Phase structure control and optical spectroscopy of rare-earth activated GdF₃ nanocrystal embedded glass ceramics via alkaline-earth/alkali-metal doping

Abstract

Hexagonal to orthorhombic phase transformation of GdF₃ nanocrystals in bulk glass ceramics was achieved through alkaline-earth/alkali-metal doping and crystallization temperature controlling. Structural characterizations and spectroscopic analyses of the Eu³⁺ probe evidenced the incorporation of rare earth emitting-centers into the precipitated GdF₃ crystals among the glass matrix. In addition, the influence of phase evolution on the upconversion luminescence of Er³⁺/Yb³⁺ co-doped glass ceramics was systematically investigated and it was evidenced that the upconversion intensity of the orthorhombic GdF₃ embedded glass ceramic was two orders of magnitude higher than that of the hexagonal GdF₃ containing glass ceramic. Benefiting from greatly enhanced upconversion luminescence after glass crystallization, the present glass ceramic composites were demonstrated to have promising applications in optical temperature sensors as well as tunable displays.

---

Introduction

Currently, rare earth (RE) ion doped optical materials have attracted great interest due to their potential applications in solid-state lasers, bio-imaging, solar cells, displays, catalysis and sensors etc.^1–10 Generally, the luminescence efficiency of these materials is bound up with the dynamics of RE activators which is restricted by the environments around the RE ions.^11,12 Hence, it is effective to enhance RE luminescence efficiency by searching for appropriate hosts. Among a large number of materials, fluorides are known to be suitable media owing to their low phonon energy and high solubility for RE dopants. Unfortunately, the instability in chemicals and the complexity in fabrication seriously restrict their practical applications. On the other hand, oxide glasses, which have high chemical and mechanical stability, suffer from large phonon energies ascribed to the stretching vibration of network-forming oxide. To satisfy these problems, oxyfluoride glass ceramics (GC) emerged as required and received widespread attention.^13–22 Such composites combine the merits of excellent mechanical strength, high chemical durability and thermal stability of oxide glasses, and low phonon energy of fluoride crystals. Oxyfluoride GC, made up with fluoride nanocrystals (NCs) embedding among an oxide glassy matrix, are generally achieved by controlled crystallization of the precursor glass with appropriate chemical compositions. And the partition of the optically active RE ions into the precipitated fluoride lattice plays an important role to improve optical performance.¹⁵

Lanthanide trifluorides (LnF₃, Ln = La–Lu) with phonon energy lower than 400 cm⁻¹, where the Ln host ions could be easily substituted by RE activators with the same valence, are very attractive media for doping RE emitting centers.^23,24 Recently, the structures and optical properties of the RE doped LaF₃ and YF₃ NCs embedded glass ceramics have been well studied.^25–29 In addition, the general nanocrystallization behaviors of the lanthanide trifluorides from LaF₃ to LuF₃ in the same aluminosilicate glass matrix were also reported.³⁰ In fact, Thoma and Brunton et al. found that the LnF₃ exhibited hexagonal structure at high temperature, and they undergone a rapid transition into orthorhombic one upon cooling.³¹ However, as far as we know, there is no investigation on the phase evolution from hexagonal to orthorhombic for RE-doped GdF₃ NCs in glass ceramics and the related optical spectroscopy so far.

In this paper, hexagonal-to-orthorhombic phase evolution of GdF₃ NCs in the glass matrix was realized by designing precursor glass composition with the addition of alkaline-earth/alkali-metal fluorides and controlling glass crystallization temperature. Using Eu³⁺ as probe ions, complete structural analyses were carried out to investigate local environment around RE activators. The influence of phase evolution on the upconversion (UC) performance of Er³⁺/Yb³⁺ codoped GC samples were systematically studied. Finally, the fabricated glass ceramics were demonstrated to have potential applications in optical thermometry as well as tunable display.

Experimental

The precursor glass (PG) compositions (in mol%) of SiO₂–Al₂O₃–GdF₃–MF₂/AF (M = Mg, Ca, Sr, Ba; A = Li, Na, K), the crystallization condition as well as the corresponding crystallization phases were tabulated in Table 1. The RE activators were introduced into glasses by addition of rare earth fluorides, such as EuF₃, YbF₃, ErF₃, HoF₃ and TmF₃. For each batch, about 15 g raw materials were thoroughly mixed and melted in a covered alumina crucible at 1590 °C for 60 min to achieve a homogeneous melting, then was poured into a 300 °C preheated copper mold to cool down to room temperature. The as-prepared PGs were then cut into 5 mm² square coupons and were crystallized by heat-treating at 550, 600, 650, 700 and 750 °C for 2 h to obtain GCs.

Table 1 Chemical compositions of the investigated glass samples, crystallization condition and the corresponding crystallization phase

Samples	Glass composition (mol%)	Crystallization condition	Crystallization phase
GC750-11Mg	70SiO2–7Al2O3–7GdF3–11MgF2	750 °C/2 h	Amorphous
GC750-11Ca	70SiO2–7Al2O3–7GdF3–11CaF2	750 °C/2 h	Amorphous
GC750-6Sr	70SiO2–7Al2O3–7GdF3–6SrF2	750 °C/2 h	Amorphous
GC750-11Sr	70SiO2–7Al2O3–7GdF3–11SrF2	750 °C/2 h	Hexagonal
GC550-16Sr	70SiO2–7Al2O3–7GdF3–16SrF2	550 °C/2 h	Amorphous
GC600-16Sr	70SiO2–7Al2O3–7GdF3–16SrF2	600 °C/2 h	Amorphous
GC650-16Sr	70SiO2–7Al2O3–7GdF3–16SrF2	650 °C/2 h	Hexagonal
GC700-16Sr	70SiO2–7Al2O3–7GdF3–16SrF2	700 °C/2 h	Hexagonal
GC750-16Sr	70SiO2–7Al2O3–7GdF3–16SrF2	750 °C/2 h	Hexagonal
GC750-11Ba	70SiO2–7Al2O3–7GdF3–11BaF2	750 °C/2 h	Hexagonal
GC600-Li	70SiO2–7Al2O3–7GdF3–22LiF	600 °C/2 h	Hexagonal
GC650-Li	70SiO2–7Al2O3–7GdF3–22LiF	650 °C/2 h	Hexagonal & orthorhombic
GC700-Li	70SiO2–7Al2O3–7GdF3–22LiF	700 °C/2 h	Hexagonal & orthorhombic
GC750-Li	70SiO2–7Al2O3–7GdF3–22LiF	750 °C/2 h	Orthorhombic
GC750-Na	70SiO2–7Al2O3–7GdF3–22NaF	750 °C/2 h	Hexagonal & orthorhombic
GC750-K	70SiO2–7Al2O3–7GdF3–22KF	750 °C/2 h	Hexagonal & orthorhombic

---

To identify the crystallization phase and determine the mean size of the crystallites, X-ray diffraction (XRD) analysis was carried out with a powder diffractometer (DMAX2500 RIGAKU) using Cu Kα radiation (λ = 0.154 nm). The microstructures of the samples were studied using a transmission electron microscope (TEM, JEM-2010) equipped with the selected area electron diffraction (SAED). The samples were ground into a very fine powder that was placed onto a carbon coated copper grid and introduced into the microscope. High resolution TEM (HRTEM) images were performed to observe atomic planes inside the crystalline lattices. Scanning transmission electron microscopy (STEM) observation in the high-angle annular dark-field (HAADF) mode, being sensitive to the atomic number (Z) of the sample (scaling proportionally to ∼Z²),³² is carried out on an FEI aberration-corrected Titan Cubed S-Twin transmission electron microscope operated at 200 kV. The emission, excitation spectra and decay curves of the Eu³⁺ doped PG and GC samples were recorded on an Edinburgh Instruments FS5 spectrofluorometer equipped with both continuous (150 W) and pulsed xenon lamps. UC emission spectra of the PG and GC samples were detected using the Hamamatsu R943-02 photomultiplier tube and the Spex 1000M monochromator with an adjustable laser diode (980 nm) as the excitation source. Visible UC decay curves of Er³⁺ in the PG and GC samples were measured with a customized UV to mid-infrared steady-state and phosphorescence lifetime spectrometer (FSP920-C, Edinburgh) equipped with a digital oscilloscope (TDS3052B, Tektronix) and a tunable mid-band OPO pulse laser as excitation source (410–2400 nm, 10 Hz, pulse width ≤ 5 ns, Vibrant 355II, OPOTEK). The corresponding lifetimes were evaluated via the equation of because of non-single-exponential characteristics of these decay curves, where I_p is the peak intensity and I(t) is the time-related luminescence intensity. The temperature dependent Er³⁺ green UC emission spectra were recorded on an Edinburgh Instruments FS5 spectrofluorometer equipped with a homemade temperature controlling stage and a 980 nm laser diode.

Results and discussion

XRD patterns of the as-quenched PG (with composition of SiO₂–Al₂O₃–GdF₃–SrF₂) and the GC samples obtained by heat-treated at various temperatures are shown in Fig. 1a. It exhibits a typical amorphous structure in the precursor glass, and still no sharp peaks can be observed after the heat treatment at 550 °C and 600 °C. With increase of heating temperature, several diffraction peaks corresponding to the hexagonal GdF₃ appear and tend to be sharper. Since the XRD data of hexagonal GdF₃ have not been reported, we indexed this phase according to the isostructural SmF₃ (JCPDS no. 05-0563). Fig. 1b presents XRD patterns of glass ceramics with different SrF₂ content heat-treated at 750 °C for 2 h. XRD pattern of GC750-6Sr sample is the amorphous diffuse humps like precursor glass while intense diffraction peaks can be observed when SrF₂ content is increased, which indicates that lower SrF₂ content goes against GdF₃ crystallization. Furthermore, the impact of different alkaline-earth fluorides (MgF₂, CaF₂, SrF₂ and BaF₂) doping on glass crystallization was investigated, as exhibited in Fig. 1c. Obviously, no crystallization phase was detected for the MgF₂ doped glass after heat-treatment at 750 °C for 2 h, and hexagonal GdF₃ crystals precipitate from glass matrix for all the other samples. Notably, the degree of GdF₃ crystallization in the CaF₂ doped glass is lower than those in the SrF₂ and BaF₂ doped glasses.

Fig. 1 XRD patterns of (a) 16 mol% SrF₂ (b) 6 mol%, 11 mol%, 16 mol% SrF₂ (c) 11 mol% MF₂ (M = Mg, Ca, Sr, Ba) doped samples heat-treated at various temperatures. The standard crystalline diffraction data of hexagonal SmF₃ (JCPDS no. 05-0563) are provided in (a).

Fig. 2a shows XRD patterns of LiF doped glass ceramics heat-treated at 600–750 °C for 2 h. Hexagonal GdF₃ crystals were obtained for crystallization temperature lower than 600 °C. Further increasing temperature (650–700 °C) resulted in the formation of both hexagonal and orthorhombic GdF₃ crystals, and finally all the hexagonal GdF₃ crystals were converted into orthorhombic ones (JCPDS no. 49-1804) when the crystallization temperature reached as high as 750 °C. In addition, XRD patterns of diverse alkali-metal fluorides (LiF, NaF, KF) doped glass ceramics prepared by heat-treated at 750 °C for 2 h are shown in Fig. 2b. Different to the case of LiF doped GC, both hexagonal and orthorhombic GdF₃ crystalline phase coexist in the NaF and KF doped GC samples obtained by heat-treating at 750 °C.

Fig. 2 XRD patterns of (a) 22 mol% LiF, (b) 22 mol% AF (A = Li, Na, K) doped glass ceramic samples heat-treated at various temperatures. The standard crystalline diffraction data of hexagonal LaF₃ and (JCPDS no. 32-0483) and orthorhombic GdF₃ (JCPDS no. 49-1804) are provided in (a).

The microstructure of the glass ceramic samples was studied by electron microscopy. TEM bright field image of GC750-11Sr sample in Fig. 3a shows that nearly spherical GdF₃ NCs with the size of 15–20 nm homogenously precipitated from the glass matrix. The corresponding selected area electron diffraction (SAED) pattern exhibits discrete polycrystalline diffraction rings, confirming the formation of hexagonal GdF₃ particles after crystallization. High-resolution TEM (HRTEM) image for an individual GdF₃ NC (Fig. 3b) verifies its single-crystalline nature with high-crystallinity. The lattice fringes are clearly resolved, and a typical d-spacing around 3.1 Å is observed, corresponding to the (111) plane of hexagonal GdF₃ phase. Similarly, TEM image and SAED pattern of the GC750-Li sample (Fig. 4a) evidence that orthorhombic GdF₃ nanoparticles with the size of 40–50 nm monodispersely distributed among glass matrix. HRTEM observation on an individual orthorhombic GdF₃ particle demonstrates its high crystallinity and well-resolved (210) lattice plane (Fig. 4b).

Fig. 3 (a) TEM image of GC750-11Sr sample and the corresponding SEAD pattern. (b) HRTEM micrograph of an individual particle. (c) STEM-HAADF image of GC750-11Sr sample with associated (d) F (red), (e) Gd (orange), (f) Yb (yellow) and (g) Er (green) elemental mappings.

Fig. 4 (a) TEM images of GC750-Li sample and the corresponding SEAD pattern. (b) HRTEM micrograph of an individual particle. (c) STEM-HAADF image of GC750-Li sample with associated (d) F (red), (e) Gd (orange), (f) Yb (yellow) and (g) Er (green) elemental mappings.

Furthermore, scanning transmission electron microscopy (STEM) operated in the high-angle annular dark-field (HAADF) mode, which is sensitive to the atomic number (Z) of the sample (scaling proportionally to ∼Z²),³² is adopted to investigate the microstructures of GC samples. Note that for STEM-HAADF images, contrary to previous TEM ones, the particles appear with a brighter contrast than the aluminosilicate glass matrix (Fig. 3c and 4c) due to the much larger atomic number of Gd (Z = 64) segregated in the GdF₃ particles than those of Si/Al (Z = 14/13) distributed in the glass matrix. The F, Gd, Yb and Er STEM elemental mappings for the GC750-11Sr (Fig. 3d–g) and GC750-Li (Fig. 4d–g) samples reveal the localization of Gd, Yb and Er in the GdF₃ NCs, whereas the equal presence of F in the particles and the glass matrix. These results indicate the incorporation of Yb³⁺ and Er³⁺ activators into GdF₃ crystalline lattice.

In the SiO₂–Al₂O₃–GdF₃–MF₂/AF (M = Mg, Ca, Sr, Ba; A = Li, Na, K) glasses, the glass network can be considered to consist of [SiO₄] and [AlO₄] tetrahedral, and the network modifying ions, such as Li⁺, Sr²⁺ and Gd³⁺, are located in the interstices of the glass framework.³³ The occurrence of crystallization of the specific phase in the glass is determined by the interactions among the ions of the glass components. The ionicity of the M–O or A–O bond is greater than that of the Gd–O one,³⁴ thus the affinity of M²⁺ and A⁺ to O²⁻ is stronger than that of Gd³⁺ in the aluminosilicate network. The ionicity of the bonds of the glass components with F⁻ increases in the sequence of Si–F < Al–F < Gd–F < M–F < A–F.³⁴ As M²⁺ and A⁺ ions tightly bond with O²⁻, Gd³⁺ is therefore the next element having the highest affinity to F⁻. This makes the GdF₃ crystallization in the interstices of the aluminosilicate framework possible. Compared with the orthorhombic one, the hexagonal phase requires smaller interstitial space owing to its dense-packing structure. As a result, in the alkaline-earth (especially Sr²⁺ and Ba²⁺ with large ionic radii) doped glasses, i.e., hexagonal GdF₃ particles are easier to precipitate from glass matrix, while in the alkali-metal (especially Li⁺ with small ionic radii) doped glasses, the interstitial spaces of the glass framework tend to be able to accommodate the nucleation and growth of orthorhombic GdF₃ particles.

To discern the surroundings of RE activators in GC samples, room temperature excitation and emission spectra of Eu³⁺, acting as a structural probe, in both hexagonal GdF₃ embedded GC and orthorhombic GdF₃ contained GC, are recorded, as presented in Fig. 5 and 6. Two sets of excitation signals can be detected in the excitation spectra according to Fig. 5a and 6a by monitoring 592 nm (Eu³⁺: ⁵D₀ → ⁷F₁) emission: one consists of several characteristic Eu³⁺ excitation peaks from the ⁷F₀ ground-state to the indexed excited-states, and the other consists of two excitation bands at 272 nm (Gd³⁺: ⁸S_7/2 → ⁶I_J) and 310 nm (Gd³⁺: ⁸S_7/2 → ⁶P_J) respectively, which demonstrates the existence of energy transfer from Gd³⁺ to Eu³⁺. Impressively, the dominant peak for Eu³⁺: ⁷F₀ → ⁵L₆ excitation band locates in 393 nm for PG while it converts into 396 nm for hexagonal GdF₃ embedded GC (GC750-Sr, Fig. 5a), indicating the modification of Eu³⁺ crystal-field environment. In the emission spectra (Fig. 5b), typical Eu³⁺ emission lines of 4f–4f transitions from the excited states to the lower ones, i.e., ⁵D_3,2,1 → ⁷F_J and ⁵D₀ → ⁷F_J (J = 0–4) are observed. Compared to PG (393 nm excitation), the spectrum of GC750-Sr (396 nm excitation) significantly changes, i.e., the emission bands become Stark-splitting and narrowed, the ⁵D_3,2,1 → ⁷F_J emission intensities greatly enhance, and the ratio of ⁵D₀ → ⁷F₂ electric dipolar transition to ⁵D₀ → ⁷F₂ magnetic dipolar one greatly decreases. All these results evidence the alternation of Eu³⁺ environment from amorphous glass to hexagonal GdF₃ crystals with high crystal-field symmetry and low phonon energy (<400 cm⁻¹) after crystallization treatment.²⁴ Considering the approximate ionic radius between Eu³⁺ and Gd³⁺, Eu³⁺ activator is believed to substitute Gd³⁺ site in the GdF₃ host. Notably, the emission spectrum of GC750-Sr obtained under 393 nm excitation is quite different to that under 396 nm excitation (Fig. 5b), suggesting that there are still some residual Eu³⁺ ions staying in glass matrix after crystallization. As for LiF-doped GCs, significantly spectral changes are found when hexagonal GdF₃ phase is transformed into orthorhombic one with increase of crystallization temperature (Fig. 6b). The crystal-like emission spectrum of Eu³⁺ in GC750-Li under indirect excitation of Gd³⁺ (272 nm) confirmed the incorporation of Eu³⁺ in orthorhombic GdF₃ crystal lattice. Finally, the experimental decay lifetimes for the Eu³⁺ excited states provides important information to evaluate the environments of the Eu³⁺ ions. Fig. 7 shows the fluorescence decay curves of the ⁵D₀ level for the PG and GC samples by detecting 592 nm emission (Eu³⁺: ⁵D₀ → ⁷F₁ transition). Apparently, the obtained lifetime increases obviously after glass crystallization owing to the partition of Eu³⁺ ions into the hexagonal/orthorhombic GdF₃ crystals with low-phonon-energy.

Fig. 5 Room temperature (a) excitation and (b) emission spectra of the Eu³⁺ doped PG and GC containing hexagonal GdF₃ NCs.

Fig. 6 Room temperature (a) excitation and (b) emission spectra of the Eu³⁺ doped GCs containing orthorhombic GdF₃ NCs.

Fig. 7 Fluorescence decay curves of the Eu^{3+ 5}D₀ level for (a) the hexagonal GdF₃ embedded GC, (b) orthorhombic GdF₃ contained GC and the corresponding PGs.

In the further experiment, the influence of the glass crystallization on the UC optical performance of the Er³⁺/Yb³⁺ co-doped PG and GC samples with the addition of alkaline-earth/alkali-metal ions is investigated, as shown in Fig. 8. Under 980 nm near-infrared (NIR) laser excitation, both green (²H_11/2, ⁴S_3/2 → ⁴I_15/2) and red (⁴F_9/2 → ⁴I_15/2) UC emissions of Er³⁺ appear in the emission spectra of all the PG and GC samples. In comparison to PG, significant enhancement of the UC intensity after glass crystallization (Fig. 8a and c) certifies the preferential partition of the optically active centers into the precipitated GdF₃ crystalline phase, which greatly reduces non-radiative relaxation probabilities of Er³⁺ owing to the low-phonon-energy crystalline environments. For example, the UC intensities of GC750-16Sr and GC750-Li GC samples are about 40 and 1000 times as high as those of the corresponding PGs. As revealed in the inset of Fig. 8a, as crystallization temperature is elevated, UC intensity gradually increases and red to green emission intensity ratio monotonously decreases, attributing to the improved crystallinity and increased size of hexagonal GdF₃ NCs. Generally, the greater non-radiative multi-phonon relaxation probabilities from the Er³⁺: ²H_11/2/⁴S_3/2 and ⁴F_9/2 states to the lower ones in GC sample with smaller NCs can be expected as they have a high surface-to-volume ratio and therefore a larger proportion of Er³⁺ emitting-centers are located on the NC surfaces and under the influence of high energy vibrations of aluminosilicate glass groups.^35,36

Fig. 8 UC emission spectra of the Yb/Er doped PG and GC samples with the addition of: (a) 16 mol% SrF₂, (b) 11 mol% MF₂ (M = Ca, Sr, Ba), (c) 22 mol% LiF, and (d) 22 mol% AF (A = Li, Na, K). Inset of (a) shows crystallization temperature dependent Er³⁺ integrated UC intensity as well as red to green UC intensity ratio. (e) Comparison of UC emission spectra between GC750-11Sr sample and GC750-Li one. Insets of (e) show the corresponding UC luminescent photographs of GC samples.

The impact of diverse alkaline-earth and alkali-metal dopants on UC performance of GC samples has also been investigated. As shown in Fig. 8b, the UC emission intensities of GC750-11Ba and GC750-11Sr samples are far higher than those of GC750-11Ca and GC750-11Mg ones, owing to the formation of a large amount of hexagonal Yb/Er: GdF₃ NCs in the former two GC samples. For the LiF doped GC samples (Fig. 8c), it can be observed that the UC emission spectra of GC700-Li and GC750-Li samples are completely different to those of GC600-Li and GC650-Li ones, confirming the formation of different phases, i.e., orthorhombic GdF₃ and hexagonal GdF₃, respectively. Clearly, with the conversion from hexagonal to orthorhombic, UC emission intensity is greatly intensified for all the LiF, NaF and KF doped GC samples (Fig. 8c and d). The comparison of UC emission spectra between GC750-Li and GC750-11Sr under the identical measuring condition (Fig. 8e) evidenced that the intensity of the former with orthorhombic GdF₃ embedded in glass matrix has two orders of magnitude (about 400 times) higher than that of the latter with hexagonal GdF₃ embedded in glass matrix. Notably, such enhanced luminescence should be attributed to the combined effects of different phase structures as well as crystal sizes.

Time-resolved decay behaviors were investigated to reveal the influence of structural variation of GC on UC luminescence. All the lifetime values are evaluated and tabulated in Table 2. For SrF₂ doped GC samples, the trend of lifetime variation is consistent with that of UC intensity variation: the lifetimes for the Er³⁺: ²H_11/2, ⁴S_3/2 green-emitting states and Er³⁺: ⁴F_9/2 red-emitting one become longer with increase of crystallization temperature, as exhibited in Fig. 9a and b. For LiF doped GC samples, the lifetimes have the same trend as the SrF₂ doped samples when temperature is not exceeding 650 °C. However, up to 700 °C, the lifetimes of samples become shorter with increase of crystallization temperature (Fig. 9c and d, Table 2). Such decay behavior is believed to attribute to the phase transition from hexagonal GdF₃ to orthorhombic one with increase of temperature. As demonstrated in the insets of Fig. 9c and d, the lifetimes of both green and red emitting states in the GC750-Li GC embedded with orthorhombic GdF₃ NCs are lower than those in the GC750-11Sr GC embedded with hexagonal GdF₃ NCs even though the UC intensity of the former GC is far higher than that of the latter one. It was well-known that the UC luminescence was highly dependent on the crystal phase structure, size and morphology. In our samples, the sizes (40–50 nm) of orthorhombic GdF₃ spherical NCs in GC750-Li are larger than those (15–20 nm) of hexagonal GdF₃ spherical NCs in GC750-Sr sample. Therefore, the shorten lifetime in GC750-Li should be due to the hexagonal-to-orthorhombic phase transition of GdF₃ in glass matrix.

Table 2 UC lifetimes for ²H_11/2, ⁴S_3/2 states and ⁴F_9/2 one for PG and GC samples

Samples	^2H11/2, ^4S3/2 (ms)	^4F9/2 (ms)
PG-16Sr	0.129	0.287
GC650-16Sr	0.359	0.506
GC700-16Sr	0.417	0.567
GC750-16Sr	0.656	0.848
PG-Li	0.104	0.201
GC600-Li	0.299	0.502
GC650-Li	0.326	0.528
GC700-Li	0.259	0.463
GC750-Li	0.221	0.365

---

Fig. 9 Green (²H_11/2, ⁴S_3/2 states, λ_em = 542 nm) and (b) red (⁴F_9/2 state, λ_em = 655 nm) UC decay curves for (a and b) SrF₂ doped and (c and d) LiF doped PG and GC samples prepared by crystallization at various temperatures. Insets of (c and d) are the comparison of decay curves between GC750-Li sample and GC750-16Sr one.

The crystal structures of hexagonal and orthorhombic GdF₃ are schematically illustrated in Fig. 10. In orthorhombic GdF₃ crystal, Gd³⁺ ion locates in the center of an irregular trigonal prism with six F⁻ ions at the corners and other three F⁻ ions are symmetrically disposed in front of the prism (Fig. 10a); hence, the coordination number (CN) of F⁻ around Gd³⁺ is nine.³⁷ On the other hand, in the hexagonal GdF₃ crystal, there are extra two F⁻ ions disposed up and down the prism,³⁷ and thus the CN of F⁻ around Gd³⁺ becomes eleven (Fig. 10b). Notably, the shorter Gd³⁺–Gd³⁺ ionic distance (3.593 Å) in orthorhombic GdF₃ than that (4.036 Å) in hexagonal GdF₃ will be beneficial for efficient energy transfer among RE (Yb/Er) activators, which results in intense UC luminescence in the orthorhombic GdF₃ embedded GC.

Fig. 10 Crystal structures of (a) orthorhombic and (b) hexagonal GdF₃. The Gd–Gd ionic distances (unit: Å) given in the unit cell are derived from ref. 37.

The ability of controllable crystallization with the incorporation of RE activators into precipitated fluorides enables new research opportunities for the applications of the prepared GCs. Here, these GCs are demonstrated to have potential application in optical temperature sensors. As shown in Fig. 11, temperature-dependent UC emission spectra in the temperature range of 293–563 K were recorded. Importantly, the relative intensity of the two green emission bands of Er³⁺ around 521 nm (²H_11/2 → ⁴I_15/2) and 542 nm (⁴S_3/2 → ⁴I_15/2) exhibits remarkable temperature dependence (Fig. 11a and b), owing to the thermal coupling competition between ²H_11/2 and ⁴S_3/2 states of Er³⁺. The fluorescent intensity ratio (FIR) of these two UC emissions can be expressed by the equation³⁸

where I₅₂₁ and I₅₄₂ are the integrated UC emission intensities corresponding to the ²H_11/2 → ⁴I_15/2 and ⁴S_3/2 → ⁴I_15/2 transitions of Er³⁺, respectively, C is the constant, ΔE is the energy gap between ²H_11/2 and ⁴S_3/2 states, k_B is the Boltzmann constant, and T is the absolute temperature. According to eqn (1), monolog plot of FIR versus inverse absolute temperature for the GC750-11Sr and GC750-Li samples are provided in Fig. 11c and d, respectively. The slopes are fitted to be 1133 and 1065, respectively. As a consequence, the corresponding energy gaps (ΔE) for GC750-11Sr and GC750-Li samples are determined to be about 788 and 741 cm⁻¹, respectively. Furthermore, the temperature sensor sensitivity was calculated according to the equation

Fig. 11 Temperature-dependent (293–563 K) UC emission spectra of the Yb/Er doped (a) GC750-11Sr (pumping power: 200 mW) and (b) GC750-Li (pumping power: 50 mW) samples in the wavelength range of 500–580 nm. Monolog plots of FIR as a function of inverse absolute temperature for (c) GC750-11Sr and (d) GC750-Li samples. Dependence of sensor sensitivity of (e) GC750-11Sr and (f) GC750-Li samples on temperature.

According to eqn (2), the maximal sensitivity can reach 0.33% K⁻¹ at 568 K for the GC750-11Sr sample and 0.25% K⁻¹ at 533 K for the GC750-Li sample, respectively (Fig. 11e and f). The sensitivities of GCs are comparable to previously reported values in silicate glasses (∼0.31% K⁻¹) and fluoride micro-/nano-crystals (∼0.40% K⁻¹).^39–44 Importantly, the present GCs show improved UC performance relative to silicate glasses due to the partition of Yb³⁺/Er³⁺ activators into GdF₃ crystalline lattice with low phonon energy, and exhibited better stability than pure fluorides owing to the protecting role of the inorganic aluminosilicate glass matrix. Therefore, benefited from its greatly enhanced UC luminescence, the present GC embedded with orthorhombic GdF₃ NCs (Yb/Er doped GC750-Li sample) is expected to be a very promising candidate for accurate optical temperature sensing.

On the other hand, color-tunable UC luminescence can be realized by doping appropriate RE emitting centers into GCs. Fig. 12a shows the UC emission spectra of the Yb/Tm/Ho triply doped GC750-Li samples under the excitation of 980 nm laser. The blue (Tm^{3+ 1}G₄ → ³H₆), green (Ho^{3+ 4}S₂, ⁵F₄ → ⁵I₈) and red (Ho^{3+ 5}F₅ → ⁵I₈; Tm^{3+ 1}G₄ → ³F₄) emissions are simultaneously generated. By modifying Ho³⁺content, the relative intensity of blue, green and red luminescence is adjustable. As demonstrated in the Commission International de I'Eclairage (CIE) 1931 chromaticity diagram (Fig. 12b), the UC luminescent color can be tuned from blue to cyan and finally to green and is clearly visible by the naked eyes (insets of Fig. 12b). In addition, pumping-power-dependent UC emission spectra of Yb/Ho and Yb/Tm doped GC samples (not provided here) evidence nearly quadratic and cubic dependence for Ho³⁺ green and Tm³⁺ blue UC luminescence, respectively. These results suggest that the electron populations in the ⁵S₂, ⁵F₄ states of Ho³⁺ and ¹G₄ one of Tm³⁺ are two-photon and three-photon UC processes, respectively. As a result, the impact of pumping power on Tm^{3+ 1}G₄ → ³H₆ (blue) and ¹G₄ → ³F₄ (red) transitions will be more significant than Ho^{3+ 5}S₂, ⁵F₄ → ⁵I₈ (green) and ⁵F₅ → ⁵I₈ (red) ones. Hence, the color of UC luminescence for GC sample is expected to finely optimize via modifying the 980 nm laser pumping power, as shown in Fig. 12c. The enhancement of both blue and red UC emissions is faster than that of green one when pumping power of 980 nm laser gradually intensifies (right of Fig. 12c). Excitation-power-dependent variation of color coordinates in CIE 1931 chromaticity diagram clearly verifies the shifting of UC emission color towards cyan region with increase of laser pumping power (left of Fig. 12c).

Fig. 12 (a) UC emission spectra of the Yb/Ho (1/0.01, mol%), Yb/Tm (1/0.005, mol%) co-doped and Yb/Tm/Ho (1/0.005/x, mol%, x = 0.001–0.1) tri-doped GC750-Li samples (with pumping power of 50 mW) and (b) the corresponding chromaticity coordinates in CIE diagram. Insets of (b) show UC luminescent photographs of GC samples. (c) Pumping-power-dependent chromaticity coordinates of UC luminescence for the Yb/Tm/Ho (1/0.005/0.005, mol%) tri-doped GC750-Li sample (left) and the corresponding UC emission spectra (right).

Conclusions

Hexagonal and orthorhombic GdF₃ NCs embedded glass ceramics were successfully prepared by melt-quenching and subsequent glass crystallization. It was found that both crystallization temperature and alkaline-earth/alkali-metal dopants played a key role on the phase evolution of the precipitated GdF₃ among glass matrix. Particularly, the addition of alkaline-earth fluorides such as SrF₂ and BaF₂ was beneficial to the crystallization of hexagonal GdF₃ while the introduction of alkali-metal fluorides such as LiF induced the formation of pure orthorhombic GdF₃ in the glass matrix. Elemental mapping in the scanning transmission electron microscope and optical spectroscopy analysis confirmed the partition of the active centers into the GdF₃ crystalline lattice. As a result, greatly enhanced Er³⁺ UC luminescence was observed after glass crystallization, ascribing to modification of Yb³⁺/Er³⁺ surrounding from amorphous aluminosilicate glass matrix to GdF₃ crystalline lattice with low phonon energy and high crystallinity. Moreover, the UC luminescence of the orthorhombic GdF₃ embedded glass ceramic was found to be about 400 times as high as that of the hexagonal GdF₃ embedded glass ceramic, relating to the combined effects of different phase structures and crystal sizes. Subsequently, rare earth doped glass ceramic containing orthorhombic GdF₃ was demonstrated to be a promising material for accurate temperature sensing as well as color tunable emitting.

Acknowledgements

This work was supported by the Natural Science Foundation of Zhejiang Province for Distinguished Young Scholars (LR15E020001), National Natural Science Foundation of China (21271170, 61372025, 51402077 and 51572065) and the 151 talent's projects in the second level of Zhejiang Province.

References

1. E. Heumann, S. Bär, K. Rademaker, G. Huber, S. Butterworth, A. Diening and W. Seelert, Appl. Phys. Lett., 2006, 88, 061108.
2. S. Gai, C. Li, P. Yang and J. Lin, Chem. Rev., 2014, 114, 2343.
3. D. Q. Chen, Y. S. Wang and M. C. Hong, Nano Energy, 2012, 1, 73.
4. X. Huang, S. Han, W. Huang and X. Liu, Chem. Soc. Rev., 2013, 42, 173.
5. E. Downing, L. Hesselink, J. Ralston and R. Macfarlane, Science, 1996, 273, 1185.
6. G. Chen, H. Qiu, P. N. Prasad and X. Chen, Chem. Rev., 2014, 114, 5161.
7. D. M. Yang, P. A. Ma, Z. Y. Hou, Z. Y. Cheng, C. X. Li and J. Lin, Chem. Soc. Rev., 2015, 44, 1416–1448.
8. Z. H. Xu, M. Quintanilla, F. Vetrone, A. O. Govorov, M. Chaker and D. L. Ma, Adv. Funct. Mater., 2015, 25, 2950.
9. Y. H. Zhang, L. X. Zhang, R. R. Deng, J. Tian, Y. Zong, D. Y. Jin and X. G. Liu, J. Am. Chem. Soc., 2014, 136, 4893.
10. D. Q. Chen, W. D. Xaing, X. J. Liang, J. S. Zhong, H. Yu, M. Y Ding, H. W. Lu and Z. G. Ji, J. Eur. Ceram. Soc., 2015, 35, 859.
11. F. Auzel, Chem. Rev., 2004, 104, 139.
12. F. Wang and X. Liu, Chem. Soc. Rev., 2009, 38, 976–989.
13. S. Chenu, E. Veron, C. Genevois, G. Matzen, T. Cardinal, A. Etienne, D. Massiot and M. Allix, Adv. Opt. Mater., 2014, 2, 364.
14. S. F. Zhou, C. Y. Li, G. Yang, G. Bi, B. B. Xu, Z. L. Hong, K. Miura, K. Hirao and J. R. Qiu, Adv. Funct. Mater., 2013, 23, 5436.
15. P. P. Fedorov, A. A. Luginina and A. I. Popov, J. Fluorine Chem., 2015, 172, 22.
16. S. L. Zhao, X. Wang, X. Sun, G. H. Jia, L. H. Huang, D. G. Deng, F. X. Xin and S. Q. Xu, CrystEngComm, 2013, 15, 7346.
17. A. Herrmann, M. Tylkowski, C. Bocker and C. Rüssel, Chem. Mater., 2013, 25, 2878.
18. Y. L. Wei, X. M. Li and H. Guo, Opt. Mater. Express, 2014, 4, 1367.
19. C. G. Lin, C. Liu, Z. Y. Zhao, L. G. Li, C. Bocker and C. Russel, Opt. Lett., 2015, 40, 5263.
20. C. G. Lin, C. Bocker and C. Russel, Nano Lett., 2015, 15, 6764.
21. X. Y. Liu, H. Guo, S. Ye, M. Y. Peng and Q. Y. Zhang, J. Mater. Chem. C, 2015, 3, 5183.
22. X. Y. Liu, H. Guo, Y. Liu, S. Ye, M. Y. Peng and Q. Y. Zhang, J. Mater. Chem. C, 2016, 4, 2506.
23. R. A. Hewes and J. F. Sarver, Phys. Rev., 1969, 182, 427.
24. M. M. Lage, A. Righi, F. M. Matinaga, J. Y. Gesland and R. L. Moreira, J. Phys.: Condens. Matter, 2004, 16, 3207.
25. Z. Chen, G. B. Wu, H. Jia, K. Sharafudeen, W. B. Dai, X. W. Zhang, S. F. Zeng, J. M. Liu, R. F. Wei, S. C. Lv, G. P. Dong and J. R. Qiu, J. Phys. Chem. C, 2015, 119, 24056.
26. D. Q. Chen, Y. L. Yu, P. Huang, H. Lin, Z. F. Shan and Y. S. Wang, Acta Mater., 2010, 58, 3035.
27. Y. Y. Bu, S. J. Cheng, X. F. Wang and X. H. Yan, Appl. Phys. A: Mater. Sci. Process., 2015, 121, 1171.
28. D. Q. Chen, Y. S. Wang, Y. L. Yu and P. Huang, Appl. Phys. Lett., 2007, 91, 051920.
29. A. C. Yanes, A. Santana-Alonso, J. Mendez-Ramos, J. del-Castillo and V. D. Rodriguez, Adv. Funct. Mater., 2011, 21, 3136.
30. D. Q. Chen, Y. L. Yu, P. Huang, F. Y. Weng, H. Lin and Y. S. Wang, CrystEngComm, 2009, 11, 1686.
31. R. E. Thoma and G. D. Brunton, Inorg. Chem., 1966, 5, 1937.
32. D. Q. Chen and P. Huang, Dalton Trans., 2014, 43, 11299.
33. P. E. McMillan, in Glass-Ceramics, Academic Press, London, 1979.
34. E. Görlich, in The Effective Charges and the Electronegativity, Polish Academy of Art and Sciences, Kraków, 1997.
35. F. Vetrone, J. C. Boyer, J. A. Capobianco, A. Speghini and M. Bettinelli, J. Appl. Phys., 2004, 96, 661.
36. F. Wang, J. Wang and X. G. Liu, Angew. Chem., Int. Ed., 2010, 49, 7456.
37. X. T. Zhang, T. Hayakawa, M. Nogami and Y. Ishikawa, J. Nanomater., 2010, 651326.
38. D. Q. Chen, Z. Y. Wan, Y. Zhou, X. Z. Zhou, Y. L. Yu, J. S. Zhong, M. Y. Ding and Z. G. Ji, ACS Appl. Mater. Interfaces, 2015, 7, 19484.
39. G. S. Maciel, L. d. S. Menezes, A. S. L. Gomes, C. B. de Araújo, Y. Messaddeq, A. Florez and M. A. Aegerter, IEEE Photonics Technol. Lett., 1995, 7, 1474.
40. Q. Zhou, C. R. Li, Z. F. Liu, S. F. Li and C. L. Song, Opt. Mater., 2007, 30, 513.
41. S. H. Zhou, K. M. Deng, X. T. Wei, G. C. Jiang, C. K. Duan, Y. H. Chen and M. Yin, Opt. Commun., 2013, 291, 138.
42. S. Jiang, P. Zeng, L. Q. Liao, S. F. Tian, H. Guo, Y. H. Chen, C. K. Duan and M. Yin, J. Alloys Compd., 2014, 617, 538.
43. H. Suo, C. F. Guo and T. Li, J. Phys. Chem. C, 2016, 120, 2914.
44. H. Suo, C. F. Guo, Z. Yang, S. S. Zhou, C. K. Duan and M. Yin, J. Mater. Chem. C, 2015, 3, 7379.

---