III. RESULTS AND DISCUSSION

Figure 1 shows the DTA curve of the present glass where T _g (480 °C), T _x1 (600 °C), and T _x2 (730 °C) stand for the glass transition, first and second crystallization temperatures, respectively. XRD results show that the first crystallization peak T _x1 is ascribed to the precipitation of SrF₂ crystals and the second crystallization peak T _x2 to the bulk crystallization process of the glass matrix.

FIG. 1.

DTA curve of the as-made oxyfluoride glass.

The XRD patterns of glass and GCs (GC2h, GC4h, and GC6h) are shown in Fig. 2. As expected, the XRD pattern of the glass contains two broad curves, typical of structures without long-range order. For the GCs, there are a number of narrow and relatively intense peaks that show a diffraction pattern of a crystalline phase, with more intense peaks for the GC6h sample, indicating its higher crystalline nature and/or larger size of the nanocrystals. Analysis of the peaks reveals that the sample contains a cubic SrF₂ crystalline phase with JCPDS standard card No. 06-0262 and from peak widths of the XRD spectra, average size of the nanocrystals have been calculated using the Scherrer formula,

(1)$$D_{hkl} = {{K{\rm{\lambda }}} \over {{\rm{\beta cos\theta }}}}\quad ,$$

where D _hkl is the crystal size in the vertical direction of (hkl), λ is the wave length of x-rays, θ is the angle of diffraction, β is the full width at half maximum of the diffraction peak and K = 0.90.^{Reference Kawamoto, Kanno and Qiu17} The average sizes of nanocrystals have been determined using the above formula and are found to be 35, 39, and 40 nm for GC2h, GC4h, and GC6h samples, respectively. A similar crystal size (38 nm) of SrF₂ nanocrystals was reported in the Er³⁺-Yb³⁺ co-doped GCs.^{Reference Qiao, Fan, Wang and Zhang13}

FIG. 2.

XRD patterns of the glass and the GCs (GC2h, GC4h, and GC6h).

The TEM morphology of 2.0 mol% Er³⁺-doped GC6h is shown in Fig. 3. It can be seen that most of the particles are in clusters of uniform size and few particles are agglomerates with an irregular shape. The agglomerates observed may be due to alcohol used for dispersion of powders during TEM analysis. A few particles show nano-sized ellipse-like morphology, with an average particle size of 40 nm. The selected area electron diffraction (SAED) pattern shows concentric rings indicating polycrystalline nature of the GCs and is shown in the inset of Fig. 3. All diffraction rings can be identified as the cubic structure and is consistent with the XRD pattern.

FIG. 3.

TEM micrograph of 2.0 mol% Er³⁺-doped GC6h sample and the inset is the SAED pattern.

Absorption spectra of 2.0 mol% Er³⁺-doped oxyfluoride glass and GCs in the UV-visible-NIR range (300–1700 nm) are shown in Fig. 4. There are 13 and 11 bands in the absorption spectra of glass and GCs, respectively, which are assigned to transitions from the ⁴I_15/2 ground state to the excited states of the Er³⁺ ion. From absorption spectra it can be found that TGCs containing SrF₂ nanocrystals can be obtained by appropriate thermal treatment of the precursor glass around the first crystallization temperature. By increasing the time of thermal treatment, however, the GCs gradually lost their transparency. In the present GCs, GC2h and GC4h are fully transparent but GC6h is slightly turbid. The decrease in transparency might be due to light scattering of larger nanocrystals. According to Rayleigh-Gans model,^{Reference Kerker18} TGCs can be obtained if the radii of nanocrystals are less than 15 nm and the refractive index difference between the glass and nanocrystalline phase is less than 0.1. In the present GCs, the size of SrF₂ nanocrystals increases gradually with increase in time of thermal treatment resulting in a gradual decrease in transparency. The absorption band positions (λ_p), experimental (f _exp) and calculated (f _cal) oscillator strengths (10⁻⁶ cm⁻¹) of 2.0 mol% Er³⁺-doped glass and GCs have been determined using the relationships described elsewhere^{Reference Renuka Devi and Jayasankar19} and are presented in Table II. Reasonably small root mean square (σ_rms) deviations (×10⁻⁶ cm⁻¹) of ±0.47, ±0.46, ±0.62, and ±0.47 are found for glass, GC2h, GC4h, and GC6h, respectively, indicating the good fit between the experimental and calculated oscillator strengths.

FIG. 4.

Absorption spectra of the Er³⁺ ions in the glass and GCs (GC2h, GC4h, and GC6h). The absorption bands were assigned to the transitions from ground state to the excited states (1) ²G_7/2, (2) ⁴G_9/2, (3) ⁴G_11/2, (4) ²G_9/2, (5) ⁴F_3/2, (6) ⁴F_5/2, (7) ⁴F_7/2, (8) ²H_11/2, (9) ⁴S_3/2, (10) ⁴F_9/2, (11) ⁴I_9/2, (12) ⁴I_11/2, and (13) ⁴I_13/2.

TABLE II.

The absorption band positions (λ_p), experimental (f _exp) and calculated (f _cal) oscillator strengths (10⁻⁶) of 2.0 mol% Er³⁺-doped glass and GCs (GC2h, GC4h, and GC6h). Absorption transitions are from the ground state, ⁴I_15/2, to the excited states.

The absorption bands, ⁴I_15/2 → ⁴G_11/2 and ⁴I_15/2 → ²H_11/2 located at 378 and 520 nm, respectively, are most intense and are called hypersensitive transitions (HSTs).^{Reference Babu, Seo, Kesavulu, Jang and Jayasankar20,Reference Gorller-Walrand, Binnemans, Gschneidner and Eyring21} The HSTs are sensitive to small changes of the environment around RE ions^{Reference Gorller-Walrand, Binnemans, Gschneidner and Eyring21,Reference Pisarski22} and follow the selection rules, |ΔS| = 0, |ΔL| ≤ 2, and |ΔJ| ≤ 2.^{Reference Gorller-Walrand, Binnemans, Gschneidner and Eyring21} From Table II, it is interesting to note that oscillator strengths of these HSTs are smaller in glass ceramics than in glass. This is an indication for the lower site symmetry^{Reference Gorller-Walrand, Binnemans, Gschneidner and Eyring21} (higher asymmetry) of Er³⁺ ions in the glass. In general, Ω₂ is sensitive to the asymmetry in the vicinity of the RE ions (short-range effects)^{Reference Tanabe, Ohyagi, Soga and Hanada23} and the changes in Ω₆ are correlated with long-range effects of glass host.^{Reference Tanabe, Hayash, Hanada and Onodera24} For many instances, Ω₂ is also found to be affected by the covalence between the RE ions and the ligand anions.^{Reference Heidepriem, Ehrt, Bettinelli and Speghini25} In traditional glasses, the Ω₂ values decrease with the change of host from oxide to fluorides.^{Reference Zou and Izumitani26}

Using the experimental oscillator strengths of various absorption bands and refractive index, the Judd-Ofelt (JO) intensity parameters (Ω_λ, λ = 2, 4, and 6) have been derived^{Reference Judd27,Reference Ofelt28} and are presented in Table III. For the calculation of JO parameters, the matrix elements, ||U^λ||², have been taken from (Ref. Reference Renuka Devi and Jayasankar19). As can be seen from Table III, with increase in the time of thermal treatment, Ω₂ gradually decreases indicating that the ligand field of Er³⁺ ion is more ionic in the GCs. This is due to progressive incorporation of Er³⁺ ions into SrF₂ nanocrystals. During the nucleation and formation of SrF₂ nanocrystals, Er³⁺ ions act as crystal nucleus. Thus the Er³⁺ ions might enter into the SrF₂ nanocrystalline phase and the Sr²⁺ ions are partially substituted by Er³⁺ ions. The Er³⁺ ions in the SrF₂ nanocrystal are coordinated by F⁻ ions and more ionic environment of the Er³⁺ ions in the GCs can be expected in comparison with the glass, which leads to the decrease of Ω₂ parameter.

TABLE III.

The JO parameters (Ω_λ, × 10⁻²⁰ cm²) of Er³⁺ ions in the glass, GCs (GC2h, GC4h, and GC6h) and other hosts.

The JO parameters (×10⁻²⁰ cm²) of glass are Ω₂ = 8.50, Ω₄ = 2.03, and Ω₆ = 2.58. These values are compared with Er³⁺: glass systems in Table III that include silicate,^{Reference Zou and Izumitani26} phosphate,^{Reference Yu, Song, Wang, Luo, Han, Cheng, Sun, Tian and Pun29} fluorophosphate,^{Reference Zemon, Lambert, Andew, Miniscalco, Hall, Wei and Folweiler30} lead borate,^{Reference Pisarski22} tellurite,^{Reference Nandi, Jose, Jayakrishnan, Debbarma, Chalapathi, Alti, Dharmadhikari, Dharmadhikari and Mathur31} fluoride,^{Reference Pisarski22} oxyfluoride,^{Reference Qiao, Fan, Wang and Zhang32} SrF₂,^{Reference Qiao, Fan, Wang and Wang12} CaF₂,^{Reference Kumar, Riman, Chae, Jang, Bae and Moon33} LaF₂,^{Reference Kumar, Riman, Chae, Jang, Bae and Moon33} and BaF₂,^{Reference Patel, Reddy and Nash-Stevenson34} crystals. The trend of JO parameters (Ω₂ > Ω₆ > Ω₄) is similar for glass and GCs, fluoride^{Reference Pisarski22} and LaF₂^{Reference Sardar, Russell and Yow35} systems whereas for other glass systems (Er³⁺: silicate,^{Reference Zou and Izumitani26} fluorophosphate,^{Reference Zemon, Lambert, Andew, Miniscalco, Hall, Wei and Folweiler30} lead borate,^{Reference Pisarski22} tellurite,^{Reference Nandi, Jose, Jayakrishnan, Debbarma, Chalapathi, Alti, Dharmadhikari, Dharmadhikari and Mathur31} oxyfluoride,^{Reference Qiao, Fan, Wang and Zhang32} SrF₂,^{Reference Qiao, Fan, Wang and Wang12} and CaF₂,^{Reference Kumar, Riman, Chae, Jang, Bae and Moon33}) the parameters follow the general trend as Ω₂ > Ω₄ > Ω₆. The trend of JO parameters for Er³⁺:BaF₂ crystals^{Reference Patel, Reddy and Nash-Stevenson34} (Ω₄ > Ω₂ > Ω₆) differs from the other two hosts (Er³⁺: LaF₂ and Er³⁺: CaF₂). This variation in trends of JO parameters can be attributed to the large and more sensitive values of oscillator strengths of the two HSTs. In the present glass, the value of Ω₂ is larger than the other two parameters. This tendency is also seen in other oxide glasses shown in Table III. The larger value of Ω₂ in the present glass suggests a larger degree of covalency of the Er–O bond and/or asymmetry of the Er³⁺ sites than other Er³⁺: glass systems that are compared in Table III.

The Ω₄ parameter is affected by the factor causing changes in both Ω₂ and Ω₆ and is not usually studied for local structure investigations.^{Reference Heidepriem, Ehrt, Bettinelli and Speghini25} However, the spectroscopic quality factor χ (=Ω₄/Ω₆), which was initially established only to quantify the IR emission of Nd³⁺ ions,^{Reference Qiao, Fan, Wang and Zhang32} is an important predictor for stimulated emission in a laser active medium.^{Reference Sardar, Russell and Yow35} The values of χ have been calculated for the present glass and GCs and are compared to those of reported Er³⁺: glass systems in Table III. The χ values of glass and GCs are within the range of 0.22–1.5, found for Nd³⁺ ions in different hosts.^{Reference Kaminiskii36} The smaller value of χ is associated with more intense emission of the ⁴S_3/2 → ⁴I_15/2 transition. From Table III, it can be found that the value of χ decreases with increasing time of thermal treatment indicating the higher intensity green emission of GCs compared to glass. The JO parameters have been used to predict radiative properties^{Reference Babu, Seo, Jang, Balakrishnaiah, Jayasankar and Joshi37} of the important luminescence levels of Er³⁺ in glass and GCs such as transition probabilities (A) and lifetimes (τ_R) and are presented in Table IV. The radiative lifetimes of Er³⁺ emission transitions are dominated by Ω₄ and Ω₆ since the square of the matrix elements ||U⁽²⁾|| of these transitions are zero or much smaller than ||U⁽⁴⁾|| and ||U⁽⁶⁾|| except for the HSTs, ²H_11/2 → ⁴I_15/2 and ²G_11/2 → ⁴I_15/2. For these transitions, Ω₂ plays an important role. The τ_R values of the luminescent levels of present systems are compared to those of reported Er³⁺: glass systems in Table V. As can be seen, the τ_R values for the present glass follow the trend as ⁴I_13/2 > ⁴I_9/2 > ⁴I_11/2 > ⁴F_9/2 > ⁴S_3/2. A similar trend was also found for Er³⁺: lead phosphate,^{Reference Santos, Guedes, Loong, Boatner, Moura, de Araujo, Jacinto and Vermelho38} and tellurite,^{Reference Lin, Liu, Pun, Ma, Peng, An, Yu and Jiang39} glass systems whereas for Er³⁺: oxyfluoride,^{Reference Qiao, Fan, Wang and Wang12} fluoride,^{Reference Pisarski22} and lead borate,^{Reference Pisarski22} glass systems, the τ_R values decrease continuously with increase in energy of the excited states. Further, the τ_R values in the present glass are less than those of Er³⁺: oxyfluoride,^{Reference Qiao, Fan, Wang and Wang12} fluoride,^{Reference Pisarski22} and lead borate,^{Reference Pisarski22} glass systems but are slightly more than Er³⁺: lead phosphate,^{Reference Santos, Guedes, Loong, Boatner, Moura, de Araujo, Jacinto and Vermelho38} and much more than tellurite,^{Reference Lin, Liu, Pun, Ma, Peng, An, Yu and Jiang39} glass systems.

TABLE IV.

Emission transitions (SLJ → S′L′J′), predicted radiative transition probabilities (A) and lifetimes (τ_R) of some important emission levels of 2.0 mol% Er³⁺-doped glass and GCs (GC2h, GC4h, and GC6h).

TABLE V.

Comparison of predicted radiative lifetimes (ms) of some important luminescence levels of Er³⁺-doped glass and GCs (GC2h, GC4h, and GC6h) with some reported Er³⁺: Glasses.

Stokes emission spectra of the glass and GCs, measured by exciting at 365 nm, are shown in Fig. 5. The luminescence bands of the Er³⁺ ions can be assigned to ²H_11/2 → ⁴I_15/2 (530 nm), ⁴S_3/2 → ⁴I_15/2 (552 nm), ⁴F_9/2 → ⁴I_15/2 (627 nm), and ⁴I_9/2 → ⁴I_15/2 (824 nm) transitions. As can be seen, red emission bands (⁴F_9/2 → ⁴I_15/2) of the GCs are split into several sharp peaks. It is well known that Stark splitting is caused by the crystal field. Splitting of emission bands shows that Er³⁺ ions have been incorporated into the SrF₂ nanocrystals. The overall increase in Stokes emission in GCs is due to lowering of phonon energy of the host, which reduces multiphonon relaxation between the energy levels due to progressive incorporation of Er³⁺ ions into SrF₂ nanocrystals. Stronger red emission compared to green is attributed to efficiency of cross-relaxation channel, ⁴F_7/2, ⁴I_11/2 →⁴F_9/2, ⁴F_9/2 due to matching of energy levels. The inset of Fig. 5 shows the variation of downconversion intensities versus glass and GC samples. However, all the GC samples show not only the enhanced green emission from the ⁴S_3/2 level but also more than ten times stronger red emission from the ⁴F_9/2 level and stronger near-infrared emission from the ⁴I_9/2 level because of reduced multiphonon relaxation. It implies that Er³⁺ ions are incorporated into SrF₂ nanocrystals with lower phonon energy environment during crystallization.

FIG. 5.

Stokes emission spectra (λ_ex = 365 nm) of the Er³⁺ ions in the glass and GCs (GC2h, GC4h, and GC6h) and the inset shows the variation of downconversion intensities for glass and GC samples.

Upconversion luminescence spectra of glass and GCs, measured with 980 nm excitation, are shown in Fig. 6. The emission bands can be assigned to ⁴S_3/2 → ⁴I_15/2 (545 nm), ⁴F_9/2 → ⁴I_15/2 (664 nm) and ⁴I_9/2 → ⁴I_15/2 (828 nm) transitions, respectively. In the glass, upconversion luminescence is very weak and can hardly be seen. However, in GCs, upconversion luminescence increases with increase in time of thermal treatment. Furthermore, with the increase in time of thermal treatment, Stark splitting of the 659 nm upconversion emission band increases gradually, which further indicates that there is progressive incorporation of Er³⁺ ions into SrF₂ nanocrystals. The inset of Fig. 6 shows the variation of upconversion intensities in glass and GC samples. There is a small increase in green and near-infrared emission intensities in GC samples compared to glass. In contrast, compared with the Stokes emission in Fig. 5, the relative intensity of red to green or NIR emission is several times enhanced in upconversion emission. It implies that the cross-relaxation of ⁴S_3/2 → ⁴F_9/2 and ⁴I_9/2 → ⁴F_9/2 feeds the ⁴F_9/2 state, which dramatically enhances the red emission intensity.

FIG. 6.

Upconversion emission spectra (λ_ex = 980 nm) of the Er³⁺ ions in the glass and GCs (GC2h, GC4h, and GC6h) and the inset shows the variation of upconversion intensities for glass and GC samples.

Partial energy level diagram of Er³⁺ ions in glass and GCs illustrating the mechanisms of upconversion emissions is shown in Fig. 7. When glass and GCs are excited with 980 nm wave length radiation, the Er³⁺ ions in the ground state absorb the pump photons and are excited to the ⁴I_9/2 level. This process is known as GSA. Then Er³⁺ ions relax nonradiatively to the ⁴I_11/2 and ⁴I_13/2 levels, in which part of the excitation energy in the ⁴I_11/2 level further relaxes radiatively and nonradiatively to the ⁴I_13/2 level. From the ⁴I_11/2 and ⁴I_13/2 levels, the Er³⁺ ions are further excited to the ⁴F_3/2 and ²H_11/2 levels by the pump photons through the process known as excited state absorption (ESA) due to good energy matching. In Fig. 7, ESAs of ⁴I_11/2 → ⁴F_3/2 and ⁴I_13/2 → ²H_11/2 are denoted as ESA1 and ESA2, respectively. Out of these two, ESA2 is more probable due to the longer lifetime of the ⁴I_13/2 level. The Er³⁺ ions pumped to the ⁴F_3/2 level relax nonradiatively and populate the ²H_11/2 level, which is also directly populated by the ESA2 process. Another possibility of excitation is by energy transfer from the ⁴I_11/2 level [shown as energy transfer upconversion (ETU1) in Fig. 7] in which two Er³⁺ ions in the ⁴I_11/2 level interact, and one ion gains energy and reaches the ⁴F_7/2 level whereas the other loses energy and goes to the ground state. Most of the electrons in the ⁴F_7/2 level then relax nonradiatively to the ²H_11/2 level. Another process that depopulates the ⁴F_7/2 level involves the interaction of two Er³⁺ ions, one of them in the ⁴I_11/2 level and the other in the ⁴F_7/2 level, both populating the ⁴F_9/2 level (shown as ETU2 in Fig. 7). Out of the two ETU processes, ETU2 is responsible mainly for the enhanced red emission. All the above processes have been illustrated in Fig. 7. Because there are other metastable energy levels between the ⁴S_3/2, ⁴F_9/2 and ⁴I_9/2 states and the ground state, the upconversion luminescence of Er³⁺ ions is usually baffled by the multiphonon relaxation. The multiphonon relaxation probability depends primarily upon the energy gap between two successive levels and phonon energy of the host.^{Reference Tanabe, Tamai, Hirao and Soga40} The lower the phonon energy of the host is, the smaller the multiphonon relaxation probability, and the longer the lifetime of the excited state. Owing to the smaller phonon energy of SrF₂ nanocrystals (300 cm⁻¹ for Sr–F),^{Reference Reisfeld and Jorgensen41} than that of silicate based glass matrix (1100 cm⁻¹ for Si–O), the upconversion luminescence of Er³⁺ in the SrF₂ nanocrystals is more intense than in glass. Thus the upconversion luminescence intensity of the GCs increases significantly with increase in time of thermal treatment due to increase in SrF₂ nanocrystalline phase around the Er³⁺ ions.

FIG. 7.

Partial energy level diagram of the Er³⁺ ions and possible upconversion luminescence mechanisms under 980 nm excitation.

In the upconversion process, the relationship between the emission intensity I _em and the IR excitation intensity I _ex is

(2)$$I_{{\rm{em}}} \propto (I_{{\rm{ex}}} )^n \quad ,$$

where n is the number of IR photons absorbed per visible photon emitted. Therefore, a plot of log (I _em) versus log (I _ex) should yield a straight line with a slope of n. Figure. 8 shows the pump power dependence of the up-conversion integral luminescence intensity corresponding to the ⁴F_9/2 → ⁴I_15/2 transition of the Er³⁺ ions in the GC6h sample. The slopes of the green and red emissions are 1.38 and 1.98, respectively. These results indicate that green (⁴S_3/2 → ⁴I_15/2) and red (⁴F_9/2 → ⁴I_15/2) luminescence mechanisms are related to the two-photon process. The possible up-conversion mechanisms can be described by ESA and/or ETU.^{Reference Fan, Wang, Qiao, Wang, Adam and Zhang42}

FIG. 8.

Log-log plot of the pump power dependence of upconversion emission intensity of the Er³⁺ ions in the GC6h sample.

The luminescence decay curves of the ⁴S_3/2 level in glass and GC6h have been measured by exciting with 355 nm from Nd-YAG laser and monitoring the emission at 552 nm and are shown in Fig. 9. The decay curves are found to be nonexponential for the glass and GC6h. The average lifetimes (τ_exp)^{Reference Lahoz, Martin, Mendez-Ramos and Nunez43} are estimated to be 1.1 and 11.6 μs for glass and GC6h, respectively. These results clearly indicate the incorporation of Er³⁺ ions into SrF₂ nanocrystals in the GC6h sample resulting in the longer lifetime of the ⁴S_3/2 level.^{Reference Fan, Wang, Qiao, Wang, Adam and Zhang42}

FIG. 9.

Fluorescence decay curves of the ⁴S_3/2 level of the Er³⁺ ions in the glass and GC6h sample.