Peasecod‐Like Hollow Upconversion Nanocrystals with Excellent Optical Thermometric Performance

Abstract

Trivalent lanthanide (Ln³⁺)‐doped hollow upconversion nanocrystals (UCNCs) usually exhibit unique optical performance that cannot be realized in their solid counterparts, and thus have been receiving tremendous interest from their fundamentals to diverse applications. However, all currently available Ln³⁺‐doped UCNCs are solid in appearance, the preparation of hollow UCNCs remains nearly untouched hitherto. Herein, a class of UCNCs based on Yb³⁺/Er³⁺‐doped tetralithium zirconium octafluoride (Li₄ZrF₈:Yb/Er) featuring 2D layered crystal lattice is reported, which makes the fabrication of hollow UCNCs with a peasecod‐like shape possible after Ln³⁺ doping. By employing the first‐principle calculations, the unique peasecod‐like hollow nanoarchitecture primarily associated with the hetero‐valence Yb³⁺/Er³⁺ doping into the 2D layered crystal lattice of Li₄ZrF₈ matrix is revealed. Benefiting from this hollow nanoarchitecture, the resulting Li₄ZrF₈:Yb/Er UCNCs exhibit an abnormal green upconversion luminescence in terms of the population ratio between two thermally coupled states (²H_11/2 and ⁴S_3/2) of Er³⁺ relative to their solid Li₂ZrF₆:Yb/Er counterparts, thereby allowing to prepare the first family of hollow Ln³⁺‐doped UCNCs as supersensitive luminescent nanothermometer with almost the widest temperature sensing range (123–800 K). These findings described here unambiguously pave a new way to fabricate hollow Ln³⁺‐doped UCNCs for numerous applications.

A class of peasecod‐like hollow Yb³⁺/Er³⁺‐doped Li₄ZrF₈ upconverting nanocrystals (UCNCs) that features a 2D layered crystal lattice is reported. Abnormal green upconverting luminescence that cannot be achieved in the solid UCNCs is observed in these peasecod‐like hollow UCNCs, thereby making them suitable as supersensitive nanothermometers with a wide temperature range for optical temperature sensing.

1. Introduction

Trivalent lanthanide (Ln³⁺) ions doped inorganic upconversion nanocrystals (UCNCs), possessing outstanding optical properties including large anti‐Stokes shifts, long excited‐state lifetimes and tunable emission colors, have drawn increasing attention for their potential applications in areas as diverse as biological imaging, detection, photonics, and temperature sensing.^{[ 1} , ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ^{13 ]} Particularly, with the rapid development of the nanocrystal synthesis technology, Ln³⁺‐doped UCNCs can now be prepared with finely controlled composition, phase, morphology, size, and output color.^{[ 14} , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ^{23 ]} However, as exemplified by the most representative β‐NaYF₄:Yb/Er UCNCs (Figure S1, Supporting Information), all the currently available UCNCs are solid in appearance, which is primarily ascribed to their 3D close‐packed crystal structures that make the preparation of hollow Ln³⁺‐doped UCNCs impossible. In contrast, it is expected that the hollow nanostructures can be formed through hetero‐valence Ln³⁺ doping in a 2D layered crystal structure, where the interlayer dilation of crystal lattice induced by different electrostatic interactions can occur and thus trigger the formation of hollow Ln³⁺‐doped UCNCs (Figure  1a), as achieved in other 2D porous nanomaterials such as carbon nanosheets,^{[ 24} , ^{25 ]} graphitic carbon nitride nanosheets,^{[ 26} , ^{27 ]} and metal chalcogenide nanocrystals.^{[ 28} , ^{29 ]} Therefore, seeking an appropriate host material that features a 2D layered crystal structure is of key importance for the fabrication of hollow Ln³⁺‐doped UCNCs. It is believed that such hollow Ln³⁺‐doped UCNCs will exhibit a totally different optical performance that cannot be realized in their solid counterparts, owing to their superhigh surface‐to‐volume (S/V) ratio.

Figure 1.

a) Proposed mechanism for the formation of hollow nanocrystals through Ln³⁺ doping in a 2D layered crystal structure. b) Crystal structure of orthorhombic‐phase Li₄ZrF₈ and c,d) the plausible localized crystal structure showing the different electrostatic interactions among Zr, Ln, and F atoms after the hetero‐valence Ln³⁺ doping in the lattice of Li₄ZrF₈ host matrix. The solid and dotted black arrows represent strong and weak electrostatic interactions, respectively. Comparison of electronic charge density for the e) pure and f) Er‐doped Li₄ZrF₈ crystals based on the first‐principle calculations, showing the slightly shortened Zr—F bonds in comparison with their pure counterparts after Ln³⁺ doping, and thus the interlayer dilation around the substituted Er³⁺ dopant.

To this end, herein, we report the first class of hollow inorganic UCNCs based on Yb³⁺/Er³⁺‐doped tetralithium zirconium octafluoride (hereafter referred to as Li₄ZrF₈:Yb/Er) that is characterized by a 2D layered crystal structure (Figure 1b). This unique 2D layered crystal structure of inorganic Li₄ZrF₈ fluoride enables the preparation of hollow Ln³⁺‐doped UCNCs after the hetero‐valence doping of Yb³⁺/Er³⁺ into the Li₄ZrF₈ host matrix (Figure 1c,d). By employing the first‐principle calculations based on density functional theory (DFT), we provide direct and solid evidence to support that the hollow nanostructure of Li₄ZrF₈:Yb/Er UCNCs is primarily due to the interlayer dilation of crystal lattice, induced by the hetero‐valence doping of Yb³⁺/Er³⁺ into the 2D layered crystal lattice of Li₄ZrF₈ host. Thanks to the novel hollow nanostructure, an abnormal upconversion luminescence (UCL) property is observed in these Li₄ZrF₈:Yb/Er UCNCs when compared with their solid counterparts, thereby allowing us to fabricate the first family of hollow Ln³⁺‐doped UCNCs that can serve as promising luminescent nanothermometers with excellent optical thermometric performance.

2. Results and Discussion

In our design, orthorhombic‐phase Li₄ZrF₈ was chosen as the host material for the synthesis of hollow Ln³⁺‐doped UCNCs due to its unique 2D layered crystal structure. The Li₄ZrF₈ has an orthorhombic crystal structure (space group Pnma, a = 9.738 Å, b = 9.747 Å, c = 5.758 Å, Z = 4, ICSD No. 80 398) with Zr⁴⁺ ion separated by LiF₆ octahedra that form the corrugated infinite layers, resulting in an overall 2D layered crystal structure. On this basis, we recognize that the hollow Li₄ZrF₈:Yb/Er UCNCs can be achieved after the hetero‐valence doping of Ln³⁺ into the crystal lattice of Li₄ZrF₈, owing to the different electrostatic interactions between the strong Zr—F and weak Ln‐F bonds, as clearly indicated in Figure 1c,d. This can be well substantiated by means of the first‐principle calculations based on DFT, where almost all the Zr—F bonds (2.271, 2.128, and 2.132 Å) in the Li₄ZrF₈:Yb/Er UCNCs were calculated to be slightly shorter than those (2.283, 2.144, and 2.131 Å) in their pure Li₄ZrF₈ counterparts (Figure 1e,f and Table S1, Supporting Information), and thereby leading to a broadened interlayer spacing of the LiF₆ octahedra layers from 4.854 Å for the pure Li₄ZrF₈ to 4.906 Å for the Er‐doped Li₄ZrF₈ (Figure 1e,f and Figure S2, Supporting Information). To probe the feasibility of such hetero‐valence Yb³⁺/Er³⁺ doping into the lattice of Li₄ZrF₈ NCs, the crystal lattice parameters and formation energies of pure, Yb³⁺ and/or Er³⁺ doped Li₄ZrF₈ NCs were further determined, assumed that Yb³⁺ and Er³⁺ ions are located at the substitutional Zr⁴⁺ lattice sites in the orthorhombic Li₄ZrF₈ crystal with an overall doping content of 12.5 mol% (Tables S2 and S3, Supporting Information). Although no apparent change in the crystal lattice parameters was observed for pure and Yb³⁺/Er³⁺‐doped Li₄ZrF₈ NCs, the formation energy per atom was calculated to decrease by about 0.07 eV when Zr⁴⁺ ions were partially replaced by Yb³⁺/Er³⁺couple, clearly indicating that the substitution of Zr⁴⁺ by Yb³⁺/Er³⁺ in the lattice of Li₄ZrF₈ NCs is thermodynamically favored regardless of their discrepancy in the valence state and ionic radius (0.84 Å for Zr⁴⁺, 0.99 Å for Yb³⁺ and 1.00 for Er³⁺ with a coordination number of eight).^{[ 30 ]}

On the basis of these first‐principle calculations, we then doped the typical upconverting lanthanide couple of Yb³⁺/Er³⁺ at precisely defined doping concentration (20/2 mol%) into the lattice of monodisperse Li₄ZrF₈ NCs via a modified high temperature coprecipitation method as previously reported.^{[ 31 ]} For comparison, Yb³⁺/Er³⁺ couple was also introduced into the host matrix of 3D close‐packed dilithium hexafluorozirconate (Li₂ZrF₆) possessing identical elemental compositions but different stoichiometric ratios as compared with Li₄ZrF₈ by using similar synthetic procedures (Figure S3, Supporting Information). All the as‐synthesized Li₄ZrF₈:Yb/Er and Li₂ZrF₆:Yb/Er UCNCs can be indexed in accordance with orthorhombic‐phase Li₄ZrF₈ (ICSD No. 80 398) and monoclinic‐phase Li₂ZrF₆ (ICSD No. 409 667) crystals, respectively (Figure  2a). Representative electron microscopy (TEM) images reveal that the as‐synthesized Li₄ZrF₈:Yb/Er UCNCs have a well‐defined peasecod‐like shape with an average length of 100 ± 15.2 nm and a width of 25 ± 5.1 nm (Figure 2b,c), and of which a large amount of internal cavities with an estimated pore size of 5–40 nm are clearly detected (Figure 2c), totally different from the case of Li₂ZrF₆:Yb/Er UCNCs possessing solid nanostructures (Figure 2d). Note that similar hollow nanostructures can be also formed in other Ln³⁺ ions (Yb/Tm, Yb/Ho, and Eu) doped Li₄ZrF₈ nanocrystals (Figure S4, Supporting Information). Regardless of their unique hollow nanoarchitecture formed, clear lattice fringes with a d‐spacing of 0.36 nm are observed at the edge of these internal cavities of peasecod‐like Li₄ZrF₈:Yb/Er UCNCs (Figure 2e), in good agreement with the lattice spacing of (201) plane of orthorhombic‐phase Li₄ZrF₈ crystal (ICSD No. 80 398) and thus reveals the formation of highly crystalline Li₄ZrF₈:Yb/Er UCNCs. Compositional analyses by energy‐dispersive X‐ray spectroscopy coupled with inductively coupled plasma atomic emission spectroscopy confirm the presence of host elements of Zr and F and the dopants of Yb and Er (Figure S5 and Table S4, Supporting Information), which clearly demonstrates the successful doping of Yb³⁺ and Er³⁺ ions in the as‐synthesized hollow Li₄ZrF₈:Yb/Er and solid Li₂ZrF₆:Yb/Er UCNCs.

Figure 2.

a) Powder XRD patterns of the as‐synthesized Li₄ZrF₈:Yb/Er and Li₂ZrF₆:Yb/Er UCNCs. b,c) TEM images of the hollow Li₄ZrF₈:Yb/Er UCNCs at different magnifications. The inset of c) schematically illustrates a peasecod‐like shape of the hollow Li₄ZrF₈:Yb/Er UCNCs. d) TEM image of the solid Li₂ZrF₆:Yb/Er UCNCs. e) High‐resolution TEM image of the hollow Li₄ZrF₈:Yb/Er UCNCs. f) N₂ adsorption–desorption isotherms and corresponding pore‐size distributions (inset) for the hollow Li₄ZrF₈:Yb/Er (red) and solid Li₂ZrF₆:Yb/Er (black) UCNCs.

To shed more light on the hollow Li₄ZrF₈:Yb/Er UCNCs, we further carried out the N₂ adsorption‐desorption experiments on Li₄ZrF₈:Yb/Er and Li₂ZrF₆:Yb/Er UCNCs. As compared in Figure 2f, the Brunauer–Emmet–Teller (BET) surface area and pore volume for the hollow Li₄ZrF₈:Yb/Er UCNCs were calculated to be 46.2 m² g⁻¹ and 0.07 cm³ g⁻¹, which are about eleven and nine times larger than those of their solid Li₂ZrF₆:Yb/Er counterparts (4.1 m² g⁻¹ and 0.008 cm³ g⁻¹), respectively. As a result, pores with sizes in the range of 2–25 nm can be clearly detected in the pore‐size distribution of Li₄ZrF₈:Yb/Er UCNCs, which differs markedly from the case of solid Li₂ZrF₆:Yb/Er UCNCs showing no any pores (Inset of Figure 2f). Note that the hollow nanostructure cannot come into being in the pure Li₄ZrF₈ NCs under otherwise identical synthetic conditions (Figure S6, Supporting Information), such significantly increased BET surface area and pore volume for the Li₄ZrF₈:Yb/Er UCNCs relative to the Li₂ZrF₆:Yb/Er counterparts unambiguously demonstrate that the hollow nanostructure observed in the Li₄ZrF₈:Yb/Er UCNCs is indeed originating from the Ln³⁺‐doping in the 2D layered crystal structure of Li₄ZrF₈, consistent well with our theoretical first‐principle calculations based on DFT (Figure 1e,f).

Another prominent feature for these hollow Li₄ZrF₈:Yb/Er UCNCs is that they can be rapidly synthesized in a relatively short period of time of ≈5 min. Figure  3a–d show the typical TEM images of the hollow Li₄ZrF₈:Yb/Er UCNCs prepared in various reaction time of 5, 10, 30, and 60 min, respectively. Despite the dramatically shortened reaction time, the well‐defined hollow nanostructure for the peasecod‐like Li₄ZrF₈:Yb/Er UCNCs prepared in 5 min keeps essentially identical to those synthesized at prolonged reaction time ranging from 10 to 60 min. Particularly, in all cases, their XRD peaks for Li₄ZrF₈:Yb/Er UCNCs prepared in different reaction time match the standard orthorhombic‐phase Li₄ZrF₈ crystal in terms of line position and full width at half maximum, which strongly support that the formation of hollow Li₄ZrF₈:Yb/Er UCNCs can be quickly completed in a very short time down to ≈5 min (Figure 3e). In addition to the rapid synthesis, we would like to emphasize that the hollow Li₄ZrF₈:Yb/Er UCNCs can be also prepared in large scale by only scaling up the reaction variables such as the amount of Zr precursor. In this way, over 1.1 g of Li₄ZrF₈:Yb/Er UCNCs were synthesized via a one‐pot reaction, where the crystal phase, morphology, nanocrystal size as well as the UCL performance for the resulting Li₄ZrF₈:Yb/Er UCNCs were found to be preserved during scaling up (Figure 3e–h and Figure S7, Supporting Information).

Figure 3.

a–d) TEM images of the hollow Li₄ZrF₈:Yb/Er UCNCs prepared in different reaction time of 5, 10, 30, and 60 min, respectively. e) XRD patterns of the hollow Li₄ZrF₈:Yb/Er UCNCs prepared in different reaction time and via a large‐scale synthesis. f) TEM and high‐resolution TEM (inset) images of the hollow Li₄ZrF₈:Yb/Er UCNCs via the large‐scale synthesis. g) Photographs of the hollow Li₄ZrF₈:Yb/Er UCNCs via the large‐scale synthesis dispersed in 500 mL cyclohexane under daylight (left) and upon excitation at 980 nm (right). h) Photographs showing the weight (1.1 g) of the resulting hollow UCNCs via a one‐pot synthesis under daylight (left) and UCL upon irradiation at 980 nm (right), respectively.

Figure  4a compares the representative room‐temperature UCL spectra for the hollow Li₄ZrF₈:Yb/Er, solid Li₂ZrF₆:Yb/Er and solid β‐NaYF₄:Yb/Er UCNCs when excited by using a 980‐nm diode laser at a power density of 10 W cm⁻². The characteristic UCL bands arising from the ²H_9/2→⁴I_15/2 (406 nm), ²H_11/2→⁴I_15/2 (525 nm), ⁴S_3/2→⁴I_15/2 (545 nm), and ⁴F_9/2→⁴I_15/2 (650 nm) transitions of Er³⁺ are readily detected in the visible spectral regions for both UCNCs. Regardless of their much larger S/V ratio, the overall UCL intensity for hollow Li₄ZrF₈:Yb/Er UCNCs was determined to be about 3.4 times stronger than that of solid Li₂ZrF₆:Yb/Er samples (Figure 4b and Figure S8, Supporting Information), which differs markedly from the previous observation showing that the large S/V ratio of Ln³⁺‐doped UCNCs usually leads to weak UCL due to severe luminescence quenching.^{[ 32 ]} Accordingly, the photoluminscence quantum yield (PLQY) was observed to stepwise decrease with the phase transformation from hollow Li₄ZrF₈:Yb/Er (≈0.50%) to solid Li₂ZrF₆:Yb/Er (≈0.29%) UCNCs (Figure 4c and Figure S8, Supporting Information). Especially, the integrated luminescence intensity ratio (LIR) between these two thermally coupled levels (²H_11/2 and ⁴S_3/2) of Er³⁺ in the hollow Li₄ZrF₈:Yb/Er UCNCs (termed as I _H/I _S) was determined to 1/2, which is about two times larger than that observed in the solid Li₂ZrF₆:Yb/Er and β‐NaYF₄:Yb/Er controls (1/4, Figure 4a). This experimental observation strongly supports that the population ratio for the ²H_11/2 state to ⁴S_3/2 state of Er³⁺ in the hollow Li₄ZrF₈:Yb/Er UCNCs is much larger than in the solid Li₂ZrF₆:Yb/Er ones (Figure 4d), as further evidenced by the significantly prolonged ²H_11/2 lifetime of Er³⁺ from the solid Li₂ZrF₆:Yb/Er (≈0.08 ms) to hollow Li₄ZrF₈:Yb/Er (≈0.29 ms) UCNCs (Figure 4e).

Figure 4.

a) Comparison of typical UCL spectra for the hollow Li₄ZrF₈:Yb/Er, solid Li₂ZrF₆:Yb/Er, and solid β‐NaYF₄:Yb/Er UCNCs under excitation of a 980‐nm diode laser. b) Comparison of UCL spectra for the hollow Li₄ZrF₈:Yb/Er, solid Li₂ZrF₆:Yb/Er, and the mixed‐phase UCNCs, and c) their corresponding PLQYs, demonstrating the gradually increased UCL performance from the solid Li₂ZrF₆:Yb/Er to hollow Li₄ZrF₈:Yb/Er UCNCs. d) Proposed mechanism for green UC emissions assigned to the ²H_11/2→⁴I_15/2 (centered at 525 nm) and ⁴S_3/2→⁴I_15/2 (centered at 545 nm) transitions of Er³⁺, showing their different population ratios for the ²H_11/2 and ⁴S_3/2 states of Er³⁺ in the hollow Li₄ZrF₈:Yb/Er and solid Li₂ZrF₆:Yb/Er UCNCs, respectively. e) Comparison of the ²H_11/2 lifetime of Er³⁺ in the hollow Li₄ZrF₈:Yb/Er and solid Li₂ZrF₆:Yb/Er UCNCs, respectively.

Owing to the small energy gap (ΔE) between the two energy levels, in theory, the ²H_11/2 level of Er³⁺ can be easily populated from its low‐lying ⁴S_3/2 state via thermal excitation. Consequently, the population ratio for the ²H_11/2 state to ⁴S_3/2 state of Er³⁺ can be described by the Boltzmann's distribution (∝‐ΔE/K _B T, here K _B is the Boltzmann's constant and T is the temperature in Kelvins) and thus scale linearly with the integrated LIR of I _H/I _S.^{[ 33} , ³⁴ , ³⁵ , ^{36 ]} For Er³⁺ in the hollow Li₄ZrF₈:Yb/Er and solid Li₂ZrF₆:Yb/Er UCNCs, the ΔE _s were calculated to be about 597.4 and 523.2 cm⁻¹ from their respective UCL spectra (Figure 4a). Theoretically, the population ratio for the ²H_11/2 state to ⁴S_3/2 state of Er³⁺ in the hollow Li₄ZrF₈:Yb/Er UCNCs with enlarged ΔE should be smaller than that in their solid Li₂ZrF₆:Yb/Er counterparts, which is obviously contrary to our experimental finding abovementioned (Figure 4d). Considering their completely identical elemental compositions of Li, Zr, and F for the hollow Li₄ZrF₈:Yb/Er and solid Li₂ZrF₆:Yb/Er UCNCs, we attribute this unusual increase in the LIR of I _H/I _S to the hollow nanostructure of the peasecod‐like Li₄ZrF₈:Yb/Er UCNCs. To reconfirm this effect of the hollow nanostructure, we further synthesized the solid β‐NaYF₄:Yb/Er (20/2 mol%) UCNCs as controls (Supporting Information). Indeed, the integrated LIR of I _H/I _S and ²H_11/2 lifetime for Er³⁺ in the solid β‐NaYF₄:Yb/Er UCNCs were also found to be virtually identical to those of the solid Li₂ZrF₆:Yb/Er UCNCs (Figure 4a and Figure S9, Supporting Information), thereby providing another evidence to turn out that the increased LIR of I _H/I _S is due to the hollow nanostructure formed in the peasecod‐like Li₄ZrF₈:Yb/Er UCNCs.

In view of the unusual population ratio of the ²H_11/2 state to ⁴S_3/2 state of Er³⁺ associated with the hollow nanostructure, we reasoned that Er³⁺ ions in these peasecod‐like Li₄ZrF₈:Yb/Er UCNCs would be more sensitive to the temperature when compared with their solid counterparts. To verify the temperature effect on UCL, we recorded the temperature‐dependent UCL spectra for the hollow Li₄ZrF₈:Yb/Er and solid Li₂ZrF₆:Yb/Er UCNCs in the temperature range from 10 to 800 K. The intensity ratios of I _H/I _S of Er³⁺ for both the Li₄ZrF₈:Yb/Er and Li₂ZrF₆:Yb/Er UCNCs increased with rising temperature, as a result of enhanced thermal population of the ²H_11/2 state from the ⁴S_3/2 state at higher temperature (Figure  5a and Figure S10, Supporting Information). However, as compared in Figure 5b, the hollow Li₄ZrF₈:Yb/Er UCNCs exhibited an excellent linear relationship between the ln(I _H/I _S) and inverse temperature (1/T) in the temperature range from 123 to 800 K, much wider than that of their solid Li₂ZrF₆:Yb/Er counterparts (223–800 K). Note that such a linear range of 123–800 K for the hollow Li₄ZrF₈:Yb/Er UCNCs is among the widest temperature ranges for the Yb³⁺/Er³⁺ codoped luminescent nanothermometers previously reported (Table  1 ).^{[ 12} , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ^{43 ]} In addition to the linear temperature response range, the thermal sensitivities for the hollow Li₄ZrF₈:Yb/Er UCNCs also outperform their solid Li₂ZrF₆:Yb/Er counterparts (Figure 5c). For example, the highest absolute temperature sensitivity (S _a, defined as ∂LIR/∂T)^{[ 44} , ^{45 ]} and highest relative temperature sensitivity (S _r, defined as (1/LIR)(∂LIR/∂T))^{[ 44} , ^{45 ]} for the hollow Li₄ZrF₈:Yb/Er UCNCs were calculated to be 0.52% K⁻¹ at 523 K and 5.65% K⁻¹ at 123 K, which are about two and 1.43 times higher than those of solid Li₂ZrF₆:Yb/Er UCNCs (0.26% K⁻¹ at 523 K and 3.95% K⁻¹ at 123 K). Particularly, when compared with other Yb³⁺/Er³⁺ codoped UCNCs, the thermal sensitivities of the hollow Li₄ZrF₈:Yb/Er UCNCs are also superior to the most previously reported temperature sensing systems (Table 1).^{[ 12} , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ^{43 ]}

Figure 5.

a) Temperature‐dependent UCL spectra for the hollow Li₄ZrF₈:Yb/Er UCNCs recorded at the temperature range of 10–800 K. The peaks are normalized at 542 nm. b) Plots of ln(I _H/I _S) versus 1/T and the fitted curves to calibrate the thermometric scale for the hollow Li₄ZrF₈:Yb/Er and solid Li₂ZrF₆:Yb/Er UCNCs, respectively. c) Calculated temperature sensitivities (S _a and S _r) versus T in 123–800 K for the hollow Li₄ZrF₈:Yb/Er and solid Li₂ZrF₆:Yb/Er UCNCs, respectively. d) Variation of the intensity ratio I _H/I _S measured at 123–800 K over a span of 20 cycles of heating and cooling processes for the hollow Li₄ZrF₈:Yb/Er UCNCs.

Table 1.

Comparison of various Yb³⁺/Er³⁺ co‐doped UCNCs as luminescent nanothermometers by using LIR technique in the green spectral region

Material	Temperature range [K]	S a, max [% K−1]	S r, max [% K−1]	References
Li4ZrF8:Yb/Er	123–800	0.52	5.65	This work
Li2ZrF6:Yb/Er	123–800	0.26	3.95	This work
KMnF3:Yb/Er	303–343	1.13	5.7	[37]
NaYF4:Yb/Er	173–353	–	3.58	[38]
CaF2:Yb/Er	293–318	–	2.3	[39]
Gd2O3:Yb/Er	300–900	0.39	0.83	[40]
SrF2:Yb/Er	303–373	–	1.21	[12]
Ba5Gd8Zn4O21:Yb/Er	260–490	0.32	1.69	[41]
csUCNP@C	293–343	–	1.1	[42]
YF3:Yb/Er	260–490	0.26	1.48	[43]

Moreover, such temperature evolution of I _H/I _S for the hollow Li₄ZrF₈:Yb/Er UCNCs was found to be reversible during the heating and cooling cycles in the temperature range of 123–800 K. The intensity ratios of I _H/I _S recorded at 123 and 800 K are nearly unchanged with deviations smaller than 0.5% over a span of 20 cycles during heating and cooling processes, indicative of the high reliability of these hollow Li₄ZrF₈:Yb/Er UCNCs as optical thermometric materials (Figure 5d). Taken together, these results unambiguously demonstrate that the as‐synthesized peasecod‐like hollow Li₄ZrF₈:Yb/Er UCNCs are highly promising candidates as luminescent nanothermometers for optical temperature sensing. Currently, further efforts on the practical use of these hollow nanocrystals as in vivo temperature sensor or drug delivery carrier are underway in our laboratory.

3. Conclusion

In summary, we reported for the first time a peasecod‐like hollow Li₄ZrF₈:Yb/Er UCNCs that were synthesized through a modified high‐temperature coprecipitation method. By utilizing the first‐principle calculations based on DFT, we revealed that the hollow nanostructure of Li₄ZrF₈:Yb/Er UCNCs arouse primarily from the hetero‐valence substituted doping of Ln³⁺ into the Li₄ZrF₈ host lattice that features a 2D layered crystal structure. Furthermore, these peasecod‐like hollow Li₄ZrF₈:Yb/Er UCNCs were able to be prepared rapidly and in large scale with preserved crystal phase, morphology, size, and UCL properties. Benefiting from the hollow nanostructure, these Li₄ZrF₈:Yb/Er UCNCs exhibited an abnormal UCL performance in comparison with their solid counterparts, thereby endowing them with excellent optical thermometric performances including a wide temperature range, relatively high thermal sensitivity and high photochemical stability. These findings unambiguously pave a new way to construct the hollow Ln³⁺‐doped inorganic UCNCs for various applications such as optical temperature sensing and bioimaging.

4. Experimental Section

General Procedure for the Preparation of Peasecod‐Like Hollow Li₄ZrF₈:Yb/Er UCNCs

In a typical procedure for the synthesis of Li₄ZrF₈:Yb/Er UCNCs (1 mmol), 2 mmol of LiOH·2H₂O, 0.78 mmol of Zr(CH₃COO)₄, 0.2 mmol of Yb(CH₃COO)₃·4H₂O, and 0.02 mmol of Er(CH₃COO)₃·4H₂O were mixed with 8 mL of OA and 16 mL of ODE in a 100 mL three‐neck round‐bottom flask. The solution was heated to 150 °C under N₂ flow with constant stirring for 60 min to form a clear solution, and then cooled down to room temperature. Thereafter, 10 mL of methanol solution containing 3 mmol of NH₄F was added and the resulting mixture was stirred for 30 min. After removal of the methanol by evaporation, the solution was heated to 280 °C under N₂ flow with vigorous stirring for 60 min, and then cooled down to room temperature. The resulting Li₄ZrF₈:Yb/Er UCNCs were precipitated by addition of ethanol, collected by centrifugation, washed with ethanol and cyclohexane for several times, and finally redispersed in cyclohexane. As for the large‐scale synthesis of 20 mmol hollow Li₄ZrF₈:Yb/Er UCNCs, identical experimental procedures were used except for proportionately scaling up the reaction variables including the amount of precursors and solvents. In addition, the general procedures for the preparation of solid Li₂ZrF₆:Yb/Er and β‐NaYF₄:Yb/Er UCNCs are provided in the Supporting Information.

Structural and Optical Characterization

Powder XRD measurements were performed on a powder diffractometer (MiniFlex2, Rigaku) with Cu Kα1 radiation (λ = 0.154187 nm) from 10° to 70° at a scanning rate of 5° min⁻¹. Both the TEM and high‐resolution TEM measurements were conducted on a TEM (TECNAI G2F20) equipped with an energy dispersive X‐ray spectroscope. The N₂ adsorption‐desorption isotherms and BET surface area of samples were determined by gas sorption analysis (Micromeritics ASAP 2020, 77 K). UCL spectra were measured upon 980‐nm NIR excitation from a continuous‐wave diode laser. UCL photographs of the UCNCs were taken by using a Canon 70D digital camera without using any filter. All the UCL decay curves for Yb³⁺/Er³⁺ co‐doped UCNCs were measured with a customized UV to mid‐infrared steady‐state and phosphorescence lifetime spectrometer (FSP920‐C, Edinburgh Instrument) equipped with a digital oscilloscope (TDS3052B, Tektronix) and a tunable mid‐band Optical Parametric Oscillator (OPO) pulse laser as the excitation source (410–2400 nm, 10 Hz, pulse width ≤5 ns, Vibrant 355II, OPOTEK). For temperature‐dependent UCL measurements, the as‐synthesized UCNCs were put inside the Linkam THMS600E heating and freezing stage with a tunable temperature range from 10 to 800 K and heating rate of 60 °C min⁻¹. The as‐prepared UCNCs were in situ cooled or heated on the stage, whose temperature was controlled to a certain temperature by a temperature controller (Linkam LNP95) and held for 5 min to keep the temperature stable, and then proceeded to a next temperature. All the spectral data were corrected for the spectral response of the spectrometer.

Computational Details

A theoretical assessment for the pure, Yb³⁺ and/or Er³⁺ doped Li₄ZrF₈ system was performed using the Vienna Ab initio Simulation Package (VASP).^{[ 46 ]} The generalized gradient approximation (GGA) with Perdew–Burke–Ernzerhof^{[ 47 ]} functional and the projector‐augmented‐wave pseudopotentials method were introduced to compute the exchange‐correlation and valence‐core interactions, respectively. The geometric optimizations and the formation energy (E _form) of the pure and Yb³⁺ and/or Er³⁺ doped Li₄ZrF₈ NCs were first evaluated to confirm the successful hetero‐valence doping of Ln³⁺ ions in the Li₄ZrF₈ host lattice. To study the doping effect, the 1 × 1 × 2 supercell containing 104 atoms was constructed, in which one and two zirconium ions were replaced by Ln³⁺ ions with doping concentrations of 12.5% and 25% respectively. Considering the strongly‐correlated character of 4f valence electrons of Yb³⁺ and Er³⁺ ions, the Heyd–Scuseria–Ernzerhof hybrid functional (HSE06) or GGA+Umethods were adopted to provide more accurate electronic properties for Ln³⁺‐doped materials.^{[ 48} , ⁴⁹ , ^{50 ]} Due to the large number of atoms, calculations with the HSE06 functional were crudely terminated. Therefore, all of the calculated results were obtained by using the GGA+U method. It should be mentioned that the U _eff values introduced by Dudarev were employed as the U values for Zr and Yb/Er atoms, in which three equations of U _eff = U − J, J = 0 and U _eff = U were set up and values of 2.0 and 6.0 eV were appointed to zirconium and lanthanide ions, respectively.^{[ 51 ]} The kinetic energy cutoff was set to 400 eV. All of the atomic positions and cell parameters were allowed to relax until the limits of 1 × 10⁻⁶ eV for energy and −0.01 eV Å⁻¹ for force, respectively. For the Brillouin zone sampling, the 3 × 3 × 3 Γ‐centered k‐point mesh was used for all calculations. Moreover, the post‐processing and graphics of the calculation results were completed by using the VASPKIT and VESTA softwares.^{[ 52} , ^{53 ]}

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Fu H., Liu C., Peng P., Jiang F., Liu Y., Hong M., Peasecod‐Like Hollow Upconversion Nanocrystals with Excellent Optical Thermometric Performance. Adv. Sci. 2020, 7, 2000731 10.1002/advs.202000731