Systematic study of the synthesis and coordination of 2-(1,2,3-triazol-4-yl)-pyridine to Fe(II), Ni(II) and Zn(II); Ion-induced folding into helicates, mesocates and larger architectures, and application to magnetism and self-selection

Article abstract of DOI:10.1039/c5dt00233h

With its facile synthesis, the pyridine-1,2,3-triazole chelate is an attractive building block for coordination-driven self-assembly. When two such chelates are bridged by a spacer and exposed to cations of octahedral geometrical preference, they generally self-assemble into dinuclear triple-stranded structures in the solid state and in solution in the presence of non-coordinating counter-ions. In solution, a wider range of architectures may nevertheless form, depending on the nature of the spacer. A systematic study of the spacer and substitution pattern is therefore presented, which allows assessing the various factors affecting self-assembly around the pyridine-1,2,3-triazole chelate, as well as the stereochemical control in these architectures. Applications to chirality, magnetism and system selection are discussed, and involve Fe(II), Ni(II), Zn(II) and Cu(I) cations.

Full text of DOI:10.1039/c5dt00233h

Dalton
Transactions
View Article Online
View Journal
PAPER
Systematic study of the synthesis and coordination
of 2-(1,2,3-triazol-4-yl)-pyridine to Fe(II), Ni(II) and
Zn(II); ion-induced folding into helicates,
Cite this: DOI: 10.1039/c5dt00233h
mesocates and larger architectures, and
application to magnetism and self-selection†
Nan Wu,a Caroline F. C. Melan,b Kristina A. Stevenson,a Olivier Fleischel,c Huan Guo,a
Fatemah Habib,d Rebecca J. Holmberg,d Muralee Murugesu,d Nicholas J. Mosey,a
Hélène Nierengartene and Anne Petitjean*a
With its facile synthesis, the pyridine–1,2,3-triazole chelate is an attractive building block for coordination-
driven self-assembly. When two such chelates are bridged by a spacer and exposed to cations of octa-
hedral geometrical preference, they generally self-assemble into dinuclear triple-stranded structures in
the solid state and in solution in the presence of non-coordinating counter-ions. In solution, a wider
range of architectures may nevertheless form, depending on the nature of the spacer. A systematic study
of the spacer and substitution pattern is therefore presented, which allows assessing the various factors
affecting self-assembly around the pyridine–1,2,3-triazole chelate, as well as the stereochemical control
in these architectures. Applications to chirality, magnetism and system selection are discussed, and
involve Fe(II), Ni(II), Zn(II) and Cu(I) cations.
Received 17th January 2015,
Accepted 17th July 2015
DOI: 10.1039/c5dt00233h
www.rsc.org/dalton
cations owing to its high efficiency and ease of purification,
functional group tolerance and compatibility with various sol-
Introduction
The copper-catalyzed azide–alkyne cycloaddition (CuAAC) reac- vents including water.1 Recently, the resulting nitrogen-rich
tion has received tremendous attention and found many appli- 1,2,3-triazole has also been recognized in supramolecular
chemistry as a folding codon2 and a ligand for metal ions.3,4
New coordination motifs based on this unit display unique
luminescence,3c,4e magnetism4f and electrochemistry.4b,d,e
aDepartment of Chemistry, Queen’s University, Chernoff Hall, 90 Bader Lane,
Multinuclear 1,2,3-triazole-based self-assembled architectures
Kingston, ON K7L3N6, Canada. E-mail: anne.petitjean@chem.queensu.ca;
such as cages are emerging.4c,h Surprisingly, helicates,5 which
Fax: +(1) 613 533 6669; Tel: +(1) 613 533 6587
bDépartement de Chimie, Université d’Angers, 2 boulevard Lavoisier, 49045 Angers
Cedex 01, France
are among the most popular self-assembled structures due to
their unique properties (e.g., chirality, energy transfer)5 have
barely been explored in the 1,2,3-triazole series,4g,6 despite the
attractive features of the nitrogen-rich triazole unit (e.g., mul-
tiple interaction modes, limited steric hindrance, moderate
ligand field). However, as with most multinuclear self-
assembled architectures, the role of the spacer bridging two
coordination units, as well as the substitution state of the
ligands, are critical in determining the folding outcome.7 We
report herein a systematic study of the self-assembly of di-
nuclear architectures based on the 1,2,3-triazolylpyridine
chelate, by varying the flexibility and length of the spacer
(L1–L6, Scheme 1), and changing the substitution pattern of
the terminal pyridine (L1M, L3M and L4M, Scheme 1), giving
insights into the impact on stereochemical, thermodynamic
and magnetic properties.
cBASF, Carl-Bosch-Straße 38, 67056 Ludwigshafen, Germany
dDepartment of Chemistry, University of Ottawa, 401 D’Iorio Hall, 10 Marie Curie,
Ottawa K1N6N5, ON, Canada
eService de Spectrométrie de masse de l’Institut de Chimie de Strasbourg, 1 rue
Blaise Pascal, 67000 Strasbourg, France
†Electronic supplementary information (ESI) available: Synthetic details for
ligands and complexes; copy of the 1H and 13C NMR of the ligands; 1H NMR of
the iron(II) complexes (1D and 2D); UV-vis titrations and Job plot for the nickel
(II) and iron(II) for all ligands; details on DFT calculations, mass spectrometry,
VPO and magnetism; simple models for L4-derived helicates and mesocates,
loop structures and hexameric assembly; revised crystallographic data for
[Fe2(L3)3](BF4)4. The latter complex was already published in ref. 4g, but its struc-
−
ture was further refined to locate all the BF4 anions, consistent with a 2+ oxi-
dation state around each iron center. 19F NMR analysis of the solubilized
−
crystals also confirms that each complex is accompanied by 4 BF4 anions (as
illustrated in the ESI). CCDC 1057879. For ESI and crystallographic data in CIF
or other electronic format see DOI: 10.1039/c5dt00233h
This journal is © The Royal Society of Chemistry 2015
Dalton Trans.

View Article Online
Paper
Dalton Transactions
some cases, heating was applied after the first reaction and
before addition of the CuAAC reagents in order to ensure full
SN2 conversion.
The nickel(II) and iron(II) complexes were either generated
in situ on a small scale for direct 1H NMR analysis, and iso-
lated by crystallization (when amenable) of an acetonitrile
solution through slow diffusion of diethyl ether (74–95% recov-
ery yields). Complexes that failed to produce crystals were iso-
lated by concentrating the acetonitrile solution to dryness,
followed by hot digestion with dichloromethane to remove any
unbound ligand (51–55% recovery yield). Complexes of L3 with
iron(II) were prepared in degassed solvents and crystallized
under argon to ensure no iron(III) would be responsible for the
observed magnetic behaviour (vide infra). Iron(II) complexes
with the other ligands were consistent with the Fe(II) redox
state and were not observed to be air sensitive. Therefore they
can be handled in air.
Self-assembly of dinuclear complexes
Self-assembly in the solid state
With two chelating groups connected through a semi-rigid
spacer, ligand L2c is designed to form dinuclear multistranded
complexes akin to those derived from 2,2′-bipyridines.5b
Coordination of L2c to FeII and NiII indeed results in isostruc-
tural helicates, with two pseudo-octahedral coordination
centers of identical chirality (Λ, Fig. 1a) bridged by a central
p-xylyl fragment of opposite twist.4g As anticipated, the Fe–N
Scheme 1 Monofunctional (L1,3b,4a,b,e,6 L1M) and bifunctional (L2a,
L2b, L2c,4b L2d,4g L3,4a,6 L4, L5 and L6) ligands used. The numbering on
the mono-chelating unit (bottom) relates to NMR assignments.
4+ 4g
4+
bond lengths in Fe(L1)3
and Fe2(L2c)3 are nearly identi-
cal, and typical of low-spin iron(II).6a Similar properties are
found in the solid-state structure of the naphthalene
Synthesis of ligands and complexes
One of the strengths of the triazole unit is its facile and
efficient synthesis starting from an organic azide using copper
catalysed alkyne–azide cycloaddition.1 In an effort to avoid iso-
lating the potentially explosive organic azides, their in situ
preparation was achieved from alkylbromides through an SN2
reaction with sodium azide. As described by Crowley’s group,4a
both the SN2 and CuAAC reactions may be carried out in a one
pot reaction, by mixing all reagents in a combination of
organic solvent (e.g., DMF or tBuOH) and water. In our study,
this method worked very efficiently for electrophiles that are
activated through the proximity of π systems, as in propargylic
and benzylic/benzylic-like positions (e.g., L2b, L2c, L2d), or
through inductive effects of the second bromide, as in L2a.
However, yields are moderate for non-activated alkyl bromides
(e.g., organic halides where the two leaving groups are separ-
ated by more than two methylene units). In order to promote a
complete SN2 reaction on these electrophiles, it is beneficial to
split this one pot conversion into two processes: the SN2 reac-
tion first efficiently takes place in the absence of any protic
solvent (e.g., in pure DMF) to complete the in situ synthesis of
the organic azide, after which the CuAAC reagents and water
are added to transform the organic azide into the triazole ring.
The overall conversion is therefore efficiently accomplished in
a one pot, although through two consecutive processes. In
4+
Fig. 1 Side and top views of the dinuclear (a) Fe2(L2c)3 (only one
4+
enantiomer shown) and (b) Fe2(L3)3 at 77 K. Solid-state self-assembly
is also illustrated (blue boxes: head-to-tail pyridine–triazole dimers; pink
boxes: head-to-head pyridine–triazole dimers, green shade: van der
Waals contacts).4g Anions and solvents are omitted for clarity. Selected
distances (Å): (a) Fe–N2: 1.939(2), Fe–N1: 1.998(2); Fe⋯Fe: 11.391; (b)
Fe–Ntriazole: 1.925–1.945, Fe–Npyridine: 1.995–2.012, Fe⋯Fe: 9.878.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
[Fe2(L2d)3](BF4)2 analog.4g Despite all efforts, attempts to between very tight binding for L3 and L4 (sharp transition at
produce single crystals from complexes derived from ethylene- the 0.67 ratio, Fig. 2a), and strong binding for L2a and L5
like spacers L2a and L2b have been unsuccessful. However, (smoother transition, saturating at a higher ratio, Fig. 2a). The
increasing the number of CH2 units, the FeII and NiII com- propylene and butylene spacers therefore accommodate the
plexes of the propylene-derived L3 ligand are found to also be triple stranded assembly in the best manner. On the contrary,
isostructural in the solid-state and result in mesocates due to the shorter ethylene spacer (L2a) may induce increased strain
the chirality reversal induced by the propylene spacer (Fig. 1b). and/or electrostatic repulsion (this aspect will be discussed
In our study, longer ligands (L4–L6) have not been amenable further below). Furthermore, the longer pentylene spacer with
to single crystal formation with either iron(II) or nickel(II).
increased degrees of freedom may be entropically more
Based on the crystal structures above illustrating the stereo- disfavoured.
chemical effect of even-numbered and odd-numbered spacers
1H NMR analysis allows us to better understand the triple-
(L2c and L3, respectively), the helicate vs. mesocate selection stranded structures of the iron(II) complexes for all ligands
rules defined for 2,2′-bipyridyl- and catechol-based systems5b (Fig. 3).
may be at work here as well. In these two families of ligands,
As seen in Fig. 3, the short spacer in L2a leads to a mixture
the zigzag conformational preference of alkyl fragments has of species among which sharp peaks correspond to the well-
4+
been proposed to favour mesocates for spacers with an odd defined Fe2(L2a)3 triple-stranded complex, as evidenced by
number of carbon atoms, and helicates for spacers with an the typical relative chemical shifts of the H3 and H6 pyridyl
even number of carbon atoms (here, a p-xylyl unit acts as an protons which are easily extracted through COSY analysis
extended CH2–CH2 bridge). However, the limited crystallo- (ESI†). The broader signals likely result from the coexisting
graphic data and possible bias introduced by packing forces in oligomers (vide infra). When the two carbon atoms of the
the solid state invite more in-depth exploration of self-assem- spacer are spread further apart, for instance with an ethynyl
bly through solution studies.
(L2b, Fig. 3), a phenyl (L2c, ESI†) or a naphthyl group (L2d,
ESI†), only one well-defined species is observed. The short
ethylene spacer in L2a is further discussed below. With a pro-
pylene spacer which is one carbon atom longer, ligand L3
mixed with [Fe(H2O)6](BF4)2 in a 3 : 2 ratio only leads to one
well-defined species. Surprisingly, the butylene spacer of L4
not only forms a major triple-stranded structure, but also
minor side-products (see small peaks in Fig. 3). The same
minor side-products are observed irrespective of the prepa-
ration method, i.e. a fresh 3 : 2 L4/Fe2+ mixture or isolation of
the iron complexes. These side-products are likely coexisting
larger architectures (vide infra). Interestingly, the pentylene
spacer shows well-defined signals consistent with two species
only (shaded differently in Fig. 3). Both species correspond to
triple-stranded dinuclear complexes similar to the other
Self-assembly in solution: influence of the spacer length
analysed by NMR and UV-vis spectroscopy
Stoichiometry as well as relative binding strength in both iron
(II) and nickel(II) families were first explored by UV-vis absorp-
tion titrations. Diamagnetic iron(II) species were then studied
by 1H NMR spectroscopy in search for structural information.
Comparing complexes derived from the aliphatic spacers
(from ethylene, L2a, to pentylene, L5), saturated 3 : 2 ligand/
cation species were exclusively formed in the 10−5 M range for
both nickel(II) and iron(II) with tetrafluoroborate counterions,
as evidenced by the plateau observed at around 0.7 ion/ligand
ratio, in the absorbance profile at 305 nm and 425 nm respect-
ively (Fig. 2). A closer look at the absorbance profiles indicates
that all ligands form very tight complexes with iron(II), as the
absorbances show a plateau starting at a 0.67 ratio for all
ligands (Fig. 2b). The iron complexes are so tight that UV-vis
titrations at these concentrations fail to provide quantitative
distinction between the relative association constants.
However, the nickel titration profiles show a differentiation
4+
Fe2(L)3 complexes. Significantly different signatures appear
Fig. 3 Comparative 1H NMR of [Fe2(L)3](BF4)4 complexes with L = L2a,
L2b, L3, L4 and L5 (CD3CN, 25 °C, c ∼ 10 mM). The complex with L3 was
freshly prepared (3 : 2 L3/Fe2+ mixture). Top: H numbering in black, tri-
azole nitrogen numbering in blue.
Fig. 2 Overlay of UV-vis titration profiles of (a) nickel(II) and (b) iron(II)
complexes of L2a, L3, L4 and L5 (CH3CN, 25 °C).
This journal is © The Royal Society of Chemistry 2015
Dalton Trans.

View Article Online
Paper
Dalton Transactions
in the aliphatic region, and in particular in the signals of the ging from the chelating unit, and leading to the linker, relative
γ protons (Fig. 3). As observed in other chelating systems,5b the to a virtual line defined by the two donor atoms in the chelate.
central CH2 group is essential in differentiating various triple- Here, the (N1)triazole–Cα bond and the NPyridine–(N3)triazole line
stranded structures. In the ‘primed species’ (Fig. 3), the (Fig. 3) are not quite parallel. However, this system is probably
γ protons experience an identical environment which is mechanically closer to a 5,5′-substituted-2,2′-bipyridyl-type
characteristic of a helicate species (each proton faces the system than to a 4,4′-substituted-2,2′-bipyridyl architecture, the
‘nitrogen’ end of one chelate unit and the CH edge of the latter suffering from greater steric limitations. These various
other, Fig. 4a).5,8a The γ protons of the ‘non-primed species’ bond orientations impact the flexibility of the linker to various
show a diastereotopic signature signifying that they perceive a degrees, and thereby impact the accessibility of helicate vs.
different environment, and therefore belong to a mesocate mesocate structures. Finally, the actual chemical nature of the
structure (the H(2) proton in Fig. 4b faces the ‘nitrogen’ ends ‘connection point’ between the chelating unit and the start of
of both chelates, whereas H(1) in Fig. 4b faces the two CH the linker plays a significant role in tuning the odd/even
edges of the chelates).
number rule; its own conformational energy profile dictates
Based on the above assignment rationale, the ratio of heli- conformational restrictions which can tune the number of
4+
cate to mesocate for Fe2(L5)3 is 1 : 2 as determined by the carbon atoms at which both helicate and mesocate start co-
intensities of the 1H NMR signals. The preference for the existing. In the present example, a simple, minimal twist of
mesocate structure agrees with the odd/even rule which has the ‘bottom’ Cα–(N1)triazole bond to the right (Fig. 4a) or to the
been rationalized based on the preferential zigzag confor- left (Fig. 4b) inverts the chirality of the 2nd (bottom) metal
mational preference of alkyl spacers.5b However, with the center, so the transition point to a helicate + mesocate mixture
pyridyl–triazole chelates, rough molecular models indicate starts as early as in the pentylene spacer. Rich examples invol-
that only a very small deviation from an ‘all-staggered’ confor- ving an amide connection point to 2,2′-bipyridine8b and hydro-
mation allows access to the helicate architecture (Fig. 4a), xypyridine8c chelates harness the weak interactions offered by
which supports a significant population of the helicate struc- the amide unit (with solvents) to restrict the available confor-
ture. Indeed, as the alkyl spacer length increases, a smaller mations and control the helicate/mesocate outcome. With
helical twist in the spacer region is required to form the heli- catechol chelating units, conformational restrictions can be
cate structure. As a result, a minute energy difference in the imposed by the ‘inner’ catechol oxygens binding to a cationic
conformational energy profile is expected between the helicate guest at the corners of the cavity, resulting in the selection of
and mesocate, leading to coexisting species. In the present helical vs. side-by-side architectures.5c Overall, in the pyridyl–
study focusing on pyridyl–1,2,3-triazole chelates, the transition triazole family connected by alkyl spacers, the present com-
to a helicate + mesocate mixture starts with the pentylene parison therefore highlights that (i) the odd–even rule only
spacer. More generally speaking, this transition point will vary holds up to the length of a butylene spacer, and (ii) the even
from one system to the next, depending on several parameters numbered spacers (L2, L4) induce oligomer formation at mM
influencing flexibility. The odd/even number of carbons within concentrations (Fig. 3), whereas the odd-numbered spacers
the linker itself is one parameter, as stressed by other investi- only favour folding into well-defined species.
gators.5 Another one is the orientation of the first bond emer-
As reported above, the ethylene spacer in L2a stands out: it
leads to less stable complexes at sub-micromolar concen-
trations (UV-vis, Fig. 2) and to a significant amount of oligo-
mers at millimolar concentrations (1H NMR, Fig. 3). The more
4+
limited stability of a Fe2(L2a)3 triple-stranded complex may
stem from one or both of the following effects: (i) the closer
proximity of the two divalent Fe2+ centers resulting in
increased mutual electrostatic repulsion, and/or (ii) the proxi-
mity of all six 1,2,3-triazolyl N2 nitrogen atoms, and conse-
quent lone pair electronic repulsion near the center of the
triple-stranded structure (Fig. 1 gives an idea of the distri-
bution of the three N2 (‘middle’) nitrogen atoms within triple-
stranded helicates and mesocates, yet in a context where both
coordination units are fairly far apart). In order to assess the
relative contribution of those two effects to the limited stability
of Fe2(L2a)34+, a comparative study of the Zn2+ complexes with
two ‘L2-type’ ligands was undertaken. The diamagnetic nature
of the Zn2+ complex facilitates structural and stability studies
Fig. 4 Cartoon representation of the environments experienced by the
1
through H NMR titrations. Ligands L2a and L2c, with ethyl-
central methylene protons of odd-numbered alkyl spacers in (a) helicate
ene and p-xylyl spacers, respectively, offer similar rigidity but a
significant difference in ion–ion distances (the Fe2+⋯Fe2+ dis-
tance is 11.4 Å in the X-ray structure of [Fe2(L2c)3](BF4)4, and
4+
and in (b) mesocate. Rough models of Fe2(L5)3 helicate (a) and meso-
cate (b) are also represented. The central methylene group is circled in
grey and specified on the cartoons.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
∼7.7 Å in a rough model of Fe2(L2a)34+). By 1H NMR, the
By comparison with the p-xylyl spacer, the ethylene spacer
p-xylyl-based L2c shows the gradual formation of a triple- is characterized by broad signals all along the titration (0.2–0.9
stranded structure first, from 0.1 to 0.7 equivalent of Zn2+ equivalent of Zn2+), and only leads to reasonably well-defined
(Fig. 5b, bottom part; note the shielding of the pyridyl proton signals after 1.0 equivalent. The species formed at 1.0 equi-
H6 and diastereotopic protons in the CH2 group, which are valent is, in all likelihood, the double-stranded [Zn2(L2a)2]4+
typical of a triple-stranded assembly), followed by dissociation complex: the chemical shifts and ordering of the aromatic
into a double-stranded species from 0.8 equivalents and on protons are similar to that in [Zn2(L2c)2]4+. In addition, water
(Fig. 5b, top). The signature of the double-stranded structure gets involved in binding after 0.8 equivalents in both cases
warrants a few comments. Contrary to the structure at 0.7 equi- (moving exchangeable water peak not shown), which confirms
valent of Zn2+, the methylene groups are not diastereotopic the availability of binding sites on Zn2+ in the 0.2–0.8 equiv.
anymore (on the NMR time-scale). Importantly, the pyridyl range. In both cases, this likely 2 : 2 species slowly precipitates
proton H6 appears at a high chemical shift, consistent with quantitatively, as anticipated with the formation of a neutral
the coordination outside the shielding cone of the neighbour- complex in a relatively polar solvent (two chelates and two tri-
ing aromatic ligands. The protons of the phenyl ring (‘a’) flates around each divalent cation center). The broadening at
however, undergo significant shielding, possibly due to a side- 1.0 equivalent and above is, however, very limited compared to
by-side arrangement.
the very broad patterns observed for L2a between 0.2 and
0.8 Zn2+ equivalent, suggestive of either polymeric materials or
intermediate exchange rates, which would be strikingly
different from the slow exchange observed with L2c.‡ It there-
4+
fore appears that L2a readily forms the Zn2(L2a)2 double-
stranded complex at 1.0 equivalent of zinc(II), but fails to form
a well-defined triple-stranded complex at 0.7 equivalent. Since
both the double- and triple-stranded species would be charac-
terized by a very similar M2+–M2+ distance, the electrostatic
repulsion between the two divalent metal ions in L2a-derived
complexes is not expected to be very different. As a result, the
main driving force behind the limited stability of the satu-
rated, triple-stranded [Zn2(L2a)3]4+ complex is very likely
the repulsion of the six converging lone pairs of the ‘middle’
N2 triazole nitrogen atoms (shaded in grey in Fig. 5c). In a
double-stranded structure, however, the possibility of organiz-
ing the two pyridyl–triazole chelates in a perpendicular
arrangement (as in a tetrahedral complex) relieves the lone
pair repulsion, leading to efficient double-stranded self-
assembly.
The dominant impact of the repulsion between the N2 tri-
azole lone pairs over the electrostatic repulsion between the
two divalent ions in the pyridyl–triazole system may be ana-
lysed by comparison with other chelate families. In the 4,4′-
substituted-2,2′-bipyridyl system, a dinuclear iron(II) complex
with ligands bearing a short ethylene spacer has been reported
with an Fe2+⋯Fe2+ distance of 7.65 Å in the crystalline state,9
which is very similar to what L2a would produce according to a
1
rough model. Although neither yields nor H NMR data were
provided for the 4,4′-substituted-2,2′-bipyridyl system,9 it is
assumed that the complex was well-defined as determined by
TLC and electrochemical analysis. Similarly, in catechol
systems, multivalent cations are held at a close distance, as the
Fig. 5 (a) 1H NMR spectrum of Fe2(L2c)34+, prepared L2c with 0.7 equi-
ethylene spacer is tolerated in
a dinuclear triple-strand
valent of Fe2+ in CD3CN, given for reference. (b) Titration of Zn(OTf)2 to
format,5 and extra monovalent cations are present in the
a solution of L2c (5 : 3 CDCl3/CD3CN, 25 °C, 500 MHz, [L2c]0 = 10 mM);
4+
grey shading signals the triple-stranded Zn2(L2c)3 complex; black rec- center of the architecture. However the compensating anionic
4+
tangles highlight the 1 : 1 double-stranded Zn2(L2c)2 species. (c) Titra-
nature of the strand renders the comparison difficult. Overall,
tion of Zn(OTf)2 to
a solution of L2a (1 : 1 CDCl3/CD3CN, 25 °C,
400 MHz, [L2a]0 = 14 mM). The numbers below the metal ion on the left
correspond to the number of equivalents of M2+ with regards to the
ligand. The triazole N2 nitrogen atom is shaded in grey on the chemical
structures of the complexes.
‡1H NMR broadening may also result from intermediate exchange rates.
However, it is unclear why the exchange rate would be so different in L2a and
L2c.
This journal is © The Royal Society of Chemistry 2015
Dalton Trans.

View Article Online
Paper
Dalton Transactions
in the present case, two divalent cations seem to be easily different chelate units merge into a symmetrical ligand (blue)
accommodated at ∼7 Å distance apart, but the close, enforced when more iron(II) is added. The other ligand form (in green)
proximity of the N2 triazole nitrogen lone pairs is more detri- shows bound chelates resembling the previously characterized
mental to the stability of the triple-stranded structure. Assem- octahedral iron complexes all along the titration. The cartoon
blies with the L2a ligand in the gas phase are discussed at the bottom right of Fig. 6 offers an explanation to these
further below.
various spin environments. At the beginning of the titration,
At the other extreme of the spacer length spectrum to L2a, a the excess ligand surrounds the metal ion by wrapping (sym-
ligand with a very long and flexible spacer composed of three metrical species, in green in Fig. 6) and partial binding (blue
consecutive ethylene-oxide units (L6, Scheme 1 and Fig. 6) was part of the unsymmetrical ligand, Fig. 6). At 0.7 equiv. of iron
also prepared and its coordination properties to iron(II), nickel (II), the spectrum simplifies, with two distinct symmetrical
(II) and zinc(II) were explored. UV-vis titration of iron(II) into a ligands in a 1 : 2 ratio, both of them fully bound to the iron(II)
solution of L6 (Fig. 6 top) shows the formation of two consecu- center (as per the chemical shifts of the triazole t and pyridyl
tive complexes. Up to 0.7 equivalent of iron(II), a first MLCT H6 protons). A likely schematic structure for this ratio is rep-
band at 420 nm corresponds to a 2 : 3 Fe2+/L6 complex (Job resented in Fig. 6, right, where each of the two iron(II) centers
plot in the ESI†). From 0.7 equivalent to 1.0 equivalent of carries a wrapped ligand (in green) and is bridged to the
iron(II), a second complex emerges, with an MLCT band second ion through a third “linear” ligand (blue, Fig. 6). The
growing at 378 nm, corresponding in all likelihood to a 1 : 1 latter structure is likely to be the 3 : 2 ligand/iron(II) complex
complex (a Job plot at 373 nm indicates a mixture of 2 : 3 and observed in the UV-vis titration (Fig. 6. top). Hence, the 3 : 2
1 : 1 complexes overall, ESI†). Because of the inherent flexibility ligand/iron(II) complex with a flexible ligand such as L6 is not
of the spacer in the L6 ligand, its wrapping into a 1 : 1 complex a triple-stranded structure of the helicate/mesocate family.
is expected to occur. However, earlier in the titration, the pres-
Complexation of nickel(II) to L6 leads to the sole formation
ence of a 2 : 3 complex could correspond to a number of folded of the 2 : 3 metal ion/ligand complex (Job plot with maximum
architectures, including triple-stranded structures. In order to at xNi2+ = 0.4, and sharp isosbestic points, ESI†). Zinc(II), on the
gain insight into this architecture, structural information was other hand, leads to the very clean formation of a 1 : 1 complex
sought by 1H NMR. The titration reported in Fig. 6 indicates which is likely an intramolecularly wrapped, tetrahedral-like
two regimes: up to 0.7 equivalent, the ligand signals corres- complex, with strong binding (ESI†).
pond to two coexisting ligand forms. One shows two different
pyridyl–triazole environments at the beginning of the titration
(blue and purple, Fig. 6). The ‘purple’ set of signals resembles
Self-assembly in the gas phase: influence of the spacer length
analysed by mass spectrometry
that of the free ligand and initially has the same signal intensi- Solution studies described above, in particular by 1H NMR
ties as the ‘blue’ signals, suggesting that they correspond to spectroscopy, give some insight into the structure of the self-
the same dissymmetrical, half-bound ligand. These two assembled architectures, yet with limitations when mixtures
are produced as in the case of L2a and L4, or when kinetic
1
exchange takes place in an intermediate rate range where H
NMR signals are broad. In order to probe the nature of the
self-assembled outputs, electrospray mass spectrometry was
applied to the 3 : 2 ligand to iron(II) ion mixtures, at 10−4
M
concentration (at lower concentrations, these architectures fall
apart even under soft desolvation methods, as has been
observed by others6a). Even at such a significant concentration,
extensive fragmentation is visible as mononuclear (in green)
and double-stranded dinuclear (in pink) complexes are present
with all ligands (Fig. 7). The ESI-MS analysis of these mixtures
deserves several comments. First, the use of tetrafluoroborate
counterions somewhat complicates the spectra because of its
partial hydrolysis yielding fluoride anions. Because of the mul-
tiply charged nature of the iron(II) assemblies, various combi-
−
nations of BF4 and F− anions often produce several signals
for the same cationic assembly. So although many peaks
appear on the spectra, only a limited number of cationic
assemblies are actually present. However, the presence of co-
ordinating fluoride anions (compared to tetrafluoroborates)
has the unintended benefit of providing some information
about the coordination sphere of the ion in each assembly:
assemblies whose m/z values point to complexes where the
iron(II) centers are fully coordinated to pyridyl–triazole units
Fig. 6 Iron(II) titration to ligand L6 monitored by UV-vis (top right,
CH3CN, 25 °C, [L6] 3.5 × 10−5 M) and by 1H NMR spectroscopy (bottom,
CD3CN, 25 °C, [L6] 16 mM). (B) stands for ‘bound’, (L) for ‘loop’, (F) for
‘free’.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
Fig. 7 ESI-MS analysis of 2 : 3 mixtures of [Fe(H2O)6](BF4)2 with ligands
L2c (a), L2a (b), L3 (c) and L4 (d), in CH3CN (25 °C). Peak assignment is
shape and colour coded, as detailed on top. Each peak assignment was
confirmed by its isotopic pattern.
Fig. 8 ESI mass spectra between 1300 and 1600 g mol−1 for 2 : 3
mixtures of [Fe(H2O)6](BF4)2 with ligands L2c, L2a, L3 and L4, in CH3CN
(25 °C). Each peak assignment was confirmed by its isotopic pattern.
−
display at least one peak corresponding to sole BF4 accompa-
of this region is given in Fig. 8, together with a cartoon rep-
resentation of the peak assignments. Considering that 2+ ions
are intrinsically less represented in ESI-MS spectra than the 1+
ion, their abundance in the gas phase analysis of L2a and L4
complexes is striking, and correlates with their complex 1H
NMR behaviour compared to L2c and L3 (Fig. 3).
Partially de-assembled hexamers are also observed in the
ESI-MS spectra in the form of dinuclear quadruple-stranded
complexes (orange triangles, Fig. 7), and tetranuclear quadru-
ple-stranded architectures (orange distorted square, Fig. 7;
note the association of four fluorides, as if one for each ion
corner).
In summary, ESI-MS analysis complements and refines the
1H NMR information above: ligands such as L2c and L3 mostly
select dinuclear triple-stranded architectures, while L2a and to
a certain extent L4 are prone to form larger architectures such
as tetranuclear sextuple-stranded assemblies. As suggested by
1H NMR, the L2a ligand seems more susceptible to poly-
association, and this may be the result of a closer proximity of
lone pairs of the 1,2,3-triazole ‘middle’ N2 nitrogen atoms, as
discussed above. In the hexameric structure, this effect would
be relieved in the ligand bridging the two dinuclear double-
stranded portions, as illustrated by a model for one possible
hexameric structure represented in the ESI.†
nying anions (as one would anticipate since no additional
ligand would be required to complete the ion coordination
sphere). However, when the m/z values correspond to complexes
bearing only one or two pyridyl–triazole chelates, at least one
fluoride accompanies the cationic assembly. This interesting
piece of information correlates with the peak assignment deter-
mined based on m/z values and isotopic patterns.
When analysing the ESI-MS spectra in detail, the following
families of assemblies are found: (i) triple-stranded dinuclear
assemblies resembling the ones found in the solid state and
characterized in solution by 1H NMR, represented by brown
diamonds in Fig. 7 are present in their 2+ state for L2c and L3
and 1+ for all the analysed ligands (L2c, L2a, L3 and L4).
However, the m/z section where 1+ signals of these triple-
stranded assemblies appear contains significant amounts of
the tetranuclear sextuple-stranded 2+ architecture (red hexa-
gons in Fig. 7),§ in particular for L2a and L4 ligands. A zoom
§The tetranuclear hexastranded complexes are represented as closed, metallo-
macrocyclic structures in the cartoons. Please note that there is no direct struc-
tural evidence that these are not open-stranded oligomers at this stage. However,
−
the consistent observation of a BF4 only peak seems to suggest that each iron
center is coordinated to pyridyl–triazole chelates only, which can only corre-
spond to a cyclic architecture.
This journal is © The Royal Society of Chemistry 2015
Dalton Trans.

View Article Online
Paper
Dalton Transactions
Self-assembly in solution: influence of the spacer length
analysed by vapour pressure osmometry
The E/c ratio measured for iron(II) complexes with L2a, L2c,
L3 and L4 is given in Fig. 9, and extrapolated to c = 0 g kg−1 by
linear regression (dotted lines).
Although the ESI-MS analysis above provides precious, detailed
information about the various architectures formed by these
complexes by displaying the various m/z peaks corresponding
to each individual assembly in the mixtures, the question of
the desolvation effect still remains. As shown by UV-vis ana-
lysis under dilute conditions (typically 10−5 M, see the ESI† for
details on each ligand), 2 : 3 iron(II)/ligand complexes with
these ligands are very stable. Yet, concentrations as high as
10−4 M are necessary for these species to be detected after elec-
trospray desolvation, and with significant fragmentation still.
Therefore solution-phase mass assessment was explored using
vapour pressure osmometry (VPO), looking for signs of larger
assemblies in solution as well. Although a fairly old method
which gives only average molecular weights and is complicated
by the fact that these are charged, associative systems, VPO
yields some additional qualitative information about the rela-
tive size of the assemblies (when comparing the various
ligands together). The reader is invited to refer to reviews cited
in ref. 10 for details on VPO thermodynamics. For the purpose
of this article, the most relevant expression relating the VPO
response (electrical voltage imbalance) to the average mole-
cular weight (Mna) and absolute charge (|z|) is given by eqn (1)
below, for solutions where no supporting electrolyte was used.
As indicated by eqn (1), the y-intercept of the E/c vs. c curves
is inversely proportional to the apparent molecular weight
Mapp. Based on the y-intercepts of the E/c curves from Fig. 9,
the Mapp (i.e. Mna/(|z| + 1)) ranking for the various complexes
is therefore:
L2a>L4 ꢀ L2c>L3 ðVPOÞ:
The expected molecular weights of the [Fe2(L)3]4+cations
follow the following trend:
L2c>L4>L3>L2a ðMwÞ:
The VPO measurements therefore highlight two features:
(i) L2a ligand yields architectures with the largest Mapp as
measured, when it would be expected to generate the lightest
dinuclear triple-stranded architectures based on the calculated
molecular weights. A larger Mapp results either from a larger
Mna or from a smaller charge |z|. However, UV-vis titrations
indicate that 2 : 3 iron(II)/ligand complexes are formed with all
ligands, so the charge of all these assemblies is at least 4+. A
larger charge would artificially reduce Mapp. However, in L2a’s
case, the Mapp is surprisingly larger than anticipated (not
smaller). This means that L2a forms heavier assemblies, a
result that correlates with species observed by ESI-MS: the
tetranuclear quadruple-stranded species (orange distorted
square, Fig. 7), and the tetranuclear sextuple-stranded complex
(red hexagons, Fig. 7, and red cartoons, Fig. 8). These two
larger species increase Mapp through their significantly larger
molecular weight, while decreasing Mapp through their higher
charge (8+). The overall effect is still a larger Mapp reflected by
a smaller y-intercept for L2a.
(ii) Some similarity is observed with L4-derived complexes,
although at a much smaller scale. Indeed, a dinuclear triple-
stranded [Fe2(L4)3]4+ complex would have a smaller calculated
molecular weight than its L2c counterpart. However their
respective VPO signals are nearly superimposable, meaning
that Mapp for L4 is larger than expected for [Fe2(L4)3]4+ cations.
Once again, this correlates with a larger proportion of higher-
degree assemblies for L4 than for L2c as observed by ESI-MS
(sextuple-stranded assemblies, Fig. 7 and 8) and by 1H NMR
(small side-product peaks). However, as anticipated from the
1H NMR signals, the proportion of these higher oligomers in
the case of L4 is minimal, which correlates with the small
effect on its VPO response compared to L2a.
ꢀ ꢁ
Kvp
E
c
Mna
jzj þ 1
lim
¼
where Mapp
¼
ð1Þ
c!0
Mapp
VPO response is defined by E/c as a function of concen-
tration. E is the electrical microvolt imbalance, which is pro-
portional to the temperature difference between the two
thermistors, c is the concentration in g of solute per kg of
solvent, Kvp is a proportionality constant dependent on the
apparatus, temperature and solvent, Mna is the average mole-
cular weight of the solute, Mapp is the apparent molecular
weight of the solute, and |z| its absolute charge.
All in all, VPO measurements confirm that the higher
assemblies observed in the gas phase by ESI-MS impact the
species distribution in solution as well. Both aliphatic ligands
with an even number of carbon atoms in their spacer are
affected, with the effect on the shorter ethylene spacer being
more substantial than that of the butylene spacer, as also
observed by 1H NMR. This ‘even number’ effect does not seem
to be restricted to the ethylene and butylene spacers, and the
hexylene spacer was also reported to form higher aggregates as
observed by 1H NMR.6a
Fig. 9 VPO analysis of 2 : 3 mixtures of [Fe(H2O)6](BF4)2 with ligands
L2a, L2c, L3 and L4 in CH3CN (38 °C) at 10–40 mM ligand concen-
trations. E is the electrical microvolt imbalance,10 c is the concentration
in g kg−1. Linear regression curves are indicated by dotted lines.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
to better accommodate the end-group strain imposed by the
methyl group. Moving away from a purely aliphatic spacer,
ligand L2cM with a p-xylyl spacer is more conformationally
restricted because of just two methylene groups. However, it
offers more stabilizing effects with possible edge-to-face π–π
interactions between the spacers, as suggested by the CPK rep-
resentation of the [Fe2(L2c)3](BF4)4 and [Ni2(L2c)3](BF4)4 crystal
data, and by the more sigmoidal UV-vis titration curves of
L2cM with Ni(II) and Fe(II) compared to the aliphatic-based
spacers (the comparison of UV-vis data between L2cM and
L4M/L3M is presented in the ESI† for the sake of clarity). This
may explain why L2cM’s binding is stronger than L3M’s, but
weaker than L4M’s. Unfortunately, in our study all attempts to
prepare single crystal suitable for X-ray crystallography to
probe the coordination sphere and spacer involvement with
methylated ligand-based complexes failed. Overall, the end-
strain induced by a methyl group in the 6-position of a pyri-
dine is easily accommodated by a flexible spacer composed of
at least four methylene groups. A more rigid propylene spacer
still forms the 2 : 3 metal/ligand complex yet with a weaker
affinity. However, rigidity can be compensated for by an
additional interaction between the spacer groups, as in L2cM.
Not surprisingly, end-group strain affects iron(II) complexes
more than their nickel(II) counterparts because of iron(II)’s
intrinsic weaker affinity to pyridyl-type ligands.
Fig. 10 Comparative UV-vis titration profiles of (a) nickel(II) and (b) iron
(II) to methylated (L3M, L4M) and non-methylated (L3, L4) bifunctional
ligands (CH3CN, 25 °C).
Self-assembly in solution: influence of the terminal pyridine
Beyond the effect of the spacer on the stability of dinuclear
triple-stranded architectures, another element which may
impact the self-assembly process is the substitution pattern of
the most robust binding site within the chelate, which is, in
the present case, the terminal pyridine. Generally speaking,
the triazole unit is a much weaker ligand than pyridine,
affinity-wise,3e as we have comparatively studied for Cu(I) com-
plexes.4b By introducing a substituent in the 6-position of the
pyridyl, we could investigate the extent of destabilization of the
derived dinuclear species, and the role that the spacer plays in
accommodating the steric hindrance to coordination that is
enforced by the methyl group. A new chelating unit was there-
fore synthesized through the cycloaddition of 2-ethynyl-6-
methylpyridine11 with diazides prepared in situ by the SN2
reaction of sodium azide onto dibromides (ESI†). The di-
bromides, which dictate the properties of the spacer in the
final ligands, were restricted to the p-xylyl, propylene and buty-
lene fragments for which the analogues without methyl groups
form the best defined and most stable Ni(II) and Fe(II) com-
plexes, as reported above. The discussion below will therefore
be limited to methyl-bearing L3M, L4M and L2cM ligands
which are methylated analogues of the propylene-derived (L3),
butylene-derived (L4) and p-xylyl-derived (L2c) ligands, respect-
ively (Scheme 1). The methylated monofunctional ligand L1M
was also prepared as a reference.
In the family of methylated ligands, L3M with the shortest
tested spacer suffers the most destabilization in both iron(II)
and nickel(II) binding, with a much more significant impact
on iron(II) coordination (Fig. 10). Job plot with maxima
between 0.4 and 0.5 confirm the coexistence of 2 : 3 and 2 : 2
complexes (ESI†).¶ Elongating the alkyl spacer by one more
methylene group (as in L4M), however, restores very good to
excellent affinity to iron(II) and nickel(II), respectively, and a
2 : 3 metal/ligand stoichiometry (Fig. 10 and Job plot in the
ESI†). It therefore appears that a four methylene unit brings
sufficient flexibility from the spacer for the dinuclear complex
ESI-MS analysis was also conducted on iron(II) complexes of
L2cM, L3M and L4M ligands at 10−4 M, but not surprisingly
the observed signals mostly correspond to dissociated assem-
blies (ESI, Fig. S64†). VPO also reflects complex dissociation,
with a smaller Mapp for both the L3M and L4M complexes
(L2cM did not give complexes that were soluble enough in
CH3CN for VPO analysis; see Fig. S65 in the ESI†).
Magnetic properties
Iron(II) complexes are of particular interest due to the richness
of their magnetic properties. Indeed, depending on the
balance between the field strength and the electron pairing
energy, these d6 transition metal ions may occupy high-spin
(paramagnetic, S = 2) and low-spin (diamagnetic, S = 0) states.
More interestingly, complexes involving ligands of intermedi-
ate field strength may switch between these two states under
the influence of external stimuli, a category known as ‘spin
cross-over’ (SCO) complexes.12a In this context, the 1,2,3-tri-
azole–pyridine diad is of particular interest since it combines
the strong-field pyridine ligand with the weak-field 1,2,3-
triazol unit. Additionally, the ligand field of a pyridine is
decreased upon introduction of
a methyl group in its
6-position, favouring a transition to a higher spin in species
prone to forming low spin complexes with ligands such as
phenanthrolines and 2,2′-bipyridine.12a
Hence, the spin state of 1,2,3-triazole–pyridine-based
iron(II) complexes was explored theoretically and experi-
mentally in mononuclear compounds derived from ligands L1
and L1M. Furthermore, the experimental study of the di-
¶Elemental analysis of the powder obtained by mixing L3M or L4M and
[Fe(H2O)6](BF4)2 or [Ni(H2O)6](BF4)2 in a 3 : 2 ratio and washing excess ligand
confirmed the formation of species consistent with the [M2(LM)3](BF4)4 formula
(M = Fe2+, Ni2+).
This journal is © The Royal Society of Chemistry 2015
Dalton Trans.

View Article Online
Paper
Dalton Transactions
nuclear systems allows assessing the impact of the spacer, its the UV-vis data confirm the formation of a low-spin mono-
rigidity or induced relative chirality at the metal ion centers on nuclear complex in [Fe(L1)3](BF4)2. This is consistent with a
potential spin cross-over to be investigated.
DFT-derived 28.9 kcal mol−1 stabilization of the low-spin
complex vs. its high-spin counterpart with 1,2,3-triazolylpyri-
dine chelates where the pyridine position 6 is occupied by a
hydrogen. However, the introduction of a methyl group ortho
to the nitrogen of the pyridine is known to weaken the ligand
field. Experimentally, the [Fe(L1M)3](BF4)2 complex was indeed
shown to be high-spin at room temperature, as indicated by its
Electronic effects studied by UV-vis spectroscopy and DFT in
mononuclear complexes
As in iron(II) tris(2,2′-bipyridine) complexes, the tris(1,2,3-tri-
azolylpyridine) iron(II) complexes are intensely coloured (dark
orange to red) as quantified by a relatively strong absorption in
the visible region (λmax at 425 nm, Fig. 11a for [Fe(L1)3]2+). DFT
calculations confirm the metal-to-ligand charge transfer
nature of the bands in the visible region for a low-spin species
(Fig. 11b). In the DFT-calculated spectrum of the low spin
complex (Fig. 11b), absorption at the highest wavelength
(around 450 nm) involves a metal-to-pyridine charge transfer,
whereas a separate metal-to-triazole charge transfer is calcu-
lated to occur at higher energy (corresponding to a λmax
around 310 nm). In the experimental spectrum of [Fe(L1)3]-
(BF4)2 (Fig. 11a), these transitions are merged into a broad
series of absorptions over about 100 nm. Together with the
well-defined 1H NMR spectrum and X-ray structure at 180 K,
1
very broad H NMR spectrum over an extended chemical shift
range (ESI†) and by SQUID measurements (see below Fig. 13a).
The absorption of the yellow complex [Fe(L1M)3](BF4)2 also dis-
plays a shoulder between 300 and 320 nm consistent with the
DFT calculations of an iron(II) tris(pyridyl-1,2,3-triazol)
complex in the high-spin state (Fig. 11b).
Although efforts to crystallize [Fe(L1M)3](BF4)2 have been
unsuccessful, geometry optimization of [Fe(L1M′)3]2+ by DFT
offers some insight into its structure (in L1M′, a methyl group
replaces the benzyl unit). Among the four forms of high-spin,
low-spin, fac and mer [Fe(L1M′)3]2+ isomers, the mer complex is
significantly more stable (the calculated relative energy levels
are given in the ESI†), with the high-spin complex favoured by
more than 1.8 kcal mol−1 (ESI†). In the fac complexes, the sub-
stituent in position 6 of one pyridine points towards the center
of another pyridine ring. This is easily accommodated in an
unsubstituted pyridine, such as in [Fe(L1)3]2+ (H on pyridine
position 6) which displays a fac arrangement in the crystal
state (Fig. 12c).4g However, 6-methyl-substituted pyridines
suffer from significant steric congestion when all three methyl
groups are stacked against the adjacent pyridines in the fac tri-
methylated complex (mer [Fe(L1M′)3]2+ is represented in
Fig. 12a, and its fac isomer appears in Fig. 12b). Similar mer
selection has been observed in the crystal state of iron(II) tris-
(6-methyl-2,2′-bipyridine) which also displays high-spin
properties.12b
Low-spin and high-spin states in iron(II) dinuclear complexes
With the characteristics of the low and high-spin iron(II) com-
plexes in hand, it is interesting to explore the behaviour of the
dinuclear species and the role of the spacer on the magnetic
properties. Dinuclear iron(II) complexes have attracted signifi-
cant attention due to possible cooperativity/anti-cooperativity
in spin cross-over around each metal center,12c–h and the
Fig. 11 (a) Measured experimental and (b) TD-DFT-derived theoretical
UV-vis spectra of high-spin and low-spin iron(II) mononuclear com-
plexes. The experimental spectra in (a) were collected for the syn-
thesized [Fe(L1)3](BF4)2 and [Fe(L1M)3](BF4)2 complexes in acetonitrile
(4 × 10−5 M, 25 °C). DFT calculations for high-spin and low-spin com-
plexes were performed based on a monofunctional ligand similar to L1
except for the benzyl group which was replaced by a methyl group to
save calculation time. Under the spectra are represented the charge
density differences between the excited and the ground states (increase
in purple, decrease in orange; details in the ESI†).
Fig. 12 Comparative CPK representation of (a) DFT-optimized high-
spin mer [Fe(L1M’)3]2+, (b) DFT-optimized high-spin fac [Fe(L1M’)3]2+ and
(c) crystallographic low-spin fac [Fe(L1)3]2+ 4g
.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
ambiguously confirm SCO. In the literature, gradual and only
partial spin cross-over with full conversion above 300 K has
also been observed with mono-nuclear triazole-based iron(II)
complexes bearing two terdentate triazole–pyridine–pyrazole
ligands and tetrafluoroborate counterions.12l,m
The unique magnetic behaviour shown by [Fe2(L3)3](BF4)4
by SQUID correlates with a particular ageing effect of solutions
of this sample, as observed by both UV-vis and 1H NMR
spectroscopy. Fresh preparation of the triple-stranded complex
by mixing L3 (3 equiv.) with [Fe(H2O)6](BF4)2 (2 equiv.) in
Fig. 13 SQUID measurements for iron(II) complexes of (a) methyl-
(L1M, L3M, L4M) and (b) non-methyl-bearing (L3) ligands. [Fe2(L3)3]
(BF4)4 was measured in the crystalline state, while [Fe(L1M)3](BF4)2,
[Fe2(L3M)3](BF4)4 and [Fe2(L4M)3](BF4)4 were studied as powders.
1
acetonitrile gives well-resolved H NMR signals (Fig. 3) as well
as a UV-vis signature typical of a low spin complex with MLCT
character (ESI†). However, letting the solution age over 3 days
induces a change to very broad 1H NMR signals and loss of the
MLCT band at 425 nm in favour of a broad band around
350 nm, consistent with a dominant high-spin species (the
comparative data are represented in the ESI, Fig. S36,† and the
potential effect of guest binding on spin cross-over in these
architectures.12i–k Many of these dinuclear systems involve the
spin cross-over prone imidazole–imine chelate, and the mag-
netic behaviour of triazolyl-pyridine iron(II) complexes has not
been reported to our knowledge.
Like their mononuclear analogue, the dinuclear iron(II)
complexes derived from the methyl bearing ligands only show
high-spin properties in the solid state, irrespective of tempera-
ture (Fig. 13a).k The presence of the methyl group on the tight-
est binding element of the chelating unit dramatically
weakens the ligand field to the point where the low-spin
species is no longer accessible. This is reminiscent of the tran-
sition to an exclusive iron(II) high-spin state when one methyl
group is introduced in 2,2′-bipyridine ligands (leading to
6-methyl-2,2′-bipyridine),12b as well as in mono- and di-nuclear
benzimidazolylpyridines.12c
Most of the non-methylated ligands induce low-spin iron(II)
complexes as evidenced by their MLCT in UV-vis spectra, well-
resolved 1H NMR signature and SQUID measurement (L2c),
similarly to their mononuclear analogue [Fe(L1)3](BF4)2.
However, the propylene bridged [Fe2(L3)3](BF4)4 complex dis-
plays a unique behaviour with possible spin cross-over pro-
perties. Polycrystalline [Fe2(L3)3](BF4)4 samples were studied
by SQUID between 1.8 and 375 K (Fig. 13b illustrates the
2.5–300 K range). At 2.5 K, the χT product is 0.01 cm3 K mol–1
which corresponds to a low-spin FeII (S = 0). As the tempera-
ture increases above 50 K, the χT product increases gradually
to 1.03 cm3 K mol–1 at 300 K which would be consistent with a
gradual spin crossover from a diamagnetic low-spin to a high-
spin S = 2 FeII ion. It is noteworthy that the magnetic suscepti-
bility data from 2.5 to 300 K are identical to that measured
from 300 to 2.5 K with no thermal hysteresis. However, the full
transition cannot be observed in the temperature range investi-
gated, and no clear sigmoidal signature is observed. As a
result, SCO may be occurring at higher temperatures; yet,
access to higher temperatures would be necessary to un-
DFT calculated high-spin spectrum for
a mononuclear
complex model is displayed in Fig. 11b). The single crystals
grown over a week under argon for X-ray crystallography and
used for SQUID experiments also show the signature of the
1
aged complex when analysed in solution by H NMR and UV-
vis spectroscopy. The exact reasons why a transition to high-
spin in aged solutions and crystalline samples is observed are
still under investigation. None of the other bifunctional
ligands bridged by aliphatic spacers (L2a, L4, L5) showed the
1
same behaviour (no ageing effect as studied by H NMR and
UV-vis spectroscopy, ESI†).
Because spin cross-over relates to the strength of the ligand
field, the UV-vis analysis of the nickel(II) complexes, which
gives access to the 10Dq parameter in pseudo-octahedral com-
plexes, provides some insight into the possibility of SCO and
relative ligand field strengths. As apparent from Fig. 14a and
Table 1, all non-methyl-bearing ligands give nickel(II) com-
plexes with maximum absorption around 870 nm and 10Dq
(Ni2+) well within the range of 11 200–12 400 cm−1 where SCO
for Fe2+ species is considered accessible.12a The introduction
of a methyl group induces a significantly weaker field (10Dq
kAs discussed earlier in this manuscript and apparent from Fig. 3, iron(II) com-
plexes derived from L2a, L4 and L5 are composed of mixtures in solution. It was
also not possible to obtain single crystals from these complexes. As a result, they
were judged not pure enough to be submitted to magnetism studies by SQUID.
Fig. 14 NIR absorption of nickel(II) complexes of (a) L1, L3, L4 and L5
(2–5 mM), and (b) L1M, L3M, L4M (8 mM) in acetonitrile, 25 °C.
This journal is © The Royal Society of Chemistry 2015
Dalton Trans.

View Article Online
Paper
Dalton Transactions
Table 1 NIR absorption data for [Ni(L)3](BF4)2 (L = L1, L1M) and [Ni2(L)3](BF4)4 (L = L3, L4, L5, L3M, L4M) in acetonitrile (25 °C)
X
L1
L3
L4
L5
L1M
970a
L3M
L4M
λmax (nm)
875a
865b
870a
11 495c
870a
11 495c
960a
970a
10DqNi (cm−1
)
11 430c
11 560d
10 310e
10 420e
10 310e
a
d
e
10 nm. b 5 nm. c 130 cm−1
.
65 cm−1
.
110 cm−1
.
(Ni2+) ∼ 10.3 × 103 cm−1 for a 970 nm absorption, Fig. 14b, stranded [Fe2(L2c)3]4+ and double-stranded [Cu2(L2cM)2]2+
Table 1) consistent with the lack of spin cross-over.
outcome. 1H NMR confirms this hypothesis, as evidenced in
1
As seen from the very similar values of NIR absorption in Fig. 15. Fig. 15a shows the H NMR signature of a 2 : 2 L2cM/
nickel(II) complexes with hydrogen-bearing pyridines, the Cu+ mixture, corresponding to the [Cu2(L2cM)2](BF4)2 chemi-
unique possible SCO behaviour of [Fe2(L3)3](BF4)4 may not be cal formula determined by elemental analysis (ESI†). The sig-
simply explained by the sole effect of the ligand field. In the nature of [Fe2(L2c)3]4+ is repeated in Fig. 15b. The
solid state, weak interactions responsible for crystal packing of 3 : 2 : 2 : 2 mixture of L2c, L2cM, Fe2+ and Cu+ analysed in
[Fe2(L3)3](BF4)4 fully described in ref. 4g induce the closest Fig. 15c confirms the sole formation of the self-selected
lateral packing of all the crystallized dinuclear complexes (and [Fe2(L2c)3]4+ and [Cu2(L2cM)2]2+ complexes, as the spectrum of
shortest ion–ion distance) in all the resolved crystal structures the mixture is exactly the superimposition of that of the two
(9.9 Å in [Fe2(L3)3](BF4)4 vs. 11.4 Å in [Fe2(L2c)3](BF4)4 and individual architectures. Such discrimination provided by a
13.4 Å in [Fe2(L2d)3](BF4)4),4g and may provide a pathway for methyl group in the 6-position of a pyridine, which destabi-
the effective ligand field to be further tuned.
lizes the formation of octahedral complexes while leaving the
stability of tetrahedral geometries where the two chelate
ligands are orthogonal to each other unaffected, has pre-
cedents in the 2,2′-bipyridine helicate literature.13d
Selection of self-assembled
architectures
Spacer-strain effect: short vs. long spacer. In the example
above, the central spacer is identical in both L2c and L2cM,
ligands, and the end-strain was used to screen octahedral vs.
tetrahedral environments, as in the 2,2′-bipyridine family.
Double-stranded vs. triple-stranded systems
Self-sorting of self-assembled architectures from a mixture of However, the pyridyl–triazole unit offers an alternative handle
components has been a subject of intense studies,7 probably to screen triple-stranded architectures: as discussed in the
driven by the awe-inspiring self-assembly of countless above sections in gas and solution phases, the very short ethyl-
materials into cells.13a–c In this context, the pyridyl–triazole ene spacer in L2a is characterized by destabilized triple-
chelate, although a weaker ligand than 2,2′-bipyridine, also is stranded dinuclear iron(II) assemblies, likely resulting from
a candidate for specific selection of double-stranded vs. triple- the proximity of the triazole ‘middle’ nitrogen of each metal
stranded architectures. The selection process may be dictated ion center. This would be expected to be particularly sensitive
by:
with metal ions with a strictly octahedral geometry such as
- the metal ion’s geometrical preference,
- the ligand’s features including (i) the steric hindrance dis-
played at its end where coordination takes place (methylated
vs. non-methylated pyridine) and (ii) the particular properties
of the spacer (short spacer as in L2a vs. longer spacer as in
L2c). Both of these aspects are explored below.
End-strain effect: methylated vs. non-methylated pyridine.
As discussed above, the presence of a methyl group on posi-
tion 6 of the terminal pyridine limits the ability of the pyridyl–
triazole unit to form very stable octahedral complexes within
dinuclear structures. On the one hand, the presence of the
methyl group in L2cM for instance is not easily accommodated
in a rigid octahedral environment (octahedral iron(II)), but
would provide limited steric hindrance in the tetrahedral
context of a double-stranded species. On the other hand, non-
methylated ligand L2c which does not display steric hindrance
gives a very stable triple-stranded species in the presence of
Fe2+. As a result, self-selection would be expected in a mixture
Fig. 15 Self-selection of double-stranded copper(I) vs. triple-stranded
iron(II) architectures evidenced by 1H NMR of a 3 : 2 : 2 : 2 mixture of L2c,
L2cM, [Cu(CH3CN)4]BF4 and [Fe(H2O)6](BF4)2 (CD3CN, 25 °C, 500 MHz
of L2c, L2cM, Fe2+ and Cu+, giving exclusively the triple- for (a) and (c), 400 MHz for (b)).
Dalton Trans.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
iron(II). However, ions with a more flexible geometry, or even with L2a bridging ligands, as observed by ESI-MS with L2a
more variety in their coordination numbers, would likely well alone (Fig. 7 and 8), although more work would need to be
tolerate the L2a ligand. Therefore we explored the possible done to explore such hybrid structures. In summary, although
self-selection of triple-stranded dinuclear helicates with well- discrimination is not perfect under the explored conditions,
spaced pyridyl–triazole chelates (such as in the p-xylyl based this example shows that pyridyl–triazole ligands with spacers
L2c ligand) away from dinuclear architectures based on the of various lengths can be used to direct preferential self-assem-
more compact L2a ligand, using iron(II) as a source of a strictly bly and self-selection.
octahedral environment, and zinc(II) as a more flexible metal
ion. H NMR analysis of the 3 : 2 : 2 : 2 mixture of L2c/L2a/ZnII/
1
Double-stranded vs. loop architectures
FeII is represented in Fig. 16.
The above example uses the spacer length as the guiding infor-
Although the selection is not as dramatic as in the previous
section due to the slow precipitation of the self-assembled zinc
(II) complex leading to loss and broadening of signals, a few
features are worth highlighting. First, in the four component
mation for selection, with L2a and L2c which are both semi-
rigid strands. In another approach, we explored the effect of
the flexibility of the spacer in directing the selection of ‘loop’
vs. helicate architectures, in the presence of ions of flexible or
mixture (Fig. 16c), a major component appears to be the triple-
stricter geometrical preferences (e.g. Zn2+ vs. Cu+). Using L2c
as semi-rigid and L6 as semi-flexible ligands, respectively, we
wanted to see if any binding selectivity would be shown
towards Cu+ and Zn2+, knowing that the semi-rigid L6 ligand
accommodates the flexible zinc ion into a loop structure very
well (see above), and that copper(I)’s fairly strict tetrahedral
geometry is well tolerated by L2c in a 2 : 2 complex. This is an
intrinsically difficult challenge, as a mono-nuclear looped
structure comes from a simple bimolecular event, whereas the
formation of a dinuclear double-stranded structure involves
bringing four independent components together, and there-
fore is entropically much more costly. So not surprisingly the
selection process is not as clear cut as above, but it still brings
some valuable information highlighted below.
stranded p-xylyl-based [Fe2(L2c)3]4+ complex, as evidenced by
the aromatic and benzylic protons at chemical shifts identical
to those of pure [Fe2(L2c)3]4+ (Fig. 16d; red bands). In the four
component mixture (Fig. 16c), ligand L2a seems to be very
much involved in the same zinc complex as in a simple
2 : 2 mixture with Zn(II) (Fig. 16b, green bands), although
partial precipitation of this complex reduces its signals’ inten-
sity. Interestingly, the very complex and broad signature of a
2 : 2 mixture between Fe2+ and L2a alone (Fig. 16e, particularly
visible in the α CH2 and H5 regions; see Fig. 3 for labelling)
does not seem to be present, suggesting that there is indeed
segregation of Fe2+ with L2c. This self-selection process does
not lead to totally orthogonal binding of Fe2+ with L2c, and
Zn2+ with L2a, however, as a well-defined minor component
appears to be present, with small signals of what seems to
involve the L2c ligand (e.g. minor triazole proton at 8.76 ppm,
and minor diastereotopic protons in the 5.2–5.6 ppm region,
Fig. 16c). These may be related to a mixed hexameric assembly
When
a
1 : 1 : 1 : 1 solution of L6/L2c/Zn(OTf)2/Cu-
(CH3CN)4BF4 in acetonitrile was analysed by ESI-mass spectro-
metry (Fig. 17),** the most intense peaks indicate the selection
of zinc complexes with the loop ligand L6 only ([ZnL6]2+ ion at
m/z 235.06 and the 1+ triflate species at 619.07, Fig. 17, con-
firmed by the isotopic pattern††). This is consistent with the
very stable 1 : 1 Zn/L6 complex observed by UV-vis titrations
with L6 alone (see above and the ESI†). Maybe not surprisingly,
the loop complex with Cu+ is also present in the mixture as
indicated by the peak at 469.11 corresponding to [CuL6]+.
Indeed, the low entropic cost for the formation of a 1 + 1
complex vs. a 2 + 2 dinuclear assembly combined likely over-
rides the enthalpic cost associated with the steric strain that
the semi-flexible loop L6 provides to the strictly tetrahedral
Cu+ ions. So overall it is not surprising to detect the [CuL6]+
ion in the gas phase. Interestingly, only the copper(I)
xylyl-based complexes appear at significant levels, with signals
at 457.09 (2+), 1001.20 and 1065.14 (1+ with BF4 and OTf coun-
terions, respectively). A Cu+ ‘corner’ involving two xylyl-based
ligands L2c is also present with a signal at 851.26 (Fig. 17a).
Fig. 16 Self-selection of zinc(II) vs. iron(II) architectures based on the
spacer length explored by 1H NMR of a 3 : 2 : 2 : 2 mixture of L2c, L2a, Zn
(OTf)2 and [Fe(H2O)6](BF4)2 (CD3CN, 25 °C, 500 MHz, [L] ∼20–30 mM):
(a) L2a alone; (b) L2a signals after addition of one equivalent of Zn(OTf)2
(with respect to L2a); (c) after further addition of L2c (1.5 equiv.) and
[Fe(H2O)6](BF4)2 (1 equiv.). For comparison, 1H NMR of [Fe2(L2c)3]4+
(400 MHz, (d), and 3 : 2 mixtures of L2a and [Fe(H2O)6](BF4)2 (300 MHz,
(e) are given in CD3CN (in e, asterisks signal what was identified as the
triple-stranded [Fe2(L2a)3]4+ complexes; see Fig. 3).
**Because of intermediate exchange rates often observed for zinc complexes
involving the loop ligand as well as the low solubility of the copper(I) complexes
in acetonitrile, mixtures were only analyzed by ESI mass-spectrometry.
††Isotopic patterns proved very useful to determine not only the charge of the
ion for each peak, but also to confirm the chemical nature of the ion (distinct Zn
vs. Cu patterns) as well as nuclearity (distinct patterns for mono-nuclear and di-
nuclear complexes).
This journal is © The Royal Society of Chemistry 2015
Dalton Trans.

View Article Online
Paper
Dalton Transactions
Conclusions
Although the pyridyl–triazole diad may in first approximation
be seen as a synthetically convenient alternative to the classical
2,2′-bipyridine coordination chelate, the systematic study
described in this manuscript highlights its own original fea-
tures. More specifically:
(1) The particular orientation of the nitrogen N2 lone-pairs
of the 1,2,3-triazole unit, pointing towards the cavity of triple-
stranded dinuclear complexes, seems to destabilize these par-
ticular architectures when small ethylene spacers are involved,
and populate larger architectures. Such spacer sensitivity dis-
tinguishes the pyridyl–triazole diad from the 2,2′-bipyridine
chelate. In this context, the intriguing cyclic hexamers
detected by mass spectrometry are certainly worthy of further
exploration, as they may allow access to increased complexity
in the self-assembly process (e.g. selection of architectures
with mixed ligands and mixed metal ions). Preliminary
examples of self-selection are already emerging.
(2) Structurally speaking, semi-rigid bifunctional ligands
yield triple-stranded dinuclear structures with iron(II) and
nickel(II) under non-competitive conditions (non-coordinating
anions), and follow the ‘even/odd’ helicate/mesocate rule
observed with bipyridine and catechol chelates. However, the
flexibility of the triazole-spacer hinge is such that both helicate
and mesocate structures start to be nearly equally populated as
early as in the pentylene spacer.
(3) Due to its weaker ligand field, the pyridyl–triazole diad
may open the door to spin-cross over properties, controlled by
the chemical nature and length of the spacer. The propylene
spacer is of particular interest, and further studies need to be
conducted to fully understand the details of its magnetic be-
haviour, for instance its ageing effect and slow magnetic con-
version. With regards to the latter point, replacing the
tetrafluoroborate with a perchlorate was shown in the literature
to transform a gradual, high-temperature spin cross-over into a
complete, lower temperature conversion of iron(II) complexes
bound to triazole-involving terdentate ligands,12l so exploring
counterion effects on iron(II) complexes of all the non-methyl-
ated ligand family would be the first measure to take.
Fig. 17 Selection of double-stranded Cu+ dinuclear complexes vs. Zn2+
and Cu+ metallo-loops observed by ESI mass spectrometry. Black lines
represent oligoether ligand L6. Grey lines with a bulge represent the p-
xylyl spacer in L2c ligands. (a) Full spectrum, (b) zoom on the region
where 2 : 2 and 2 : 3 zinc(II)/L2c species would appear. In the boxes, ‘X’
stands for ‘counteranion’ which is then specified at the bottom left and
right of each box.
One could argue that the zinc complexes with L2c may just not
be observed by mass spectrometry due to lower stability.
However, a close inspection of the area where the 2 : 2 and 2 : 3
zinc(II)/L2c species would appear leads to the detection of
several expected zinc complexes, although at very low abun-
dance, showing that they do withstand ESI-MS analysis, but
are very minor species. All the peaks of very low abundance in
the dinuclear zinc complex region could be assigned on the
basis of the m/z values and isotopic patterns. A zoomed spec-
trum on the relevant area with 100 times magnification com-
pared to Fig. 17a is represented in Fig. 17b, and shows some
interesting features. First, very minute amounts of triple-
stranded zinc complexes are present, in an all-L2c homoleptic
[Zn2(L2c)3]4+or heteroleptic [Zn2(L2c)2(L6)]4+ format. At least
two semi-rigid L2c are required to form this type of triple-
stranded dinuclear complex. Among these very minor pro-
ducts, double-stranded mixed complexes (one Cu+ and one
Zn2+) including at least one semi-rigid L2c ligand are more
abundant. Finally, what was assigned as a loop dimer bridged
by triflates is also detected (this interpretation is inspired by
zinc’s intrinsic oxophilicity as well as by the coordination pro-
perties of L6 with Fe2+, as seen above, Fig. 6).
Overall, with such convenient synthetic access and rich
physical and chemical properties, the pyridyl–triazole diad
embedded in polyfunctional ligands is certainly only begin-
ning to reveal its potential.
Acknowledgements
The authors thank Prof. Louis Cuccia for his precious help
with and access to the VPO equipment, and the reviewers for
their time and attention provided to this long manuscript, as
well as their suggestions. The authors are also indebted to
Dr Gabriele Schatte for her assistance in the analysis of
X-ray diffraction data. Funding from the Natural Sciences
and Engineering Research Council of Canada (USRA for
N. W. and D. G. for A. P.), French Région Rhône-Alpes
Dalton Trans.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
(Explo’ra Sup scholarship for C. F. M.), the Ontario Ministry of
Innovation and Research, and Queen’s University is
acknowledged.
7 (a) M. M. Safont-Sempere, G. Fernández and F. Würthner,
Chem. Rev., 2011, 111, 5784; (b) M. L. Saha and
M. Schmittel, Org. Biomol. Chem., 2012, 10, 4651.
8 (a) M. Albretch and S. Kotila, Angew. Chem., Int. Ed. Engl.,
1995, 34, 2134; (b) N. Ousaka, Y. Takeyama and
E. Yashima, Chem. Sci., 2012, 3, 466; (c) J. Xu, T. N. Parac
and K. N. Raymond, Angew. Chem., Int. Ed., 1999, 38, 2878.
9 B. R. Serr, K. A. Andersen, C. M. Elliott and O. P. Anderson,
Inorg. Chem., 1988, 27, 4499.
10 (a) E. T. Adams, P. J. Wan Jr. and E. F. Crawford, Methods
Enzymol., 1978, 48, 69; (b) E. T. Adams Jr., J. Lipid Res.,
1986, 27, 1233.
Notes and references
1 (a) H. C. Kolb, M. G. Finn and K. B. Sharpless, Angew.
Chem., Int. Ed., 2001, 40, 2004; (b) H.-F. Chow, K.-N. Lau,
Z. Ke, Y. Liang and C.-M. Lo, Chem. Commun., 2010, 46,
3437; (c) Y. Hua and A. H. Flood, Chem. Soc. Rev., 2010, 39,
1262.
11 D. Alagille, R. M. Baldwin, B. L. Roth, J. T. Wroblewski,
E. Grajkowska and G. D. Tamagnan, Bioorg. Med. Chem.,
2005, 13, 197.
2 (a) R. M. Meudtner and S. Hecht, Angew. Chem., Int. Ed.,
2008, 47, 4926; (b) H. Juwarker, J. M. Lenhardt,
12 (a) P. Gütlich and H. A. Goodwin, Spin Crossover in Tran-
sition Metal Compounds I, in Topics in Current Chemistry,
Springer-Verlag, Berlin, Heidelberg, 2004, ch. 3;
(b) D. Onggo, J. M. Hook, A. D. Rae and H. A. Goodwin,
Inorg. Chim. Acta, 1990, 173, 19; (c) L. J. Charbonnière,
A. F. Williams, C. Piguet, G. Bernardinelli and E. Rivara-
Minten, Chem. – Eur. J., 1998, 4, 485; (d) F. Tuna,
M. R. Lees, G. J. Clarkson and M. J. Hannon, Chem. – Eur.
J., 2004, 10, 5737; (e) M. H. Klingele, B. Moubaraki,
J. D. Cashion, K. S. Murray and S. Brooker, Chem.
Commun., 2005, 987; (f) Y. Garcia, C. M. Grunert,
S. Reiman, O. van Campenhoudt and P. Gütlich,
Eur. J. Inorg. Chem., 2006, 3333; (g) D. Pelleteret, R. Clérac,
C. Mathonière, E. Harté, W. Schmitt and P. E. Kruger,
Chem. Commun., 2009, 221; (h) Y. Sunatsuki, R. Kawamoto,
K. Fujita, H. Maruyama, T. Suzuki, H. Ishida, M. Kojima,
S. Iijima and N. Matsumoto, Inorg. Chem., 2009, 48, 8784;
(i) J. J. M. Amoore, C. J. Kepert, J. D. Cashion,
B. Moubaraki, S. M. Neville and K. S. Murray, Chem. – Eur.
J., 2006, 12, 8220; ( j) J. J. M. Amoore, S. M. Neville,
B. Moubaraki, S. S. Iremonger, K. S. Murray, J.-F. Létard
J.
C.
Castillo,
E.
Zhao,
S.
Krishnamurthy,
R. M. Jamiolkowski, K.-H. Kim and S. L. Craig, J. Org.
Chem., 2009, 74, 8924; (c) Y. Hua and A. H. Flood, J. Am.
Chem. Soc., 2010, 132, 12838.
3 (a) H. Struthers, T. L. Mindt and R. Schibli, Dalton Trans.,
2010, 39, 675; (b) D. Schweinfurth, R. Pattacinin, S. Strobel
and B. Sarkar, Dalton Trans., 2009, 9291; (c) D. Urankar,
B. Pinter, A. Pevec, F. De Proft, I. Turel and J. Košmrlj,
Inorg. Chem., 2010, 49, 4820; (d) A. M. Najar, I. S. Tidmarsh
and M. D. Ward, CrystEngComm, 2010, 12, 3642;
(e) J. D. Crowley and D. A. McMorran, Top. Heterocycl.
Chem., 2012, 28, 31.
4 (a) J. D. Crowley and P. H. Bandeen, Dalton Trans., 2010, 39,
612; (b) O. Fleischel, N. Wu and A. Petitjean, Chem.
Commun., 2010, 46, 8454; (c) S. O. Scott, E. L. Gavey,
S. J. Lind, K. C. Gordon and J. D. Crowley, Dalton Trans.,
2011, 40, 12117; (d) J. T. Fletcher, B. J. Bumgarner,
N. D. Engels and D. A. Skoglund, Organometallics, 2008, 27,
5430; (e) B. Happ, D. Escudero, M. D. Hager, C. Friebe,
A. Winter, H. Görls, E. Altuntas, L. González and
U. S. Schubert, J. Org. Chem., 2010, 75, 4025; (f) G. Aromí,
L. A. Barrios, O. Roubeau and P. Gamez, Coord. Chem. Rev.,
2011, 255, 485; (g) K. A. Stevenson, C. F. C. Melan,
O. Fleischel, R. Wang and A. Petitjean, Cryst. Growth Des.,
2012, 12, 5169; (h) U. R. Pokharel, F. R. Fronczek and
A. W. Maverick, Dalton Trans., 2013, 42, 14064.
5 (a) C. Piguet, G. Bernardinelli and G. Hopfgartner, Chem.
Rev., 1997, 97, 2005; (b) M. Albrecht, Chem. – Eur. J., 2000,
6, 3486; (c) M. Albrecht, Chem. Rev., 2001, 101, 3457;
(d) M. J. Hannon and L. J. Childs, Supramol. Chem., 2004,
16, 7; (e) M. Albrecht and R. Froehlich, Bull. Chem. Soc.
Jpn., 2007, 80, 797.
6 (a) S. K. Vellas, J. E. M. Lewis, M. Shankar, A. Sagatova,
J. D. A. Tyndall, B. C. Monk, C. M. Fitchett, L. R. Hanton
and J. D. Crowley, Molecules, 2013, 18, 6383;
(b) S. V. Kumar, W. K. C. Lo, H. J. L. Brooks and
J. D. Crowley, Inorg. Chim. Acta, 2015, 425, 1; (c) B. Akhuli,
L. Cera, B. Jana, S. Saha, C. A. Schalley and P. Ghosh, Inorg.
Chem., 2015, 54, 4231.
and C. J. Kepert, Chem.
– Eur. J., 2010, 16, 1973;
(k) R. J. Archer, C. S. Hawes, G. N. L. Jameson, V. McKee,
B. Moubaraki, N. F. Chilton, K. S. Murray, W. Schmitt and
P. E. Kruger, Dalton Trans., 2011, 40, 12368;
(l) N. Chandrasekhar and R. Chandrasekar, Dalton Trans.,
2010, 39, 987; (m) C. Cook, F. Habib, T. Aharen, R. Clérac,
A. Hu and M. Murugesu, Inorg. Chem., 2013, 52, 1825.
13 For examples of cell-related self-sorting, see: (a) D. Voet
and J. D. Voet, Biochemistry, Wiley, New York, 2011, chap-
ters on nucleic acids (29) and on viruses (33), and an inter-
esting commentary in (b) S. Ventegodt, T. D. Hermansen,
T. Flensbord-Madsen, M. L. Nielsen, B. Clausen and
J. Merrick, Scientific World Journal, 2006, 6, 1132. For an
example on the self-organization of cells into tissues
(review in neurology), see: (c) M. Karus, S. Blaess and
O. Brüstle, J. Comp. Neurol., 2014, 522, 2831. (d) R. Krämer,
J.-M. Lehn and A. Marquis-Rigault, Proc. Natl. Acad.
Sci. U. S. A., 1993, 90, 5394.
This journal is © The Royal Society of Chemistry 2015
Dalton Trans.