Cation-Directed Synthetic Strategy Using 4f Tungstoantimonates as Nonlacunary Precursors for the Generation of 3d–4f Clusters

The first synthetic pathway using a series of four nonlacunary 4f-heterometal-substituted polyoxotungstate clusters Na21[(Ln(H2O)(OH)2(CH3COO))3(WO4)(SbW9O33)3]·nH2O (NaLnSbW9; Ln = TbIII, DyIII, HoIII, ErIII, YIII) as precursors for the directed preparation of nine new 3d–4f heterometallic tungstoantimonates K5Na12H3[TM(H2O)Ln3(H2O)5(W3O11)(SbW9O33)3]·nH2O (KTMLnSbW9; TM = CoII, NiII; Ln = TbIII, DyIII, HoIII, ErIII, YIII) has been developed. Systematic studies revealed an increased K content in the aqueous acidic reaction mixture to be the key step in the cation-directed preparation of 3d–4f compounds; among those, the Co-containing members represent the first examples of KCoLnSbW9 (Ln = TbIII, DyIII, HoIII, ErIII, YIII) heterometallic tungstoantimonates exhibiting the SbW9 building block. All 13 compounds have been characterized thoroughly in the solid state by powder and single-crystal X-ray diffraction (XRD), revealing a cyclic trimeric polyoxometalate architecture with three SbW9 units encapsulating a planar triangle of LnIII ions in the case of NaLnSbW9 and a heterometallic core of one TMII and three LnIII for KTMLnSbW9 (TM = CoII, NiII; Ln = TbIII, DyIII, HoIII, ErIII, YIII). The results obtained by XRD are supplemented by complementary characterization methods in the solid state such as IR spectroscopy, thermogravimetric analysis, and elemental analysis as well as in solution by UV–vis spectroscopy. Detailed magnetic studies on the representative compounds KTMDySbW9 (TM = CoII, NiII) and KCoYSbW9 of the series revealed field-induced slow magnetic relaxation.


■ INTRODUCTION
Polyoxometalates (POMs) 1 are polyanions constructed of early-transition-state metals mostly in their highest oxidation states (V and VI), which are bridged by O atoms. Changing the size, shape, or composition of POMs enables their tuning for different kinds of applications ranging from catalysis, 2 materials science, 3 and molecular magnetism 4 to bio-and nanotechnology 5 as well as macromolecular crystallography. 6 Lacunary POMs have attracted significant attention in the past decades because they can be regarded as inorganic multidentate ligands toward electrophiles like transition metals (TMs) or lanthanides (Ln) to form more stable metal−oxo cluster structures. 7 While 3d-substituted POMs form the largest POM class and the field of 4f-containing POMs is significantly expanding at present, little has been reported on the field of heterometallic 3d−4f POMs 8,9 despite their interesting magnetic properties owing to the combination of 4f ions with large magnetic anisotropy exhibiting slow magnetic relaxation if coupled to 3d paramagnets 10 (Table S1).
This scarcity of 3d−4f heterometallic POM species can be explained by competitive reactions in the TM/Ln/POM systems, thereby most often resulting in only TM or Ln blocking the vacant sites of the cluster, leaving the remaining second metal type as a mere surface decoration. 10 As a matter of fact, a variety of different synthetic strategies have been developed to circumvent this problem. 11 For instance, the use of preformed 3d−4f metal complexes with labile ligands has been established by various working groups, 12 whereas the reaction of 3d-metal-substituted POM clusters with lanthanides to avoid the formation of pure 4f metal clusters has been applied as well. 13 However, the use of preformed 4f POM clusters as precursors for the preparation of 3d−4f clusters under mild conditions has not been reported yet, although such a synthetic approach may open new perspectives for the preparation of new 3d−4f-substituted POM compounds by shielding the overreactive 4f metal centers with multidentate lacunary ligands. Moreover, the use of preformed 4f-doped clusters as nonlacunary precursors 14 in combination with 3d metals may facilitate the directed generation of high-nuclear POMs with a defined heterometallic core.
Herein, we report on the first step-by-step synthetic approach using four preformed high-nuclear 4f-doped t u n g s t o a n t i m o n a t e c l u s t e r s N a 2 1 [ ( L n ( H 2 O ) -(OH) 2 (CH 3 COO)) 3 (WO 4 )(SbW 9

■ RESULTS AND DISCUSSION
Synthesis. The first step in the preparation of the 3d−4f clusters was the synthesis of the nonlacunary 4f precursors. Among the Keggin-type polyoxotungstates (POTs), the trilacunary derivatives {XW 9 O 33 } (X = As III , Sb III , Bi III ) 15 represent ideal candidates for constructing POM-based metal clusters regarding their well-defined vacant sites and high negative charges. 16 Given the low reported number of only nine 3d−4f compounds based on the Na 9 [SbW 9 O 33 ] 19.5H 2 O {SbW 9 O 33 } unit (Table S1) and the lone pair of the trilacunary precursor, which prevents the formation of a closed Keggin {XW 12 O 40 } sphere, the {SbW 9 O 33 } unit 15 was chosen as a building block. Initially, the 4f POT precursor was prepared upon the addition of the corresponding Ln salt to a solution of {SbW 9 O 33 } in a 1:2 mixture of a H 2 O/NaOAc buffer (pH 5.5, 2 M). In order to prevent the formation of insoluble lanthanide hydroxides, a low pH value in the range of 5.5 was combined with high acetate concentrations considering the templating effect of organic ligands, which temporarily coordinate to the Ln III metal ions to decrease their overreactivity. 16 For the second reaction step, a lower reaction temperature of 50°C was chosen to ensure the integrity of the trimeric cluster. The corresponding TM was introduced as a chloride salt, TMCl 2 (TM = Ni II , Co II ), together with a solution of K 2 CO 3 . As for the KCoLnSbW 9 compounds, the addition of CoCl 2 to the reaction mixture instantaneously led to a pink solution, indicating the presence of octahedrally coordinated Co II centers, whereas blue reaction solutions were observed at temperatures higher than 80°C, which could be explained by the formation of square-pyramidally coordinated Co II metal centers, as present in the Herve-type sandwich POT. 17 Initial crystallization attempts upon slow evaporation of the reaction mixture in a pure NaOAc buffer resulted in three different crystal phases consisting of the desired 3d−4f cluster KTMLnSbW 9 , unreacted 4f precursor, and the corresponding Krebs POM Na 10 [(TM(H 2 O) 3 ) 2 (WO 2 ) 2 (SbW 9 O 33 ) 2 ] (TM = Co II , Ni II ). 18 Regarding the 3:1 stoichiometric ratio of TM to 4f precursor, 1 equiv of TM was grafted onto the 3d−4f compound, whereas 2 equiv led to the formation of Na 10 [(TM(H 2 O) 3 ) 2 (WO 2 ) 2 (SbW 9 O 33 ) 2 ], leaving the unreacted 4f precursor to crystallize ( Figure S17). Single-crystal XRD studies on the crystal phases revealed K countercations originating from the added K 2 CO 3 exclusively in the 3d−4f cluster (Scheme 1). Given the well-documented directing effect of countercations in POM reaction systems, 19 we chose to slightly increase the K content in the reaction mixture to favor formation and crystallization of the desired 3d−4f compounds. Initial attempts to increase the K content of the reaction mixture by merely adding KCl resulted in precipitates, which despite all our efforts could not be characterized. The appearance of insoluble precipitates upon the addition of KCl to the reaction mixture could be explained by a change of the ionic strength in the solution. As a matter of fact, the reaction medium was changed from pure NaOAc to a KOAc/NaOAc (5%, v/v) mixture, leading to an accelerated selective crystallization of the desired 3d−4f POM clusters by elevating the K content of the solution, while avoiding the introduction of further anions like chloride in the system and maintaining the initial pH value of 5.5. It should be mentioned that a onepot synthetic approach by combining the corresponding 3d and 4f metals with the SbW 9 building block exclusively resulted in the formation of a Krebs POM. 20 Exchanging K with Cs as a countercation in an attempt to increase the product yield by further decreasing the solubility of the POM resulted in precipitates, which could not be identified. Despite all our efforts, single crystals of KNiHoSbW 9 with sufficient quality for single-crystal XRD measurements could not be obtained. However, elemental analysis, IR spectroscopy, thermogravimetric analysis (TGA), and powder XRD measurements clearly indicate the successful synthesis of pure K 5 Na 12 H 3 [Ni-(H 2 O)Ho 3 (H 2 O) 5 (W 3 O 11 )(SbW 9 O 33 ) 3 ]·87H 2 O. To probe the versatility of the step-by-step synthesis protocol, the synthetic conditions were applied using the diamagnetic N a 1 6 ( 21 as a nonlacunary precursor in combination with Co II ions, leading to the isolation of isostructural KCoYSbW 9 . When this paper was under preparation, three crystal structures identical with our NaLnSbW 9 (Ln = Dy III , Ho III , Er III ) 4f precursor have been Scheme 1. Schematic Representation of the Step-by-Step synthesis of KTMLnSbW 9 (TM = Co II , Ni II ; Ln = Tb III , Dy III , Ho III , Er III , Y III ) a a The synthesis starts with the preparation of the 4f-doped compounds NaLnSbW 9 upon the addition of Ln III salts to a solution of SbW 9 in an aqueous NaOAC buffer at 80°C. Subsequent addition of the 3d metal salt TM II to a 5% KOAc/NaOAc buffer (v/v) solution of NaLnSbW 9 leads to the desired 3d−4f heterometallic clusters. Color legend: WO 6 , fairy floss octahedra; Ln III , dark-green balls; TM II , gray balls; Sb III , dark-gray balls; K + , light-blue balls; O, red balls; C, white balls.  9 }/Ln ratio of 1:2, whereas the synthesis protocol reported in this work applies a lower pH value of 5.5 and a {SbW 9 }/Ln ratio of 1:1, which prevents the formation of insoluble lanthanide hydroxides, while using lower amounts of Ln salts, ultimately giving higher yields of 52−70% (this work) versus 18−25% based on Ln. 22 Structure. Crystallographic studies on the in total 12 compounds showed that the four 4f precursors NaTbSbW 9 , NaDySbW 9 , NaHoSbW 9 , and NaErSbW 9 crystallize in the rhombohedral space group R 3m , whereas the 3d−4f polyanions KTMTbSbW 9 , KTMDySbW 9 , KCoHoSbW 9 , KTMErSbW 9 , and KCoYSbW 9 (TM = Co II , Ni II ) belong to the triclinic space group P1̅ . The Ln III centers are arranged in a triangular shape with a Ln−Ln distance of 5.891 Å and a bond angle of 60°. Each Ln center exhibits a distorted square-antiprismatic coordination environment, with an acetate and a water ligand occupying the three peripheral positions of the metal centers and Ln−O bond lengths ranging from 2.20(3) to 2.43(2) Å ( Figure S18 and Tables S4−S12).
For the 3d−4f counterparts, crystallographic studies showed a trimeric architecture composed of three {SbW 9 } units, which are linked by a metal core comprising three 8-coordinated Ln III metal centers connected to an octahedrally coordinated central W VI position (Figure 1). The central W VI unit is connected to two W VI and a TM center (TM = Co II , Ni II ), which act as capping moieties for the polyanion, resulting in an idealized C 3v symmetry. Each Ln III exhibits a distorted square-antiprismatic coordination geometry. All three Ln centers are arranged in a triangular shape, with Ln−Ln distances ranging from 5.898 to 6.561 Å and an angle of 61.163°. The octahedrally coordinated TM center shows bond lengths from 1.957(19) to 2.083(16) Å, with a terminal water ligand occupying the free metal position (Figure 1 and Tables S13−S28). Powder XRD measurements were performed on the new compounds NaTbSbW 9 ( Figure S19) and KTMLnSbW 9 (TM = Co II , Ni II ; Ln = Tb III , Dy III , Ho III , Er III , Y III ; Figures S20 and S21) and compared to the corresponding simulated spectra, thereby showing the homogeneity of all bulk samples.
Besides XRD, all 13 compounds were characterized by attenuated-total-reflectance IR spectroscopy showing the terminal WO and bridging W−O−W vibrations typical for the Keggin-type POT framework (Figures S1 and S2). The bands at 1541, 1460, and 1409 cm −1 are attributed to the stretching bands of the acetate−carboxylate groups, whereas a band belonging to the bending vibrations of the acetate methyl groups can be observed at 1348 cm −1 in NaLnSbW 9 lacking in KTMLnSbW 9 (Figure S3), which is in accordance to the crystal structures of the compounds.
The number of water molecules in the compounds NaTbSbW 9 ·71H 2 O ( Figure S4 Figure S16) was determined using TGA (Table S3). All compounds exhibit, in general, two to four weight-loss regions that are attributed to losses of water and acetate ligands, respectively. The TGA spectra indicate no major decomposition up to 600°C.
The UV−vis spectra of NaLnSbW 9 ( Figure S22) and KTMLnSbW 9 ( Figure S23) are characterized by an absorption maximum at 272 nm attributed to the p π (O b ) → d π *(W) ligand-to-metal charge transfer typical for the Keggin-type framework. 23 Magnetism. Information storage represents one of the most important uses of magnets nowadays. However, the superparamagnetic size, below which information cannot be permanently stored because magnetization freely fluctuates, poses a lower limit to the size of the memory elements. 24 This occurs at room temperature for particles in the range of 10− 100 nm, owing to the nature of the material. A solution to this problem would be single-molecule magnets (SMMs) being molecules with slow relaxation of their magnetization. The combination of both 3d and 4f metal ions within a heterometallic core represents a promising approach in the preparation of SMMs because the 3d ions can provide significant spin (S), while 4f ions such as Dy III contribute the anisotropy necessary to hinder spin inversion. 25 The magnetic behavior of the heterometallic Dy III -containing members KCoDySbW 9 and KNiDySbW 9 was studied by measuring the direct-current (dc) and alternating-current (ac) magnetic properties of the compounds. To show the importance of the heterometallic core in the compounds, comparative magnetic studies were carried out on KCoYSbW 9 .
The dc magnetic properties of the three representative compounds out of the reported series of POMs KCoDySbW 9 , KNiDySbW 9 , and KCoYSbW 9 are shown in Figure 2 (χ Μ T vs T plot). Magnetic measurements were carried out on microcrystalline powder samples between 2 and 300 K under an applied field of 1000 Oe. The χ Μ T values are 45.06 and 44.74 cm 3 K mol −1 for KCoDySbW 9 and KNiDySbW 9 , respectively, at room temperature. These values are in good agreement with what is expected for three noninteracting Dy III ions (14.17 cm 3 K mol −1 , S = 5 / 2 , 6 H 15/2 , g = 4 / 3 ) 26 and one isolated Co II or Ni II ion (1.877 cm 3 K mol −1 and g = 2 for S = 3 / 2 and 1.00 cm 3 K mol −1 and g = 2 for S = 1). 27 With decreasing temperature, the χ M T products gradually decrease until T = 2.0 K and reach values of 35.98 and 37.27 cm 3 K mol −1 for KCoDySbW 9 and KNiDySbW 9 , respectively, which Inorganic Chemistry pubs.acs.org/IC Article can be associated with the depopulation of MJ (Stark) sublevels of the Dy III centers. 28 In the case of KCoDySbW 9 , an additional contribution to decreasing the temperature dependence of magnetic susceptibility can be associated with the strong spin orbital component characteristic of octahedral Co II ions. 27,29,30 This contribution is present in the heterometallic compound KCoYSbW 9 , in which the strong paramagnetic ions are replaced by the diamagnetic yttrium(III) analogue. At room temperature, the χ Μ T value for KCoYSbW 9 is 3.54 cm 3 K mol −1 (Figure 2), which is significantly higher than the spinonly value for S = 3 / 2 (g = 2). At 2 K, the product of χ Μ T takes a value of 2.08 cm 3 K mol −1 . Low-temperature measurements (2 K) of the field dependence of magnetization exhibit nearsaturation at 5 T, suggesting the presence of a well-isolated ground state in KCoYSbW 9 . For KCoDySbW 9 and KNi-DySbW 9 , the evolution of magnetization versus field has typical shapes and values, which correspond to the presence of three isolated Dy III ions with significant magnetic anisotropy ( Figure S29). 26, 31 The dynamic properties of compounds KCoDySbW 9 ( Figure S24), KNiDySbW 9 ( Figure S26), and KCoYSbW 9 ( Figure S25) were studied by measuring the temperature-and field-dependent ac magnetic susceptibility (Figures 3 and 4). For KCoYSbW 9 , no signal was observed under zero dc field at 2.0 K in the out-of-phase component (χ ac ″) ac susceptibility. After the application of small dc fields (up to 4000 Oe), a frequency-dependent out-of-phase signal becomes visible ( Figure S25A,B). Such behavior can be a consequence of the suppression of quantum tunneling of magnetization (QTM), indicating the presence of field-induced slow magnetic relaxation. The intensity of the out-of-phase signal gradually increases until about 2000 Oe and remains stable. To investigate the nature of this slow magnetic relaxation, the ac susceptibility data were collected under a fixed dc field and stable temperatures between 2.0 and 7.0 K (with a 0.2 K increment; Figure S25D,E). Evaluation of the field and temperature dependence of relaxation times was achieved by fitting the out-of-phase (χ ac ″; eq S3) and in-phase (χ ac ′; eq S2) components of the ac susceptibility using a generalized Debye model for a one-relaxation process. 32,33 The dynamic properties of KCoDySbW 9 and KNiDySbW 9 are similar. At 2.0 K under zero dc field, a signal without a maximum was detected in the out-of-phase component of the ac susceptibility. With a change in the magnetic field, a clear maximum appears after 1500 Oe in the out-of-phase component of the ac susceptibility. The 2500 and 2000 Oe static fields were chosen for the temperature-dependent measurements ( Figures S24D,E and S26D,E). Considering the shapes of the Cole−Cole plots derived from the temperature-and field-dependent measurements (Figures S24C,F and S26C,F), a two-step relaxation process is suggested for KCoDySbW 9 and KNiDySbW 9 and the ac susceptibility data were fitted with generalized Debye equations including a two-step relaxation process (eqs S4 and S5). 34 The two relaxation times (τ 1 and τ 2 ) and distribution parameters (α 1 and α 2 ) occur along with two isothermal susceptibilities (χ T1 and χ T2 ) and one common adiabatic susceptibility (χ s ; eqs S4 and S5). In both compounds, the first relaxation process at low frequency (LF) is well-defined, while the second one at high frequency (HF) has large uncertainty estimates on variable parameters. The distributions of relaxation times for the LF relaxation process is rather broad (α 1 = 0.3−0.5) but well-defined. This is most likely related to the presence of three crystallographically different Dy III ions in the molecular structure (Figure 1), which can exhibit slightly different relaxation properties.
In the case of KCoYSbW 9 , three different mechanisms of relaxation, namely, QTM, direct, and Raman or Orbach mechanisms, respectively, are suggested ( Figure 5). It should be mentioned that the presence of either a Raman or Orbach mechanism, together with QTM and direct mechanisms of relaxation, is sufficient in order to obtain a satisfactory fit of data. For compounds KCoDySbW 9 and KNiDySbW 9 , only the LF signals were analyzed because the second process is poorly defined. In both compounds, QTM is well-defined and indispensable for fitting relaxation data of the LF signal. The   Inorganic Chemistry pubs.acs.org/IC Article second mechanism may be Raman or Orbach with a satisfactory agreement factor (R; Table S29 and Figures S27  and S28).

■ CONCLUSIONS
In summary, a facile synthetic protocol for the directed stepby-step preparation of a 3d−4f heterometallic series of tungstoantimonates via preformed high-nuclear 4f-doped POMs as nonlacunary precursors is reported for the first time. Systematic studies on the reaction system revealed increased K contents as a key step in the directed preparation and isolation of pure KTMLnSbW 9 . The compounds have been characterized thoroughly both in the solid state by singlecrystal and powder XRD, IR spectroscopy, TGA, and elemental analysis as well as in solution by UV−vis spectroscopy. The versatility of the step-by-step protocol was shown by the successful preparation and characterization of isostructural compound KCoYSbW 9 upon application of the synthetic conditions discussed in this work on the literatureknown diamagnetic Y-containing compound. Variable-temperature and variable-field ac susceptibility measurements show that the Dy III -containing representatives of the compound family exhibit field-induced slow magnetic relaxation, which is supplemented by comparative studies on isostructural KCoYSbW 9 lacking the strong paramagnetic Dy III centers. This work shows the importance of the countercations in solution and represents a first step toward the directed preparation of 3d−4f heterometallic POM clusters of a defined composition for magnetic studies by using preformed 4f tungstoantimonates as nonlacunary precursors.
■ ASSOCIATED CONTENT