Coupled crystallographic order-disorder and spin state in a bistable molecule: Multiple transition dynamics

Article abstract of DOI:10.1002/chem.201003197

A novel bispyrazolylpyridine ligand incorporating lateral phenol groups, H₄L, has led to an Fe^II spin-crossover (SCO) complex, [Fe(H₄L)₂][ClO₄]₂.H₂O.2 (CH₃)₂CO (1), with an intricate network of intermolecular interactions. It exhibits a 40K wide hysteresis of magnetization as a result of the spin transition (with T_0.5 of 133 and 173K) and features an unsymmetrical and very rich structure. The latter is a consequence of the coupling between the SCO and the crystallographic transformations. The high-spin state may also be thermally trapped, exhibiting a very large T_TIESST (≈104K). The structure of 1 has been determined at various temperatures after submitting the crystal to different processes to recreate the key points of the hysteresis cycle and thermal trapping; 200K, cooled to 150K and trapped at 100K (high spin, HS), slowly cooled to 100K and warmed to 150K (low spin, LS). In the HS state, the system always exhibits disorder for some components (one ClO₄^- and two acetone molecules) whereas the LS phases show a relative ≈9 % reduction in the Fei-N bond lengths and anisotropic contraction of the unit cell. Most importantly, in the LS state all the species are always found to be ordered. Therefore, the bistability of crystallographic order-disorder coupled to SCO is demonstrated here experimentally for the first time. The variation in the cell parameters in 1 also exhibits hysteresis. The structural and magnetic thermal variations in this compound are paralleled by changes in the heat capacity as measured by differential scanning calorimetry. Attempts to simulate the asymmetric SCO behaviour of 1 by using an Ising-like model underscore the paramount role of dynamics in the coupling between the SCO and the crystallographic transitions.

Full text of DOI:10.1002/chem.201003197

DOI: 10.1002/chem.201003197
Coupled Crystallographic Order–Disorder and Spin State in a Bistable
Molecule: Multiple Transition Dynamics
Gavin A. Craig,[a] Josꢀ Sꢁnchez Costa,*[a] Olivier Roubeau,*[b] Simon J. Teat,[c] and
Guillem Aromꢂ*[a]
Abstract: A novel bispyrazolylpyridine
ligand incorporating lateral phenol
groups, H4L, has led to an FeII spin-
crystal to different processes to recre-
ate the key points of the hysteresis
cycle and thermal trapping; 200 K,
cooled to 150 K and trapped at 100 K
(high spin, HS), slowly cooled to 100 K
and warmed to 150 K (low spin, LS).
In the HS state, the system always ex-
hibits disorder for some components
in the LS state all the species are
always found to be ordered. Therefore,
the bistability of crystallographic
order–disorder coupled to SCO is dem-
onstrated here experimentally for the
first time. The variation in the cell pa-
rameters in 1 also exhibits hysteresis.
The structural and magnetic thermal
variations in this compound are paral-
leled by changes in the heat capacity as
measured by differential scanning calo-
rimetry. Attempts to simulate the
asymmetric SCO behaviour of 1 by
using an Ising-like model underscore
the paramount role of dynamics in the
coupling between the SCO and the
crystallographic transitions.
crossover (SCO) complex, [Fe
ACHTUGNRTNE(NUGN H4L)2]-
A
ACHTUNGTRENNUNG
intricate network of intermolecular in-
teractions. It exhibits a 40 K wide hys-
teresis of magnetization as a result of
the spin transition (with T0.5 of 133 and
173 K) and features an unsymmetrical
and very rich structure. The latter is a
consequence of the coupling between
the SCO and the crystallographic trans-
formations. The high-spin state may
also be thermally trapped, exhibiting a
very large TTIESST (ꢀ104 K). The struc-
ture of 1 has been determined at vari-
ous temperatures after submitting the
À
(one ClO4 and two acetone mole-
cules) whereas the LS phases show a
À
relative ꢀ9% reduction in the Fe N
bond lengths and anisotropic contrac-
tion of the unit cell. Most importantly,
Keywords: bistability · hysteresis ·
iron · magnetic properties · order–
disorder transitions · spin crossover
Introduction
their transition occurs between a diamagnetic low-spin (LS,
S=0) and a paramagnetic high-spin (HS, S=2) state and is
often accompanied by drastic changes in optical or dielectric
responses.[2,3] The focus of the renewed interest in SCO sys-
tems for applications in nanotechnology is directed toward
their nanostructure, such as incorporation into soft-matter
phases,[4,5] nanoparticles[6] or surfaces.[7] In this context,
issues of supramolecular or long-range interactions are of
special relevance. Indeed, although the origin of the SCO
phenomenon is molecular, the kinetics of this transforma-
tion are intimately related to the mode of connection be-
tween the centres that undergo the transition. The latter is
crucial because it determines the conditions or feasibility of
bistability by controlling both the abruptness of the transi-
tions and the properties of any possible hysteretic behav-
iour. It is, therefore, of paramount importance to understand
the fundamentals of this relationship as a possible means of
controlling the dynamics of the SCO and its various mani-
festations. In most systems, various types of intermolecular
interactions occur simultaneously to produce the whole net-
work. The modelling of these systems, however, is usually
oversimplified by combining them into one single phenom-
enological parameter that describes the interaction between
the spin-active centres.[8] The structures of SCO solids nor-
mally also contain spin-inactive components, such as sol-
vents and counterions. In some cases, temperature-induced
The phenomenon of spin crossover (SCO) provides for a
versatile means of achieving molecular bistability, driven or
addressed through external stimuli, and thus constitutes a
promising avenue for research into molecular devices.[1]
Compounds based on FeII ions are particularly attractive;
[a] G. A. Craig, Dr. J. Sꢀnchez Costa, Dr. G. Aromꢁ
Departament de Quꢁmica Inorgꢂnica
Universitat de Barcelona
Diagonal 647, 08028, Barcelona (Spain)
Fax : (+34)93-4907725
E-mail: j.sanchezcosta@qi.ub.es
guillem.aromi@qi.ub.es
[b] Dr. O. Roubeau
Instituto de Ciencia de Materiales de Aragꢃn
CSIC and Universidad de Zaragoza
Plaza San Francisco s/n, 50009, Zaragoza (Spain)
Fax : (+34)976-761229
E-mail: roubeau@unizar.es
[c] Dr. S. J. Teat
Advanced Light Source
Berkeley Laboratory
1 Cyclotron Road, Berkeley, CA 94720 (USA)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201003197.
3120
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 3120 – 3127

FULL PAPER
crystallographic ordering of some of these components has
been detected in concomitance with the spin transition.[9–12]
In a remarkable example, two degrees of ordering have
been detected at one given temperature (25 K), depending
on whether the sample has been brought to that tempera-
ture by slow cooling or by quench-cooling.[13] Recently, this
coupling has been modelled for the cases in which the spin-
active and inactive components belong to either one or two
sublattices (i.e., for one- and two-step SCO, respectively).[8]
The issue of bistability in the context of coupled phase tran-
sitions involving crystallographic ordering has so far not
been treated at this level, perhaps because it has not yet
been revealed experimentally.
One of the most versatile groups of ligands in SCO re-
search is that based on 2,6-bis(pyrazol-x-yl)pyridine (x-bpp;
x=1, 3).[14,15] Both regioisomers are capable of chelating
transition metals in an N-based tridentate fashion. Within
the broad family of resulting complexes, especially those of
FeII, many exhibit SCO behaviour of varying natures, often
depending on spin-inactive components (counterions, sol-
vent molecules and the extent of hydration).[11,16] Also, de-
rivatives of 1-bpp and 3-bpp have been prepared by intro-
ducing substituents onto almost every accessible site of the
ligand framework.[17] This has allowed correlations between
local geometric distortions and the SCO properties of relat-
ed chromophores to be established[17] and permitted the
study of the effect of order–disorder phase transitions on
the spin transition.[11,18] Very recently, several intermediate
states were observed upon warming in the thermal SCO of
Results and Discussion
Synthesis: Ligand H4L has been prepared through a less
conventional method for making bpp derivatives, that is, the
aza-cyclation of bis-1,3-diketones[21] by excess hydrazine as
previously described by us.[22] The aerobic reaction in ace-
tone of H4L with FeACHTUNRGTNE(UNG ClO4)2·4H2O in the presence of a small
amount of ascorbic acid leads to an orange solution that,
upon layering with diethyl ether, gives the complex [Fe-
AHCNUTTGERG(NUNN H4L)2]ACHTNURTGEGNN[UN ClO4]2·H2O·2ACHTNUGTREN(NUGN CH3)2CO (1) after approximately one
week. It is interesting to note that the presence in this reac-
tion system of anions (XÀ; X=SCN, Cl, Br) that can poten-
tially act as terminal ligands triggers the oxidation of FeII to
FeIII and the degradation of ascorbate to oxalate, which
leads to oxalate-bridged [FeIII2] complexes that incorporate
XÀ as axial ligands.[20] Thus, the poor coordination ability of
perchlorate seems crucial here for the stability of complex 1.
Description of the structure: At 200 K, compound 1 crystal-
¯
lizes in the triclinic space group P1 (Table 1) and the unit
cell is composed of the main complex cation [Fe
(H4L)2]2+
À
(Figure 1) together with two ClO4 ions, two lattice mole-
cules of acetone and one lattice molecule of water. The
cation of 1 consists of an FeII centre bound to two neutral
H4L ligands, each of which uses its three central N-atoms
for coordination, which leads to a distorted octahedral ge-
ometry. Each ligand lies in approximately one plane (espe-
cially the bpp fragments). In one ligand, both phenol groups
point in the same direction (syn,syn conformation with re-
spect to the central coordination pocket) whereas in the
other ligand the phenol groups are oriented opposite to
each other (syn,anti; Figure 1). The angle (q) between the
idealized [Fe1,N2,N3,N4] and [Fe1,N7,N8,N9] meridian
planes is 84.698, whereas the N3-Fe1-N8 angle (f) is
the [Fe
adopted by the [NiACHTUNGTRENNUNG
ACHTUNGTRENNUNG
(1-bpp)2]2+ cation, resulting from the different forms
(mnt)2]À anions (mnt=maleonitriledithio-
late) in the system and leading to multiple bistability.[19]
Indeed, the coupling of simple SCO molecular complexes,
such as the [FeACHTUNGTRENNUNG
(bpp)2]2+ type, with lattice components con-
À
stitutes a potential way to design cooperative SCO systems.
With the aim of studying in detail the interrelation between
the phenomenon of spin transition and the properties of the
crystal lattice hosting it, we have prepared a functionalised
derivative of 3-bpp (2,6-bis{5-(2-hydroxyphenyl)pyrazol-3-
yl}pyridine, H4L)[20] by introducing two hydroxyphenyl
groups on the sides of the bpp unit. These substituents were
expected to facilitate the establishment of a large variety of
intermolecular interactions. Indeed, extensive variable-tem-
perature structural, magnetic and thermal studies of the mo-
177.00(5)8. The Fe N bond lengths are in the range
2.118(2)–2.206(2) ꢅ, with an average of 2.167 ꢅ. These
values fall within the range expected for FeII centres in the
HS state (see Tables 1 and S1 for other metric parameters).
The components of 1 are engaged in an extensive network
of intermolecular interactions in the three crystal dimen-
À
sions, involving p···p and C H···p contacts as well as hydro-
gen bonds (Table 1). In this network, the 2-hydroxyphenyl
groups introduced as substituents of the bpp units play a
very significant role (vide infra). Thus the cations are
aligned in the ac direction with their pyridyl rings approxi-
mately oriented towards each other and establishing interac-
tions through p···p stacking (Figures 2 and S1). In this direc-
tion, the complexes are connected by means of hydrogen
lecular complex [FeCATHGNURTENGUNN(H4L)2]ACHNUTRGTEGN[NNU ClO4]2·H2O·2ACHTNGUTRENUN(NG CH3)2CO (1)
reveal the occurrence of SCO behaviour and a large bistable
domain. Its striking asymmetric nature is related to a crys-
tallographic order–disorder transition involving most of the
intermolecular interactions at work in the compound.
À
bonds through the intermediacy of ClO4 anions to form
supramolecular chains. Each ClO4 group establishes hydro-
À
À
À
gen bonds with one [N H] of one H4L ligand and one [O
H] of another ligand from an opposite complex. This pattern
occurs twice between each pair of adjacent complex cations
(Figures 2 and S1), and takes place alternately within each
of the two approximate planes defined by the ligands along
the chain. In the ab direction zig-zag chains are formed
through p···p interactions between pairs of phenol/pyrazolyl
Chem. Eur. J. 2011, 17, 3120 – 3127
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3121

J. S. Costa, O. Roubeau, G. Aromꢁ et al.
Table 1. Crystallographic data and crystallographic parameters related to the intermolecular interactions (See
Figure 3 and for letter codes) and to the local environment of the SCO centres of complex
(C52H48Cl2FeN10O15), and occupancies of disordered species at various temperatures and spin states.
hibit crystallographic disorder
(Figure 2) over two positions
with various occupancy ratios
(see Table 1).
4
1
T [K]
200 (HS)
triclinic
150 (HS)
triclinic
150 (LS)
triclinic
100 (LS)
triclinic
100 (HS, trapped)
triclinic
crystal system
space group
a [ꢀ]
b [ꢀ]
c [ꢀ]
¯
¯
¯
¯
¯
P1
P1
P1
P1
P1
Magnetic properties: The bulk
magnetization of a polycrystal-
line sample of 1 was measured
in the 5–300 K temperature
range under the influence of a
constant magnetic field of 5 kG
(Figure 5). At 300 K, the prod-
uct cT is 3.45 cm3 KmolÀ1 (c is
the molar magnetic susceptibili-
ty), which indicates the pres-
ence of FeII ions in the HS (S=
2) state (cT=3.0 cm3 KmolÀ1 is
expected for g=2). The value
of cT stays almost constant
upon cooling the system at
12.310(3)
13.442(3)
17.399(4)
104.57(3)
99.29(3)
12.275(3)
13.407(3)
17.386(4)
104.61(3)
99.25(3)
12.364(3)
13.523(3)
17.312(4)
106.69(3)
98.53(3)
12.326(3)
13.513(3)
17.269(4)
106.63(3)
98.50(3)
12.247(2)
13.385(3)
17.355(4)
104.60(3)
99.12(3)
a [8]
b [8]
g [8]
105.44(3)
105.39(3)
106.26(3)
106.36(3)
105.47(12)
V [ꢅ3]
2604.1(13) 2588.7(13) 2578.0(13) 2560.5(13) 2574.4(12)
0.520
m [mmÀ1
]
0.522
0.524
0.528
0.525
reflns
9823
9952
9955
9720
10009
R1 (all data)
wR2
0.0407
0.1178
1.064
12.48(5)
145.85
2.125
0.0506
0.1403
1.071
12.43(5)
144.72
2.123
0.0537
0.1555
1.044
9.77(4)
100.13
1.932
0.0495
0.1404
1.020
9.76(4)
100.52
1.929
0.0642
0.1779
1.035
12.42(5)
144.26
2.123
S
octahedral volume [ꢅ3]
S [8]
À
av. Fe Npy [ꢅ]
À
av. Fe Npz [ꢅ]
2.187
2.182
1.985
1.984
2.181
Fe···Fe (ac) [ꢅ]
9.743(3)
2.747(3)
2.934(3)
2.876(3)
2.893(2)
4.802(2)
4.566(2)
4.465(2)
4.733(2)
4.113(2)
3.598(2)
3.722(3)
3.412(3)
–
9.730(3)
2.751(4)
2.921(3)
2.867(3)
2.882(3)
4.794(2)
4.570(2)
4.453(2)
4.702(2)
4.095(2)
3.592(2)
3.711(3)
3.392(3)
–
9.850(3)
2.908(3)
2.834(3)
2.818(3)
2.934(3)
4.748(2)
4.628(2)
4.269(2)
4.464(2)
3.786(2)
3.558(2)
–
3.356(3)
3.712(3)
1:0
1:0
1:0
9.829(3)
2.939(3)
2.856(3)
2.818(3)
2.938(3)
4.730(2)
4.607(2)
4.251(2)
4.448(2)
3.767(2)
3.547(2)
–
3.336(3)
3.691(3)
1:0
1:0
1:0
9.723(3)
2.758(5)
2.910(4)
2.818(3)
2.938(3)
4.786(2)
4.562(2)
4.439(2)
4.679(2)
4.075(2)
3.590(2)
3.699(5)
3.377(4)
–
À
[O3 H3···O8] [ꢅ]
1 KminÀ1 until it initiates
a
À
[N10 H10A···O6] [ꢅ]
À
[O1 H1···O10] [ꢅ]
sudden decline near 150 K,
which takes place in various
(reproducible) small steps and
more gradually towards the
end, down to 0.19 cm3 KmolÀ1
at 100 K. This value is consis-
tent with an almost complete
conversion of the FeII centres to
the LS state, and remains prac-
tically constant until the lowest
measured temperature. Upon
re-heating, the cT versus T data
are superimposable on the
values obtained as T was de-
creased until about 90 K, at
À
[N5 H5B···O9] [ꢅ]
py···py A [ꢅ][a]
py···py B [ꢅ][a]
phen···pz C [ꢅ][a]
phen···pz D [ꢅ][a]
phen···pz E [ꢅ][a]
phen···pz F [ꢅ][a]
[b]
À
C H···pz a [ꢅ]
[b]
À
C H···pz b [ꢅ]
[b]
À
C H···pz c [ꢅ]
ClO4À occupancy (O5,O8:O55,O88) 0.8:0.2
0.8:0.2
0.4:0.6
0.3:0.7
0.8:0.2
0.4:0.6
0.3:0.7
acetone 1 occupancy (C2S:C3S)
acetone 2 occupancy (C7S:C8S)
0.4:0.6
0.3:0.7
[a] Centroid-to-centroid distance. [b] Distance from H to centroid.
which the former diverge by describing a small decline
down to a plateau at 0.10 cm3 KmolÀ1 (corresponding to vir-
tually 100% of LS centres). When T reaches ꢀ170 K there
is a very abrupt, one-step symmetric increase in cT up to
3.39 cm3 KmolÀ1, at which it again joins the curve seen in
the cooling mode, up to 300 K. These results show that com-
plex 1 experiences a process of almost complete thermal
SCO and describes a large hysteresis loop (DT0.5 ꢀ40 K
wide), which implies the existence of bistability of the
system over a significant temperature range. The shape of
this loop is markedly asymmetrical; the cooling branch
(with T0.5fl=133 K) exhibits several small steps resulting in
various slope maxima (Figure 6, left) whereas the warming
branch shows a very sudden jump (T0.5›=173 K), and only
one maximum of the first derivative of cT with respect to T.
The residual amount of HS centres still present after cooling
below 90 K presumably corresponds to marginal portions of
the sample that have remained thermally trapped in that
state. The small decline in cT observed near that tempera-
ture upon warming of the sample is attributed to the relaxa-
tion of some of these trapped residual centres to the LS
Figure 1. Representation of the cation of 1: [FeACTHNUTRGNEUNG
(H4L)2]2+. Grey non-la-
belled atoms are C, and lighter grey atoms are H. For clarity, only H
atoms on non-C atoms are shown.
fragments from adjacent molecules. Each connection is rein-
À
forced by two complementary C H···p interactions
(Figure 3). Stacking p···p interactions between phenol/pyra-
zolyl fragments, or between phenol rings serve to connect
the zig-zag chains to each other forming sheets (Figure 4).
À
In this structure, one ClO4 and the acetone molecules ex-
3122
www.chemeurj.org
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 3120 – 3127

Multiple Transition Dynamics
FULL PAPER
Figure 4. Representation of the connection (at 200 K) between two zig-
zag chains of 1 (one darker than the other) that results from two types of
À
p···p stacking interactions (dashed lines E and F) and one C H···p inter-
action (ellipses c). In fact, the latter is only operative in the LS state (see
main text and Table 1). The cell directions are indicated.
Figure 2. Representation of two cations of
1 (at 200 K) interacting
through two py rings (thick dashed line A) and hydrogen bonds (thin
À
dashed lines) through ClO4 anions. The crystallographic disorder of
À
ClO4 (two equivalent anions shown) and two acetone molecules (also
hydrogen-bonded to the cations) is emphasized; second positions are
shown as lighter grey. Selected atoms are labelled and hydrogen atoms
are not shown.
Figure 5. cT vs. T curve per mol of 1 in the cooling (blue) and warming
mode (red) and after quenching the compound at 10 K and increasing
À
the temperature (black) (Tr.=trapped). The ClO4 anions in the boxes
represent the species with order–disorder transitions and the tempera-
tures and spin state at several crucial points (see main text).
Figure 3. Representation of a zig-zag chain formed along the ab direction
by the cations of 1 (at 200 K), by means of two types of p···p stacking in-
À
teractions (dashed lines C and D) and two types of C H···p interactions
(ellipses a and b). The cell directions are indicated.
fluences differently in one sense than in the other. In view
of this behaviour, the possibility of fully thermally trapping
this system in a metastable state was explored. A polycrys-
talline sample of 1 was cooled rapidly to 10 K under a field
of 5 kG, and its susceptibility was measured while the
system was warmed (Figure 5). At 10 K, the value of cT is
2.70 cm3 KmolÀ1 and slowly increases upon warming at
1 KminÀ1 to a constant value of 3.29 cm3 KmolÀ1 near 50 K.
When the temperature reaches about 99 K, an abrupt de-
crease to a constant value of 0.79 cm3 KmolÀ1 takes place in
the curve, which then suddenly jumps to 3.39 cm3 KmolÀ1 at
around 173 K. This behaviour is clear evidence that the
state. Clearly, the spin transition is coupled to other changes
that influence the dynamics of the process. On the one
hand, the kinetics of the direct process (cooling mode) dif-
fers significantly from that of the inverse transit. On the
other hand, the transformations are highly affected by coop-
erative phenomena. The latter is certainly a consequence of
the extensive network of intermolecular interactions in 1
(vide supra). The asymmetry of the hysteresis must be relat-
ed to some sort of crystallographic phase transition that in-
Chem. Eur. J. 2011, 17, 3120 – 3127
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3123

J. S. Costa, O. Roubeau, G. Aromꢁ et al.
were reproducible over various temperature cycles. The
excess heat capacity (DCp) due to the phase transitions oc-
curring in 1 is shown in Figure 6, right, and coincide with
the maxima of the derivative of cT (Figure 6, left). The cor-
responding excess enthalpy and entropy of the transition
(i.e., DtransH/DtransS) amount to 9.29 kJmolÀ1/54.1 JmolÀ1 KÀ1
and 3.53 kJmolÀ1/26.7 JmolÀ1 KÀ1 for cooling and warming,
respectively, also reproducing the difference in sharpness of
the SCO process. The values of both DtransH and DtransS for
the warming branch are typical of cooperative SCO sys-
tems.[26] Remarkably, even in the cooling regime the excess
entropy is much higher than the value expected solely from
the spin-state change (Rln5=13.4 JmolÀ1 KÀ1), which indi-
cates an additional vibrational component that can be as-
cribed to intermolecular interactions in 1. Because DSC is a
dynamic technique, the smaller values obtained in the cool-
ing regime may be a consequence of the slower dynamics of
one of the processes that shows up because the transition
occurs at a lower temperature upon cooling.
Figure 6. Left: The first derivative of cT vs. T for 1 in the cooling and
warming mode. Middle: Thermal hysteresis of the cell angles of 1 from
single-crystal X-ray diffraction. Right: Excess heat capacity of 1 with de-
creasing and increasing temperature. Blue is the cooling mode, red is the
warming mode and black is the thermal trapping.
compound may be fully trapped into a metastable HS state
through a process of thermally induced excited spin state
trapping (TIESST),[23] remaining in this state until the tem-
perature approaches TTIESST (ꢀ104 K, see Figure S2), at
which the system exhibits 75% thermal relaxation. Restora-
tion of the HS state occurs upon further warming, in the
same conditions as seen before. The value of TTIESST mea-
sured for 1 is among the highest ever observed[24,25] and
could reflect the fact that the SCO is coupled to another
phase transition.
Correlation of the SCO with the structure: To understand
the details of the structural phase transition(s) accompany-
ing the SCO, an extensive variable-temperature crystallo-
graphic investigation of complex 1 was undertaken. Single
crystal X-ray diffraction data were collected at 100 K. The
basic structure is the same as that observed at 200 K, al-
À
though with significant differences. At 100 K, the Fe N
bond lengths in complex 1 now range from 1.928(2) to
1.985(2) ꢅ (av. 1.967 ꢅ), which represents approximately a
9% reduction with respect to the structure at 200 K. This is
consistent with the thermal process of spin transition from
HS to LS observed for 1, which leads to a volume reduction
in the coordination polyhedron from 12.48(5) to 9.76(4) ꢅ3
(a 21.8% reduction) as calculated with IVTON.[27] The dis-
tortion of this polyhedron from octahedral, as characterised
by the parameter S,[28] is reduced from 145.85 to 100.528 on
lowering the temperature. The large values of S are in line
with these observed in FeII/bpp complexes, which is attribut-
ed to both the narrow bite angle of the chelating ligands
and the peculiar Jahn–Teller distortions experienced by
these chromophores.[15] The significant reduction is consis-
tent with the HS to LS transition, which also causes a
change in the q and f angles to 86.43 and 178.53(7)8, respec-
tively. Comparison of both datasets reveals anisotropic
changes to the unit cell parameters (Tables 1 and S2) attrib-
uted to a combination of thermal contraction and the effects
of the crystallographic phase transitions accompanying the
SCO. Thus, the HS to LS transformation results in the fol-
lowing changes: Da=+0.016, Db=+0.071, Dc=À0.13 ꢅ;
Da=+2.06, Db=À0.79, Dg=+0.928. The peculiar fact that
the various cell dimensions evolve in different directions[29]
causes an overall volume change of only DV=À43.6 ꢅ3
(1.6%). It is hard to establish a correlation between the ex-
pansion tensors and the directions of the various types of in-
termolecular interactions because the latter constitute a
complex network (Figures 2–4 and S1) and, therefore, proj-
ect on the three cell dimensions. In fact, the spin and crystal-
Thermal properties: Variations in the thermal properties of
1 associated with these magnetic and structural transitions
were investigated by using differential scanning calorimetry
(DSC) using polycrystalline material. The raw DSC traces
(Figure 7) are consistent with the asymmetric nature of the
SCO of 1, exhibiting a very broad hump in the 145–120 K
region upon cooling and a very sharp and more energetic
anomaly at 172 K detected upon warming. These features
Figure 7. DSC trace for crystals of compound 1 upon cooling (light gray
line) and warming (dark grey line, 10 KminÀ1). Similar results were ob-
served for repeated cycles and for several batches. Note that cooling was
performed at 2 KminÀ1 to obtain a satisfactory evaluation of the LT lat-
tice baseline, which resulted in an apparently even broader anomaly.
3124
www.chemeurj.org
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 3120 – 3127

Multiple Transition Dynamics
FULL PAPER
lographic transitions involve changes to the structural pa-
rameters in all the intermolecular interactions, which leads
to a reinforcement of most of them. A complete list of pa-
rameters describing these interactions at the various temper-
atures is collected in Table 1. These variations notwithstand-
ing, the most remarkable structural changes of the crystallo-
graphic phase transition concern those species that exhibit
disorder at 200 K, which become perfectly ordered at 100 K.
Examples in which the spin state of FeII has been correlated
with the crystallographic disorder of spin-inactive[9–11] or
-active[12] components of the system have been reported. To
shed light on the relationship between the dynamics of the
SCO of complex 1 and its structure, single-crystal X-ray dif-
fraction data were collected under conditions that led to the
various metastable magnetic states identified in this system.
Thus, the structure of 1 trapped into its metastable HS state
was determined after rapidly cooling the sample from room
temperature to 100 K (100 K Tr. HS; C). The structure was
also obtained in the magnetic bistability region, that is, after
slowly cooling the sample from room temperature down to
150 K (150 Kfl, HS; D) and after slowly warming the system
up to 150 K (150 K›, LS; E) from its low-temperature LS
state. The expected magnetic state and temperature of the
system for these structural determinations are indicated in
Figure 5. The main observation from these experiments is
that for the cases in which the system is expected to be in its
HS metastable magnetic state (C and D), all the structural
parameters distinctive of this magnetic state are maintained,
Simulating the SCO of 1: Although a number of phenom-
enological models have been used successfully to describe
the phenomenon of SCO, only the Ising-like model was re-
cently used to introduce the coupling of SCO and the order-
ing of spin-inactive components, such as solvent molecules
or anions.[8] In addition to the usual term describing the cou-
pling between spin-active centres (denoted J), the model in-
cludes terms for the coupling between disordered spin-inac-
tive species (denoted K) and between the spin-active and
spin-inactive moieties (denoted I). It successfully repro-
duced the peculiar transition curves of two SCO complexes
that exhibit such disorder: [Fe
ACHTUNGTRNE{NUNG 2,6-bis(3-methylpyrazol-1-yl)-
pyrazine}2][ClO4]2 with an asymmetric SCO curve that
AHCTUNGTRENNUNG
shows a small abrupt hysteresis for HS fractions above 0.5
and a gradual tail for fractions below 0.5,[8,9] and [Fe(2-pico-
lylamine)3]Cl2·EtOH that has a two-step SCO curve with an
intermediate ordered phase.[8,11] Therefore, we attempted to
use the model with one spin-active and one disordered spin-
inactive sub-lattice [Eqs (5a) and (5b) in reference [8]] to re-
produce at least qualitatively the peculiar SCO of com-
pound 1. As pointed out by the authors of the model, we
found that the coupling between the disordered spin-inac-
tive species and the SCO centres results in asymmetric SCO
curves, the low-temperature (low HS fractions) part being
more gradual. Although this allows reasonable simulation of
the gradual and asymmetric cooling mode for compound 1
(Figure 8, black line) with parameters I=200, J=20, K=
À250, no hysteresis is predicted and the warming curve is
identical to the cooling curve. On the other hand, reproduc-
ing the hysteresis width and its abrupt warming branch re-
quires a completely different set of parameters (I=370, J=
20, K=À250; Figure 8, red and blue lines). In addition, the
resulting warming branch remains less abrupt than the ex-
perimental findings, due to the coupling between spin-active
species and the disorder of the spin-inactive ones, whereas
the cooling branch is more abrupt than the warming branch.
We have indeed found that no set of parameters allows pre-
À
including the crystallographic disorder of one ClO4 anion
and two acetone molecules (Table 1). Likewise, the structure
of the LS compound at the bistability temperature (E) ex-
hibits the features determined for the system in its stable LS
state (D). It is thus interesting to note that complex 1 is
found to exhibit crystallographic disorder when trapped at
100 K, whereas it might be perfectly ordered when prepared
at a higher temperature (150 K) if this state is reached after
warming the LS system from 100 K. To the best of our
knowledge, complex 1 is the first system in which bistability
of the crystallographic order–disorder coupled to a spin
transition has been experimentally demonstrated. This
aspect of the crystallographic phase transition could help to
explain the asymmetry of the magnetic hysteresis. Indeed,
the transition taking place during the cooling process occurs
from a disordered initial crystallographic state. Therefore,
this process is marked by the coexistence of different dy-
namics of the transformation reflected by a complicated
curve for the magnetic HS to LS transition. By contrast, the
changes occurring on warming the system occur from an or-
dered state and, therefore, lead to a much more simple mag-
netic response of the transformation. Further details of the
thermal transition were gathered by determining the cell pa-
rameters of 1 at various temperatures for both branches of
the hysteresis loop. The evolution of these parameters re-
flects the structural changes expected from the SCO
(Table S2), and also exhibit hysteretic behaviour (Figure 6,
middle).
Figure 8. Fraction of HS molecules as a function of temperature calculat-
ed by using Equations (5a) and (5b) from ref. [8]. Black line (warming
and cooling identical): I=200, J=20, K=À250; light and dark grey lines:
I=370, J=20, K=À250. I is the coupling between the spin-active and
spin-inactive moieties, J is the coupling between spin-active centres and
K is the coupling between disordered spin-inactive species.[8]
Chem. Eur. J. 2011, 17, 3120 – 3127
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3125

J. S. Costa, O. Roubeau, G. Aromꢁ et al.
X-ray crystallography: Data were collected at l=0.7515 ꢅ by using a
single-axis HUBER diffractometer on station BM16 of the European
Synchrotron Radiation Facility (Grenoble, France). The crystal habit was
a block that was either orange or purple depending on the temperature.
Measurements were performed at 200 K (orange), 150 K (orange), 100 K
(purple) cooled from 200 K at 5 KminÀ1, 150 K (purple) warmed from
100 K at 5 KminÀ1 and 100 K (orange) by mounting the crystal directly at
100 K from RT. Cell refinement, data reduction and absorption correc-
tions were done by using the HKL-2000 suite.[30] The structure was
solved by using SIR92[31] and the refinement and all further calculations
were carried out by using the SHELX-TL suite.[32,33] All non-hydrogen
atoms were refined anisotropically. At 200, 150 (cooling) and 100 K
(trapping), one of the perchlorate ions presents disorder of two of its
oxygen atoms (O5, O8) over two positions with an occupancy of 0.8:0.2.
Both acetone lattice molecules also present disorder of one of their
methyl carbons over two positions (C2S:C3S and C7S:C8S) with occu-
pancy factors 0.4:0.6 and 0.3:0.7, respectively. At 100 and 150 K (warm-
ing), this disorder is not present. Hydrogen atoms were found by using
difference Fourier maps, except for those of the disordered methyl
groups. Water and phenol hydrogen atoms were refined with thermal pa-
rameters of ꢆ1.5 that of their carrier oxygen and distance restraints. The
rest of the hydrogen atoms were placed geometrically on their carrier
atom and refined by using a riding model. Monitoring of the cell parame-
ters variation with temperature was done on a block crystal of 1 by using
a Bruker APEX II CCD diffractometer on station 11.3.1 of the Ad-
vanced Light Source at Lawrence Berkeley National Laboratory, with
l=0.7749 ꢅ from a silicon 111 monochromator. Datasets were also ac-
quired at both 200 and 100 K to confirm the similarity of the structure.
Crystallographic data for the various structures and several local and lat-
tice parameters of 1 in the different states studied are given in Table 1. A
selection of bond lengths and angles for the FeII coordination sphere is
given in Table S1. Table S2 gathers the cell parameters at various temper-
atures in the range 85–200 K both upon cooling and warming.
diction of hysteresis behaviour with a sharper warming
branch than the cooling one, such as that exhibited by com-
pound 1. Because using two sets of parameters to reproduce
the experimental behaviour of one same compound is sense-
less, modelling the comparatively more gradual cooling
branch of the hysteresis in the SCO curve of 1 necessitates
additional parameters not included in the model. The most
likely explanation is related to the influence of temperature
on the ordering of the spin-inactive species (not included in
the model in reference [8]), which would result in slower dy-
namics in the cooling branch.
Further development of the theoretical model will be re-
quired so as to include the dynamics of the order–disorder
process and ascertain how its coupling with the SCO causes
the asymmetric nature of the cooperativity and hysteresis as-
sociated with these phase transitions in 1 or related com-
plexes.
Conclusion
The functional groups built into the new 3-bpp ligand H4L
are at the origin of the intricate network of packing forces
in 1 and are, therefore, responsible for the large hysteresis
and the dynamic behaviour of the SCO in this compound.
The latter is indeed coupled to a complex crystallographic
phase transition that involves a process of order–disorder,
which results in the asymmetric nature and rich structure of
these thermal transformations. The bistability of the spin
state of 1 is fully correlated with the crystallographic order-
ing of some of its components. This situation is shown here
experimentally for the first time in both states of the bista-
ble region, as well as in the thermally trapped metastable
state. This system represents a promising window for the
study of the dynamics of SCO in relation to its coupling
with crystallographic phase transitions. For example, work is
currently in progress to investigate the dynamics of the HS
state of 1 and its structure within both the bistability range
and the thermally trapped state.
CCDC-793027 (200 K HS), -793028 (150 K HS), -793029 (100 K LS),
-793030 (150 K LS) and -793031 (100 K Tr.) contain the supplementary
crystallographic data for this paper. These data can be obtained free of
charge from The Cambridge Crystallographic Data Centre via
www.ccdc.cam.ac.uk/data_request/cif.
Physical measurements: Infrared spectra were recorded as Nujol mulls
by using a Perkin–Elmer FT-IR instrument. Microanalyses were carried
by using a Perkin–Elmer Series II CHNS/O Analyzer 2400 at the Servei
de Microanꢂlisi of CSIC (Barcelona, Spain). Magnetic measurements (5–
300 K) were performed by using a Quantum Design MPMS-XL SQUID
magnetometer. Background corrections for the sample holder assembly
were applied and diamagnetic corrections of the complexes were per-
formed by using Pascalꢇs constants. Differential scanning calorimetry
(DSC) was performed by using a Q1000 differential scanning calorimeter
with the LNCS accessory from TA Instruments. The temperature and en-
thalpy scales were calibrated with a standard sample of indium by using
its melting transition (156.68C, 3296 JmolÀ1). The measurements were
carried out at scanning rates of 10 or 2 KminÀ1 for several batches of 3
to 8 mg of undamaged crystals taken directly out of the mother liquor
and sealed in aluminium pans with a mechanical crimp, with an empty
pan as reference. To determine the heat capacity, the zero-heat flow pro-
cedure described by TA Instruments was followed, with a synthetic sap-
phire of similar mass to the samples used as the reference compound. An
overall accuracy of ꢀ0.2 K for temperature and ꢀ10% for the heat ca-
pacity was thus estimated over the whole temperature range. The excess
heat capacity, DCp, associated with the phase transitions, was obtained by
estimating a normal heat capacity curve with the high- and low-tempera-
ture data, and subtracting it from the total heat capacity. In this estima-
tion no heat capacity step at the transition temperature was considered.
The deduced calorimetric figures associated with the SCO/order-disorder
transition DtransH (integration of DCp over T) and DtransS (integration of
DCp over lnT) feature the observed difference in sharpness between the
warming and cooling modes. Note, however, that the figures correspond-
ing to the cooling mode are subject to higher error due to the difficulty
of correctly separating the normal heat capacity from the anomaly for
this broader and less energetic peak. Moreover, the low-temperature part
Experimental Section
Synthesis
2,6-Bis{5-(2-hydroxyphenyl)-pyrazol-3-yl}pyridine (H4L): This ligand was
prepared as previously described by our group.[20]
[Fe
acetone (10 mL) was added dropwise with continuous stirring to a solu-
tion of Fe(ClO4)2·H2O (0.0165 g, 0.065 mmol) in acetone (5 mL) in the
ACHTUNGTRENNUNG(H4L)2]ACHTUNGTRENNUNG[ClO4]2 (1): A suspension of H4L (0.0504 g, 0.128 mmol) in
ACHTUNGTRENNUNG
presence of ascorbic acid (0.003 g, 0.017 mmol). The orange solution that
formed was stirred for 1 h at RT before being filtered to remove excess
ascorbic acid. The filtrate was used to prepare layers with ether (volume
1:1). Large, dark orange, polycrystalline aggregates were obtained after a
week, which gave crystals suitable for single-crystal X-ray diffraction
(yield: 12.3 mg, 16%). IR (KBr pellet) n˜ =3414 (s), 3319 (s), 1690 (w),
1613 (s), 1486 (m), 1467 (s), 1355 (w), 1290 (m), 1114 (s), 1087 (s),
755 cmÀ1 (s); elemental analysis calcd (%) for C52H48FeN10O15Cl2: C
52.94, H 4.10, N 11.87; found: C 52.79, H 4.08, N 11.87.
3126
www.chemeurj.org
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 3120 – 3127

Multiple Transition Dynamics
FULL PAPER
used for the determination of the normal heat capacity for the cooling
mode covers a very limited temperature range due to its closeness to the
low-temperature limit of the apparatus.
[13] B. A. Leita, S. M. Neville, G. J. Halder, B. Moubaraki, C. J. Kepert,
J. F. Lꢈtard, K. S. Murray, Inorg. Chem. 2007, 46, 8784–8795.
[14] M. A. Halcrow, Polyhedron 2007, 26, 3523–3576.
[15] M. A. Halcrow, Coord. Chem. Rev. 2009, 253, 2493–2514.
[16] J. M. Holland, J. A. McAllister, C. A. Kilner, M. Thornton-Pett,
A. J. Bridgeman, M. A. Halcrow, J. Chem. Soc. Dalton Trans. 2002,
548–554.
[17] M. A. Halcrow, Coord. Chem. Rev. 2005, 249, 2880–2908.
[18] J. M. Holland, J. A. McAllister, Z. B. Lu, C. A. Kilner, M. Thornton-
Pett, M. A. Halcrow, Chem. Commun. 2001, 577–578.
[19] M. Nihei, H. Tahira, N. Takahashi, Y. Otake, Y. Yamamura, K.
Saito, H. Oshio, J. Am. Chem. Soc. 2010, 132, 3553–3560.
[20] G. A. Craig, L. A. Barrios, J. S. Costa, O. Roubeau, E. Ruiz, S. J.
Teat, C. C. Wilson, L. Thomas, G. Aromꢁ, Dalton Trans. 2010, 39,
4874–4881.
Acknowledgements
G.A. thanks the Generalitat de Catalunya for a ICREA Academia 2008
prize. The authors thank the Spanish MCI through CTQ2009-06959
(J.S.C., G.A.C. and G.A.), support for access to BM16 at ESRF ( EXP16-
01-739), and for a “Juan de la Cierva” (J.S.C.) research fellowship. The
Advanced Light Source (S.J.T.) is supported by the U.S. Department of
Energy under contract no. DE-AC02-05CH11231.
[21] G. Aromꢁ, P. Gamez, J. Reedijk, Coord. Chem. Rev. 2008, 252, 964–
989.
[1] P. Gamez, J. S. Costa, M. Quesada, G. Aromꢁ, Dalton Trans. 2009,
7845–7853.
[22] L. A. Barrios, G. Aromꢁ, J. Ribas, J. S. Uber, O. Roubeau, K. Sakai,
S. Masaoka, P. Gamez, J. Reedijk, Eur. J. Inorg. Chem. 2008, 3871–
3876.
[23] M. Marchivie, P. Guionneau, J. F. Lꢈtard, D. Chasseau, J. A. K.
Howard, J. Phys. Chem. Solids 2004, 65, 17–23.
[24] V. A. Money, C. Carbonera, J. Elhaik, M. A. Halcrow, J. A. K.
Howard, J. F. Lꢈtard, Chem. Eur. J. 2007, 13, 5503–5514.
[25] A. F. Stassen, O. Roubeau, I. F. Gramage, J. Linarꢊs, F. Varret, I.
Mutikainen, U. Turpeinen, J. G. Haasnoot, J. Reedijk, Polyhedron
2001, 20, 1699–1707.
[26] M. Sorai, M. Nakano, Y. Miyazaki, Chem. Rev. 2006, 106, 976–1031.
[27] T. B. Zunic, I. Vickovic, J. Appl. Crystallogr. 1996, 29, 305–306.
[28] P. Guionneau, M. Marchivie, G. Bravic, J. F. Lꢈtard, D. Chasseau,
Top. Curr. Chem. 2004, 234, 97–128.
[29] H. Daubric, J. Kliava, P. Guionneau, D. Chasseau, J. F. Lꢈtard, O.
Kahn, J. Phys. Condens. Matter 2000, 12, 5481–5494.
[30] Z. Otwinowski, W. Minor in Methods in Enzymology, Vol. 276, Aca-
demic Press, New York, 1997, pp. 307–326.
[31] A. Altomare, M. C. Burla, G. Camalli, G. Cascarano, C. Giacovazzo,
A. Gualiardi, G. Polidori, J. Appl. Crystallogr. 1994, 27, 435.
[32] SHELXTL, G. M. Sheldrick, Bruker AXS Inc., Madison, Wisconsin,
2001.
[2] S. Bonhommeau, T. Guillon, L. M. L. Daku, P. Demont, J. S. Costa,
J. F. Lꢈtard, G. Molnar, A. Bousseksou, Angew. Chem. 2006, 118,
1655–1659; Angew. Chem. Int. Ed. 2006, 45, 1625–1629.
[3] P. Gꢉtlich, H. A. Goodwin in Spin Crossover in Transition Metal
Compounds I, Vol. 233, Springer, Berlin, 2004, pp. 1–47.
[4] A. B. Gaspar, M. Seredyuk, R. Gꢉtlich, Coord. Chem. Rev. 2009,
253, 2399–2413.
[5] O. Roubeau, A. Colin, V. Schmitt, R. Clꢈrac, Angew. Chem. 2004,
116, 3345–3348; Angew. Chem. Int. Ed. 2004, 43, 3283–3286.
[6] R. V. Martꢁnez, F. Garcia, R. Garcia, E. Coronado, A. Forment-
Aliaga, F. M. Romero, S. Tatay, Adv. Mater. 2007, 19, 291–295.
[7] M. Cavallini, C. Albonetti, F. Biscarini, Adv. Mater. 2009, 21, 1043–
1053.
[8] D. Chernyshov, N. Klinduhov, K. W. Tornroos, M. Hostettler, B.
Vangdal, H. B. Burgi, Phys. Rev. B 2007, 76, 014406.
[9] D. Chernyshov, M. Hostettler, K. W. Tornroos, H. B. Burgi, Angew.
Chem. 2003, 115, 3955–3960; Angew. Chem. Int. Ed. 2003, 42, 3825–
3830.
[10] G. S. Matouzenko, A. Bousseksou, S. A. Borshch, M. Perrin, S. Zein,
L. Salmon, G. Molnar, S. Lecocq, Inorg. Chem. 2004, 43, 227–236.
[11] V. A. Money, J. Elhaik, I. R. Evans, M. A. Halcrow, J. A. K.
Howard, Dalton Trans. 2004, 65–69.
[33] G. M. Sheldrick, Acta Crystallogr. Sect. A 2008, 64, 112–122.
[12] C. F. Sheu, S. Pillet, Y. C. Lin, S. M. Chen, I. J. Hsu, C. Lecomte, Y.
Wang, Inorg. Chem. 2008, 47, 10866–10874.
Received: November 7, 2010
Published online: February 15, 2011
Chem. Eur. J. 2011, 17, 3120 – 3127
ꢄ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3127