Stabilization of Supramolecular Networks of Polyiodides with Protonated Small Tetra-azacyclophanes

: Polyiodide chemistry is among the ﬁrst historically reported examples of supramolecular forces at work. To date, owing to the increasingly recognized role of halogen bonding and the incorporation of iodine-based components in several devices, it remains an active ﬁeld of theoretical and applied research. Herein we re-examine azacyclophanes as a class of ligands for the stabilization of iodine-dense three-dimensional networks, showing how we devised novel possible strategies starting from literature material. The new set of azacyclophane ligands affords novel crystal structures possessing intriguing properties, which develop on a double layer. At a macroscopic level, the obtained networks possess a very high iodine packing density (less than 2 times more diluted than crystalline I 2 ): a simple parameter, I N , is also introduced to quickly measure and compare iodine packing density in different crystals. On the microscopic level, the present study provides evidence about the ability of one of the ligands to act as a three-dimensional supramolecular mold for the template synthesis of the rarely observed heptaiodide (I 7 − ) anion. Therefore, we believe our approach and strategy might be relevant for crystal engineering purposes.


Introduction
The earliest observations of some of the most recently recognized intermolecular forces, like halogen-halogen interactions, trace back to Gay Lussac's pioneering assertions about the ability of iodide to weakly retain molecular iodine [1,2]. To date, supramolecular chemistry of polyiodide systems remains a challenging and interesting topic, as these species generally manifest both a hypervalent behavior and a marked ability to catenate between themselves through intermolecular donor-acceptor interactions, which are heavily dependent onto their local chemical environment and on the nature of counterions [3]. Such features are gradually coming into the spotlight the more we become able to recognize and understand halogen bonding in these systems and direct polyiodide assembly [3][4][5][6][7][8][9][10][11]. Moreover, polyiodide systems have recently encountered a broad technological interest, mainly connected with solar energy harvesting and storage, e.g., dye-sensitized solar cells and batteries, where they have been successfully implemented as electrolyte in various forms [12,13]. Solid-state conductors based on the Grotthus conduction mechanism [3,14] are also a major field of research [15][16][17]; on this side some of us recently reported about the possibility of using supramolecular forces, such as the strong anion-π interactions exerted by s-tetrazines [18], to direct the assembly of polyiodide networks [19]. Azacyclophanes have long been in the spotlight as pH-dependent anion receptors [20]. Moreover, some interesting results concerning the ability of these ligands to stabilize polyiodides, including the elusive I 4 2− anion, when in their protonated form [21], prompted us to investigate further concerning their binding possibilities. According to Ilioudis and Steed, stabilization of such species were made possible by the good structural complementarity of the rather large 3,10,17-Triazabicyclo[17.3.1]tricosa-1(23),19,21-triene with the iodide anion, which proved effective in stabilizing highly charged species such as I 4 2− . However, hydrogen bonding was not found to be a prominent force and the organization of less charge dense (I 8 2− ) polyiodide network was also observed around the ligand. In this work, instead, we discuss the use of small protonated azacyclophanes for the stabilization of polyiodide networks. Size mismatch between the two interacting partners and nesting of the charged ammonium sites deep within the azacyclophane cage appealed to us as a way to foster the formation of large hydrophobic iodine-rich cages surrounding relatively small ligands, actively promoting the formation of iodine-dense packings.
The experimental design followed in the present work was mainly guided by two literature sources, the results of which deserve a quick analysis here for the purpose of contrasting and comparing them with our findings.
The main literature source, as anticipated above, is ref [21]. The ligand proposed by Ilioudis and Steed (Figure 1a) presents an almost perfect size/shape complementarity to the iodide anion ( [21] and Figure 1b,c), resulting in its effective coordination by the ligand. It is worth noticing that, arguably due to said complementarity, charged iodine species (namely the tip of I 4 2− ion or iodide ion, Figure 1b,c) are always found interacting with the macrocycle. Neutral iodine molecules and/or less charge-dense higher polyiodide species (I 8 2− ) preferentially interact with the hydrophobic hydrocarbon backbone of the ligand instead ( Figure 1d). Other helpful insights were drawn from the structure of the hydrobromic salt of 3,6,9-triaza-1(2,6)-pyridinecyclodecaphane (L1) which became the starting point of current work [22]. As can be seen from Figure 2, the marked basicity of the ligand [22][23][24] (the extra aza-type nitrogen helping to compensate for the loss of basicity due to the proximity of charged groups compared to the azacyclophane used in [21]) pairs with its small size, so that even in the case of the significantly smaller (compared to I − ) bromide, only one anion can be accommodated above the macrocycle; yet it may not engage all N atoms into salt-bridge/hydrogen bonding (Figure 2a,b). Specifically, triprotonated L1 is observed to only manage to use proximal aliphatic ammonium groups as convergent binding sites ( Figure 2a) (proximal N···N distance 3.079(3) Å), but not the two distal ones (i.e., those separated by the pyridine ring) (distal N···N distance 5.471(3) Å). To accommodate all the necessary counterions, the ligand is forced to use its ammonium groups as divergent binding sites, so that further bromide anions can be trapped within three-dimensional cavities formed in between the ligand molecules ( Figure 2c). This was considered as an indication of the ability of this ligand to force the localization of charged species outside its cage and stabilize them in three-dimensional supramolecular networks.
The main literature source, as anticipated above, is ref [21]. The ligand proposed by Ilioudis and Steed (Figure 1a) presents an almost perfect size/shape complementarity to the iodide anion ( [21] and Figure 1b,c), resulting in its effective coordination by the ligand. It is worth noticing that, arguably due to said complementarity, charged iodine species (namely the tip of I4 2− ion or iodide ion, Figure  1b,c) are always found interacting with the macrocycle. Neutral iodine molecules and/or less chargedense higher polyiodide species (I8 2− ) preferentially interact with the hydrophobic hydrocarbon backbone of the ligand instead ( Figure 1d).
Other helpful insights were drawn from the structure of the hydrobromic salt of 3,6,9-triaza-1(2,6)-pyridinecyclodecaphane (L1) which became the starting point of current work [22]. As can be seen from Figure 2, the marked basicity of the ligand [22][23][24] (the extra aza-type nitrogen helping to compensate for the loss of basicity due to the proximity of charged groups compared to the azacyclophane used in [21]) pairs with its small size, so that even in the case of the significantly smaller (compared to I − ) bromide, only one anion can be accommodated above the macrocycle; yet it may not engage all N atoms into salt-bridge/hydrogen bonding (Figure 2a,b). Specifically, triprotonated L1 is observed to only manage to use proximal aliphatic ammonium groups as convergent binding sites (Figure 2a) (proximal N···N distance 3.079(3) Å), but not the two distal ones (i.e., those separated by the pyridine ring) (distal N···N distance 5.471(3) Å). To accommodate all the necessary counterions, the ligand is forced to use its ammonium groups as divergent binding sites, so that further bromide anions can be trapped within three-dimensional cavities formed in between the ligand molecules ( Figure 2c). This was considered as an indication of the ability of this ligand to force the localization of charged species outside its cage and stabilize them in three-dimensional supramolecular networks. In the case of iodine-based species the hydrogen bonding contribution was expected, and found [21] to be more modest than for bromide, while hydrophobic contacts are expected to play a major role in stabilizing scarcely charged polyiodides. Accordingly, we adopted methylation as a tool to  In the case of iodine-based species the hydrogen bonding contribution was expected, and found [21] to be more modest than for bromide, while hydrophobic contacts are expected to play a major role in stabilizing scarcely charged polyiodides. Accordingly, we adopted methylation as a tool to gradually modify the ligands, eventually forcing the hydrogen bond network nesting inside the cyclophane structure, leaving a hydrophobic external surface suitable for iodine interaction. The ligands employed in this work are reported in Figure 3. gradually modify the ligands, eventually forcing the hydrogen bond network nesting inside the cyclophane structure, leaving a hydrophobic external surface suitable for iodine interaction. The ligands employed in this work are reported in Figure 3. Regrettably, despite considerable efforts, only protonated forms of L1 and L2-Me3 furnished single crystals of polyiodide complexes suitable for X-ray diffraction (XRD) studies. Nevertheless, the obtained crystalline samples present some fascinating structural features.

Crystal Structure of [(H 2 L1) 2 I 2 (I 5 )(I 3 ) 3 ]
This structure consists of diprotonated L1 ligands, polyiodide anions and iodine molecules. The asymmetric unit contains two non-equivalent ligands, one I 5 − , three I 3 − anions and two halves of non-equivalent I 2 molecules ( Figure 4). Each ligand assumes a bent conformation featuring 148 • and 145 • dihedral angles between the aromatic ring plane and the mean plane defined by the aliphatic nitrogen atoms, respectively. All the aliphatic nitrogen atoms interact via H-bonds with the polyiodide anion (Table S1). Most of these contacts are rather long (minimum 3.566 Å, maximum 3.922 Å, and average 3.7 ± 0.1 Å) and it is worth noticing how they rarely (with the exception of I9) involve the central, formally charge-bearing atoms of polyiodides (even considering a I 3 − ·I 2 description for the pentaiodide anion, see below).
This indicates either that classic charge-charge interactions have little influence in orchestrating the overall packing or that valence in the polyiodide network goes beyond what can be predicted on a qualitative level, or both. Relevant information on this point is provided by the iodine molecules ( . This creates a succession of alternating long-short inter-and intramolecular distances, displayed in Figure 5, which is typical of iodine-iodine secondary bonds. This term, introduced in its general usage by Alcock [25], is employed in polyiodide chemistry to describe I···I intermolecular interactions in the 3.4-3.7 Å range [3], i.e., supramolecular interactions which nevertheless involve a certain degree of orbital interaction. Accordingly, the formation of secondary bonds is always accompanied by a weakening of intramolecular bonds of the involved polyiodide species. The alternating long-short inter and intra molecular distances of Figure 5 clearly indicate orbital interactions between polyiodides (I 3 − , I 5 − ) and I 2 molecules ( Figure 5), so that, as strong secondary bonds are formed, covalent interactions are loosened. Each ligand assumes a bent conformation featuring 148° and 145° dihedral angles between the aromatic ring plane and the mean plane defined by the aliphatic nitrogen atoms, respectively. All the aliphatic nitrogen atoms interact via H-bonds with the polyiodide anion (Table S1). Most of these contacts are rather long (minimum 3.566 Å, maximum 3.922 Å, and average 3.7 ± 0.1 Å) and it is worth noticing how they rarely (with the exception of I9) involve the central, formally charge-bearing atoms of polyiodides (even considering a I3 − ·I2 description for the pentaiodide anion, see below). This indicates either that classic charge-charge interactions have little influence in orchestrating the overall packing or that valence in the polyiodide network goes beyond what can be predicted on a qualitative level, or both. Relevant information on this point is provided by the iodine molecules ( Figure 5). Beyond being stabilized by CH···I interactions from neighboring ligand molecules, both I2 molecules are sandwiched between I3 − (I16-I16') or I5 − (I13-I13') anions ( Figure 5). Close inspection reveals I-I distances in the iodine molecules of 2.812(1) Å (I13-I13') and 2.816(2) Å (I16-I16'), i.e., longer than those of molecular iodine (2.715 Å) [3], while also the interacting polyiodide anions are found asymmetric (I3 − : I6-I2 2.8188(8) Å, I2-I5 3.0707(8) Å; I5 − : I4-I1 2.9234(7) Å, I1-I3 2.8905(7) Å, I10-I12 2.7723(8) Å, I3-I12 3.2541(8) Å). This creates a succession of alternating long-short inter-and intramolecular distances, displayed in Figure 5, which is typical of iodine-iodine secondary bonds. This term, introduced in its general usage by Alcock [25], is employed in polyiodide chemistry to describe I···I intermolecular interactions in the 3.4-3.7 Å range [3], i.e., supramolecular interactions which nevertheless involve a certain degree of orbital interaction. Accordingly, the formation of secondary bonds is always accompanied by a weakening of intramolecular bonds of the involved polyiodide species. The alternating long-short inter and intra molecular distances of Figure 5 clearly indicate orbital interactions between polyiodides (I3 − , I5 − ) and I2 molecules ( Figure 5), so that, as strong secondary bonds are formed, covalent interactions are loosened. The high asymmetry observed for the pentaiodide anion suggests this species is better rationalized, as often observed in the literature, [3] as a I3 − anion (I3-I1-I4) strongly interacting with an iodine molecule (I12-I10). As a matter of fact, this disguised iodine molecule is also sandwiched between the strongly interacting (I3-I1-I4) triiodide and the I11-I7-I14 triiodide, which in turn is also interacting with the last triiodide (I15-I9-I8) found in the crystal structure. As a consequence of the The high asymmetry observed for the pentaiodide anion suggests this species is better rationalized, as often observed in the literature, [3] as a I 3 − anion (I3-I1-I4) strongly interacting with an iodine molecule (I12-I10). As a matter of fact, this disguised iodine molecule is also sandwiched between the strongly interacting (I3-I1-I4) triiodide and the I11-I7-I14 triiodide, which in turn is also interacting with the last triiodide (I15-I9-I8) found in the crystal structure. As a consequence of the network formed by secondary bonds, all I 3 − anions in the crystal structure are found asymmetric (I11-I7 3.1083(8) Å, The result of these significant iodine-iodine secondary bonds is a polyiodide network constituted of zig-zag chains involving all the iodine-based species in the crystals. These chains leave void open channels where the protonated ligands are hosted, to the point that the whole structure appears as a sort of self-assembled supramolecular clathrate ( Figure 6).

Figure 5.
Detail of the long-short alternating I-I interatomic distances involving the two I2 molecules found within the crystal structures. Inter-and intramolecular interatomic distances which are found shorter than reference values are displayed with solid black lines, those which are found longer are displayed with dotted red lines. Arbitrary color is used to mark symmetry equivalent polyiodides.
The high asymmetry observed for the pentaiodide anion suggests this species is better rationalized, as often observed in the literature, [3] as a I3 − anion (I3-I1-I4) strongly interacting with an iodine molecule (I12-I10). As a matter of fact, this disguised iodine molecule is also sandwiched between the strongly interacting (I3-I1-I4) triiodide and the I11-I7-I14 triiodide, which in turn is also interacting with the last triiodide (I15-I9-I8) found in the crystal structure. As a consequence of the network formed by secondary bonds, all I3 − anions in the crystal structure are found asymmetric (I11- The result of these significant iodine-iodine secondary bonds is a polyiodide network constituted of zig-zag chains involving all the iodine-based species in the crystals. These chains leave void open channels where the protonated ligands are hosted, to the point that the whole structure appears as a sort of self-assembled supramolecular clathrate ( Figure 6).

Crystal Structure of [H 2 L2-Me 3 (I 7 ) 2 ]
The structure consists of diprotonated L2-Me 3 ligands and I 7 − anions, each lying on the crystallographic planes ( Figure 7). The ligand assumes a bent conformation, typical of small macrocycles incorporating rigid aromatic units [22][23][24]26,27], which can be described by the 84 • dihedral angle between the aromatic ring plane and the mean plane defined by the aliphatic nitrogen atoms. The acidic protons are located on the benzylic nitrogen atoms [22][23][24], yet their interaction with the polyiodide anions are very limited, consisting of a sole contact, and moreover quite long to be considered a typical salt bridge (N2(H2A)···I6 3.753 Å). The same can be said for the OH group, giving an additional long H···I contact (O1(H1)···I5 3.779 Å). Each I 7 − anion assumes a pyramidal conformation with bond angles at the apical iodine atom of 84 • , 84 • and 93 • (A Figure 7) and 83 • , 83 • and 102 • (B Figure 7). The I-I terminal bond distances are in the range 2.75(1)-2.815(3) Å while the I-I distances from the apical iodine vary from 3.061(3) to 3.233(2) Å. The overall symmetry (angles and distances) of the heptaiodides is higher than those observed for reported I 7 (cf. Discussion section), however, this aspect cannot be discussed meaningfully because such symmetry originates from the fact that the anions lay on crystallographic planes. Both the terminal and the apical iodine atoms are involved in short I···I intermolecular contacts which can be considered as secondary bonds (I3···I10 3.46(1) Å) or van der Waals interactions (below 4 Å): as can be seen in Figure 8, the central iodine atom of B, formally bearing the charge, forms the strongest interaction with neighboring I 7 − A anions linearly along I-I covalent bonds, i.e., respecting the typical geometry expected from σ-hole  3 ] crystals, the whole crystal structure can be defined as a iodine-based clathrate structure trapping the diprotonated H 2 L2-Me 3 2+ ligand molecules, but in this case the formed architecture is even more striking: as shown in Figure 9 the contacts between I 7 − anions define a slightly distorted void cubic lattice hosting the countercations. A global view of the packing is reported in Figure 10.
distances) of the heptaiodides is higher than those observed for reported I7 (cf. Discussion section), however, this aspect cannot be discussed meaningfully because such symmetry originates from the fact that the anions lay on crystallographic planes. Both the terminal and the apical iodine atoms are involved in short I···I intermolecular contacts which can be considered as secondary bonds (I3···I10 3.46(1) Å) or van der Waals interactions (below 4 Å): as can be seen in Figure 8, the central iodine atom of B, formally bearing the charge, forms the strongest interaction with neighboring I7 − A anions linearly along I-I covalent bonds, i.e., respecting the typical geometry expected from σ-hole interactions. As in the case of the [(H2L1)2I2(I5)(I3)3] crystals, the whole crystal structure can be defined as a iodine-based clathrate structure trapping the diprotonated H2L2-Me3 2+ ligand molecules, but in this case the formed architecture is even more striking: as shown in Figure 9 the contacts between I7 − anions define a slightly distorted void cubic lattice hosting the countercations. A global view of the packing is reported in Figure 10.

Discussion
Analysis and comparison with literature data of the obtained structures can be undertaken both at a macroscopic and a microscopic level.
On a macroscopic level, density of iodine is an important parameter for possible practical applications, especially as solid-state conductors. This is because, as logical and reported [3], chance of short I···I distances allowing for orbital interactions and electron conduction via a Grotthus-like mechanism [3,14], increases the more the iodine atoms are densely packed.
We propose as an easy macroscopic descriptor, allowing for direct comparison between different crystal structures, a quantity IN, or Iodine Number. This is easily calculated. Counting the I atoms found within a unit cell and dividing for the cell volume gives the crystal average iodine number

Discussion
Analysis and comparison with literature data of the obtained structures can be undertaken both at a macroscopic and a microscopic level.
On a macroscopic level, density of iodine is an important parameter for possible practical applications, especially as solid-state conductors. This is because, as logical and reported [3], chance of short I···I distances allowing for orbital interactions and electron conduction via a Grotthus-like mechanism [3,14], increases the more the iodine atoms are densely packed.
We propose as an easy macroscopic descriptor, allowing for direct comparison between different crystal structures, a quantity I N , or Iodine Number. This is easily calculated. Counting the I atoms found within a unit cell and dividing for the cell volume gives the crystal average iodine number density per Å 3 , a rather inconvenient parameter to utilize as it is (we get to know that very small decimal fractions of I atoms can be included in a volume as small as 1 Å 3 ). The I N is defined as the ratio of the iodine number density per Å 3 of the crystal under examination to the iodine number density per Å 3 of crystalline molecular iodine (reference state in standard conditions).
Iodine number density in I 2 crystals has a value of 0.02346 Å −3 . It is instructive to show the qualitative meaning of such a number. An isolated I atom treated as a hard sphere of radius 2.04 Å [28], i.e., occupying 35.6 Å 3 , would have an I number density of 0.02812 Å −3 . Dense packing of spheres has a packing efficiency of ≈0.74, meaning that the maximum theoretical I number density under hard spheres packing approximation would be 0.02079 Å −3 . Crystalline molecular iodine possesses an even higher I number density than that (0.02346 Å −3 , i.e., it is roughly 12% more dense than the theoretical hard spheres packing): this is due to the molecular nature of I 2 , which allows spheres to interpenetrate significantly. According to its high packing density and significance, we decided to use I number density of crystalline I 2 as a reference limit value for comparing iodine packing densities within different crystals: by definition of I N , crystalline molecular I 2 has an I N = 1.
When we move from the I 4 2− complex of NABWOD (Figure 1b (Figure 1b) [21], to three-dimensional ribbons of polyiodides interacting with the ligands (Figure 1d) [21], to polyiodide structures forming channels hosting small azacyclophanes (Figure 6), up to clathrate-like cages completely engulfing the ligand ( Figure 10). According to these observations it is safe to conclude that the choice of small azacyclophanes actively promotes a dense packing of the polyiodide anions, rather than the coordination of discrete anions observed with a cyclophane matching the size of the iodide anion [21].
The microscopic description level has a twofold importance. On the applicative side, having an iodine packing which is dense on average (e.g., as view from a macroscopic descriptor like I N ), does not necessarily imply that an uninterrupted I···I secondary bond network capable of functioning as an electron conductor exists within the crystal. This is the reason we dedicated much of the crystallographic section to the description of the I···I intermolecular contacts. On a theoretical side, which is more of concern at present, evaluation of forces in play through the observed contacts are paramount for the understanding of these packings and the recognition of novel crystal engineering strategies.
As anticipated from the beginning, indeed the nesting of the hydrogen bond network within the frame of small azacyclophanes resulted in a lesser dependence of the packing on salt bridge interactions. This can be easily observed from the comparison of Figure 11 with Figures 1 and 2. paramount for the understanding of these packings and the recognition of novel crystal engineering strategies.
As anticipated from the beginning, indeed the nesting of the hydrogen bond network within the frame of small azacyclophanes resulted in a lesser dependence of the packing on salt bridge interactions. This can be easily observed from the comparison of Figure 11 with Figures 1 and 2. As can be seen, the hydrophilic face of the azacyclophanes presenting the ammonium groups is found interacting either with a triiodide or with the iodine molecule strongly bound to a triiodide within the pentaiodide anion. Neither convergence of ammonium groups towards a specific anionic site, nor the encapsulation induced by the perfect fit observed in Figure 1 are possible here. Conversely, the hydrophobic face of the ligands, forced to interact with a bromide anion in Figure 2b [22], is here observed to be a prominent docking site for neutral iodine molecules. Contrary to what is observed with simpler protonated amines [11], where an active dialogue is established between halogen and hydrogen bonding, here the interactions between the charged ligands and iodine-based species mainly involves the external hydrophobic surface of the azacyclophane, bringing its shape into the spotlight.
Perhaps the most prominent crystal engineering piece of information that we can derive from the present study is the possibility to use charged azacyclophanes as templating agents for the synthesis of polyiodides. Despite some of this being reported for linear polyiodides (I3 − , I4 2− , I6 2− ) in the case of aliphatic [11] or aromatic [6] linear ammonium derivatives, in the current case we appear to be in the presence of the first three-dimensional mold reported for the synthesis of the I7 − anion, moreover a species which can be considered of quite rare occurrence (only 10 structures reported so far in the CSD feature I7 − [29][30][31][32][33][34][35][36][37][38], only half of them containing selectively I7 − rather than a mixture of As can be seen, the hydrophilic face of the azacyclophanes presenting the ammonium groups is found interacting either with a triiodide or with the iodine molecule strongly bound to a triiodide within the pentaiodide anion. Neither convergence of ammonium groups towards a specific anionic site, nor the encapsulation induced by the perfect fit observed in Figure 1 are possible here. Conversely, the hydrophobic face of the ligands, forced to interact with a bromide anion in Figure 2b [22], is here observed to be a prominent docking site for neutral iodine molecules. Contrary to what is observed with simpler protonated amines [11], where an active dialogue is established between halogen and hydrogen bonding, here the interactions between the charged ligands and iodine-based species mainly involves the external hydrophobic surface of the azacyclophane, bringing its shape into the spotlight. Perhaps the most prominent crystal engineering piece of information that we can derive from the present study is the possibility to use charged azacyclophanes as templating agents for the synthesis of polyiodides. Despite some of this being reported for linear polyiodides (I 3 − , I 4 2− , I 6 2− ) in the case of aliphatic [11] or aromatic [6] linear ammonium derivatives, in the current case we appear to be in the presence of the first three-dimensional mold reported for the synthesis of the I 7 − anion, moreover a species which can be considered of quite rare occurrence (only 10 structures reported so far in the CSD feature I 7 − [29][30][31][32][33][34][35][36][37][38], only half of them containing selectively I 7 − rather than a mixture of different polyiodide species [31][32][33][34][35]). As we tried to portray in Figure 12, the diprotonated H 2 L-Me 3 ligand possesses a stunning complementarity with the two interacting heptaiodide anions. No matter the point of view, we are always able to distinguish complementary and stabilizing structural elements, such as the curvature of the ligand with respect to the polyiodide angles (Figure 12a,d), the perfect fitting of the phenolic O in the assembly of the anions (red, Figure 12a-d), the templating role of the methyl substituents (Figure 12b-d) and that of the charged crevice defined by the ammonium groups, perfectly matched by one of the I 7 − axes (blue, Figure 12c). According to the extraordinary complementarity observed, we are led to believe that the selective observation of I 7 − anions in the crystal structure, and of them only, is due to the capacity of the diprotonated form of this ligand to act as a supramolecular templating agent, molding the heptaiodide anion around itself.
the perfect fitting of the phenolic O in the assembly of the anions (red, Figure 12 a,b,c,d), the templating role of the methyl substituents (Figure 12 b,c,d) and that of the charged crevice defined by the ammonium groups, perfectly matched by one of the I7 − axes (blue, Figure 12c). According to the extraordinary complementarity observed, we are led to believe that the selective observation of I7 − anions in the crystal structure, and of them only, is due to the capacity of the diprotonated form of this ligand to act as a supramolecular templating agent, molding the heptaiodide anion around itself.

Software
CCDC Mercury [56] and UCSF Chimera [57] programs were used for analysis and graphical presentation of data.

Conclusions
Halogen bonding [58][59][60][61][62][63][64][65], incorporation of iodine molecules within three-dimensional networks [66][67][68] and template synthesis of specific polyiodides [4][5][6]9,10] are current topics in polyiodide chemistry. We contribute to the discussion with the proposition of small protonated azacyclophanes as a meaningful class of stabilizing ligands capable of producing high-density iodine packings, of which we discuss two examples. I N is proposed as a simple quantitative parameter to compare the macroscopic density of iodine packing among different crystal structures. Moreover, diprotonated form of the L2-Me 3 ligand is suggested as the first templating agent capable of promoting selectively the formation of the I 7 − anion. We can safely anticipate the crystal engineering directed assembly of iodine-based clathrate-like compounds to produce technologically relevant materials in the near future.