Lanthanide‐Based Thermometers: At the Cutting‐Edge of Luminescence Thermometry

Present technological demands in disparate areas, such as microfluidics and nanofluidics, microelectronics and nanoelectronics, photonics and biomedicine, among others, have reached to a development such that conventional contact thermal probes are not accomplished anymore to perform accurate measurements with submicrometric spatial resolution. The development of novel noncontact thermal probes is, then, mandatory, contributing to an expansionary epoch of luminescence thermometry. Luminescence thermometry based on trivalent lanthanide ions has become very popular since 2010 due to the unique versatility, stability, and narrow emission band profiles of the ions that cover the entire electromagnetic spectrum with relatively high emission quantum yields. Here, a perspective overview on the field is given from the beginnings in the 1950s until the most recent cutting‐edge examples. The current movement toward usage of the technique as a new tool for thermal imaging, early tumor detection, and as a tool for unveiling the properties of the thermometers themselves or of their local neighborhoods is also summarized.

From the invention of thermoscope by Galileo until our days, many new methods and temperature sensors have been developed. [13] Generally, the temperature reading is achieved from an invasive probe material in direct physical contact with the body for which temperature is unknown. Thermocouples and thermistors dominate the market but are inappropriate below 10 µm, as the required thermal connection with the sample disturbs the measurements at those small scales. [10,[14][15][16][17][18][19][20][21] Furthermore, these conventional contact thermometers depend upon an electrical link hampering their applications where the electromagnetic noise is significant and sparks are dangerous. [4] Hence, such restrictions of contact thermometers for systems of small dimensions have been stimulating the development of new noncontact precise thermometers with spatial resolution at the micrometric and nanometric scales, an exciting research topic under continuous expansion in the last decade. [16][17][18][19]21,22] High-resolution noncontact thermometers operating at those scales have been grouped using distinct criteria, as whether they make use of optical or electrical signals or they are based on near-field or far-field techniques. However, each method possesses several advantages as well as drawbacks and exhibits different spatial, temporal, and temperature resolution (see, for instance, Table 1 of ref. [10] for details).
Organic dyes are the most available and used thermal probes; for an exhaustive review of the subject see the works of Hoogenboom and co-workers. [33,34] However, recently QDs and Ln 3+ -based materials are gaining importance, due to their higher photostability. For instance, QDs were employed in submicrometer thermometry due to its temperature-dependent luminescence features (intensity changes or emission peak shifts). [35,36] One of the most appealing application areas for QDs is nanomedicine, since its bioconjugation can make them target selective. Nevertheless, QDs usage in future use in clinical trials may be difficult, as they generally include highly toxic elements (e.g., Cd). [37,38] The work of Jaque and co-workers gives a complete review of the application of QDs in microthermometry and nanothermometry. [39] Ln 3+ -based materials are stable and narrow band emitters covering the entire electromagnetic spectrum with, in general, high emission quantum yields (>50% in the visible). [40][41][42][43][44][45][46][47] In the last decade, many Ln 3+ -based thermometers have been reported covering a wide temperature range, from cryogenic (T < 100 K) to physiological (298-323 K) values, and including chelate complexes, [14,15,48] metal organic frameworks (MOFs), [49][50][51][52] polymers, [53,54] organic-inorganic hybrids, [48,55] upconverting, [56][57][58][59] downconverting, [60] and downshifting [61][62][63][64] nanoparticles (NPs), and multifunctional heater-thermometer nanoplatforms. [65][66][67] The implementation of these Ln 3+ -based phosphors as ratiometric thermometers in diverse applications was extensively revised in the past decade, [7,9,10,[16][17][18][19]21,50,52,57,[68][69][70][71][72][73][74][75][76][77][78][79][80][81][82][83] including in two books. [84,85] The story of thermographic phosphor thermometry began in 1937 with Neubert [86,87] during the development of the fluorescent lamp (for a review see the works of Allison and Gillies [4] and Khalid and Kontis [6] ). The observed loss of luminescence intensity with increasing temperature suggested the use of phosphor emission as a nonintrusive technique for monitoring the temperature of hot bodies. In short, the bodies were optically projected on an excited luminous screen and the temperature is rendered visible analyzing the image formed by the quenching of the luminescence of the screen produced by the infrared radiation emitted from the hot body. [86,87] Twelve years later, Urbach et al. showed that the temperaturedependence of the luminescent efficiency of numerous phosphors could be used for the measurement of temperatures and temperature distributions, Figure 2. [88] The first applications, in aeronautics and medicine, date back to the 1950s and 1960s. In the former example, a phosphor was painted on the wing surfaces of a wind-tunnel model in order to probe the temperature, Figure 3, [89] while a couple of years later, and during the course of some studies on drug-induced tumor pain, it became clear that most breast cancers were characterized by an increase in temperature. [90] This heat elevation could be detected in the skin over the tumors and Lawson and Alt in 1965 reported the employment of a thermally sensitive ZnCdS-based phosphor to record the human skin temperature, opening the avenue to apply thermography to clinical diagnosis, Figure 4. [90,91] After these pioneering works, the interest on luminescent thermometers remains essentially flat until the 1990s, with less than 10 publications by year, Figure 5. In this period, mention must be done to some intriguing works on Ln 3+based luminescent thermometers [92][93][94] and their applications in thermal imaging of surfaces [95,96] and high-speed integrated circuits, [97] and to fiber tip thermometry systems. These last systems, commonly known as fluoroptic sensors, were initially proposed by Wickersheim and Alves [98] by applying a phosphor at the tip of an optical fiber, for a review see the work of Wickersheim and Sun. [99] In 1978, Luxtron (now LumaSense Technologies) industrialized the idea through the creation of its Fluoroptic technology working through the 5 D 0 decay time of Gd 2 O 2 S:Eu 3+ . [100] From 1995 to date, In 2002, a major breakthrough on the subject aroused with the inspiring work of Wang et al. about using luminescent NPs for thermometry. [104] Ratiometric luminescent thermometers based on ZnS:Mn 2+ , Eu 3+ semiconductor NPs were introduced taking the temperature-dependent ratio of the emission intensities of the two dopants-the so-called fluorescence intensity ratio (FIR). This concept was generalized a few years later to NPs doped exclusively with Ln 3+ ions (e.g., BaTiO 3 :Er 3+ NPs). [105] Adv. Optical Mater. 2019, 7, 1801239 The front view of the device displays the thermometers used for calibration, temperature increases from right to the left. The top image shows the phosphor upon UV excitation, with the brighter region corresponding to the lowest temperature. b) Arrangement for studying air stream impinging upon phosphor screen. c) High-contrast illustrative print of thermal pattern produced by the air stream. d) Isotherms on the screen obtained from the set of prints like that presented in (c). Numbers are temperatures in degrees centigrade. Adapted with permission. [88] Copyright 1949, OSA Publishing.

www.advopticalmat.de
Adv. Optical Mater. 2019, 7, 1801239 Figure 3. a) Schematic diagram of optical arrangement for calibration and use of temperature-sensitive ZnCdS phosphors in aeronautics. [89] b) Photograph of the half-wedge (surface recovered with the phosphor) after 2 s of supersonic flow. The marked bright line is a crack in the phosphor coverage. The brighter parts of the surface are cooler by about 2° than the darker parts and cooler by about 5° than at the beginning of the flow. The bright streaks at the top and bottom of the picture are an indication of the transition between the laminar and turbulent flowing regimes. The marked bright streak at the right center are believed to be due to the beginning of natural turbulence. c) Detail of the construction of the phosphor coated half-wedge. Adapted with permission. [89] Copyright 1953, AIP Publishing. . a) General view of a pioneering thermal imaging setup based on the employment of thermally sensitive phosphors which glow when exposed to ultraviolet illumination, in inverse proportion to the underlying temperature. The thermal image can be directly observed or more critically analyzed and photographed on a simple closed-circuit television monitor. b) Calibrated gray scale of the heat-quenched of a ZnCdS-based phosphor (in the form of an aerosol spray). Dark areas mean increased temperature. c) A black-and-white image is converted to a thermogram by switching from room light to ultraviolet. Top view shows an ordinary image of encircled breast lump, whereas bottom is a view under UV irradiation of heat patterns in the skin coated by the ZnCdSbased phosphor. The dark streaks over the tumor (marked with arrows) depict hot veins typical of carcinoma. d) Breast showing various cold areas (in white, marked with arrows) corresponding to non-malign cysts that can be aspired. The same measurement using a thermistor probe is presented in e) for comparison, being evident the lower spatial discrimination in comparison with the image in (d). Adapted with permission. [91] Copyright 1965, Can. Med. Assoc.
In the last couple of years, we are observing a gradual shift of the emphasis of luminescence thermometry from the synthesis and general characterization of new thermographic phosphors toward the use of the technique for thermal imaging, early tumor detection, and as a tool for unveiling thermometer's features or details of their local surroundings. In the former case, examples include recording of in vivo thermal images, [123][124][125][126] acquisition of subcutaneous thermal videos, [125] and in vivo ischemia recognition in small animals. [126] Early tumor detection becomes possible by transient thermometry using near-infrared (NIR) emitting Ag 2 S nanocrystals. [127] Examples of the later approach are the analysis of heat flux assessment in heater-thermometer nanoplatforms, [128] the estimation of the absorption coefficient and of the thermal diffusivity of tissues, [129] the quantification of the instantaneous ballistic velocity of Brownian nanocrystals suspended in aqueous and organic solvents, [130] the determination of the thermal conductivity of porous silica and titania nanostructures, [131] and the measurement of the thermal resistance of NPs (in air). [132] Among these recent examples, the present manuscript considers essentially those based on Ln 3+ ions, being a follow up of the two books [84,85] and the review papers published since 2015 on Ln 3+ -based luminescent thermometers. [9,10,[16][17][18][19]21,50,52,74,79,80,82,83] The few examples involving other emitting centers are included in the text either in the context of historical reasons, e.g., the ZnCdS-based phosphors, [88][89][90][91] or due to its enormous impact for the development of the area, e.g., the Ag 2 S nanocrystals. [127]

Classifying Thermometers: Primary and Secondary Thermometers
There are many ways of sorting thermometers. Here, we list the most used classification regarding the way how the sensor is in physical contact with the probe and the intensity-to-temperature calibration relation.
Concerning the physical interaction between the thermal probe and the measurand, the thermometers can be classified  : ((luminescence OR fluorescence) AND (thermometry OR nanothermometry)) OR ((luminescent OR fluorescent) AND (thermometer OR nanothermometer)) OR ((upconversion) AND (thermometry OR nanothermometry)) OR (phosphor thermometry) OR (phosphor temperature-measurements) OR (thermographic phosphors), topic, OR (temperature recording with phosphors), title. Panels (c) and (d) display the corresponding normalized data to the keywords (luminescent OR luminescence), topic. The search was performed on 20/10/2018 and includes all the different phosphors reported up to now providing a contactless thermal reading through their light emission properties.
www.advopticalmat.de as contact (evasive, e.g., thermocouples and thermistors) or noncontact (minimally evasive, e.g., infrared pyrometers). In between, semi-invasive thermometers are those in which there is contact between the sample and the probe, but the temperature is remotely interrogated. While the contact thermometers are well-suited for routine punctual measurements of nonmoving systems, noncontact examples are appropriate for temperature measurements or mappings on moving objects or on objects in hazardous locations (e.g., thermal inspection of devices). The suitability of a given thermometer for an application is related to its sensing performance (e.g., operating range, thermal sensitivity, temperature uncertainty, acquisition rate, and spatiotemporal resolution), but also with the material's properties (e.g., physical state, simple and easy processable synthesis, facility to be implemented, and mechanical and thermal stability). Contact thermal characterization methods for small scales applied to active and passive devices and interconnects were reviewed by Christofferson et al. [133] Depending on the calibration relation, the temperature probes can be classified into primary and secondary thermometers. While the temperature is determined based on the knowledge of thermodynamic laws and quantities, the thermometer is termed as primary. On the contrary, if the temperature is calculated via comparison with a reference thermal probe, a calibration process is required and the thermometer is labeled as secondary. [134] Basically, whenever the knowledge of a measurable physical quantity is not enough to calculate the temperature from an equation of state (relating temperature with other measurable physical quantities), the thermometer must be referred to an external temperature reference and it is called secondary.
Up to know, five thermodynamic measurable quantities are used to determine temperature in primary thermometry: the pressure of a gas in a constant volume, the speed of sound in a monatomic gas, the dielectric constant of a gas, the black-body emission, and the power spectral density of Johnson-noise in an electrical resistor. [134,135] Moreover, although primary thermometers are generally complex and mostly employed for metrology purposes, they are currently engaged in the redefinition of the international temperature scale (1990s International Temperature Scale or ITS-90) in terms of the Boltzmann constant (k B ). [136] Examples of primary luminescent thermometers are, however, very scarce. Up to now, the reported examples are based on CdSe(ZnS) QDs, [137] Si NPs functionalized with 1-dodecene, [138] Y 2 O 3 :Eu 3+ microparticles and nanoparticles, [139] and SrF 2 :Yb 3+ /Er 3+ UCNPs. [140] This last work, discussed in detail in Section 4.1.1, is based on the Boltzmann law as the equation of state and is a major breakthrough on the subject. It demonstrates that any luminescent thermometer based on a ratio of intensities arising from two thermally coupled emitting levels can be used to determine the temperature without a previous calibration procedure.
The most known secondary thermometers are electrical probes, such as platinum resistance thermometers, thermocouples, thermistors, capacitance thermometers, and silicon diodes. [57] Although in general secondary thermometers are less complex than primary ones, recurrent calibrations are required, namely when the thermometers are used in a medium different than the one in which they were calibrated. This is a tedious and time-consuming task that is not always possible to be executed for luminescent thermometers, as, for instance, in living cells and in operating electronic devices. Indeed, many of the secondary thermometers reported in the literature assume a valid unique calibration relation, independent of the medium, a procedure potentially inaccurate (see Section 4.1.1).

Thermometer's Performance
The quantitative comparison of the performance of any temperature probe is critical to evaluate its use and for the comparison between distinct techniques. Thermometer's performance can be evaluated based on its relative thermal sensitivity, temperature uncertainty, repeatability, reproducibility, and spatiotemporal resolution. All the performance parameters for luminescent thermometers were recently reviewed [10] and, then, in what follows we only present a short summary highlighting the most important aspects. Table 1 summarizes the relative thermal sensitivity and temperature uncertainty values (including the working temperature and spectral ranges) of illustrative examples of the distinct classes of Ln 3+ -based luminescent thermometers discussed in the review.

Thermal Sensitivity
The thermal sensitivity is the rate of change of the thermometric parameter (generally, designated by Δ), in response to the variation of temperature. The absolute thermal sensitivity (S a ) is expressed as [141] S T a = ∂Δ ∂ (1) depending only on the degree of the thermally induced variations in Δ. However, it is meaningless to quantitatively compare the thermal sensitivity among thermometers of different nature operating by different physical principles (e.g., optical, electrical, or mechanical thermometers) or, using the same physical principle, operating using different materials. To compare the performance of distinct thermometers, irrespectively of their nature or the material employed, the relative thermal sensitivity (S r ) should be adopted This parameter was introduced in 1998 by Collins et al., [141] and has been extensively adopted as a figure of merit for the comparison of the thermometer's performance after the proposal of Brites et al. in 2012. [17] S r is usually expressed in units of % change per degree of temperature change (% K −1 ), being the maximum value of S r denoted by S m (occurring at a temperature designated as T m ). [10]

Δλ
Material  [125] www.advopticalmat.de thermometer, depending not only on the material but also on the experimental setup used. The uncertainty in the temperature arises from several factors, such as the experimental detection setup, the acquisition conditions, and the signal-to-noise ratio. Usually, δT is estimated by the time-dependent output fluctuations of the thermometer calculating the evolution of the temporal fluctuations on the thermometric parameter. Using a calibration curve, the temperature that corresponds to each Δ is obtained (calculated from an equation of state in primary thermometers or empirically obtained in secondary examples), allowing the construction of a histogram of the temperature readouts during a certain time interval. The experimental δT of the thermometer is the standard deviation of the resulting temperature histogram. When this strategy is not possible to be implemented, an estimate of the temperature uncertainty is given by [10] T S where δΔ/Δ is the relative uncertainty in the determination of the thermometric parameter (depending on the acquisition setup and estimated from the errors in Δ). This value is controlled adjusting the signal-to-noise ratio of the emission spectrum used to calculate each Δ value, e.g., larger integration times and/or averaging consecutive measurements. Clearly, there is a compromise between dropping δT and increasing the acquisition time. In fact, the longer is the acquisition time, the lower is the temperature uncertainty, resulting the minimum δT value from the temperatures histogram's standard deviation in the limit t → ∞.
One interesting strategy to quantify the minimum temperature uncertainty of any thermometer was reported by Alicki and Leitner applying the spin-boson model and using size and systemdependent properties. [142] For solid-state nanothermometers, the relative fluctuation in temperature is related to the number of atoms in the sample (N A ) and its Debye temperature (T D ) For T D in the range 100-2000 K, the term in parenthesis changes between 0.9 and 1.3, meaning that the order of magnitude of the temperature uncertainty is determined by [142] δ ≈ which means that the minimum achievable δT is fundamentally controlled by the size of the thermal probe. In quantum metrology, it is known that for nonentangled particles the precision δθ of a general quantity θ scales with the inverse of the number of particles (N P ) [143] δθ ≈ 1 a relation called shot-noise scaling (for entangled states, however, Heisenberg-scaling applies and δθ is inversely proportional to N P ). As N P is proportional to N A , Equation (6) supports the result of the model derived by Alicki and Leitner [142] (Equation (5)).
There are very few examples reporting the thermal resolution of luminescent thermometers as a function of its size. One case is Alaulamie's work [144] that examined experimentally the correlation between particle size and the temperature uncertainty based on the temperature readouts of Er 3+ -doped UCNPs clusters of different sizes (ranging from 1 to 9 µm). Briefly, the larger the cluster size the higher the signal-to-noise ratio, leading to smaller temperature uncertainties (low standard deviation value). The experimental data present an unequivocal increase of the temperature uncertainty as the cluster size decreases, in accord with Equations (5) and (6). For the examples discussed below in Section 5.2.1 for which S r increases as the size of the crystals decreases, Equations (3) and (5) are only compatible if the decrease in 1/S r is less than the increment of δΔ/Δ, resulting in an overall increase of δT with the decrease of N A .

Resolution, Reproducibility, and Repeatability
The spatial (δx) and temporal (δt) resolutions of a measurement are defined as the minimum distance or time interval between measurements, respectively, presenting a temperature change larger than δT.
The thermometer's reproducibility is the change of the same measurement carried out under modified circumstances (e.g., different equipment in use, different measurement methods, different observers, etc.).
The repeatability, R, describes the thermometer ability to provide repeatedly the same result, under the same circumstances and is computed by where Δ c and Δ i represent, respectively, the thermometric parameter's mean value and the thermometric parameter measured at each temperature.

Sensing Temperature with Luminescence
Luminescence is affected by the temperature, among other external stimuli, and the induced changes can be monitored measuring distinct parameters of the emitting center, such as i) the integrated emission intensity of a single transition or of a pair of transitions, ii) the spectral shift, bandshape, or bandwidth of a given transition, and iii) lifetime measurements, using the time-decay intensity profiles of emitting excited states ( Figure 6). As the emission properties are characteristic of the emitting center itself, no fundamental limitations preclude the development of thermal probes with nanometric size. Luminescence thermometry exploits those emission temperatureinduced changes either following the spectral changes of a given emission spectra (time-integrated scheme) or the temporal changes of a given transition (time-resolved scheme), as detailed in the next two sections.

www.advopticalmat.de
It must be pointed out that when the thermometric parameter Δ is calculated based on integrated intensities, spectral shifts, or bandwidths, the emission spectra must be represented as a function of energy and not wavelength, to circumvent misleading and incorrect conclusions. For this (and besides the normal correction for the instrumental response of the equipment), the photon flux per constant wavelength interval function, φ(λ), must be converted to photon flux per energy interval, φ(E), according to [145,146] where h is the Planck constant, c is the speed of the light, and the minus sign-that can be ignored-simply indicates the distinct directions of integration in λ and E. This correction (known as Jacobian transformation) was seldom performed in the field of luminescence thermometry (in fact, more generally in luminescence spectroscopy) but can be critical, especially when the spectrum displays multiple peaks over a wide energy range and with different bandwidths. [146] Despite the Jacobian transformation will have little effect on 4f transitions, especially when those enrolled in the calculus of Δ are closely spaced, its usage is strongly encouraged for an exact description of the materials' electronic structure and a proper evaluation of the thermometer's performance.

Time-Integrated Schemes
Generally, the emission intensity of a given transition is sensitive to temperature changes due to the following mechanisms: • Population redistribution over electronic levels according to the Boltzmann statistics; • Temperature-activated quenching mechanisms (e.g., crossrelaxation between electronic levels); • Nonradiative deactivations (the electrons relax from excited states to the ground state generating heat, instead of light); • Phonon-assisted Auger conversion processes.
Luminescent thermometers based on the intensity of a single transition are highly dependent of eventual illumination oscillations, signal-to-noise detection, absorption and scatter cross-sections, and local fluctuations on the phosphor concentration. As recursive calibration procedures are not compatible with end-user applications, a ratio of intensities must be employed. [10,18]

Intensity Ratio (FIR) or Bandshape
The bandshape-based nanothermometry exploits the relative change in the intensity ratio of two independent energy-close transitions. Both emission lines can be generated from a single luminescent center (single-center thermometers) or they can result from two distinct emitting centers (dual-center thermometers). [16,17] The use of single-center thermometers was introduced by Kusama et al. [92] in a seminal paper (Collins et al. [147] and Wade et al. [148] give a review of the technique).
In single-center ratiometric thermometers, Δ (or FIR) is defined using the emission intensities of the |2〉 → |0〉 (I 02 I 2 ) and |1〉 → |0〉 (I 01 I 1 ) transitions, where |0〉 denotes the ground level and |1〉 and |2〉 the two thermally coupled excited levels (level |2〉 is more energetic than level |1〉) [148][149][150] where N 1 and N 2 are the populations of the |1〉 and |2〉 levels, ν 01 and ν 02 are the frequencies of the |1〉 → |0〉 and |2〉 → |0〉 transitions, and A 01 and A 02 are the total spontaneous emission rates from levels |1〉 and |2〉 to level |0〉. Note that Equation (9) assumes that the intensities I 2 and I 1 are corrected by the instrument response and, as mentioned above, should also consider the Jacobian transformation. Furthermore, if the depopulation of the |1〉 and |2〉 energy levels involves other energy levels beyond |0〉, Equation (9) must be corrected by the β 2 /β 1 ratio, where β i (i = 1,2) are the branching ratios of the |i〉 level (i.e., the percentage of the total emission from the thermalized level (|1〉 or |2〉) to the final |0〉 state). If the two high-energetic levels are in thermal equilibrium (they are called by "thermally coupled levels," with energetic separations of the order of the thermal energy k B T), N 1 and N 2 are related by where g 1 and g 2 are the degeneracies of the two levels and δE is the energy gap between the barycenters of the |1〉 → |0〉 and |2〉 → |0〉 emission bands. Equation (9) is, thus, written as Adv. Optical Mater. 2019, 7, 1801239 . The two emitting levels cannot be too separated in energy, otherwise its thermalization is not detected. Typically, they are considered "thermally coupled" (e.g., in a thermodynamically quasi equilibrium state) for δE ranging from 200 to 2000 cm −1 . [151] Up to know, in most of the examples involving the temperature determination in single-center thermometers based on the FIR of "thermally coupled" levels, the δE and B parameters result from a fit (e.g., are not determined independent of the temperature), and, therefore, an external thermal calibration of the thermometric parameter is required. The typical calibration process needs an independent temperature measurement, using a thermocouple, a pyrometer, or an infrared thermal camera, for instance, to permit the intensity-to-temperature conversion. As follows, a new calibration procedure is mandatory whenever the thermometer works in an alternate medium or a different environment, with altered ionic strength, pH, pressure, ions local neigborhoods, or atmosphere composition, that disturb the thermometric parameter and, thus, the calibration. Moreover, recording several calibration curves in various medium is a tedious task that is not always possible, as mentioned previously, and, generally, a single calibration relation is assumed to be valid, independent of the medium. This ad hoc assumption is a fundamental bottleneck of secondary luminescent thermometers.
However, recently Balabhadra et al. [140] realized that δE and B can be measured independently of any experimental calibration procedure, demonstrating that single-center thermometers based on two thermally coupled electronic levels are intrinsically luminescent primary thermometers. This was a significant step forward because it means that any thermometer based on Equation (11) is, intrinsically, a primary thermometer, and, therefore, the equation can be used to predict the temperature calibration curve independently of the medium. [10,140,152] The δE value is determined using the formal definition of barycenter of a J-J′ transition (J stands for the total angular momentum) or, when experimental difficulties in assigning precisely the Stark-Stark transitions exist, by an envelope fitting to the I 1 and I 2 transitions, [61] see Figure 7a for the example of the 2 H 11/2 → 4 I 15/2 and 4 S 3/2 → 4 I 15/2 Er 3+ transitions. The parameter B is empirically inferred from the plot of Δ versus laser excitation power, Figure 7b. [10,140] When the laser-induced heating is negligible, (i.e., in the limit of null excitation power), the temperature, T 0 , corresponds the room-temperature [153,154] and the thermometric parameter Δ 0 is given by The absolute temperature value is determined by the ratio Δ/Δ 0 (calculated as the ratio between Equations (11) and (12)) resulting This approach was validated for the seminal example of the 2 H 11/2 → 4 I 15/2 (I H I 2 ) and 4 S 3/2 → 4 I 15/2 (I S I 1 ) Er 3+ transitions ( 2 H 11/2 and 4 S 3/2 are thermally coupled energy levels [147,149] ), using as an illustrative example SrF 2 :Yb 3+ / Er 3+ UCNPs (Yb 3+ /Er 3+ is by far the widely reported pair in Ln 3+ -luminescent thermometry). [140] The temperature calculated through Equation (13) was compared with the readout of conventional thermometers positioned in contact with the particles (for powders) or immersed in aqueous suspensions (Figure 7c). A remarkable agreement between the predicted and measured temperature values is observed, irrespectively of the NPs' size and of the dispersion media, demonstrating that for this example no other variables apart temperature impact the thermometric parameter value. We should note that this approach is general and is extensible to any other thermometer based on two "thermally coupled" levels, albeit up to now was solely applied to the 2 H 11/2 and 4 S 3/2 emitting levels.
The parameter B can be also calculated using the Judd-Ofelt theory. [155,156] For that, let us consider the integrated coefficient of spontaneous emission of a transition between J and J′ manifolds as [42,157,158] Adv. Optical Mater. 2019, 7, 1801239 Figure 7. a) Emission spectrum (points) of SrF2:Yb 3+ /Er 3+ NPs (41 nm) recorded at 300 K using a laser power density of 42 ± 5 W cm −2 . The spectrum was fitted to a set of two and five Gaussian peaks in the 2 H 11/2 → 4 I 15/2 and 4 S 3/2 → 4 I 15/2 spectral regions, respectively (not shown). The magenta line assigns the envelope of the sum of the two transitions. b) Evolution of the thermometric parameter with the laser power density for SrF2:Yb 3+ /Er 3+ NPs of distinct sizes. The value of Δ at no-laser excitation (Δ 0 ) is determined from the intercept. c) Calculated temperature using Equation (13) (y) versus temperature reading using a thermocouple (x) of SrF2:Yb 3+ /Er 3+ NPs (41 nm) in powder and water suspension. The dashed line is a guide for the eyes corresponding to y = x. The horizontal error bars represent the thermocouple accuracy and the vertical ones the error in the calculated temperature. Adapted with permission. [140] Copyright 2017, ACS Publishing.  (14) where e is the electronic charge, and n is the refractive index of the medium. The electric (S ed ) and magnetic (S md ) dipole strengths are given (in units of e 2 ) by where the quantities Ω λ (λ = 2, 4, 6) are the so-called Judd-Ofelt intensity parameters [155,156] and m is the electron mass.

Bandwidth
Generally, the emission lines of phosphors broaden as temperature increases. This is ascribed to the intrinsic vibrations of the lattice (homogeneous broadening, highly temperature dependent) or to the presence of different optical centers and defects (inhomogeneous broadening, slightly temperature dependent). Henderson and Imbusch described the temperature dependence of the bandwidth (W) of emission and absorption bands as [164] where W 0 is the full width at half maximum (FWHM) of the band at 0 K, and hΩ is the phonon energy (lattice vibration) that interacts with the electronic transitions.
There are few reports using intra-4f line's emission bandwidth to measure the temperature, despite all of them present the functional form predicted by Equation (18). Strangely, none of these reports used Equation (18) to extract the energy of the phonon responsible for the broadening. Peng et al., for instance, used the 5 D 0 → 7 F 2 transition in the Y 2 O 3 :Eu 3+ phosphor to determine the temperature in the 10-670 K range. [165] The transition bandwidth remains essentially unaltered for T < 70 K, broadening linearly for higher temperatures at a rate of 0.078 cm −1 K −1 , corresponding to S m = 0.78% K −1 (at 70 K). [165] In another example, Wang et al. analyzed the bandwidth of several Tm 3+ emission lines in NaYbF 4 :Tm 3+ @SiO 2 core@shell microparticles (Figure 8). [166] Among the transitions studied, the 1 D 2 → 3 F 4 (350 nm) and 3 H 4 → 3 H 6 (798 nm) ones broaden linearly with increasing temperature (100-700 K). [166] Using a YAlO 3 :Nd 3+ nanoperovskite, Hernández-Rodríguez et al. [167] compared the thermometric performance of the material using the FIR method ( 2 H 9/2 , 4 F 5/2 → 4 I 9/2 / 4 F 3/2 → 4 I 9/2 ) and the FWHM of a Stark component of the 4 F 3/2 → 4 I 13/2 transition. The corresponding S m values are 1.8 and 3.3% K −1 , for the FIR method and the FWHM change, respectively (at 293 and 370 K, respectively). The corresponding temperature uncertainty values are 0.9 and 0.4 K. Although this example points out the benefit of using the spectral bandwidth approach for temperature determination, more studies are required to infer if this conclusion can be generalized for other systems.
Tm 3+ -doped crystalline TiO 2 films [168] and the [Eu(keto) 3 (H 2 O)] [169] (keto = ketoprofen) coordination compound are interesting and atypical examples of Ln 3+ -based luminescent thermometers reporting simultaneously temperature-induced bandwidth increase and wavelength shift. In the former example, the 3 F 3,0-1 → 3 H 6,0-1 Tm 3+ transition (at 676 nm) exhibits a linear wavelength blue-shift dλ/dT of −2.2 pm K −1 (+0.048 cm −1 K −1 ) with a 1.25 nm (55 cm −1 ) bandwidth increase, over the ≈85-750 K range. This linear behavior is in contrast with that exhibited by the convential Al 2 O 3 :Cr 3+ optically based thermometers in which the peak-shift linear behavior only covers a limited region (≈300-600 K), being the bandwidth and temperature precision highly affected at increasing temperatures. [168] For [Eu(keto) 3 (H 2 O)], whereas the FWHM of the 5 D 0 → 7 F 0 Eu 3+ transition shows small variations (≤3.0 cm −1 , the experimental resolution), its energy displays a linear blueshift as temperature increases from 25 to 300 K, the maximum splitting is 16.6 ± 3.0 cm −1 . [169] In both examples, the electronphonon coupling seems to play a relevant role in the mechanism beyond the temperature dependence of the Ln 3+ transitions.

Time-Resolved Scheme: Lifetime
Unlike the luminescence intensity methods, the lifetime-based technique holds crucial advantage of virtually not being affected by the size, geometry, and the concentration of the luminescent probe. Moreover, the value of lifetime shown to be independent on the effects of light scattering, reflection, and intensity fluctuations of the excitation source. However, lifetime determination needs a pulsed excitation source with long illumination and acquisition time which in turn leads to time-consuming measurements limiting the use of this technique. Nevertheless, www.advopticalmat.de and although the thermal readout of large temperature gradients at time intervals shorter than or equal to the lifetime of the luminescence is impracticable using this technique, the recent technological advances made simpler and less expensive the employment of the technique.

Comparing the Time-Integrated and Time-Resolved Schemes
Albeit interesting, the comparison between the performance of the time-integrated and time-resolved temperature methods has not been done systematically. The pioneer work of Collins et al. [147] in Cr 3+ -, Er 3+ -, and Pr 3+ -based crystals was one of the first examples doing it. The FIR response provides higher S r values (e.g., 5 times larger for the Pr 3+ -based crystal).
As lifetime values are essentially temperature-independent in cryogenic temperatures range this is quite evident. The same conclusion was also inferred by Rai and Rai in Pr 3+ -doped lithium tellurite glass, [170] Paviolo et al. [29] in fluorescent molecular thermometers based on rhodamine B (RhB) and by Gálico et al. [171] in the [Eu(bzac) 3 (H 2 O) 2 ] complex (where bzac − stands for tris(1-phenyl-1,3-butanedione). Paviolo et al. realized that the temperature in an organic tissue's cytoplasm measured via RhB emission intensity is more precise and reliable than that measured using the RhB lifetime, [29] while Gálico et al. reported S m = 5.25% K −1 (at 303 K) and S m = 1.35% K −1 (at 293 K) [171] through the temperature dependence of the integrated intensity of the 5 D 0 → 7 F 2 transition and the 5 D 0 lifetime, respectively. Recently, Gharouel et al. [172] performed a systematic comparison of the performance of distinct Pr 3+ -based thermometers operating over 298-363 K using either the thermal dependence of the ratio of two 4f transitions or of the decay time of an excited state. Similar S r values ranging from 0.25 to 0.60% K −1 were obtained and, again, although the lifetime approach requires more complex detection systems, the thermal sensitivity outcome is, at best, the same that the one obtained by the FIR method. [172] Therefore, in general, the wisest approach should involve a ratiometric intensity response to temperature variations.

Molecular Complexes and Organic-Inorganic Hybrids
Ln 3+ -doped molecular systems (essentially with Eu 3+ and Tb 3+ ions) have been widely explored in luminescent thermometry, since the pioneering work of Sato et al. in 1989. [173] For a review on molecular Ln 3+ -based thermometers, see the work of Uchiyama et al. [14] More recent noteworthy and illustrative examples are the works of Suzuki et al., [174] Hatanaka et al., [175] and Khudoleeva et al. [176] The first example showed real-time thermogenesis in a single HeLa cell using the Eu(tta) 3 complex (tta − stands for 3-thenoyltrifluoroacetonate) dissolved in dimethyl sulfoxide. Temperature variations as small as 1 K in the physiological range were detected. [174] The second example used [Tb 0.99 Eu 0.01 (hfa) 3 (linker)] n polymers (where hfa − stands for hexafluoro acetylacetonate and four phosphine oxides were used as linkers)-so-called chameleon emitters, because the emission color gradually changes with the temperature-to demonstrate how the thermal sensitivity is controlled by the linker, as well as by the hfa ligand. [175] In the last example, surface modified tph@Tb 0.999 Eu 0.001 F3 fluoride (tph − = terephthalate) demonstrated nontoxicity and cellular permeability, exhibiting in vitro Eu 3+ and Tb 3+ luminescence with S m = 0.35% K −1 at 313 K.
The chemical and optical instabilities of the isolated molecular systems with the increasing of te temperature, [177] however, preclude effective thermometry applications and composite materials formed by polymers or organic-inorganic hybrid hosts incorporating the lanthanoid complexes rapidly became an attractive alternative. An illustrative and pioneering Adv. Optical Mater. 2019, 7, 1801239   Figure 8. a) Temperature-dependent Tm 3+ emissions in NaYbF 4 :Tm 3+ @SiO 2 core@shell microparticles. b) Temperature-dependent bandwidth of distinct Tm 3+ transitions in NaYbF 4 :Tm 3+ / SiO 2 . Adapted with permission. [166] Copyright 2013, OSA Publishing.

www.advopticalmat.de
example is the thermal imaging of a metal stripe covered with a thin film of poly(methyl methacrylate) (PMMA) doped with the Eu(tta) 3 complex. [95] Despite the low temperature uncertainty (δT = 0.08 K) and the quite interesting spatial resolution (δx = 0.7 µm), the system is not ratiometric.
Using organic-inorganic hybrid materials as hosts, Brites et al. [15] proposed groundbreaking thermometers based on the Tb 3+ ( 5 D 4 → 7 F 5 ) and Eu 3+ ( 5 D 0 → 7 F 2 ) emissions, at 545 and 612 nm, respectively. The key point on this approach is the rational design of the host permitting the occurrence of thermally driven 5 D 4 → host energy transfer, keeping the population of the 5 D 0 emitting level unaffected. Then, the ratio between the intensity of the 5 D 4 → 7 F 5 and 5 D 0 → 7 F 2 transitions, I( 5 D 4 → 7 F 5 )/I( 5 D 0 → 7 F 2 ), gives the absolute temperature, with a spatial resolution determined by the size of the clusters to which the Ln 3+ complexes are attached. [15] We should note that this mechanism is dissimilar than that frequently reported in Tb 3+ /Eu 3+ -doped examples in which the temperature dependence of the I( 5 D 4 → 7 F 5 )/I( 5 D 0 → 7 F 2 ) ratio is controlled by Tb 3+ -to-Eu 3+ energy transfer mechanisms. [178] The use of Tb 3+ -to-Eu 3+ energy transfer as a tool to temperature determination was first proposed by Sato et al. in 1989 [173] and later revisited by Liu et al. in 2005. [179] Nevertheless, and undoubtedly inspired by the works of Sato et al. [173] and Brites et al., [15] the I( 5 D 4 → 7 F 5 )/I( 5 D 0 → 7 F 2 ) ratio is currently, by far, the widely used thermometric parameter allowing in the 298-333 K range typical values of S m = 7.1 ± 0.2% K −1 , δT = 0.09 ± 0.01 K, R > 99.2%, [131] δx > 0.5 µm, and δt > 0.005 s ( Table 1). [118] For a review on luminescent thermometers based on organicinorganic hybrids, see ref. [77].
Recent examples on Tb 3+ /Eu 3+ -doped systems not discussed in this review and deserving remark are the works of Piñol et al., [128] Brites et al., [131] Rodrigues et al., [180] and Ramalho et al. (Table 1). [181] The first two examples are discussed in more detail in Section 6.2. Rodrigues et al. [180] reported the thermometric functionalization of a Si surface with Tb 3+ and Eu 3+ complexes leading to S m = 1.45% K −1 , δT = 0.3 K (at 323 K), and cycle-recycle reliability of 98.6%. The hysteresis of the luminescence of the material results in a dual-sensitive temperature regime. The observed reversible bistability permits the Si-functionalized surface to operate as an optically active two-module molecular demultiplexer logic circuit, opening the possibility of using this computing molecule in medical and biotechnologies, such as blood diagnostics, "labon-a-molecule" systems, and molecular computational identification of small objects. On the other hand, Ramalho et al. [181] used a PMMA substrate coated with Tb 3+ /Eu 3+ -doped organicinorganic hybrids to fabricate luminescent quick response (QR) codes. [181] QR codes have gained increased attention, as they offer a simple physical tool for quick access to web sites for advertising and social interaction. As QR codes are a widespread technology, the adding of functionalities simultaneously with the increase of the storage capacity is a relevant technological issue. Luminescent QR codes based on transparent plastic substrates coated with Tb 3+ /Eu 3+ -doped organic-inorganic hybrids were reported, demonstrating the increase of storage capacity per unit area by a factor of two. A novel methodology to decode the multiplexed QR codes is developed based on a color separation threshold where a decision level is calculated through a maximum-likelihood criterion to minimize the error probability of the demultiplexed modules, maximizing the foreseen total storage capacity. The luminescent QR codes ability to sense temperature was demonstrated, as the Tb 3+ /Eu 3+based hybrid emission color coordinates vary with temperature (reproducibility higher than 93%), opening new fields of applications for QR codes as smart labels for sensing. [181] MOFs are a class of porous hybrid materials consisting of metal ions or clusters coordinated to organic ligand linkers. [182] As the materials' building blocks (metal ions, linkers, and guest ions or molecules) are all potential sources of light emission, MOFs are appropriate platforms to engineering luminescence. [183][184][185] Because the light-emitting centers of certain Ln 3+ -bearing MOFs change considerably with temperature, these materials have been explored in the last years as luminescent thermometers, especially based on the intensity ratio of two intra-4f transitions. The first reports on the subject by Cui et al. [183,184] and Cadiau et al. [51] (first Ln 3+ -bearing nanoMOF working as luminescent thermometer) used the intensity ratio of the Tb 3+ ( 5 D 4 → 7 F 5 ) and Eu 3+ ( 5 D 0 → 7 F 2 ) emissions, following up closely the ideas reported by us a couple of years before. [15] Nonetheless the MOF characteristic limited thermal stability limits effective thermometry applications much above roomtemperature, they are promising and can compete with other thermometer materials in the cryogenic (<100 K) and biological (298-323 K) temperature ranges. In fact, we should note that among Ln 3+ -based materials, Ln 3+ -containing MOFs present the highest relative thermal sensitivity values reported so far, both in the cryogenic, S m = 31% K −1 (at 4 K), [186] and in the physiological, S m = 16% K −1 (at 300 K), [187] ranges (Table 1). Cui et al. [50] and Rocha et al. [52] reviewed recently the main concepts and ideas assisting the design of MOF-based thermometers.
Recent examples on Tb 3+ /Eu 3+ -doped systems not discussed in these reviews and deserving remark are the works of Liu et al. [188] and Li et al. [189] The first paper reported an in situ reduction and crystallization route for preparing Eu 2+ / Eu 3+ -doped MOFs. The materials exhibit intrinsic-and sensitized-emissions of Eu 2+ and Eu 3+ ions, besides a long-lived luminescence from ligand-to-metal charge transfer. A ratiometric luminescent thermometer was demonstrated based on the linear relation between temperature and the intensity ratio of Eu 3+ (at 618 nm) and Eu 2+ (at 427 nm) emissions between 9 and 293 K. [188] The second example described a MOF-based thermometer based on Eu 3+ -to-ligand back energy transfer and operative over a wide temperature range, including the physiological (12-320 K), upon excitation with visible light (450 nm). [189]

Upconversion (UC) and UCNPs
UC emission is a nonlinear process developed by Bloembergen, [190] Auzel, [191] and Ovsyankin and Feofilov. [192] The distinct mechanisms of UC emission were extensively reviewed in the last years [193] and there are a great number of Ln 3+ -based UCNPs that were proposed for luminescence nanothermometry. The UC emission can be distinguished as single-center and multicenter, depending on whether the UC signal is generated by a single type of Ln 3+ ion or by a combination of different Ln 3+ ions. Interestedly, and as far as we know, there is www.advopticalmat.de no example reported in the literature of downconverting luminescent thermometers.

Single-Center UC Nanothermometers
The most common UC systems are based on Yb 3+ as a sensitizer and Er 3+ , Ho 3+ , or Tm 3+ as activators. Yb 3+ acts as an effective sensitizer owing a large absorption cross-section at 980 nm, a wavelength easily available due to its use in telecommunications technology. Furthermore, the Yb 3+ excited state energy level matches well with the excited states of the Er 3+ , Tm 3+ , and Ho 3+ , thus allowing an efficient resonant energy transfer.
As an activator, Er 3+ is one of the widely used ions due to its strongly temperature-dependent intense green emission arising from the 2 H 11/2 → 4 I 15/2 (515-525 nm) and 4 S 3/2 → 4 I 15/2 (535-545 nm). [147,149] Based on that emission, Rodrigues et al. measured one of the highest S r values reported so far for upconversion thermal sensing using β-NaGd 0.94 Pr 0.02 Er 0.02 Yb 0.02 F 4 @ 3NaY 0.8 Yb 0.2 F 4 core@shell NPs. [194] Under 980 nm irradiation, the visible Er 3+ emission shows a relative thermal sensitivity in the 83-323 K range, reaching S m = 9.52% K −1 (at 83 K, Table 1). The intensity of those Er 3+ transitions together with that of the Eu 3+ 5 D 0 → 7 F 2 (603-643 nm) line was used by Nigoghossian et al. [195] to develop an intriguing example of a dual-mode nanothermometer. NaGdF 4 :Yb 3+ /Er 3+ UCNPs were synthesized and coated with a silica shell to which a Eu 3+ complex with tta − was incorporated. Whereas the Er 3+ UC emission was excited at 980 nm, the downshifting (DS) Eu 3+ signal (not ratiometric) was excited at 352 nm through the tta ligands. Measurements were recorded near the physiological temperature range (293-323 K), revealing S m = 2.48% K −1 /δT = 0.47 K (980 nm) and S m = 2.67% K −1 /δT = 0.06 K (352 nm), at 303 K. Moreover, because the Eu 3+ luminescence decreases with the increasing of the UV-light exposure time, the Eu(tta) 3 -based complex anchored in the silica shell was tested as a UV sensor. One of the limitations of the great majority of the nanothermometers fabricated up to now (including Er 3+ -doped UCNPs) is a limited operating temperature range (typically not above 400 K), which prevents use in high-temperature applications, such as, for example, thermal barrier coatings and chemical reactors. Illustrative examples widening the temperature range of luminescent thermometers up to ≈900 K are the works of Geitenbeek et al. in silica-coated NaYF 4 :Yb 3+ /Er 3+ NPs, [196] Brites et al. in Sr 2 GeO 4 :Pr 3+ crystalline powders, [197] and Kolesnikov et al. in YVO 4 :Nd 3+ NPs. [198] Tm 3+ and Ho 3+ are the other activator ions most used for temperature sensing. [199][200][201] An illustrative example is Y 2 O 3 :Yb 3+ / Ho 3+ submicrometric porous powders (using 978 nm excitation) reported by Lojpur et al. [200] presenting a maximum absolute sensitivity of 0.097 K −1 , at 85 K corresponding to S m = 1.61% K −1 . Using the ratio between the intensity of the 5 F 4 , 5 S 2 → 5 I 8 transition and that of distinct Stark components of the 5 F 4 , 5 S 2 → 5 I 7 level in Ho 3+ yields to a maximum absolute sensitivity of 0.078 K −1 , that corresponds to S m = 0.55% K −1 (at 275 K).
When the intensity ratio is taken from two transitions arising from thermally coupled levels, the thermometric parameter is given by Equation (11) and accordingly to the definition of S r , Equation (2), results If B is temperature independent, S r only depends on δE, decreasing monotonically with temperature increasing. On the contrary, if B is function of temperature, S r also depends on the temperature dependence of the branching ratios of the two thermally coupled levels, Equation (17) (both the Ω 2,4,6 parameters and the frequencies of the two transitions involved in the thermometric parameter are temperature independent). This dependence on the branching ratios of the two thermally coupled levels might justify eventual changes on the S r values with system's parameters, such as size, morphology, Ln 3+ doping, or matrix's phonon energy.
The impact of nanoparticle's size on S r values has been discussed in a few works. [105,202,203] The first reference dates back to 2004 reporting a size dependence on S r values for BaTiO 3 :Er 3+ nanocrystals in which the thermal parameter is the intensity ratio between the 4 S 3/2 → 4 I 15/2 and 2 H 11/2 → 4 I 15/2 Er 3+ transitions. [105] As the size of the crystals decreases (from 60 to 26 nm) S r increases, while the energy gap between the barycenters of the two transitions remains unchanged. Marciniak et al. [203,204] also discussed a similar tendency in NaYF 4 :Yb 3+ /Er 3+ and LiLaP 4 O 12 :Yb 3+ /Er 3+ nanophosphors. At 200 K, S m raises from 1.1 to 2.1% K −1 , as the size of fluoride nanocrystals decreases from 200 to 8 nm, and from 1.1 to 1.8% K −1 , as the size of tetraphosphate nanocrystals decreases from 240 to 20 nm. An opposite trend, however, was reported by Dong et al. [202] on NaYF 4 :Yb 3+ /Er 3+ microspheres (size from 0.7 to 2 µm) showing that S a decreases as size increases, S m = 1.4% K −1 (700 nm) and S m = 1.0% K −1 (1600 nm) (at 223 K), as calculated by us. Intriguingly, both trends were attributed to the larger specific surface area of the smaller crystals. As the size of the crystal decreases, a relatively large number of optically active ions are located at the surface and its contribution becomes increasingly important, being influenced by nonraditive relaxation channels related to electronphonon interactions that are stronger with increasing temperature. [105,202,203] Moreover, the nonradiative rates of the two emitting levels might present a distinct temperature dependence and when the size of the nanoparticle decreases the material's phonons density changes inducing a dependence of those rates with the crystal size. However, Balabahadra et al. [140] and Brandão-Silva et al. [205] do not observe significant changes on S r values with the decrease of the particle size in SrF 2 :Yb 3+ / Er 3+ (size from 10 to 41 nm) and Y 2 O 3 :Er 3+ (21 to 86 nm) NPs, respectively. Clearly, the influence of the nanoparticle's size on the thermal sensitivity of luminescent thermometers seems to be dependent of the host and of the electron-phonon interactions requiring a deep and systematic analysis.
In summary, single-centered UC thermometers are essentially being reported based on the analysis of the emission intensity of two thermally coupled energy levels. Although these systems are intrinsically primary thermometers (see Section 2), the tight restriction in the energy gap δE to ensure the strong coupling between www.advopticalmat.de the two levels, typically δE < 2000 cm −1 , [83] implies S r values typically smaller than those characteristic of the downshifting examples (Section 5.3 and Table 1). Larger energetic separation between the thermally coupled levels decreases the thermalization of the upper energetic level, resulting in poorer luminescence intensity. To overcome this limitation, strategies for further S r enhancement other than thermally coupled levels for designing FIR-based thermometers should be considered, besides playing with the size of the NPs or with the phonon energy of the hosts (as discussed above). A wisest approach consists in using two distinct (and thermally decoupled) emission lines of the same Ln 3+ emitting center, or to use two emitting levels of distinct Ln 3+ emitting centers, as discussed in the next section. The recent review by Cheng et al. [83] discusses strategies to improve S r values by using thermometric parameters based on "fully decoupled" or "moderately coupled" emitting levels or emitting levels in which the energy transfer is mediated or thermally assisted by a host or a ligand level.

Multicenter UC Nanothermometers
Multicenter UC thermometry exploits the different thermal dependence of the intensity of two lines generated in distinct Ln 3+ ions. The transitions can be fully thermally decoupled (e.g., both transitions decrease upon temperature raise) or can present an indirect thermal coupling (e.g., energy exchange mediated by the ion host or ligands). Both the possibilities have been reported extensively in the literature, being the energy transfer engineered via smart-chemical design of the thermal probe. The core@shell structure architecture allows a facile incorporation of dopants to guide an efficient energy transfer among different ions. Although multicore@shell structures have been considerably reported for noncontact temperature measurements, [206,207] there are only few works on multicenter UCNPs. [208][209][210][211] An intriguing example was reported by Xu et al. using a Yb/Ho/Ce:NaGdF 4 @Yb/Tm:NaYF 4 core@shell structure, presenting S m = 2.4% K −1 (at 298 K) in the absence of Ce 3+ . The thermometric parameter is defined using the red (Ho 3+ : 5 F 5 → 5 I 8 and Tm 3+ : 1 G 4 → 3 F 4 transitions) and the green (Ho 3+ : 5 S 2 , 5 F 4 → 5 I 8 transition) bands. Introducing Ce 3+ in the core the authors can tune S m between 0.7 and 4.4% K −1 via the efficient cross relaxation processes between Ce 3+ and Ho 3+ (Figure 9). [210] Recently Savchuk et al. [132] reported KLuWO 4 :Tm 3+ /Ho 3+ NPs as tunable multifunctional heater-thermometer nanoplatforms under 808 nm excitation. The intensity ratio is defined using the intensity of the Adv. Optical Mater. 2019, 7, 1801239   Figure 9. a) Schematic representation of the Yb/Ho/Ce:NaGdF 4 @Yb/Tm:NaYF 4 core@shell nanostructure. b) Temperature-dependent UC emission spectra of the nanostructures upon 980 nm excitation, normalized to the emission line at 450 nm. c) Simplified energy level diagrams of Ce 3+ , Ho 3+ , Yb 3+ , and Tm 3+ with the proposed energy transfer mechanisms. Adapted with permission. [210] Copyright 2016, Royal Society of Chemistry.

DS and Downshifting Nanoparticles
DS is a single photon process where upon excitation with a highenergy photon (typically in the UV region of the electromagnetic spectrum) a nonradiative relaxation occurs pursued by radiative relaxation, thereby resulting in the emission of a lowerenergy photon. As an explanatory example, we can mention the work of Ishiwada et al. [212] that reported Tb 3+ /Tm 3+ :Y 2 O 3 NPs as visual thermosensors, working in the 323-1123 K range. The intensity ratio between the Tm 3+ ( 1 G 4 → 3 H 6 ) and the Tb 3+ ( 5 D 4 → 7 F 5 ) lines is strongly temperature-dependent, under 355 nm (UV) excitation with S m = 0.33% K −1 (at 750 K). Lifetime based examples were also discussed, although in minor number, for the reasons addressed in Section 4.3. The temperature dependence of the 4 S 3/2 lifetime in the NaY 2 F 5 O:Yb 3+ /Er 3+ NPs [56] presents S m = 2.75% K −1 (at 330 K), Figure 10. The authors tentatively attributed the shortening of the lifetime with the temperature increasing to nonradiative relaxation and multiphonon phenomena.

Complex Systems
In this section, we discuss illustrative examples of complex thermometric systems comprising multiple core@shell nanostructures, [207,[217][218][219] polymer-and hybrid-based materials, [30,154,[220][221][222][223] and heater-thermometer nanoplatforms. [154,224] Multiple core@shell architectures are attracting much interest in recent years due to the possibility of combining distinct functions on a single nanostructure (see ref. [219] for an updated review). An interesting approach is the use of these structures as double ratiometric nanothermometers. The first Adv. Optical Mater. 2019, 7, 1801239   Figure 10. a) 4 S 3/2 decay curves in NaY 2 F 5 O:Yb 3+ /Er 3+ NPs at 298 and 333 K. b) Temperature dependence of the normalized 4 S 3/2 and 4 F 9/2 lifetime values. Points represent experimental data and solid lines are the best to the experimental points using straight lines. c) 4 S 3/2 decay curves measured in NPs injected 1 mm below the surface of a chicken breast when the heating laser was on (1.2 W) or off (0 W). d) Subcutaneous temperature measured using the lifetime values for distinct heating laser power values. Reproduced with permission. [56] Copyright 2014, Royal Society of Chemistry.

www.advopticalmat.de
As mentioned previously, an important research line in Ln 3+ -based luminescent thermometers explores the potential of polymers and organic-inorganic hybrids as host materials. The interesting example by Huang et al. [220] reported a hybrid nanocarrier consisting of Yb 3+ /Er 3+ codoped NaGdF 4 NPs encapsulated in the aqueous core of liposomes. The 980 nm excitation is used simultaneously to induce the Er 3+ green UC emission and for drug delivery, by coloading the doxorubicin (DOX) model anticancer drug with the UCNPs in the liposome nanocarrier. Additionally, a quenching on the green emission was ascribed to the donor UCNPs-to-acceptor DOX energy transfer, allowing the spectral monitoring of the DOX loading and release from the liposome nanocarriers. [220] Although the authors do not report temperature determination the emission spectra of the nanoplatform can be also used with that purpose.
The combination of Ln 3+ -based phosphors with other emitting centers was also exploited. Cerón et al. [54] developed a complex system joining in a hybrid nanostructure Nd 3+doped NaGdF 4 NPs, semiconductor PbS/CdS/ZnS QDs, and poly(lactic-co-glycolic acid). The thermometric parameter is based on the strongly temperature-dependent emission at 1220 nm (arising from the QDs) and a temperature-independent reference peak at 1060 nm (the Nd 3+ 4 F 3/2 → 4 I 11/2 line). The coexistence of these two emissions allows to obtain S m = 2.5% K −1 (at 303 K, Table 1).
Heater-thermometer nanoplatforms are other interesting systems because they combine the heat release and temperature monitoring in a single nanostructure. One of the most exploited strategies consists in using plasmonic nanostructures and luminescent thermometers to achieve a fully noncontact Adv. Optical Mater. 2019, 7, 1801239   Figure 12. a) Schematic representation of UC and DS transitions in NaLuF 4 :Gd 3+ /Yb 3+ /Er 3+ @NaLuF 4 :Yb 3+ @NaLuF 4 :Nd 3+ /Yb 3+ @NaLuF 4 core@shell nanostructures, depicted schematically in the inset. b) Emission spectra of the core@shell nanostructures under 808 nm excitation in the 223-423 K range. c) Overlap between a photograph and a NIR image under 808 nm laser excitation of a mouse subcutaneously injected with aqueous dispersion of the PEG modified core@shell nanostructures. Adapted with permission. [225] Copyright 2015, Royal Society of Chemistry.
www.advopticalmat.de control of the system. One of the first examples was reported by Debasu et al. [154] that developed a single nanoplatform integrating laser-induced heat generation by Au NPs and temperature sensing up to 2000 K via Gd 2 O 3 :Yb 3+ /Er 3+ nanorods. The temperature determination is done based on the green Er 3+ upconversion emission and Boltzmann's law, for temperatures until 1200 K, and using the blackbody emission and the Planck's law of blackbody radiation for temperatures until 2000 K.
Other illustrative example combines Au nanorods and NaGdF 4 :Yb 3+ /Er 3+ UCNPs. [224] The longitudinal surface plasmon resonance of the nanorods is tuned to 980 nm, in resonance with the Yb 3+ absorption wavelength, so luminescence thermometry and heating can be simultaneously excited. As an added benefit the authors observe a luminescence enhancement until ninefolds due to the proximity of the Er 3+ ions with the plasmonic structures. The temperature is controlled by the 980 excitation power and measured in situ by the Er 3+ thermally sensitive luminescence. [224] In a step forward, the same group reported a nanocomposite consisting of NaGdF 4 :Yb 3+ / Er 3+ nanoparticle, mesoporous SiO 2 , Au nanorods, and a photosensitizer. Under 980 nm irradiation, plasmonic and UC NPs are simultaneously excited leading to a plasmonic enhancement of the UC luminescence and a simultaneous temperature increasing. Upon laser irradiation, and after loading a Zn phthalocyanine photosensitizer into the mesoporous SiO 2 , the UC visible light subsequently activates the photosensitizer to release reactive oxygen species. [226]

Current Trends in Lanthanide-Based Luminescent Thermometers
As highlighted in the introduction, in the last couple of years the focal point of luminescence thermometry has progressively moved from the synthesis and general characterization of new thermographic phosphors toward the use of the technique as a tool in bioimaging, early tumor detection, and for unveil properties of the thermometers themselves or of their local surroundings. The present section covers meaningful examples demonstrating this shift. Moreover, the section also summarizes recent developments to boost the thermal sensitivity of luminescent thermometers. [227]

Luminescence Thermometry Using Ln 3+ Ions as a Biomedical Tool
Eu 3+ -based thermometers were used to visualize the temperature-distribution within living organisms at microscopic www.advopticalmat.de scales. A polymer incorporating Eu-tris(dinaphthoylmethane)bis-trioctylphosphine oxide as a temperature sensitive probe and rhodamine 800 (Rh800) to provide a reference emission band enabled the temperature monitoring and mapping with S m = 3.6% K −1 (at 298 K) and 1.4 < δT < 1.0 K. The polymer embedding the thermal probes successfully detected in vivo temperature shifts at localized sites in the muscle of a living beetle, either due to an external heat source (980 nm laser heating) or due to the animal's voluntary muscle-activation (preflight preparation, Figure 14a-c). The Eu 3+ /Rh800 intensity ratio was recorded in 5 distinct positions of the muscle (Figure 14d,e) presenting a dynamics similar to that monitored by an IR thermal camera. [124] Moreover, the same research group developed free-standing nanosheets embedding the same emitting centers with superior flexibility and transparency, compared with the previous report, enabling the attachment onto the uneven surfaced living tissues without any glue (Figure 15a,b) and the densification of muscle fibers covered by the thermometric film. The film was used to measure temperature shifts in a beetle's dorsal longitudinal muscle (Figure 15c,d) during the voluntary heat production, with S m = 3.75% K −1 at 302 K and δT up to 0.75 K. Endogenous heat production transfer between muscle fibers was studied using the thermal mapping recorded using the Eu 3+ -Rh800 stacked nanosheets with a spatial resolution at least at the single muscle fiber level (impossible to be achieved through IR thermal cameras), leading to further understand physiological activities in three stimulations: before, during, and after the beetle's preflight preparation (Figure 15e). [123] The absorption coefficient and thermal diffusivity of tissues were calculated by Ximendes et al. [129] using Yb 3+ /Nd 3+based thermal probes. The authors follow the time-dependent temperature relaxation process to calculate the thermal diffusivity of living tissues by luminescence thermometry (0.13 ± 0.04 mm 2 s −1 ) in good agreement with that reported for breast tissue at 810 nm (0.142 mm 2 s −1 ). [129] Zhou et al. demonstrated a programmed combination of chemotherapy (CT) and photothermal therapy (PTT) combining in a single system, NaLuF 4 :Yb 3+ /Er@NaLuF 4 @yolkshell-SiO 2 -PdPc@DPPC-DOX UCNPs, a photothermal agent (octabutoxyphthalocyanine palladium (II), abbreviated as PdPc), and a thermal responsive drug release unit (1,2-dipalmitoyl-snglycero-3-phosphocholine, abbreviated as DPPC), as the vehicle of the DOX chemodrug (Figure 16). [228] By controlling the nanostructure's temperature (using a 730 nm laser), the CT-PTT sequence can be designed to achieve programmed combination cancer therapy. When the dosages of DOX and heat are kept at low level (2.5 × 10 −6 m and 0.150 W cm −2 , respectively), programmed combination therapy can achieve 39-fold improvement in therapeutic effect in vitro in comparison with conventional combination therapy that simultaneously initiates CT and PTT. [228] Adv. Optical Mater. 2019, 7, 1801239   Figure 14. a) Photograph of a Dicronorrhina derbyana beetle with a detail of the dorsal longitudinal muscle (a major flight muscle of the beetle) in which the ratiometric thermosensor was loaded (dashed triangle). Inset shows a photo of the muscle and a schematic representation of the nanoparticle. b) Pseudocolor images of the intensity of the Eu 3+ and Rh800 channels before and after the deposition of the thermometer. c) Intensity ratio of the Eu 3+ and Rh800 channels that is function of the temperature. d) Temporal thermal profile performed by an IR thermometer of the region presented in the inset of (d). e) Corresponding temporal profiles of the normalized intensity ratio in the distinct regions marked in the inset of (d). Adapted with permission. [123,124] Copyright 2016, American Chemical Society.

www.advopticalmat.de
IR-emitting core@shell Er-Yb@Yb-Tm LaF 3 UCNPs were reported for subcutaneous thermal probing operating in the second and third transparency biological windows of tissues with thermal sensitivity of 5% K −1 . The acquisition of timeresolved 2D thermal images in a small animal model was reported taking advantage of the high thermal sensitivity and large optical penetration into tissues, scanning the subcutaneous temperature images in a minimally invasive way. Moreover in vivo subcutaneous thermal videos opened the possibility to develop new diagnosis and control techniques with potential impact in existing biomedicine methods (Figure 17). [125] Without using Ln 3+ ions, Santos et al. [127] use the temporal decay temperature profiles monitored by Ag 2 S IR nanothermometers to early-tumor diagnosis. Changes in tissues are discerned through changes in their thermal dynamics measured by thermal transient thermometry. Experimental data obtained in a murine model of melanoma reveal evident differences between the thermal relaxation dynamics of tumoral and healthy tissues even at the early stages of the tumor development (e.g., up to 7 days before changes could be detected by optical inspection, Figure 18). The same group used NIR emitting PbS/CdS/ZnS QDs as deep tissue nanothermometers to discriminate between the ischemic and inflammatory phases (that occur after artery ligation), using exclusively thermal transient thermometry. The authors distinguish between the ischemic and the healthy tissues by distinct temporal decays of the NIR luminescence. Furthermore, the QDs were used for time monitoring the revascularization of tissues after temporal restriction of blood supply. [126]   , and after (right) beetle's preflight preparation using the stacked nanosheets. e) The corresponding temporal evolution of the normalized Eu 3+ /Rh800 intensity ratio in the three stimulations, followed in the ROI areas presented in (c). Adapted with permission. [123,124] Copyright 2016, American Chemical Society.

www.advopticalmat.de
and Tb 3+ complexes and further polymeric layers to render aqueous stability. [128] The thermometer presents a high relative thermal sensitivity (5.8% K −1 ) and low temperature uncertainty (0.5 K), whereas the NP demonstrated a relatively small heat capacitance. Taking advantage of the physical contact between the thermometer and the nanoheaters, the existence of an unexpected temperature gradient between the nanoheaters and the surrounding media for t > 10 s and relatively low heat powers (10 −16 W per heater) generated by an alternated magnetic field was grasped. Moreover, previously unrevealed changes in the local temperature during the first few seconds of heating (t < 10 s) are observed opening intriguing possibilities to study heat flow at sub-micrometric scales. [128] In a recent work, Baral et al. [229] measured the local temperature of lithographically fabricated Au nanostructures or drop-casted Au nanoparticles (40 nm) deposited on the top of AlGaN:Er 3+ films combining near-field scanning optical microscopy (SNOM) with Er 3+ luminescence thermometry (in downshifting). The thermal images present spatial resolution limited only by the SNOM cantilever tip aperture (200 nm) allowing to measure the effective local temperature of the nanostructures, with temperature uncertainty near 1 K and temporal resolution of 130 ms (still out of the sub-micrometer, sub-millisecond region, not covered so far by any example). The steady-state thermal profiles of distinct sized clusters made from the Au NPs (excited with a 532 nm CW laser) show that the maximum Adv. Optical Mater. 2019, 7, 1801239   Figure 16. a) Temperature responsive NaLuF 4 :Yb 3+ /Er@NaLuF 4 @yolk-shell-SiO 2 -PdPc@DPPC-DOX nanosystem. The temperature is monitored by the UC emission (980 nm excitation). The power density of the photothermal excitation source (730 nm) can be tuned accurately to initiate a thermalcontrolled drug release and PTT stepwise. b) Temperature increase of an aqueous suspension of the NPs (0.5 g L −1 ) at various power densities. c) DOX release rate with time for an aqueous suspension of the NPs (50 W cm −2 ). Control experiments at distinct temperatures are also presented. d) Emission of the UCNPs and propidium iodide (denoted by UCL and PI, respectively) and e) temperature imaging during PTT with 730 nm laser irradiation (0.140 W cm −2 ). The temperature imaging was performed using the Er 3+ UC emission in the green spectral region. The cells were dead as their nucleuses could be stained with propidium iodide. Adapted with permission. [228] Copyright 2018, Springer Nature. www.advopticalmat.de temperature change and temperature decay length into the surrounding medium increases linearly with cluster radius.
Using Yb 3+ /Er 3+ -doped NaYF 4 nanocrystals of distinct sizes suspended in aqueous and organic solvents, Brites et al. [130] measured the instantaneous Brownian velocity of the suspended nanocrystals. A heat flux in the nanofluid medium was externally induced through the Joule effect and real-time temperature monitoring by upconversion thermometry allowed to calculate the time instant associated to the onset of the change in the thermometric parameter induced by temperature variation. This critical time was measured irradiating the nanofluids (980 nm) at different positions along the predesigned path of the heat flux permitting to determine the instantaneous Brownian velocity of the nanocrystals for distinct volume fractions, solvents, and nanoparticle sizes, Figure 19. [130] This technique ought to give a more deep comprehension of the factors governing thermal conductivity, convective heat, and mass transport in nanofluids.
Guided by the same general purpose of evaluating the material's properties through luminescence thermometry, the thermal conductivity of mesoporous SiO 2 and TiO 2 nanofilms was calculated using the luminescent temperature readout of Eu 3+ /Tb 3+ -codoped organic-inorganic hybrid probes. [131] Using first a 980 nm laser beam to promote the plasmonic heating of the nanostructures, the temperature decay after turning the heating beam off was followed by luminescence thermometry to further calculate the thermal conductivity of the films. The authors demonstrated that the reported noncontact and Adv. Optical Mater. 2019, 7, 1801239   Figure 17. a) Scheme of the in vivo 2D subcutaneous dynamic thermal imaging experiment. Thermal images obtained by dividing the luminescence images at 1000 and 1200 nm during b) heating and c) cooling processes. An optical figure of the anesthetized mouse was superimposed. d) Time evolution of the average temperature of the injection area as measured by the luminescent thermometers during heating and thermal relaxation processes. Adapted with permission. [125] Copyright 2017, Wiley-VCH. Figure 18. a) Optical and NIR fluorescence images of a representative mouse. The NIR emission evidences the presence of Ag2S nanocrystals both at the tumor site and in a healthy tissue area. b) Time evolution after tumor induction (day zero) of Δτ (normalized difference between the thermal relaxation time of tumoral tissue with respect to that of the healthy tissue). c) Time evolution of tumoral volume. d) Time evolution after tumor induction of the surface temperature difference between tumor and healthy tissue as obtained with a thermographic camera. In all cases, the dots correspond to the experimental data obtained as an average of the 6 mice analyzed in the work and solid lines are included as guides for the eyes. The background colors and vertical lines indicate the time when the tumor becomes detectable by the different methods. Adapted with permission. [127] Copyright 2018, Wiley-VCH.

www.advopticalmat.de
nondestructive strategy is able to estimate the nanolayers' thermal conductivity in accordance with the values reported for similar nanostructures using evasive electrical methods, Figure 20. [131] The thermal conductivity of an AlGaN:Er 3+ film was also recently evaluated using luminescence thermometry. [230] The authors positioned an Au microdisk over a AlGaN:Er 3+ film deposited on a silicon substrate. Upon continuous 532 nm laser irradiation (power density of 6 × 10 9 W m −2 ), the Au microdisks have a temperature increase of ≈20°. By temporal-modulating of the 532 nm laser intensity, the authors induced a temperature increase followed by the relaxation to the room-temperature, which is monitored by the luminescence decay of the 2 H 11/2 → 4 I 15/2 and the 4 S 3/2 → 4 I 15/2 transitions. The temperature is calculated using the ratio of the intensity decay curves and thus the heat-transfer dynamics was mapped with spatial resolution below 100 nm in the microsecond range. This work is able to follow the cooling of a single Au micro disk and to measure the rate of heat dissipation to the environment, directly calculating the absolute thermal conductivity of the AlGaN:Er 3+ film (1.7 W m −1 K −1 ) in agreement with the literature values. [230] The thermal resistance of KLuWO 4 :Tm 3+ /Ho 3+ NPs was also recently estimated using a fully noncontact approach through upconversion thermometry. [132] The thermal resistance of powder NPs in contact with air was computed from the instantaneous temperature profiles and it was found to be of the same order of magnitude than the interfacial thermal resistance between Si and NiSi 2 in Si-based heterostructured nanowires. [132]

Exploring New Strategies to Enhance The Thermal Sensitivity
It is well known that temperature increase causes, in general, a loss in the light emission intensity that is known as thermal Adv. Optical Mater. 2019, 7, 1801239   Figure 19. a) Schematic of the experimental set-up used to determine the instantaneous Brownian velocity of the NaYF 4 :Yb 3+ /Er 3+ @NaYF 4 nanofluid. The inset shows the solvent-mixing effect arising from the Brownian motion of the nanoparticle located at the interface between the cold (T 1 ) and hot (T 2 ) regions of the nanofluid. b) Emission spectra of the water-based nanofluid (volume fraction of 0.0068%) recorded at 300 and 330 K. The spectra were normalized to the maximum intensity of the 4 S 3/2 → 4 I 15/2 emission at 300 K. The inset presents a scheme of the core@shell NP. c) Reduced temperature profiles of the NPs dispersed in water as measured by laser excitation from different positions x i along the x direction (depicted in (a)). The dashed line refers to the critical onset time t 0i when the onset of change in the intensity ratio is observed due to temperature variation upon turning on the heater. d) The corresponding linear correlation (r 2 > 0.994) between x i and t 0i , as measured in water for the nanofluids with different nanoparticle volume fractions. e) Measured Brownian velocities of the NPs in water as deduced from (c). The horizontal line marks the velocity in the limit of low dilution. Adapted with permission. [130] Copyright 2016, Springer Nature. www.advopticalmat.de quenching, that limits the application of luminescent materials at higher temperatures. The thermal quenching is attributed to nonradiative relaxation pathways that are thermally activated because the material's atomic vibrations (phonons) are heightened with the temperature increase.
For NPs, however, there is a quenching channel, the socalled energy migration-induced surface quenching, that does not exist in their bulk counterparts. Recently, Cui et al. [231] demonstrated that the energy migration-induced surface quenching in Yb 3+ -doped NPs can be suppressed by increasing temperature, resulting in an uncommon luminescence thermal enhancement. A mechanism based on the effect of thermal lattice expansion on Yb 3+ -mediated energy migration is proposed to be beyond that unusual effect. Furthermore, since 2015, it was understood that the thermal effects on the light emission of organic-capped UCNPs is a size-dependent phenomenon. Systematic studies on the emission intensity of the particles upon distinct temperatures were performed by Li et al. [232,233] and Shao et al. [234,235] The authors showed that at higher temperatures (e.g., 300-500 K) whereas the emission intensity is enhanced for particles smaller than a critical size (e.g., 20-30 nm), it is quenched for bigger particles. This effect was used to control of NPs' emission color via modulation of the irradiation laser power or directly varying the temperature (through complex heating systems). [232,233,235] Although several models were advanced to explain these intriguing thermal effects, a recent explanation purposed by Zhou et al. [227] is being discussed in the literature. The authors observed an ≈2000-fold enhancement in blue emission for 9.7 nm Yb 3+ /Tm 3+ -codoped organic-capped UCNPs at 453 K, justifying their observations by a phonon-assisted energy transfer from sensitizers to activators, populating up the intermediate excited state of the UC process. The authors ascribed the phonon to existence of an oxygen moiety chelating Yb 3+ ions, [Yb⋅⋅⋅O], present in the NPs due to the chelation Yb 3+ ions at the nanoparticle's surface by carboxylic groups (of the capping molecules). A physical mechanism to fully explain this surface mechanism remains, however, absent. [236] Adv. Optical Mater. 2019, 7, 1801239   Figure 20. a) Scaled model of the mesoporous nanostructures with the Au NPs. b) Linear dependence of the temperature increment with the 980 nm laser power density. Comparison between the thermal conductivity values calculated using luminescence thermometry (filled symbols) and the thermal conductivity values reported in the literature computed by conventional electrical methods. Reproduced with permission. [131] Copyright 2017, American Chemical Society.

www.advopticalmat.de
This new design principle opens the possibility to design novel size-dependent thermal probes that, using two emitting centers, can present higher performance in comparison to the existing ones. We notice that single-center thermometers based on Equation (13) do not benefit from the thermal enhancement effect in S r . The minimum δT value, however, is generally reduced because the relative uncertainty δΔ/Δ will be lower for small particles at higher temperatures. Moreover, the primary nature of these thermometers can be used to determine the absolute temperature of the nanoparticle, despite the presence of thermal enhancement or quenching effects. We can foresee that multicenter UC nanothermometers will have their thermal sensitivity increased from the current 1-3% K −1 to values one order of magnitude higher just by taking the ratio between thermally enhanced and thermally quenched emissions.
Martinez et al. [152] described an electrothermal device built by the assembly of nanocomposite transparent films combining a PMMA matrix and a percolating network of Ag nanowires [237,238] covered by UCNPs of distinct sizes and compositions. The devices are excellent platforms to thermally fine-tune the particles' emission intensity and the NPs' local temperature and, thus, the emission color is easily controlled through the heat dissipation in the electrothermal device (by simply applying different DC voltages). This strategy provides a guide to fabricate innovative and versatile multichromatic upconversion nanostructures whose spectral emission is externally adjusted by electrothermal control. From thermometric point of view, the devices incorporating Er 3+ -doped NPs are inherently primary thermometers and the particle's local temperature can be accessed directly, using the intensity ratio of the green emission bands ( 2 H 11/2 → 4 I 15/2 / 4 S 3/2 → 4 I 15/2 ), Equation (13). Devices composed of small and large Ln 3+doped NPs containing distinct dopants allow to estimate the temperature using the primary thermometer based on the Er 3+ I H and I S bands. The calculated temperature values were used to calibrate the secondary thermometer defined using a mixture of thermally enhanced and thermally quenched emission lines resulting in a relative thermal sensitivity of about 6% K −1 . [152]

Summary and Perspectives
The field of luminescence thermometry is growing intensively showing significant breakthroughs in sensing, imaging, diagnostics, and therapy, among other areas. This interest has been mostly encouraged because many of the present technological demands in areas such as microelectronics and nanoelectronics, photonics, nanomedicine, and microfluidics and nanofluidics have reached a point such that the use of traditional contact thermal probes (liquid-filled and bimetallic thermometers, thermocouples, pyrometers, and thermistors) is not capable to make reliable measurements when spatial resolution enters to the sub-micrometer range. This limitation of conventional thermometers for small systems has spurred the development of luminescent microthermometers and nanothermometers, a research topic leaving its inflationary epoch accounting nowadays for more than 2.5% of the total publications in luminescence materials. Moreover, and from an industrial point of view, this predicted miniaturization is expected to bring to the market new nanoscale thermal probes. Despite the recently developed luminescent nanothermometers are radically more sophisticated encompassing complex synthesis procedures, the fundamental problems and the applications that are being addressed are analogous to those reported in the field's infancy: the understanding of heat transfer and energy transfer mechanisms, the optimization of temperature readout, and the developing of efficient and cost-effective sensors for front-edge medical and engineering tools.
Among the distinct emitting centers used in luminescence thermometry, including proteins nucleic acids and other biomolecules, thermosresponsive polymers, organic dyes, and QDs, many Ln 3+ -based luminescent thermometers have been reported, essentially in the last decade. Systems comprise molecular complexes, MOFs, polymers, and organic-inorganic hybrids, multifunctional heater-thermometer nanoplatforms, and UC, DC, and DS NPs. This review describes the use of these Ln 3+ -based phosphors as luminescent ratiometric thermometers in diverse applications, with focus on what the authors believe that will be the emergent new research areas in this fascinating research field: the use of luminescence thermometry for thermal imaging, early tumor detection, and as a tool for unveiling properties of the thermometers themselves or of their local surroundings.
Finally, to become a consolidated subject and not a temporary fashion, the research on luminescent thermometry must settle down as a strong node of fertile interactions among disparate communities, such as chemists, physicists, engineers, theoreticians, biologists, and physicians. The cross-fertilization of ideas and experiences at these interfaces will certainly induce important and exciting breakthroughs in future years.