Tailoring Energy Transfer in Mixed Eu/Tb Metal–Organic Frameworks for Ratiometric Temperature Sensing

Abstract

Eu/Tb metal–organic frameworks (Eu/Tb-MOFs), exhibiting Eu³⁺ and Tb³⁺ emissions, stand out as some of the most fascinating luminescent thermometers. As the relative thermal sensitivity model is limited to its lack of precision for fitting ratio of Eu³⁺ and Tb³⁺ emissions, accurately predicting the sensing performance of Eu/Tb-MOFs remains a significant challenge. Herein, we report a series of luminescent Eu/Tb-MOF thermometers, Eu_xTb_1−xL, with excellent thermal sensitivity around physiological levels, achieved through the tuning energy transfer from ligands to Eu³⁺ and Tb³⁺ and between the Ln ions. It was found that the singlet lowest-energy excited state (S₁) of the ligand and the higher triplet energy level (T_n) are crucial in the energy transfer processes of ligand→Tb³⁺ and ligand→Eu³⁺. This enables Eu_xTb_1−xL to serve as an effective platform for exploring the impact of these energy transfer processes on the temperature-sensing properties of luminescent Eu/Tb-MOF thermometers. The relative thermal sensitivity is comparable to that of dual-center MOF-based luminescent thermometers operating at physiological levels. This study provides valuable insights into the design of new Eu/Tb thermometers and the accurate prediction of their sensing performance.

1. Introduction

Precision temperature sensing is vital in various fields, such as large-area temperature mapping, biology, and chemical reaction monitoring [1,2,3,4,5,6,7,8,9,10]. Lanthanide metal–organic frameworks (Ln-MOFs), crystalline porous materials constructed from lanthanide nodes and organic ligands, have garnered significant interest in luminescence thermometry [11,12,13,14,15,16,17,18]. Among the various Ln-MOF-based luminescent thermometers, Eu/Tb-MOFs [19,20,21,22,23,24,25,26,27,28,29,30,31] are frequently used to determine the temperature based on the ratio between the integrated intensities of the ⁵D₄→⁷F₅ and ⁵D₀→⁷F₂ transitions of Tb³⁺ and Eu³⁺, respectively. Due to the diverse organic ligands available, many Eu/Tb-MOFs with high relative thermal sensitivity have been reported, functioning across temperature ranges from cryogenic to physiological. Over the past decade, some efforts have been dedicated to fine-tuning the temperature-sensing properties of Eu/Tb-MOFs. However, the observed thermal-responsive luminescence involves complex energy transfer processes between ligands and Tb³⁺/Eu³⁺ ions, as well as from Tb³⁺ to Eu³⁺. Consequently, a general model with which to precisely predict the temperature-sensing properties of these thermometers has not yet been proposed. As a result, many exceptional Eu/Tb-MOF thermometers are still being discovered serendipitously, and rationalizing their relative thermal sensitivity and operating temperature range remains a significant challenge in the field.

Given that excited-state calculations of Eu/Tb-MOFs are time-consuming, the Mott–Seitz model is frequently employed as a rapid experimental approach to rationalize the thermal quenching of luminescence in these materials [32,33,34]. However, a significant limitation of the Mott–Seitz model is its lack of precision when used to fit the ratio between the integrated intensities of Tb³⁺ and Eu³⁺ emissions in Eu/Tb materials. This shortcoming is often attributed to the empirical observation [35,36,37,38,39,40] that Tb³⁺ emission is influenced by the energy difference (ΔE_T) between the triplet lowest-energy excited state (T₁) of the ligand and the lowest emitting levels of Tb³⁺. Conversely, the model assumes that the Eu³⁺ transition is unaffected by temperature variations. Consequently, the integrated intensity ratio is predominantly determined by ΔE_T. However, this prediction is not consistently observed in practical applications. In addition to T₁, higher energy levels of ligands, such as the singlet lowest-energy excited state (S₁), may also play a crucial role in the energy transfer processes between the ligand and Tb³⁺/Eu³⁺ in certain Eu/Tb-MOFs. Moreover, recent findings indicate that Eu³⁺ content is another significant structural factor. Therefore, considerable efforts are still needed to explore the contributions of energy transfers between the higher energy levels of ligands and Tb³⁺/Eu³⁺, as well as the impact of Eu³⁺ content on the temperature-sensing performance of Eu/Tb-MOF thermometers.

Herein, we report the investigation of TbL and Eu_xTb_1−xL (x = 0.0001, 0.0005, and 0.001). Aiming to bridge the gap between the composition of Eu/Tb-MOFs and their desired thermometric performance, we examined the temperature-dependent luminescence of TbL and Eu_xTb_1−xL. This study demonstrates that energy transfer likely occurs from S₁ and higher triplet energy levels (T_n) of the ligand to Tb³⁺ or Eu³⁺ and establishes how to optimize the thermometric performance of Eu_xTb_1−xL by selecting appropriate ligands and adjusting Eu³⁺ content. The results offer valuable insights for further improvements in the rational design of Eu/Tb-MOF luminescent thermometers, enabling their effective operation across temperature ranges from cryogenic to physiological levels.

2. Results and Discussion

2.1. Structural Properties of TbL and Eu_xTb_1−xL

In recent years, we have reported the structural and luminescence properties of a Tb-MOF and a series of Eu/Tb-MOFs based on a tetracarboxylic acid ([1,1:4,1-terphenyl]-2,4,4,5-tetracarboxylic acid, H₄L) [41], referred to as TbL and Eu_xTb_1−xL, respectively. Despite the triplet excited-state energy of H₄L (20,661 cm⁻¹) being very close to the ⁵D₄ energy level of Tb³⁺ (20,500 cm⁻¹), TbL emits bright green light originating from Tb³⁺ upon UV excitation. For Eu_xTb_1−xL, both the ⁵D₄→⁷F₅ transition of Tb³⁺ and the ⁵D₀→⁷F₂ transition of Eu³⁺ are observed in the emission spectra. These results make Eu_xTb_1−xL an excellent platform to investigate the effects of S₁, T_n, and Eu³⁺ content on temperature-sensing properties.

TbL and Eu_xTb_1−xL are in soluble in water and common organic solvents such as methanol and ethanol. Figure 1 shows the coordination environment of Tb³⁺ ions, L⁴⁻ ligands, and supramolecular interactions within the crystal structure of TbL. The asymmetric unit of TbL includes two crystallographically unique Tb³⁺ ions (Tb1 and Tb2), two L⁴⁻ ligands (one of which is on the inversion center), and two coordinated NMP molecules. As shown in Figure 1a,b, there exist μ₇ and μ₈ coordination modes for the ligand. Two Tb1 and two Tb2 are linked by 12 carboxylates and 4 NMP to form the Tb₄(COO)₁₂·4NMP cluster (Figure 1c). Importantly, a C-H···π hydrogen bond with a 2.456 Å H···Cg length (C38-H38A→Cg8) was observed between the ligand on the inversion center and the NMP molecule coordinated with Tb1 (Figure 1d). In contrast, π-π stacking was not observed. This result indicates that Tb³⁺ emission is sensitized by excited ligand monomers. The structures of TbL and Eu_xTb_1−xL were confirmed by powder X-ray diffraction (PXRD) (Figure S1), thermogravimetric analysis (TGA) (Figure S2), and Fourier transform infrared (FT-IR) spectroscopy (Figure S3). For TbL and Eu_xTb_1−xL, TGA curves were very similar in the temperature range from room temperature to 723 K. At higher temperatures, the TGA curves of two compounds with x = 0.0001 and 0.0005 were different from the other ones. This is probably due to complex phase changes in combustion of the ligand molecules in the temperature range. The characteristic IR absorption bands of amide C=O vibration and sp³ C−H vibration in NMP are observed around 1670 and 2890 cm⁻¹, respectively. The IR spectra difference, near 3500 cm⁻¹, between the compounds resulted from adsorbed water on the surface of particles. These results demonstrate that the samples of TbL and Eu_xTb_1−xL (x = 0.0001, 0.0005, and 0.001) are essentially pure phases.

2.2. Photoluminescence Properties of TbL and Eu_xTb_1−xL

As reported in prior research, the ΔE_T of TbL is only 161 cm⁻¹, which contradicts the empirical rule. To investigate the excited states of the ligand, we examined the luminescence properties of the ligand. Concentration-dependent 3D photoluminescence (PL) spectra of H₄L were recorded at room temperature and identified as ligand monomer emissions. As depicted in Figure 2a–d, the maximum excitation wavelengths of the H₄L ligand in solutions shift from 318 nm to 353 nm as the concentrations increase from 10⁻⁴ to 10⁻² M, while the emission peaks remain at 387 nm. The S₁ state of H₄L molecules is thus determined to be 27,400 cm⁻¹. For solid H₄L, the emission maximum redshifts to 442 nm and the excitation maximum to 392 nm, likely due to molecular packing effects.

Next, phosphorescence spectra of the ligand were recorded at 77 K in the solid state upon excitation from 260 nm to 365 nm (Figure 2e,f). The band at 526 nm was assigned to T₁, and the weak band around 420 nm, ~23,800 cm⁻¹, was attributed to the ligand’s T_n state. The energy difference between the T_n state and the emitting level ⁵D₄ is approximately 3300 cm⁻¹, suggesting that the T_n→⁵D₄ energy transfer might also contribute to the observed bright TbL emission. Therefore, we speculate that TbL emission may result from S₁ and higher triplet energy levels of the ligand. A recent example of S₁→Ln³⁺ energy transfer was reported by Jérôme Long and Luís D. Carlos et al., in which the intramolecular energy transfer rates of [Ln(bpy)₂(NO₃)₃] (bpy—2,2′-bipyridine; Ln—Tb or Eu) [26] were determined. At temperatures above 125 K, the S₁→Eu³⁺ channel dominates the sensitization of the ⁵D₀ level.

2.3. Temperature-Dependent Photoluminescence Properties of TbL and Eu_xTb_1−xL

To rationalize the energy diagram of this system, we consider the energy transfer from the ligand to Tb³⁺ and the backward energy transfer from Tb³⁺ to the ligand in relation to the temperature-dependent ⁵D₄ emissions and lifetime. Upon excitation at 351 nm, TbL shows typical emissions around 488, 542, 583, and 621 nm, corresponding to the ⁵D₄→⁷F_6–3 transitions (Figure 3a). The relative intensities of the ⁵D₄→⁷F₅ (I_Tb) transitions were quantified by integrating the emission spectra between 530 and 570 nm. The I_Tb exhibited negative thermal quenching in the range of 77–225 K, with a 63% increase, and mild thermal quenching in the range of 22–353 K, with an ≈42% decrease (Figure 3b). The temperature sensitivity of TbL is much lower than that of (Me₂NH₂)₃[Ln₃(FDC)₄(NO₃)₄]·4H₂O (H₂FDC = 9-fluorenone-2,7-dicarboxylic acid) [36], with a prior sample showing low ΔE_T, suggesting a low backward energy transfer rate for TbL. As the first ΔE_T-driven single-lanthanide organic framework ratiometric luminescent thermometer, (Me₂NH₂)₃[Ln₃(FDC)₄(NO₃)₄]·4H₂O showed extreme Eu³⁺ emission thermal quenching with increasing temperature. This result aligns with the small energy difference (553 cm⁻¹) between the H₂FDC triplet excited state (17,794 cm⁻¹) and the ⁵D₀ Eu³⁺ level (17,241 cm⁻¹), indicating a strong thermally activated ion-to-ligand backward energy transfer. We obtained the activation energies for the nonradiative channels of (Me₂NH₂)₃[Ln₃(FDC)₄(NO₃)₄]·4H₂O and found that the values match the experimentally observed energy difference, proving that ion-to-ligand backward energy transfer is the dominant pathway for Eu³⁺ emission thermal quenching. These results suggest that the empirical observations overestimate the contribution of T₁ level to energy transfer between ligands and Ln³⁺ in certain Ln-MOFs.

Figure 3c shows the temperature-dependent ⁵D₄ decay curves of TbL, which can only be accurately represented by biexponential decay functions, possibly due to the presence of two distinct Tb³⁺ local sites. The determined lifetimes range from 1118 to 198 μs between 77 and 353 K. As it is beyond the scope of the present paper to solve the energy transfer rate and backward energy transfer rate between the S₁, T_n, T₁, and ⁵D₄ levels, we adopted the empirical Mott–Seitz model here.

$$\tau (T)={\displaystyle \frac{{\tau }_0}{1+\alpha ·\mathrm{e}\mathrm{x}\mathrm{p}\left(\frac{-∆E}{k_{B}T}\right)}}$$

(1)

We used the empirical Mott–Seitz model to fit the temperature-dependent ⁵D₄ lifetimes (Figure 3d) in the 77–353 K range with R² = 0.98. Here, τ₀ represents the lifetimes of the Tb³⁺ local sites at T = 0 K, α is the ratio between the nonradiative (T = 0 K) and radiative rates, ΔE represents the activation energy for the nonradiative channels of Tb³⁺, k_B is the Boltzmann constant, and T is the absolute temperature. τ₀, α, and ΔE are determined to be 1115 μs, 36.8, and ≈500–600 cm⁻¹, respectively. Comparing the energy barrier value extracted from the phosphorescence spectrum (161 cm⁻¹) with that resulting from the Mott–Seitz analysis, we observed a factor of ≈3 between them. This discrepancy might have arisen because of (i) the presence of forward and backward energy transfer between the S₁ or higher triplet energy levels of the ligand and Tb³⁺ in TbL and (ii) the fact that the Mott–Seitz model generally overestimates the energy barrier value.

According to the results, it is reasonable to assume that the S₁ and T_n are involved in the luminescence of Eu_xTb_1−xL as much as the T₁ state is. Thus, exploring Eu_xTb_1−xL could provide a good platform to construct luminescent thermometers with tunable relative thermal sensitivity and operating temperature ranges and to study the contributions of multiple energy transfers and Eu³⁺ content to the temperature-sensing performance of Eu/Tb-MOF thermometers. The room temperature photoluminescence (PL) excitation spectra of Eu_xTb_1−xL were monitored at the ⁵D₄→⁷F₅ (Tb³⁺) transition (Figure S3a, Supporting Information), and all excitation spectra were dominated by a broad band around 340 nm, attributed to the ligand’s singlet excited state. The emission spectra of Eu_xTb_1−xL (77–353 K) consisted of characteristic ⁵D₄→⁷F_6–3 transitions of Tb³⁺ and ⁵D₀→⁷F₀₋₄ transitions of Eu³⁺ (see Figure S3b). The relative intensities of the ⁵D₄→⁷F₅ (I_Tb) and ⁵D₀→⁷F₂ (I_Eu) transitions were quantified by integrating the emission spectra between 530–570 nm and 605–640 nm, respectively. To determine absolute temperature, I_Tb serves as the temperature probe while I_Eu acts as the reference due to their strong PL intensities, thus enabling precise temperature determination.

For the Eu_xTb_1−xL samples, while the Tb³⁺ emissions followed typical Mott–Seitz model curves across the entire temperature range, the Eu³⁺ emissions displayed very distinct temperature dependence. For the sample with x = 0.0001 (Figure 4a,b), the Tb³⁺ and Eu³⁺ emissions remained relatively constant in the range of 77–175 K, followed by decreases of approximately 92% and 75% up to 353 K, respectively. For the sample with higher Eu³⁺ content (x = 0.0005), the Tb³⁺ emissions underwent typical thermal quenching, decreasing by approximately 88% up to 353 K. Meanwhile, the Eu³⁺ emissions displayed negative thermal quenching in the range of 77–253 K (Figure 4c,d), with an increase of approximately 26% followed by a decrease of approximately 23% up to 353 K. With increasing Eu³⁺ content (x = 0.001), the Eu³⁺ emissions showed stronger negative thermal quenching, rising by ≈270% up to 353 K (Figure 4e,f), while the Tb³⁺ emissions decreased by ≈90% up to 353 K. This indicates that the thermal sensitivity of Eu³⁺ emissions progressively increases with rising Eu³⁺ content, making Eu_xTb_1−xL excellent candidates for ratiometric thermometers with tunable thermometric performances.

Here, the ratio of ⁵D₄→⁷F₅ transition to I_Eu was used to define the thermometric parameter Δ = I_Tb/I_Eu. We note that while I_Tb corresponds only to the ⁵D₄→⁷F₅ transition, the I_Eu integration range includes a small contribution from the Tb^{3+ 5}D₀→⁷F₃ transition. Nevertheless, all the thermometric parameters of the Eu_xTb_1−xL samples were fitted to an empirical sigmoidal Boltzmann function,

$$∆={\displaystyle \frac{I\left(\mathrm{T}\mathrm{b}\right)}{I\left(\mathrm{E}\mathrm{u}\right)}}={\displaystyle \frac{A_1-A_2}{1+e^{\frac{T-T_0}{-dT}}}}+A_2$$

(2)

where I(Tb) is the emission intensity in the wavelength range of 530–570 nm, I(Eu) is the emission intensity in the wavelength range of 605–640 nm, A₁ represents the maximum emission intensity and A₂ represents the minimum emission intensity, T₀ is the temperature at which the emission intensity reaches half of A₁, and T is the absolute temperature. The fit results show R² > 0.999, implying that the thermometer performance was unaffected by the presence of the ⁵D₀→⁷F₃ transition. This holds true for all the Eu³⁺ contents studied in this study. Therefore, in the following analysis, we considered the commonly assumed labeling of I_Eu as being solely due to the ⁵D₀→⁷F₂ contribution. Figure 5a displays the calibration curves of Eu_xTb_1−xL samples within the temperature range of 77–353 K under 335 nm excitation. The fitting parameters for the Eu_xTb_1−xL compounds are listed in

Table 1. Fitting results of Δ (T) to Equation (2).

Items	A1	A2	T0	R2
Eu0.0001Tb0.9999L	6.70	1.20	278.00	0.9990
Eu0.005Tb0.999 L	6.80	0.42	240.00	0.9993
Eu0.001Tb0.999L	6.40	0.10	183.00	0.9997

and show a nearly constant Δ₀ and an increase in T₀ as the Eu³⁺ content increased. By increasing the amount of Eu³⁺ from x = 0.0001 to 0.001, a shift of 95 K toward lower temperatures was observed in T₀.

The relative sensitivity was employed to compare the performance of this model with those of similar thermometers reported in previous studies, defined as

$$S_r={\displaystyle \frac{1}{∆}}·\left|{\displaystyle \frac{\partial ∆}{\partial T}}\right|$$

(3)

The maximum relative temperature sensitivities of the thermometers were 0.86% K⁻¹ at 333 K, 1.42% K⁻¹ at 293 K, and 2.11% K⁻¹ at 273 K for, respectively, the samples with x from 0.0001 to 0.001 (Figure 5b). Intriguingly, in the physiological range, the relative sensitivities were still above 1.3% K⁻¹ for the sample with x = 0.0005 and 1.5% K⁻¹ for the sample with x = 0.001, while the corresponding temperature uncertainties were estimated from

$$\delta T={\displaystyle \frac{1}{S_r}}·{\displaystyle \frac{\delta ∆}{∆}}$$

(4)

yielding the values of 0.08 K for the sample with x = 0.0005 and 0.07 K for the sample with x = 0.001 at 313 K (Figure 5c). This variation in the S_r is partially explained by the increase in the nonradiative decay rate of the Tb^{3+ 5}D₄→⁷F₅ transition (relatively to the radiative one) with the increase in Eu³⁺ doping.

2.4. Energy Transfer

To rationalize the thermal dependence of the luminescence of the materials, we further analyzed the energy transfer from Tb³⁺ ions to Eu³⁺ ions using the sample with x = 0.001 as an example. Generally, the long distances between these ions may result in low rates that cannot compete with the ligand-to-Ln³⁺ rates, which are orders of magnitude higher. The thermal dependence of the Tb³⁺ and Eu³⁺ lifetimes was investigated by monitoring the emission decay curves of the ⁵D₄→⁷F₅ and ⁵D₀→⁷F₂ transitions (Figure 5d,e), respectively. All the decay curves are well modeled and show that the ⁵D₀ lifetime remains stable up to 1289 μs at 175 K, slightly decreasing to 1104 μs at 353 K. As the temperature rose from 77 K to 353 K, the ⁵D₄ lifetime progressively decreased from 1239 μs to 127 μs, remaining nearly independent of the Eu³⁺ doping until T > 175 K (Figure 5f). Notably, a rise time, dependent on the temperature, occurred in the ⁵D₀ emission decay curves in the temperature range from 175 K to 353 K (Figure 5e). This rise time was also observed in [Ln(bpy)₂(NO₃)₃] by Jérôme Long and Luís D. Carlos et al. [26] and was found to be similar to the ⁵D₄ lifetime. This implies that the ⁵D₄ level could support Tb³⁺-Eu³⁺ energy transfer within Eu_0.001Tb_0.999L above 175 K.

3. Materials and Methods

3.1. Materials and Characterization

All chemicals used in this work were commercially available and used without further purification. The X-ray powder diffraction (PXRD) patterns were collected using a D/MAX 2500/PC powder diffractometer (Rigaku, Tokyo, Japan) equipped with a Cu Kα radiation source, covering a 2θ range of 5–50°. The FT-IR spectra of samples embedded in KBr pellets were recorded using a PerkinElmer FT-IR spectrometer. Thermogravimetric analysis (TGA) was conducted on a TG/DTA 6300 thermal gravimetric analyzer (Hitachi, Tokyo, Japan) at a constant rate of 10 K/min. Room temperature luminescence spectra were obtained using an FS5 steady-state transient fluorescence spectrometer with a 150 W CW ozone-free xenon lamp. Temperature-dependent luminescence spectra were collected using an FLS1000 photoluminescence spectrometer (Edinburgh Instruments Ltd., Livingston, UK) with a 300 W CW ozone-free xenon lamp.

3.2. Preparation of [Tb₂L_1.5(NMP)₂]_n (TbL)

A solution of H₄L was prepared by dissolving H₄L (0.2 mmol, 81.2 mg) in 24 mL of N-methyl-2-pyrrolidone (NMP), and a Tb(NO₃)₃ solution was prepared by dissolving Tb(NO₃)₃·6H₂O (0.4 mmol, 180.5 mg) in 24 mL of H₂O. These two solutions were then mixed in a 100 mL Teflon-lined stainless-steel autoclave, followed by the addition of 200 μL of concentrated hydrochloric acid. The mixture was sealed and maintained at 433 K for 3 days. After the autoclave was gradually cooled to room temperature, colorless rhombic crystals of TbL (Yield: 65%) were obtained by filtration after washing with ethanol three times.

3.3. Preparation of Eu_xTb_1−xL (x = 0.0001, 0.0005, and 0.001)

Eu_xTb_1−xL (x = 0.0001, 0.0005, and 0.001) samples were prepared using a method similar to that used for TbL, with the exception that the Tb(NO₃)₃·6H₂O solution was replaced with a mixture of Eu(NO₃)₃·6H₂O (x⋅24 mL) and Tb(NO₃)₃·6H₂O (24(1−x) mL) solutions.

4. Conclusions

This study offers a thorough analysis and explanation of the luminescence mechanisms of TbL and the fine-tuning of the temperature sensitivity of Eu_xTb_1−xL thermometers. Different from the luminescence mechanisms of similar mixed Eu/Tb-MOFs, the S₁ and higher triplet energy levels of the ligand are involved in the sensitization of Tb³⁺ emission. Therefore, the judicious choice of ligand and Eu³⁺ content makes Eu_xTb_1−xL excellent thermometers, with a relative thermal sensitivity comparable to those of similar mixed Eu/Tb-MOF-based luminescent thermometers operating across a temperature range from cryogenic to physiological. Additionally, the analysis of the luminescence lifetimes of the sample with x = 0.001 provides a clear understanding of the temperature dependence of the luminescence spectra. Furthermore, this approach serves as a platform that can be leveraged to guide the design of new Eu/Tb thermometers based on dual-emissive centers. The results enable lots of neglected similar materials to show luminescent temperature-sensing properties. Notably, since the precise prediction of multiple-energy-level-involved luminescence is complex within mixed Eu/Tb-MOFs, it is still a challenge to design similar materials.