Accurate In Vivo Nanothermometry through NIR-II Lanthanide Luminescence Lifetime. Accurate In Vivo Nanothermometry through NIR-II Lanthanide Luminescence Lifetime

: Luminescence nanothermometry is promising for non-invasive probing of temperature in biological microenvironment at nanometric spatial resolution. Yet, wavelength- and temperature-dependent absorption and scattering of tissues distort measured luminescence spectral profile, rendering conventional luminescence nanothermometers (ratiometric, intensity, band shape, or spectral shift) problematic for in vivo temperature determination. Here, we describe a class of lanthanide-based nanothermometers, which are able to provide precise and reliable temperature readouts at varied tissue depths through NIR-II luminescence lifetime. To achieve this, we have devised an inert core /active shell/inert shell structure of tiny nanoparticles (size, 13.5 nm), in which thermosensitive lanthanide ion pairs (ytterbium and neodymium) are spatially confined in the thin middle shell layer (sodium yttrium fluoride, 1 nm), ensuring being homogenously close to the surrounding environment while meanwhile protected by the outmost calcium fluoride shell (CaF 2 , ~ 2.5 nm) that shields out bioactive milieu interferences. This ternary structure endows the nanothermometers with the ability to consistently resolve temperature changes at depths of up to 4 mm in biological tissues, having a high relative temperature sensitivity of 1.4-1.1% °C -1 in the physiological temperature range of 10 - 64 o C. These lifetime-based thermosensitive nanoprobes allow for in vivo diagnosis of murine inflammation, mapping out the precise temperature distribution profile of nanoprobes-interrogated area. A class of Nd@CaF 2 lifetime nanothermometers were developed with a high relative temperature sensitivity of 1.4% °C -1 in the physiological temperature range of 10 - 64 o C. These nanothermometers are able to map out the precise temperature distribution profile of nanoprobes-interrogated area for in vivo diagnosis of murine inflammation.


Introduction
Temperature is a critical factor that governs or affects manifold behaviors of physical, chemical, and biological systems, accurate measurement of which is, therefore, of fundamental importance to unravel the complexity of these systems and to impact a multitude of technological applications ranging from integrated photonic devices to precise medicine [1][2][3] .
Thermocouples and thermistors dominate the market but are inappropriate to probe temperature in live biological systems as physical contact with measured samples is a prerequisite, which disturbs the measurements at sub-millimeter scales [4] . Alternatively, luminescence nanothermometry is emerging as a non-invasive spectroscopic method that allow to probe temperature variation at nanometric spatial resolution and in remote distance, spurring wide interests [5,6] . The contactless and high-resolution nature makes them ideal candidates for temperature evaluation in the early diagnosis of several diseases as well as for providing real-time temperature feedback in thermal (hypothermia or hyperthermia) therapies of malignant cancers [7][8][9] .
Tissue-penetrating luminescence enables these nanothermometers for temperature sensing in deep abdomen [24] , video-rate recording of subcutaneous thermal alternations [16] [25] , and identifying incipient diseases (ischemia or tumors) in vivo [25,26] . However, these demonstrations assume that the spectral profile collected at the detector is identical to the one generated in situ by the nanothermometers located in deep tissues.
The optical properties of tissue in-between the nanoparticle probes and the photon detector are often overlooked. Though tissues have low attenuation coefficient in the biological windows, the wavelength-dependent photon absorption and scattering processes still distort the recorded luminescence spectral profile, resulting in inaccurate nanothermometric reading (Figure 1a). In vivo experiments demonstrate that these tissueinduced spectral distortions can, indeed, lead to erroneous temperature readings of few degrees [27] . For instance, luminescence spectral profile from nanoparticles codoped with neodymium (Nd 3+ ) and ytterbium (Yb 3+ ) at a fixed temperature (room temperature), after passing through varied thicknesses of biological tissues (chicken breast), were observed to be distorted (Figure 1b, top). As a consequence, the observed LIR between the emission at1050 nm and the emission at 1000 nm, previously established for in vivo ratiometric luminescence nanothermometry, is observed to be proportional to the tissue thickness (Figure 1b, bottom; Figure S1), creating artifacts for temperature measurement in vivo [28] . Moreover, tissue optical properties vary from tissue type-to-type [11] , and are dependent on the felt temperature ( Figure   1b), adding complexities to in vivo luminescence nanothermometry. Spectral profile deformation demands complicated calibration procedures in order to attain an accurate temperature in vivo [27] .
Alongside spectral profile, lifetime is a temporal parameter that characterizes the decay process of nanoprobe luminescence, and is independent of nanoprobe concentration and laser irradiance [29][30][31] . Importantly, luminescence lifetime of nanoprobes remains unaltered after passing through varying thicknesses of biological tissues (Figure 1c), providing possibilities to implement accurate luminescence nanothermometry in vivo in the biological windows [32] .
Luminescence lifetimes of certain types of carbon dots, quantum dots and organic dyes are temperature-sensitive, but often falling in nanoseconds scale on the same order as that of tissue autofluorescence [21,30] . Note that in coexistence with high photostability and sharp emission bands, the protection of intra 4f-4f transitions by the outer complete 5s and 5p orbitals make lanthanide luminescence possess have ultralong lifetime on millisecond or submillisecond scale, about six orders of magnitude longer than that of tissue autofluorescence [33] .
Herein, we describe an approach to enable in vivo accurate thermographic mapping through thermosensitive lanthanide luminescence lifetime in the NIR-II window. We designed a ternary structure of inert core/active shell/inert shell NaYF4@NaYF4: Yb 3+ , Nd 3+ @CaF2 nanoparticles (tiny size, 13.5 nm), in which thermosensitive ion pairs of neodymium (Nd 3+ ) and ytterbium (Yb 3+ ) ions are incorporated into the thin middle layer (thickness, 1 nm). This architecture allows energy harvesting and utilization within the confined nanoscopic domain, minimizing deactivation processes from both the crystal defects in the core and surface quenching centers. Indeed, the inert core/active shell/inert shell NaYF4@NaYF4: Yb 3+ , Nd 3+ @CaF2 nanoparticles present higher luminescence and longer lifetime than the regular NaYF4: Yb 3+ , Nd 3+ @CaF2 core/ shell structure due to reduced dactivation processes ( Figure   S2) [34,35] . Importantly, this ternary structure also places the protected thermoactive domain (the middle layer) close to the probed milieu, thus maximizing the thermosensitivity of the designated nanoprobe.

Results and Discussions
Synthesizing ternary domain thermosensitive nanoprobes. A fluorite crystal structure of sodium yttrium fluoride (NaYF4) was adopted to build the inert core/active shell/inert shell NaYF4@NaYF4: Yb 3+ , Nd 3+ @CaF2 nanostructure, in which NaYF4 is known to be one of the most efficient host materials for lanthanide luminescence, having low lattice phonon energy (~ 350 cm -1 ) that minimizes multiphonon nonradiative processes of excited states [36] . Calcium fluoride (CaF2) was selected as the outmost inert shell material to shield the optically active middle layer from surface-related quenching processes, as it has high optical transparency range (0.13-10 µm) and low lattice mismatch with NaYF4. Moreover, CaF2 shell can enhance the nanoprobe biocompatibility, as calcium and fluoride ions are common endogenous components and constituents of calcified tissues like bone and teeth [37] . Nd 3+ ions are utilized to harvest the excitation light at 800 nm in the NIR-I window, while Yb 3+ ions act as the lanthanide activators to emit luminescence at ~ 1000 nm in the NIR-II window ( Figure 2a).
Optical excitation at 800 nm is preferred in in vivo experiments as it leads to a minimum thermal loading. [38] Nonradiative energy transfer processes between Nd 3+ and Yb 3+ ions, bridging the absorption and the emission process, are responsive to temperature and able to impact observed luminescence lifetimes [16] .
A set of ternary domain core/shell/shell nanoparticles designated as NaYF4@NaYF4: Yb 3+ , Nd 3+ @CaF2 with defined amounts of Yb 3+ and Nd 3+ dopants were synthesized using a seed-mediated layer-by-layer approach through thermal decomposition of metal trifluoroacetates at high temperature. Nanoparticles of NaYF4 were first used as seeds for the epitaxial growth of the interlayer NaYF4: Yb 3+ , Nd 3+ , after which the inert CaF2 shell was further grown as the outmost protecting layer. Representative transmission electron microscopic (TEM) images of the core NaYF4, the core/shell NaYF4@NaYF4: Yb 3+ , Nd 3+ , and the core/shell/shell NaYF4@NaYF4: Yb 3+ , Nd 3+ @CaF2 nanoparticles are shown in Figure   2b, indicative of the near-uniform size and shape for all the resultant samples. Z-contrast high-angle annular dark field scanning TEM (HAADF-STEM) image (Figure 2b, bottom right), which is sensitive to variations in the atomic number of elements in the sample, distinguishes the designated CaF2 shell (grey) from the NaYF4 host lattice (white), underlying the formation of the desginated core/shell/shell nanoparticles. The size of parent core nanoparticles was measured to be about 6.5 nm, while the size of the core/shell nanoparticles was about 8.5 nm, suggesting the middle layer thickness of ~ 1 nm. Moreover, the size of the resultant core/shell/shell nanoparticles was estimated to be about 13.5 nm, evaluating the outer CaF2 layer thickness to be about 2.5 nm ( Figure S3), in good agreement with the HAADF-STEM result. Measured x-ray diffraction (XRD) patterns confirmed that the core, the core/shell, and the core/shell/shell nanoaprticles are of the designated fluorite crystal structure (cubic phase) ( Figure S4). The pristine ligand of oleic acid on the surface of asprepared NaYF4@NaYF4: Yb 3+ , Nd 3+ @CaF2 nanoparticles was then substituted by a hydrophilic ligand of poly (acrylic acid) (PAA, MW= 18, 000), converting nanoprobes into aqueous medium for thermosensitivity and thermographic imaging studies. Compared Fourier transform infrared (FTIR) spectra between surface modified nanoparticles and as-prepared nanoparticles ( Figure S5) confirm the successful ligand replacement.
While concentration tuning with Nd 3+ ions showed no obvious size change of the core/shell and the core/shell/shell nanoparticles ( Figure S11). When fixing Nd 3+ dopant concentration at 50%, variation of Yb 3+ dopant concentrations (from 2, 5, 10, 20, 30, 40 to 50%) produces pronounced effects on luminescence decay curves of the core/shell and the core/shell/shell nanoparticles at room temperature. Gradual lifetime descending was observed at high Yb 3+ dopant concentrations in the core/shell NaYF4@NaYF4:Yb 3+ , Nd 3+ nanoparticle ( Figure S7), being ascribed to the increased surface quenching and surface-related dopant concentration quenching (associated with energy migration process from inside Yb 3+ ions to surface quenching sites through Yb 3+ sub-lattice). Shelling NaYF4@NaYF4:Yb 3+ , Nd 3+ with inert CaF2 layer protracted the decay process substantially, and revealed the longest lifetime at Yb 3+ dopant concentration of 30%, due to the effective alleviation of both surface-related quenching processes. Moreover, all the core/shell/shell samples showed a remarkable temperature-dependent luminescence lifetime at 980 nm (corresponding to the 2 F5/2 → 2 F7/2 transition of Yb 3+ ions, termed as Yb 3+ luminescence lifetime below) ( Figure S8), while the lifetime at 1060 nm (corresponding to the 4 F3/2 → 4 I11/2 transition of Nd 3+ ions, termed as Nd 3+ luminescence lifetime below) ( Figure S9) is temperature-independent in a temperature range of 10 -64 o C. This indicates a strong temperature dependence of both Yb 3+ → Nd 3+ back energy transfer (BET) process and the Nd 3+ →Yb 3+ energy transfer process that involves phonon assistance ( Figure 2e). However, the rates of both processes are supposed to be faster than or on the same level of the radiative rate of Yb 3+ luminescence (1/τ, ~ 2,000 s -1 ) while lower than the radiative rate of Nd 3+ luminescence (1/τ, ~ 50,000 s -1 ), thus resulting in a temperature-dependent and -independent luminescence lifetimes for Yb 3+ and Nd 3+ , respectively. The Yb 3+ luminescence lifetime for all the samples was observed to decrease with an increase of temperature ( Figure S8). The derivative of Yb 3+ luminescence lifetime with respect to temperature, i.e., the thermosensitivity, can be evaluated using the following equations [1] : Sr=100% (2) Where Sa and Sr represent the absolute and relative temperature sensitivities, respectively; τ and T stand for the luminescence lifetime and the temperature, respectively. For the entire temperature range investigated here, maximum Sr was observed at Yb 3+ concentration of 20%.
And maximum Sa was observed at Yb 3+ concentration of 20% for temperature below 42 o C, while at Yb 3+ 30% for tempeature above 42 o C ( Figure S10).  Figure S12 and Figure 2e). All the samples presented a similar temperature dependence for Yb 3+ luminescence lifetime, which was shortened at higher temperatures ( Figure S13). The Nd 3+ luminescence lifetime was observed to be virtually temperature-independent for all the Nd 3+ contents ( Figure S14). The relative thermal sensitivity (Sr for Yb 3+ luminescence lifetime) was shown to ascend when increasing Nd 3+ concentration from 30 to 60% in the invested temperature range of 10-65 o C ( Figure S15). As a result, taking both thermal sensitivities and biological applications into consideration, the optimal doping levels for Yb 3+ and Nd 3+ in the NaYF4@NaYF4: Yb 3+ , Nd 3+ @CaF2 core/shell/shell nanoparticles were determined to be 20% and 60%, respectively, presenting a sensitive luminescence lifetime to temperature (Figure 2c). An increase of temperature from 10 to 64 °C results in a decrease of lifetime from 898 µs to 450 µs, about 49% decrease. Both Sr and Sa were found to descend with temperature increase in the investigated temperature range (Figure 2d). Importantly, the Sa reaches 13 μs·°C -1 , while Sr reaches as high as 1.4 %°C -1 at 10 °C, about~ 14 fold higher than that of previously reported ratiometric luminescence nanothermometers based on Nd 3+ and Yb 3+ ions [16] . We also measured the temperature-dependent spectra of NaYF4@NaYF4: 20%Yb 3+ , 60%Nd 3+ @CaF2 core/shell/shell nanoparticles, the emission intensity of which gradully decreased with an increase of temperature from 10 to 64 °C ( Figure S16).  S13). Depopulation of the emitting 2 F5/2 sate of Yb 3+ ions can be caused by two possible processes: One is the multiphonon-assisted nonradiative process [39] , while the other one is the BET process from Yb 3+ to Nd 3+ ions. Measured Yb 3+ luminesce lifetime from Yb 3+ -only NaYF4@ NaYF4: 20% Yb 3+ @CaF2 and Nd 3+ luminesce lifetime from Nd 3+ -only NaYF4@

Mechanisms
NaYF4: 60% Nd 3+ @CaF2 nanoparticles remain invariant when elevating temperature, ruling out the first possible process ( Figure S17). Observation of Nd 3+ luminescence when performing direct excitation of Yb 3+ in NaYF4@NaYF4:Yb 3+ , Nd 3+ @CaF2 nanoparticles confirmed the occurrence of the BET process ( Figure S18). Importantly, when coexisting with Nd 3+ ions, Yb 3+ luminesce lifetime was observed to be thermosensitive no matter exciting Nd 3+ dopants at 800 nm (Figure 2c) or Yb 3+ dopants at 980 nm (Figure 2f), demonstrating the essential role of Nd 3+ -enacted BET mechanism to regulate the thermal depopulation process of the emitting 2 F5/2 state of Yb 3+ ions. Indeed, an elevated temperature can result in a thermal population of higher 2 F5/2 Stark levels that follow a Boltzmann distribution, [40] minimizing the energy mismatch involved in BET. This minimization favors energy extraction from Yb 3+ to Nd 3+ and thus result in temperature-dependent depopulation of the 2 F5/2 sate of Yb 3+ ion [41] . 9 The temperature dependence of Yb 3+ luminescence lifetime is governed by the temperature dependence of both ET and BET rate, the radiative and nonradiative transition (multiphonon process) probabilities, as well as the energy migration (EM) process (among Yb 3+ ions). As commented before, the luminescence lifetime of Yb 3+ ions in single doped systems is temperature-independent. This means that the radiative and nonradadiative transition probabilities from the excited state to the ground state of Yb 3+ are temperatureindependent. Energy migration probability is also expected to be temperature independent as it depends on, mainly, the Yb 3+ -Yb 3+ distance. Thus, these processes contribute with a temperature-independent background to the luminescence lifetime of Yb 3+ Figure S15b).

Assessment of ternary domain nanoprobes for thermal sensing.
We chose the optimized NaYF4@ NaYF4: Yb 3+ 20%, Nd 3+ 60%@CaF2 nanoprobes to assess their potential use for luminescence lifetime thermal sensing. We first calibrated the Yb 3+ luminescence lifetime versus temperature in a range of 10-70 o C (Figure 3a and 3b). Attained experimental data can be well fitted by the following equation (R 2 =0.998): τ=0.06347T 2 -13.0927T+10109.817 where τ is the Yb 3+ luminescence lifetime (μs) and T is the temperature ( o C). . It is known that after oral or intravenous administration, nanoparticles can be accumulated in vivo in internal organs (such as stomach and liver) due to pharmacokinetic biodistribution, or into malignant tumor sites due to targeting delivery or enhanced permeability and retention (EPR) effect [42] . Microenvironments in the body can also be harsh: The normal pH of the stomach is roughly 1.5 -2.0, while tumor tissue is featured to have extracellular acidic microenvironment (pH 6.5 -6.9) [43] . The independence of NIR-II luminescence lifetime versus the nanoprobe concentration and pH values makes them ideal candidates for in vivo remote thermal sensing in internal organs.
Thermographic imaging through tissue-mimicking phantom. We performed thermographic imaging of the optimized ternary domain NaYF4@ NaYF4: Yb 3+ 20%, Nd 3+ 60%@CaF2 nanoprobes (aqueous dispersion, 10 mg/ml) at a set of prefixed temperature and through tissue-mimicking phantom with defined thicknesses. As illustrated in Figure 4a, nanoprobes in a quartz cuvette were placed in between a temperature-controlled heating platform and the tissue phantom with varied thicknesses. A thermocouple was positioned in colloidal solution for temperature control. An NIR-sensitive InGaAs camera was utilized to attain NIR-II lifetime-encoded images, which was synchronized but triggered at a precisely Thermographic imaging in vivo. We then utilized the thermosensitive ternary domain nanoprobes to image temperature distribution in living mice. An inflammation in mouse is often accompanied by an increase of body temperature [44,45] . In order to demonstrate the ability of our nanoprobes for accurate thermal monitoring of inflammation we injected the subcutaneous temperature. Note that the acquired temporal luminescence lifetime images are also able to present the precise temperature distribution profile in the nanoprobes-interrogated area.

Conclusions
In summary, we describe a class of ternary domain NaYF4@ NaYF4:Yb 3+ , Nd 3+ @CaF2 inert core/active shell/inert shell nanoprobes with a tiny size of ~ 13.5 nm, in which NIR-II luminescence lifetime at 1000 nm (from Yb 3+ ) is sensitive to temperature. This structure confines the thermosensitive lanthanide ion pair to the middle layer, being close to the surrounding environment while shielding deleterious effects from both the core and the surrounding environment. We revealed that both Nd 3+ and Yb 3+ concentrations play an important role in tuning nanoprobe temperature sensitivity, through regulation of the temperature-dependent back energy transfer process from Yb 3+ to Nd 3+ ions and the energy migration process between Yb 3+ and Yb 3+ ions. The luminescence lifetime thermosensitive nanoprobes were shown to be stable against intense laser exposure, independent of nanoprobe concentration, having reliable repeatability, and being able to work in harsh environment with pH ranging from 1 to12. Importantly, the nanoprobes are able to probe precise temperature at varied tissue depths and allow an accurate thermographic mapping of temperature distribution in nanoparticles-probed area in vivo. These superior advantages render the described nanoprobes as ideal nanothermometers for temperature measurement in live mammals.

Synthesis of α-
The Re (TFA)3 shell precursors were first synthesized using the identical procedure as that for min, and further to 300 o C for 30 min before a natural cooling down to room temperature. The resultant core/shell nanoparticles were collected following the identical procedure as that for collecting the α-NaYF4 core nanoparticles, and also dispersed in in 10 ml hexane for further uses.
Synthesis of α-NaYF4@NaYF4: Yb 3+ x%, Nd 3+ 50% (x=2, 5, 10, 20, 30, 40, and 50) @CaF2 core/ shell/ shell nanoparticles: The procedure for the preparation of the core/shell/shell nanoparticles is similar to the one for preparing the α-NaYF4@ NaYF4: Yb 3+ x%, Nd 3+ 50% core/shell nanoparticles. Instead, the α-NaYF4@NaYF4: Yb 3+ , Nd 3+ core/shell nanoparticles were used to subsitute the α-NaYF4 particles for a seed-mediated growth of the shell. In brief, a mixture of α-NaYF4@NaYF4: Yb 3+ , Nd 3+ core/shell nanoparticles (5 mL, hexane dispersion), 7 ml OA, and 7 ml ODE were first heated to 310 o C and maintained at this temperature under argon gas protection. Subsequently, the Ca-TFA-OA (0.5 mmol/ml, 1.6 ml) shell precursors were injected into the solution for two times with an interval of about 25 min. The resultant core/shell/shell nanoparticles were precipitated, washed with ethanol, and finally dispersed in 10 mL hexane. Technology. The inflammatory model was established by injecting 10% yeast on the back of the mouse (10 ml/kg), and the same volume normal saline was injected to another mouse as a control [46] . The temperature of the mice was monitored by the infrared thermal camera (Fotric 280s, Shanghai, China) every one hour, and after 16 hours, the luminescence lifetime-encoded images were collected.

Synthesis of α-NaYF4@
Lifetime imaging and decoding. Optical excitation was performed using a fiber-coupled 800 nm laser diode, which run at 1 Hz and with a 40% duty cycle and delivered an averge power density of 50 mW/cm 2 . The luminescence images were collected by an NIR-II InGaAs camera (C-Red 2), which was synchronized with the excitation laser at triggered at a precisely defined delay time. The camera kept for integration before the arrival of next laser pulse. A series of time delays were set to sample separate sections of the luminescence decay curve of the nanoprobe. For each pixel of the luminescence lifetime image, the lifetime was fitted by the equation, It=I0 , where t is delay time, I0 is the luminescence intensity at t=0, τ is 15 the lifetime of the nanoprobes. All the image processing was carried out in the commcercialized MATLAB software.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author. windows [14][15][16]19] . b. Normalized luminescence spectra of ternary domain NaYF4@  Representative transmission electron microscopy (TEM) images of the core NaYF4(top left), the core/shell NaYF4@ NaYF4:Yb 3+ , Nd 3+ (top right), and the core/shell/shell NaYF4@  c. Thermographic luminescence lifetime-hued imaging of the optimized nanoprobes in the inflamed and normal mouses, revealing the precise local temperature distribution surrounding the nanoprobes. 25

The table of contents entry
A class of NaYF4@ NaYF4: Yb, Nd@CaF2 lifetime nanothermometers were developed with a high relative temperature sensitivity of 1.4% °C -1 in the physiological temperature range of 10 -64 o C. These nanothermometers are able to map out the precise temperature distribution profile of nanoprobes-interrogated area for in vivo diagnosis of murine inflammation.