**1. Introduction**

According to Kitaigorodskii's principle of close packing [1–5], molecules in crystals tend to dovetail and pack as e fficiently as possible in order to maximize attractive dispersion forces and to minimize free energy. In other words, void space in crystals is always unfavorable. Thus, the construction of porous materials from discrete organic molecules (i.e., molecular porous materials (MPMs)) demands some special tactics [6–11]. For example, the packing of molecules specifically designed to bear su fficiently large and dimensionally fixed inner cavities or clefts (e.g., molecular cages and bowl-shaped compounds) can lead to porous structures [12–14].

Another viable synthetic strategy towards MPMs is to employ molecules with bulky, divergent and/or awkward shapes so that they no longer have the ability to pack tightly. Molecules such as 4-*p*-Hydroxyphenyl-2,2,4-trimethylchroman (Dianin's compound) [15,16], tris(*o*-phenylenedioxy) cyclotriphosphazene (TPP) [17–19] and 3,3,4,4-tetrakis(trimethylsilylethynyl)biphenyl (TTEB) [20] are well-known for producing MPMs merely as a consequence of frustrated packing, even though they do not have pre-fabricated molecular free volumes.

We have now expanded this idea to a family of tetrahedral molecules substituted at the four vertices with bulky groups. Here, we report the synthesis and structural investigation of tetrakis(4-((trimethylsilyl)ethynyl)phenyl)methane (**TMS4TEPM**) and 1,3,5,7-tetrakis (4-((trimethylsilyl)ethynyl)phenyl)adamantane (**TMS4TEPA**) (Scheme 1). By affixing large trimethylsilylethynyl (TMS-acetylenyl) moieties to the parent tetraphenylmethane (TPM) and 1,3,5,7-tetraphenyladamantane (TPA) core units, our aim was to disturb close-packing and to realize more open crystalline solids.

**Scheme 1.** Structural formulas of tetrakis(4-((trimethylsilyl)ethynyl)phenyl)methane (**TMS4TEPM**) and 1,3,5,7-tetrakis(4-((trimethylsilyl)ethynyl)phenyl)adamantane (**TMS4TEPA**).

Even though molecular shape is of primary importance in crystal packing, it is not the only structure-directing factor. The presence of functional units that can partake in directional and energetically significant non-covalent interactions has a major influence on molecular arrangement. With tectons (i.e., molecules featuring multiple peripheral binding sites) [21–24], the structure is built up so as to saturate the maximum amount of interactions, which is usually accompanied by compromises regarding dense-packing. Their association induces the assembly of networks where each molecule is positioned, through directional molecular recognition events, in a definite way with respect to its neighbors. Moreover, unlike van der Waals contacts, intermolecular point contacts consume only a limited amount of molecular surface, thereby leaving more usable surface. In this context, a grea<sup>t</sup> body of work has been done with hydrogen-bonding tectons to build so-called hydrogen-bonded organic frameworks (HOFs) [25–28]. Some notable examples include triptycenetrisbenzimidazolone (TTBI) [29], triaminotriazine-functionalized spirobifluorene [30,31] and polyfluorinated triphenylbenzene equipped with pyrazole [32].

Molecular tectonics based on halogen bonding (XB) is still in its infancy [33,34]. We therefore decided to modify the TPM and TPA scaffolds and transform them into new tecton-like entities, tetrakis(4-(iodoethynyl)phenyl)methane (**I4TEPM**) and 1,3,5,7- tetrakis(4-(iodoethynyl)phenyl)adamantane (**I4TEPM**) (Scheme 2). When iodine is directly bonded to an *sp*-hybridized carbon, it is strongly polarized, resulting in a more pronounced electron-deficient region (i.e., σ-hole) at the tip along the C–I bond axis [35–38]. The iodoethynyl functionality is, therefore, a perfect candidate for σ-hole/XB interactions. Alhough largely overlooked in molecular tectonics and crystal engineering, it can direct the assembly of network structures through C≡C–I···(C≡C) interactions (wherein the ethynyl π system acts as the XB acceptor) [39–41]. These T-shaped contacts frequently lead to *zigzag* chain motifs and are topologically parallel to those formed by C≡C–H···(C≡C) and C≡C–Br···(C≡C) contacts [37,42–52], but preferably serve as a stronger counterpart. Additional features that make the iodoethynyl unit well-suited for devising molecular building blocks include its structural rigidity, steric openness and core expanding ability.

**Scheme 2.** Structural formulas of tetrakis(4-(iodoethynyl)phenyl)methane (**I4TEPM**) and 1,3,5,7- tetrakis(4-(iodoethynyl)phenyl)adamantane (**I4TEPM**).

### **2. Results and Discussion**

The four molecules of interest were obtained according to the synthetic pathways shown in Schemes 3 and 4. Starting with commercially available tetraphenylmethane, **TMS4TEPM** was prepared in two steps (tetra-*para*-bromination followed by coupling with trimethylsilylacetylene) with an overall yield of 78%. The synthesis of **TMS4TEPA** required three steps (Friedel-Crafts reaction of 1-bromoadamantane and benzene, tetra-*para*-iodination followed by coupling with trimethylsilylacetylene), and the yield over these three steps was 50% (with respect to 1-bromoadamantane).

**Scheme 3.** Synthetic route to **TMS4TEPM** and **I4TEPM**.

Both **I4TEPM** and **I4TEPA** were accessible from the corresponding TMS derivatives, **TMS4TEPM** and **TMS4TEPA**, via one-pot/in situ desilylative iodination using silver(I) fluoride and *N*-iodosuccinimide. This direct trimethylsilyl-to-iodo transformation allowed us to avoid potentially unstable ethynyl intermediates and to achieve the target compounds in moderate yields (56% and 63%, respectively). Even though the 1H and proton-decoupled 13C-NMR spectra of these four-fold symmetric tetraiodoethynyl species are quite simple, the signals display considerable solvent dependency due to

their XB-based complexation ability, with the alkynyl carbon bonded to iodine being most strongly affected (**I4TEPM**: 7.0 ppm in CDCl3 versus 18.4 ppm in DMSO-*d*6, **I4TEPA**: 6.2 ppm in CDCl3 versus 17.0 ppm in DMSO-*d*6). It is also worth mentioning that the 1H-NMR spectrum of **I4TEPA** exhibits conspicuous second order (leaning/roofing) effects.

**Scheme 4.** Synthetic route to **TMS4TEPA** and **I4TEPA**.

Crystals of **TMS4TEPM** suitable for single-crystal X-ray analysis were obtained by slow evaporation of either tetrahydrofuran/ethanol or chloroform/ethanol solution. For **TMS4TEPA**, X-ray quality crystals could be harvested from hexane, heptane, heptane/dichloromethane or chloroform/ethanol. As anticipated, structural determination revealed that both are somewhat porous in nature (14.9% and 14.5% free volume, respectively). They, however, do not form empty-channel structures; instead, they have disconnected spatial voids or "porosity without pores", as described by Barbour (Figure 1) [53]. The overall packing is mainly mediated by extensive phenyl embraces.

In order to ge<sup>t</sup> some insight about the electron density/charge distribution over the free tetraiodoethynyl tectons and the degree of activation of XB donor sites (i.e., iodine atoms) delivered by *sp*-hybridized carbons [35–38], their molecular electrostatic potential (MEP) maps were generated (Figure 2). As expected, both **I4TEPM** and **I4TEPA** were found to have well-built σ-holes (+172.4 and +170.7 kJ/mol, respectively) on each iodine atom. Indeed, these σ-hole potential values are significantly higher than those of other closely-related tetra-halogenated molecules (see Supplementary Materials, Figure S33).

We then tried to grow crystals of **I4TEPM** and **I4TEPA** but were successful only with the former. The structural analysis of **I4TEPM** crystals (harvested from hexanes) showed that the molecules are arranged in stacks which, in turn, are linked together by C≡C–I···(C≡C) halogen bonds, with near orthogonal approach of C–I donors towards C≡C triple bonds (detailed geometrical data are given in Table 1). In each **I4TEPM** molecule, only two iodoethynyl arms participate in these T-shaped contacts, and the remaining two form weak C≡C–I···<sup>π</sup>(phenyl) interactions. The extended (and possibly cooperative) *zigzag* arrays of the C≡C–I···<sup>π</sup>(ethynyl) interactions ultimately make ladder-like motifs between individual molecular rows, leading to an infinite two-dimensional network (Figure 3 left). **I4TEPM**shares these packing features with its bromo analog, tetrakis(4-(bromoethynyl)phenyl)methane (**Br4TEPM**) [42], but not with tetrakis(4-ethynylphenyl)methane (**TEPM**), which forms an interwoven diamondoid net [44].

**Figure 1.** Crystal structures of **TMS4TEPM** and **TMS4TEPA**. (from left) Single molecules, overall packing and phenyl embraces (representative structures are shown from disordered structures).

**Figure 2.** Molecular electrostatic potential (MEP) surfaces of the free tetraiodoethynyl tectons, **I4TEPM** and **I4TEPA**. Both plots have been set to the same color scale for visual comparison. Range: from −80 kJ/mol (red) to +175 kJ/mol (blue).

**Figure 3.** Crystal structure of **I4TEPM**, showing halogen bonding (XB)-driven network formation (left) and void space in overall packing (right).

In contrast to the structure of **TMS4TEPM** with isolated voids, **I4TEPM** possesses one-dimensional channels along the crystallographic *b* axis (Figure 3 right). These channels account for 26.5% of the crystal volume, which is roughly twice as high as that of **TMS4TEPM**. Another point worth emphasizing is that the precursor molecules, tetraphenylmethane (**TPM**), tetrakis(4-bromophenyl)methane (**Br4TPM**) and tetrakis(4-iodophenyl)methane (**I4TPM**), all form non-porous structures (see Supplementary Materials, Figure S34), highlighting the effectiveness of our strategy.

Since MPMs are usually held together by relatively weak interactions, they are not as rigid and robust as zeolites, metal-organic frameworks (MOFs) or covalent-organic frameworks (COFs). In most cases, attempts at activation (i.e., removal of entrapped gues<sup>t</sup> molecules) cause structural disintegration. Hence, the real challenge lies in attaining permanently porous molecular materials that can behave analogously to framework-type solids. Most importantly, **I4TEPM**, sustained primarily by the iodoethynyl catemer motif (i.e., the infinite C≡C–I··· C≡C–I··· synthon), can maintain its structural integrity upon gues<sup>t</sup> solvent loss, indicating its potential to exhibit permanent porosity.

In addition to tectonic construction, we also wanted to test the suitability of **I4TEPM** in modular construction by co-crystallizing it with appropriate Lewis basic (i.e., XB-accepting) co-formers, in order to realize multicomponent architectures. With tetraphenylphosphonium halide salts (Ph4P<sup>+</sup>X<sup>−</sup>; X− = Cl<sup>−</sup>, Br<sup>−</sup>, I−), it readily a fforded diamondoid (**dia**) frameworks, but interpenetration and the inclusion of bulky Ph4P+ cations gave rise to highly compact arrangements within those solids [54]. As a charge-neutral co-crystallizing partner, our first choice was pyridine, one of the simplest XB acceptors, even though it cannot lead **I4TEPM** to a polymeric assembly. We managed to ge<sup>t</sup> a binary crystalline material (confirmed by IR, NMR and TGA) but the structural characterization was not successful, as those crystals were quite fragile and rapidly deteriorated during data collection. This intrigued us to try out other Lewis basic/coordinating solvents with multiple bond forming ability. In three cases, with tetrahydrofuran (THF), dimethyl sulfoxide (DMSO) and 1,4-dioxane, **I4TEPM** afforded crystalline binary solids.

Crystallization of **I4TEPM** in THF/methanol afforded crystals of **I4TEPM**·2THF where each THF molecule forms two halogen bonds in a bifurcated manner and connect adjacent **I4TEPM** molecules together, thereby forming a one-dimensional twisted ribbon-like architecture (Figure 4a left). The resulting lattice comprises isolated voids that account for 14.4% of unit cell volume (Figure 4a right).

**Figure 4.** Crystal structures of (**a**) **I4TEPM**·2THF, (**b**) **I4TEPM**·2DMSO and (**c**) **I4TEPM**·2Dioxane, showing XB-directed chain/net formation (left) and void space in overall packing (right).

Crystallization of **I4TEPM** from neat DMSO or DMSO/methanol yielded crystals of **I4TEPM**·2DMSO which has XB interactions analogous to those observed in **I4TEPM**·2THF. Once again, the coordinating solvent acts as a bridging ligand and gives rise to a twisted-ribbon supramolecular chain (Figure 4b left), with one-dimensional channels of 21.0% free volume in the overall packing (Figure 4b right).

By using 1,4-dioxane/dichloromethane as the solvent system, crystals of **I4TEPM**·2Dioxane could be obtained. As expected, dioxane serves as a linear ditopic ligand, so the structure propagates into two dimensions (Figure 4c left). As in **I4TEPM**·2DMSO, the structure creates one-dimensional channels parallel to the crystallographic *c* axis, holding 21.0% free volume (Figure 4c right).

Unfortunately, as is the case with many other crystalline solvates, all these binary crystals are unstable at room temperature. Once removed from the mother liquor, they gradually become opaque because of the partial loss of halogen-bonded and freely-occupying solvent molecules. The DSC and TGA thermograms (Figure 5), however, show that the solvents are somewhat strongly attached to the crystal lattice. In particular, for **I4TEPM**·2THF and **I4TEPM**·2Dioxane, the removal temperatures are noticeably higher than their respective boiling points.

**Figure 5.** (**left**) DSC traces (Tzero aluminum pan, 1–2 mg sample size, 5 ◦C·min−<sup>1</sup> heating rate, nitrogen atmosphere) and (**right**) TGA traces (platinum pan, 5–10 mg sample size, 10 ◦C·min−<sup>1</sup> heating rate, nitrogen atmosphere).

Table 1 presents XB distances and angles of **I4TEPM** and its binary crystals/solvates, along with the normalized distance (*ND*) and the percent radii reduction (%*RR*) values, which are two common indicators used as rough measures of the XB strength. In **I4TEPM**, C≡C–I···(C≡C) interactions are not symmetric and the C–I donors reach more toward terminal acetylenic carbons. Consequently, one I···C separation is significantly longer (with a low %*RR* value) and deviates from linearity. The %*RR* values calculated for XBs observed in the three solvates are greater than 15% (except in one case), reflecting the moderate strength of those interactions. Moreover, all bonds have near-linear (> 170◦ angles, again one exception) arrangements, reflecting their high directionality.


**Table 1.** XB interaction parameters in the studied crystal structures.

*a* Normalized distance, *ND* = *dxy*/(*rx* + *ry*), where *dxy* is the crystallographically determined XB distance, and *rx* and *ry* are the van der Waals radii for the two involved atoms (I = 1.98 Å, C = 1.70 Å, O = 1.52 Å). *b* Percent radii reduction, *%RR* = (1 − *ND*) × 100. Symmetry transformations used to generate equivalent atoms: *c* 1−x, 12<sup>+</sup>y, 1.5−z. *d* 1 2+x, 1.5−y, 1−z. *e* −12<sup>+</sup>x, −12−y, −12<sup>+</sup>z. *f* x, 1+y, z. *g* −12<sup>+</sup>x, −2.5+y, −12<sup>+</sup>z.
