![Fig. 14. Electron density (in c A - 3) at the Fe-N bond critical points as a function of bond length, in lFe(pbenh(NCS)2] and IFe(htrh(NCS)z] · H20.](/figures/fig-14-electron-density-in-c-a-3-at-the-fe-n-bond-critical-1sk61otr.webp)
![Fig. 10. Molecular structure and labelling scheme of [Fe(phenh(NCSh] in the HS-2 metastable state.](/figures/fig-10-molecular-structure-and-labelling-scheme-of-fe-phenh-2gdcm9af.webp)
![Fig. 3. Evolution of the HS fraction Yus(t) between two laser excitations at T = 32 K: derived from ditiraetion data (open squares) and photo-magnetic measurements (full triangles) [22]. The line is a leastsquares tit to the diffraction data.](/figures/fig-3-evolution-of-the-hs-fraction-yus-t-between-two-laser-1q7kvike.webp)
![Fig. 2. Temperature dependence of XM T for powder samples prepared using lhe diffusion method [22], HS-1 denotes lhe high temperature HS .'itate and HS-2 the light-induced mctastable, low temperature HS state.](/figures/fig-2-temperature-dependence-of-xm-t-for-powder-samples-1py40njt.webp)
HAL Id: hal-01007170
https://hal.archives-ouvertes.fr/hal-01007170
Submitted on 19 Feb 2017
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entic research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diusion de documents
scientiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Distributed under a Creative Commons Attribution| 4.0 International License
Out-of-equilibrium charge density distribution of spin
crossover complexes from steady-state
photocrystallographic measurements: experimental
methodology and results
Sebastien Pillet, Vincent Legrand, Hans-Peter Weber, Mohammed Souhassou,
Jean-François Létard, Philippe Guionneau, Claude Lecomte
To cite this version:
Sebastien Pillet, Vincent Legrand, Hans-Peter Weber, Mohammed Souhassou, Jean-François Létard,
et al.. Out-of-equilibrium charge density distribution of spin crossover complexes from steady-state
photocrystallographic measurements: experimental methodology and results. Zeitschrift für Kristal-
lographie - Crystalline Materials, De Gruyter, 2008, 223 (4-5), pp.235-249. �10.1524/zkri.2008.0023�.
�hal-01007170�
1
Out-of-equilibrium charge density distribution
of
spin crossover
complexes from steady-state photocrystallographic measurements:
experimental methodology and results
Sebastien Pillet*·'. Vincent Legrand
1
•
11
, Hans-Peter Weber
111
, Mohamed Souhassou
1
,
Jean-Fran9ois Letard
1
v,
Philippe Guionneau
1
v
and
Claude Lecomte
1
1
Lahoratoire
de
Cristallographie
et
Modelisation des
Materiaux
Mineraux
et
Biologiques,
UMR
CNRS 7036, Nancy-Universite,
BP
239,
54506 Vandoeuvrc-les-Nancy,
France
11
Institut Laue Langevin, 6
rue
Jules Horowitz, BP 156, 38042 Grenoble Cedex 9, France
111
ACCE,
Grand
Vivier,
3R9fi0
St
Aupre,
France
IV
lnstitut
de
Chi
mic
de
la
MatiCrc
CondcnsCc
de
Bordeaux,
ICMCB
CNRS,
Universite
Bordeaux
I,
87
avenue
du
Doctcur Schwcitzcr,
33608 Pcssac Ccdex,
France
Dedicated to the memory
of
Niels
K.
Hansen, our friend and colleague
Electron density I Photocrystallography I
Molecular magnetic material I Spin crossover
Abstract. The electron density distribution of a light-in-
duced molecular excited state,
i.e. the high spin metastable
state of [Fe(phenh(NCS),], was determined from steady-
state photocrystallographic measurements.
We
defined
the
experimental
conc:Htions
under which the accuracy
of
the
measured diffraction
data
is
compatible with
an
electron
density analysis. These include:
(i)
a large structural
and
electronic contrast between high spin (HS)
and
low
spin
(LS) states, (ii)
an
efficient photoconversion under light
irradiation
and
(iii) slow relaxation of
the
HS
metastable
state. Multipolar mode ling
of
the electron density yielded
a deformation density
and
3d-orbital populations for Fe(II)
characteristic of a high spin
(t
2
:e/) electron configuration
and
support
the
assumption
of
significant a-donation
and
n-backbonding of
the
Fe-N
interactions. The electron
density distribution
in
the
intermolecular regions confirms
anisotropic intermolecular interactions with possibly a
layer topology parallel
to
the orthorhombic (ab) plane,
re-
lated
to
the
system cooperativity.
1. Introduction
Single crystal X-ray diffraction under optical excitation,
termed photocrystallography,
is
a promising
and
rapidly
developing field (see
e.g. [1-4]). Many solid-state pro-
cesses can
be
light-triggered or -driven which renders this
technique very appealing for
the
study of out-of-equili-
brium
phenomena such
as
structural
relaxation processes
[5], long-lived metastable states
[1],
short-lived excited
states
[6,
7]
and solid-state photochemical reactions
[4,
8].
Depending on the reversibility
and
time scale (ms to ps)
of
the
solid-state physical phenomenon investigated, sev-
eral photocrystallographic experimental setups have been
designed specifically for laboratory [9], synchrotron X-ray
[10,
11]
and
neutron sources [12]. For
very
long-lived spe-
cies (t > hours), a conventional
laboratory
setup suffices,
while for short-lived species, more sophisticated pump-
probe methods have
to
be
considered. All of these experi-
ments take advantage of
the
fast data collection opportu-
nities that CCD area detectors afford. The usual experi-
mental procedure
for
a photocrystallographic measurement
consists
in
shining light from a lamp or a laser onto a
single crystal;
the
exposure
is
timed empirically to reach a
photo-stationary state,
at
which point diffraction data are
collected
as
quickly
and
accurately
as
possible. The choice
of suitable excitation conditions
in
terms of wavelength,
bandwidth (broad-band or monochromatic), power
and
duration
is
a pre-requisite for
any
accurate photocrystallo-
graphic measurement.
We
recently drew attention
to
the accuracy
and
preci-
sion of the structural parameters derived from a typical
steady-state photocrystallographic experiment.
We
showed
that the accuracy
is
lower than usually assumed;
and
in
order
to
remedy this state
of
affairs we proposed ways
to
improve
the
experimental set-up [13].
In
the
present paper
we
go a step further
and
define
the
conditions under
which
the
quality of
the
diffraction data
is
such
so
as
to
permit a charge density study. The reconstruction
of
the
periodic electron density (ED) distribution of light-induced
metastable states from single crystal diffraction data
is
in-
deed a challenging task.
We
have already reported the ED
distribution of a metastable state obtained
by
rapid ther-
mal quenchinli
to
cryogenic temperature
[3].
This study
2
demonstrated that detailed and accurate
EO distributions
of
out-of-equilibrium states are experimentally within
reach. The results obtained
in
that study encouraged
us
to
pursue
our·
efforts towards the characterization
of
lixht-in-
duced
metastable states,
the
investigation
of
which is
much more delicate.
For
reasons detailed below, we chose
the
spin cross-
over coordination complex
[Fe(phenh(NCS),
l
as
a
proto-
type
of
a photo-active molecular material. Magnetic and
Mossbauer measurements show that
[Fe(phen),(NCShJ
undergoes
at
T
=
176
K
an
abrupt first-order thermal spin
transition from a high spin (HS, S
=
2,
t
2
.'e/)
to
a
low
spin
(LS, S
=
0,
t2/e,'
1
)
electron configuration [14].
In
due
course, the crystal structures
of
both
HS
and
LS
phases
of
this
compound were
determined
from diffraction measure-
ments at 298 K and
130
K
[15];
no
space group change
was evidenced, the corresponding phase transition is iso-
symmetric with orthorhombic
Pbcn
(No. 60)
space group.
The structural differences characterized
are
pronounced
and representative
of
thermal
HS-LS
transitions [16]: a
drastic shortening
of
the
Fe-N(phen)
(from
2.206
A
to
2.009
A)
and
Fe-N(CS)
bond distances (from
2.057
A
to
1.958
A),
and a far more regular FeN,
octahedral
environ-
ment
in
the
LS
state.
At
the origin
of
the structural
changes
is
the redistribution
of
electrons taking place
in
the magnetic spin-active metal ion upon cooling through
the transition.
In
the present work we
will focus on deter-
mining experimentally just this subtle electronic
redistribu-
tion which
in
the present case
is
not thermally induced,
but caused
by
light excitation.
At
very low temperature,
[Fe(phenh(NCShJ exhibits
the
well-known Light-Induced Excited
Spin State Trap-
ping phenomenon ("LIESST"),
by
means
of
which a
HS
metastable state can be efficiently (completely) populated
using filtered white light or a
He-Ne
laser excitation
source [17
-19
j.
This process consists
of
optical pumping
from the thermodynamically stable
LS
state
to
an
intermedi-
ate
1
MLCT state (Metal Ligand Charge Transfer), which
in
turn rapidly decays
to
a quintet
HS
state through
inter-
system crossings
[20-21]
(Fig.
1).
At
very
low
tempera-
ture,
HS
to
LS relaxation occurs extremely slowly through
a multiphonon non-adiabatic tunnel process,
and
much fas-
ter at higher temperature
by
means
of
a thermally acti-
Energy
1
MLCT
states
hv
LS ('A,)
Fe-N bond distance
Fig.
t.
Schematic
energy
diagram
and
LIESST
process.
vated mechanism. The relaxation
rate
kHL
as
a function
of
temperature was derived from photomagnetic measure-
ments, and a
LIESST
relaxation temperature T
LIESST
of
62 K deduced [22]. The structure
of
the light-induced
me-
tastable
HS
state
of
[Fe(phenh(NCShl
(denoted
HS-2
in
the following,
to
distinguish
it
from the room temperature
HS-1
state, Fig.
2)
was
first
detennined
at
30
K
by
single
crystal
ditl'raction,
consecutive to
He-
Ne laser excitation
[23]. The observed structural changes were as large
as
those occurring
at
the thermal
HS-LS transition.
It
is
therefore tempting,
on
the basis
of
all the
charac-
teristics just described,
i.e.:
(i) large structural and
electm-
nic contrast between both spin states, (ii) efficient
photo-
conversion and (iii) slow relaxation
of
the
HS
metastable
state,
to
demonstrate the feasibility
of
an
EO study of a
light-induced metastable state.
This paper
is
organized
as
follows.
In
Section
2,
we
recall methodological aspects
of
photocrystallographic
ex-
periments
and
in
particular the influence
of
inappropriate
experimental conditions on the accuracy or
the
derived
crystal structure. Electron density modeling and analysis
of
the light-induced
HS-2
metastable state
is
described
in
Section 3. Section 4 summarizes our results
and
outlines
future work.
2. Methodology for steady-state photocrystallo-
graphic measurements
2.1
Synthesis and magnetic characterization
It
is
now
well established that the spin transition
characteris-
tics
of
[Fe(phenh(NCSh].
particularly its abruptness
and
completeness (i.e. amount
of
residual paramagnetic species
at
low
temperature), are highly dependent
on
the sample
preparation method as demonstrated
by
magnetic
suscept-
ibility measurements and Mossbauer studies [24, 25].
Sev-
eral synthesis procedures have been described
in
the
litera-
ture, leading
to
two distinct magnetic behaviors. The slow
solvent diffusion procedure results
in
incomplete thermal
spin transition (nearly
14-17%
residual
HS
species), while
the extraction method leads
to
a more abrupt and complete
transition.
On
the other hand, the diffusion method yields
bigger and better quality single crystal samples; reason for
which it has been preferred
in
the present study. Following
previously published procedures
[15],
a H-shaped
double-
tube was filled
on
one side with a methanolic solution
of
I, I 0-phenanthroline and
on
the other side with
stoechio-
metric amount
of
KNCS
and
Fe(S0
4
) ·
7
H,O
dissolved
in
methanol. Large dark single crystals
of
very
good quality
grew then
by
slow solvent diffusion over several weeks. For
the
diffraction measurements, single crystals, with size or
0.05
mm
to
0.25
mm,
of
good quality were then embedded
in
vacuum grease
and
mounted on glass fibers.
Samples from the same preparation than those used for
the
diffraction experiments were characterized
by
magnetic
measurements; they exhibited spin transition properties
in
agreement with published results,
with
residual
HS
species
amounting
to
nearly
19%
at
low
temperature [22,
23]
(Fig.
2).
This incompleteness
of
the transition precludes
any determination
of
the
LS EO
distribution.
3
4
HS-1
HS-2
~·
.......
.:"'
3
0
·-
•
0
•
e
•
•
•
~
2
•
"
hv
•
e
"
•
•
~
•
""~
• LS
,;
:
....
~
.......
><
0
0
50 100 150
200
250 300
Temperature
(K)
Fig. 2. Temperature
dependence
of
XM
T
for powder samples prepared
using
lhe
diffusion method
[22],
HS-1
denotes
lhe
high temperature
HS
.'itate
and
HS-2
the light-induced mctastable, low temperature
HS
state.
2.2
Defining appropriate experimental conditions for
steady-state photocrystallographic measurements
The structural determination and analysis
of
a metastable
state obtained from steady-state photocrystallographic
measurements
is
a difficult task. It requires a careful
choice
of
several crucial experimental parameters,
espe-
cially the light excitation conditions. We have recently
ad-
dressed this point
in
a separate publication l13]; we
briefly discuss below its main conclusions, together with a
justification for the parameters selected in our present
X-ray diffraction experiment, keeping in mind that the
ulti-
mate goal is to get
a
d~ta
set
of
an accuracy suitable for
an ED analysis.
2.2.1
Light excitation and relaxation parameters
Excitation conditions such as wavelength and power
of
the
light source, exposure duration, continuous/pulsed
excita-
tion and
diffraction
measurement temperature, were cho-
sen according to previously published spectroscopic and
photomagnetic results.
In the LS state, the electronic spectrum
of
[Fe(phenh(NCS),j
exhibits
a weak band at
960
nm attrib-
uted to the spin forbidden
1
A
1
~
3
T
1
transition, and an
intense broad absorption band at
580
nm due to a
1
MLCT
(metal to
lig·and
charge transfer) transition
[14].
We
se-
lected the red line
of
an
Ar-Kr
gas laser
(A
=
647 nm)
and alternatively a
He-Ne
laser
(A
=
633 nm), the
wave-
lengths
of
which are close enough to the absorption max-
imum at
580 nm to ensure a large excitation yield. but
at
the same time far enough as to warrant an adequate light
penetration depth
[26].
Monochromatic excitation was
pre-
ferred to broad-band
filtered
white light, even though both
source types have proven to be quite
eftkient
[17-19].
As
a function
of
time, the populations
of
the
mole-
cular ground-state (LS) and
of
the metastable state (HS-
2) result
trom
the competing intluence
of
laser
exci-
tation and relaxation through tunnelling and thermally
activated processes. In the mean field approximation,
according to Enachescu
et al.
[27, 28], the corresponding
macroscopic evolution equation for a spin transition sys-
1.04
1.02
1.00
l
~
,._
0.98
0.96
l
0.94
I
~
l
\
XRO
prese
nt
study
netism
Photomag
.-
0 1000 2000 3000 4000 5000 6000 7000
time (s)
Fig. 3. Evolution
of
the
HS
fraction
Yus(t)
between two laser excita-
tions at
T =
32
K:
derived from
ditiraetion
data (open squares) and
photo-magnetic measurements (full triangles) [22].
The line is a least-
squares tit to the diffraction data.
tern is written as
dyHS
dt
=
cpphoto-ex.citation
-
<Prelax.ation
ff
=
r
a(
I-
YHs)-
YHskuL(y,
T)
(I)
where
YHs
is the fraction
of
HS species. The first term on
the right-hand side
of
(I)
is the linear photo-conversion
rate, written as the product
of
the incident light intensity
J'ff
at the laser wavelength and
a
response factor
a
de-
pending on the absorption cross section and the quantum
yield
of
the photo-conversion process. The second term
of
(I)
is the HS to LS self-accelerated relaxation
of
rate con-
stant
kHL
(y,
T),
expressed as the product
of
a temperature
dependent rate constant
kHL(T)
and an acceleration
expo-
nential factor exp
[a(T)(l
-
YHs)]
[20].
In the photo-sta-
tionary state. when the population
of
the metastable state
has stopped evolving, the conversion percentage results
from the opposing terms
in
equation 1 and quantitatively
depends on
/'rr,
a
and
kHL(y,
T).
Once the optimum exci-
tation wavelength is chosen,
a
thus being essentially fixed,
1"
1
is
tuned so as to counter the HS-LS relaxation. This
latter has been experimentally determined as a function
of
temperature from photomagnetic and retlectivity
measure-
ments [22]. For illustrative purposes, the relaxation rate at
T
=
32
K,
i.e.
the decrease in
YHs.
is
shown in Fig. 3.
This relaxation is fairly slow but still significant; we thus
anticipate that this effect has to be accounted for with
great care in both the diffraction measurements and subse-
quent structural
refinements.
2.2.2 Diffraction measurement strategy
and the photo-stationary state
Diffraction measurements were performed in two different
experimental environments, the influence
of
which on the
photo-conversion process is described below. Common to
both environments was the use
of
an open flow He cryo-
system (Helijet, Oxford Diffraction) and a CCD detector.
Both low temperature and fast data collection afforded a
maximal reduction
of
relaxation efiects.
For the first measurement, the crystal was mounted on
our laboratory Xcalibur diffractometer (Oxford Diffraction)
4
Cyclic excitation
Laboratory
32K
data
collection
(l)
A
(2)
v"'<:C:=:::::::::::::~::::::::::;::=
(3)
Mean
relaxation
'==::;;;;:;;;=========:;
Time
(I)
2h
Continuous
excitation
Synchrotron
ISK
data
collection
:(23,:)
r
.....................
'
L '
I .
··11
.~;,.'f~
..•
;i
~==============:•
Time
(I)
__
.
___
.........
~hoto-stationary
state
(
1)
laser
excitation
(2)
Metastable
state
population
y
11
,(t)
(3)
Diffraction measurement
Fig.
4.
Excitation and
diffraction
measurement
cycles
for the labora-
tory
(32
K)
and ESRF (15 K) data collections.
and cooled down
to
32 K (well below
T
LIESST
=
62 K)
by
!lash cooling
to
prevent any sample degradation upon pas-
sing through the thermal transition. The single crystal
was
then exposed at 32 K during
150
s
to
the 64 7 nm line
of
an
Ar-Kr
laser operating
at
50
mW,
while the sample was
continuously rotated
tG
ensure a spatially homogeneous
excitation. Lattice parameters were determined afterwards;
their values confirmed the
LS
to
HS-2
photo-conversion
when compared with those reported
by
Marchivie
et al.
[23]. A complete set
of
X-ray
diffraction
data was subse-
quently collected over nearly 22 hours, using
an
w-scan
over
337',
a
1'
frame-width and
lOO
s exposure-time per
frame.
Since
at
'this temperature the HS-2 lo
LS
relaxation
is
not negligible, the laser excitation (647 nm, 50
mW)
was repeated every two hours during
15
s,
which according
to
Fig. 3 is supposed
to
reduce relaxation effects to less
than 1
%.
This
cyclic excitation/data collection procedure
is
shown in Fig. 4 (top).
In
spin crossover materials, the cell volume
of
the
HS-
2 and
LS
states differs
by
a large amount, which corre-
lates to a first approximation linearly with the fraction
YHS·2
of
HS-2
species, according
to:
V(YHs-2)
=
YHS-2
·
VIIS-2
+
(I -
YHs-2)
· Vts ·
(2)
By
monitoring the evolution
of
the unit cell volume
during the course
of
the data collection
of
the
HS-2 state,
the stability
of
this photo-stationary slate can be estimated
in situ, directly from the ditlraction measurement, the cell
volume at time
t
being approximated by:
V(t)
=
YHs-2(1)
·
VHS-2
+
(1-
YHs-2(t))
·
VLs
·
(3)
This method has already been successfully used during the
X-ray diffraction measurements
of
the metastable state
of
the spin-crossover complex
[Fe(btr)z(NCS)zj ·
H,O [3].
In
the case
of
[Fe(phen),(NCS)zJ. the relaxation rate derived
from the time dependence
of
the unit cell volume
at
T
= 32 K
is
depicted in Fig. 3 [29]. The agreement with
the photomagnetic results is satisfactory; the continuous
YHs(t)
decay observed indicates a slight but significant
relaxation between the 2-hourly laser ex:citations. This
time interval
is
therefore certainly not sufficiently short
to
prevent significant HS-2 to
LS
relaxation
at
32
K.
In
other
words, the diffraction measurement in our home laboratory
has not been performed on a purely photo-stationary state
as represented schematically on Fig. 4 (top). Accordingly,
the complementary population
of
metastable and ground-
state species
is
changing during the course
of
the
ditl'rac-
tion measurement
as
metastable state molecules progres-
sively convert back
to
the ground-stale.
A second data collection was then performed using
im-
proved excitation conditions and
an
increased X-ray
flux
provided
by
a synchrotron source. Since HS-2
to
LS
re-
laxation strongly affects the quality
of
our diffraction data,
a lower temperature
(T
=
15
K)
than for the laboratory
case
(T
= 32
K)
was selected. The relaxation term
in
Eq.
(I)
is therefore reduced further; hence we expect a higher
photo-conversion throughout the diffraction measurement.
Data were collected on beam line
BMOJ
A at the
ESRF,
using a
KUMA
6-circles diffractometer equipped with
an
Onyx (Oxford
Diffraction) CCD area detector and a Heli-
jet cryosystem. The wavelength
(.i.
=
0.7100(1)
A)
was
ca-
librated with Fit2D software
[301
using
LaB, NJST
stan-
dard. Light excitation was performed with a
He-Ne
laser
operating at
1.25
mW.
Since
the
HS-2
to
LS
relaxation
was nol negligible
if
excitation took place every
two
hours, laser excitation was applied continuously (see
Fig. 4 bottom). Two detector positions
(26
equal
50'
and
130') were used with a scan width
!'>.r:p
=
I'
and a 4 s
exposure time per frame, providing a higher resolution
(
1.0
A
1
)
than in the laboratory measurement
(0.8
A-I).
However, due to the shadowing
of
the detector
by
the
He-
lijet nozzle, the completeness was reduced to 72.2% for
the whole
26
range, but reached 96%
up
to
0.8
A
-I
reso-
lution.
For both 32 K laboratory and
15
K
ESRF
data sets, the
diffraction
profiles were integrated using
CRYSALIS
[31]
and Gaussian analytical absorption corrections were ap-
plied. The data were reduced using
SORTAV
[32]. The
crystal structures were first relined with SHELX97
[33].
Non-hydrogen atoms were refined anisotropically, hydro-
gen atoms were refined isotropically without any con-
straint. More details on data collection and structural
re-
finement results are collected in Table I. For the ESRF
data set, the good
R;"'
agreement factor
is
evidence
of
the
high precision
of
the corresponding structure factor ampli-
tudes.
2.2.3
Accuracy
of
the derived
structural
parameters
We
have shown previously that for the 32 K laboratory
measurement, the probed sample corresponds
to
a continu-
ously relaxing system.
Tn
such a case, or for
an
incom-
plete photo-conversion, the probed single crystal contains
both ground-state and metastable state species, with time
dependent population and unknown spatial distribution.
The corresponding structural analysis
of
the
HS-2 meta-