Thermally Activated Photophysical Processes of Organolanthanide Complexes in Solution.

The effect of temperature upon the lanthanide luminescence lifetime and intensity has been investigated in toluene solution for the complexes LnPhen(TTA)3 (Ln = Eu, Sm, Nd, Yb; Phen = 1,10-phenanthroline; TTA = thenoyltrifluoroacetonate). Thermally excited back-transfer to a charge transfer state was found to occur for Ln = Eu and can be explained by lifetime and intensity back-transfer models. The emission intensity and lifetime were also quenched with increasing temperature for Ln = Sm, and the activation energy for nonradiative decay is similar to that for the thermal population of Sm3+ excited states. Unusual behavior for lifetime and intensity was found for both Ln = Nd, Yb. The usually assumed equivalence of τ/τ0 = I/I0 (where τ is lifetime and I is intensity) does not hold for these cases. We infer that for these lanthanide systems the intensity decreases with temperature in the stage prior to population of the luminescent state. The lifetime changes are discussed.

For many decades, the unique sharp emission bands and long lifetimes of trivalent lanthanide ions (Ln3+) have received attention in diverse fields of application,[1] including light emitting diodes (LEDs)[2,3] and imaging probes for the biomedical field.[4,5] More recently, the use of Ln3+ ions in thermal sensing, employing homonuclear[6,7] or heteronuclear[8−10] complexes, has been a burgeoning field of study. To further this application, the mechanism of thermal quenching needs to be thoroughly investigated and understood. This present work utilizes different Ln3+ ions in the same organometallic complex in order to understand and compare the temperature quenching mechanisms.

In organolanthanide complexes, the sensitization of Ln3+ is often taken to follow the S0 → S1 → T1 → Ln3+ pathway,[11,12] although other mechanisms have been put forward. Emission from a lanthanide ion is observed when the excited state relaxes with the radiative rate kr and nonradiative rate knr, with kr > knr. The increase of nonradiative rate with temperature may be due to (i) increased vibrational relaxation to a lower state,[13] (ii) thermally induced energy transfer,[14] or (iii) electron transfer processes.[15] In addition, other processes such as those affecting excitation may change the intensity of emission. It is important to distinguish temperature-dependent and -independent nonradiative losses—whether the quenching occurs before or after the excitation reaches the luminescent state—and this may be possible by comparing emission intensity with lifetime measurements to see if they follow a similar trend.

The emission quenching can be modeled by a simple Arrhenius process with activation energy E to a higher state such as a triplet in the case of Tb3+.[16] Alternatively, a scenario for thermal quenching of luminescence lifetime τ or intensity I due to a single-barrier back-transfer to a triplet state, conduction band,[17] or a charge-transfer (CT) state[18,19] (Figure a) has been modeled by eqs and 2and assuming that τ(T)/τr = I(T)/I0, where τr = 1/kr[20]where Γ0 is the attempt nonradiative rate, k is the Boltzmann constant, and T is the temperature.

Figure 1

(a) Illustration of the thermally activated single-barrier back-energy transfer (BET) using Eu3+ as an example. (b) FTIR spectra of LnPhen(TTA), Ln = Y, La, Nd, Sm, Eu, Gd, and Yb. (c) Optimized structure of SmPhen(TTA) using the PBE0/def2-TZVP level of theory. C: gray; H: white; N: blue; O: red; F: green; S: yellow; Sm: Cyan. (d) Absorption spectra of 10 μM SmPhen(TTA)3 in toluene at different temperatures corrected with the thermal expansion of toluene (0.00108 mL/K).

Various strategies have been applied to optimize the antenna design for the triplet state in light-harvesting for lanthanide ions. These include the chemical variation of the triplet state energy of the lanthanide complex and the tuning of the location of the CT energy level. Since eqs and 2 describe single-barrier back-energy transfer, the location of the triplet state or the CT state energy level can in principle be estimated using thermal quenching experiments.

In this work, by using the same complex LnPhen(TTA) (Phen = 1,10-phenanthroline; TTA = thenoyltrifluoroacetonate), we enquire how the mechanisms of thermal quenching properties vary for different Ln3+ ions. The relevant energy level diagrams for the ions are included in Figure S1. The luminescence lifetime and intensity have been recorded at different temperatures for each complex dissolved at low concentration in toluene solution. We have chosen this complex because it is well-researched[21] and gives efficient luminescence for Ln3+.[22−24] The materials, syntheses, and instrumental details are presented in the Supporting Information.

Our complexes have been characterized spectroscopically (Figure b) and by DFT[25] (Figure c). The FTIR spectra of solid LnPhen(TTA) (Figure b) are similar and show that the coordination of the antennae to the Ln3+ does not change across the series. The calculated[25] vibrational frequencies of SmPhen(TTA) are in good agreement with the experimental data (Figure S2). The electronic absorption spectra of LnPhen(TTA) are similar across the series, with a peak maximum at ∼340 nm and a shoulder at ∼360 nm (Figure S3a). A minor blue shift of the absorption maximum can be observed with a decreasing ionic radius of Ln3+ (Figure S3b). The absorption spectra do not exhibit remarkable wavelength or absorbance shifts with temperature (for example, as in Figures d and S4). Such shifts may account for luminescence intensity quenching when using a fixed excitation wavelength.

Generally, the luminescence intensity of LnPhen(TTA) (Ln = Eu, Sm, Yb, and Nd) decreases with increasing temperature. The ligand singlet level lies below the lanthanide energy levels for Ln = Y, La, and Gd, and hence, ligand–metal ion energy transfer does not occur in those cases. We now discuss the results from individual lanthanide ions in turn, starting with Eu3+, for which the 5D0 level is situated at more than 11 900 cm–1 above the next-lowest level, 7F6, so that multiphonon relaxation is slow. The dominant mechanism for temperature quenching should then involve back-energy transfer to an excited state. The variation of the emission intensity of EuPhen(TTA) with temperature is shown in Figure a, using 342 nm excitation into the ligand absorption band. The luminescence decay curves using 355 nm excitation are shown in Figure b. Over the temperature range from 283 to 333 K, the intensity decreases by a factor of 3.8, whereas the emission lifetime decreases by the factor of 2.1. Figure S5 shows that the lifetime and intensity data follow the same trend only after about 305 K so that then both apply to processes occurring solely in the 5D0 state. Our measurement of the emission lifetime at 77 K yields 0.708 ms, and this leads to the estimated value of kr = 1412 s–1. We have treated the data in Figure assuming temperature-assisted back-energy transfer from 5D0 to a single upper state. Using the above value of kr, the value of activation energy from a three-parameter fit to the intensity data using eq is displayed in the inset of Figure a, with fits to the entire data set (red) or to the data points above 303 K (blue), which show a similar trend to lifetime data in Figure S5a. The correction for the volume expansion of toluene has a negligible effect upon the fits and parameters and is neglected here and subsequently. The fit to the lifetime data using eq with three parameters gives the values E = 5282 ± 101 cm–1, Γ0 = (1.4 ± 0.6) × 1013 s–1, and kr = 1460 ± 2 s–1, with the latter value not far from the measured 77 K value. Considering the additional back-energy transfer from 5D0 to the 5D1 state situated at 1746 cm–1 higher energy, two-level back-transfer fits are discussed in the Supporting Information, section E, and found to be inapplicable. In conclusion, the back-energy transfer is directed dominantly to one excited level situated about 5000 cm–1 above 5D0.

Figure 2

(a) Evolution of the emission spectrum of 10 μM EuPhen(TTA) in toluene with respect to temperature. The inset shows the integrated area of the 5D0 → 7F2 Eu3+ emission with temperature after a Jacobian correction of each spectrum to an energy scale. Alternative fittings using eq with the value kr = 1412 s–1 are displayed in red and blue in the inset. (b) Evolution of the lifetime of 10 μM EuPhen(TTA) in toluene with respect to temperature. The inset shows the lifetime variation of Eu3+ with temperature, and the fit using eq is displayed in purple. In both figures, purple represents the lowest temperature (282 K), and red represents the highest temperature (333 K) with steps of 3 K.

The nature of this excited level is now considered. Figure S7 shows that for Ln = Gd, Y, and La, the ligand triplet state is located at ∼488 nm (20 480 ± 50 cm–1). The energy gap from 5D0 at 17242 cm–1 is only 3238 cm–1, which is considerably smaller than the calculated activation energy. Back-energy transfer to 5D2 (at 4180 cm–1) is also discounted. Berry et al.[26] have attributed the back-energy transfer in solid-state europium tris(2,2,6,6-tetramethyl-3,5-heptanedionato) (hereafter abbreviated to Eu(thd)3) to a charge transfer state. We plot their data according to the simple Arrhenius model in Figure S8a. The 5D0 nonradiative rate is very small and constant up to 236 K, and then, it increases. We fitted the rising linear portion to give the activation energy of 4076 ± 41 cm–1. There is no evidence for dominant back-transfer to a second state above the temperature of 290 K. The analogous plot of our lifetime data is presented in Figure S8b, and a linear fitting with an activation energy of 4712 ± 50 cm–1 is obtained above 297 K. The portion to lower temperature in this figure exhibits a lower activation energy (roughly 2580 ± 166 cm–1 from five data points), just like the portion between 236–290 K in Figure S8a. Presumably, the lower activation energy in this temperature range corresponds to back-transfer to 5D1 and to the triplet state.

Generally, although other parameters may be involved, the charge transfer energy can be related to the electric dipole/magnetic dipole ratio of the emission spectrum, since the 5D0 → 7F2 Eu3+ emission intensity can be enhanced, whereas 5D0 → 7F1 is not, to first order. The degree of enhancement is inversely proportional to the energy separation of the CT band from that of 5D0. Taking the ratio (5D0 → 7F2)/(5D0 → 7F1) = R as an illustration, the value of R for Eu(thd)3 is 19.9, whereas it is 13.0 for EuPhen(TTA). The charge transfer band is therefore expected at higher energy in the present case than for Eu(thd)3.

The charge transfer vertical transition state for EuPhen(TTA) is estimated at 385 nm (25 974 cm–1) from the subtraction of the EuPhen(TTA) and GdPhen(TTA) electronic absorption spectra in Figure S9. As noted by Blasse et al.,[27] the location of the vertical charge transfer transition energy may differ appreciably from that of the intersection of the 4f6 7F level manifold with the charge transfer state given by E.

We move on to the quenching of Sm3+ emission in both the visible (below 565 nm) and near-infrared (below ∼880 nm) regions, which occurs from the same luminescent state, 4G5/2 at ∼17 700 cm–1. The emission spectra of SmPhen(TTA) in toluene are displayed in Figure a,b as a function of temperature. The charge transfer state of SmPhen(TTA) is calculated to lie at 9103 ± 1231 cm–1 above that of EuPhen(TTA)(28) so that back-energy transfer to this state does not occur at the temperatures investigated. The next-lowest level, at ∼9200 cm–1 below 4G5/2, is 6F9/2,[29] and using the equation for the temperature dependence of multiphonon relaxation rate for weak couplingwhere ν̅ is an effective phonon frequency and p is the order of the process, it is not possible to obtain a data fit using sensible values for ν̅ and p. The 4F3/2 and 4G7/2 levels of Sm3+ are situated at ∼1000 and ∼2000 cm–1 above 4G5/2, respectively,[30,31] so that, just as for the following case of Nd3+ subsequently discussed, the multiphonon decay rates from these states to 4G5/2 are very high. The Arrhenius plot of the variation of nonradiative rate with temperature (Figure S10) shows that the activation energy changes with temperature, with the slope of the red line for the highest temperatures measured in Figure S10a giving the activation energy of 866 ± 93 cm–1. The insets in Figures a,b show plots using eq for the temperature dependence of emission intensity, and the value E = 1326 ± 75 cm–1 is obtained in Figure a, with a poorer fit to the lower data quality in Figure b. We therefore conclude that within the range of our measurements, spectral changes in lifetime and intensity involve an activation energy of ∼1000 ± 500 cm–1. This energy can be associated with the thermal occupation of the 4F3/2 level above 4G5/2. The thermal population of 4F3/2 reduces both the lifetime and the intensity of the spectrum. No new hot band spectral features are observed within the temperature range studied.

Figure 3

(a) Visible and (b) NIR emission spectra of 10 μM SmPhen(TTA) in toluene as a function of temperature after a Jacobian correction to energy scale. The insets show the integrated area of 4G5/2 → 6H9/2 (visible) and 4G5/2 → 6F5/2 (NIR) Sm3+ emission with temperature. The fits using eq are displayed in olive. Purple: lowest temperature; red: highest temperature, as in Figure .

Just as for SmPhen(TTA), the thermal quenching of the luminescence intensity of NdPhen(TTA) is not as remarkable as that in EuPhen(TTA). The emission occurs from the two Kramers doublets of the 4F3/2 level at ∼11 400 cm–1, which is ∼5100 cm–1 above the next-lowest term 4I15/2 and ∼960 cm–1 below 4F5/2.[32,33] The emission spectra between 282 and 333 K are shown in Figure a. Three red arrows mark the positions of hot band emission. In the range from 282 to 333 K, the emission intensity decreases by a factor of 1.1, whereas the lifetime increases by the same factor. The activation found from eq for the decrease in the 4F3/2 → 4I11/2 peak area of NdPhen(TTA) with temperature in Figure a is found to be 693 ± 502 cm–1. The lifetime of the 4F3/2 state in NdPhen(TTA) is less than 1 μs and increases slightly with increasing temperature. The lifetime is much shorter than, for example, Y3Al5O12:Nd3+ (1 at. %), where the radiative and measured lifetimes are 250 and 228 μs, respectively,[34] because the 5100 cm–1 gap to 4I15/2 can be spanned by three ∼1600 cm–1 vibrations in the present case. The decrease in emission intensity and yet increase in lifetime is intriguing, because they follow the same trend in Figure S11a but in fact work in opposite directions (Figure S11b). The excitation wavelength is similar in these experiments: 349 nm for the intensity measurement and 355 nm for the lifetime measurement. The difference between these lifetime and intensity plots shows an energy loss prior to entering the 4F3/2 state, presumably during transfer from the ligand. The state above 4F3/2 (i.e., 4F5/2) has a greater oscillator strength for the emission transition to the electronic ground state and a shorter luminescence lifetime than 4F3/2 (for example, see refs (35) and (36)). In fact, the decrease in 4F3/2 lifetime with decreasing temperature has been found by other authors, particularly in crystals with a higher concentration of Nd3+, and attributions to cross-relaxation, self-absorption, and multisite effects have been discounted as the major reason by Turri et al.[37] We consider that the change in lifetime with temperature could result from (i) thermalization within the two Kramers doublets of 4F3/2 and/or (ii) back-energy transfer to Nd3+4F3/2 from a trap, such as Yb3+ present in the sample. The lifetime of the 2F5/2 state of Yb3+ in YbPhen(TTA) (see below) is longer than that of the 4F3/2 Nd3+ state in NdPhen(TTA), and it is situated at ∼1150 cm–1 to lower energy (i.e., −1150 cm–1). The three-parameter single-barrier fit eq to the lifetime data in Figure b in the inset gives the values kr = (9.6 ± 0.1) × 105 s–1, Γ0 = 22 ± 34 s–1, and E = −1699 ± 293 cm–1. However, the adjusted coefficient of determination, Radj2 = 0.99072 and the value of kr would infer a quantum yield around 90%. Restraining the quantum yield to less than 50% and using separate fits to the two fairly linear portions of the curve gives much smaller values of E around −150 cm–1.

Figure 4

Emission spectra of 10 μM (a) NdPhen(TTA) and (c) YbPhen(TTA) in toluene as a function of temperature after a Jacobian correction to energy scale. The insets show the integrated areas of the emission transitions 4F3/2 → 4I11/2 Nd3+ and 2F5/2 → 2F7/2 Yb3+, respectively, with temperature with fits using eq . The decay curves of 10 μM (b) NdPhen(TTA) and (d) YbPhen(TTA) emission in toluene with varying temperature. The insets show the variation in lifetime with temperature with fits by eq . Purple: lowest temperature; red: highest temperature, as in Figure .

The 4f13 system Yb3+ comprises only two 2L multiplet terms: the ground state term 2F7/2 (comprising four Kramers doublets) and the excited 2F5/2 term at ∼10 200 cm–1 (comprising three Kramers doublets) (Figures S1 and S12a). The temperature variation of the YbPhen(TTA) emission spectrum and lifetime are displayed in Figure c,d. The spectrum has previously been reported at slightly higher resolution, with an estimated quantum yield of 0.16%.[38] The lifetime does not show a significant change when the concentration in toluene is increased by a factor of 10 (Figure S12b) showing that solvent quenching effects may not be very important. Although the trends of intensity and lifetime are similar in Figure S12c, over the temperature range studied, the lifetime only increases by a factor of 1.01, whereas the intensity decreases by a factor of 1.38 (Figure S12d). The single-barrier quenching model, eq , gives the activation energy for Figure c of 954 ± 210 cm–1, with the fit shown in the inset. Similar to the trap in NdPhen(TTA), this is in reasonable agreement with the separation of ∼1150 cm–1 from the 4F3/2 level of Nd3+. However, the fit of lifetime data in Figure d using eq gives the activation energy of −1663 ± 331 cm–1, with a quantum yield of nearly 100%. Using the above quantum yield to estimate kr in an Arrhenius equation gives a poor fit (Radj2 = 0.955) with E = −17 cm–1.

In conclusion, we have investigated the occurrence of thermally activated processes occurring near room temperature for LnPhen(TTA) complexes. Dominant back-transfer from Eu3+ to a charge-transfer state has been demonstrated for Ln = Eu, analogous to the solid-state study of Berry et al.[26] The derived energies from the back-transfer model for Ln = Sm, Nd, and Yb cannot be linked to the ligand or charge transfer states. Ln = Sm3+ exhibits quenching of intensity and lifetime with increasing temperature, and this has been associated with thermal occupation of higher levels. By contrast, Ln = Nd, Yb exhibit intensity quenching and lifetime lengthening with increasing temperature. The two effects cannot be consistently modeled using the common back-transfer models. We infer that the intensity decrease with increasing temperature is related to the population mechanism from the ligand, whereas the lifetime change with temperature subsequently occurs in the lanthanide excited state. A detailed investigation of the ligand–lanthanide energy transfer is therefore required for these cases.