Adapting (4,4) Networks through Substituent Effects and Conformationally Flexible 3,2’:6’,3”-Terpyridines

Coordination networks formed between Co(NCS)2 and 4’-substituted-[1,1’-biphenyl]-4-yl-3,2’:6’,3”-terpyridines in which the 4’-group is Me (1), H (2), F (3), Cl (4) or Br (5) are reported. [Co(1)2(NCS)2]n·4.5nCHCl3, [Co(2)2(NCS)2]n·4.3nCHCl3, [Co(3)2(NCS)2]n·4nCHCl3, [Co(4)2(NCS)2]n, and [Co(5)2(NCS)2]n·nCHCl3 are 2D-networks directed by 4-connecting cobalt nodes. Changes in the conformation of the 3,2’:6’,3”-tpy unit coupled with the different peripheral substituents lead to three structure types. In [Co(1)2(NCS)2]n·4.5nCHCl3, [Co(2)2(NCS)2]n·4.3nCHCl3, [Co(3)2(NCS)2]n·4nCHCl3, cone-like arrangements of [1,1’-biphenyl]-4-yl units pack through pyridine…arene π-stacking, whereas Cl…π interactions are dominant in the packing in [Co(4)2(NCS)2]n. The introduction of the Br substituent in ligand 5 switches off both face-to-face π-stacking and halogen…π-interactions, and the packing interactions are more subtly controlled. Assemblies with organic linkers 1–3 are structurally similar and the lattice accommodates CHCl3 molecules in distinct cavities; thermogravimetric analysis confirmed that half the solvent in [Co(3)2(NCS)2]n·4nCHCl3 can be reversibly removed.


Crystal Growth and Bulk Material Characterization
Ligands 1-5 were prepared as previously reported [12]. Reactions between Co(NCS)2 and each ligand were carried out under ambient conditions by layering a methanol solution of Co(NCS)2 over a chloroform solution of 1, 2, 3, 4 or 5. Depending on the ligand, single crystals of X-ray quality were obtained within one day and two weeks (see  (5)2(NCS)2]n . nCHCl3. All the complexes assemble into 2D-networks with the Co atoms acting as 4-connecting nodes, and the nets may be categorized into three structure types as detailed below.
After selection of a crystal for single crystal X-ray diffraction, the remaining crystals were analyzed by solid-state IR spectroscopy and powder X-ray diffraction (PXRD). The IR spectra are shown in Figures S1-S5 in the Supporting Information. The strongest absorption band in each spectrum is assigned to the thiocyanate C≡N stretching mode Although the correspondence between the PXRD patterns for [Co(2)2(NCS)2]n . 4.3nCHCl3 ( Figure S6) is good, it was not possible to generate a fit using the program FULLPROF (see Section 3.7).

Crystal Growth and Bulk Material Characterization
Ligands 1-5 were prepared as previously reported [12]. Reactions between Co(NCS) 2 and each ligand were carried out under ambient conditions by layering a methanol solution of Co(NCS) 2 over a chloroform solution of 1, 2, 3, 4 or 5. Depending on the ligand, single crystals of X-ray quality were obtained within one day and two weeks (see Sections 3.2-3.6), and structures were determined for [Co(1) 2 2 ] n , and [Co(5) 2 (NCS) 2 ] n · nCHCl 3 . All the complexes assemble into 2D-networks with the Co atoms acting as 4connecting nodes, and the nets may be categorized into three structure types as detailed below.
After selection of a crystal for single crystal X-ray diffraction, the remaining crystals were analyzed by solid-state IR spectroscopy and powder X-ray diffraction (PXRD). The IR spectra are shown in Figures S1-S5 in the Supporting Information. The strongest absorption band in each spectrum is assigned to the thiocyanate C≡N stretching mode (2072 cm −1 for [Co(2) 2 (NCS) 2 ] n ·4.3nCHCl 3 , 2070 cm −1 for [Co(1) 2 (NCS) 2 ] n ·4.5nCHCl 3 , 2069 cm −1 for [Co(3) 2 (NCS) 2 ] n ·4nCHCl 3 and [Co(5) 2 (NCS) 2 ] n ·nCHCl 3 , and 2063 cm −1 for [Co(4) 2 (NCS) 2 ] n . Fits between the powder pattern predicted from the single crystal structure and the experimental pattern for [Co(3) 2 (NCS) 2 ] n ·4nCHCl 3 and [Co(5) 2 (NCS) 2 ] n ·nCHCl 3 ( Figure 1) confirmed that the single crystals were representative of the bulk materials. For the remaining compounds, comparisons of the experimental PXRD and those predicted from the single crystal measurements are shown in Figures S6-S8. Although the correspondence between the PXRD patterns for [Co(2) 2 (NCS) 2 ] n ·4.3nCHCl 3 ( Figure S6) is good, it was not possible to generate a fit using the program FULLPROF (see Section 3.7). n . nCHCl3, with fitting to the predicted patterns from the single-crystal structures. The black lines are the best fits from the Rietveld refinements, and green lines show the Bragg peak positions. Each blue plot gives the difference between calculated and experimental points, and the differences in intensities are due to differences in the preferred orientations of the crystallites in the bulk samples.  4.3nCHCl3 were disordered and were modelled over two positions of fractional occupancies 0.5/0.5 (for the ring containing C15) and 0.6/0.4 (for the terminal ring). In each structure, the Co atom is octahedrally sited with a trans-arrangement of thiocyanato ligands, and the 3,2':6',3"tpy domain adopts conformation A (Scheme 2). Bond lengths in the cobalt(II) coordination spheres are given in Table 1. Despite significant variation in the angles between the planes of the aromatic rings in coordinated ligands 1, 2 and 3 (Table 1) (5) 2 (NCS) 2 ] n ·nCHCl 3 , with fitting to the predicted patterns from the single-crystal structures. The black lines are the best fits from the Rietveld refinements, and green lines show the Bragg peak positions. Each blue plot gives the difference between calculated and experimental points, and the differences in intensities are due to differences in the preferred orientations of the crystallites in the bulk samples.
[Co (3) 4.3nCHCl3 ( Figure 2 and Table 1) and the cone assembly is built up around a 2-fold axis (Figure 3c,d).     3 , adjacent sheets pack with the cones of biphenyl substituents stacked inside one another along the crystallographic c-axis (Figure 4a). Face-to-face π-stacking occurs between the pyridine ring containing N2 (see Figures S9-S11 for atom numbers) in one sheet with the peripheral arene ring in the next sheet. In [Co(1) 2 (NCS) 2 ] n ·4.5nCHCl 3 , the πstacking interaction is characterized by a centroid distance of 3.75 Å and an angle between the ring planes of 9.6 • . These parameters are 3.69 Å and 11.9 • in [Co(2) 2 (NCS) 2 ] n ·4.3nCHCl 3 for one modelled site of the disordered phenyl ring. The cone . . . cone π-stacking contacts in [Co(3) 2 (NCS) 2 ] n ·4nCHCl 3 are effective for only one of the two independent ligands with a centroid . . . centroid distance of 3.63 Å and an inter-ring plane angle of 10.9 • . Although the nets and packing of the nets are similar in the compounds with ligands 1, 2 and 3, the number of CHCl 3 molecules per Co differ. Part of the solvent region in each of [Co(1) 2 (NCS) 2 ] n ·4.5nCHCl 3 and [Co(2) 2 (NCS) 2 ] n ·4.3nCHCl 3 was treated with a solvent mask because of disordering, and the small difference between the final formulations was carefully checked. There was insufficient residual electron density in the coordination network with 2 to permit a formula of [Co(2) 2 (NCS) 2 ] n ·4.5nCHCl 3 . Further confirmation came from an analysis of the void spaces (calculated in Mercury [23] using a contact surface map with probe radius = 1.2 Å), which revealed 31.7% in [Co(1) 2 (NCS) 2 ] n and 29.8% in [Co(2) 2 (NCS) 2 ] n . This difference may be attributed to variations in the twist angles between the arene rings of ligands 1 and 2 (Table 1), and the fact that ligand 2 is disordered over two positions. Figure 4b illustrates that the four refined CHCl 3 molecules (colored red and yellow in Figure 4b) per Co occupy two types of cavities. The solvent accessible voids are illustrated in Figure 4c. Figure 4c was generated in Mercury [23] using a solvent-free lattice, while Figure 4d shows the residual voids after taking into account the refined CHCl 3 molecules. These essentially closed cavities account for 4.2% of the solvent accessible void in the lattice and accommodate 0.5CHCl 3 per Co. Figure 5 illustrates the lattice in [Co(3) 2 (NCS) 2 ] n ·4nCHCl 3 . Similarities between Figures 4b and 5a are clear, and the CHCl 3 molecules are again distributed equally between two types of cavities in the lattice. The remaining closed voids that run along the c-axis account for only 1.4% of the total void and host no solvent, in contrast to analogous voids in [Co(1) 2 (NCS) 2 ] n ·4.5nCHCl 3 and [Co (2)    was generated in Mercury [23] using a solvent-free lattice, while Figure 4d shows the residual voids after taking into account the refined CHCl3 molecules. These essentially closed cavities account for 4.2% of the solvent accessible void in the lattice and accommodate 0.5CHCl3 per Co. Figure 5 illustrates the lattice in [Co (3)   As detailed in Figure 1, PXRD confirmed that the single crystal structure of [Co(3)2(NCS)2]n . 4nCHCl3 was representative of the bulk material, and therefore this coordination network was selected for an analysis of solvent removal from, and re-entry into, the crystal lattice. Crystals of [Co(3)2(NCS)2]n . 4nCHCl3 were heated to 80 °C for 30 min. Figure 6 illustrates that loss of CHCl3 was detected with mass peaks at m/z 83.0, 85.0

Thermogravimetric Analysis (TGA) of [Co(3) 2 (NCS) 2 ] n ·4nCHCl 3
As detailed in Figure 1, PXRD confirmed that the single crystal structure of [Co(3) 2 (NCS) 2 ] n · 4nCHCl 3 was representative of the bulk material, and therefore this coordination network was selected for an analysis of solvent removal from, and re-entry into, the crystal lattice. Crystals of [Co(3) 2 (NCS) 2 ] n ·4nCHCl 3 were heated to 80 • C for 30 min. Figure 6 illustrates that loss of CHCl 3 was detected with mass peaks at m/z 83.0, 85.0 and 87.0 corresponding to CHCl 2 + as the dominant fragment [24]. In [Co(3) 2 (NCS) 2 ] n ·4nCHCl 3 , four molecules of CHCl 3 correspond to 32.7% of the molecular weight. The mass loss in the TGA (Figure 6 caption) corresponded to ca. 17%, indicating the loss of two CHCl 3 per Co atom. This is in accord with the structural data, which revealed that the solvent molecules in [Co(3) 2 (NCS) 2 ] n ·4nCHCl 3 are equally distributed within two different types of cavities (Figure 5a). The same part of the lattice viewed down the b-axis; the residual voids (calculated using a contact surface map with probe radius = 1.2 Å, and drawn using Mercury 2020.1 [23]) that are aligned along the c-axis account for 1.4% of the total void.

Thermogravimetric Analysis (TGA) of [Co(3)2(NCS)2]n . 4nCHCl3
As detailed in Figure 1, PXRD confirmed that the single crystal structure of [Co(3)2(NCS)2]n . 4nCHCl3 was representative of the bulk material, and therefore this coordination network was selected for an analysis of solvent removal from, and re-entry into, the crystal lattice. Crystals of [Co(3)2(NCS)2]n . 4nCHCl3 were heated to 80 °C for 30 min. Figure 6 illustrates that loss of CHCl3 was detected with mass peaks at m/z 83.0, 85.0 and 87.0 corresponding to CHCl2 + as the dominant fragment [24]. In [Co(3)2(NCS)2]n . 4nCHCl3, four molecules of CHCl3 correspond to 32.7% of the molecular weight. The mass loss in the TGA (Figure 6 caption) corresponded to ca. 17%, indicating the loss of two CHCl3 per Co atom. This is in accord with the structural data, which revealed that the solvent molecules in [Co(3)2(NCS)2]n . 4nCHCl3 are equally distributed within two different types of cavities (Figure 5a).  After the initial TGA cycle, the sample was cooled to room temperature and was exposed to CHCl 3 vapor for 24 h. TGA analysis was repeated (cycle 2) and loss of CHCl 3 was again detected as shown in Figure S13 in the Supporting Material. A weight loss corresponding to ca. 19% was similar to that in cycle 1. In order to distinguish between the residual lattice solvent after cycle 1, and solvent re-entering the lattice during exposure to CHCl 3 vapor, a third cycle was carried out using CDCl 3 . The same crystalline material was placed in contact to CDCl 3 vapor for 24 h and then analyzed by TGA (Figure 7). Mass peaks at m/z 84.0, 86.0 and 88.0 were detected at ca. 80 • C and were assigned to the CDCl 2 + ion, and the ca. 19% weight loss (see caption to Figure 7) was consistent with the loss of ca. 2 molecules of CDCl 3 per Co atom. In turn, this is consistent with a formulation of [Co(3) 2 (NCS) 2 ] n ·2nCHCl 3 ·2nCDCl 3 after exposure to CDCl 3 vapor and before TGA cycle 3. The process was finally repeated using CH 2 Cl 2 vapor. The product of cycle 3, [Co(3) 2 (NCS) 2 ] n ·2nCHCl 3 , was placed in contact with CH 2 Cl 2 vapor for 24 h. TGA and mass spectrometric analysis of this material showed mass peaks at m/z 49.0, 51.0, 84.0 and 86.0 arising from CH 2 Cl + and CH 2 Cl 2 + at ca. 40 • C ( Figure S14), and the 12% weight loss was consistent with the removal of two molecules of CH 2 Cl 2 per Co atom.
ion, and the ca. 19% weight loss (see caption to Figure 7) was consistent with the loss of ca. 2 molecules of CDCl3 per Co atom. In turn, this is consistent with a formulation of [Co(3)2(NCS)2]n . 2nCHCl3 . 2nCDCl3 after exposure to CDCl3 vapor and before TGA cycle 3. The process was finally repeated using CH2Cl2 vapor. The product of cycle 3, [Co(3)2(NCS)2]n . 2nCHCl3, was placed in contact with CH2Cl2 vapor for 24 h. TGA and mass spectrometric analysis of this material showed mass peaks at m/z 49.0, 51.0, 84.0 and 86.0 arising from CH2Cl + and CH2Cl2 + at ca. 40 °C (Figure S14), and the 12% weight loss was consistent with the removal of two molecules of CH2Cl2 per Co atom.

Crystal Structures of [Co(4)2(NCS)2]n and [Co(5)2(NCS)2]n . nCHCl3
The compounds [Co (4) Table 2. The units shown in Figure 8 propagate into 2-dimensional (4,4) nets, consistent with the coordination networks assembled with ligands 1, 2 and 3,. However, whereas the The results of the TGA experiments illustrate that the coordination network in [Co(3) 2 (NCS) 2 ] n ·4nCHCl 3 is sufficiently robust to allow half of the solvent to be reversibly removed. Inspection of Figure 4b,c indicates that these are most likely the molecules accommodated in the open channels, which run along the c-axis adjacent to the stacked cone-assemblies.

Crystal Structures of [Co(4) 2 (NCS) 2 ] n and [Co(5) 2 (NCS) 2 ] n ·nCHCl 3
The compounds [Co(4) 2 (NCS) 2 ] n and [Co(5) 2 (NCS) 2 ] n ·nCHCl 3 both crystallize in the monoclinic space group P2 1 /n. The structures of the asymmetric units showing the atom numbering are displayed in Figures S15 and S16 in the Supporting Material. In each, the octahedral Co(II) center exhibits a typical trans-arrangement of thiocyanato ligands. In [Co(4) 2 (NCS) 2 ] n , the Co atom lies on an inversion center (Figure 8a), while [Co(5) 2 (NCS) 2 ] n ·nCHCl 3 contains two independent 3,2':6',3"-terpyridine ligands (Figure 8b). Relevant bond lengths and twist angles between arene rings are presented in Table 2. The units shown in Figure 8 propagate into 2-dimensional (4,4) nets, consistent with the coordination networks assembled with ligands 1, 2 and 3,. However, whereas the 3,2':6',3"tpy domains in coordinated 1-3 exhibit conformation A, that in 4 possesses conformation B (Scheme 2 and Figure 8a), while the two independent ligands in [Co(5) 2 (NCS) 2 ] n ·nCHCl 3 adopt conformations A and B, respectively (Figure 8b). The two structures are consequently distinct from one another, and from those with ligands 1-3.   In the (4,4) net in [Co(4) 2 (NCS) 2 ], the Co atoms lie in a plane, and crystallographic symmetry dictates that all rhombi are identical with internal angles of 82.4 • and 97.6 • , and the 4'-chloro-[1,1'-biphenyl]-4-yl units are directed up/up/down/down around each rhombus. This contrasts with the cone-assemblies in the compounds containing 1-3. A consequence of the conformational switch of the 3,2':6',3"-tpy on going from 1-3 to 4 is that the 4'-chloro-[1,1'-biphenyl]-4-yl domains of the ligands lie over the rhombi (Figure 9a,b) rather than projecting directly above the plane. The Cl atoms decorate the outer surfaces of the 2-dimensional sheet in [Co(4) 2 (NCS) 2 ], and each Cl engages in a Cl . . . π interaction with a pyridine ring in the adjacent sheet. The shortest contacts are Cl1 . . . C10 vi = 3.422(2) and Cl1 . . . C11 vi = 3.383(2) Å (symmetry code vi = 1 / 2 +x, 3 / 2 -y, -1 / 2 +z). These distances are within the cut-off value of 3.62 Å applied by Prasanna and Row [25], and the interaction in [Co(4) 2 (NCS) 2 ], which involves the Cl atom directed at a specific arene π-bond, is classified as semi-localized [26,27]. In contrast to the coordination networks incorporating ligands 1, 2, 3 and 5, that with ligand 4 contains no solvent of crystallization; the void (calculated using a contact surface map with probe radius = 1.2 Å) is <2%. We note that the network contains no face-to-face π-stacking interactions.  In contrast to the structures described above, the (4,4) net in [Co(5)2(NCS)2]n . nCHCl3 (Figure 10a) has a corrugated profile (Figure 10b) with each rhombus adopting a folded conformation with an internal dihedral angle of 124.3°. Of the two independent terpyridine ligands, that with conformation A has greater twist angles between the pyridine rings (38.8° and 48.3°) than that with conformation B (29.1° and 21.1°). The orientations of the ligands with respect to the (4,4) net defined by the Co atoms is shown in Figure 10b. The 4'-bromo-[1,1'-biphenyl]-4-yl substituents in adjacent sheets are closely associated (Figure 10c) but there are no π-stacking interactions, nor Br...π interactions. The CHCl3 molecules occupy channels that follow the crystallogrphic b-axis. The difference between the coordination network on going from chloro-substituted ligand 4 to bromo-substituted 5 is striking, but the origins behind the change are unclear. We emphasize that the PXRD (Figure 1b) confirmed that the single crystal structure was In contrast to the structures described above, the (4,4) net in [Co(5) 2 (NCS) 2 ] n ·nCHCl 3 (Figure 10a) has a corrugated profile (Figure 10b) with each rhombus adopting a folded conformation with an internal dihedral angle of 124.3 • . Of the two independent terpyridine ligands, that with conformation A has greater twist angles between the pyridine rings (38.8 • and 48.3 • ) than that with conformation B (29.1 • and 21.1 • ). The orientations of the ligands with respect to the (4,4) net defined by the Co atoms is shown in Figure 10b. The 4'-bromo-[1,1'-biphenyl]-4-yl substituents in adjacent sheets are closely associated (Figure 10c) but there are no π-stacking interactions, nor Br . . . π interactions. The CHCl 3 molecules occupy channels that follow the crystallogrphic b-axis. The difference between the coordination network on going from chloro-substituted ligand 4 to bromo-substituted 5 is striking, but the origins behind the change are unclear. We emphasize that the PXRD (Figure 1b) confirmed that the single crystal structure was representative of the bulk sample.
( Figure 10a) has a corrugated profile (Figure 10b) with each rhombus adopting a folded conformation with an internal dihedral angle of 124.3°. Of the two independent terpyridine ligands, that with conformation A has greater twist angles between the pyridine rings (38.8° and 48.3°) than that with conformation B (29.1° and 21.1°). The orientations of the ligands with respect to the (4,4) net defined by the Co atoms is shown in Figure 10b. The 4'-bromo-[1,1'-biphenyl]-4-yl substituents in adjacent sheets are closely associated (Figure 10c) but there are no π-stacking interactions, nor Br...π interactions. The CHCl3 molecules occupy channels that follow the crystallogrphic b-axis. The difference between the coordination network on going from chloro-substituted ligand 4 to bromo-substituted 5 is striking, but the origins behind the change are unclear. We emphasize that the PXRD (Figure 1b) confirmed that the single crystal structure was representative of the bulk sample.

General
A PerkinElmer UATR Two instrument (Perkin Elmer, 8603 Schwerzenbach, Switzerland) was used to record FT-infrared (IR) spectra. Cobalt(II) thiocyanate was bought from Alfa Aesar and used as received. Compounds 1-5 were prepared and characterized as previously described [12].
All crystal growth experiments were carried out under ambient conditions using identical crystallization tubes (i.d. = 13.6 mm, 24 mL).
Thermogravimetric analysis (TGA) was carried out under nitrogen using a TGA5500 instrument (TA Instruments, New Castle, DE 19720, USA) coupled to a Discovery II MS, Cirrus 3, Mass Spectrometer, DMS (TA Instruments, New Castle, DE 19720, USA). A Barchart scanning method in the mass range 10-125 was applied. For each experiment, the temperature of the TGA instrument was initially stabilized at 30 • C, and then the samples were heated to 80 • C and maintained at this temperature for 30 min. The nature of the solvent lost was identified by mass spectrometry. Afterwards the sample was cooled to room temperature (ca. 22 • C) and placed in contact with vapors of CHCl 3 , CDCl 3 or CH 2 Cl 2 for 24 h. TGA was then repeated using the above procedure.

[Co(1) 2 (NCS) 2 ] n ·4.5nCHCl 3
A solution of Co(NCS) 2 (5.3 mg, 0.030 mmol) in MeOH (5 mL) was layered over a CHCl 3 solution (4 mL) of 1 (11.9 mg, 0.030 mmol). Pink block-like crystals grew within two weeks. A single crystal was selected for X-ray diffraction and the remaining crystals were washed with MeOH and CHCl 3 , dried under vacuum and analyzed by PXRD and FT-IR spectroscopy.

[Co(2) 2 (NCS) 2 ] n ·4.3nCHCl 3
A solution of Co(NCS) 2 (5.3 mg, 0.030 mmol) in MeOH (5 mL) was layered over a CHCl 3 solution (5 mL) of 2 (11.6 mg, 0.030 mmol). Pink plate-like crystals grew within two weeks. A single crystal was selected for X-ray diffraction and the remaining crystals were washed with MeOH and CHCl 3 , dried under vacuum and analyzed by PXRD and FT-IR spectroscopy.

[Co(4) 2 (NCS) 2 ] n
A solution of Co(NCS) 2 (5.3 mg, 0.030 mmol) in MeOH (6 mL) was layered over a CHCl 3 solution (6 mL) of 4 (12.6 mg, 0.030 mmol). Pink block-like crystals grew after 5 days. A single crystal was selected for X-ray diffraction and the remaining crystals were washed with MeOH and CHCl 3 , dried under vacuum and analyzed by PXRD and FT-IR spectroscopy.

[Co(5) 2 (NCS) 2 ] n ·nCHCl 3
A solution of Co(NCS) 2 (5.3 mg, 0.030 mmol) in MeOH (5 mL) was layered over a CHCl 3 solution (4 mL) of 5 (13.9 mg, 0.030 mmol). Light pink plate-like crystals grew within two weeks. A single crystal was selected for X-ray diffraction and the remaining crystals were washed with MeOH and CHCl 3 , dried under vacuum and analyzed by PXRD and FT-IR spectroscopy.
PXRD patterns were collected at room temperature in transmission mode using a Stoe Stadi P diffractometer (STOE & Cie GmbH, 64295 Darmstadt, Germany) with Cu Kα1 radiation (Ge(111) monochromator) and a DECTRIS MYTHEN 1K detector. Wholepattern decomposition (profile matching) analysis [34][35][36] of the diffraction patterns was performed with the package FULLPROF SUITE [36,37] (v. September 2020) using a previously determined instrument resolution function based on a NIST640d standard. The structural models were taken from the single crystal X-ray diffraction refinements. Refined parameters in Rietveld were: scale factor, zero shift, lattice parameters, Co and S atomic σ positions, background points and peaks shapes as a Thompson-Cox-Hastings pseudo-Voigt function. Preferred orientations were included in the analysis as a March-Dollase multi-axial phenomenological model. adjacent sheets in [Co(5) 2 (NCS) 2 ] n ·nCHCl 3 are closely associated, there are no Br . . . π interactions. Face-to-face π-stacking interactions are not observed in either [Co(4) 2 (NCS) 2 ] n or [Co(5) 2 (NCS) 2 ] n ·nCHCl 3 .
This investigation demonstrates that a combination of the conformational flexibility of the 3,2':6',3"-tpy metal-binding domain with a change in the peripheral group (Me, H, F, Cl or Br) leads to significant structural variation while retaining a 2D-network defined by 4-connecting Co nodes . With ligands 1, 2 and 3, pyridine . . . arene π-stacking is dominant, whereas Cl . . . π interactions are important for packing in [Co(4) 2 (NCS) 2 ] n . The introduction of the Br substituent in ligand 5 switches off both face-to-face π-stacking and halogen . . . π-interactions, and the packing interactions are more subtly controlled.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly accessible at present.