The title compound, [Ni(1,10-phen)₂(H₂O)₂](NO₃)₂ 1, crystallizes in the monoclinic space group P2₁/c with a large b and cell volume; a = 19.032(1) Å, b = 47.979(3) Å, c = 17.596(1) Å, β= 116.066(4)°, V = 14433.3(14) ų, Z = 24 (Z′ = 6). Basic crystal information is contained within Table 1. Good CHN analysis data were obtained for 1 as follows. Ni₁C₂₄H₂₀N₆O₈ analysis: observed (calculated) %; C 49.76% (49.77), H 3.21% (3.48), N 14.31% (14.51).

The unit cell and space group were determined uniquely from the diffraction data. Although the cell is rather large, it is clear from the diffraction images that this cell does not arise because of incorrect indexing due to twinning or the presence of multiple crystals. The large cell arises for good structural reasons; here small but significant differences in the orientation of [Ni(1,10-phen)₂(H₂O)₂]²⁺ cations and associated nitrate anions give rise to the large unit cell.

There are six crystallographically distinct [Ni(1,10-phen)₂(H₂O)₂]²⁺ cations in the asymmetric unit, each of which comprises a distorted octahedral Ni center coordinated by four N atoms from two 1,10-phen ligands and two O atoms from the ligated H₂O molecules. (Figure 1a) There exist, therefore, twelve discrete 1,10-phen and twelve nitrate counter anions in the asymmetric unit. Figure 1b shows the coordination environments for the [Ni(1,10-phen)₂(H₂O)₂]²⁺ cations and the close neighboring nitrate anions. Atomic numbering is presented for only Ni atoms in this figure for clarity, but ORTEP representations of each unique [Ni(1,10-phen)₂(H₂O)₂]²⁺ cation are provided separately in the Supplementary Information.

The 1,10-phen ligands adopt the expected chelating mode of coordination with the chelating angles in the range 79.41–80.29° with a mean 79.90(11)°. It is also common that two 1,10-phen ligands of the same cationic cluster adopt cis configuration with any two co-centric molecular planes set nearly orthogonal; here the angles vary between 83.68° and 89.28°. The conformation about each Ni atom can therefore be described as symmetrical-cis (sy-cis) configuration. The six bond lengths about each nickel cation are rather similar; the Ni-N bond lengths lie in the range 2.065(3)−2.101(3) Å with a mean of 2.082 Å and standard deviation 0.009 Å, while Ni-O bond lengths vary between 2.058(3)−2.102(3) Å with a mean of 2.080Å and a standard deviation 0.014Å. Selected bond lengths and angles are listed in Table 2. The as-described configuration, bond lengths and angles are in good agreement with those reported for the closely related Ni-1,10-phen structures [7-9]. According to the refined bond lengths and the assumption for parameters R₀(Ni^II-O^-II) = 1.654, R₀(Ni^II-N^-II) = 1.700, and b= 0.370 [10,11], the bond valence sums have been calculated and the results confirm a divalent state for Ni atoms: 2.058, 2.050, 2.060, 2.066, 2.054 and 2.061 for Ni1, Ni2, Ni3, Ni4, Ni5 and Ni6, respectively. The presence of divalent nickel within the [Ni^II(1,10-phen)₂(H₂O)₂]²⁺ cation is in accordance with the crystallographic observation of two nitrate anions for every nickel complex.

Three other closely related compounds built up of identical cationic complex as 1 have been previously reported, including [Ni(1,10-phen)₂(H₂O)₂](ClO₄)₂.4H₂O [7], [Ni(1,10-phen)₂(H₂O)₂](SO₄)·5.6H₂O [8] and [Ni(1,10-phen)₂(H₂0)₂](ZnCl₄)·0.4H₂O [9]. These structures have much simpler asymmetric units that comprise either one-half or a single cationic cluster with counter anions of tetrahedral geometry. The patterns of hydrogen bonding interactions and the appearance of π-π interactions are similar in these structures. The employment of the nitrate anion with unit charge instead of the divalent anions in the previously reported compounds results, at a first glance, in a different cation-to-anion ratio; 1:2 in 1 compared with 1:1 in two of the previously reported structures. The triangular planar geometry of the nitrate can, in addition, arrange more flexibly in smaller space compared to the tetrahedral ones. The hydrogen bonding analysis listed in Table 3 reveals that every nitrate is hydrogen bonded to the neighboring cationic clusters in such a way that the spatial orientation of the clusters is directed through the hydrogen bonding interactions. Figure 2 shows the O-H⋯O interactions about each cluster formed between the O-H groups of the water ligated to the [Ni^II(1,10-phen)₂(H₂O)₂]²⁺ cations and the O atoms of the nearby nitrate anions. It is clear that subtle changes in the orientation of each [Ni^II(1,10-phen)₂(H₂O)₂]²⁺ cation facilitate hydrogen bonding to the nitrate anions which themselves occupy positions designed to maximize interactions with the surrounding cations. Relatively short π-π distances between the 1,10-phen ligands of different complexes in the range 3.45 Å to 3.5 Å are present.

The structure of 1 is therefore remarkable for the beautiful chains, formed from classical O-H⋯O hydrogen bonds, which run parallel to the crystallographic c axis. Within 1 there are four independent types of chain which, while similar, have subtle differences. These chains are illustrated A, B, C, and D in Figure 2. In each case, the repeat unit has length 17.596(1) Å, which is a unit translation along c. These can be classified into two pairs of chains; A&B and C&D. For each of the first pair, identical metal complexes are assembled into chains through hydrogen bonding by nitrate anions. For the second pair, while each chain has the same basic arrangement of ions, two different metal complexes make up each chain, bridged by hydrogen bonds to nitrate. However, there is additional nitrate bound by hydrogen bonds which decorates each of the chains, and the differences in this decoration make each chain unique. These chains are described further below.

Complexes of Ni1 are assembled into saw tooth chains through classical hydrogen bonds between bound water and nitrate with central atom N100 forming a C 2 2 ( 8 )hydrogen bonding pattern (chain A). Successive complexes are related by the c-glide operation. The chain can be labeled as having an MnMn repeat in which there is only one crystallographically unique metal complex (signified M) and one unique nitrate (signified n). This chain is further decorated by nitrate centered on N200 which does not form part of the chain link. This nitrate forms classical hydrogen bonds to the two bound water molecules on Ni1 through O201 and O202 forming an embrace that may be described as R 2 2 ( 8 )using the Graph Set Notation [12]. The second chain (B) is composed of complexes of Ni2 and nitrate (N180) and is extremely similar to the first, presenting a C 2 2 ( 8 )pattern, differing only in the way it is decorated by further nitrate. Adjacent nickel complexes are again related by the c-glide operation and the chain can be labeled MnMn as for the first. However, while there is a further nitrate anion (central atom N160) bound to this chain at every metal centre through hydrogen bonding, it does so through only one O atom. In contrast to chain A, each bound water is still involved in localizing this nitrate, so this embrace can be described by the different graph set, R 2 1 ( 6 ).

A third independent chain (labeled C) is formed from two metal complexes Ni3 and Ni6 which are bridged by hydrogen bonds from two nitrates (N130 and N150). Each of the four species involved in the formation of this chain is crystallographically unique, but there is a regular alternating arrange of metal complexes and nitrates to give a chain with form MnM′n′MnM′n′. The formal description of this chain as C 4 4 ( 16 )demonstrates the larger crystallographically unique repeat unit. The chain has a very similar form to the two described above but is not generated by the c-glide. This chain is decorated by two unique nitrate groups: the nitrate with central atom N120 forms an R 2 2 ( 8 )embrace at Ni3, which the nitrate N140 binds through a R 2 2 ( 8 )ring at Ni6.

Finally, the fourth chain, labeled D, is similar to the third; it is composed of the complexes Ni4 and Ni5 bridged by classical hydrogen bonds to the nitrates N110 and N190. These four symmetry independent units alternate to generate a chain with form MnM′n′MnM′n′ running parallel to the c-axis with notation C 4 4 ( 16 ). This is decorated by two independent nitrate anions that bind through one oxygen atom only, forming R 2 1 ( 6 )embraces. The nitrate N170 forms two hydrogen bonds at Ni4 through the oxygen O170. Very similarly, the nitrate N220 binds at Ni5 forming two hydrogen bonds through oxygen O221.

The subtle differences in the nitrate orientation and associated hydrogen bonds give rise to the complicated crystallographic asymmetric unit. The structure arrangement adopted by the metal complexes and nitrate make it possible for four different hydrogen bonded chains to form that rely upon small differences in orientation. The choice of data collection strategy (two overlapping datasets of different counting times) was important because obtaining reliable estimates of the intensity of the weak reflections is important in determining small differences in groups of light atoms such as nitrate. We note also that the interpretation of the hydrogen bonding present was greatly assisted by the presence of electron density corresponding to hydrogen atoms within final difference Fourier maps.

The hydrogen bonding chains are further packed with regular alternation, leading to the ABAB and CDCD layer registries in the direction of a. While there are no O⋯H distances indicative of classical hydrogen bonds, these chains are held together by a very large number of C-H⋯O non-classical hydrogen bonding interactions between nitrate and 1,10-phen. Applying the Jeffrey criterion for hydrogen bonds [13], there are 46 independent C-H⋯O hydrogen bonds with a mean H⋯O distance of 2.47 Å and standard deviation 0.08 Å. These generate a 2-D layered network in the ac plane, as shown in Figure 3. Full details of these interactions are given within the Supplementary Information. The 3-D hydrogen bonding supramolecular assembly with the (CDCD-ABAB-CDCD-DCDC-BABA-DCDC)_n stacking sequence is further generated according to the 2₁ and 1̄ operations in the direction of b (Figure 4). The subtle differences in both number and spatial orientation of weak but abundant C-H⋯O hydrogen bonding interactions apparently depend on the arrangement of the neighboring nitrate anions. The assembly directing function of the nitrate anions which also account for the minor deviations in the asymmetric [Ni(1,10-phen)₂(H₂O)₂]²⁺ cations is therefore illustrated.

We are confident in the correctness of the structure solution, despite the large asymmetric unit for a number of reasons. It was not necessary to apply restraints to the bond lengths within the metal complexes, only to restrain the O-H distances in water to suitable values. The root mean square (rms) deviation of pseudo-equivalent bonds lengths is small, which is a good indicator that the structure is reliable. Full details of all of the bond lengths are given in the Supplementary Information. The final difference Fourier maps after heavy atoms had been located provided information about the location of hydrogen atoms present. In this case the refinement of hydrogen atom positions subject to sensible restraints proceeded smoothly. The location and refinement of hydrogen atoms has been cited as a reliable indicator of a correct structure [14]. Although the structure displays pseudosymmetry, there are good structural reasons why a smaller asymmetric unit is inappropriate, namely the four independent hydrogen-bonding chains, which arise from differences in the orientation of the cations and nitrate. The six sheets of chains in the ac plane represented in Figure 3 that are stacked along b are close to being related by a translation of b/3 (2b/3 as identified by PLATON). However this is a rather poorly satisfied psuedosymmetry operation; the differences in the nature of the chains and their decoration enforce the need for a long b axis. It is also notable that the six independent cations within the asymmetric unit are related in pairs by a translation of 0.45 along the a direction. These are Ni1&Ni2, Ni3&Ni4, Ni5&Ni6 as can be seen in Figure 1. The nitrate positions and orientations are important in breaking this psuedosymmetry; careful inspection of the pairs of metal complexes down [100] reveals that they do not overlie due to significant rotation. The Supplementary Information contains a plot of the six independent [Ni(1,10-phen)₂(H₂O)₂]²⁺ cations overlaid under the action of the pseudosymmetry operations present. There is a clear deviation from a truly symmetric array.

The FT-IR spectrum of 1 and a complete list of assignments are provided within the Supplementary Information. The FT-IR spectrum of 1 shows characteristic vibrational bands of each of the ligated species [15-18], which is in good agreement with the crystal structure obtained at 120 K. The strong, broad absorption centered on 1382 cm⁻¹ with three shoulders at 1359, 1341 and 1305 cm⁻¹ can be assigned to asymmetric stretching vibration (ν₃) of the nitrate anions present. The local D_3h symmetry of the non-coordinated nitrate, means the strong absorption due to the out-of-plane ν₂ and in-plane ν₄ deformations are clearly visible at 847 and 726 cm⁻¹, respectively. The presence of a weak band at 1051 cm⁻¹, which can be due to the otherwise IR-inactive symmetric stretching ν₁ suggests the involvement of the nitrate in hydrogen bonding, in accordance with the structure of 1 described above. The broad and strong band centered about 3220 cm⁻¹ confirms the existence of the ligated water. There is clear evidence for coordinated 1,10-phen within the spectrum, although peaks due to this are much weaker than those from nitrate. The band at 1516 cm⁻¹ and the less intense band at 1494 cm⁻¹ and the other weak band at 775 cm⁻¹ with a shoulder at 785 cm⁻¹ are clear in the spectrum. These bands are characteristic of the 1,10-phen within coordination complexes.

Upon heating under nitrogen, 1 loses mass in two well-defined steps and then decomposes over a very broad temperature range. Between 80 and 155 °C, the sample loses an amount of mass in good agreement with the loss of two molecules of water. Thereafter, the composition Ni(1,10-phen)₂(NO₃)₂ is remarkably stable from 155 °C to 365 °C. X-ray diffraction data collected from a sample heated to 220 °C show this phase does not display long range order. At about 365 °C there is a very sharp mass loss to generate a phase with RMM ≈ 320 g mol⁻¹ which is likely to be due to loss of 1,10-phen and loss/decomposition of nitrate. Above 370 °C, the sample loses mass steadily up to 1000 °C, until it approaches the final endpoint, NiO. Thermogravimetry and differential scanning calorimetry data are contained within the Supplementary Information.

The initial data collection was carried out at 120 K to minimize atomic thermal motion. We thought it possible that the structure might convert to a much simpler one at higher temperatures, particularly where there is sufficient thermal energy for the nitrates to rearrange by rotation. A single crystal of 1 was examined at 291 K and the structure determined by X-ray diffraction. The structure remained essentially unchanged upon warming to 291 K with space group P2₁/c. The unit cell parameters at this temperature were a = 19.1244(13) Å, b = 48.654(4) Å, c = 17.9030(12) Å, β = 116.949(5)°, V = 14,849.3(19) ų. The cell shows a 2.88% increase in volume from 120 K to 291 K. At this higher temperature the scattering of the crystal was markedly weaker and the final solution poorer so that hydrogen atoms were not apparent in final difference Fourier maps. There are small changes in the structure upon heating, but the large cell is still uniquely identified and there is no evidence for the asymmetric unit becoming closer to a simpler atomic arrangement.

Heating the crystal to 333 K in an attempt to collect data led to catastrophic breakdown of the crystal. This was apparent from the diffraction images and from the cracking of the crystal into myriad fragments. However, it was possible to collect a limited amount of useful diffraction data from a crystal held at 308 K. At this temperature, the crystal ultimately decomposed during data collection, but it was possible to collect twenty diffraction images of reasonable quality before the sample ceased to be crystalline. These twenty images showed evidence of a dominant, crystalline domain within the sample without the long b-axis of 1 (See Supplementary Information). Spots due to this domain were relatively sharp and it was possible to index the unit cell of this phase, hereafter labeled 1b. The initial unit cell obtained was monoclinic with parameters a = 9.565(5), b = 16.246(9), c = 15.979(6) Å and β = 95.06(3)°; the unit cell volume was found to be 2474 ų. Systematic absences in the data suggested the space group was P2₁/n. The unit cell volume is almost exactly one sixth of the volume of 1 at 291 K. Furthermore this unit cell has very similar dimensions to one reported by Liu et al. for diaqua(1,10-phenanthrolin-2-ol) nickel(II) dinitrate at 298 K [19], which crystallises in P2₁/n with a single metal complex in the asymmetric unit. The components of this compound are extremely similar to those of 1b, differing only in the intramolecular hydrogen-bonding about the Ni²⁺ cation. This information strongly suggests that 1b is a high temperature form of 1 which crystallises with a single [Ni(1,10-phen)₂(H₂O)₂]²⁺ cation in the asymmetric unit (Figure 5). It is possible to reset 1b into the standard setting of space group 14, P2₁/c, with cell parameters a = 9.565(5) Å, b = 16.246(9) Å, c = 17.864(7) Å and β = 116.97(3)°. The relationship between the unit cells of 1 and 1b then becomes apparent: a′ = a/2; b′ = b/3; c′ = c; β′ = β. The sixfold reduction in unit cell volume does not alter the symmetry but leads to a greatly simplified asymmetric unit. It is important to note that there was no evidence in the diffraction images for the presence of the larger cell at 308 K.

Although the crystal was weakly diffracting beyond d = 1.05 Å and only a very limited number of data frames could be collected before the crystal decomposed, it was possible to refine a structure for 1b against the available data. Approximately 30% of the available data were recorded to a resolution of 1.05 Å. Limitations in the data quality and coverage necessitated the use of relatively severe restraints on the molecular geometry, including rigid body restraints on the 1,10-phen molecules. Hydrogen atoms were all placed at calculated positions. The final structure determination is rather poor quality (R = 12.53% for the data with I > 4σ_I) but gives very useful chemical information about the structure of 1b. Full details are given in the Supplementary Material.

The structure of 1b is in good agreement with the prediction of a Z′ = 1 structure composed of hydrogen bonding chains similar to those in 1. The overall structure is similar to that of 1. The key feature of 1b is that it contains only one chemically unique chain, which is illustrated in Figure 6. Using the notation for the chains in 1, these chains are similar to those labelled type A in Figure 2. Inter-chain π-π interactions are still present between 1,10-phen ligands at a distance of about 3.43 Å. It is notable that the differences in the structures are due to ordering of nitrate positions within 1, while in 1b only two nitrate positions are observed. However, the nitrates in 1b were rather poorly located, suggesting a high degree of positional and rotational disorder in 1b, reinforcing the idea that the phase change from 1 to 1b happens when there is sufficient energy to remove ordering of nitrate positions. It may be possible to undertake further studies employing synchrotron radiation to obtain a better structure for 1b in the short time (ca. 60 minutes) available before it decomposes at 308 K.

In order to grow single crystals of 1, an ethanolic solution of the organic ligand (solution A) was first prepared by dissolving 0.495 g (2.47 mmol) of 1,10-phenanthroline (C₁₂H₈N₂; Fluka, 99%) in 15.00 cm³ of ethyl alcohol (C₂H₅OH; Merck, 99.9%). The aqueous solution of Ni²⁺ (solution B) was prepared by dissolving 1.4947 g (5.09 mmol) nickel(II)nitrate (Ni(NO₃)₂·6H₂O;Carlo Erba, 99%) in 30.00 cm³ of aqueous potassium hydroxide solution (KOH; Merck, 85%) with the concentration of 0.167 mol dm⁻³. A portion of 1.50 cm³ of solution B was gently loaded into a glass tube of 5 mm in diameter and 10 cm in length, followed by another 1.50 cm³ of solution A. The purple crystals of 1 appeared at the boundary between solution A and solution B after leaving the tube undisturbed for about a month at room temperature. Following drying, the resulting crystals were visible homogeneous and well dispersed, rather than forming aggregates.

A good sized crystal, visibly representative of the bulk product, was selected for diffraction measurements. A data set of 28015 independent reflections was collected in a range of 1.36 ≤ θ/° ≤ 25.89 from a 0.60 × 0.32 × 0.24 mm³ purple block crystal of 1 using a Stoe IPDS2 diffractometer and Stoe X-Area software [20]. The temperature of the crystal was controlled during data collection using an Oxford Instruments nitrogen gas cryostream and set to 120 K for the first data collection. Two tranches of data with different counting times were collected. Each frame employed a 1° ω-rotation. The first run consisted of 158 images with a counting time of 48 s per frame. The second run consisted of 123 images with a counting time of 390 s per frame. There was significant overlap between the two runs in reciprocal space to allow reliable scaling of the two tranches to give a single data file. The use of two different counting times was designed to obtain the most reliable signal to noise ratio for both the strong and weak reflections.

A face indexed absorption correction was applied to the collected data within the Stoe X-RED software using the Tompa method [21,22]. The data were then reduced and merged giving 15,842 reflections with I > 2σ(I) and an internal R of 0.0862. The structure was solved by direct methods using SHELXS-97 and full-matrix least-squares refinement was carried out with SHELXL-97 via the WinGX program interface [23,24]. All non-hydrogen atoms were refined with anisotropic displacement parameters. Despite the complexity of the structure it was not necessary to restrain bond distances or angles involving any of the non-hydrogen atoms. Hydrogen atoms were unambiguously located within Fourier difference maps once all other atoms had been identified. Those of the 1,10-phen ligands were places at geometrically calculated positions using a riding model with C-H bond length 0.95 Å. The hydrogen atoms attached to water were placed from the Fourier difference maps, subject to sensible O-H bond length restraints (target distance 0.86 Å) and where necessary geometric restraints to generate chemically sensible hydrogen bonds. Displacement parameters for the 24 independent hydrogen atoms of water were divided into two groups dependent upon the magnitude of the electron density at that point in a Fourier map. Within each group, a single isotropic displacement parameter was applied for all atoms. Crystallographic details for structural solution and refinement are summarized in Table 1.

The rms deviation of the bond lengths in the pseudo-equivalent fragments is a good indicator of the quality of the refinement [25] with large deviations indicative of an incorrect model. Here the rms deviation is 0.018 Å for the 18 equivalent bonds in the about (Ni-O and Ni-N) in the [Ni(1,10-phen)₂ (H₂O)₂]²⁺ cations. Similarly, the rms deviation of equivalent bond lengths from their mean, may also be used as a guide to the quality of fit. Here, the rms deviations from the mean are 0.014 Å for Ni-O bonds and 0.009 Å for Ni-N bonds. These small discrepancy values give confidence in the model.

The data collection strategy for the higher temperature X-ray diffraction measurements was similar. For the data collection at 291 K, three tranches of data with significant overlap but different counting times were collected, corrected for absorption, and merged to give a single set of intensity data. The data collection at 308 K recorded a single set of 20 frames. The unit cell was determined from these and the data were integrated using standard techniques and corrected for crystal decay.

The Fourier-transform IR spectrum of ground crystals of 1 was recorded using a Perkin-Elmer Paragon 1000 FT-IR instrument (4,000–400 cm⁻¹, resolution 0.5 cm⁻¹) using KBr (BDH 98.5%) disks.

Analytical data were collected at the University of Hull by Mrs C. Kennedy using a Fisons EA 1108 CHN analyzer.

Thermogravimetry: samples were placed into platinum pans, loaded into a Mettler-Toledo TGA/DSC1 Thermogravimetric Analyzer and heated under a flow of nitrogen from room temperature to 1000 °C at a ramp rate of 10 °C min⁻¹.