Nano-Crystallization of Ln-Fluoride Crystals in Glass-Ceramics via Inducing of Yb³⁺ for Efficient Near-Infrared Upconversion Luminescence of Tm³⁺

Abstract

Transparent glass-ceramic composites embedded with Ln-fluoride nanocrystals are prepared in this work to enhance the upconversion luminescence of Tm³⁺. The crystalline phases, microstructures, and photoluminescence properties of samples are carefully investigated. KYb₃F₁₀ nanocrystals are proved to controllably precipitate in the glass-ceramics via the inducing of Yb³⁺ when the doping concentration varies from 0.5 to 1.5 mol%. Pure near-infrared upconversion emissions are observed and the emission intensities are enhanced in the glass-ceramics as compared to in the precursor glass due to the incorporation of Tm³⁺ into the KYb₃F₁₀ crystal structures via substitutions for Yb³⁺. Furthermore, KYb₂F₇ crystals are also nano-crystallized in the glass-ceramics when the Yb³⁺ concentration exceeds 2.0 mol%. The upconversion emission intensity of Tm³⁺ is further enhanced by seven times as Tm³⁺ enters the lattice sites of pure KYb₂F₇ nanocrystals. The designed glass ceramics provide efficient gain materials for optical applications in the biological transmission window. Moreover, the controllable nano-crystallization strategy induced by Yb³⁺ opens a new way for engineering a wide range of functional nanomaterials with effective incorporation of Ln³⁺ ions into fluoride crystal structures.

1. Introduction

Trivalent lanthanide (Ln³⁺) ions doped upconversion (UC) luminescent materials have been extensively investigated for applications in lighting, solar cells, solid-state laser, biological imaging, and temperature sensing due to the multiple luminescence wavelength from ultraviolet to near-infrared (NIR) regions [1,2,3,4,5,6,7,8,9]. Among these, Tm³⁺ doped materials possess UC luminescence in the NIR region at around 800 nm (³H₄→³H₆), which is located in the biological transmission window, and the light in that region can easily penetrate biological tissues [10,11,12,13]. Moreover, the UC emissions of Tm³⁺ are dramatically enhanced via the energy transfer (ET) from Yb³⁺, which possesses a large absorption section for the commercial 980 nm lasers. Thus, Yb³⁺-Tm³⁺ co-doped materials were more efficient candidates for NIR-to-NIR UC luminescence and were significant light sources for biologically non-destructive detection. In the past decades, a large number of investigations about Yb³⁺-Tm³⁺ co-doped UC materials have been widely reported for achieving high-efficiency NIR UC luminescence [14,15,16,17,18,19,20].

Glass exhibits high transmittance, easy fabrication, and excellent processability due to its amorphous and softening characteristics. These properties make glass an appealing matrix for the design and fabrication of special optics devices, particularly of optical fiber lasers [21,22], which can hardly be achieved by using crystals. Furthermore, nano-crystallized glass ceramic (GC) composites can be obtained via the precipitation of functional nanocrystals in a glass matrix via heat treatments. The nano-crystallized GCs still possess high optical transmittance and the UC luminescence efficiency in GCs can be greatly enhanced, as compared to glass, when Ln³⁺ ions are successfully incorporated into the crystal structures featuring lower probability of non-radiative transition ascribed to the lower phonon energy. So far, Yb³⁺-Tm³⁺ co-doped GCs have been considered as significantly optical gain materials for UC NIR fiber lasers, noncontact optical thermometers, and bio-imaging [23,24,25,26]. Generally, Ln³⁺ ions were expected to incorporate into the crystal structures via the ionic substitution for Y³⁺, La³⁺, Lu³⁺, Sc³⁺, and Sr²⁺ in the GCs embedded with Na(Y/La/Lu)F₄, LaF₃, YF₃, KSc₂F₇, and SrF₂ crystals, etc [27,28,29,30,31]. However, the ionic substitution process was uncontrollable and the amount of Ln³⁺ incorporated into the crystal structures was usually small due to the large mismatch between Ln³⁺ and the replaceable sites. The enhancements of luminescence in GCs were limited. Moreover, a lot of remaining crystals not occupied by Ln³⁺ ions made no contribution to the enhancement of luminescence but will trigger severe issues of optical scattering and low optical transmittance in glass. These made the traditional GCs difficult to be used in the practical applications. Accordingly, it is highly desirable for design and fabrication of novel GCs for controllably incorporating a large number of Ln³⁺ into crystal structures to achieve high-efficiency NIR UC luminescence of Tm³⁺.

Ln-fluoride crystal, intrinsically containing Ln³⁺ (for example, Yb³⁺ or Er³⁺) in the crystal lattices, is a significant host for the incorporation of other active Ln³⁺ ions because the mismatches in ionic radius between different Ln³⁺ ions are smaller and active Ln³⁺ ions enter Ln-fluoride crystals more easily [32]. In this work, GCs containing Ln-fluoride nanocrystals were designed to greatly enhance the UC luminescence of Tm³⁺ and achieve pure, high-efficiency NIR-to-NIR emissions. Two kinds of Ln-fluoride nanocrystals (KYb₃F₁₀ and KYb₂F₇) were controllably precipitated from a glass matrix via the heat treatments dependent on the doping of Yb³⁺. Tm³⁺ ions entered the fluoride crystal lattices easily by replacing Yb³⁺ sites. As a result, a large number of Ln³⁺ were incorporated into the fluoride crystals featuring extremely low phonon energy and the UC emission of Tm³⁺ in the GCs were dramatically enhanced. Moreover, the pure NIR UC emissions were also obtained by adjusting the doping concentration of Tm³⁺. Thus, the designed GCs provide an efficient material for pure NIR UC luminescence and offer highly promising developments for UC fiber lasers, non-contact optical thermometers, and bio-imaging.

2. Materials and Methods

2.1. Materials Preparation

Samples with a molar composition of 70SiO₂-15KF-15ZnF₂-xYbF₃-yTm₂O₃ (x = 0~3.5, y = 0~0.3) were prepared using a melt quenching method. First, 30 g reagent grade stoichiometric mixtures were mixed thoroughly in an agate mortar and melted in covered quartz crucibles in an electric furnace at 1550 °C for 30 min; the melts were poured onto a brass plate and then pressed using another brass plate to obtain precursor glasses (PGs). PG samples were heated at 540 °C for 5 and 10 h to obtain GC samples according to the differential scanning calorimetry (DSC) results in ref. [33]. PG and GC samples were cut and polished to 2 mm thick for measurements.

2.2. Characterizations

To identify the crystalline phase in GCs, X-ray diffraction (XRD) patterns were performed on a X-ray diffractometer (Bruker, Fällanden, Switzerland) with Cu/Ka (λ = 0.1541 nm) radiation. The morphology and size distribution of the nanocrystals in GCs were measured via high-resolution transmission electron microscopy (HRTEM) (FEI, Hillsboro, OR, USA). UC emission spectra of samples were recorded using an Edinburgh FLS980 fluorescence spectrometer (Edinburgh Instruments, Edinburgh, UK). A 980 nm laser diode (LD) was used as the exciting source for the measurement of UC emission spectra. The emission decay curves were measured using the same spectrometer with a microsecond lamp as the excitation source. All measurements were performed at room temperature.

3. Results and Discussion

Figure 1a shows the XRD patterns of Yb³⁺-Tm³⁺ co-doped PG and GCs. A broad band is observed in the XRD pattern of PG due to the amorphous characteristic of glass. Sharp peaks are observed in the XRD patterns of the two GCs. The main peaks at 26.99°, 31.27°, 44.81°, and 53.1° are ascribed to the (222), (400), (440), and (622) crystal facets of KYb₃F₁₀ (No: 74-2204), respectively, indicating that KYb₃F₁₀ crystals have been precipitated in the GCs. Moreover, weak peaks at 18.86°, 31.04°, and 38.28° attributed to the diffraction peaks of K₂SiF₆ (No: 85-1382) crystals are also observed in the XRD patterns of GC heated at 540 °C for 5 h. It is also found that the intensities of diffraction peaks for KYb₃F₁₀ crystal are all increased when the heating time increases from 5 to 10 h. However, the diffraction peaks of K₂SiF₆ crystal all weaken due to the further heat treatment to 10 h. These results indicate that KYb₃F₁₀ crystals prefer to precipitate in the 1.0Yb³⁺-0.005Tm³⁺ co-doped GCs via the further heat treatment. Additionally, the average size of crystals can be calculated by the Scherrer’s equation. The diffraction peak, around 2θ = 27.01° and 18.42°, was selected for the calculation, and the average size of the KYb₃F₁₀ and K₂SiF₆ nanocrystals in the GC heated at 540 °C for 5 h was calculated to be approximately 13.35 and 31.34 nm, respectively.

The HAADF-TEM image shown in Figure 1b reveals that the nanocrystals are in-situ, precipitated among the glass matrix. The measured size of these nanoparticles is from 10 to 30 nm. The HR-TEM image is shown in Figure 1c. The crystal lattice fringes are obvious, which is different from that of the amorphous glass matrix. The interval of the crystal lattice fringes d can be measured directly, and its value is about 0.286 nm, which corresponds to the (400) crystal facet of cubic KYb₃F₁₀. These also prove that KYb₃F₁₀ nanocrystals are precipitated in the GCs.

The optical transmission spectra of the 1.0Yb³⁺-0.1Tm³⁺ co-doped samples are shown in Figure 2a. The PG sample possesses high transmittance (~90%) from 300 to 800 nm with a thickness of 2 mm. Though nanocrystals are precipitated in the GCs, the optical transmittances of GCs are still as high as 85%. Interestingly, the transmittance of the GC heated for 10 h is higher than that heated for 5 h. As calculated from the XRD patterns, the average size of KYb₃F₁₀ crystals is smaller than that of K₂SiF₆ crystals. When the heat treatment time increases from 5 to 10 h, more KYb₃F₁₀ crystals and less K₂SiF₆ crystals are precipitated in the GC, resulting in the increase of transmittance.

Figure 2b shows the emission spectra of 1.0Yb³⁺-0.005Tm³⁺ co-doped PG and GCs recorded at room temperature upon the excitation of a 980 nm laser diode (LD). The spectra all include a NIR emission peak at 802 nm attributed to the ³H₄→³H₆ transition of Tm³⁺. Meanwhile, the blue (450 and 480 nm), red (650 nm), and deep red emission peaks (680 and 700 nm) attributed to ¹D₂→³F₄, ¹G₄→³H₆, ¹G₄→³F₄, ³F₂→³H₄, and ³F₃→³H₄ transition of Tm³⁺ [34,35], respectively, are all observed in the emission spectra of PG and GC samples. Since Tm³⁺ exhibits no absorption for 980 nm light, these emissions are all obtained via the ET processes from Yb³⁺ to Tm³⁺. In addition, the emission peaks of GCs are narrower than that of PG and the splitting of spectrum by crystal fields are observed in the spectra of GCs, indicating the incorporation of Tm³⁺ into crystal coordinated sites. It is also found from the spectra that the emission intensity at 450 and 802 nm increases 15 and 6 times after the heat treatment for 10 h, respectively. Figure 2c,d show the emission decay curves monitored at 450 and 802 nm, respectively. The emission lifetime at 450 nm increases from 441 to 601 and 719 μs, and that at 802 nm increases from 393 to 658 and 887 μs when the sample is heated at 540 °C for 5 and 10 h, respectively. These results are attributed to the fact that the low phonon energy environment reduces the possibility of non-radiative transition and enhances the emission intensity and lifetime in the sample, which in turn prove the incorporation of Tm³⁺ into the fluoride nanocrystals in GCs. Actually, it is difficult for Tm³⁺ to enter K₂SiF₆ crystal structures due to the lack of appropriate sites for the substitution of Tm³⁺. However, Tm³⁺ can easily incorporate into KYb₃F₁₀ crystal structures by occupying the sites of Yb³⁺ because the ionic radiuses of Tm³⁺ (R = 0.086 nm) and Yb³⁺ (R = 0.085 nm) are very similar. Accordingly, the incorporation of Tm³⁺ into KYb₃F₁₀ nanocrystals is responsible for the enhancements of emission intensities and lifetimes in GCs.

The double-logarithmic plots of the excitation power dependency on the emission intensities are presented in Figure 3a. The fitted slope (n) of the plot of Ln (I_UC) versus Ln (Power) is used to determine the absorbed photon numbers per UC emitted [36,37]. The obtained value of n was 1.67, 1.73, 2.67, and 3.40 corresponding to the 802,700, 650, and 450 nm emission of 1.0Yb³⁺-0.005Tm³⁺ co-doped GC, respectively. These prove that the emissions are all attributed to UC emission of Tm³⁺ and obtained via the ET from Yb³⁺. Two, two, three, and four pump photons are needed to pump the electrons to ³H₄, ³F_2,3, ¹G₄, and ¹D₂ energy levels of Tm³⁺ to achieve the corresponding UC emissions, respectively.

Based on these, the energy levels of Tm³⁺ and Yb³⁺ and the relative transitions corresponding to the above UC emissions are illuminated in Figure 3b [38,39,40]. Under 980 nm LD excitation, the electrons of Yb³⁺ are excited from the ground state ²F_7/2 to the ²F_5/2 excited state, then the energy is transferred to Tm³⁺ through the ET1 process: [²F_5/2 (Yb³⁺) + ³H₆ (Tm³⁺)→²F_7/2 (Yb³⁺) + ³H₅ (Tm³⁺) + phonons]. The electrons of Tm³⁺ are excited from the ground state to the ³H₅ excited state and then undergo a non-radiative relaxation transition to the ³F₄ energy level. Via the ET2 process, [²F_5/2 (Yb³⁺) + ³F₄ (Tm³⁺)→²F_7/2 (Yb³⁺) + ³F_2,3 (Tm³⁺) + phonons], the electrons are populated to the ³F_{2, 3} energy level. A part of the electrons undergo a non-radiative relaxation transition to ³H₄ energy level. In this process, ³H₄→³H₆ (802 nm) and ³F_2,3→³H₆ (700 and 680 nm) transitions are obtained. Next, part of the electrons at ³H₄ energy levels are excited to ¹G₄ energy through the process of ET3: [²F_5/2 (Yb³⁺) + ³H₄ (Tm³⁺)→²F_7/2 (Yb³⁺) + ¹G₄ (Tm³⁺) + phonons]. This produces ¹G₄→³H₆ (480 nm) and ¹G₄→³F₄ (650 nm) transitions. Finally, the electrons of the ¹G₄ energy level are populated to the ¹D₂ energy level through the process of ET4, [²F_5/2 (Yb³⁺) + ¹G₄ (Tm³⁺)→²F_7/2 (Yb³⁺) + ¹D₂ (Tm³⁺) + phonons], producing a ¹D₂→³F₄ (450 nm) transition.

Furthermore, the transitions of Tm³⁺ are also modulated via the interaction between neighboring Tm³⁺ ions, which is directly determined by the doping concentration of Tm³⁺ ions [41,42]. The dependence of UC emission spectrum on the doping concentration of Tm³⁺ in the co-doped PG and GC is presented in Figure 4a,b. It can be observed from the spectra that the blue emission intensities at 450 nm and 480 nm both decrease monotonically with the increase of Tm³⁺ concentration, while the NIR emission intensity at 802 nm experiences first an enhancement and then a decrease due to the further increase of Tm³⁺ concentration. Actually, as a result of the increase of Tm³⁺ content, the distance between Tm³⁺ becomes shorter. The interactions between neighboring Tm³⁺ ions get larger, which benefits the cross-relaxation (CR) process, ¹G₄ + ³H₆→³F_2,3 + ³F₄, as shown in Figure 3b. The CR process depopulates electrons in the ¹G₄ level, while the electrons in ³F_2,3 and ³F₄ levels are populated via the CR process. Thus, blue emission intensities decrease quickly with the increase of Tm³⁺ even though the doping concentration is low. The NIR emission increases firstly and then decreases when the doping concentration is further increased due to the concentration quenching effect. Compared to a glass matrix, crystal exhibits a more compact structure. The distance between Tm³⁺ in GC is shorter than that in PG at the same doping concentration, resulting in the lower luminescence-quenching concentration in GC as compared to PG. The emission intensity at 802 nm reaches a maximum with the doping of 0.025 and 0.01Tm³⁺ in PG and GC, respectively. As shown in Figure 4b and the inset, the blue emissions around 450 and 480 nm in PG and GC both disappear and the UC spectra almost exhibit pure NIR emissions when the contents of Tm³⁺ exceed 0.01 mol%. A blue UC emission only being obtained in GC at a low doping concentration indicates most Tm³⁺ ions are incorporated into the KYb₃F₁₀ crystal structure with short interionic distance. The pure NIR emissions that are obtained at a very low doping concentration of Tm³⁺ indicates that the GC is an excellent candidate for pure NIR UC emission and is a significant UC luminescent material for promising applications in NIR fiber lasers and high-resolution biological imaging.

As mentioned above, Yb³⁺ works as sensitizer ion for the UC emissions of Tm³⁺. More importantly, Yb³⁺ also participates in the construction of fluoride nanocrystals and provides appropriate crystal sites for the incorporation of Tm³⁺. The doping of Yb³⁺ plays a vital role in the precipitation of nanocrystals and enhancement of the UC emissions. The Yb³⁺-concentration-dependent XRD patterns of the Yb³⁺-Tm³⁺ co-doped and no-doped GCs are presented in Figure 5. It is found that only K₂SiF₆ crystals are precipitated in the no-doped GC and these make no contribution to the enhancement of UC emission. Via the doping of Yb³⁺, KYb₃F₁₀ are precipitated in the GCs apart from K₂SiF₆ crystals as the Yb³⁺ content increases from 0.5 to 1.0 mol%. When Yb³⁺ concentration is set to 1.5 mol%, only the diffraction peaks of KYb₃F₁₀ crystals are observed in the XRD pattern of GC. These results indicate that the precipitation of KYb₃F₁₀ nanocrystals in the GCs is governed by the doping concentration of Yb³⁺ induced by Yb³⁺. More interestingly, KYb₂F₇ crystals are also precipitated in the GCs when the Yb³⁺ concentration is increased from 2.0 to 3.5 mol%. In the XRD pattern of the 3.5Yb³⁺-0.1Tm³⁺ co-doped sample, only the diffraction peaks of KYb₂F₇ crystals are observed. Therefore, the nano-crystallization in GC is induced by the doped Yb³⁺ and the crystalline phase in the GC can be controllably regulated from K₂SiF₆ to KYb₃F₁₀ and further changes to KYb₂F₇ crystals by adjusting the concentration of Yb³⁺ from 0 to 3.5 mol%.

Via the nano-crystallizations of KYb₃F₁₀ and KYb₂F₇, Yb³⁺ ions are spontaneously confined within fluoride crystal structures. More importantly, Tm³⁺ can easily incorporate these two crystal structures by the substitution of Yb³⁺. Owing to the extremely low phonon energies in the fluoride crystals, the probabilities of non-radiative transitions are low, which is beneficial for the efficient ET from Yb³⁺ to Tm³⁺ and enhanced UC emission of Tm³⁺. The UC emission spectra of Yb³⁺-Tm³⁺ co-doped GC with various concentrations of Yb³⁺ are shown in Figure 6a. Excited by a 980 nm LD, pure NIR UC emissions around 802 nm of Tm³⁺ are observed in the spectra of GCs. As the Yb³⁺ concentration increases from 0.5 to 1.5 mol%, the quantity of KYb₃F₁₀ nanocrystals precipitated in the GCs is increased, and the amount of Tm³⁺ and Yb³⁺ ions located in the fluoride crystal structures are increased. The emission intensity enhances firstly, reaching a maximum at 1.0 mol% Yb³⁺, and then decreases when the Yb³⁺ concentration is increased to 1.5 mol% due to the concentration quenching effect caused by the severe interactions between Ln³⁺ ions. When the Yb³⁺ concentration is further increased to 2.0 mol%, KYb₂F₇ crystals start to precipitate in the GCs. A part of the Tm³⁺ enter KYb₂F₇ nanocrystals in the 2.0Yb³⁺-0.1Tm³⁺ co-doped GC, which disperses the distribution of Tm³⁺ and increases the NIR UC emission intensity again. When the Yb³⁺ concentration increases from 2.0 to 3.5 mol%, more KYb₂F₇ crystals are precipitated in the GCs, more Tm³⁺ ions are incorporated into KYb₂F₇ crystals, and the UC emission intensity at 800 is increased monotonically. It is also found that the profile of the emission spectrum for the 3.5 Yb³⁺-0.1Tm³⁺ co-doped GCs is different from that of the 1.0Yb³⁺0.1Tm³⁺ co-doped samples (inset of Figure 6a), which in turn proves that Tm³⁺ are incorporated in two different crystal environments in the two GCs containing different nanocrystals. Compared to KYb₂F₇ crystal, more Yb³⁺ ions are distributed in the crystal structure in KYb₃F₁₀. The luminescent-quenching concentration of the UC emission in KYb₂F₇ crystal is higher than that in KYb₃F₁₀. The emission intensity in 3.5Yb³⁺-0.1Tm³⁺ co-doped GC is 7 times higher than that in the 1.0Yb³⁺-0.1Tm³⁺ sample, which indicates that the nano-crystallization of KYb₂F₇ provides more excellent crystal environments for achieving more efficient NIR UC emission of Tm³⁺.

The UC emission decay curves of the Yb³⁺-Tm³⁺ co-doped GCs monitored at 802 nm are shown in Figure 6b. The lifetime of the NIR emission decreases as Yb³⁺ concentration increases from 0.5 to 1.5 mol%. Especially, the lifetime of the 1.5Yb³⁺-0.1Tm³⁺ co-doped GC decreases dramatically to 373 μs due to the severe interaction between the neighboring Tm³⁺ in KYb₃F₁₀ crystals. However, the lifetime increases from 16.5 to 39.8 μs when the Yb³⁺ concentration is increased from 2.5 to 3.5 mol%. These results also prove that the coordinated environments of Tm³⁺ in high-doping GCs (2.5 and 3.5 mol% Yb³⁺-0.1Tm³⁺) is distinct to that in low-doping GCs (0.5, 1.0, and 1.5 mol% Yb³⁺-0.1Tm³⁺). Therefore, the precipitation of KYb₂F₇ nanocrystals in GCs further enhances the NIR UC emission of Tm³⁺ and provides a significant matrix for applications in photonic devices, in particular of high-efficiency UC fiber lasers.

4. Conclusions

In summary, transparent Yb³⁺-Tm³⁺ co-doped GCs were prepared in this work for greatly enhancing the UC luminescence of Tm³⁺. The GCs possessed high transmittance (>85.0%) though fluoride nanocrystals were precipitated among the glass matrices. K₂SiF₆ crystals are precipitated in the no-doped GC. However, KYb₃F₁₀ and KYb₂F₇ nanocrystals were controllably and successively precipitated in the GCs via the inducing of Yb³⁺ when the doping concentration of Yb³⁺ changed from 0.5 to 3.5 mol%. UC emissions of Tm³⁺ were dramatically enhanced when Ln³⁺ ions were incorporated into KYb₃F₁₀ crystal structures as the Yb³⁺ concentration changed from 0.5 to 1.5 mol%. Pure NIR UC emissions were obtained by adjusting the concentration of Tm³⁺. KYb₂F₇ nanocrystals were precipitated in GCs when the Yb³⁺ concentration exceeded 2.0 mol%. More efficient UC emissions were achieved via the precipitation of KYb₂F₇ than that of KYb₃F₁₀ in GCs. The designed GCs offer potential optical gain materials for highly-efficient NIR UC photoluminescence. More importantly, the Yb³⁺-induced nano-crystallization strategy paves a new way for the design and fabrication of emerging GCs to provide excellent crystal environments for Ln³⁺.