Time-of-flight neutron powder diffraction study on the third row transition metal hexafluorides WF6, OsF6, and PtF6

Article abstract of DOI:10.1063/1.471473

A neutron diffraction study on the third-row transition metal hexafluorides MF₆ (M≡W, Os, Pt) has been performed using the high resolution neutron powder diffractometer (HRPD) at the spallation source ISIS, England. The previously unknown structures of the low-temperature phases of OsF₆ and PtF₆ are reported. WF₆, OsF₆, and PtF₆, which exhibit a (5dt_2g)⁰, (5dt_2g)², and (5dt_2g)⁴ electronic configuration, respectively, are found to be isostructural and crystallize in the UF₆ structure, space group Pmnb, (No. 62). The geometry of the MF₆ molecules is to good approximation octahedral for each compound, the mean M-F bond length increasing only slightly from 182.5 (W) to 185.0 (Pt). For WF₆ deviations from ideal octahedral geometry are only marginally significant [181.8(2) to 183.2(2) pm] and may be interpreted on the basis of packing effects. Deviations for the d₂ complex OsF⁶ are somewhat larger [181.5(2) to 184.4(3) pm] and may be assumed to be caused by packing effects essentially the same as for WF₆, in addition to a first-order Jahn-Teller effect arising from the (5dt_2g)² electronic configuration. While eliminating the effects of packing by a comparison of individual M-F bond lengths for WF₆ and OsF₆, the OsF₆ molecule shows to have D_4h symmetry with two apical M-F bonds about 1.8 pm longer than the four equatorial bonds as a result of the Jahn-Teller distortion. Only small deviations from ideal octahedral geometry [184.4(3) to 185.8(3) pm] are found for the d⁴ complex PtF₆. Within the series W to Pt a substantial shortening of the F...F van der Waals contact distances is observed. This shortening more than compensates for the increase in the M-F bond lengths and leads to unit cell volumes and cell parameters decreasing continuously from W to Pt. The variation of F...F contact distances and M-F bond lengths may be rationalized in terms of polarization of the F-ligands in the field of the highly charged nuclei of the central atoms which are only incompletely shielded by the 5 d electrons.

Full text of DOI:10.1063/1.471473

Timeofflight neutron powder diffraction study on the third row transition metal
hexafluorides WF6, OsF6, and PtF6
R. Marx, K. Seppelt, and R. M. Ibberson
Citation: The Journal of Chemical Physics 104, 7658 (1996); doi: 10.1063/1.471473
View online: http://dx.doi.org/10.1063/1.471473
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/104/19?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Site occupancy and lattice changes on nitrogenation in Nd3Fe29−x Ti x N y
J. Appl. Phys. 80, 2955 (1996); 10.1063/1.363152
Coupled HartreeFock electric polarizabilities of secondrow fluorides
AIP Conf. Proc. 330, 52 (1995); 10.1063/1.47845
A netron scattering study of premartensitic phases of TiNi (Fe)
AIP Conf. Proc. 53, 223 (1979); 10.1063/1.31867
A neutronscattering study of the modulation phase transitions of Rb2ZnBr4
AIP Conf. Proc. 53, 207 (1979); 10.1063/1.31861
Structure determination of plagioclase (Labradorite): I. Al/Si distribution in an An66 (neutron diffraction): II.
Superstructure of an An54 (Xray diffraction)
AIP Conf. Proc. 53, 311 (1979); 10.1063/1.31844
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
216.165.95.72 On: Sat, 22 Nov 2014 23:56:04

Time-of-flight neutron powder diffraction study on the third row transition
metal hexafluorides WF6, OsF6, and PtF6
R. Marx and K. Seppelt
¨
¨
Freie Universitat Berlin, Institut fur Anorganische und Analytische Chemie, Fabeckstr. 34/36, D-14195
Berlin, Germany
R. M. Ibberson
ISIS Science Division, Rutherford Appleton Laboratory, Chilton, Didcot, Oxon; OX11, 0QX, England
͑Received 11 October 1995; accepted 12 February 1996͒
A neutron diffraction study on the third-row transition metal hexafluorides MF6 ͑MϵW, Os, Pt͒ has
been performed using the high resolution neutron powder diffractometer ͑HRPD͒ at the spallation
source ISIS, England. The previously unknown structures of the low-temperature phases of OsF6
4
and PtF6 are reported. WF6, OsF6, and PtF6, which exhibit a (5dt2g)0, (5dt2g)2, and (5dt2g
)
electronic configuration, respectively, are found to be isostructural and crystallize in the UF6
structure, space group Pmnb, ͑No. 62͒. The geometry of the MF6 molecules is to good
approximation octahedral for each compound, the mean M–F bond length increasing only slightly
from 182.5 ͑W͒ to 185.0 ͑Pt͒. For WF6 deviations from ideal octahedral geometry are only
marginally significant ͓181.8͑2͒ to 183.2͑2͒ pm͔ and may be interpreted on the basis of packing
effects. Deviations for the d2 complex OsF6 are somewhat larger ͓181.5͑2͒ to 184.4͑3͒ pm͔ and may
be assumed to be caused by packing effects essentially the same as for WF6, in addition to a
first-order Jahn–Teller effect arising from the (5dt2g)2 electronic configuration. While eliminating
the effects of packing by a comparison of individual M–F bond lengths for WF6 and OsF6, the OsF6
molecule shows to have D4h symmetry with two apical M–F bonds about 1.8 pm longer than the
four equatorial bonds as a result of the Jahn–Teller distortion. Only small deviations from ideal
octahedral geometry ͓184.4͑3͒ to 185.8͑3͒ pm͔ are found for the d4 complex PtF6. Within the series
W to Pt a substantial shortening of the F•••F van der Waals contact distances is observed. This
shortening more than compensates for the increase in the M–F bond lengths and leads to unit cell
volumes and cell parameters decreasing continuously from W to Pt. The variation of F•••F contact
distances and M–F bond lengths may be rationalized in terms of polarization of the F-ligands in the
field of the highly charged nuclei of the central atoms which are only incompletely shielded by the
5d electrons. © 1996 American Institute of Physics. ͓S0021-9606͑96͒00119-1͔
I. INTRODUCTION
tion just below the melting point from a cubic, I-centered
plastic phase with disordered MF6 molecules to a primitive
orthorhombic phase with ordered MF6 units. Complete struc-
tural data are available only in case of WF6 for both
phases,6,7 however the variation in the lattice parameters as
well as the reflection intensities suggests an isotypic relation
for all MF6 compounds.8
The third-row transition metal hexafluorides are both
chemically highly reactive and volatile molecular com-
pounds. Within the series the number of nonbonding
d-electrons increases from d0 for W to d4 for Pt. The physi-
cal properties in the vapor phase of these model compounds
are of considerable theoretical interest and have been the
subject of numerous investigations by different methods, in-
cluding vibrational,1,2 visible and ultraviolet spectroscopy,3,4
and gas phase electron diffraction.5 Irrespective of the num-
ber of d-electrons involved the MF6 molecules exhibit octa-
hedral geometry with M–F bond lengths of 183.3͑8͒,
183.1͑8͒, and 183.0͑8͒ pm for WF6, OsF6, and IrF6,
respectively.5 PtF6 was not measured but it was expected to
continue the trends in this series. Characteristic features in
the Ir spectra of ReF6 and OsF6 were interpreted in terms of
dynamic distortions due to a first-order Jahn–Teller effect.1
The colors of the transition metal hexafluorides ͑WF6 is col-
orless, OsF6 is bright yellow, and PtF6 is deep red͒, are ex-
plained by charge transfer bands, implying a narrowing of
the energy gap between the nonbonding F orbitals and the
partially filled nonbonding metal d-orbitals.3,4
The MF6 series therefore appears an ideal model system
for studying the small Jahn–Teller distortions arising from
the nonbonding (dt2g
)n electronic configuration as intermo-
lecular forces due to the packing in the solid state will be
small and comparable for different members of the series. A
detailed comparison may thus enable the distortions arising
from electronic effects alone to be determined ͑see Ref. 9
and references cited therein for other model systems, for ex-
ample the transition metal sulfate hexahydrates͒.
In the present investigation OsF6, PtF6, and for com-
parative purposes WF6 were chosen for detailed structural
study and determination of their molecular geometry. Since
the geometry is likely to be dependent on the number of
d-electrons attached to the metal atom, the highest possible
structural accuracy is required. There are a number of experi-
mental difficulties associated with x-ray single crystal stud-
In the solid state, the hexafluorides show a phase transi-
7658
J. Chem. Phys. 104 (19), 15 May 1996
0021-9606/96/104(19)/7658/7/$10.00
© 1996 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
216.165.95.72 On: Sat, 22 Nov 2014 23:56:04

Marx, Seppelt, and Ibberson: Neutron diffraction study of WF6 , OsF6 , and PtF6
7659
TABLE I. The R-values of the Rietveld refinements of WF6 , OsF6 , and
PtF6 in the space group Pmnb ͑No. 62͒.
ies, for example, twinning and unsuitability of usual capil-
lary materials due to the high reactivity of the compounds. In
such cases, high resolution neutron powder diffraction is an
ideal alternative.
Compound
Rp
Rwp
Re
RI
WF6
OsF6
PtF6
7.1
4.2
4.4
7.1
4.8
5.0
7.8
4.3
4.8
5.2
3.9
4.7
II. EXPERIMENT
A. Sample preparation
WF6 ͑Merck, zur Synthese͒ was used without further
purification.
OsF6 was prepared by heating Os powder ͑Alfa, M3N͒
and F2 ͑1:2͒ in a stainless steel autoclave at 300 °C for 12 h.
Excess fluorine and volatile impurities such as SiF4 or HF
were removed at Ϫ78 °C using a stainless steel vacuum line.
PtF6 was prepared by the original method of Weinstock
et al.10 A platinum wire ͑Heraeus, technisch rein͒, electri-
cally heated until ignition was burnt in a F2 atmosphere ͑1:2
F2 excess͒, the Monel reaction vessel was maintained at liq-
uid nitrogen temperature ensuring the PtF6 vapor condenses
on the cold reactor walls. Excess F2 was pumped off at
Ϫ196 °C, possible impurities such as SiF4 or HF were re-
moved at Ϫ78 °C.
tional data on the empty FEP can were collected for 12 h.
The raw diffraction data were corrected for wavelength de-
pendent factors such as neutron flux and detector efficiency.
This standard normalization procedure is described
elsewhere.15 Background arising from the FEP sample can
was accounted for by subtracting a background function. The
latter was determined from the empty can data using a two-
channel maximum entropy technique that enabled the reli-
able separation of broad, smoothly-varying background fea-
tures of the spectrum from sharp features, in this case
represented by vanadium Bragg peaks from the cryostat.16
Sample environment attenuation effects arising from the va-
nadium parts of the cryostat were also accounted for at this
stage. Remaining background and sample attenuation effects
were accounted for as part of the profile refinement proce-
dure.
For the neutron diffraction experiments the metal
hexafluorides once checked for impurities by Ir spectros-
copy, were condensed into FEP-tubes ͑perfluorinated
ethylen-propylene copolymerisate, л 9 mm, wall thickness
1
2
mm͒, again using a stainless steel vacuum line. In order to
avoid condensing large crystalline pieces with possible pre-
ferred orientation effects successive small amounts of MF6
were collected in the upper part of the tube and tapped to the
bottom. The FEP tubes were sealed under vacuum and stored
in liquid nitrogen.
C. Data analysis
The corrected data were analyzed by the Rietveld
method of profile refinement using the local program suite
adapted for TOF data.17 First the orthorhombic cell param-
eters of WF6, OsF6, and PtF6 established by neutron and
x-ray powder diffraction work at 77 K and about 250 K,
respectively7,8 were refined for the 5 K data. An intensity-
only ͑Pawley͒ refinement in space group Pmnb ͑No. 62͒ did
not show any peak fitting discrepancies confirming the origi-
nal indexing and supporting the isotypic relation suggested
by Siegel et al.8 for WF6, OsF6, PtF6 and other hexafluo-
rides. Secondly, the Rietveld refinement was carried out, the
atomic arrangement of WF6 at 77 K given by Lewy et al.7
serving as a starting model. The diffraction data for WF6,
OsF6, and PtF6 were refined to Ri values of 5.2%, 3.9%, and
4.7%, respectively, in space group Pmnb ͑No. 62͒ ͑Tables I,
II, and Fig. 1͒. Attempts to refine the data in a lower sym-
metry space group ͑P21nb, molecular symmetry c1͒ failed.
The final structural model involved isotropic displacement
parameters for the central atoms W, Os, and Pt and indepen-
dent anisotropic displacement parameters for the ligand at-
oms. In case of W and Pt slightly nonpositive definite ellip-
soids of displacement were obtained for some atoms
probably due to overparameterization. However, atomic co-
ordinates and molecular geometry obtained are found to be
largely independent on the kind of thermal displacement pa-
rameters applied.
B. Data collection
The neutron diffraction measurements on WF6, OsF6,
and PtF6 were carried out at the neutron spallation source
ISIS, Rutherford Appleton Laboratory ͑UK͒ using the High
Resolution Powder Diffractometer ͑HRPD͒.11 HRPD is a
time-of-flight ͑TOF͒ instrument and thus operates using a
pulsed polychromatic neutron beam, produced by the spalla-
tion process and moderated by water at ambient temperature.
The incident flux ranges from 50 to 450 pm and shows an
exponential fall with increasing wavelength. The scattered
neutrons are recorded by fixed-angle detectors with neutrons
of different wavelengths discriminated by their time of ar-
rival since tϳ1/ nϳ␭nϳd, where t is the neutron time-of-
v
flight,
is the neutron velocity, ␭ the neutron wavelength,
vn
n
and d is the spacing of the lattice planes. In contrast to con-
ventional constant wavelength diffractometers the high in-
strumental resolution, ⌬d/dϷ8•10Ϫ4, is effectively constant
over the whole spectrum and thus the powder patterns have a
high information content, particularly by virtue of the short
d-spacings. HRPD has been shown to be a competitive alter-
native to conventional single crystal studies, see for example
Refs. 12–14.
Data on WF6, OsF6, and PtF6 were collected at 5 K,
collection times were 7, 14, and 14 h, respectively. Addi-
J. Chem. Phys., Vol. 104, No. 19, 15 May 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
216.165.95.72 On: Sat, 22 Nov 2014 23:56:04

7660
Marx, Seppelt, and Ibberson: Neutron diffraction study of WF6 , OsF6 , and PtF6
TABLE II. Atomic coordinates and equivalent isotropic temperature factors
for WF6 , OsF6 , and PtF6 at 5 K refined in the spacegroup Pmnb ͑for W, Os,
and Pt isotropic temperature factors are given͒.
TABLE III. The effect of mean M–F bond lengths (M–F) and mean F•••F
intermolecular distances (F•••F) on lattice parameters and unit cell volumes
of WF6 , OsF6 , and PtF6 ͑distances in pm, volume in 106 pm3͒.
Phase Atom in
x/a
y/b
z/c
U ͑pm2͒
Compound Volume
a
b
c
M–F F•••F
WF6
OsF6
PtF6
MFa6
W
F1
F2
F3
F4
Os
F1
F2
F3
F4
Pt
F1
F2
F3
F4
M
4c
4c
4c
8d
8d
4c
4c
4c
8d
8d
4c
4c
4c
8d
8d
1/4
1/4
1/4
0.0990͑1͒
0.0990͑1͒
1/4
1/4
1/4
0.0980͑1͒
0.0990͑2͒
1/4
1/4
1/4
0.0959͑2͒
0.0951͑2͒
1/4
1/4
1/4
0.1283͑2͒
0.0126͑2͒
0.2429͑2͒
0.0166͑1͒
0.2386͑1͒
0.1278͑1͒
0.0094͑2͒
0.2458͑2͒
0.0168͑1͒
0.2387͑1͒
0.1269͑1͒
0.0061͑2͒
0.2456͑2͒
0.0138͑1͒
0.2393͑2͒
0.0852͑4͒
Ϫ0.2102͑3͒
0.3836͑3͒
0.2409͑2͒
Ϫ0.0676͑2͒
0.0872͑2͒
Ϫ0.2122͑4͒
0.3895͑6͒
0.2463͑2͒
Ϫ0.0685͑4͒
0.0934͑3͒
Ϫ0.2093͑4͒
0.3954͑6͒
0.2537͑2͒
Ϫ0.0662͑3͒
5͑3͒
41
47
46
43
9͑1͒
56
56
48
45
1͑2͒
43
51
33
46
WF6
OsF6
PtF6
MFa6
396.930͑4͒ 853.968͑5͒ 937.738͑6͒ 495.668͑3͒ 182.5 301.3
387.112͑3͒ 849.193͑5͒ 930.518͑4͒ 489.898͑3͒ 182.7 297.5
384.509͑4͒ 846.131͑6͒ 928.149͑5͒ 489.611͑3͒ 185.0 294.7
385
849.3
924.6
490.3
aIdeal orthorhombic lattice parameters for a unit cell volume of 385•106 pm3
assuming an ideal ABAC stacked hexagonal close packing of F atoms
͑aϭ3x, bϭ4/3ͱ6x, cϭͱ3x, xϭF–F distance͒.
the F•••F van der Waals contacts between neighboring MF6
molecules. This reduction more than compensates the in-
crease in bond lengths and leads to a decrease in the lattice
parameters and unit cell volumes. These results are summa-
rized in Table III and may be rationalized in terms of polar-
ization of the F ligands in the field of the highly charged
nuclei of the central atoms that are only shielded in part by
the 5d electrons. In the series W to Pt the F nuclei are sub-
ject to increasingly repulsive forces, whilst the F electron
clouds are subject to increasingly attractive forces leading to
an elongation of the M–F ͑nuclei–nuclei͒ distances and al-
lowing closer F•••F contacts. ͑From a more chemical point of
view one might try to explain the varying M–F bond lengths
by their bond strengths. However qualitative arguments us-
ing the MO theory do not allow to rationalize the results of
Table III in a simple manner. Within the series W to Pt on
one hand the M–F ␴-bond is expected to gain in strength due
to an increasing covalency of the bond while on the other
hand the M–F ␲-bond is expected to lose in strength due to
the successive filling of the metal t2g orbitals. So focusing on
the ␴-bonds one would predict decreasing M–F and F•••F
distances while focussing on the ␲-bonds one would predict
increasing M–F and F•••F distances.͒
1/8
1/12
Ϫ1/4
5/12
1/4
Ϫ1/12
F1
F2
F3
F4
0
1/4
0
1/12
1/12
1/4
aFor means of comparison atomic coordinates are given for an idealized
MF6 structure with a perfect ABAC stacked hexagonal close packing of F
atoms and M atoms occupying 1/6 of the octahedral holes.
III. RESULTS AND DISCUSSION
A. Volume considerations
WF6, OsF6, and PtF6 are molecular solids comprising
isolated MF6 molecules. Irrespective of the metal atom in-
volved these MF6 units approach almost perfect octahedral
geometry. The mean M–F bond lengths are similar, increas-
ing only slightly from W ͑182.5͒ to Os ͑182.7͒ and Pt ͑185.0
pm͒. ͓Compare M–F bond lengths of 183.3͑8͒ and 183.1͑8͒
pm for W and Os obtained by means of gas phase electron
diffraction.͔ In contrast there is a substantial shortening of
B. The packing of the MF6 molecules
Within the MF6 structure the packing of the MF6 octa-
hedra is defined by n- and b-glide planes perpendicular to
the orthorhombic b- and c-axis ͑Fig. 2͒. The fluorine atoms
form an ABAC-stacked hexagonal close packed array with
the hexagonal layers lying in yϷ0, 1/4, 1/2, 3/4, 1,... perpen-
dicular to the b-axis. The MF6 octahedra with metal atoms in
yϷ1/8, 3/8, 5/8, 7/8,... can be divided into two subsets, one
constituting the B-layer of F atoms in yϷ1/4 and one con-
stituting the C-layer in yϷ3/4. Within each subset, one half
of the octahedra ͑M atoms in xϭ1/4͒ point down the b-axis
and the other half ͑M atoms in xϭ3/4͒ point up ͑Fig. 2͒.
In Table IV the atomic coordinates determined by
Rietveld refinement are given for W, Os, and Pt together
with ideal coordinates calculated for hexagonal close pack-
ing of F atoms as described above. A comparison shows
small ͑р5 pm͒ but significant deviations; the z coordinate is
found to increase whereas the y coordinate decreases. These
shifts of the metal atoms ͑and their associated F6 shells͒ may
be explained by the action of van der Waals forces between
FIG. 1. The observed, calculated, and difference profiles of OsF6 at 5 K
obtained using HRPD.
J. Chem. Phys., Vol. 104, No. 19, 15 May 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
216.165.95.72 On: Sat, 22 Nov 2014 23:56:04

Marx, Seppelt, and Ibberson: Neutron diffraction study of WF6 , OsF6 , and PtF6
7661
FIG. 2. Schematic view of the structure of WF6 , OsF6 , and PtF6 along the
orthorhombic b-axis. The MF6 structure may be regarded as an ABAC
stacked hexagonal close packing of F atoms where the metal atoms occupy
1/6 of the octahedral holes. This idealized F arrangement with A-layers in
yϭ0, 1/2, B-layers in yϭ1/4 and C-layers in yϭ3/4 is given on the right-
hand side of the figure ͑metal atoms omitted͒. The arrangement of the MF6
units with metal atoms in 1/8, 3/8, 5/8, 7/8 is established by the glide planes
n and b as shown on the left-hand side of the figure. The MF6 octahedra are
all orientated with a face parallel to the a–c plane and are symbolized by
six lines pointing towards the F ligands. Of these only the atoms of the
A-layers are given, atoms of the B- or the C-layers are omitted.
FIG. 3. Variation of the F•••F van der Waals contact distances within an
ideal variant of the MF6 structure derived from an ABAC stacked close
packing of F atoms ͑F•••F distance: 283 pm͒ by contracting the F6 units
around the metal atoms in order to obtain suitable M–F bond lengths ͑185
pm͒. The structure section shown in a projection along b contains MF6
molecules with M in yϭϪ1/8 and yϭ1/8 and F atoms in yϷϪ0.25 ͑empty
circles͒, yϷ0 ͑shaded circles͒, and yϷ0.25 ͑full circles͒. Shifts and rotations
necessary to reduce the imbalance in the F•••F contacts are indicated by
arrows.
tween neighboring octahedra along the c-axis. In order to
compensate for this the MF6 units are shifted along ͓001͔ and
rotated about the a-axis to obtain more uniform van der
Waals contacts ͑Fig. 3͒. In a third step similar shifts along
Fatoms of neighboring MF6 molecules and M␦ϩ–M␦ϩ repul-
sive forces. The refined structures of WF6, OsF6, and PtF6
are compared with a structure derived from an ideal close
packing as follows. Starting from an ideal packing in a first
trivial step the F atoms around a central atom are adjusted so
as to form an ideal octahedron with suitable M–F bond
lengths ͑185 pm͒. As a consequence, intramolecular F•••F
contacts of about 260 pm and a number of intermolecular
F•••F contacts of around 300 pm are formed. Inspection of
Fig. 3 shows a large imbalance of these van der Waals con-
tacts. Compare for example, F3–F3 contacts of 285 and 305
pm between neighboring molecules along the a-axis and
F1–F2 contacts of 285 and F1–F3 contacts of 294 pm be-
TABLE IV. Fractional atomic coordinates for the metal atoms W, Os, and
Pt and their deviations ⌬y and ⌬z from the ideal coordinates xϭ1/4, yϭ1/8,
zϭ1/12 within a hexagonal close packing of F atoms. ⌬Y and ⌬z are dis-
cussed as a result of packing forces ͑see text͒.
Atom
x/a
y/b
z/c
⌬y
⌬z
FIG. 4. Schematic view of the arrangement of the metal atoms within the
W
Os
Pt
1/4
1/4
1/4
0.1283͑2͒
0.1278͑1͒
0.1269͑1͒
0.0852͑4͒
0.0872͑2͒
0.0934͑3͒
0.0033
0.0028
0.0019
0.0019
0.0039
0.0101
MF6 structure along the b-axis ͑M atoms in yϭϪ1/8, 1/8, 3/8, 5/8 repre-
sented by 1, 1, 3, 5͒. Shifts caused by M␦ϩ–M␦ϩ repulsion forces between
¯
M and pairs of noncollinear M atoms roughly along the a-axis ͑dotted lines͒
and along the b-axis ͑broken lines͒ are indicated by arrows.
J. Chem. Phys., Vol. 104, No. 19, 15 May 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
216.165.95.72 On: Sat, 22 Nov 2014 23:56:04

7662
Marx, Seppelt, and Ibberson: Neutron diffraction study of WF6 , OsF6 , and PtF6
FIG. 5. Postulated orbital correlation diagram for the distortion of a d4 MF6 molecule from Oh to C4v symmetry.
͓010͔ and ͓001͔ have to be applied to account for M␦ϩ–M␦ϩ
repulsive forces between a metal atom and pairs of noncol-
linear M atoms along the a- and the b-axis, respectively ͑Fig.
4͒. As this M␦ϩ–M␦ϩ repulsion depends on the effective
positive charge on the metal atom it may be assumed largest
for tungsten which is the most electropositive element in the
series W–Os–Pt. Thus the largest shift along the b-axis is
obtained for W while along the c-axis the largest shift ͑pre-
dominantly caused by van der Waals forces and reduced by
M–M repulsive forces͒ is obtained for Pt ͑Table IV͒.
are the roughly trigonal prismatic molecules W͑CH3͒6 or
Re͑CH3͒6.21 Replacing the less electronegative ␲-innocent
CH3 ligand by the strongly electronegative ␲-donor ligand F
increases the HOMO-LUMO gap so that SOJT arguments do
not apply for WF6.22 Charge transfer spectra however show
that the energy gap between the T1u,T2u ligand ␲ orbitals
and the T2g metal orbitals continuously decreases from WF6
͑58 350 cmϪ1͒ to PtF6 ͑23 500 cmϪ1͒.4 PtF6 may therefore be
a candidate for a SOJT distortion.23 In Fig. 5 is shown a
speculative distortion initiated by the T1u vibration mode
leading to a molecule with C4v symmetry with one longer
and one shorter apical M–F bond.
¯
C. Electronic aspects
WF6, OsF6, and PtF6 are examples of 5d0, 5d2, and 5d4
MF6 complexes. Of these compounds only OsF6 shows an
orbitally degenerate electronic ground state, Jϭ2, suitable
for a first order Jahn–Teller ͑FOJT͒ distortion.18 Anomalous
features of the vibrational spectrum ͑compared to the other
5dn MF6 members͒ such as frequency shift and broadening
of the eg band and combination bands involving eg are con-
sistent with a ͑dynamic͒ D4h distortion of the OsF6
octahedron.3,19 The d4 system PtF6 with an electronic ground
state Jϭ0 and the d0 system WF6 do not show degeneracy
allowing for a FOJT effect. ͕The electronic structures of the
5d transition metal hexafluorides have been evaluated by
Moffitt et al.3 and modified by Holloway et al.4 from Ir, vis-
ible and UV spectroscopic data analyzed by means of ligand
field theory. The (dt2g)n configuration ͑strong ligand field͒
was shown to be formally identical with that of the atomic
configuration p6Ϫn ͑pn formalism͒. For the latter Russell–
Saunders coupling was encountered leading to total orbital
and total spin angular momenta, L and S, with JϭLϩS. So
D. Distortions from ideal octahedral geometry for
WF6 , OsF6 , and PtF6
In Table V or Fig. 6 the individual geometries obtained
by Rietveld refinement are given for WF6, OsF6, and PtF6.
As can be seen bond lengths and angles deviate only slightly
from those of a perfect octahedron. In the simplest ͑and least
ambiguos͒ interpretation these deviations may be assumed to
be caused by systematic errors in the powder profile refine-
ment ͑particularly in case of PtF6, which shows the smallest
deviations͒. However these deviations can also be explained
by packing forces in addition to the electronic state of the
central atom.
As discussed in Sec. III B within the MF6 structure
M␦ϩ–M␦ϩ Coulomb forces and van der Waals forces act on
a MF6 molecule. The former affect only the central atoms,
the latter however affect the ligand atoms. These forces are
dependent on the effective positive charge on the metal atom
and must give rise to some level of distortion of the MF6
molecule. One can deduce that the metal atom is shifted
towards the F1F4 face of the octahedron ͑Figs. 3, 4͒. This is
exactly the geometry obtained for WF6 by refinement.
For OsF6, packing effects essentially the same as for
WF6 and electronic effects have to be considered. The latter
are most easily extracted by comparing the individual M–Fi
bond lengths obtained for WF6 and OsF6. It is found that the
four equatorial M–F distances ͑F3 and F4͒ are shortened by
3
for example for (dt2g)4ϵp2 the ground state is P with a
normal multiplet structure ͓3P0(Jϭ0) lying lowest͔;
(dt2g)2ϵp4 shows again a 3P ground state however with an
inverted multiplet structure ͓3P2(Jϭ2) lying lowest͔.͖
While the first order Jahn–Teller effect is connected to
degenerate electronic states, a second order Jahn–Teller
͑SOJT͒ effect may arise for molecules with HOMOs and
LUMOs lying close to each another.20 Prominent examples
J. Chem. Phys., Vol. 104, No. 19, 15 May 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
216.165.95.72 On: Sat, 22 Nov 2014 23:56:04

Marx, Seppelt, and Ibberson: Neutron diffraction study of WF6 , OsF6 , and PtF6
7663
TABLE V. Bond lengths and bond angles of the WF6 , OsF6 , and PtF6 molecules with the crystallographic
molecular symmetry Cs ͑bond lengths in pm, bond angles in °͒.
Compound
M–F1
M–F2
M–F3
M–F4
F1–M–F2
F3–M–F5
WF6
OsF6
PtF6
182.2͑2͒
183.5͑2͒
185.8͑3͒
182.8͑2͒
184.4͑3͒
184.4͑3͒
183.2͑2͒
182.8͑1͒
184.9͑2͒
181.8͑2͒
181.5͑2͒
184.8͑2͒
179.5͑1͒
179.7͑1͒
179.6͑2͒
179.5͑1͒
179.6͑1͒
179.7͑2͒
F1–M–F3
89.9͑1͒
90.1͑1͒
F1–M–F4
90.2͑1͒
90.3͑1͒
F2–M–F3
89.7͑1͒
89.7͑1͒
F2–M–F4
90.2͑1͒
89.9͑1͒
F3–M–F3
89.5͑1͒
88.9͑1͒
F4–M–F4
90.4͑1͒
90.0͑1͒
WF6
OsF6
PtF6
89.8͑1͒
90.2͑2͒
89.9͑1͒
90.1͑1͒
89.7͑1͒
90.3͑1͒
0.4 pm while the two apical distances ͑F1, F2͒ are elongated
by 1.4 pm. As stated above, such a ͑4ϩ2͒ coordination is
indeed expected for a (dt2g)2 electronic state metal atom.
For PtF6 the distortion of the molecular geometry by
packing forces tends to be less severe as polarization effects
may reduce the effective positive charge of the central atom
giving rise to the M␦ϩ–M␦ϩ repulsion. Observed Pt–F bond
lengths are therefore considered without any correction for
packing forces, keeping in mind however, that M–F1 and
M–F4 distances are found shortened for WF6 due to packing
effects. The PtF6 molecule then comprises four equatorial
bonds equal in length and two apical M–F bonds of which
the one is slightly longer than the other. This small distortion
away from octahedral geometry is consistent with a SOJT
effect resulting from a mixing of the T1u ligand orbital with
the T2g metal orbital.
IV. CONCLUSIONS
The transition metal hexafluorides represent an ideal
model system for studying the small Jahn–Teller distortions
arising from the nonbonding (dt2g)n electronic configuration
of the central atom. The structural trends observed in the
present study, although subtle, are consistent with theoretical
considerations and demonstrate that high resolution neutron
powder diffraction is a powerful tool, useful in the determi-
nation of small changes in the geometry of model molecular
compounds. It seems desirable to extend this type of detailed
study to the remaining third-row transition metals Re and Ir
as well as the second-row transition metals Ru and Rh.
ACKNOWLEDGMENTS
The authors would like to thank Professor Dr. O. Stein-
¨
born, Institut fur Physikalische und Theoretische Chemie,
¨
Universitat Regensburg, for helpful discussions.
1 B. Weinstock, H. H. Claassen, and J. G. Malm, J. Chem. Phys. 32, 181
͑1960͒.
2 B. Weinstock and G. L. Goodman, Adv. Chem. Phys. 9, 169 ͑1965͒.
3 W. Moffitt, G. L. Goodman, M. F. Weinstock, and B. Weinstock, Mol.
Phys. 2, 109 ͑1959͒.
4 J. H. Holloway, G. Stanger, E. G. Hope, W. Levason, and J. S. Odgen, J.
Chem. Soc. Dalton Trans. 1988, 1341.
5 M. Kimura, V. Schomaker, D. W. Smith, and B. Weinstock, J. Chem.
Phys. 48, 4001 ͑1968͒.
6 J. H. Levy, J. C. Taylor, and P. W. Wilson, J. Less Common Metals 45,
155 ͑1976͒.
7 J. H. Levy, J. C. Taylor, and A. B. Waugh, J. Fluorine Chem. 23, 29
͑1983͒.
8 St. Siegel and D. A. Northrop, Transition Metal Fluorides 5, 2187 ͑1966͒.
9 T. Kellersohn, R. G. Delaplane, J. Olovsson, and G. J. McIntyre, Acta
Crystallogr. B 49, 179 ͑1993͒.
10 B. Weinstock, H. H. Claassen, and J. G. Malm, J. Am. Chem. Soc. 79,
5832 ͑1957͒.
FIG. 6. The octahedral OsF6 molecule ͑molecular symmetry cs , ellipsoids
of 90% probability, distances in pm͒. The individual Os–F bond lengths
range from 181.5 to 184.4 pm and are caused by packing effects in addition
to a first-order Jahn–Teller effect arising from the 5d2 electronic state of Os.
Comparing with the corresponding W–F bond lengths ͑numbers in brackets͒
may give an idea of the magnitude and the kind of Jahn–Teller distortion as
for the 5d0 complex WF6 only packing effects have to be considered.
11 W. I. F. David, D. E. Akporiaye, R. M. Ibberson, and C. C. Wilson,
Rutherford Appleton Report No. RAL-88-103 ͑1988͒.
12 W. I. F. David and J. D. Jorgensen, The Rietveld Method in IUCr Mono-
graphs on Crystallography, edited by R. A. Young ͑Oxford University,
Oxford, 1993͒.
13 R. Marx, H. Dachs, and R. M. Ibberson, J. Chem. Phys. 93, 5972 ͑1990͒.
J. Chem. Phys., Vol. 104, No. 19, 15 May 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
216.165.95.72 On: Sat, 22 Nov 2014 23:56:04

7664
Marx, Seppelt, and Ibberson: Neutron diffraction study of WF6 , OsF6 , and PtF6
14 R. Marx, A. R. Mahjoub, K. Seppelt, and R. M. Ibberson, J. Chem. Phys.
101, 585 ͑1994͒.
18 H. A. Jahn and E. Teller, Proc. R. Soc. A 161, 220 ͑1937͒.
19 A. A. Kiseljov, J. Phys. B ͑At. Mol. Phys.͒ 2, 270 ͑1969͒.
15 R. M. Ibberson, W. I. F. David, and K. S. Knight, Rutherford Appleton
Laboratory Report No. RAL-92-031 ͑1992͒.
20
¨
U. Opik and M. H. L. Pryce, Proc. R. Soc. A 238, 425 ͑1957͒.
21 V. Pfennig and K. Seppelt, Science 271, 626 ͑1996͒.
22 S. K. Kang, H. Tang, and T. A. Albright, J. Am. Chem. Soc. 115, 1971
͑1993͒.
16 D. S. Silvia, Maximum Entropy and Bayesian Methods, edited by P. F.
Foug’ere ͑Kluwer Academic, The Netherlands, 1990͒, pp. 195–209.
17 W. I. F. David, R. M. Ibberson, and J. C. Mathewman, Rutherford Apple-
ton Laboratory Report No. RAL-92-032 ͑1992͒.
23 R. G. Pearson, J. Am. Chem. Soc. 91, 4947 ͑1969͒.
J. Chem. Phys., Vol. 104, No. 19, 15 May 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
216.165.95.72 On: Sat, 22 Nov 2014 23:56:04