Dual mode temperature sensing through luminescence lifetimes of F- and O-coordinated Cr³⁺ sites in fluorosilicate glass-ceramics

Abstract

Luminescence lifetime based temperature sensing has an intrinsic immunity to the influence of external conditions, and dual mode thermometry is highly accurate due to its “self-calibration” merit. To develop thermometry with both features, we investigated the phase and microstructural evolution of Cr³⁺-doped calcium-fluorosilicate glass and glass-ceramics, which revealed different luminescent behavior relating to the different Cr³⁺ sites in the materials. From the photoluminescence (PL) spectra, the emission at 717 nm was derived from the O-coordinated octahedral sites, while the 1 μm super-broad emission was assigned to the F-coordinated octahedral sites. After an annealing treatment, cubic CaF₂ nanocrystals were homogeneously precipitated in the glass-ceramics; thus, both the O-coordination in the residual glass phase and F-coordination in the CaF₂ crystalline phase were strengthened. This led to the enhancement of both the emissions at 717 nm and 1 μm. The O-coordinated sites were relatively strong-field sites in which the fluorescence of Cr³⁺ originated from the radiative transitions of the two thermally coupled energy levels, ²E and ⁴T₂, while the F-coordinated sites were relatively weak-field sites. Hence, the Cr³⁺ exhibits only one excited state ⁴T₂, which is inactivated by radiative transitions and non-radiative transitions from the thermal quench. Based on the obtained results, the maximum relative temperature sensitivity coefficients are 0.76% K⁻¹ at 498 K for the 717 nm emission and 0.47% K⁻¹ at 351 K for the 1 μm emission. This provides the possibility of developing a dual mode temperature sensor with high precision only using a single material.

---

1. Introduction

In recent years, optical temperature sensors have been widely applied in electrical power, chemical, metallurgical and other fields due to their advantages in anti-electromagnetic interference and high sensitivity.^1–3 In order to further improve the accuracy and extend the operating temperature regions of optical temperature sensors, researchers have attempted to utilize multi-active-center doped material systems or new host materials for doping active-center ions.^4–8 These attempts involve a variety of temperature sensing mechanisms, providing ample valuable inspiration for the design of new types of temperature sensors. Among the various types of optical temperature sensors, luminescence-lifetime-based sensors⁴ make up a large proportion due to their intrinsic immunity to the influence of external conditions such as those of the excitation beams and optical fiber transmission efficiencies. Trivalent chromium (Cr³⁺) ions, with the [Ar]3d³ electron configuration, have been widely applied as temperature sensing probes due to their diversity in temperature-dependent lifetimes, which vary with the different host materials used.^9–11 Three equivalent d-electrons in the free Cr³⁺ ion lead to the spectral terms ⁴F (ground state), ⁴P and ²G (excited states). In crystals or glasses, the electrostatic interactions between the d-electrons and ligand atoms can modify the spectral terms and induce the splitting of the d-orbital, where ⁴F would split into the ⁴T₁ and ⁴T₂ states and ²G would split into the ²E, ²T₁, ²T₂ and ²A₂ states. On the basis of crystal field theory,^12,13 the optical properties of Cr³⁺ ions can be manipulated by the coordination of the host lattice or matrix particularly the atoms in the first coordination sphere. This leads to adjustable lowest excited states for the Cr³⁺ ions in various crystal fields. According to the Tanabe–Sugano diagram, ²E acts as the lowest excited state of Cr³⁺ at a high crystal field strength and is related to a narrow-band red emission (∼700 nm) due to the parity and spin doubly forbidden ²E → ⁴A₂ transition with a long photo-luminescence (PL) lifetime. Furthermore, ⁴T₂ would act as the lowest energy state at a low crystal field strength and results in a broad-band near-infrared (NIR) emission (∼1 μm) due to the parity-forbidden but spin-allowed ⁴T₂ → ⁴A₂ transition with a relatively short PL lifetime. Therefore, Cr³⁺ could be a good candidate for dual mode temperature sensing by coordination into different phases with significantly different strength crystal fields in multi-phase materials, such as glass-ceramics.^5,14

Oxyfluoride glass-ceramics^15–17 have been considered as ideal hosts for various luminescence-active ions due to their high thermal and mechanical stability originating from the oxide glass matrix as well as their excellent spectroscopic merits related to the homogenously precipitated fluoride nanocrystals. Using a convenient heat treatment strategy, the crystal size and crystallinity can be easily controlled. Along with the evolution of both the phase and microstructure, the coordination of Cr³⁺ ions in different sites can be executed. Using this manipulation, one can design the crystal field around the Cr³⁺ ions and then adjust the spectroscopic behavior of the resulting materials.¹⁸

In this paper, Cr³⁺-doped oxyfluoride glass-ceramics with homogeneously distributed cubic CaF₂ : Cr³⁺ nanocrystals were prepared upon annealing 50SiO₂–20Al₂O₃–30CaF₂ glass. Two significantly different Cr³⁺ sites in the glass-ceramics were identified via PL spectroscopy and assigned to the O- and F-coordinated sites with red and NIR emissions, respectively. It follows that the temperature dependent PL lifetimes of Cr³⁺ at the different sites are both suitable for temperature sensing and can be proposed as a dual mode temperature sensing method requiring only the Cr³⁺-singly-doped glass-ceramics.

2. Experimental procedure

Oxyfluoride glass with the composition 50SiO₂–20Al₂O₃–29.95CaF₂–0.05CrF₃ (in mol%, named as G) was prepared using a melt-quenching method. The appropriate batch of high purity (3 N) raw materials (silica, alumina, calcium fluoride and chromium fluoride) was placed in a covered corundum crucible and melted at 1500 °C for 45 min in air. Plain glasses were obtained by quenching the melt between two brass plates. Then, the subsequent crystallization temperatures (580 °C, 600 °C, 620 °C and 640 °C) were selected between the glass transition temperature (T_g) and the first crystallization temperature (T_x1) (Fig. S1†). The glass-ceramics GC580, GC600, GC620 and GC640, named after their annealing temperature, were obtained by annealing the glass at the abovementioned temperatures for 2 h in the air.

Differential thermal analysis (DTA) measurements were carried out on a CDR-1 differential thermal analyzer with a fixed specimen weight of 60 mg. X-ray diffraction (XRD) measurements were performed on a DIMAX-RA X-ray diffractometer (Rigaku Corporation, Tokyo, Japan) using Cu-K_α radiation at a scan rate of 2° min⁻¹. Transmission electron microscopy (TEM) and high resolution transmission electron microscopy (HRTEM) measurements were conducted to check the fine crystal structure using a CM200 (Philips, Eindhoven, the Netherlands) microscope. Photoluminescence (PL) spectra, including the excitation spectra, emission spectra and temperature dependent luminescence decay curves, were collected using a FLSP920 spectrometer (Edinburgh Instrument Ltd., Livingston, UK) equipped with a TAP-02 temperature controller.

3. Results and discussion

3.1 Phase and microstructure

The glass-ceramics have a typical two-phase structure, where the cubic CaF₂ lattice grew up to the nanoscale in the glass matrix and Cr³⁺ enlarged the lattice as an interstitial dopant. As shown in the XRD patterns (Fig. 1(a)), the as-quenched glass (sample G) exhibits a typical amorphous feature without any sharp diffraction peaks. In contrast, the glass-ceramics show sharp XRD peaks and the diffraction intensity became stronger upon increasing the annealing temperature, indicating the precipitation of FCC (face centered cubic) structured CaF₂ (JCPDS # 35-0816). Calculated using the Scherrer equation, the average crystal size could be estimated to be about 10, 13, 15 and 20 nm for GC580, GC600, GC620 and GC640, respectively. Moreover, the higher annealing temperatures also led to a higher crystallinity of CaF₂ in the glass-ceramics. Furthermore, the diffraction peaks of the FCC CaF₂ crystals in the investigated glass-ceramics shifted to lower diffraction angles (2θ) as compared to those of the standard JCPDS card (# 35-0816). For GC620, the calculated lattice parameter of the precipitated nanocrystals was 5.490 Å, which was higher than the standard value (5.463 Å). This was clear evidence of crystal lattice expansion, indicating the enrichment of Cr³⁺ ions into the CaF₂ lattice. It was unlikely that the Cr³⁺ (0.62 Å) would substitute Ca²⁺ (1.2 Å) due to the related lattice shrinkage effect. From the data obtained for the Ca²⁺ radius (1.2 Å in octahedral coordination) and F⁻ radius (1.23 Å in tetrahedral coordination), the largest radii of the ions occupying the tetrahedral and octahedral interstices in cubic CaF₂ were calculated to be 0.27 Å and 0.50 Å, respectively. Hence, Cr³⁺ ions exhibit a higher probability to occupy the octahedral interstice and enlarge the octahedron. In addition, the glass and the glass-ceramics obtained upon annealing at temperatures lower than 600 °C (GC580 and GC600) maintain high transparency in the visible-light spectral region (Fig. 1(b)). However, in annealing at temperatures higher than 620 °C (GC620 and GC640), the glass-ceramics apparently lose their transparency. This was due to the large crystal size and crystallinity of the precipitated CaF₂ in GC620 and GC640 (Table 1). In the transmittance spectra, two visible absorption bands were assigned to the ⁴A₂ → ⁴T₁ (centered at 440 nm) and ⁴A₂ → ⁴T₂ (centered at 640 nm) energy level transitions of the Cr³⁺ ions Fig. 1(b).

Fig. 1 X-ray diffraction patterns (a) and transmittance spectra (b) of the Cr³⁺-doped glass and glass ceramics fabricated using different annealing procedures. The TEM (c) and HRTEM (d) images of the Cr³⁺-doped glass-ceramics obtained after annealing at 600 °C for 2 h (GC600).

Table 1 The crystal size and crystallinity of the Cr³⁺ ion doped glass-ceramics

	GC580	GC600	GC620	GC640
Crystal size (nm)	10.4 ± 1.1	13.4 ± 1.0	15.5 ± 1.1	20.5 ± 1.3
Crystallinity (%)	17.0 ± 0.5	24.8 ± 0.5	31.6 ± 0.5	35.6 ± 0.6

---

The transmission electron microscopy (TEM, Fig. 1(c)) and high-resolution TEM (HRTEM, Fig. 1(d)) images display the precipitated CaF₂ nanocrystals in GC600 with crystal sizes of ∼10–15 nm, which are dispersed homogeneously in the glass host. This is consistent with the crystal size calculated from the XRD patterns using the Scherrer equation. The CaF₂ nanocrystals in the HRTEM image exhibited a well-defined lattice structure, and the interplanar spacing was calculated to be 0.3215 nm using a fast Fourier transform (FFT) algorithm. The interplanar spacing corresponds to the (111) plane of cubic CaF₂ (0.3155 nm, JCPDS # 35-0816). Similar to the XRD results, the calculated interplanar spacing (d = 0.3215 nm in Fig. 1(d)) was larger than the standard value. This indicates that a number of Cr³⁺ ions have entered the precipitated CaF₂ crystalline phase, occupied the octahedral interstice, and enlarged the interplanar spacing.

3.2 Photoluminescence (PL)

The PL emission spectra (Fig. 2(a) and (c)) of the glass and glass-ceramics show different broad emission bands centered at 717 nm and 1 μm when excited at 296 nm and 464 nm, respectively. Here, the sharp zero-phonon lines (R-lines at ∼700 nm), observed due to the ²E → ⁴A₂ transition, could not be found in the spectra (Fig. 2(a)), which could be because the glass matrix inhomogenously broadened the zero-phonon lines.¹⁹ Thus, the emission may be assigned to ²E, ²T₁ → ⁴A₂, while the 1 μm emission may be attributed to ⁴T₂ → ⁴A₂. It was believed that these two clearly distinguishable emissions were related to the different Cr³⁺ sites coordinated with O²⁻ (717 nm) and F⁻ (1 μm) in the glass and glass-ceramics, respectively.²⁰ This will be further discussed in Section 3.3. By monitoring the emission at 717 nm, three broad excitation bands were recorded centered at 296 nm, 419 nm and 586 nm, corresponding to charge transfer band, Cr³⁺: ⁴A₂ → ⁴T₁ transition and Cr³⁺ :  ⁴A₂ → ⁴T₂ transition, respectively. By monitoring at 1 μm, the three excitation bands corresponding to charge transfer band, Cr³⁺ : ⁴A₂ → ⁴T₁ transition and Cr³⁺ : ⁴A₂ → ⁴T₂ transition still exist. However, all three bands were red-shifted to longer wavelengths and located at 306 nm, 438 nm and 617 nm. As the annealing temperature was increased, the 717 nm emission was improved very limitedly; however, the 1 μm emission intensity showed a significant enhancement. This could be due to the enhancement of the O²⁻ or F⁻ coordination of Cr³⁺ during the growth of the new phase (CaF₂ nanocrystals).

Fig. 2 The PL emission and excitation spectra (a, c) as well as the PL decay curves (b, d) of the Cr³⁺-doped glass-ceramics fabricated using the different annealing procedures: (a and b) by exciting at 296 nm and monitoring at 717 nm; (c and d) by exciting at 464 nm and monitoring at 1 μm.

The PL decay curves were recorded by monitoring the emission at 717 nm or 1 μm, as shown in Fig. 2(b) and (d). The PL decay curves were best fitted to double-exponential functions, which could be described by the following equation:

where τ₁ and τ₂ are the long- and short-decay components respectively, and parameters A₁ and A₂ are fitting constants. According to eqn (1), the average lifetime 〈τ〉 is given by:

According to eqn (2), the average lifetimes of the 717 nm and 1 μm luminescence were evaluated to be stable at about 1.6–1.7 ms and 22–23 μs, respectively. The average lifetime of 717 nm was much longer than that of 1 μm, which was consistent with assigning the 717 nm emission to the ²E, ⁴T₂ → ⁴A₂ transition (partiy- and spin-forbidden for ²E) and assigning the 1 μm emission to the ⁴T₂ → ⁴A₂ transition (parity-forbidden but spin-allowed).

3.3 O²⁻- or F⁻ coordination

Fluorosilicate glass probably constitutes both silicate and fluoride glassy matrices.²¹ Hence, the glass and glass-ceramics could provide F- and O-coordinated sites for Cr³⁺. On one hand, the silicate matrix is a random network of [SiO₄] tetrahedra, to which modifiers such as CaO were added to break up the networks and stabilizers such as Al₂O₃ were added to prevent crystallization.¹⁹ Breakage of the Si–O covalent bonds by the modifiers produce approximately octahedral arrangements of O²⁻. The Cr³⁺ ions prefer to occupy these octahedral sites rather than the tetrahedral sites. On the other hand, the fluoride matrix formed by the [AlF₆] octahedral and polyhedral [CaF₈] also exist in the fluorosilicate glass host, where the Cr³⁺ ions can easily substitute Al³⁺ to occupy the octahedral sites. Alkaline earth fluorides such as CaF₂ act as modifiers in the fluoride matrix and are easily crystallized from the matrix. According to the phase and microstructural analysis (Fig. 1), the Cr³⁺ ions also enter the precipitated CaF₂ crystalline phase of the glass ceramics, occupying the octahedral interstice and enlarging the interplanar spacing. As a result, the Cr³⁺ ions were octahedrally coordinated with O²⁻ and F⁻, and displayed two different luminescence behaviors. Upon increasing the annealing temperature, the crystallinity of the precipitated CaF₂ gradually increases, so that more Cr³⁺ ions are surrounded by the CaF₂ lattice and subsequently, the 1 μm emission was enhanced (Fig. 2(c)). The F/O ratio of the residual glassy phase was reduced to a lower level due to the crystallization of CaF₂ and thus, the O²⁻ coordination environment was significantly enhanced. However, such an O²⁻ coordination enhancement was too small to significantly influence the PL intensity (Fig. 2(a)).

The energy level structure of Cr³⁺ is highly dependent on the crystal field strength and determines the different PL behaviour of the O-coordinated and F-coordinated Cr³⁺ ions. The energy level structures of transition metal ions are usually determined by the relative strengths of the octahedral crystal ligand field splitting parameter, D_q, and the Racah parameters, B and C. Solutions to the multi-electron crystal-field Hamiltonian are represented on Tanabe–Sugano diagrams,^22,23 in which the normalized multiplet energies, E(Γ)/B, are plotted as a function of D_q/B, for a constant value of C/B, where Γ denotes the irreducible representation of the electronic state. In inorganic glass or crystals, the Cr³⁺ions prefer to occupy the sites exhibiting a nearly perfect octahedral symmetry because of the strong ligand field stabilization energy of Cr³⁺ in a six-fold coordination geometry.²⁴ As illustrated in the Tanabe–Sugano diagram (Fig. 3(a)), in an octahedral crystal field the ⁴F of Cr³⁺ will split into ⁴A₂, ⁴T₂ and ⁴T₁, while ²G would split into ²E, ²T₁, ²T₂ and ²A₁. The ⁴A₂ state has the lowest energy and serves as the ground state. The energy difference between ²E and ²T₁ or ²T₂ and ⁴A₂ was almost constant or varies slightly in all the fields; however, the energy difference between ⁴T₂ and ⁴A₂ varies significantly upon changing the crystal field. The studies of both Tanabe^25,26 and Casalboni²⁷ clearly show the relationship between the energy of the different excited states of Cr³⁺ ions and D_q/B. In weak crystal-field sites, where D_q/B < 2.3, the lowest excited state was an orbital triplet ⁴T₂, from which broadband PL is observed due to the enhancement of the phonon assisted ⁴T₂ → ⁴A₂ transitions. For strong crystal fields (D_q/B > 2.3), the lowest excited state was changed to ²E and the spectrum consists of narrow zero-phonon lines (R lines) with vibrationally induced sidebands due to the ⁴T₂ → ⁴A₂ transitions. For intermediate crystal fields (D_q/B ≈ 2.3), mixing between ⁴T₂ and ²E occurs and the observed photoluminescence spectrum, even at low temperature, is a superposition of the broad ⁴T₂ → ⁴A₂ band on the ²E → ⁴A₂ R line and its phonon sideband.

Fig. 3 Tanabe–Sugano diagram of Cr³⁺ ions in the octahedral crystal field (a); energy levels of the Cr³⁺ ions coordinated with F⁻ (b) in a low crystal field strength and with O²⁻ in an intermediate crystal field strength (c).

The strength D_q of the octahedral crystal field and the Racah parameter B can be determined from the peak energies of the ⁴A₂ → ⁴T₂ and the ⁴A₂ → ⁴T₁ transitions.²⁸ In octahedral symmetry, the energy difference between the ⁴A₂ and ⁴T₂ states is equal to 10D_q, which is measured from the peak energy (ν₁) of the ⁴A₂ → ⁴T₂ absorption band:

The value of B is determined from the energy value (ν₁) of ⁴A₂ → ⁴T₂ and the energy value (ν₂) of ⁴A₂ → ⁴T₁, is given by

According to the absorption spectra (Fig. 1(b)), the glass and the glass-ceramics have almost the same average values for ν₁ and ν₂, and thus have the same values of D_q = 1571.3 cm⁻¹, B = 734.1 cm⁻¹ and D_q/B = 2.14.

As a matter of fact, the PL spectra (Fig. 2) revealed two types of Cr³⁺ sites: the O²⁻ octahedrally coordinated sites (centered at 717 nm) and the F⁻ octahedrally coordinated sites (centered at 1 μm) in the glass and glass-ceramics. Herein, in order to evaluate the crystal field strengths of the [CrO₆] and [CrF₆] octahedra, the PL excitation peak wavelengths were used to deduce the values of ν₁ and ν₂; then, the D_q/B values were evaluated as 2.5 and 2.3 for [CrO₆] and [CrF₆], respectively. Thus, the energy level diagrams of the F⁻ octahedrally coordinated (Fig. 3(b)) and O²⁻ octahedrally coordinated (Fig. 3(c)) sites were elicited from the Tanabe–Sugano diagram (Fig. 3(a)). O²⁻ coordination has an intermediate crystal field with first excited states of ²E and ²T₁, while F⁻ coordination has a weak crystal field with a lower first excited state, ⁴T₂. This was consistent with the increasing trends of D_q/B correlated with the anion packing densities along the following sequence: fluoride → silicate → borate. Silicate glasses provide relatively strong-field sites, while fluoride and fluorozirconate glasses provide weak-field sites only and the luminescence observed is only a broad ⁴T₂ → ⁴A₂ band with a large Stokes red-shift.^19,20,28 Accordingly, when compared with O-coordinated Cr³⁺ (717 nm), the PL emission of F-coordinated Cr³⁺ ions show a red-shift to 1 μm and shorter PL lifetimes. Such a situation was also clearly observed with the excitation bands. When monitored at 1 μm, the three broad excitation bands show red-shifts to longer wavelengths when compared with those monitored at 717 nm. In addition, the charge transfer band monitored by the 1 μm emission was much weaker than those monitored by the 717 nm emission. This can be deduced from the larger electron density related to O²⁻ compared to F⁻.

3.4 Temperature sensing performance

To explore the possible applications of dual-model temperature-dependent decay lifetimes in optic temperature sensors, the PL decay curves for the Cr³⁺-doped GC640 sample from room temperature to about 300 °C (Fig. 4(a) and (b)) were recorded by monitoring at 717 nm and 1 μm. The evaluated lifetimes for both Cr³⁺ sites appear to decrease with temperature (Fig. 4(c) and (d)). Such temperature dependent relationships are mainly due to the thermally activated repopulation between the ²E and ⁴T₂ states. The ²E → ⁴A₂ transition is doubly forbidden by parity and spin, which has a longer decay time than the ⁴T₂ → ⁴A₂ transition (spin-allowed).

Fig. 4 The 296 nm excited 717 nm luminescence decay curves (a) and the 464 nm excited 1 μm luminescence decay curves (b) obtained for GC640 at different temperature. The fitted relationships of the luminescence lifetimes of Cr³⁺ and temperature (c and d), where the insets show the sensitivity at the investigated temperature ranges.

For O-coordinated sites, the ²E, ²T₁, ⁴T₂ and ⁴A₂ energy states were proposed as interrelated, as Fig. 5(a) illustrates, in a configuration coordinate model, where ²E serves as the lowest excited state and easily intersects with the upper state, ⁴T₂, by the assistance of only a few phonons. Upon increasing the temperature, more electrons repopulate from ²E to ⁴T₂, resulting in short PL decay lifetimes.⁹ Theoretically, the total number of excited ions can be expressed as: n = n_E + n_T, where n_E and n_T represent the number of ions at ²E and ⁴T₂,respectively. The particles at the two levels follow the Boltzmann distribution: where C is the degeneration ratio of ²E to ⁴T₂, with a value of 3. The decay rate of the total excited state ions is represented by the following expression: Therefore, by solving the equation: the lifetime was obtained as:

Fig. 5 Simplified energy level diagrams of Cr³⁺ at the different sites in the investigated glass-ceramics: (a) O-coordinated sites and (b) F-coordinated sites.

The sensitivity could be calculated using the following equation:

Thereby, the PL lifetime data of 717 nm from 298.35 K to 573.25 K were well fitted as the solid line using the least-square method, as shown in Fig. 4(c). The correlation coefficient (R²) reached 0.99. The parameters τ_E, τ_T and ΔE were determined to be 1.835 ms, 0.008 ms and 2044.5 cm⁻¹, respectively. In addition, the sensitivity curve reaches a maximum of 0.76% K⁻¹ at 498 K, as shown in the inset of the Fig. 4(c).

For the F-coordinated sites, the ²E, ²T₁, ⁴T₂ and ⁴A₂ energy states were proposed as interrelated using a configuration coordinate model as illustrated in Fig. 5(a), where ⁴T₂ serves as the lowest excited state. The large Stokes shift and weak crystal field at the F-coordinated sites enhance the decay rate of (W_NR) from the ⁴T₂ excited state to the ⁴A₂ ground state. It thus becomes a predominant factor to significantly reduce the 1 μm PL lifetime as shown in Fig. 4(b). Theoretically, the Mott–Seitz model was used to describe this mechanism quantitatively.²⁹ The total transition probabilities of the ⁴T₂ state are comprised of radiative (W_R) and non-radiative (W_NR) transition probabilities:

where τ_R is the radiative lifetime and τ_NR is the non-radiative lifetime. As illustrated in Fig. 5(b), an electron in the ⁴T₂ state would cross the barrier ΔE to the ground state by absorbing a certain amount of phonons.³⁰ We assumed that τ_R was independent of temperature, while τ_NR is dependent of temperature; the relationship between τ_NR and temperature can be expressed using the Arrhenius equation:where τ_NR(0) is the non-radiative decay time at 0 K, ΔE is the energy barrier mentioned above and k_B is the Boltzmann constant. Accordingly, when the temperature increases, the non-radiative decay will become stronger. Through the deformation of the formulae (7) and (8), the total (experimentally measured) lifetime can be written as:

Then, the sensitivity was calculated using the following equation:

We used this equation to fit the lifetime data of the 1 μm emission observed for the GC640 sample. The solid line shown in Fig. 4(d) was the least-squares curve fitting to the lifetime data from 298.25 K to 498.15 K. Impressively, the correlation coefficient (R²) reached as high as 0.999. This not only expounded the temperature-dependence of the lifetime in the experimental temperature range but also guaranteed the effectiveness of the temperature sensing. Through the fitting, the corresponding parameters τ_R, τ_NR and ΔE were determined to be 32.814 μs, 0.997 μs and 898.3 cm⁻¹, respectively. In addition, the sensitivity curve reaches a maximum of 0.47% K⁻¹ at 351 K as shown in the inset of the Fig. 4(c).

Based on the abovementioned temperature sensing parameters, comparable performances could be expected for optical thermometries upon introducing the Cr³⁺-doped fluorosilicate glass-ceramic materials. As a comparison, the temperature sensing parameters of some typical materials reported in the literature are listed in Table 2. The O-coordinated Cr³⁺ in the glass-ceramics exhibit maximum sensitivity (S_max = 0.76% K⁻¹), which was slightly lower than that observed for LiAl₅O₈ : Cr³⁺, but still better than those reported for Y₃Al₅O₁₂ : Cr³⁺, Al₂O₃ : Cr³⁺ (ruby) and Ga₂O₃ : Cr³⁺ in glass-ceramics and with a higher work temperature (T_max = 498 K). The F-coordinated Cr³⁺ in the glass-ceramics showed the maximum sensitivity (S_max = 0.47% K⁻¹) at T_max = 351 K, which was similar to that observed in Al₂O₃ : Cr³⁺ (ruby), so it could also be a good candidate material for optical thermometry. When compared with crystalline and ceramic materials, the glass-ceramics have the further advantages of good designability, easy molding, cost-efficient and so on. Therefore, these types of Cr³⁺-doped glass-ceramics are very promising candidates for developing a type of dual mode optical thermometry.

Table 2 The optical thermometry performance of some Cr³⁺ ion-doped materials

Materials	ΔE (cm^−1)	Tmax (K)	Smax (% K^−1)	Ref.
Y3Al5O12 : Cr^3+	—	—	0.50	31
LiAl5O8 : Cr^3+	—	447	0.83	32
Ruby (Al2O3 : Cr^3+)	1637	390	0.48	33
LiSrAlF6 : Cr^3+	4557	333	1.80	34
Ga2O3 : Cr^3+ (in glass-ceramics)	1215	386	0.59	5
O-coordinated Cr^3+ (in glass-ceramics)	2044	498	0.76	This work
F-coordinated Cr^3+ (in glass-ceramics)	898	351	0.47	This work

---

4. Conclusion

In fluorosilicate glass-ceramics comprising homogenous cubic CaF₂ nanocrystals, Cr³⁺ions occupied two different types of sites: one was the O-coordinated octahedral sites in the residual glass phase and the other was the F-coordinated octahedral sites were distributed not only in the residual glass phase but also in the nanocrystalline CaF₂ phase. A certain number of Cr³⁺ions entered the precipitated CaF₂ nanocrystals, occupying the octahedral interstice and induced an expansion of the fluorite cube lattices. The 717 nm emission was derived from the O-coordinated octahedral sites, while the 1 μm super-broad emission was assigned to the F-coordinated octahedral sites. Upon annealing, the crystal size and crystallinity of the CaF₂ nanocrystals could be enlarged, thus the F-coordinated octahedral Cr³⁺ sites were enhanced and the PL at 1 μm became stronger.

The O-coordinated sites were relatively strong crystal field sites, so the 717 nm emission originates from the radiative transitions of the two thermally coupled ²E and ⁴T₂ energy levels of Cr³⁺. In contrast, the F-coordinated sites were relatively weak-field sites, thus ⁴T₂ alternatively served as the lowest excited state and a low vibrational level of ⁴T₂ was easily tunneled to a high vibrational level of the ground state as a result of a thermal quench in non-radiative transition. Accordingly, the temperature dependent PL lifetimes could be theoretically described by eqn (5) and (9). By mean square fitting methods, the maximum relative temperature sensitivity coefficients were 0.76% K⁻¹ at 498 K for the 717 nm PL lifetime and 0.47% K⁻¹ at 573 K for the 1 μm PL lifetime. These results are comparable with some other typical optical temperature sensing materials, providing evidence of the potential to apply glass-ceramics in highly sensitive, dual mode, self-calibrated PL-lifetime-based temperature sensing.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (No. 51672243), the Program for International S&T Cooperation Projects of China (No. 2014DFB50100), Zhejiang Provincial Natural Science Foundation of China (No. LY16E020003) and the Fundamental Research Funds for the Central Universities (No. 2016QNA4005; No. 2016FZA4007). The authors are very grateful to the support of their research team and for involving one of the authors (Shuo Cui) in the sample preparation and spectral measurements. Associate Prof. Xvsheng Qiao gave a detailed and patient guidance to this work.

References

1. X. Wang, Q. Liu, Y. Bu, C. S. Liu, T. Liu and X. Yan, RSC Adv., 2015, 105, 86219–86236.
2. X. D. Wang, O. S. Wolfbeis and R. J. Meier, Chem. Soc. Rev., 2013, 42, 7834–7869.
3. A. H. Khalid and K. Kontis, Sensors, 2008, 8, 5673–5744.
4. M. D. Chambers and D. R. Clarke, Annu. Rev. Mater. Res., 2009, 39, 325–359.
5. D. Chen, Z. Wan and Y. Zhou, Opt. Lett., 2015, 40, 3607–3610.
6. A. K. Singh, P. Shahi, S. B. Rai and B. Ullrich, RSC Adv., 2015, 21, 16067–16073.
7. H. Lu, H. Hao, G. Shi, Y. Gao, R. Wang, Y. Song, Y. Wang and X. Zhang, RSC Adv., 2016, 60, 55307–55311.
8. S. Liu, S. Liu, M. Zhou, X. Ye, D. Hou and W. You, RSC Adv., 2017, 59, 36935–36948.
9. Z. Zhang, K. T. V. Grattan and A. W. Palmer, Phys. Rev. B, 1993, 48, 7772–7778.
10. Y. L. Hu, Z. Y. Zhang, K. T. V. Grattan, A. W. Palmer and B. T. Meggitt, Sens. Actuators, A, 1997, 63, 85–90.
11. Y. R. Shen and K. L. Bray, Phys. Rev. B, 1997, 56, 10882–10891.
12. H.-H. Schmidtke and J. Degen, in Stereochemistry and Bonding, Springer, 1989, pp. 99–124.
13. A. Luci, T. Castrignano, U. Grassano, M. Casalboni and A. Kaminskii, Phys. Rev. B, 1995, 51, 1490.
14. D. Chen, Z. Wan, Y. Zhou, X. Zhou, Y. Yu, J. Zhong, M. Ding and Z. Ji, ACS Appl. Mater. Interfaces, 2015, 7, 19484–19493.
15. M. J. Dejneka, J. Non-Cryst. Solids, 1998, 239, 149–155.
16. Y. Wang and J. Ohwaki, Appl. Phys. Lett., 1993, 63, 3268–3270.
17. C. Liu and J. Heo, J. Am. Ceram. Soc., 2012, 95, 2100–2102.
18. P. I. Macfarlane, K. Holliday, J. Nicholls and B. Henderson, J. Phys.: Condens. Matter, 1995, 7, 9643.
19. B. Henderson, M. Yamaga, Y. Gao and K. O'Donnell, Phys. Rev. B, 1992, 46, 652.
20. F. Rasheed, K. O'Donnell, B. Henderson and D. Hollis, J. Phys.: Condens. Matter, 1991, 3, 3825.
21. J. Zhao, R. Ma, X. Chen, B. Kang, X. Qiao, J. Du, X. Fan, U. Ross, C. Roiland and A. Lotnyk, J. Phys. Chem. C, 2016, 120, 17726–17732.
22. S. Sugano, Multiplets of transition-metal ions in crystals, Elsevier, 2012.
23. B. Henderson and G. F. Imbusch, Optical spectroscopy of inorganic solids, Oxford University Press, 2006.
24. J. Fernández, M. Illarramendi, R. Balda, M. Arriandiaga, J. Lucas and J. Adam, J. Non-Cryst. Solids, 1991, 131, 1230–1234.
25. Y. Tanabe and S. Sugano, J. Phys. Soc. Jpn., 1954, 9, 753–766.
26. Y. Tanabe and S. Sugano, J. Phys. Soc. Jpn., 1954, 9, 766–779.
27. M. Casalboni, V. Ciafardone, G. Giuli, B. Izzi, E. Paris and P. Prosposito, J. Phys.: Condens. Matter, 1996, 8, 9059.
28. F. Rasheed, K. O'Donnell, B. Henderson and D. Hallis, J. Phys.: Condens. Matter, 1991, 3, 1915.
29. L. Bøtter-Jensen, S. W. McKeever and A. G. Wintle, Optically stimulated luminescence dosimetry, Elsevier, 2003.
30. M. Ren, C. D. Brites, S.-S. Bao, R. A. Ferreira, L.-M. Zheng and L. D. Carlos, J. Mater. Chem. C, 2015, 3, 8480–8484.
31. H. Aizawa, T. Katsumata, Y. Kiyokawa, T. Nishikawa, T. Sasagawa, S. Komuro, T. Morikawa, H. Ishizawa and E. Toba, Measurement, 2006, 39, 147–152.
32. X. Li, G. Jiang, S. Zhou, X. Wei, Y. Chen, C. K. Duan and M. Yin, Sens. Actuators, B, 2014, 202, 1065–1069.
33. H. Seat and J. Sharp, IEEE Trans. Instrum. Meas., 2004, 53, 140–154.
34. Z. Zhang, K. Grattan and A. Palmer, Optical Fiber Sensors Conference, IEEE Xplore, 1992, pp. 93–96.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c7ra10864h

---