Syntheses, crystal structures, and thermal stabilities of polymorphs of Cr(thd)3

Article abstract of DOI:10.1002/zaac.201000160

The syntheses, crystal structures, and thermal stabilities of six polymorphs of Cr(thd)₃ [(thd)^- = anion of H(thd) = C ₁₁H₂₀O₂ = 2,2,6,6-tetramethylheptane-3,5-dione] were studied using X-ray diffraction and differential scanning calorimetry. Compound 1 is thermodynamically stable below ca. 110 °C and forms plate-shaped crystals [a = 9.927(5), b = 18.010(5), c = 21.427(5) A, and β = 95.461(5)° at 295 K; space group C2/c]. Its crystal structure is isotypic with Mn(thd)₃. Polymorph 4 is metastable to 1, and the only way to prepare 4 appears to be by precipitation from solution. Crystals of 4 are needle shaped [a = 28.54(3), b = 19.14(2), c = 21.92(2) A, and β = 97.31(2)° at 295 K; space group C2/c] and this structure is isotypic with monoclinic Co(thd)₃. Modifications 2, 2*, and 5 also appear as needle-shaped crystals [orthorhombic: a = 18.97(7), b = 18.69(7), and c = 10.62(5) A for 2 and tetragonal: a = b = 18.93(5) and c = 10.57(4) A for 5, both at 110 K]. Crystals of 2 and 5 have limited lifetime, which depends on storage temperature and exposure to X-rays. The conversion sequence is: 5 → 2 → 1, where the first step takes 3 to 4 h and the second some 20 h. Attempts to establish detailed accounts of the orthorhombic Co(thd) ₃-type structure of 2 and 2* by SXD were unsuccessful due to disorder. The structural distinction between 2 and 2* appears to be associated with rotational disorder among the Cr(thd)₃ molecules. 2* and 3 are high-temperature modifications with stability range ca. 155-200 and ca. 200-235 °C, respectively [Cr(thd)₃ melts at ca. 235 °C]. Quenched 2* can be retained as metastable at room temp. for some weeks, whereas 3 is not quenchable. High-temperature PXD data show that 3 is cubic [a = 13.357(4) A at 225 °C; with strong rotational disorder of the molecules]. There are only minor variations in bond lengths and angles between 1 and 4 or from 100 to 295 K, and disorder is limited.

Full text of DOI:10.1002/zaac.201000160

ARTICLE
DOI: 10.1002/zaac.201000160
Syntheses, Crystal Structures, and Thermal Stabilities of Polymorphs of
Cr(thd)3
Mohammed A. K. Ahmed,[a] Helmer Fjellvåg,[a] Arne Kjekshus,*[a] Renie K. Birkedal,[a]
Poul Norby,[a] David S. Wragg,[a] and Nalinava Sen Gupta[a]
Keywords: Chromium; Cr(thd)3; Crystal structure; Polymorphism; Thermal stability
Abstract. The syntheses, crystal structures, and thermal stabilities of depends on storage temperature and exposure to X-rays. The conver-
six polymorphs of Cr(thd)3 [(thd)– = anion of H(thd) = C11H20O2
=
sion sequence is: 5 → 2 → 1, where the first step takes 3 to 4 h and
the second some 20 h. Attempts to establish detailed accounts of the
orthorhombic Co(thd)3-type structure of 2 and 2* by SXD were unsuc-
cessful due to disorder. The structural distinction between 2 and 2*
appears to be associated with rotational disorder among the Cr(thd)3
molecules. 2* and 3 are high-temperature modifications with stability
range ca. 155–200 and ca. 200–235 °C, respectively [Cr(thd)3 melts at
ca. 235 °C]. Quenched 2* can be retained as metastable at room temp.
for some weeks, whereas 3 is not quenchable. High-temperature PXD
data show that 3 is cubic [a = 13.357(4) Å at 225 °C; with strong
rotational disorder of the molecules]. There are only minor variations
in bond lengths and angles between 1 and 4 or from 100 to 295 K,
2,2,6,6-tetramethylheptane-3,5-dione] were studied using X-ray dif-
fraction and differential scanning calorimetry. Compound 1 is thermo-
dynamically stable below ca. 110 °C and forms plate-shaped crystals
[a = 9.927(5), b = 18.010(5), c = 21.427(5) Å, and β = 95.461(5)° at
295 K; space group C2/c]. Its crystal structure is isotypic with
Mn(thd)3. Polymorph 4 is metastable to 1, and the only way to prepare
4 appears to be by precipitation from solution. Crystals of 4 are needle
shaped [a = 28.54(3), b = 19.14(2), c = 21.92(2) Å, and β = 97.31(2)°
at 295 K; space group C2/c] and this structure is isotypic with mono-
clinic Co(thd)3. Modifications 2, 2*, and 5 also appear as needle-
shaped crystals [orthorhombic: a = 18.97(7), b = 18.69(7), and c =
10.62(5) Å for 2 and tetragonal: a = b = 18.93(5) and c = 10.57(4) Å
for 5, both at 110 K]. Crystals of 2 and 5 have limited lifetime, which and disorder is limited.
to be filled by the present ongoing project. Recent findings
1. Introduction
([6] and references cited therein) suggest that the occurrence
of polymorphs is more the rule than the exception for these β-
diketonato complexes.
The 3d metal complexes of beta diketones are both useful
and interesting. They are used as reagents for NMR spectro-
scopic purposes [1–3], catalysts for oligomerization and po-
lymerization [4] and precursors in atomic layer deposition
(ALD) of thin films [5] among other things. From a structural
point of view, they provide an excellent example of tempera-
ture dependent rotational disorder. This paper aims at clarify-
ing basic chemistry issues [6] involved with synthesis and pu-
rification of β-diketonato complexes. It is specifically focused
on Cr(thd)3, which is frequently mentioned in the literature [1–
3, 7–21].
Our research group [6] has recently given a brief overview
on accumulated structural knowledge on (acac)– [= anion of
H(acac) = acetylacetone = C5H8O2 = pentane-2,4-dione] and
(thd)– [= anion of H(thd) = C11H20O2 = 2,2,6,6-tetramethylhep-
tane-2,4-dione] complexes of the 3d metals (M). Some of the
several completely blank spots on structural information along
the M(acac)2, M(acac)3, M(thd)2, and M(thd)3 series are aimed
The incomplete structural data available for these complexes
makes it difficult to see a pattern along the series from M =
Sc to Zn [6]. Insight into the structural features at and around
room temp. is limited, and there is virtually no information
available for higher temperatures. The structural situation just
below the melting point of molecular complexes is of general
interest since the molecular units may be expected to exhibit
appreciable orientational (rotational) disorder just before melt-
ing sets in [22]. A first theoretical foundation for a research
specialty on “rotational motion of molecules in crystals” was
laid by Pauling [23] and Stern [24] in 1930/31, and the experi-
mental activities really flourished in the late 1930’s. Signifi-
cant contributions to the progress came from this department
[25–27], inter alia demonstrating that owing to steric hindrance
the rotation of the molecular units cannot be entirely free [27].
With access to unit-cell dimensions and atomic co-ordinates
of sufficiently high precision it becomes highly meaningful to
search for structural details as well as coarser traits, which
differ between polymorphs of Cr(thd)3. It is of particular inter-
est to explore the atomic arrangement around the central chro-
mium atom (viz. identify variations in Cr–Oket bond lengths
and Oket–Cr–Oket bond angles). The β-diketonato skeletons are
* Prof. Dr. A. Kjekshus
Fax: +47-22855441
E-Mail: arne.kjekshus@kjemi.uio.no
[a] Center for Materials Science and Nanotechnology
Department of Chemistry
University of Oslo
P.O. Box 1033 Blindern
0315 Oslo, Norway
View this journal online at
wileyonlinelibrary.com
2422
© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Z. Anorg. Allg. Chem. 2010, 636, 2422–2432

Polymorphs of Cr(thd)3
Several procedures were tried for preparation of crystals of the various
polymorphs of Cr(thd)3. Crystals grown from solutions appeared, in
general, to provide specimens of better quality for SXD work than
those obtained by sublimation. The specific crystal specimens used for
collection of SXD data were prepared as follows.
also natural targets for careful attention. Focal points of inter-
est are here planarity versus non-planarity of the chelate ring
and whether there are significant variations in Oket–Cket and
C
ket–Cket distances and associated angles. Another aspect,
which deserves attention is variations in disorder either in the
connection between the diketonato ligands and the central
chromium atom (viz. a major source of rotational disorder) or
within the tert-butyl groups. Temperature induced changes in
disorder as well as distinctions in modes of disorder between
polymorphs are of interest in this context. However, designing
appropriate models for such disorder is challenging and, in
general, the extent of disorder is difficult to quantify. More-
over, disorder is intimately correlated with the thermal motions
of the atoms.
1 and 4 were crystallized from solutions of Cr(thd)3 in acetonitrile
(either from saturated solution at b.t. and left for slow cooling to room
temp. or from diluted solution left for slow evaporation of the solvent
at room temp.). Suitable single crystals of these polymorphs were se-
lected on inspection of the products with an optical microscope.
2 (and its tetragonal variant 5 in favorable cases) was (were) obtained
in the form of seemingly single-crystal specimens by sublimation of
Cr(thd)3 at 180 °C (s.t.) and deposition at 90 °C (d.t.). Initial SXD
tests of several of these specimens also suggested that they could be
good single crystals. However, attempts to solve the structure demon-
strated that these crystals were burdened with severe disorder. Another
difficulty for SXD work with 2 is its limited lifetime on exposure to
X-ray.
2. Experimental Section
2.1 Reactants and Solvents
2* was obtained in attempts to improve the crystal quality of 2 by heat
treatment of the specimens in sealed, thin-walled silica-glass tubes at
130–160 °C for a few days, followed by quenching in ice-water, but
these endeavors resulted in even more severely disordered specimens.
CrCl3·6H2O (Aldrich, ≥ 96 %), H(thd) (Aldrich, purum, ≥ 98 %), urea
(Aldrich, 99 %), absolute (abs.) ethanol (Arcus, prima), methanol
(Merck, p.a.), 1-propanol (Aldrich, 99.5 %), 2-propanol (Fluka, ≥
99.5 %), tert-BuOH (Merck, p.a.), acetone (Prolabo, for HPLC), 1,4-
dioxane (Fluka, p.a., ≥ 99.5 %), acetonitrile (Merck, for GC, ≥
99.8 %), toluene (Merck, p.a.), tetrahydrofuran (Prolabo, p.a.), ben-
zene (Fluka, p.a., ≥ 99.5 %), and hexane (Fluka, p.a., ≥ 99.5 %) were
used as reactants and/or solvents, mostly without further purification.
Methanol, ethanol, and toluene were dried, distilled and stored over
molecular sieves.
2.3 Powder X-ray Diffraction (PXD)
All samples were characterized by PXD at room temp. with a Siemens
D5000 diffractometer (capillary geometry) using monochromatic Cu-
Kα1 radiation (λ = 1.540598 Å) from an incident-beam germanium
monochromator and a position-sensitive detector (Braun); Si (a =
5.431065 Å) served as internal standard. The diffraction patterns were
collected over the 2θ range 3–90° and auto-indexed with help of the
DICVOL [28] and TREOR [29] program packages. Unit-cell dimen-
sions were obtained by least-squares refinements using the METRIC
program [30].
2.2 Syntheses
All syntheses were performed in round-bottomed flasks equipped with
reflux condenser and magnetic stirrer. The reaction mixtures were re-
fluxed at boiling temperature (b.t.) or heated at fixed lower tempera-
tures for different periods of time. Unless otherwise mentioned the
solid products were filtered through a sinter-glass funnel, washed,
dried under vacuum, and finally subjected to sublimation (Büchi-type
B-580 apparatus at ca. 0.05 mbar pressure) followed by deposition on
the cold finger at 50–90 °C and/or recrystallized from various solvents.
The phase purity of all final products was ascertained by PXD.
The variable-temperature PXD measurements were collected on the
same instrument. Samples were packed between plugs of silica wool in
0.7 mm internal diameter silica-glass capillaries. The capillaries were
mounted in a flow cell [31] and nitrogen was flowed over the sample
during heating. The temperature at the position of the sample was cali-
brated using the thermal expansion of silver as external standard.
Apart from a scale-up by a factor of 3 and use of proper 1:3 stoichiom-
etry for the CrCl3·6H2O to H(thd) content of the reaction mixture, the
preparation of Cr(thd)3 mostly followed reference [14]. The reaction
2.4 Single-crystal X-ray Diffraction (SXD) Analysis
mixture of CrCl3·6H2O (0.033 mol), H(thd) (0.100 mol), urea Crystals were mounted on thin glass fibers clamped on brass pins.
(1.0 mol) distilled H2O (75 mL), and abs. EtOH (195 mL) was heated Diffraction data were collected at temperatures between 100 and 485 K
under reflux at boiling temperature for 1 to 5 d with stirring. After on a Bruker D8 Apex II diffractometer (Mo-Kα radiation) equipped
cooling to room temp., distilled water (300 mL) was added and the with an Oxford Cryosystems Cryostream Plus cooling device. The data
mixture was stirred for 1 h. The liquid phase was filtered off through were integrated with SAINT [32] and corrected for absorption using
a sinter-glass funnel and the solid product washed several times with SADABS [33]. Structures were solved by direct methods with the pro-
distilled water and then dried under vacuum at 60 °C for 1 d, sublimed gram SIR2004 [34] or SHELXS [35] and refined using full-matrix
at 90–180 °C and/or recrystallized from various solvents (overall total least-squares against |F|2 with SHELXL as implemented in Farrugia’s
yield ca. 77 %). Cr(thd)3 was obtained as a deep purple crystalline WinGX suite [36]. All non-hydrogen atoms were refined allowing for
solid in three polymorphs [plate- (1) and/or needle-shaped (2 and 4) anisotropic displacement parameters (introduced for chromium already
crystals] depending on the purification treatment; bold-faced numbers from the beginning and for carbon and oxygen at the penultimate stage
designate the polymorphs. Elemental analysis for various phase-pure before the hydrogen atoms were added). Hydrogen atoms were as-
samples of modifications 1, 2, and 4 of Cr(thd)3 (C33H57O6Cr) gave signed to idealized positions and refined with isotropic thermal param-
virtually identical results; typical: C 65.88 (calcd. 65.86), H 9.48 eters proportional to the thermal parameters for the atoms to which
(9.55), O 15.88 (15.95), Cr 8.64 (8.64) %.
they are attached (riding model).
Z. Anorg. Allg. Chem. 2010, 2422–2432
© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.zaac.wiley-vch.de
2423

A. Kjekshus et al.
ARTICLE
Table 1. Single-crystal data and relevant parameters used in structure refinements of complexes 1 and 4. Estimated standard deviations are given
in parentheses.
1 (295)
4 (295)
4 (110)
Empiric formula
M
C33H57O6Cr
601.033
293
C33H57O6Cr
601.033
293(2)
C33H57O6Cr
601.033
110
T /K
Crystal system
Space group
a /Å
Monoclinic
C2/c
Monoclinic
C2/c
Monoclinic
C2/c
9.927(5)
18.010(5)
21.427(5)
90.00
28.54(3)
19.14(2)
21.92(2)
90.00
27.924(3)
18.6192(19)
21.627(2)
90.00
b /Å
c /Å
α /°
β /°
95.461(5)
90.00
97.312(16)
90.00
98.296(1)
90.00
γ /°
V /Å3
3813(2)
4
11877(21)
12
11126.7(19)
12
Z
D /g·cm–3
μ(Mo-Kα) /mm–1
F(000)
1.048
1.010
1.078
0.335
0.322
0.344
1308
3924
3924
Crystal size /mm
Radiation /Å
θ min, max / °
No. reflections meas.
No. unique reflect.
Rint
0.30 × 0.60 × 0.80
0.71073
1.9, 21.6
8616
0.30 × 0.40 × 1.00
0.71073
1.6, 22.1
41984
0.21 × 0.27 × 0.91
0.71073
1.9, 26.5
43366
2213
7304
11477
0.206
0.040
0.023
Obs. data [I > 2σ(I)]
No. parameters
R1 [I > 2σ(I)]
wR2 (all data)
GOF
686
4722
9155
183
560
542
0.0755
0.2463
0.91
0.0652
0.2222
1.04
0.0473
0.1442
1.02
Min, max resid /e·Å–3
–0.29, 0.18
–0.22, 0.31
–0.40, 0.55
Information on the crystals, data collection, and reliability factors for ent samples rarely differed more than 0.3 °C. Enthalpy was calcu-
the reported structures is recorded in Table 1 together with unit-cell lated on the bases of the melting enthalpy of the Perkin–Elmer indium
dimensions and space-group assignments.
standard. Transition enthalpies were determined by integration of the
heat-flow peaks, values obtained in parallel runs agree within
2 J·g–1; data mentioned in text and illustrations refer to rounded
averages.
CCDC-786169, -786170, and -786171 contain the supplementary crys-
tallographic 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.
2.7 Magnetic Susceptibility
Magnetic susceptibility measurements were performed (temperature
range 2–300 K, magnetic field 1 kOe) with a Magnetic Property Meas-
urement System; Quantum Design. The sample mass was 15–30 mg
and the samples were contained in gelatine holders during the measure-
ments.
2.5 Elemental Analysis
Elemental analysis was done by standard combustion technique at Ilse
Beetz or Birmingham University (UK).
2.6 Thermogravimetric and Thermal Analysis
3. Results and Discussion
3.1 Syntheses and Product Purification
Thermogravimetric (TG) analysis was performed with a Perkin–Elmer
TGA7 system in a nitrogen atmosphere. Silica-glass containers were
used as sample holders. The heating rate was 5 and/or 10 °C·min–1,
the temperature interval covered was 30–700 °C, and the sample mass
was 15–30 mg.
The as-synthesized batches of Cr(thd)3 according to the rec-
ipe in reference [14], contained two (occasionally three) phases
with 4 as the major constituent, the mixed ligand complex
[Cr(thd)2(OEt)]2 as by-product, and irregularly small amounts
of 1 (vide infra). The instructions in reference [14] certainly
provide a good yield of Cr(thd)3 (some 90–95 %), but the by-
product [Cr(thd)2(OEt)]2 may cause some trouble if sublima-
tion is chosen for purification. An account of synthesis (includ-
ing sublimation), crystal structure, etc. for [Cr(thd)2(OEt)]2 is
Differential scanning calorimetry (DSC) was made with a Perkin–
Elmer Pyris 1 system. Coarsely powdered specimens were contained
in sealed 50 μL aluminum pans during the measurements and (as an
extra measure against oxidation) the containers were surrounded by
flowing nitrogen gas. The heating rate was 5 and/or 10 °C·min–1, the
temperature range covered was 30–500 °C, and the sample mass was
2–3 mg. The temperature scale was calibrated against various stand-
ards. Transition temperatures assessed from parallel runs on 4–6 differ- to be published separately [37]. An efficient means to clean
2424 www.zaac.wiley-vch.de
© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Z. Anorg. Allg. Chem. 2010, 2422–2432

Polymorphs of Cr(thd)3
Cr(thd)3 contaminated by [Cr(thd)2(OEt)]2 is to take advantage 3.2 Magnetic Susceptibility
of the high solubility of Cr(thd)3 and low solubility of
The temperature dependence of the reciprocal magnetic sus-
[Cr(thd)2(OEt)]2 in acetone. With acetone as solvent virtually
complete separation is attained already after one recrystalliza-
tion. An alternative approach could be to shift to MeOH–H2O
as solvent since [Cr(thd)2(OMe)]2 [38] appears to cause less
trouble than [Cr(thd)2(OEt)]2 for sublimation of Cr(thd)3.
Attempts to perform these syntheses at a temperature be-
tween room temp. and ca. 50 °C were unsuccessful, almost
certainly, owing to insufficient access to the NH3 at tempera-
tures where the hydrolysis has not really commenced [37]. The
visual-inspection-based description of the products in reference
[14] refers to the same phases as observed in this study, viz.
green fluffy powder, purple platelets, and ruby-red needles re-
fer to [Cr(thd)2(OEt)]2, 1, and 4, respectively. Reference [14]
does not provide information about the relative amounts of the
three phases. The present findings strongly suggest that reflux-
ing of the reaction mixtures for 1 d at ca. 90 °C is sufficient
to ensure that the reaction is completed and equilibrium estab-
lished. Modification 4 is evidently the kinetically preferred
polymorph of Cr(thd)3 under these conditions. The unsystem-
atic appearance of the small amount of 1 in the as synthesized
batches may possibly be attributed to temperature fluctuations
in the reaction mixtures, whereas the formation of the
[Cr(thd)2(OEt)]2 as by-product is controlled by a different reac-
tion scheme [37].
Several solvents for growth of crystals were tested in this
study, e.g., acetone, ethanol, ethanol–water, methanol, metha-
nol–hexane, 2-propanol, and acetonitrile. The crystal growth
was performed at selected temperatures between room temp.
and boiling temp., starting from nearly saturated solution,
which was left for gradual evaporation of the solvent. The
trend in these experiments is that when the crystallization took
place below 50 °C mostly phase-pure 1 was obtained, mixtures
of 1 and 4 occurred when the crystallization took place at 55–
80 °C, whereas essentially phase-pure 4 was found at and
above 90 °C.
The choice of deposit between 50 and 90 °C in the sublima-
tion-induced crystal growth of Cr(thd)3 appeared as a conve-
nient coincidence. Polymorphs 1 and 2 divide the accessible
d.t. interval between them, compound 1 rules from 50 to ca.
85 °C and compound 2 from 85 to 90 °C. Small amounts of
2, which occur sporadically in the batches of 1 and vice versa
are attributed to temperature fluctuations at the cold-finger (in
particular above 80 °C). Two significant observations should
be emphasized: (i) Modification 4 has never been observed in
any of the numerous batches of Cr(thd)3 treated by sublima-
tion. The present findings show that preparation of phase-pure
4 demands dissolving and subsequent crystallization of
Cr(thd)3 from a suitable solvent. (ii) A systematic re-examina-
tion of the hundreds of PXD diagrams of as-synthesized, crys-
tallized, and sublimed products shows that 2 undergoes a
change to 1 on storing at room temp.
ceptibility for Cr(thd)3 is virtually identical for the polymorphs
1, 2, and 4 and closely follow (correction for induced diamag-
netism [39]) the Curie law with a paramagnetic moment μp =
3.85 μB (S = 1.49 ≈ 3/2 according to the “spin-only” approxi-
mation). The findings concur with the expectations for Cr(thd)3
as an ideal mononuclear CrIII complex with regard to valence.
3.3 Structure Determinations and Refinements
Already from the unit-cell dimensions it is inferred that 1, 2,
and 4 are isotypic (concept defined as in reference [40]) with
Mn(thd)3 [41], o-Co(thd)3 (o = orthorhombic) [6], and m-
Co(thd)3 (m = monoclinic) [6], respectively. Owing to the sim-
ilar atomic sizes and the almost equal X-ray scattering factors
of the metal atoms of these complexes, the virtually identical
appearance (with regard to peak locations as well as intensity
distribution) of corresponding pairs of observed PXD diagrams
further strengthen the claim of isotypism, and the structures of
the prototypes should provide quite reasonable approximations
for the corresponding chromium complexes.
The refinements of the crystal structures of 1 and 4 (with
starting models from direct methods or the Mn(thd)3- and m-
Co(thd)3 structures, respectively) run rather smoothly and the
structures show little evidence of disorder. This result has been
confirmed from data collected with several specimens of each
type.
The situation for 2 and 2* is quite different. Several of the
numerous crystal specimens of these variants gave data with
promising good internal reliability factors (around 0.03) and,
to begin with, we expected that refinements should proceed as
smoothly as for 1 and 4. As soon as the six oxygen atoms of
the ligands-to-central-atom attachments had been identified
and accepted, another set (B) of six oxygen atoms turned up,
related to the first set (A) by a rotational shift of approximately
45° (see Figure 1). Despite numerous attempts with different
models to account for the implied disorder it proved impossi-
Figure 1. Mode of rotational disorder in the Cr-to-O attachment in 2*
as obtained in attempts to solve the structure by direct methods (see
text). The abundance of disorder-form A relative to B is roughly 1:1,
rotation angle between A and B is approximately 45°, and preliminary
Cr–O distances range from 1.88 to 1.96 Å. The octahedral arrange-
ments in A (black lines) and B (grey, narrower lines) are emphasized
on the right.
The limited insight on polymorphs 1, 2, and 4 gained from
the synthesis work will be carried on to following sections
where also the perspective of polymorphs 2*, 3, and 5 are
considered to get a more general picture.
Z. Anorg. Allg. Chem. 2010, 2422–2432
© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.zaac.wiley-vch.de
2425

A. Kjekshus et al.
ARTICLE
Table 2. Range and average bond lengths (dmin to dmax and dav in Å) and
bond angles (φmin. to φmax. and φav. in °) together with individual bond
valences (vi = exp[(Di–di) / b]; Di and b from reference [42]) and bond
valence sums (V = ∑vi) for modifications 1 and 4 of Cr(thd)3. Intra- and
interligand Oket–Cr–Oket bond angles are differentiated and data for the later
category appear in square brackets. Bonds outside chelate ring not included.
ble to bring the structure-determination/refinement process sig-
nificantly beyond the rather trivial Cr–O stage. This result was
surprising in view of the close match between PXD data for
o-Cr(thd)3 and o-Co(thd)3.
Several nice-looking crystals of 2 were tested, but apart from
providing unit-cell dimensions, the SXD data for these proved
to be unsuitable for structure determination. Similar observa-
tions were made for 2* (which differs from 2 by post-anneal-
ing at 130–160 °C). The crystal imperfections, which render
the structure determination/refinement difficult, appear to dif-
fer for 2 and 2*. The crystals of 2 are less burdened with
rotational disorder of the Cr(thd)3 molecules than 2*, but in-
stead 2 undergoes structural rearrangements as a function of
time (while retaining the external crystal shape).
1 (295 K)
Cr–Oket
Oket–Cr–Oket
Range: 1.932(6)–1.938(8)
Average: 1.936
Range: 90.6(3)–91.3(5) [86.5(3)–92.6(4)]
Average: 90.8 [89.7]
vCr–O = 0.56; VCr = 6·vCr–O = 3.36
Oket–Cket
Cr–Oket–Cket
Range: 126.9(8)–128.5(10)
Average: 127.2
Range: 1.189(14)–1.262(11)
Average: 1.244
vO–C = 1.48; VO = vO–Cr + vO–C = 2.04
Cket–Cket
Oket–Cket–Cket
Range: 1.393(15)–1.440(18)
Average: 1.413
Range: 122.4(16)–124.0(12)
Average: 123.1
Cket–Cket–Cket
Range: 120.9(15)–126.0(11)
Average: 123.3
Somewhat surprisingly the SXD efforts unambiguously es-
tablished that there also exists a tetragonal (shorthand designa-
tions 5 and t) modification of Cr(thd)3. 5 comes upon the scene
as a short-lived forerunner for 2 in sublimation-treated samples
(s.t. = 180 °C, d.t. = 90 °C). 5 is accordingly the first stage in
a time-ruled conversion process 5 → 2 → 4 and does not occur
as a symmetry-motivated (t between o and c) intermediate in
the temperature-determined succession 2* → 3. Time and ef-
forts were wasted in our search for 5 because we overlooked
the clear message that the axial a/b ratio of 2 approaches one
with decreasing temperature (e.g., a/b = 1.07, 1.02, and 1.01
at 410, 295, and 110 K, respectively). Besides the usual sour-
ces of error in determinations of unit-cell dimensions by SXD,
evaluations for 2 and 5 pick up additional uncertainties through
the temperature- and time-variations of the lattices. For 5 in
particular, as the first rather short-lived member of the genesis
series 5 → 2 → 4, one must work relatively fast and cool to
increase the lifetime to observe its existence. The structural
evolution from 5 via 2 to 1 necessarily has a significant impact
on the crystal quality of 5 and 2. This is probably the main
reason for the obstacles we have experienced in attempts to
resolve the crystal structure of 2 from SXD data. Representa-
tive unit-cell dimensions at 110 K are proposed as a = b =
18.93(5) and c = 10.57(4) Å for 5 and a = 18.97(7), b =
18.69(7) and c = 10.60(5) Å for 2. These may vary depending
on the synthesis process used. [5 was occasionally noticed, but
not measured at 295 K; a = 19.43(5), b = 19.08(5), and c =
10.79(3) Å for 2 at 295 K.] There is reasonable consistency
between unit-cell dimensions derived by SXD and PXD.
Apart from possible implications of the disorder (e.g., altera-
tion of the space group symmetry), the assignment of 2 as
isotypic with o-Co(thd)3 is appropriate. This relation (viz. the
implicit close match between corresponding atomic co-ordi-
nates) can be utilized to derive fairly accurate estimates for
interatomic distances and angles in the inner parts of these
molecular complexes. Like o-Co(thd)3 [6], 2 carries disorder,
but the disorder in these complexes appears to be limited to
the outer tert-butyl groups. The disorder in 2* appears to be
of a different character and concerns essentially the entire com-
plex. In fact, coupled orientational disorder and large aniso-
tropic thermal motions are precisely the ingredients which fa-
vor formation of spherical shaped molecules. This, in turn,
4 (295 K)
Domain of Cr1
Cr1–Oket
Range: 1.944(3)–1.954(3)
Average: 1.949
Oket–Cr1–Oket
Range: 89.71(14)–89.91(14) [89.93(14)–90.29(14)]
Average: 89.8 [90.1]
vCr1–O = 0.54; VCr1 = 6·vCr1–O = 3.24
Oket–Cket
Cr1–Oket–Cket
Range: 128.1(3)–128.5(3)
Average: 128.4
Range: 1.265(15)–1.270(5)
Average: 1.267
vO–C = 1.39; VO = vO–Cr1+ vO–C = 1.93
Cket–Cket
Oket–Cket–Cket
Range: 1.374(7)–1.385(5)
Average: 1.381
Range: 123.3(5)–124.3(5)
Average:123.9
Cket–Cket–Cket
Range: 125.1(5)–126.4(7)
Average: 125.9
Domain of Cr2
Cr2–Oket
Range: 1.942(4)–1.959(4)
Average: 1.947
Oket–Cr2–Oket
Range: 89.70(15)–90.00(16) [89.13(15)–91.04(15)]
Average: 89.8 [90.1]
vCr2–O = 0.55; VCr2 = 6·vCr2–O = 3.30
Oket–Cket
Cr2–Oket–Cket
Range: 1.253(5)–1.281(6)
Average: 1.267
Range: 127.5(3)–128.8(3)
Average: 128.4
Cket–Cket
Oket–Cket–Cket
Range: 1.375(7)–1.397(6)
Average: 1.386
Range: 123.0(5)–124.4(5)
Average: 123.6
Cket–Cket–Cket
Range: 124.6(5)–127.0(5)
Average: 125.7
4 (110 K)
Domain of Cr1
Cr1–Oket
Oket–Cr1–Oket
Range: 89.82(6)–90.10(8) [89.75(6)–90.41(6)]
Average: 89.9 [90.0]
Range: 1.9453(14)–1.9487(14)
Average: 1.947
vCr1–O = 0.55, VCr1 = 6·vCr1–O = 3.30
Oket–Cket
Cr1–Oket–Cket
Range: 128.50(13)–128.74(13)
Average: 128.6
Range: 1.267(2)–1.275(2)
Average: 1.271
vO–C = 1.38; VO = vO–Cr1 + vO–C = 1.93
Cket–Cket
Oket–Cket–Cket
Range: 1.395(3)–1.399(3)
Average: 1.397
Range: 124.2(2)–124.6(2)
Average: 124.4
Cket–Cket–Cket
Range: 124.0(2)–124.4(2)
Average: 124.2
Domain of Cr2
Cr2–Oket
Range: 1.941(2)–1.952(2)
Average: = 1.947
Oket–Cr2–Oket
Range: 89.64(6)–89.98(6) [88.48(7)–91.50(6)]
Average: 89.8 [90.0]
vCr2–O = 0.55; VCr2 = 6·vCr2–O = 3.30
Oket–Cket
Cr2–Oket–Cket
Range: 126.66(14)–129.10(14)
Average: 128.0
Range: 1.255(2)–1.281(3)
Average: 1.271
vO–C = 1.38; VO = vO–Cr2 + vO–C = 1.93
Cket–Cket
Oket–Cket–Cket
Range: 1.385(3)–1.399(3)
Average: 1.394
Range: 122.2(3)–124.5(2)
Average: 123.7
Cket–Cket–Cket
Range: 124.0(2)–125.3(2)
Average:124.5
2426
www.zaac.wiley-vch.de
© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Z. Anorg. Allg. Chem. 2010, 2422–2432

Polymorphs of Cr(thd)3
strongly influences the intra- and intermolecular interactions and Cket–Cket bonds can retain lengths characteristic for their
[22], which govern melting, sublimation, decomposition etc.
bonding situation in the chelate ring (for illustration see fig-
All structure information on 3 [c-Cr(thd)3] was obtained ure 8 of reference [43]).
through the use of high-temperature PXD within its narrow
Valence-state accounting according to the bond-valence
stability region (200–235 °C). Preparation and examination of scheme [42] shows (Table 2) an acceptable balance. The bond-
single crystals of 3 would necessarily be subject to similar valence sums (V) for chromium came out consistently some
temperature restrictions. Even if single crystals of 3, against 10 % too high (3.24–3.36), whereas the corresponding sums
considerable odds, should become available their utilization for for Oket only deviated from two by about 3 % (1.93–2.05). The
structure determination is likely to be hampered by extensive valence states for Cket cannot be assessed in the same way, but
disorder which is evidently more complex and pronounced in considerate estimates on the basis of the empirical bond length
3 than 2*.
A juxtaposition of the interatomic distances in 1 and 4 (Ta- effective valences of Cket and Calk are close to four.
ble 2) shows a few smaller irregularities, which indicate that The use of structural data for Cr(acac)3 and Cr(thd)3 from
versus bond-order relation for carbon [44, 45] show that the
the experimental data are impeded with certain shortcomings. gas-phase electron diffraction (ED) [46] and density-func-
The possibility of collecting new SXD data was carefully con- tional-theory (DFT) calculations [46–48] gives a bond valence
sidered, but was rejected in view of the very few atomic co- of nearly three for chromium (VCr = 3.03), whereas the devia-
ordinates that appear to be affected by such defects. To gain tions for Oket increase somewhat. These changes reflect the
insight into how such errors can arise, attention was focused 0.01–0.02 Å elongation of the bonds in free molecules com-
on the electron-density distribution. For example, the co-ordi- pared to the solid state (viz. a measure of the perturbations
nates of atom C4 in the crystal structure of 1 appear to be caused by the intermolecular van der Waals type interactions).
incorrect since the O–Cket and Cket–Cket bonds, which involve However, a further meditation on various aspects of bond-va-
this atom are, respectively, shorter and longer than the corre- lences falls outside the scope of this paper.
sponding bonds in other chromium acac and thd complexes.
The following knowledge is gained on the structure and
Gradients in the charge distribution at and near the C4 site bonding in molecular Cr(thd)3. (1) The valence states of the
indicate how the misplacement of this atom has come about. individual constituents are CrIII, OII, and CIV. (2) The CrO6
For utilization of the present structural data for bonding con- configuration in the central part is close to regular octahedral
siderations a couple of obviously incorrect interatomic distan- and the Cr–Oket bond length [individual values within the
ces and angles may simply be neglected.
range 1.932(6)–1.967(6) Å] is adequately accounted for by
bond-valence accounting. (3) The very similar and nearly 90°
intra- and interligand Oket–Cr–Oket angles imply that there can-
not be a significant bonding interaction across the ca. 2.75 Å
Oket···Oket bite and non-bite separations. (4) The atomic ar-
3.4 Crystal and Molecular Structures
Unit-cell dimensions and space groups are included in Ta-
ble 1. As for m-Co(thd)3 [6], the crystal structure of 4 com-
prises both the left (Λ) and right (Δ) helical forms of the chiral
isomers of Cr(thd)3. A complete listing of the refined posi-
tional and thermal parameters for the chromium, oxygen, and
carbon atoms in 1 and 4 are given in the electronic deposit.
rangement (viz. interatomic distances and angles) in the Oket
–
C
ket–Cket–Cket–Oket part of the chelate ring is virtually the same
in Cr(acac)3 [47, 49, 50] and Cr(thd)3. (5) The adjusting influ-
ence of the tert-butyl side groups on the shape of the chelate
ring to which it belongs is quite marginal, whereas the influ-
ence on packing modes in the solid state is more decisive. (6)
Points (1)–(5) can be generalized to cover the entire M(acac)2,
M(acac)3, M(thd)2, and M(thd)3 series.
The supplementary material also contains the Cr–Oket, Oket
–
C
ket, and Cket–Cket bond lengths and corresponding bond angles
specified according to the numbering of the atoms in Figure 2.
Table 2 gives a surveying extract of these data, in which the
various categories of bonds are represented by the range over
which the individual distances and the averages are distributed.
The molecular structure of Cr(thd)3 is nearly identical in 1
and 4 and also essentially unchanged between 110 and 295 K.
Octahedral CrO6 co-ordination rules in the central part of the
complex, whereas the diketonato chelate ring with its side
groups dominates in the outskirts. However, perfectly regular
octahedral arrangement for the Cr–O bonds appears to be in-
compatible with bonding demands of the chelate ring, the mu-
tual competition resulting in a compromise through which both
parts relinquish a little.
Least-squares fitted planes through the whole or parts of the
chelate ring can unveil deviation from planarity as atoms lo-
cated significantly above or below the mean plane. Already
the representation of the molecular structures of 1 and 4 in
Figure 2 suggests that some of the chelate rings of the Cr(thd)3
complexes are non-planar. The deviation from planarity varies
with the particular complex and chelate ring as well as with
temperature, but the actual magnitude of the departures is
rather small. For one of the non-equivalent rings of complex
1 (295 K), the maximum deviation from the mean plane is
0.12(2) Å, whereas the departure in the non-equivalent ring
only amounts to 0.03(2) Å. We are unable to see a pattern in
the available information on non-planarity of chelate rings in
acac and thd complexes. Therefore, we hesitate to speculate
on possible impacts on the π-conjugated electrons of the rings.
The molecular structures of Cr(thd)3 in 1 and 4 are very
It is well established that bond angles, in general, are more
flexible than bond lengths toward perturbations. Applied to the
molecular structures of the Cr(thd)3 polymorphs this guidance
anticipates that the Cr–Oket bonds are fairly free to adopt
lengths adjusted to the size of CrIII and OII, while the Oket–Cket similar at and below room temp. and the disorder within their
Z. Anorg. Allg. Chem. 2010, 2422–2432
© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.zaac.wiley-vch.de
2427

A. Kjekshus et al.
ARTICLE
Figure 2. (a) X-ray molecular structures of Cr(thd)3 in 1 (295 K), 4 (295 K), and 4 (110 K). Illustrations of 4 at both 110 and 295 K are included
to visualize increased thermal motion. Two identical units of 1 are shown to display the absence of chirality compared to the two independent
units of 4. Note that the separation between the units for 1 and 4 does not represent the real separation in the crystal structure. Hydrogen atoms
are systematically omitted. Thermal displacement ellipsoids are drawn at 50 % probability level. See references [41, 6] for illustrations of the
isotypic complexes Mn(thd)3 and m-Co(thd)3 of 1 and 4, respectively. (b) Atomic labeling schemes (identical for 4 at 110 and 295 K) for
chromium, oxygen, and carbon atoms in the asymmetric units of 1 and 4.
ligands (which certainly plays a major role in 2, 2*, 3, and 5) slightly different starting points and initial routes (see section
is small. Nevertheless the packing of the molecular units in the 3.4) which Cr(thd)3 can take to gradually approach spherical
crystal structures of 1 and 4 differs. On the basis of our find- shape with disorder as the main means. The progressive re-
ings it is tempting to suggest that the distinctions between moval of restrictions on symmetry is instrumental in this proc-
these packing arrangements along the a axes directions reflect ess. It is indeed sad that it has hitherto proved impossible to
2428
www.zaac.wiley-vch.de
© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Z. Anorg. Allg. Chem. 2010, 2422–2432

Polymorphs of Cr(thd)3
obtain single crystals suitable for full SXD exploration of the
crystal and molecular structures of 2, 2*, 3, and 5. Detailed
insight into the structural features of these polytypes is proba-
bly the key to a better understanding of properties of rotational
disordered crystals.
3.5 Relation Between the Polymorphs
The present examination of the phase relationship between
the polymorphs above room temp. was performed with TG,
DSC, and HT-PXD (used in scanning and fixed temperature
modes as well as room temp. PXD of specimens quenched
from selected annealing temperatures). The dynamic nature of
scanning techniques makes such methods sensitive to instru-
mental design and in particular to the experimental settings for
heating/cooling rate and atmosphere. Careful calibrations are
therefore needed to obtain reasonable consistency between
transition temperatures derived by different scanning tech-
niques, which inevitably only provide approximations to the
equilibrium transition temperatures. Considerable experience
may be needed to make the most correct choice from a set of
scattered values from different sources.
The onset s.t. ≈ 140 °C according to TG is higher than the
lowest s.t. ≈ 80 °C observed for incipient evaporation with the
sublimation apparatus (s.t. = 90 °C). Moreover, the TG find-
ings ensure that Cr(thd)3 can be subjected to sublimation with-
out decomposition below a certain temperature (guaranteed
300 °C under the present conditions).
The DSC curves for modifications 1, 2, and 4 (Figure 3)
show that Cr(thd)3 undergoes several phase transition with
temperature. Essentially identical DSC characteristics were ob-
tained above ca. 155 °C regardless which of these [stable (1)
or long-time metastable (2 and 4)] polymorphs provided the
test specimen is phase pure (especially with regard to
[Cr(thd)2(OEt)]2 [37]). The present DSC curves contain one
feature more than the DTA curve for Cr(thd)3 reported by Yo-
Figure 3. Differential scanning calorimetric (DSC) data for modifica-
tions (a) 1, (b) 2, and (c) 4 of Cr(thd)3 at 5 °C·min–1 heating rate. (d)
DSC cooling curve (5 °C·min–1) for Cr(thd)3 from melt. Temperatures
(in °C; averages of 3–5 runs) for the start and end of phase changes
are indicated (below the peaks) together with the amount of energy
involved (in J·g–1; average of 3–5 runs; printed in slanted numbers).
shida et al. [9], who only recorded an un-assigned peak at ca. phase changes in Cr(thd)3 as the DSC scans. (Note that the
200 °C in addition to the signature of melting at 236 °C. temperature intervals cited on the illustrations in Figure 4 refer
HT-PXD scans of the same polymorphs (Figure 4) confirm to stability regions assigned to the specified polymorphs as
that each of these polymorphs is subject to three phase changes opposed to transition regions stated on Figure 3.) However, the
above room temp. After indexing of the HT-PXD data (Fig- ability of the present HT-XPD setup to detect transient phases
ure 4), these transitions could be identified as 1 → 2*, 2 → and pinpoint phase transition temperature is poorer than that
2*, and 4 → 2*, for the first steps, followed by the common of the DSC equipment. Accordingly it is appropriate to attrib-
sequence of transitions 2* → 3 → melt. As seen from the start ute more weight to transition temperatures derived by DSC
and end temperatures for the later transitions (recorded below than assessments based on inspection of HT-PXD scans. How-
the peaks in Figure 3) the DSC data exhibit excellent repro- ever, when supplementary information from fixed-temperature
ducibility. [The transition enthalpies (which were included be- HT-PXD and room temp. PXD of quenched specimens are in-
cause they were available) also show good reproducibility.] troduced, stipulation of transition temperatures becomes a mat-
The transitions labeled 1 → 2*, 2 → 2*, and 4 → 2* are ter of discernment. Figure 5 summarizes, on a temperature
necessarily free to adopt unequal start temperatures since the scale, all available information on relations between the poly-
initial phases are different. Neither are the end temperatures morphs of Cr(thd)3. Efforts were made to provide the best pos-
obliged to be equal since the advancing transformation to the sible estimates for transition temperatures. Polymorph 5 is not
strongly disordered 2* state may take different routes. The incorporated in Figure 5 because of its obviously special for-
broad, smeared look of the peaks appears to support such a mation and evolution progression. The absence of the reverse
point of view.
of the 2 → 2* in the DSC and HT-PXD cooling curves pro-
HT-PXD scans of polymorphs 1, 2, and 4 (Figure 4) covey vides circumstantial evidence for the claimed substantial en-
essentially the same general view of the temperature induced hancement of disorder at the 2 → 2* transition.
Z. Anorg. Allg. Chem. 2010, 2422–2432
© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.zaac.wiley-vch.de
2429

A. Kjekshus et al.
ARTICLE
Figure 4. HT-PXD scans for modifications 1, 2, and 4 of Cr(thd)3. The
combined effect of heating rate, diffractometer speed, and scan range
amounts to a temperature difference of 12 min between two successive
scans. To ease inspection of the illustrations, scans which contain re-
flections from two adjacent modifications (regarding 2* see text) are
omitted, and the thus created interruptions are artificially expanded.
Temperatures (in °C) for the extension of the phase regions according
to HT-PXD are marked.
Figure 5. Summary of relations between the different modifications of
Cr(thd)3 as extracted from Figure 3 and Figure 4, as well as supple-
mentary results collected under fixed temperature conditions or derived
from quenched samples. Transition temperatures (in °C) are marked.
4. Conclusions
On combining the evidences presented in Figures 3–
3<figr4><figr5><figr6><figr7> we arrive at the conclusion
that the 2 → 2* transition belongs together with 1 → 2* and
4 → 2* in the first-order category, whereas the 2* → 3 and 3
→ melt transitions are of second order. It may be mentioned
that it is the progressively increasing rotational disorder which
facilitate the second order character of the latter transitions.
The stability of the various polymorphs at room temp. com-
plies with the evolution pattern monoclinic–orthorhombic–cu-
bic–melt on increasing temperature (Figure 5). These findings
are in accordance with the conceptions for rotationally disor-
dered molecules in crystals. The increase in crystal symmetry
is a means to facilitate increase in rotational disorder within a
molecular complex like Cr(thd)3 and the disorder together with
the thermal motions within the molecular units serve to accu-
mulate energy, which ultimately leads to sublimation, melting
or thermal decomposition. The conversion temperature gives a
measure of disorder a given structure can tolerate.
Approximate representations of the thermal expansion of the
lattices of the different polymorphs of Cr(thd)3 (Figure 6) have
been extracted from the HT-PXD scans. The juxtapositioned
curves for the temperature variation of the unit-cell volumes
per formula unit [V/Z, Figure 6(d)] is of particular interest in
the present context. The succession 1, 4, and 2 of the volume
expansion curves supports the inference that 1 is stable relative
to 4, which in turn is more stable than 2. The discontinuities
in the volume expansion, which is evident at the 1 → 2* and
4 → 2* transitions comply with first-order character nature of
these transitions. The steep increase in volume above ca.
150 °C is compatible with the rapid increase in disorder and
lattice defects in general anticipated as the melting temperature
is approached.
There are clear relationships between the unit cells of the
different polymorphs (Table 1 and Figure 7). Expressed in
terms of polymorph 2, vectorial equations for the relations are:
a1 = –a2/3 + 2c2/3, b1 = b2, c1 = –a2 + c2, for 2 → 1; a3 = a2/
2 + b2/2, b3 = –a2/2 + b2/2, c3 = 5c2/4 (a3 = b3 = c3 imposed)
for 2 → 3; a4 = a2 + 2c2, b4 = b2, c4 = -a2 + c2 for 2 → 4;
and a1 = a4/3, b1 = b4, c1 = c4 for 4 → 1 (5 represents a special
configuration of 2, in which a2 = b2). These equations describe
the structural correspondence between the unit cells of the
Acknowledgement
The authors are grateful to cand. scient. Ole Bjørn Karlsen and Dr.
Hiroshi Okamoto for help with DSC and magnetic susceptibility meas-
polymorphs with a relative accuracy of better than some 5 %. urements, respectively.
2430 www.zaac.wiley-vch.de
© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Z. Anorg. Allg. Chem. 2010, 2422–2432

Polymorphs of Cr(thd)3
Figure 7. Relation between the unit cells of the different polymorphs
of Cr(thd)3; 1 or 4 versus 2 and 3 versus 2 (Vectorial equations for the
relations are given in the text. Note that the cell indicated for 3 is only
pseudo-cubic; oblique angle 89.1°).
References
[1] G. C. Levy, U. Edlund, J. G. Hexem, J. Magn. Reson. 1975, 19,
259.
[2] G. C. Levy, U. Edlund, C. E. Holloway, J. Magn. Reson. 1976,
24, 375.
[3] G. C. Levy, J. J. Dechter, J. Kowalewski, J. Am. Chem. Soc. 1978,
100, 2308.
[4] <lit4>J. W. Freeman, J. L. Buster, R. D. Knudsen, US Patent
5856257, 1999.
[5] G. Carta, M. Natali, G. Rossetto, P. Zanella, G. Salmaso, S. Rest-
ello, V. Rigato, S. Kaciulis, A. Mezzi, Chem. Vap. Dep. 2005, 11,
375.
[6] M. A. K. Ahmed, H. Fjellvåg, A. Kjekshus, P. D. C. Dietzel, Z.
Anorg. Allg. Chem. 2008, 634, 247.
[7] G. S. Hammond, D. C. Nonhebel, C.-H. S. Wu, Inorg. Chem.
1963, 2, 73.
[8] E. W. Berg, N. M. Herrera, Anal. Chim. Acta 1972, 60, 117.
[9] I. Yoshida, H. Kobayashi, K. Ueno, Bull. Chem. Soc. Jpn. 1974,
47, 2203.
[10] K. Ueno, H. Kobayashi, I. Yoshida, Mem. Fac. Eng. Kyushu.
Univ. 1978, 38, 83.
[11] F. Wilkinson, C. Tsiamis, J. Phys. Chem. 1981, 85, 4153.
[12] H. Dreeskamp, H. Kluge, I. Koch, Miner. Kohlechem. 1985, 324.
[13] H. Dreeskamp, H. Kluge, Fuel 1985, 64, 1696.
[14] D. Stille, J. R. Doyle, J. E. Finholt, G. E. Mcguire, Inorg. Synth.
1986, 24, 183.
[15] M. A. V. Ribeiro da Silva, M. L. C. C. H. Ferrao, J. Chem. Ther-
modyn. 1987, 19, 645.
[16] C. Tsiamis, C. C. Hadjikostas, S. Karageorgiou, G. Manoussakis,
Inorg. Chim. Acta 1988, 143, 17.
[17] T. Matsuura, H. Takenawa, T. Suzuki, R. Hirota, K. Sasaki, J.
Radioanal. Nucl. Chem. Lett. 1990, 146, 223.
[18] J. C. Machado, A. O. Porto, C. F. Carvalho, W. F. Magalhaes, A.
Marques-Netto, J. Phys. 1993, 3, 201.
[19] F. Magalhaes, A. O. Porto, J. C. Machado, G. Duplâtre, Mater.
Sci. Forum 1995, 175–178, 719.
Figure 6. Temperature variation (slightly smoothed) of relative unit-
cell dimensions for modifications 1, 2 and 2*, and 4 of Cr(thd)3. The
monoclinic angle β for 1 and 4 contracts slightly on increasing temper-
ature (95.2 and 96.9° at 110 °C, respectively). (d) Juxtapositioned mo-
lar unit-cell volume versus temperature for Cr(thd)3. The volume ex-
pansion above ca. 200 °C should be regarded as tentative
(experimental data of relatively low precision in relation to rapid expo-
nential-like changes).
[20] A. O. Porto, W. F. Magalhaes, C. F. Carvalho, J. C. Machado, J.
Radioanal. Nucl. Chem. 1995, 198, 151.
[21] O. Nilsen, H. Fjellvåg, A. Kjekshus, Thermochim. Acta 2003,
404, 187.
[22] W. I. Dunning, in: The Plastically Crystalline State, (Ed.: J. N.
Sherwood) Wiley, Chichester, 1979.
[23] L. Pauling, Phys. Rev. 1930, 36, 430.
[24] T. E. Stern, Proc. Roy. Soc. London 1931, A130, 551.
Z. Anorg. Allg. Chem. 2010, 2422–2432
© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.zaac.wiley-vch.de
2431

A. Kjekshus et al.
ARTICLE
[25] C. Finback, O. Hassel, Z. Phys. Chem. 1936, B32, 130; C. Fin- [38] H. R. Fischer, D. J. Hodgson, E. Pedersen, Inorg. Chem. 1984,
back, O. Hassel, Z. Phys. Chem. 1936, B32, 433; C. Finback, O.
Hassel, Z. Phys. Chem. 1937, B35, 25; C. Finback, O. Hassel, Z.
Phys. Chem. 1937, B36, 25; C. Finback, O. Hassel, Z. Phys.
Chem. 1937, B37, 75; C. Finback, O. Hassel, Z. Phys. Chem.
1937, B37, 301.
23, 4755.
[39] P. E. Selwood, Magnetochemistry, Interscience, New York, 1956.
[40] U. Müller, Inorganic Structural Chemistry, Wiley, Chichester,
2007, p. 10.
[41] P. Magnus, A. H. Payne, M. J. Waring, D. A. Scott, V. Lynch,
Tetrahedron Lett. 2000, 41, 9725.
[26] O. Hassel, A. M. Sommerfeldt, Z. Phys. Chem. 1938, B40, 391.
[27] C. Finbak, Arch. Math. Naturvidenskab. 1938, B42, No. 1; C.
Finbak, Tidssk. Kjemi Bergvesen 1937, 145; C. Finbak, Tidssk.
Kjemi Bergvesen 1938, 101.
[42] N. E. Brese, M. O’Keeffe, Acta Crystallogr., Sect. B 1991, 47,
192.
[43] B. A. Kolesov, I. K. Igumenov, Spectrochim. Acta 1984, 40A,
233.
[28] A. Boultif, D. J. Louer, J. Appl. Crystallogr. 1991, 24, 987.
[29] P.-E. Werner, Program TREOR, Version 5, University of Stock-
holm, Stockholm, 1988; P.-E. Werner, L. Eriksson, M. Westdahl,
J. Appl. Crystallogr. 1985, 18, 367.
[44] L. Pauling, The Nature of the Chemical Bond, 3rd ed., Cornell
University Press, Ithaca 1960.
[45] C. A. Coulson, Valence, Oxford University Press, Oxford, 1953.
[46] S. Samdal, Private communication.
[30] WIN-INDEX Professional Powder Indexing, Version 3.0, Siemens
AG Analytical,System AUT, Karlsruhe, 1996.
[31] P. Norby, J. Am. Chem. Soc. 1997, 119, 5215.
[32] Area-Detector Integration Software, V7.12A; Bruker-Nonius, WI,
2004.
[47] I. Diaz-Acosta, J. Baker, W. Cordes, P. Pulay, J. Phys. Chem.
2001, A105, 238.
[48] I. Diaz-Acosta, J. Baker, J. F. Hinton, P. Pulay, Spectrochim. Acta
2003, 59A, 363.
[33] Bruker-Nonius, Madison, WI, 2004.
[34] M. C. Burla, R. Caliandro, M. Camalli, B. Carrozzini, G. L. Cas-
carano, L. D. Caro, C. Giacovazzo, G. Polidori, R. Spagna, J.
Appl. Crystallogr. 2005, 38, 381.
[49] E. A. Shugam, L. M. Shkol'nikova, Dokl. Akad. Nauk SSSR 1960,
133, 386.
[50] B. Morosin, Acta Crystallogr. 1965, 19, 131.
[35] G. Sheldrick, Acta Crystallogr., Sect. A 2008, 64, 112.
[36] L. J. Farrugia, J. Appl. Crystallogr. 1999, 32, 837.
[37] M. A. K. Ahmed, H. Fjellvåg, A. Kjekshus, N. Sen Gupta, D. S.
Wragg, Z. Anorg. Allg. Chem. 2010, 636, accepted.
Received: April 7, 2010
Published Online: August 16, 2010
2432
www.zaac.wiley-vch.de
© 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Z. Anorg. Allg. Chem. 2010, 2422–2432