Unraveling the Electronic Structures of Neodymium in LiLuF4 Nanocrystals for Ratiometric Temperature Sensing

Abstract Nd3+‐doped near‐infrared (NIR) luminescent nanocrystals (NCs) have shown great promise in various bioapplications. A fundamental understanding of the electronic structures of Nd3+ in NCs is of vital importance for discovering novel Nd3+‐activated luminescent nanoprobes and exploring their new applications. Herein, the electronic structures of Nd3+ in LiLuF4 NCs are unraveled by means of low‐temperature and high‐resolution optical spectroscopy. The photoactive site symmetry of Nd3+ in LiLuF4 NCs and its crystal‐field (CF) transition lines in the NIR region of interest are identified. By taking advantage of the well‐resolved and sharp CF transition lines of Nd3+, the application of LiLuF4:Nd3+ NCs as sensitive NIR‐to‐NIR luminescent nanoprobes for ratiometric detection of cryogenic temperature with a linear range of 77–275 K is demonstrated. These findings reveal the great potential of LiLuF4:Nd3+ NCs in temperature sensing and also lay a foundation for future design of efficient Nd3+‐based luminescent nanoprobes.


Introduction
Trivalent neodymium (Nd 3+ ) ion doped luminescent nanocrystals (NCs) have recently attracted considerable attention owing to their superior optical properties in the near-infrared (NIR) spectral region. [1] These Nd 3+ -doped NCs are able to emit NIR light under excitation with a low-cost 808 nm diode laser, which feature a series of advantages such as large penetration depth, minimal background interference, and little damage to the targeted samples, and thus are regarded as excellent NIR-to-NIR luminescent nanoprobes for various bioapplications. [2] Specifically, Nd 3+ -doped NCs have been frequently used as sensitive

Results and Discussion
The LiLuF 4 crystal has a scheelite structure (space group I4 1 /a) with Lu 3+ ions surrounded by eight F − ions that form the edges of a slightly distorted dodecahedron. All Lu 3+ ions occupy a single crystallographic site of S 4 symmetry (Figure 1a). Highquality LiLuF 4 :Ln 3+ (Ln = Eu and Nd) NCs were synthesized through a thermal decomposition method as we previously reported. [6b] The as-synthesized NCs are hydrophobic and can be readily dispersed in a variety of nonpolar organic solvents such as cyclohexane. X-ray diffraction (XRD) patterns (Figure 1b) show that all diffraction peaks of the NCs can be well indexed into tetragonal LiLuF 4 (JCPDS No. 027-1251) without any additional impurities. Transmission electron microscopy (TEM) images show that both LiLuF 4 :Nd 3+ and LiLuF 4 :Eu 3+ NCs are rhomboid with mean sizes of (28.4 ± 1.2) × (33.1 ± 1.5) and (28.0 ± 0.9) × (34.2 ± 1.0) nm, respectively (Figure 1c-e). Highresolution TEM images (insets of Figure 1c,d) exhibit clear lattice fringes with an observed d spacing of 0.46 nm for the (101) plane of tetragonal LiLuF 4 , confirming pure phase and high crystallinity of the resulting NCs. Compositional analyses through energy-dispersive X-ray spectrum and inductively coupled plasma-atomic emission spectroscopy reveal 1.8 mol% of Nd 3+ and 4.6 mol% of Eu 3+ in LiLuF 4 matrix ( Figure S1, Supporting Information), which are generally consistent with their nominal dopant concentrations (2 mol% of Nd 3+ and 5 mol% of Eu 3+ ).
Eu 3+ ion is a sensitive spectroscopic probe, which can provide site symmetry information because of its nondegenerate emissive state of 5 D 0 and the ground state of 7 F 0 . [8] To probe the practical local site symmetry of Ln 3+ dopants in LiLuF 4 NCs, we measured the high-resolution PL spectra of Eu 3+ in LiLuF 4 NCs. Emission and excitation spectra and PL decays were recorded at 10 K to avoid thermal broadening of spectral lines at room temperature ( Figure S2, Supporting Information). [9] Figure 2 shows the high-resolution PL spectra of LiLuF 4 :5%Eu 3+ NCs at 10 K, which enables a detailed assignment of the CF transition lines of Eu 3+ . By monitoring the Eu 3+ emission at 613.8 nm, a series of CF transition lines of Eu 3+ from the 7 F 0 ground state to the excited multiplets ( 5 D J , 5 L J , 5 G J , 5 H J , and 5 F J ) were observed ( Figure 2a). [10] Upon excitation to 5 L 6 of Eu 3+ at 393.0 nm, the CF emission peaks from 5 D 0 and 5 D 1 to 7 F J (J = 0, 1, 2, 3, and 4) with full-width at half-maximum (FWHM) smaller than 0.5 nm were detected ( Figure 2b). PL decay measurements show that both 5 D 0 and 5 D 1 display a single exponential decay with PL lifetimes of 11.3 and 2.7 ms, respectively (Figure 2c), suggesting a homogeneous CF environment around Ln 3+ dopants in LiLuF 4 lattice. [11] The distinct PL lifetimes of 5 D 0 and 5 D 1 allow us to distinguish the emission peaks of 5 D 0 from those of 5 D 1 by means of TRPL spectroscopy. Figure 2d shows the TRPL spectra of LiLuF 4 :5%Eu 3+ NCs at 10 K with different Adv. Sci. 2019, 6, 1802282 delay times. It was observed that the emission peaks from the short-lived 5 D 1 level declined gradually with increasing the delay time and totally vanished when the delay time was longer than 6 ms, while the emission peaks from the long-lived 5 D 0 level remained explicitly observed in the TRPL spectra even at a delay time of 10 ms. As a result, total numbers of 0, 2, 3, 4, and 4 CF transition lines of Eu 3+ from 5 D 0 to 7 F 0 , 7 F 1 , 7 F 2 , 7 F 3 , and 7 F 4 can be discerned in LiLuF 4 NCs. To check whether all these transition lines arise from the same site, site-selective excitation spectra were measured by monitoring the three peaks of 5 D 0 → 7 F 2 at 610.4, 613.8, and 620.8 nm. The obtained excitation spectra were coincident ( Figure S3, Supporting Information), indicating that the PL originated from Eu 3+ ions occupying a single spectroscopic site, as also evidenced by the essentially identical site-selective emission spectra upon excitation to 5 L 6 at 393.0, 395.8, 399.4, and 400.6 nm and the same PL lifetimes of the 5 D 0 → 7 F 1 emissions at 590.6 and 593.9 nm ( Figure 2e and Figure S4, Supporting Information). According to the branching rules and the transition selection rules of the 32 point groups (Table S1, Supporting Information), [8b,c] the spectroscopic site symmetry of Eu 3+ in LiLuF 4 NCs was determined to be S 4 , which agrees well with the crystallographic site symmetry of Lu 3+ in LiLuF 4 . These results suggest that Ln 3+ ions are prone to occupy a single spectroscopic site of S 4 symmetry in LiLuF 4 NCs at low doping levels (<5 mol%) due to the close ionic radii and chemical properties of Ln 3+ ions.
With definite local site symmetry of Ln 3+ dopants, we are able to assign the CF transition lines of Nd 3+ in LiLuF 4 NCs by means of high-resolution PL spectroscopy. Figure 3a shows the PL excitation spectrum of Nd 3+ in LiLuF 4 NCs at 10 K by monitoring the Nd 3+ emission at 1053.2 nm, from which a series of CF transition lines of Nd 3+ from the 4 I 9/2 ground state to the excited multiplets ( 4 F J , 2 H J , 4 S J , 2 G J , 4 G J , 2 D J , and 4 D J ) were identified. [12] Specifically, two excitation peaks at 861.4 and 865.9 nm were clearly observed (inset of Figure 3a), ascribing to the CF transitions of Nd 3+ from the 4 I 9/2 ground state to the upper (R 2 ) and lower (R 1 ) Stark sublevels of 4 F 3/2 , respectively. This implies that the CF levels of Nd 3+ are doubly degenerate in LiLuF 4 NCs, as expected for a Kramers ion. [13] 10 K PL emission spectrum (Figure 3b) shows that the NCs exhibit a set of characteristic and sharp emission peaks (FWHM < 0.9 nm) from the two Stark sublevels of 4 F 3/2 (R 1 and R 2 ) to those of 4 I 9/2 , 4 I 11/2 , and 4 I 13/2 of Nd 3+ under xenon lamp excitation at 791.3 nm. To confirm that all these transition lines arise from a single site, we recorded PL emission spectrum of the NCs by exciting them with an 808 nm diode laser, whereby all possible spectroscopic sites of Nd 3+ could be excited in view of the high power density (50 W cm −2 ) and relatively wide FWHM (3.2 nm) of the laser source ( Figure S5, Supporting Information). It turned out that the emission pattern of the NCs under 808 nm diode laser excitation was exactly identical to that under xenon lamp excitation at 791.3 nm ( Figure S6, Supporting Information), inferring that the PL originates from Nd 3+ ions occupying a single spectroscopic site. From the emission spectrum, total numbers of 10, 12, and 14 CF transition lines of Nd 3+ from 4 F 3/2 to 4 I 9/2 , 4 I 11/2 , and 4 I 13/2 were discerned, which agree well Adv. Sci. 2019, 6, 1802282   Figure 2. a) 10 K PL excitation spectrum of LiLuF 4 :5%Eu 3+ NCs by monitoring the Eu 3+ emission at 613.8 nm and b) their emission spectrum upon excitation at 393.0 nm. The inset in (b) enlarges the 5 D 0 → 7 F 4 emissions including four CF transition lines, and the asterisks represent the 5 D 1 → 7 F 4 emissions of Eu 3+ . c) PL decay curves of LiLuF 4 :5%Eu 3+ NCs by monitoring the 5 D 0 → 7 F 2 and 5 D 1 → 7 F 1 emissions of Eu 3+ at 613.8 and 582.8 nm, respectively. d) 10 K TRPL spectra of LiLuF 4 :5%Eu 3+ NCs with different delay times under excitation at 393.0 nm. The asterisks denote the CF transition lines from the 5 D 1 multiplet of Eu 3+ . e) PL decay curves of LiLuF 4 :5%Eu 3+ NCs by monitoring the 5 D 0 → 7 F 1 emissions of Eu 3+ at 590.6 and 593.9 nm.
with the theoretically predicted numbers for Nd 3+ in LiLuF 4 with doubly degenerate CF levels. [13] These results demonstrate unambiguously that Nd 3+ ions occupy a single spectroscopic site of S 4 symmetry in LiLuF 4 NCs, as revealed by using Eu 3+ as the structural probe.
Because the energy difference between the R 1 and R 2 Stark sublevels of 4 F 3/2 is only 58 cm −1 , the higher Stark sublevel (R 2 ) is easily thermally populated from the lower one (R 1 ). As a result, PL intensity ratio between the emissions from R 2 and R 1 would increase with the temperature rise, enabling discrimination of the CF transition lines of R 2 from those of R 1 through the temperature-dependent PL emission spectra (Figure 3c). It was observed that the intensities of the CF emission peaks at 861.9, 872.2, 875.7, 882.3, and 904.9 nm increased significantly with the temperature rise, corresponding to the transitions from the R 2 sublevel of 4 F 3/2 to the five CF levels (Z 1 , Z 2 , Z 3 , Z 4 , and Z 5 ) of 4 I 9/2 (Figure 3d). [12c] By contrast, the intensities of the CF emission peaks at 866. 2, 876.9, 880.4, 886.4, and 909.8 nm showed only a slight increase with the temperature rise, corresponding to the transitions from the R 1 sublevel of 4 F 3/2 . The slight increase in R 1 lines is caused by the spectral overlap with R 2 lines. Based on the defined Adv. Sci. 2019, 6, 1802282   Figure 3. a) 10 K PL excitation spectrum of LiLuF 4 :2%Nd 3+ NCs by monitoring the Nd 3+ emission at 1053.2 nm and b) their emission spectrum upon excitation at 791.3 nm. The inset in (a) shows two CF transition lines from the 4 I 9/2 ground state to the upper and lower Stark sublevels of 4 F 3/2 . c) Temperature-dependent PL emission spectra (10-300 K) for the 4 F 3/2 → 4 I J (J = 9/2, 11/2, and 13/2) CF transitions of Nd 3+ in LiLuF 4 NCs upon 808 nm diode laser excitation at a power density of 1 W cm −2 . The spectra were normalized at the maximum intensities around 880.4, 1053.1, and 1325.1 nm for the emissions from 4 F 3/2 to 4 I 9/2 , 4 I 11/2 , and 4 I 13/2 , respectively. The dashed lines denote the CF transitions from the R 1 (black) and R 2 (red) Stark sublevels of 4 F 3/2 to those of 4 I J . d) CF energy levels of the 4 F 3/2 and 4 I J multiplets of Nd 3+ in LiLuF 4 NCs, showing all CF transitions observed in (c).
CF transition lines, the CF levels of 4 I 9/2 can be unequivocally identified, as listed in Table 1. Similarly, CF transition lines from the R 1 and R 2 sublevels of 4 F 3/2 to those of 4 I 11/2 and 4 I 13/2 can be specified by virtue of the temperature-dependent PL emission spectra, from which the CF levels of 4 I 11/2 and 4 I 13/2 were experimentally assigned (Table 1). Besides, we also found a redshift in CF transition lines of Nd 3+ with the temperature rise, especially for the transitions from R 2 of 4 F 3/2 to Z 1 and Z 2 of 4 I 11/2 (Figure 3c), as a result of enhanced electron-phonon coupling at higher temperatures. [4b] Importantly, we found that the CF transition lines from the thermally coupled R 1 and R 2 Stark sublevels of 4 F 3/2 to Z 1 of 4 I 9/2 at 862 nm (R 2 →Z 1 ) and 866 nm (R 1 →Z 1 ) are well resolved with little interference from other Stark components at temperatures below 300 K. This feature makes LiLuF 4 :Nd 3+ NCs an ideal nanoprobe candidate for ratiometric luminescent detection of temperature below 300 K by using the temperature-dependent PL intensity ratio between the R 2 →Z 1 and R 1 →Z 1 transitions at 862 and 866 nm (I 862 /I 866 ), respectively.
To validate the applicability of LiLuF 4 :Nd 3+ NCs for temperature sensing, we deconvoluted the 4 F 3/2 → 4 I 9/2 emission spectra of Nd 3+ into ten Gaussian components according to the CF transitions between their Stark sublevels (Figure 4a), from which the PL intensity ratio between R 2 →Z 1 and R 1 →Z 1 (I 862 /I 866 ) was derived. Further temperature-correlated PL emission spectra showed that the PL intensity ratio I 862 /I 866 increased gradually with increasing the temperature from 77 to 575 K ( Figure S7, Supporting Information), as a result of enhanced thermal population of the R 2 sublevel from the R 1 sublevel of 4 F 3/2 at higher temperatures. Specifically, the ratio of I 862 /I 866 displayed a linear dependence on temperature in the range of 77-275 K, with its value increased from 1.46 at 77 K to 3.23 at 275 K ( Figure 4b). Moreover, such temperature evolution of I 862 /I 866 was found to be reversible during the heating and cooling cycle between 77 and 275 K. The ratios of I 862 /I 866 recorded at 77, 175, and 275 K were nearly unchanged with deviations smaller than 0.5% over a span of 20 cycles of heating and cooling processes Adv. Sci. 2019, 6, 1802282  Figure 4. a) PL emission spectrum for the 4 F 3/2 → 4 I 9/2 transitions of Nd 3+ in LiLuF 4 NCs at 275 K and its Gaussian fit according to the CF transitions. b) PL intensity ratio between the R 2 →Z 1 and R 1 →Z 1 CF transitions at 862 and 866 nm (I 862 /I 866 ) as a function of temperature during a heating and cooling cycle between 77 and 275 K. Each data point represents the mean (±standard deviation) of three independent measurements. c) Variation of the intensity ratio I 862 /I 866 recorded at 77, 175, and 275 K measured over a span of 20 cycles of heating and cooling processes. d) The relative temperature sensitivity (S r ) of LiLuF 4 :2%Nd 3+ nanoprobes as a function of temperature. The error bars result from error propagation in the determination of S r .
( Figure 4c), as a merit of high photochemical stability of the NCs. [14] This indicates that the PL of LiLuF 4 :Nd 3+ NCs is fully reversible without any observable thermal hysteresis in the temperature range of 77-275 K ( Figure S8, Supporting Information), which is of key importance for temperature sensing by using a luminescent nanoprobe. [15] The absolute temperature sensitivity (S a ) of the nanoprobe, defined as the change of response R with temperature, namely, ∂R/∂T where R is the PL intensity ratio I 862 /I 866 and T the absolute temperature, [16] was calculated to be a constant of 0.00913 K −1 , which is among the highest S a values for Nd 3+ -activated luminescent nanothermometer ever reported. [3b,17] The relative temperature sensitivity (S r ), defined as the fractional rate of the change of response R with temperature, namely, (1/R)(∂R/∂T), [16] was plotted in Figure 4d, from which the highest S r was determined to be 0.62% K −1 at 77 K. The highest S r obtained in LiLuF 4 :Nd 3+ NCs is comparable to the best S r values for Nd 3+ -activated luminescent nanothermometers previously reported (Table S2, Supporting Information). [18] The temperature uncertainty (δT), defined as the relative error of the response (δR/R) versus the relative temperature sensitivity (S r ), [1i,19] was calculated to be lower than 0.6 K for temperatures below 250 K ( Figure S9, Supporting Information). It is worth mentioning that the spectral overlap between different Stark or CF components in the emission spectrum of Nd 3+ is unavoidable at high temperatures because of the CF line broadening and shifting and the multiple sites of Nd 3+ with distinct CF surroundings in NCs. Therefore, the assignment of CF transition lines in Nd 3+ -doped NCs should be judiciously carried out by taking into account the interference from different CF or Stark components, which is a prerequisite for temperature sensing by using Nd 3+ -doped luminescent nanoprobes. In this regard, the superior features combined with the single photoactive site symmetry, well-resolved CF transition lines, and high photostability of LiLuF 4 :Nd 3+ NCs, make them excellent NIR-to-NIR luminescent nanoprobes for ratiometric temperature sensing in practical application.

Conclusions
In summary, we have systematically investigated the local site symmetry and electronic structures of Nd 3+ in LiLuF 4 NCs through low-temperature and high-resolution PL spectroscopy. By employing Eu 3+ as the structural probe, a single spectroscopic site of S 4 symmetry for Ln 3+ dopants was identified in LiLuF 4 NCs, which is consistent with the crystallographic site symmetry of Lu 3+ in LiLuF 4 lattice. By means of temperature-dependent PL spectroscopy, a total number of 36 CF transition lines of Nd 3+ in LiLuF 4 NCs in the NIR region were unequivocally assigned. Furthermore, by employing the sharp and well-resolved CF transitions from the thermally coupled Stark sublevels of 4 F 3/2 of Nd 3+ , we have demonstrated the application of LiLuF 4 :Nd 3+ NCs as sensitive NIR-to-NIR luminescent nanoprobes for ratiometric detection of temperature with a wide linear range of 77-275 K. The unambiguous revelation of photoactive site symmetry and electronic structures of Nd 3+ in inorganic NCs is of vital importance for future design and development of Nd 3+ -based NIR luminescent nanoprobes toward versatile applications such as cryogenic temperature sensing for space and energy exploration.