Middle infrared fiber lasers at 3 μm region have potential application in eye-safe laser radar, monitoring atmospheric pollutants and high-resolution spectroscopy of low-pressure gases1,2,3. In the recent years, the use of erbium-doped bulk solid state lasers for high-quality cutting or ablation of biological tissue caused by the high absorption of 3 μm radiation in water has been demonstrated4,5. The well-known 2.7 μm emission of Er3+ due to the 4I11/2→4I13/2 transition can be obtained pumped at 800 nm or 980 nm6 and the lack of commercial high-power pump diodes that emit at 1100 nm or 1300 nm limits the application of holmium and dysprosium doped materials.

Fluoride glasses are potential candidates for IR optical fibers because of their middle infrared region transparency and low theoretical loss <0.01 dB/Km in the wavelength range of 2 μm to 5 μm, which is 10 times less than the silica fibers7. Most investigations focused on the fluoride glass host at the pioneer time and up to now only fluoride glass fiber lasers have been reported in the region of 3 μm. Over the past decades, the 10-W-level diode pumped 2.7 μm Er-doped ZBLAN fiber lasers was obtained8,9,10. However, the poor chemical durability, low mechanical strength, along with the low glass transition temperature (Tg), have limited the application of ZBLAN glass and scaling up the output power further is fundamentally difficult11. Thus, exploring effective host matrix becomes a challenge to the researchers. Oxide glass is one candidate with easy preparation, good physical and mechanical performance and good chemical durability. There has been attempt pouring into tellurite glass12,13, bismuthate glass14 and germinate glass15 at present. On the other hand, new high thermally and chemically durable fluoride glasses should be considered and investigated based on the significant advantages of high solubility for rear earth ions, low maximum phonon energy as well as high mid-infrared transparency. By contrast, AlF3-based glasses appear to be less susceptible to those problems that limit the further application of fluorozironate glasses7,16. H.Yanagita has reported the laser performances of Er3+-doped fluorozircoaluminate17. However, no investigation has focused on the 2.7 μm emission properties of Er3+-doped fluoroaluminate glasses so far.

The present work aims to investigate the various compositions of high thermally and chemically durable AlF3- based glasses and compares their thermal and mechanical properties with the well-known ZBLAN glass. In addition, 2.7 μm emission of Er3+ in these AlF3-based fluoride glasses is investigated for future application in mid infrared lasers, Judd-Ofelt intensity parameters, spontaneous transition probability and stimulated emission cross section were also calculated and discussed.

The investigated glasses have the following molar compositions: 35AlF3-15YF3-20CaF2-10BaF2-10SrF2-10MgF2-1ErF3 (AYF1), 40AlF3-15YF3-15CaF2-10BaF2-10SrF2-10MgF2-1ErF3 (AYF2), 37AlF3-15YF3-15CaF2-13BaF2-10SrF2-10MgF2-1ErF3 (AYF3), 37AlF3-15YF3-15CaF2-13BaF2-10SrF2-10MgF2-2ErF3 (AYF3-2), 37AlF3-15YF3-15CaF2-13BaF2-10SrF2-10MgF2-4ErF3 (AYF3-4) and 53ZrF4-20BaF2-4LaF3-3AlF3-20NaF-1ErF3 (ZBLAN).

The samples were prepared using high-purity AlF3, YF3, CaF2, BaF2, SrF2, MgF2, ZrF4, LaF3, NaF and ErF3 powder. Well-mixed 25 g batches of the samples were placed in platinum crucibles and melted at about 1000°C for 30 min. Then the melts were poured onto a preheated copper mold and annealed in a furnace around the glass transition temperature. The annealed samples were fabricated and polished to the size of 20 mm×15 mm×1 mm for the optical property measurements.

The characteristic temperatures (temperature of glass transition Tg and temperature of onset crystallization peak Tx) of samples were determined using a NetzschSTA449/C differential scanning calorimetry at a heating rate of 10 K/min. Phases of the samples were characterized by X-ray diffraction (XRD, Rigaku RINT-2000) with Cu Kα radiation. The density and refractive indices of the samples were measured by the Archimedes method using distilled water as an immersion liquid and the prism minimum deviation method respectively. The concentration of the cationic species in water was measured by Inductively Coupled Plasma (ICP). Furthermore, the absorption spectra were recorded with a Perkin-Elmer Lambda 900 UV/VIS/NIR spectrophotometer in the range of 100 nm–1900 nm and the emission spectra were measured with a Triax 320 type spectrometer (Jobin-Yvon Co., France). All the measurements were carried out at room temperature.

The transparent and homogeneous samples with the different compositions mentioned before have been obtained. Figure 1 shows the XRD (X-ray diffraction) spectra of the samples, all the samples have two dispersion peaks at approximately 25° and 47° which are the characteristic peaks of the fluoride glasses18. It is demonstrated that the prepared AlF3-based fluoride glasses have good glass-forming ability and have not shown devitrification tendencies when being formed.

Figure 1
figure 1

XRD spectra of the present samples.

Full size image

Table 1 presents the DSC results for AYF1, AYF2, AYF3 and ZBLAN glasses in this study. The glass criterion, ΔT = Tx-Tg (temperature of glass transition Tg, temperature of onset of crystallization Tx) was introduced by Dietzel19,20 and is often regarded as an important parameter for evaluating the glass forming ability. ΔT has been frequently used as a rough criterion to measure glass thermal stability; a large ΔT means strong inhibition of nucleation and crystallization. Hruby21 developed the Hr criterion, Hr = ΔT/(Tm-Tp), which includes the characteristic temperatures, where Tm is the melting temperature of the glass and Tp is the temperature of peak of crystallization. The glass formation factor of the materials is given by the parameter kgl = (Tx- Tg)/(Tm-Tg)20. Compared with ΔT, it is more suitable to estimate the glass thermal stability. The larger the kgl, the better forming ability the glass will have. The glass forming ability can be estimated by the given characteristic temperatures. The values of these characteristic temperatures among these three AlF3-based glasses changed slightly. When AlF3 was replaced by CaF2 or MgF2 partly, the values decreased slightly in some degree. However, the thermal stability of these AlF3-based glasses is better than that of the ZBLAN sample. In addition, Tg is also an important factor for laser glasses, a high one such as 420+°C of the AlF3-based glasses compared with other various fluorozirconate glasses (270°C–350°C)22,23 provides good thermal stability to resist thermal damage at high pumping intensities.

Table 1 Characteristic temperatures (Tg, Tx, Tp, Tm) and ΔT, Hr, kgl of present glasses
Full size table

The chemical durability of the sample was measured as follows: (3.1 shows that the three AlF3-based glasses have similar thermal stability. The results of AYF2 and ZBLAN glasses are listed because they have the same regular pattern.). First, the weighted sample (W1) was placed into the distilled water. Second, the sample was kept in a thermostatic water bath at 98°C for 5 h and then the samples were cooled and dried in a dying box at 70°C for 1 h. Finally, the dry sample was weighed again (W2). The chemical durability of AYF2 and ZBLAN glasses was evaluated by the values of 24 and . The ΔW%(111.5 mg/g)and ΔW’% (488.6 mg/cm3) of ZBLAN were 30 times larger than those of the AYF2 sample which were 3.3 mg/g and12.6 mg/cm3, respectively.

The surface layer of the ZBLAN corroded and became white after placing into distilled water for 24 hours at room temperature and the corrosion layer tended to peel off. The layer in contact with the water remained the same by naked-eye observation in the case of AlF3-based glass. The cationic species Zr, Ba, La, Al and Na (ZBLAN) and Al, Ba, Ca, Y, Sr and Mg (AYF2) were measured through ICP (Inductively Coupled Plasma) and the results are shown in Table 2. The concentration of the cationic of ZBLAN in water was much larger than that of AYF2. Therefore, NaF,AlF3,CaF2, MgF2, SrF2 and BaF2 dissolve at faster rates than ZrF4, LaF3 and YF3, as was also reported by T. Iqbal et al7. These results prove that the chemical stability of AYF2 is much better than that of ZBLAN glass.

Table 2 Concentration of cationic ions of based glass in water
Full size table

Figure 2(a) shows the transmittance spectra of the 1.0 mm thick AYF2 and ZBLAN samples. The IR cut-off wavelength for ZBLAN is ~7 μm. The cut-off occurs instead at ~6 μm when Zr is largely replaced by light Al atoms in the AYF glass matrix, beacuse the Al-F fundamental vibrations shifted toward slightly high frequencies. The transmittance of AYF and ZBLAN samples reached as high as 93% and 90%, respectively. The ~10% loss contained the Fresnel reflections dispersion and absorption of the glass. An OH− absorption peak (~2.9 μm) exists in AYF2 sample, but the transmittance still reached above 83% at 2.9 μm. More study should be carried out further to deduce the concentration of the OH− in the fluoroaluminate glasses in the future. Figures 2 (b) and (c) exhibits the IR transmission spectra of AYF2 and ZBLAN glasses before and after the water treatment, respectively. Change in the absorption band at 2.9 μm was because of the fundamental OH− absorption. The transmittance of the ZBLAN sample dropped dramatically and was even completely light-tight in the region of 3 μm. The absorption around 6.2 μm because of the fundamental of fluoride glass also became serious. Meanwhile, the phenomenon of light-tight in AYF2 glass was much weaker compared with that of ZBLAN. The poor chemical stability against water attack of ZBLAN glass can degrade its otherwise outstanding optical properties.

Figure 2
figure 2

(a): Transmittance spectra of the 1.0 mm thick AYF2 and ZBLAN samples. (b): The IR transmission spectra of ZBLAN glass before and after water treatment. (c): The IR transmission spectra of AYF2 glass before and after water treatment.

Full size image

Figure 3 indicates the absorption spectra of the samples at room temperature in the wavelength region of 300 nm–1900 nm. The shape and peak positions of each transition for the Er3+-doped AlF3-based glass are very similar to those in ZBLAN and other Er3+-doped glasses25,26. The absorption bands corresponding to the transitions starting from the 4I15/2 ground state to the higher levels 4I13/2,4I11/2,4I9/2,4F9/2,4S3/2,2H11/2,4F7/2 were labeled. The absorption peak at 980 nm because of the 4I15/2 →4I11/2 transition indicates the samples can be pumped by 980 nm laser efficiently.

Figure 3
figure 3

Absorption spectra of the samples at room temperature in the wavelength region of 300 nm–1900 nm.

Full size image

Important spectroscopic and laser parameters of rare earth doped glasses have been commonly analyzed by many researchers using the Judd–Ofelt theory27,28, hence, only the results will be presented. Table 3 shows the J-O parameters Ωt of Er3+ in various glasses. Good agreement was found between the calculated and experimental values, the room mean-square error deviation of intensity parameters is ×10−6, indicting the validity of the Judd-Ofelt theory for predicting the spectral intensities of Er3+ and the reliable calculations. Notably, that the ascending order of the parameters is Ω2<Ω4<Ω6 for the presented glasses and the trend appeared similar to that of other rare earth doped fluoride glasses27,29. The Ω2 is strongly dependent on the local environments of rare earth ion sites. The values of Ω2 in fluoride glasses were smaller than those of heavy metal oxide glasses because the O2− possesses higher covalency than F− and the fluoroaluminate glasses have lower covalency and higher asymmetry among the two main system of fluoride glasses.

Table 3 J-O parameters Ωt of Er3+ in various glasses
Full size table

Table 4 shows the calculated predicted spontaneous transition probability (A), radiaive lifetime (τrad) and branching ratio (β) of certain optical transitions for Er3+ - doped present glasses. The predicted spontaneous emission probabilities for Er3+:4I11/2→4I13/2 transition in fluoroaluminate glass was a little smaller than that of fluorozirconate glass owing to the lower refractive index, but owns higher branching ratio, which is beneficial to the 2.7 μm emission.

Table 4 Calculated predicted spontaneous transition probability (A), radiaive lifetime (τrad) and branching ratio (β) of certain optical transitions for Er3+ - doped present glasses
Full size table

Under 980 nm diode laser excitation, the 4I11/2→4I13/2 fluorescence around 2.7 μm was obviously observed for the samples, as shown in Fig. 4. For Er3+-doped fluoride glasses with different compositions, the emission peak occurs at similar wavelength. Although the AlF3-based glasses possess lower intensity compared with the ZBLAN glass, their calculated emission cross sections at 2710 nm were 8.9, 9.4 and 8.8(×10−21 cm−2) similar to the value of ZBLAN (9.5×10−21 cm−2). The lower intensity attributes to the OH− absorption in AlF3 –based glass. Enhanced 2.7 μm emission will be observed in our following work with the diminishment of the OH− content and the adjustment of the composition. Moreover the intensity significantly increased when the concentration of Er3+ ions were added to the AYF3-based-glass. It may be inferred that minimized concentration quenching occured in the 4 mol% Er3+- doped fluoroaluminate glass. Higher concentration Er3+-doped glasses can be studied further and it is reasonable to believe these high chemically durable glasses may be considered to be a hopeful host for a 2.7 μm microchip laser and other optical laser application.

Figure 4
figure 4

The 2.7 emission spectra of present glasses under 980 nm diode laser excitation.

Full size image

The emission cross section was calculated according to the Fuchtbauer-Ladenburg theory30,31 and the absorption cross section (σabs) can be derived from the calculated σem using the McCumber equation30. On the basis of the σabs and σem, the wavelength dependence of the net gain32 can be calculated as a function of population inversion for the upper laser level to determine qualitatively the gain property as follows:

where population inversion P is assigned to the concentration ratio of Er3+ in the 4I11/2 and 4I13/2 levels. As shown in Figure 5, the gain coefficients with various P values ranging from 0 to 1 were calculated for 4I11/2—4I13/2 transition of the 1 mol% Er3+-doped AYF2 glass (other AlF3-based glasses and ZBLAN have similar spectra). Evidently, the positive gain was obtained when P>0.4, similar to the case in the ZBLAN glass33 indicting that a low pumping threshold was achieved for the Er3+:4I11/2→4I13/2 laser operation.

Figure 5
figure 5

Gain coefficients with the various P values ranging from 0 to 1 for 4I11/2—4I13/2 transition of the 1 mol% Er3+-doped AYF2 glass.

Full size image

AlF3-based glasses (AlF3-YF3-CaF2-BaF2-SrF2-MgF2) with enhanced thermal and chemical stability were synthesized and compared with the well-known ZBLAN. All the prepared samples exhibited good glass-forming ability and did not show devitrification tendencies when being formed. Higher Tg (420+°C) of the AlF3-based glasses compared with other various fluorozirconate glasses (270°C to 350°C) provides the glass good thermal stability to resist thermal damage at high pumping intensities. In addition, the characteristic temperatures of AlF3-based glasses are larger than those of ZBLAN glass. The weight loss of ZBLAN in water was 30 times larger than that of AYF2 sample and the concentration of cationic of ZBLAN in water was much larger than that of AYF2, which proves that the chemical stability of AYF2 was much better than that of ZBLAN glass. The emission properties and relevant parameters of the present samples were investigated and discussed. The fluoroaluminate glasses possessed large branching ratio (20%) along with emission cross section (9.4×10−21 cm−2) of the Er3+:4I11/2→4I13/2 transition. Meanwhile, enhanced 2.7 μm emission in highly Er3+-doped AYF glass was obtained. These results suggest that this kind of AlF3-based fluoride glass has potential application in compact 2.7 μm lasers.