Crystal structures, electron spin resonance, and thermogravimetric analysis of three mixed-valence copper cyanide polymers

The crystal structures of three mixed-valence copper cyanide polymers with similar polymeric CuCN networks are described, together with the results of electron spin resonance and thermogravimetric analyses.


Chemical context
Polymeric CuCN compounds with organic ligands have continued to excite interest in light of their varied structures (Pike, 2012), the magnetic exchange or photoluminesence exhibited by many of them, and other potentially useful physical properties (Lim et al., 2008).Many hundreds of crystal structures are now known (e.g.Nicholas et al., 2019;Xu et al., 2019;Etaiw et al., 2016).One class of such polymers comprises anionic CuCN frameworks with guest cations providing charge neutrality and we have made systematic studies of such compounds containing cations derived from amines and ethanolamines (Koenigsmann et al., 2020;Corfield et al., 2022).Mixed-valence CuCN polymers containing bases coordinated to the Cu II atoms are also well known (Liu et al., 2017;Qin et al., 2016), though fewer in number than the Cu I CN complexes.Such networks would be neutral not anionic, and might therefore be capable of crystallizing with neutral molecules as guests.We have made studies of several such complexes involving diamines (Corfield et al., 2016), but until recently we were less successful at isolating crystalline complexes of mixed-valence CuCN networks involving N-substituted ethanolamines.In the present article, we describe the isolation and structural characterization of crystals of three such mixed-valence CuCN networks, along with powder ESR data and thermogravimetric analyses: poly[bis(�-3-aminopropanolato)tetra-�-cyanido-dicopper(I)dicopper(II)], 1 (Scheme 1), with the base propanolamine, coordinated to Cu II as an alkoxide; poly[bis(2-aminopropanol)tetra-�-cyanido-dicopper(I)copper(II)], 2 (Scheme 2), with the base 2-aminopropan-1-ol; and poly[bis(2-aminoethanol)tetra-�-cyanidodicopper(I)copper(II)], 3 (Scheme 3), with the base ethanolamine.

Synthesis and crystallization
Syntheses were carried out by air oxidation of solutions containing NaCN/CuCN in the presence of the amine ligand.Results varied from batch to batch.Specific details of representative syntheses follow.
For the preparation of 1, 17 mmol (0.833 g) of NaCN were dissolved in 25 ml of distilled water and 15 mmol (1.343 g) of CuCN were added and the mixture stirred and filtered, as not quite all of the CuCN had dissolved.Then 35 mmol of 3-aminopropan-1-ol (2.629 g) were added with stirring, and the colorless mixture was covered.Red crystals separated after about two weeks (yield: 0.393 g, or 21%, based upon Cu).IR spectra (cm À 1 ): 2124 (s), 2135 (s) (CN stretch); 3255 (s), 3298 (s) (N-H stretch); 3452 (m) (O-H stretch, broad, probably due to moisture contaminant).Preparations under similar conditions did not always produce homogeneous samples, but crystals of 1 were usually present.
For the preparation of 2, 20 mmol (0.995 g) of NaCN were dissolved in 20 ml of distilled water and 10 mmol (0.893 g) of CuCN were added.The mixture was stirred until a clear solution was obtained.20 mmol (1.502 g) of 2-aminopropan-1ol were added and the mixture stirred.After about three months, 38 mg of a brown product composed of gold-brown plates were obtained.Based upon a molecular formula of Cu 3 (CN) 4 L 2 , where L = 2-aminopropan-1-ol, this corresponds to a percentage yield of 2.6%.IR spectra (cm À 1 ): 2104 (s), 2116 (s) (CN stretch); 3255 (m), 3319 (m), 3347 (w) (N-H stretch); 3506 (m), 3543 (m) (O-H stretch).We were surprised to also obtain 184 mg of a crystalline material that gave the same IR spectrum in a separate synthesis designed to give a Cu I CN complex only.In this case, we used the same procedure, with 5.0 mmol of CuCN instead of 10 mmol, but the base was neutralized before addition to the aqueous NaCN/CuCN mixture, which we presumed would hinder air oxidation of Cu I .We do not have an explanation for this synthesis when so many other attempts had been unsuccessful.

Refinement
Crystal data, data collection and structure refinement details are summarized in Table 1.For structures 2 and 3, the tensor analysis in XABS2 (Parkin et al., 1995) was used to improve the absorption correction, leading to a somewhat less noisy final difference Fourier map.In all structures, C-H protons were restrained to the expected geometry, with C-H distances of 0.97 A ˚, while positional coordinates for N-H and O-H protons were refined.For C-H hydrogens, the U iso (H) values were constrained to 1.2U eq of the adjacent atoms for 1, and 1.5U eq for 2 and 3.In 1, the low-angle 020 reflection was omitted as it was partially obscured by the beamstop.For 3, data sets from two crystals were merged using SORTAV (Blessing, 1989).The size of the second smaller crystal was 0.15 � 0.10 � 0.06 mm.

Figure 1
General scheme for the title compounds.
In 1, atoms C12, C13, and C14 of the chelate ring Cu2/O11/ C12-C14/N15/Cu2 were treated as disordered above and below the central plane of the six-membered ring, and the A/B occupancies refined to 74 (2) and 26 (2)%.In both 2 and 3, the chelating ethanolamine ligands were modeled as disordered between � and � conformations.All the ligand atoms were counted as disordered, except that the tightly bound NH 2 atoms of the two disorder mates were constrained in both structures to the same positions, whereas the more loosely bound OH atoms were allowed to refine independently, along with the ligand C atoms, with constraints on the displacement parameters for the O atoms.The A/B occupancies refined to 53.4 ( 9) and 46.6 (9)% for 2, and to 64.3 (16) and 35.7 (16)% for 3.

Structural commentary
In each of the three title structures, a Cu II moiety coordinated by chelated alkanolamine bases links zigzag Cu I CN chains into a diperiodic network via bridging CN groups, with the Cu II moiety situated at a crystallographic center of symmetry, as seen in the general scheme (Fig. 1).In every case, the C atoms of the chelated rings are disordered.For convenience, we have labeled Cu I atoms as Cu1 and Cu II atoms as Cu2 in the discussion that follows.
In the structure of 1, shown in Fig. 2, the propanolamine bases have lost their hydroxyl H atoms, and coordinate as chelates to two Cu2 atoms to form a dimeric Cu II  Figs. 3 and 4 show the asymmetric units for structures 2 and 3, where single Cu2 atoms play the role of the dimeric moieties in 1 in linking the Cu1 chains into diperiodic structures.Cu2 atoms are octahedrally coordinated by two cyanide groups and by two chelating ligands, which are disordered between � and � conformations.The coordination is illustrated in Fig. 5, and comparisons of bond lengths and angles with 1 are given in Table 2.The coordination geometries are almost identical for compounds 2 and 3.In these two structures, the ligands have not lost their OH protons and the Cu-O distances are much longer than in 1 where the bonding is to an alkoxide.The Cu-NH 2 and Cu-NC bonds are also slightly longer in 2 and  The asymmetric unit of 2, showing the atom numbering and 50% displacement ellipsoids.The colors are as in Fig. 2.
The asymmetric unit of 3, showing the atom numbering and 50% displacement ellipsoids.The colors are as in Fig. 2.
3 than in 1.The bond angles at Cu1 in 2 range from 113.4 (2) to 121.8 (2) � , while in 3 they range from 118.33 (12) to 120.78 (11) � .In both cases, the largest angle is trans to the CN bridging to Cu II , in contrast to 1, where there is a larger deviation of angles from 120 � and the smallest angle is trans to the CN-Cu2 bridge.
The structure of 3 has been reported previously (Jin et al., 2006), but was redone in our laboratory for consistency.In Jin et al., the space group was given as C2/m.We chose C2/c, as in all crystals that we have studied, reflections with k + l odd are present, though systematically weak; in C2/m, k + l reflections would be absent, leading to a slightly different structure.Jin et al. did not record the presence of the OH proton, so that database searches based upon ethanolamine do not lead to their structure.

Supramolecular features
In 1, the dimeric Cu2 moieties bridge monoperiodic Cu I CN chains to form diperiodic networks parallel to (102), as shown in Figs. 6 and 7.The Cu I CN zigzag chains extend in the direction of the b axis, out of the plane of Fig. 7.The plane of the Cu I CN chain network makes an angle of 85.5 (1) � with the eight-atom dimeric Cu2 plane.Cu1 atoms from neighboring sheets are within 3.103 (1) A ˚of one another, in roughly axial positions with regard to their trigonal planar coordination.Also, the Cu1 atom lies 0.075 (2) A ˚out of the plane of its three ligands, which brings it closer to the neighboring Cu1 atom in the neighboring sheet.This weak cuprophilic interaction links the sheets into a triperiodic network, and is shown as a dashed line in Fig. 7. Atom H14B is found on the other side of the Cu1 coordination plane, at 3.10 A ˚from Cu1.Otherwise, there are no other short intermolecular contacts of note in this structure.
The structures of 2 and 3 have the same space group and very similar unit-cell dimensions.In both structures, the Cu2 moieties bridge monoperiodic Cu I CN chains to form diperiodic networks parallel to ( 102

Figure 5
The Cu II coordination in 3. The coordination in 2 is the same, with a methyl group added to position C12.

Figure 6
A diperiodic sheet in 1.The colors are as in Fig. 2.

Figure 7
The  3 and 4 for compounds 2 and 3, respectively.No contacts in 1 fit these criteria.For both compounds 2 and 3, the NH 2 group is donor to a screwrelated O atom, and one OH disorder component is donor to a C N group.

Electron spin resonance (ESR)
ESR spectra of the powdered sample materials were recorded at room temperature using a Bruker EMXnano operating at 9.63 GHz (X-band).For all samples reported here, spectral line shape was unaffected by incident microwave power (i.e.no saturable component of the inhomogeneously broadened line), and the general operating parameters were 0.3 mW (incident power); 2 Gauss (field modulation).Spectra were recorded without using the spectrometer's digital filter in 0.5 Gauss steps for a 1000 Gauss field sweep; the receiver signal acquisition time per step corresponded to four time constants.
The structure of 1 features two Cu II ion centers bridged by oxygen, and the resultant ESR spectrum is a broad asymmetric singlet with a crossing point at g = 2.24 [Fig.12(a)].The absence of discernable g-anisotropy in the line shape is indicative of spin exchange, as is expected for a bridged binuclear center.The shape and g-value determined for this isotropic line is nearly identical to that obtained when a crystal of calcium copper acetate, which ordinarily has a well-defined anisotropic spectrum, is decomposed at 750 � C; the resulting ESR spectrum of this mixed metal oxide is a broad singlet with g = 2.22 (Bender, unpublished).
In contrast to the structure of 1, the Cu II centers of both structures 2 and 3 are mononuclear, and the corresponding polycrystalline ESR spectra possess line shape features that are rhombic in character (cf.Hathaway, 1971).The turning points in the spectrum of 2 [Fig.12(b)], corresponding approximately to the diagonal components of the g-tensor, are g 1 = 2.06, g 2 = 2.09, and g 3 = 2.20.The three unequal pairs of coordinate bonds [i.e. bond lengths: 2.60 (Cu-O), 2.056 (Cu-N), and 1.967A ˚(Cu-NC)], lead us to expect an elongated-rhombic octahedral configuration, and our data compare favorably with the literature values (cf.Hathaway, 1971).
The ESR spectrum of polycrystalline 3 [Fig.12(c)] differs from that of structure 2, presumably due to the reduced   Cu-O distance.The shape of the line is rhombic in character, but lacks the added feature [labeled in Fig. 12(b) as 'g 3 '] that is associated with elongation or its counterpart, compression (Hathaway, 1971).The g-values determined from the two turning points are g 0 = 2.06 and g 00 = 2.16; the crossing point occurs at g 000 = 2.10.

Thermogravimetric analysis (TGA)
TGA was carried out with a TA Instruments Q500 device.Samples of each compound weighing 5-20 mg were heated under nitrogen gas at 3 � min À 1 to 600 � C or more.The TGA plots up to 500 � C for the three compounds are shown in Fig. 13.(For 1, analyses were complicated, as most samples were heterogeneous.The plot shown is for crystals handsorted under the microscope.)In all three compounds, there is a sharp drop in mass of 10-15% at 150-200 � C, followed by slower rates of mass loss.The absence of clear steps in the decomposition curves indicates overlapping of incremental decomposition changes.Not shown in the figure are the continued slow mass losses after 500 � C and the occasional subsequent mass increase, presumably due to the formation of an oxide of copper by reaction with residual oxygen in the system.Microscopic examination of residues obtained at the higher temperatures appeared to show mixtures of a black substance and metallic copper.
The decomposition curves for 2 and 3 differ more from each other than might have been expected from the similarity of the structures.In particular, there is a clear change in slope after a mass loss of about 8% for 2, while the corresponding change is more gradual for 3, and also the mass differences in the 350-400 � C range are more than would be anticipated based upon their molecular formulae.
From previous experiments in our laboratory, we expect any CuCN(s) formed to decompose to Cu(s) in the temperature range 400-500 � C. For this reason, each experiment was repeated with termination at 400 � C, in every case leaving black powdery residues.We obtained IR spectra and elemental analyses for each residue at this point, and the results are given in Table 5.All of the 400 � C residues showed IR peaks indicating the presence of CuCN(s) and, in all cases, the residues were richer in both Cu and C than expected for pure CuCN(s), while showing negligible presence of H.Total percentages ranged from 99.3 to 100.4%, precluding the presence of any significant amount of O.We have interpreted the residue composition in terms of mixtures of CuCN(s), C(s), and Cu(s), since it is assumed that any copper(I) acetylide formed would have decomposed by this tempera-   ture, and the percentages calculated from the assumed mixtures are also given in Table 5.The observed %Cu values were calculated by dividing the %Cu expected from the molecular formula by the fraction of mass remaining at 400 � C. This, of course, assumes that the starting material was pure.
To check for HCN(g) emission at the initial stage, we heated 15-20 mg samples of each of the compounds in turn in a test tube, and looked for cloudiness in a drop of AgNO 3 (aq) held over the mouth of the tube.Cloudiness in the AgNO 3 drop was clearly seen at sand-bath temperatures of 190-200 � C for 2 and 205-215 � C for 3, but no evidence for HCN(g) emission was noted for 1, even when the sample was held at or above 250 � C for several minutes.
It is possible that the CN groups bridging Cu1 and Cu2 in 2 and 3 are lost first, combining with the OH protons from the ligands coordinated to Cu2 to form HCN(g), which is not possible in 1.This would leave the Cu I CN chains intact, while the deprotonated ligands would bind to Cu2 more tightly, as they do in 1.Thereafter, for all three compounds, the bound ligands apparently lose their O and N moieties, while leaving at least some of the carbon present in elemental form.Apparently, this begins to happen in 1 before any cyanide loss.

Database survey
The neutral diperiodic mixed-valence CuCN network of the three structures described here has not often been noted.A search in the Cambridge Structural Database (CSD, Version 5.35; Groom et al., 2016) found two similar structures (Kim et al., 2005;Trivedi et al., 2014).The first involves a CuCN network with Cu II atoms coordinated by cyclam units and the second a more complex cyanide/azide network with Cu II coordinated by NH 3 .
A search for Cu compounds containing propanolamine, with or without the OH proton, yielded 48 hits.None of these structures involved Cu bonded to cyanide, however.Of these structures, 44 involved Cu 2 O 2 units with chelating propanolate chelates, as found in 1.Only seven of the structures contained the propanolamine ligand with the OH group intact bonded to Cu, with three of these also containing the Cu 2 O 2 unit.A separate, more general, search of the CSD for dimeric Cu 2 O 2 moieties with Cu bonded to two O and two N atoms yielded 258 hits, so that the four-membered ring of Cu 2 O 2 is not uncommon.
A search for Cu coordinated by 2-aminopropan-1-ol gave seven hits.In all but one case (CSD refcode BOYPIO; Nieuwpoort et al., 1983;Marsh, 2005), the base coordinates with the OH proton intact, as in Podjed et al. (2022).
A search for Cu coordinated by two ethanolamine ligands, with or without the OH proton, produced only three examples, i.e.Tudor et al. (2006), Vasileva et al. (1994), and the work by Jin et al. (2006) cited earlier.

Figure 13
The thermal analysis results for the three compounds.The mass percentage remaining is plotted as a function of temperature.

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.

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
moiety bridged by the O atoms of the bases to form a central fouratom ring.The eight central Cu, N, and O atoms of the dimeric moiety are roughly coplanar, with r.m.s.deviation from the best plane of 0.023 A ˚.The Cu2 atoms are in a distorted square-planar coordination, with the O-Cu-O angle in the four-membered ring equal to 76.91 (8) � and the other bond angles ranging from 91.15 (11) to 96.64 (9) � .The fourth ligand to each Cu2 atom is a CN group which bridges to a Cu1 atom bonded to two other CN groups.The bond angles at Cu1 are 111.63(10), 122.02 (10), and 125.89 (10) � , with the smallest angle trans to the CN bridging to Cu2.

Figure 2
Figure 2The structure of 1, showing the atom numbering and 50% displacement ellipsoids, with H atoms depicted as small spheres.Only the majordisorder component for the chelate ring is shown.Cu atoms are shown in green, N atoms in blue, O atoms in red, and C and H atoms in black.The asymmetric unit is highlighted in bold.

Figure 3
Figure 3 ), as shown in Figs. 8 and 9 for 2, and in Figs. 10 and 11 for 3. Inversion-related Cu1 atoms from neighboring sheets are 2.7400 (15) A ˚apart in 2 and 2.7734 (7) A ˚apart in 3.In each structure, Cu1 atoms are packing in 1.The sheets shown in Fig. 6 are viewed edge on.Putative cuprophilic bonds are shown as dashed double lines.pulled out of the plane of their three coordinated atoms towards the neighboring inversion-related Cu1 atom, 0.302 (4) A ˚in 2 and 0.1774 (16) A ˚in 3.In 2, the C2 atom of cyanide C2N2 is also positioned somewhat close to the Cu1 atom of the neighboring sheet, at 2.502 (6) A ˚, which may indicate that the CN group could be regarded as � 3 -bonded, with very different Cu-C/N distances, rather than the � 2bonding that we have assumed.A similar situation exists for 3, although here the Cu-C/N distance to the neighboring sheet is even longer, at 2.686 (3) A ˚. Putative hydrogen bonds based on D� � �A � 3.30 A ˚and D-H� � �A > 130 � are listed in Tables

Figure 8 A
Figure 8A diperiodic sheet in 2. The colors are as in Fig.2.

Figure 9
Figure 9 Packing in 2. The sheets shown in Fig. 8 are viewed edge on.Putative cuprophilic bonds are shown as dashed double lines and putative � 3 -C-Cu bonds as blue single-dashed lines.

Figure 10 A
Figure 10A diperiodic sheet in 3. The colors are as in Fig.2.

Figure 11
Figure 11The packing in 3. The sheets shown in Fig.10are viewed edge on.Putative cuprophilic bonds are shown as dashed double lines and putative � 3 -C-Cu bonds as blue single-dashed lines.

Table 1
Experimental details.

Table 5
Thermogravimetric analysis data.
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.