A Shape‐Complementarity Approach to Inducing Short Au(III)···Au(III) Contacts in Tetracyanoaurate(III) Salts

Abstract

Six sets of tetracyanoaurate(III) salts were synthesized and structurally characterized using the metal‐ligand complex cations [RE(bipyO₂)₄]³⁺ (RE = Sc, Y, La; bipyO₂ = 2,2′‐bipyridine‐N,N’‐dioxide), [Fe(bipyO₂)₃]³⁺, [Ln(dmbipyO₂)₄]³⁺ (Ln = Ce, Eu, Yb; dmbipyO₂ = 4,4′‐dimethyl‐2,2′‐bipyridine‐N,N’‐dioxide), [Ca(tcmc)]²⁺ (tcmc = 1,4,7,10‐tetrakis‐(carbamoylmethyl)‐1,4,7,10‐tetraazacyclododecane), and [Ca(12‐crown‐4)₂]²⁺. Noncovalent assembly of the [Au(CN)₄]⁻ anions tended to occur via Au···N_cyano interactions; however, rare Au(III)···Au(III) contacts between the [Au(CN)₄]⁻ groups — suggesting aurophilicity — could be induced when certain cation shape requirements were met. Specifically, cations with shape, size, and symmetry that allowed for packing in a complementary fashion with Au(III)···Au(III) aligned [Au(CN)₄]⁻ dimers or trimers — providing efficiently close‐packed layers — were found to be sufficient for manifesting Au(III)···Au(III) contacts. Modifying the [RE(bipyO₂)₄]³⁺ cation with peripheral methyl groups (the [Ln(dmbipyO₂)₄]³⁺ cation) caused isoreticular replacement of a {[Au(CN)₄]₃}³⁻ trimer with a dumbbell‐shaped {[Au(CN)₄]₂Cl}³⁻ tri‐anion featuring an unusual Au···Cl···Au bridge — illustrating that the assembly of the anionic groups will adapt to conserve the same close packing. Au(III) aurophilicity between the [Au(CN)₄]⁻ groups was studied using computational methods, crystal packing of the structures was probed using Hirshfeld surface analysis, and the emission properties of compounds containing the [Eu(dmbipyO₂)₄]³⁺ luminophore were investigated, showing high quantum yields of ca. 50%.

In a series of tetracyanoaurate(III) salts, careful consideration of cation shape — with the goal of shape‐complementary crystal packing — can allow for the assembly of [Au(CN)₄]⁻ anions to be controlled, and in some cases Au(III)···Au(III) aurophilic interactions to be rationally accessed.

1. Introduction

The attraction of d ¹⁰ or d ⁸ metal centers to each other and the consequent tendency toward self‐assembly has been well established, being termed metallophilicity, and this phenomenon has been used for many applications.^{[ 1} , ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ^{10 ]} Most notably, the influence of metallophilic interactions on the luminescent properties of Au(I) and Pt(II) systems is well known and has been leveraged in the design of light‐emitting diodes (LEDs) and chemical, pressure, and temperature sensors.^{[ 11} , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ^{17 ]} In contrast are aurophilic interactions between d ⁸ Au(III) centers, which have historically been far less studied, with only seventeen cases of unsupported Au(III)···Au(III) contacts reported to date.^{[ 18} , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ^{29 ]} However, the recent implication of Au(III) aurophilicity in the aggregation and luminescent properties of cyclometallated Au(III) compounds has sparked its closer consideration.^{[ 21} , ³⁰ , ³¹ , ^{32 ]}

Our group has previously explored the self‐assembly of tetracyanoaurate(III)‐containing materials; for example, aggregation of the [Au(CN)₄]⁻ units via Au···N_cyano “coinage bonds” ³³ , ^{34 ]} was probed through the synthesis of a series of [Cat]ⁿ⁺[Au(CN)₄]_n double salts (Cat = [ ⁿ Bu₄N]⁺, [AsPh₄]⁺, [N(PPh₃)₂]⁺, [Co(phen)₃]²⁺ (phen = 1,10‐phenanthroline), [Mn(2,2′:6′,2″‐terpyridine)₂]²⁺, trans‐[Co(en)₂Cl₂]⁺ (en = ethane‐1,2‐diamine), [Co(NH₃)₆]³⁺, [Ni(en)₃]²⁺, and [1,4‐diazabicyclo[2.2.2]octane‐H]⁺).^{[ 35 ]} The first five cations yielded structures with isolated [Au(CN)₄]⁻ anions, while use of the latter four cations generated structures featuring 1‐ and 2D [Au(CN)₄]⁻ structural motifs supported by Au···N_cyano interactions between adjacent anions and H···N_cyano H‐bonds involving the cations.

From this it was concluded that cations capable of H‐bonding can assist in inducing [Au(CN)₄]⁻ aggregation via Au···N_cyano interactions — in contrast to Au(I) and Pt(II) double salt systems where metallophilic interactions between anions are promoted through the use of H‐bond donor cations.^{[ 36} , ³⁷ , ^{38 ]} Since the comparatively weak Au(III) aurophilic interactions are delicate, these results suggest that the presence of any strong H‐bonding disrupts the potential for Au(III) aurophilicity to be observed.

Accordingly, we posited that Au(III) aurophilicity might be coaxed out by removing the potential for strong H‐bonding. Further, we noted the absence of non‐H‐bonding cations with charges greater than 2+ in our prior study. Based on this impetus, we initiated a new study of tetracyanoaurate(III) salts, targeting 3+ cations. For this purpose, we chose the [Ln(bipyO₂)₄]³⁺ (bipyO₂ = 2,2′‐bipyridine‐N,N’‐dioxide) cation since Ln‐bipyO₂ complexes have been researched as LEDs, single‐molecule magnets, and optical thermometers; in addition, many previous Ln‐bipyO₂‐cyanometallate systems have featured metallophilic interactions.^{[ 16} , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ , ^{44 ]} Indeed, we recently showed that the combination of [Ln(bipyO₂)₄]³⁺ with [Au(CN)₄]⁻ yielded compounds exhibiting the shortest unsupported Au(III) aurophilic interactions reported to date.^{[ 28 ]}

The structure of the [Ln(bipyO₂)₄][Au(CN)₄]₃·H₂O (Ln(bipyO₂)₄Au₃ ; Ln = Ce, Eu, Tb, Lu) compounds features [Ln(bipyO₂)₄]³⁺ cations in a square close‐packed arrangement with trimeric {[Au(CN)₄]₃}³⁻ anions sitting in the cavities within the layer (Figures 1a,b). Therein the aurophilic interactions occur, with Au···Au distances of 3.3603(4) Å and 3.4354(4) Å in the Ln = Lu case (shorter than the previous benchmark distances of 3.4435(3), 3.4949(9), and 3.507(3) Å between neutral, cationic, and anionic gold(III) compounds, respectively).^{[ 18} , ²¹ , ^{22 ]} The structural data was supported by computations at the MP2 level, with an energy minimum of roughly 2 kJ mol⁻¹ at a Au···Au distance of ca. 3.5 Å between two [Au(CN)₄]⁻ anions predicting weak aurophilicity. We postulated that the cation's high 3+ charge might facilitate the interactions, inductively alleviating the inherent Coulombic repulsion between the [Au(CN)₄]⁻ anions. Further, we noted the crystal structure's efficient packing and postulated a ‘shape complementarity’ concept (illustrated in Figures 1c,d) where the formation of the [Au(CN)₄]³⁻ trimer might be partially driven by its complementary shape with the [Ln(bipyO₂)₄]³⁺ cation: four of the cations can assemble to create a cavity in which three stacked [Au(CN)₄]⁻ molecules can appropriately nestle, providing a closely packed structure. Understanding the roles that cation charge, shape, and size play in coaxing out very weak Au(III) aurophilicity has importance in understanding the supramolecular chemistry of the [Au(CN)₄]⁻ group.

Figure 1.

Top: the structure of Lu(bipyO₂)₄Au₃ , exhibiting a) square close‐packing and b) short aurophilic contacts between [Au(CN)₄]⁻ anions.^{[ 28 ]} Bottom: a schematic representation of the structure, with green crosses representing the [Lu(bipyO₂)₄]³⁺ cations and gold circles representing the [Au(CN)₄]⁻ anions. The packing of these shapes c) is facilitated if d) the three [Au(CN)₄]⁻ groups stack into a column and fill the holes between the cations.

Additionally, the ideas of molecular shape and complementary packing — broadly captured by Kitaigorodskii's^{[ 45 ]} Principle of Close Packing; the projection of one molecule fitting into the hollow of another — are now being implicated in growing fields, including protein aggregation, molecular separation, and pharmaceutical co‐crystal design.^{[ 46} , ⁴⁷ , ⁴⁸ , ^{49 ]} However, while the consequences of co‐crystal shape and molecular complementarity have been well demonstrated in many organic multi‐component systems, examples of these concepts applied to inorganic systems are isolated and rare.^{[ 50} , ⁵¹ , ⁵² , ^{53 ]} This is because inorganic systems often feature highly directional interactions that govern the assembly process;^{[ 45 ]} however, in the study of very weak interactions like Au(III) aurophilicity, it is not unreasonable to begin to consider shape as a significant factor.

Therefore, this concept was investigated using the Ln(bipyO₂)₄Au₃ structure as a jumping‐off point and selectively and systematically varying the cation shape and charge. Herein, we report the results, featuring the synthesis and structures of sixteen tetracyanoaurate(III) salts, each illustrating important points for the aggregation of [Au(CN)₄]⁻ groups through noncovalent interactions (aurophilic or otherwise).

2. Results and Discussion

2.1. Synthesis and Structure of Sc(bipyO₂)₄Au₃

First, the limits of the prototypical Ln(bipyO₂)₄Au₃ structure were probed by seeing whether the series could extend to the smaller group 3 metal center Sc(III). One equiv of Sc(OTf)₃, 4 equiv of bipyO₂, and 3 equiv of KAu(CN)₄ were combined in water, yielding single crystals after a few days. Single crystal X‐ray diffraction (SCXRD) analysis revealed the compound [Sc(bipyO₂)₄][Au(CN)₄]₃·H₂O (Sc(bipyO₂)₄Au₃ ), isomorphous with the recently reported Ln(bipyO₂)₄Au₃ compounds^{[ 28 ]} and displaying the Au(III) aurophilic‐bound {[Au(CN)₄]₃}³⁻ structural motif.

The formation of isomorphous Sc(bipyO₂)₄Au₃ is interesting, as while Sc(III) has similar chemistry to the lanthanides, due to its smaller size it is most commonly six‐coordinate.^{[ 54 ]} Like the cations in the Ln(bipyO₂)₄Au₃ structures, the Sc(III) center has dodecahedral geometry (0.404% deviation via SHAPE analysis^{[ 55} , ^{56 ]}) and the cation is S₄ symmetric. To the best of our knowledge, this is the first scandium‐bipyO₂ complex reported to date.

Due to the smaller size of Sc(III) and thus the metal‐ligand complex, the cell dimensions of the crystal structure are reduced, and the Au(III)···Au(III) distances in the structure are even shorter than those reported by us previously for the Ln(bipyO₂)₄Au₃ series. Specifically, the Au2···Au3 distance is 3.4160(3) Å and the Au1···Au2 distance is 3.3351(3) Å, which is, to the best of our knowledge, the shortest Au(III) aurophilic distance reported to date. Previously with the Ln(bipyO₂)₄Au₃ series, we showed that the Au1···Au2 and Au2···Au3 distances were dependent on the ionic size of the lanthanide used. This dataset has been expanded to include the Sc(bipyO₂)₄Au₃ structure, providing the plot of Au···Au distances versus ionic radius shown in Figure 2. While the longer Au2···Au3 distance still appears to have a linear correlation with ionic size, it was found that the shorter Au1···Au2 distance could not be fit linearly and was best fit versus ionic size using a second‐order polynomial curve. We speculate that as smaller metal cations are used the Au1···Au2 distance eventually reaches a limit, where either packing effects or repulsion between the anions prohibits a closer approach of the Au(III) centers.

Figure 2.

A plot of Au(III)···Au(III) distance versus RE³⁺ ionic radius^{[ 57 ]} for the RE(bipyO₂)₄Au₃ (RE = Sc, Y, La, Ce, Eu, Tb, Lu) structures. The Au1···Au2 data is fitted with a second‐order polynomial, as represented by the red line. The Au2···Au3 data are fitted linearly, as represented by the blue line. Error bars are within the points.

To our understanding, this is the first report of a series of compounds that crystallize isomorphously for all of the rare earth elements (RE = Sc, Y, Ln), which was confirmed by synthesizing the RE = Y and La compounds (Au···Au distances therein also plotted in Figure 2). However, we note that Sc is underexplored in comparison to the lanthanides and other series spanning the Ln elements likely crystallize isomorphously for Sc and Y as well. The slight differences observed as a function of lanthanide ionic radius have been harnessed to control material properties such as negative thermal expansion,^{[ 58 ]} proton conduction,^{[ 59 ]} and others,^{[ 60} , ^{61 ]} and thus accessing the full gamut of rare‐earth metals would suggest a compelling range of tunability.

To describe the crystal packing of the Sc(bipyO₂)₄Au₃ structure and characterize the size and shape of the [Sc(bipyO₂)₄]³⁺ and {[Au(CN)₄]₃}³⁻ groups, the box model^{[ 50} , ^{62 ]} can be used, where the van der Waals volume of each molecule is enclosed in a rectangular box with dimensions L, M, and S (corresponding to the long, medium, and short axes, respectively; see Supporting Information for details of the bounding box calculation). The molecular size is approximated by the size of the box, while the shape can be described using the ratios of the axes: S/L, S/M, and M/L. The [Sc(bipyO₂)₄]³⁺ cation enclosed in its bounding box is shown in Figure S1; the box has L, M, and S axes lengths of 13.05, 12.28, and 12.04 Å, respectively, — with the L axis being in the direction of the S₄ rotation axis — and S/L, S/M, and M/L ratios of 0.92, 0.98, and 0.94, respectively (Table 1). The {[Au(CN)₄]₃}³²⁻ tri‐anion has a bounding box with dimensions of 11.05, 9.22, and 9.07 Å, providing S/L, S/M, and M/L ratios of 0.82, 0.98, and 0.83, respectively. While the box of the [Sc(bipyO₂)₄]³⁺ cation is larger than that of the tri‐anion, both groups have approx. the same shape, being square prisms with S/L ≈ M/L < S/M ≈ 1, in line with the square close‐packing of the ions observed in the structure.

Table 1.

Size and shape of crystallographic groups approximated from the box model, where the large, medium, and small (L, M, and S) box axes lengths (Å) and box volume (ų) describe the size, and S/L, S/M, and M/L ratios describe the shape.

Group [a]	L	M	S	Volume	S/L	S/M	M/L
[Sc(bipyO2)4]3+	13.05	12.28	12.04	1929.33	0.92	0.98	0.94
[Fe(bipyO2)3]3+	11.09	10.73	10.44	1242.12	0.94	0.97	0.97
[Ce(dmbipyO2)4]3+ [b]	16.77	16.60	7.38	2054.45	0.44	0.44	0.99
[Ce(dmbipyO2)4]3+ [c]	16.08	13.59	13.59	2967.67	0.85	1.00	0.85
[Ca(tcmc)]2+	10.00	9.36	7.75	725.40	0.78	0.83	0.98
[Ca(12‐crown‐4)2]2+	9.02	8.57	8.53	659.48	0.95	1.00	0.95
{[Au(CN)4]3}3−	11.05	9.22	9.07	923.94	0.82	0.98	0.83
{[Au(CN)4]2}2−	9.17	9.05	7.21	597.99	0.79	0.80	0.99

^[a]Where bipyO₂ = 2,2′‐bipyridine‐N,N’‐dioxide, dmbipyO₂ = 4,4′‐dimethyl‐2,2′‐bipyridine‐N,N’‐dioxide, and tcmc = 1,4,7,10‐tetrakis‐(carbamoylmethyl)‐1,4,7,10‐tetraazacyclododecane.

^[b]From the Ce(dmbipyO₂)₄Au₃·H₂O structure.

^[c]From the Ce(dmbipyO₂)₄Au₂Cl structure.

2.2. Synthesis and Structure of Fe(bipyO₂)₃Au₃

To provide a comparison to the RE(bipyO₂)₄Au₃ structure, where the cation has the same charge but a different shape, the combination of Fe(III) centers with bipyO₂ was targeted. Combining 1 equiv of FeCl₃, 3 equiv of bipyO₂, and 3 equiv of KAu(CN)₄ in water yielded single crystals of [Fe(bipyO₂)₃][Au(CN)₄]₃ (Fe(bipyO₂)₃Au₃ ), revealed by SCXRD analysis to contain octahedral [Fe(bipyO₂)₃]³⁺ cations alongside [Au(CN)₄]⁻ anions self‐aggregated into sheets via Au···N_cyano interactions (Figure 3).

Figure 3.

The structure of Fe(bipyO₂)₃Au₃ , showing a) the packing of [Fe(bipyO₂)₃]³⁺ cations into 1D chains via π‐ π interactions (orange dashed lines; calculated centroid positions shown as orange spheres), b) the intercalation of cations (represented by green spheres) in between 2D [Au(CN)₄]⁻ sheets, and c) the structure of the 2D sheets, generated via Au···N_cyano interactions (light blue and pink lines) between [Au(CN)₄]⁻ molecules. Color scheme: Au, yellow; C, gray; Fe, orange; N, blue; O, red. Hydrogen atoms omitted for clarity.

The [Fe(bipyO₂)₃]³⁺ cation is disordered in the structure of Fe(bipyO₂)₃Au₃ , with both C₃‐symmetric Λ/Δ‐[Fe(bipyO₂)₃]³⁺ enantiomers occupying the same site. Accordingly, for clarity when depicting the structure and to allow for Hirshfield analysis of the crystal packing (vide infra), an idealized nondisordered structure was created. The idealized packing of two of the cations is shown in Figure 3a: the cations pack in 1D chains along the a‐axis via π–π interactions between the aromatic rings of neighboring cations (ring centroid‐centroid distances of 3.858(13) and 3.59(2) Å).

As shown in Figures 3b,c, arrays of 1D chains of [Fe(bipyO₂)₃]³⁺ cations lie sandwiched between 2D sheets of [Au(CN)₄]⁻ anions, which interact via Au···N_cyano interactions. These sheets can be described as 1D ribbons (formed via highly directional Au2···N2 and Au1···N5 interactions, with d_Au···N = 3.038(19) and 2.97(2) Å, respectively, shown as light blue lines in Figure 3c), which are then joined together into a staircase‐like 2D sheet by longer, less directional Au1···N1 contacts (d_Au···N = 3.43(2) Å, shown as pink lines in Figure 3c). The latter interactions are just outside of the sum of the van der Waals radii of Au and N (3.06–3.36 Å)^{[ 63 ]} indicating very minimal contacts, while the first set of interactions are clearly shorter and stronger. However, both Au···N_cyano structural motifs are common and thus noteworthy, with propagation of 1D chains or 2D sheets often occurring through the first mode where the Au‐CN···Au atoms are nearly colinear, and dimerization of [Au(CN)₄]⁻ commonly occurring through the second mode where the intermolecular Au‐CN···Au angle is ca. 90 °. For ease of description, we will refer to the former mode as Type I and the latter mode as Type II.

In comparison with the Au(III) aurophilicity‐featuring RE(bipyO₂)₄Au₃ series, Fe(bipyO₂)₃Au₃ represents the result where [Au(CN)₄]⁻ is combined with a 3+ charged but differently shaped cation. As can be seen, no Au(III) aurophilic interactions are observed in the structure of Fe(bipyO₂)₃Au₃ , with only Au···N_cyano interactions occurring. This result clearly indicates that a 3+ charged cation is not sufficient for inducing Au(III) aurophilicity in the [Au(CN)₄]⁻ system. Although there could be many small factors contributing to the emergence of one structural motif over another — especially when such weak interactions are at play — in this case, we suppose that the size and shape of the cation play a major role.

Using the box model to compare the two cations reveals that the [Fe(bipyO₂)₃]³⁺ has a smaller size than the [Sc(bipyO₂)₄]³⁺ cation in the Sc(bipyO₂)₄Au₃ structure (box volumes of 1242.12 versus 1929.33 ų, respectively), but very similar axes ratios, indicating a similar overall shape. However, the biggest difference in shape between the two cations is not suitably captured by the box model: namely, the symmetry of the cation. That is, there is an obvious mismatch between the threefold symmetry of the C₃‐symmetric [Fe(bipyO₂)₃]³⁺ cations and the fourfold symmetry of the [Au(CN)₄]⁻ groups; this difference likely supports the observed partitioning of the ions and self‐aggregation, with the cations packing together into 1D chains and the anions forming 2D sheets.

2.3. Synthesis and Structure of Ce(dmbipyO₂)₄Au₃·H₂O

To see how the archetypical RE(bipyO₂)₄Au₃ series survives perturbations to the cation shape, the replacement of the bipyO₂ ligand with similar 4,4′‐dimethyl‐2,2′‐bipyridine‐N,N’‐dioxide (dmbipyO₂) was targeted. Thus, 1 equiv of CeCl₃·7H₂O was combined with 4 equiv of dmbipyO₂ and 3 equiv of KAu(CN)₄ in water to yield single crystals of Ce(dmbipyO₂)₄[Au(CN)₄]₃·H₂O (Ce(dmbipyO₂)₄Au₃·H₂O). SCXRD analysis revealed a structure composed of discrete [Ce(dmbipyO₂)₄]³⁺ cations and 1D chains of Au···N_cyano‐linked [Au(CN)₄]⁻ anions (Figure 4). Repeating this reaction with Ln(III) centers smaller than Ce(III) gave precipitates with powder X‐ray diffraction (PXRD) patterns that did not match the simulated pattern derived from the Ce(dmbipyO₂)₄Au₃·H₂O structure (Figure S2). A formula of Ln(dmbipyO₂)₄[Au(CN)₄]₃·xH₂O (Ln = Eu, x = 4.7; Ln = Yb, x = 5.7; Ln(dmbipyO₂)₄Au₃·xH₂O) could tentatively be assigned to the precipitates based on CHN elemental analysis and X‐ray fluorescence spectroscopy.

Figure 4.

The structure of Ce(dmbipyO₂)₄Au₃·H₂O, showing a) the [Ce(dmbipyO₂)₄]³⁺ cation featuring intra‐ligand π–π interactions (orange dashed lines; calculated centroid positions shown as orange spheres), b) the interstitial water molecule O9, which is H‐bonded with O1 (red dashed line) and coordinated to Au2 (light blue line), c) the overall packing of cations (represented by green spheres) and 1D [Au(CN)₄]⁻ chains, and d) the structure of the 1D chains featuring Au···N_cyano interactions (light blue and pink lines). Color scheme: Au, yellow; C, gray; Ce, tan; N, blue; O, red. Hydrogen atoms except those of O9, omitted for clarity.

With the Ce(dmbipyO₂)₄Au₃·H₂O structure, a striking difference in cation shape can be seen in comparison to the [RE(bipyO₂)₄]³⁺ cations in the RE(bipyO₂)₄Au₃ structure (Figure 4a). Instead of the globular, S₄ symmetric shape of the [RE(bipyO₂)₄]³⁺ cation, the [Ce(dmbipyO₂)₄]³⁺ is disk‐shaped and C₄ symmetric (with S/L and M/L = 0.43 and M/L = 0.99), supported by π–π interactions between the aromatic rings of the ligands (ring centroid‐centroid distances of 3.743(4), 3.778(4), 3.571(3), and 3.769(4) Å). Instead of the dodecahedral coordination geometry in [RE(bipyO₂)₄]³⁺, the Ce(III) center has rare cubic geometry (0.286% deviation via SHAPE analysis). This difference in geometry may be why the same structure is not obtained when smaller lanthanides are used, as steric crowding around a smaller metal center might force a change to either square antiprismatic or dodecahedral geometry – the two lowest‐energy configurations for eight‐coordinate systems.^{[ 64 ]} The same behavior has been observed with the perchlorate salts of the [Ln(bipyO₂)₄]³⁺ cation, where cubic geometry is observed with Ln = La, but all other lanthanides have structures with the coordination geometry distorted toward square antiprismatic.^{[ 65} , ⁶⁶ , ^{67 ]} Photoluminescence measurements of Eu(dmbipyO₂)₄Au₃·4.7H₂O — the precipitate formed from the same reaction when using EuCl₃·6H₂O — showed tentative crystal field splitting of the ⁵D₀→⁷F_J (J = 0, 1, and 2) into 1, 3, and 4 visible peaks, in line with either C_2v or lower site symmetry of the Eu(III) centers (Figure S3).^{[ 68 ]}

Another feature of the Ce(dmbipyO₂)₄Au₃·H₂O structure is an interstitial water molecule adjacent to the cation H‐bonded with an O atom exposed at the face of the cation disk (Figure 4b; O1···O9 = 2.909(7) Å). Due to the prominent position of this interstitial water molecule, it appears to direct the assembly of the [Au(CN)₄]⁻ groups: the water molecule is both H‐bonded with a nearby N_cyano group (O9···N12 = 3.083(10) Å) and also loosely coordinated to the axial position of the Au2 center (Figure 4b; O9···Au2 = 2.875(5) Å). Accordingly, Au···N_cyano interactions between the [Au(CN)₄]⁻ groups form 1D chains that run between the cations (Figure 4c). As shown in Figure 4d, the chain features alternating dimeric and monomeric [Au(CN)₄]⁻ groups, with weak Type II Au1···N2 and Au3···N11 interactions within the dimers (d_Au···N = 3.345(6) and 3.258(6) Å, respectively), and Type I Au1···N5 and Au3···N7 interactions linking the groups (d_Au···N = 2.953(6) and 2.950(6) Å, respectively). As it is difficult in this case to ascertain the relationship between cation shape and [Au(CN)₄]⁻ aggregation mode — due to the H‐bonding with the interstitial water molecule — a solvent molecule‐free structure featuring the [Ln(dmbipyO₂)₄]³⁺ and [Au(CN)₄]⁻ ions was sought.

2.4. Synthesis and Structure of Ce(dmbipyO₂)₄Au₃·0H₂O

Thus, dehydration of the Ce(dmbipyO₂)₄Au₃·H₂O crystal was explored, on the basis that removal of the interstitial water molecule might induce a conformational change and yield a new structure — as previously observed with Ln‐bipyO₂‐dicyanoaurate(I) systems.^{[ 43 ]} As can be seen from its thermogravimetric curve (Figure S4), Ce(dmbipyO₂)₄Au₃·H₂O loses one water molecule per formula unit (ca. 1% weight loss; 19 g mol⁻¹) at ca. 100 °C, which is consistent with the water content from the crystal structure and CHN elemental analysis. Accordingly, the same crystal used for the initial structure determination was subsequently heated in situ to 400 K using a dry nitrogen stream and held at temperature for two hours. SCXRD analysis at 400 K revealed the same structure but with no electron density located at the previous water molecule sites — indicating topotactic dehydration of the crystal. The crystal was then cooled back to 100 K and a low‐temperature structure of Ce(dmbipyO₂)₄[Au(CN)₄]₃ (Ce(dmbipyO₂)₄Au₃·0H₂O) was determined.

Consistent with the preservation of the single crystallinity, the crystal structure was not greatly perturbed by the water loss, and no significant changes in structure occurred (Figure S5). The Au···N_cyano interactions are of generally similar length, and no measurable changes are seen with the cation. This result shows that crystal growth from aqueous solution leads to a structural energetic minimum, which is unaffected after removal of the interstitial water molecules. Therefore, to observe the ideal arrangement of the [Ce(dmbipyO₂)₄]³⁺ and [Au(CN)₄]⁻ ions, without influence from interstitial water molecules, crystal growth from a nonaqueous solution was targeted.

2.5. Synthesis and Structure of Ln(dmbipyO₂)₄Au₂X

Combining 1 equiv of CeCl₃·7H₂O with 4 equiv of dmbipyO₂ and 3 equiv of ( ⁿ Bu₄N)[Au(CN)₄] in 8:1 MeCN:H₂O gave single crystals of [Ce(dmbipyO₂)₄][Au(CN)₄]₂Cl (Ce(dmbipyO₂)₄Au₂Cl) – a different compound than Ce(dmbipyO₂)₄Au₃·H₂O and containing no interstitial water molecules, thus providing a suitable comparison with the RE(bipyO₂)₄Au₃ series. In fact, the compound has a very similar structure to the RE(bipyO₂)₄Au₃ crystals, exhibiting square close‐packing of cations and anions in the ab plane, layers of which stack along the c‐axis (Figure 5). The [Ce(dmbipyO₂)₄]³⁺ cation also has essentially the same shape as the [RE(bipyO₂)₄]³⁺ cation — globular, S₄ symmetric, with a dodecahedral Ce(III) center (0.063% deviation by SHAPE analysis) — but now with proximal methyl groups present (Figure 5a). Comparing this with the earlier [Ce(dmbipyO₂)₄]³⁺ structure, there is clearly a minimal energy cost for Ce(III) to adopt either a cubic or dodecahedral geometry, which allows the cation to readily assume these two different shapes. As for the anionic group, the aurophilic {[Au(CN)₄]₃}³⁻ trimer observed in the RE(bipyO₂)₄Au₃ structure is replaced with a dumbbell‐shaped {[Au(CN)₄]₂Cl}³⁻ anion, where two [Au(CN)₄]⁻ anions are spanned by a single bridging chloride (Au1···Cl1 = 3.0563(16) Å and Au2···Cl1 = 2.9954(16) Å; Figure 5a). One rationale for the formation of this ‘dumbbell’ anion could be attributed to the protruding methyl groups of the cation; to borrow the language of Kitaigorodskii^{[ 45 ]} the methyl groups are ‘bumps’ requiring new ‘hollows’ in the tri‐anion group for close packing. Alternatively, the cavities left behind by square close‐packed [Ce(dmbipyO₂)₄]³⁺ cations can no longer accommodate three [Au(CN)₄]⁻ groups, but are now appropriately sized to fit the dumbbell‐shaped group.

Figure 5.

The structure of Ce(dmbipyO₂)₄Au₂Cl, showing a) the [Ce(dmbipyO₂)₄]³⁺ cation and the {[Au(CN)₄]Cl}³⁻ anion (with Au···Cl interactions represented by light blue lines) and b) the overall packing of the ions into a square close‐packed arrangement. Color scheme: Au, yellow; C, gray; Ce, tan; N, blue; O, red. Hydrogen atoms omitted for clarity.

This structural motif — a colinear, triatomic Au···X···Au bridge — is rare, with the most similar example being the {[AuBr₄]₂Br}³⁻ anion in the structure of (PyrH)₂[AuBr₄]Br (Pyr = pyrrole) with d_Au···Br = 3.4585(3) Å.^{[ 69 ]} Some bent Au···X···Au bridges or bridges not perpendicular to the AuL₄ planes are also known, and weakly interacting infinite Au···Cl···Au···Cl chains have been reported in the (dpaH₂)[Au(Cl)₄]Cl and (dpeH₂)[AuCl₄]Cl (dpa = 4,4′‐dipyridylethane; dpe = 4,4′‐dipyridylethene) structures (d_Au···Cl = 3.670(1) and 3.640(1) Å, respectively; longer than the sum of the van der Waals radii: 3.41 Å).^{[ 63} , ⁷⁰ , ^{71 ]} The Au···Cl distances observed in the Ce(dmbipyO₂)₄Au₂Cl structure are much shorter, although still longer than a typical equatorial Au‐Cl bond in a square‐planar [AuCl₄]⁻ anion, which is ca. 2.30 Å. The dumbbell anions are further aligned along the c‐axis due to direct stacking of the {[Ce(dmbipyO₂)₄][Au(CN)₄]₂Cl}^∞ layers (in contrast to the parallel displaced stacking of layers in the RE(bipyO₂)₄Au₃ structures); however, the distance between Au centers in neighboring {[Au(CN)₄]₂Cl}³⁻ groups is 5.1947(5) Å, precluding the presence of aurophilic contacts.

Taken together, the RE(bipyO₂)₄Au₃ and Ce(dmbipyO₂)₄Au₂Cl structures illustrate an important idea: given a suitably close‐packed multi‐component crystal structure, any change in the shape of one component could induce a likewise adaptation in the other's shape. In other words, shape complementarity is conserved. In this case, the shape adaptation occurs in situ — with a halide replacing an [Au(CN)₄]⁻ group to give a differently shaped tri‐anion — but this idea could be used in an a priori fashion, informing molecule replacement strategies. This is similar to Desiraju's^{[ 72 ]} shape and size mimicry strategy for the design of ternary and higher‐order multicomponent crystals.

Isomorphous Ln(dmbipyO₂)₄Au₂Cl compounds (Ln = Eu, Yb) could also be obtained using the appropriate lanthanide chloride salts; it was also found that the bromide analogues (Ln(dmbipyO₂)₄[Au(CN)₄]₂Br (Ln(dmbipyO₂)₄Au₂Br; Ln = Ce, Eu, Yb) could be synthesized using lanthanide bromide salts. Thus, a family of six compounds Ln(dmbipyO₂)₄[Au(CN)₄]X (Ln(dmbipyO₂)₄Au₂X; Ln = Ce, Eu, Yb; X = Cl, Br) was realized.

In both the chloride‐ and bromide‐containing structures, replacing Ce(III) with a smaller lanthanide caused the c‐axis to decrease in length and the a/b axes to increase slightly. Concomitantly, the Au···X contacts become slightly shorter, similar to the RE size‐dependent Au(III)···Au(III) distances observed in the RE(bipyO₂)₄Au₃ series. The structural information of the Ln(dmbipyO₂)₄Au₂X family is summarized in Table 2.

Table 2.

Cell lengths (Å) and selected interatomic distances (Å) for the Ln(dmbipyO₂)₄Au₂X (Ln = Ce, Eu, Yb; X = Cl, Br) compounds.

Ln	X	a/b‐axes length	c‐axis length	Au1···X	Au2···X
Ce	Cl	16.6109(7)	11.2464(7)	3.0563(16)	2.9954(16)
Ce	Br	16.6536(12)	11.2433(13)	3.1533(6)	3.1027(6)
Eu	Cl	16.6341(9)	11.1413(8)	3.0289(11)	2.9873(11)
Eu	Br	16.6670(6)	11.1171(7)	3.1257(12)	3.0883(12)
Yb	Cl	16.6819(12)	11.0431(12)	3.006(2)	2.985(2)
Yb	Br	16.6984(8)	11.0269(10)	3.1110(13)	3.0806(13)

2.6. Synthesis and Structure of Ca(tcmc)Au₂

We also sought to determine if Au(III) aurophilicity with [Au(CN)₄]⁻ units could persist in a di‐cationic analogue of the RE(bipyO₂)₄Au₃ structure, that is, to demonstrate whether a 3+ cation charge is necessary to induce Au(III) aurophilicity in the [Au(CN)₄]⁻ system or if simply the correct cation shape is sufficient. To target such a compound, comparison can be made to the Sc(bipyO₂)₄Au₃ structure, which is described as a square close‐packed array of similarly sized and shaped cation and anion groups. Therefore, to mimic this structure using 2+ cations, a metal‐ligand complex di‐cation that has the same size and shape as two stacked [Au(CN)₄]⁻ anions would be required, with the assumption that square close‐packing of such cations would generate cavities that are perfectly sized to contain aurophilic‐bound {[Au(CN)₄]₂}²⁻ dimers; effectively, the length and width of the cavity should be similar to that in Sc(bipyO₂)₄Au₃, but the height should be approximately 2/3 as much. To find cations meeting this criteria, the box model was used and a comparison was made to the box of a mock {[Au(CN)₄]₂}²⁻ di‐anion, obtained using the staggered Au1 and Au2 [Au(CN)₄]⁻ groups in the structure of Sc(bipyO₂)₄Au₃ . Further — as the importance of symmetry was illustrated by the RE(bipyO₂)₄Au₃ and Fe(bipyO₂)₃Au₃ structures — di‐cations with eight coordinate metal centres were sought to best match the symmetry of the [RE(bipyO₂)₄]³⁺ group.

According to the above outlined criteria, the Cambridge Structural Database (CSD) was searched, and the cation [Ca(tcmc)]²⁺ was identified (CSD Refcode ZAGLOI), where the 1,4,7,10‐tetrakis‐(carbamoylmethyl)‐1,4,7,10‐tetraazacyclododecane (tcmc) ligand consists of a cyclen macrocycle with four pendant amide groups that wrap around the Ca center.^{[ 73 ]} The [Ca(tcmc)]²⁺ cation's bounding box has L, M, and S dimensions of 10.00, 9.36, and 7.75 Å and S/L, S/M, and M/L ratios of 0.78, 0.83, and 0.98, respectively — very closely matching the size and shape of the mock {[Au(CN)₄]₂}²⁻ group: L, M, and S of 9.17, 9.05, and 7.21 Å and S/L, S/M, and M/L ratios of 0.79, 0.80, and 0.99, respectively.

Hence, 1 equiv of Ca(OTf)₂, 1 equiv of tcmc, and 2 equiv of KAu(CN)₄ were combined in water, generating single crystals of Ca(tcmc)[Au(CN)₄]₂ (Ca(tcmc)Au₂ ) that by SCXRD indeed revealed a di‐cationic analogue of the RE(bipyO₂)₄Au₃ structure (Figure 6). That is, square close‐packed layers of [Ca(tcmc)]²⁺ cations and {[Au(CN)₄]₂}²⁻ dimeric anions were observed, homologous to the arrangement of [RE(bipyO₂)₄]³⁺ and {[Au(CN)₄]₃}³⁻ groups in the RE(bipyO₂)₄Au₃ structure. The {[Au(CN)₄]₂}²⁻ dimers have a Au1···Au1’ separation of 3.6555(5) Å (Figure 6a), comparable to the aurophilic interactions found in the structures of [AuCl₂(dpa)][AuCl₄] (dpa = 2,2′‐dipyridylamine), [Au(tbpp)]₂[AuCl₄][AuCl₂]·2HOAc (H₂tbpp = 5,10,15,20‐tetrakis(4‐butoxyphenyl)porphyrin), and [AuCl₂(bpm)][AuCl₄] (bpm = 2,2′‐bipyrimidine) of 3.7467(1), 3.718, and 3.6540(3) Å, respectively.^{[ 23} , ^{24 ]} This result indicates that a 3+ cation is not required to induce alignment of [Au(CN)₄]⁻ anions and short Au(III)···Au(III) contacts, and that such assembly can be engineered simply through careful consideration of molecular shapes. A similar concept has been demonstrated using Ln[DOTA]⁻ complexes (H₄DOTA = 2,2′,2″,2″’‐(1,4,7,10‐tetraazacyclododecane‐1,4,7,10‐tetrayl)tetraacetic acid, which is the carboxylate analogue of tcmc); such complexes are often nine‐coordinate with an apical aqua ligand, breaking the C₄ symmetry of the Ln(III) center and limiting their magnetic properties, but shape‐complementary close‐packing with tetragonally shaped Me₄N⁺ cations allowed the uncapped Ln[DOTA]⁻ complexes to be preferentially accessed.^{[ 74 ]}

Figure 6.

The structure of Ca(tcmc)Au₂ , showing a) the [Ca(tcmc)]²⁺ cation and {[Au(CN)₄]₂}²⁻ dimer (Au(III)···Au(III) contact shown as gold line), b) the 1D chain of [Au(CN)₄]⁻ groups, where both short Au···N and Au···C contacts are shown with pink lines, to exemplify the Au···π nature of the interaction, c) the overall packing of the ions into a square close‐packed arrangement, akin to the packing observed with the RE(bipyO₂)₄Au₃ structures, and d) the H‐bonding network (red dashed lines) between the [Ca(tcmc)]²⁺ NH₂ moieties and the [Au(CN)₄]⁻ N_cyano groups. Color scheme: Au, yellow; Ca, green; C, gray; N, blue; O, red. Hydrogen atoms omitted for clarity except for the NH₂ groups in Figure 6d.

In the Ca(tcmc)Au₂ structure, the Ca center has a coordination geometry distorted between cubic and square anti‐prismatic (lying 59% along the minimal distortion path from cubic to square anti‐prismatic, as measured via SHAPE analysis), due specifically to a 24.53(6) ° rotation of the square of O‐donor amide ligands with respect to the base square of N‐donor amine ligands. Accordingly, the cation has overall C₄ symmetry. The Ca center lies 1.5407(16) Å above the plane of the amine N atoms, but only 1.0435(11) Å below the plane of the amide O atoms.

In contrast to the RE(bipyO₂)₄Au₃ structure, where the Au centers of the [Au(CN)₄]⁻ anions are aligned and the cyanide groups are fanned out in different directions, the two [Au(CN)₄]⁻ anions in the Ca(tcmc)Au₂ structure are parallel displaced and eclipsed. The {[Au(CN)₄]₂}²⁻ dimers stack infinitely in 1D chains, connected by interactions that at first glance appear to be Type II Au···N_cyano interactions; however, the closest approach to the Au centers is in fact from the adjacent C atoms (Au1···N1 and Au1···C1 distances of 3.636(2) and 3.359(2) Å, respectively; Figure 6b). Such close contacts have been observed before in the packing of cyclometallated Au(III) compounds,^{[ 31} , ⁷⁵ , ^{76 ]} with distances of 3.247–3.310 Å occurring between Au centers and the C atoms of aryl groups (sum of the van der Waals radii: 3.44 Å).^{[ 63 ]} Based on this precedent, this can be considered a Au···π type interaction, with the Au center interacting with the π cloud of the cyanide group. Au···π aryl interactions have been reviewed previously.^{[ 77 ]} Also, as shown in Figure 6d, three of the N_cyano groups H‐bond with the amide NH₂ groups of the [Ca(tcmc)]²⁺ cations: N1···H6B‐N6, N3···H6A‐N6, N4···H8A‐N8, and N4···H8B‐N8, with d_N···N = 3.101(3), 3.134(3), 3.150(3), and 3.167(3) Å, respectively.

One of our initial assumptions for inducing Au(III) aurophilicity was that the potential for H‐bonding should be removed, allowing for the weaker interactions to manifest in the structure. The [Ca(tcmc)]²⁺ cation suggests this to not be necessarily correct; however, the importance of the location of the H‐bond donor sites is readily evident. In the Ca(tcmc)Au₂ structure, the H‐bonds are situated such that they support the {[Au(CN)₄]₂}²⁻ dimer (Figure 6d), but this synergistic behavior can only occur if the cation first has the right shape. Comparison can be made to the previously reported [Ni(en)₃][Au(CN)₄]₂, which also makes use of a 2+ cation with peripheral NH₂ H‐bond donor groups.^{[ 35 ]} However, like the octahedral [Fe(bipyO₂)₃]³⁺ cation, the [Ni(en)₃]²⁺ cation has threefold symmetry, and no Au(III)···Au(III) contacts occur in the structure – only self‐aggregation by Au···N_cyano interactions is observed.

2.7. Synthesis and Structure of Ca(12‐crown‐4)₂Au₂

To show that cation shape plays a decisive role in the manifestation of short Au(III)···Au(III) contacts in Ca(tcmc)Au₂ , a compound featuring a similar di‐cation — differing only in shape — was sought. Searching the CSD provided such a candidate, the [Ca(12‐crown‐4)₂]²⁺ cation (CSD Refcode ECOQUI), consisting of a Ca(II) center sandwiched between two 12‐crown‐4 ligands.^{[ 78 ]} Like the [Ca(tcmc)]²⁺ cation, [Ca(12‐crown‐4)₂]²⁺ also features a Ca(II) center and a similar fourfold symmetric, macrocyclic base; however, its bounding box has L, M, and S dimensions of 9.02, 8.57, and 8.53 Å, respectively, corresponding to a taller square prism with a smaller base – mismatched with the shape of the {[Au(CN)₄]₂}²⁻ dimer such that even if the cations assembled into a square close‐packed pattern (as may be in their nature, due to their internal fourfold symmetry), the cavity would be too small and mismatched in height to efficiently fit the {[Au(CN)₄]₂}²⁻ dimer.

Thus, 1 equiv of Ca(OTf)₂ was combined in water with 2 equiv of KAu(CN)₄ and an excess of 12‐crown‐4, providing single crystals of Ca(12‐crown‐4)₂[Au(CN)₄]₂ (Ca(12‐crown‐4)₂Au₂ ). SCXRD analysis found this compound to have a structure similar to that of Fe(bipyO₂)₃Au₃ : layers of [Ca(12‐crown‐4)₂]²⁺ cations arranged between 2D sheets of [Au(CN)₄]⁻ anions (Figure 7). This result confirms that the specific shape of the [Ca(tcmc)]²⁺ cation is important in allowing Au(III)···Au(III) alignment of the [Au(CN)₄]⁻ groups to emerge.

Figure 7.

The structure of Ca(12‐crown‐4)₂Au₂ , showing a) the [Ca(12‐crown‐4)₂]²⁺ cation and {[Au(CN)₄]₂}²⁻ dimer (Au···π interactions shown with pink lines), b) the stacking of cations (represented by green spheres) in between 2D [Au(CN)₄]⁻ sheets, and c) the structure of the 2D sheets, generated via Type I Au···N_cyano interactions (light blue lines) between {[Au(CN)₄]₂}²⁻ dimers. Color scheme: Au, yellow; Ca, green; C, gray; N, blue; O, red. Hydrogen atoms omitted for clarity.

The [Ca(12‐crown)₂]²⁺ cation sits on an inversion center and exhibits whole molecule disorder. Thus, an idealized nondisordered structure was created to allow for Hirshfeld surface analysis. The coordination geometry of the Ca center is square antiprismatic (1.095% deviation by SHAPE analysis), providing a cation with overall S₈ symmetry. In contrast to the [Ca(tcmc)]²⁺ cation, the Ca center sits 1.413(2) and 1.476(2) Å above and below the planes of the 12‐crown‐4 oxygen atoms. This difference — as well as the lack of pendant NH₂ groups — is why the [Ca(12‐crown‐4)₂]²⁺ cation is much taller with a smaller square base than the otherwise similar [Ca(tcmc)]²⁺ cation.

The 2D sheets are composed of {[Au(CN)₄]₂}²⁻ dimers assembled via Au···π interactions (Au1···N1 and Au1···C1 distances of 3.421(2) and 3.388(3) Å, respectively; Figure 7c). Lastly, the dimers are connected into sheets via Type I Au···N_cyano interactions (Figure 7c; Au1···N3 distance = 3.077(3) Å).

2.8. Computational Analysis

The nature of the short Au(III)···Au(III) contacts in the Sc(bipyO₂)₄Au₃ and Ca(tcmc)Au₂ crystal structures can be interrogated computationally. Previously, we showed with ab initio calculations at the MP2 level that weak aurophilic interactions were predicted within a {[Au(CN)₄]₂}²⁻ dimer derived from the crystal structure of Lu(bipyO₂)₄Au₃ ^{[ 28 ]}; generally, the treatment of electron correlation at the MP2 level of theory gives a good prediction for the existence of interactions that are dispersive in nature.^{[ 32 ]} To expand on this using the new structures, a combination of MP2, DFT, and CCSD(T) methods were harnessed to understand the nature of the interactions.

First, using DFT (ωB97X‐D/def2‐TZVP, def2‐TZVP ECP for Au), a potential energy curve (PEC) was constructed for the {[Au(CN)₄]₃}³⁻ trimer from the Sc(bipyO₂)₄Au₃ structure (Figure 8a), starting at the crystal structure geometry and symmetrically varying the distances from the two outer [Au(CN)₄]⁻ molecules to the middle [Au(CN)₄]⁻ molecule (see Supporting Information for details). In contrast to the earlier MP2 result, this gave a PEC that exhibited no minimum, likely due to strong Coulombic repulsion between the anions. However, upon inclusion of the [Sc(bipyO₂)₄]³⁺ cation, a minimum of ca. 24 kJ mol⁻¹ was found, located within ca. 0.2 Å of the crystal structure geometry and corresponding to Au···Au distances of 3.53 and 3.61 Å. This highlights the critical role of the cation in stabilizing the observed trimeric arrangement.

Figure 8.

Calculated DFT potential energy curves (PECs) for a) {[Au(CN)₄]₃}³⁻ trimers from the Sc(bipyO₂)₄Au₃ crystal structure and b) {[Au(CN)₄]₂}²⁻ dimers from the Ca(tcmc)Au₂ crystal structure, with PECs excluding the cation group shown in red and those including the cation group shown in blue (shown in the insets).

A relevant comparison can be drawn from the study on Au(III) aurophilicity conducted by Blasco et al,^{[ 32 ]} where both MP2 and CCSD(T) calculations predicted a minimum for neutral Au(CH₃)₃(NH₃) dimers, but PECs for anionic [Au(N₃)₄]⁻ or cationic [Au(CH₃)₂(NH₃)₂]⁺ pairs were either purely repulsive or exhibited minima only at the MP2 level. Electrostatic repulsion typically dominates interactions between charged fragments, obscuring potential Au(III)···Au(III) aurophilic stabilization. To probe this, local energy decomposition (LED)^{[ 79 ]} at the DLPNO‐CCSD(T)/def2‐TZVP level was performed on [Au(CN)₄]⁻ dimers (excluding counterions) at the PEC minimum. While the electrostatic term was strongly repulsive (+138 kJ mol⁻¹), a dispersion stabilization of –8 kJ mol⁻¹ was identified, indicating a weak but noticeable aurophilic interaction between two Au(III) atoms.

Next, PECs for the Ca(tcmc)[Au(CN)₄]₂ system were constructed by varying the distance between the two [Au(CN)₄]⁻ units, both with and without the cation (Figure 8b). In the absence of the cation, the PEC was repulsive at all distances, whereas inclusion of the cation introduced a shallow minimum (ca. 10 kJ mol⁻¹) at an Au···Au separation of ca. 3.75 Å (compared to crystallographic value of 3.6555(5) Å). This again underscores the critical role of the cation in stabilizing the assembly. However, LED analysis at the DLPNO‐CCSD(T) level revealed a nearly negligible (<1 kJ/mol) dispersion component, suggesting that aurophilic stabilization is absent in this system —consistent with the significantly larger Au(III)···Au(III) distance compared to Sc(bipyO₂)₄[Au(CN)₄]₃ .

To further validate this, a PEC for the Ca(tcmc)Au₂ system was constructed at the MP2/def2‐TZVP level with a second‐order DKH correction, analogous to the previously reported analysis of Lu(bipyO₂)₄Au₃ .^{[ 28 ]} Contrary to the Lu(bipyO₂)₄Au₃ system, the interaction was repulsive at all distances (Figure S6). Thus, although the crystal structure shows alignment of [Au(CN)₄]⁻ anions and Au(III)···Au(III) contacts within the sum of the van der Waals radii, computational efforts predict that this is not stabilized by aurophilicity but is a result of packing. Indeed, we suggest that aurophilicity between the [Au(CN)₄]⁻ anions in the Ca(tcmc)Au₂ structure may be discouraged by their eclipsed configuration, which itself is enforced by H‐bonding with the pendant amides of the tcmc macrocycle. Additionally, it may be that a high cation charge (e.g., 3+ for the cation in the Lu(bipyO₂)₄Au₃ structure) does assist in manifesting Au(III) aurophilic interactions, by inductively relieving the repulsion between the anions.

2.9. Hirshfeld Surface Analysis

To understand broadly the crystal packing of the six archetypical crystal structures, Hirshfeld surface analysis was conducted using the software program CrystalExplorer (see Supporting Information for details).^{[ 80 ]} Hirshfeld surface analysis can provide a wealth of visual and quantitative information on intermolecular contacts and crystal packing, thus giving a useful, holistic perspective of the crystal structures. Surfaces were generated for the cations in the Sc(bipyO₂)₄Au₃ , Fe(bipyO₂)₃Au₃ , Ce(dmbipyO₂)₄Au₃·H₂O, Ce(dmbipyO₂)₄Au₂Cl, Ca(tcmc)Au₂ , and Ca(12‐crown‐4)₂Au₂ structures and the results of the analysis are shown in Figure S7 (see Supporting Information for descriptions of surface properties and plots). The volumes of the Hirshfeld surfaces are also tabulated in Table 3, alongside the min, mean, and max values of the d _norm surface property. Each point on the Hirshfeld surface has a d _norm value; if the distance from the nearest surface interior atom to the nearest exterior atom is less than the sum of their van der Waals radii, then d _norm < 0 and vice versa. Calculated packing coefficients (β) of the corresponding crystal structure are also included for each cation; packing coefficients are a non‐Hirshfeld surface property that simply quantifies how efficiently the crystal structures are packed, being the ratio of the sum of the molecular van der Waals volumes to the unit cell volume (the fraction of space occupied by the molecules; see Supporting Information for calculation details).^{[ 45 ]}

Table 3.

Hirshfeld surface volumes (Å), d _norm min, mean, and max values, and packing coefficients (β).

Group	Volume	d norm Min < Mean < Max	Β
[Sc(bipyO2)4]3+	874.85	−0.339 < 0.440 < 1.311	0.680
[Fe(bipyO2)3]3+	696.63	−0.386 < 0.526 < 2.268	0.644
[Ce(dmbipyO2)4]3+ [a]	1089.72	−0.563 < 0.527 < 1.755	0.662
[Ce(dmbipyO2)4]3+ [b]	1122.41	−0.260 < 0.532 < 1.963	0.632
[Ca(tcmc)]2+	502.47	−0.374 < 0.489 < 1.313	0.680
[Ca(12‐crown‐4)2]2+	460.78	−0.132 < 0.569 < 1.441	0.650

^[a]From the Ce(dmbipyO₂)₄Au₃·H₂O structure.

^[b]From the Ce(dmbipyO₂)₄Au₂Cl structure.

The analysis shows that H···N contacts account for between 23% and 49% of contacts and are the primary mode of interaction in all but the [Ce(dmbipyO₂)₄]³⁺ structures, where the larger size and methyl‐group decoration of the cation extremities cause an increase in the number of H···H contacts. The H···N contacts can be understood as weak H‐bonds of the type C‐H···N_cyano, except in the case of the [Ca(tcmc)]²⁺ cation, where N‐H···N_cyano H‐bonds involving the peripheral amide groups are also present.

To analyze the packing efficiency of the structures, the Hirshfeld surface d _norm values can be compared (Table 3). The [Fe(bipyO₂)₃]³⁺ and two [Ce(dmbipyO₂)₄]³⁺ cations show the largest d _norm max values, indicating looser packing around the molecules, although the cation of the Ce(dmbipyO₂)₄Au₃·H₂O structure also shows the most negative d _norm min of all the compounds, in line with strong H‐bonding involving the aqua ligand. On the other hand, the [Ca(12‐crown‐4)₂]²⁺ cation has the least negative d _norm min, revealing a lack of any strong interactions; likewise, the cation has the largest d _norm mean. The [Sc(bipyO₂)₄]³⁺ and [Ca(tcmc)]²⁺ cations — representing the two isoreticular compounds featuring stacks of [Au(CN)₄]⁻ anions — have the smallest d _norm max and d _norm mean values, both indicating efficient packing. These packing trends are corroborated by a visual inspection of the fingerprint plots of the compounds, where the [Fe(bipyO₂)₃]³⁺ and two [Ce(dmbipyO₂)₄]³⁺ cations have ‘smeared’ fingerprints — with contacts receding into the top‐rightmost corner of the plot — but the [Sc(bipyO₂)₄]³⁺, [Ca(12‐crown‐4)₂]²⁺, and [Ca(tcmc)]²⁺ cations have compact, blot‐like fingerprints. For the cations in Fe(bipyO₂)₃Au₃ and Ce(dmbipyO₂)₄Au₃·H₂O the inefficient packing stands to reason, as the ions therein do not have complementary shapes. However, a smeared fingerprint is unexpected for the case of Ce(dmbipyO₂)₄Au₂Cl, which has analogous packing to Sc(bipyO₂)₄Au₃ and Ca(tcmc)Au₂ . Inspection of the crystal structure reveals the reason for this discrepancy: the ‘wings’ of the dmbipyO₂ ligands form C‐shaped claws, and the cations stack together such that opposing claws interlock to enclose a ca. 28 ų void space (Figure S8), which is too small to contain solvent molecules (MeCN molecular volume = ca. 57 ų).^{[ 81 ]} This void space would contribute to the large d _norm max and the smeared fingerprint.

Comparing the packing coefficients of the compounds, the isoreticular Sc(bipyO₂)₄Au₃ and Ca(tcmc)Au₂ structures both have β = 0.680, higher than the rest of the crystal structures. To further examine this point, packing coefficients were calculated for 30 other tetracyanoaurate(III) salts/coordination polymers assembled using metal‐ligand complex cations, as found in the CSD (Table S2). It was found that the average packing coefficient was 0.64(2), with Sc(bipyO₂)₄Au₃ and Ca(tcmc)Au₂ having higher packing coefficients than 93% of those compounds – further indicating the tightly packed nature of the aurophilic‐based structures. Taken together, this analysis suggests that for sufficiently close‐packed crystal structures (small d _norm max/mean values; high β), conservation of the efficient packing will be advantageous. Using Sc(bipyO₂)₄Au₃ as the example: the [Sc(bipyO₂)₄]³⁺ cation can be substituted with a [Ce(dmbipyO₂)₄]³⁺ cation and the packing is conserved — at the expense of a modified tri‐anion — or it can be substituted with a smaller, less charged (but similarly shaped) [Ca(tcmc)]²⁺ cation, and the packing is still conserved, due to packing of the di‐cation with only a di‐anion.

2.10. Photoluminescence Measurements

Short Au(III)···Au(III) contacts in organogold(III) compounds are of interest in that they might impart unique photoluminescent properties to the complexes; however, the [Au(CN)₄]⁻ groups in the RE(bipyO₂)₄Au₃ structures are nonemissive, as previously shown in the RE = Lu case.^{[ 28 ]} This is further verified using the RE = Y and La compounds herein (photoluminescence spectra shown in Figure S9), which like Lu(bipyO₂)₄Au₃ exhibit only phosphorescence from the bipyO₂ ligands at 77 K. Sc(bipyO₂)₄Au₃ is nonemissive, likely due to the decreased heavy metal effect from the Sc center.

However, as a comparison to our previous investigation of the emission properties of the Eu(bipyO₂)₄Au₃ compound, the photoluminescent properties of the new Eu(dmbipyO₂)₄Au₂X (X = Cl, Br) series were examined. Accordingly, crystals of the Eu(dmbipyO₂)₄Au₂X (X = Cl, Br) compounds were ground, their phase purity confirmed using powder X‐ray diffraction (Figure S10), and their photoluminescence studied at both 300 and 77 K (Figure 9).

Figure 9.

Normalized photoluminescence spectra of the Eu(dmbipyO₂)₄Au₂X (X = Cl, Br) compounds. Top (X = Br): excitation spectra shown in blue (300 K; λ _em = 617 nm) and light blue (77 K; λ _em = 617 nm); emission spectra shown in red (300 K; λ _ex = 320 nm) and pink (77 K; λ _ex = 310 nm). Bottom (X = Cl): excitation spectra shown in blue (300 K; λ _em = 617.5 nm) and light blue (77 K; λ _em = 617.5 nm); emission spectra shown in red (300 K; λ _ex = 320 nm) and pink (77 K; λ _ex = 310 nm).

Both compounds show essentially the same spectral features and exhibit similar photoluminescence properties to the related compound Eu(bipyO₂)₄Au₃ , being classified as ligand‐sensitized emission via Eu(III) ⁵D₀→⁷F_J (J = 1–4) transitions.^{[ 28 ]} In the 300 K excitation spectra, excitation via the ligand π–π* transition is seen at 320 nm (cf. 325 nm for Eu(bipyO₂)₄Au₃ ) and a charge transfer band is seen at ca. 380 nm. The charge transfer band occurs also at 380 nm for Eu(bipyO₂)₄Au₃ , but the intensity of this band is noticeably reduced for the Eu(dmbipyO₂)₄Au₂X (X = Cl, Br) compounds. Particularly in the X = Cl case, the charge transfer appears as only a shoulder of the ligand‐based transition. Like Eu(bipyO₂)₄Au₃ , the ligand‐based π–π* band shifts to 310 nm upon cooling to 77 K. As for the emission spectra, the Eu(III) f‐f transitions show their characteristic line‐like structure regardless of the X anion and the temperature; the ⁵D₀→⁷F₂ transition at 617–617.5 nm is the most prominent, and the ⁵D₀→⁷F₀ transition at 570—585 nm is absent, in line with the D_2d symmetry of the Eu centers.^{[ 68 ]}

The 300 K photoluminescence quantum yield (PLQY) of each compound was measured using an excitation wavelength of 325 nm, measuring 53(3) and 47(3) % for X = Cl and Br, respectively. To the best of our knowledge, the performance of the dmbipyO₂ ligand as an Eu photoluminescence sensitizer has not yet been studied, but such high PLQY values show improvement over similar Eu‐bipyO₂ compounds like Eu(bipyO₂)₄Au₃ , [Eu(bipyO₂)₄][ClO₄]₃, and [Eu(tta)₃(bipyO₂)], which have PLQYs of 29(3), 15, and 46%, respectively, (Htta = 2‐thenoyltrifluoroacetone), suggesting that exploring the photoluminescent properties of Eu‐dmbipyO₂ compounds further may be a promising venture.^{[ 28} , ⁴¹ , ^{66 ]}

3. Conclusion

In summary, sixteen compounds have been synthesized and characterized, each providing important insights into Au(III) aurophilicity between [Au(CN)₄]⁻ anions through comparison to the archetypical RE(bipyO₂)₄Au₃ structure. The RE(bipyO₂)₄Au₃ system exhibits uncommonly short Au(III)···Au(III) interactions and is able to accommodate rare earth elements from Sc to La at the cation metal site. This distinctive behavior is likely due to the tight packing bestowed by the complementary shapes of the [RE(bipyO₂)₄]³⁺ cation and {[Au(CN)₄]₃}³⁻ tri‐anions, as confirmed using packing coefficients and Hirshfeld surface analysis. Aurophilic interactions cannot be accessed using other 3+ cations that have different shapes (i.e., the [Fe(bipyO₂)₃]³⁺ and disk‐shaped [Ce(dmbipyO₂)₄]³⁺ cations), but similar alignment of the [Au(CN)₄]⁻ anions and consequently short Au(III)···Au(III) contacts did manifest in the Ca(tcmc)Au₂ structure because of the carefully chosen cation shape. Overall, viable Au(III)···Au(III) aurophilic interactions were generated in Sc(bipyO₂)₄Au₃ , Y(bipyO₂)₄Au₃ , and La(bipyO₂)₄Au₃ , with the shortest such distance to date of 3.3351(3) Å reported herein. The stabilization of structures with the 3+ charged [RE(bipyO₂)₄]³⁺ cation through Au(III) aurophilic interactions was also supported by calculations at the CCSD(T) level of theory.

Further, we note that the results here have implications that are applicable beyond just the topic of Au(III) aurophilicity. The RE(bipyO₂)₄Au₃ structure can clearly act as a structural guide for the creation of isoreticular analogues, as illustrated by Ca(tcmc)Au₂ and the Ln(dmbipyO₂)Au₂X series, structures which share the same 2D tiling motif as RE(bipyO₂)₄Au₃ . In general, it can be predicted that an isoreticular compound will likely be generated if based on a parent system that has a sufficiently high packing efficiency, and if components are chosen to have complementary shape such that the same efficient packing motif can be realized. This ultimately is based on a shape‐complementarity concept, facilitated by an understanding of tiling patterns and close‐packing, and informed by molecular bounding boxes as descriptors of molecular shape. A similar approach has been demonstrated by Mirkin et al.^{[ 82 ]} who focused on creating space‐filled colloidal crystals through the assembly of shape‐complementary polyhedral nanocrystals. This work also shows the significance of the “cavities” left behind by packed cations, which can template the assembly of anions and host those anion groups in close proximity. Similar synergy has been shown with metal‐organic framework (MOF) systems, where templating from guests can allow certain MOF structures to arise, but reactivity between guests can also be mediated by the pore environment of MOFs.^{[ 83 ]}

Overall, the results describe a shape complementarity approach for inorganic crystal systems that is useful for targeting molecules to aggregate via weak and often inaccessible noncovalent interactions.

Supporting Information

Experimental details, synthetic procedures, crystallographic tables and figures, additional luminescence spectra, thermogravimetric curves, and PXRD data are included within the Supporting Information. Deposition Numbers 2375664–2375677 contain the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service. The authors have cited additional references within the Supporting Information.^{[ 84} , ⁸⁵ , ⁸⁶ , ⁸⁷ , ⁸⁸ , ⁸⁹ , ⁹⁰ , ⁹¹ , ⁹² , ⁹³ , ⁹⁴ , ⁹⁵ , ⁹⁶ , ⁹⁷ , ⁹⁸ , ⁹⁹ , ¹⁰⁰ , ¹⁰¹ , ¹⁰² , ¹⁰³ , ¹⁰⁴ , ¹⁰⁵ , ¹⁰⁶ , ¹⁰⁷ , ¹⁰⁸ , ¹⁰⁹ , ¹¹⁰ , ¹¹¹ , ¹¹² , ^{113 ]}

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

Supporting Information

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.