentration was kept constant at 6·10²⁰ Yb³⁺/cm³ which corresponds to about 1.3 mol% Yb₂O₃.

Batches of 300 g were melted in covered dispersion hardened platinum crucibles at a temperature of 1500 °C using an inductive furnace under argon atmosphere. The melts were bubbled with argon for 1 to 3 h and afterwards kept at 1560 °C for another 30 min for refining. Then, the melts were cast into steel moulds preheated to 800 °C and transferred into a muffle furnace, preheated to the same temperature. The cooling furnace was then switched off to enable samples to cool down to room temperature (rate: approximately 3 K/min).

Table 1 summarizes the chemical compositions of all prepared glasses. The Li₂O, MgO, La₂O₃ and SiO₂ concentrations were kept constant for most compositions. The Al₂O₃ was replaced by equimolar quantities of B₂O₃ in the range from 0 to 15 mol% B₂O₃. For the two glasses B3F9 and B9F9, Li₂O and MgO were partially replaced by LiF and MgF₂. Sample B3F9 was melted as a 10 kg batch in a semi-industrial scale.

Table 1. Chemical compositions of the studied glasses. All glasses, except B10Yb0 and B12Yb0 are doped with 6·10²⁰ Yb³⁺/cm³

View Table | View all tables in this article

The glass transition temperatures T_g and the CTE were determined in the temperature range from 20 to 1000 °C using a dilatometer (dilatometer DIL 402 PC, NETZSCH Gerätebau GmbH, Germany) and supplying a heating rate of 5 K/min. The thermal properties were further characterized using dynamic scanning calorimetry (DSC PT-1600, Linseis, Germany) at a heating rate of 10 K/min. The low temperature viscosities were determined using a beam bending viscometer (VIS 401, Bähr Thermoanalyse GmbH, Germany) and samples with a geometry of 4 × 4 × 50 mm³; the high temperature viscosities were measured using rotation viscometry (VIS 403, Bähr Thermoanalyse GmbH, Germany) and rotation frequencies of 10 and 150 min⁻¹. The obtained viscosity data were fitted to the Vogel-Fulcher-Tammann equation.

A helium pycnometer (AccuPyc 1330, Micromeritics GmbH, Germany) was used for the determination of the densities. The FTIR spectra were recorded with an IRAFFINITY-1 spectrometer (Shimadzu Corp., Japan) using sheets with a thickness of 1 mm, polished on both sides.

The luminescence lifetimes were measured as described in [3] for the Yb^{3+ 2}F_5/2→²F_7/2 transition at an emission wavelength of approximately 1030 µm.

Results and discussion

Most of the prepared glasses were of very good quality and fully transparent. Exceptions were the samples with relatively high borate content: sample B15 (15 mol% B₂O₃) was white and intransparent and the samples B10 (10 mol% B₂O₃) and B12Yb0 (12 mol% B₂O₃, undoped) had a light whitish haze, while the sample B10Yb0 (10 mol% B₂O₃, undoped) was perfectly transparent. All samples were X-ray amorphous. Figure 1 shows two SEM micrographs of samples B10 (left) and B15 (right). For this, sample B10 was additionally annealed at 800°C for 1h to increase the effect. Mostly spherical particles of 100-300 nm can be seen in both pictures. The bright colour of the particles indicates a higher mean atomic number in these particles, most likely due to accumulation of the Yb³⁺ and/or La³⁺ ions. A similar behaviour was already reported for borosilicates with high rare earth concentrations [17]. Here phase separation in a rare earth rich borate phase and a SiO₂ rich matrix was observed. The same phase separation process is assumed to occur for the borate rich aluminosilicate glasses presented here. A further indication is the sample B10Yb0 without Yb³⁺ doping. For this sample a phase separation is not detected due to the lower overall RE concentration (1.3 mol% La₂O₃ only).

 

Fig. 1 SEM micrographs of samples B10 (left) and B15 (right).

Download Full Size | PDF

Table 2 summarizes some properties of the samples such as glass transition temperature, density, refractive index and coefficient of thermal expansion (CTE). The most obvious differences are found for the glass transition temperatures which were determined by dilatometry. While T_g is 703 °C for the glass without boron, a steady and nearly linear decrease of T_g with increasing B₂O₃ concentration is found. At the maximum B₂O₃ concentration of 15 mol%, T_g is as low as 558 °C. For the samples additionally containing fluoride, the T_g values decrease even to 530 °C. The density values and refractive indices generally decrease with increasing B₂O₃ concentration because of the decrease of the mean atomic weight [18]. The lowest density is found for the undoped samples. The addition of fluoride seems to decrease the refractive indices even more, while an effect on the density is not observed. The CTEs do not vary much and are almost constant at around 5·10⁻⁶ K⁻¹.

Table 2. Physical properties of the studied glasses

View Table | View all tables in this article

In Fig. 2, the viscosities of the glass without B₂O₃ and of four glasses where Al₂O₃ was partially substituted by B₂O₃ are shown. The viscosity for all glasses decreases over decades with decreasing temperature. Furthermore, a steady decrease of the viscosity with increasing B₂O₃ concentration is observed. The glass without B₂O₃ possesses a viscosity of about 5·10¹² dPa·s at 750 °C, while that of the sample with 10 mol% B₂O₃ has a viscosity of less than 10⁹ dPa·s, i.e. more than three orders of magnitude smaller. The glasses, where oxides of the raw materials were partially replaced by fluorides show even lower viscosities. For some samples, the viscosity at 760 °C was even lower than 10⁸ dPa·s, i.e. more than five orders of magnitude smaller than that of the glass without boron and without fluoride. Hence, a drastic effect of both, the replacement of Al₂O₃ by B₂O₃ and the replacement of oxide by fluoride is observed which should decrease the melting temperature notably. No reliable rotation viscometry measurement was possible for the glass B0, since its viscocity was too high. The same problem occured for the sample B5 at temperatures below 1240°C (Fig. 2).

 

Fig. 2 Glass melt viscosities as a function of temperature and sample composition.

Download Full Size | PDF

Figure 3 shows the FTIR absorption spectra of glasses with different B₂O₃ concentrations. Two different absorption bands are observed. One band is at 3,600 cm⁻¹; it increases slightly with increasing B₂O₃ concentration and is caused by weakly bonded OH^- groups, i.e. physically dissolved water in the glass structure [19, 20]. The other band at about 2,700 cm⁻¹ does not occur in the glass without boron and increases steadily with increasing B₂O₃ concentration. However, in this wavenumber range also the absorption of more strongly associated OH^- groups is found [19, 21]. The peak positions of both absorption bands are not affected by the B₂O₃ concentration. Additionally, Fig. 3 includes the two fluoride containing samples: both samples show the two absorption bands as already observed in the glasses without fluoride. However, while the absorption band at 3,600 cm⁻¹ is very weak for all other glasses, it is highest for the sample B3F9. As already mentioned in the experimental section, this sample was melted in a much larger quantity, i.e. in a larger platinum crucible with a much smaller surface/volume ratio. Furthermore, bubbling was performed using the same platinum tube as for the other samples and hence is assumed to be much less efficient. Therefore this glass has the highest water content. For the other fluoride containing sample B9F9, the band at 3,600 cm⁻¹ almost disappears. Here, the substitution of oxides by fluorides results in a drastic decrease of the weakly bonded OH^- groups. Obviously, the weakly bonded OH^- groups are removed efficiently by the fluoride [19, 21]. The narrow absorption bands at around 3,600 cm⁻¹ are due to water in the atmosphere. Also a slight shift of the OH^- absorption band at 3,600 cm⁻¹ to lower wavenumbers can be noted for the fluoride containing samples.

 

Fig. 3 FTIR absorption spectra as a function of the B₂O₃ concentration.

Download Full Size | PDF

Figure 4 shows the absorptivities of the absorption band at 2,700 cm⁻¹ as a function of the B₂O₃ concentration. An almost linear increase of the absorptivity with increasing B₂O₃ concentration is observed. Therefore at first sight it is difficult to decide whether this band is directly related to the B₂O₃ addition or to strongly associated OH^- groups as stated earlier. In pure SiO₂ glass, only one band occurs in the O-H vibration range, at 3,600 cm⁻¹ [19]. By analogy, predominantly that band is also observed in binary alkaline (Li⁺, Na⁺, K⁺) silica glasses of low alkaline concentrations. With increasing alkaline concentration an additional band at about 2,700 cm⁻¹ occurs [19]. When introducing Al₂O₃ to these alkaline silica glasses, the intensity of this band decreases and the absorption band at 3,600 cm⁻¹ becomes predominant again. The band at 2,700 cm⁻¹ disappears if the molar alkaline (i.e. Na₂O) concentration is equal to the molar Al₂O₃ concentration. This effect is explained by the non-occurrence of non-bridging oxygen sites in this glass composition, since the Na⁺ ions balance the negative charge of the [AlO₄]^- tetrahedra. Hence, in glasses without non-bridging oxygen, the OH^- groups are weakly bonded, resulting in the typical band at 3,600 cm⁻¹. In the glasses with higher alkaline concentrations non-bridging oxygen sites occur and the dissolved OH^- groups can form hydrogen bonds with these. These strongly bonded OH^- groups are reported to absorb at 2,700 cm⁻¹ [19].

 

Fig. 4 Absorption coefficient at 2,700 cm⁻¹ as a function of the B₂O₃ concentration.

Download Full Size | PDF

In the glasses, studied in this paper, the intensity of the band at 2,700 cm⁻¹ is negligible at a B₂O₃ concentration of 0%, i.e. an Al₂O₃ concentration of 20 mol%. In this glass, non-bridging oxygen does not occur, because all alumina should occur in fourfold coordination and a sufficient quantity of ionic species occurs to compensate the negative charge of the [AlO₄]^- tetrahedra. This is in analogy to reports in the literature (see above). If, however, Al₂O₃ is successively replaced by B₂O₃, the glass structure changes. If both boron and aluminum occurred in fourfold coordination, the concentration of the network modifiers would be high enough to compensate the negative charge of the [BO₄]^- and [AlO₄]^- tetrahedra. However, in the case of boron, not all boron would occur as [BO₄]^- tetrahedra and hence, the replacement of Al₂O₃ against B₂O₃ in the glass composition gives rise to the occurrence of non-bridging oxygen [22]. In principle, the second harmonic of the fundamental B-O stretching vibration at around 1350 cm⁻¹ may contribute to the band observed at 2,700 cm⁻¹. This could be proved by shifting the fundamental B-O vibrations due to changes in the glass structure. However, this was not the aim of this work. The main contribution to the band at 2,700 cm⁻¹ should be the OH-vibration of strongly bonded OH^- groups. A further hint to this is the absorption of the sample with high fluoride concentration (B9F9): although the boron concentration is the same as in sample B9, the peak at 2,700 cm⁻¹ is notably decreased. Obviously, fluoride can partially remove the strongly bonded OH^- groups. Hence, the correlation of the B₂O₃ concentration with the intensity of the band at 2,700 cm⁻¹ is due to the increase in concentration of non-bridging oxygen sites and hence due to the attributed OH-vibration.

Figure 5 shows the luminescence lifetimes of Yb³⁺ as a function of the OH^- absorption at 3,600 cm⁻¹. As already reported earlier, the water concentration has a strong influence on the lifetimes and quantum efficiency of Yb³⁺ especially at high doping concentrations as for the samples presented here [21]. Concerning the B₂O₃ concentration series, the substitution of Al₂O₃ against B₂O₃ has only a minor effect; all lifetimes are close to 0.84 ms. The sample with the highest OH^- concentration clearly has the shortest lifetime (B3F9). Now, the question arises why dissolved water which gives rise to the band at 3,600 cm⁻¹ results in a strong decrease in the luminescence lifetime, while water which causes the band at 2,700 cm⁻¹ does not. This can be explained by the so called energy gap law [23] which is given by:

 

Fig. 5 Luminescence lifetime as a function of the OH^- absorption at 3,600 cm⁻¹.

Download Full Size | PDF

Here k_MPR is the quench rate due to multi-phonon relaxation, ΔE is the energy gap in the energy level scheme of the luminescent ion (around 10,000 cm⁻¹ for Yb³⁺), ħω is the phonon energy, T is the temperature, k is the Boltzmann constant and α and β are constants. With this formula, the quenching rate due to multi-phonon relaxation can be estimated with an accuracy of about one order of magnitude [23]. Even for a phonon energy as high as 2,700 cm⁻¹ the multi-phonon quenching rate is around 5 orders of magnitude lower than the luminescence emission rate of Yb³⁺ (around 1 ms⁻¹), while for a phonon energy of 3,700 cm⁻¹ (onset of the 3,600 cm⁻¹ band), both rates are in the same order. That means, that the higher phononic energy attributed to the vibrations at 3,600 cm⁻¹ quenches the luminescence efficiently, while the quenching due to vibrations at around 2,700 cm⁻¹ is negligible.

For the sample B3F9, the fluoride addition could not counteract the much less efficient bubbling process as explained earlier. But generally, the partial replacement of oxides against fluorides (sample B9F9) results in a significant increase in the luminescence lifetime. This, on the one hand, is an effect of the more efficient water removal in this sample but also, most likely to an even larger extent, an effect of a change in the local surrounding of the Yb³⁺ ions due to the fluoride addition [3, 11, 24]. Despite a relatively high B₂O₃ concentration of 9 mol%, a luminescence lifetime of almost 1 ms at a doping concentration of 6·10²⁰ Yb³⁺/cm³ could be achieved for sample B9F9. This is the same value that was also achieved for boron free lithium alumino silicate glasses [21]. However the lithium alumino silicate glasses had a lower fluoride concentration.

Conclusion

A series of glasses in the system Li₂O/MgO/La₂O₃/Al₂O₃/B₂O₃/Yb₂O₃/SiO₂ was prepared and doped with 6·10²⁰ Yb³⁺/cm³ which corresponds to about 1.3 mol% Yb₂O₃ in the glass composition. In this series Al₂O₃ was partially replaced by B₂O₃ up to a concentration of 15 mol%. Furthermore, LiF and MgF₂ were partially substituted for Li₂O and MgO. A single sample was prepared as a 10 kg batch in semi-industrial scale. All samples were bubbled with dry argon gas to ensure low OH^- concentrations. The B₂O₃ addition results in a drastic decrease of the glass transition temperature T_g. Only a slight decrease in the density and the refractive index with increasing B₂O₃ concentration was observed. The water content and the coefficient of thermal expansion remain almost unaffected. Fluoride additions decrease the OH^- concentration even further and slightly decrease the refractive index. Despite its relatively high phonon energy, the luminescence lifetime of Yb³⁺ is hardly affected by the B₂O₃ addition. However, the Yb³⁺ lifetimes are very sensitive to the OH^- and fluoride concentrations. Interestingly, the IR vibration band of strongly bonded OH^- groups at around 2,700 cm⁻¹ shows no observable effect on the Yb³⁺ lifetime; the decisive factor is the concentration of weakly bonded OH^- groups. From these experiments, it can be concluded that additions of up to 9 mol% of B₂O₃ can significantly decrease the melting temperature and enable a better homogenization of the glasses and help to avoid dissolution of impurities such as platinum. For that reason, a better optical quality of the glasses can be achieved without losing the good thermo-mechanical and luminescence properties of these glasses.