Magnetic properties of cationic tungsten(IV) half sandwich compounds: Experimental and theoretical study of a solvent and ligand stabilized singlet ground state leading to a thermally induced singlet-triplet spin state interconversion

Article abstract of DOI:10.1021/ja028313c

A series of novel neutral tungsten(III) and cationic tungsten(IV) complexes with disubstituted 4,4′-R,R-2,2′-bipyridyl (R₂-bpy) ligands of the type [Cp*W(R₂-bpy)Cl₂]ⁿ⁺ (n = 0, 1) were prepared and chararacterized by X-ray crystallography. Susceptibility measurements of the tungsten(IV) complexes revealed an intrinsic paramagnetism of these compounds and evidenced different magnetic properties of the dimethylamino and methyl (R = NMe₂, Me) substituted tungsten(IV) compounds in solution and in the solid state. In dichloromethane solution, singlet ground states with thermally populated triplet states were observed, whereas triplet (R = Me) and singlet ground states (R = NMe₂) were observed in the solid state. Using both experimental and theoretical techniques (DFT) allowed to establish solvation and ligand effects to account for the different magnetic behavior. Thermodynamic parameters were derived for the spin equlibria in solution by fits of the temperature dependent ¹H NMR shifts to the Van Vleck equation and were found to be in excellent agreement with the DFT calculations.

Full text of DOI:10.1021/ja028313c

Published on Web 06/03/2003
Magnetic Properties of Cationic Tungsten(IV) Half Sandwich
Compounds: Experimental and Theoretical Study of a
Solvent and Ligand Stabilized Singlet Ground State Leading
to a Thermally Induced Singlet-Triplet Spin State
Interconversion
Christian Cremer and Peter Burger*
Contribution from the Anorg.-chem. Institut, UniVersita¨t Zu¨rich,
Winterthurerstr. 190, 8057 Zu¨rich, Switzerland
Received August 28, 2002; Revised Manuscript Received March 10, 2003; E-mail: chburger@aci.unizh.ch
Abstract: A series of novel neutral tungsten(III) and cationic tungsten(IV) complexes with disubstituted
4,4′-R,R-2,2′-bipyridyl (R2-bpy) ligands of the type [Cp*W(R2-bpy)Cl2]n+ (n ) 0,1) were prepared and
chararacterized by X-ray crystallography. Susceptibility measurements of the tungsten(IV) complexes
revealed an intrinsic paramagnetism of these compounds and evidenced different magnetic properties of
the dimethylamino and methyl (R ) NMe2, Me) substituted tungsten(IV) compounds in solution and in the
solid state. In dichloromethane solution, singlet ground states with thermally populated triplet states were
observed, whereas triplet (R ) Me) and singlet ground states (R ) NMe2) were observed in the solid state.
Using both experimental and theoretical techniques (DFT) allowed to establish solvation and ligand effects
to account for the different magnetic behavior. Thermodynamic parameters were derived for the spin equlibria
1
in solution by fits of the temperature dependent H NMR shifts to the Van Vleck equation and were found
to be in excellent agreement with the DFT calculations.
Scheme 1
Introduction
The spin state transition between low and high spin states in
organometallic and coordination compounds is in general a well-
known process.1-8 In particular, a large number of detailed
studies have been devoted to Fe(II) based systems1-8 while
group 6 transition metal complexes on the other hand have been
only rarely investigated.9,10 Besides earlier NMR-based studies,
which have been pursued mostly in the late 1960s and early
1970s,6-9,11-13 the majority of the magnetic measurements were
carried out in the solid state. Therefore, it is not surprising that
little attention has been paid to solvation effects on the energetics
of the two spin states and was addressed in a quite limited
number of experimental studies.7,8,12-14 Linert et al. for instance
recently showed that the high/low spin energy gap in amine
donor Fe(II) complexes in solution is strongly influenced by
solvation effects.12,13
Herein, we report on the magnetic properties of half sandwich
tungsten(III, IV) compounds, which have been studied both in
solution and in the solid state. The position of the singlet-
triplet spin equilibrium observed for some of the W(IV)
complexes is strongly influenced by solvent and ligand effects
and was rationalized through a combination of both experimental
and theoretical methods. The presented methodology is con-
ceived as a case study to provide a general guideline for future
investigations of this type of systems/phenomena.
(1) Goodwin, H. A. Coord. Chem. ReV. 1976, 18, 293.
(2) Gu¨tlich, P. Struct. and Bond. 1981, 44, 83.
(3) Gu¨tlich, P.; Hauser, A.; Spiering, H. Angew. Chem. Int. Ed. Engl. 1994,
33, 2024.
(4) Kahn, O.; Krober, J.; Jay, C. AdV. Mater. 1992, 4, 718.
(5) Ko¨nig, E. Struct. Bond. 1991, 76, 51.
(6) Turner, J. W.; Schultz, F. A. Inorg. Chem. 2001, 40, 5296.
(7) Tweedle, M. F.; Wilson, L. J. J. Am. Chem. Soc. 1975, 98, 4824.
(8) Dose, E. V.; Murphy, K. M. M.; Wilson, L. J. Inorg. Chem. 1976, 15,
2622.
(9) Kriley, C. E.; Fanwick, P. E.; Rothwell, I. P. J. Am. Chem. Soc. 1994,
116, 5224.
Results
(10) Poli, R. Chem. ReV. 1996, 96, 2135.
(11) Gu¨tlich, P.; McGarvey, B. R.; Kla¨ui, W. Inorg. Chem. 1980, 19, 3704.
(12) Linert, W.; Enamullah, M.; Gutmann, V.; Jameson, R. F. Monats. Chem.
1994, 125, 661.
The investigated complexes are presented in Scheme 1; their
syntheses is described in the next section.
Complex Syntheses. The synthetic access to the cationic
pentamethylcyclopentadienyl (Cp*) tungsten(IV) complexes
(13) Strauss, B.; Gutmann, V.; Linert, W. Monats. Chem. 1993, 124, 515.
(14) LaMar, G. NMR of Paramagnetic Molecules; Academic Press: New York,
1973.
9
7664
J. AM. CHEM. SOC. 2003, 125, 7664-7677
10.1021/ja028313c CCC: $25.00 © 2003 American Chemical Society

Cationic Tungsten(IV) Half Sandwich Compounds
A R T I C L E S
Table 1. Data Collection Parameters of Complexes 1-NMe2 and 2-Me
1-NMe2
2-Me
formula
C48H53BCl2N4W‚CH2Cl2
1036.43
green rect. parallel epiped
0.2 × 0.2 × 0.3
monoclinic
P21/n (no. 14)
12.9300(10)
C22H27Cl2N2W‚1/2 Et2O
611.27
violet rect. parallel epiped
0.15 × 0.2 × 0.3
triclinic
P-1 (no. 2)
8.369(2)
10.952(2)
13.977(3)
102.84(3)
101.01(3)
101.60(3)
1185.7(4)
2
Fw
habitus
crystal dimension [mm]
crystal system
space group
a [Å]
b [Å]
c [Å]
20.888(2)
17.632(2)
R [°]
â [°]
108.450(10)
γ [°]
V [Å3]
Z
4517.3(8)
4
1.524
2.83
2096
183(2)
4-52
18000, 5421
F
calc [g‚cm-3
]
1.712
5.11
604
µ [mm-1
F(000)
T [K]
]
183(2)
2 QΘ [°]
4-52
no. of measured
reflections (total unique)
no. of parameters
R1(F2 > 2σF2)
20705, 10705
532
260
0.0266
0.0609
1.185
0.0318
0.0776
1.042
wR2(F2 > 2σF2)
GOOF, S
res. el. dens. [e-Å-3
]
1.703, -1.043
1.16, -1.44
Cp*W(R2-bpy)Cl2+, 1-Me and 1-NMe2, where R2-bpy is a 4,4′-
disubstituted 2,2′-bipyridyl donor (R ) Me and NMe2) was
previously developed by us from the corresponding trichloride
starting materials by chloride abstraction.15
Due to difficulties with the purification of the rather insoluble
unsubstituted trichloride complex, Cp*W(4,4′-H2-bpy)Cl3, the
latter route proved to be cumbersome for complex 1-H. We
have therefore decided to access 1-H from the corresponding
hitherto unknown tungsten(III) complex, 2-H, by one-electron
oxidation.
Both complexes 2-H and 2-Me are paramagnetic and display
identical magnetic moments µeff ) 1.67 µB at room temperature.
The corresponding cationic tungsten(IV) complex BArF1-H
was prepared by one-electron oxidation of complex 2-H with
one equiv. of ferrocenium tetraphenylborate followed by me-
tathesis with NaBArF4 (BArF4 ) B(3,5-(CF3)2-C6H3)4) (eq 2).
Syntheses of 2-H and 2-Me. The novel tungsten(III)
complexes 2-H and 2-Me were obtained by ligand induced
splitting of the dimeric µ-chloro bridged tungsten(III) complex,
(Cp*WCl2)2 in excellent yields. Because the latter compound
was previously described just as a cocrystallized complex with
a phosphorus byproduct,16 we established high yield synthetic
access to pure (Cp*WCl2)2 by Zn reduction of Cp*WCl4 (see
experimental). It should be noted that a similar strategy was
applied with success by Poli et al. for the synthesis of related
mono- and bidentate bisphosphine cyclopentadienyl Mo(III)
complexes.17
Complex BArF1-H was thus obtained as a microcrystalline,
analytically pure dark-green solid.
X-ray Crystal Structure of Complex 1-NMe2. We have
recently reported the X-ray crystal structure analysis of the
dimethyl substituted cationic (Cp*) tungsten(IV) complex,
1-Me.15 For comparison, an X-ray crystal structure analysis of
the corresponding dimethylamino complex, 1-NMe2, was sought
during the course of this study. The details of the data collection
and selected bond distances and angles are summarized in Tables
1 and 2, the molecular structure of 1-NMe2 is presented in
Figure 1.
The coordination geometry of the cationic complex 1-NMe2
is best described with a square-pyramidal (sq.-py.), i.e., a four-
legged pianostool geometry. The closest distance of the tungsten
center to the tetraphenylborate counterion is 5.38 Å (W-H),
which clearly suggests that complex 1-NMe2 has a formal 16-
electron configuration. As expected, the bipyridyl rings and
NMe2 unit are essentially coplanar; this also explains the
The dark red tungsten(III) complexes 2-R were obtained as
microcrystalline, analytically pure materials. Their four-legged
pianostool geometry was unambiguously confirmed by single-
crystal X-ray structure analysis of complex 2-Me (vide infra).
1
inequivalence of the NMe2 methyl groups observed in the H
(15) Cremer, C.; Burger, P. J. Chem. Soc., Dalton Trans. 1999, 1967.
(16) Harlan, C. J.; Jones, R. A.; Koschmieder, S. U.; Nunn, C. M. Polyhedron
1990, 9, 669.
(17) Abugideiri, F.; Brewer, G. A.; Desai, J. U.; Gordon, J. C.; Poli, R. Inorg.
Chem. 1994, 33, 3745.
NMR spectrum due to hindered rotation about the N-Caryl bond.
X-ray Crystal Structure of the Tungsten(III) Complex,
2-Me. In addition, we have carried out an X-ray crystal structure
9
J. AM. CHEM. SOC. VOL. 125, NO. 25, 2003 7665

A R T I C L E S
Cremer and Burger
Table 2. Selected Distances [Å] and Angles [deg] of Complexes
1-Me, 1-NMe2, and 2-Me
1-Me2
1-NMe2
2-Me1
W-Clavg
2.361(1)
2.139(3)
2.02
73.6(1)
83.1(4)
84.5
134.2
113.4
112.2
2.389
2.12
1.97
74.7
81.7
2.427
2.106
1.99
73.5
82.3
W-Navg
W-Zavg
N-W-N
Cl-W-Cl
cis-Cl-W-Navg
trans-Cl-W-Navg
N-W-Zavg
Cl-W-Zavg
84.2
83.5
133.8
113.3
114.6
132.3
114.9
112.6
Figure 3. Curie plot of [Cp*W(Me2bipy)Cl2][BPh4] 1-Me in CD2Cl2/CCl3F
(3:1). A similar deviation from linearity is observed for the remaining
resonances.
the major difference between the tungsten(III) and tungsten-
(IV) complexes are the tungsten chloride bond distances, which
are 0.06 Å shorter in the cationic complex 1-Me than in the
corresponding neutral compound 2-Me suggesting stronger
π-donation of the chloride ligands to the more Lewis-acidic
cationic tungsten center. As will be elaborated later on in the
discussion section, the large deviation of the W-CCp* distances
observed in the complexes described herein is a major feature
of these compounds.
1
Temperature, Solvent and Anion Dependent H NMR
Studies of Complexes 1-R and 2-Me. (a) Temperature
1
Dependence in CD2Cl2 Solution. The dispersion of the H
-
Figure 1. Ortep Plot (50%) level of complex 1-NMe2. The BPh4
counterion has been omitted for clarity.
NMR chemical resonances in complexes 1-R clearly evidenced
the intrinsic paramagnetism of these compounds. This is best
exemplified with the chemical shift range of (100 ppm observed
1
for the H NMR resonances of 1-Me in CD2Cl2 at RT. This
prompted us to investigate the temperature dependence of the
1H NMR chemical shifts. In exemplary fashion this is shown
for the vt 1H NMR chemical shift of the Cp* methyl resonance
in 1-Me plotted vs reciprocal temperature (Figure 3), which was
recorded in a CD2Cl2/CCl3F mixture to reach lower tempera-
tures. A comparison with data obtained in pure CD2Cl2 showed
no significant influence of the solvent mixture on the 1H NMR
chemical shifts.
The strong deviation of the chemical shifts vs T-1 from
linearity, which was also observed for the remaining 1H NMR
resonances (not shown), immediately demonstrated the non-
Curie behavior of complex 1-Me (Figure 3). Similar results were
obtained for the dimethylamino substituted complex 1-NMe2
1
although the dispersions of its H NMR chemical shifts were
appreciably smaller. The magnetic behavior of the substituted
bipyridyl complexes 1-Me and 1-NMe2 in CD2Cl2 solution was
therefore clearly distinguishable from a pure triplet state. Prior
to our DFT calculations of these systems, which also prompted
us to investigate the unsubstituted bipyridyl complex 1-H, we
intended to establish that a Curie behavior could be observed
in these square pyramidal tungsten halfsandwich complexes after
all. Therefore, we studied the temperature dependence of the
1H NMR shifts of the corresponding neutral tungsten(III)
complex 2-Me in solution which is shown for one of the five
resonances in Figure 4.
Figure 2. Ortep Plot (50%) level of the neutral tungsten(III) complex 2-Me.
analysis of the neutral tungsten(III) complex 2-Me. The mo-
lecular structure of 2-Me is shown in Figure 2, details of the
data collection and selected distances and angles are compiled
in Tables 1 and 2.
The molecular structure of the neutral tungsten(III) complex
2-Me strongly resembles its cationic tungsten(IV) congener
1-Me and also 1-NMe2. This is best shown by a comparison of
the geometrical parameters involving the metal centers of 1-Me,
2-Me, and 1-NMe2, which are compiled in Table 2.
While essentially identical metal centered angles were noticed,
As anticipated, the 1H NMR shifts of the open shell S ) 1/2
tungsten(III) complex 2-Me displayed an excellent linear
9
7666 J. AM. CHEM. SOC. VOL. 125, NO. 25, 2003

Cationic Tungsten(IV) Half Sandwich Compounds
A R T I C L E S
Figure 4. 1/T dependence of the 1H NMR resonances of tungsten(III)
complex 2-Me in CD2Cl2. (R2 ) 0.999).
Figure 6. 1/T-dependence of the Me2bipy chemical shifts Protonen in
BArF
[Cp*W(Me2bipy)Cl2][X] 1-Me (X ) BPh4-) and
(9) in CD2Cl2.
1-Me (X ) BArF)
Figure 5. Curie behavior of 1-H (shown for the Cp* methyl proton shift)
in CD2Cl2. Linear regression (- -) with R2 ) 0.999.
Figure 7. 1/T dependence of the 1H NMR chemical shifts of the Me2bipy
BArF
protons in [Cp*W(Me2bipy)Cl2][BArF]
1-Me in CD2Cl2 (2) and in a
toluene-d8/CD2Cl2 (2:1) mixture (9) (R2 ) 0.978).
dependence on the reciprocal temperature (Figure 4). This
clearly substantiated that an agreement with the Curie-law could
be in principle also observed in complexes 1-Me and 1-NMe2,
provided that the cationic tungsten(IV) systems had a pure triplet
state.
1
that the temperature dependent H NMR chemical shifts are
essentially independent of the counterion.
(c) Mixtures of CD2Cl2/Toluene (1:2 v/v). To probe for the
polarity dependence of the spin-state transition equilibrium in
Prior to the measurements of the unsubstituted complex 1-H,
this rationale was confirmed by the results of our DFT
calculations. The latter also suggested the possibility to induce
a change of the position of the spin-state equilibrium through
different substitutents R of the 4,4′-R2-bpy ligands (vide infra).
1
complex 1-Me, we recorded vt H NMR spectra of complex
BArF1-Me in a CD2Cl2/toluene 1:2 (v/v) mixture. Due to the
rather low solubility of the cationic complexes, the usage of
the partially fluorinated BArF anion was a prerequisite for these
measurements. The plot of the chemical shift for the bipyridyl
methyl group vs reciprocal temperature is shown in Figure 7.
1
We have therefore also recorded vt H NMR spectra of the
unsubstituted bipyridyl tungsten(IV) complex 1-H (Figure 5).
The inspection of the temperature dependent 1H NMR
chemical shifts vs T-1 data of 1-H clearly revealed the Curie
behaVior of this complex (Figure 5), which was fully consistent
with the trend predicted by our theoretical calculations.
(b) Anion Dependence. The temperature dependence of the
1H NMR chemical shifts of the analogue of 1-Me, complex
BArF1-Me with the more weakly coordinating anion was also
investigated in CD2Cl2 solvent (Figure 6).
Despite some minor difference in the curvatures and changes
of the chemical shifts, the δ vs T-1 plots for the cationic 4,4′-
Me2-bipyridyl complexes 1-Me and BArF1-Me presented in
Figure 6 are essentially identical. A strong deviation from the
Curie behavior (linearity) is therefore also attributable to
complex BArF1-Me. Hence, this measurement clearly revealed
Although the plot presented in Figure 7 displayed a slight
curvature, a linear T-1 dependence of the chemical shifts can
be conceived with good confidence (R2 ) 0.978). Therefore,
in sharp contrast to the results observed in pure CD2Cl2 solvent
(data are included for comparison in Figure 7), a Curie behaVior
F
can be attributed to complex BAr 1-Me in this significantly less
polar solVent mixture.
(d) Concentration Dependence including UV/Vis Spectral
1
and Molar Conductivity Data. The H NMR chemical shifts
of complex 1-Me in CD2Cl2 displayed a negligible concentration
dependence (( 2 ppm), which was evidenced through measure-
ments at 10-1 and 10-3 M concentrations. Further support for
the absence of a chemical equilibrium was established through
the perfect Lambert-Beer behavior of the absorptions of the
9
J. AM. CHEM. SOC. VOL. 125, NO. 25, 2003 7667

A R T I C L E S
Cremer and Burger
deeply green colored complex 1-Me in the UV/Vis spectra
measured in the concentration range of 10-5-10-3 M. In
addition, concentration dependent molar conductivity measure-
ments (not shown) were performed for 1-Me15 The molar
conductivity of Λm ) 58 S‚cm2‚mol-1 at 10-4 molar concentra-
tion,15 together with the observed [1-Me]1/2 concentration
dependence (range of 10-3-10-5 M) in dichloromethane clearly
allowed to establish that 1-Me behaves as an 1:1 electrolyte. It
can be therefore described with a formal 16-electron configured
tungsten center.
Determination of Thermodynamic Parameters. At the
outset of this section, it should be emphasized that the observed
1
temperature induced H NMR spectral changes for all com-
pounds described herein are fully reversible. The determination
of the thermodynamic parameters for the singlet triplet equi-
librium, S f T, in complexes 1-Me and 1-NMe2 was initially
attempted by a fit of the temperature dependence of the observed
1H NMR chemical shifts, δobs (Hobs) to eq 3 using the Van-
Vleck equation in the form applied by Rothwell et al. for their
successful fit of a tungsten(IV) singlet-triplet spin-equilibrium
in toluene solution.9
Figure 8. Temperature dependence of the 1H NMR chemical shifts δobs in
[Cp*W(Me2bipy)Cl2][BPh4] 1-Me: -Mebipy (2), Harom (9, b, [) und Cp*
(1). The fitting curves were derived using eq 4.
HO‚A‚g‚â‚(6‚e-E/kT
(γH/2π)‚3‚k‚T‚(1 + 3‚e-E/kT
)
Hobs ) Hdia
+
(3)
)
However, severe problems with the fits to this equation for
complex 1-Me (details see discussion section) let us resort to
the more general eq 4. The relation between eqs 3 and 4 arises
from the proper treatment of entropy terms. In addition, the free
enthalpy change of solvation is not properly accounted for in
eq 3. Equation 4 was originally derived by Eaton et al.18 and
revised later by Horrocks et al.19 It has to be stressed that this
treatment is only appropriate if the observed paramagnetic shifts
can be solely attributed to a contact-shift, i.e., through bond
term.14 The justification of this assumption for complexes 1-Me
and 1-NMe2 will be provided in the discussion section
Figure 9. Temperature dependence of the proton chemical shifts in [Cp*W-
((NMe2)2bipy)Cl2][BPh4] 1-NMe2: Harom (9, b, 2), -NMe2 (+, /) und
Cp* (1). The fitting are based on eq 4.
Table 3. Diamagnetic Chemical Shifts δdia Hyperfine Coupling
Constants A with esd’s for the Protons in [Cp*W(R2Bipy)Cl2][BPh4]
1-Me and 1-NMe2
A‚γe‚g‚â
γH‚3‚k‚T‚(1 + e(∆H-T∆S)/RT
δobs ) δdia
+
(4)
)
1-Me
dδia [ppm]
A [MHz]
1-NMe2
δ
dia [ppm]
A [MHz]
Here γe, γH are the gyromagnetic ratios of the electron and
proton. â is the Bohr magneton, k is the Boltzmann constant, g
is the g-factor of the unpaired electron, A is the hyperfine
splitting constant of the observed proton, ∆H is the enthalpy,
and ∆S is the entropy change for the singlet triplet spin
equilibrium.
Harom
Harom
Harom
9 ( 3 + 0.56 ( 0.03 Harom 8.49 ( 0.04 + 0.05 ( 0.01
8 ( 2 - 0.12 ( 0.01 Harom 6.99 ( 0.01 + 0.01 ( 0.01
6 ( 3 - 0.28 ( 0.03 Harom 6.71 ( 0.02 - 0.04 ( 0.01
-Mebipy 2 ( 4 - 0.56 ( 0.03 -NMe2 3.25 ( 0.01 - 0.07 ( 0.01
Cp* 3 ( 1 - 0.09 ( 0.01 -NMe2 3.22 ( 0.02 - 0.06 ( 0.01
Cp* 1.98 ( 0.01 - 0.04 ( 0.01
The quality of the fit can be estimated from the optimized
values for the proton resonances, which should lead to different
values for δdia,i and Ai for the individual proton sites i, but to
the same values for ∆H and ∆S. Additionally, the fit can be
checked by inspection of the derived optimized chemical shifts
Table 4. Averaged Thermodynamic Parameters with esd’s for the
Singlet-Triplett Spin State Equilibrium and Estimated Based
Thereon and Experimental Effective Magnetic Moments of
[Cp*W(R2Bipy)Cl2][BPh4] 1-Me, 1-NMe2
complex
∆H [kJ mol-1
]
∆S [Jmol-1K-1
]
µeff, calc. [µB]
µeff, exp [µB]
δ
dia,i. For a reasonable fit, the values of δdia,i are expected to be
4-Me
4-NMe2
4.5 ( 0.1
14 ( 1
29 ( 1
37 ( 7
2.10
0.58
2.02b
0.64a
in the range of the corresponding diamagnetic compounds/free
ligands and should display small esd’s. The fits of the temper-
ature dependent 1H NMR shifts for complexes 1-Me and
1-NMe2 are presented in Figures 8 and 9; the derived values
for the optimized parameters are summarized in Tables 3
and 4.
Using the derived thermodynamic parameters, spin equilib-
rium constants (S f T) of K ) 106 for 1-Me and 5‚10-5 for
complex 1-NMe2 can be derived at RT. Using these values and
µeff,triplet ) 2.49 µB from the solid-state measurements for the
pure triplet and µsinglet ) 0 for the singlet states, this allowed to
estimate the effective magnetic moments µeff,calc in CD2Cl2
(18) Eaton, D. R.; Phillips, W. D.; Cadwell, D. J. J. Am. Chem. Soc. 1963, 85,
397.
(19) Pignolet, L. W.; Horrocks, W. D., Jr. J. Am. Chem. Soc. 1969, 91, 3976.
9
7668 J. AM. CHEM. SOC. VOL. 125, NO. 25, 2003

Cationic Tungsten(IV) Half Sandwich Compounds
A R T I C L E S
Table 5. DFT Derived Total and Relative Energies (in
parentheses) of Complexes 1-R in [Hartree] and [kJ/mol]
S ) 0
S ) 1
1-Mea)
-1952.160651
-1952.167339 (-17.6)
1-Merotaa
)
-1952.159857 (+2)
-1952.1352275 (+67)
-1952.1252866 (+92)
-1951.1220302
1-Metbpy(II)a
1-Meiso(III)a
1-Meb
1951.1315935 (-25.0)
1873.497976 (-16.7)
-2141.56705 (-12.5)
-2140.4016170 (-16.7)
1-Ha
-1873.49162912
-2141.561919
-2140.3946289
a
1-NMe2
1-NMe2
b
a BP86 functional. b B3LYP functional.
formalisms at the density functional theory (DFT) level and the
nonlocal Becke-Perdew-86 functional.22,23 The RI-J method
implemented in the RIDFT24,25 program of Turbomole26 program
suite was applied with the spin occupations fixed at either the
singlet (S ) 0) or triplet (S ) 1) spin state. The amount of spin
contamination was carefully monitored by the inspection of the
expectation value of the spin operator S2 ‚ S2 was found in
all cases to be 2.00 ( 0.01 for the triplet states and 0.00 (
0.01 for the singlet states, thus evidencing negligible contribu-
tions of higher spin states. A Stuttgart-Dresden (ecp-60-mwb)
pseudopotential27 was used throughout in these calculations for
the tungsten center and the structures were initially optimized
using a TZVP28 basis set for W and SVP29 basis sets for the
residual atoms. After convergence, optimization was continued
with the large TZVP basis applied to all atoms; usually this
required just a few steps to converge. For the methyl substituted
complex 1-Me the singlet and triplet state optimized geometries
(TZVP basis) were characterized as true minima through the
calculation of the numerical second derivatives (finite differ-
ences), which displayed the absence of imaginary vibrations.
In addition, we performed geometry optimizations with the
unrestricted hybrid UB3LYP functional30 using a TZVP basis
for all atoms and the structures optimized with the BP86
functional as starting geometries. Final relative energies of these
calculations are summarized in Table 5.
Figure 10. Temperature dependence of the effective magnetic moment
µeff of [Cp*W(Me2bipy)Cl2][BPh4] 1-Me in the solid state (9) and in
solution (2, Evans method).
solution. A comparison of the experimentally determined values
in solution (Evans method) and the estimated magnetic moments
(µeff,exp and µeff,calc) established the excellent agreement of these
data (Table 4). This clearly validated the experimentally
determined thermodynamic parameters.
VT magnetic susceptibility measurements of 1-Me and
1-NMe2 in the solid state and solution (Evans method):
Solid-state magnetic susceptibility measurements for complexes
1-Me and 1-NMe2 were performed in the temperature range
from 4 to 300 K. The temperature-dependent effective magnetic
moment of complex 1-Me presented in Figure 10 clearly
evidenced a triplet ground state with zero-field splitting and an
effective magnetic moment, µeff ) 2.49 µB in the solid state at
room temperature.
The vt magnetic moment data of complex 1-Me in dichlo-
romethane were derived from magnetic susceptibility data in
solution by the Evans method,20,21 (see exp. for details) and are
also included in Figure 10. An inspection of Figure 10
immediately revealed that the CH2Cl2 solution data are in sharp
contrast to the solid-state measurements. In solution, a strong
deviation from linearity and significant smaller values of the
effective magnetic moment µeff vs temperature data were
observed. Note, that this is consistent with the aforementioned
temperature dependence of the 1H NMR chemical shifts of 1-Me
in CD2Cl2 solution.
Finally, vt solid-state magnetic susceptibility data for complex
1-NMe2 showed that the dimethylamino substituted complex
is diamagnetic in the temperature range from 10 to 300 K. This
result slightly deviated from our magnetic susceptibility data
measured at 298 K with a Johnson-Matthey laboratory magnetic
balance, which gave a reproducible value of µeff ) 0.52 µB.
Although we have no explanation for this slight difference, both
results clearly provided evidence for the reduced magnetic
moment of complex 1-NMe2 compared to its methyl substituted
congener 1-Me.
Further evidence for the higher stability of the triplet states
of complexes 1-H, 1-Me, and 1-NMe2 in the gasphase was
provided through SCF-instability calculations of the S ) 0 states,
which were performed with the ESCF module31 of the Turbo-
mole program package. This was manifested through the
observation of triplet instabilities for the converged wave
functions of the singlet states of these complexes.
Furthermore, we performed calculations for the hypothetical
pseudo-trigonal biypyramidal (tbpy) isomer, 1-Metbpy presented
in Figure 11. The geometry was constrained to Cs symmetry
containing the chlorine and tungsten atoms in the symmetry
plane, since the geometry optimization collapsed otherwise to
the pianostool geometry, i.e., 1-Me. In addition, we performed
(22) Becke, A. D. Phys. ReV. A 1988, 38, 3098.
(23) Perdew, J. P. Phys. ReV. B 1986, 33, 8822.
(24) Eichkorn, K.; Treutler, O.; O¨ hm, H.; Ha¨ser, M.; Ahlrichs, R. Chem. Phys.
Lett. 1995, 242, 652.
DFT Calculations and MO Analyses of Complexes 1-R.
The coordinates of the cationic metal complexes 1-Me and
1-NMe2 observed in the X-ray crystal structures, with the C-H
bonds set to 1.1 Å, were used as starting geometries. These
structures were optimized using the open shell unrestricted
(25) Eichkorn, K.; Treutler, O.; O¨ hm, H.; Ha¨ser, M.; Ahlrichs, R. Chem. Phys.
Lett. 1995, 240, 283.
(26) Ahlrichs, R.; Ba¨r, M.; Ha¨ser, M.; Horn, H.; Ko¨lmel, C. Chem. Phys. Lett.
1989, 162, 165.
(27) Andrae, D.; Ha¨ussermann, U.; Dolg, M.; Stoll, H.; Preuss, H. Theor. Chim.
Acta 1990, 77, 123.
(28) Scha¨fer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1994, 100, 5829.
(29) Scha¨fer, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97, 1992.
(30) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
(31) Weiss, H.; Ahlrichs, R.; Ha¨ser, M. J. Chem. Phys. 1993, 99, 1262.
(20) Evans, D. F. J. Chem. Soc. 1959, 2003.
(21) Schubert, E. M. J. Chem. Educ. 1992, 69, 62.
9
J. AM. CHEM. SOC. VOL. 125, NO. 25, 2003 7669

A R T I C L E S
Cremer and Burger
Table 6. Relative Energies in kJ/mol of the Singlet and Triplet
Spin States of Complex 1-Me in the Gas Phase and in
Dichloromethane Solution
S ) 1
S ) 0
gas phasea
0
0
+15
CH2Cl2 solutiona
+6.7
T ) 298 K, ꢀ ) 8.9, d ) 1.325 a
CH2Cl2 solutiona
0
0
+1
T ) 180 K, ꢀ ) 16.3, d ) 1.573a
dichloromethane solution (RT, ꢀ ) 8.9)b
-12
a Jaguar 4.1 calculations (see text). b DMOL calculations (see text).
Figure 11. DFT derived relative energies of geometry optimized isomers
of [Cp*W(Me2bipy)Cl2]+ 1-Me (S ) 0) (DFT).
version 5.3 of Turbomole, the density functional program
package DMOL was sought as an alternative. Unfortunately,
usage of relativistic ECP’s is not available for COSMO
calculations in the current version of DMOL.33 We therefore
performed all-electron calculations, in which scalar relativistic
effects for all atoms were included using the approach described
by Delley et al. (VPSR keyword).34 The methylenechloride
solvent was modeled with a dielectric constant of ꢀ ) 8.9 (at T
) 298 K) and using a probe radius of 2.3 Å.35 Both the singlet
and triplet states of 1-Me were fully optimized applying dnp
basis sets for all atoms using the COSMO implementation and
a BP86 functional with the standard convergence criteria of
DMOL. Final total energies are tabulated in Table 6.
In addition, we performed calculations with the Jaguar rel.
4.1 program package36 using the self-consistent reaction field
(SCRF) approach implemented in Jaguar.37 A LACV3P basis
set including a Hay-Wadt scalar relativistic pseudopotential for
the tungsten metal center and 6-31G** basis sets for the residual
atoms and the BP86 functional were used. The aforementioned
BP86 and UBP86 optimized gasphase structure (TZVP basis)
of complex 1-Me were used as starting geometries and required
just a few geometry optimization cycles to converge with the
BP86 (S ) 0) and UBP86 (S ) 1) functionals. In a next step,
the geometries were also optimized including solvation by
dichloromethane at 298 K. The temperature dependent change
of the solvent was modeled at 180 K using experimental density
and permittivity data of dichloromethane at this temperature
from the literature.38 The derived relative energies for both spin
states are also summarized in Table 6.
Figure 12. 12. HOMO and LUMO for the S ) 0 state of 1-Me.
a geometry optimization of another pseudo-trigonal biypyra-
midal isomer, i.e., 1-Meiso using again Cs symmetrical con-
straints with the N and W atoms spanning the symmetry plane.
The relative energy of the geometry optimized structure of
1-Meiso is also included in Figure 11.
In addition, we inspected the frontier orbitals of complex
1-Me. The molecular orbitals of the HOMO and LUMO of the
singlet state (S ) 0) are presented in Figure 12.
We also analyzed the equivalent molecular orbitals (MOs)
of the corresponding triplet state of 1-Me, i.e., the second
highest, single occupied molecular orbital, SSHOMO, and the
SHOMO. As anticipated, the same ordering and shape of the
orbitals was observed; therefore the MOs shown in Figure 12
for the singlet state of 1-Me are also a applicable to the
corresponding orbitals of the S ) 1 state.
Furthermore, we analyzed the HOMO-LUMO gaps (∆S)0),
of the singlet states and the energy differences ∆S)1 between
the SSHOMO and the SHOMO levels in the S ) 1 states (∆S)1
) ESHOMO - ESSHOMO). The values of ∆S)0 in complexes 1-R
are appreciably small and in the narrow range of 0.28-0.30
eV. For the energy differences, ∆S)1 the order 1-H (0.47) <
1-Me (0.48) < 1-NMe2 (0.64) with the energies given in
parentheses in [eV] is observed.
Discussion
In a recent account Poli reviewed the chemistry of open-
shell paramagnetic organometallic complexes.10 This clearly
evidenced that little attention has been paid to the magnetic
properties, in particular spin state transitions, of tungsten d2
metal complexes, i.e., in their tetravalent oxidation state, while
molybdenum(IV) complexes were studied in some detail by the
author.10,17 In principle, one might infer that halfsandwich
complexes with a d2-electron configuration could be intrinsically
diamagnetic. In fact, however, quite frequently just the opposite
is observed, and in the aforementioned review R. Poli discusses
(33) Andzelm, J.; Ko¨lmel, C.; A., K. J. Chem. Phys. 1995, 103, 9312.
(34) Delley, B. Int. J. Quantum Chem. 1998, 69, 423.
DFT Calculationsssolvent Effects. The effect of solvation
on the energy differences between the triplet and singlet state
were probed using a dielectric continuum approach. Because
the COSMO solvation method developed by Klamt32 et al. was
not implemented for unrestricted open-shell calculations in
(35) This value was estimated using the formalism prescribed in Jaguar, i.e.,
from r3
) 3/4π‚(M/Na)‚∆/F, where M is the molecular weight of
dichloromethane (85.9 g/mol), Na is Avogadro’s number, ∆ is the packing
density, which was set to 0.5, and F is the density of the dichloromethane.
(36) Jaguar 4.1, Schro¨dinger., Inc., Portland, OR, 1991-2001.
(37) Tannor, D. J.; Marten, B.; Murphy, R.; Friesner, R. A.; Sitkoff, D.; Nicholls,
A.; Rignalda, M.; Goddard, W. A., III; Honig, B. J. Am. Chem. Soc. 1994,
116, 11 875.
(32) Klamt, A. J. Phys. Chem. 1995, 99, 2224.
(38) Morgan, S. O.; Lowry, H. H. J. Phys. Chem. 1930, 34, 2385.
9
7670 J. AM. CHEM. SOC. VOL. 125, NO. 25, 2003

Cationic Tungsten(IV) Half Sandwich Compounds
A R T I C L E S
Figure 13. Molecular orbital scheme in sq.-pyramidal geometry.
aspects when to expect open shell systems.10 At the risk to
reiterate their arguments, it is deemed nevertheless instructive
to consider the following scenario. In a d2-configured C4V
symmetrical ML5 transition metal complex with a sq.-pyramidal
coordination geometry, the initial small energy difference of
the dxy (b2) and the degenerate dxz and dyz (e) orbitals is raisedon
widening the metal centered angle æ between the apical and
basal ligands from 90° (Figure 13).39 Increase of the æ angle
also significantly lowers the energy of the dz2 orbital in that it
is eventually lying below the degenerate set of the dxz and dyz
orbitals (not presented in Figure 13). Considering the Cp ligand
as a monodentate apical ligand L, this picture holds in principle
also for CpML4 complexes. As has been shown by R. Hoffmann
et al. in their EHT analysis on the latter system,40 the dxz and
dyz orbitals are further raised in energy with respect to the dz2
orbital due to additional dπ-pπ interactions with the e1′ set (D5h
symmetry) of the cyclopentadienyl π-system. Therefore, just
the rather low lying metal based dxy and dz2 orbitals will be
occupied in CpML4 systems; as will be shown below this is
also observed for the complexes described herein.
solution by fits of the observed temperature dependent 1H NMR
data to the Van-Vleck equation (eq 3).
The strong deviation from the Curie behavior of the temper-
1
ature dependent H NMR resonances (Figure 4) in the formal
16-electron configured cationic complexes 1-Me and 1-NMe2
and the similarity of the plots for the individual chemical shifts
vs temperature (Figures 10 and 11) with the results of Rothwell
et al. suggested also a singlet-triplet spin-state equilibrium for
1-Me and 1-NMe2 in CD2Cl2 solution. This contrasted the
results observed for the related cationic molybdenum halfsand-
+
wich complexes with a pianostool geometry, i.e., CpMoL2X2
,
where L2 is a mono- or bidentate phosphine donor and X is a
halide ligand, which were found to have a pure triplet state in
the temperature range of 4 K to 293 K.17
To establish that a Curie behavior could be observed for the
cationic Cp*WCl2(R2-bpy)+ system after all, we prepared and
investigated the magnetic properties of the novel corresponding
neutral d3-W(III) complex, 2-Me. The linear dependence of the
1H NMR chemical shifts vs reciprocal temperature presented
in Figure 4, clearly established a Curie behavior for 2-Me. This
supported the view that provided complexes 1-Me and 1-NMe2
inherited pure triplet ground states, one might observe a Curie
dependence of their 1H NMR shifts. Note, that this was
eventually confirmed for complex 1-H (Figure 5) in the course
of our investigations.
In the absence of additional splitting of these levels due to
π-acceptor ligands or, (to smaller extent) due to π-donors, it
becomes readily apparent that these complexes display just small
HOMO-LUMO gaps, ∆. In principle, the multiplicity of the
ground state of the complex, i.e., singlet vs triplet state is
determined by the relative sizes of ∆ and the pairing energy,
PE. If ∆ < PE a triplet state is preferred, which is likely to be
the case for the system described above with an anticipated small
∆. Although this simplified picture is not rigorously correct,
and readers are referred to R. Poli’s excellent review for
details,10 it is quite self-instructive to explain the presence of
thermally populated spin state equilibria in these systems with
∆ ≈ PE.
1
To rule out other possible explanations for the observed H
NMR temperature dependence in complexes 1-Me and 1-NMe2,
we studied counterion (b) and concentration (equilibrium, (d))
effects for complex 1-Me. The topic under item (b) was
BArF
investigated through 1H NMR measurements of 1-Me
,
whereas (d) was studied by 1H NMR spectroscopy at two
different concentrations (10-1 and 10-3 M) of complex 1-Me.
In both cases, essentially unchanged H NMR spectra were
1
In an elegant study, Rothwell et al. established singlet-triplet
spin state equilibria of this type for a pseudo-octahedral pyridyl
phenoxy tungsten(IV) system with a formal 14-electron con-
figuration.9 These complexes display singlet ground states with
the triplet state being partially thermally populated. The singlet-
triplet energy gaps ∆E were determined quantitatively in toluene
observed (Figure 6), which evidenced that deviations from the
Curie behavior could be attributed to neither item (b) nor (d).
This view was also supported through the perfect Lambert-
Beer behavior observed for the bands in the UV/Vis spectrum
of 1-Me. In addition, we analyzed the ν(W-Cl) Raman
spectroscopic data in complex 1-Me both in the solid state and
in solution to probe for possible isomers. Because just negligible
(39) Albright, T. A.; Burdett, J. K.; Whangbo, M.-W. Orbital Interactions in
differences of the ν(W-Cl) stretching frequencies (<5 cm-1
)
Chemistry; John Wiley & Sons: New York, Chichester, Brisbane, Toronto,
Singapore, 1985.
were detectable in the solid state and solution spectra, there
was no evidence for an additional species in solution. Although
(40) Kubacek, P.; Hoffmann, R.; Havlas, Z. Organometallics 1982, 1, 180-
188.
9
J. AM. CHEM. SOC. VOL. 125, NO. 25, 2003 7671

A R T I C L E S
Cremer and Burger
this did not unambiguously rule out the presence of other
isomers with a different coordination geometry, e.g., a pseudo
trigonal-bipyramidal structure, together with the results of the
DFT calculations (vide infra) and previous studies by Hoffmann
et al.,40 we take this as a good hint that the square pyramidal
structure observed in the solid state is also maintained in
solution.
paramagnetic contributions to the chemical shifts originate
exclusively from the contact (through bond) rather than from
the pseudo-contact (dipolar, through space) shift term. Following
a previously devised method,14 the presence of a prevailing
contact-shift term in the investigated compounds was confirmed
by comparison of the bipyridyl proton resonances in the
unsubstituted complex 1-H and its methyl substituted congener,
1-Me: Although the same signs of the paramagnetic shifts were
observed for the proton resonances in the 3, 5, and 6 positions
of the bipyridyl ring in complexes 1-H and 1-Me, the
paramagnetic shifts of the bipyridyl methyl protons and the ring
proton in the 4-position were found to display opposite signs.
This clearly suggested that the prerequisite for the fit procedure
described above was indeed fulfilled. It should be noted that it
has been additionally conceived9 that a successful fit of the NMR
data to the Van-Vleck equation is also a reasonable indication
for an exclusive/prevailing contact shift term. As shown below,
this criterion was also fulfilled for complexes 1-Me and 1-NMe2.
Initially, fits to the Van-Vleck equation in the form used by
Rothwell et al. in the aforementioned successful study of
tungsten(IV) complexes were attempted (eq 3).9 It is noteworthy
that fits to eq 3 were also used with success by other groups
for other complexes, e.g., by Poli et al. for a Nb halfsandwich
system.41
Using eq 3 the fits for complex 1-NMe2 were apparently
successful, which was evidenced through the small standard
deviations for the derived values for ∆E and the diamagnetic
shifts δdia for the individual protons, which were in the expected
chemical shift range. However, for compound 1-Me we
encountered severe difficulties with the fit procedure. Although
the fits converged and small standard deviations for the singlet
triplet energy gaps were observed, we had to note that the
derived δdia parameters converged not to their anticipated values.
This is best explained for the chemical shift for the Cp* methyl
protons, which converged at δdia ) -20(( 2) ppm. Similar large
deviations from their expected diamagnetic values (∆δdia ) 7-9
ppm) were also observed for the shifts of the residual aromatic
and methyl protons of the bipyridyl ring. This clearly indicated
an intrinsic problem with this fit procedure, respectively, with
the description of the magnetic properties of these molecules.
This was further substantiated by the aforementioned solid-state
magnetic susceptibility data.
In sharp contrast to our previous findings and interpretation
of the solution 1H NMR data in dichloromethane, the magnetic
susceptibility data of complex 1-Me clearly evidenced a pure
triplet state with zero-field splitting in the solid state (Figure
10). Because the bulk compound used in these measurements
was essentially single crystalline material, it was deemed
unlikely that its structure was different from the one observed
by single-crystal X-ray diffraction. Further evidence for the
distinct difference between the solution and solid-state magnetic
susceptibility data was provided through magnetic susceptibility
measurements in dichloromethane solution using the Evans
method.20,21 This is best illustrated by Figure 10, which includes
both the solution and solid-state data for complex 1-Me. The
values of the temperature-dependent effective moment µeff vs
T data in solution were not only significantly reduced, but also
display a different temperature dependence. Obviously, these
evidently contradicting results required a different rationale or,
respectively, a refinement of the proposed singlet-triplet spin
state equilibrium. Because we had no evidence for the presence
of structural isomers in solution (vide supra) or further chemical
equilibria, solVation effects, not present in the solid state, were
accounted for the distinct difference. This was also hinted
through the strong temperature dependence of the dielectric
constant of the dichloromethane solvent used in these measure-
ments, which almost doubles at -90 °C compared to its value
at RT.38
To address this point, a solution study in a significantly less
polar solvent than CH2Cl2 was sought. For these measurements,
BArF
we resorted to
1-Me, which displays a significantly better
solubility in low polar solvents owing to its partially fluorinated
anion. This allowed to record vt 1H NMR spectra in a 1:2 (v/v)
mixture of CD2Cl2/toluene-d8. As shown in Figure 7, despite
some small curvature, an essentially linear dependence of the
BArF
1H NMR shifts of
1-Me on the reciprocal temperature was
observed. In contrast to neat CD2Cl2 solvent, a Curie behaVior
F
with a triplet ground state could be therefore ascribed to BAr 1-
This prompted us to examine the Van-Vleck equation used
in the form by Rothwell et al. in more detail (eq 3). The latter
equation is clearly related to the more general form shown in
eq 4. However, rather than using the entropy ∆S as a fourth
fittable parameter, it is kept at a fixed value of ∆S ) R ln3 )
9.1 J mol-1 K-1 in eq 3. This value thus just equals the
electronic term, ∆Sel, for the total entropic change of a singlet-
triplet spin equilibrium with contributions from spin degeneracy.
Additional terms of ∆S, e.g., vibronic contributions, ∆Svib, and
from solvation (vide infra) are thus ignored. For the free energy
change ∆GSfT of the singlet-triplet spin equilibrium in solution,
∆GSfT can be written as
Me in this substantially less polar solvent mixture. Hence, this
experiment provided strong support for the view that solvation
strongly participated in the energetics of the two spin states.
As will be shown below in the theoretical section, it is deemed
that substantially better solvation of the S ) 0 state is the
prevailing mechanism.
In conclusion, items (a) - (d) pointed strongly toward a
singlet-triplet spin state crossover with thermodynamically
favorable singlet ground states for compounds 1-Me and
1-NMe2 in CD2Cl2 solution. For complex 1-H on the other hand,
a triplet ground state can be ascribed to this complex in both
dichloromethane solution and the solid state.
∆GSfT ) ∆Ggasphase, SfT + ∆Gsolv, SfT
(5)
For complexes 1-Me and 1-NMe2, the quantitative determi-
nation of the singlet triplet energy gap in solution was therefore
deemed possible and was attempted through fits of the vt H
where ∆Ggasphase and ∆Gsolv are the free energy changes in the
1
NMR data to the Van-Vleck equation. As has been noticed in
detail elsewhere,14 such a fitting procedure is only valid, if the
(41) Fettinger, J. C.; Keogh, D. W.; Kraatz, H.-B.; Poli, R. Organometallics
1996, 15, 5489.
9
7672 J. AM. CHEM. SOC. VOL. 125, NO. 25, 2003

Cationic Tungsten(IV) Half Sandwich Compounds
A R T I C L E S
gasphase and the free energy change of solvation of the singlet
and triplet state. The latter can be estimated using the solvation
model derived by Kirkwood et al., in which the solvent is
described as a dielectric continuum with dielectric constant ꢀ.42
Including contributions from both monopoles (ions) and dipoles
eq 6 is obtained. For a successful application and the discussion
of the model the reader is referred to an excellent publication
by T. J. Meyer et al., in which this model was used to describe
the solvent dependence of metal-to-ligand-charge-transfer (MLCT)
transitions43
the groups of Rothwell and Poli.9,41 The situation is completely
changed, however, if dꢀ(Τ)/dT for a solvent is large, which will
lead to a significant temperature dependence of ∆Gsolv. From
the Debye law47 it can be deduced that ꢀ is proportional to 1/T,
i.e., ꢀ(T) ) C′′T-1 + C′′′, where C′′ and C′′′ are constants.48
This is best illustrated with the data for dichloromethane using
the ꢀ(T) values determined by Morgan et al. already in the 1930s
(C′′ ) 3372.6 K, C′′′ ) -2.48).38 Most importantly, the
dielectric constant of dichloromethane displays a significant
temperature gradient with a sharp increase upon lowering of
the temperature; i.e., almost doubling of the dielectric constant
is observed in going from 25 °C (ꢀ ) 8.9) to - 90 °C (ꢀ )
16.8).38
q3
1
µ2 1 - ꢀ
∆Gsolv(T) )
‚
- 1 +
‚
(6)
(
)
(
)
R3
2R ꢀ
2‚ꢀ + 1
Using ꢀ(T) ) C′′T-1 + C′′′ eq 8 can be written as
The first term in eq 6 is the familiar Born model for the solvation
of charged species,44 which uses a point charge q at the center
of the sphere with radius R, whereas the second equals the
Onsager dipole solvation energy term for a compound with
dipole moment µ.45
(µT(µT - µS)cos φ)
C
∆Gsolv,SfT(T) )
‚
+ C′
RS,T
C′′T-1 + C′′′
3
(
)
(9a)
For the contribution of the solvation energy term to the
energetics of the singlet-triplet spin equilibrium, the change
of free enthalpy of solvation ∆Gsolv,SfT ) ∆Gsolv,T - ∆Gsolv,S
has to be considered. As will be shown below in the theoretical
section, small geometrical changes are observed between the
singlet and triplet states. This view is supported through the
essentially identical volumes of the cationic complexes in 1-Me
(S ) 0, V ) 465.0 S ) 1, V ) 466.7 [Å3] based on DFT
calculations), which evidenced that the radii of the S ) 0 and
S ) 1 ions are approximately equal, i.e., RS)1 ≈ RS)0. Because
the charge q is unchanged in the singlet and triplet states, the
first term of eq 6 therefore cancels out in the evaluation of
∆Gsolv,SfT leaving alone the dipolar term (eq 7)
(µT(µT - µS)cos φ)
∆Gsolv,SfT(T) ≈
‚(AT + B) (9b)
3
RS,T
where A and B are constants.
Using the dielectric constants of dichloromethane at 8.9 and
16.8 at 298 and 193 K and with the assumption that the dipole
moments and radii of the triplet and singlet state µT and µS and
R
S,T are temperature independent, an increase of 7% is estimated
for the absolute value of |gDGsolv,SfT| upon lowering the
temperature. Whether this interaction becomes more un/favor-
able depends on the relative size the dipole moments of the S
) 0 and S ) 1 solutes.
Furthermore, inspection of eq 9b immediately reveals that
there is a temperature dependent and an independent term. Using
the assumption that µ and R are temperature independent and
grouping the terms to the apparent enthalpy and entropy of
(µT(µT - µS)cos φ)
1 - ꢀ(T)
2‚ꢀ(T) + 1
∆Gsolv,SfT(T) )
‚
(7)
(
)
3
RS,T
solvation ∆Hsolv,app ) B/R3 µT(µT-µS)‚cos(φ) and ∆Ssolv,app
)
-
Because we want to focus on the temperature dependence of
the free enthalpy of solvation with respect to the dielectric
constant, the latter is denoted as temperature dependent, i.e.,
ꢀ(T). The dependence of ∆Gsolv,SfT(T) on ꢀ(T) seems to be rather
complicated at first glance. Fortunately, the second term can
be simplified for solvents displaying a somewhat higher polarity.
As has been pointed previously,43,46 this is based on the fact
that the (1 - ꢀ(T))/(2ꢀ(T) + 1) term becomes essentially linearly
dependent on 1/ꢀ(T) for solvents with ꢀ > 5. For dichlo-
romethane, which fulfills this criterion, eq 7 can be therefore
written as (eq 8)
-A/R3 µT(µT - µS) ‚ cos(φ) leads to ∆Gsolv,SfT ) ∆Hsolv,app
T‚∆Ssolv,app
.
Using, ∆Ggasphase ) ∆Hgasphase - T∆Sgasphase, the free enthalpy
of the spin transition process ∆GSfT in solution can be written
as (eqs 10-11)
∆GSfT ) (∆Hgasphase - ∆Hsolv,app) -
T(∆Sgasphase + ∆Ssolv,app) (10)
∆GSfT ) ∆Happarent - T∆Sapparent
(11)
On the basis of eq 11 a four-parameter fit of eq 5 with δdia,i, Ai,
∆Happ and ∆Sapp as free variables was therefore deemed possible.
It should be noted, however, that ∆Happ and ∆Sapp cannot be
interpreted in terms of a pure (gasphase or solid state) electronic
singlet-triplet energy gap, because a contribution from solvation
of unknown size is included. Nevertheless, the derived values
allow to estimate equilibrium constants for the singlet-triplet
spin equilibrium from the Van’t-Hoff equation.
(µT(µT - µS)cos φ)
C
ꢀ(T)
∆Gsolv,SfT(T) )
‚
+ C′
(8)
3
(
)
RS,T
For a solvent with a negligible (small) temperature gradient of
its dielectric constant, i.e., dꢀ(Τ)/dT ≈ 0, ∆Gsolv(T) is essentially
temperature independent, i.e., ∆Gsolv,SfT (T) ) constant. It
should be noted that this criterion is essentially met for toluene,
which has been used in the aforementioned measurements by
(47) Atkins, P. W., de Paula, J. Physical Chemistry; 7th ed.; Oxford University
Press: New York, 2002.
(48) Alternatively, a numerical expression of the (1 - ꢀ(T))/(2 ꢀ(T) - 1) term
of dichloromethane can be obtained from a linear regression of the (1 -
ꢀ(T))/(2 ꢀ(T) - 1) vs T data. For dichloromethane, this gave: (1 - ꢀ(T))/
(2 ꢀ(T) - 1) ) - 0.511(1) + 3.06(6) 10-4 [K - 1] T with R2) 0.997.
(42) Kirkwood, J. G. J. Chem. Phys. 1934, 2, 351.
(43) Kober, E. M.; Sullivan, B. P.; Meyer, T. J. Inorg. Chem. 1984, 23, 2098.
(44) Born, M. Z. Phys. 1920, 1, 45.
(45) Onsager, L. J. Am. Chem. Soc. 1936, 58, 1486.
(46) Reichardt, C. Angew. Chem. 1965, 77, 30.
9
J. AM. CHEM. SOC. VOL. 125, NO. 25, 2003 7673

A R T I C L E S
Cremer and Burger
The results from fits to eq 4 for complexes 1-NMe2 and 1-Me
presented for all chemically inequivalent protons in Figures 8
and 9 clearly established the necessity to include an explicit
entropic term as fourth fittable parameter. This can be im-
mediately seen by inspection of the derived entropies ∆Sapp for
the spin-state equilibria (Table 4). The derived values for ∆Sapp
are in the order of 30 J mol-1 K-1 for both 1-Me (29 eu) and
1-NMe2 (37 eu) and are thus significantly larger than ∆Sel ) R
ln 3 ≈ 9.1 J mol-1 K-1. The most prominent difference between
the magnetic properties of the two complexes manifested itself
through the fitted values for ∆Happ. In accordance with the
experimental and theoretical results, the dimethylamino substi-
tuted complex 1-NMe2 displayed a substantially larger enthalpy
for the spin equilibrium in comparison with the methyl
substituted complex 1-Me. The origin for this difference will
be traced in the next section.
DFT Calculations. Further support for this view was sought
from DFT calculations of the relative energies of the singlet
and triplet states. However, an inherent problem of this
procedure is that excited states are not correctly described by
single Slater determinant methods. MCSCF/MRSCF DFT
methods, on the other hand, which could appropriately deal with
these systems are (currently) not affordable to deal with large
systems.49 It should be noted, however, that despite this inherent
problem, DFT calculations were carried out in a number of cases
to estimate the relative energies of different spin states. In the
cases studied, they were found to produce rather reasonable
results.50-52 An additional inherent problem in the calculation
of electronic/magnetic properties of 5d transition metals is the
appropriate treatment of relativistic effects. Although scalar
relativistic corrections were included through the usage of a
relativistic Stuttgart-Dresden pseudopotential, spin-orbit (s-
o) coupling was not accounted for in our calculations.53
However, since the magnetic moments observed in our systems
are close to the expected spin-only values it is deemed that the
s-o contributions are negligibly small, thus validating our
theoretical approach.
The calculations were carried out using the unrestricted
formalism (UDFT) for the triplet and both restricted and
unrestricted for the S ) 0 states. With the aforementioned
shortcomings of the applied DFT methodology kept in mind,
the results of these calculations came nevertheless as a surprise.
Although the calculated energy gaps between the singlet and
triplet states in the gasphase were appreciably small and were
hence in accordance with the experimental data, the ground
states of complexes 1-R were predicted to have thermodynami-
cally faVorable (12-21 kJ/mol) triplet rather than singlet states
(Table 5). Prior to the measurements of the magnetic properties
of 1-Me in the solid state and further experimental data in
solution, we accounted for this apparent contradiction between
theory and experiment for 1-Me and 1-NMe2 by the inadequacy/
unreliable accuracy of the DFT method.
calculated to be 67 and 92 kJ/mol uphill from the square-
pyramidal complex 1-Me (S ) 0). Because these values are far
beyond the typical errors of DFT calculations, this clearly
evidenced that these isomers do not play a role in the
magnetochemistry of the tungsten(IV) system. In addition, we
calculated the energy of the conformational isomer 1-Merota
,
in which the Cp* ligand was rotated by 180° about the tungsten-
centroid vector with respect to the geometry-optimized structure
of 1-Me. In accordance with our expectation, the total energy
of 1-Merota was found to be just 2 kJ/mol above 1-Me.
Solvation Effects. Dichloromethane coordination to electro-
philic cationic metal centers in transition metal complexes is
well-established.54 On the basis ofthe observation of a weak
paramagnetism and weak W-Cl bonds in the corresponding
neutral trichloride complex of 1-Me, Cp*W(Me2-bipy)Cl3;15
however, we did not expect that coordination of the CH2Cl2
chlorine lone pair could have lead to a stabilization of the singlet
state in 1-Me. Support for this was indeed provided through
the failure to locate an energy minimum for such an η1 or η2
dichloromethane adduct, i.e., Cp*(Me2-bpy)Cl2W(ClCH2Cl)+ in
the DFT calculation.
Therefore, we attributed the influence of the solvent to mere
(undirected) solvation effects, i.e., a dielectric continuum. The
influence of solvation on the relative energies of the spin S )
0 and S ) 1 states in complexes 1-Me was studied by DFT
calculation using two different solvation models. We applied
the COSMO dielectric continuum model of Klamt et al.32
implemented in DMOL33 as well as the self-consistent reaction
field (SCFRM) method available in Jaguar.36 In the latter model,
the solvent was modeled at two different temperatures, i.e., at
180 and 298 K using the experimentally determined dielectric
constants and densities. The derived relative energies of the two
spin states for the geometry optimized CH2Cl2 solvated struc-
tures of complex 1-Me compiled in Table 6 are in full
accordance with the experimental findings. Although the triplet
state is just slightly energetically favored over the S ) 0 state
in the SCFRM model, the DMOL/COSMO calculation predicts
the singlet state to be the ground state of complex 1-Me in
CH2Cl2 at RT (Table 6).
MO Analysis. Previously, CpML4 systems have been studied
by several research groups at different levels of theory and were
used to explain the observed preference for the pseudo sq.-py.
coordination geometry and geometrical distortions found in a
number of these complexes.40,55,56 It deserves special mentioning
that aside from the research of Poli et al.,41,55 these investigations
mostly centered on d4 transition metal systems and bearing at
least one ligand with a sizable π-acceptor propensity, i.e.,
carbonyl groups and, to a lesser extent phosphine donors. This
contrasts the d2-configured complexes 1 described herein, which
incorporate excellent σ- and π-donor ligands. However, due to
the absence of good acceptor groups in compounds 1, the
bipyridyl and cyclopentadienyl ligands have to “fill this gap”
in the quite electron rich metal compounds - a fact which is
clearly reflected in their structural and magnetic properties (vide
infra).
In addition, we performed DFT calculations for the trigonal-
bipyramidal isomers 1Meiso, 1-Metbp (Figure 11). The latter were
At the outset of the discussion it should be noticed that the
frontier orbitals of the S ) 0 state, the HOMO and LUMO of
(49) Young, D. Computational Chemistry: A Practical Guide for Applying
Techniques to Real World Problems; Wiley: New York, 2001.
(50) Smith, K. M.; Poli, R.; Legzdins, P. Chem. Commun. 1998, 1903.
(51) Rodriguez, J. H.; Wheeler, D. E.; McCusker, J. K. J. Am. Chem. Soc. 1998,
120, 12 051.
(54) For an example see: Arndtsen, B. A.; Bergman, R. G. Science 1995, 270,
1970.
(55) Poli, R. Organometallics 1990, 9, 1892.
(56) Lin, Z.; Hall, M. B. Organometallics 1993, 12, 19-23.
(52) Legzdins, P.; McNeil, W. S.; Smith, K. M.; Poli, R. Organometallics 1998,
17, 615.
(53) Figgis, B. N.; Lewis, J. Prog. Inorg. Chem. 1964, 6, 37.
9
7674 J. AM. CHEM. SOC. VOL. 125, NO. 25, 2003

Cationic Tungsten(IV) Half Sandwich Compounds
A R T I C L E S
Figure 14. X-ray crystallographically determined and DFT calculated geometries (S ) 0 and 1 states) of complex 1-NMe2 (top trace) and complex 2-Me
(lower trace).
complexes 1 are also applicable for the triplet state (SHOMO
and SSHOMO) and differ just by their electron occupation
numbers. For comparison with the MO scheme of the C4V
symmetrical sq.-py. system presented in Figure 13, this analysis
will be pursued with the coordinate system used there, i.e., the
z-axis is pointing toward the Cp* ligand.
η2/η4, respectively, η1/η3-binding of the Cp ligand to maximize
The inspection of the LUMO (S ) 0) revealed contributions
from both the metal center as well as the chloro and bipyridyl
ligands (Figure 12). The metal character is constituted through
the dz2 orbital with some admixture of the dxy orbital. In context
with the further analysis presented below, we would like to draw
the attention to the dπ-pπ* back-donation to the bipyridyl π*-
acceptor orbital, which partly compensates for the antibonding
combination of the chlorine-tungsten pπ-dπ donor interaction.
It is noteworthy that in a previous analysis of d2-configured
Cp*ML4 complexes by Poli et al., it was recognized that the
strength of the πC-dπmetal-acceptor interaction strongly influ-
ences the HOMO-LUMO gap, which in turn is dependent on
the Cpcentroid-W-Cl angle.41
The metal character in the HOMO is easily recognized as
the dxy orbital (Figure 12). It displays an antibonding (π-donor)
interaction between the chloro ligands and the tungsten center.
In analogy to the LUMO, the latter is partly compensated by
strong δ-bonding between the filled dxy orbital and a Cp δ*-
acceptor orbital (e2′′ in local D5h symmetry), which is schemati-
cally shown below.
the overlap between the dxy orbital and the Cp acceptor orbital
(e2′′* in local D5h symmetry). In the singlet state, this leads to
a pronounced asymmetry for the binding of the Cp* ligand and
is clearly reminiscent of an η3-binding mode. This is best
illustrated through the W-C distances observed in the optimized
geometry of the S ) 0 state of complexes 1. This is also nicely
reflected in the X-ray crystal structure of complex 1-NMe2,
which according to the solid and in solution susceptibility
measurements has a strong thermodynamic preference for the
S ) 0 state. This is further illustrated by the experimementally
and DFT derived geometrical parameters for complex 2-Me,
which are also shown in Figure 14.
On the basis of the analysis of the frontier orbitals, a stronger
stabilization of the LUMO is expected for bipyridyl ligands with
a better π-accepting propensity, e.g., such as the unsubstituted
ligand 2,2′-bipyridyl in complex 1-H. Since the HOMO is not
affected through substitution of the bipyridyl ligand, a smaller
HOMO/LUMO gap and therefore a higher preference for the
triplet state would be expected for this ligand metal combination.
For weaker π-acceptors on the other hand, such as the dimethyl-
amino bipyridyl ligand in complex 1-NMe2, a smaller stabiliza-
tion of the LUMO (SHOMO for S ) 1) thus leading to a
thermodynamically more favorable singlet state would be
anticipated. This rationale provides an excellent explanation for
the experimentally observed reduced paramagnetism in 1-NMe2
and, also, the triplet state found for complex 1-H bearing the
unsubstituted 2,2′-bipyridyl ligand.
This type of interaction was first discussed by Hoffmann et
al.40 in the aforementioned EHT study on CpML4 complexes
and leads to a significant slippage of the metal center toward
9
J. AM. CHEM. SOC. VOL. 125, NO. 25, 2003 7675

A R T I C L E S
Cremer and Burger
X-ray Crystal Structure Analyses: General Remarks. Suitable
single crystals were mounted on glass fibers in polyisobutylene oil
(Aldrich, 38,896-6), transferred on the goniometerhead to the diffrac-
tometer and the crystal cooled to -90 °C in a N2-cryostream. The data
sets were collected with graphite monochromated Mo-KR-radiation
(0.70713 Å) on a Stoe IPDS image plate diffractometer. Intensities
were corrected for Lorentz and polarization effects. Absorption cor-
rections were performed numerically with the faces and crystal
dimensions determined using the STOE Faceit-Video CCD camera
system. The structures were solved using direct methods with the
SHELXS-93 program package.61 The refinements were carried out with
In addition, it allows to explain the structural differences
observed in X-ray crystal stuctures of complexes 1-Me and
1-NMe2. In the S ) 0 state, back-donation proceeds exclusively
through δ-bonding, leading to a pronounced η3-binding of the
Cp* ligand in order to maximize overlap between the dxy and
the e2′′* Cp based orbital. In the triplet state, the latter interaction
is significantly reduced due to additional dπ-pπ* back-donation,
which consequently leads to more symmetrical (η5) bonding of
the Cp* ligand. This trend is very well reflected in the
calculations (see Figure 14) and also experimentally through
the X-ray crystal structure of complex 1-Me.
2
SHELXL-9762 using all unique Fo . All non-hydrogen atoms were
treated anisotropically with the positions of the hydrogen atoms
calculated in idealized positions (C-H bonds fixed at 0.96 Å) and
refined as riding model. The details of the data collections and
refinements including R-values are summarized in Table 1.
Conclusion
The combination of experimental and theoretical methods,
allowed to unambiguously unravel the magnetic properties of
the novel cationic tungsten(IV) complexes both in solid state
and in solution. The presented methodology is sought to provide
strong guidelines for future studies in the field of solution and
solid-state magnetochemistry.
Synthesis of Cp*W(bpy)Cl2, 2-H. 280 mg (0.359 mmol)
(Cp*WCl2)2 and 112 mg (0.718 mmol) bpy were dissolved in 20 mL
THF in a high vacuum Young Teflon tap sealed Schlenk tube and
degassed by three freeze pump thaw cycles. After heating in a vacuum
for 2 d at 120 °C (behind a safety shield) a precipitate formed. Upon
cooling to RT, the supernatant solvent was decanted off and the
remaining solid washed 3× with toluene, followed by pentanes, and
finally dried in high vacuum. The microcrystalline black, analytically
pure product is insoluble in THF and slightly soluble in dichlo-
Experimental Section
Reactions were carried out under a dinitrogen atmosphere using
glovebox and Schlenk techniques. The halogenated solvents were
thoroughly dried over P4O10 and saturated with N2. The other solvents
were distilled under nitrogen from violet sodium benzophenone ketyl
and stored under nitrogen. In addition, residual traces of water in the
solvents and on the glassware, used in the synthesis and spectroscopic
analysis of the highly water sensitive cationic complexes, were removed
by either filtration over alumina (for CH2Cl2, Woelm, act. Super-I,
neutral, previously dried at 10-5 Torr at 350 °C), or stirred over
sodium-potassium alloy followed by filtration (ether, THF). The
deuterated solvents were dried over sodium or P4O10 and transferred
in vacuo using high vacuum techniques. 2,2′-bipyridyl was purchased
from Fluka and used as received. Na(B(C6H3(CF3)2)4) and [Cp2Fe]-
[BPh4] were prepared according to published syntheses.57,58 The known
dimer [Cp*WCl2]2 was obtained from Cp*WCl459 using 1.5-fold excess
of Zn dust in thf in analogy to the synthesis of [(C5H4iPr)WCl2]2 reported
by Green et al.60 Complexes 1-Me and 1-NMe2 were prepared according
to a route previously reported by us.15 1H-, 19F-, and 31P NMR- spectra
were recorded on Varian Gemini 200 and 300 spectrometers. Chemical
shifts are given in ppm and referenced to the residual 1H-solvent shift
of a deuterated external sample or H3PO4 (85%, 31P) or trifluorotoluene
(19F). Raman spectra were recorded on a Renishaw Ramascope
spectrometer using HeNe (632.8 nm) excitation; UV/Vis-spectra were
taken on a Cary 1E UV-Visible spectrometer. Mass spectra were
measured with a Finnigan/MAT 8320 (MS) spectrometer and the peak
assignments confirmed by a simulation of the isotope patterns.
Molecular conductivities were determined in dichloromethane with an
Amel-160 conductometer (glass cell, Pt, K ) 1.0). Magnetic moments
at RT were recorded with a Johnson-Matthey laboratory susceptibility
balance and are corrected for diamagnetic contributions. Variable
temperature solid state magnetic susceptibility measurements were
performed with a Foner magnetometer at the University of Mainz by
Dr. F. Tuczek. CHN-analyses were carried out with a LECO CHNS-
932 elemental analyzer in our institute. The stoichiometric incorporation
of solvent molecules into some of the analytically pure compounds
1
romethane. Yield: 343 mg, (0.628 mmol), 88%. H NMR (CD2Cl2,
298K), δ [ppm]: 24.5 (br, ω1/2 ≈ 70 Hz); 8.2 (br, ω1/2 ≈ 130 Hz); 4.5
(br, ω1/2 ≈ 60 Hz) -23 (vbr, ω1/2 ≈ 500 Hz); -34 (vbr, ω1/2 ≈ 450
Hz). MS (EI) m/z 546 (M+). Magnetic moment (298 K): µeff ) 1.67
µB (corrected). Elemental analysis (C20H23Cl2N2W): Calcd. C, 43.98;
H, 4.24; N, 5.13. Found. C, 43.75; H, 4.16; N, 5.13.
Synthesis of Cp*W(Me2bpy)Cl2, 2-Me. 315 mg (0.404 mmol)
(Cp*WCl2)2 and 149 mg (0.810 mmol) Me2bpy were dissolved in 40
mL THF and transferred to a high vacuum Young Teflon tap sealed
Schlenk. The mixture was degassed, heated for 4 d to 120 °C and then
allowed to cool to RT. The supernatant solution was decanted off from
the precipitate, which formed during the reaction. The solid was washed
with toluene and pentanes and finally dried in high vacuum. The black
microcrystalline material is moderately soluble in THF and dissolves
well in dichloromethane. Yield: 390 mg, 0.680 mmol, 84%. 1H NMR
(CD2Cl2, 298 K), δ [ppm]: 84 (vbr, ω1/2 ≈ 1300 Hz), 22.6 (br, ω1/2
≈
80 Hz); 3.8 (br, ω1/2 ≈ 60 Hz); 2.3 (br, ω1/2 ≈ 80 Hz) -45.5 (vbr, ω1/2
≈ 800 Hz). MS (EI): m/z 574 (M+). Magnetic moment (298 K): µeff
) 1.65 µΒ. Elemental analysis (C22H27Cl2N2W): Calcd. C, 46.02; H,
4.74; N, 4.88. Found. C, 45.95; H, 4.82; N, 4.86. Single crystals suitable
for X-ray diffraction were obtained by slow diffusion of ether into a
solution of 2-Me in 1,2-dichloroethane at RT.
F
BAr
Synthesis of [Cp*W(bpy)Cl2][B(C6H3(CF3)2)4],
1-H. To a
suspension of 216 mg (0.396 mmol) Cp*W(bpy)Cl2, 2-H, in 20 mL
dichloromethane, 201 mg (0.396 mmol) [Cp2Fe][BPh4] was added in
small portions under vigorous stirring upon which a color change from
black to dark green and the formation of a green precipitate was
observed. The volume of the solvent was concentrated to 5 mL in high
vacuum and the supernatant solvent finally decanted off. The solid green
residue was washed with toluene and pentanes and then dried in high
-
vacuum giving 340 mg of the crude BPh4 salt, which was sparingly
soluble in dichloromethane. The suspension of this material in 20 mL
CH2Cl2 was reacted with 350 mg (0.396 mmol) Na(B(C6H3(CF3)2)4),
stirred for 30 min at RT, then filtered through a sintered glassfrit. The
1
F
was confirmed by either X-ray analysis or H NMR integration.
BAr
product,
1-H, was precipitated by slow addition of pentane to the
filtrate, collected by filtration, washed with pentane and finally dried
F
BAr
(57) Brookhart, M.; Grant, A.; Volpe, F. J. Organometallics. 1992, 11, 3920.
(58) Brauer, G. Handbuch der Pra¨paratiVen Anorganischen Chemie; Ferdinand
Enke Verlag: Stuttgart, 1981; Vol. 3.
in high vacuum giving
1-H as an analytically pure green solid.
(59) Murray, R. C.; Blum, L.; Liu, A. H.; Schrock, R. R. Organometallics 1985,
4, 953.
(61) Sheldrick, G. M. SHELXS-97; UniVersity of Go¨ttingen: Go¨ttingen, Germany
1997.
(62) Sheldrick, G. M. SHELXL-97, program for crystal structure solution nd
refinement; UniVersita¨t Go¨ttingen, Germany 1993 and 1997.
(60) Green, M. L. H.; Hubert, J. D.; Mountford, P. J. Chem. Soc., Dalton Trans.
1990, 3793.
9
7676 J. AM. CHEM. SOC. VOL. 125, NO. 25, 2003

Cationic Tungsten(IV) Half Sandwich Compounds
A R T I C L E S
1
Overall yield: 66% (370 mg, 0.263 mmol). H NMR (CD2Cl2, 298
1,2-dichloroethane as reference and a mixture of 1,2-dichloroethane
and CD2Cl2 in a sealed capillary in the same NMR-tube as an external
standard. The mass susceptibility was calculated from the chemical
shift differences ∆f of the 1,2-dichloroethane reference in solution and
in the sample using the following equation21
K), δ [ppm]: 47.3 (br, ω1/2 ≈ 60 Hz); 37.8 (br, ω1/2 ≈ 45 Hz); 19.6
(br, ω1/2
≈
40 Hz); 7.72 (br, [B(C6H3(CF3)2)4]-); 7.56 (br,
B(C6H3(CF3)2)4]-); -106.1 (br, ω1/2 ≈ 180 Hz); -109.5 (br, ω1/2
≈
200 Hz). 19F NMR (CD2Cl2, 298 K),
δ [ppm]: -63.8 (s,
[B(C6H3(CF3)2)4]-). MS (FAB+, CH2Cl2): m/z 545 (M+). Magnetic
moment (298 K): µeff ) 2.41 µB. Elemental analysis (C52H35-
BCl2F24N2W): Calcd. C, 44.32; H, 2.50; N, 1.99. Found. C, 43.98; H,
2.55; N, 1.94.
ø0(d0 - dS)
3∆f
4πfm
øg )
+ ø0 +
m
Corrections for temperature induced change of the solvent density were
included using the data for CH2Cl2 given in reference.38 Diamagnetic
corrections of the molar susceptibilities øm were applied using either
the independently determined susceptibilities of the free ligands or with
the aid of the Pascal constants. The effective magnetic moments, µeff,
F
BAr
Synthesis of [Cp*W(Me2bpy)Cl2][B(C6H3(CF3)2)4],
1-Me. A
solution of 90 mg (0.10 mmol) NaB(C6H3(CF3)2)4 in 5 mL THF was
slowly added to a suspension of 61 mg (0.10 mmol) Cp*W(Me2bpy)-
Cl3 in 5 mL THF under stirring upon which a brown solution was
obtained. The solvent was removed in high vacuum and the green solid
residue dried for several hours in high vacuum. The product was
extracted into 10 mL CH2Cl2 and filtered through a sintered glassfrit.
Upon slow addition of pentane to the dark green solution a slightly
oily precipitate was obtained, which was isolated by decanting the
were calculated from µeff ) 2.828(ømT)1/2
.
DFT-Calculations. The applied methodology is described in the
text. The restricted and unrestricted DFT calculations were performed
with the Turbomole program suite and using the RI-DFT (BP-86
functional) and B3LYP implementation for the geometry optimizations
and the calculation of the second derivatives. Essentially identical
energy differences were obtained with the BP-86 functional using the
pure DFT rather than the RI-DFT method. The parallel version of the
Turbomole program package was used on our 24 CPU Linux cluster
in all cases. For the DFT calculations including solvation the DMOL
rel 3.5 (COSMO) and Jaguar rel. 4.1 (SCFRM) program packages were
used.
supernatant solvents off. The solid was dried in high vacuum to give
F
BAr
1
91 mg, (0.063 mmol) 63%
1-Me as a dark green solid. H NMR
(CD2Cl2, 293 Κ, δ [ppm]): 113 (br s); 45 (br s); 30 (br s); 18.3 (br s);
7.72 (m, 8 H, [B(C6H3(CF3)2)4]-); 7.56 (m, 4 H, [B(C6H3(CF3)2)4]-);
-92 (br s).
VT H NMR Spectroscopy. The temperature calibration of the vt
unit of the NMR spectrometer was performed using a calibration curve
for the chemical shift differences of the methyl and OH protons of a
1
1
standard methanol-H4 sample. Temperature equilibration in the vt H
Acknowledgment. We are indebted to Prof. Heinz Berke for
his generous support. We would like to thank Professor Rinaldo
Poli for pointing out the Van-Vleck fitting method for the
determination of singlet triplet energy gaps. Comments and
suggestions by Professor Ian Rothwell and Dr. Gerrit Luinstra
on this work are appreciated. We would particularly like to thank
Professor Felix Tuczek for the solid state magnetic susceptibility
measurements and very stimulating discussions. Funding of this
project by the Swiss National Science Foundation is gratefully
acknowledged.
NMR experiments was allowed for at least 20 min. Measurements were
performed in NMR tubes sealed with a Young Teflon tap with the
samples being prepared in a glovebox under N2 atmosphere. (<1 ppm
O2, H2O) and finally degassed.
1
Fit of the VT H NMR Data to the Van-Vleck Equation. The
4-parameter fits using eq 5 were performed with the Origin 5.0
(Microcal) program package using a Levenberg-Marquardt algorithm
(Figures 10 and 11). Starting with different initial parameters for the
hyperfine coupling constant A, δdia, ∆E (∆Happ and ∆Sapp), the fits
converged also to the same values.
Determination of the Magnetic Moments: (a) Solid State. Room-
temperature molar magnetic susceptibilities øm in the solid state were
determined with a Johnson-Matthey laboratory magnetic balance in
quartz tubes. The crystalline samples were crushed in an achat mortar,
transferred to the tube and then sealed with a thick layer (3 mm) of
Parafilm (preparations in glovebox). The given values are averaged
values of several measurements, which showed only a small variance.63
(b) Solution. The magnetic susceptibility of complex 1-Me was
determined with the Evans-method in CD2Cl2 using a small amount of
Supporting Information Available: Coordinates of DFT
optimized structures and crystallographic information (PDF).
This material is available free of charge via the Internet at
http://pubs.acs.org.
JA028313C
(63) McMillan, J. A. Electron Paramagnetism; Reinhold Book Corporation:
New York, 1968.
9
J. AM. CHEM. SOC. VOL. 125, NO. 25, 2003 7677