The Amine Group as Halogen Bond Acceptor in Cocrystals of Aromatic Diamines and Perﬂuorinated Iodobenzenes

: In order to study the proclivity of primary amine groups to act as halogen bond acceptors, three aromatic diamines ( p -phenylenediamine ( pphda ), benzidine ( bnzd ) and o -tolidine ( otol )) were cocrystallised with three perﬂuorinated iodobenzenes (1,4-tetraﬂuorodiiodobenzene ( 14tﬁb ), 1,3-tetraﬂuorodiiodobenzene ( 13tﬁb ) and 1,3,5-triﬂuorotriiodobenzene ( 135tﬁb )) as halogen bond donors. Five cocrystals were obtained: ( pphda )( 14tﬁb ), ( bnzd )( 13tﬁb ) 2 , ( bnzd )( 135tﬁb ) 4 , ( otol )( 14tﬁb ) and ( otol )( 135tﬁb ) 2 . In spite of the variability of both stoichiometries and structures of the cocrystals, in all the prepared cocrystals the amine groups form exclusively I ··· N halogen bonds, while the amine hydrogen atoms participate mostly in N–H · · · F contacts. The preference of the amine nitrogen atom toward the halogen bond, as opposed to the hydrogen bond (with amine as a donor), is rationalised by means of computed hydrogen and halogen bond energies, indicating that the halogen bond energy between a simple primary amine (methylamine) and a perﬂuorinated iodobenzene (pentaﬂuoroiodobenze ne) is ca. 15 kJ mol − 1 higher than the energy of the (H)NH ··· NH 2 hydrogen bond between two amine molecules.

The most commonly employed neutral halogen bond donors in crystal engineering to date have been perfluorinated iodobenzenes, namely, 1,2-tetrafluorodiiodobenzene (12tfib), 1,4-tetrafluorodiiodobenzene (14tfib), 1,3-tetrafluorodiiodobenzene (13tfib) and 1,3,5-trifluorotriiodobenzene (135tfib) [37][38][39][40][41][42]. These substances are stable in ambient conditions, soluble in most organic solvents, easy to handle and commercially available. The presence of the electron-withdrawing fluorine atoms in the molecule increases the positive electrostatic potential of the σ-holes of the iodine atoms, making them reliable halogen bond donors for a wide range of organic and metal-organic Lewis bases [43,44]. However, their reliability as halogen bond donors can be severely reduced by the formation of competing hydrogen bonds. The balance between halogen and hydrogen bonds in systems where the two can compete is often quite delicate, and can be influenced by various factors, such as weak interactions and overall crystal packing [45,46], the choice of crystallisation solvent [47] and even the amount of solvent used in the cocrystal synthesis [48]. For these reasons, the result of a supramolecular synthesis in systems where there is competition between halogen and hydrogen donors for the (same) acceptor sites generally cannot be unequivocally predicted.
A typical acceptor group where the reliability of halogen bond fails is the primary amine group (-NH 2 ). Normally, the nitrogen atom is one of the most reliable bases for the formation of halogen bonds in general, and with perfluorinated iodobenzenes in particular. According to the data deposited with the Cambridge Structural Database (CSD) [49], there have been 686 crystal structures containing a perfluorinated iodobenzene and a nitrogen atom reported to date; of these, 430 (63%) feature the C−I···N halogen bond. However, out of the (only) 50 structures containing perfluorinated iodobenzene and a primary amine group, only in 5 of them (a mere 10%) is a C−I···N(amino) halogen bond present. It is tempting to attribute this to the ability of the primary amine nitrogen to also act as both a hydrogen bond donor and an acceptor. However, the amine group is a rather poor hydrogen bond donor-the R-N(H)-H···NH 2 hydrogen bond can be found to be present in only about 5% of all the structures containing (either aromatic or aliphatic) a primary amine group (1965 out of 42,021), while generally in the presence of NH or OH hydrogen bond donors it acts as a hydrogen bond acceptor in about 13% of cases (4434 out of 33,964). Therefore, the more probable reason for the low occurrence of halogen bonds involving the primary amine group is the presence of better halogen bond acceptor sites in the molecule (such as heterocyclic aromatic nitrogen atoms). This conclusion is also borne out by an earlier study of aromatic amines as halogen bond acceptors performed within our group [50], which has demonstrated that the primary amine group acted as a halogen bond acceptor only if no stronger halogen bond acceptors (such as heterocyclic nitrogen, carboxylic and even nitro groups) were present.
In this work, we endeavoured to study the halogen bonding proclivity of the aromatic primary amino group in the absence of other electronegative atoms or functional groups which might compete with the primary amine nitrogen as halogen bond acceptors. For this purpose, we selected three aromatic primary diamines (p-phenylenediamine (benzene-1,4-diamine, pphda), benzidine (1,1'-biphenyl-4,4'-diamine, bnzd) and o-tolidine (3,3'-dimethyl-[1,1'-biphenyl]-4,4'-diamine, otol); Scheme 1) as (potential) halogen bond acceptors. The crystal structures of all three have been previously reported [51][52][53], demonstrating that almost every nitrogen atom in each of the structures of the pure diamines does act as a hydrogen bond acceptor (the exception being two out of the four polymorphs in bnzd-each with multiple molecules of bnzd in the asymmetric unit-where some of the NH 2 nitrogen atoms do not). This makes them an excellent platform for a study of the competition of-R-N(H)-H···NH 2 hydrogen bonds and C−I···NH 2 halogen bonds. The competition was studied by attempting to cocrystallise each of the three selected diamines with three selected perfluorinated iodobenzenes (14tfib, 13tfib and 135tfib) as halogen bond donors (Scheme 1). The cocrystallisation screening was performed mechanochemically by liquid-assisted grinding (LAG) in order to determine in how many instances the cocrystals would be formed. The cocrystallisation experiments were also performed from solution in order to produce single crystals of halogen-bonded cocrystals, to be studied by single-crystal X-ray diffraction in order to determine which intermolecular interactions govern their assembly.

Scheme 1.
Halogen bond donors and acceptors used in this study.

Solution and Single-Crystal Synthesis of Cocrystals
The corresponding halogen bond donor and diamine were dissolved in 5.0 mL of a solvent (ethanol or ethanol/acetone mixture), whereupon the solutions were left to cool and slowly evaporate (experimental details are given in the Supplementary Materials, Table S1). The obtained products were characterized by powder X-ray diffraction (PXRD) in order to examine their phase purity. Single crystals suitable for single-crystal X-ray diffraction (SCXRD) experiments were prepared by crystallization from ethanol and appeared after three to seven days. ORTEP plots of the obtained compounds are shown in Figures S1-S5 in the Supplementary Materials.

Powder X-ray Diffraction Measurements
Powder X-ray diffraction experiments on the samples were performed on an Aeris X-ray diffractometer (Malvern Panalytical, Malvern Worcestershire, UK) with CuKα1 (λ = 1.54056 Å) radiation. The scattered intensities were measured with a PIXcel-1D-Medi-pix3 detector. The angular range was from 5° to 40° (2θ) with a continuous step size of 0.02° and measuring a time of 0.5 s per step.
Data collection methods were created using the program package START XRDMP CREATOR (Malvern Panalytical, Malvern Worcestershire, UK) while the data were analysed using X'Pert HighScore Plus (Version 2.2, Malvern Panalytical, Malvern Worcestershire, UK) [54]. The comparison of measured and calculated PXRD patterns of the prepared compounds are shown in Figures S6-S10 in the Supplementary Materials.

Single-Crystal X-ray Diffraction Measurements
Single-crystal X-ray diffraction experiments were performed using an Oxford Diffraction Xcalibur Kappa CCD X-ray diffractometer (Oxford Diffraction Ltd., Abingdon, UK) with graphite-monochromated MoKα (λ = 0.71073 Å) radiation. The data sets were collected using the ω-scan mode over the 2θ-Range up to 54°. The programs CrysAlis PRO CCD and CrysAlis PRO RED were employed for data collection, cell refinement and data reduction [55,56]. The structures were solved and refined using SHELXS (Version 2013, Göttingen, Germany), SHELXL programs (Version 2013, Göttingen, Germany) and

Solution and Single-Crystal Synthesis of Cocrystals
The corresponding halogen bond donor and diamine were dissolved in 5.0 mL of a solvent (ethanol or ethanol/acetone mixture), whereupon the solutions were left to cool and slowly evaporate (experimental details are given in the Supplementary Materials, Table S1). The obtained products were characterized by powder X-ray diffraction (PXRD) in order to examine their phase purity. Single crystals suitable for single-crystal X-ray diffraction (SCXRD) experiments were prepared by crystallization from ethanol and appeared after three to seven days. ORTEP plots of the obtained compounds are shown in Figures S1-S5 in the Supplementary Materials.

Powder X-ray Diffraction Measurements
Powder X-ray diffraction experiments on the samples were performed on an Aeris X-ray diffractometer (Malvern Panalytical, Malvern Worcestershire, UK) with CuKα1 (λ = 1.54056 Å) radiation. The scattered intensities were measured with a PIXcel-1D-Medipix3 detector. The angular range was from 5 • to 40 • (2θ) with a continuous step size of 0.02 • and measuring a time of 0.5 s per step.
Data collection methods were created using the program package START XRDMP CRE-ATOR (Malvern Panalytical, Malvern Worcestershire, UK) while the data were analysed using X'Pert HighScore Plus (Version 2.2, Malvern Panalytical, Malvern Worcestershire, UK) [54]. The comparison of measured and calculated PXRD patterns of the prepared compounds are shown in Figures S6-S10 in the Supplementary Materials.

Single-Crystal X-ray Diffraction Measurements
Single-crystal X-ray diffraction experiments were performed using an Oxford Diffraction Xcalibur Kappa CCD X-ray diffractometer (Oxford Diffraction Ltd., Abingdon, UK) with graphite-monochromated MoKα (λ = 0.71073 Å) radiation. The data sets were collected using the ω-scan mode over the 2θ-Range up to 54 • . The programs CrysAlis PRO CCD and CrysAlis PRO RED were employed for data collection, cell refinement and data reduction [55,56]. The structures were solved and refined using SHELXS (Version 2013, Göttingen, Germany), SHELXL programs (Version 2013, Göttingen, Germany) and SHELXT programs (Version 2013, Göttingen, Germany) respectively [57][58][59]. The struc-tural refinement was performed on F 2 using all data. The hydrogen atoms were placed in calculated positions and treated as riding on their parent atoms [C-H = 0.93 Å and U iso (H) = 1.2 U eq (C) for aromatic hydrogen atoms; C-H = 0.97 Å and U iso (H) = 1.2 U eq (C) for methyl hydrogen atoms]. The amino group hydrogen atoms were located from the electron difference map and then refined with the following restraints, where necessary: d(N-H) = 0.920 Å, d(H···H) = 1.500 Å, U iso (H) = 1.5 U eq (N). All calculations were performed using the WinGX crystallographic suite of programs [60]. The figures were prepared using Mercury 2020.2.0 (CCDC, Cambridge, UK) [61]. Crystallographic data of the prepared compounds are shown in Table S2 in the Supplementary Materials.

Thermal Analysis
Differential scanning calorimetry (DSC) and thermogravimetric (TG) measurements were performed simultaneously on a Mettler-Toledo TGA/DSC 3+ module (Mettler Toledo, Greifensee, Switzerland). Samples were placed in alumina crucibles (40 µL) and heated in the temperature range 25 to 300 • C, at a heating rate of 10 • C min −1 under a nitrogen flow of 150 mL min −1 .
Data collection and analysis were performed using the program package STARe Software (Version 15.00, Mettler Toledo, Greifensee, Switzerland) [62]. TG and DSC curves of the prepared compounds are shown in Figures S11-S15 in the Supplementary Materials.

Calculations
All calculations were performed using the Gaussian 09 (Revision D.01) software package [63]. Geometry optimizations were performed using the B3LYP/def2-TZVP [64][65][66] level of theory with the D3 version of Grimme's dispersion [67] and an ultrafine integration grid (99 radial shells and 590 points per shell). This method was shown to reproduce experimental halogen bond lengths, complexation energies and vibrational frequencies in the gas phase with good accuracy [68]. The harmonic frequency calculations were performed on the optimized geometries to ensure that they correspond to minima on the potential energy surface.

Results and Discussion
The crystallisation from solution of the selected diamines and the halogen bond donors yielded five cocrystals, whereas in four cases only the starting materials were isolated. An overview of the obtained cocrystals is given in Table 1. Table 1. An overview of the obtained cocrystals of aromatic diamines and perfluorinated iodobenzenes.

Figure 1.
In the structure of (pphda)(14tfib), layers are formed by the linking of halogen bonded (yellow) chains through N−H⋯F hydrogen bonds (green) in the structure of (pphda)(14tfib).
As can be seen from the above structure descriptions, in all of the five obtained cocrystals, the N-H⋯N hydrogen bond (which is the dominant interaction in the crystal structures of the amines) is entirely replaced by I···N halogen bonds in the cocrystals, with amine hydrogen atoms participating mostly only in rather ephemeral N-H⋯F contacts. In order to investigate whether this coincides with a corresponding difference in the R-N(H)-H···NH2 hydrogen bond and the C−I···NH2 halogen bond energies, we attempted to calculate the binding energies of chosen amine molecules in hydrogen-bonded dimers and halogen-bonded complexes with iodopentafluorobenzene (ipfb), and thus determined the relative strength of these interactions in vacuo. To avoid the presence of other acceptor sites in the molecules, we chose simple primary amines: aniline (PhNH2) and methylamine (MeNH2). From the results (Table 2), it can be noticed that the obtained binding energy for the aniline acceptor in hydrogen-bonded dimers is slightly higher (0.8 kJ mol −1 ) than for the complex with ipfb, indicating marginally larger stabilization of the hydrogenbonded complexes. However, the optimised structure of the hydrogen-bonded (PhNH2)2 complex reveals the presence of a C-H⋯π contact (approximately the orthogonal position of the two aromatic rings), which almost certainly further stabilizes the aniline dimer (Figure 6a) [70]. In the optimized structure of (PhNH2)(ipfb), such additional interactions are Figure 4. (a) Layers formed by linking bnzd and 13tfib molecules via I···NH 2 and I· · · C π halogen bonds (yellow) in the structure of (otol)(14tfib); (b) layers connected through N-H· · · F hydrogen bonds (green) into a 3D structure.
As can be seen from the above structure descriptions, in all of the five obtained cocrystals, the N-H· · · N hydrogen bond (which is the dominant interaction in the crystal structures of the amines) is entirely replaced by I···N halogen bonds in the cocrystals, with amine hydrogen atoms participating mostly only in rather ephemeral N-H· · · F contacts. In order to investigate whether this coincides with a corresponding difference in the R-N(H)-H···NH 2 hydrogen bond and the C−I···NH 2 halogen bond energies, we attempted to calculate the binding energies of chosen amine molecules in hydrogen-bonded dimers and halogen-bonded complexes with iodopentafluorobenzene (ipfb), and thus determined the relative strength of these interactions in vacuo. To avoid the presence of other acceptor sites in the molecules, we chose simple primary amines: aniline (PhNH 2 ) and methylamine (MeNH 2 ). From the results (Table 2), it can be noticed that the obtained binding energy for the aniline acceptor in hydrogen-bonded dimers is slightly higher (0.8 kJ mol −1 ) than for the complex with ipfb, indicating marginally larger stabilization of the hydrogenbonded complexes. However, the optimised structure of the hydrogen-bonded (PhNH 2 ) 2 complex reveals the presence of a C-H· · · π contact (approximately the orthogonal position of the two aromatic rings), which almost certainly further stabilizes the aniline dimer (Figure 6a) [70]. In the optimized structure of (PhNH 2 )(ipfb), such additional interactions are absent (Figure 6b), the only interaction between the molecules being a single I···N halogen bond almost perpendicular (∠ (I···N-C) = 106.6 • ) to the plane of the aniline ring (which closely corresponds to the conformation of halogen-bonded complexes also observed in the cocrystal structures). Therefore, the calculated difference in binding energies for (PhNH 2 ) 2 and (PhNH 2 )(ipfb) does not pose a reliable representation of the difference in hydrogen-and halogen-bond energies with the primary amine as an acceptor (and donor of the former). In order to avoid the additional stabilisation of the hydrogenbonded complex by a C-H· · · π interaction, we opted for an additional set of computations using methylamine as the model primary amine molecule. In the (MeNH 2 ) 2 dimer, the (H)NH···NH 2 hydrogen bond is indeed the only contact within the dimer (Figure 6c). Significantly, here the binding energy in the halogen-bonded (ipfb)(MeNH 2 ) complex ( Figure 6d) is ca. 15 kJ mol −1 higher than the one for the hydrogen-bonded (MeNH 2 ) 2 dimer. There is a difference in the basicities of the aliphatic and the aromatic primary amine groups-the aromatic amines being weaker bases-which makes the halogen bond in the (ipfb)(MeNH 2 ) complex stronger by ca. 8.4 kJ mol −1 than the corresponding bond in the (PhNH 2 )(ipfb) complex. As this reduction in the basicity of the acceptor should also proportionally influence the hydrogen bond energy, it can therefore be expected that the contribution of the R-N(H)-H···NH 2 hydrogen bond to the overall binding energy of the (PhNH 2 ) 2 dimer is ca. 10 kJ mol −1 less than the overall energy itself, the rest being due to the C-H· · · π bond. Therefore, the computational results do indicate that the halogen bond between the primary amine group and a perfluorinated iodobenzene as a halogen bond donor is indeed more favourable than the (H)NH···NH 2 hydrogen bond between the amine groups, even in the case of aromatic amines, which is in line with the observed total absence of the said hydrogen bonds in the obtained cocrystals. However, they also indicate that weak interactions (e.g., C-H· · · π contacts) can have a sufficient contribution to the overall energy to alter this preference, which might explain our failure to prepare cocrystals of pphda with 13tfib and 135tfib; bnzd with 14tfib; and otol with 13tfib.
Crystals 2021, 11, x FOR PEER REVIEW 8 of 12 absent (Figure 6b), the only interaction between the molecules being a single I‧‧‧N halogen bond almost perpendicular (∠ (I···N-C) = 106.6°) to the plane of the aniline ring (which closely corresponds to the conformation of halogen-bonded complexes also observed in the cocrystal structures). Therefore, the calculated difference in binding energies for (PhNH2)2 and (PhNH2)(ipfb) does not pose a reliable representation of the difference in hydrogen-and halogen-bond energies with the primary amine as an acceptor (and donor of the former). In order to avoid the additional stabilisation of the hydrogen-bonded complex by a C-H⋯π interaction, we opted for an additional set of computations using methylamine as the model primary amine molecule. In the (MeNH2)2 dimer, the (H)NH···NH2 hydrogen bond is indeed the only contact within the dimer (Figure 6c). Significantly, here the binding energy in the halogen-bonded (ipfb)(MeNH2) complex (Figure 6d) is ca. 15 kJ mol −1 higher than the one for the hydrogen-bonded (MeNH2)2 dimer. There is a difference in the basicities of the aliphatic and the aromatic primary amine groups-the aromatic amines being weaker bases-which makes the halogen bond in the (ipfb)(MeNH2) complex stronger by ca. 8.4 kJ mol −1 than the corresponding bond in the (PhNH2)(ipfb) complex. As this reduction in the basicity of the acceptor should also proportionally influence the hydrogen bond energy, it can therefore be expected that the contribution of the R-N(H)-H···NH2 hydrogen bond to the overall binding energy of the (PhNH2)2 dimer is ca. 10 kJ mol −1 less than the overall energy itself, the rest being due to the C-H⋯π bond. Therefore, the computational results do indicate that the halogen bond between the primary amine group and a perfluorinated iodobenzene as a halogen bond donor is indeed more favourable than the (H)NH···NH2 hydrogen bond between the amine groups, even in the case of aromatic amines, which is in line with the observed total absence of the said hydrogen bonds in the obtained cocrystals. However, they also indicate that weak interactions (e.g., C-H⋯π contacts) can have a sufficient contribution to the overall energy to alter this preference, which might explain our failure to prepare cocrystals of pphda with 13tfib and 135tfib; bnzd with 14tfib; and otol with 13tfib. Figure 5. (a) Halogen bonding between bnzd and 135tfib molecules in the structure of (bnzd)(135tfib)4; (b) a layer formed by linking bnzd and 135tfib molecules via I···NH2, I⋯Cπ and I···I halogen bonds in the structure of (bnzd)(135tfib)4. Figure 5. (a) Halogen bonding between bnzd and 135tfib molecules in the structure of (bnzd)(135tfib) 4 ; (b) a layer formed by linking bnzd and 135tfib molecules via I···NH 2 , I· · · C π and I···I halogen bonds in the structure of (bnzd)(135tfib) 4 .

Conclusions
In the absence of competing halogen acceptors, the amine group has proven to be a reliable halogen bond acceptor, forming I···N halogen bonds up to the total exclusion of N-H⋯N hydrogen bonds, leaving the amine hydrogen atoms participating only in weakly bonding contacts. This is in line with the difference in the corresponding calculated bond energies with methylamine as the halogen/hydrogen bond acceptor. However, as only in 5 out of the 9 tested donor/acceptor combinations yielded cocrystals, it seems that this difference in halogen-and hydrogen-bond energies is not always sufficient to lead to the formation of a cocrystal. It is therefore probable that the deciding factors as to whether or not a halogen-bonded cocrystal of a primary (di)amine will be obtainable, will in many cases be from other contributions, such as the overall crystal packing of the reactants and the products.

Conclusions
In the absence of competing halogen acceptors, the amine group has proven to be a reliable halogen bond acceptor, forming I···N halogen bonds up to the total exclusion of N-H· · · N hydrogen bonds, leaving the amine hydrogen atoms participating only in weakly bonding contacts. This is in line with the difference in the corresponding calculated bond energies with methylamine as the halogen/hydrogen bond acceptor. However, as only in 5 out of the 9 tested donor/acceptor combinations yielded cocrystals, it seems that this difference in halogen-and hydrogen-bond energies is not always sufficient to lead to the formation of a cocrystal. It is therefore probable that the deciding factors as to whether or not a halogen-bonded cocrystal of a primary (di)amine will be obtainable, will in many cases be from other contributions, such as the overall crystal packing of the reactants and the products.