Deciphering the Influence of Meridional versus Facial Isomers in Spin Crossover Complexes

Article abstract of DOI:10.1002/chem.201804161

Chelate coordination of non-symmetrical didentate pyrazine-benzimidazole (L1) or pyridine-benzimidazole (L2) N-donor ligands around divalent iron in acetonitrile produces stable homoleptic triple-helical spin crossover [Fe(Lk)₃]²⁺ complexes existing as mixtures of meridional (C₁-symmetry) and facial (C₃-symmetry) isomers in slow exchange on the NMR timescale. The speciation deviates from the expected statistical ratio mer/fac=3:1, a trend assigned to the thermodynamic trans-influence, combined with solvation effects. Consequently, the observed spin state Fe^II_low-spin?Fe^II_high-spin equilibria occurring in [Fe(Lk)₃]²⁺ refer to mixtures of complexes in solution, an issue usually not considered in this field, but which limits rational structure-properties correlations. Taking advantage of the selective and quantitative formation of isostructural facial isomers in non-constrained related spin crossover d-f helicates (HHH)-[LnFe(Lk)₃]⁵⁺ (Ln is a trivalent lanthanide, Lk=L5, L6), we propose a novel strategy for assigning pertinent thermodynamic driving forces to each spin crossover triple-helical isomer. The different enthalpic contributions to the spin state equilibrium found in mer-[Fe(Lk)₃]²⁺ and fac-[Fe(Lk)₃]²⁺ reflect the Fe?N bond strengths dictated by the trans-influence, whereas a concomitant solvent-based entropic contribution reinforces the latter effect and results in systematic shifts of the spin crossover transitions toward higher temperature in the facial isomers.

Full text of DOI:10.1002/chem.201804161

A Journal of
Accepted Article
Title: Deciphering the Influence of Meridional vs Facial Isomers in Spin
Crossover Complexes
Authors: Claude Piguet, Timothée Lathion, Laure Guénée, Céline
Besnard, and Azzedine Bousseksou
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.201804161
Link to VoR: http://dx.doi.org/10.1002/chem.201804161
Supported by

Chemistry - A European Journal
10.1002/chem.201804161
1
Deciphering the Influence of Meridional vs Facial Isomers in Spin Crossover
Complexes.
Timothée Lathion,[a] Laure Guénée,[b] Céline Besnard,[b] Azzedine Bousseksou[c] and Claude
Piguet*[a]
_
_________________________________________________________________
[
a] Dr T. Lathion, Prof. Dr C. Piguet. Department of Inorganic and Analytical Chemistry,
University of Geneva, 30 quai E. Ansermet, CH-1211 Geneva 4 (Switzerland). E-mail:
Claude.Piguet@unige.ch
[
b] Dr L. Guénée, Dr C. Besnard. Laboratory of Crystallography, University of Geneva, 24 quai
E. Ansermet, CH-1211 Geneva 4 (Switzerland)
[
c] Dr A. Bousseksou. Laboratory of Coordination Chemistry (LCC), CNRS & Université de
Toulouse (UPS, INP), 205 route de Narbonne, Toulouse 31077 Cedex 4 (France).
Supporting information and the ORCID identification number(s) for the author(s) of this article can
be found under https://doi.org/10.1002/chemxxxxx
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
2
Abstract
Chelate coordination of non-symmetrical didentate pyrazine-benzimidazole (L1) or pyridine-
benzimidazole (L2) N-donor ligands around divalent iron in acetonitrile produces stable homoleptic
2
+
triple-helical spin crossover [Fe(Lk)
3
] complexes existing as mixtures of meridional (C -symmetry)
1
and facial (C -symmetry) isomers in slow exchange on the NMR time scale. The speciation deviates
3
from the expected statistical ratio mer/fac = 3:1, a trend assigned to the thermodynamic trans-
influence combined with solvation effects. Consequently, the observed spin state FeIIlow-spin ↔ FeIIhigh-
2
+
spin equilibria occurring in [Fe(Lk) ] refer to mixtures of complexes in solution, an issue usually not
3
considered in this field, but which limits rational structure-properties correlations. Taking advantage
of the selective and quantitative formation of isostructural facial isomers in non-constrained related
5
+
spin crossover d-f helicates (HHH)-[LnFe(Lk) ] (Ln is a trivalent lanthanide, Lk = L5, L6), we
3
propose a novel strategy for assigning pertinent thermodynamic driving forces to each spin crossover
triple-helical isomer. The different enthalpic contributions to the spin state equilibrum found in mer-
] and fac-[Fe(Lk)
]2+ reflect the Fe-N bond strengths dictated by the trans-influence, while
2
+
[Fe(Lk)
3
3
a concomitant solvent-based entropic contribution reinforces the latter effect and results in systematic
shifts of the spin crossover transitions toward higher temperature in the facial isomers.
Introduction
[
1]
4
Almost nine decades after the discovery of the spin state equilibrium phenomenon, a plethora of d -
d7 complexes has been identified for their spin crossover properties, which can be induced by thermal,
pressure, optical or pH stimuli.[2] Among them, the d pseudo-octahedral six-coordinate [Fe N
]
6
II
6
[
3]
chromophore found in [FeL
3
] (L is a didentate N-donor ligand) and in [Fe(L’)
2
] (L’ is a tridentate
N-donor ligand) plays a central role.[4] Taking pure octahedral symmetry (O
) as a guide for [FeN ]
h
6
II
entities, the electronic ground state of the Fe center ‘hesitates’ between low-spin (ls) and high-spin
(hs) configurations depending on the balance between the magnitude of the ligand-field strength oct
and that of the electron spin pairing energy P (Scheme 1).
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
3
II
II
[
Fe N ]
[Fe N ]
6
6
O (low-spin)
O (high-spin)
h
h
Free-FeII
3
/5oct
R3
3/5oct
-2/5oct
-2/5oct
5D
1A1
5T2
O
ls
R
Oh
hs
R3
h
3
E  E  2P 12 5
E  E  2 5
oct
oct
O
hs
O
ls
h
h
E  E  2

oct  P

II
Scheme 1. The thermodynamic origin of Fe spin state equilibria in octahedral symmetry (P is the
electron spin pairing energy).
[
5]
Since both the spectrophotometric series (which controls oct) and the nephelauxetic series (which
[
6]
controls P via the Racah parameters B and C) depend on the strength and on the nature of the metal-
ligand interactions, it is not so surprising that, for [FeN ] units, the coordination sphere of which is
6
systematically made up of six heterocyclic nitrogen donor, some simple metric parameters were found
to be valuable reporters for predicting the emergence of either pure low-spin state (
O
hs
Oh
ls
O
ls
Oh
hs
O
h
hs
Oh
ls
h
h
E  E  RT ), pure high-spin state ( E  E  RT ) or spin state equilibria ( E  E





RT, Eq. (1)).[7]
II
KSCO
II


GSCO /RT

 eSSCO /RHSCO /RT



KSCO  xhs xls  e
Fe
  Fe
(1)
low-spin
high-spin
2
+
In the famous family of homoleptic tris-diimine complexes [FeL
3
] , the interatomic distance
d(NN) between the two N-donor atoms in the non-coordinated didentate diimine ligands proved to
[
7]
be the searched indicator for predicting spin crossover (SCO) behaviors. Values of d(NN) larger
than 2.91 Å, as found in non-constrained didentate ligands made up of connected five-membered
ring/five-membered ring (for instance 2,2’-biimidazole), produce loose Fe-N bonding and high-spin
ground states. On the contrary, d(NN) distances smaller than 2.75 Å as found in 2,2’-bipyridine,
the archetype of two connected six-membered ring/six-membered ring, strongly stabilize low-spin
configurations. Spin crossover behaviors match the intermediate regime (2.75 ≤ d(NN) ≤ 2.91 Å)
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
4
provided by non-symmetrical didentate ligands made up of connected five-membered ring/six-
[
7]
membered ring as found in 2,2’-pyridine-imidazole or derivatives of it. Taking now for granted that
favorable metric parameters are in hand, some further tuning of the thermodynamic SCO parameters
(
Eq. (1)) results from the modulation of specific -donating (mainly affecting oct) and/or -accepting
characteristics (mainly affecting the spin pairing energy P via the Racah parameter B) encoded in the
bound heterocyclic rings. Reminiscent of Tanabe-Sugano diagrams,[8] the trivial thermodynamic
II
analysis summarized in Scheme 1 implies that the spin crossover properties in Fe complexes depend
on the oct/B ratio where (i) large values favor low-spin configurations, (ii) small values stabilize
high-spin states and (iii) intermediate values induce SCO behaviors. A precise and rational correlation
between molecular ligand structures and thermodynamic HSCO, SSCO parameters (especially the
equilibrium temperature T1/2 = HSCO/SSCO) thus goes through the oct/B ratio, a crucial strategy for
the rational programming of accessible spin state transitions, which can be implemented in
magnetic[9] and/or optical[10] (nano)molecular switches for practical applications. With this goal in
mind, major efforts were focused on the systematic synthesis of large collections of didentate and
tridentate polyimine binding units with variable coordinating properties for tuning the SCO transition
temperatures and thermodynamic cooperativities in [FeL
3
]2+ and [Fe(L’)
2
]2+ complexes.[2]-[4],[9],[10]
II
Although unsymmetrical didentate ligands have been often reacted with Fe to give triple-helical
2
+
homoleptic [FeL ] complexes, the influence of standard meridional and facial isomerization on the
3
spin state equilibrium mainly escaped attention and interest (Figure 1). Only a few salts containing a
single isomer could be isolated in the solid state (meridional or facial as certified by their X-ray crystal
structures),[11] but the associated SCO parameters did not give any clear evidences or trends, probably
because intermolecular packing interactions responsible for the stabilization of the crystal structures
largely dominate minor molecular effects. To the best of our knowledge, no attempt was made to
address the meridional/facial challenge in solution where the limited solvation effects offer the
opportunity to establish direct links between molecular geometries and SCO properties.
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
5
Figure 1. a) Perspective view along the helical axis for the meridionalfacial isomerization
processes (flip of the yellow ligand) operating in homoleptic [M(L2a)
]z+ complex where L2a is the
3
archetype of a non-symmetrical didentate ligand (see Scheme 2) and M is a pseudo-octahedral six-
coordinate metal center. b) Schematic views of the two enantiomeric pairs of mer/fac geometrical
isomers.
Pioneer attempts to address this challenge aim at connecting three unsymmetrical didentate units to a
short hydrocarbon-containing tripod, thus producing sexidentate ligands L”, which force the
2
+
[4d],[7],[12]
formation of the facial isomer fac-[FeL”] as the unique complex in solution.
However, this
strategy induces drastic intramolecular thermodynamic constraints during the formation of the [FeN
6
]
chromophore, an effect which can be quantified with the help of the concept of effective molarity.[13]
Applied to related podands bound to more flexible trivalent lanthanides, free-energy penalties as large
as 70-80 kJ/mol are common for the preorganization of three unsymmetrical binding units by short
to medium-size tripods. Compared with a total binding energy of 110-120 kJ/mol accounting for pure
intermolecular processes, this facial organization in a single tripodal ligand corresponds to c.a. 60%
reduction of the affinity for the entering metal.[14] One therefore understands why such constrained
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
6
tripodal complexes were not further considered as fair models for extracting the SCO properties of
]2+ isomers. However, some recent progresses made in the
analogous, but unconstrained fac-[FeL
3
understanding of the thermodynamic trans-influence occuring in simple mononuclear mer/fac-
]2+ isomers (Lk= L1, L2; Scheme 2 top)[15] combined with the design of flexible non-
[Zn(Lk)
3
5
+
covalent tripods displaying minor intramolecular cost (≤ 20 kJ/mol) in HHH-[ZnLn(Lk)
3
] helicates
(
Lk= L5, L6; Scheme 2 bottom),[16] reactivate the search for a rational treatment of the influence of
] ↔ fac-[FeL
]2+ isomerization on the SCO properties of these complexes in
2
+
molecular mer-[FeL
3
3
solution. We report here on the discovery that the meridional and facial isomers, due to their structural
differences indeed display synergestic enthalpic and entropic contributions, which may shift the spin
crossover transition temperature T1/2 by up to 60 degrees. A thermodynamic analysis assigns this
effect to trans-influence and to changes in (dipole-based) solvation energies accompanying the
change in volume produced by the SCO reaction.
N
N
N
N
N
N
N
R
R
R = CH : L1a
L2a
L2b
L3
L4
3
R = C H : L1b
2
5
L5
L6
L7
Scheme 2. Chemical structures of ligands L1-L7 considered in this work.
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
7
Experimental section
Chemicals were purchased from Sigma-Aldrich and Acros and used without further purification
unless otherwise stated. The didentate ligands L1,[15] L2,[15] L3,[17] and segmental ligands L5,[15] L6
[
18]
and L7 [19] were prepared according to literature procedures. Dichloromethane, diethyl ether and
N,N-dimethylformamide were dried through an alumina cartridge. Silica-gel plates (Merck, 60 F254
)
®
were used for thin-layer chromatography, SiliaFlash silica gel P60 (0.04-0.063 mm,) and Acros
silica gel 60 (0.035-0.07 mm) were used for preparative column chromatography.
Preparation of N-methyl-2-nitroaniline (1). 1-chloro-2-nitrobenzene (31.75 g, 201.5 mmol, 1.0 eq)
and methylamine (198 mL, 40% wt in H
tube equipped with a magnetic stirrer, and heated at 120 °C for 48 h. Excess of methylamine was
rotatory evaporated and the residual brown oil was partitioned between CH Cl (300 mL) and half
sat. aq. NH Cl (300 mL). The organic layer was separated and the aq. phase was further extracted
with CH Cl (3 x 150 mL). The combined organic extracts were dried over anhydrous Na SO
filtered and the solvent evaporated to dryness. The resulting red oil was purified by column
2
O, 2295.6 mmol, 11.4 eq) were introduced into a Carius
2
2
4
2
2
2
4
,
chromatography (silica, CH
2
Cl
2
) to give 28.99 g of N-methyl-2-nitroaniline (1, 190.5 mmol, yield
1
9
2
=
4.5 %) as a deep red orange oil, which slowly crystallized within hours. H NMR (CDCl
3
, 400 MHz,
3
4
3
3
98 K) δ ppm : 8.13 (1H, dd, J = 8.8 Hz, J = 1.6 Hz), 8.00 (1H, bs), 7.43 (1H, ddd, J = 8.8 Hz, J
4
3
4
3
3
7.2 Hz, J = 1.6 Hz), 6.81 (1H, dd, J = 8.6 Hz, J = 1.0 Hz), 6.62 (1H, ddd, J = 8.4 Hz, J = 7.2
4
Hz, J = 1.2 Hz), 2.99 (3H, s).
Preparation of 3-methyl-picolinic acid (2). 3-methyl-picolinonitrile (1.53 g, 13.0 mmol, 1.0 eq) in
aq. 5M NaOH (5 mL) were refluxed for 4 h. After cooling, the mixture was neutralized (pH = 2.0)
with concentrated hydrochloric acid (12 M) and evaporated to dryness. The residual solid was
suspended in ethyl acetate (250 mL), refluxed for 3 h and filtered when hot through a Büchner funnel.
The filtrate was evaporated to dryness to give 1.61 g of 3-methyl-picolinic acid (2, 11.7 mmol, yield
1
, 400 MHz, 298 K) δ ppm: 9.97 (1H, bs), 8.46 (1H, dd, 3J
9
1 %) as a white powder. H NMR (CDCl
3
4
3
4
3
3
=
4.8 Hz, J = 0.8 Hz), 7.72 (1H, dd, J = 7.8 Hz, J = 0.6 Hz), 7.48 (1H, dd, J = 7.8 Hz, J = 4.6
1
3
Hz), 2.78 (3H, s). RMN H ((CD ) SO, 400 MHz, 298 K) δ ppm: 13.07 (1H, s), 8.46 (1H, dd, J =
3
2
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
8
4
3
3
3
4
.6 Hz, J = 1.0 Hz), 7.77 (1H, d, J = 8.0 Hz), 7.47 (1H, dd, J = 7.8 Hz, J = 4.6 Hz), 2.46 (3H, s).
13C NMR (CDCl
37.67 (C ), 129.79 (CH), 20.30 (CH
, 100 MHz, 298 K) δ ppm: 164.40 (C
), 145.32 (CH), 143.39 (C ), 141.80 (CH),
3
q
q
1
q
3
).
Preparation of N,3-dimethyl-N-(2-nitrophenyl)picolinamide (3). Thionyl chloride (4.8 mL, 66.1
mmol, 6.0 eq) were dropwise added to a mixture of 3-methyl-picolinic acid (2, 1.51 g, 11.0 mmol,
1
.0 eq) and N,N-dimethylformamide (200 l) in dry 1,2-dichloroethane (10 mL). The solution was
refluxed for 4 h under an inert atmosphere, then cooled and evaporated to dryness. The residue was
dissolved in dry CH Cl (15 mL) and a solution of N-methyl-2-nitroaniline (1, 4.41 g, 27.2 mmol, 2.5
eq) in dry CH Cl (10 mL) was added dropwise. The final mixture was refluxed for 14 h during which
two portions of (iPr) NEt (3 mL, 17.2 mmol, 1.6 eq) were added after 20 min and 12 h, respectively.
Evaporation to dryness provided a solid, which was partitioned between CH Cl (200 mL) and half
sat. aq. NH Cl (200 mL). The organic layer was separated and the aq. phase was further extracted
with CH Cl (3 x 120 mL). The combined organic extracts were dried over anhydrous Na SO
2
2
2
2
2
2
2
4
2
2
2
4
,
filtered and the solvent evaporated to dryness to give 2.38 g of N,3-dimethyl-N-(2-
1
nitrophenyl)picolinamide (3, 8.8 mmol, yield 80 %) as a brownish solid. H NMR (CDCl
3
, 400 MHz,
2
98 K) δ ppm: mixture of rotamers A (2/3) et B (1/3): 8.50-7.50 (4H, m), 7.48-6.95 (3H, m), 3.56
(
3H, s, A), 3.28 (3H, s, B), 2.44 (3H, s, B), 2.38 (3H, s, A). ESI-MS (soft positive mode, m/z): [3+H]+:
+
272.1 (exp.), 272.3 (calc.); [(3)
2
+H] : 543.8 (exp.), 543.6 (calc.).
Preparation of ligand L4. N,3-dimethyl-N-(2-nitrophenyl)picolinamide (3, 2.38 g, 8.8 mmol, 1.0
eq) and activated powdered iron (9.77 g, 174.9 mmol, 19.9 eq) were refluxed for 24 h in a mixture of
ethanol (400 mL), water (115 mL) and concentrated hydrochloric acid (37 % in H O, 25 mL, 294.3
2
mmol, 33.5 eq). Excess iron was removed with a magnet and ethanol was evaporated. Water (50 mL)
and EDTA disodium dihydrate (65.3 g, 175.4 mmol, 20.0 eq) in water (250 mL) were successively
added to the solution and its pH was adjusted to 9.5 with concentrated aq. ammonia. CH
2
Cl
2
(300
mL) was finally added together with concentrated hydrogen peroxide (30 % in H O, 5 mL, 44.1
2
mmol, 5.0 eq) in small portions under vigorous stirring. The organic layer was separated and the aq.
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
9
phase was further extracted with CH
2
Cl
2
(3 x 300 mL). The combined organic extracts were dried
over anhydrous Na
2
SO , filtered and the solvent evaporated to dryness. The solid residue was purified
4
by column chromatography (silica, CH
2
Cl /MeOH 98:2) to give 1.45 g of L4 (6.5 mmol, yield 74 %)
2
1
3
4
as a cream powder. H NMR (CDCl
3
, 400 MHz, 298 K) δ ppm: 8.58 (1H, dd, J = 4.6 Hz, J = 0.6
3
Hz), 7.85-7.83 (1H, m), 7.70 (1H, d, J = 7.6 Hz), 7.45-7.43 (1H, m), 7.38-7.29 (3H, m), 3.88 (3H,
1
3
s), 2.56 (3H, s). C NMR (CDCl
42.62 (C ), 139.02 (CH), 135.83 (C
CH), 109.69 (CH), 31.39 (CH ), 19.58 (CH
3
, 100 MHz, 298 K) δ ppm: 150.94 (C
q
), 148.69 (C ), 146.52 (CH),
q
1
q
q
), 135.04 (C
q
), 123.79 (CH), 123.03 (CH), 122.24 (CH), 120.11
+
(
(
3
3
). ESI-MS (soft positive mode, m/z) : [L4+H] : 224.4
exp.), 224.3 (calc.). Elemental analysis: calcd: %C 75.31, %H 5.87, %N 18.82 ; found: %C 75.16,
H 5.83, %N 19.02.
Preparation of mononuclear Fe complexes with L1-L4. In a typical synthesis, 1.47 mmol of
ligand Lk (3.0 eq) in CH Cl (8 mL) were added to 0.49 mmol (1.0 eq) of Fe(CF SO or
Fe(BF 6H O in acetonitrile (48 mL). The resulting mixture was stirred for 12 h under argon, then
%
II
2
2
3
)
3 2
4
)
2
2
evaporated to dryness. The solid residue was dissolved in acetonitrile (4 mL) and diethylether was
slowly diffused until the formation of microcrystalline powders or monocrystals suitable for X-ray
diffraction studies. Filtration followed by washing with diethyl ether gave 62-85 % of
[Fe(Lk)
3
](CF
3
SO
3
)
2
xSolvent or [Fe(Lk)
3
](BF
4
)
2
xSolvent. For the hexafluorofluorophosphate salt,
n
the same procedure was repeated in presence of ( Bu)
4
NPF (25 eq) to yield 59-82% of
6
[
Fe(Lk)
3
](PF
6
)
2
xSolvent. All complexes were characterized by elemental analyses (Table S1
](PF (4) and
CN (5). The latter complex was crystallized in presence of a large excess of
Supporting Information) and X-ray crystal structures could be solved for [Fe(L2a)
3
)
6 2
[Fe(L4)
3
](ClO
4
)
2
CH
3
(nBu)
NClO
.
4
4
Caution! Dry perchlorates may explode and should be handled in small quantities and with the
necessary precautions.20
III
II
III
II
Preparation of dinuclear Ln Fe and Ln Zn complexes with L5-L7. In a typical synthesis, 27.6
μmol of Ln(CF SO (1.0 eq; Ln = La or Eu) and 27.6 μmol of M(CF SO (1.0 eq; M = Fe or Zn)
3
3
)
3
3
)
3 2
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
10
in acetonitrile/water (99:1, 1.38 mL) were added to 82.8 μmol of ligand Lk (3 eq) in dichloromethane
(700 l) and the resulting clear solution was stirred for 4 h under an inert atmosphere. The solvent
were removed and the solid residue was dissolved in acetonitrile (2 mL). Diethylether was slowly
diffused until the formation of microcrystalline powders or monocrystals suitable for X-ray
diffraction studies. Filtration followed by washing with diethyl ether gave 68-93 % of
[LnFe(Lk)
3
](CF
3
SO
3
5
) xSolvent
or
[LnZn(Lk)
3
](CF
3
SO
3
)
5
xSolvent.
For
the
n
hexafluorofluorophosphate, the same procedure was repeated in presence of ( Bu)
4
NPF (25 eq) to
6
yield 69-87% of [LnFe(Lk)
analyses (Table S2 Supporting Information). Single crystal X-ray crystal structures could be solved
for [LaFe(L5) ](ClO 5CH CN (6), [EuFe(L5) ](ClO 6CH CN (7),
](CF SO 2.5CH CN(CH C(OCH ) (8), {[LaFe(L7) ](ClO 6CH CN1.5H
9), {[EuFe(L7) ](CF SO 4CH CN3(CH C(OCH ) (10) and [LuFe(L7) 2CH CN
NClO
3
](PF
6
)
5
xSolvent. All complexes were characterized by elemental
3
4
)
5
3
3
4
)
5
3
[
(
(
EuFe(L6)
3
3
3
)
5
3
3
)
3
3
3
4
)
5
}
2
3
2
O
3
3
3
)
5
}
2
3
3
)
3
3
3
](CF
3
SO
3
)
5
3
n
11). The perchlorate complexes were crystallized in presence of a large excess of ( Bu)
4
4
.
Spectroscopic and analytical measurements
1H and 13C NMR spectra were recorded at 293 K on a Bruker Avance 400 MHz spectrometer.
Chemical shifts are given in ppm with respect to TMS. Spectrophotometric titrations were performed
with a J&M diode array spectrometer (Tidas series) connected to an external computer. In a typical
3
-4
experiment, 25 cm of ligand in acetonitrile (210 M) were titrated at 298 K with a solution of
-3
Fe(CF
3
SO
3
) (210 M) in acetonitrile under an inert atmosphere. After each addition of 75 μL, the
2
absorbance was recorded using Hellma optrodes (optical path length 0.1 cm) immersed in the
thermostated titration vessel and connected to the spectrometer. Mathematical treatment of the
spectrophotometric titrations was performed with factor analysis[21] and with ReactLabTM Equilibria
(
previously Specfit/32).[22] Pneumatically-assisted electrospray (ESI-MS) mass spectra were recorded
-
4
-3
from 10 M (ligands) and 10 M (complexes) solutions on an Applied Biosystems API 150EX
LC/MS System equipped with a Turbo Ionspray source. Elemental analyses were performed by K. L.
Buchwalder from the Microchemical Laboratory of the University of Geneva. Electronic spectra in
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
11
the UV-Vis region were recorded at 293 K from solutions in CH CN with a Perkin-Elmer Lambda
3
1050 using quartz cells of 0.1 or 1.0 mm path length. Solid-state magnetic data were recorded on a
MPMS 3 or MPMS 5 QUANTUM DESIGN magnetometers using magnetic fields of 1000-5000 Oe
and at 2 K/min rates within the 2-400 K range. The magnetic susceptibilities were corrected for the
magnetic response of the sample holder and for the diamagnetism of the compounds.[23] Magnetic
data for samples in acetonitrile were obtained by the Evans method[24] using methanol for temperature
calibration.[25] The method was modified for application with a superconducting magnet.[26] Specific
problems associated with the ‘solvent correction term’[27] were overcome by determining the
diamagnetic contribution under the same experimental conditions.[28] Measurements were carried out
on degassed CD CN solutions containing 1% (v/v) and using tert-butanol as an internal reference. All
3
the data were corrected for changes in solvent density with temperature.[29]
X-ray crystallography
Single crystals. Summary of crystal data, intensity measurements and structure refinements for
compounds (4) to (11) were collected in Tables S3-S10. The crystals were mounted on MiTeGen
kapton cryoloops with protection oil. X-ray data collections were performed with an Agilent
SuperNova Dual diffractometer equipped with a CCD Atlas detector (Cu[Kα] radiation). The
structures were solved by using direct methods.[30] Full-matrix least-square refinements on F were
performed with SHELX2014.[31] CCDC 1862027-1862034 contain the supplementary
crystallographic data. These data can be obtained free of charge from the Cambridge Crystallographic
Data Centre via www.ccdc.cam.ac.uk/.
2
Powder crystals. Variable temperature X-ray powder diffraction experiments were performed on an
Empyrean (PANalytical) diffractometer in capillary mode, with a focusing X-ray mirror for CuKα
radiation and a PIXcel3D area detector. Variable temperatures were achieved with a nitrogen
cryostreamer (Oxford cryosystem). Temperature cooling ramps from 340 K to 140 K were performed
-
1
at 3 K∙min and data collections were performed at fixed temperature plateau (2theta range from 7 to
30° with continuous scan total time 26 min/diagram)
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
12
Results and Discussions
Solid-state SCO Fe(II) complexes with ligands L1-L7: an empirical approach.
The didentate ligands L1-L2[15] and L3[17] were previously obtained according to standard Phillips-
modified strategies similar to that used for the novel ligand L4 (Scheme 3). 3-methyl-picolinic acid
(
(
2) was in situ activated as its acid chloride, which was further reacted with N-methyl-2-nitroaniline
1) to give the ortho-nitroamide (3) in good yield. Final reductive cyclization provided the target
constrained ligand L4.
Scheme 3. Multistep synthesis of ligand L4.
Stoichiometric reactions of L1-L4 (3.0 eq) with Fe(CF
3
SO
3
)
2
or Fe(BF
4
)
2
(1.0 eq) in
dichloromethane/acetonitrile mixtures followed by slow diffusion of volatile ether provided fair
yields (62-85%) of microcrystalline [Fe(Lk) ](CF SO and [Fe(L4) ](BF complexes with various
3
3
3
)
2
3
)
4 2
amounts of co-crystallized solvent molecules (Table S1 in the Supporting Information). Subsequent
n
metathesis of the triflate anions with the help of ( Bu)
4
NPF in acetonitrile/ether mixtures additionally
6
yielded [Fe(Lk)
3
](PF
6
2
) (Lk = L1a, L2a, L4, Table S1). Intense efforts for producing mono-crystals
of sufficient quality for X-ray diffraction studies were only successful for [Fe(L2a)
3
](PF
6
) (4) and
2
n
[Fe(L4)
3
](ClO
4
2
) CH
3
CN (5, after methathesis with ( Bu)
4
NClO , Figure 2 top, Tables S3-S4 and
4
Figures S1-S2 in the Supporting Information). The six-coordinate homoleptic tris-diimine
2
+
]2+ cations in 4 and 5 adopt meridional arrangements, in which the [FeN
]
[Fe(L2a)
3
] and [Fe(L4)
3
6
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
1
3
chromophores only slightly deviate from perfect octahedrons according to SHAPE’s scores (Table 1,
entries 5-6).[32] The Fe-N bond lengths measured at 180 K are typical for a low-spin electronic
configuration in mer-[Fe(L2a)
]2+ (1.96-2.03 Å), whereas the 10% expansion found in mer-
3
2
+
[Fe(L4)
3
] (2.14-2.26 Å, Table 1) is diagnostic for alternative high-spin configuration in the latter
complex.[2]-[4]
II
Figure 2. Molecular structures of the Fe -containing cations in the crystal structures of
[
Fe(L2a)
EuFe(L5)
[LaFe(L7)
10) and [LuFe(L7)
3
](PF
](ClO
](ClO
6
)
2
(4), [Fe(L4)
3
](ClO
4
)
2
CH
[EuFe(L6)
O (9), {[EuFe(L7)
CN (11)
3
CN (5) [LaFe(L5)
](CF SO 2.5CH CN(CH
](CF SO 4CH CN3(CH
3
](ClO
4
)
5
5CH
3
CN (6),
(8),
C(OCH
[
3
4
)
5
6CH
3
CN
(7),
3
3
3
)
5
3
3
)
3
C(OCH
3
)
{
3
4
)
5
2
} 6CH
3
CN1.5H
2
3
3
3
)
5
}
2
3
3
)
3
3
)
(
3
](CF
3
SO
3
5
) 2CH
3
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
1
4
II
II
Table 1. Structural data for the six-coordinate metallic (M = Fe or Zn ) centers in complexes 4-11 in the solid state at 180 K.
Complexes
[Zn(L2a)
2.11(2)
2.24(4)
12(9)
87(2)
1.66
3
]2+
[Fe(L2a)
1.96(1)
2.04(3)
11(3)
3
]2+
[Fe(L4)
2.14(2)
2.26(5)
41.1(2)
76(5)
3
]2+
[EuZn(L5)
2.09(3)
2.26(7)
17(7)
3
]5+
[LaFe(L5)
1.979(8)
1.98(1)
9(6)
3
]5+ [EuFe(L5)
1.973(9)
1.970(8)
8(2)
3
]5+
d
d
M–N(bz) /Å [a]
M–N(py/pz) /Å [a]
α /° [b]
/° [c]

88.3(4)
0.829
88(1)
85.5(5)
0.71
87(2)
Octahedron [d]
Trig. prism [d]
1.957
14.922
-
1.61
0.68
13.11
-
14.064
-
13.12
8.92
15.29
14.96
d
M–Ln /Å [e]
9.19
9.17
Reference
[15]
This work
]5+ [LaFe(L6)
1.95(3)
2.00(4)
9(4)
This work
[EuFe(L6)
1.968(5)
1.988(4)
10.(2)
[15]
This work
This work
Complexes
[EuZn(L6)
2.05(6)
2.3(2)
23(16)
81(8)
3
3
]5+
3
]5+ [LaFe(L7)
2.14(3)
3
]5+
[EuFe(L7)
2.13(2)
3
]5+ [LuFe(L7)
2.13(3)
2.33(7)
19(15)
3
]5+
d
d
M–N(bz) /Å [a]
M–N(py) /Å [a]
2.33(7)
2.34(7)
α /° [b]
/° [c]
30(4)
23(11)

86(2)
85(1)
76(4)
80(3)
80(5)
Octahedron [d]
Trig. prism [d]
1.88
0.91
0.75
3.0(4) [f]
14.9(9) [f]
8.5(3) [f]
This work
2.7(2) [f]
14.3(7) [f]
9.0(2)[f]
This work
2.79
14.40
15.07
15.54
13.53
d
M–Ln /Å [e]
8.96
9.03
9.14
8.94
Reference
[18a]
[18b]
This work
This work
[
a]
Bz = benzimidazole, py = pyridine, pz = pyrazine. [b] Interannular intraligand angle. [c] Interchelate angle. [d] SHAPE’s score for comparison with
perfect octahedrons and trigonal prisms. [f] Average for the two independent complexes in the asymmetric unit.
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
15
The origin of these different spin states can be traced back to the considerable ligand-field splitting
produced by heterocyclic nitrogen atoms, which favors low-spin configurations as found in mer-
] . However, sterical constraints induced by the 3-methyl substituents in mer-[Fe(L4)
]2+
2
+
[Fe(L2a)
3
3
prevent coplanarity within the didentate polyaromatic rings (interannular angle = 41° in mer-
]2+ compared with  = 11° in mer-[Fe(L2a) ]2+, Table 1). This reduces the ligand-field
[Fe(L4)
3
3
2
+ [7]
strength to such an extent that high-spin configuration becomes more stable in mer-[Fe(L4) ] .
3
Unfortunately, we were not able to isolate monocrystals of the facial isomers in the solid state for
mononuclear complexes, but stoichiometric mixing of segmental ligands L5-L7 (3.0 eq) with
M(CF
dichloromethane/acetonitrile followed by slow diffusion of volatile ether provided fair yields (68-
3%) of microcrystalline [LnM(Lk) ](CF SO complexes (Table S2), among which monocrystals
suitable for X-ray crystal structure analyses could be obtained after a second crystallization
3
SO
3
)
2
(M = Fe, Zn; 1.0 eq) and Ln(CF
3
SO
3
)
3
(Ln = La, Eu, Lu; 1.0 eq) in
9
3
3
)
3 5
(
sometimes
EuFe(L5)
[LaFe(L7)
10) and [LuFe(L7)
in
presence
6CH CN
6CH
](CF SO
of
(nBu)
[EuFe(L6)
O (9), {[EuFe(L7)
CN (11) (Figure 2 bottom, Tables S5-S10 and Figures S3-S8).
4
NClO
4
)
for
](CF SO
](CF
[LaFe(L5)
2.5CH CN(CH
SO 4CH CN3(CH
3
](ClO
4
)
5
5CH
3
CN
(6),
[
3
](ClO
4
)
5
3
(7),
3
3
3
)
5
3
3
)
3
C(OCH
3
)
(8),
{
3
](ClO
4
)
5
}
2
3
CN1.5H
2CH
2
3
3
3
)
5
}
2
3
3
)
3
C(OCH )
3
(
3
3
3
)
5
3
5
+ [15]
]5+ [18a] and [LaFe(L6)
5+ [18b]
As previously reported for [EuZn(L5)
3
] ,
[EuZn(L6)
3
3
] ,
the three
helical strands adopt a head-to-head-to-head (HHH) arrangement[33] compatible with the formation
II
II
of purely facial organization of the three didentate binding units around Fe or Zn (Figure 2 bottom).
The intramolecular intermetallic distance (average dM-Ln = 9.0(2) Å) is roughly insensitive to the
choice of the metals and to the nature of the terminal six-membered rings (substituted pyridine in L6
and L7 or pyrazine in L5). The Ln-O and Ln-N distances follow the expected lanthanide contraction
along the La > Eu > Lu series (Figure 3 and Table S11 in the Supporting Information).[34] On the
contrary, the Fe-N distances in a given helicate are insensitive to the size of the various capping
trivalent lanthanide cations (Figure 3), a trend in line with the lack of measurable constraints induced
by the flexible non-covalent lanthanide tripod.[16] In agreement with the analogous mononuclear mer-
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
16
[
[
(
Fe(L2a)
3
]2+ complex, the Fe-N distances (1.95-2.00 Å, Table 1) observed at 180 K for HHH-
5
+
LnFe(L6)
3
] points to low-spin configuration, this whatever the meridional (mononuclear) or facial
dinuclear helicate) arrangements of the three (5-methylpyridin-2-yl)-benzimidazole binding units
II
]5+ gives similar Fe-
around Fe . Extension to (5-methylpyrazin-2-yl)-benzimidazoles in [LnFe(L5)
3
N distances (1.97-1.98 Å) and low-spin characteristics. On the contrary, constrained (6-
5
+
methylpyridine-2-yl)-benzimidazole units in [LnFe(L7) ] lead to high-spin configurations at 180 K
3
2
+
with larger Fe-N distances (2.13-2.34 Å, Table 1) as previously discussed for mer-[Fe(L4)
di,j /Å
3
] .
2.8
2.6
2.4
2.2
2.0
dLn-N(py)
dLn-N(bz)
dLn-O
dFe-N(py)
dFe-N(bz)
Lu
Eu
La
1.00
1.10
1.20
CN=9
RLn / Å
5
+
Figure 3. Interatomic distances dLn-O, dLn-N and dFe-N measured for the HHH-[LnFe(L7) ] cations
3
CN=9
in the crystal structures of 9-11 at 180 K (py = pyridine, bz = benzimidazole; RLn = nine-coordinate
lanthanide ionic radius) .
The molar magnetic susceptibilities (
M
) corrected for diamagnetism and for FeIII contamination
recorded at variable temperature for solid state samples confirm the spin states deduced by X-ray
3
-1
diffraction studies at 180 K with negligible 0.00 ≤ Cls ≤ 0.8 cm Kmol Curie constants for the low-
spin FeII complexes with ligands L1, L2, L5 and L6 at this temperature (Figure 4;
2
2
2
C   T 

N 3k

g S

S 1



g 8

S

S 1
, where N is Avagadro number,  is Bohr

M
B
magneton, k
B
is Boltzmann constant, g = 2 is Landé factor, S is the total spin quantum number).[35]
II
On the other hand, the homoleptic tris-didentate Fe complexes with sterically demanding ligands L4
3
-1
and L7 possess significant Curie constants (3.5 ≤ Cls ≤ 3.8 cm Kmol , Figure 4) diagnostic for high-
II
[36]
spin Fe (S = 2) configuration at this temperature. Upon warming the samples,  T for complexes
M
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
17
with ligands L4 and L7 do not significantly change, a trend in line with the existence of pure high-
spin configuration. The abrupt decrease of  T at very low temperature is diagnostic for the existence
M
-
1
)
.[37]
of axial zero-field splittings fitted to D = 1.37 cm and g = 2.26 for [Fe(L4)
3
](PF
6
2
a)
1
80 K
4
3
2
1
0
[Fe(L4) ](PF )
3
6 2
[
Fe(L2a) ](PF )
3
6 2
[Fe(L1a)
](Otf)
2
3
[
Fe(L1a) ](PF )
3
6 2
Fe(L2a) ](Otf)2
[

3
0
100
200
T / K
300
400
b)
1
80 K
[
4
3
2
1
0
LaFe(L7) ](ClO )
3
4 5
[
LaFe(L6) ](Otf)5
3
[
[
LaFe(L6) ](PF )
3
6 5
LaFe(L5) ](PF )
3
6 5
[LaFe(L5)
](Otf)
5
3

0
100
200
300
400
T / K
Figure 4. 
M M
T versus T plot of the molar magnetic susceptibilities ( ) corrected for diamagnetism
III
II
and for Fe contamination recorded for a) mononuclear and b) dinuclear homoleptic tris-diimine Fe
-
SO -
). The temperature at which the X-ray crystal structures have
complexes (solid state, Otf = CF
3
3
been measured is highlighted (180 K).
For complexes with ligands L1, L2, L5 and L6, 
M
T values smoothly increase with temperature, a
behavior diagnostic for the operation of thermodynamic temperature-driven spin state equilibria. All
curves can be satisfyingly fitted with standard Eq. (2), where Cls, Chs and CFe(III) correspond to the
Curie constants of the incriminated electronic configurations and TIPhs and TIPhs refer to the
temperature independent paramagnetic contributions, see next section.[2d],[38] The mole fraction of
high-spin complexes (xhs) can be further deduced with Eq. (3) when one takes into account the mass
balance xls + xhs + xFe(III) = 1.
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
18
MT  xhs 

C T TIP

 xls 

C T TIP

 xFeIII CFeIII
(2)
hs
hs
ls
ls
MT T TIP C  x
 C C
T TIPls

ls
ls
Fe III



ls
Fe III


x 
(3)
hs
C C T 

TIP TIP

hs
ls
hs
ls
II
In the solid state, the spin transition leads to 25 to 100% of high-spin Fe at the highest accessible
temperature (400 K, Figure S9, Supporting information), but no rational correlation can be drawn
between the local molecular structures (mer/fac isomerism or pyridine versus pyrazine donor groups)
and the SCO properties. Major variations are observed upon exchange of non-coordinated (i. e. ionic)
counter-anions, which indicates that intermolecular contributions and packing effects largely
dominate local effects in the solid state. The record of heating/cooling cycles shows no evidence of
cooperativity for the SCO processes in the solid state. A thorough study of the temperature
dependence of the volume of the crystallographic unit cell of [Fe(L2a)
3
](PF
6
) (4) confirms the
2
operation of a complete smooth spin transition for this complex in the 200-350 K range with an unit
cell expansion of 31.2(6) Å (Appendix 1).
Solution SCO processes for Fe(II) complexes with ligands L1-L4: a molecular approach to spin-
state equilibria. In solution, the minor second-sphere interactions between the solvent molecules and
2
+
six-coordinate homoleptic tris-diimine [Fe(Lk) ] cations are expected to produce only minor energy
3
contributions, thus allowing the deciphering of chemically-programmable structural effects.
]2+ with the didentate ligands
L1a, L1b, L2a and L2b,[15] spectrophotometric titrations of these ligands with Fe(CF
Following the procedure previously detailed for analogous [Zn(Lk)
3
3
SO
3
) in dry
2
acetonitrile (Figure 5a,b and Figures S10a,b-S12a,b) show the successive formation of the expected
2
+
three absorbing complexes [Fe(Lk) ] (n = 3, 2, 1) as ascertained by their independent eigenvectors
n
found in the factor analyses (Figures 5c and S10c-S12c)[21] and their satisfying re-constructed
absorption spectra (Figures 5d and S10d-S12d).
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
19
a)
b)
25
20
15
10
5
0
25
20
15
10
5
3
20 nm
3
2
3
09 nm
90 nm
58 nm
MLCT
256 nm


5
36 nm
0
240 320 400 480 560 640
0.0 0.5 1.0 1.5 2.0 2.5 3.0

/nm
[Fe]tot / [L1a]tot
d)
c)
Fe(L1a)3]2+
[
[Fe(L1a)]2+
60
50
40
30
20
10
0
1
.0
.8
0
+
[
[
[
Fe(L1a) ]2
0.6
0.4
0.2
0.0
3
2+
Fe(L1a)2]
[Fe(L1a) ]2
+
Fe(L1a)]2+
2

L1a
L1a
-0.2
240
340
440
/nm
540
640
0.0 0.5 1.0 1.5 2.0 2.5 3.0

[Fe]tot / [L1a]tot
e)
2
.5
.0
.5
.0
.5
.0
[
Fe]2+
2
1
1
0
0
L1a
Fe(L1a) ]2
+
[
3
[
Fe(L1a)]2+
[
Fe(L1a)2]2+
0
.0 0.5 1.0 1.5 2.0 2.5 3.0
[
Fe]tot / [L1a]tot
Figure 5. a) Variation of absorption spectra and b) corresponding variation of observed molar
extinctions at different wavelengths recorded for the spectrophotometric titration of L1a with
-
4
-3
.
) (total ligand concentration: 2.3ꞏ10 mol dm in acetonitrile, 298 K). c) Evolving factor
Fe(CF
3
SO
3
2
analysis using four absorbing eigenvectors,[21] d) re-constructed individual electronic absorption
spectra[22] and e) associated speciation.[39]
-
1
*
Beyond (i) the diagnostic 2000 cm red-shift of the ligand-centered   transition observed upon
II
[40]
complexation to Fe (Figure 5a,d and S10a,d-S12a,d) and (ii) the stepwise increase of the molar
2
+
]2+  [Fe(Lk) ]2+ series (Figure 5d and S10d-
absorption coefficients along the [Fe(Lk)]  [Fe(Lk)
2
3
S12d), the absorption spectrum of [Fe(Lk)
3
]2+ displays a supplementary broad absorption band
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
20
covering the visible domain (440-600 nm), which can be unambiguously assigned to the Lk FeII
metal-to-ligand charge transfer transition (MLCT) occurring in the low-spin iron(II) complex.[41]
Non-linear least-square fits[22] of the spectrophotometric data to equilibria (4)-(6) provided the
macroscopic cumulative formation constants gathered in Table S12 together with speciation curves
(
Figures 5e and S10e-S12e),[39] which indicate that the target complexes [Fe(Lk)
3
]2+ are quantitatively
formed for 1:3 stoichiometric ratio at sub-millimolar concentrations.
Fe2+
Fe2+
Fe2+
+
+
+
Lk
[FeLk]2+
1,1
Fe,Lk
(4)
(5)
(6)
]2+
Fe,Lk
1,2
2 Lk
3 Lk
[Fe(Lk)
[Fe(Lk)
2
3
]2+
Fe,Lk
1,3
Application of the site-binding model summarized in Eqs (7)-(9)[15] dissects the free energy changes
accompanying the complexation processes into (1) a pure entropic contribution produced by the
change in rotational entropies accompanying the transformation of the reactants into products, a
parameter often referred to as the statistical factor,[42] (2) a simple intermolecular affinity f Fe,Lk
II
between the didentate ligand and the entering Fe cation (including the change in solvation) and (3)
Lk,Lk
Lk,Lk
allosteric cooperativity factors ui, j
measuring the extra energy cost (ui, j 1), respectively
Lk,Lk
energy benefit (ui, j 1) produced by an already bound ligand for the connection of the next one
II [43]
]2+ and [Fe(Lk) ]2+
around Fe .
Since different microspecies contribute to the [Fe(Lk)
2
3
macrospecies in solution (Figure S13),[15] various intramolecular interligand cooperativity factors
Lk,Lk
ui, j should be considered (Eqs (7)-(9)).
Fe,Lk
Fe,Lk
1,1
 24 f
(7)
(8)
2
Fe,Lk
1,2
Fe,Lk
Fe,Lk
Lk,Lk
Lk,Lk
trans,mer
Lk,Lk
cis, fac
Lk,Lk
cis,mer
12 f

  trans, fac
u
 u
 4u
 4u

3
2
2
Fe,Lk
1,3
Lk,Lk
cis, fac

Lk,Lk
Lk,Lk
cis,mer

16

f

u

u

 3

u

cis, fac
(9)




This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
21
With the only resort to the three macroscopic formation constants obtained by spectrophotometric
Lk,Lk
cis, fac
Lk,Lk
cis,mer
Lk,Lk
trans, fac
Lk,Lk
trans,mer
Lk,Lk
titrations, one roughly assumes that u
 u
 u
 u
 u
, and the multi-linear
Fe,Lk
fits of the experimental 1,1
to Eqs (7)-(9) gives the intrinsic metal-ligand affinities
Fe,Lk
Fe,Lk
Lk,Lk
Lk,Lk

G
 RT ln

f

and allosteric interligand interactions Ei, j  RT ln(ui, j ) illustrated
in Figure 6 and tabulated in Table S12.
a)
0
Zn/Fe Zn/Fe Zn/Fe Zn/Fe
-
-
-
-
10
20
30
40
L1b
L2b
L1a
L2a
b)
2
.0
.0
Zn/Fe
Zn/Fe Zn/Fe Zn/Fe
0
-
-
-
2.0
4.0
6.0
L1a
L1b
L2b
L2a
M,Lk
M,Lk
Figure 6. a) Intrinsic affinities G
 RT ln f
and b) global interligand interactions


Lk,Lk
Lk,Lk

E
 RT ln(u
) for the complexation of M(CF
with the help of the
; acetonitrile at 298 K).
3
SO
3
2
) with ligands Lk obtained by fitting
equilibria
(4)-(6)
site-binding
model
(Eqs
(7-9);
Lk,Lk
cis, fac
Lk,Lk
cis,mer
Lk,Lk
trans, fac
Lk,Lk
trans,mer
Lk,Lk
u
 u
 u
 u
 u
In line with the Irving-Williams series,[44] the affinities of ligands Lk for Fe is marginally smaller
II
II
than those with Zn (Figure 6a). The ethyl substituents connected to the benzimidazole units in L1b
and L2b reduce the affinity for any entering metal by circa 10 kJ/mol because of the sterical
constraints produced in the bound cis-cis-chelate conformation.[15] Remarkably, the recurrent global
II
positive cooperativity accompanying the successive complexation of didentate ligands to Fe (-5 ≤
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
22
Lk,Lk
EFe
Lk,Lk

≤ -3 kJ/mol) is at least twice as large as that measured in the Zn(II) analogues (-2 ≤ EZn
bipy,bipy
≤
1 kJ/mol, Figure 6b). This trend is reminiscent of EM
M(bipy)
becomes much more negative in [Fe(bipy)
estimated with Eqs. (7)-(9) for
[
n
]2+ in water, where the minor values found for M = Zn, E Zb ni py,bipy = -0.57(7) kJ/mol
2
+
bipy,bipy
Fe
n
] ( E
= -18(7) kJ/mol) because the coordination
2
+
of a third ligand to high-spin [Fe(bipy) ] (possessing two anti-bonding electrons) gives low-spin
2
[
Fe(bipy)
3
]2+ with no anti-bonding electron.[45] The less drastic effect observed in going from
2
+
]2+ can be tentatively assigned to the smaller ligand-field strength produced
[Fe(Lk)
2
] and [Fe(Lk)
3
2
+
by L1 or L2, compared with bipy, thus leading to spin-crossover [Fe(Lk)
3
] complexes instead of
pure low-spin entities.
II
II
[46]
1
Since Fe complexes are kinetically less labile than their Zn analogues, separated H NMR signals
for the microspecies mer-[Fe(Lk)
3
]2+ and fac-[Fe(Lk) ]2+ (Lk = L1a, L2a; Figure S14) can be
3
observed in the 233-293 K range (Figure S15), while coalescence due to fast exchange processes on
II
[15]
1
the NMR time scale already occurs at 250 K for Zn analogues. The integration of the H NMR
Fe,Lk
mer fac
Fe,Lk
and
mer fac
signals at variable temperatures gives K
(Eq. (10)), from which the enthalpic H
Fe,Lk
mer fac
entropic S
contributions are deduced with the help of van’t Hoff plots (Table S13 and Figure
S16).
Fe,Lk
mer fac
mer-[Fe(Lk)
3
]2+
fac-[Fe(Lk)
3
]2+
K
(10)
II
In agreement with an identical analysis performed for Zn analogues, a weak, but non-negligible
Fe,Lk
mer fac
enthalpic driving force (-5 ≤ H
≤ -3 kJ/mol) favors the formation of the facial isomers fac-
] , the signature of a weak thermodynamic trans-influence. Specific solvation effects [15]
2
+
[15]
[Fe(Lk)
3
Fe,Lk
mer fac
produce negative entropic contributions (-17 ≤ S
≤ -13 J/molK), which (i) exceed the
Fe,Lk,stat
expected statistical factor Smer fac  Rln

1 3
= -9.1 J/molK and (ii) oppose the enthalpic trend

2
+
thus making mer-[Fe(Lk) ] the major isomer in acetonitrile at room temperature. Application of the
3
site-binding model to Equilibrium (10) yields Eq. (11),[43] which can be combined with Eqs (7)-(9) to
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
23
Lk,Lk
Lk,Lk
trans, fac
Lk,Lk
trans,mer
Lk,Lk
 utrans , u
Lk,Lk
cis, fac
Lk,Lk
cis,mer
partition the rough cooperativity factors u
into u
 u
and u
(
Table S12).
2
Fe,Lk
fac
Lk,Lk
cis, fac
1,3,
1
 u

Fe,Lk
mer fac
K

 


(11)
Fe,Lk
Lk,Lk
cis,mer

1
,3,mer
3
u


With this thermodynamic toolkit in hand, we are able to catch the meridional/facial speciation of
]2+ at any temperature in acetonitrile solution (Figure S17), a pre-requisite for assigning
[Fe(Lk)
3
specific magnetic SCO properties to each isomer. Moreover, the stability constants collected for FeII
II
and Zn (Table S12) are strong support for assuming that the closely related homoleptic pseudo-
2
+
octahedral [Ni(Lk)
3
] complexes (existing as mixtures of mer/fac isomers) are quantitatively formed
in acetonitrile at 0.1 M, where well-resolved absorption spectra can be recorded for the
Ni(3T2g A2g), Ni(3T1g A2g) and Ni(1E
3
3
 A2g) d-d transitions (the symmetry labels assume O
3
g
h
symmetry, Figure S18). Non-linear least-square fits of the experimental energies at the maximum of
the absorption bands (Table S14) with the ligand-field Eqs (12-15) pertinent to octahedral d8
systems[47],[48] give the ligand-field strengths oct and Racah parameters B and C collected in Table 2.
3
E
E


A

= 0
(12)
(13)
2g
2
6
B
1
Eg

= 8B + 2C -
oct
3
3
E
E


T

= oct
(14)
(15)
2g
2
2
T1g

=1.5 +7.5B-0.5 225B +  -18 B
oct
oct
oct
This well-established procedure overcomes the difficult detection of weak d-d transitions masked by
intense charge transfer band found in Fe complexes. Ligand-field strength in the 11200 ≤ oct(NiII)
II
[49]
-
1
II
≤
12400 cm range corresponds to the accepted limits for inducing accessible SCO behavior in Fe
analogues,[2a] but it may induce some uncertainties for the calculation of the exact values of oct(NiII)
due to the mixing of spin-allowed 3T2g A2g with the spin-forbidden 1E  A2g.[50] Further
3
3
g
refinements using the improved equation proposed by Hart et al[50] provide oct(Ni(L1a)
) = 11730
3
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
24
-
1
-1
) = 11680 cm and oct(Ni(L2b)
) = 11780 cm-1,
-1
cm , oct(Ni(L1b)
3
) = 11880 cm , oct(Ni(L2a)
3
3
which deviate by less than 2% of those estimated in absence of mixing (eq. 14, column 2 in Table 2).
Table 2. Ligand-field strengths (oct), Racah parameters B et C and nephelauxetic parameters  for
2
+
homoleptic [Ni(Lk) ] complexes (acetonitrile, 293 K).
3
complexes
Ni(bipy)
]2+
]2+

oct /cm-1 B /cm-1 C /cm-1
C/B
3.78
3.64
3.50
3.48
3.48
2.63
3.05

oct/B
β [a]
0.71
0.83
0.84
0.85
0.85
0.95
0.88
0.83
0.80
0.89
Reference
This work
This work
This work
This work
This work
This work
This work
This work
[48]
[
3
12699
11476
11628
11423
11567
9644
741
866
878
889
882
994
918
864
830
929
2801
3149
3072
3096
3067
2614
2800
17.1
13.3
13.2
12.9
13.1
9.7
[
Ni(L1a)
3
[
Ni(L1b)
3
]2+
]2+
[
Ni(L2a)
3
[
Ni(L2b)
3
]2+
]2+
]2+
]2+
[
Ni(L3)
3
3
[
Ni(L4)
10681
10100
10730
8580
11.6
11.7
12.9
9.2
[
Ni(CH
3
CN)
6
3708
4.29
]2+
]2+
[b]
[b]
[
Ni(NH
3
)
6
6
[b]
[b]
[Ni(H O)
2
[48]
[
a]
o
o
-1
2+ [6b] [b]
Nephelauxetic parameter = B/B with B = 1042 cm for free Ni .
Not computed in
reference [48].
L2 L1
L3 L4
bipy
E/B
0
1T2g
1Eg
1A1g
6
5
4
3
2
1
1S
0
0
0
0
0
0
3
T
1
1
2
g
g
g
1
1
1
3
T
T
A
T1g
1
g
1G
3P
1D
3F
3T2g
1Eg
3A2g
0
2 4 6 8 10 12 14 16 18 20
oct/B
8
o
o
Figure 7. Tanabe-Sugano diagram built for the d electronic configuration (C /B = 4.71 taken for
the free Ni ion) highlighting the oct/B ratio observed for [Ni(Lk)
]2+ in acetonitrile solution.[8]
2
+
[8]
3
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
25
] and [Fe(L2)
]2+ display SCO behavior
2
+
Neglecting this minor effect, Table 2 predicts that [Fe(L1)
3
3
] and [Fe(L4)
]2+ adopt pure high-spin state. Furthermore, the slightly
2
+
in solution, whereas [Fe(L3)
3
3
larger nephelauxetic effect produced by the remarkable -accepting properties of terminal pyrazine
groups in L1[51] shifts oct/B toward high values. The associated SCO behavior is thus expected to
occur at higher temperature for [Fe(L1)
documented for related replacements of pyridine rings with pyrazine analogues.[52] Since the NMR
]2+ in acetonitrile are complicated by mer/fac isomerization and by
3
]2+ than for [Fe(L2)
3
]2+ (Figure 7), a trend previously
spectra recorded for [Fe(Lk)
3
paramagnetic line broadening (Figure S15), we gave up chemical shifts techniques[53] and preferred
to use the Evans’ method[24],[27] adapted for systems with non-negligible diamagnetic contributions
(
Eq. 16).[28] The corrected magnetic susceptibilities 
are obtained from the balance between the
M
para
shifts in frequency of the paramagnetic sample 
(Hz) with respect to tert-butanol taken as the
]2+ complex of molecular mass Mpara (gmol ) and at
-1
reference (measured for the [Fe(Lk)
3
concentration mpara (gcm )) and that of the diamagnetic analogue  (measured for [Zn(Lk)
3
dia
]2+
3
dia
dia
of molecular mass M at concentration m ). In Eq (16),  stands for the operating frequency of the
0
NMR spectrometer in Hz and S
f
= 4/3 is the shape factor of the superconducting magnet.[28]
para
para
dia
dia
1
 
M
 M



M 0

(16)


para
dia
Sf
m
m
Plots of 
M
T products versus T collected in Figure 8 show pure high-spin character for [Fe(L3)
]2+
3
] and pure low-spin character for [Fe(bipy)
]2+ in complete agreement with previous
2
+
and [Fe(L4)
3
3
solid state measurements (Figure 4a).[45b],[54] Non-linear least-squares fits using Eq. (2) provide Curie
3
-1
]2+ and [Fe(L4)
range 3.13-3.65 cm Kmol ) and Cls = 0.00(8) cm Kmol typical for low-spin [Fe(bipy)
contributions of temperature-independent paramagnetism are opposite, but in the same range as the
]2+ (expected
constants of 3.41 ≤ Chs ≤ 3.44 cm Kmol typical for high-spin [Fe(L3)
3
3
3
-1 [55]
3
-1
]2+. The
3
] (TIPhs = 358(57)10 cm Kmol ) and for [Fe(bipy)
]2+ (TIPls
2
+
-6
3
-1
diamagnetic response for [Fe(L4)
3
3
-
6
3
-1
=
351(55)10 cm Kmol , Table 3). Using these values for pure high-spin, respectively pure low
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
26
spin configurations allows the non-linear least-square fit of the  T versus T curves recorded for the
M
] and [Fe(L2)
]2+ with the help of Eqs (1)-(2) (see Appendix 2
2
+
spin crossover complexes [Fe(L1)
3
3
for details). In solution, the estimated enthalpic (HSCO) and entropic (SSCO) contributions to the
spin state equilibria (Table 3) are essentially independent from the choice of the counter anions (Table
3 [
). However, the replacement of pyrazine rings in [Fe(L1)
56]
]2+ with pyridine rings in [Fe(L2) ]2+
3
3
significantly decreases HSCO, while SSCO is poorly affected. As a result, the ‘transition’ temperature
2
+
T
1/2 at which xls = xhs = 0.5 is shifted by 40 K toward high temperature for [Fe(L1) ] (Figure 8 and
3
Table 3), a trend in line with stronger Fe-N interactions and a larger oct/B parameter found for this
ligand (larger -backbonding, vide infra).
[
Fe(L3) ](Otf)2
3
4
3
2
.0
.0
.0
[
Fe(L4) ](Otf)2
3
[Fe(L2a) ](Otf)2
3
[Fe(L2a) ](PF )
3
6 2
[Fe(L1a) ](Otf)2
3
[
Fe(L1a) ](PF )
3
6 2

1.0
[
Fe(bipy) ](Otf)2
3
0
.0
2
23
263
303
343
T / K
Figure 8. Corrected 
M
T products plotted as a function of the temperature for 0.01 M solution of
](PF (squares) in CD CN. The dotted traces correspond to
[
Fe(Lk) ](CF SO (discs) or [Fe(Lk)
3
3
3
)
2
3
6
)
2
3
non-linear least-square fits obtained with Eqs (1)-(2) and using the parameters gathered in Table 3.
Two additional points merit to be highlighted here. Firstly, an especially large TIPhs = 1321(88)10-6
cm Kmol is found for high-spin [Fe(L3)
]2+ because the sterically hindered approach of the 6-
3
-1
3
methyl-pyridine rings produces very weak ligand-field (Table 2) and, consequently, large second-
order Zeeman contributions via the spin-orbit coupling mechanism.[35] Secondly, the thermodynamic
parameters HSCO and SSCO recorded for [Fe(L1)
3
]2+ and [Fe(L2) ]2+ refer to average values
3
characterizing a mixture of temperature-dependent mer/fac isomers, probably both displaying
specific and different SCO behaviors (see next section).
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
27
Solution SCO processes for LnFe(II) helicates with ligands L5-L7: deciphering
meridional/facial contributions. Stoichiometric mixing of segmental ligands L5-L7 (3 eq.) with
II
II
III
III
M(CF SO
3 3
)
2
(1 eq, M = Fe , Zn ) and Ln(CF
3
SO
3
) (1 eq, Ln = La , Eu ) in acetonitrile at
3
millimolar concentrations quantitatively form C
3
-symmetrical HHH-[LnM(Lk)
3
)]5+ triple-stranded
1
helicates as ascertained by ESI-MS data (qualitative speciation, Figure S19 and Table S15) and H
NMR spectra (quantitative speciation, Figures S20-S24). These results corroborate detailed solution
]5+ [15] and [LnM(L6)
] ,
5+ [18]
and they unambiguously
studies previously reported for [EuZn(L5)
3
3
II
II
establish that the divalent d-block Fe or Zn cations are facially coordinated by three unsymmetrical
diimine units in these C -symmetrical dinuclear complexes in solution, as depicted in their crystal
structures (Figure 2). The associated corrected magnetic susceptibilities  , recorded using Evans’
3
M
method in CD
3
CN (Eq. (16)) and plotted as 
M
T versus T curves (Figure 9), show pure high-spin
behavior for [LaFe(L7)
3
]5+ with the constrained ligand L7. The magnetic behavior of the latter
3
-1
-6
3
-
complex provides reliable estimates for Chs = 3.44(4) cm Kmol and TIPhs = 310(15)10 cm mol
1
using Eq. (2), two parameters very similar to those found for the mononuclear parent high-spin
5
+
complex [Fe(L4) ] (Table 3).
3
4
3
2
1
0
.0
.0
.0
.0
.0
[
LaFe(L7) ](ClO )
3 4 5
[LaFe(L6) ](Otf)
3
5
[YFe(L5) ](Otf)
3
5

[CaFe(L5) ](Otf)
3
4
[
LaFe(L5) ](Otf)
3
5
2
23
263
303
343
T / K
Figure 9. Corrected 
M
T products plotted as a function of the temperature for 0.01 M solution of
5
+
]4+, [YFe(L5) ]5+, [LaFe(L6) ]5+ and [LaFe(L7) ]5+in CD
[LaFe(L5)
3
] , [CaFe(L5) CN. The dotted
3
3
3
3
3
traces correspond to non-linear least-square fits obtained with Eqs (1)-(2) and using the parameters
gathered in Table 3.
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
2
8
M
Table 3. Magnetic and thermodynamic parameters obtained by non-linear least-square fits using Eqs (1)-(2) for the  T versus T plots recorded for the
homoleptic [Fe(Lk)
3
]X
2
complexes in solution (CD CN).
3
TIP 10-6
TIP 10
-6
-1
[a]
X
C
hs
C
ls
HSCO
SSCO
/JKmol-1
-
T
1/2
Reference
hs
ls
complexes
x
Fe(III)
3
-1
3
-1
3
-1
3
-1
/
cm Kmol /cm mol /cm Kmol /cm mol  /kJmol
/K
-
[
Fe(bipy)
3
]X
2
CF
3
3
SO
3
3
-
-
0.00(8)
0.00[c]
0.00[c]
0.00[c]
0.00[c]
-
351(55)
351[c]
351[c]
351[c]
351[c]
-
-
0.00
This work
[
Fe(L1a)
Fe(L1a)
Fe(L2a)
Fe(L2a)
3
]X
]X
]X
]X
2
CF
SO
3.44[b]
3.44[b]
3.44[b]
3.44[b]
3.41(5)
3.44(2)
3.44[b]
3.44[b]
3.44[b]
3.44[b]
3.44[b]
3.44(4)
358[b]
358[b]
358[b]
358[b]
1321(88)
358(57)
358[b]
358[b]
358[b]
358[b]
358[b]
310(15)
35.0(3)
35.7(3)
28.3(4)
28.6(3)
-
100(1)
101(1)
91(1)
91.5(8)
-
349(5) 0.032 This work
[
3
3
3
2
2
2
PF
CF SO
PF
6
354(5)
0.00
0.01
0.00
0.01
0.00
This work
This work
This work
This work
This work
[
[
3
3
309(6)
6
313(4)
[Fe(L3)
3
]X
2
CF
CF
CF
CF
CF
CF
3
3
3
3
3
3
SO
SO
SO
SO
SO
SO
3
3
3
3
3
3
-
-
[Fe(L4)
3
]X
2
-
-
-
-
[
LaFe(L5)
YFe(L5)
CaFe(L5)
3
]X
5
0.00[c]
0.00[c]
0.00[c]
0.00[c]
0.00[c]
-
351[c]
351[c]
351[c]
351[c]
351[c]
-
37.9(5)
34(2)
36.8(2)
29.6(2)
30.0(2)
-
92(1)
83(5)
90.4(6)
86.2(5)
88(1)
-
412(8) 0.01[d] This work
412(33) 0.08[d] This work
407(3) 0.05[d] This work
[
3
]X
5
[
3
]X
4
[LaFe(L6)
[LaFe(L6)
[LaFe(L7)
3
3
3
]X
]X
]X
5
5
5
344(3)
331(5)
-
0.03
-
This work
[49d]
ClO
4
4
ClO
0.02
[19]
[
a]
1/2 = HSCO/SSCO. [b]
) Cls and TIPls found for
are used for fitting the SCO behaviors. [c]
T
Chs and TIPhs found for [Fe(L4)
3
](CF
3
SO
3
2
[
Fe(bipy)
3
](CF
3
SO
3
)
2
are used for fitting the SCO behaviors. [d] Deduced from low-temperature data fitted for pure low-spin complex.
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201804161
29
4
+
]5+ [LaFe(L5)
]5+ and [LaFe(L6)
]5+, the 
T versus T plots are
For [CaFe(L5)
3
] , [YFe(L5)
3
3
3
M
diagnostic for spin-crossover processes (Figure 9), the thermodynamic characteristics of which could
be extracted with Eqs (1)-(2) (see Appendix 2) and collected in Table 3. Since only a limited
4
+
5+
temperature range is available for collecting SCO data for [CaFe(L5)
3
] , [YFe(L5)
3
] and
5
+
[LaFe(L5)
3
] in solution (Figure 9), HSCO and SSCO are the only adjusted parameters within the
SCO domain using Eqs (1)-(2) for these complexes, while TIPhs, TIPls and xFe(III) are fixed (see caption
in Table 3). Taking into account that no cooperativity factor can influence mononuclear SCO
processes, the known linear dependence of the van’t Hoff equation correlating HSCO and SSCO with
ln(KSCO) implies that both parameters can be obtained with a similar precision, obviously limited by
experimental dispersion and by the restricted accessible temperature range.[57]
Considering that HSCO and SSCO measured for C
3
-symmetrical helicates HHH-[LaFe(Lk)
3
]5+
strictly refer to pure facial isomer (Lk = L5-L6), it is not so surprising that the enthalpic contributions
2
+
are 1-2 kJ/mol larger than those found in related mixtures of mer/fac-[Fe(Lk) )] complexes (Lk =
3
L1-L2, Table 3) because the latter mixtures are largely dominated by the meridional isomers (see
Figure S17). It is indeed the facial arrangement which benefits from the largest thermodynamic trans-
influence (Table S13) and consequently possesses the strongest Fe-N bonds. Interestingly, the
-
1
-1
entropic contributions SSCO is reduced by 6-10 Jmol K in the pure facial isomer found in the
[LaFe(Lk)
3
]5+ helicate, a trend which can be assigned to the larger dipole moments of the facial
isomers (compared with the meridional isomers) and their detrimental effect on the solvation free
energy changes upon molecular size expansions accompanying low-spin  high-spin transition
(
Eq(18), vide supra).[58] Let’s stress here that changing the size or the charge of the innocent (i. e.
2
+
3+
3+
]4+, [YFe(L5)
] and
5+
non-paramagnetic) capping cations (Ca , Y , La ) in the helicates [CaFe(L5)
3
3
5
+
II
[LaFe(L5)
3
] has very limited impact on the Fe -centered spin-crossover processes, a consequence
II
of the large flexibility of the capping unit and its minor influence on the Fe coordination sphere,
except for the preorganization of a single facial isomer (Figure S25).[18b],[49d],[59] Altogether, the
opposite trends displayed by HSCO (i. e. increase) and SSCO (i. e. decrease) in going from the
This article is protected by copyright. All rights reserved.