Trends in luminescence thermometry

Is luminescence thermometry a mature technology? This question is not easy to answer. The technique has long been known, having been introduced as early as the 1930s,^1,2 and many of its modern adaptations were envisaged by some of the earliest researchers in the field.^3,4 Nonetheless, the number of related publications and citations has increased almost exponentially with time,⁵ and the number of published patents follows a similar trend. A recent Google search for patents using the terms “luminescence thermometry” and “phosphor thermometry” retrieved over 2800 results. Furthermore, these results typically are not incremental advancements. On the contrary, each year, we are witnessing new breakthrough advances in luminescence thermometry, in diverse scientific and technological fields. Moreover, in the last few years, two thematic books have appeared^6,7 and the first thematic conference was held in Scotland in 2018 (The Inaugural International Conference on Phosphor Thermometry—ICPT 2018). However, the widespread use of the method among non-experts and professionals has not been seen, even when its potential for the betterment of many measurements is obvious. These reasons explain why luminescence thermometry is not yet a mature discipline. In his fascinating paper on the history of luminescence thermometry, Allison³ describes three eras in the development of the technique. At the present time, in my opinion, the field of luminescence thermometry is entering a fourth era, the era of maturity. To that end, further developments of the technique should be aimed at removing, or at least alleviating, inherent methodological problems and faults, to facilitate its widespread use.

Temperature measurement devices are abound. Again, a simple web search for commercial products using the term “temperature sensor” gives almost 1000 different possibilities, as stated by Talghader et al.⁸ in their review article on thermal history sensors. Why, then, should we invest resources and effort into advancing luminescence thermometry? The answer to this question is simple. Luminescence thermometry succeeds in environments where many other, if not all the existing, thermometry techniques fail. These environments include some that are of immense importance in nanotechnology, biomedicine, and optoelectronics. In biomedicine, for example, there is a strong need for precise temperature measurements at the intra- and inter-cellular levels. In addition, luminescence thermometry can be easily adapted to aid theranostics and nanotheranostics, since luminescence probes are inherently polyvalent materials. Besides thermometry, luminescence probes can be used for luminescence bioimaging, and many of them may also facilitate magnetic resonance imaging and x-ray computed tomography.⁹ In addition, it has been demonstrated that up-converted light emitted by specially tailored nanoparticles can be used to trigger drug-release.^10–12 There are only a few thermometry methods with spatial resolutions of <1 μm, with luminescence thermometry being one of them. Hence, the well-known historic Richard Feynman speech title on the emerging field of nanoscience, “There's plenty of room at the bottom,” is appropriate as a description of the need for nanoscale thermometry methods. Today, we are able to routinely produce luminescent particles and structures of merely a few tens of atoms in size. These can be used not only to measure temperature but also to answer fundamental questions regarding the continuum and thermodynamic equilibrium at the nanoscale.

Being essentially a semi-contact method, luminescence thermometry can be used in environments that other methods cannot access. It can be exploited over a wide temperature range, having potential applications in cryogenic environments and in extremely high temperature settings, and it is usually possible to operate luminescence thermometry probes in harsh environments. Many engineering applications, such as flow temperature measurements or temperature measurements at engine surfaces, have benefited by using the method. Since the redefinition of the kelvin in 2018, primary thermometry has gained an additional significance at high (>1300 K) and low temperatures (around 1 K), because traceability may be directly linked to the redefined kelvin, instead of, as formerly, to a defined scale. Primary thermometry presents significant advantages over defined temperature scales in these temperature regions, whether owing to lower uncertainty or simply because of the ease of access to reliable thermometry.¹³ Thus, the opportunity has arisen for the development and application of primary luminescence thermometry, especially at very high and low temperatures.

The scientific background of luminescence thermometry is well-established as it is mainly derived from existing extensive knowledge from the fields of luminescence and materials sciences. It is known that different applications require different approaches and probes in luminescence thermometry. However, the complex and interdisciplinary nature of many contemporary applications requires further acquisition of background scientific expertise and knowledge, which may be very demanding and far from trivial. In biomedicine, for example, the science behind in vivo measurements should include knowledge of real interactions between luminescence probes and biological constituents. In addition, in the same application field, there are serious concerns about the health and environmental impacts of materials used in luminescence temperature sensing and imaging, which have only been superficially considered so far. The long-term evolution of the properties of these probes also deserves further consideration. The widespread use of the method will only become a reality when these concerns have been properly addressed. Similar reasoning applies to all the various possible application fields, though each may face specific individual challenges.

This perspective follows the recent publication of reviews^5,14–35 and books^6,7 that comprehensively cover the techniques of temperature measurements via luminescence, materials used to prepare luminescence probes, and luminescence thermometry applications. For this reason, here, the aim is to outline recent trends in the development of the field, highlight innovative approaches in temperature read-out and probe materials, and point out novel and prospective applications. Note that the intention is not to provide a complete description of all the recent findings and developments but to discuss some recent results as indicators of the paths taken within the field of luminescence thermometry. Many parts of the text reflect the author's personal opinions and perspectives on the matters under discussion. The reporting of these opinions stems from the desire to help articulate current thinking, both about the goals to which this field should aspire in the forthcoming years—of which the overarching aim should be, in my opinion, to become a mature widespread technology—and regarding the strategy for achieving them.

There are many ways to determine a temperature (temperature readouts) from luminescence, depending on the application. Likewise, many different materials can be used to construct a luminescence thermometry probe. The choice is mainly made according to the required working environment (e.g., bio-medical, nanoscale, or high-temperature engineering), operating temperature range, precision, thermometer complexity, and cost. But, in each case, the core of the luminescence thermometry application is a measurement. The objective of a measurement is to determine the value of the measurand, that is, the value of the particular quantity to be measured. In the case of thermometry, the measurand is the thermodynamic temperature, T. A measurement begins with an appropriate specification of the measurand, the method of measurement, and the measurement procedure.³⁶ Luminescence thermometry does not provide the value of the temperature directly; instead, it provides an indication (Q), which could be, for example, a ratio of emission intensities, luminescence lifetime, or a band shift. To be able to assess the quality of a measurement and to compare different types of measurements, one must determine the parameters which quantify measurement performance (figures of merit). The most important of these are the measurement range, absolute and relative sensitivities, temperature resolution, temporal resolution, spatial resolution, and the repeatability and reproducibility of the measurements. (The definitions of each of these parameters are given in Table I.) The difference between the smallest and the largest temperature that can be reliably measured by a thermometer determines its dynamic range (measurement range). The absolute sensitivity $( S a)$ of a thermometer is defined as a quotient of the change in an indication and the change in the temperature value, and it is expressed in terms of the indication unit per kelvin. The relative sensitivity (S_r) is defined as the ratio of the absolute sensitivity and the indication value. The temperature resolution $(δT)$ is the smallest change in a temperature that causes a perceptible change in the indication. The spatial resolution $(δ x min)$ of a measuring system is defined as the minimum distance between the points of a measurement that can be resolved under the temperature resolution of the system, while the temporal resolution $(δ t min)$ is the minimum period of time between measurements capable of resolving a temperature higher than the temperature resolution.

The repeatability is the measurement precision achieved during replicate measurements over a short period of time, under a repeatability measurement condition that comprises the same measurement procedure, measuring system, operator, operating conditions, and location; this, in fact, is the ability of the thermometer to provide the same result. Similarly, the reproducibility is the measurement precision under a reproducibility measurement condition that includes repeated measurements at distinct locations, using different measuring systems and operators. Both repeatability and reproducibility may be expressed quantitatively in terms of the dispersion characteristics of the results.³⁶ 

Contrary to the situation in the earlier stages of luminescence thermometry development, recent papers generally report all figures of merit except for reproducibility. This considerably eases comparison between different luminescence thermometry methods and probe materials. However, the community should be aware of the importance of reproducibility in providing confidence in research results.³⁷ Even though such work may be challenging, in future, efforts must be made to undertake measurement comparison, establishment of consensus values, the optimization of reference material selection, and the determination of confidence limits, in order to advance the practice of luminescence thermometry.

To date, many different principles of temperature measurement from luminescence, so-called temperature readouts, have been demonstrated. Since practically all luminescence features are temperature-dependent to some extent, the inventive combination of these features may lead to new T-readouts, such as the recently reported dual-excited single-band and time-resolved single-band ratiometric readouts which will be discussed in more detail later in this section. Depending on the temporal nature of the observed luminescence phenomena, temperature readouts can be realized in steady-state, time-resolved, and frequency-domain measurement configurations. They differ in the complexity of the components needed for their realization (here, the term complexity also comprises the cost of instrumentation), their sensitivity to external disturbance, processing time, precision, and whether or not they require an external temperature reference element; read-out methods that do not require an external temperature reference are termed self-referencing. Table II presents a comparison of the typical characteristics of frequently used temperature readouts from luminescence.

Considering the number of papers published over the last two decades, spectral-shape-based thermometry realized by measurements of the luminescence intensity ratio (LIR) between two emission bands (ratiometric temperature readout) is the most popular choice for luminescence thermometry. This method is of paramount importance since it is self-referencing and hence problems caused by changes in measurement conditions are circumvented by relying on measurements of the ratios of absolute intensities. Compared to lifetime-based temperature readouts, which are also a very popular method, LIR T-readout is faster, simpler, requires less sophisticated instrumentation, and can be easily adapted for the thermal imaging. LIR can be monitored from luminescent probes containing one and two or more luminescence centers. LIR probes with a single emission center are usually lanthanide-ion-activated phosphors, and, in recent times, transition-metal-ion-activated phosphors, in which the ratio of emission intensities from two adjacent and thermally coupled excited states is utilized as a measure of temperature (Fig. 1).

Higher values of relative sensitivity for LIR can be achieved by utilizing a lanthanide emission from an excited state of higher energy than the first thermally coupled state, because of the larger energy difference, $Δ E ′>ΔE$ (see Fig. 1). It is necessary, however, that a thermal equilibrium exists between these states. This approach is feasible, for example, in Dy³⁺-activated phosphors, where a temperature-dependent intensity ratio between the 480-nm emission from ⁴F_9/2 excited state and the deep-blue 430-nm emission from the ⁴G_11/2 state, rather than the blue 458-nm emission from the ⁴I_15/2 state, may be used (the states are indicated in Fig. 1). The energy difference between ⁴G_11/2 and ⁴F_9/2 is approximately 2400 cm⁻¹, which provides a relative LIR sensitivity of 3.84% K⁻¹, a much higher value than in the conventional LIR utilizing emissions from the ⁴I_15/2 and ⁴F_9/2 states, as recently demonstrated by Li et al.⁴² using Dy³⁺ activated CaWO₄ powders. In addition to the much exploited green and red emissions, an Er³⁺ ion also provides a weak blue emission from the ⁴F_5/2 → ⁴I_15/2 (≃450 nm), ²P_3/2 → ⁴I_11/2 (≃470 nm), and ⁴F_7/2 → ⁴I_15/2 (≃490 nm) transitions.⁴³ The intensity ratio between emissions originating from the ⁴F_7/2 → ⁴I_15/2 (≃490 nm) and ⁴S_3/2 → ⁴I_15/2 (≃550 nm) transitions is more sensitive to temperature changes than the conventional ratio of emissions originating from the ²H_11/2 → ⁴I_15/2 (≃530 nm) and ⁴S_3/2 → ⁴I_15/2 transitions,^44–46 (see Fig. 1), and can reach 3.6% K⁻¹ at 300 K, as the energy difference between ⁴F_7/2 and ⁴S_3/2 is approximately 2260 cm⁻¹. Despite the greater sensitivity, the ratio between the emissions from ⁴F_7/2 and ⁴S_3/2 has been shown to provide about half the measurement resolution with respect to that given by the ²H_11/2 and ⁴S_3/2 ratio, 0.7 K vs 0.3 K.⁴⁶ An additional possibility is the LIR between Er³⁺ emissions from the ²G_9/2 and ²H_9/2 states that are separated by ΔE ∼ 2700 cm⁻¹; however, these emissions may be very weak. Energy differences between excited states larger than 2000 cm⁻¹, suitable for achieving a high relative measurement sensitivity, can be also found in Sm³⁺ (⁴G_7/2 and ⁴G_5/2 states) and Nd³⁺ (⁴F_7/2 and ⁴F_3/2 states), as shown in Fig. 1. Highly sensitive temperature readings using Nd³⁺ activated nanoparticles are favorable for biothermal imaging.^47,48 However, the higher relative sensitivity does not necessarily provide an overall benefit for temperature sensing from luminescence. Emissions from the high-energy excited states usually have lower intensities than those from the lower ones, so the uncertainty in measurements may be higher, which reduces temperature resolution, as was the case for the Dy³⁺ activated CaWO₄ powders.⁴² Thermalization between f and d excited states in divalent lanthanide ion luminescence centers facilitates luminescence thermometry with a high relative sensitivity.⁴⁹ For example, the energy difference between 4f⁵5d¹ and ⁵D₀ (4f⁶) states in Sm²⁺-doped SrB₄O₇ is relatively high (about 4000 cm⁻¹) (Fig. 1), which provides a relative sensitivity for LIR using the d- and f-emissions of about 6.4% K⁻¹ at 300 K. Thermal coupling between these excited levels is possible due to strong electron–phonon coupling.

LIR T-readouts have been recently extended to monitor the ratio of two intensities of emission, $I EX 1$ and $I EX 2$⁠, corresponding to the same upper state being excited by two different energies in succession.⁵⁰ This method is called dual-excited single-band ratiometric thermometry (LIR_GE), and it is schematically depicted in Fig. 2.

LIR methodology can be further advanced if performed with time-resolved measurements. The ratio of emission intensities for the same emission band taken at the different time-delays after excitation [“time-resolved single-band ratiometric luminescence thermometry (TSBR)”] has been demonstrated by Qiu et al.⁵⁶ In this work, hybrid upconversion nanoclusters (UCL-NCs) comprised of PbS quantum dots (3.5 nm in size) and NaYbF₄:0.5%Tm@NaYF₄:10%Yb@NaYF₄:50%Nd (12 nm in size) were used to provide a short-lifetime emission centered at 814 nm and a long lifetime emission centered at 804 nm, respectively. Upon 865-nm excitation, the ratio of the 810-nm emission intensity acquired after different time delays revealed a relative sensitivity of up to 5.6% K⁻¹ and a thermal resolution of 0.5 K at 45 °C. Since both excitation and emission fall in the spectral region of the first biological window, the hybrid nanoprobe is suitable for use in biothermal imaging, as demonstrated by the authors using pork tissue slices of different thicknesses to simulate the rise in temperature of the nanoprobe in biological tissue. In the same study, the authors also performed in vivo thermal mapping of tumors in mice (Fig. 3). The breakthrough of this temperature readout is its insensitivity to tissue absorption and scattering. These two factors can considerably affect readouts based on the intensity ratio of emissions centered at different wavelengths since the magnitudes of both absorption and scattering vary with the light wavelength, and these variations can also vary between the same tissue type for different individuals (in-group variation).

Regarding dual-activated luminescence temperature probes, in recent times, high-sensitivity LIR temperature readouts have been sought among different combinations of lanthanide (Ln) and transition metal (TM) ion emissions. These ions may be co-doped into a single host material or separated in different hosts comprising a binary mixture. Ln/TM-based LIR usually utilizes Ln ions for temperature-independent or slow-changing emission and TM ions for rapidly quenched emissions. Some recent examples include probes based on a Ho³⁺:Y₂O₃ + Mn⁴⁺:Mg₂TiO₄ binary mixture,⁵⁷ Dy³⁺,Mn⁴⁺:BaLaMgNbO₆,⁵⁸ Er³⁺,Ni²⁺:SrTiO₃,⁵⁹ Nd³⁺,Cr³⁺:Gd₅Al_5–xGa_xO₁₂,⁶⁰ Ce³⁺,Mn⁴⁺:Lu₃Al₅O₁₂,⁶¹ Eu³⁺,Mn⁴⁺:Y₃Al₅O₁₂,⁶² and Eu³⁺:GdVO₄ + Cr³⁺:Al₂O₃ hybrid particles.⁶³ The high sensitivity of these probes is a result of the strong T-quenching of the TM emission. Highly T-sensitive LIR probes can be constructed using this approach, taking advantage of the emissions of two lanthanide ions such as those in Nd³⁺,Eu³⁺:YVO₄,⁶⁴ Ce³⁺,Tb³⁺:YBO₃,⁶⁵ Ho³⁺:Y₂O₃ + Er³⁺:Y₂O_3, and Ho³⁺:Y₂O₃ + Nd³⁺:Y₂O₃ binary mixtures.⁶⁶ Kolesnikov et al.⁶⁴ obtained higher relative thermal sensitivity by using a Eu³⁺,Nd³⁺ co-doped YVO₄ probe and better thermal resolution by using the Eu³⁺:YVO₄ + Nd³⁺:YVO₄ binary mixture. The list of possible combinations of activator ions is long, and each combination can provide specific values for a thermometry measurement in terms of the spectral positions of the utilized emissions, achieved sensitivity, and resolution in different temperature ranges, etc. An example of such ingenuity in the optimization of dual-activated luminescence probes can be found in a paper by Pan et al.⁶⁷ They prepared Eu²⁺,Eu³⁺:Sc₂O₃ nanoparticles via a thermal decomposition method using oleylamine as a solvent and a surfactant [Fig. 4(a)].

Oleylamine, however, also reduced the proportion of Eu³⁺ to Eu²⁺ ions, which had the effect of the nanoparticles exhibiting both characteristic Eu²⁺ d–f and Eu³⁺ f–f emissions, the former changing dramatically in intensity with temperature and the latter serving as a reference [Fig. 4(b)]. The strong temperature sensitivity of the Eu²⁺/Eu³⁺ LIR (3.06% K⁻¹ at 267 K) is easily distinguished by the naked eye from the temperature-induced color variations observed under 254-nm excitation [Fig. 4(c)].

The use of the host emission as the reference in LIR is another option when developing highly sensitive LIR probes. This can be accomplished by doping a TM metal into a semiconductor quantum dot (QD) and using the LIR of the TM and QD emissions. Examples of this type of probe include Mn²⁺-doped ZnS QDs, presented in the paper by Wang et al.,⁶⁸ and Mn²⁺-doped perovskite CsPbCl₃ QDs.⁶⁹ Similarly, the use of defect host emissions as the reference is also possible, and Eu³⁺- and Sm³⁺-doped anatase TiO₂ nanoparticles,^70,71 Dy³⁺-activated Gd₂Ti₂O₇,⁷² Eu³⁺-activated SrZrO₃,⁷³ and Mn²⁺-activated Zn₂SiO₄⁷⁴ have all been demonstrated as such. However, one should keep in mind that the reproducible preparation of probes having identical defect emission intensities is a difficult task.

The usability of the temperature dependences of emission and excitation band positions and bandwidths for luminescence thermometry has long been a subject of debate. It had been thought that temperature-induced band shifts and bandwidth broadenings are too small to provide precise temperature measurements. However, such assumptions were based on early findings in lanthanide-activated phosphors, including the very small T-induced shift of the Eu^{3+ 5}D₀ → ⁷F₀ emission band in Y₂O₂S⁷⁵ and the Nd^{3+ 4}F_3/2 → ⁴I_9/2 emission band in LaF₃,⁷⁶ as well as an only 50-cm⁻¹ bandwidth increase over a 630-K range for the Eu³⁺:Y₂O₃ ⁵D₀ → ⁷F₂ emission band.⁷⁷ Somewhat larger T-induced band shifts and broadenings have been observed for semiconductor quantum dots, for example, in CdTe QDs⁷⁸ and CdSe QDs coated with ZnS,⁷⁹ respectively, and in ZnO:Zn and ZnO:Ga.⁸⁰ However, recent studies^81–84 have shown that Ln- and TM-activated phosphors may present band shifts and band broadenings with adequate T-sensitivities, and hence their applicability for temperature sensing ought to be reconsidered, especially in light of the fact that the uncertainties in spectral measurements are quite low compared to, for example, uncertainties in intensity measurements. Ćirić et al.⁸¹ have measured and analyzed T-dependent band shifts and band broadenings in Eu³⁺- and Dy³⁺-activated YVO₄. They found high relative sensitivities (>1% K⁻¹ in the physiological range of temperatures—300–350 K) for the Eu^{3+ 5}D₀ → ⁷F₁ and Dy^{3+ 4}F_9/2 → ⁶H_15/2 band positions. Kolesnikov et al.⁸² have shown that the Nd³⁺(2.4 at. %):YVO₄ ⁴F_3/2 → ⁴I_9/2 band shift has a temperature sensitivity of 0.75% K⁻¹, while Marciniak et al.⁸³ found relative sensitivities of 0.32 and 0.46% K⁻¹ for the band shifts in Nd³⁺-activated LiLaP₄O₁₂ and LiNdP₄O₁₂, respectively. Finally, in a very recent paper, Amarasinghe and Rabuffetti⁸⁴ showed a blueshift of 2.5% K⁻¹ for the red emission band maximum (⁴T₂ → ⁴A₂) in Mn⁴⁺:Na₄Mg(WO₄)₃ at 300 K.

The analysis presented in Ref. 94 shows that the excited-state lifetimes of Mn⁴⁺-activated phosphors (the same analysis may be applied to Cr³⁺-activated phosphors in which Cr³⁺ is in the so-called strong crystal-field environment) can be used for sensitive thermometry in different temperature ranges, depending on the properties of the host material (Fig. 5). Hosts having low phonon-coupling energies to the ²E → ⁴A₂ transition and low values of the cross-over energy are favorable for use at low temperatures, while hosts with large phonon-coupling and cross-over energies can be exploited at high-temperatures. The high non-radiative rate results, unfavorable for traditional phosphor applications (LEDs), facilitate luminescence thermometry with a high relative sensitivity. The choice between Mn⁴⁺ and Cr³⁺ phosphors may be arbitrary. Generally, excited-state lifetimes of Mn⁴⁺ phosphors are shorter than those of Cr³⁺ phosphor, which means that the former are favored in applications that require high temporal resolution (for example, in luminescence thermometry of fluid flows). However, Cr³⁺ phosphors emit at longer wavelengths, well inside the first biological window, which suits biomedical and chemical sensing applications.

Progress in luminescence thermometry must be associated with the development of theoretical models and a careful re-examination of those currently in use. For example, the Boltzmann-type LIR method is frequently used and interpreted with the use of Eq. (1) without prior checks as to whether a Boltzmann equilibrium exists between the excited levels responsible for the LIR emissions. Meijerink and his collaborators^95,96 analyzed the conditions required for the existence of the Boltzmann equilibrium between two excited levels and found that these conditions were not met in the experiments presented in many reports. The absence of the Boltzmann equilibrium leads to a deviation of the T-dependence of the experimental data from that described by Eq. (1). As stated in Ref. 95, the Boltzmann equilibrium exists only when the equilibration of two thermally coupled states is fast in comparison with other processes affecting their population, notably, optical feeding and decay channels to the ground state. Thus, three different temperature behaviors of LIR of emissions from adjacent Ln excited states may occur, as illustrated in Fig. 6. The first, in which nonradiative relaxation is much faster than radiative decay, $1/ τ NR≫1/ τ R$⁠, for which a Boltzmann equilibrium exists (left-hand side of Fig. 6). The second, in which nonradiative and radiative decay rates are similar, $1/ τ NR≈1/ τ R$ (middle section of Fig. 6), and the third, in which the radiative decay is faster than the nonradiative decay, $1/ τ NR≪1/ τ R$ (right-hand side of Fig. 6), do not follow Boltzmann statistics. For larger energy differences between excited levels ΔE, the relaxation between thermally coupled levels becomes slower, so deviations from Boltzmann equilibria occur more easily. Note that Ln activators with large ΔE values are frequently favored for LIR when high measurement sensitivity is needed. Therefore, it is of paramount importance to check for the existence of a Boltzmann equilibrium when exploiting LIR. The simplest way to do that is to perform time-resolved measurements and inspect emission decays from two excited levels. A single exponential decay, identical for emissions from both levels, is expected if a true Boltzmann equilibrium exists.⁹⁵ One way to overcome this problem and ensure a Boltzmann equilibrium is realized is to increase the concentration of Ln activators in the phosphor probe, which increases the non-radiative relaxation rate.

The use of theoretical modeling may reduce the requirements for experimental work, as demonstrated by Ćirić et al.⁹⁷ They showed that Judd–Ofelt intensity parameters derived from the room-temperature optical spectrum of a Ln³⁺-activated phosphor can be used to calculate the thermometric properties of materials, as an LIR readout, with an error of less than 5%. Moreover, one can use the tabulated Judd–Ofelt intensity parameter values, which are abundantly available in the literature, to rapidly search for appropriate materials for sensor probes. Recently, this theoretical approach was extended to dual-excited single-band ratiometric thermometry.⁵¹ Finally, it is important to mention that there are no reports on the use of materials informatics and data-driven approaches in luminescence thermometry research, to the best of author's knowledge. Given the immense developments in this field, it is reasonable to expect that such approaches may provide benefits in further luminescence thermometry developments.

Today, luminescence thermometry has many applications in diverse research fields and environments, including electrical and mechanical engineering, biomedicine, nanotechnology, and microfluidics; for a comprehensive insight into luminescence thermometry applications, the reader may refer to two books^6,7 and the references therein. However, unsurprisingly, since temperature is one of the most commonly measured physical quantities and has effects on every aspect of our daily life, the field of applications of luminescence thermometry continues to expand. In this section, some of the novel and critical applications of luminescence thermometry are highlighted.

Recognized as one of the key chemical reactions for the advancement of civilization, catalysis has been extensively researched and, since the 1950s, has been a component of about 85%–90% industrial chemical processes.⁹⁸ Catalysis technologies have had, and continue to have, a strong positive impact on the preservation of the environment—for example, in atmospheric and wastewater depollution processing—and are backbones of the development of green chemistry for sustainable societies and the transition to carbon-neutral operations. The catalytic technology for methanol-to-olefins/aromatics/gasoline conversion is a promising alternative to classical production routes for high-demand chemicals and intermediates, helping to eliminate reliance on oil, to point-out just one among the many emerging uses of catalysis that are highly beneficial for the environment and promotion of sustainable development.⁹⁹ In catalysis research and applications, temperature is an essential parameter since it governs the reaction speed and influences the activation energy. Therefore, accurate temperature measurements of both the catalyst sample itself, as well as the reactants and products, are vital for comprehensive studies of catalytic reactions and processes.¹⁰⁰ Indeed, there is a particular requirement for temperature measurements to be performed under working conditions, which can be challenging. Previously, such measurements have been performed by using a variety of methods, such as IR thermography, multiple thermocouples, and NMR thermometry (see Ref. 101 and the references therein). For catalytic applications, however, these thermometry methods have serious drawbacks, such as low spatial resolution and complicated data analysis (as in the case of NMR thermometry). Recently, the great potential of luminescence thermometry for in situ temperature measurements during catalytic processes has been demonstrated in several papers.

Pfaff et al.¹⁰⁰ have demonstrated the use of luminescence thermometry for measurement of the temperature of a catalyst sample in a flow reactor, for heterogeneous catalysis, and showed that there was excellent agreement between this temperature and that determined by IR-thermometry. In this study, a Mg₃F₂GeO₄:Mn⁴⁺ red-emitting phosphor was mixed with a magnesium aluminum silicate hydroxypropyl cellulose (HPC) binder and applied to the heating element in the reactor. The temperature was derived from the lifetime of the phosphor emission (excited by 266-nm Nd:YAG laser radiation), and a single shot accuracy of ±2 K was achieved. Geitenbeek et al.¹⁰¹ mixed NaYF₄:Yb³⁺,Er³⁺ microcrystalline upconversion particles with a commercial zeolite H-ZSM-5 to monitor the temperature at different heights in a reactor bed during a methanol-to-hydrocarbon catalytic reaction. By exploiting the ratio of the green upconversion emission intensities corresponding to the ⁴I_15/2 → ⁴I_15/2 and ⁴S_3/2 → ⁴I_15/2 transitions under 980-nm excitation, the authors were able to visualize a front of increasing temperature migrating down the fixed reactor bed, which occurs due to the exothermic nature of the catalytic reaction. A fiber probe, positioned in succession at different heights in the catalyst bed, was used to both excite and to collect the emission from the NaYF₄:Yb³⁺,Er³⁺ particles. The uncertainty in measurement varied from 0.3 K, at the start of the catalytic reaction, to ca. 22 K, at later measurement times, due to the lowering of the emission signal intensity. Hartman et al.¹⁰² used novel catalyst extrudate sensors that facilitate both luminescence thermometry and shell-isolated nanoparticle-enhanced Raman spectroscopy (SHINERS)^103,104 during the catalytic process of direct conversion of the synthesis gas (a fuel gas mixture consisting of hydrogen and carbon monoxide) into hydrocarbons. The catalyst extrudate sensors [Fig. 7(a)] consist of millimeter-sized SiO₂ extrudates supporting attached quasi-spherical Au@SiO₂ nanoparticles [Fig. 7(b), Au diameter ≈88 nm; SiO₂ shell thickness ≈2.6 nm], NaYF₄:Yb³⁺,Er³⁺@SiO₂ nanoparticles [Fig. 7(c), NaYF₄:Yb³⁺,Er³⁺ core ≈25.4 nm, SiO₂ shell thickness ≈12.2 nm, Yb³⁺ concentration = 18 mol. % and Er³⁺ concentration =2 mol. % Er³⁺], and the Rh catalyst nanoparticles [Fig. 7(d)]. As for the previous example, the ratio of green upconversion emission intensities was used to monitor the temperature during an operando study of the catalytic reaction; a temperature measurement range extending to 900 K and a standard deviation of 0.3 K (calculated for >50 emission spectra) were achieved. Luminescence thermometry revealed the presence of a large offset of 50 K between the set reactor temperature and the catalyst temperature during CO hydrogenation, and a dependence on the CO/H₂ gas feed ratio at a constant flow speed resulting in temperature differences of up to 40 K.

Thermal history sensors, also known as off-line thermometers, are used to provide forensic evidence of the temperatures that they experience during thermal events of interest. Such temperature recording is important for monitoring the integrity of pharmaceutical products, ensuring the quality of foods and beverages, and monitoring the degradation of consumer electronics, batteries, and even textiles.¹⁰⁵ It is also much needed in high-temperature environments for the development and safe use of chemical reactors, engines, and gas turbines. Furthermore, in some cases, off-line measurements are the only option to assess the temperature distribution at the location of interest during a thermal event, for example, in harsh environments, such as during laser annealing or an explosion.⁸ The same is true for areas that are not physically accessible, such as the interior of high-temperature solid oxide fuel cells. In contrast to the abundance of traditional thermometers, thermal history sensors are rare and mostly still at the research stage, owing to the stringent requirements for their effective functioning. Such sensors should be based on an irreversible change of sensor properties with temperature and should be capable of following rapid temperature changes without damage, measuring and storing temperature and time in some manner without internal or external sources of power and free of communication, which requirements are not very compatible with the thermal event to be monitored. Temperature-sensitive paints and coatings (or thermal paints and coatings) are the best known and exploited thermal history sensors. These paints are made of metallic compounds (Pb, Co, Mg, etc.), binders, and solvents. Due to the chemical reactions of the metallic components at elevated temperatures, the paint permanently changes its color, and this color change indicates the highest temperature to which the paint was exposed. For thermal coatings, the inclusion of a binder is avoided to allow the thermal sensor to endure longer in harsh environments, but, as a result, its application requires complex deposition techniques. In general, the use of thermal paints and coatings requires cumbersome interpretation of data, the toxicity of the metal components in paints is problematic, color measurements may be compromised by metal fragments or other environmental components, and the method provides a temperature resolution that is at best 10 K.¹⁰⁶ 

Thermographic phosphors in the form of powders, paints, and coatings, are an alternative to thermal history paints and coatings with exceptional potential, and may provide a solution for off-line temperature measurements in many, if not all, the above-mentioned fields. Contrary to “on-line” thermographic phosphors, whose emission properties must reversibly and reproducibly change with temperature, thermal history phosphors (TH-phosphors) should undergo irreversible changes in their emission properties with temperature change; the emission properties being any of those already mentioned that are exploited in luminescence thermometry (changes in emission shape, lifetime, etc.). So far, four processes that can promote thermally driven permanent changes in the luminescence of a phosphor during its exposure to elevated temperatures have been identified: (i) amorphous-to-crystalline transitions, i.e., the crystallization of the phosphor, (ii) phase-change processes, (iii) diffusion of ions into the host material, usually emission killer center ions, and (iv) the oxidation of dopant activator ions, for example, the oxidation of Eu²⁺ to Eu³⁺. Compared to color measurements, luminescence measurements are unaffected by ambient conditions, the presence of metal fragments and environmental artifacts, and can provide considerably higher temperature and spatial resolutions. For off-line measurements, temporal resolution is not of particular importance.

Rabhiou¹⁰⁷ prepared amorphous Y₂SiO₅:Tb phosphor via a solgel technique. It was shown that upon heating, the crystallization of the phosphor resulted in a monotonic increase in the Tb³⁺ excited-state lifetime. A thermal history phosphor paint was prepared from the amorphous Y₂SiO₅:Tb phosphor and a commercial binder, applied to a 180-mm stainless-steel disk, and then exposed to heating for a period of 40 min using a propane torch [Fig. 8(a)]. During the heat exposure, the temperature of the upper surface of the disk was measured using thermocouples in contact with the surface; excellent agreement between the temperatures obtained from the emission decays and the thermocouples was demonstrated, as shown in Fig. 8(b).^107,108 Copin et al.¹⁰⁹ have shown that temperature-induced crystallization of solgel-prepared erbium-activated yttria-stabilized zirconia powder (YSZ:Er³⁺) enables thermal history reading on account of large increases in the emission intensity and decay time over the 1223–1423 K and 1173–1373 K ranges, respectively. Estimated theoretical resolutions for these off-line measurements are between 0.3 and 1 K for the intensity and between 6 and 2 K for the decay time readouts—much higher than that achievable using thermal paints in the same temperature range. The different mechanisms causing irreversible changes in the phosphor emission, i.e., the thermally activated oxidation of activator ions, have been investigated by Rabhiou et al.¹¹⁰ They showed that the oxidation of valence ions in the +2 state in BaMgAl₁₀O₁₇:Eu²⁺, BaMgAl₁₀O₁₇:Eu²⁺,Mn²⁺, and SrAl₁₄O₂₅:Eu²⁺ can be used for emission-intensity ratio based thermal history recording over the 600–1300 °C range. In addition, because annealing phosphor powders in an Ar atmosphere at 1400 °C reduces the activator ions to their initial valence state, these sensors may be reusable.¹¹¹ 

Over the years, luminescence thermometry has been successfully used to substitute or complement measurement techniques for the sensing and measurement of various important physical properties or processes other than temperature. For example, the use of luminescence thermometry for temperature measurements in gas and liquid flows has been demonstrated and is still being extensively researched, as recently reviewed by Abram et al.³⁵ Future developments, as suggested in that review, may be concerned with three-dimensional “tomographic” temperature and velocity measurements, and the identification and/or production of more suitable phosphor particles, having near-spherical shapes, narrow size distributions, and higher brightness, to improve measurement accuracy. Considering the immense importance of controlling and measuring temperatures in microfluidic devices,⁷ exploration of the possibilities for luminescence thermometry in this application field continues. The ideal temperature measurement technique for a microfluidic device should possess the following characteristics: (i) high spatial and temperature resolutions, (ii) high acquisition rate (temporal resolution), (iii) non-invasiveness, (iv) inertness in a solvent, and (v) the thermometer material should be light- and heat-resistant.¹¹² Luminescence thermometry meets these criteria since it can combine the high relative thermal sensitivity (>1% K⁻¹) and spatial resolution (<10 μm) in short acquisition times (<1 ms). So far, organic dyes have mainly been used for luminescence thermometry in microfluidic devices. However, dyes suffer from photobleaching, may react with fluids, and have limited measurement ranges due to degradation. These drawbacks may be overcome by adopting inorganic probes, as recently shown by Geitenbeek et al.,¹¹³ who showed that the ratio of Er³⁺ emission intensities in NaYF₄ nanoparticles can be used for in situ temperature mapping with a temperature resolution of 0.34 K and a spatial resolution of ca. 1 mm. Further progress is expected in measurements of fast thermal transients by high-speed phosphor thermometry (HSPT). Such measurements are needed in internal combustion engines, swirl-stabilized gas turbine model combustors, turbulent flows, etc., where conventional thermometry methods, such as optical pyrometry, are infeasible.^114,115 The improved quality and reduced cost of high-speed photodetectors, such as high-frame-rate CMOS cameras,¹¹⁶ as well as the increased availability of bright fast-decaying phosphors, should prove beneficial for future HSPT applications. In addition, new approaches for temperature determination are welcomed, and some, such as the use of the anti-Stokes luminescence of YAG:Ce³⁺,¹¹⁵ are actively being investigated.

Research also continues on the development of luminescence thermometry for extremely high temperatures, which is important for modern turbine engines where the temperature limits for luminescence thermographic materials are tested.¹¹⁷ Recently, Allison et al.¹¹⁷ developed high-temperature luminescence-based fiber optic thermometry for temperatures up to 1700 °C. The optical system designed in this study was comprised of a Nd:YAG 355-nm laser injected into one leg of a flexible bifurcated optical fiber connected to a sapphire light pipe positioned in proximity to YAG:Dy³⁺ or YAG:Dy³⁺,Er³⁺ crystalline cylindrical probes (Fig. 9). The Dy³⁺ emission was measured after passing through another arm of the bifurcated fiber, a filter, and a lens, by a PMT detector. Temperature values were determined from the emission decay times, with sensitivities of 1% K⁻¹ at 1400 °C and 0.9% K⁻¹ at 1700 °C. At approximately the same time, Anderson et al.¹¹⁸ demonstrated thermometry based on emission intensity ratios up to 1773K, using a YAG:Dy³⁺ thin film as a phosphor probe, and detecting Dy³⁺ luminescence to 2033K. Moreover, with this technique, high-speed temperature measurements (at rates up to 80 kHz) were shown to be possible under laboratory conditions.

Runowski et al.¹¹⁹ converted a luminescent thermometer (YVO₄:Yb³⁺,Er³⁺) into a remote vacuum sensor. First, they demonstrated enormous enhancement of light-to-heat conversion and heating of the YVO4:Yb³⁺,Er³⁺ particles under irradiation by a 975-nm laser in a vacuum, compared to the heating effects observed under ambient conditions, when air molecules transfer the generated heat to the surroundings and cool the particles [Fig. 10(a)]. Then, they showed that on account of pressure–temperature interdependence, pressures can be measured from the temperature-induced Er³⁺ emission intensity ratio change (525/550 nm) [Fig. 10(b)], in the pressure range from 0.07 to 10 mbar; sensitivities ranging from 500% mbar⁻¹ to 1% mbar⁻¹ and resolutions from 0.1 to 10 mbar, depending on the nominal pressure were recorded. Besides being a highly innovative application, the importance of this result lies in the fact that it suggests that luminescence probe heating must be considered for luminescence thermometry testing and applications in low-pressure environments, for example, when performing measurements at low temperatures in vacuum cryostats. Another recent example of luminescence thermometry for measurements of physical properties other than temperature was reported by Miandashti et al.,¹²⁰ who used luminescence thermometry to study the photothermal chirality of Au helicoid nanoparticles. They spin-coated helicoid nanostructures on an AlGaN:Er³⁺ thin film and used a 532-nm laser for photothermal and photoluminescence excitation of gold nanostructures and Er³⁺. In the course of this experiment, the optical and photothermal handedness, and an absolute temperature difference of 6 K between right- and left-circularly polarized light, were identified.

Recent sensor strategies have foreseen future designs of compact, lightweight, portable, and low-cost sensor systems.¹²¹ Smartphones may become a platform for the construction of next-generation cost-effective hand-held sensors, due to their portability and ubiquitous availability (with an estimated more than two billion users worldwide). Such immense availability provides a large number of potential subscribers for smartphone-based sensor systems.¹²¹ Smartphone-based measurements of optical spectra and emission lifetimes have been already demonstrated in several papers.^122,123 Pan et al.¹²² demonstrated temperature measurements by observing changes in the TiO₂ thin film reflection spectrum using a smartphone. Zhu¹²³ presented a method to record upconversion emission decays with a low-cost and miniaturized apparatus consisting of a smartphone equipped with a 980-nm CW laser and motor.¹²³ As well as the basic display by smartphones of methods for luminescence measurements, several smartphone-based luminescence sensing applications for consumer goods monitoring have been published recently. Araque et al.¹²⁴ published a technique for the determination of gaseous oxygen concentrations inside packed foods based on the use of a luminescent membrane sensitive to O₂, optically excited and read by a smartphone. Ramalho et al.¹²⁵ introduced smart luminescence quick response (QR) codes, which can be used to store information as well as to sense the temperature in real time via the acquisition of a photograph using the charge-coupled device of a smartphone (Fig. 11). Organic–inorganic hybrids with europium (Eu³⁺) and terbium (Tb³⁺) ions used in the preparation of the QR codes provide information on the temperature to which the QR code is exposed, based on the intensity ratio of red and green pixels of an image obtained by the smartphone.

The redefinition of the kelvin in terms of the Boltzmann constant has no effect on the temperature values or realization uncertainties of the two currently defined temperature scales (ITS-90 and PLTS-2000).¹²⁷ However, with the new definition, primary realizations of the kelvin can be established at any point of the temperature scale, which is significant for thermometry at low (<20 K) and high (>1300 K) temperatures and opens space for the development of primary luminescence thermometry for different environments.

A similar approach for obtaining the equation of state can be performed with other models that describe temperature variations of luminescence lifetimes, such as those that include multiphonon relaxation or quenching via charge-transfer band,¹⁴ but with more complex equations and a larger number of parameters that must be independently derived.

Regarding the luminescence intensity ratio (LIR) Boltzmann-type temperature readout, the realization of the primary thermometer may present a considerable challenge, even though the equation relating LIR and T is very simple [Eq. (1)]. The challenge arises due to the fact that parameters in Eq. (1) (B and $ΔE$⁠) cannot really be considered to be temperature independent over a relatively wide temperature range. B incorporates the radiative transition probabilities and emission barycenter energies of two transitions, which vary with temperature, although weakly. The same stands for the energy difference between the excited levels, $ΔE$⁠. This problem may be overcome by introducing an approximation for the temperature dependences. Finally, one should consider that if the equation of state is based on an approximation of a complex theory, it must at least be possible to estimate the order of magnitude of the deviation from theory.¹²⁸ 

Recently, several attempts at the realization of primary luminescence thermometry have been made, and three examples from among these are highlighted here. Botas et al.¹³² showed that the thermal dependence of the emission peak position of luminescent silicon nanoparticles may provide primary thermometry over the 13–480 K range. Previously, Thomas et al.¹³³ observed the temperature-induced change in the emission wavelength of a CdSe(ZnS):SiO₂ thin film nanocomposite (∼0.11 nm K⁻¹) over the 295–525 K range. In both cases, the Varshni relation was used as an equation of state. The upconversion emission intensity ratio of SrF₂:Yb³⁺,Er³⁺ powder has been exploited by Balabhadra et al.¹³⁴ to propose a primary thermometer for which the need for temperature calibration is overcome by using the value of the thermometric parameter in the limit of zero optical pump excitation power.

The last decade has witnessed a nearly exponential growth in the number of published papers and patents in the field of luminescence thermometry. This reflects extensive work carried out by a large scientific community and indicates the importance of the field for current science and technology research and development. There are still many ways in which luminescence thermometry might be advanced in the future. The recent reporting of new temperature-read-out techniques and high-sensitivity probe constructions are proof that there is plenty of space for improvements, even in these basic methodological elements. Likewise, there are yet many unexplored applications for which luminescence thermometry may prove beneficial. However, the adaptation of the method for novel applications requires further extension of background scientific knowledge, due to the interdisciplinary nature of these applications, in most cases. The search for more sensitive readouts and probes should be followed by a detailed evaluation of other thermometric parameters, the temperature resolution in particular. But, above all, future development of the method must resolve its inherent problems and faults, as well as facilitate its widespread use among experts and professionals. Thus, issues such as the assessment of measurement reproducibility and health and environmental effects should be resolved so that the scientific and engineering community as a whole gains the confidence in the method.

The research was funded by the Ministry of Education, Science and Technological Development of the Republic of Serbia and the National Recruitment Program of High-end Foreign Experts (Grant No. GDT20185200479) offered by the Chongqing University of Posts and Telecommunications (CQUPT), People’s Republic of China.

The data that support the findings of this study are available from the corresponding author upon reasonable request.