Synthesis and Structure

The novel discrete, neutral polyoxo‐16‐palladium(II) cluster [Pd₁₆O₂₄(OH)₈((CH₃)₂As)₈] (Pd₁₆) was synthesized by room‐temperature stirring of a mixture of palladium(II) acetate Pd(OAc)₂ in a sodium dimethylarsinate (also known as cacodylate, from here on abbreviated as cac) buffer solution at pH 7 for 2 days (after reaction pH was ≈5.7), followed by filtration and crystallization (see Supp. Info for Exp. Section). Single‐crystal X‐ray analysis revealed that the Pd₁₆ comprises 16 square‐planar oxo‐coordinated palladium(II) ions, which can be subdivided in a central [Pd₈O₈(OH)₈]⁸⁻ square‐antiprismatic unit, encircled by a cyclic [Pd₈O₁₆((CH₃)₂As)₈]⁸⁺ unit, resulting in the neutral, discrete metal‐oxo cluster Pd₁₆ (Figure 1). All Pd²⁺ ions in Pd₁₆ exhibit a square‐planar coordination geometry with Pd‐O distances in the range of 1.978(9)–2.053(9) Å (Supporting Information, Table S2). The novel Pd₁₆ has idealized D _4d point group symmetry with the C₄ principal rotation axis passing through the central [Pd₈O₈(OH)₈]⁸⁻ square‐antiprismatic unit. Bond valence sum (BVS) calculations on the μ₂‐OH groups yields values of 1.02–1.16 (Table S4), confirming that these oxygens are monoprotonated. To the best of our knowledge, this is the first example of a discrete, neutral Pd‐oxo cluster. To date, only a handful of neutral noble‐metal‐based oxo‐clusters have been reported. ^[21]

Figure 1

Structural representation of the disk‐shaped Pd₁₆; (top) top view and (bottom) side view. Color code: Pd (violet), O (red), C (gray), H (white), (CH₃)₂AsO₂ (cyan tetrahedra).

In the solid‐state lattice, each Pd₁₆ is further linked through weak C−H⋅⋅⋅O hydrogen bonds to other Pd₁₆ units (as well as to the co‐crystallized cacodylates) to form a supramolecular 2D layered assembly (Supporting Information, Figure S1, Table S5). This labile supramolecular arrangement of Pd₁₆ results in high solubility of the compound in water. In fact, the crystalline nature of the material is lost rapidly upon filtration and exposure to air turning it amorphous.

The second novel discrete palladium(II)‐oxo cluster Na₂[Pd₁₆O₂₆(OH)₃Cl₃((CH₃)₂As)₈] (Pd₁₆Cl) was synthesized by room‐temperature stirring of a mixture of palladium(II) chloride PdCl₂ in a sodium cacodylate buffer solution at pH 7 for 2 days (after reaction pH was ≈5.7), followed by filtration and crystallization (see Supp. Info for Exp. Section). Single‐crystal X‐ray analysis demonstrated that Pd₁₆Cl has an overall identical structure as Pd₁₆, but careful analysis revealed the presence of some chloro ligands. The use of palladium(II) chloride (rather than acetate) resulted in partial substitution of the μ₂‐OH groups O1A, O2A, O3A, and O4A (each having a crystallographic sof of 0.625) by Cl⁻ (Cl1, Cl2, Cl3 and Cl4 each having a crystallographic sof of 0.375). Thus, the structure of Pd₁₆Cl can be described as a central [Pd₈O₁₀(OH)₃Cl₃]¹⁰⁻ square‐antiprismatic unit encircled by a ring‐shaped [Pd₈O₁₆((CH₃)₂As)₈]⁸⁺ unit, leading to the formation of a discrete dianionic assembly [Pd₁₆O₂₆(OH)₃Cl₃((CH₃)₂As)₈]²⁻ (Figures 2 (left); Figure S2). Elemental analysis indicated that this negative charge is balanced by two sodium counter cations, which could not be located by single‐crystal XRD due to disorder. Nonetheless, their presence was also suggested by theoretical computations (vide infra). However, ESI‐MS studies (vide infra) indicate that in solution, the two deprotonated hydroxo groups reprotonate, yielding the neutral free acid H₂Pd₁₆Cl. In essence, the main differences between Pd₁₆Cl and Pd₁₆ are that in the former (i) three hydroxo groups are replaced by Cl⁻ ions and (ii) two of the hydroxo groups are deprotonated and the negative charge is balanced by two Na⁺ ions. The presence of a strong hydrogen bond acceptor in the form of Cl⁻ in Pd₁₆Cl leads to strong C−H⋅⋅⋅Cl interactions involving the μ₂‐OH/Cl⁻ groups (O1A/Cl1, O2A/Cl2, O3A/Cl3, O4A/Cl4) and the C−H bonds of the cacodylate methyl groups (Table S9, Figure S2). ^[22] Furthermore, the introduction of the Cl⁻ groups introduces a certain degree of hydrophobicity in Pd₁₆Cl (see the computational study in the proceeding text), which in turn leads to stronger intermolecular interactions in 3D space. Thus, each Pd₁₆Cl unit is linked to six others leading to a stable 3D hydrogen‐bonded organic–inorganic framework (HOIF). This 3D arrangement exhibits a uninodal acs framework topology featured by the Schläfli symbol 4⁹.6⁶, ^[23] giving rise to hexagonal 1D‐channels running along the “c” direction (Figure 2). There are only a handful of such types of frameworks reported in the literature, probably due to the inherent difficulty of controlling hydrogen bond interactions between the organic and inorganic components. ^[24] Pd₁₆Cl is the first such example of a noble‐metal‐oxo‐cluster‐based 3D HOIF. The formation of a stable 3D framework in Pd₁₆Cl is accompanied by a higher isolated yield and crystallinity as well as lower aqueous solubility as compared to Pd₁₆.

Figure 2

(left) Structural representation of the disk‐shaped Pd₁₆Cl. Color code: Pd (violet), Cl/O (disordered; light‐green), O (red), C (gray), H (white), (CH₃)₂AsO₂ (cyan tetrahedra). (middle) 3D lattice structure of Pd₁₆Cl. (right) Uninodal acs topology in Pd₁₆Cl. The dark‐green ellipsoids represent individual Pd₁₆Cl units.

The stability and crystallinity of Pd₁₆Cl was further corroborated by PXRD studies (Figure 3 (top)). The PXRD pattern of freshly prepared Pd₁₆Cl matched well with the simulated PXRD pattern, indicating phase‐purity. Upon dehydration of Pd₁₆Cl by heating to 70 °C for 1 h, only minimal changes were observed in the PXRD pattern, indicating that even upon loss of the lattice water molecules, the framework retains its stability and long‐range order. Rehydration in a wet atmosphere at room temperature again keeps the PXRD spectrum intact. Thermogravimetric analysis (TGA) studies on Pd₁₆Cl reiterated the stability of the compound upon dehydration and rehydration (Figure 3 (bottom); Supporting Information). Thus, the introduction of Cl⁻ ions into the Pd₁₆ induces a drastic improvement in the stability and crystallinity of the compound due to the formation of stable extended supramolecular assembly.

Figure 3

(top) PXRD patterns and (bottom) TGA curves of freshly prepared, dehydrated, and rehydrated compound Pd₁₆Cl.

The third novel discrete, neutral palladium(II)‐oxo cluster [Pd₂₄O₄₄(OH)₈((CH₃)₂As)₁₆] (Pd₂₄) was synthesized by room‐temperature stirring of a mixture of palladium(II) acetate Pd(OAc)₂ in a sodium cacodylate buffer solution at pH 7 for 2 days (after reaction pH was ≈5.8), followed by adjustment of pH to ≈7 by NaOH_aq solution and then stirred further for 1 day before filtering. This step is crucial because without such pH readjustment after the reaction Pd₁₆ is formed. Attempts to synthesize a chloro‐derivative of Pd₂₄ by adjusting the pH of the reaction mixture of Pd₁₆Cl to ≈7 after 2 days of stirring at room temperature failed, and instead Pd₁₆Cl was obtained in low yield. The structure of Pd₂₄ contains Pd₁₆ as a core, but then four of the eight μ₂‐hydroxo groups (two on either side of the molecule) are deprotonated and bind to two cationic, tetranuclear [Pd₄O₈(OH)₂((CH₃)₂As)₄]²⁺ units, one on each side of the cluster, resulting in the bicapped [Pd₂₄O₄₄(OH)₈((CH₃)₂As)₁₆] (Pd₂₄) (Figure 4; Figure S3). Pd₂₄ has idealized point group symmetry C₁ and crystallizes in the triclinic space group P $\bar{1}$ (Table S1). The deprotonation step appears to be the key for the formation of Pd₂₄ and this is accomplished by pH 7 adjustment after reaction. The bond valence sum (BVS) calculations on the μ₂‐OH groups yield values of 1.08–1.18 (Table S12), which are typical for hydroxo groups. In the solid state, each Pd₂₄ cluster is further linked to four other Pd₂₄ clusters through weak C−H⋅⋅⋅O interactions involving the C−H bonds of the methyl groups, the oxygens of the cacodylates and the μ₂‐OH groups, resulting in a 4‐connected hydrogen‐bonded SBU (Figure S4, Table S13). This leads to a 3D HOIF with a dia topology with the Schläfli symbol 6⁶. ^[25] However, in spite of forming such a framework, the weak C−H⋅⋅⋅O hydrogen bonds impart significant lability and flexibility to the 3D assembly leading to the loss of crystallinity upon exposure to air and high aqueous solubility (Figure S4d).

Figure 4

Structural representation of Pd₂₄, which can be viewed as a bicapped Pd₁₆. Color code: Pd (violet), O (red), C (gray), (CH₃)₂AsO₂ (cyan tetrahedra). Hydrogen atoms are omitted for clarity.

Solid‐state, Solution, and DOSY NMR Spectroscopy

Although the molecular entities of Pd₁₆Cl and Pd₁₆ are isostructural, the former has a significantly higher yield and therefore it was utilized for NMR measurements. The ¹H 30 kHz MAS spectrum of cacodylic acid (H‐cac) exhibits two broad peaks at ≈1.8 and ≈12.3 ppm, corresponding to the protons of the methyl and those of the acidic groups, respectively (Figure 5 (top)). The broadness of the peaks reflects a strong H‐H dipolar interaction between the different cacodylic acid molecules stacked in the crystal structure, which is typical for small organic molecules. ^[26] The ¹H 30 kHz MAS spectrum of the sodium salt of cacodylic acid (Na‐cac), however, exhibits two sharp peaks at ≈1.6 and ≈5.3 ppm, corresponding to the protons of the methyl groups and the crystal water molecules, respectively. The sharpness of the peaks is due to the fact that sodium cacodylate liquefies itself in its water of crystallization at ca 60 °C, leading to a solution‐like behavior. ^[27a] Indeed, the high spinning frequency (ca 30 kHz) is known to induce sample heating during solid‐state NMR experiments (the estimated temperature inside the NMR rotor is ≈57 °C at 30 kHz). ^[27b] This was further proven by ¹H MAS NMR spectroscopy of Na‐cac at spinning frequencies below 20 kHz, which resulted in broad peaks at ≈1.2 and ≈5.9 ppm, respectively (Figure 5 (top); Figure S6a). The ¹H solid‐state NMR spectra of Pd₁₆Cl and Pd₂₄ reveal broad peaks at ≈2.4 and ≈1.9 ppm, respectively, corresponding to the protons of the cacodylate methyl groups. In addition, a small peak at −0.7 ppm is observed for Pd₂₄, which corresponds to the protons of the μ₂‐OH groups. ^[28] A similar peak is not observed in the ¹H‐solid‐state spectrum of freshly prepared Pd₁₆Cl, probably due to fast exchange of these protons with the lattice water molecules. Indeed, upon thermal dehydration at 100 °C for 2 hours, a small peak at −1.9 ppm corresponding to the μ₂‐OH groups is now visible in the spectrum, indicating a frozen regime after complete removal of the lattice water molecules (Figure 5 (bottom)). The situation is different for Pd₂₄, where some of the μ₂‐OH groups are located in hydrophobic pockets lined by the cacodylate groups, which prevents such an exchange regime (Figure 4). In the solid state, the ¹H NMR peak corresponding to the hydroxo group can span a wide window depending on the extent of hydrogen bonding with acceptor molecules such as water and therefore, in turn, the extent of shielding–deshielding. Finally, we note the absence of the deshielded signal (12.3 ppm) due to the acidic proton indicating the complete deprotonation of the cac ligands in Pd₁₆Cl and Pd₂₄.

Figure 5

(top) ¹H MAS NMR (30 kHz) spectra of Pd₁₆Cl and Pd₂₄ compared to spectra of cacodylic acid (H‐cac) and its sodium salt (Na‐cac, rotation speeds 20 and 30 kHz) as references. (bottom) ¹H MAS NMR (30 kHz) spectra of as‐prepared (hydrated) Pd₁₆Cl and Pd₂₄, as well as a thermally dehydrated (100 °C for 2 h) Pd₁₆Cl sample.

The ¹³C{¹H} CPMAS (10 kHz) solid‐state NMR spectrum of H‐cac exhibits two narrow peaks at 15.6 and 20.3 ppm that correspond to the two crystallographically inequivalent cacodylic acid molecules present in the crystal structure of cacodylic acid (Figure 6). ^[26] The ¹³C solid‐state NMR spectrum of Na‐cac also exhibits two peaks at 19.3 and 19.9 ppm for the same reason. ^[26] The ¹³C solid‐state NMR spectra of Pd₁₆Cl and Pd₂₄, however, exhibit multiple overlapping broad peaks in the region 18–28 ppm and 15–30 ppm, respectively. These resonances could be attributed to the presence of 5 (Pd₁₆Cl) and 16 (Pd₂₄) crystallographically inequivalent cacodylates in their respective single crystal structures, which leads to several unresolved NMR lines of the methyl groups (Figure 6).

Figure 6

¹³C{¹H} CPMAS NMR (10 kHz) spectra of Pd₁₆Cl and Pd₂₄ compared to spectra of cacodylic acid (H‐cac) and its sodium salt (Na‐cac) as references.

The ¹H (D₂O) liquid‐state NMR spectra of H‐cac and Na‐cac exhibit sharp peaks at 1.9 and 1.6 ppm, respectively, corresponding to the protons of the methyl groups (Figure S6b). The low solubility of Pd₁₆Cl in water prevented us from acquiring a proper ¹H‐NMR spectrum. The ¹H (D₂O) NMR spectrum of Pd₂₄, however, exhibits peaks at 1.6 and 1.8 ppm that correspond to the co‐crystallized cacodylate and acetate, respectively, and a set of resonances in the region 1.7–3.1 ppm that likely corresponds to the methyl protons of the coordinated cacodylates in Pd₂₄ (Figure S6b). The ¹H‐DOSY NMR spectrum of Pd₂₄ (Figure S7) confirms these assignments where the signal of acetate (1.8 ppm) indicates a diffusion coefficient D of 840 μm² s⁻¹, and the free cacodylate signal (1.6 ppm) a value of 670 μm² s⁻¹, comparable to that of Na‐cac observed at 640 μm² s⁻¹. All other small peaks (1.7–3.1 ppm) are aligned around a value of D 215 μm² s⁻¹. Such a decrease of D by a factor of around three is fully consistent with coordination of the cac ligands to a nanoscopic object, such as Pd₂₄.

Mass spectrometry. All the peaks observed in the ESI‐MS spectrum of Pd₁₆Cl could be clearly assigned to molecular species related to the neutral free acid form of the 16‐palladium‐oxo cluster (Figure 7 (top); Table S14a and experimental and simulated isotope distributions in Figure S8a), with the main peak at m/z=1581.496 corresponding to the species {Na₂[Pd₁₆O₂₄(OH)₅Cl₃((CH₃)₂As)₈]}²⁺ (denoted as {Na₂(Pd₁₆)}²⁺), the peak at m/z=2101.329 corresponding to the species {Na₃(Pd₁₆)₂}³⁺, and the peak at m/z=3140.006 corresponding to {Na(Pd₁₆)}⁺. Thus, the ESI‐MS spectrum of Pd₁₆Cl corroborates the solid‐state structural analysis. Similarly, the ESI‐MS spectrum of Pd₂₄ (Figure 7 (bottom); Table S14b and experimental and simulated isotope distributions in Figure S8b) exhibits peaks that can be clearly assigned to the molecular species related to the 24‐palladium‐oxo cluster, with the main peak at m/z=1713.991 corresponding to {Na₃[Pd₂₄O₄₄(OH)₈((CH₃)₂As)₁₆]}³⁺ (denoted as {Na₃(Pd₂₄)}³⁺) and the peak at m/z=2559.988 corresponding to {Na₂(Pd₂₄)}²⁺. The peak at m/z=2057.769 is consistent with a partially dissociated species having the formula {Na₂[Pd₂₀O₁₀(OH)₈((CH₃)₂AsO₂)₁₂]}²⁺ (Table S14b), which indicates the fragility of the Pd₂₄‐moiety in the gas‐phase. We can envisage this partially dissociated species to be formed by removal of one of the Pd₄(cac)₄ capping units from one side of the Pd₁₆(cac)₈ (Figure 4).

Figure 7

ESI‐MS spectra of (top) Pd₁₆Cl and (bottom) Pd₂₄. Major species labeled in the Figure correspond to singly, doubly, and triply sodiated Pd₁₆Cl and Pd₂₄ clusters, respectively, as well as to a partially dissociated species of Pd₂₄.

Computational Studies on Pd₁₆ and Pd₁₆Cl

We carried out atomistic MD simulations with explicit solvent molecules to compare the behavior in solution of Pd₁₆ and the dianionic form Pd₁₆Cl with an aim to provide an explanation for the different supramolecular assemblies observed in the X‐ray crystal structures of both species and their significantly different aqueous solubility. To do so, we simulated 15 Pd₁₆ clusters and 15 Pd₁₆Cl anions, with their respective Na⁺ counter cations for 250 ns and analyzed their collective behavior. Visual exploration of the MD trajectories revealed that, at a cluster concentration of ca. 90 mM, both species are capable of forming large agglomerate structures in a similar manner, although of a different nature. Figure 8 compares the Pd₁₆⋅⋅⋅Pd₁₆ radial distribution functions (RDFs) for the simulated systems. The distribution of Pd₁₆ around each other (Figure 8 a (red line)) shows a main peak that covers an array of distances between ca. 12 and 14.5 Å, indicating that these are the preferred intermolecular distances for the Pd₁₆⋅⋅⋅Pd₁₆ interactions in solution. This corresponds to an interaction mode in which the clusters bring together the methyl groups of their respective cacodylate moieties (Figure 8 b) and thus, it can be classified as a hydrophobic interaction. In addition, the clusters tend to interact in a kind of layered disposition that resembles the one observed in the crystal (Figure S1). Other observed contacts correspond to interactions of the same nature, although less structured or involving three or more clusters (Figure S10). Conversely, Pd₁₆Cl anions do not show such preference for interacting in a single fashion, but the RDF indicates three well‐differentiated interaction modes (Figure 8 a (green dashed line)), represented in Figure 8 c. Notably, this is in good agreement with three different interaction modes observed in the trigonal crystal structure, in which every dianion is surrounded by six neighbors (Figure S2). Thus, it is reasonable to think that the formation of these agglomerates can be related to the initial nucleation steps towards the crystal formation, as previously observed in MD simulations with Wells‐Dawson‐type heteropolytungstate ions. ^[29] It is also worth mentioning that unlike the dianionic cluster, the protonated form of Pd₁₆Cl exhibits intercluster interactions that do not differ significantly from those observed for Pd₁₆ (Figure S11), in agreement with the fact that Pd₁₆Cl has to lose two protons during the crystallization process to yield a different supramolecular assembly than Pd₁₆ (trigonal P $\bar{3}$ c1 vs. triclinic P$\bar{1}$ crystal structures, vide supra). Moreover, these simulations also served to identify the most likely positions for the Na⁺ ions in the crystal structure of Pd₁₆Cl, which could not be determined by X‐ray diffraction. As shown in the volumetric densities of Figure S12, the sodium cations tend to sit between the oxygen atoms of vicinal cacodylate groups. Considering the vast number of equivalent sites, one might not expect that the X‐ray crystal structure can show a single preferred position for them with high occupancy.

Figure 8

a) Radial distribution function (RDF) between Pd₁₆ clusters taking the respective centers of mass as reference, averaged over the last 10 ns of a 250 ns simulation and over 15 clusters, with data sampling every 2 ps. The red solid line corresponds to Pd₁₆ whereas the green dashed line corresponds to the sodium salt of the dianionic form of Pd₁₆Cl. b,c) Representative snapshots associated with the maxima of the RDFs, representing agglomerated species formed during the simulations with Pd₁₆ (b) and Pd₁₆Cl (c). The cacodylate moieties of different clusters are represented in different colors for clarity, and sodium cations are shown as blue spheres.

We performed another set of MD simulations with one cluster each of Pd₁₆, Pd₁₆Cl and the protonated form H₂ Pd₁₆Cl (expected to be the dominant species in solution) in water to determine the distribution of water molecules around each cluster and the strength of their interactions (Figure S13 and associated text, Table S15). The different interaction modes of Pd₁₆ and Pd₁₆Cl can be ascribed to an increasing hydrophobicity of the inner Pd₈ core in moving from Pd₁₆ to the Cl‐containing Pd₁₆Cl. Therefore, the core of Pd₁₆Cl might be more prone to interacting with the hydrophobic methyl groups of other clusters, whereas the more hydrophilic core of Pd₁₆ with 8 hydroxo ligands is more likely surrounded by water molecules available, with which it can establish a greater number of hydrogen bonds. The comparison of the Pd₁₆⋅⋅⋅water RDFs represented in Figure S13 suggests that indeed, the incorporation of chloride ligands in the structure prevents the association of water molecules and, in consequence, can modulate the 3D structure of the supramolecular assembly incurred upon crystallization. In fact, this is not surprising since synthesis of chloride‐ and other halide‐containing molecules has been extensively employed as a strategy to enhance the hydrophobicity of organic drugs and in turn, their membrane penetration in cells and binding to pathological proteins. ^[30]