1D-Zigzag Eu3+/Tb3+ Coordination Chains as Luminescent Ratiometric Thermometers Endowed with Multicolor Emission

Two homometallic Coordination Polymers (CPs) with composition [Ln(hfac)3bipy]n (Ln3+ = Eu3+, 1, and Tb3+, 2; hfac = hexafluoroacetylacetonato, bipy = 4,4′-bipyridine) were used to develop a family of ratiometric luminescent thermometers containing Eu3+ and Tb3+ as red and green emitters, respectively. The thermometric properties of pure CPs and of their mixtures having an Eu3+/Tb3+ molar ratio of 1:1, 1:3, 1:5, and 1:10 (samples: Eu1Tb1, Eu1Tb3, Eu1Tb5, and Eu1Tb10) were studied in the 83–383 K temperature range. Irrespective of the chemical composition, we observed similar thermometric responses characterized by broad applicative temperature ranges (from 100 to 165 K wide), and high relative thermal sensitivity values (Sr), up to 2.40% K−1, in the physiological temperature range (298–318 K). All samples showed emissions endowed with peculiar and continuous color variation from green (83 K) to red (383 K) that can be exploited to develop a colorimetric temperature indicator. At fixed temperature, the color of the emitted light can be tuned by varying composition and excitation wavelength.


Introduction
Nowadays, luminescence thermometry plays a relevant role due to its high importance in many societal needs. For example, contactless temperature measurements with a sub-micrometric spatial resolution are required in countless technological applications and industrial fields, such as microelectronics, microoptics, photonics, microfluidics, and nanomedicine (e.g., in intracellular temperature measurements, or in microcircuit thermal map) [1][2][3]. Among the different types of luminescent thermometers, intensity-based ones are particularly useful for many applications such as real-time measurements on large moving systems [4,5]. However, using a single emitter, fluctuations of the light source or changes in the emitter local concentration are potential sources of error in temperature readout [6][7][8].
The stated drawbacks can be easily overcome using the intensity ratio of different emission bands of a given luminescent material, thus creating self-referencing thermometers [6,7,9,10]. These thermometers, also called "ratiometric thermometers", have been developed using a variety of luminescent probes [7,8,11]. Among others, europium and [Ln(hfac) 3 (H 2 O) 2 ] according to the literature [33]. 4,4 -bipyridine (bipy, Aldrich) was used as received. All reactions have been carried out under inert Ar atmosphere using standard Schlenk techniques.
FTIR spectra in solid phase were recorded with a Perkin-Elmer "Spectrum One" spectrometer, equipped with an ATR accessory (Perkin Elmer, Waltham, MA USA). Elemental analyses (C, H, N) were performed at the Dipartimento di Chimica e Chimica Industriale, Università di Pisa (Pisa, Italy). The metal content was determined by treating the samples with diluted HNO 3 in a platinum crucible. The resulting solution was then evaporated to dryness. After calcination at 850 • C, solid residues were weighted as Eu 2 O 3 or Tb 2 O 3 .
Photoluminescence experiments have been performed on KBr pellets (Sigma-Aldrich, spectroscopic grade, Sigma-Aldrich, Burlington, MA, USA) containing pure CPs or their mixtures. KBr has been previously heated at 100 • C overnight to remove H 2 O traces. Room temperature luminescence spectra were recorded in a front-face acquisition geometry with a Horiba JobinYvon Fluorolog-3 spectrofluorimeter (Horiba JobinYvon, Edison, NJ, USA) equipped with double-grating monochromator in both the excitation and emission sides, coupled to a R928P Hamamatsu photomultiplier and a 450 W Xe arc lamp as the excitation source. Emission spectra were corrected for detection and optical spectral response of the spectrofluorimeter supplied by the manufacturer. The excitation spectra were corrected for the spectral distribution of the lamp intensity using a photodiode reference detector.
Temperature-dependent experiments (83-383 K) were carried out in backscattering geometry using a Horiba T64000 triple spectrometer equipped with a Peltier-cooled chargecoupled device detector (Horiba Synapse). A Xe arc lamp (450 W) has been used as excitation source. The scattered radiation was collected through a 10× microscope objective (Olympus MPLAN, 10 × 0.25, Olympus, Tokyo, Japan). The spectrograph, equipped with 300 lines/mm gratings, was used as a single-stage imaging monochromator. Temperaturedependent experiments were performed by means of a Linkam THMS600 heating/freezing microscope (Linkam, Tadworth, UK) stage having temperature stability <0.1 K over the 83-873 K temperature range.

Synthesis and Characterization
In our previous work, we carefully described the synthesis of monodimensional lanthanide β-diketonate coordination polymers based on bipy as connector ligand [33]. The synthesis of these CPs must be carried out in anhydrous toluene, avoiding the use of oxygenated solvents. Indeed, the presence of O-donor molecules in the reaction environment led to the formation of low-nuclearity products due to the higher affinity of lanthanide ions towards oxygenated ligands than to N-donor ones [41][42][43]. We also showed that 1D zigzag CPs were obtained regardless of the employed [Ln(β-dike) 3 ] precursors, using an equimolar amount of bipy per metal center [33]. We also reported the synthesis and the characterization of [Eu(hfac) 3 (bipy)] n (1). Its structure consists of 1D zigzag chains extended along the b crystallographic axis ( Figure 1). Each europium center is coordinated to six oxygen atoms from the three hfac ligands and two N atoms of two different bipy in a distorted square antiprismatic geometry [33]. The terbium CP [Tb(hfac)3(bipy)]n (2) has been similarly synthesized, re [Tb(hfac)3] and bipy in a 1:1 molar ratio, and characterized through elemental analys infrared spectroscopy ( Figure S1). The terbium CP [Tb(hfac) 3 (bipy)] n (2) has been similarly synthesized, reacting [Tb(hfac) 3 ] and bipy in a 1:1 molar ratio, and characterized through elemental analysis and infrared spectroscopy ( Figure S1).

Photoluminescence and Thermometric Studies
Upon irradiation, 1 and 2 emissions are excited over a wide wavelength range, from UV to visible, up to ≈450 nm for 1 and ≈400 nm for 2 ( Figure S2). The lanthanidessensitized emission can be achieved through the so-called "antenna effect" exploiting the absorption properties of the organic ligands. The process can be summarized in the three following steps: (i) absorption of light by the organic ligand and population of its first excited singlet state (S 1 ); (ii) intersystem crossing (ISC) from the singlet to the triplet (T 1 ) level, and (iii) energy transfer (E.T.) from T 1 to the lanthanide excited level [44]. Both the compounds showed intense light emission, also visible at the naked eye ( Figure 2), red for 1 and green for 2. The terbium CP [Tb(hfac)3(bipy)]n (2) has been similarly synthesized, reacting [Tb(hfac)3] and bipy in a 1:1 molar ratio, and characterized through elemental analysis and infrared spectroscopy ( Figure S1).

Photoluminescence and Thermometric Studies
Upon irradiation, 1 and 2 emissions are excited over a wide wavelength range, from UV to visible, up to ≈450 nm for 1 and ≈400 nm for 2 ( Figure S2). The lanthanidessensitized emission can be achieved through the so-called "antenna effect" exploiting the absorption properties of the organic ligands. The process can be summarized in the three following steps: (i) absorption of light by the organic ligand and population of its first excited singlet state (S1); (ii) intersystem crossing (ISC) from the singlet to the triplet (T1) level, and (iii) energy transfer (E.T.) from T1 to the lanthanide excited level [44]. Both the compounds showed intense light emission, also visible at the naked eye ( Figure 2), red for 1 and green for 2. and 2 (c) exposed to UV light (λexc =320 nm) at room temperature.
In particular, the emission spectrum of 1 ( Figure 3a) is characterized by the typical Eu 3+ 5 D 0 → 7 F J (J = 0-4) transitions corresponding to the bands centered at 587, 595, 614, 652, and 693 nm. Conversely, compound 2 emission spectrum ( Figure 3b) showed Tb 3+ 5 D 4 → 7 F J (J = 6-3) transitions centered at 490, 543, 583, and 619 nm, respectively [45]. The spectra ( Figure 3) of 1 and 2, dominated by the 5 D 0 → 7 F 2 (Eu 3+ ) and 5 D 4 → 7 F 5 (Tb 3+ ), have a shape that is typical for β-diketonato complexes [38,46].  The photoluminescence excitation (PLE, see Figure S2) spectrum of predominantly localized at λ < 380 nm with a low-intensity tail extending to the boun with the visible region. Instead, the PLE spectrum of 1 is redshifted and extends i visible up to about 450 nm. Based on the shape of PLE spectra, we decided to study and their mixtures, exciting their luminescence at 320 and 370 nm. While at 320 nm emitters are effectively excited, PLE signals of 1 at 370 nm are more intense than f The photoluminescence excitation (PLE, see Figure S2) spectrum of 2 is predominantly localized at λ < 380 nm with a low-intensity tail extending to the boundary with the visible region. Instead, the PLE spectrum of 1 is redshifted and extends in the visible up to about 450 nm. Based on the shape of PLE spectra, we decided to study 1, 2 and their mixtures, exciting their luminescence at 320 and 370 nm. While at 320 nm both emitters are effectively excited, PLE signals of 1 at 370 nm are more intense than for 2. However, the use of 370 nm excitation wavelength has distinctive applicative and technological advantages compared to 320 nm, being compatible with organic substrates used in flexible electronics and with low glass transition temperature glasses having a cutoff wavelength around 350 nm. Moreover, light-emitting devices (LEDs) emitting at 370 nm are quite common, and generally cheaper than UV LED.
The temperature-dependent luminescence properties of compounds 1 and 2 were studied in the 83-383 K temperature range. As visible from Figure 4c-e, the emission intensity of both compounds decreases as temperature increases. The integrated areas of Eu 3+ 5 D 0 → 7 F 2 and Tb 3+ 5 D 4 → 7 F 5 transitions were chosen as thermometric parameter (∆) for 1 and 2, respectively, since they are the main contributions in the total emission spectra. The ∆ curves of 1 and 2, normalized at 83 K, are reported in Figure 4.  The shape of Δ curves depends on the employed metal ion. In 1 (Figure 4a), the thermometric parameter has small variations up to ≈223 K, and then decreases, keeping 25% ca. of the initial 5 D0→ 7 F2 transition intensity at 383 K (Figure 4c,d). This trend does not depend on the employed excitation wavelength. Conversely, compound 2 showed a The shape of ∆ curves depends on the employed metal ion. In 1 (Figure 4a), the thermometric parameter has small variations up to ≈223 K, and then decreases, keeping 25% ca. of the initial 5 D 0 → 7 F 2 transition intensity at 383 K (Figure 4c,d). This trend does not depend on the employed excitation wavelength. Conversely, compound 2 showed a well-defined S-shaped profile that gradually decreases above ≈113 K. Under the adopted experimental conditions, 2 did not show residual emission intensity at 383 K (Figure 4b,e,f). The emission of terbium-CP is partially quenched at room temperature, although still clearly observable, and gains intensity lowering the temperature. Europium emission is less effected and is very bright from physiological down to cryogenic temperatures (T < 100 K) [27,39]. The shape of ∆ vs. T curves depends on the balance between radiative and non-radiative deactivation pathways that contribute to the modulation of the luminescence properties. The most encountered non-radiative processes determining the overall thermometric response of the luminescent molecular thermometers are: (i) back energy transfer from Ln 3+ to the ligand triplet level, (ii) multiphonon relaxation promoted by high energy oscillators (mainly −OH and −NH groups) directly bonded to the metal center, and (iii) energy transfer to LMCT states. Depending on the activation energies, more than one process can operate. Other deactivation paths can also be encountered but are less common (ion-ion interaction, donor-acceptor phenomena, etc.). The different behavior displayed by 1 and 2 is mainly correlated to the energy difference between Eu 3+ or Tb 3+ emitting levels and antenna triplet level (ca. 4700 and 1500 cm −1 for Eu 3+ and Tb 3+ , respectively) that affect the efficiency of the energy transfer from T 1 to the lanthanide excited level, i.e., from the back energy transfer that results more efficient in 2. A contribution from LMCT states cannot be unambiguously ruled out [44].
For a better comparison of compounds 1 and 2 thermometric properties, we used the relative thermal sensitivity (S r , % K −1 ) [6], a figure of merit commonly adopted to compare thermometers was used: In Equation (1), ∆ is the thermometric parameter and δ∆ is the variation of the thermometric parameter on a certain temperature variation (δT). A value of S r ≥ 1 was assumed as quality criterion to determine the applicative temperature range of these CPs [3].
Below 263 K, compound 1 (Figure 5a) has S r close to zero. Instead, S r is >1 above 323 K and its temperature-dependence is almost independent from the excitation wavelengths. Conversely, 2 ( Figure 5b) has S r = 0 in almost all the studied temperature range, with an application range from 243 to 383 K. more than one process can operate. Other deactivation paths can also be encountered but are less common (ion-ion interaction, donor-acceptor phenomena, etc.). The different behavior displayed by 1 and 2 is mainly correlated to the energy difference between Eu 3+ or Tb 3+ emitting levels and antenna triplet level (ca. 4700 and 1500 cm −1 for Eu 3+ and Tb 3+ , respectively) that affect the efficiency of the energy transfer from T1 to the lanthanide excited level, i.e., from the back energy transfer that results more efficient in 2. A contribution from LMCT states cannot be unambiguously ruled out [44]. For a better comparison of compounds 1 and 2 thermometric properties, we used the relative thermal sensitivity (Sr, % K −1 ) [6], a figure of merit commonly adopted to compare thermometers was used: Sr = (1) In Equation (1), Δ is the thermometric parameter and δΔ is the variation of the thermometric parameter on a certain temperature variation (δT). A value of Sr ≥ 1 was assumed as quality criterion to determine the applicative temperature range of these CPs [3].
Below 263 K, compound 1 (Figure 5a) has Sr close to zero. Instead, Sr is >1 above 323 K and its temperature-dependence is almost independent from the excitation wavelengths. Conversely, 2 ( Figure 5b) has Sr ≠ 0 in almost all the studied temperature range, with an application range from 243 to 383 K.
Known amounts of the two homometallic CPs (see Table 1) were mixed in KBr,
Known amounts of the two homometallic CPs (see Table 1) were mixed in KBr, obtaining homogeneous dispersions. The mixtures have been finely grinded in an agate mortar, adding a few drops of n-pentane to facilitate the milling process. The organic solvent was then removed under N 2 flow. Finally, the resulting powders have been pressed, forming non-friable easy-to-handle pellets. Four samples with different europium/terbium molar ratios have been prepared, keeping constant the Ln 3+ /KBr molar ratio (Ln 3+ = Eu 3+ + Tb 3+ ). The samples were labelled Eu1Tb1, Eu1Tb3, Eu1Tb5, and Eu1Tb10 according to the Eu 3+ /Tb 3+ relative amount (1:1, 1:3, 1:5, and 1:10 respectively).
At room temperature, the samples showed a bright homogeneous emission also visible to the naked eye. The color of the emitted light is a combination of the Eu 3+ (red) and Tb 3+ (green) emissions, which varies from orange to pale green, increasing the terbium content in the mixture (see Figure 6, Figure 7a,b, Figures S3 and S4).
Switching the excitation wavelength from 320 nm to 370 nm, we modified the relative intensity of Eu 3+ and Tb 3+ emissions (compare Figure 7a,b, Figures S3 and S4), since the longer wavelengths favor europium emission with respect to the terbium one (see Figure S2). To check the sample homogeneity, we measured the emission spectra (λ exc = 320 nm and 370 nm) on different spots on each pellet (sampling size = 200 µm). We observed small differences in the I Tb /I Eu intensity ratio, within 10%, thus confirming a homogeneous dispersion of the two CPs inside the specimens.
The temperature-dependent luminescent properties of the mixed samples were studied in the 83-383 K range. As an example, emission spectra of Eu1Tb1 at 320 nm and 370 nm recorded at different temperatures are reported in Figure 7c,d (see also Figures S5 and S6 for the other samples).
Noteworthy, in all the samples the intensity of terbium 5 D 4 → 7 F 5 transition is strongly influenced by temperature variations, while the europium 5 D 0 → 7 F 2 band is less affected (Figure 7c,d, Figures S5 and S6), as also previously observed in the homometallic CPs 1 and 2. The color of the emitted light changes from green/yellow to orange/red depending on the sample composition and the excitation wavelength going from low to high temperature.
For all samples, ∆ curves have a well-defined S-shape. Irrespective of the excitation wavelength, the higher the terbium content in the sample, the higher the ∆ value at 83 K ( Figure S7). To get qualitative information on the thermometric response among the mixed samples, the ∆ curves were normalized at 83 K (Figure 8a,b).
In the mixed samples, the normalized ∆ curves showed similar S-shapes regardless of the relative europium/terbium amount and excitation wavelength. As a result, the S r vs. T curves, and hence the thermometric properties, of the different samples are quite similar (Figure 8c,d). All the S r curves showed an asymmetric bell shape with a S r maximum (S rmax ) of 2.4% K −1 ca. between 303 and 323 K with an extended applicative temperature range (Sr ≥ 1, 100 K ÷ 165 K wide), depending on the sample composition (see Table 2). Interestingly, excellent S r values in the physiological temperature range (298-318 K) [27,47] were found for these ratiometric thermometers. The reproducibility of the thermometric response of the samples has been tested performing three different heating-cooling cycles on each sample. All the studied materials showed a highly reproducible thermometric response as well as the absence of thermal hysteresis (Figure 9). and Tb 3+ (green) emissions, which varies from orange to pale green, increasing the terbium content in the mixture (see Figures 6, 7a,b, S3 and S4). Switching the excitation wavelength from 320 nm to 370 nm, we modified the relative intensity of Eu 3+ and Tb 3+ emissions (compare Figures 7a,b, S3 and S4), since the longer wavelengths favor europium emission with respect to the terbium one (see Figure S2). To check the sample homogeneity, we measured the emission spectra (λexc= 320 nm and 370 nm) on different spots on each pellet (sampling size = 200 µm). We observed small differences in the ITb/IEu intensity ratio, within 10%, thus confirming a homogeneous dispersion of the two CPs inside the specimens.
The temperature-dependent luminescent properties of the mixed samples were studied in the 83-383 K range. As an example, emission spectra of Eu1Tb1 at 320 nm and 370 nm recorded at different temperatures are reported in Figure 7c  Noteworthy, in all the samples the intensity of terbium 5 D4→ 7 F5 transition is strongly influenced by temperature variations, while the europium 5 D0→ 7 F2 band is less affected (Figures 7c,d, S5 and S6), as also previously observed in the homometallic CPs 1 and 2. The color of the emitted light changes from green/yellow to orange/red depending on the sample composition and the excitation wavelength going from low to high temperature.
For all samples, Δ curves have a well-defined S-shape. Irrespective of the excitation wavelength, the higher the terbium content in the sample, the higher the Δ value at 83 K ( Figure S7). To get qualitative information on the thermometric response among the mixed samples, the Δ curves were normalized at 83 K (Figure 8a,b). In the mixed samples, the normalized Δ curves showed similar S-shapes regardless of the relative europium/terbium amount and excitation wavelength. As a result, the Sr vs. T curves, and hence the thermometric properties, of the different samples are quite similar (Figure 8c,d). All the Sr curves showed an asymmetric bell shape with a Sr maximum (Srmax) of 2.4% K −1 ca. between 303 and 323 K with an extended applicative temperature range (Sr ≥ 1, 100 K ÷ 165 K wide), depending on the sample composition (see Table 2).
Interestingly, excellent Sr values in the physiological temperature range (298-318 K) [27,47] were found for these ratiometric thermometers. The reproducibility of the thermometric response of the samples has been tested performing three different heating-cooling cycles on each sample. All the studied materials showed a highly reproducible thermometric response as well as the absence of thermal hysteresis (Figure 9). The temperature resolution of our thermometers, i.e., the smallest detectable temperature variation (δT), was determined by the heating/freezing cycles, according to literature [2,3,18], using the following equation: (2) The temperature resolution of our thermometers, i.e., the smallest detectable temperature variation (δT), was determined by the heating/freezing cycles, according to literature [2,3,18], using the following equation: where δ∆ is the ∆ uncertainty at a given temperature [3,6] determined by calculating the standard deviation of three heating/freezing cycles. All the samples showed comparable resolution in the order of 0.01 K (average value over the thermometer operational range), in line with literature values for Ln-based luminescence thermometers [2,3]. The change of the excitation wavelength produces a variation of the samples color outputs. Indeed, at 370 nm, the green component due to terbium emission is less intense than at 320 nm, thus producing an overall chromatic shift towards red ( Figure 10).

Conclusions
In this work, the temperature-dependent luminescent properties of the two homometallic 1D zigzag [Ln(hfac)3bipy]n CPs (Ln 3+ = Eu 3+ ,1, and Tb 3+ , 2) were studied in the 83-383 K temperature range. 1 and 2 showed Sr values higher than 1 starting from 323 K and 223 K, respectively. The two homometallic compounds were then used to prepare a family of Eu 3+ /Tb 3+ ratiometric thermometers in which terbium acts as probe and europium as reference. Four different thermometers were prepared by mixing different amounts of 1 and 2 in KBr matrix with Eu 3+ /Tb 3+ molar ratio of 1:1, 1:3, 1:5, and 1:10, respectively. Sample composition and the excitation wavelength influenced the color of the emitted light but not the thermometric response. All the thermometers were sensitive to temperature variation between 83-383 K, showing a wide applicative range (Sr ≥ 1) from 100 K to 165 K depending on the sample composition. The samples showed high relative thermal sensitivity values in the physiological temperature range (298-318 K), with a Sr maximum value of about 2.40% K −1 . The thermometric response of the samples did not vary significantly in changing the excitation wavelength that, instead, influences the color of the emitted light. The well-defined and peculiar color variation of heterobimetallic materials is used as a colorimetric thermometer. Using the eyes as a detector, temperature difference in the order of 10-20 K can be appreciated. To improve the temperature resolution, using the color of the emitted light as probe, calibrated luminance/color meters  Figure 10 shows a schematic diagram representing the color variation of the different samples over the 83-383 K temperature range. The color at the different temperatures is defined by the CIE coordinates determined from the emission spectra obtained exciting the samples at 320 and 370 nm. Note how the temperature can be mapped by a chromatic reading generating something similar to the litmus paper, the famous pH testing papers, that can work with one or more chromatic channels simply defined by the composition of the mixture of 1 and 2. Noteworthy, the correlation between the color of the emitted light and sample composition can also be used for the development of luminescent QRcodes [24] in which the temperature can be associated to a single color or a combination of two or more tints. In principle, the tuning of the europium/terbium relative amount should enable the optimization of the color gradient, thus allowing the detection of even narrower temperature intervals.

Conclusions
In this work, the temperature-dependent luminescent properties of the two homometallic 1D zigzag [Ln(hfac) 3 bipy] n CPs (Ln 3+ = Eu 3+ , 1, and Tb 3+ , 2) were studied in the 83-383 K temperature range. 1 and 2 showed S r values higher than 1 starting from 323 K and 223 K, respectively. The two homometallic compounds were then used to prepare a family of Eu 3+ /Tb 3+ ratiometric thermometers in which terbium acts as probe and europium as reference. Four different thermometers were prepared by mixing different amounts of 1 and 2 in KBr matrix with Eu 3+ /Tb 3+ molar ratio of 1:1, 1:3, 1:5, and 1:10, respectively. Sample composition and the excitation wavelength influenced the color of the emitted light but not the thermometric response. All the thermometers were sensitive to temperature variation between 83-383 K, showing a wide applicative range (S r ≥ 1) from 100 K to 165 K depending on the sample composition. The samples showed high relative thermal sensitivity values in the physiological temperature range (298-318 K), with a S r maximum value of about 2.40% K −1 . The thermometric response of the samples did not vary significantly in changing the excitation wavelength that, instead, influences the color of the emitted light. The well-defined and peculiar color variation of heterobimetallic materials is used as a colorimetric thermometer. Using the eyes as a detector, temperature difference in the order of 10-20 K can be appreciated. To improve the temperature resolution, using the color of the emitted light as probe, calibrated luminance/color meters are required. The results here reported pose the basis for future activity oriented to bring luminescent molecular thermometer into daily life by measuring temperature by reading the color of the emitted light with a smartphone camera.