Introduction

Nitrogen heterocycles are of considerable interest for their ability to act as ligands in coordination chemistry, notably supporting multimetallic compounds and, in particular, compounds having metal-metal bonds (Chipman & Berry, 2020;Beach et al., 2021;Kerru et al., 2020). Examples of these types of ligands can be seen in 2-naphthyridylphenylamine (Ding et al., 2015;Liu, Wang et al., 2009;Liu, Chen et al., 2009;Tsai et al., 2013), 1,8-naphthyridin-2(1H)-one (Chang et al., 2017), 2-anilinopyridinate (Roy et al., 2022) and 2,2 0 -dipyridylamine (Hdpa) (Chipman & Berry, 2018a,b;Lescoue ¨zec et al., 2001;Berry et al., 2003;Hsiao et al., 2008).

We have recently explored the ability of the ligand 2,2 0dipyridylamine (Scheme 1 shows the structures of Hdpa, succinimide, the proposed 'succinimidine' structure, and the observed structure of 1) to support linear trimetallic metalmetal-bonded compounds (Brogden & Berry, 2016). In the search for other multitopic N-donor ligands that might support similar structures, our attention was drawn to the class of compounds called 'imidines', first described by Pinner in 1883 (Pinner, 1883) and then later by Elvidge and Linstead in the 1950s. In particular, we focus on the heterocyclic compounds 'succinimidine' and 'glutarimidine' (Elvidge et al., 1959;Elvidge & Linstead, 1954). These compounds were so named because of their proposed structural analogy to succinimide (Scheme 1) and the corresponding six-memberedring analog glutarimide. Since imidines represent a relatively rare functional group, these structures, proposed solely on the basis of elemental analysis results, have been propagated in prominent organic chemistry textbooks (March, 1992). We show here that although solution-based studies agree with the historically predicted imidine tautomers, in the solid state, the compounds 'succinimidine' and 'glutarimidine' adopt a different tautomeric form from those originally proposed. In the solid state, the structures are unsymmetric imino-amines and are better named systematically as 2-imino-3,4-dihydro-2Hpyrrol-5-amine (1) and 6-imino-3,4,5,6-tetrahydropyridin-2amine (2).

The 1950s syntheses involved the reaction of methanol solutions of terminal dinitriles (succinonitrile, glutaronitrile, or adiponitrile) with liquid ammonia before heating (Elvidge & Linstead, 1954;Elvidge et al., 1959). We have found that similar results can be obtained by saturating a methanol solution of succinonitrile with anhydrous ammonia. This solution, when heated for 18 h in a sealed bomb flask, yielded 1 in >50% yield. The product is easily separated from the mother liquor by precipitation via the addition of excess diethyl ether. The synthesis of 2 was performed in an almost identical manner; however, to achieve a useful yield, the reaction mixture was heated for 40 h total. The solvent was then removed by rotary evaporation and yellow crystals separated from the residual oil, which was washed away with ether. The modified Pinner reaction conditions result in protio-neutral ring closing to yield the N-heterocycle with two additional N-atom-based functional groups. Both the original article from Pinner and the later articles from Elvidge and Linstead draw all three N-atom sites as being singly protonated in a symmetric 'imidine' form (Pinner, 1883;Elvidge & Linstead, 1954;Elvidge et al., 1959). Elvidge and Linstead additionally reported that reaction of the imidines with water sequentially replace one and then both terminal N-atom functional groups with carbonyl groups, such that 'succinimidine' could be fully hydrolyzed to form succinimide (Elvidge & Linstead, 1954;Elvidge et al., 1959). While the symmetric structure of succinimide in the solid state is well established (Yu et al., 2012;Mason, 1961), the monohydrolyzed forms of 1 and 2 have not been investigated before, and they are structurally characterized here (Scheme 2 shows the structures of the most stable solid-phase tautomers of the species described . in this article, with only one resonance structure being shown for the protonated species found in 1•HCl). A combination of solid state, solution, and computational studies are employed to best describe the various possible tautomers of these species.

Experimental

General methods

Methanol (Sigma-Aldrich) was distilled from CaH 2 under N 2 and used immediately. Succinonitrile and glutaronitrile were purchased from Sigma-Aldrich and used as received. Inhibitor-free anhydrous diethyl ether was purchased from Sigma-Aldrich and used as received. All deuterated solvents were purchased from Sigma-Aldrich, used as received, and stored long term in air. Unless otherwise noted, all manipulations were performed in air. Electrospray ionization mass spectrometry was performed with a Thermo Q Exactive Plus mass spectrometer. IR spectra were recorded with a Bruker Tenser 27 spectrometer using an ATR adapter. 1 H NMR spectra were recorded on a 400 MHz Bruker Avance III spectrometer. Caution! The synthetic procedures for the preparation of 1 and 2 involve heating a sealed reaction vessel and should only be performed at or below the scale described here using rated thick-walled glassware, with a protective blast shield.

Synthesis and crystallization

Synthesis of 1.

Imidine 1 was synthesized through a modification of the literature procedure of Elvidge & Linstead (1954). Anhydrous methanol (70 ml), succinonitrile (4.02 g, 50.1 mmol), and a Teflon stirrer bar were combined in a 250 ml heavy-walled threaded glass vessel. The solid was fully dissolved and the resulting solution was sparged with anhydrous ammonia gas until saturated. The flask was then tightly sealed and partially submerged in an oil bath. The oil bath was programmed to heat to 70 C for 18 h before automatically cooling to room temperature. A blast shield was placed in front of the flask and the heating cycle was started. Upon cooling to room temperature, the pressure flask containing a black solution was removed from the oil bath. Activated carbon ($3 g) was added to the solution, which was sparged with nitrogen for 10 min. The solution was then filtered through Celite to yield a pale-yellow filtrate. This filtrate was added to diethyl ether (300 ml), resulting in precipitation of the product. The suspension was filtered through a glass frit and the off-white solid was washed several times with ether.

The solid was dried under high-vacuum overnight and stored in a nitrogen glove-box without further purification. X-rayquality crystals were obtained by slow diffusion of diethyl ether into a saturated solution of 1 in MeOH under an inert atmosphere. ESI (m/z): ([M + H] + ) 98.0712. IR (ATR, cm À1 ): 3289,3157,3077,2935,2847,1829,1772,1749,1686,1662,1654,1636,1532,1473,1453,1418,1328,1296,1265,1241,1223,1190, Experiments were carried out at 100 K using a Bruker SMART APEXII (Quasar) diffractometer. H atoms were treated by a mixture of independent and constrained refinement. Absorption was corrected for by multi-scan methods (SADABS; Bruker, 2016;Krause et al., 2015). , 1129, 1115, 996, 936, 919, 851, 822, 783, 665, 651, 641. 1 H NMR (400 MHz, DMSO): 7.37 (s, 3H), 2.46 (s, 4H). Crystals of 1•HCl were fortuitously obtained by slow diffusion of diethyl ether into a deuterated chloroform solution containing 1 (yield: 2.46 g, 25.3 mmol, 50.6%).

Synthesis of 2.

Imidine 2 was synthesized by a modified literature method (Elvidge & Linstead, 1954). Anhydrous methanol (70 ml), glutaronitrile (2.0299 g, 21.568 mmol), and an oven-dried stirrer bar were added to an oven-dried pressure flask under a constant stream of nitrogen gas. The resulting solution was sparged with nitrogen gas for 5 min and then saturated with ammonia gas. The flask was then sealed and heated at 70 C for 40 h while stirring. Once the flask had cooled, the clear solution was sparged with nitrogen for $20 min. The solvent was removed via rotary evaporation. The resulting yellow powder was washed with diethyl ether and filtered to remove residual glutaronitrile. X-ray-quality crystals were obtained by evaporation of a saturated MeOH solution (yield: 0.760 g, 31.7%). ESI (m/z): ([M + H] + ) 112.0868. IR (ATR, cm À1 ): 3254, 3004, 2954, 1666, 1605, 1543, 1457, 1418, 1373, 1334, 1316, 1316, 1187, 1145, 1103, 1061, 967, 909, 886, 791, 758, 676. 1 H NMR (400 MHz, DMSO): 7.05 (s, 3H), 2.20 (t, J = 6.5 Hz, 4H), 1.80-1.57 (q, 2H).

Synthesis of 3.

A scintillation vial was charged with 1 (1.0 g, 0.010 mol). Milli-Q water (3.4 ml, 0 C) was then added to the vial, immediately turning the solution faint brown. The vial was stored in a 0 C refrigerator overnight. The next day, white crystals (yield: 0.68 g, 0.0069 mol, 69%) suitable for X-ray diffraction analysis were collected from the solution. ESI (m/z): ([M + H] + ) 99.0552. IR (ATR, cm À1 ): 3220, 3135, 3019, 2938, 2918, 2851, 2360, 2341, 1686, 1627, 1526, 1456, 1437, 1418, 1397, 1338, 1294, 1251, 1221, 1161, 1009, 929, 866, 852, 827, 765, 677. 1 H NMR (400 MHz, DMSO): 8.30 (s, 1H), 8.07 (s, 1H), 2.67-2.56 (m, 2H), 2.34-2.25 (m, 2H).

2.2.4. Synthesis of 4. A scintillation vial was filled with 2 (0.10 g, 0.90 mmol) and the solid was subsequently dissolved in a minimal amount of Milli-Q water. The resulting solution was cooled overnight before allowing ether vapor to diffuse into the solution. The product precipitated out as white crystals (yield 0.048 g, 47%) suitable for X-ray diffraction, with a minor impurity of 6-hydroxy-4,5-dihydropyridin-2(3H)-one. ESI (m/z): ([M + NH 4 ] + ): 130.0975. IR (ATR, cm À1 ): 3381, 3185, 2967, 2947, 2920, 2886, 2823, 2774, 1644, 1534, 1506, 1458, 1426, 1418, 1349, 1299, 1274, 1222, 1153, 1120, 1071, 1056, 948, 917, 864, 807, 756, 671, 638. 1 H NMR (400 MHz, DMSO-d 6 ): , 7.35 (s, 1H), 6.80 (s, 1H), 2.24 (t, J = 7.7 Hz, 2H), 1.88 (t, J = 7.3 Hz, 2H), 1.78 (quint, J = 7.4 Hz, 2H).

Refinement

Crystal data, data collection and structure refinement details are summarized in Table 1. For the structures of 1 and 4, the diffraction data were consistent with a triclinic unit cell. The E-statistics for 1 and 4 strongly suggested the centrosymmetric space group P1, which yielded chemically reasonable and computationally stable refinements. For the structures of 2, 3, and 1•HCl, a combination of systematic absences in the diffraction data and the E-statistics were used to assign the centrosymmetric space groups P2 1 /c, P2 1 /n, and C2/c, respectively.

The structures were solved via intrinsic phasing and refined by least-squares refinement on F 2 , followed by difference Fourier synthesis. All non-H atoms above 70% occupancy were refined with anisotropic displacement parameters. Unless otherwise stated, all H atoms were included in the final structure-factor calculations at idealized positions and were allowed to ride on their neighboring atoms with relative isotropic displacement coefficients. In the structure 1•HCl, all amine H atoms were fixed at idealized locations, where as the imidine and water H atoms were allowed to freely refine.

The coordinates of the H atoms bound to N atoms in 1, 3, and 4 were allowed to refine freely. In 2, residual electron density provided strong evidence for the coordinates of the N-atom-bound H atoms; however, there was not sufficient electron density to allow the H atoms to refine freely. As such, the coordinates of the H atoms bound to N atoms in 2 were fixed at idealized positions.

In the structure of 2, the three methylene C atoms of the ring are disordered over two positions, with a major occupancy of 85.4 (6)%. The lesser fraction of the disordered part of the ring was restrained to the geometry of the major fraction of the same ring. One of the methanol solvent molecules exhibited disorder of the CH 3 protons.

Results and discussion

Structural commentary

Three of the title N-heterocycles, namely, 1, 3, and 4, crystalize with only one molecule in the asymmetric unit, with no disorder or solvent molecules. The crystal structure for 2 includes two N-heterocycles and two methanol solvent molecules in the asymmetric unit. The two independent molecules of 2 (denoted 'upper' and 'lower') interact via a set of two N-HÁ Á ÁN hydrogen bonds to form a dimeric structure. A similar structural motif is seen in the structure of succinimide (Yu et al., 2012;Mason, 1961) and for some of the other compounds described here, when looking at the structures beyond just the asymmetric unit (vide infra). Additionally, one of the molecules of 2 displays disorder across the three -CH 2units in the backbone, and one methanol molecule shows disorder of the H atoms on the -CH 3 group. The structure of 1•HCl contains one neutral five-membered heterocycle, its protonated species, a Cl À counter-ion, and one solvent water molecule. The asymmetric unit of each structure is shown in Fig. 1.

In 1, the NH protons are distributed such that one terminal N atom is doubly protonated as an amine, the N atom in the ring is not protonated, and the other terminal N atom is singly protonated, as an imine, with the proton pointing towards the hydrophobic backbone. In 2, the H atoms are distributed in a nearly identical manner. However, due to intermolecular O-HÁ Á ÁN hydrogen-bonding interactions with the solvent methanol molecules, the imine N atom of each of the two independent molecules of 2 has its single H atom pointed away from the hydrophobic backbone. In both 3 and 4, the O atom binds as a carbonyl group, as indicated by the short C O distances of 1.231 (1) and 1.238 (1) A ˚. As in the NNN structures, the N atom in the ring is not protonated, and the terminal N atom is doubly protonated as an amine. Crystals of 1•HCl were obtained fortuitously from slow diffusion of diethyl ether into a solution of deuterated chloroform containing 1. In the structure of 1•HCl, there exists both a neutral species, comparable to the heterocycle found in 1, and a protonated cationic species where both terminal N atoms are doubly protonated, with the N atom in the ring being left unprotonated. The protonated species in 1•HCl is balanced by a Cl À anion. The protonation states of all the complexes can be seen in Fig. 1. Notably, the protonation states of all the compounds differ from the structure of succinimide, which remains symmetric despite forming similarly asymmetric hydrogen-bonded dimers (Yu et al., 2012;Mason, 1961). The structures of 1 and 2 are also notably inconsistent with their earlier structural proposals as 'succinimidine' and 'glutarimidine', and it is particularly notable that protonation of 1 to form the HCl salt occurs at a terminal imine rather than the Figure 2 A generic structure used to define the bonds of interest.

Table 2

Selected bond lengths and comparisons (A ˚) of the structures. 299 (3) 1.343 (3) 1.349 (3) 1.294 (2) 0.005 (5) 0.006 (6) 1•HCl (neutral species) 1.314 (3) 1.323 (3) 1.393 (3) 1.274 (3) 0.040 (6) 0.070 (6) 2 (upper) 1.329 (2) 1.316 (2) 1.386 (1) 1.280 (2) 0.049 (4) 0.070 (3) 2 (lower) 1.325 (2) 1.323 (2) 1.381 (2) 1.289 (2) 0.036 (4) 0.058 (4) 3 1.311 (2) 1.333 (1) 1.379 (2) 1.231 (1) -0.046 (3) 4 1.315 (2) 1.334 (1) 1.366 (2) 1.238 (1) -0.032 (3) internal ring position. These observations are consistent with pK a data for terminal versus internal imines: (Ph) 2 C NH (pK a = 31.0) (Bordwell & Ji, 1991) and PhCH 2 N C(Ph) 2 (pK a = 24.3) (Bordwell, 1988).

See Fig. 2 for definitions of distances

The proposed protonation states of 1-4 are further supported by the bond lengths across the heteroatoms, as seen in Fig. 2. These bond distances, as well as relevant comparisons, are given in Table 2. We note the neutral compounds show statistically meaningful differences between the A/D and B/C bond pairs defined in Fig. 2. Specifically, these differences appear to indicate a localized -system with alternating single and double bonds, where the shorter bonds are localized to B

Table 5

Hydrogen-bond geometry (A ˚, ) for 2. and D. In contrast, these differences in the structure of the protonated species of 1•HCl are statistically insignificant. Thus, the structure of the protonated species in 1•HCl is best described by a delocalized electronic structure which could be represented by the two limiting resonance forms shown in Scheme 3. Notably, the neutral species in 1•HCl shows nearly identical differences in the bond lengths to those in 1. Notably, the neutral molecule in 1•HCl and in 2 show an alternate binding motif for the imine-bound proton observed in 1. This alternative binding motif likely arises from the hydrogenbonding interaction blocking the other side of the imine.

To gain further insights into the protonation states of 1, computational studies were performed. All calculations were carried out using GAUSSIAN16 (Frisch et al., 2016), Hartree-Fock theory, and the 6-31g(d) basis set. Input geometries were constructed from modified crystallographic coordinates. The geometry-optimized xyz coordinates for all structures are provided in the supporting information (Tables S1 andS2).

The calculations indicate that, in the gas phase, the Gibbs free energy of the symmetric 'succinimidine' tautomer is $1.9 kcal mol À1 more stable than the asymmetric form observed crystallographically. The energy difference is small enough to allow for the network of hydrogen bonds in the crystal structure to dictate which tautomer of the compound is observed in the solid state. This packing-influenced tautomerization also aligns with previous tautomer-based studies that utilized variable-temperature crystallography and thermal evolution to better understand the tautomer ratios in keto-amine/iminoenol systems (Godsi et al., 2004). To examine which tautomer is preferred in solution, we examined a solution of 1 in DMSO-d 6 by 1 H NMR spectroscopy. The main signal observed is a singlet at 2.46 ppm assignable to the CH 2 protons, consistent with the symmetric 'succinimidine' tautomer. This provides evidence that in solution, the imidine structure, as historically drawn in textbooks (March, 1992), dominates, yet in the solid state, the asymmetric tautomer is prevalent. Additionally, the singlet at 7.37 ppm likely indicates rapid exchange between all three of the NH protons. For reference, the 1 H NMR spectrum of succinimide in CDCl 3 -d 1

Table 7 Hydrogen-bond geometry (A ˚, ) for 4. electronic reprint consists of a singlet at 2.769 ppm (https://www.chemicalbook. com/SpectrumEN_123-56-8_1HNMR.htm).

Crystal packing

Unsurprisingly, the large number of hydrogen-bond donors and acceptors in the molecules examined here result in significant intermolecular hydrogen-bonding interactions throughout the crystal structures (Tables 34567). In 1, 2, and 4, the hydrogen-bonding interactions result in oligomerization of the planar dimer units formed by the hydrophilic section of the molecules being paired together (Fig. 3). Each pair involves a double-hydrogen-bonded eight-membered ring reminiscent of the structural motifs seen for carboxylic acid dimers in the solid (Jasinski et al., 2009), solution (Kolbe et al., 1997), or gas phase (Emmeluth et al., 2003). The linking of these hydrogen-bonded dimers through further lateral hydrogen bonds creates long two-dimensional ribbons throughout the crystal lattice. These ribbons stack together to form the three-dimensional crystal structures. For both 1 and 4, there are no hydrogen-bonding interactions between ribbons either in the same plane or in between planes, as seen in Fig. 4. This pattern is broken with 2, where the methanol solvent molecule hydrogen bonds in between sheets. This additional hydrogenbonding interaction perpetuates throughout the packed crystal structure, making a series of interlaced sheets, as seen in Fig. 5. Compounds 1 and 2 contain a mismatch in the number of hydrogen-bond-donating and -accepting groups, leading to structures in which one of the potential hydrogenbond donors remains unsatisfied.

In the structure of 1•HCl, hydrophobic backbone and hydrophilic heteroatoms alternate in the plane, as seen in Fig. 1. Additionally, the solvent water molecule in 1•HCl hydrogen bonds between sheets, bridging pairs of these sheets, as seen in Fig. 6. The major exception to the planar molecular sheets stabilized by a hydrogen-bond network is found in the crystal packing of 3. Compound 3 does not form discrete carboxylic acid-style dimers. Instead, each molecule of 3 has hydrogen-bonding interactions with four other molecules of 3 that form an interconnected three-dimensional lattice as the molecules stack perpendicular to each other, as seen in Fig. 7. The introduction of the three-dimensional hydrogen-bonding lattice is likely what aids in the crystallization of 3 from aqueous conditions.

Summary

Through careful analysis of solid-state and solution phase measurements of the historical imidines, the apparent experimental disagreement between whether their structures are asymmetric or symmetric tautomers has been resolved. The crystallographic data provide evidence for the solid-state asymmetric tautomer for both the five-and six-membered ring compounds, whereas solution phase NMR spectroscopy data strongly indicate a more symmetric form. The energetic differences between the symmetric and asymmetric forms were calculated to be sufficiently small to allow for tauto-merization to reasonably occur in solution at room temperature. The synthetic methods and characterization of these compounds have been modernized and safety issues associated with the synthesis have been clarified.

Figure 7

The stacked hydrogen-bonding network observed in 3. All atoms are drawn with 50% probability displacement ellipsoids and dotted lines are used to indicate hydrogen-bonding interactions. [Symmetry codes: 3), (4); APEX3 (Bruker, 2017) for ( 2). Cell refinement: SAINT-Plus (Bruker, 2016) for 1HCl, (1), ( 3), (4); SAINT (Bruker, 2017) for (2). Data reduction: SAINT-Plus (Bruker, 2016) for 1HCl, (1), ( 3), (4); SAINT (Bruker, 2017) for (2). Program(s) used to solve structure: SHELXT2014 (Sheldrick, 2015a) for 1HCl, (3), (4); olex2.solve (Bourhis et al., 2015) for (1); SHELXT2018 (Sheldrick, 2015a) for ( 2). For all structures, program(s) used to refine structure: SHELXL2018 (Sheldrick, 2015b); molecular graphics: OLEX2 (Dolomanov et al., 2009); software used to prepare material for publication: OLEX2 (Dolomanov et al., 2009). where P = (F o 2 + 2F c 2 )/3 (Δ/σ) max < 0.001 Δρ max = 0.44 e Å -3 Δρ min = -0.24 e Å -3

5-Amino

Special details

Geometry. All esds (except the esd in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s. planes.

Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å

Hydrogen-bond geometry (Å, º)

2-Imino-3,4-dihydro-2H-pyrrol-5-amine (1)

Crystal data

Special details

Geometry. All esds (except the esd in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s. planes.

Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å

3.04 ( 11)

Data collection

Bruker SMART APEXII diffractometer Radiation source: sealed X-ray tube, Siemens, K FFCU 2K 90 Equatorially mounted graphite monochromator Detector resolution: 7.9 pixels mm -1 0.60\ ω and 0.6\ φ scans Absorption correction: multi-scan (SADABS; Bruker, 2016)

T min = 0.853, T max = 0.915 7398 measured reflections 886 independent reflections 775 reflections with

Special details

Geometry. All esds (except the esd in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s. planes. Hydrogen-bond geometry (Å, º) where P = (F o 2 + 2F c 2 )/3 (Δ/σ) max < 0.001 Δρ max = 0.44 e Å -3 Δρ min = -0.27 e Å -3

Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å

Special details

Geometry. All esds (except the esd in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s. planes. Hydrogen-bond geometry (Å, º)

Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å

Special details

Geometry. All esds (except the esd in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s. planes.

Fractional atomic coordinates and isotropic