Spectroscopic Properties of TmF₃-Doped CaF₂ Crystals

Abstract

In this study, we report the growth and comprehensive spectroscopic analysis of TmF₃-doped CaF₂ crystals, grown using the vertical Bridgman method. The optical absorption and photoluminescence properties of both trivalent (Tm³⁺) and divalent (Tm²⁺) thulium ions were investigated. Optical absorption spectra in the UV-VIS-NIR range reveal characteristic transitions of Tm³⁺ ions, as well as weaker absorption bands corresponding to Tm²⁺ ions. The Judd–Ofelt (JO) formalism was applied to determine the intensity parameters Ω₂, Ω₄, and Ω₆, which were used to calculate radiative transition probabilities, branching ratios, and radiative lifetimes for the Tm³⁺ ions. The emission spectra showed concentration-dependent quenching effects, with significant emissions observed for the concentration of 0.1 mol% TmF₃ under excitation at 260 nm and 353 nm for Tm³⁺ ions and at 305 nm for Tm²⁺ ions. A new UV emission associated with divalent Thulium is reported. The results indicate that higher TmF₃ concentrations lead to increased non-radiative energy transfer, which reduces luminescence efficiency. These findings contribute to the understanding of the optical behavior of Tm-doped fluoride crystals, with implications for their application in laser technologies and radiation dosimetry.

1. Introduction

The spectroscopic properties of TmF₃-doped CaF₂ crystals have attracted significant interest due to their potential applications in laser technologies and optical materials [1,2,3,4]. Rare-earth (RE) ions dissolved in CaF₂, excluding Sm, Eu, and Yb, typically exist in the trivalent state. However, it is known that a certain fraction of these RE³⁺ ions can be reduced to the divalent state through various methods: X-ray irradiation [5,6]; gamma irradiation [7], additive coloration [8], or directly in the as-grown crystals [9,10,11,12,13,14].

Tm³⁺-doped CaF₂ crystals exhibit efficient near-infrared (NIR) emission, making them suitable for various laser applications [15,16,17,18,19]. Incorporating Tm³⁺ ions into fluoride hosts enhances their spectroscopic properties due to the low phonon energy environment, which reduces non-radiative losses. This makes Tm³⁺-doped CaF₂ a promising material for developing efficient laser sources.

On the other hand, Tm²⁺-doped CaF₂ crystals have been investigated especially for the red and near-infrared emission [5,8,20,21].

Previous studies have demonstrated that Tm³⁺-doped CaF₂ crystals exhibit significant spectroscopic characteristics, including a broad emission band around 1.8 μm under excitation at 795 nm, which are essential for laser applications [16]. The clustering behavior of Tm³⁺ ions in CaF₂ crystals has also been studied, revealing that clustering can affect the spectroscopic properties of the material [18,19,22]. The energy transfer mechanisms between Tm³⁺ ions in these crystals are crucial for optimizing their performance in laser applications.

The laser performance of Tm³⁺-doped CaF₂ crystals has been extensively investigated. Studies have shown that these crystals can achieve high slope efficiencies and output powers when used as laser gain media [16,17,18]. Additionally, the co-doping of CaF₂ crystal with other rare earth ions, such as La³⁺ [23], Y³⁺ [24] or Lu³⁺ [25], can further enhance the near-infrared luminescence performance by enabling efficient energy transfer processes.

Tm³⁺-doped CaF₂ crystals also exhibit unique thermoluminescent properties, making them suitable for applications in radiation dosimetry [26]. The thermoluminescent response of these crystals under gamma radiation exposure indicates their potential for use in radiation monitoring and measurement.

Studies on Tm³⁺-doped CaF₂ crystals also reveal their potential for self-Q-switched lasing, demonstrating their versatility in laser technology applications [4,27].

The optical properties of Tm³⁺-doped CaF₂ have been further studied in different host materials, showing a significant influence of the host in luminescence characteristics [28,29,30,31]. Fluoride crystals with an ordered structure stand out among various host matrices due to their unique chemical, mechanical, and thermal properties. These crystals exhibit a large band gap, broad transparency spectra extending from the vacuum UV to the far-infrared regions, possess low phonon energies that enhance activator luminescence, and feature extended fluorescence lifetimes, among other advantageous characteristics [32,33,34,35,36,37,38]. In the cubic CaF₂ crystal structure, the F⁻ ions are located at the corners of the cubes. Alternatively, these cubes are occupied with Ca²⁺ ions. Trivalent RE ions substitute for the Ca²⁺ ions and there is a need for charge compensation to maintain electrical neutrality. It has been shown that such charge compensation is enabled by F⁻ ions taking place in the interstitial positions of the empty cubes. At low dopant concentrations (<0.1 wt%), isolated Tm³⁺ ions can be found in several sites resulting from the possible positions for the compensating F⁻ ions in the rare earth ion vicinity. Strickland and Jones [22] identified three major centers in lightly doped 0.05% Tm³⁺:CaF₂: one center consists of a trivalent ion without any local charge compensation (O_h symmetry) and two centers that involve F⁻ charge compensating ions, located as near neighbors or as next near neighbors, the symmetry of these centers being tetragonal (C_4v) and trigonal (C_3v), respectively. At slightly higher rare earth dopant concentrations, adjacent rare earth ions appear forming ion pairs (or dimers) and more complicated clusters such as trimers and tetramers depending on the size, nature, and concentration of the considered RE ion [19]. Such a diversity of sites has important consequences on the spectroscopic characteristics of the material, and the very low concentrated systems (<0.05 wt%) must be distinguished from the higher concentrated ones for which ion clusters may be predominant. Despite the fact that there are numerous examples of research on the spectroscopic properties of Tm³⁺ ions in different host materials, there are only a few, older studies on the spectral properties of Tm²⁺ ions in fluorite crystals [19]. A recent study shows the luminescence properties of the Tm²⁺ ions in fluorite crystals. It is reported that an emission in the 680–800 nm range corresponds to 4f-4f transitions of Tm²⁺ upon excitation by 450 nm laser radiation [20].

In this study, we investigate the spectroscopic properties of TmF₃-doped CaF₂ crystals in the UV-VIS spectral region, with a focus on both Tm³⁺ and Tm²⁺ ions, respectively. To achieve this goal, optical absorption and photoluminescence (PL) measurements were performed for both Tm³⁺ and Tm²⁺ ions, and the Judd–Ofelt (JO) model was used to obtain information on the spectroscopic properties of Tm³⁺:CaF₂ crystals. We focused on the influence of TmF₃ concentration on the optical absorption and UV-VIS luminescence behavior. Our investigation builds on previous research and aims to provide a comprehensive understanding of the mechanisms that govern the optical behavior of these doped crystals.

2. Materials and Methods

TmF₃-doped CaF₂ crystals were grown in our Crystal Growth Laboratory by the vertical Bridgman method using a shaped graphite furnace which ensures appropriate temperature distribution [39]. The mentioned dopant concentrations refer to the raw material concentration added to the melt. The crystals were grown under a vacuum of approximately 10⁻¹ Pa in a graphite crucible. Suprapure grade (Merck) calcium fluoride and thulium fluoride were used as the starting materials. It is known that good charge conversion can be obtained in high deoxidization conditions during the growth process. The use of a graphite crucible during the synthesis can provide a reducing environment that influences the Tm³⁺ → Tm²⁺ charge conversion in the as-grown samples [40]. Figure 1 shows the typical preparation stages of the Bridgman setup.

The whole process takes place in four stages: (i) the heating stage to reach the melting temperature, T_m = 1381 °C (approximately 6 h). The nominal power to reach the melting temperature is P = 4.7 kW. (ii) The melting stage which is necessary to ensure that the entire amount of raw material is melted (approximately 3 h). In stage (iii) the crystal pulling process takes place at a pulling rate of 4 mm/h, at constant power and extends for almost 20 h (AB in Figure 1). The crystal growth begins by lowering the crucible in the temperature distribution in the furnace. The last stage (iv) corresponds to the crystal cooling stage with a cooling rate of approximately (3 ± 0.03) °C/min, which ensures a dislocation density of ~(1 ± 0.1) × 10⁴ dislocations/cm² [41]. The grown crystals are approximately 8 cm long and 10 mm in diameter. The accuracy in measuring the length and diameter of the crystal is 0.01 mm. The crystals were cleaved along the (111) crystallographic direction into several slices. For this study, crystals with thicknesses varying between 1.65 and 2.91 mm were used. They are transparent and free of any visible inclusions or cracks, as shown in the inset of Figure 1a–c. It is observed that with the increase in the concentration of TmF₃ added in the melt, from 0.1 mol% TmF₃ (Figure 1a) to 1 mol% TmF₃ (Figure 1b) and 5 mol% TmF₃ (Figure 1c), the crystals have an increasingly green color.

The room temperature optical absorption spectra in the UV-VIS-NIR spectral range were recorded using a Shimadzu 1650PC from Shimadzu, Kyoto, Japan and Nexus 470 FTIR spectrophotometer from Thermo Scientific, Cambridge, UK. The baseline correction was applied to the measured absorption spectra in order to eliminate distortions caused by instrument noise, reflectance losses, and stray light. Without this correction, the absorption data would be distorted, leading to overestimated absorption coefficients and inaccurate Judd–Ofelt (JO) analysis. This correction ensured that only the absorption due to the Tm³⁺ and Tm²⁺ ions was considered, providing a true representation of the material’s optical properties. The raw absorption spectra were first plotted, and baseline points were identified automatically based on the minimal absorption values. A linear fitting method was then applied to define the baseline across the spectrum, ensuring that both broad and narrow features were accounted for. The baseline was subtracted from the original data using the “Subtract” function in OriginLab 9.0, which allowed us to obtain the corrected absorption spectrum accurately representing the TmF₃-doped CaF₂ crystals’ true absorption characteristics. To measure the room temperature luminescence spectra, in the UV-VIS domain the PerkinElmer LS55 spectrofluorimeter was used. The luminescence spectra of Tm³⁺ ions were measured by excitation at two wavelengths, 260 and 353 nm respectively, which correspond to the ³H₆ → ¹D₂ and ³H₆ → ³P₂ transition [42]. The branching ratios, the emission transition probabilities, and the radiative lifetimes were obtained using the JO analysis (Judd–Ofelt analysis) [43,44]. A comparison with results obtained by other authors was provided. The influence of Tm³⁺ ion concentration on the JO parameters and the radiative lifetime was also investigated. The emission spectra of Tm²⁺ ions were registered under excitation at 305 nm corresponding to the 4f–5d transitions.

3. Results

3.1. Optical Absorption Spectra

In Figure 2a, the absorption spectra of CaF₂:x mol% TmF₃ (where x = 0.1, 1, and 5 mol%) are presented, revealing multiple absorption bands corresponding to both trivalent and divalent thulium ions. To express the absorption spectra in absorption coefficient units, $\mathsf{\alpha }\left(\mathsf{\lambda }\right)$, the Beer–Lambert law was applied: $\mathsf{\alpha }(\mathsf{\lambda })={\displaystyle \frac{2.303 lo{g}_{10}(I/I_0)}{d}}$, where d is the thickness of the sample in cm and $lo{g}_{10}(I/I_0$) is the absorbance. In order to minimize errors due to optical losses, absorption baseline correction was applied. To be visible, the absorption spectrum corresponding to the concentration of 0.1 mol% TmF₃ is multiplied by 10 (black line). The observed peaks are attributed to well-known transitions from the ground state, ³H₆, of Tm³⁺ ions (on the basis of Hund’s rules) to the excited states ³P₂ (256 nm) ³P_1,0 (273 nm), ¹D₂ (353 nm), ¹G₄ (460 nm), ³F₂ (650 nm), ³F₃ (674 nm), ³H₄ (766 nm), ³H₅ (1207 nm) and ³H₆ (1608 nm) of Tm³⁺ ions [42]. For a free Tm³⁺ ion, its electronic configuration is 4f¹². Along with spectral absorption bands of Tm³⁺ ions in the CaF₂ host a few weaker absorption bands which correspond to the 4f–5d transition of Tm²⁺ ions are observed, and they are indicated in Figure 2a. These bands are sensitive to the crystal field, which varies with the host lattice (CaF₂, SrF₂, BaF₂), leading to shifts in the energy of the absorption peaks [20]. These maxima are centered around 305, 409, and 593 nm. A linear trend of increasing the absorption coefficient at 353 nm and 260 nm (associated with Tm³⁺ ions) and 305 nm (associated with Tm²⁺ ions) with the concentration of TmF₃ was observed. To establish a definitive linear dependence, more measurements across a wider range of concentrations would be necessary. The correlation coefficients are shown in Figure 2b. Also, the error bars were added to each data point, representing the standard deviation from multiple measurements to illustrate the uncertainty in the absorption coefficient values at different TmF₃ concentrations, thereby providing a clearer depiction of the data’s reliability and supporting the observed linear trends.

3.2. Judd–Ofelt Analysis

The standard Judd–Ofelt (JO) analysis was employed for CaF₂ doped with 1 and 5 mol% TmF₃ to calculate the spectroscopic parameters based on the optical absorption spectra for Tm³⁺ ions [42,43]. By analyzing six absorption bands corresponding to specific transitions of Tm³⁺ ions from the ground state ³H₆ to excited states: ³F₄, ³H₅, ³H₄, ²F_2,3, ¹G_4, and ¹D₂, the JO intensity parameters Ω₂, Ω₄ and Ω₆ were determined. Due to the very low intensity of the absorption bands in the CaF₂ crystal doped with 0.1 mol% TmF₃, the JO analysis could not be applied. The parameter Ω₂ is highly sensitive to the local environment surrounding the doped ions and is related to the asymmetry and covalency of the rare-earth ion sites. The very large difference between the values of Ω₂ can be explained by the cluster formation with the increase in TmF₃ concentration. To obtain the JO parameters, Ω_t (where t = 2, 4, 6), and the corresponding experimental line strengths (

Table 1. The transition, mean wavelength and the integrated absorption cross-section of the measured and calculated absorption line strengths of the selected absorption peaks of CaF₂:TmF₃ crystals.

Sample		CaF2:1 mol% TmF3			CaF2:5 mol% TmF3
Transition, 3H6→	λm [nm]	Σ [×10−20 cm2·nm]	Smeas [×10−20 cm2]	Scalc [×10−20 cm2]	Σ [×10−20 cm2·nm]	Smeas [×10−20 cm2]	Scalc [×10−20 cm2]
3F4	1787	4.027	1.437	1.211	15.605	0.933	0.783
3H5	1234	14.1428	1.248	1.093	8.125	0.702	0.885
3H4	785	6.960	0.944	1.098	6.806	0.923	0.978
2F2 + 2F3	677	9.089	1.429	1.267	9.146	1.436	1.148
1G4	460	1.148	0.264	0.087	1.402	0.323	0.089
1D2	353	1.722	0.513	0.181	1.669	0.498	0.368
			ΔSrms = 0.298 × 10−20 cm2			ΔSrms = 0.267 × 10−20 cm2

), a system of six equations was solved using the Levenberg–Marquardt algorithm, based on the six selected transitions. The experimental line strengths were calculated using the following expression:

$$S_{\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{s}}={\displaystyle \frac{3\hslash c\left(2J+1\right)n\left(\lambda \right)}{8{\pi }^3{N}_0{\lambda }_{\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{n}}{e}^2}}\left[{\displaystyle \frac{9}{{\left({n\left(\lambda \right)}^2+2\right)}^2}}\int \alpha \left(\lambda \right)\mathrm{d}\lambda \right]$$

(1)

where J is the total angular momentum quantum number of the initial state, n(λ) is the refractive index, N₀ is the Er³⁺ ions concentration (added to the melt), λ_mean is the mean wavelength of the specific absorption bands, $\Sigma =\int \alpha \left(\lambda \right)\mathrm{d}\lambda$ is the integrated absorption coefficient as a function of λ, α(λ) is absorption coefficient, c is the vacuum speed of light and ħ is Planck’s constant. The refractive index, n, was determined from the Sellmeier dispersion equation [45]. The determination of the JO parameters involved solving a set of six equations for the transitions between the J and J′ manifolds using the experimental line strengths. This calculation is based on the Expression (1) and on the electric dipole (ed) line strength:

$$S_{{JJ}^{\prime }}^{\mathrm{e}\mathrm{d}}=\sum _{t=\mathrm{2,4},6}{\mathsf{\Omega }}_{\mathrm{t}}{\left|⟨\left(\mathrm{S},\mathrm{L}\right)\mathrm{J}\Vert {\mathrm{U}}^{\left(\mathrm{t}\right)}\Vert \left({\mathrm{S}}^{\prime },{\mathrm{L}}^{\prime }\right){\mathrm{J}}^{\prime }⟩\right|}^2$$

(2)

and the magnetic dipole (md) line strength given in [29]:

$$S_{{JJ}^{\prime }}^{\mathrm{m}\mathrm{d}}={\left|⟨\left(\mathrm{S},\mathrm{L}\right)\mathrm{J}\left|\right|\mathrm{L}+2\mathrm{S}\left|\right|\left({\mathrm{S}}^{\prime },{\mathrm{L}}^{\prime }\right){\mathrm{J}}^{\prime }⟩\right|}^2$$

(3)

where $⟨\Vert U^{\left(t\right)}\Vert ⟩$ are the reduced matrix elements of rank t (t = 2, 4 and 6) of tensor operators between states characterized by the quantum numbers (S, L and J) and (S′, L′ and J′). For selected Tm³⁺ transitions, the values of the reduced matrix elements are those tabulated in Kaminskii’s paper [42]. The calculated JO parameters, by applying a least-squares fitting of S_meas and S_calc, and the spectroscopic quality factor, χ, are given in

Table 2. The Judd–Ofelt parameters, Ω_t (expressed in 10⁻²⁰ cm²) and the spectroscopic quality factors, χ.

Ωt (10−20 cm2)	CaF2:1 mol% TmF3 (This Paper)	CaF2:5 mol% TmF3 (This Paper)	CaF2: 0.5 at.% TmF3 [24]	CaF2:1.5 at.% TmF3, 4 at.% YF3 [3]	CaF2:3 at.% TmF3, 2 at.% LaF3 [23]	CaF2:3 at.% TmF3, 2 at.% LaF3 [27]
Ω2	1.25043	0.20294	3.28	0.891	1.429	1.283
Ω4	0.15075	0.88504	0.99	1.251	1.485	1.323
Ω6	1.44999	1.03299	1.52	0.959	1.163	2.159
χ	0.10397	0.85677	0.65	1.304	1.277	0.590

. A comparison of the obtained JO parameters, Ω_t, with those reported by other authors [3,23,24,27].

The root mean square deviation, defined by $\Delta S_{\mathrm{rms}}={\left[{\left(q-p\right)}^{-1}\sum {\left(S_{\mathrm{calc}}-S_{\mathrm{meas}}\right)}^2\right]}^{1/2}$ is a measure of the accuracy of the fitting procedure; q = 6 is the number of analyzed spectral bands and p = 3 is the number of sought parameters. In this case, ΔS_rms = 0.298 × 10⁻²⁰ cm² for CaF₂:1 mol% TmF₃ and ΔS_rms = 0.267 × 10⁻²⁰ cm² for CaF₂:5 mol% TmF₃ which are comparable with those obtained by other authors [3,23,24]. To calculate the radiative lifetime (τ_rad) for an excited state, J′, the relationship τ_rad = 1/ΣA_JJ′ can be used, where:

$$A_{{JJ}^{\prime }}={\displaystyle \frac{64{\pi }^2{e}^2}{3\hslash (2J+1){\lambda }_{\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{n}}^3}}\left[{\displaystyle \frac{n{(n^2+2)}^2}{9}}{S}_{{JJ}^{\prime }}^{\mathrm{e}\mathrm{d}}+n^2{S}_{{JJ}^{\prime }}^{\mathrm{m}\mathrm{d}}\right]$$

(4)

is the spontaneous emission probability and the sum is taken over all final lower-lying states, $S_{{JJ}^{\prime }}^{\mathrm{e}\mathrm{d}}$ and $S_{{JJ}^{\prime }}^{\mathrm{m}\mathrm{d}}$ are defined by Equations (2) and (3). The calculation of fluorescence branching ratios, β_JJ′ involves utilizing the expression β_JJ′= A_JJ′·τrad.

Table 3. The transitions corresponding to the wavelengths where luminescence was observed (Figure 3a,b), along with the calculated radiative lifetimes, radiative emission probabilities, and branching ratios.

Transition	λem. (nm)	Ref.	CaF2:1 mol% TmF3			CaF2:5 mol% TmF3
			τcalc. (ms)	AJJ′ (s−1)	βJJ′	τcalc. (ms)	AJJ′ (s−1)	βJJ′
3F3→3H6	690	[3,32]	0.87	1010	0.87 ± 0.173	0.95	915.4	0.87 ± 0.129
1D2→3H5	510	[1,32]	0.20	78.7	0.02 ± 0.003	0.20	58.5	0.01 ± 0.002
1G4→3H6	482	[1,3,32,46,47,48,49,50]	0.83	175.7	0.15 ± 0.030	1.23	157.5	0.22 ± 0.033
1D2→3F4	449	[1,3,32,46,47,48,49,50,51]	0.20	2048.8	0.41 ± 0.081	0.20	1137.6	0.23 ± 0.034
3P0→3H5	378	[32]	0.78	0.10	-	0.95	0.10	-
1D2→3H6	358	[3,46]	0.20	1508.8	0.30 ± 0.060	0.20	3074.3	0.63 ± 0.093
3P0→3F4	343	[32]	0.78	11.4	0.01 ± 0.002	0.95	66.8	0.06 ± 0.009

shows the values of the radiative emission probabilities, the branching ratios, and the radiative lifetimes for transitions where luminescence was observed (Figure 3). It should be mentioned that magnetic-dipole transition matters in transitions that obey the selection rules: ΔJ = 0 or ±1, ΔS = 0 and ΔL = 0. Taking absorption spectra into account $S_{{JJ}^{\prime }}^{\mathrm{m}\mathrm{d}}$ is non-zero for ³F₂ → ³F₃, ³F₃→ ³F₄, ³H₄ → ³H₅ and ³H₅ → ³F₆ transitions.

The results obtained using the JO theory suggest that the radiative transition probability of the ¹D₂ state is greater than that of other states, as shown in Table 2. This conclusion is in good agreement with Wang et al. [32]. The Judd–Ofelt parameters for different molar contents of TmF₃, presented in Table 2, provide information on how the local environment of Tm³⁺ ions and the symmetry of the crystal field evolve with increasing TmF₃ concentration. These parameters allow the evaluation of optical transition probabilities and radiative lifetimes of the material, which are directly related to the performance of Tm-doped CaF₂ crystals in applications such as lasers and phosphors. As the mole fraction of TmF₃ increases, we can observe changes in the values of Ω₂, which is sensitive to the asymmetry of the crystal field, indicating how the local structure of Tm³⁺ ions evolves with concentration. The results obtained in this work are compared with those obtained by other authors to provide a basis for understanding how the optical properties of Tm³⁺ in CaF₂ crystals vary depending on the dopant concentration. [3,23,24,27]. From Table 2, the values of Ω_t indicate distinct differences between the samples doped with 1 mol% and 5 mol% TmF₃. For the 1 mol% TmF₃ sample, χ = 0.104, which suggests that higher-order transitions Ω₆ dominate, reflecting a relatively symmetric local environment around the Tm³⁺ ions. In contrast, the 5 mol% TmF₃ sample shows a significantly higher χ value of 0.857, indicating an increased contribution from moderate angular momentum transitions (Ω₄), which may be attributed to increased distortion in the local symmetry due to higher dopant concentration. These results suggest that as the concentration of TmF₃ increases, the local environment becomes more distorted, leading to changes in the relative strength of optical transitions.

3.3. Emission Spectra of Tm³⁺ Ions

To obtain the room temperature emission spectra, two absorption bands were pumped, namely λ_exc. = 260 nm (corresponding to the ³H₆ → ³P₂ transition) and λ_exc. = 353 nm (³H₆ → ¹D₂ transition). The emission spectra of the studied samples under 260 nm excitation are shown in Figure 3a,b. The emission peaks are observed at 343, 358, 378, 449, 482, 510 nm, and 690 nm. These peaks correspond to the following transitions: ³P₀ → ³F₄, ¹D₂ → ³H₆, ³P₀ → ³H₅, ¹D₂ → ³F₄, ¹G₄ → ³H₆, ¹D₂ → ³H₅, and ³F₂ → ³H₆. Two emission bands around 449 and 482 nm corresponding to the ¹D₂ → ³F₄ and ¹G₄ → ³H₆ are observed under excitation at 353 nm.

The energy level diagram, in Figure 4, provides a comprehensive overview of the electronic transitions that correspond to the emission spectra observed in Figure 3a,b. This diagram allows us to understand the mechanisms behind the different emission peaks and how they relate to the energy levels of the Tm³⁺ ions in the CaF₂ crystal.

The International Commission on Illumination (CIE) chart of the CaF₂:Tm³⁺ crystal corresponding to the visible emissions is shown in Figure 5a,b for both excitation wavelengths. The color coordinates are shown in

Table 4. The CIE coordinates chart of the CaF₂:Tm³⁺ crystal corresponding to the observed visible emissions under excitation at 260 and 356 nm.

CaF2:x mol% TmF3	λexc. (nm)	X-Coordinate	Y-Coordinate	λexc. (nm)	X-Coordinate	Y-Coordinate
x = 0.1	260	0.28	0.20	356	0.18	0.10
x = 1		0.25	0.22		0.18	0.16
x = 5		0.25	0.24		0.17	0.19

. The results are in good agreement with those reported by Taikar et al. for La₂O₃:Tm³⁺ [1]. The calculated color coordinate nearly matched with National Television Standard Committee (NTSC) standard values for blue color. The CIE coordinates were obtained using Gocie V2 software [51].

3.4. Emission Spectra of Tm²⁺ Ions

It is known that the spectroscopic behavior of Tm²⁺ ions is influenced by the host crystal’s structure, the local environment around the ion, and external factors like temperature and irradiation [5,7,20,21,35,37]. Emission spectra of Tm²⁺ ions were investigated in fluorite crystals, especially for the red emission, in the 680–800 nm region [20] under excitation at 450 nm, or infrared emission at 1116 nm [21] and 1250 nm [35]. Figure 6 shows the emission spectra of CaF₂ crystals doped with various concentrations of TmF₃ (0.1, 1, and 5 mol%) under 305 nm excitation, corresponding to the Tm²⁺ transitions. The prominent emission peak at 353 nm is observed, with the intensity drastically decreasing as the concentration of TmF₃ increases.

Figure 6b presents the emission spectrum (black line) and the excitation spectrum (blue line) for CaF₂ doped with 0.1 mol% TmF₃. The excitation spectrum, which corresponds to monitoring the emission at 353 nm, shows a clear peak around 305 nm. This suggests that excitation at 305 nm effectively populates the energy levels that lead to the emission observed at 353 nm. The presence of the 305 nm excitation peak is indicative of Tm²⁺ ions, as this excitation wavelength aligns with known absorption bands of Tm²⁺ rather than Tm³⁺. Together, these spectra suggest that the 353 nm emission observed in the figure primarily originates from Tm²⁺ ions when excited at 305 nm, supporting the conclusion that the emission at 353 nm is due to transitions of Tm²⁺ ions.

4. Discussion

The absorption spectrum is a usual method for characterizing the transition properties from the ground state to all excited states of luminescent centers. Additionally, luminescent centers can be identified through their absorption spectra, as each specific center has distinct absorption peaks within a given host material. Figure 2 provides a comprehensive overview of the absorption properties of CaF₂ crystals doped with various concentrations of TmF₃, highlighting the distinct absorption behavior of Tm³⁺ and Tm²⁺ ions in the 250–1800 nm spectral region.

The absorption intensity generally increases with the concentration of TmF₃, particularly evident in the higher concentration samples (x = 5 mol%), which show more pronounced absorption peaks. The spectra demonstrate the successful doping of TmF₃ into the CaF₂ matrix, with distinct absorption features consistent with Tm³⁺ and Tm²⁺ ions. The presence of Tm²⁺ bands suggests a partial reduction in Tm³⁺ during crystal growth, which could be of interest for further studies exploring the control of valence states in doped fluoride crystals [18,40]. Figure 2b provides a quantitative analysis of the absorption coefficients at three specific wavelengths: 353 nm (associated with ³H₆ → ¹D₂ transition of Tm³⁺ ions), 260 nm (corresponding to the ³H₆ → ³P₂ transition of Tm³⁺ ions), and 305 nm (associated with Tm²⁺ ions). These absorption bands were selected for measuring the emission spectra (see Section 3.3 and Section 3.4). The graph suggests a linear trend between the absorption coefficient and the TmF₃ concentration, with correlation coefficient values between 0.98 and 0.998. This linear trend can help to estimate the optical properties by adjusting the concentration of TmF₃ in the host materials. On the other hand, the linearity observed suggests uniform incorporation of Tm ions into the CaF₂ lattice and minimal clustering, which is essential for maintaining the optical quality of the material. The study of the distribution of trivalent and divalent thulium ions along the CaF₂ crystals and the determination of the effective segregation coefficients corresponding to both ions in fluorite crystals will be analyzed in a separate paper. The strongest absorption at 674 nm for the highest concentration of TmF₃ indicates that Tm³⁺ ions dominate the absorption features. On the other hand, the slower increase in the absorption coefficient at 305 nm for Tm²⁺ ions with TmF₃ concentration suggests a relatively smaller Tm²⁺ ions population increase with TmF₃ concentration This balance between Tm³⁺ and Tm²⁺ states could be exploited in designing materials for specific photonic applications, such as tunable lasers or optical amplifiers, where controlled absorption is desired.

The emission intensity is highest at the lowest doping concentration (x = 0.1 mol%), particularly for the 343 nm peak, which corresponds to the ³P₀→³F₄ transition. As the doping concentration increases, the emission intensity decreases significantly, indicating a concentration-quenching phenomenon. Taikar et al. [1] reported that after 1 mol% Tm₂O₃ was doped in La₂O_3, the emission intensity of Tm³⁺ ions decreases because of concentration quenching. In CaF₂ crystals, the 1450 nm emission of Tm³⁺ ions is quenched for 1.34% TmF₃ for which the clusters are predominant. At low doping concentrations (around 0.1 mol%), Tm³⁺ ions are mostly isolated, occupying different lattice sites due to the various possible positions of the compensating F⁻ ions. These isolated Tm³⁺ ions, which occupy distinct lattice sites with specific local symmetries, exhibit sharp and well-defined emission spectra because their luminescence is not significantly influenced by neighboring ions. As the concentration of Tm³⁺ ions increases beyond this low doping regime, adjacent Tm³⁺ ions start to form ion pairs (dimers) or more complex clusters (trimers, tetramers, hexamers, etc.). This proximity between Tm³⁺ ions facilitates energy transfer between them, which is one of the primary mechanisms of concentration quenching. This behavior is demonstrated by the experimentally observed fluorescence quenching reported by Renard et al. [19] and Camy et al. [18], which arises directly from these non-radiative processes.

The emission peaks of Tm³⁺ ions observed at 343, 358, 378, 449, 482, 510 nm, and 690 nm, corresponding to the ³P₀ → ³F₄, ¹D₂ → ³H₆, ³P₀ → ³H₅, ¹D₂ → ³F₄, ¹G₄ → ³H₆, ¹D₂ → ³H₅, and ³F₂ → ³H₆ transitions were reported previously in La₂O₃:Tm³⁺ [32]. To our knowledge, the intense emission band at 343 nm characteristic of the Tm³⁺ ions doped in the CaF₂ crystal has not been reported before.

By excitation at 353 nm, the emission spectrum is characterized by two emission bands: one of them centered at 449 nm (Figure 3a) which corresponds to the ¹D₂ → ³F₄ transition, and the other centered at 482 nm (with a much weaker intensity) corresponding to the ¹G₄ → ³H₆ transition of Tm³⁺ ions.

In order to determine manifold-to-manifold radiative emission probabilities, radiative lifetimes and branching ratios the JO theory was applied to the room-temperature absorption spectra. A computer program was designed to perform the least squares fitting between measured and calculated electric dipole line strength of 6 transitions corresponding to the Tm³⁺ ions shown in Table 1. Due to the very low intensity of the absorption bands in the CaF₂ crystal doped with 0.1 mol% TmF₃ (see Figure 2), the JO analysis could not be applied. The differences between the theoretical and experimental values can be attributed to the fact that the theoretical values are based on the Judd–Ofelt (JO) theory, which involves certain approximations, such as the assumption of ideal local symmetry around the rare-earth ions (Oh site-symmetry) and homogeneous distribution of dopants within the crystal lattice. In reality, the local environment of Tm³⁺ ions may deviate from these ideal conditions due to lattice defects, clustering, or site distortions, leading to differences between calculated and measured values. At higher concentrations of Tm³⁺ ions, interactions such as cross-relaxation, energy transfer, and ion clustering become significant, affecting the measured transition probabilities. These effects are not typically included in the simplified theoretical calculations, leading to further discrepancies between theoretical and experimental values, especially for transitions like the ¹G₄ state. On the other hand, significant errors usually occur in the estimation of JO parameters because it is difficult to obtain accurate absorption line strengths in the case of broad and structured absorption bands (as in our case), and due to the JO model itself [52]. These large errors have also been reported in other papers [24,30] for Tm³⁺ ions.

The JO parameters are listed in Table 2 along with the JO parameters found in previous studies [3,23,24,27]. The results obtained suggest that the radiative transition probability of the ¹D₂→³F₄ transition (~2049 s⁻¹) is greater than that of other states, as shown in Table 2. This result is in good agreement with the emission spectrum under excitation at 353 m (Figure 3b). However, the strong emission around 343 nm under excitation at 260 nm (³H₆→³P₂) cannot be explained by JO analysis. Further investigations are needed to clarify this.

The UV emission spectra of Tm²⁺-doped CaF₂ crystals under 305 nm excitation (Figure 6) show a pronounced peak at 353 nm for CaF₂:0.1 mol% TmF₃. To our knowledge, the UV emission of Tm²⁺ ions has not been reported before. The intensity of this peak decreases significantly with increasing TmF₃ concentration, from 0.1 mol% to 5 mol%, suggesting the presence of concentration quenching. This quenching effect can be attributed to the increased probability of non-radiative energy transfer between Tm²⁺ and Tm²⁺/Tm³⁺ ions, which becomes more pronounced at higher doping levels. The observed spectral characteristics align with previous studies on Tm²⁺ in CaF₂, which highlight the role of crystal field splitting and the importance of maintaining optimal doping concentrations to preserve luminescence efficiency [35]. These findings contribute to our understanding of the spectroscopic behavior of Tm²⁺ in fluoride hosts, with implications for their use as a luminescent material.

5. Conclusions

In this work, we successfully grew TmF₃-doped CaF₂ crystals using the vertical Bridgman method and investigated their spectroscopic properties. The optical absorption spectra revealed distinct transitions of both trivalent (Tm³⁺) and divalent (Tm²⁺) thulium ions, providing insights into the nature of the doped ions and their interactions within the host matrix. The application of the Judd–Ofelt theory enabled us to determine the spectroscopic parameters, including intensity parameters, radiative transition probabilities, branching ratios, and radiative lifetimes which are useful for understanding the luminescence mechanisms in these materials. Our photoluminescence studies demonstrated that TmF₃-doped CaF₂ crystals exhibit strong UV-VIS emissions, with significant emission quenching observed at higher dopant concentrations due to increased non-radiative energy transfer. This quenching effect highlights the importance of optimizing dopant concentrations to maximize the luminescence of Tm³⁺ and Tm²⁺ efficiency in CaF₂ crystals. A new UV emission around 353 nm, associated with divalent Thulium, was reported. To our knowledge, the intense emission band at 343 nm characteristic of the Tm³⁺ ion doped in the CaF₂ crystal has not been reported before. The observed luminescent properties, particularly the emission peaks associated with both ions, suggest that these materials hold great potential for applications in laser technologies. Overall, our findings provide an understanding of the spectroscopic behavior of TmF₃-doped CaF₂ crystals, contributing to the knowledge of the field of rare-earth-doped fluoride materials. Future work will focus on exploring the control of valence states in these crystals and optimizing their optical properties for specific photonic applications.