Self-Assembly of Cyclohelicate [M3L3] Triangles Over [M4L4] Squares, Despite Near-Linear Bis-terdentate L and Octahedral M

Article abstract of DOI:10.1002/chem.201702153

Self-assembly of 1:1:2 M^II(BF₄)₂ (M=Zn or Fe), pyrazine-2,5-dicarbaldehyde (1) and 2-(2-aminoethyl)pyridine gave trimetallic triangle architectures rather than the anticipated tetrametallic [2×2] squares. Options for the n

Full text of DOI:10.1002/chem.201702153

A Journal of
Accepted Article
Title: Self-Assembly of Cyclohelicate [M3L3] Triangles over [M4L4]
Squares, despite near Linear bis-terdentate L and Octahedral M
Authors: Sally Brooker, Ross Hogue, Sebastien Dhers, Ryan M
Hellyer, Jingwei Luo, Garry S Hannan, David S Larsen, and
Anna L Garden
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.201702153
Link to VoR: http://dx.doi.org/10.1002/chem.201702153
Supported by

10.1002/chem.201702153
Chemistry - A European Journal
FULL PAPER
Self-Assembly of Cyclohelicate [M3L3] Triangles over [M4L4]
Squares, despite near Linear bis-terdentate L and Octahedral M
Ross W. Hogue,[a] Sebastien Dhers,[a] Ryan M. Hellyer,[a] Jingwei Luo,[b], [c] Garry S. Hanan,[c] David S.
Larsen,[a] Anna L. Garden*[a] and Sally Brooker*[a]
Abstract: Self-assembly of 1:1:2 MII(BF4)2 (M = Zn or Fe), pyrazine-
2,5-dicarbaldehyde (1) and 2-(2-aminoethyl)pyridine gave trimetallic
triangle architectures rather than the anticipated tetrametallic [2x2]
squares. Options for the non-trivial synthesis of 1 are considered, and
synthetic details provided for both preferred routes. Rare cyclohelicate
triangle architectures are observed for the pair of structurally
characterized yellow-brown [Zn3L3](BF4)6 and dark green
[Fe3L3](BF4)6 complexes of the neutral bis-terdentate Schiff base L. In
order to form these pyrazine-edged triangles, the octahedral metal
ions – with all 6 N-donors provided by the terdentate binding pockets
of two L – are located 0.4-0.5 Å out of the plane of the bridging
pyrazines, towards the centre of the triangle. Density functional theory
calculations confirm that simple particle counting entropic arguments,
which predict triangles over squares, are correct here, with the
triangles shown to be energetically favored over the corresponding
squares. However, importantly, DFT analysis of these and related
triangle vs square systems also show that vibrational contributions to
entropy dominate and may significantly influence the preferred
architecture, such that simple particle counting cannot in general be
reliably employed to predict the observed architecture.
Here we report on the self-assembly of L, the di-imine analogue
of the diamide H2L2,5-Et (Scheme 1) with octahedral metal ions
(Scheme 2). Whilst L is neutral and expected to be far softer than
the deprotonated diamido analogues, it was expected to also form
[2x2] square cyclohelicates with octahedral metal ions.
Surprisingly, this was not the case: instead triangular
cyclohelicates are obtained (Scheme 2), contrary to expectations
based on the directional bonding approach[1c] and the molecular
library model.[1d]
Scheme 1. Self-assembly of MII(BF4)2 and pyrazine-based ditopic bis-terdentate
diamide ligands H2L2,5-Et, H2L2,3-Me and H2L2,3-Et results in [2x2] grids (2,3-
isomers) or squares (2,5-isomer).
Introduction
Coordination-driven self-assembly has been used by many
chemical disciplines to target specific molecular architectures.[1]
The outcomes of self-assembly are, to a large degree, governed
by the directionalities of carefully designed ligands and the
predictable coordination geometries of metal centers, and by
entropy, which is simply expected to favor smaller polygons. Of
particular interest to this research group are tetrametallic [2x2]
grids and squares as they can provide an array of addressable
magnetically interesting and/or redox-active metal centers.[2]
Previously we have reported a family of bis-terdentate pyrazine-
diamide ligands which have generated [2x2] grids and squares
upon self-assembly with octahedral metal ions (Scheme 1).[3]
Scheme 2. Self-assembly of a 1:1:2 mixture of metal(II) tetrafluoroborate,
pyrazine-2,5-dicarbaldehyde
[MII3L3](BF4)6 triangles.
and
2-(2-aminoethyl)pyridine
results
in
Most reported triangle metal complexes fall into one of two
categories[1h, 1i, 1k] (Figure 1): (a) those with linear metal edges and
ditopic angular linking ligand corners (Figure 1, a), and (b) those
with ditopic linear edge ligands and metal ion corners ‘capped’
with ancillary ligands (Figure 1, b). In contrast, a rare category of
molecular triangle is that in which the three bridging ligands
provide a complete donor set to the metal ions (Figure 1, c).
Within this last class, in which all donors are provided by just the
three bridging ligands, there are only two examples containing
octahedral metal ions.[4] The linker ligands of the triangle typically
bind one metal ion from above and one metal ion from below the
M3 plane, thus forming a chiral trinuclear cyclohelicate (Figure 1,
c), usually obtained as a racemic mixture. A triangular complex
with all donor atoms supplied by the three linker ligands is the
most appealing design to generate stable complexes for magnetic
and redox applications.
[a]
Ross W. Hogue, Dr Sebastien Dhers, Ryan M. Hellyer, Prof David
S. Larsen, Dr Anna L. Garden and Prof Sally Brooker
Department of Chemistry and MacDiarmid Institute for Advanced
Materials and Nanotechnology, University of Otago
PO Box 56, Dunedin 9054, New Zealand
E-mail: sbrooker@chemistry.otago.ac.nz
[b]
[c]
Jingwei Luo
Department of Chemistry, University of Victoria, Victoria, Canada
Prof Garry S. Hanan, Département de Chimie, Université de
Montréal, Quebec, Canada
Supporting information for this article is given via a link at the end of
the document.
This article is protected by copyright. All rights reserved.

10.1002/chem.201702153
Chemistry - A European Journal
FULL PAPER
Figure 1. (a): Triangles with angular corner ligands (black). (b): Triangles with
linear linker ligands (red) and metal ions which also bind ‘capping’ ancillary
ligands (pink). (c): Trinuclear cyclohelicates with all donors from 3 ligands.
Results and Discussion
Pyrazine-2,5-dicarbaldehyde (1) is of significant interest as a
synthetic building block. Recently it has been made via ozonolysis
of 2,5-distyrylpyrazine and used as the basis of ligands designed
for supramolecular chemistry.[5] The synthesis of 1 is non-trivial
(Scheme 3, see SI for discussion of the routes),[5a, 6] but is key to
the triangle complexes reported herein. Hence detailed, reliable
protocols for preparing 1 via ozonolysis and also via pyrazine-2,5-
diester, were established (see SI for experimental details). The
ester route has the advantage of being safer to scale up, but has
four steps with an overall yield of 20% (Scheme 3, blue). The
ozonolysis route has its associated hazards, but only requires two
steps and proceeds in a better overall yield of 31% (Scheme 3,
red).
Scheme 3. The available synthetic routes to pyrazine-2,5-dicarbaldehyde.
6b]
Conditions (yield): (i) V2O5, MoO3, AgO2 catalyst, O2 (10%).[6a,
(ii)
benzaldehyde, benzoic anhydride, reflux (79%).[6c, 6d] (iii) a: O3, MeOH, -78 °C,
b: Na2S2O5, continuous extraction (38%).[5a, 6e] (iv) OsO4, NaIO4, THF/H2O (2:1),
Ar (40±5%).[6f] (v) m-CPBA, EtOAc (90%).[6g] (vi) Ac2O, 158 °C (20%).[6g] (vii)
NaOMe, dry MeOH (94%).[6g] (viii) MnO2, CHCl3, reflux (84%).[6g] (ix) SeO2,
pyridine/water (10:1), reflux (61%). (x) a: NaBH4, MeOH, b: H2O, continuous
extraction (40%).[6h]
Both [Zn3L3](BF4)6 and [Fe3L3](BF4)6 are made by one pot
reactions of a 1:2:1 ratio of 1, 2-(2-aminoethyl)pyridine and
MII(BF4)2·xH2O (M = Zn or Fe), in MeCN. Subsequent vapor
diffusion of Et2O into these solutions over a few days gives pale-
tan block-like single crystals for Zn(II), and green plate-like single
crystals for Fe(II). Single crystal X-ray diffraction studies on both
complexes unexpectedly revealed the structures to be rare
examples of triangular cyclohelicates (Figure 2), rather than the
anticipated [2x2] squares. Cryospray mass spectrometry was
X-ray crystal structure analysis showed that coordination of L
to Fe(II) results in a trimetallic triangular complex, [Fe3L3](BF4)6.
Here the asymmetric unit contains one and a half FeIIL units, with
the other half of the triangle formed by 2-fold rotation (Figure 2).
All of the Fe-N bonds are ca. 2.0 Å and Σ = 45.4° (Table 1),
characteristic of low spin (LS) Fe(II). As with the Zn(II) complex,
[Fe3L3](BF4)6 is a chiral cyclohelicate in which the three Fe(II) ions
within one complex cation have the same configuration, and a
50:50 mixture of (ΔΔΔ) and (ΛΛΛ) enantiomers are observed in
the crystal lattice due to inversion symmetry (C2/m).
Ligands providing a bridging pyrazine moiety provide these two
N donor atoms at 180° (Scheme 2), so, according to the
directional bonding approach[1c] and the molecular library
model,[1d] assembly with octahedral metal ions is expected to
produce molecular squares. The formation of molecular triangles
instead must be the result of either (a) the pyrazine directional
angle deviating from 180° (edges) and/or (b) the Npz-M-Npz corner
angles deviating from 90° (square) towards 60° (triangle). Indeed
the Npz-M-Npz corners are less than 90° for both compounds
{[Zn3L3], 79.6°(1); [Fe3L3], av. 85.1°(1)}, but these modest
deviations clearly cannot account for how these triangles have
been obtained. Rather, it is the deviation of the binding of pyrazine
edges away from 180° directionality that is the key here.
consistent with MII L3 triangles, with peaks corresponding to
3
{[M3L3](BF4)5}+ and {[M3L3](BF4)4}2+ observed for both compounds
(Figures S4-S6 and S10-S11): in no case were MII L4 or higher
4
order species detected (Figures S7-S9 and S12-S13). Both
compounds have been further characterized by elemental
analysis, IR, UV-vis, and 1H and 13C NMR spectroscopies (ESI).
The asymmetric unit of [Zn3L3](BF4)6 contains one ZnIIL
component, which represents one third of the triangle, with the
other two thirds generated by a 3-fold rotation axis through the
center of the triangle, at right angles to the Zn3 plane. All donors
to each octahedral Zn(II) are provided by two N3 pockets from two
different L ligands. The cyclohelical architecture imparts inherent
chirality to the [Zn3L3](BF4)6 triangle. All Zn(II) ions are chiral-at-
metal with all ions within one complex cation having the same
chirality. The complex crystallized in the R-3 space group, so is
racemic, with inversion generating a 50:50 mixture of (ΔΔΔ) and
(ΛΛΛ) enantiomers.
This article is protected by copyright. All rights reserved.

10.1002/chem.201702153
Chemistry - A European Journal
FULL PAPER
Table 1. Comparison of selected distances (Å) and angles (°) for compounds
[Zn3L3](BF4)6, [Fe3L3](BF4)6, and [CoIII4(L2,5-Et)4](BF4)4 (See Table S2 for more
details).
[Zn3L3] (BF4)6
[Fe3L3](BF4)6
[CoIII4(L2,5-Et)4]
[a]
(BF4)4
Reference
This work
This work
[3b]
cis-N-M-N
range; [av.]
73.70(10)-
105.26(10);
[89.70]
81.24(9)-95.32(9);
[89.99]/
80.82(9)-96.24(9);
[90.03]
81.94(14)-
94.38(15); [89.98]
Npz-M-Npz
79.6(1)
-0.41
84.33(13)/85.87(9)
-0.46
87.52(14)
0.07
Av. M out of
plane pz[b]
[a] Fe1/Fe2. [b] Negative value indicates M is on the inside face of the plane,
closer to center of triangle/square.
In each case a molecular square was anticipated by the authors,
however, much like the present cyclohelicates, Npz-M-Npz angles
closer to 80° (80.47°-85.98°, av. 82.05°) in combination with the
metal ions being located out of the pyrazine planes (0.23-0.66 Å,
av. 0.44 Å), facilitated the formation of triangles instead. All four
of these literature examples are ‘capped’ triangles, and the bulky
‘caps’ had been postulated to predispose the metal corners
towards angles more suited to triangles over squares.[1h]
In contrast, here we report the first examples of structurally
characterized pyrazine-edged triangular cyclohelicates (Figure 1
c), in which all six donor atoms to each metal ion are provided by
the di-imine bis-terdentate pyrazine-edge ligands L. This triangle
architecture contrasts with the molecular squares that the closely
analogous diamide bis-terdentate ligands, H2L2,5-Et [3b] and H2L2,3-
Figure 2. Perspective view of the hexacations of [Zn3L3](BF4)6 (top, including an
expansion of one corner, displaying Npz-M-Npz and pyrazine-pyrazine mean
plane angles, and the M out of plane pyrazine distances) and [Fe3L3](BF4)6
(bottom). Zn (light indigo), Fe (red), N (blue) and C (grey). Hydrogen atoms
omitted for clarity. Symmetry operations (top) 3-fold rotation axis running
through the center of the triangle perpendicular to the M3 plane, and (bottom) 2-
fold rotation axis running through Fe1 only in the M3 plane.
Me or Et [3a, 3c, 11]
(Scheme 1), generated upon complexation with
octahedral metal ions. The only pyrazine-edged triangular
cyclohelicate, [Zn3(bbppz)3](PF6)6 (bbppz is 2,5-bis([2,2’]bipyridin-
6-yl)pyrazine), reported previously was only observed in solution
as the minor product in equilibrium with the (structurally
characterized) ZnII4 square.[4a]
To gain further insight into the subtle factors contributing to
whether a triangle or square architecture is realized, density
functional theory (DFT) calculations on the [Zn3L3]6+ and [Fe3L3]6+
triangles, and the corresponding (but not observed) [Zn4L4]8+ and
[Fe4L4]8+ squares were performed with a continuum acetonitrile
solvent model. The GGA functional BP86,[12] which is known to
perform well with transition metal complexes,[13] was used with a
def2-SVP basis set.[14]
The total electronic energies, E, are reported as ‘per ML unit’,
i.e. the total energies of the triangles and squares are divided by
three and four respectively to account for the stoichiometry.
Optimized structures of [Zn3L3]6+ and [Fe3L3]6+ have very similar
geometries to the crystallographic data (Tables S5-S6). Despite
showing a significant deviation from the ideal 180° directional
bonding in the bridging pyrazine unit, the optimized triangles both
have lower E (per ML unit) than the corresponding [Zn4L4]8+ and
[Fe4L4]8+ squares by 3.42 and 4.05 kJ mol-1 respectively (Tables
2 and S4). The results of the DFT calculations are therefore
consistent with the experimentally observed triangle architectures.
In all cases the metal center sits significantly out of the plane of
the attached pyrazine rings towards the center of the triangle
(Figure 2), on average by 0.41 Å for [Zn3L3] and 0.46 Å for [Fe3L3].
In contrast the square [CoIII4(L2,5-Et)4](BF4)4 obtained when the
analogous pyrazine di-amide ligand was employed (Scheme
1),[3b] has Npz-M-Npz angles (av. 87.52°) closer to 90° and metal
ions only 0.07 Å out the plane of the two coordinated pyrazine
rings (Table 1). In summary, these results show that the 2,5-
pyrazine bridging moiety in these imine and amide bis-terdentate
ligands is not geometrically predisposed towards only one
architecture, as either triangles or squares can be achieved.
A survey of the literature [CSD search for triangles comprising
three metal ions bridged by any bonds to three pyrazine rings
(Figure S26); CSD version 5.38 updates (November 2016)]
revealed only four other structurally characterized pyrazine-edge
triangle complexes: two with square planar metal ions,
[8]
[Rh3(PPh3)6(μ-pz)3](ClO4)3[7] and [{Pt(PMe3)2(μ-pz)}3] (CF3SO3)6
and two with octahedral metal ions, [{trans,cis-RuCl2(dmso-S)2(μ-
pz)}3][9] and [{ZnCl2(PPh3)2(μ-bppz)}3][10] (Table S3).
This article is protected by copyright. All rights reserved.

10.1002/chem.201702153
Chemistry - A European Journal
FULL PAPER
The same DFT analysis was applied to the related pyrazine-
preference to such an extent that particle counting arguments are
not sufficient to explain the observed product.
edged squares, [Co4(L2,5-Et)4]4+,[3b] [Zn4(bbppz)4]8+,[4a] and
[FeII (bbppz)4]8+,[15] as well as the corresponding (but not
4
structurally characterized and/or observed) triangles, and the total
electronic energies were also consistent with the reported
architecture outcomes (Table S4).
Conclusions
Rather than the square [2x2] cyclohelicates that were anticipated,
and indeed seen for the analogous di-amide ligand, the first
examples of pyrazine-based triangle cyclohelicates were
unexpectedly obtained on self-assembly of a new bis-terdentate
pyrazine di-imine ligand with octahedral metal ions. Careful
analysis of the structures of these triangles reveals that the
pyrazine bridging moiety of this bis-terdentate ligand is quite
flexible, accommodating the binding of octahedral metal ions
either in (square, related ligands), or well out (triangles, this study),
of the plane of the pyrazine ring.
DFT analysis of both the observed triangles and the
corresponding squares, as well as of related reported pyrazine
cyclohelical triangles/squares, correctly predicts the observed
architectures. However, these DFT calculations also show that
vibrational contributions, which are difficult to model, dominate the
overall entropy, so, in general, simple counting arguments (which
favor more smaller assemblies) cannot be reliably employed in
the prediction of architecture outcomes.
Table 2. Total electronic energy, ΔH, and ΔS relative to the square complex
for each M+L combination. Rotational, translational, and vibrational
contributions to overall entropy are detailed. All are per ML (kJmol-1)
[Zn3L3]6+
[Zn4L4]8+
[Fe3L3]6+
[Fe4L4]8+
Erel
Hrel
TSrel
TSrot
TStrans
TSvib
-3.42
-4.37
≈8
16.50
19.63
≈100
0 (by def.)
0 (by def.)
0 (by def.)
12.74
14.99
≈100
-4.05
-7.53
≈8
16.34
19.60
≈100
0 (by def.)
0 (by def.)
0 (by def.)
12.61
14.96
≈100
To probe the competing energetic influences, the relative
enthalpic and entropic stabilities were evaluated using the DFT-
calculated data and ideal gas statistical mechanics (Table 2 and
Table S4). In all five cases, the enthalpic product was the same
as that predicted by consideration of electronic energy E alone.
Despite the subcomponents deviating from their ideal bonding
geometries, the triangles [Zn3L3]6+ and [Fe3L3]6+ are calculated to
have lower enthalpy than the corresponding squares.
Explanations for this are difficult as enthalpic contributions are
often complex and can involve geometric strain,[16] internal void
space,[4a] and hydrophobic effects upon changing surface area to
volume ratios.[17] For these reasons we refrain from using ΔH to
calculate Gibb’s energies.
Subtle variations of these bis-terdentate donor sets are
currently underway as fine-tuning of both the field strength and
geometric control will enable further probing of this assembly
algorithm and the formation of magnetically interesting, rather
than diamagnetic, metal complexes.
Experimental Section
For all five cyclohelicates the triangle was found to be the
entropically-favored product (Table 2 and Table S4). This is no
surprise, as it is generally accepted that dynamic mixtures of
molecular polygons entropically favor smaller assemblies due to
the greater number of discrete molecules formed (4 triangles vs 3
squares from a mixture of 12M + 12L).[1d, 1h, 1j, 1l] However, we note
that the calculation of entropy for these complexes is difficult as
they have very many low-frequency modes which are not well-
described within standard thermochemical models such as those
used here. Despite these inherent uncertainties, it is clear that the
vibrational contributions, TSvib≈100 kJ mol-1 (Table 2), are much
larger than the rotational and translational contributions, TSrot and
TStrans, so are the dominant contributor to the overall entropy of
these cyclohelicates. So, whilst DFT indicates that in all cases the
triangles are entropically favored over the squares, in fact this
may or may not be the case.
This observation may provide some insight into some
previously puzzling results, where larger cyclohelicate assemblies
were reported to be entropically favored,[4a, 18] a result described
by those authors as counterintuitive, but one which might be
explained by low frequency vibrational modes tipping the overall
entropy in favor of the larger molecular assemblies, i.e. the
dominance of the vibrational contribution to the overall entropy,
as shown herein, indicates that the different vibrational properties
of the different sized cyclohelicates may influence the entropic
General experimental details
The solvents used were reagent grade and were used as received from
commercial suppliers. NaBH4 (97%) and SeO2 (97%) were purchased
from Ajax Finechem. MnO2 (85%, particle size < 5µm) was purchased from
Sigma-Aldrich. All other chemicals were reagent grade and were used as
received from commercial suppliers. All reactions were conducted in air
unless otherwise stated.
IR spectra were recorded as solids on a Bruker Alpha FT-ATR IR
spectrometer with a diamond anvil Alpha-P module between 400 and 4000
1
cm-1. The H and 13C NMR spectra at 25 °C were recorded on a Varian
400 MHz spectrometer. Standard microanalysis was carried out by the
Campbell Microanalytical Laboratory at the University of Otago. Mass
spectrometry was recorded by Professor Garry Hanan at the University of
Montreal, Canada using a MicrOTOF II mass spectrometer from Bruker in
positive ion mode using pneumatically assisted CryoSpray source:
Capillary voltage, 4500 V; End Plate Offset -500V; Capillary Exit voltage,
20 V; dry gas temperature, -20 to 0°C; Nebulizer gas temperature, -60 to -
20°C; Nebulizer gas flow, 0.150 MPa; dry gas flow, 1.5 L/min.
X-ray crystallographic data were collected on Bruker Kappa Apex II
CCD area detector diffractometer (University of Otago). In both cases
graphite-monochromated Mo KR radiation (λ = 0.71073 Å) was used. The
data were collected at 85 K. All data sets were corrected for absorption
using SADABS.[19] Structures were solved using SUPERFLIP and refined
against all F2 data (except for the Zn structure in which OMIT 0 1 2 was
used due to this reflection being partially obscured by the beamstop) using
This article is protected by copyright. All rights reserved.

10.1002/chem.201702153
Chemistry - A European Journal
FULL PAPER
SHELX-2014.[20] All non-H atoms were refined with anisotropic thermal
parameters. All H atoms were inserted at calculated positions and rode on
the C atoms to which they were attached with thermal parameters equal
to 1.2 times that of the parent atom. Owing to the presence of badly
disordered lattice solvent molecules and significant unassignable Q peaks,
the structures were treated with SQUEEZE[21] (see SI for details). High-
resolution pictures were prepared using Mercury[22] and POVray[23]
software. CCDC 1530644-1530645.
dimethylpyrazine-2,5-dicarboxylate. Found: C 49.24, H 3.87, N 14.25%.
Calculated for C8H8N2O4 (196.16 gmol-1): C 48.98, H 4.11, N 14.28%. 1H
NMR (300 MHz, CD3Cl3): δ(ppm) = 9.39 (s, 2H, PzH), 4.07 (s, 6H, CH3).
13C NMR (100 MHz, CD3Cl3): δ(ppm) = 163.6 (PzCO), 145.6 (PzH), 145.3
(PzCO), 53.6 (OCH3).
2,5-Bis(hydroxymethyl)pyrazine.[6h] Dimethylpyrazine-2,5-dicarboxylate
(1.0 g, 5.1 mmol) was dissolved in methanol (40 mL). The solution was
chilled in an ice bath then NaBH4 (1.54 g, 40.8 mmol) was added slowly
with stirring over 15 minutes. The ice bath was removed after around an
hour, when all the ice was melted. After stirring overnight, 5 mL of water
was added and the solution left to stir for 10 minutes then taken to dryness
under reduced pressure. The residue was then dissolved in 2.5 mL of
water and continuously extracted with chloroform for 2 days. 2,5-
Bis(hydroxymethyl)pyrazine was collected as a pale yellow solid (280 mg,
40%). Found: C 50.26, H 5.53, N 19.11%. Calculated for C6H8N2O2∙H2O
(143.36 gmol-1): C 50.13, H 5.89, N 19.41%. 1H NMR (500 MHz, CD3OD):
δ(ppm) = 8.65 (s, 2H, PzH), 4.76 (s, 4H, CH2), 4.85 (s, 2H, OH). 13C NMR
(500 MHz, CD3OD): δ(ppm) = 156.1 (PzCH2), 142.8 (PzH), 64.0 (CH2).
Synthesis details
2,5-Distyrylpyrazine.[6d] A mixture of 2,5-dimethylpyrazine (6.27 g, 58.0
mmol), benzaldehyde (20.37 g, 192.0 mmol) and benzoic anhydride (22.19
g, 97.9 mmol) was refluxed overnight. After cooling to room temperature,
it was filtered and the black/yellow filter cake washed with methanol (3 x
50 mL) then diethyl ether (1 x 25 mL) to give the product as a yellow
powder (8.95 g, 54%, which should be stored in the dark). Found: C 84.50,
H 5.78, N 10.04%. Calculated for C20H16N2 (284.36 gmol-1): C 84.48, H
5.67, N 9.85%. 1H NMR (400 MHz, CDCl3): δ(ppm) = 8.60 (s, 2H, HA), 7.74
(d, J = 7.7 Hz, 1H, HB), 7.61 (d, J = 7.3 Hz, 2H, HD), 7.40 (t, J = 7.3 Hz, 2H,
HE), 7.34 (d, J = 7.3 Hz, 1H, HF), 7.19 (d, J = 7.7 Hz, 1H, HC).
Pyrazine-2,5-dicarbaldehyde (1). To
a yellow solution of 2,5-
bis(hydroxymethyl)pyrazine (280 mg, 2.0 mmol) in dry chloroform (50 mL)
was added manganese dioxide (1.7 g, 19.5 mmol). The resulting
suspension was stirred and refluxed for 3 hours. The resulting mixture was
filtered (celite on sinter) and the solid washed with chloroform (2 x 50 mL).
The combined filtrate was taken to dryness in vacuo yielding 275 mg of the
product as a yellow solid (1.69 mmol, 84%). Found: C 52.21, H 3.41, N
19.76%. Calculated for C6H4N2O2∙1.5H2O (163.13 gmol-1): C 51.92, H 3.12,
N 20.18%. 1H NMR (500 MHz, CDCl3): δ(ppm) = 10.232, s (CHO), 9.302,
s (PzH). 13C NMR (125 MHz, CDCl3): δ(ppm) = 191.63 CHO, 148.60 Pz,
143.24 PzH.
Pyrazine-2,5-dicarbaldehyde.[6e] Caution: ozonolysis is a reaction that
proceeds through the formation of an explosive intermediate, an organic
ozonide, making it an extremely dangerous reaction to scale up.
Consequently, the reaction should be done on a small scale (never use
more than 2 g of starting material), a blast shield should be used during
the reaction and extreme precaution should be taken. Behind a blast shield,
a pale yellow suspension of 2,5-distyrylpyrazine (1.90 g, 6.68 mmol) in
methanol (300 mL) was chilled in a dry ice/acetone bath and then treated
with ozone for 10 minutes, turning the suspension blue. Nitrogen was then
bubbled through for a period of 30 minutes to remove the excess ozone,
giving back the original yellow colour. Next, 2 mL of a freshly prepared
aqueous solution of sodium metabisulfite (23% by mass) was added
dropwise and the reaction mixture allowed to warm up to room temperature
whilst still under nitrogen. The mixture was filtered and the filtrate was
taken to dryness under reduced pressure. Saturated NaCl solution (10 mL)
was added and used to rinse the resulting yellow solid into a separating
funnel. The aqueous layer was washed with diethyl ether (3 x 50 mL) to
remove the benzaldehyde byproduct, and then continuously extracted with
chloroform for 1 day. The chloroform fraction was taken to dryness to give
the product as a yellow solid (350 mg, 2.52 mmol, 38%). Found: C 53.20,
H 3.26, N 20.33%. Calculated for C6H4N2O2 (139.11 gmol-1): C 52.95, H
2.96, N 20.58%. 1H NMR (400 MHz, CDCl3): δ(ppm) = 10.22 (s, 2H, CHO),
9.29 (s, 2H, PzH).
[Zn3L3](BF4)6. To a stirred solution of pyrazine-2,5-dicarbaldehyde (10.0
mg, 73.5 μmol) in CH3CN (10 mL) was added 2-(2-aminoethyl)pyridine
(17.2 μL, 147 μmol) in CH3CN (155 μL). After 20 min, a solution of
Zn(BF4)2·H2O (17.6 mg, 73.5 μmol) in CH3CN (1 mL) was added, giving a
pale tan-brown coloured solution which darkened on stirring for 30 min.
Vapour diffusion of Et2O into this solution over 4 days gave brown crystals
in a yellow coloured solution. The solvent was decanted off and the solids
dried in vacuo affording [Zn3L3](BF4)6 as a tan-brown coloured solid (yield
36 mg, 84.0%). 1H NMR (CD3NO2, 500 MHz): δ = 3.16 (m, 2H, CH2), 3.36
(d, 2H, CH2, 3J = 16 Hz), 4.49 (m, 2H, CH2), 4.65 (d, 2H, CH2, 3J = 18 Hz),
7.35 (m, 2H, PyH), 7.56 (d, 2H, PyH, 3J = 8 Hz), 7.99 (m, 2H, PyH), 8.08
(d, 2H, PyH, 3J = 5.5 Hz), 8.34 (s, 2H, PzH) and 8.80 ppm (d, 2H, ImH, 3J
= 2 Hz). 13C NMR (CD3NO2, 133 MHz): δ = 34.77 (CCH2), 57.521 (CCH2),
125.373 (CPy), 127.911 (CPy), 142.31 (CPy), 145.75 (CPz), 146.937 (CPz),
149.44 (CPy), 161.324 (CPy), 164.165 ppm (CIm). Elemental analysis.
Found: C, 41.87; H, 3.70; N, 14.67%. Calculated for
C60H60N18Zn3B6F24·CH3CN: C, 41.57; H, 3.54; N, 14.86%. IR υCN (ATR,
cm-1): 1645 (m) 1609 (m) 1572 (w) 1488 (m) 1445 (m) 1404 (m) 1309 (m)
1188 (m) 1019 (s) 879 (m) 841 (m) 784 (w) 767 (m). UV/Vis (CH3CN): λmax
(ε) = 240 (45389), 324 (41259 L mol-1cm-1). Cryospray-MS (+): m/z =
Dimethylpyrazine-2,5-dicarboxylate. Caution: SeO2 is toxic and should
be handled with care. If spilled, use dry sand and collect the spilled
material to put it in a sealed container, and then dispose of it in a secured
sanitary landfill. To a light yellow solution of 2,5-dimethylpyrazine (8.11 g,
75.0 mmol) in pyridine (150 mL) was added solid selenium dioxide (37.5 g,
338 mmol) and water (15 mL) and the resulting suspension refluxed for 18
hours. The resulting dark red-brown mixture was evaporated to dryness
under reduced pressure. Water (250 mL) was added and the solid
elemental selenium was filtered off. The red filtrate was evaporated to
dryness. The resulting dark red-brown solid was taken up in methanol (150
mL), thionyl chloride (4.6 mL) added and the suspension refluxed for 8
hours. The resulting suspension was hot filtered and the solid residue
washed with dichloromethane (5 x 20 mL). The combined organic phases
were reduced to 100 mL under reduced pressure to give pale yellow
orange feathery crystals. The solid was filtered off and washed with ice-
cold methanol (30 mL) to give 8.98 g (45.8 mmol, 61%) of analytically pure
1662.3074
{[Zn3L3](BF4)5}+
(calcd
=
1662.3276),
788.1550
{[Zn3L3](BF4)4}2+ (calcd = 778.1610), 496.4363 {[Zn3L3](BF4)3}3+ (calcd =
496.4393), 350.5771 {[Zn3L3](BF4)2}4+ (calcd = 350.5785).
[Fe3L3](BF4)6. To a stirred solution of pyrazine-2,5-dicarbaldehyde (10.0
mg, 73.5 μmol) in CH3CN (10 mL) was added 2-(2-aminoethyl)pyridine
(17.2 μL, 147 μmol) in CH3CN (155 μL). After 30 min, a solution of
Fe(BF4)2·6H2O (49.6 mg, 147 μmol) in CH3CN (1 mL) was added, giving
an intense dark green coloured solution which was left to stir overnight
before concentration down to half the volume in vacuo. Vapour diffusion of
Et2O into this solution over 3 days gave green single plate-like crystals in
a colourless solution. The solvent was decanted off and the solids dried in
vacuo giving [Fe3L3](BF4)6 as a dark green coloured solid (yield 22 mg,
This article is protected by copyright. All rights reserved.

10.1002/chem.201702153
Chemistry - A European Journal
FULL PAPER
1
52 %). H NMR (CD3CN, 400 MHz): δ = 2.76 (m, 2H, CH2), 3.38 (d, 2H,
supporting this research. We also thank the Natural Sciences and
Engineering Research Council (NSERC) for financial support of
the visit of JL (Victoria) to GSH (Montreal), during which the MS
data were obtained, and Prof Scott McIndoe (Victoria) and Dr
Joseph Lane (Waikato) for helpful advice. The authors wish to
acknowledge NeSI high-performance computing facilities. NZ's
national facilities are provided by the NZ eScience Infrastructure
and funded jointly by NeSI's collaborator institutions and through
the Ministry of Business, Innovation & Employment's Research
Infrastructure programme. URL https://www.nesi.org.nz.
CH2, 3J = 16.0 Hz), 4.32 (m, 2H, CH2), 5.18 (d, 2H, CH2, J = 16.0 Hz),
3
7.11 (m, 2H, PyH), 7.42 (d, 2H, PyH, 3J = 7.3 Hz), 7.56 (d, 2H, PzH, 3J =
5.0 Hz), 7.83 (m, 4H, PyH) and 9.76 ppm (s, 2H, ImH). 13C NMR (CD3CN,
133 MHz): δ = 34.51 (CCH2), 57.18 (CCH2), 123.53 (CPy), 127.79 (CPy),
140.16 (CPy), 151.03 (CPy), 153.37 (CPy), 155.58 (CPz), 165.12 (CPz),
172.67 ppm (CIm). Elemental analysis calcd for C60H60N18Fe3B6F24: C
41.86, 3.51, 14.64%; found: C 41.68, H 3.74, N, 14.84%. IR υCN (ATR, cm-
1): 1608 (m) 1569 (w) 1484 (m) 1443 (m) 1363 (w) 1317 (w) 1303 (w) 1206
(m) 1030 (s) 927 (m) 761 (m). UV/Vis (CH3CN): λmax (ε) = 254 (16775), 300
(12750), 433 (6250), 623 (5250), 721 nm (4750 L mol-1cm-1). Cryospray-
MS (+): m/z = 1635.3403 {[Fe3L3](BF4)5}+ (calcd = 1635.3479), 774.1794
{[Fe3L3](BF4)4}2+ (calcd = 774.1719).
DFT calculations
Conflict of Interest
All DFT calculations were performed using the ORCA program version
3.0.3.[24] Complexes were fully optimized using the BP86[12] functional with
a def2-SVP basis set.[14] The resolution of identity approximation[25] was
also used with a def2-SVP/J auxillary basis set.[26] Calculations were
performed in a polarizable continuum solvent using both the COSMO[27]
and[28] SMD solvation models, and acetonitrile as the solvent. The starting
The authors declare no competing financial interests.
Keywords: supramolecular chemistry; self-assembly; triangles;
cyclohelicates; entropy
coordinates for [Zn3L3]6+, [Fe3L3]6+, [CoIII4(L2,5-Et)4]4+, [FeII (bbppz)4]8+, and
4
[1]
a) D. L. Caulder, K. N. Raymond, J. Chem. Soc., Dalton Trans. 1999,
1185-1200; b) D. L. Caulder, K. N. Raymond, Acc. Chem. Res. 1999, 32,
975-982; c) B. J. Holliday, C. A. Mirkin, Angew. Chem. Int. Ed. 2001, 40,
2022-2043; d) S. Leininger, B. Olenyuk, P. J. Stang, Chem. Rev. 2000,
100, 853-908; e) J.-M. Lehn, Science (Washington D.C.) 2002, 295,
2400-2403; f) M. Fujita, M. Tominaga, A. Hori, B. Therrien, Acc. Chem.
Res. 2005, 38, 369-378; g) J. R. Nitschke, Acc. Chem. Res. 2007, 40,
103-112; h) E. Zangrando, M. Casanova, E. Alessio, Chem. Rev. 2008,
108, 4979-5013; i) B. Lippert, P. J. Sanz Miguel, Chem. Soc. Rev. 2011,
40, 4475-4487; j) R. J. Chakrabarty, P. S. Mukherjee, P. J. Stang, Chem.
Rev. 2011, 111, 6810-6918; k) N. J. Young, B. P. Hay, Chem. Commun.
2013, 49, 1354-1379; l) W. Wang, Y.-X. Wang, H.-B. Yang, Chem. Soc.
Rev. 2016, 45, 2656-2693.
[ZnII (bbppz)4]8+ were those from the cif file of the x-ray crystallographic
4
data. The starting coordinates of [Zn4L4]8+ and [Fe4L4]8+ were those from
the cif file of [CoIII4(L2,5-Et)4]4+, and the starting coordinates of [CoIII3(L2,5-
Et)3]3+, [FeII3(bbppz)3]6+ and [ZnII (bbppz)3]6+ were those of [Zn3L3]6+
.
3
Frequency calculations were carried out on all complexes to confirm they
were local minima by the existence of no significant imaginary frequencies.
Imaginary frequencies below 35 cm-1 were assumed to be due to
numerical noise. No corrections were made for zero point energies or
dispersion. The calculations produce total electronic energies, which are
adjusted for the complex stoichiometry and reported as total energy per
ML unit. All complexes were re-optimized using the B3LYP functional and
the results were qualitatively the same, confirming that the relative
stabilities were not artefacts of the chosen density functional.
The enthalpy, H, and entropy, S, of the cyclohelicates were derived
using DFT-calculated energies and ideal gas statistical mechanics within
the harmonic approximation. Vibrational modes with frequencies below 35
cm-1 were excluded, the number of which was between 5-18, depending
on the complex. It should be noted that there remain very many low-
frequency modes. Low-frequency vibrational modes are not well-described
with a harmonic description therefore there exists a large degree of
uncertainty in the calculated vibrational entropy. The calculated rotational,
vibrational and translation entropies of the cyclohelicates are presented in
Table S4. Clearly, TSvib is approximately an order of magnitude larger than
TSrot and TStrans, therefore it has the largest effect on the overall entropy
of the cyclohelicates. Given that this value is highly uncertain, the overall
entropy and any quantity derived from it (e.g. Gibb’s energy, G) is also
uncertain therefore we do not present calculated values of G here.
Considering the relative enthalpy and entropy of the cyclohelicates (Table
S4), it is clear that the enthalpic product is the same as predicted by the
relative electronic energies, for all M3L3/M4L4 pairs. In all cases, the
triangle is the entropically favoured cyclohelicate.
[2]
[3]
a) M. Ruben, J. Rojo, F. J. Romero-Salguero, L. H. Uppadine, J.-M. Lehn,
Angew. Chem. Int. Ed. 2004, 43, 3644-3662; b) M. Ruben, J.-M. Lehn,
P. Müller, Chem. Soc. Rev. 2006, 35, 1056–1067; c) L. N. Dawe, T. S.
M. Abedin, L. K. Thompson, Dalton Trans. 2008, 1661-1675; d) L. N.
Dawe, K. V. Shuvaev, L. K. Thompson, Chem. Soc. Rev. 2009, 38, 2334-
2359; e) J. G. Hardy, Chem. Soc. Rev. 2013, 42, 7881-7899.
a) J. Hausmann, G. B. Jameson, S. Brooker, Chem. Commun. 2003,
2992-2993; b) J. Hausmann, S. Brooker, Chem. Commun. 2004, 1530-
1531; c) J. Klingele (née Hausmann), J. F. Boas, J. R. Pilbrow, B.
Moubaraki, K. S. Murray, K. J. Berry, K. A. Hunter, G. B. Jameson, P. D.
W. Boyd, S. Brooker, Dalton Trans. 2007, 633-645 and front cover
feature.
[4]
[5]
a) T. Bark, M. Duggeli, H. Stoeckli-Evans, A. von Zelewsky, Angew.
Chem. Int. Ed. 2001, 40, 2848-2851; b) W. Meng, T. K. Ronson, J. K.
Clegg, J. R. Nitschke, Angew. Chem., Int. Ed. 2013, 52, 1017-1021.
a) A.-M. Stadler, F. Puntoriero, S. Campagna, N. Kyritsakas, R. Welter,
J.-M. Lehn, Chem. Eur. J. 2005, 11, 3997-4009; b) A.-M. Stadler, F.
Puntoriero, F. Nastasi, S. Campagna, J.-M. Lehn, Chem. Eur. J. 2010,
16, 5645-5660; c) A.-M. Stadler, J. Ramírez, J.-M. Lehn, Chem. Eur. J.
2010, 16, 5369-5378; d) A.-M. Stadler, L. Karmazin, C. Bailly, Angew.
Chem. Int. Ed. 2015, 54, 14570-14574; e) J. Ramirez, A.-M. Stadler, G.
Rogez, M. Drillon, J.-M. Lehn, Inorg. Chem. 2009, 48, 2456-2463.
a) V. V. Kastron, I. G. Iovel, I. P. Skrastyn'sh, Y. S. Gol'dberg, M. V.
Shimanskaya, G. Y. Dubur, Chemistry of Heterocyclic Compounds 1986,
22, 915-917; b) I. G. Iovel, I. Jansone, Y. S. Gol'dberg, M. V.
Shimanskaya, Khim. Geterotsikl. 1990, 4, 532-537; c) R. Franke, Ber.
Deutsch. Chem. Ges. 1905, 38, 3724-3728; d) M. Hasegawa, Y. Asusuki,
F. Susuki, H. Nakanishi, J. Polym. Sci., Part A: Polym. Chem. 1969, 7,
Acknowledgements
[6]
We thank the University of Otago (PhD scholarship to RWH; and
Versalab magnetometer purchase), the MacDiarmid Institute for
Advanced Materials and Nanotechnology (PhD scholarship to
SD), and the Marsden Fund (RSNZ; PhD scholarship to RMH) for
This article is protected by copyright. All rights reserved.

10.1002/chem.201702153
Chemistry - A European Journal
FULL PAPER
743-752; e) R. H. Wiley, Journal of Macromolecular Science: Part A -
Chemistry 1987, 24, 1183-1190; f) R. Coufal, M. Prusková, I. Císařová,
D. Drahoňovský, J. Vohlídal, Synth. Commun. 2016, 46, 348-354; g) S.
K. Das, J. Frey, Tetrahedron Lett. 2012, 53, 3869-3872; h) H. A. Staab,
W. K. Appel, Liebigs Ann. Chem. 1981, 1065-1072.
[17] a) C. Tanford, Science 1978, 200, 1012-1018; b) W. Cullen, C. A. Hunter,
M. D. Ward, Inorg. Chem. 2015, 54, 2626-2637.
[18] O. Mamula, F. J. Monlien, A. Porquet, G. Hopfgartner, A. E. Merbach, A.
von Zelewsky, Chem. Eur. J. 2001, 7, 533-539.
[19] a) R. H. Blessing, Acta Crystallogr., Sect. A: Found. Crystallogr. 1995,
51, 33-38; b) G. M. Sheldrick, Göttingen, 1996.
[7]
[8]
[9]
Y. Xiao-Yan, M. Masahiko, K. Mitsuru, K. Susumu, J. Guo-Xin, Chem.
Lett. 2001, 30, 168-169.
[20] G. Sheldrick, Acta Crystallographica, Section C 2015, 71, 3-8.
[21] P. van der Sluis, A. L. Spek, Acta Crystallogr., Sect. A: Found. Crystallogr.
1990, A46, 194-201.
M. Schweiger, S. R. Seidel, A. M. Arif, P. J. Stang, Angew. Chem. Int.
Ed. 2001, 40, 3467-3469.
S. Derossi, M. Casanova, E. Iengo, E. Zangrando, M. Stener, E. Alessio,
Inorg. Chem. 2007, 46, 11243-11253.
[22] C. F. Macrae, I. J. Bruno, J. A. Chisholm, P. R. Edgington, P. McCabe,
E. Pidcock, L. Rodriguez-Monge, R. Taylor, J. van de Streek, P. A. Wood,
J. Appl. Crystallogr. 2008, 41, 466-470.
[10] A. Neels, H. Stoeckli-Evans, Inorg. Chem. 1999, 38, 6164-6170.
[11] D. S. Cati, J. Ribas, J. Ribas-Arino, H. Stoeckli-Evans, Inorg. Chem.
2004, 43, 1021-1030.
[23] POVray, Persistence of Vision Raytracer (Version 3.6) ed., Persistence
of Vision Raytracer (Version 3.6), 2004.
[12] a) J. P. Perdew, W. Yue, Phys. Rev. B 1986, 33, 8800-8802; b) J. P.
Perdew, Phys. Rev. B 1986, 33, 8822-8824; c) A. D. Becke, Phys. Rev.
A 1988, 38, 3098-3100.
[24] F. Neese, WIREs Comput. Mol. Sci. 2012, 2, 73-78.
[25] F. Neese, J. Comput. Chem. 2003, 24, 1740-1747.
[26] a) K. Eichkorn, O. Treutler, H. Öhm, M. Häser, R. Ahlrichs, Chem. Phys.
Lett. 1995, 240, 283-290; b) K. Eichkorn, O. Treutler, H. Öhm, M. Häser,
R. Ahlrichs, Chem. Phys. Lett. 1995, 242, 652-660; c) K. Eichkorn, F.
Weigend, O. Treutler, R. Ahlrichs, Theor. Chem. Acc. 1997, 97, 119-124.
[27] a) A. Klamt, G. Schüürmann, J. Chem. Soc., Perkin Trans. 2 1993, 799-
805; b) A. Klamt, J. Phys. Chem. 1995, 99, 2224-2235; c) A. Klamt, V.
Jonas, J. Chem. Phys. 1996, 105, 9972-9981.
[13] a) M. Bühl, H. Kabrede, J. Chem. Theory Comput. 2006, 2, 1282-1290;
b) L. J. Kershaw Cook, R. Kulmaczewski, R. Mohammed, S. Dudley, S.
A. Barrett, M. A. Little, R. J. Deeth, M. A. Halcrow, Angew. Chem. Int. Ed.
2016, 55, 4327-4331.
[14] A. Schäfer, H. Horn, R. Ahlrichs, J. Chem. Phys. 1992, 97, 2571-2577.
[15] T. Bark, A. von Zelewsky, D. Rappoport, M. Neuburger, S. Schaffner, J.
Lacour, J. Jodry, Chem. Eur. J. 2004, 10, 4839-4845.
[28] A. V. Marenich, C. J. Cramer, D. G. Truhlar, The Journal of Physical
Chemistry B 2009, 113, 6378-6396.
[16] a) M. Fujita, O. Sasaki, T. Mitsuhashi, T. Fujita, J. Yazaki, K. Yamaguchi,
K. Ogura, Chem. Commun. 1996, 1535-1536; b) M. Scherer, D. L.
Caulder, D. W. Johnson, K. N. Raymond, Angew. Chem., Int. Ed. Engl.
1999, 38, 1587-1592; c) P. N. W. Baxter, J.-M. Lehn, G. Baum, D. Fenske,
Chem. Eur. J. 2000, 6, 4510-4517; d) F. A. Cotton, C. A. Murillo, R. Yu,
Dalton Trans. 2006, 3900-3905.
This article is protected by copyright. All rights reserved.

10.1002/chem.201702153
Chemistry - A European Journal
FULL PAPER
Layout 2:
FULL PAPER
Ross W. Hogue, Sebastien Dhers, Ryan
M. Hellyer, Jingwei Luo, Garry S.
Hanan, David S. Larsen, Anna L.
Garden* and Sally Brooker*
Page No. – Page No.
Self-Assembly of Cyclohelicate [M3L3]
Triangles over [M4L4] Squares,
despite near Linear bis-terdentate L
and Octahedral M
Rare cyclohelicate triangle architectures, not squares as seen for the bis-
terdentate di-amide ligand, self-assemble when the di-imine analogue is employed.
In the triangles, the octahedral metal ions are located 0.4-0.5 Å out of the plane of
the bridging pyrazine rings, towards the centre of the triangle. DFT calculations
correctly predict the observed triangle architectures to be energetically favoured.
This article is protected by copyright. All rights reserved.