Gold-Based Coronands as Hosts for M3+ Metal Ions: Ring Size Matters

The controlled, self-assembled synthesis of multinuclear coordination compounds can be performed via different approaches. Frequently, steric, geometric and/or electronic factors located at the ligand systems predefine the way in which metal ions can assemble them to large aggregates. For the compounds in the present paper, also the Pearson’s acidities and preferred coordination geometries of the metal ions were used as organization principles. The ligand under study, 2,6-dipicolinoylbis(N,N-diethylthiourea), H2L1ethyl, possesses ‘soft’ sulfur and ‘hard’ nitrogen and oxygen donors. One-pot reactions of this compound with [AuCl(tht)] (tht = tetrahydrothiophene) and M3+ salts (M = Sc, Y, La, Ln, Ga, In) give products with gold-based {Au3(L1ethyl)3}3+ or {Au2(L1ethyl)2}2+ coronands, which host central M3+ ions. The formation of such units is templated by the M3+ ions and the individual size of the coronand rings is dependent on the ionic radii of the central ions in a way that small ions such as Ga3+ form a [Ga⊂{Au2(L1ethyl)2}]+ assembly, while larger ions (starting from Sc3+/In3+) establish neutral [M⊂{Au3(L1ethyl)3}] units with nine-coordinate central ions.


Introduction
Molecular self-assembly or molecular self-organization are common organization principles in all branches of chemistry and many supramolecular systems found in the biological chemistry or material sciences are products of such processes. The thermodynamic drive to lower energy is regarded as the general principle of such processes, which frequently work with a combination of strong and weak intra-and intermolecular interactions [1][2][3][4][5]. Similar mechanisms apply for metal-based compounds, where mainly coordinative bonds are responsible for the self-assembled formation of supramolecular aggregates. Such systems are of widespread interest in different fields of research, e.g., as molecular nanocontainers, molecular magnets, catalysts and as models (or even mimics) of reactive centers in bioinorganic systems [6][7][8][9][10][11][12][13][14][15][16].
Well-defined supramolecular aggregates are commonly obtained in one-pot reactions by mixing metal precursors and appropriate ligands in defined ratios and suitable solvents. A rational design of the desired products becomes possible by a careful selection of the ligand systems concerning their coordination abilities and inherent constraints defined by the arrangement of the donor atoms, their chelating abilities and/or by geometrical parameters, which control the possible coordination sphere(s) of the metal ion(s). Several successful approaches were reported, e.g., the directional binding approach by P. Stang and coworkers [17], Fujita's molecular paneling approach [18], Raymond's use of symmetry-triggered interactions [19], Mirkin's weak-interaction approach [20], or the use of bimetallic building blocks as was demonstrated by Cotton and coworkers [21]. Well-defined supramolecular compounds with targeted shapes and sizes are accessible following these synthetic principles and two-dimensional assemblies such as molecular triangles or squares [22,23], but also three-dimensional units such as tetrahedral or octahedral cages were reported [24,25]. Similar organization principles are applicable for other large molecular aggregates as in the case of the compounds of the present paper.
In addition to topological aspects of the ligand design and preferred coordination environments of the used metal ions, also simple principles, e.g., Pearson's classical concept of 'hard and soft' acids and bases [26], can be applied to direct metal ions to desired positions in multidentate ligands. This is particularly useful when ligands have 'hard' and 'soft' donor atoms well separated in different positions of their scaffold. Such ligands can be designed on the basis of aroylthioureas, which represent in their most simple form, the benzoylthioureas (HL R ), a bidentate S,O chelator, which forms [M(L R ) 2 ] complexes (type I in Figure 1) with a large variety of transition metal ions [27][28][29][30]. It is interesting to note that corresponding bipodal phthaloylbis(N,N-dialkylthioureas) with two S,O chelating units form multinuclear assemblies with divalent metal ions such as Ni 2+ , Pt 2+ or Cu 2+ . The nuclearity of the resulting complexes strongly depends on the relative positions of the aroylthiourea units. Trinuclear products are formed when two of the acylthiourea units substitute the central phenyl ring in para-positions, while meta substitution (ligands H 2 L0 R in Figure 1) result in dimeric products [31][32][33][34][35][36][37][38].
Molecules 2023, 28, x FOR PEER REVIEW 2 of 30 of the ligand systems concerning their coordination abilities and inherent constraints defined by the arrangement of the donor atoms, their chelating abilities and/or by geometrical parameters, which control the possible coordination sphere(s) of the metal ion(s). Several successful approaches were reported, e.g., the directional binding approach by P. Stang and coworkers [17], Fujita's molecular paneling approach [18], Raymond's use of symmetry-triggered interactions [19], Mirkin's weak-interaction approach [20], or the use of bimetallic building blocks as was demonstrated by Cotton and coworkers [21]. Welldefined supramolecular compounds with targeted shapes and sizes are accessible following these synthetic principles and two-dimensional assemblies such as molecular triangles or squares [22,23], but also three-dimensional units such as tetrahedral or octahedral cages were reported [24,25]. Similar organization principles are applicable for other large molecular aggregates as in the case of the compounds of the present paper.
In addition to topological aspects of the ligand design and preferred coordination environments of the used metal ions, also simple principles, e.g., Pearson's classical concept of 'hard and soft' acids and bases [26], can be applied to direct metal ions to desired positions in multidentate ligands. This is particularly useful when ligands have 'hard' and 'soft' donor atoms well separated in different positions of their scaffold. Such ligands can be designed on the basis of aroylthioureas, which represent in their most simple form, the benzoylthioureas (HL R ), a bidentate S,O chelator, which forms [M(L R )2] complexes (type I in Figure 1) with a large variety of transition metal ions [27][28][29][30]. It is interesting to note that corresponding bipodal phthaloylbis(N,N-dialkylthioureas) with two S,O chelating units form multinuclear assemblies with divalent metal ions such as Ni 2+ , Pt 2+ or Cu 2+ . The nuclearity of the resulting complexes strongly depends on the relative positions of the aroylthiourea units. Trinuclear products are formed when two of the acylthiourea units substitute the central phenyl ring in para-positions, while meta substitution (ligands H2L0 R in Figure 1) result in dimeric products [31][32][33][34][35][36][37][38].  [27][28][29][30].
Although the bi-or trinuclear complexes of the composition [M II (meta-L0 R )] 2 (see Figure 1) or [M II (para-L0 R )] 3 [31][32][33] possess large central voids, these voids are empty and cannot accommodate any guest atoms or molecules. No suitable donor atoms are available and the effective space is restricted by the hydrogen atoms of the central phenyl rings pointing toward the centers. The replacement of the central phenyl ring in ligand H 2 L0 R by a pyridine ring or the introduction of ether units in the side chains results in the formation of the ligand systems H 2 L1 R and H 2 L2 R , which have such donor atoms available.
Moreover, they represent a class of ligands with bifunctional coordination sites (moderately 'soft' S,O and 'medium hard' O,N,O or 'hard' O,O,O,O donors), each of which favorably binds to a particular type of metal ions. This gives access to the syntheses of a large variety of multinuclear, bimetallic complexes (types II, II, IV, VI and VII in Figure 1). They are usually formed by simple one-pot reactions in solutions containing the ligands and mixtures of two metal ions with different Pearson's acidity [39][40][41][42][43][44][45][46][47]. Particularly interesting bonding features are established with 'soft' metal ions such as Ag + or Au + , where no chelate formation was found, but a monodentate coordination to the sulfur atoms of the ligands is observed. As a result, macrocyclic, coronand-type ligands were formed, which can accommodate one or two alkaline earth metal ions [30,41]. The ready formation of such assemblies raises questions concerning the mechanism of their formation, the role of the (central) metal ion(s) and if the resulting ring size is exclusively determined by the used precursors H 2 L1 R or H 2 L2 R .
In the present study, we report the results of reactions of the pyridine-based ligand H 2 L1 ethyl , the common Au(I) precursor [AuCl(tht)] (tht = tetrahydrothiophene) and salts of M 3+ ions. The trivalent metal ions were taken from group 3, group 13 of the periodic table and from the lanthanide series.

General Synthetic Aspects
One-pot reactions of each three equivalents of [AuCl(tht)] (tht = tetrahydrothiophene) and H 2 L1 ethyl , and one equivalent of the chlorides or nitrates of the trivalent metal ions shown in the left part of Scheme 1 in methanol give tetranuclear complexes of the composition [M⊂{Au 3 (L1 ethyl ) 3 }]. The use of labile gold(I) starting material such as [AuCl(tht)] is essential for the isolation of the products in good yields and high purity. Although the use of (NEt 4 )[Au III Cl 4 ] also results in the formation of [M ⊂ {Au 3 (L1 ethyl ) 3 }] compounds, the products obtained in this way contain a significant amount of less-defined side-products resulting from the reduction of the Au(III) precursor, which are hard to remove from the host/guest compounds. Attempted reactions with [AuCl(PPh 3 )] instead of [AuCl(tht)] did not form the mixed-metal assemblies, most probably due to the stability of the Au-P bond in this precursor (vide infra).
Moreover, they represent a class of ligands with bifunctional coordination sites (moderately 'soft' S,O and 'medium hard ' O,N,O or 'hard' O,O,O,O donors), each of which favorably binds to a particular type of metal ions. This gives access to the syntheses of a large variety of multinuclear, bimetallic complexes (types II, II, IV, VI and VII in Figure  1). They are usually formed by simple one-pot reactions in solutions containing the ligands and mixtures of two metal ions with different Pearson's acidity [39][40][41][42][43][44][45][46][47]. Particularly interesting bonding features are established with 'soft' metal ions such as Ag + or Au + , where no chelate formation was found, but a monodentate coordination to the sulfur atoms of the ligands is observed. As a result, macrocyclic, coronand-type ligands were formed, which can accommodate one or two alkaline earth metal ions [30,41]. The ready formation of such assemblies raises questions concerning the mechanism of their formation, the role of the (central) metal ion(s) and if the resulting ring size is exclusively determined by the used precursors H2L1 R or H2L2 R .
In the present study, we report the results of reactions of the pyridine-based ligand H2L1 ethyl , the common Au(I) precursor [AuCl(tht)] (tht = tetrahydrothiophene) and salts of M 3+ ions. The trivalent metal ions were taken from group 3, group 13 of the periodic table and from the lanthanide series.

General Synthetic Aspects
One-pot reactions of each three equivalents of [AuCl(tht)] (tht = tetrahydrothiophene) and H2L1 ethyl , and one equivalent of the chlorides or nitrates of the trivalent metal ions shown in the left part of Scheme 1 in methanol give tetranuclear complexes of the composition [M⊂{Au3(L1 ethyl )3}]. The use of labile gold(I) starting material such as [AuCl(tht)] is essential for the isolation of the products in good yields and high purity. Although the use of (NEt4)[Au III Cl4] also results in the formation of [M ⊂ {Au3(L1 ethyl )3}] compounds, the products obtained in this way contain a significant amount of less-defined side-products resulting from the reduction of the Au(III) precursor, which are hard to remove from the host/guest compounds. Attempted reactions with [AuCl(PPh3)] instead of [AuCl(tht)] did not form the mixed-metal assemblies, most probably due to the stability of the Au-P bond in this precursor (vide infra).
The selection of the M 3+ precursors is not crucial as long as the used salts are sufficiently soluble in methanol. The use of the corresponding chlorides and nitrates is perfectly suitable and the anions do not have any influence on the obtained yields. Alternatively, also labile chelate complexes of the M 3+ ions (e.g., acetylacetonates) can be used as precursors. This was proven with [Sc(acac)3] and [Ga(acac)3]. The corresponding one-pot reactions give pure products in similar yields as reactions starting from salts. For lanthanides, this approach was not attempted since the reactions starting from the nitrates or chlorides gave good yields and the products precipitated in high purity. The selection of the M 3+ precursors is not crucial as long as the used salts are sufficiently soluble in methanol. The use of the corresponding chlorides and nitrates is perfectly suitable and the anions do not have any influence on the obtained yields. Alternatively, also labile chelate complexes of the M 3+ ions (e.g., acetylacetonates) can be used as precursors. This was proven with [Sc(acac) 3 ] and [Ga(acac) 3 ]. The corresponding one-pot reactions give pure products in similar yields as reactions starting from salts. For lanthanides, this approach was not attempted since the reactions starting from the nitrates or chlorides gave good yields and the products precipitated in high purity.
The [M⊂{Au 3 (L1 ethyl ) 3 }] complexes 1-17 precipitate directly from the reaction mixtures upon stirring the components at room temperature, either directly or after the addition of a few drops of triethylamine as a supporting base. For the isolation of [Ga⊂{Au 2 (L1 ethyl ) 2 }](BF 4 ), which was prepared from [Ga(acac) 3 ], the addition of one equivalent of KBF 4 was required. The products are readily soluble in CH 2 Cl 2 or CHCl 3 . Single crystals were commonly produced by slow diffusion of MeOH or diethyl ether into CH 2 Cl 2 solutions of the complexes. The products were characterized by elemental analysis, IR and UV-vis spectroscopy, mass spectrometry and (where appropriate) by NMR spectroscopy.
The first information about the success of the reactions can be derived from two indicative features in the IR spectra of the products: (1) double deprotonation of the organic ligands can be concluded from the disappearance of the NH stretches of the ligands in the region above 3200 cm −1 and (2) the C=O bands, which appears at 1674 cm −1 in the uncoordinated H 2 L1 ethyl [42], are bathochromically shifted in the spectra of the complexes by approximately 100 cm −1 . This is a clear sign for the formation of a delocalized π-system upon coordination. Stronger shifts by more than 120 cm −1 are observed when the aroylthiourea ligands are involved in O,S-chelate formation [47]. More structural and spectroscopic features of the complexes will be discussed within the group of metals in the following sections.
The formation of large ring systems with two, three or four [30] {Au(L1 ethyl )} − units is not intuitive and requires an explanation. As already mentioned above, the selection of reactive gold starting materials is important and the use of [AuCl(tht)] with an only weakly bonded thioether ligand proved to be optimal. Corresponding one-pot reactions between H 2 L1 ethyl , [AuCl(PPh 3  addition of a few drops of triethylamine as a supporting base. For the isolation of [Ga⊂{Au2(L1 ethyl )2}](BF4), which was prepared from [Ga(acac)3], the addition of one equivalent of KBF4 was required. The products are readily soluble in CH2Cl2 or CHCl3. Single crystals were commonly produced by slow diffusion of MeOH or diethyl ether into CH2Cl2 solutions of the complexes. The products were characterized by elemental analysis, IR and UV-vis spectroscopy, mass spectrometry and (where appropriate) by NMR spectroscopy.
The first information about the success of the reactions can be derived from two indicative features in the IR spectra of the products: (1) double deprotonation of the organic ligands can be concluded from the disappearance of the NH stretches of the ligands in the region above 3200 cm -1 and (2) the C=O bands, which appears at 1674 cm -1 in the uncoordinated H2L1 ethyl [42], are bathochromically shifted in the spectra of the complexes by approximately 100 cm -1 . This is a clear sign for the formation of a delocalized π-system upon coordination. Stronger shifts by more than 120 cm -1 are observed when the aroylthiourea ligands are involved in O,S-chelate formation [47]. More structural and spectroscopic features of the complexes will be discussed within the group of metals in the following sections.
The formation of large ring systems with two, three or four [30] {Au(L1 ethyl )} -units is not intuitive and requires an explanation. As already mentioned above, the selection of reactive gold starting materials is important and the use of [AuCl(tht)] with an only weakly bonded thioether ligand proved to be optimal. Corresponding one-pot reactions between H2L1 ethyl , [AuCl(PPh3)] and M 3+ ions gave only mixtures of poorly defined products. Therefore, we performed reactions of H2L1 ethyl with [AuCl(tht)] or [AuCl(PPh3)] in absence of additional metal ions (Scheme 2).  (20). They precipitate as colorless solids from the reaction mixtures either after a short heating period (19) or after the addition of NEt3 (20). The necessity of a supporting base for the latter reaction arises from the required deprotonation of the ligand, which is not the case during the formation of the chlorido complex 19. It becomes evident that in both reactions only the sulfur atoms of 2,6-dipicolinoylbis(N,N-diethylthiourea) are involved in the coordination and neutral, linear gold(I) complexes are formed. Charge compensation in complex 20 is achieved by a double deprotonation of H2L1 ethyl , which can readily be seen in the IR spectrum of the compound, where the ν(NH) stretch (3302 cm -1 in 19) is absent. The 1 H NMR spectrum of 19 provides additional evidence for the proposed composition. It shows a singlet related to the NH proton at 10.83 ppm. One doublet and one triplet at 8.42 and 8.20 ppm are assigned to the protons of the pyridine ring. Two quartets due to the CH2 protons (4.00 and 3.75 ppm) and two triplets of the CH3 protons (1. 49 (20). They precipitate as colorless solids from the reaction mixtures either after a short heating period (19) or after the addition of NEt 3 (20). The necessity of a supporting base for the latter reaction arises from the required deprotonation of the ligand, which is not the case during the formation of the chlorido complex 19. It becomes evident that in both reactions only the sulfur atoms of 2,6-dipicolinoylbis(N,N-diethylthiourea) are involved in the coordination and neutral, linear gold(I) complexes are formed. Charge compensation in complex 20 is achieved by a double deprotonation of H 2 L1 ethyl , which can readily be seen in the IR spectrum of the compound, where the ν(NH) stretch (3302 cm −1 in 19) is absent. The 1 H NMR spectrum of 19 provides additional evidence for the proposed composition. It shows a singlet related to the NH proton at 10.83 ppm. One doublet and one triplet at 8.42 and 8.20 ppm are assigned to the protons of the pyridine ring. Two quartets due to the CH 2 protons (4.00 and 3.75 ppm) and two triplets of the CH 3 protons (1. 49 (20). Figure 2 depicts the molecular structures of both compounds. Although the H 2 L1 ethyl or {L1 ethyl } − ligands are only bonded to gold via their sulfur atoms, the coordination results in some bond length modifications in the skeleton of the ligands with regard to the non-coordinated form. The corresponding values are summarized in Table 1. The values clearly show that the C-S bonds are lengthened upon coordination irrespective of the deprotonation of the ligand. Furthermore, a considerable bond length equalization is observed for the C-N bonds. This is a common result for this type of ligands and is explained by the formation of an extended π-system as the consequence of the chelate formation [27][28][29]. In the present case, it is found for monodentate coordinated aroylthioureas as well as for non-deprotonated ligands of this type.
Molecules 2023, 28, x FOR PEER REVIEW 5 of 30 The spectroscopic information could be confirmed by single-crystal X-ray diffraction studies on [(AuCl)2(H2L1 ethyl )] (19) and [{Au(PPh3)}2(L1 ethyl )] (20). Figure 2 depicts the molecular structures of both compounds. Although the H2L1 ethyl or {L1 ethyl } -ligands are only bonded to gold via their sulfur atoms, the coordination results in some bond length modifications in the skeleton of the ligands with regard to the non-coordinated form. The corresponding values are summarized in Table 1. The values clearly show that the C-S bonds are lengthened upon coordination irrespective of the deprotonation of the ligand. Furthermore, a considerable bond length equalization is observed for the C-N bonds. This is a common result for this type of ligands and is explained by the formation of an extended π-system as the consequence of the chelate formation [27][28][29]. In the present case, it is found for monodentate coordinated aroylthioureas as well as for non-deprotonated ligands of this type.   [42].
Au1-S1 S1-C12 C12-N12 C12-N11 N11-C11 Au2-S2 S2-C22 C22-N22 C22-N21 N21-C21 19 2.266(1) 1.710(4) 1.305 (5)  The Au-Cl and Au-S distances in compound 19 are very similar to those observed in the structure of N,N-diethyl-N'-camphanylthioureatogold(I) chloride obtained by Koch [48] and N,N-diethyl-N'-benzoylthioureatogold(I) chloride reported by Bensch [49]. Weak Au-Au interactions of 3.5493(5) Å are established in the solid-state structure of 19. They are accompanied by π-π interactions between the pyridine rings of two adjacent molecules of 3.709(1) Å as is illustrated in Figure 2a. The values of the Au···Au as well as the offset parallel orientation of the π-π interactions are consistent with the values determined by Pathaneni [50] and Janiak [51], respectively. The shortest Au-Au distance in compound 20 is 9.535(8) Å. Thus, no bonding interactions can be discussed. The solid state structure of this compound exhibits a perfect disorder of the carbonyl groups and pyridine ring over two positions. A similar disorder is observed in another gold(I) complex with an isophtaloylbis(thiourea) ligand [52]. As in the chlorido complex 19, the gold atoms in 20 are in an almost linear environment with a P-Au-S angle of 171.4(1)°. The Au-P distances of 2.267(3) Å and the Au-S bond lengths of 2.296(3) Å are similar to the values in the structures published by Schwade [52] and Gimeno [53].
The ready formation of [(AuCl)2(H2L ethyl )] (19) during the reaction of [AuCl(tht)] with H2L1 ethyl without any sign for the parallel formation of cyclic units as well as the dependence of the obtained ring sizes on the template ion (Scheme 1 and [30]) strongly indicate a mechanism involving templating for the formation of the title compounds.   (20) complexes and in the uncoordinated ligand H 2 L1 ethyl [42]. The Au-Cl and Au-S distances in compound 19 are very similar to those observed in the structure of N,N-diethyl-N -camphanylthioureatogold(I) chloride obtained by Koch [48] and N,N-diethyl-N -benzoylthioureatogold(I) chloride reported by Bensch [49]. Weak Au-Au interactions of 3.5493(5) Å are established in the solid-state structure of 19. They are accompanied by π-π interactions between the pyridine rings of two adjacent molecules of 3.709(1) Å as is illustrated in Figure 2a. The values of the Au···Au as well as the offset parallel orientation of the π-π interactions are consistent with the values determined by Pathaneni [50] and Janiak [51], respectively. The shortest Au-Au distance in compound 20 is 9.535(8) Å. Thus, no bonding interactions can be discussed. The solid state structure of this compound exhibits a perfect disorder of the carbonyl groups and pyridine ring over two positions. A similar disorder is observed in another gold(I) complex with an isophtaloylbis(thiourea) ligand [52]. As in the chlorido complex 19, the gold atoms in 20 are in an almost linear environment with a P-Au-S angle of 171.4(1) • . The Au-P distances of 2.267(3) Å and the Au-S bond lengths of 2.296(3) Å are similar to the values in the structures published by Schwade [52] and Gimeno [53].

Lanthanum and the Lanthanide Series
The ready formation of [(AuCl) 2 (H 2 L ethyl )] (19) during the reaction of [AuCl(tht)] with H 2 L1 ethyl without any sign for the parallel formation of cyclic units as well as the dependence of the obtained ring sizes on the template ion (Scheme 1 and [30]) strongly indicate a mechanism involving templating for the formation of the title compounds.

Lanthanum and the Lanthanide Series
Chlorides or nitrates of the lanthanide elements readily undergo the reactions illustrated in Scheme 1 and described above. The resulting [M⊂{Au 3 (L1 ethyl ) 3 }] complexes precipitate from the reaction mixtures essentially as colorless solids. Only the cerium Molecules 2023, 28, 5421 6 of 29 complex has an orange-red color, while the holmium one is pale pink. The IR spectra of the products show the features described in the previous section and allow for a fast assessment of the colorless solids obtained.
NMR spectra of sufficient quality were recorded for the diamagnetic lanthanum complex. Deprotonation of the organic ligand is indicated by the absence of NH signals, which are found in the spectrum of uncoordinated H 2 L1 ethyl at 9 ppm. The protons of the methylene groups are not equivalent and appear as three multiplets at 3.7, 3.5 and 3.4 ppm. This non-equivalence is due to the rigidity of the formed assemblies and hindered rotation around the C-NEt 2 bonds. Similar effects were found for other complexes with {L1 ethyl } 2− ligands [39][40][41][42][43], but are also typical for chelate complexes with the parent benzoylthioureato ligands, where a rotational barrier of approximately 650 kJ/mol was found [54]. The 13 C NMR spectrum shows the signals in the expected regions. The C=O and C=S signals appear in the complexes at 185.5 and 159.3 ppm, while they are found in the spectrum of H 2 L1 ethyl at 177.2 and 156.3 ppm, respectively. The observed downfield shift of the signals upon coordination is not unusual for the class of ligands under study.
An indicative method for the characterization of the complexes is mass spectrometry. ESI+ mass spectra of the host/guest complexes show the molecular peaks of the bimetallic [M⊂{Au 3 (L1 ethyl ) 3 } + H] + assemblies with high intensity. They are frequently accompanied by the corresponding [M⊂{Au 3 (L1 ethyl ) 3 } + Na] + and/or [M⊂{Au 3 (L1 ethyl ) 3 } + K] + cluster ions. Figure 3 shows the spectrum of the lanthanum complex together with the simulation of the isotopic pattern of the molecular ion, which perfectly matches the experimental one. The mass spectra of the other compounds can be found in the Supplementary Materials. It is interesting to note that in almost all spectra of such compounds, fragments of low intensities are observed, which can be assigned to a subsequent loss of {Au(L1 ethyl } units, while the lanthanide ions remain bonded. This type of fragmentation is doubtlessly attributed to the high-energy conditions inside the mass spectrometer, but might be regarded as a hint for general degradation pathways for such compounds under thermal conditions. Chlorides or nitrates of the lanthanide elements readily undergo the reactions illustrated in Scheme 1 and described above. The resulting [M⊂{Au3(L1 ethyl )3}] complexes precipitate from the reaction mixtures essentially as colorless solids. Only the cerium complex has an orange-red color, while the holmium one is pale pink. The IR spectra of the products show the features described in the previous section and allow for a fast assessment of the colorless solids obtained.
NMR spectra of sufficient quality were recorded for the diamagnetic lanthanum complex. Deprotonation of the organic ligand is indicated by the absence of NH signals, which are found in the spectrum of uncoordinated H2L1 ethyl at 9 ppm. The protons of the methylene groups are not equivalent and appear as three multiplets at 3.7, 3.5 and 3.4 ppm. This non-equivalence is due to the rigidity of the formed assemblies and hindered rotation around the C-NEt2 bonds. Similar effects were found for other complexes with {L1 ethyl } 2ligands [39][40][41][42][43], but are also typical for chelate complexes with the parent benzoylthioureato ligands, where a rotational barrier of approximately 650 kJ/mol was found [54]. The 13 C NMR spectrum shows the signals in the expected regions. The C=O and C=S signals appear in the complexes at 185.5 and 159.3 ppm, while they are found in the spectrum of H2L1 ethyl at 177.2 and 156.3 ppm, respectively. The observed downfield shift of the signals upon coordination is not unusual for the class of ligands under study.
An indicative method for the characterization of the complexes is mass spectrometry. ESI+ mass spectra of the host/guest complexes show the molecular peaks of the bimetallic [M⊂{Au3(L1 ethyl )3} + H] + assemblies with high intensity. They are frequently accompanied by the corresponding [M⊂{Au3(L1 ethyl )3} + Na] + and/or [M⊂{Au3(L1 ethyl )3} + K] + cluster ions. Figure 3 shows the spectrum of the lanthanum complex together with the simulation of the isotopic pattern of the molecular ion, which perfectly matches the experimental one. The mass spectra of the other compounds can be found in the Supplementary Materials. It is interesting to note that in almost all spectra of such compounds, fragments of low intensities are observed, which can be assigned to a subsequent loss of {Au(L1 ethyl } units, while the lanthanide ions remain bonded. This type of fragmentation is doubtlessly attributed to the high-energy conditions inside the mass spectrometer, but might be regarded as a hint for general degradation pathways for such compounds under thermal conditions. The availability of a large number of lanthanide complexes with a presumably similar or identical composition makes it interesting to study the solid-state structures of the compounds. Therefore, we attempted to grow single crystals of the products for an X-ray study. Eventually, we obtained single crystals for almost the entire series with the exception of the complex with the radioactive promethium. The compounds could be crystallized by a careful covering of CH2Cl2 solutions with n-hexane and subsequent diffusion of the solvents. Irrespective of the central M 3+ ion, the compounds have a composition of [M⊂{Au3(L1 ethyl )3}] with a coordination number of nine for the central M 3+ ions. This is The availability of a large number of lanthanide complexes with a presumably similar or identical composition makes it interesting to study the solid-state structures of the compounds. Therefore, we attempted to grow single crystals of the products for an X-ray study. Eventually, we obtained single crystals for almost the entire series with the exception of the complex with the radioactive promethium. The compounds could be crystallized by a careful covering of CH 2 Cl 2 solutions with n-hexane and subsequent diffusion of the solvents. Irrespective of the central M 3+ ion, the compounds have a composition of [M⊂{Au 3 (L1 ethyl ) 3 }] with a coordination number of nine for the central M 3+ ions. This is remarkable considering they encompass ions with radii ranging from 1.216 Å (La 3+ ) to 1.032 Å (Lu 3+ ) [55]. Figure 4 shows the molecular structures of the lanthanum and lutetium complexes as representatives of the full series. The La 3+ and Lu 3+ ions are depicted with their respective ionic radii and it becomes evident that the sizes of these ions do not have a significant influence on how they are hosted by the {Au 3 (L1 ethyl ) 3 } 3− coronands. This is a clear sign that the 36-membered {Au 3 (L1 ethyl ) 3 } 3− ring system is flexible over a wide range for the coordination of central metal ions. The structures of the other compounds are similar and are depicted in the Supplementary Materials. remarkable considering they encompass ions with radii ranging from 1.216 Å (La 3+ ) to 1.032 Å (Lu 3+ ) [55]. Figure 4 shows the molecular structures of the lanthanum and lutetium complexes as representatives of the full series. The La 3+ and Lu 3+ ions are depicted with their respective ionic radii and it becomes evident that the sizes of these ions do not have a significant influence on how they are hosted by the {Au3(L1 ethyl )3} 3-coronands. This is a clear sign that the 36-membered {Au3(L1 ethyl )3} 3-ring system is flexible over a wide range for the coordination of central metal ions. The structures of the other compounds are similar and are depicted in the Supplementary Material. Although the coordination environments of the central M 3+ ions in the [M⊂{Au3(L1 ethyl )3}] complexes seem to be independent of their ionic radii, clear changes in the M-N and M-O bond lengths are found. They decrease almost steadily from the lanthanum to the lutetium compound, following the well-documented lanthanide contraction as is shown in Figure 4c, where they are depicted along with the radii of the central ions for coordination number nine [55]. A summary of selected bond lengths and angles is given in Table 2. Some anomalies become evident with the three elements with the smallest ionic radii (thulium, ytterbium and lutetium), where differences between the individual M-O bond lengths are apparently larger (Tm, Lu) or smaller (Yb) than in the other compounds. On the basis of the available data, however, it is not justified to attribute a single reason to this observation, especially considering that unlike all other [M⊂{Au3(L1 ethyl )3}] complexes in this series, the thulium and lutetium compounds do not crystallize in the trigonal space group 3 . The M-N bonds are not affected to the same extent. Although the coordination environments of the central M 3+ ions in the [M⊂{Au 3 (L1 ethyl ) 3 }] complexes seem to be independent of their ionic radii, clear changes in the M-N and M-O bond lengths are found. They decrease almost steadily from the lanthanum to the lutetium compound, following the well-documented lanthanide contraction as is shown in Figure 4c, where they are depicted along with the radii of the central ions for coordination number nine [55]. A summary of selected bond lengths and angles is given in Table 2. Some anomalies become evident with the three elements with the smallest ionic radii (thulium, ytterbium and lutetium), where differences between the individual M-O bond lengths are apparently larger (Tm, Lu) or smaller (Yb) than in the other compounds. On the basis of the available data, however, it is not justified to attribute a single reason to this observation, especially considering that unlike all other [M⊂{Au 3 (L1 ethyl ) 3 }] complexes in this series, the thulium and lutetium compounds do not crystallize in the trigonal space group P3. The M-N bonds are not affected to the same extent. It is also interesting to note that the M 3+ -O and M 3+ -N bond lengths in the goldcontaining coronates of the present study are remarkably short compared with another series of nine-coordinate lanthanide chelates with {L1 ethyl } 2− ligands: trinuclear cobalt(II) complexes of the composition [Co 2 Ln(L1 ethyl ) 2 (µ-acetate) 2 Cl] with Ln = Ce, Nd, Sm, Dy, Er, Yb [43]. The M 3+ ions in the latter compounds are bonded by the N and O donor atoms of each two {L1 ethyl } 2− chelators and additional acetato and chlorido ligands (see compounds II in Figure 1). Their M 3+ -N bonds also regularly decrease within the lanthanides series, but span a range from 2.650 (3)  Considerable progress in the exploration of the fundamental chemistry of lanthanides was achieved in recent years [57][58][59][60], and specific properties of their compounds make them interesting for applications particularly in fields such as catalysis [61][62][63][64][65][66][67][68][69][70], optical [71][72][73][74] and magnetic materials [75][76][77], but also in pharmacy and life science [78][79][80][81].
Although the present series of compounds would give a unique opportunity for a comprehensive comparative study, it cannot be the aim of the present, introducing paper to cover all these interesting issues of lanthanide chemistry. Some magnetic properties of lanthanide/transition metal complexes containing {L1 ethyl } 2− ligands were published recently, showing clear differences in the temperature dependence of the magnetic susceptibility for the different lanthanide ions [43]. Gold(I) and gold(III) complexes found increasing interest in medicinal chemistry for cancer treatment [82][83][84][85][86][87] and, recently, the potential use of self-assembled host-guest compounds was addressed for such applications [87][88][89][90]. Consequently, also the bimetallic [M⊂{Au3(L1 ethyl )3}] complexes of the present study may have a considerable potential for such purposes, particularly when it suc- Considerable progress in the exploration of the fundamental chemistry of lanthanides was achieved in recent years [57][58][59][60], and specific properties of their compounds make them interesting for applications particularly in fields such as catalysis [61][62][63][64][65][66][67][68][69][70], optical [71][72][73][74] and magnetic materials [75][76][77], but also in pharmacy and life science [78][79][80][81]. Although the present series of compounds would give a unique opportunity for a comprehensive comparative study, it cannot be the aim of the present, introducing paper to cover all these interesting issues of lanthanide chemistry. Some magnetic properties of lanthanide/transition metal complexes containing {L1 ethyl } 2− ligands were published recently, showing clear differences in the temperature dependence of the magnetic susceptibility for the different lanthanide ions [43]. Gold(I) and gold(III) complexes found increasing interest in medicinal chemistry for cancer treatment [82][83][84][85][86][87] and, recently, the potential use of self-assembled host-guest compounds was addressed for such applications [87][88][89][90]. Consequently, also the bimetallic [M⊂{Au 3 (L1 ethyl ) 3 }] complexes of the present study may have a considerable potential for such purposes, particularly when it succeeds to increase the cytotoxic activity of the gold compounds either through cytotoxic effects induced by the guest metal ions or by ionizing radiation of appropriate β − -emitting isotopes (e.g., 198 Au, 177 Lu, 90 Y, 153 Sm, 149 Tb). Suitable radioactive gamma or positron emitters (e.g., 44 Sc, 68 Ga, 111 In) can also be used to monitor the therapeutic progress. The first studies on the evaluation of [M⊂{Au 3 (L1 R ) 3 }] complexes are currently underway [91] and shall not be treated more in detail in the present communication. Instead, we briefly want to describe exemplarily the luminescent properties of the europium compound.
In a recent review, the fluorescence properties of Eu 3+ complexes were discussed with regard to the influence of the site symmetry in the compounds and used to derive structural information about the coordination geometry around the metal ion [92]. It is interesting to compare the derived information with the results of the single-crystal X-ray structure of the compound. In order to reach the necessary resolution of the single transitions, a grating of 1200 mm −1 was chosen. Figure 6a shows the emission spectrum of [Eu⊂{Au 3 (L1 ethyl ) 3 }] measured in CH 2 Cl 2 at room temperature with a grating of 1200 mm −1 . The europium(III) spectrum is composed of the red luminescence of the Eu 3+ ion, which results from transitions from the first excited state ( 5 D 0 ) to all the lower J levels of the ground term 7 F J (J = 0-4) [93]. Emission bands can be observed at 16835, 16207, 15290 and 14205 cm −1 , which are attributed to the f-f transitions 5 D 0 → 7 F 1 , 5 D 0 → 7 F 2, 5 D 0 → 7 F 3 and 5 D 0 → 7 F 4 , respectively. An additional band is observed in the region of the 5 D0→ 7 F0 transition, but its intensity is very low and, thus, this transition can be regarded as forbidden. Transitions arising from 5 D1 are not observed or they are too weak for an interpretation.
In the first and most intense transition, 5 D0→ 7 F1, the selection rules forbid an electricdipole, but allow a magnetic-dipole transition [95]. It consists of two strong lines, where the lower frequency is split into two components. The 5 D0→ 7 F2 transition is a so-called "hypersensitive transition", the intensity of which is markedly influenced by the environment of the europium(III) ion and the nature of the ligand [96]. In this case, we observed a relatively intense band with the higher frequency being split into two lines. A shoulder is observed in the second one. Its high intensity can be attributed to a low symmetry around the Eu 3+ ion or the presence of highly polarizable ligands. Similar effects were found before, especially in compounds with chelating rings, such as β-diketonates or 2,6pyridinedicarboxylato ligands [96], which are similar to H2L1 ethyl .
No band is observed in the region of 5 D0→ 7 F3 transitions, while two broad bands of medium intensity appear in the region of the 5 D0→ 7 F4 transition. The highest frequency band is split into five lines, although it is quite difficult to determine their exact positions. The second one is split into two lines.
The derived relatively low symmetry of the Eu 3+ ion in [M⊂{Au3(L1 ethyl )3}] is in ac- An additional band is observed in the region of the 5 D 0 → 7 F 0 transition, but its intensity is very low and, thus, this transition can be regarded as forbidden. Transitions arising from 5 D 1 are not observed or they are too weak for an interpretation.
In the first and most intense transition, 5 D 0 → 7 F 1 , the selection rules forbid an electricdipole, but allow a magnetic-dipole transition [95]. It consists of two strong lines, where the lower frequency is split into two components. The 5 D 0 → 7 F 2 transition is a so-called "hypersensitive transition", the intensity of which is markedly influenced by the environment of the europium(III) ion and the nature of the ligand [96]. In this case, we observed a relatively intense band with the higher frequency being split into two lines. A shoulder is observed in the second one. Its high intensity can be attributed to a low symmetry around the Eu 3+ ion or the presence of highly polarizable ligands. Similar effects were found before, especially in compounds with chelating rings, such as β-diketonates or 2,6-pyridinedicarboxylato ligands [96], which are similar to H 2 L1 ethyl .
No band is observed in the region of 5 D 0 → 7 F 3 transitions, while two broad bands of medium intensity appear in the region of the 5 D 0 → 7 F 4 transition. The highest frequency band is split into five lines, although it is quite difficult to determine their exact positions. The second one is split into two lines.
The derived relatively low symmetry of the Eu 3+ ion in [M⊂{Au 3 (L1 ethyl ) 3 }] is in accordance with the experimental X-ray data. An analysis of the coordination polyhedron around europium with the continuous shape measures algorithm [97,98] gives no clear preference of one of the idealized coordination polyhedra for coordination number nine [94]. The bonding situation in the solid state can best be described as lying in between a capped square antiprism, a tricapped trigonal prism and a muffin structure, and a fluxional behavior in solution cannot be ruled out. A more detailed, comparative evaluation of the coordination polyhedra for the entire series of nine-coordinate [M⊂{Au 3 (L1 ethyl ) 3 }] complexes will be given vide infra.

Group 3 and Group 13 Elements
The formation of isostructural nine-coordinate compounds having the same donor atom constellation, which are recruited from one single, self-assembled ring system over the entire lanthanide series is unique. It became evident that the formation of such aggregates proceeds via a templated reaction, which requires effective interactions between the donor atoms and the central M 3+ ions during their formation. The bond length values of Table 2 and particularly some anomalies found for the smaller ions suggest that the size of the M 3+ ion seems to be important for the optimization of the metal-donor atom interactions. This raises the questions of the limiting factors for the formation of such nine-coordinate assemblies and what happens when [M⊂{Au 3 (L1 ethyl ) 3 }] complexes can no longer be stabilized. The latter question also becomes interesting in the light of the fact that divalent metal ions such as Ca 2+ or Ba 2+ form large {Au 4 (L1 ethyl ) 4 } 4− rings, which can accommodate two alkaline earth ions [30]. Good candidates to answer these questions might be given with the ions of the group 3 and group 13 elements having ion radii smaller than that of lutetium: e.g., Sc 3+ (0.87 Å for coordination number C.N. 8), In 3+ (0.92 Å for C.N. 8), Ga 3+ (0.62 Å for C.N. 6). Additionally, the group 3 element yttrium with an ionic radius of 1.075 Å for C.N. 9 seem to be a good supplement to the values obtained for the lanthanides series given in Figure 2c and Table 1. Indeed, the bonding situation in [Y⊂{Au 3 (L1 ethyl ) 3 Table 3) is very similar to those in the lanthanides compounds with ionic radii close to that of Y 3+ (Ho 3+ : 1.072 Å and Dy 3+ : 1.082 Å). On the other hand, the very small Ga 3+ ion (unlike the other M 3+ ions regarded in this study) was not able to establish bonds with the nine donor atoms provided by a {Au 3 (L1 ethyl ) 3 } 3− ring system. Such a result is not completely surprising, considering the fact that the gallium complexes with coordination numbers larger than six are extremely rare. Apart from compounds with π-bonded rings, carbaboranes and cluster compounds, there are only a few entries of such compounds in the Cambridge Structural Database [99]. These examples involve polymeric carboxylates [100,101], crown ethers or cryptands [102]. It should be mentioned that the latter compounds contain gallium in its oxidation state "+1". Reactions of common Ga(III) starting materials (Ga(NO 3 [30]. This suggests that the energetically favored interaction for the Ga 3+ ions involves "chelate formation" with the 24-membered {Au 2 (L1 ethyl ) 2 } 2− ring system, rather than charge compensation with ligands such as nitrate.

}] (see
Crystalline materials with nitrate or BF 4 − counter ions were obtained directly from the reaction mixture in the case of [Ga⊂{Au 2 (L1 ethyl ) 2 }](NO 3 ) (18a) or after the addition of KBF 4 to the reaction mixture with [Ga(acac) 3 ] (18b). The structures of the complex cations are not influenced by their counter ions and the structure of the nitrate salt is shown in Figure 7. Selected bond lengths and angles for both salts of the six-coordinate gallium complex are summarized in Table 3 3+ and In 3+ ions are adopted from their ionic radii given in [55], the Au, S, P and N atoms are shown with arbitrary sizes.
The formation of complexes with coordination numbers < 9 in the present series is restricted to gallium. This is remarkable since, at least for In 3+ and Sc 3+ , coordination compounds with C.N. 9 are extremely rare [103][104][105][106][107][108][109][110] and, particularly, the few scandium compounds contain a considerable number of aqua ligands or small chelators such as nitrate or boronhydride in their coordination spheres [106][107][108][109][110]. The molecular structures of the indium and scandium compounds are shown in Figures 7 and 8.  3+ and In 3+ ions are adopted from their ionic radii given in [55], the Au, S, P and N atoms are shown with arbitrary sizes.
The formation of complexes with coordination numbers < 9 in the present series is restricted to gallium. This is remarkable since, at least for In 3+ and Sc 3+ , coordination compounds with C.N. 9 are extremely rare [103][104][105][106][107][108][109][110] and, particularly, the few scandium compounds contain a considerable number of aqua ligands or small chelators such as nitrate or boronhydride in their coordination spheres [106][107][108][109][110]. The molecular structures of the indium and scandium compounds are shown in Figures 7 and 8.
The formation of complexes with coordination numbers < 9 in the present series is restricted to gallium. This is remarkable since, at least for In 3+ and Sc 3+ , coordination compounds with C.N. 9 are extremely rare [103][104][105][106][107][108][109][110] and, particularly, the few scandium compounds contain a considerable number of aqua ligands or small chelators such as nitrate or boronhydride in their coordination spheres [106][107][108][109][110]. The molecular structures of the indium and scandium compounds are shown in Figures 7 and 8.   3 }] complexes illustrating the good fit of the two lanthanide structures and the differences between the structures of the lanthanum/scandium complexes due to their different ionic radii. The sizes of the central ions are adopted from their ionic radii given in [55], the Au, S, P and N atoms are shown with arbitrary sizes.
Despite the retention of the coordination number, some structural changes are observed when switching from the lanthanide series to the elements of group 3 and group 13. While the M 3+ -N distances are generally larger than the M 3+ -O bonds for the lanthanides (see the graph in Figure 4 and the values in Table 2), this situation gradually changes for the main group and transition metals (Table 3)

More Structural Aspects and Potential Implication for the Reactivity
There exist no regular Platonic, Archimedean or Catal polyhedra with nine vertices and no prisms or antiprisms can be constructed with an odd number of vertices. S. Alvarez and co-workers address this point in an excellent, scholarly written paper and derive a number of shapes to be relevant for compounds with C.N. 9 [94]. Among others, they include an octagonal pyramid (OPY-9), a heptagonal bipyramid (HBPY-9), a spherical relaxed capped cube (CCU-9), a spherical capped square antiprism (CSAPR-9), a spherical tricapped trigonal prism (TCTPR-9), a tridiminished icosahedron (JTDIC-9) and a muffin (MFF-9). Some of the classical Johnson polyhedra [111] were adopted to better meet the description of coordination compounds, i. e. providing approximately identical vertex to center distances instead of identical edge lengths, as used for the description of the pure geometrical shapes [94,112]. This results in 'relaxed' or 'spherical' polyhedra, which are also used in the present considerations and relevant shapes are shown in Figure 9. The 'Continuous Shape' approach [94,97,98] allows the evaluation of the bonding situations, even in the case of simple visualization giving no unambiguous results (as in the present case).  Figure 9 illustrates the above mentioned selection of idealized polyhedra applicable for coordination compounds having C.N. 9 [113] together with the coordination polyhedra around some M 3+ ions derived from the X-ray diffraction studies. The examples were selected in order to cover the entire range of ionic radii, but also to include lanthanides, main group and transition metals. It becomes evident that a plain visual inspection of the polyhedra does not allow to derive differences within the series. This becomes possible by means of the continuous shape algorithm [114,115]. Table 4 contains the continuous shape measures derived for the [M⊂{Au3(L1 ethyl )3}] complexes of the present study, considering the six coordination polyhedra regarded as relevant for coordination compounds [113]. It becomes clear that capped cubes, Hula hoop shapes and tridiminished icosahedra do not play any role for any of the complexes. The preferred polyhedron for all [M⊂{Au3(L1 ethyl )3}] complexes of the present study is a tricapped trigonal prism (see the corresponding values in Table 4 highlighted in red). However, it also becomes evident that the r-CSAPR-9 (highlighted in orange) and MFF-9 (highlighted in green) shapes play a considerable role in the description of these compounds. It is noteworthy that the fits of the experimental data with the three idealizes shapes are best for the compounds, which have the smallest M 3+ ions. The best fits are found for the indium compound with continuous shape values of 0.404 for s-TCTPR-9 and 0.910 for r-CSAPR-9. This might have to do with the fact that the M-O and M-N bonds in these compounds are almost equal, while they are clearly different in most of the other compounds (particularly in the representatives of the lanthanide series, see Figure 4 and Tables 2 and 3).  Figure 9 illustrates the above mentioned selection of idealized polyhedra applicable for coordination compounds having C.N. 9 [113] together with the coordination polyhedra around some M 3+ ions derived from the X-ray diffraction studies. The examples were selected in order to cover the entire range of ionic radii, but also to include lanthanides, main group and transition metals. It becomes evident that a plain visual inspection of the polyhedra does not allow to derive differences within the series. This becomes possible by means of the continuous shape algorithm [114,115]. Table 4 contains the continuous shape measures derived for the [M⊂{Au 3 (L1 ethyl ) 3 }] complexes of the present study, considering the six coordination polyhedra regarded as relevant for coordination compounds [113]. It becomes clear that capped cubes, Hula hoop shapes and tridiminished icosahedra do not play any role for any of the complexes. The preferred polyhedron for all [M⊂{Au 3 (L1 ethyl ) 3 }] complexes of the present study is a tricapped trigonal prism (see the corresponding values in Table 4 highlighted in red). However, it also becomes evident that the r-CSAPR-9 (highlighted in orange) and MFF-9 (highlighted in green) shapes play a considerable role in the description of these compounds. It is noteworthy that the fits of the experimental data with the three idealizes shapes are best for the compounds, which have the smallest M 3+ ions. The best fits are found for the indium compound with continuous shape values of 0.404 for s-TCTPR-9 and 0.910 for r-CSAPR-9. This might have to do with the fact that the M-O and M-N bonds in these compounds are almost equal, while they are clearly different in most of the other compounds (particularly in the representatives of the lanthanide series, see Figure 4 and Tables 2 and 3). Table 4. Continuous shape measures derived for the nine-coordinate [M⊂{Au 3 (L1 ethyl ) 3 }] species. Preferred polyhedra are highlighted in color (the used colors refer to those used in Figure 9).  The continuous shape studies on the [M⊂{Au 3 (L1 ethyl ) 3 }] complexes indicate a widespread independence of the established coordination spheres from the size of the M 3+ ions. Thus, geometrical factors should also not be discussed as driving forces for a potential reactivity of the bimetallic assemblies. Nevertheless, some information about potential reaction patterns of such compounds is desired for the evaluation of potential applications. This particularly concerns their reactivity with other metal ions. Thus, we performed first of such studies with [Sc⊂{Au 3 (L1 ethyl ) 3 }]. As the complex with the smallest M 3+ ion and a low tendency for stabilization in a nine-coordinate environment, which is documented by the small number of such compounds [106][107][108][109][110], we regarded it as a good candidate for potential transmetalations. Another advantage of scandium is that 45 Sc NMR is feasible. 45 Sc has 100% natural abundance, an NMR frequency close to that of 13 C and a high receptivity (approximately 0.3 relative to 1 H) [116]. As a quadrupole nucleus with I = 7/2 (Q = −22 fm 2 ) [117], 45 Sc frequently produces NMR spectra with large line widths, particularly for compounds having a low local symmetry around the nucleus. This point might also be a considerable drawback with the nine-coordinate compounds of the present study. Nevertheless, a broad signal at 8.1 ppm (line width 1670 Hz) could be resolved for a CH 2 Cl 2 solution of [Sc⊂{Au 3 (L1 ethyl ) 3  In a first reaction, we exposed [Sc⊂{Au3(L1 ethyl )3}] to a solution of zinc acetate in methanol, which resulted in the rapid formation of colorless crystals of [Zn⊂{Au2(L1 ethyl )2}] (21). This means that in addition to the release of the Sc 3+ ions from the potentially unfavorable coordination situation with C.N. 9, gold-sulfur bonds of the {Au3(L1 ethyl )3} 3-rings were cleaved. This comes not completely unexpected in the light of the fact that Zn 2+ ions preferably form complexes with the coordination numbers four or six. However, the mechanism of the ring cleavage of the {Au3(L1 ethyl )3} 3-unit and the re-formation of an {Au2(L1 ethyl )2} 2-ring system under the influence of Zn 2+ ions is not yet completely clear and requires further studies.
Considering that tris(acetylacetonato)scandium(III), [Sc(acac)3], can be used as a precursor for the synthesis of [Sc⊂{Au3(L1 ethyl )3}] (15) (Scheme 3), it is interesting to note that Sc 3+ can also be released from compound 15 by interacting with [Zn(acac)2] and [Sc(acac)3] is re-formed. The latter reaction was chosen for a corresponding NMR experiment. Figure  10 depicts a reaction sequence with the subsequent addition of [Zn(acac)2] aliquots. The broad signal of the starting material at 6.8 ppm gradually disappears and two new signals appear: a transient, small signal at 17.0 ppm and that of the final scandium-containing product [Sc(acac)3] at 90.1 ppm. Unfortunately, we could not isolate the intermediate with the 45 Sc NMR signal at 17 ppm. Indicated by its small line width, it most probably contains the Sc 3+ ion in a sterically less restricted form than in the starting compound 15 or [Sc(acac)3], where the metal ion establishes a distorted octahedron due to the limitations given by the three chelating acetylacetonato ligands [118]. In a first reaction, we exposed [Sc⊂{Au 3 (L1 ethyl ) 3 }] to a solution of zinc acetate in methanol, which resulted in the rapid formation of colorless crystals of [Zn⊂{Au 2 (L1 ethyl ) 2 }] (21). This means that in addition to the release of the Sc 3+ ions from the potentially unfavorable coordination situation with C.N. 9, gold-sulfur bonds of the {Au 3 (L1 ethyl ) 3 } 3− rings were cleaved. This comes not completely unexpected in the light of the fact that Zn 2+ ions preferably form complexes with the coordination numbers four or six. However, the mechanism of the ring cleavage of the {Au 3 (L1 ethyl ) 3 } 3− unit and the re-formation of an {Au 2 (L1 ethyl ) 2 } 2− ring system under the influence of Zn 2+ ions is not yet completely clear and requires further studies.
Considering that tris(acetylacetonato)scandium(III), [Sc(acac) 3 ], can be used as a precursor for the synthesis of [Sc⊂{Au 3 (L1 ethyl ) 3 }] (15) (Scheme 3), it is interesting to note that Sc 3+ can also be released from compound 15 by interacting with [Zn(acac) 2 ] and [Sc(acac) 3 ] is re-formed. The latter reaction was chosen for a corresponding NMR experiment. Figure 10 depicts a reaction sequence with the subsequent addition of [Zn(acac) 2 ] aliquots. The broad signal of the starting material at 6.8 ppm gradually disappears and two new signals appear: a transient, small signal at 17.0 ppm and that of the final scandium-containing product [Sc(acac) 3 ] at 90.1 ppm. Unfortunately, we could not isolate the intermediate with the 45 Sc NMR signal at 17 ppm. Indicated by its small line width, it most probably contains the Sc 3+ ion in a sterically less restricted form than in the starting compound 15 or [Sc(acac) 3 ], where the metal ion establishes a distorted octahedron due to the limitations given by the three chelating acetylacetonato ligands [118].

Spectroscopic Methods
Elemental analyses of carbon, hydrogen, nitrogen and sulfur were determined using a Heraeus vario EL elemental analyzer. IR spectra were measured as KBr pellets on a Shimadzu IR Affinity-1 spectrometer between 400 and 4000 cm −1 or a Thermo Scientific Nicolet iS10 ATR spectrometer. The NMR spectra were recorded on a JEOL 400 MHz spectrometer. ESI TOF mass spectra were measured with an Agilent 6210 ESI TOF (Agilent Technologies, Santa Clara, CA, USA). All MS results are given in the form: m/z, assignment. UV-vis spectra were recorded on a Specord 40 (Analytik Jena) spectrophotometer in the wavelength range 200-700 nm. Time-Resolved Laser Fluorescence (TRLFS) was performed at T < 20 K using a pulsed Nd:YAG (Spectra Physics) pumped dye laser system (Radiant Dyes Narrow Scan K). The detection was carried out with a spectrograph (Shamrock 303i) equipped with a polychromator with 300, 600 and 1200 lines/mm gratings and an ICCD camera system (Andor iStar). Common concentrations were applied for the spectroscopic studies. In some cases, they were limited by the low solubility of the compounds.

X-ray Crystallography
The intensities for the X-ray determinations were collected on STOE IPDS-2T or Bruker CCD instruments with Mo/Kα radiation. The various temperatures applied are due to the experimental setup of the different diffractometers. Semi-empirical or numerical absorption corrections were carried out by the SADABS or X-RED32 programs [126,127]. Structure solution and refinement were performed with the SHELX programs [128,129] included in the WinGX [130] program package or OLEX2 [131]. Hydrogen atoms were calculated for the idealized positions and treated with the 'riding model' option of SHELXL. Since some of the lanthanide compounds crystallized together with disordered solvent molecules (partially close to special positions), the refinements of such structures were undertaken with the removal of the disordered solvent molecules using the SQUEEZE option installed in the program PLATON or the solvent mask option of OLEX2. Details are given in the Supplementary Materials. The representation of molecular structures was done using the program DIAMOND [132]. coronands, which host M 3+ ions from the lanthanide series or the 'group 3 and 'group 13 elements. Except for the very small Ga 3+ ions, nine-coordinate environments are established around the M 3+ ions. This is particularly surprising for small ions such as In 3+ or Sc 3+ , as nine-coordinate complexes are extremely rare for them.

Conclusions
Transmetalation reactions starting from the scandium complex gives access to corresponding compounds with bivalent transition metal ions. The mechanism of such reactions is hitherto not clear and require more detailed studies, which should also give information about the stabilities of such assemblies and their synthetic potential.
Supplementary Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/molecules28145421/s1, Table S1. Crystallographic data and data collection parameters; Figure S1. Ellipsoid representation of [La⊂{Au 3 (L1 ethyl ) 3 }] (1) including the positional disorder in the peripheral ethyl groups. The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity. A solvent mask was calculated and 53 electrons were found in a volume of 218 A 3 in 3 voids per unit cell. This is consistent with the presence of 2.5 H 2 O per formula unit which account for 50 electrons per unit cell; Figure S2. Ellipsoid representation of [Ce⊂{Au 3 (L1 ethyl ) 3 }] (2) including the positional disorder in the peripheral ethyl groups. The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity; Figure S3: Ellipsoid representation of [Pr⊂{Au 3 (L1 ethyl ) 3 }] (3) including the positional disorder in the peripheral ethyl groups. The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity; Figure S4. Ellipsoid representation of [Nd⊂{Au 3 (L1 ethyl ) 3 }] (4) including the positional disorder in the peripheral ethyl groups. The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity. A solvent mask was calculated and 54 electrons were found in a volume of 247 A 3 in 4 voids per unit cell. This is consistent with the presence of 2.5 H 2 O per formula unit which account for 50 electrons per unit cell; Figure S5. Ellipsoid representation of [Sm⊂{Au 3 (L1 ethyl ) 3 }] (5). The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity. A solvent mask was calculated and 55 electrons were found in a volume of 240 A 3 in 4 voids per unit cell. This is consistent with the presence of 2.5 H 2 O per formula unit which account for 50 electrons per unit cell; Figure S6. Ellipsoid representation of [Eu⊂{Au 3 (L1 ethyl ) 3 }] (6). The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity. A solvent mask was calculated and 46 electrons were found in a volume of 227 A 3 in 3 voids per unit cell. This is consistent with the presence of 2.5 H 2 O per formula unit which account for 50 electrons per unit cell; Figure S7. Ellipsoid representation of [Gd⊂{Au 3 (L1 ethyl ) 3 }] (7). The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity. A solvent mask was calculated and 65 electrons were found in a volume of 179 A 3 in 2 voids per unit cell. This is consistent with the presence of 3.5 H 2 O per formula unit which account for 70 electrons per unit cell; Figure S8. Ellipsoid representation of [Tb⊂{Au 3 (L1 ethyl ) 3 }] (8) including the positional disorder in the peripheral ethyl groups. The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity. A solvent mask was calculated and 48 electrons were found in a volume of 252 A 3 in 3 voids per unit cell. This is consistent with the presence of 2.5 H 2 O per formula unit which account for 50 electrons per unit cell; Figure S9. Ellipsoid representation of [Dy⊂{Au 3 (L1 ethyl ) 3 }] (9). The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity. A solvent mask was calculated and 43 electrons were found in a volume of 138 A 3 in 1 void per unit cell. This is consistent with the presence of 2 H 2 O per formula unit which account for 40 electrons per unit cell; Figure S10. Ellipsoid representation of [Ho⊂{Au 3 (L1 ethyl ) 3 }] (10) including the positional disorder in the peripheral ethyl groups. The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity. A solvent mask was calculated and 43 electrons were found in a volume of 282 A 3 in 3 voids per unit cell. This is consistent with the presence of 2 H 2 O per formula unit which account for 40 electrons per unit cell; Figure S11. Ellipsoid representation of [Er⊂{Au 3 (L1 ethyl ) 3 }] (11) including the positional disorder in the peripheral N(ethyl) 2 groups. The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity; Figure S12. Ellipsoid representation of [Tm⊂{Au 3 (L1 ethyl ) 3 }] (12) including the positional disorder in the peripheral N(ethyl) 2 groups. The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity; Figure S13. Ellipsoid representation of [Yb⊂{Au 3 (L1 ethyl ) 3 }] (13) including the positional disorder in the peripheral ethyl groups. The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity. A solvent mask was calculated and 33 electrons were found in a volume of 235 A 3 in 3 voids per unit cell. This is consistent with the presence of 1.5 H 2 O per formula unit which account for 30 electrons per unit cell; Figure S14. Ellipsoid representation of [Lu⊂{Au 3 (L1 ethyl ) 3 }] (14). The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity; Figure S15. Ellipsoid representation of [Sc⊂{Au 3 (L1 ethyl ) 3 }] (15) including the positional disorder in the peripheral ethyl groups. The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity; Figure S16. Ellipsoid representation of [Y⊂{Au 3 (L1 ethyl ) 3 }] (16) × H 2 O. The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity; Figure S17: Ellipsoid representation of [In⊂{Au 3 (L1 ethyl ) 3 }] (17) including the positional disorder in the peripheral ethyl groups. The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity; Figure S18. Ellipsoid representation of [Ga⊂{Au 2 (L1 ethyl ) 2 }](NO 3 ) (18a). The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity. A solvent mask was calculated and 334 electrons were found in a volume of 2100 A 3 in 1 void per unit cell. This is consistent with the presence of 2 CH 2 Cl 2 per formula unit which account for 336 electrons per unit cell; Figure S19. Ellipsoid representation of [Ga⊂{Au 2 (L1 ethyl ) 2 }](BF 4 ) (18b). The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity; Figure S20. Ellipsoid representation of [(AuCl) 2 (H 2 L1 ethyl )] (19). The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity; Figure S21. Ellipsoid representation of [Zn⊂{Au 2 (L1 ethyl ) 2 }] (21). The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity. A solvent mask was calculated and 220 electrons were found in a volume of 946 A 3 in 1 void per unit cell. This is consistent with the presence of 6 H 2 O per formula unit which account for 240 electrons per unit cell; Figure S22. Ellipsoid representation of [Sc(H 2 O) 2 {Au(L1 ethyl ) 2 }] (22) including the positional disorder in the peripheral ethyl groups. The thermal ellipsoids are set at a 50% probability level. Hydrogen atoms are omitted for clarity; Figure S23. IR (KBr) spectrum of [La⊂{Au 3 (L1 ethyl ) 3 }] (1); Figure S24. 1 H NMR spectrum of [La⊂{Au 3 (L1 ethyl ) 3 }] (1) in CDCl 3 ; Figure S25. 13