Substituent Effects on Fluoride Binding by Lanthanide Complexes of DOTA-tetraamides

Fluoride binding by a series of europium and ytterbium complexes of DOTA-tetraamide ligands derived from primary, secondary and tertiary amides has been studied by NMR and luminescence spectroscopies. In all the systems studied, fluoride binding results in a change in the nature of the 10 magnetic anisotropy at the metal centre from an easy axis, to an easy plane anisotropy. This results in reversal of the peaks in the NMR spectra, and in changes to the fine structure of the luminescence spectra. Furthermore, changes to the periphery of the binding cavity are implicated in determining the affinity constant for fluoride. There are clear differences in the entropic contribution to the free energy of activation between systems with benzylic amides and those with 15 methylamides.

magnetic anisotropy at the metal centre from an easy axis, to an easy plane anisotropy. This results in reversal of the peaks in the NMR spectra, and in changes to the fine structure of the luminescence spectra. Furthermore, changes to the periphery of the binding cavity are implicated in determining the affinity constant for fluoride. There are clear differences in the entropic contribution to the free energy of activation between systems with benzylic amides and those with 15 methylamides.
The spectroscopic and magnetic properties of lanthanide complexes have been widely exploited over the course of many years, particularly in magnetic resonance imaging 20 contrast agents and in time-resolved bioassays, and more recently in the field of molecular magnetism. In all of these areas, optimisation of the properties of these lanthanide containing systems is contingent on the fundamental understanding of their electronic structure, and the interaction 25 of open shell f-electrons with their surroundings.
Bleaney's approach to understanding the effect of lanthanide magnetism on surrounding nuclear spins has underpinned the development of the field for almost fifty years. 1 Defining these theories required a number of 30 approximations to be made; and, while many have pointed out discrepancies between theoretical prediction and experimental observations (particularly in low symmetry systems), it is only recently that supporting theoretical methods have become available to study and interpret more unusual aspects of 35 lanthanide behaviour in molecular systems.
We recently investigated the binding of fluoride ions to lanthanide complexes of the tetraamide ligand DTMA and determined that the replacement of water with fluoride induces a change in the nature of the magnetic anisotropy and 40 in the sign of the crystal field parameter ! ! . 2,3 The change resulted in dramatic changes to NMR, EPR and luminescence spectra, and the phenomenon was further understood using ab initio calculations. These results, taken together with a number of recent studies on single molecule magnets 4 and 45 luminescent complexes, 5 provide a rapidly growing body of data that reveal the subtleties inherent to the magnetic behaviour of lanthanide complexes.
In this manuscript we seek to determine how the nature of the ligand affects the fluoride-binding event in symmetric 50 DOTA tetraamide complexes. To this end we have synthesised and studied the fluoride binding by a number of complexes related to DTMA (L 1 , L 3 -L 7 ), shown in Scheme 1. This array of related complexes has allowed us to explore the effect of variations in amide structure and hydrophobicity on 55 fluoride binding and its spectroscopic consequences.

Results and Discussion
Synthesis A series of tetraamide ligands derived from a variety of 60 different amines, were synthesised by the well-established procedure outlined in Scheme 1. The chloroacetamides were synthesised from the appropriate primary or secondary amine and reacted with cyclen to give the 8-coordinate ligands. The complexes were formed by reaction of the ligands with the 65 appropriate lanthanide triflate salts and characterised by NMR, mass spectrometry and CHN analysis (see SI).
Scheme 1. Synthesis of the complexes studied herein.

70
Proton NMR studies focussed on the Yb 3+ complexes of ligands L 1-7 due to the dominance of the pseudocontact shift for this ion. The 1 H NMR spectra in D 2 O, and the effect of addition of fluoride on the spectra are shown in Figure 1 2 In all cases, EXSY spectra show a reversal of the peak order between hydrated and fluoride-bound forms, 5 implying a change of sign of ! ! (see SI). In our earlier studies, we rationalised these observations by a change in the order of the m J states arising from ligand field splitting of the 2 F 7/2 ground state giving rise to a change in the nature of the magnetic anisotropy at the lanthanide centre, and it is 10 reasonable to infer that this phenomenon is general across this series of complexes. The observation of such dramatic changes to the NMR spectra upon fluoride coordination must be explained in terms of changes to the Boltzmann populations within the m J manifold as a result of changing the 15 relative energies of the various m j states, and it is clear that the overall anisotropy of the metal ion must reflect all of these states. Inspection of the spectra in Figure 1 reveals that while the range of chemical shifts observed for the hydrated complexes are fairly similar, those for the fluoride-bound species show significant differences between complexes. Differences in the spread of observed 1 H chemical shifts between complexes 5 reflect the relative magnitudes of the axial magnetic anisotropy resulting from the effects of different ligand fields. For systems with axial symmetry and negligible contact shift contribution, the effect of the crystal field on the observed shifts is conveniently expressed using the Bleaney equation 1 10 and the crystal field parameter, ! ! : The crystal field coefficient, C J , and the parameters β, k and T are often grouped together into one parameter labelled 15 D 1 to give: Plots of δ PC vs. (3cos 2 ϑ-1)/r 3 , yield lines with gradients corresponding to D 1 (Eqn 2), which are proportional to ! ! .
δ PC are approximated from δ obs by subtraction of 2.9 ppm to 20 take account of the diamagnetic contribution and (3cos 2 ϑ-1)/r 3 values are taken from closely related crystal structures. 6,7 These experimentally derived D 1 values are given in Table 1 for each complex in both hydrated and fluoride-bound forms (plots are given in SI) along with the difference between the 25 two. For the hydrated species, the magnitudes of D 1 fall within a relatively narrow range. Observed differences correlate with variations in electron demand, although solvation is also expected to play a role. The nature of the para-benzyl 35 substituent has a small but significant influence on the crystal field, with the NO 2 group giving the smallest ! ! of the benzyl-substituted complexes. It is possible to use the data in Table 1 to estimate the degree of anisotropy in each of the complexes. Since and It is possible to define both ∥ and ! from D 1 provided we know !" . In previous studies, we modelled !" for Yb.L 2 -45 OH 2 and for Yb.L 2 -Fobtaining values for !" T at 300K of 2.49 and 2.51 cm 3 mol -1 K respectively, while the free ion !" would be expected to be 2.57 cm 3 mol -1 K. Thus the crystal field of the ligands undoubtedly has an effect upon the value of !" , albeit a small one. For the purposes of this study, a 50 qualitative map of the anisotropy can be obtained by estimating !" as 2.5 cm 3 mol -1 K for the systems studied in this manuscript. The results of applying this approach can be seen in Figure 2, which represents the magnetic susceptibilites as a series of spheroids in which the z-axis defines the 55 molecular axis. Values for ∥ and ! are tabulated in the supplementary information to this paper. From Figure 2, it should be clear that the anisotropy takes very different forms in the water-bound and fluoride-bound complexes, while much smaller (though still significant) differences in 60 anisotropy are observed between complexes with different macrocyclic ligands. It should be noted that, at the ambient temperatures studied in this work, free rotation about the Ln-O bond in the aquated complex results in averaging of ! and ! on the timescale of the NMR experiment (meaning that the 65 anisotropy can be treated as a spheroid rather than an ellipsoid on these timescales). YbL 1 and YbL 2 behave similarly on addition of fluoride. YbL 3 -F has anomalous chemical shift ranges and a much less negative D 1 , which are likely to be a consequence of YbL 3 being the only tertiary tetraamide studied. The change in slope of the Bleaney plot on binding fluoride is however similar to 5 YbL 1,2 , being in the range 3800-4100 ppmÅ 3 , so although the magnetic anisotropy in YbL 3 -F is smaller than the other complexes, this is due to the axial fluoride offsetting a stronger equatorial ligand field. The fluoride bound benzylamide derivatives, YbL 4-7 , have more negative D 1 than 10 YbL 1-3 while all show changes in D 1 in the range 4600-4700 ppmÅ 3 . This would argue that exchange of water for fluoride has a greater impact on the crystal field for YbL 4-7 compared with YbL 1-3 . This is likely to be a consequence of reduced solvation of the fluoride in the binding cavity making the 15 fluoride a better donor. The observed value of D 1 does, however, vary from complex to complex as a consequence of the electronic influences of the peripheral substituents on the benzyl groups. YbL 6 -F has a more negative gradient than the other benzylamides, however this is a consequence of a 20 weaker equatorial ligand field (as indicated by the trend in the series of hydrated benzylamide complexes). Analogous effects of fluoride binding are also observed in the NMR spectra of the europium complexes, although chemical shifts are significantly affected by the contact shift and so 30 Equation 1 cannot be employed. Furthermore, EuL 1-3 are present as mixtures of SAP and TSAP isomers in solution. 7 The exchange between isomers in these cases causes broadening of the 1 H NMR spectra and cooling is required to distinguish the peaks. The benzyl-substituted Eu(III) 35 complexes however display much sharper NMR spectra since only the SAP isomer appears to be present (see SI), as has been previously noted for larger secondary substituents of this nature. 8 EuL 3 is a mixture of SAP and TSAP in ca. 1:2 ratio and two new sets of 1 H signals appear on addition of fluoride, 40 in a ratio of ca. 1:3 (Figure 3), presumably corresponding to fluoride bound SAP and TSAP isomers. During our previous study we observed the 19 F signals corresponding to bound fluoride for several different lanthanide complexes of ligand L 2 . 3 In the case of the pseudo-50 lanthanide yttrium the signal is a doublet due to coupling between 19 F and 89 Y nuclei, as confirmed by HMQC. 3 Bound fluoride resonances are reported here for Y 3+ , Eu 3+ and Yb 3+ complexes ( Table 2). In the case of EuL 3 , two bound fluoride signals are observed and integration indicates that the peak at 55 -500 ppm (298 K) corresponds to the major isomer of the fluoride-bound form.
The yttrium complexes allow us to assess the diamagnetic contributions to the 19 F chemical shift. Complexes of L 1-3 have bound fluoride shifts of a similar magnitude to one another, 60 while complexes of L 4-7 give less negative bound fluoride shifts. Y-F coupling constants are also separated into these two groups. The larger coupling constants and less negative chemical shifts (a larger shift from free fluoride) would imply a significant difference to the nature of the interaction 65 between fluoride and the benzylamide-based complexes.
In the paramagnetic complexes, the diamagnetic contribution to the shift is small relative to the effect of the paramagnetic metal centres. The europium complexes have bound fluoride chemical shifts that are broadly similar, with 70 the exception of EuL 3 . Here the lanthanide induced shift (LIS) is dominated by the contact shift, especially considering that the fluoride is directly coordinated to the metal. The phenomenon of spin-spin coupling between fluorine and spatially adjacent nuclei by virtue of orbital overlap 75 ("through-space couplings") is well known and has been extensively studied. 9 Note that this does not imply any "formal" bond between the fluoride ion and the metal, nor does it necessarily tell us anything about the contribution of the contact shift for other nuclei (such as protons) in the 80 complex.The similarity of the lanthanide induced fluoride shifts therefore indicates that the contact shift felt by the fluoride ion is relatively constant across the series of complexes studied and implies that the nature of the contact interaction is insensitive to the substituents. 85 Conversely, the bound fluoride shifts of the ytterbium

This journal is © The Royal Society of Chemistry [year]
Journal Name, [year], [vol], 00-00 | 5 complexes are dominated by the pseudocontact shift, 3 and therefore vary with the size of the crystal field coefficient. A plot of D 1 for the fluoride-bound Yb 3+ complexes taken from Table 1 vs. LIS of the ytterbium bound fluoride gives a straight line (Figure 4). The shifts of the yttrium analogues are 5

Figure 4. Graph of D 1 (derived from Bleaney plots of 1 H shifts of YbL 1-7 ) vs. the observed LIS of bound fluoride for YbL 1-7 with linear fit in red.
used to subtract the diamagnetic contributions. The linear 10 correlation would imply that the geometric factor, (3cos 2 ϑ-1)/r 3 , remains constant across the series of complexes and since we can assume that ϑ is also constant, we can infer that the distance between fluoride and lanthanide does not vary with the identity of the ligand.

Luminescence Studies
We also studied the effects of fluoride binding on the luminescence properties of the europium complexes since information regarding the ligand field can readily be extracted 20 from fine structure. Changes to the shape of the spectrum are observed in all bands (see SI) upon addition of fluoride. The ΔJ = 1 band is the most appropriate to analyse since in axial symmetry we would expect the band to consist of two peaks as a result of the 2J + 1 degeneracy being lifted by the ligand 25 field. The parameter ! ! can be extracted directly from the splitting of this band. The fine structure of the ΔJ = 1 bands in the presence and absence of fluoride are show for EuL 1-7 in Figure 5.
In the absence of fluoride, the ΔJ = 1 regions of Eu 1-3 30 appear to consist of more than two transitions, especially for EuL 3 which displays a particularly broad band. For EuL 4-7 , ΔJ = 1 is split into two easily distinguishable peaks. This correlates with the presence of two conformational isomers with significantly different crystal field splitting, SAP and 35 TSAP, for Eu 1-3 and the predominance of SAP in EuL 4-7 . Therefore, while the shape of the ΔJ = 1 band for Eu 1-3 is complicated by the presence of multiple isomers, the shape of this band for Eu 4-7 is determined only by ! ! of the SAP isomer. It is instantly apparent from the splitting of ΔJ = 1 40 that L 6 invokes a smaller crystal field splitting than L 4,5,7 , which correlates with the 1 H NMR studies of the Yb 3+ complexes (Table 1).  (Table 1). The magnitude of the splitting is unresolvable in the Eu 1-3 emission spectra. For the benzyl substituted complexes in the presence of fluoride, the splitting of ΔJ = 1 is resolvable and is small compared with the original complex, in line with the 60 observations from 1 H NMR of the Yb complexes. The fact that the splitting is resolvable for EuL 4-7 but not EuL 1-3 tallies with the larger magnitude of D 1 observed for the former ( Table 1). The peak separation is larger for EuL 6 than EuL 4-5 in line with the more negative D 1 for the former. Furthermore, 65 while it is not possible to clearly identify the A and E components of the ΔJ = 1 band in EuL 4,5,7 , for EuL 6 the transition from 5 D 0 to the doubly degenerate E level of 7 F 1 (which appears broader in the spectrum) is now at higher

This journal is © The Royal Society of Chemistry [year]
Journal Name, [year], [vol], 00-00 | 6 energy than the A level. The energetic ordering of the A and E levels is thus reversed with respect to the hydrated complex, presenting further direct evidence of a change in sign of ! ! .
The luminescence lifetimes of the Eu-based emission were measured in the presence and absence of fluoride in both H 2 O 5 and D 2 O and are given along with q values in Table 3. 10 Lifetimes are lengthened on addition of fluoride corresponding to a change in the number of bound water molecules from one to zero, confirming the displacement of water by fluoride in each case.

Association Constants
The association constants for fluoride with EuL 1-7 were determined both by tracking changes in the emission spectra and by following changes in 1 H and/or 19 F NMR intensities as a function of fluoride concentration, which provide a direct 20 measure of the concentrations of each species. The data was fitted using a one to one binding model in Dynafit 11 and the values are given in Table 4 with associated confidence intervals. NMR titrations provided K values within error of those obtained from the luminescence titrations.
25  Figure S17), although a lack of linearity would imply that there are additional factors 45 contributing. The introduction of hydrophobic benzyl substituents does not appear to have a significant effect of the magnitude of K. 50 We previously determined that the rate of exchange of fluoride at the metal centre is relatively slow compared with, for example, water exchange. Qualitative information about the relative rates of exchange can be gleaned by observing the change in the proton NMR spectra of the Lu 3+ complexes as A quantitative assessment of the exchange rate is achieved by using a selective inversion NMR technique appropriate for a system in slow exchange. In a sample containing a mixture of the fluoride-bound and hydrated complexes, the effects of 70 exchange on a pair of resonances is monitored following selective inversion of one of the signals. The evolution of the magnetisations is governed by both the exchange rate (k) and the spin-lattice relaxation rates (R 1 = 1/T 1 ). A second experiment monitors the magnetisation following a non-75 selective inversion pulse and fitting of all data is performed by varying k and R 1 using the CIFIT2 program. 12 In order to extract k, an appropriate pair of resonances is required and in this case the main difficulty is that R 1 is very fast for nuclei near the paramagnetic centres. The 19 F nuclei of the L 7 ligand 80 are ideal for this study since they are far enough from the lanthanide to have reasonable R 1 s and other 19 F resonances do not overlap. The methyl protons of L 2 have much faster R 1 s than the 19 F of L 7 , but fortunately the rate of exchange is large enough to compensate in the case of YbL 2 . 85 The rate of fluoride exchange measured with EuL 7 is significantly slower than with YbL 7 at 298 K (  There is also a significant difference in k measured between Yb 3+ complexes of ligands L 2 and L 7 , with the aromatic substituents conferring a slower rate of exchange between species. This corroborates our qualitative observations with 10 Lu complexes of L 2 and L 4 above. A similar observation is made for the exchange rate of water at lanthanide centres in related complexes-hydrophobic groups tend to slow the rate of water exchange. 14 In order to explore the origins of the ligand effect on the rate, k was measured for YbL 2 and YbL 7 15 at a range of temperatures (see SI). Eyring plots display good linear correlations and are shown in Figure 6. The thermodynamic parameters extracted from the slopes and intercepts of the plots are given in Table 5. It is clear from 25 the data that the ΔH ‡ values for the two complexes are the same within error and thus that the electrostatic interaction between Yb 3+ and Fis independent of the ligand framework in this instance. However, the ΔS ‡ values extracted from the plots are significantly different, with the benzyl-appended 30 complex having a more negative value. This implies that the interaction of fluoride with lanthanide tetraamide complexes is appreciably affected by the nature of the ligand substituents and that the rearrangement of solvent in the vicinity of the metal centre plays a key role. The change in solvation around 35 the complex decorated with hydrophobic groups incurs a greater entropic cost during fluoride binding, resulting in a slower rate of exchange. Calculated ΔG ‡ values at 298 K are within error of one another.

40
From these results, several things become clear. Firstly, the effect of fluoride on the magnetic anisotropy of lanthanide tetraamide complexes appears to be a general phenomenon: in all the systems studied fluoride binding results in a change in the anisotropy from a prolate to an oblate electron distribution 45 as a consequence of the effect of the axial fluoride donor atom. Further, Eyring analysis of the data shows that entropy plays a large part in defining the free energy of activation. The results highlight the complexity of the influences on the fluoride binding event, even within complexes of similar 50 chemical structure, with electrostatics, sterics and solvation clearly playing intricate roles in the nature of the interaction. This study adds further weight to the increasing body of evidence that the crystal field is important in lanthanide coordination chemistry, and that the relative populations (and 55 indeed ordering) of the Stark sub-levels of the ground state are critical to defining the spectroscopic properties of the complexes.