1. Introduction

2~5μm mid-infrared lasers are located in the three atmospheric transmission windows and cover a great number of important molecular characteristic spectral lines; therefore they have extensive application prospects in the fields of remote sensing, range finding, environmental monitoring, bio-engineering, medical treatment, and etc [1–3].

One efficient and feasible method of generating the 2~5μm mid-infrared lasers is pumping the rare-earth ions in proper hosts by using a near-infrared laser diode (LD). This method promotes a good prospect of all-solid-state mid-infrared laser possessing many advantages such as high power output, good stability, low cost in production and operation, and etc. However, because of the small gaps between the involved and intermediate energy levels of the rare-earth ions, non-radiative quenching of the mid-infrared emissions is substantial. Low phonon-energy materials are required to act as laser working medium.

In the past several years, chalcogenide, fluoride glasses or crystals, as well as chloride crystals have received considerable interests for their low phonon-energy. The longest laser wavelength of 7.24μm-produced in rare-earth ions doped solid-state laser working at room temperature was obtained in the Pr³⁺-doped LaC1₃ crystal [4]. But the crystals are restricted to poor chemical durability. For the glass medium, the only continuous excitation at room temperature presently is the 3.5μm laser produced by Er³⁺-doped ZBLAN fiber [5], and longer wavelength mid-infrared lasers are hard to be generated. The Ho³⁺-doped ZBLAN fiber laser can generate 3.95μm laser, but must be operated at low temperature to avoid non-radiative quenching [6]. Finding other more suitable glass materials as laser working medium which can generate 2~5μm waveband mid-infrared laser at room temperature is the main emphasis of our research. Chalcogenide and chalcohalide glasses presently seem to be the best candidate materials due to their unique features.

Compared with other glass hosts, Chalcogenide and chalcohalide glasses exhibit two distinctive features which are the higher refractive indices and lower phonon energies. The high refractive indices of greater than 2.1 result in large stimulated emission cross sections and the typically low phonon energies of 250~450cm⁻¹ result in low non-radiative decay rates.

Many works in previous reports have proved the probability of mid-infrared light emitted in some well-established chalcogenide glasses such as Ge_16.5As_18.5Ga_0.5Se_64.5 [7], Ge₃₀As₁₀S₆₀ [8], Ge₂₅Ga₅S₇₀ [8] and 70Ga₂S₃·30La₂S₃ [9], but little effort was put on the compositional optimization since else sulfides or halides introduced in glasses can largely effect their physical and optical properties as well as their micro-structures.

In this paper, we selected Ge-Ga-S-CdI₂ glass as the host system. The spectral properties of Dy³⁺-doped and Dy³⁺/Tm³⁺-codoped glasses excited at 808nm are presented. Judd-Ofelt strength parameters Ω_t (t = 2, 4, 6), the transition probabilities, exited state lifetimes, branching ratios, and emission cross-sections are calculated from absorption and emission measurements by using the Judd-Ofelt analysis. Effects of CdI₂ content and Tm³⁺ level on the intensities of the 2.9μm (Dy³⁺: ⁶H_13/2→⁶H_15/2) and 4.3μm (Dy³⁺: ⁶H_11/2→⁶H_13/2) mid-infrared emissions are investigated and discussed. The glasses in Ge-Ga-S-CdI₂ system were chosen because of their wide transparency in the visible region and good thermal stability [10]. With addition of the heavy metal iodide CdI₂, the density and refractive index increase almost linearly, therefore larger stimulated emission cross section, lower non-radiative decay rate and higher radiative quantum efficiency in mid-infrared waveband are expected.

Our work was aimed at the elucidation of the glass composition’s influence on the spectroscopic properties in this novel chalcohalide system and the discovery of a new material for applications in fiber-amplifier and mid-infrared laser devices.

2. Experimental procedure

The compositions of the host chalcohalide glasses were (100-x)(0.8GeS₂·0.2Ga₂S₃)·xCdI₂ with x = 5, 10, 15 and 20, while x is the mole percent. More precisely, these glasses were labeled as GGC5, GGC10, GGC15, GGC20, respectively. All investigated samples were prepared by heating a mixture (typically 6 grams) of the required amounts of the elemental materials (5N for Ge, Ga and S, 3N for CdI₂, Dy₂S₃ and Tm₂S₃) at 950°C in an evacuated quartz ampoule (10mm inner diameter ampoule) for 12 hours and were subsequently quenched in an ampoule containing the liquid into a cold water bath. After being annealed near the glass transition temperature for 2 hours, the glass rod was cut into plates. Both surfaces of the plates were polished to mirror smoothness. The thicknesses of the obtained glass samples were 4.0mm.

All optical and spectroscopic measurements were carried out at room temperature. Absorption spectra of the samples were recorded using the Shimadzu UV-3101PC spectrophotometer between 400 to 3000nm wavelength. Fluorescence spectra were obtained using 808nm excitation from a laser diode (LD) (the average power is 1W). An InGaAs detector (Judson, U.S.A.) and an InSb detector (Judson, U.S.A.) cooled with liquid nitrogen were used to measure the intensities of fluorescence within the wavelength region of 1000~1500nm and 1650~4700nm, respectively. During the whole test, the locations of samples and the pumping source were kept unchanged and the excitation power was unvaried.

3. Results

3.1 Dy³⁺-doped glasses

3.1.1 Absorption spectra and Judd-Ofelt calculation

In this section, we will focus on the optical properties of Dy³⁺ sole-doped GGC chalcohalide glasses. The Dy³⁺ doping content is fixed to 0.2wt%. Figure 1 is the Vis-NIR absorption spectrum of the GGC5 glass in the wavelength region of 500~3000nm. The short-wavelength absorption edge of the glass lies at about 505nm. Good visible light transmittance is helpful for being pumped with proper excitation source. The absorption bands corresponding to internal 4f-4f electronic transitions of Dy³⁺ are indicated in the figure, in which the intensities of 910, 1108, 1298, 1700, 2826nm are strong, whereas that of 808nm is weak. This indicates that the pumping efficiency of 808nm LD will be low. On the other hand, as for the (100-x)(0.8GeS₂·0.2Ga₂S₃)·xCdI₂ (x = 5, 10, 15 and 20) serial glasses, the absorption coefficients of Dy³⁺ bands increase linearly as CdI₂ content increases. This slightly increases the absorption and efficiency of the pumping.

 

Fig. 1 Absorption spectrum of GGC5 glass doped with 0.2wt% Dy³⁺ (thickness 4mm).The insert is the energy level diagram of Dy³⁺ ion.

Download Full Size | PDF

The oscillator strengths of the GGC glasses were derived by integrating each absorption band area, and the results are summarized in Table 1 . Some representative host glasses are also listed for comparison. It is apparent that the oscillator strengths of chalcogenide and chalcohalide glasses are significantly larger than those of oxide and fluoride glasses. Particularly, in the oxide and fluoride glasses, oscillator strengths related to Dy³⁺ energy levels are not quite different. Whereas in the chalcogenide and chalcohalide glasses, oscillator strength of the electric dipole transition ⁶H_15/2→⁶H_9/2 + ⁶F_11/2 is much greater than other transitions, therefore the intensity of the corresponding absorption at 1298nm is the greatest within the absorption spectrum, as shown in Fig. 1. Furthermore, the oscillator strength of this transition also increases with the CdI₂ addition in the GGC serial glasses. It is obvious that the oscillator strength of the ⁶H_15/2→⁶H_9/2 + ⁶F_11/2 transition has a strong compositional dependence.

Table 1. Measured oscillator strengths and Judd-Ofelt intensity parameters of Dy³⁺-doped GGC serial glasses in comparison to other hosts [8].

View Table | View all tables in this article

Table 1 also compares the Judd-Ofelt intensity parameters of Dy³⁺ among various glass hosts. The calculated values of Ω ₂ for Dy³⁺ in GGC glasses are larger than those of other investigated hosts; and with the increase of CdI₂ content, the Ω ₂ is enhanced rapidly from 11.82 × 10⁻²⁰ to 14.26 × 10⁻²⁰cm². The value of Ω ₆ also has a similar trend, whereas that of Ω ₄ decreases markedly.

3.1.2 Fluorescence spectra and Radiative properties

The fluorescence spectra of Dy³⁺-doped GGC serial glasses under 808nm LD excitation in the ranges of 1000~1500 and 1650~1900nm are respectively shown in Fig. 2 . Three major emission bands were observed at around 1140, 1330 and 1750nm, corresponding to the optical transitions ⁶H_7/2 + ⁶F_9/2→⁶H_15/2, ⁶H_9/2 + ⁶F_11/2→⁶H_15/2 and ⁶H_11/2→⁶H_15/2, respectively. From GGC5 to GGC20, the intensities of the 1140 and 1330nm fluorescence increase. Simultaneously, the band center of the 1330nm fluorescence shifts to longer wavelength. The intensity and center location of 1750nm fluorescence have no obvious change. It should be noted that the effective line width (Δλ_eff) of 1330nm fluorescence also increases, while that of 1750nm has no obvious change. Unfortunately, the intensities of mid-infrared emissions at 2900 and 4300nm were too weak to be detected in the Dy³⁺ sole-doped glasses.

 

Fig. 2 Fluorescence spectra of Dy³⁺-doped GGC serial glasses (a) Detected with InGaAs detector in the range of 1000~1500nm (b) Detected with liquid nitrogen cooled InSb detector in the range of 1650~1900nm.

Download Full Size | PDF

Radiative transition probabilities (A), branching ratios (β) and radiative lifetimes (τ _rad) of the excited states of Dy³⁺ were evaluated following Judd-Ofelt theory and summarized in Table 2 .

Table 2. Radiative properties of Dy³⁺-doped GGC serial glasses.

View Table | View all tables in this article

The emission cross sections (σ_emi) of 1330 and 1750nm were calculated basing on the McCumber equation [11]

Radiative transition probabilities in the GGC glasses increase almost linearly with the addition of CdI₂ and the values are similar to that of Ge₂₅Ga₅S₇₀ glass but considerably greater than those of tellurite glasses. This is due to the higher refractive indices of GGC chalcohalide glasses.

3.2 Dy³⁺, Tm³⁺-codoped glasses

3.2.1 Absorption spectra

Figure 3 illustrates the Vis-NIR absorption spectra of the GGC20 glasses codoped with 0.2wt% Dy³⁺ and 0.3~0.7wt% Tm³⁺ in the wavelength region of 500~3000nm. Each assignment corresponds to the excited level of Dy³⁺ and/or Tm³⁺. It is apparent that the absorption at 808nm was enhanced with the addition of Tm³⁺.

 

Fig. 3 Absorption spectra of GGC20 glasses codoped with 0.2wt% Dy³⁺ and 0.3~0.7wt% Tm³⁺ (thickness 4mm).

Download Full Size | PDF

3.2.2 Fluorescence spectra

Figure 4 shows the evolutions of fluorescence spectra with the increase of Tm³⁺ concentration in Dy³⁺, Tm³⁺-codoped GGC20 glasses. Emissions at 2900 and 4300nm due to the Dy³⁺: ⁶H_13/2→⁶H_15/2 and Dy³⁺: ⁶H_11/2→⁶H_13/2 transitions were clearly observed. The full widths at the half maximum (FWHM) intensities were ~170 and ~230nm, respectively. Introduction of Tm³⁺ also resulted in an obvious enhancement of 1330nm emission which is ascribed to Dy³⁺: ⁶H_9/2 + ⁶F_11/2→⁶H_15/2 transition. On the other hand, when the Tm³⁺ amount was more than 0.5wt%, the intensity of emission at 1450nm due to the Tm³⁺: ³H₄→³F₄ transition shows no increase.

 

Fig. 4 The evolutions of fluorescence spectra with the increase of Tm³⁺ concentration in Dy³⁺, Tm³⁺-codoped GGC20 glasses (a) Detected with InGaAs detector in the range of 1000~1600nm (b) Detected with liquid nitrogen cooled InSb detector in the range of 1650~4700nm.

Download Full Size | PDF

It should be noted that during the preparation of the experiment, we found that the glassy system became very unstable when the Tm³⁺ concentration was further increased, and the quartz ampoule would easily explode in the furnace. On the other hand, it can be seen from Fig. 4 that the intensities of emissions at 2900 and 4300nm were greatly improved when the Tm³⁺ concentration was enhanced from 0.3 to 0.5wt%, but no prominent improvements were achieved when the Tm³⁺ concentration was enhanced from 0.5 to 0.7wt%. Considering the above factors comprehensively, 0.5wt% is a suitable concentration for Tm³⁺. Therefore, in the following text for investigating the effect of CdI₂ content on mid-infrared emissions, the concentrations of Dy³⁺ and Tm³⁺ were fixed to 0.2 and 0.5wt%, respectively.

The fluorescence spectra of Dy³⁺,Tm³⁺-codoped GGC serial glasses under 808nm LD excitation in the range of 1000~1600 and 1650~4700nm are shown in Fig. 5(a) and (b) respectively. It is worth noting that for the concerned two mid-infrared emissions at 2900 and 4300nm, the intensities’ evolutions have similar trends with that of the 1450nm emission, which is the intensities decrease first and then increase as the CdI₂ content increasing. But the intensity ratio of I₁₄₅₀/I₁₈₀₀ decreases remarkably, as shown in Fig. 5(c). The calculated emission cross sections (σ_emi) of 2900 and 4300nm for GGC20 glass are 1.68 × 10⁻²⁰ and 1.20 × 10⁻²⁰cm², respectively. These values are higher than those of the Dy³⁺-doped chalcogenide glasses in previous studies [7,9].

 

Fig. 5 (a) Fluorescence spectra of Dy³⁺,Tm³⁺-codoped GGC glasses detected with InGaAs detector in the range of 1000~1600nm (b) Detected with liquid nitrogen cooled InSb detector in the range of 1650~4700nm (c) Dependence of I₁₄₅₀/I₁₈₀₀ intensity ratio on CdI₂ content.

Download Full Size | PDF

4. Discussion

4.1 Oscillator strengths and intensity parameters

Oscillator strengths and intensity parameters are related to the local environments of rare-earths in hosts. Generally, lower symmetry of polyhedra surrounding the rare-earths and higher covalency of bonds inside the hosts result in larger oscillator strengths. Compared to oxide and fluoride glasses, oscillator strengths of Dy³⁺ ion in chalcogenide and chalcohalide glasses have larger values resulting from more significant distortion of sulfur-rare-earth polyhedra [8] and higher covalency of bonds. It can also be seen from Table 1 that in the GGC serial chalcohalide glasses, the values of Dy³⁺ oscillator strengths were enhanced almost linearly as the content of CdI₂ increased. According to our previous work on GGC glasses’ structure [10], with the addition of CdI₂, the amount of the original [GeS₄] tetrahedra in glassy network decreases whereas the number of [S_4-xGeI_x] (x = 1, 2) tetrahedra becomes larger. The symmetry decreases and the covalence has also been reduced because of larger electronegativity of the I^- ion. Considering the enhancement of oscillator strengths in the GGC serial glasses, the decrease of symmetry is therefore suggested to play a dominant role whereas the influence of covalence’s decrease can be ignored. Large Ω ₂ values for GGC serial glasses also provide good evidence of the large asymmetry of the Dy³⁺ ion site.

Furthermore, value of Ω ₆ increases with the addition of CdI₂, reflecting the fact that more ionic of bonds and lower rigidity of glass are presented.

4.2 Effect of CdI₂ content on the emission spectra of Dy³⁺-doped GGC glasses

From the fluorescence measurement, it can be seen that higher content of CdI₂ in GGC glass system induces stronger fluorescence of Dy³⁺ ions, especially at 1330nm. This phenomenon accords with the variation of spontaneous emission probability summarized in Table 2 and has a similar trend with those of halogen or alkali halide introduced chalcogenide glasses [12]. Lower nonradiative decay rate rooting from lower phonon energy and the enhancement of pumping efficiency rooting from the increase of Dy³⁺ absorption coefficients with the addition of CdI₂ can effectively populate ⁶H_7/2-⁶F_9/ and ⁶H_9/2-⁶F_11/2 levels, resulting in the intensified fluorescences at 1140 and 1330nm.

The glassy structure’s evolution following with the CdI₂ addition induces more stark level splitting in Dy³⁺ [13], resulting in the band center’s red-shift and shape’s broaden of 1330nm fluorescence.

4.3 Energy transfer and effect of Tm³⁺ concentration on the emission spectra of Dy³⁺, Tm³⁺-codoped GGC glasses

In the Dy³⁺ sole-doped glasses, we did not observe the 2900 and 4300nm emissions because of two main reasons. The intrinsical one is that the absorption of Dy³⁺at 808nm is quite low and the radiative transition probabilities of mid-infrared emissions are small. Also, the relatively low power of 808nm LD and low sensitivity of detecting equipment used in this experiment are further causes of the absences of 2900 and 4300nm emissions.

In the Dy³⁺, Tm³⁺-codoped GGC glasses, emissions at 2900 and 4300nm were clearly observed. Preliminary study by J. Heo’s indicated the presence of energy transfer process from the Tm³⁺: ³F₄ to Dy³⁺: ⁶H_11/2 (as shown in Fig. 6 ) and the increase of the 2900nm fluorescence accordingly [14]. Evolutions of spectra in present work also proved this process. Furthermore, occurrence of another energy transferring from the Tm³⁺: ³H₅ to Dy³⁺: ⁶H_9/2 + ⁶F_11/2 is suggested based on the obvious enhancement of 1330nm fluorescence shown in Fig. 4 (a). This provides a new idea for the development of a 1.3μm fiber amplifier. A cross relaxation process, i.e. ³H₄, ³H₆→³F₄, ³F₄ is well-known to exist in the high concentration Tm³⁺ doped chalcogenide glasses [15]. The corresponding exhibition in fluorescence spectrum is that the intensity ratio of I₁₄₅₀/I₁₈₀₀ becomes smaller with increasing Tm³⁺ concentration. It can therefore be concluded from Fig. 4 that the ³H₄, ³H₆→³F₄, ³F₄ cross relaxation process has occurred in the 0.2wt% Dy³⁺ and 0.5wt% Tm³⁺ co-doped glasses. While, considering the aim of our present work is to obtain strong 2900 and 4300nm mid-infrared emissions, large electron populations of Tm³⁺: ³F₄ is expected since it acts as source in the Tm³⁺: ³F₄→ Dy³⁺: ⁶H_11/2 energy transfer process. Therefore, in the Dy³⁺, Tm³⁺-codoped GGC glasses, the concentration of Tm³⁺ should be no less than 0.5wt%.

 

Fig. 6 Energy transition processes in Dy³⁺, Tm³⁺-codoped glasses

Download Full Size | PDF

4.4 Effect of CdI₂ content on the emission spectra of Dy³⁺, Tm³⁺-codoped GGC glasses

J.Heo et al. [16] figured out that the existence of the ethane-like units is responsible for the relatively higher solubility of rare-earth ions in Ge-Ga-S glasses as compared to the As-based glasses. Structural investigation of Ge-Ga-S-CdI₂ glasses in our previous work based on the Raman spectral evolution indicated that in the glasses with little CdI₂, some ethane-like units [S₃(Ga)Ge-Ge(Ga)S₃] exist because of the lack of sulfur, but the amount of these units will decrease with the addition of CdI₂ [10]. When the content of CdI₂ is higher than 10mol%, the ethane-like units disappear. It makes the clustering degree of Tm³⁺ ions increases, resulting in the decrease of intensity ratio of I₁₄₅₀/I₁₈₀₀ from GGC5 to GGC20 glass. This is helpful for the Tm³⁺: ³F₄→ Dy³⁺: ⁶H_11/2 energy transfer process and relatively strong mid-infrared emissions are achieved in Dy³⁺, Tm³⁺-codoped GGC20 glasses.

5. Conclusion

In the search for novel materials with potential practical applications in fiber-amplifiers and mid-infrared laser devices, a serial Dy³⁺-doped and Dy³⁺, Tm³⁺-codoped (100-x)(0.8GeS₂·0.2Ga₂S₃)·xCdI₂ (0≤x≤20) chalcohalide glasses were elaborated. Judd-Ofelt strength parameters Ω_t (t = 2, 4, 6) were evaluated based on the absorption spectra. The transition probabilities, exited state lifetimes and the branching ratios were calculated by using the Judd-Ofelt theory. For the Dy³⁺-doped glasses, increase of CdI₂ had a positive effect on the 1330nm fluorescence, yielding from the lower non-radiative decay rate and higher pumping efficiency. The largest emission cross sections (σ_emi) of 1330nm were estimated to be 4.19 × 10⁻²⁰cm² for the 0.2wt% Dy³⁺-doped 64GeS₂·16Ga₂S₃·20CdI₂ glass. For the Dy³⁺, Tm³⁺-codoped glasses, an increase of CdI₂ diminished the amount of ethane-like units [S₃(Ga)Ge-Ge(Ga)S₃], improved the Tm³⁺: ³F₄→ Dy³⁺: ⁶H_11/2 energy transfer efficiency and intensified the mid-infrared emissions. The largest emission cross sections (σ_emi) of 2900 and 4300nm were respectively estimated to be 1.68 × 10⁻²⁰ and 1.20 × 10⁻²⁰cm² for the 0.2wt% Dy³⁺, 0.5wt% Tm³⁺-codoped 64GeS₂·16Ga₂S₃·20CdI₂ glasses.