Implementing Defects for Ratiometric Luminescence Thermometry

Abstract

In luminescence thermometry enabling temperature reading at a distance, an important challenge is to propose new solutions that open measuring and material possibilities. Responding to these needs, in the nanocrystalline phosphors of yttrium oxide Y₂O₃ and lutetium oxide Lu₂O₃, temperature-dependent emission of trivalent terbium Tb³⁺ dopant ions was recorded at the excitation wavelength 266 nm. The signal of intensity decreasing with temperature was monitored in the range corresponding to the ⁵D₄ → ⁷F₆ emission band. On the other hand, defect emission intensity obtained upon 543 nm excitation increases significantly at elevated temperatures. The opposite thermal monotonicity of these two signals in the same spectral range enabled development of the single band ratiometric luminescent thermometer of as high a relative sensitivity as 4.92%/°C and 2%/°C for Y₂O₃:Tb³⁺ and Lu₂O₃:Tb³⁺ nanocrystals, respectively. This study presents the first report on luminescent thermometry using defect emission in inorganic phosphors.

1. Introduction

A whole range of optically active emitters can be successfully used in luminescence thermometry due to the susceptibility of their optical properties to the temperature changes, like: organic dyes, quantum dots (QDs), metal–organic frameworks (MOFs), polymers, nanodiamonds (ND), inorganic nanoparticles doped with lanthanides (Ln³⁺) and transition metal ions (TM) etc. Organic dyes [1,2,3,4] usually emit in the visible range upon excitation with ultraviolet light. Their emission is dependent on temperature, because it is affected by the electronic transitions between vibrational states responsible for luminescence. The luminescent properties of QDs, which result from the recombination of the electron–hole pair are strongly dependent both on the host material composition as well as their size [5,6,7]. In this case, the phonon-related depopulation processes and the energy gap are strongly temperature dependent. Therefore, QDs are often implemented to spectral position- and intensity-based thermometric approaches. However, their use as the source of one of the luminescent signals in more complex ratiometric nanothermometric systems can also be found [8,9]. Luminescent thermometers (LTs) based on MOFs take advantage from the fact that probability of host-to-metal and metal-to-metal energy transfer probabilities are thermally induced, to enable noncontact temperature readout [10,11,12,13]. The unlimited number of combinations of various ions and links, enables designing structures with intentional properties. Polymers [14,15,16,17,18], which most often reveal visible luminescence upon UV excitation, are another group of materials useful in LTs. In spite of their relatively low quantum efficiency, they constitute a significant group of nanothermometers due to their good solubility in water and the strong dependence of their luminescence on local environment changes. By combining polymers, for example with organic dyes or QDs, it is possible to obtain very sensitive sensors that use polymer conformational changes. In luminescence thermometry, examples of the use of color centers found in diamonds, such as luminescent tin-vacancy [19], have also been noted. Thermally induced shifting of the zero-phonon line as well as a change in the intensity of the emission of these intentionally introduced optically active centers was observed.

Interesting alternatives to already described LTs pose inorganic nanocrystals doped with optically active ions, most often lanthanides [20] or transition metals [21,22,23,24]. A number of their advantages, over other groups of material used for LTs, like high chemical and thermal stability and the lack of photobleaching and photoblinking, lead to the great research interest that they have attracted. In this case, the difference in the thermal dependence of particular emission bands enables the implementation of the ratiometric approach. The optimization of the host material composition and the optically active ions enables the alteration of the thermometric properties of this type of luminescent thermometers according to the requirements. Due to the fact that each of those already reported types of LTs possess some drawbacks and there is no universal luminescent temperature sensor, new approaches and new materials that may be implemented to noncontact temperature sensors are still searched for.

In the nanocrystalline lanthanide oxides, broad bands from surface oxygen defects have been noted. They were investigated in e.g., Y₂O₃ doped with Eu³⁺ [25] or Tb³⁺ [26], in which the defect emission intensity was comparable to that of dopant ions in a certain temperature range. The influence of temperature on the emission of defects has been demonstrated, which led us to the conclusion that such structures could be intentionally used in optical thermometry. Usually, the emission of such structures is quenched with temperature, however, for energy mismatches of excitation radiation, it is possible to note an increase in the intensity of defect emission, which is associated with satellite phonons [25]. In this work, we show for the first time the exploitation of emission of oxygen defects, occurring in rare earth oxides, in excitation-ratiometric approach to luminescence thermometry using Tb³⁺ ions emission as the luminescent reference. It has been shown that the Tb³⁺ emission band obtained upon UV excitation decreases at elevated temperatures. On the other hand, the defect emission upon green excitation reveals thermal enhancement. The opposite monotonicity of these two signals integrated in the same spectral range enables the development of single band ratiometric (SBR) luminescent thermometers based on intrinsic defect emission (Figure 1).

2. Materials and Methods

The nanocrystalline rare earth (RE) oxides phosphors (Y₂O₃ and Lu₂O₃) doped with the trivalent terbium Tb³⁺ ions (0.05%, 0.1%, 0.2% and 0.5% mole in respect to Y³⁺ and Lu³⁺, respectively) were synthesized using the citric acid-assisted method. The following compounds were used as starting materials: yttrium oxide (Y₂O₃ of 99.999% purity from Stanford Materials Corporation (SMC, Lake Forest, CA, USA.), lutetium oxide (Lu₂O₃ of 99.995% purity from SMC), terbium oxide (Tb₄O₇ of 99.99% purity from SMC), nitric acid (HNO₃ of 65% purity from Avantor, Gliwice, Slaskie, Poland), and citric acid (C₆H₈O₇ of 99% purity from Sigma-Aldrich, Poznan, Wielkopolskie, Poland). The amounts of all the precursors used in the synthesis were stoichiometrically calculated per 0.5 g of product. The first step was the production of the rare earth nitrates by addition of a significant excess of HNO₃ to the water solutions of the oxides. Then, the residual HNO₃ was eliminated by a recrystallization process. The citric acid was then added, and the mixture was kept at the heating plate at around 100 °C in ambient atmosphere for about a day to obtain a resin. The ratio of citric acid moles to metal moles was 1:2. The resultant substance was annealed at 900 °C for 8 h in air.

Powder diffraction studies were carried out on a PANalytical X’Pert Pro diffractometer equipped with an Anton Paar TCU 1000 N temperature control unit using Ni-filtered Cu Kα radiation (V = 40 kV, I = 30 mA). Rietveld’s analysis was performed using the X’Pert HighScore Plus software (version 2.2d (2.2.4), Malvern Panalytical, Malvern, Worecestershire, UK).

Raman spectra were measured on inVia confocal microscope from Renishaw supplied with the Si CCD camera for detection and an 830 nm excitation line. The spectra were taken at room temperature under 20× objective. The spatial resolution was lower than 1 µm.

Transmission electron microscope images were taken using the FEI Tecnai G2 20 X-TWIN microscope supplied with the CCD FEI Eagle 2K camera with a HAADF detector and electron gun with a LaB6 cathode. Moreover, the microscope was supplied with the X-ray microanalyzer EDAX.

The excitation spectra, the photoluminescence decay curves and the quantum yield measurements were measured at room temperature using the FLS980 fluorescence spectrometer from Edinburgh Instruments (version 980, Edinburgh Instruments, Livingston, UK). The measurements were carried out with R928P side window photomultiplier tube from Hamamatsu detector and an integrating sphere for quantum yield measurements. The excitation line was obtained using a 450 W halogen lamp and the micro-flash lamp.

The temperature-dependent emission spectra were measured using the 266 nm excitation line from a laser diode or the 543 nm line from OPOLLETE 355 LD OPO and measured using a Silver-Nova Super Range TEC Spectrometer form Stellarnet (1 nm spectral resolution, Tampa, FL, USA.). The temperature of the sample was controlled using a THMS 600 heating stage from Linkam (0.1 °C temperature stability and 0.1 °C set point resolution). Measurements of emission over a wide temperature range with both of the excitation lines were carried out using band pass filters 400–500 nm from Thorlabs. The temperature was reduced from 200 °C to 20 °C with 20 °C steps, whereas the time needed to stabilize the temperature controller between the double emission measurements was about 2 min.

3. Results

The obtained RE oxide nanocrystals underwent structural and morphological characterization. As for the crystal structure, all the studied materials crystallize in the space group I a −3 (206) with a cubic unit cell. Rare earth atoms located inside the ordered volume of nanocrystals are placed in the octahedral sites, having six atoms of oxygen as their nearest neighbors. The X-ray powder diffraction results are shown in Figure 2a. An adequate phase purity was confirmed by the consistency of the received diffractograms with the ICSD pattern data (#193042, #428543 for Y₂O₃ and Lu₂O₃, respectively). No shifts or changes in the relative peak intensities were observed for the increase in the Tb³⁺ dopant concentration. This is because the Tb³⁺ ion easily and without a noticeable impact substitutes for Y³⁺ and Lu³⁺ ions, due to the fact that those ions have similar ionic radii: 106.3 pm, 104.0 pm and 100.1 pm, respectively [27]. Based on the Rietveld refinement with respect to the reference data, the average sizes of the obtained nanocrystals were estimated (Figure S1). Despite the uncertainty of such analysis associated with the probably significant size distribution, some trend is visible: Y₂O₃ nanocrystals are larger than Lu₂O₃, which agrees with the values of RE ionic radii. For larger ions, a larger unit cell is obtained (see Figure S2), which can result in such a dependence of crystalline sizes. In fact, a smaller unit cell parameter was determined for Lu₂O₃ than for Y₂O₃, which were about 1.039 nm and 1.061 nm, respectively. Only a slight increase in this parameter was recorded in both nanocrystalline matrices when the Tb³⁺ content was increased. However, due to the chosen synthesis method, the surface of nanocrystals is highly defective. The value of the shortest metal–oxygen distances (M–O) is also larger in the case of Y₂O₃ (0.2206 nm) with respect to the Lu₂O₃ (0.2189 nm), which indicates that stronger crystal field strength affects the dopant ion in Lu₂O₃ (Figure 2b) [28]. Because of slight changes in the electron–phonon coupling, different phonon energies were observed in Raman spectra (Figure 2c). In both RE oxides, the arrangement of the peaks is similar and the appearing modes are marked on the graph (A-singly degenerated, E-doubly degenerated, T-triply degenerated, g-even) [29]. In the case of Y₂O₃, the corresponding peaks are shifted towards the lower energies with respect to Lu₂O₃, which results from the smaller unit cell of the Lu₂O₃ host. The energy of the dominant Raman peak is observed around 378 for Y₂O₃ while around 392 cm⁻¹ for Lu₂O₃. Moreover, its width increases slightly for Lu₂O₃ relative to Y₂O₃, which is consistent with the tendency shown by the average size of crystallites (Figure S1), because the smaller the grains, the larger the peak width. The analysis of the TEM images indicates that Y₂O₃:Tb³⁺ and Lu₂O₃:Tb³⁺ nanocrystalline powders consist of well crystalized and highly agglomerated particles of the average size of around 56 and 52 nm, respectively (Figure 2d, see also Figure S3). Obtained results are also consistent with average crystallite sizes calculated from XRD data.

The optical properties of the studied nanocrystallites were also investigated. The excitation spectra measured at λ_em = 540 nm matching ⁵D₄ → ⁷F₅ transition is presented in Figure 3b. In the case of both RE oxides, two broad bands in the ultraviolet range are visible. The first one around 260 nm corresponds to the spin-allowed 4f⁸ → 4f⁷5d¹ transition of the Tb³⁺ ion, while the band around 300 nm corresponds to the spin-forbidden 4f → 5d transition [30]. Because of the large width of the observed bands, and the fact that the measurements were carried out at room temperature, it is not possible to accurately determine whether the absorption resulting from the presence of defects is visible in the excitation spectrum. For both RE oxides, an increase in the relative intensity of the excitation bands of Tb³⁺ ions was observed for higher concentrations of Tb³⁺ ions in the crystalline matrix. This is due to the fact that a larger amount of Tb³⁺ ions results in the increase of the absorption cross-section, which in turn causes stronger emission. Due to the relatively large energy separation between ⁵D₄ and next lower laying energy state (⁷F₀) and the distinctive energy level scheme of Tb³⁺ ions, the ⁵D₄ emitting state is barely affected by either multiphonon depopulation or cross relaxation processes. Therefore, an increase of the dopant concentration results in the enhancement of the emission intensity. The representative emission spectra measured at λ_exc = 266 nm for RE oxides with 0.5% Tb³⁺ shown in Figure 3c consist of a series of bands around 490, 544, 584, 623, 650, 666 and 685 nm that correspond to the transitions from the ⁵D₄ level to the ⁷F_J sublevels (J = 6, 5, 4, 3, 2, 1 and 0, respectively). Although the emission spectra of Tb³⁺ doped RE oxides are very similar qualitatively, there is a large quantitative difference in the absolute brightness of emissions. Upon 266 nm resonant excitation, stronger emission intensity is observed for Y₂O₃:Tb³⁺, which is confirmed by quantum yield measurements presented in Figure S4. Moreover, with the increase of Tb³⁺ ion concentration, the quantum yield increases for each RE oxide. This effect is related to the fact that, in the case of nanocrystals, Ln³⁺ ions prefer to occupy defective surface positions, which are strongly quenched by the surface-related nonradiative processes [31]. The increase of Tb³⁺ concentration leads to the higher probability of less surface-affected sites’ occupation in the core part of the nanocrystals. Additionally, as was already mentioned, the concentration quenching is not very probable in the case of the ⁵D₄ state. Synergy of these two effects causes the enhancement of emission intensity and luminescence quantum yield with enlargement of the Tb³⁺ amount. It is therefore evidenced that the surface is strongly defected in the case of the investigated nanocrystals. The obtained luminescence decay curves are presented in Figure 3d for the representative 0.5% of Tb³⁺ concentration. The lifetime elongation of the ⁵D₄ energy state of Tb³⁺ ions in the Lu₂O₃, with respect to Y₂O₃, is associated with the higher spin-orbital coupling in the case of Lu₂O₃ nanocrystals. For the lowest Tb³⁺ concentration under investigation, the lifetimes of 0.90 ms and 0.93 ms were obtained, for Y₂O₃ and Lu₂O₃, respectively. For a concentration of 0.5% Tb³⁺, for which the luminescence decay curves are presented in Figure 3d, the obtained lifetime values were 0.86 for Y₂O₃ and 0.89 for Lu₂O₃. Decay times also show a dependence on Tb³⁺ concentration. The lifetimes determined by single-exponent fit are shown in Figure S5. It was observed that for the tenfold increase in concentration of Tb³⁺, shortening of average lifetimes reached only 4–7% of the initial value obtained for the lowest dopant concentration. For Y₂O₃, the change from 0.90 ms to 0.86 ms was noted for the increase in Tb³⁺ content in the matrix from 0.05% to 0.5%, whereas for Lu₂O₃ a change from 0.93 ms to 0.89 ms was observed. The shortening of lifetimes as the Tb³⁺ concentration increases results from the fact that the average distance between ions decreases, which in turn opens the paths of energy migration among excited states of Tb³⁺ ions to the surface quenchers. However, in this case, the shortening of the lifetimes for higher Tb³⁺ concentrations have no direct effect on the intensity of emissions, because the competitive process associated with the increase in the number of emitters located in the nanocrystal volume is dominant. That is why in the excitation spectrum the intensity of the emission band at 540 nm increases for higher Tb³⁺ concentrations (Figure 3b).

The luminescent properties of the obtained RE oxides nanocrystals doped with Tb³⁺ ions were investigated in a wide temperature range from 20 to 200 °C upon λ_exc = 266 nm and λ_exc = 543 nm and an appropriate band pass 400–500 nm filter. It was estimated from the excitation spectra that the 266 nm excitation line is suitable for the direct resonant excitation of Tb³⁺ ions. This ultraviolet line causes the 4f → 5d transitions of Tb³⁺ ions, and then leads to the population of ⁵D₃ and ⁵D₄ levels from which the emission occurs. However, the emission from the ⁵D₃ level is negligible, which may result from the probable {⁵D₄, ⁷F₆}↔{⁵D₃, ⁷F₀} cross-relaxation processes. The representative thermal evolution of the Y₂O₃:0.5%Tb³⁺ emission spectra in the monitored spectral range is shown in Figure 4a. Only one emission band associated with the ⁵D₄ → ⁷F₆ transition is observed in this spectral range, the intensity of which decreases with increasing temperature. This effect may be associated with either multiphonon depopulation of the ⁵D₄ state, of which probability is dependent on temperature, or with a decrease in the absorption cross-section as the temperature increases resulting from the increase in electron–phonon coupling [32]. However, due to the large energy separation between ⁵D₄ and ⁷F₆ states and the required number of phonons to bridge this gap, the thermally-induced enhancement of multiphonon nonradiative processes is not sufficiently efficient to explain the observed thermal dependence. Therefore, as it was proved in the course of our previous studies [32], the temperature-dependent change of the absorption cross section is the dominant process responsible for the lowering of the ⁵D₄ → ⁷F_J emission of Tb³⁺ ions. On the other hand, when an excitation wavelength was used that was not in resonance with the absorption from the ground state of Tb³⁺, a peculiar broadband emission was found (Figure 4b). The intensity of this emission band cut-off by the used shortpass optical filter increases at elevated temperatures. To verify the source of this broadband emission, a similar measurement was performed on the Y₂O₃ microcrystalline powder. The performed studies indicate that the similar shape of emission bands was noticed in this case, however of significantly lower intensity (Figure S6). This excluded the contribution of Tb³⁺ ions in the generation of broadband emission. As reported previously, RE oxides exhibit intrinsic luminescent properties associated with optically active defects [25,26]. Huang et al. [33] reported that, in the case of nanocrystalline Y₂O₃, the peak observed in the X-ray photoelectron spectroscopy (XPS) corresponding to binding energy of O1s was significantly broader with respect to those observed for their microcrystalline counterpart. This clearly indicated the presence of different chemical states of O1s ions in the Y₂O₃ nanocrystals related to the defect states. Broad defect emission was previously reported for other oxides i.e., ZnO, for which a heterogeneously broadened band ranging from 400 to 600 nm was reported by Karthikeyan et al. [34]. Foch et al. reported that the position of the defects band maximum in Si+ implemented SiO₂ can be altered by the preparation condition. The enhancement of the defect emission intensity in the Y₂O₃ and Lu₂O₃ nanocrystals with respect to the microcrystalline counterpart may suggest that mainly surface defects are involved in this process. Interestingly the emission intensity of defect states increases with temperature. This is probably associated with the fact that excitation wavelength reaches the sideband of the defects absorption band and, at an elevated temperature, the broadening of the absorption band leads to the more efficient pumping of defect emission. Confirmation of this hypothesis is the fact that in the case of the Y₂O₃ microcrystals, for which the presence of much narrower absorption band is expected, defect emission can be observed only above 140 °C. The abovementioned emission characteristics for each of the used laser lines were tested for both RE oxide matrices and for different concentrations of Tb³⁺. The Tb³⁺ emission intensity obtained upon λ_exc = 266 nm was integrated in the range corresponding to the ⁵D₄ → ⁷F₆ band and its temperature dependence was depicted in Figure 4c. There is a gradual decrease in the intensity of emission when the temperature increases from 20 to 200 °C: for Y₂O₃, a decrease to about 60% of the initial value was recorded, while for Lu₂O₃, this decrease is greater, to about 40–50%, which is related to the crystal field strength. On the other hand, upon λ_exc = 543 nm, significantly, over 5-fold enhancement of defect emission intensity at elevated temperature was observed (Figure 4d), while for the Y₂O₃:0.05%Tb³⁺ nanocrystals even 10-fold enhancement of emission intensity was found. The opposite thermal monotonicity of two signals: (1) luminescence of Tb³⁺ ions and (2) defect emission, which occur in the same spectral range upon different λ_exc, enables development of the SBR luminescent thermometers based on Y₂O₃:Tb³⁺ and Lu₂O₃:Tb³⁺ nanocrystals. In order to verify the performance of nanocrystals under investigation to noncontact temperature sensing, the ratio of integral luminescence intensities (LIR) upon λ_exc = 266 nm and λ_exc = 543 nm calculated in the same spectral range was determined:

$$LIR=\frac{I\left[T{b}^{3+}\right]}{I\left[defects\right]}=\frac{{\displaystyle \underset{400nm}{\overset{500nm}{\int }}I{(T)}_{\lambda exc=266nm}d\lambda }}{{\displaystyle \underset{400nm}{\overset{500nm}{\int }}I{(T)}_{\lambda exc=543nm}d\lambda }}$$

(1)

As shown in Figure 4e, a 12-fold enhancement of the LIR for Y₂O₃:Tb³⁺ nanocrystals can be observed in the 40–200 °C temperature range. In the case of the Lu₂O₃:Tb³⁺ nanocrystals observed, LIR’s thermal enhancement was only slightly lower due to the fact that although the thermally induced defect emission intensity was significantly lower with respect to the Y₂O₃:Tb³⁺ counterpart, the Tb³⁺ emission intensity was quenched faster in this case. The quantitative analysis of the thermally induced LIR’s changes is provided by the calculation of the relative sensitivity (S_r) of SBR LT to temperature changes defined as follows:

$$S_r=\frac{1}{LIR}\frac{\Delta LIR}{\Delta T}\cdot 100\%$$

(2)

where ΔLIR represents the change of LIR corresponding to ΔT change of temperature. The maps presenting the thermal dependence of S_r for different dopant concentration shown in Figure 4g indicate that in the case of Y₂O₃:Tb³⁺ nanocrystals, there is no correlation between dopant concentration and the S_r. The highest S_r values were found in the 60–100 °C temperature range, for which with the maximum S_r = 4.92%/°C at 40 °C for Y₂O₃:0.05%Tb³⁺. On the other hand, in the case of Lu₂O₃:Tb³⁺ nanocrystals, the increase of dopant concentration leads to the gradual enhancement of the S_r value. This effect may result from the difference in the ionic radii of Lu³⁺ and Tb³⁺. The increase of the dopant ion concentration of larger ionic radii with respect to the substituted one results in the generation of the larger number of defect states. The analogous difference in the ionic radii in the case of Y₂O₃:Tb³⁺ nanocrystals is only slight.

4. Conclusions

In this work, to the best of our knowledge, for the first-time defect emission of oxide nanoparticles was implemented to noncontact temperature sensing. In order to determine the usefulness for luminescence thermometry, the spectroscopic properties of Y₂O₃:Tb³⁺ and Lu₂O₃:Tb³⁺ nanoparticles were examined in a wide temperature range. In was found that the emission intensity of Tb³⁺ bands associated with the ⁵D₄ → ⁷F_J electronic transition obtained upon λ_exc = 266 nm decreases with temperature. This thermal quenching process was explained in terms of lowered absorption cross section at the excitation wavelength at elevated temperatures. On the other hand, emission intensity of the defect states increases significantly by one order of magnitude in the temperature range under investigation. The comparative studies with the micro sized counterparts reveal the dominant contribution of the surface states in the generation of the emission. The opposite thermal dependence of these two independently excited luminescent signals enabled the development of a single band ratiometric luminescent thermometer. The high thermometric performance of the presented approach was confirmed by the high S_r values as 4.92%/°C for Y₂O₃:0.05%Tb³⁺ and 2%/°C for Lu₂O₃:0.5%Tb³⁺ nanocrystals. Additionally, it was found that the increase of the Tb³⁺ concentration in Lu₂O₃ nanocrystals enhances the S_r values that were discussed in terms of higher probability of the defect creation in this host material associated with the difference in the ionic radii between Lu³⁺ and Tb³⁺ ions. The use of this new type of optically active center opens up a number of new possibilities and the high relative sensitivities achieved indicate that this achievement should not be overlooked and ought to be further developed.