TiF2: Linear or bent?

Article abstract of DOI:10.1002/anie.200704667

Calling nonlinearity into question: When Ti atoms are isolated in fluorine-doped argon matrices, TiF₄, TiF₃, TiF ₂, and TiF are identified by their IR spectra. The Ti isotope pattern observed for the ν₃ mode of TiF₂ is indistinguishable from that of a linear geometry. Therefore, there is now no reliable evidence for the nonlinearity of any 3d transition-metal difluorides or dichlorides. (Chemical Equation Presented)

Full text of DOI:10.1002/anie.200704667

Communications
DOI: 10.1002/anie.200704667
Transition-Metal Halides
TiF2: Linear or Bent?**
Antony V. Wilson, Alexander J. Roberts, and Nigel A. Young*
The shapes and geometries of the molecular transition-metal
halides have interested and intrigued experimental and
computational chemists for a long time. The majority of
such halides are high-temperature vapor-phase species, which
makes them challenging to study by structural and spectro-
scopic techniques. They are also a challenge to theoretical/
computational chemists. The vapor composition is often
complex, and spectral interpretation and assignment are not
always straightforward.[1,2] The naturally occurring chlorine
isotopes greatly enhance the reliability of spectral assign-
ments, but monoisotopic fluorine can make rigorous assign-
ments more difficult, especially in the absence of metal
isotopic data. There is a consensus from the experimental data
that all of the first-row transition-metal dichlorides, as well as
CaCl2 and ZnCl2, are linear, that CrF2 to ZnF2 are linear, and
that CaF2 is bent (ca. 150–1558), but owing to the lack of
Figure 1. Plot of SVFF force constants (frÀfrr) for MF2 versus MCl2. See
isotopes for ScF2 and VF2, no inference can be made from
their matrix IR spectra.[1,2] Therefore, because of the Ti
isotopes, TiF2 is the key molecule in understanding the
geometric and electronic structures of the first-row transition-
metal dihalides.
text for descriptions regarding the Ti data points.
considered to be almost certainly due to TiF3.[4] Hence, the
reported 1308 bond angle of TiF2 is actually that of 1208 in
TiF3, and analysis of the Ti isotope pattern confirmed this.[4]
An early (1969) matrix isolation IR study of the vapor-
ization of TiF3/Ti mixtures indicated a bond angle of about
1308 for TiF2.[3] However, in 1989 Beattie et al. showed that
this value was unreliable[4] by using a plot of the simple
valence force field (SVFF) force constants (frÀfrr) from the n3
values of MF2 versus those of MCl2, as indicated in Figure 1
(this is essentially identical to Figure 1in reference [4] except
that the data is limited to linear species, and to those which
are considered reliable[1,2,5]). The straight line is a fit to the
solid circles of the Cr, Mn, Fe, Co, Ni, and Zn data. The Ti
data point marked + is for the original n3 value of TiF2,
assuming linearity; it moves further away from the line if the
molecule is bent. Given the good fit to the other elements, this
was a good indication that the IR data supposedly for TiF2 was
erroneous, probably owing to the complexity of the titanium
fluoride vapor-phase system and high volatility of TiF4. The
IR bands originally assigned to TiF3 at about 790 cmÀ1 were
reassigned to TiF4 on the basis of gas-phase[6] and matrix
data[4] obtained from the evaporation of TiF4. Therefore, the
bands at about 740 cmÀ1 originally assigned to TiF2 were
Although not explicitly stated in the previous paper,[4]
À1
Figure 1gives an estimate of about 670 cm
for the n3
mode of linear TiF2. The only other experimental report on
molecular TiF2 is a matrix ESR report published in 1977 in
which the data were reported to be consistent with a bent
triplet structure; however, the signals were weak and broad
with no observable Ti or F hyperfine couplings.[7]
A detailed DFT study found the difluorides and dichlor-
ides of Mn to Zn were linear, but with soft, low-energy, large-
amplitude, bending vibrations.[5] The dihalides of Ca to Cr
(and especially the fluorides) were quasi-linear with large-
amplitude vibrations over a linear-geometry saddle point,
leading to imaginary harmonic frequencies for the bending
mode of the linear molecules.[5] The w3 mode of TiF2 was
744 cmÀ1 for the linear (saddle-point) geometry and 722 cmÀ1
for the bent (132.98) ground-state structure. More recent
multireference configuration interaction methods (icMRCI)
found TiF2 to be linear, but the near degeneracy of the ground
3
and first excited states (3ꢀgÀ, Dg) meant the ground state
could not be determined.[8] The w3 value was 705 cmÀ1 for the
À
3
3ꢀg state and 695 cmÀ1 for the Dg state. The better agree-
ment between calculated and experimental vibrational data
for TiCl2 than TiF2 was commented on,[8] but the unreliabil-
ity[4] of the published TiF2 experimental data[3] was not noted.
Therefore, as highlighted by Beattie,[1] TiF2 is the key
molecule in understanding the geometric and electronic
structures of the first-row transition-metal dihalides, and the
aim of this investigation was to obtain the first reliable
experimental values of the n3 vibrational mode of molecular
[*] A. V. Wilson, A. J. Roberts, Dr. N. A. Young
Department of Chemistry, The University of Hull
Kingston upon Hull, HU6 7RX (UK)
Fax: (+44)1482-466-410
E-mail: n.a.young@hull.ac.uk
[**] This work was supported by an EPSRC research grant (GR/T09651)
and a DTA studentship to A.V.W. Prof. Ian Beattie is thanked for
many helpful discussions.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
1774
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 1774 –1776

Angewandte
Chemie
TiF2 from matrix IR experiments and to use the Ti isotope
pattern to determine the bond angle.
When Ti atoms were trapped in 0.6% F2/Ar matrices
(Figure 2a), four sets of IR bands were observed in the nTi-F
ment (see the Supporting Information). With TiF4 assigned to
band A and TiF to band D, the only obvious assignment of the
remaining peaks, B and C, is to TiF3 and TiF2, respectively.
Therefore, the peaks at 743.68 and 739.70 cmÀ1 in 0.2%F2/Ar
matrices are assigned to 48TiF3 in
two different sites. The peaks at
677.73 and 673.81cm À1 are assigned
to 48TiF2 in two different sites. These
values are in excellent agreement
with the value predicted from
Figure 1, and an average of
675 cmÀ1 for linear TiF2 is marked
*
with an open circle ( ; if the
molecule is bent, this point will
move up vertically). Furthermore,
these values are also in excellent
agreement with the most recent
icMRCI-calculated
harmonic
values of w3 of 705 cmÀ1 for the
3ꢀg state and 695 cmÀ1 for the Dg
state,[8] especially as anharmonicity
values of À6.5 cmÀ1 have been
determined for CrF2.[14]
À
3
Figure 2. Matrix isolation IR spectra of Ti atoms isolated in F2/Ar matrices. a,b) 0.6% F2/Ar on
deposition at about 10 K (a) and after broad-band Hg–Xe photolysis (b). c,d) 0.2% F2/Ar on
deposition at about 10 K (c) and after broad-band photolysis and annealing to 25 K (d).
While the site effects (as well as
the gas-phase CO2 bands) poten-
tially complicate the determination
region at 792.0 (A); 743.4 and 739.7 (B); 678.0 and 674.0 (C);
and 635.7 cmÀ1 (D). After broad-band Hg–Xe photolysis
(Figure 2b) bands C and D reduced in intensity while A and B
increased. In these spectra the titanium isotope pattern
(central intense peak with a pair of weaker bands of similar
intensity on each side) on bands A, B, and C is masked by site
effects, and the presence of these is not surprising as they have
been observed previously for TiF4,[3,4] TiF3,[7] TiCl2,[9] and
of the bond angle, Figure 3 shows that the spectral motif
arising from the overlap of the Ti isotope patterns from each
of the sites is actually very diagnostic as the peak separation
decreases with a reduction in bond angle. In particular, the
double hump at about 680 cmÀ1 and the lack of any resolved
peaks between the two main peaks at 677.73 and 673.81cm À1
indicate that the bond angle is close to linearity. The vertical
lines marking the positions of the 46TiF2 peaks at 684.15 and
680.14 cmÀ1 and the 48TiF2 peaks at 677.73 and 673.81cm À1
imply that the bond angle is slightly less than linear. The best
peaks for an accurate bond angle determination are the 46TiF2
peak at 684.15 cmÀ1 and the corresponding 48TiF2 peak at
677.73 cmÀ1 of the a site, and an SVFF calculation yields a
bond angle of 1668. Analysis of the second-site (b) peaks of
46TiF2 and 48TiF2 at 680.14 and 673.81 cmÀ1 yields a bond angle
[10]
VF3 in both Ar and Ne matrices. In 0.2% F2/Ar matrices,
(Figure 2c) the same four bands were observed, but with
bands C (677.73 and 673.81cm À1) and D (635.55 cmÀ1) having
greater relative intensity than A (791.88 cmÀ1) and B (743.68
and 739.70 cmÀ1). After broad-band Hg–Xe photolysis and
annealing to 25 K (Figure 2d), the most marked difference is
the large increase in intensity of peak A. Despite the
continued presence of site effects, the Ti isotope pattern is
now present on bands A, C, and D, but the presence of
multiple sites for band B is still problematic. In 10% F2/Ar
matrices, only two broad bands corresponding to A and B
were observed at 792 and 738 cmÀ1. The behavior at different
F2 concentrations (0.16%, 0.2%, 0.3%, 0.6%, 1%, 2%,
10%), as well as on annealing and photolysis, confirmed that
the four sets of bands belonged to different species.
Based on the previous gas-phase[6] and matrix work[4,11] it
is straightforward to assign band A at 791.88 cmÀ1 in the 0.2%
F2/Ar matrix spectra to the n3 (T2) mode of 48TiF4. It should be
remembered that in the original work this band was assigned
to TiF3.[3] The nTi-F mode of gas-phase TiF has been observed
at about 650 cmÀ1,[12] with calculated harmonic values of
around 640 cmÀ1 for the F ground state.[13] Therefore, after
4
allowing for a reasonable matrix shift, band D at 635.55 cmÀ1
can be assigned to argon matrix isolated 48TiF, and the
titanium isotope pattern is also consistent with this assign-
Figure 3. Expansion of Figure 2d in the nTi-F region of TiF2 (top) and
SVFF-calculated spectra at different angles for the two sites of TiF2.
Angew. Chem. Int. Ed. 2008, 47, 1774 –1776ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1775

Communications
of 1608; however, owing to overlapping peaks, this result is
The Ti isotope pattern for the n3 mode is incompatible with a
substantially bent geometry (< 1608), and is indistinguishable
from a linear geometry for TiF2, which is also in agreement
with the recent icMRCI calculations.[8] Therefore, there is
now no reliable experimental evidence for the nonlinearity of
any first-row transition-metal difluoride or dichloride.
expected to be less accurate. Therefore, a bond angle of 1658
was used to calculate the overlapping isotope pattern for both
sites, and the result is shown in Figure 4. The fit is very good,
especially for the peaks above 672 cmÀ1, and while the fit
below this value is still good, the presence of the vapor-phase
CO2 components causes some interference.
Experimental Section
The general features of our matrix isolation experimental method-
ology have been described previously.[16] The titanium atoms were
evaporated at about 40 A from Ti wire (0.25 mm, Goodfellows)
wound on Ta wire (0.50 mm, Goodfellows). The F2/Ar mixtures were
prepared by using standard manometric procedures in a well-
passivated metal vacuum line using 10% F2/Ar supplied by Air
Liquide. The Ti evaporation conditions were checked using matrix
electronic absorption spectroscopy, and there was no evidence for
Ti2.[17] The SVFF calculations used SOTONVIB.[18]
Received: October 9, 2007
Published online: January 18, 2008
Keywords: halides · IR spectroscopy · matrix isolation · titanium
.
[1] I. R. Beattie, Angew. Chem. 1999, 111, 3494; Angew. Chem. Int.
Ed. 1999, 38, 3294.
[2] M. Hargittai, Chem. Rev. 2000, 100, 2233.
[3] J. W. Hastie, R. H. Hauge, J. L. Margrave, J. Chem. Phys. 1969,
51, 2648.
Figure 4. a) Expansion of Figure 2d in the nTi-F region of TiF2; b–
d) SVFF-calculated spectra for TiF2 with a 1658 bond angle: combined
sites (b), site a only (c), site b only (d).
[4] I. R. Beattie, P. J. Jones, N. A. Young, Angew. Chem. 1989, 101,
322; Angew. Chem. Int. Ed. Engl. 1989, 28, 313.
[5] S. G. Wang, W. H. E. Schwarz, J. Chem. Phys. 1998, 109, 7252.
[6] L. E. Alexander, I. R. Beattie, J. Chem. Soc. Dalton Trans. 1972,
1745; I. R. Beattie, P. J. Jones, J. Chem. Phys. 1989, 90, 5209.
[7] T. C. DeVore, W. Weltner, Jr., J. Am. Chem. Soc. 1977, 99, 4700.
[8] M. Vogel, W. Wenzel, Chem. Phys. Lett. 2005, 413, 42.
[9] J. W. Hastie, R. H. Hauge, J. L. Margrave, High Temp. Sci. 1971,
3, 257.
Isotopic substitution at the central element, as in this case,
is usually considered to give the lowest estimate of the bond
angle.[1,2,15]Therefore, it can be concluded from these IR data
that the titanium isotope pattern is inconsistent with a bent
species having a bond angle of less than 1608. The sine
function is notoriously insensitive for linear and near-linear
geometries. Although, because of the lightness of Ti and
separation of two mass units, this is a favorable case, the
calculated difference between the 1808 and 1658 n3 modes of
46TiF2 with 48TiF2 at 677.73 cmÀ1 is only 0.06 cmÀ1, which is
well below the accuracy with which band positions can be
measured in these types of experiments. Even a difference of
0.11 cmÀ1 for 1808 and 1608 bond angles is at the margin of
detection. Therefore, as for all of these experiments which
yield bond angles of 1608 and above, it is not possible to
distinguish the angles from linearity. Hence, these data
indicate that TiF2 is indistinguishable from linearity in argon
matrices. There is an interesting philosophical point that even
if the equilibrium geometry is linear, then the molecule
actually spends most of its time bent.
[10] V. N. Bukhmarina, A. Yu. Gerasimov, Yu. B. Predtechenskii,
Vib. Spectrosc. 1992, 4, 91 .
[11] B. S. Ault, J. Phys. Chem. A 1998, 102, 7245.
[12] T. Imajo, Y. Kobayashi, Y. Nakashima, K. Tanaka, T. Tanaka, J.
Mol. Spectrosc. 2005, 230, 139; R. S. Ram, P. F. Bernath, J. Mol.
Spectrosc. 2005, 231, 165; R. S. Ram, J. R. D. Peers, Y. Teng,
A. G. Adam, A. Muntianu, P. F. Bernath, S. P. Davis, J. Mol.
Spectrosc. 1997, 184, 186.
[13] A. I. Boldyrev, J. Simons, J. Mol. Spectrosc. 1998, 188, 1 38; C.
Koukounas, S. Kardahakis, A. Mavridis, J. Chem. Phys. 2004,
120, 11500.
[14] V. N. Bukhmarina, A. Y. Gerasimov, Y. B. Predtechenskii, V. G.
Shklyarik, Opt. Spectrosc. 1988, 65, 518.
[15] M. Allavena, R. Rysnik, D. White, V. Calder, D. E. Mann, J.
Chem. Phys. 1969, 50, 3399.
[16] A. J. Bridgeman, G. Cavigliasso, N. Harris, N. A. Young, Chem.
Phys. Lett. 2002, 351, 319.
[17] D. M. Gruen, D. H. W. Carstens, J. Chem. Phys. 1971, 54, 5206;
R. Busby, W. E. Klotzbꢀcher, G. A. Ozin, J. Am. Chem. Soc.
1976, 98, 4013; H. J. Himmel, A. Bihlmeier, Chem. Eur. J. 2004,
10, 627; O. Hꢀbner, H. J. Himmel, L. Manceron, W. Klopper, J.
Chem. Phys. 2004, 121, 7195.
In conclusion, this work has provided the first unambig-
uous IR characterization of TiF2 with the n3 mode of 48TiF2 at
675.73 and 673.81cm À1 for two sites in argon. These values
compare very well to the approximate value of 670 cmÀ1
predicted for linear TiF2 in Figure 1. They are also in good
agreement with the recent icMRCI-calculated w3 values of
[18] I. R. Beattie, N. Cheetham, M. Gardner, D. E. Rogers, J. Chem.
Soc. A 1971, 2240.
3
À
3
705 and 695 cmÀ1 for the ꢀg and Dg states, respectively.[8]
1776
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 1774 –1776