24476
J. Phys. Chem. B 2006, 110, 24476-24479
Quenching of IR Luminescence of Erbium, Neodymium, and Ytterbium β-Diketonate Complexes by Ligand C-H and C-D Bonds Rendy H. C. Tan,† Majid Motevalli,‡ Isaac Abrahams,‡ Peter B. Wyatt,*,‡ and William P. Gillin*,† Department of Physics and School of Biological and Chemical Sciences, Queen Mary, UniVersity of London, Mile End Road, London, U.K. E1 4NS ReceiVed: August 23, 2006; In Final Form: October 3, 2006
The effect of CH and CD quenching on the luminescence lifetime of Er3+, Nd3+, and Yb3+ in the Cs[Ln(HFA)4] system has been quantified, and we have shown that for Er3+ ions the quenching is dominated by the nearest neighbor CH oscillators, whereas for Nd3+ ions the roles of more distant CH oscillators and nearest neighbor CD oscillators are important.
1. Introduction Lanthanide-doped glasses have been widely studied, in part due to their use as optical gain media. Lanthanides have also been widely incorporated into organic hosts as the organic ligands can act as photosensitizers, resulting in much higher absorption cross sections, with broader absorption bands than for the free ions.1-4 However, ligands and coordinated solvent molecules usually contain C-H and O-H bonds that can cause vibrational quenching of the excited ions, and this limits any application of such complexes in infrared emitting devices. It is well-known that deuteration or fluorination of hydrogencontaining ligands, together with exclusion of coordinated water, can extend the lifetime of infrared luminescence from lanthanide complexes; however, there has been little quantitative work on the sensitivity of these ions to quenching through these processes. Nd(III) has been a subject of particular study5 although even here there is little quantitative data. Recent work by Quochi et al.6 has attempted to address this point for Er3+ ions through a theoretical model based on a continuous medium approximation. In order to quantitatively study the effect of C-H bonds on the radiative lifetime of Er3+, Nd3+, and Yb3+ ions and to determine the spatial distance over which de-excitation can occur, we have investigated the role of deuteration on the luminescence lifetimes of the tetrakis complexes Cs[Ln(HFA)4] (H-HFA ) 1,1,1,5,5,5-hexafluoro-2,4-pentanedione; Ln ) Er, Nd, Yb) to study the effect of removing CH oscillators. This system results in each ion residing in a symmetrical environment with only four equivalent CH bonds in the nearest neighbor positions. We have shown that for Er3+ ions the luminescence lifetime is dominated by the CH oscillators in the nearest neighbor positions, whereas for Nd3+ ions the more distant CH and nearest neighbor CD oscillators play an important role. For Yb3+ it is suggested that, due to the large energy gap for the ion, a cooperative energy transfer may be operating in addition to nonradiative energy transfer between ions. * Corresponding authors. E-mail:
[email protected] (W.P.G.);
[email protected] (P.B.W.). † Department of Physics. ‡ School of Biological and Chemical Sciences.
Figure 1. Molecular structure of Cs[Er(HFA)4].
2. Experimental Methods Cs[Er(HFA)4] was prepared by mixing aqueous Cs+ HFAwith the calculated amount of aqueous erbium(III) chloride; the pink precipitate was recrystallized from aqueous methanol to give pink needles, mp 276-278 °C; single-crystal X-ray diffraction (Figure 1) showed that this material was isomorphous with Cs[Eu(HFA)4].7 Elemental analysis found C, 20.9; H, 0.3; C20H4CsErF24O8 requires C, 21.3; H, 0.4; m/z (ES-) found 993.8828; C20H4166ErF24O8 (M-Cs+) requires 993.8831; νmax/ cm-1 (KBr) 3161 (C-H) and 1653 (CdO). Crystal data for Cs[Er(HFA)4]: C20H4CsErF24O8, M ) 1128.40, orthorhombic, a ) 8.534(3), b ) 21.189(5), c ) 17.251(8) Å, R ) 90.00°, β ) 90.00°, γ )90.00°, V ) 3119(2) Å3, space group Pbcn, Z ) 4, Dc ) 2.403 Mg/m3, λ ) 0.71073 Å, µ ) 4.027 mm-1, reflections measured 3057, reflections unique 2890 with Rint ) 0.0102, T ) 160(2) K, final R indices [I > 2σ(I)] R1 ) 0.0430, wR2 ) 0.1044 and for all data R1 ) 0.1010, wR2 ) 0.1288. CCDC 292801. Cs[Er(HFA)4]-d4 was prepared by using D2O and CH3OD as solvents in the above procedure; ErCl3‚6H2O was evaporated
10.1021/jp0654462 CCC: $33.50 © 2006 American Chemical Society Published on Web 11/14/2006
IR Luminescence of Er, Nd, and Yb β-Diketonates from D2O several times before starting the experiment, and the H-HFA was subjected to a preliminary H/D exchange by stirring with Et2O and D2O for 24 h, followed by distillation of the Et2O extract from P2O5 according to the method of Ogoshi and Nakamoto.8 Cs[Er(HFA)4]-d4 had a mp of 270-272 °C; m/z (ES-) found 997.9076; C20D4166ErF24O8 (M-Cs+) requires 997.9082; m/z (MALDI) found 994 (1.0%), 995 (1.5), 996 (5.9), 997 (13.7), 998 (100), 999 (92.5), 1000 (94.9), 1001 (26.5), 1002 (47.5), 1003 (9.6), 1004 (1.5); C20D4ErF24O8 with 98% deuteration requires 994 (0.4), 995 (0.4), 996 (4.7), 997 (8.7), 998 (100), 999 (91.2), 1000 (93.0), 1001 (22.7), 1002 (45.1), 1003 (9.4), 1004 (1.6); νmax/cm-1 (KBr) 1653 (CdO). Nd and Yb complexes were prepared similarly to the Er analogues. Cs[Nd(HFA)4]: pale purple solid, mp 191-192 °C (lit.9 191-192 °C) m/z (ES-) found 969.8607; C20H4F24142NdO8 (M-Cs+) requires 969.8606; νmax/cm-1 (KBr) 3157 (C-H), 1647 (CdO). Cs[Nd(HFA)4]-d4: mp 196-200 °C, m/z (ES-) found 973.8845; C20D4F24142NdO8 (M-Cs+) requires 973.8857; m/z (MALDI) found 972 (0.9%), 973 (10.8), 974 (100), 975 (70.4), 976 (95.7), 977 (53.7), 978 (71.0), 979 (16.1), 980 (23.9), 981 (6.2), 982 (22.2), 983 (0.6), 984 (0.1); C20D4F24NdO8 with 98% deuteration requires 972 (0.2), 973 (7.8), 974 (100), 975 (71.0), 976 (99.8), 977 (54.8), 978 (70.5), 979 (16.4), 980 (22.6), 981 (6.1), 982 (20.6), 983 (4.2), 984 (0.8); νmax/cm-1 (KBr) 1647 (CdO). Cs[Yb(HFA)4]: white solid, mp ca. 280 °C (decomp), m/z (ES-) found 1001.8937; C20H4F24O8174Yb (M-Cs+) requires 1001.8917; νmax/cm-1 (KBr) 3151 (C-H), 1654 (CdO). Cs[Yb(HFA)4]-d4: mp ca. 280 °C (decomp), m/z (ES-) found 1005.9152; C20D4F24O8174Yb (M-Cs+) requires 1005.9157; m/z (MALDI) found 1000 (3.7%), 1001 (5.5), 1002 (15.8), 1003 (46.3), 1004 (75.1), 1005 (64.8), 1006 (100), 1007 (17.0), 1008 (33.2), 1009 (5.9); C20D4F24O8Yb with 96% deuteration requires 1000 (0.4), 1001 (0.5), 1002 (10.8), 1003 (44.9), 1004 (72.0), 1005 (64.4), 1006 (100), 1007 (24.0), 1008 (38.7), 1009 (8.0), 1010 (1.3); νmax/cm-1 (KBr) 1647 (CdO). Partially deuterated Cs[Ln(HFA)4] samples were prepared by dissolving Cs[Ln(HFA)4] (0.03 mmol) in 0.3 mL of a CH3OH/ CH3OD mixture of the desired composition and adding H-HFA (0.02 mmol). The solutions were kept at room temperature for 24 h, then rotary evaporated. The residues was triturated with Et2O, then dried over P2O5 in a vacuum desiccator. The extent of deuteration in each sample was determined by analysis of the MALDI mass spectrometry data (Bruker Daltonics Autoflex). Photoluminescence from the complexes was excited using ∼7 ns pulses from a Continuum Panther optical parametric oscillator (OPO) pumped with a Surelite I laser. For erbium, an excitation wavelength of 520 nm was chosen to provide direct excitation into the 4S3/2 level of the Er3+ ion which rapidly decays to the 4I13/2 level with a measured decay time of the order of 25 ns. For neodymium and ytterbium excitation, wavelengths of 420 and 460 nm were used, respectively. The luminescence was dispersed in a Triax 550 spectrometer and detected using a Hamamastu R5509-72 infrared photomultiplier tube. Lifetime data were recorded at the peak of the photoluminescence spectra at a temperature of 300 K. For erbium this corresponded to the 4I13/2 f 4I15/2 transition (1542 nm) and for ytterbium the 2F5/2 f 2F7/2 transition (1032 nm). For neodymium, data were recorded for the 4F3/2 f 4I9/2 (920 nm), 4F3/2 f 4I11/2 (1045 nm), and 4F3/2 f 4I13/2 (1340 nm) transitions;
J. Phys. Chem. B, Vol. 110, No. 48, 2006 24477 TABLE 1: Measured Lifetimes of the 1542 nm Emission from the Er3+ 4I13/2 f 4I15/2 Transition in Deuterated Cs[Er(HFA)4]a deuteration level (%)
τ1 (µs)
τ2 (µs)
τ3 (µs)
0 30 47 58 81 87 90 94 98
1.8 (100) 1.8 (37) 2.4 (46) 2.5 (31) 3.3 (17) 3.7 (11) 3.5 (10) 3.6 (6) 11 (96)
3.0 (55) 5.5 (54) 6.0 (68) 8.4 (83) 9.5 (89) 10 (90) 10.6 (94) ∼106 (4)
10 (8)
a The figures in parentheses are the percentage contribution from each component.
only the 4F3/2 f 4I9/2 (1045 nm) transition is presented because all the lifetimes were identical. Lifetime data are fitted with either a one-, two-, or three-exponential decay model using a Marquardt-Levenberg algorithm to find the best fit. 3. Results and Discussion It has been reported that Nd(HFA)3(H2O)2 underwent deuteration of the HFA ligand when it was kept in CD3OD solution for 6 h.10 We were surprised to find that the H/D isotopic exchange of Cs[Er(HFA)4] in CH3OD solution was extremely slow (90% exchange to be achieved in 24 h at 25 °C. For Cs[Er(HFA)4] nine samples were prepared with deuteration levels of 0%, 30%, 47%, 58%, 87%, 90%, 94%, and 98%. All deuteration levels were calculated from the mass spectra of the [Er(HFA)4]- anion using laser desorption/ionization mass spectrometry and have an error of ( 2%. The results for the lifetime fits to the luminescence of these samples are given in Table 1. Only the undeuterated sample data could be fitted with a single-exponential fit, which is indicative of all the ions being in identical environments. For the 98% deuterated sample there was a small component (∼4%) to the lifetime decay which had a very long lifetime, ∼106 µs. In an earlier paper we proposed that this is due to a small number of Er3+ ions being in environments where there are no CH oscillators within ∼20 Å of the ion.11 However, for the intermediate levels of deuteration it can be seen that only the 30% deuteration sample gave three lifetime components, and for all the other samples the data could be adequately fitted using a dual-exponential model. These results are in contradiction to the theoretical result proposed by Quochi et al.6 who estimated the rate constant for quenching of erbium ions in an organic matrix by assuming that the nonradiative decay could be modeled using a continuous medium approximation with a minimum distance, Rmin, between the ions and the acceptors. This gave a simple expression for the rate constant, knr,
knr =
λem4 kr〈RA〉Er (2πn)4 Rmin3
For our deuterated samples Rmin effectively does not change as even at 70% deuteration >75% of the lanthanide ions still have at least one CH bond within their four directly coordinated ligands, and for our 30% deuterated sample >99% of the
24478 J. Phys. Chem. B, Vol. 110, No. 48, 2006 complexes contain at least one hydrogen atom. The Quochi model naturally gives an “average” deactivation rate constant and hence would predict a single lifetime decay that would be proportional to the strength of the vibrational absorption averaged over the erbium spectral emission window, 〈RA〉Er. This term would be directly proportional to the average concentration of CH oscillators in the material. Hence, for as long as Rmin remains unchanged the lifetime will be single exponential and linearly proportional to the deuteration level. For the undeuterated sample it is clear that all Er3+ ions are in equivalent positions. These ions each have four CH oscillators in nearest neighbor positions within the Er(HFA)4- anion, at 4.7 Å from XRD data, and then some distribution of CH oscillators corresponding to neighboring anions at distances greater than 6.962 Å, the next nearest neighbor position. This sample has a single-exponential decay with a lifetime of 1.8 µs. If we assume that it is only the four CH oscillators on the Er(HFA)4- anion which are contributing to this lifetime, this gives a rate constant for each CH oscillator of ∼140 000 s-1. Using this rate constant for a single CH oscillator at 4.7 Å we can calculate the predicted lifetimes for anions with between one and three hydrogen atoms present. This gives a predicted lifetime of 2.4 µs for [Er(HFA)4-d1]-, 3.6 µs for [Er(HFA)4d2]-, and 7.2 µs for [Er(HFA)4-d3]-. From Table 1 it can be seen that for each of the partially deuterated samples the fastest lifetime observed corresponds to the predicted lifetime for either the d0 (30%), d1 (47% and 58%), or d2 (81%, 87%, 90%, and 94%) species. It should be noted that for each sample up to 81% deuteration the lowest lifetime that could be measured corresponded to the site with the largest number of hydrogen atoms present, providing that the concentration was greater than ∼10%. For deuteration levels >81% we were able to measure the 3.6 µs component, corresponding to the [Er(HFA)4-d2]anion, even when it was present at levels down to 3%. This suggests that for Er3+ ions, providing there are at least two CH oscillators in the nearest neighbor positions, it is the contribution of these nearest neighbor CH oscillators which dominates the quenching. Once there is only one CH oscillator in this nearest neighbor position the contribution of the more distant oscillators becomes much more significant. For these samples, except in the case of the 30% deuterated sample, we were always able to accurately model the lifetime data using a double-exponential decay. This does not mean that there were only two lifetimes present in these samples but rather that, apart from the fast component, there is not a unique solution to the deconvolution of the various lifetime components and our approach gives the “best-fit” with the minimum number of fitting parameters. We did try fitting all the samples with up to five exponential decay processes forcing the lifetimes for each process to the values calculated for each local environment (i.e., d0-d3) and using the measured value of 11 µs for [Er(HFA)4-d4]-. With the use of this approach, the prefactors were the only free fitting parameter. Excellent fits were obtained for each sample, and the percentage contribution calculated from the fits were in reasonable agreement with those predicted (Table 2). For the Cs[Nd(HFA)4], six samples were prepared with deuteration levels of 0%, 49%, 59%, 60%, 85%, and 98%. For all of these samples a dual-exponential fit was needed to fit the data and the results are given in Table 3. For all the samples it can be seen that the minor lifetime component is very fast (i.e.,
