Amending the Anisotropy Barrier and Luminescence Behavior of Heterometallic Trinuclear Linear [M II -Ln III -M II ] (Ln III = Gd, Tb, Dy; M II = Mg / Zn) Complexes by Change from Divalent Paramagnetic to Diamagnetic Metal ions

The sequential reaction of a multisite coordinating compartmental ligand LH [2-(2-hydroxy-3-(hydroxymethyl)-5-methylbenzylideneamino)-2-methylpropane-1,3-diol] salts followed by the addition of stoichiometric ratio in the presence of triethylamine affords a series of isostructural heterometallic trinuclear complexes containing [Mg 2 Ln] 3+ [Ln = Dy ( 1 ), Gd ( 2 ) and Tb ( 3 )] and [Zn 2 Ln] 3+ [Ln = Dy ( 4 ), Gd ( 5 ) and Tb ( 6 )] cores. The formation of 1-6 is demonstrated by X-ray crystallography as well as ESI-MS spectra. All complexes are isostructural possessing a linear trimetallic core with a central lanthanide ion. In this article we have discussed the comprehensive studies, involving synthesis, structure, magnetism and photophysical properties on this family of trinuclear [Mg 2 Ln] 3+ and [Zn 2 Ln] 3+ heterometallic complexes. Complexes 1 and 4 show slow relaxation of the magnetization below 12 K under zero applied direct-current field, but without reaching a neat maximum which is due to the overlapping with a faster quantum tunnelling relaxation mediated through dipole-dipole and hyperfine interactions. Under a small applied direct-current field of 1000 Oe the quantum tunneling was almost suppressed and temperature and frequency dependent peaks were observed, thus confirming the SMM behavior of complexes 1 and 4 . The fit of the high-temperature relaxation times to the Arrhenius equation affords an effective energy barrier for the reversal of the magnetization of U eff =72(2) K with  o = 8 x 10 -9 s for the SR process and U eff an effective energy barrier for the reversal of the magnetization U eff = 67(3) K with  o = 4.5 x 10 8 s. To rule out the involvement of intermolecular collaborative interactions in the dynamic of relaxation, we have performed ac susceptibility measurements on 1:10 Dy:Y magnetic diluted samples of of 1 and 4 , named as 1' and 4' . Interestingly, the diluted compounds 1' and 4' exhibits SMM behavior under zero magnetic field, thus suggesting that their relaxation processes are single molecular in origin and arise from the M-Dy-M unit. Ab initio CASSCF+RASSI calculations carried out on 1 and 4 confirm that the magnetic anisotropy is axial along the M-Dy-M axis and that the relaxation process occurs through the first excited energy level. Furthermore, the chromophoric [LH 3 ] 2– ligand is able to act as an “ antenna ” group which was found to be effective in the selective sensitization of the emissions of Tb III -based complexes 3 and 6. The emission quantum yields and the luminescence lifetimes at room temperature are 11.7 % and 0.606 ms for 3 , 22.7 % and 0.799 ms for 6


Introduction
Coordination compounds involving lanthanide metal ions have been attracting interest in view of their potential ability to be behave as single-molecule magnets, which have been proposed for applications in molecular spintronics,[1] ultrahigh density magnetic information storage,[2] and quantum computing [3]. The utility of the lanthanide ions is due to the fact that some of them have large unquenched orbital angular momenta and consequently large intrinsic magnetic anisotropy besides carrying a significant magnetic moment (i.e., DyIII, TbIII, HoIII, ErIII). In spite of the attractiveness of lanthanide ion complexes in single-molecule magnets, [4] it is to be noted that fast quantum tunnelling mechanism (QTM)-induced relaxation processes mediated through dipolar interactions, transverse anisotropy, or hyperfine interactions can reduce the energy barrier to an effective value (Ueff), thus diminishing the SMM properties of the lanthanidecontaining species. [5] To overcome this, there are a few techniques such as the dilution of such complexes within a diamagnetic matrix to eliminate dipolar interactions [6] and/or the application of a small static magnetic field[7] to partly or fully suppress the QTM relaxation processes. We, [8] and others, [9] have experimentally shown that the very weak JM-Ln value observed for 3d/4f dinuclear (MII = Mn, Co, Ni, and Cu) [8c-e] and trinuclear (MII = Co and Ni) [8a], [8b] complexes display small effective energy barriers for the separations of the low-lying split sublevels and consequently to a smaller energy barrier for magnetization flipping. In this regard, an effective plan to boost the SMM properties of the 3d/4f aggregates would be get rid of the very weak MII-LnIII interactions that split the ground sublevels of the LnIII ion by substituting the paramagnetic MII ions by a diamagnetic ion [6c], [9a], [10]. In addition to their interest in magnetism, lanthanide complexes are also of interest in photoluminescence with potential applications ranging from photophysical properties on a family of trinuclear [Mg2Ln]3+ and [Zn2Ln]3+ heterometallic complexes.

Molecular Structures of 1-6
Single-crystal X-ray diffraction studies reveal that complexes 1-6 are isostructural (only with the change in the divalent metal ions; MgII in complexes 1-3 and ZnII in complexes 4-6) and crystallize in the monoclinic space group P21/n. Complexes 1-6 are tricationic, containing three nitrates as the counter anions.
The molecular structure of 1 will be discussed herein to illustrate the common structural features of the six complexes. The molecular structure of 1 is shown in Figure 2 and those of 2-6 are given in Supporting Information ( Figure S6). Bond parameters associated with complex 1 are c) given in the caption of Figure 2 and those of complexes 2-6 are given in the Supporting information (Tables S1 to S5 respectively).  and Mg2-O6-Dy-O11) are generated ( Figure 3). The two four-membered rings bisect each other with an angle of 59.7°. Some more details of the coordination behaviour are as follows. The imino-nitrogen of the ligand binds with MgII ion with a distance of 2.016-2.144 Å. The benzylic hydroxyl group of the ligand is not deprotonated and coordinates with the dysprosium ion with a distance of Dy-Oavg~ 2.398 Å. The alcoholic hydroxyl group of the ligand also remains protonated and one alcoholic hydroxyl group coordinates with the MgII ion with a distance of 2.062-2.124 Å. The other protonated alcoholic hydroxyl group remains free and takes part in an intramolecular hydrogen bonding with the benzylic hydroxyl group ( Figure 2). Further, a two-dimensional hydrogen bonding network is seen along the ab plane; each trinuclear unit interacts with four neighbouring molecules through the nitrate anions ( Figure S7). The bond parameters associated with the intra and inter molecular hydrogen bonding are given in Table S6. Based on the various coordination actions as discussed above, MgII ion is hexacoordinated (2N, 4O), in a distorted octahedral geometry; DyIII ion is eight coordinated (8O) and in a distorted square antiprismatic geometry ( Figure 4).

Magnetic properties
The direct-current (dc) magnetic susceptibilities (M) of complexes 1-6 has been measured in the 2-300 K temperature range under an applied magnetic field of 0.1 T and are given in Figure 5 in the form MT vs T. The room temperature χMT values for these complexes are very close to those calculated for isolated LnIII ions in the free-ion approximation (Table 1).
On cooling, the MT product of the DyIII (1 and 4) and TbIII (3 and 6) complexes steadily decreases down to 2 K, which is due to the depopulation of the excited mj sublevels of the DyIII and TbIII ions, which arise from the splitting of the 6H15/2 and 7F6 ground terms, respectively, by the ligand field, and/or possible very weak intermolecular interactions between the Ln3+ ions.
The MT product for the GdIII compounds (2 and 4) remains almost constant from room temperature to 2 K, as expected for such an isotropic ion.
The field dependence of the magnetization for complexes 1-6 are given in Figure S8. The M versus H plot at 2 K for the DyIII (1 and 4) and TbIII (3 and 6) complexes shows a relatively rapid increase in the magnetization at low field to reach almost saturation for magnetic fields of 5T.
The observed saturation values for the DyIII and TbIII complexes are rather lower than the calculated ones, which is due to crystal-field effects leading to significant magnetic anisotropy. [7a], [20] The relatively simple point-charge model [5b, 5d] predicts just that DyO8 coordination environments with an axially elongated square-antiprism D4d symmetry, which can be achieved by increasing the electronic density near to the S8 axis, favor the SMM behavior. On the other hand, the axially compressed square-antiprism D4d symmetry, which can be reached by increasing the electronic density near to the basal plane, does not favour the SMM behavior in DyO8 complexes but in the Er3+ counterparts. Taking into account exclusively the above symmetry criterion, compounds 1 and 4, both possessing axially compressed square-antiprism D4d symmetry, should not exhibit SMM behavior. It is worth mentioning that this symmetry criterion applies quite well to homoleptic DyO8 systems with almost equivalent oxygen atoms. However, in heteroleptic DyO8 systems with distorted square-antiprism D4d symmetry and non-equivalent oxygen atoms, the symmetry criterion is unsuitable for predicting SMM behavior. In these cases, the differences in charge between the oxygen atoms coordinated to the Dy3+ ion play a central role in dictating the SMM behavior. In this regard, the free ion electron density for the DyIII ion has an oblate shape, which can be stabilized by an axial crystal field, where the donor atoms with the largest electron densities are located above and below the equatorial plane, thus minimizing the repulsive interactions between the ligands and  [21]. Therefore, the magnetic moment that is perpendicular to the electron density disc is found in the direction of the shortest Dy-O bonds. In the case of the heteroleptic complexes 1 and 4, the shortest Dy-O distances involve the phenoxideoxygen donor atoms at opposite positions of the DyIII ion and therefore an easy-axis anisotropy could be expected, the magnetic moment lying parallel to the direction defined by the phenoxide- In order to know if compounds 1-6 exhibit slow relaxation of the magnetization and SMM behavior, ac magnetic susceptibility measurements as a function of the temperature and frequency were performed under zero and 1000 Oe dc fields. The results of these measurements demonstrate that only compounds 1 and 4 exhibit frequency dependence the out-of-phase ("M) signals typical of thermally activated relaxation process ( Figure S9 and S10). However, no neat maxima are observed in the temperature dependence of the out-of-phase ("M) at different frequencies, which can be due to overlapping of different relaxation processes, including a faster quantum tunnelling relaxation process even at frequencies as high as 1400 Hz. This behaviour seems to indicate that 1 and 4 shows slow relaxation of the magnetization and possibly SMM behaviour. The increase of the out-of-phase ("M) signals at very low temperature can be taken as a clear indication of the existence of fast quantum tunneling of magnetization.
When the ac measurements were performed in the presence of a small external dc field (Figure 6) of 1000 G to fully or partly suppress the quantum tunneling relaxation of the magnetization (QTM), broad peaks appear for 1 with maxima in the temperature range 9.0 K (1488 Hz)-4.5 K (10 Hz). For frequencies upper than 280 Hz two maxima begin to be visible in the 6-9 K range. It should be mentioned that the existence of several thermally activated relaxation processes for crystallographically equivalent DyIII ions is not unprecedented, [4g], [7a], [9b] demonstrating once again the complexity of the relaxation processes occurring for 4f-containing complexes. The Cole-Cole diagram for 1 in the temperature range 5-6 K ( Figure S11) exhibits semicircular shapes but the semicircles become distorted between 6.2 K and 8.2 K indicating the presence of two relaxation process. The fitting of the Cole-Cole plot to the generalized Debye model for two thermally activated processes allowed the extraction of their corresponding relaxation times.  It is well known that when different processes contribute to the relaxation, the Arrhenius plot usually deviates from linearity [22]. Despite the fact that the  values extracted from 1 and 4 under a 1000 Oe static magnetic field indicate the existence of a distribution of relaxation processes, the 20 relaxation times for these complexes do not deviate from the Arrhenius linear plot in the temperature range where the curves corresponding to the frequency dependence of the out-of-phase signals show maxima (curves used to extract accurate relaxation times).
We have performed magnetization hysteresis loop measurements on powder samples of 1 and 4 at 2 K and using a sweeping rate of 0.25 T with the aim of confirming the SMM properties of these compounds (Figure 8 and S13). The compounds exhibit at 2 K butterfly shaped hysteresis loops with a large step near zero field, which is consistent with the QTM generally found for 4f containing complexes and with the tail that these compound exhibits at low temperature in the M″ vs T plot. In order to know how the intermolecular magnetic dipolar interactions influence the relaxation of the magnetization in these complexes and to unequivocally demonstrate that the relaxation process is single molecular in origin, we have performed ac susceptibility measurements on the magnetic diluted samples 1' and 4'. These samples were prepared through crystallization with the diamagnetic and isostructural Mg2Y and Zn2Y complexes using a Dy/Y molar ratio of 1:10 (the amount of Dy present in the dilute sample was calculated to be 10.8 and 10.2% for 1' and 4', respectively, from the magnetization values at 5 T and 2 K of the magnetic dilute and neat compounds). X-Ray powder spectra for 1' and 4' clearly show that theses complexes are isostructural between themselves and with the neat compounds ( Figure S14). Interestingly, the diluted compounds 1' and 4' shows under zero-field ( Figure S15 and Figure 9) out-of-phase peaks in the ranges 6.5 (2000)-8.5K (1000 Hz) and 6.5 (300)-8K (1400 Hz  The relaxation times were extracted from the fitting of the frequency-dependent ac data for 1' and 4' to the Debye model ( Figure S16). The results were then used in constructing the Arrhenius plot shown in the insets of Figure S15 and Figure 9. The fit of the high temperature data (above 6.5 K and 5.5 K for 1' and 4', respectively) afforded an effective energy barrier for the reversal of the magnetization of 56.8 (4) K with o = 3.4 x 10-8 s for 1' and 54(4) K with o = 1.7 x 10-7 s for 4'. The Arrhenius plots, constructed from the temperatures and frequencies of the maxima observed for the "M signals in Figure S15 and Figure 9, lead virtually to the same results, as expected. As the data deviate from linearity in the low temperature region due to the existence of the QTM relaxation process, we have fitted the temperature dependence of the relaxation time to the following equation th at considers the simultaneous occurrence of both the thermal and QTM processes: The fit afforded the following parameters: Ueff = 66(7) K with o = 1.7 x 10-8 s and QT = 0.0017(1) s is for 1' and Ueff = 72(3) K with o = 1.2 x 10-8 s and QT = 0.0004(2) s is for 4'. The Ueff values are very close to that obtained for 1 and 4 under a static magnetic field of 1000 Oe when the QTM process is almost fully suppressed. The Cole-Cole plots for 1' and 4' ( Figure   S17) show in the 6-9.5 K and 5.5-8 K temperature regions semicircular shapes with  values in the range 0.44-0.33 and 0.14-0.38, respectively, thus indicating the presence of a distribution of relaxation processes in those regions. These results and the fact that compounds 2 and 5 do not exhibit out-of-phase ac signals in the temperature dependence of the out-of-phase ac susceptibility plot, clearly point out that the relaxation processes observed in 1 and 4 arise from the M-Dy-M unit rather than from intermolecular interactions and long-range ordering. After applying a small static field of 1000 Oe, the QTM is almost suppressed due a combination of field and dilution effects (Figures S18, S19 and S20) and the fit of the relaxation times vs 1/T data for 1' and 4' in the 9-6 K and 9.8-7 K temperature ranges to the Arrhenius law leads (insets Figure S18 and S19), as expected, to a considerable increase of the thermal energy barrier and a decrease of o (Ueff = 90(7) K and o = 1.1 x 10-9 for 1' and Ueff = 106(4) K and o = 5.2 x 10-10 for 4'. As expected, the relaxation processes for 1' and 4' under an applied field of 1000 Oe are slower than those for 1 and 4 under the same applied field. The above results for the dilute (1' and 4') and undiluted (1 and 4) complexes suggest that the application of a magnetic field of 1000 Oe and the 1/10 Dy/Y dilution process slow the magnetization almost in the same extent. Therefore, it is not surprising that the combination of both effects results in an additional significant slowing of the magnetization relaxation process.
Some experimental results have shown that the substitution of a paramagnetic ion by ZnII improves the SMM properties. [9a], [10e-h] This is mainly due to the following facts: (i) As the M-Dy interactions are very small, the first excited state is of only a few wavenumbers above the ground state and therefore a small effective thermal energy barrier is expected. (ii) a paramagnetic ion could create a random transversal field for the DyIII ions which would favour the faster QTM process and mask of the slow relaxation process [6a], [c], [23] (iii) a diamagnetic ion would mitigate the intermolecular interactions that favour the fast QTM.
Complexes 1 and 4 represent additional good examples of the benefit effects of such substitution as the isostructural analogues Co2Dy and Ni2Dy do not show maxima in the out-ofphase signal even in the presence of an applied dc field, [8a, b]  In order to support the presence of axial anisotropy and to provide a good description of the parameters involved in the spin relaxation processes of 1 and 4 [21,24], we have performed electronic structure calculations based on CASSCF methods. Table 2 presents the calculated energies and g factors for the four lowest Kramers' doublets. The excitation energies between the ground and second Kramers' doublets for 1 and 4 are 165.8 and 147.2 cm-1, respectively, which fall within the typical range for DyIII CASSCF+RASSI calculations. It is worth mentioning that the experimental temperature dependence of the MT product can be well reproduced from the energy levels obtained in the ab initio calculations ( Figure S21) The ground states are strongly axial (gz around 19.8), with almost vanishing transversal components of g. The easy-axis anisotropy of the DyIII ion favors the slow relaxation of the magnetization and the SMM behavior [19a, 21]. Furthermore, the calculated excitation energies in the CASSCF step are also favorable for a strongly anisotropic magnetic moment, with an almost two-fold degenerate ground state (first excitation energies 2.5 and 1.6 cm-1 for complexes 1 and 4, respectively ) and a higher second excited state (139.6 and 105.6 cm-1, respectively). This energy profile favors the mixing of mostly the first two CASSCF states in the ground-state wavefunction obtained with the RASSI method, both of similar shape (Figure 10 and Supporting Information Figure S22) resulting in an oblate beta electron density for such state [21]. The plotted beta density of the 4f DyIII electrons obtained in the CASSCF step for the ground state of the two complexes is represented in Figure 9 (DyIII is 4f9 and the 7 alpha electrons give an isotropic spherical electron density showing the expected oblate beta electron density for such state). [21]   The calculated magnetic moment of the ground state for the two complexes is aligned with the direction of the three metal atoms present in the structure. In order to verify the influence of the ligand potential on the anisotropy of the 4f electronic density of the DyIII cation, we constructed electrostatic potential maps of the ligand environment projected on the DyIII position by means of DFT calculations ( Figure S23). The differences in the electrostatic potential are very small due to the presence of eight relatively similar oxygen atoms coordinated to the metal. There is a not a clear preferential orientation to accommodate the oblate density of the DyIII with the lowest electron repulsion with the ligands. Nevertheless, as indicated elsewhere, the distances with the phenoxo oxygen atoms are shorter than those with the alkoxo groups, so that the beta electron density ( Figure S23) is accommodated in the equatorial region where the phenoxo groups are located. Despite the small differences in the electrostatic potential created by the oxygen atoms of the ligand, we have been able to calculate the direction of the anisotropy axis of the DyIII ion by using the simple electrostatic model recently reported by Chilton et al. [25] As it can be observed in Figure (Figures S24 and S25), the orientation of the anisotropic axis on each Dy3+ ion compare rather well with that obtained by ab initio methods.
To shed light on the mechanism of the magnetic relaxation in complexes 1 and 4 we computed the transversal magnetic moments between the connecting pairs of opposite magnetization ( Figure 11). As can be observed, the transversal magnetic moment between the ground state Kramers' doublet is very small in both complexes (around 10-3 B), which suggest that QTM is almost suppressed in the ground state. This must be the reason why 1 and 4 exhibit SMM behavior at zero field. The off diagonal terms of the transversal moments between the ground state and the excited states of opposite magnetization (related with the Orbach process) are only slightly larger than those involving the ground state (around 10-2 B) and therefore the relaxation takes place through the first excited state via mainly a thermal assisted QTM process, with transversal moments between the two level of the first excited Kramers doublet of 0.55 and 0.48 B for complexes 1 and 4, respectively. Finally, the fact that the effective thermal energy barriers for 1 and 4 are almost one third of the calculated energy gap between ground and first Kramers' doublets is most likely due to the existence of QTM promoted by dipole-dipole and hyperfine interactions, which cannot be fully suppressed by dilution and/or by the applications of a small static magnetic field. It is worth mentioning that a much greater reduction of the experimental Ueff with regard to the calculated one (from 200 cm-1 to 23 cm-1) has been recently observed for a bipyramidal trigonal complex [DyIII(NHPhiPr2)3(THF)2] [26]. This large reduction of the Ueff could be due to the fact that the transversal moment between the ground state Kramers' doublet is greater than those observed in 1 and 4, thus favouring the QTM and the reduction of the Ueff.

Electronic Spectra of the Complexes (1-6)
The UV-visible absorption spectra of the free ligand (LH4) and those of the corresponding complexes 1-6 were recorded in CH3OH solution (c = 1 × 10-5 M) at 298 K and are depicted in Figure 12. The ligand-centered absorption properties of complexes 1-6 are listed in Table 3. shifted in all the complexes and are found respectively at ~240, ~270 and ~345nm which can be attributed to the stabilization of the π* orbitals of the ligand upon complexation with the metal ions. It is also noteworthy that the large molar absorption coefficients observed for the LH4 implies that they have a strong ability to absorb light. The magnitudes of molar absorption coefficient values for the complexes were approximately four times higher than that of the ligand, and this trend is consistent with the presence of four ligands in each complex. In this context it is important to mention that in the metal complexes a higher extinction coefficient is observed in comparison to the ligand which indicates the possibility that the ligand can be involved in for the sensitization of lanthanide luminescence.

Photoluminescent properties
In order to understand the energy migration pathways for complexes 1-6, it was necessary to determine the singlet and triplet energy levels of the ligand (LH4). The singlet (1ππ*) energy level of the ligand (LH4) was determined by reference to the wavelength of the UV-vis upper absorption edge of Gd3+ complexes 2 and 5 ( Figure 12) and the relevant values are 363 nm (27548 cm-1) and 372 nm (26881 cm-1) respectively. In order to determine the triplet energy (3ππ*) level we have carried out the low temperature (77 K) phosphorescence measurement of the same gadolinium derivative and the values are found to be 442 nm (22,575 cm-1) and 459 nm (21,778 cm-1) respectively, for the complexes 2 and 5. It is well known that Gd3+ complexes are ideal for determining triplet energy levels (3ππ*) of the ligand for the following reasons: i) the lowest-lying excited energy level (6P7/2) for Gd3+ is located at 32150 cm−1 which prohibits any energy transfer to the gadolinium ion from the ligand ii) the heavy paramagnetic ion effect of Gd3+ enhances the possibility of intersystem crossing from the singlet to the triplet state. [27] It is therefore concluded that the luminescence observed for the gadolinium derivative is explicitly ligand-oriented. It is well documented that the singlet and triplet energy gap {ΔE(1ππ*-3ππ*)} of the ligand should be close to 5000 cm−1 for an effective inter system crossing (ISC). [28] Thus, in the present study, it amounts to 4973 cm-1 for 2 and 5103 cm-1 for 5, and therefore suggests that this ligand has a good capability for intersystem-crossing efficiency. The low-temperature phosphorescence spectra of the Gd3+ complexes 2 and 5 are shown in Figure 13. Among 1-6, the only Tb(III) derivatives, 3 and 6 exhibit metal-centered luminescence ( Figure   S26, for Dy(III) derivative). The combined steady state excitation and emission spectra for the Tb3+derivatives 3 and 6 in the solid state at room temperature are depicted in Figure 14a and 14b, respectively. The excitation spectra for 3 and 6 exhibit a broad band in the 300-400 nm region (centered at ~338 nm for 3 and ~350 nm for 6) because of the π-π* transitions of the ligand.
Moreover, it is worth mentioning that in the cases of 3 and 6 the ligand-centered emission is not detected implying an efficient energy transfer process to metal from the ligand excited states.
Upon excitation at the ligand energy level (λex= 340 nm), 3 and 6 exhibit a series of characteristic sharp emission bands of Tb3+centred at 490, 545, 585, and 620 nm, which result from deactivation of the 5D4 excited state to the corresponding 7FJ ground state of the Tb3+ ion (J = 6, a) b)

5, 4, 3).[29]
The more intense transition centred at 545 nm corresponds to the transition of 5D4 → 7F5. The room temperature excited state 5D4 (Tb3+) luminescence life time values were measured (monitored at 545 nm) and were found to be τ0 = 0.606 ms for compound 3 and 0.799 ms for compound 6. Actually in both the instances a single exponential decay curve was found highlighting the presence of a single terbium emitting center. Even though there exists a less favourable Franck Condon overlap factor with the fourth vibrational overtone of the proximate OH oscillators (υOH∼3300 to 3500 cm-1) to that of the Tb3+ emitting centre, the 5D4 lifetime values of Tb3+ complexes 3 and 6 are found to be essentially temperature dependent, with τRAD varying by more than twice (Table 4)  from the 5D4 level of terbium to the triplet states of the ligand. For 3, the energy of the ligand triplet state (ca. 22,575 cm-1) lies ca. 2075 cm-1 above the 5D4 level of terbium (20,500 cm-1), while the same for complex 6 is found to be 1278 cm-1. Hence the low energy difference between the ligand triplet and metal centered emissive excited states in complex 6 will trigger the rate of thermally activated back energy transfer and is evidenced from the nearly thrice enhanced decay profile at 77K for complex 6 (Table 4).

Figure 15.
Phosphorescence decay of (a) complex 3 and (b) complex 6. The emission was monitored at 545 nm (5D4 → 7F5) respectively in solid state at 298 K.
To quantify the ability of the ligands designed to sensitize the luminescence of lanthanides, and to draw conclusions concerning the relationship between the structure and the properties, it was appropriate to analyze the emission in terms of eq 1 (below) where Φoverall and ΦLn, represent the ligand-sensitized andintrinsic luminescence quantum yields of Ln3+; Φsen represents the

a) b)
efficiency of the ligand-to-metal energy transfer and τobs/τRAD are the observed and the radiative lifetimes of Ln3+. [30] Φoverall=Φsen×ΦLn=Φsen× (τobs/τRAD) The intrinsic quantum yield for Tb3+ (ΦTb) was estimated using eq 2 with the assumption that the decay process at 77 K in a deuterated solvent is purely radiative. [31] ΦTb=τobs(298 K)/τobs(77 K) (2) Table 4 summarizes the Φoverall, ΦLn, and Φsen. In the case of terbium luminescence, solid-state measurements gave an absolute quantum yield of 11.7 % for complex 3 compared with 22.7 % for complex 6. The closeness of the ligand triplet energy towards terbium emitting centre in complex 6 provide efficient energy transfer as evidenced from the impressive sensitization efficiency of 66.7 % which in turn give rise to a 22.7 % quantum yield when compared to complex 3, where the sensitization efficiency and quantum yields are is found to be 28% and 11.7%, respectively.

Conclusion
The efficacy of the ligand LH4 in synthesizing trinuclear heterometallic complexes allowed us to synthesize a new series of isostructural complexes containing [Mg2Ln]3+ [Ln = Dy (1), Gd (2) and Tb (3)] and [Zn2Ln]3+ [Ln = Dy (4), Gd (5) and Tb (6)] cores. Unlike the previous cases [8a, 8b], we have deliberately incorporated diamagnetic metal ions viz. Zn2+ and Mg2+ within the cluster hoping to enhance the SMM properties by suppressing quantum tunnelling as well as increasing the energy gap between ground state and excited state. Complexes 1 and 4 show SMM behavior with the following parameters: Ueff =72(2) K with o = 8 x 10-9 s for the SR process and Ueff = 61(2) K with o = 4 x 10-7 s for the FR process in complex 1, and Ueff = 67 (3) K with o = 4.5 x 10-8 s in complex 4. Ac susceptibility measurements on the magnetic diluted samples of 1 and 4, named as 1' and 4', which were prepared through crystallization with the diamagnetic and isostructural Mg2Y and Zn2Y complexes using a Dy/Y molar ratio of 1:10, clearly show that the relaxation dynamics is not due to intermolecular interactions and/or long range ordering and therefore has single molecular origin. Interestingly, the diluted compounds 1' and 4' exhibits SMM behavior under zero magnetic field and QTM at low temperature. The fit to a combination of activated and QTM relaxation processes afforded the following parameters: Ueff = 66 (7)  with 365 nm as the excitation wavelength. Absolute quantum yield was calculated on the basis of the de Mello method [33] using the equation In eq 2 Where Ei(λ) and E0(λ) are respectively the integrated luminescence as a result of direct excitation of sample and secondary excitation. A is the absorbance of the sample calculated using eq 1.
Li(λ) is the integrated excitation when the sample is directly excited, L0(λ) is the integrated excitation when the excitation light first hits the sphere and reflects to the sample, and Le(λ) is the integrated excitation profile for an empty sphere. Hydrogen atoms were fixed at calculated positions and their positions were refined by a riding model. All non-hydrogen atoms were refined with anisotropic displacement parameters. The crystallographic figures have been generated using Diamond 3.1e software.
[34f] The crystal data and the cell parameters for compounds 1-6 are summarized in Table 5 and Table 6. The crystal data and the cell parameters for 1' and 4' are summarized in Table S9. CCDC-1020216 (for 1),   Low energy spectra and g factors of the four lowest Kramer's doublets of the two studied complexes were obtained by means of CASSCF+RASSI calculations, as implemented in the MOLCAS 7.8 software package. [24] The approach is divided in two steps: (i) CASSCF (7,9) calculations for three different multiplicities (sextet, quartet and doublet) (ii) The effect of spinorbit coupling on the basis of the converged wave functions obtained in the previous step is . The structure of the model was extracted from the corresponding X-ray structure without any ligand simplification. Electrostatic potential maps were obtained by B3LYP calculations as implemented in the Gaussian09 [35] using a TZVP basis set and the geometry for the ligand environment of the previous CASSCF+RASSI calculations and removing the central DyIII ion.
Then, triethylamine was added to the above solution and stirred it for 15 minutes. After that a methanolic solution of Mg(NO3)2·6H2O or Zn(NO3)2·6H2O was added drop wise, resulting in a yellow solution. The reaction mixture was stirred for a further period of 12 h to afford a clear solution. This was filtered and the filtrate evaporated to dryness. The residue obtained was washed with diethyl ether, dried, dissolved in methanol/chloroform (1:1) and kept for crystallization under vapour diffusion conditions. After 4-7 days, pure crystalline products suitable for X-ray diffraction were isolated. Specific details of each reaction and the characterization data of the products obtained are given below. Quantities: Mg(NO3)2·6H2O (0.036 g, 0.14 mmol), Tb(NO3)3·5H2O (0.03 g, 0.07 mmol), LH4 (0.08 g, 0.29 mmol) and Et3N (0.06 ml, 0.59 mmol). Yield: 0.069 g, 66% (based on Tb). Mp: