A comparison of the lability of mononuclear octahedral and dinuclear riple-helical complexes of cobalt(II)

Article abstract of DOI:10.1021/ja963693l

The lability of the mononuclear octahedral complex tris(5-methyl-2-(1'-methylbenzimidazol-2-yl)-pyridine)cobalt(II), [Co(2)₃]²⁺, is compared with the dinuclear triple-helical complex tris[bis[1-methyl-2-(5'-methylpyrid-2'yl]benzimida

Full text of DOI:10.1021/ja963693l

2488
J. Am. Chem. Soc. 1997, 119, 2488-2496
A Comparison of the Lability of Mononuclear Octahedral and
Dinuclear Triple-Helical Complexes of Cobalt(II)
Lo1ıc J. Charbonnie`re,1a Alan F. Williams,*,1a Urban Frey,1b Andre´ E. Merbach,1b
Philippe Kamalaprija,1c and Olivier Schaad1d
Contribution from the Section of Chemistry, UniVersity of GeneVa, 30 quai Ernest Ansermet,
CH 1211 GeneVa 4, Switzerland, and Institute of Inorganic and Analytical Chemistry,
UniVersity of Lausanne, BCH, CH 1015 Lausanne, Switzerland
ReceiVed October 23, 1996X
Abstract: The lability of the mononuclear octahedral complex tris(5-methyl-2-(1′-methylbenzimidazol-2-yl)-pyridine)-
cobalt(II), [Co(2)3]2+, is compared with the dinuclear triple-helical complex tris[bis[1-methyl-2-(5′-methylpyrid-2′-
yl]benzimidazol-5-yl]methane]dicobalt(II), [Co2(1a)3]4+. [Co(2)3]2+ undergoes rapid isomerization between mer and
fac forms (k298(mer-fac) ) 1.6 ( 0.2 s-1) in acetonitrile while the racemization of (-)589-[Co2(1a)3]4+ is roughly
105 times slower (k298 ) 1.4 ( 2 10-5 s-1). The pressure dependence of the isomerization of [Co(2)3]2+ suggests
a dissociatively activated process. The racemization of (-)589-[Co2(1a)3]4+ is found to be independent of pH above
pH 4, and is not affected by added cobalt(II) or a change of solvent. Ligand exchange between [Co2(1b)3]4+ (1b )
bis[1-ethyl-2-(5′-methylpyrid-2′-yl)benzimidazol-5-yl]methane) and free 1a may be followed by electrospray mass
spectrocopy and establishes the mechanism of formation of the triple helix to be initial formation of mononuclear
[Co(1a)3]2+ followed by capping by a second cobalt to give [Co2(1a)3]4+. The slow racemization of (-)589-[Co2-
(1a)3]4+ is attributed to the very slow dissociation of a cobalt ion from the triple helix. This inertness is attributed
to the rigidity of the ligand and the tight pitch of the helix.
Introduction
Syntheses of helical complexes (helicates) containing two or
more metal ions have attracted considerable attention in recent
years as examples of self-assembly reactions.2,3 In the general
strategy, multidentate ligands wrap around two or more metal
ions placed along the helical axis to give double4-7 - or triple8-12
-helical structures, depending on the denticity of the ligating
units and the coordination requirements of the metal.13 Much
attention has been paid to ligand design and to the choice of
metal in order to obtain the desired structure, usually in almost
quantitative yield, and the structural principles of the syntheses
are now well established. Thus, if oligobidentate ligands such
as 1 are used, tetrahedral metal ions will give rise to double
helices, while octahedral ions will form triple helices (Figure
1).
Figure 1. Strategy for the synthesis of triple-helical complexes using
two octahedral metal cations.
Less attention has been paid however to the properties of
these self-assembled complexes, and in particular to their
chemical reactivity. In this field, only the electron transfer
properties have been studied to any degree, usually by electro-
chemistry;7,10,14 very little is known of the lability of the
helicates. Clearly, if the complexes are to be used as starting
points for further syntheses (as for example in the synthesis of
a molecular knot from a dinuclear double-helical complex15 ),
they must possess a degree of thermodynamic and kinetic
stability. While some work has appeared on the thermodynamic
stability,13,16 the reactivity of helicates has received less attention.
In the majority of cases the metal ion used for the assembly is
substitutionally labile, and one might therefore reasonably expect
that the resulting complexes would display the lability of their
component metal ions. It was therefore surprising to discover
that the enantiomerically pure triple-helical complex (-)589[Co2-
(1a)3]4+ (Figure 1) containing the labile Co(II) ion racemized
X Abstract published in AdVance ACS Abstracts, February 15, 1997.
(1) (a) Department of Inorganic, Analytical and Applied Chemistry,
University of Geneva; (b) Institute of Inorganic and Analytical Chemistry,
University of Lausanne; (c) Department of Organic Chemistry, University
of Geneva; (d) Department of Biochemistry, University of Geneva.
(2) Lindsey, J. S. New J. Chem. 1991, 15, 153.
(3) Lehn, J.-M. Supramolecular Chemistry; VCH Weinheim, 1995; pp
144-154.
(4) Lehn, J.-M.; Rigault, A.; Siegel, J. S.; Harrowfield, J. MacB.;
Chevrier, B.; Moras, D. Proc. Natl. Acad. Sci. U.S.A., 1987, 84, 2565.
(5) Constable, E. C.; Ward, M. D.; Tocher, D. A. J. Chem. Soc., Dalton
Trans. 1991, 1675.
(6) Constable, E. C. Tetrahedron 1992, 48, 10013.
(7) Potts, K. T.; Keshavarz-K, M.; Tham, F. S.; Abrun˜a, H. D.; Arana,
C. R. Inorg. Chem. 1993, 32, 4422 and following papers.
(8) Scarrow, R. C.; White, D. L.; Raymond, K. N. J. Am. Chem. Soc.
1985, 107, 6540.
(9) Libman, J.; Tor, Y.; Shanzer, A. J. Am. Chem. Soc. 1987, 109, 5880.
(10) Serr, B. R.; Andersen, K. A.; Elliott, C. M.; Andersen, O. P. Inorg.
Chem. 1988, 27 4499.
(11) Williams, A. F; Piguet, C.; Bernardinelli, G. Angew. Chem., Int.
Ed. Engl. 1991, 30, 1490.
(12) Kra¨mer, R.; Lehn, J.-M.; DeCian, A.; Fischer, J. Angew. Chem.,
Int. Ed. Engl. 1993, 32, 703
(13) Piguet, C.; Bernardinelli, G.; Bocquet, B.; Quattropani, A.; Williams,
A. F. J. Am. Chem. Soc. 1992, 114, 7440.
(14) Ferrere, S.; Elliot, C. M. Inorg. Chem. 1995, 34, 5818.
(15) Dietrich-Bucheker, C. O.; Sauvage, J.-P. Angew. Chem., Int. Ed.
Engl. 1990, 29, 1154; Dietrich-Bucheker, C. O.; Sauvage, J.-P.; DeCian,
A.; Fischer, J. J. Chem. Soc., Chem. Commun. 1994, 2231.
(16) Pfeil, A.; Lehn, J.-M. J. Chem. Soc., Chem. Commun. 1992, 838.
S0002-7863(96)03693-1 CCC: $14.00 © 1997 American Chemical Society

Lability of Mononuclear and Dinuclear Co(II) Complexes
J. Am. Chem. Soc., Vol. 119, No. 10, 1997 2489
Scheme 1. Ligand Synthesis
1
Figure 2. Part of the H NMR spectrum of [Co(2)3]2+ showing the
signals corresponding to the methyls bound to benzimidazole in the
fac (f) and mer (m1, m2, m3) forms.
Scheme 2. mer-fac Equilibrium of [Co(2)3]2+ a
a mer and fac isomers are present as racemic mixtures of ∆ and Λ
isomers.
The complexes [Co2(1)3]4+ and [Co(2)3]2+ were prepared by
mixing stoichiometric amounts of ligand in dichloromethane
or acetonitrile solution with cobalt(II) perchlorate hexahydrate
in acetonitrile solution. After the solvent was stripped off under
reduced pressure, the complexes could be crystallized from
acetonitrile solution by slow diffusion of methanol. Enantio-
merically pure (-)589[Co2(1a)3](ClO4)6 was obtained by oxida-
tion of racemic [Co2(1a)3](ClO4)4 with hydrogen peroxide and
perchloric acid in acetonitrile solution in the presence of
ferrocenium tetrafluoroborate as catalyst, followed by resolution
of the cobalt(III) triple helix by crystallization with (+)589-[Sb2-
very slowly (t1/2 ) 13 h) at room temperature.17 This suggested
that self-assembled dinuclear complexes might indeed be quite
inert, but two very recent studies of triple-helical systems using
biscatechol ligands have shown racemization to be rapid (t1/2
) 0.1 s) on the NMR time scale at or around room tempera-
ture.18,19
We have therefore sought to investigate the origin of the
inertness of [Co2(1a)3]4+. Our first step was to study the lability
of the mononuclear cobalt(II) complex [Co(2)3]2+ in which the
coordination sphere of the cobalt is essentially the same as that
found in [Co2(1a)3]4+; this not only establishes the inherent
lability of the cobalt(II) ion in this coordination sphere but also
enables a quantitative comparison with the dinuclear triple helix.
Secondly, we have studied the effect on the racemization of
(-)589[Co2(1a)3]4+ of pH, added cobalt(II), and solvent. Finally
we have used electrospray mass spectroscopy (ESMS) to study
the exchange of ligands in the triple-helical complex. The
results establish clearly the formation pathway of the triple helix
and give insight into the origin of the relative inertness of this
complex.
(C4H2O6)2]2- 21
Ion exhange gave the perchlorate salt. The
.
enantiomerically pure cobalt(II) helix was generated in situ by
reduction of (-)589-[Co2(1a)3](ClO4)6 with sodium dithionite
before kinetic runs.17
Behavior of the Mononuclear Complex [Co(2)3]2+. Spec-
troscopic titration13 of a solution of cobalt(II) perchlorate in
acetonitrile with 2 showed successive formation of [Co(2)]2+
,
[Co(2)2]2+, and [Co(2)3]2+ with macroscopic22 stability constants
log â1 ) 7.8(1), log â2 ) 14.9(2), and log â3 ) 21.2(3). By
virtue of the asymmetric nature of the ligand 2 the trisbidentate
complex [Co(2)3]2+ may exist in two isomeric forms, the facial
(fac) isomer in which the three pyridines occupy one face of
the octahedron and the meridional (mer) complex in which they
occupy a meridian of the octahedron (Scheme 2); both forms
exist as pairs of ∆ and Λ enantiomers. The 1H NMR spectrum
of a solution of [Co(2)3](ClO4)2 in acetonitrile-d3 shows a total
of 36 signals spread out between +108 and -50 ppm as a result
of interaction with the paramagnetic Co(II) ion. Although the
rapid relaxation of the paramagnetic complex obscures proton-
proton couplings, double resonance measurements allowed the
differentiation of signals arising from mer and fac isomers except
for eight signals arising from protons closest to the paramagnetic
center which were too broad.
Results
Synthesis. The ligands 1a, 1b, and 2 were synthesized using
the synthetic strategy developed by Piguet20 in which a pyridine
acid chloride is reacted with an aromatic nitroamine to give a
nitroamide which is then reduced and cyclized in one step with
iron/hydrochloric acid to give the pyridine-benzimidazole
(Scheme 1). This route is cleaner than the direct Philips
coupling of a diamine and a carboxylic acid.13
(17) Charbonnie`re, L. J.; Gilet, M.-F.; Bernauer, K.; Williams, A. F. J.
Chem. Soc., Chem. Commun. 1996, 39.
(18) Kersting, B.; Meyer, M.; Powers, R. E.; Raymond, K. N. J. Am.
Chem. Soc. 1996, 118, 7221.
(19) Albrecht, M.; Kotila, S. Angew. Chem., Int. Ed. Engl. 1996, 35,
1208.
Integration of the signals arising from benzimidazole methyl
groups of the two isomers (Figure 2) allowed the determination
of the equilibrium constant K ) [fac]/[mer] for the mer-fac
(21) Yoshikawa,Y.; Yamasaki, K. Coord. Chem. ReV. 1979, 28, 205.
(22) Martell, A. E.; Motekaitis, R. J. Determination of Stability Constants
2nd ed.; VCH: New York, 1992; Chapter 7.
(20) Piguet, C.; Bocquet, B.; Hopfgartner, G. HelV. Chim. Acta 1994,
77, 931.

2490 J. Am. Chem. Soc., Vol. 119, No. 10, 1997
Charbonnie`re et al.
isomerization as 0.23 ( 0.01 at 298 K. Studies in the
temperature range 240-350 K showed a variation of this value
of only a few percent and led to the thermodynamic parameters
∆H° ) +0.6 ( 0.3 kJ mol-1 and ∆S° ) -10.1 ( 1 J mol-1
K-1. Note that the equilibrium constant is slightly below the
statistical value of 1/3; this may arise from steric interactions
between the three benzimidazoles in the fac isomer.
During the NMR experiments it was observed that irradiation
of a signal of the fac isomer perturbed the intensity of signals
of the mer isomer, implying chemical exchange between
isomers. This suggested that NMR could be used to study the
kinetics of the mer-fac exchange, and that this might be used
to characterize the lability of the Co(II) center in this coordina-
tion sphere. Initial experiments used the method of Forse´n and
Hoffmann23 in which the transfer of polarization between two
sites is followed. However, the experiments were complicated
by the short relaxation times in this paramagnetic system. The
results obtained were broadly in agreement with the rates
expected for a cobalt(II) system, but the precision was low. This
method was therefore abandoned in favor of the NMR line
broadening method in which the four proton signals (one facial,
f, and three meridional, m1, m2, and m3) arising from the methyl
group bound to the benzimidazole nitrogens were followed
(Figure 2).
There are in principle six rate constants and three equilibrium
constants needed to describe the exchange among these four
different sites. However, the mer-fac isomerization requires
the transformations m1 f f, m2 f f, and m3 f f to have the
same rate, described by the rate constant kmf. Similarly the rate
constants for exchange among m1, m2, and m3 are taken to be
equal and described by kmm. In the slow exchange limit, the
transverse relaxation rate of the 1H NMR signal near the
paramagnetic center, 1/T2, can be expressed as the sum of a
term due to the paramagnetic relaxation, 1/T2m, and terms due
to chemical exchanges. In this case this reduces to24
Figure 3. Temperature dependence of 1/T2 from the 1H NMR signals
of [Co(2)3]2+ in acetonitrile solution: (9) signal m1, (b) signal m2,
(4) signal m3 and ([) signal f.
the Supporting Information). A plot of the temperature
dependence of the relaxation rate in the form of ln(1/T2) against
1/T is shown in Figure 3, and the derived kinetic, thermody-
namic, and NMR parameters are given in Tables 1 and 2.
To gain further information on the isomerization we have
studied the effect of pressure. The pressure dependence of the
rate constants was described by the linear eq 6, where the
ln k ) ln k0 - P∆Vq/RT
(6)
volumes of activation, ∆Vq, were assumed to be pressure
independent; k0 is the exchange rate at zero pressure. Equation
7 describes the pressure dependence of the equilibrium constant
ln K ) ln K0 - P∆V°/RT
(7)
(1/T2)mi ) (1/T2m)mi + kmf + kmm (i ) 1, 2, 3)
(1/T2)f ) (1/T2m)f + kfm
(1)
(2)
as a function of the reaction volume ∆V° and the value at zero
pressure K0. The variable pressure measurements were made
at 349.6 K, a temperature at which the paramagnetic contribution
(1/T2m) to the observed relaxation rate (1/T2) was low (<5%).
Therefore, the very small pressure dependence of 1/T2m could
be neglected, and the values of 1/T2m at zero pressure obtained
from the variable temperature measurements were used. The
rate and the equilibrium constants (Table S2 in the Supporting
Information) were fitted to eqs 6 and 7 (Figure 4), and the results
are reported in Table 1.
with
K ) [fac]/[mer] ) kmf/kfm
(3)
Furthermore, the temperature dependence of the pseudo-first-
order exchange rate constants, k, can be expressed by eq 4 where
pH, Metal Ion, and Solvent Effects on the Racemization
.
of (-)589[Co2(1a)3]4+ We have previously reported17 the
∆Sq ∆Hq
∆Hq
R
1
1
T
kbT
h
k298T
kinetics of racemization of the complex (-)589[Co2(1a)3]4+ as
k298 ) (1.4 ( 0.2) × 10-5 s-1, ∆Hq ) 99 ( 2 kJ mol-1, and
∆Sq ) -6 ( 7 J mol-1 K-1 (I ) 0.1 M, pH 4.75). In order to
obtain further insight into the mechanism, we have investigated
the influence of pH, added cobalt(II), and addition of a
coordinating solvent (acetonitrile) on the kinetics of racemiza-
tion.
k )
exp
-
)
exp
-
(
)
(
(
))
R
RT
298.15
298.15
(4)
∆Sq and ∆Hq are the entropy and enthalpy of activation,
respectively. An Arrhenius temperature dependence for the
paramagnetic relaxation rate has been assumed (eq 5, where
(1/T2m 298
)
is the value at 298.15 K and Em is the corresponding
Before studying the kinetics, the electronic spectrum of
[Co2(1a)3]4+ was studied as a function of pH in the range 1.2-
11.2. The electronic spectrum was unchanged in the range pH
4-11, confirming the stability of the complex, but changed
below pH 4 to yield the spectrum of the protonated ligand, the
change being complete at pH 1.5. Two features are worthy of
note: firstly, the change in the spectrum below pH 4 was rapid,
being essentially complete within the time of mixing; secondly,
the spectrum did not show any isosbestic points, and an analysis
of the data in terms of only two absorbing species was not
satisfactory. We conclude that, in acidic conditions, when the
activation energy).
1/T2m ) (1/T2m)298 exp[Em/R(1/T - 1/298.15)]
(5)
Equations 1-5 were simultaneously adjusted to the transverse
relaxation rates and to the equilibrium constants (Table S1 in
(23) Forse´n, S.; Hoffmann, R. A. J. Chem. Phys. 1963, 39, 2892; 1964,
40, 1189.
(24) Pople, J. A.; Schneider, W. G.; Bernstein, H. J. High Resolution
Nuclear Magnetic Resonance; McGraw-Hill Book Co.: New York, 1959;
p 224.

Lability of Mononuclear and Dinuclear Co(II) Complexes
J. Am. Chem. Soc., Vol. 119, No. 10, 1997 2491
Table 1. Derived Kinetic and Thermodynamic Parameters for the Variable Temperature and Pressure Studies
mer-fac
isomerization
mer
mer-fac
isomerization
mer
interconversion
interconversion
k298/s-1
1.6 ( 0.2
68.7 ( 3
7.6 ( 0.7
62.0 ( 2
∆Sq/(J mol-1K-1
)
-10.8 ( 8
+6.3 ( 0.4
-20.1 ( 6
+5.4 ( 0.2
a
∆Hq/(kJ mol-1
)
∆Vq/(cm3 mol-1
)
mer-fac
isomerization
mer-fac
isomerization
K298
0.23 ( 0.01
0.6 ( 0.3
∆S°/(J mol-1 K-1
)
-10.1 ( 1
+0.8 ( 0.3
a
∆H°/(kJ mol-1
)
∆V°/(cm3 mol-1
)
a At 349.6 K: (kmf)0 ) 107 ( 2 s-1, (kmm)0 ) 327 ( 3 s-1, K0 ) 0.25 ( 0.01.
Ligand Exchange in [Co2(1)3]4+ Followed by Electrospray
Table 2. Derived NMR Parameters for the Variable Temperature
Study
Mass Spectrometry. It has been established previously13 that
replacement of the methyl substituents of the benzimidazoles
by ethyls in ligands related to 1a does not affect formation of
m1
m2
m3
f
(1/T2m
)
298/s-1
37 ( 1
6.4 ( 0.6
36 ( 1
5.2 ( 0.6
26 ( 1
3.3 ( 0.6
26 ( 1
3.7 ( 0.7
1
the triple helix, and the H NMR spectrum of [Co2(1b)3]4+
Em/(kJ mol-1
)
confirms that a triple-helical complex is formed. The exchange
of ligand 1b for 1a offers therefore a means of studying the
exchange of two practically identical ligands in a [Co2(1)3]4+
complex. Attemps to follow the exchange by 1H NMR
spectroscopy were unsuccessful because of the complexity of
the spectra. We therefore turned to electrospray mass spec-
trometry (ESMS) which distinguishes between ligands 1a and
1b by their different molecular masses.
For ESMS to be applicable to this problem, one must be able
to identify unambiguously the different peaks observed in the
spectrum and to relate their intensities to the relative concentra-
tion of the complexes. We have previously reported26 that the
triple-helical complexes of cobalt(II) give clean mass spectra
in solution under ESMS conditions, the only significant peaks
being the molecular ions Mn+ (the base peak) and M(X)(n-1)+
where X- is the monovalent counter anion with an intensity
typically 10% that of the base peak. In agreement with this,
the ESMS spectrum of [Co2(1a)3]4+ has the base peak at m/z )
373.1 and that of [Co2(1b)3]4+ at m/z ) 394.2. Neither
compound shows significant fragmentation in the conditions
used. We may therefore attribute peaks between m/z ) 373
m/z ) and 394 to the formation of mixed ligand species.
The correlation of peak intensities to relative abundance in
solution is in general less straightforward. For the compounds
studied here, however, it is quite reasonable, since we are
comparing species of the same charge, with the same basic
structure, and very similar size (the difference in mass between
the heaviest and lightest ion is just over 5%), and none undergo
fragmentation. Results in the literature27,28 show good agree-
ment between speciations obtained by ESMS and those obtained
from NMR and spectroscopic titration for this type of com-
pound. To confirm the validity of this assumption, we have
carried out a scrambling experiment.
Figure 4. Pressure dependence of the rate constants kmm + kmf (solid
line, (9) signal m1, (b) signal m2, and (4) signal m3) and kfm (dashed
line, (] signal f) and pressure dependence of the equilibrium constant
K (0) for the intramolecular exchanges of [Co(2)3]2+ in acetonitrile
solution.
triple helix is thermodynamically unstable, the cobalt(II) ion
shows its expected lability, and that the decomplexation takes
place Via an intermediate.
Further kinetic experiments were carried out at 323 K, where
the half-life for racemization is on the order of 30 min.
Measurements at pH 5.07, 4.75, 4.26, and 4.02 (acetate buffer,
I ) 0.1M) gave rate constants of (3.8 ( 0.3), (3.7 ( 0.4), (3.2
( 0.3), and (4.6 ( 0.4) × 10-4 s-1 for racemization. Only the
last value, measured at pH 4.02, shows any significant change,
and we assume that this is due to the onset of the dissociation
of the complex in acid; above pH 4, however, the racemization
rate is independent of pH. If the mechanism of racemization
requires a rapid preequilibrium dissociation of one of the cobalt-
(II) ions, then the addition of free Co2+ would be expected to
slow the rate. Measurements with 0, 1.3, and 100 equiv of
cobalt(II) perchlorate hexahydrate added to the solution gave
rate constants of (4.0 ( 0.4), (2.9 ( 0.5), and (4.2 ( 0.2) ×
10-4 s-1, respectively, showing no retardation effect. Finally,
the racemization was studied in aqueous solution diluted with
an equal volume of acetonitrile, giving a rate constant of (2.8
( 0.4) × 10-4 s-1, slightly lower than that for the purely
aqueous solution (4.0 × 10-4 s-1). The rate is thus not very
sensitive to the presence of coordinating solvent contrary to our
observations with a copper(I) double helix.25
A volume of [Co2(1a)3](ClO4)4 (4.1 × 10-5 M in acetonitrile)
was added to an equal volume of [Co2(1b)3](CF3SO3)4 (3.1 ×
10-5 M in acetonitrile), and the ESMS spectrum was recorded
at regular intervals. Figure 5 shows the evolution of the
[Co2(1a)3-x(1b)x]4+ region of the spectrum with time. It will
be seen that the mixed ligand species [Co2(1a)2(1b)]4+ and [Co2-
(1a)(1b)2]4+ grow in slowly and no other signals are seen. The
ratio of the observed peak intensities after 3 h corresponds to
the statistical ratio expected if the intensities are proportional
to concentration. Throughout the experiment the only peaks
(26) Hopfgartner, G.; Piguet, C.; Henion, J. D.; Williams, A. F. HelV.
Chim. Acta 1993, 76, 1759. Hopfgartner, G.; Piguet, C.; Henion, J. D. J.
Am. Soc. Mass Spectrom. 1994, 5, 748.
(27) Piguet, C.; Hopfgartner, G.; Bocquet, B.; Schaad, O.; Williams, A.
F. J. Am. Chem. Soc. 1994, 116, 9092.
(25) Ru¨ttimann, S.; Piguet, C.; Bernardinelli, G; Bocquet, B.; Williams,
A. F. J. Am. Chem. Soc. 1992, 114, 4230.
(28) Marquis-Rigault, A.; Dupont-Gervais, A.; Van Dorsselaer, A., Lehn.
J.-M. Chem. Eur. J. 1996, 2, 1395.

2492 J. Am. Chem. Soc., Vol. 119, No. 10, 1997
Charbonnie`re et al.
Figure 7. Volume profile for the intramolecular exchanges.
Scheme 3. Formation Routes for [M2L3]
Discussion
The value for the equilibrium constant (Table 1) for the mer-
fac isomerization of [Co(2)3]2+ is slightly lower than the
statistical value of 1/3. We attribute this to a greater steric
repulsion when the three benzimidazole moieties are constrained
to lie on the same face of the octahedron, and this is supported
by the positive values of ∆H° and ∆V° for the reaction (Figure
7). The kinetics of the isomerization show the expected lability
of a high-spin cobalt(II) center, and the rate constants for the
exchange processes are comparable with the value of 6.9 s-1
reported29 for the racemization of [Co(phen)3]2+ in water. The
activation parameters for the mer-mer and mer-fac processes
are sufficiently close to suggest that they pass through a similar
transition state, which, given the positive ∆Vq values, is
dissociatively activated, as observed for solvent exchange at
cobalt(II) centers.30 Although the small negative value of the
entropy of activation is in contradiction with a dissociative mode
of activation, the difficulties encountered in using ∆Sq to assign
a reaction pathway have been discussed elsewhere.31 The
factors which affect this parameter are poorly understood, and
nonrandom errors are often associated with the determination
of ∆Sq. The kinetic data for [Co(2)3]2+ clearly establish that
the coordination sphere of three benzimidazole-pyridine units
does not possess an inherent inertness which might explain the
slow racemization of [Co2(1a)3]4+. At 298 K the ratio of the
rate constant for racemization of triple-helical [Co2(1a)3]4+ to
that for exchange between two sites in the mer isomer of [Co-
(2)3]2+ is 1.8 × 10-6. Examination of the activation parameters
for the two processes shows that this difference arises essentially
from the considerable increase in the activation enthalpy from
Figure 5. Base peak of the ESMS spectrum of a mixture of equal
volumes of [Co2(1a)3](ClO4)4 (4.1 × 10-5 M) and [Co2(1b)3](CF3SO3)4
(3.1 × 10-5 M) in acetonitrile after (a) 5, (b) 67, (c) 190, and (d) 368
min.
Figure 6. Intensities of the peaks due to [Co2(1a)x(1b)3-x]
4+ (x ) 0-3)
as a function of time in the ESMS spectrum: (×) [Co2(1b)3]4+, (b)
[Co2(1a)(1b)2]4+, (+) [Co2(1a)2(1b)]4+, (9) [Co2(1a)3]4+
.
of importance were [Co2(1a)3-x(1b)x]4+ or the adduct with one
counter anion.
With the significance of the peak intensities established, it is
now possible to follow the exchange of free ligand with the
complex. In a second experiment, equal volumes of a solution
of [Co2(1b)3](CF3SO3)4 (3.1 × 10-5 M in acetonitrile) and a
solution of 1a (2.3 × 10-4 M in acetonitrile, i.e., 7.4 mol of
1a/mol of [Co2(1b)3](CF3SO3)4) were mixed together, and the
ESMS spectra recorded as a function of time. Figure 6 shows
69(3) kJ mol-1 for [Co(2)3]2+ to 99(2) kJ mol-1 for [Co2(1a)3]4+
.
Before addressing the question of the inertness of [Co2-
(1a)3]4+, we note that the ESMS data allow us to clarify the
pathways for complex formation and dissociation in the triple-
helical system. We have previously noted that formation of
the triple-helical complex, as followed by spectroscopic titration,
is rapid.32 The possible routes to a triple-helical complex M2L3
are shown in Scheme 3. The final step of formation may occur
in two ways: (i) route f, in which a preformed fac-ML3 complex
the variation of the intensity of the peaks corresponding to
4+
[Co2(1a)x(1b)3-x
]
(x ) 0-3) as a function of time. As
expected, in view of the great excess of 1a, the signal due to
[Co2(1b)3]4+ falls away rapidly, but the surprising observation
is that it is [Co2(1a)3]4+ which grows fastest, the mixed species
[Co2(1a)x(1b)3-x]
4+ (x ) 1 or 2) only appearing later and more
slowly. Throughout the experiment, the spectrum is dominated
4+
by the peaks due to [Co2(1a)x(1b)3-x
]
(x ) 0-3), but weak
4+
peaks are now observed for species [Co2(1a)x(1b)4-x
0-4) and [Co(1a)x(1b)3-x
]
(x )
(29) Blinn, E. L.; Wilkins, R. G. Inorg. Chem. 1976, 15, 2952.
(30) Lincoln, S. F; Merbach, A. E. AdV. Inorg. Chem. 1995, 42, 1
(31) Newman, K. E.; Meyer, F. K.; Merbach, A. E. J. Am. Chem. Soc.
1979, 101, 1470
(32) Piguet, C.; Bernardinelli, G.; Bocquet, B.; Schaad, O.; Williams,
A. F. Inorg. Chem. 1993, 33, 4112.
]
2+ (x ) 0-3). The presence of these
species, typically a few percent of the base peak, is a result of
the considerable excess of ligand present in solution during this
experiment.

Lability of Mononuclear and Dinuclear Co(II) Complexes
J. Am. Chem. Soc., Vol. 119, No. 10, 1997 2493
binds a second metal to give M2L3 (we will call this the keystone
mechanism), and (ii) route h, in which a third ligand is added
to a preformed M2L2 complex (we will call this the braiding
mechanism). The braiding mechanism might at first sight
appear attractive since we have shown13 that a rapid equilibrium
exists between M2L2 and M2L3 for M ) Zn(II).
Chart 1
If we assume microscopic reversibility, then the dissociation
of the M2L3 complex will occur either by eq 8 according to the
M2L3 f ML3 + M
(8)
keystone mechanism or by eq 9 according to the braiding
mechanism. The ESMS results for [Co2(1b)3]4+ in the presence
M2L3 f M2L2 + L
(9)
of excess 1a allow us to eliminate the braiding mechanism, since
in the presence of an excess of a second ligand L′ (in practice
1a), the M2L2 intermediate will react with L′ to form the mixed
ligand species M2L2L′. We would therefore expect the rapid
apparition of mixed ligand species, with M2L2L′ appearing
before M2L′3, contrary to experiment. The keystone mechanism
does however explain this result. After dissociation of M
according to eq 8, the free metal ion will be complexed by the
excess free ligand, which is exclusively L′ at the beginning of
the experiment:
increase in ∆Hq. It is only when the pH is sufficiently low
(<4) for the proton to compete with the cobalt for nitrogen
donors that the recombination reaction is prevented, and the
complex dissociates rapidly.
The Bailar twist offers an alternative mechanism for racem-
ization of a trisbidentate complex and has recently been observed
for another triple-helical complex.18 The triple-helical complex
[Ga2(10)3]6- (Chart 1) studied by Raymond and co-workers was
shown to undergo inversion via a Bailar twist mechanism
passing through the ∆,Λ-complex (or meso-helicate) as an
intermediate with ∆Gq ) 79 ( 2 kJ mol-1, against ∆Gq
M + 3L′ f ML′3
(10)
As more M is released into the solution (by reaction 8 and the
reverse of reaction 10), the ML′3 will capture it to give M2L′3;
it is only as the experiment proceeds, and the amount of free L
in solution increases to where it is comparable with free L′,
that mixed ligand species will be formed. This is exactly what
is observed in Figure 6. Thus, the ESMS data establish the
formation route of [Co2(1)3]4+ to be a-b-c-f as shown in
Scheme 3. The first step is formation of mononuclear [Co-
(1)3]2+ which is then locked into an inert structure by com-
plexation of a second metal ion in the same way that the
298
298
298
) 101 ( 4 kJ mol-1 for [Co2(1a)3]4+. In this case the ∆Gq
was only 12 kJ mol-1 higher than that of the equivalent
mononuclear complex,33 in sharp contrast to the difference
between [Co2(1a)3]4+ and [Co(2)3]2+. Albrecht and Kotila19
have reported the value of ∆Gq233 ) 43 kJ mol-1 for the triple
helix [Ti2(11)3]4-; since the structure reported for this compound
shows the titanium coordination to be quite close to trigonal
prismatic, the transition state geometry for the Bailar twist, it
seems most probable that this complex also racemizes by a
Bailar twist. We may confidently exclude a Bailar twist
mechanism for [Co2(1a)3]4+. The trigonal prismatic transition
state requires the two coordinating atoms of the chelate ring to
lie approximately parallel to the C3 symmetry axis of the triple
helix. It is not possible for ligand 1a to achieve this at one
metal site while coordination of the cobalt is maintained at the
other. By virtue of the repulsions between the hydrogen atoms
in position 4 of the benzimidazole units, ligand 1 cannot be
planar, and the distribution of the four coordinating atoms in
space forms a spiral: this spiral can be right handed or left
handed, but the chirality cannot be changed while the integrity
of the [Co2(1a)3]4+ complex is maintained.
keystone of an arch confers stability on the structure as a whole.
4+
The weak peaks in the ESMS spectrum due to [Co2(1a)x(1b)4-x
(x ) 0-4) and [Co(1a)x(1b)3-x
with this mechanism, the first species arising from interaction
of free ligand with mer-[Co2(1a)x(1b)3-x
4+ (x ) 0-3) in which
]
]
2+ (x ) 0-3) are also consistent
]
one cobalt is octahedrally coordinated and the second is bound
to only two benzimidazole-pyridines, and is thus coordinatively
unsaturated. There is some precedent for this in the literature:
a study of the formation of trinuclear double helicates by
spectroscopic titration showed the successive formation of ML2,
M2L2, and M3L2.16
The slow racemization of the triple helix may be understood
in terms of a dissociative mechanism (eq 11). The rate-
fast
slow
fast
Λ-M2L3 (11)
∆-M2L3
∆-fac-ML3 + M
Λ-fac-ML3 + M
The constraints due to the ligand may also be seen in the
solid state stuctures, most particularly in the pitch of the helix.
If an and bn are the coordinates of the two atoms of the chelate
unit bound to metal Mn, then we may define a midpoint of the
chelate xn as (an + bn)/2, and estimate the twisting of the helix
by the torsion angle x1-M1-M2-x2, divided by the metal-
metal distance |M1-M2|. The pitch, defined as the displacement
along the axis required for a complete turn, is given by 360/θ
Å. For [Ga2(10)3]6- the pitch calculated in this way is 58.3 Å
compared to 19.9 Å for [Co2(1a)3]4+, differing by a factor of
3, while for the even more rapidly inverting [Ti2(11)3]4- the
rac-mer-ML3
determining step is thus the loss of M from the triple-helical
complex (eq 8), which is expected to show first-order kinetics.
This possibility was not considered in our previous communica-
tion.17 Although the individual cobalt-nitrogen bonds are
labile, the rigid structure of the helix will maintain the
dissociated nitrogen atom very close to the metal ion, and
consequently a greater degree of bond breaking is necessary to
allow the dissociation of cobalt from a pyridine-benzimidazole
unit than for the mononuclear species. This explains the
(33) Kersting, B.; Telford, J. R.; Meyer, M.; Raymond, K. N. J. Am.
Chem. Soc. 1996, 118, 5712.

2494 J. Am. Chem. Soc., Vol. 119, No. 10, 1997
Charbonnie`re et al.
pitch is still greater (ca. 220 Å).19 The shorter the pitch, the
harder it will be to untwist the helix to pass through the trigonal
prismatic transition state required by the Bailar mechanism. This
leads to the interesting hypothesis that introducing a certain
curvature into the disposition of the ligating atoms will shorten
the pitch and slow the racemization.
with 20 mL of CH2Cl2 and the organic layers gathered, filtered over
cellulose, and evaporated to dryness. The resulting solid was purified
by column chromatography over silica (50 g, CH2Cl2/MeOH, 99/1) to
give 9a as a pale brown solid which decomposes above 62 °C. Yield:
378 mg (86%). Rf ) 0.54 on Alox TLC (CH2Cl2/MeOH, 99/1). Anal.
Calcd for C29H26N6O6 : C, 62.80; H, 4.73; N, 15.16. Found: C, 62.69;
H, 5.00; N, 14.89. EI/MS: m/z ) 508 (M - NO2), 462 (M - 2NO2).
Preparation of Bis[3-nitro-4-[ethyl[(5-methylpyridin-2-yl)carbo-
nyl]amino] phenyl]methane (9b). 9b was obtained by the same
procedure as for 9a, starting from 500 mg of the acid 3 and 2.7 mL of
SOCl2 in 65 mL of dry CH2Cl2. In the second step, the solid was
dissolved in 30 mL of CH2Cl2 containing 420 mg (1.22 mmol) of the
diamine 8b and 2.6 mL of Et3N. After workup and chromatography
on silica (CH2Cl2/MeOH, 99/1) 9b was obtained as a pale brown solid
which decomposes above 62 °C. Yield: 650 mg (92%). Rf ) 0.36
on Alox TLC (CH2Cl2/MeOH, 99/1). EI-MS: m/z ) 583 (M + H+),
The only other triple helix for which racemization data are
available is the trinuclear triple helix [Ni3(12)3]6+ which was
reported by Lehn et al. to racemize with a rate constant of 1.5
× 10-6 s-1 at room temperature.12 The trinuclear species is
indeed slower than the mononuclear species [Ni(bipy)3]2+ 34
, but
the factor is only on the order of 500, 3 orders of magnitude
less than observed for [Co2(1a)3]4+. The racemization of [Ni-
(phen)3]2+ has been assigned a dissociative pathway,35 and
assuming this also to be true for [Ni3(12)3]6+ we might anticipate
a deceleration similar to that observed in our results. We
attribute the small magnitude of the effect to the greater degree
of flexibility in 12 than in 1, and to the much greater pitch of
[Ni3(12)3]6+, 48.2 Å, more than twice that of [Co2(1a)3]4+. The
lesser twisting, and the greater degree of flexibility, will facilitate
the loss of a nickel from one site without disruption of the other,
and will also lower the energetic barrier to a ∆,Λ-complex as
an intermediate. It will be noted that the two metal ions with
partially filled d-shells, Co(II) and Ni(II), are both much slower
to racemize than the ions Ga(III) (d10) and Ti(IV) (d0) studied
by Raymond18 and Albrecht.19
In conclusion, we have shown that the coordination sphere
of the mononuclear complex [Co(2)3]2+ shows the lability
expected for a cobalt(II) center. The inertness of [Co2(1a)3]4+
is thus a supramolecular effect, arising from the rigidity of the
structure as a whole. We have shown how electrospray mass
spectroscopy may be used to follow ligand exchange reactions
by the use of ligands with different peripheral substituents. The
results show the complex to be formed by initial assembly of
three ligands around one metal followed by the complexation
of the second to the preorganized hexadentate site. Each metal
thus holds in place the coordination site for the other, leading
to remarkable kinetic stability for the cobalt(II) ion. Comparison
with other triple-helical systems suggests that [Co2(1a)3]4+ owes
its remarkable inertness to the rigidity of the ligand 1 which
results in a very tight pitch. Careful design of the spacing group
may therefore be used to control the lability of the helical
complex.
3
536 (M - NO2). 1H-NMR in CDCl3: δ 1.12, 1.27 (2t, 6H, J ) 7
Hz); 2.20, 2.40 (2s, 6H, br); 3.71, 4.22 (m, 4H, br); 4.06 (s, 2H, br);
7.30-8.50 (m, 12H, br). The spectrum shows the presence of more
than one conformation as a result of restricted rotation of the amide20
and results in broad poorly resolved signals.
Preparation of Bis[1-methyl-2-(5-methylpyrid-2-yl)benzimidazol-
5-yl] methane (1a). A 369 mg sample of the diamide 9a (0.67 mmol)
was dissolved in 100 mL of absolute ethanol with 20 mL of distilled
water, 3.9 mL of concentrated HCl (47 mmol), and 1.12 g of activated
iron (20.1 mmol). The solution was refluxed under nitrogen for 14 h.
The iron was filtered and ethanol evaporated under reduced pressure.
A 20 mL sample of water containing 10 g of Na2H2EDTA (26.9 mmol)
was added to the solution and the mixture stirred vigorously with 100
mL of CH2Cl2 while the pH was increased to 9 with concentrated NH4-
OH. The aqueous layer was then further extracted with three portions
of 100 mL of CH2Cl2. The combined organic layers were dried over
Na2SO4, filtered over cellulose, evaporated to dryness, and purified by
column chromatography over silica (30 g, CH2Cl2/MeOH, 95/5). 1a
is obtained as a white solid. Yield: 265 mg (87%). The analysis and
spectra correspond to the results of the literature.32
Preparation of Bis[1-ethyl-2-(5-methylpyrid-2-yl)benzimidazol-
5-yl] methane (1b). 1b was obtained analogously to 1a starting from
diamide 9b in 87% yield after purification. Anal. Calcd for
C31H30N6: C, 76.50; H, 6.23; N, 17.27. Found: C, 76.24; H, 6.25; N,
17.36. EI/MS: m/z ) 486 (M+); 471 (M - CH3); 457 (M - C2H5).
3
1H-NMR in CDCl3: δ 1.44 (t, 6H, J ) 7 Hz), 2.40 (s, 6H), 4.26 (s,
2H), 4.81 (q, 4H, 3J ) 7.2 Hz), 7.19 (dd, 2H), 7.23 (dd, 2H), 7.62 (m,
2H), 7.68 (m, 2H), 8.27 (dd, 2H), 8.51 (m, 2H). 13C-NMR in CDCl3:
δ 15.3, 18.4 (primary C); 40.4, 42.2 (secondary C); 109.7, 119.7, 124.0,
124.4, 137.2, 149.0 (tertiary C); 133.3, 134.7, 136.2, 142.9, 148.1, 149.9
(quaternary C).
Preparation of 5-Methylpyridine-2-carboxylic Acid N-Methyl-
N-(2-nitrophenyl)amide (7). A 1 g sample of acid 3 (7.3 mmol) was
dissolved in 100 mL of dry CH2Cl2 with a few drops of DMF and 5.3
mL of SOCl2 (73 mmol) added dropwise over 10 min. A white
precipitate formed which slowly disappeared while the solution was
refluxed for 1 h with a CaCl2 tube capping the reflux condenser. The
solution was then evaporated to dryness and the solid coevaporated
twice with 30 mL of CH2Cl2 and finally dried under vacuum (10-2
Torr, 30 min). The solid was dissolved in 60 mL of dry CH2Cl2 and
a solution containing 1.22 g of 6a (7.3 mmol) and 5.2 mL of Et3N (73
mmol) in 60 mL of dry CH2Cl2 added dropwise for 15 min under
constant stirring. The deep green solution was refluxed for 5 h under
nitrogen and evaporated to dryness. The solid was dissolved in 250
mL of CH2Cl2 and washed with 250 mL of 10% (NH4)2SO4 in water.
The aqueous layer was washed twice with 150 mL of CH2Cl2. The
combined organic layers were dried over Na2SO4, filtered, and
evaporated to dryness. The oily residue was purified by column
chromatography over alumina (CH2Cl2/MeOH, 99.5/0.5). After drying,
7 was obtained as a pale brown solid which decomposes above 127
°C. Yield: 1.43 g (72%). 1H-NMR in CDCl3: δ 2.39 (s, 3H, br);
3.47 (s, 3H, br); 7.29-8.43 (m, 7H, br). EI-MS: m/z ) 225 (M-NO2).
Rf ) 0.26 (silica TLC, CH2Cl2/MeOH, 98/2).
Experimental Section
Materials. Solvents and starting materials were purchased from
Fluka AG (Buchs, Switzerland) and used without further purification
unless otherwise stated. Aluminum oxide (Merck activity I or II-III,
0.063-0.200 mm) and silica gel (Merck, 0.063-0.200 mm) were used
for preparative column chromatography. 2-Carboxy-5-methylpyridine
(3),32 bis[3-nitro-4-(N-methylamino)phenyl]methane (8a),13 bis[3-nitro-
4-bis(N-ethylamino)phenyl]methane (8b),13 and the complex [Co2(1a)3]-
32
(ClO4)4 were prepared according to literature procedures.
Ligand Synthesis. Preparation of Bis[3-nitro-4-[methyl[(5-me-
thylpyridin-2-yl)carbonyl]amino]phenyl]methane (9a). A 300 mg
(2.18 mmol) sample of acid 3 was dissolved in 45 mL of dry CH2Cl2
(filtered on Alox Act. I), and 1.6 mL (21.7 mmol) of SOCl2 was added
with a few drops of DMF. A white precipitate formed which slowly
disappeared on refluxing the solution for 1 h, with a CaCl2 tube capping
the reflux condenser. The solution was evaporated to dryness and the
oily residue dried under vacuum for 30 min. The solid was dissolved
in 25 mL of dry CH2Cl2 containing 280 mg (0.79 mmol) of diamine
8a and 1.14 mL of Et3N, and the solution was heated to reflux for 18
h. After cooling, the solution was extracted with 50 mL of water and
50 mL of 10% (NH4)2SO4 in water. The aqueous fractions were washed
Preparation of 5-Methyl-2-(1-methylbenzimidazol-2-yl)pyridine
(2). A 604 mg sample of 7 (2.2 mmol) was dissolved in 100 mL of
absolute ethanol containing 30 mL of distilled water, 6.4 mL of
(34) Schweitzer, G. K; Lee, J. M J. Phys. Chem. 1952, 56, 195.
(35) Lawrance, G. A.; Stranks, D. A. Inorg. Chem. 1978, 17, 1804.

Lability of Mononuclear and Dinuclear Co(II) Complexes
J. Am. Chem. Soc., Vol. 119, No. 10, 1997 2495
concentrated HCl, and 2.47 g of activated iron (44 mmol). The solution
was refluxed for 12 h under a nitrogen atmosphere. The remaining
iron was filtered, the ethanol evaporated under reduced pressure, and
a solution of 20.6 g of Na2H2EDTA (55.3 mmol) in 60 mL of water
added. The pH was raised to 9 with concentrated NH4OH while the
aqueous layer was extracted with 500 mL of CH2Cl2. The aqueous
layer was washed with 100 mL of CH2Cl2, and the combined organic
layers were dried over Na2SO4, filtered, and evaporated to dryness.
The solid was dissolved in 50 mL of CH2Cl2/MeOH (95/5) and filtered
over a thin layer of Alox. The Alox was washed with 200 of mL CH2-
Cl2/MeOH (95/5). The combined solutions were evaporated to dryness,
and the solid was recrystallized in a mixture of CH2Cl2/hexane to give
white crystals of 2, melting at 129-131 °C. Yield: 425 mg (86%).
1H-NMR in CDCl3: δ 2.39 (s, 3H); 4.24 (s, 3H); 7.20-7.36 (m, 2H);
7.41 (dd, 1H, 3J ) 6.1 Hz); 7.63 (dd, 1H, 3J ) 7.9 Hz); 7.79 (dd, 1H,
temperature. Fine red needles formed and were filtered. The mother
liquor was then concentrated to give a second crop of crystals. The
first crop was dissolved in 1 mL of acetonitrile and 10 mL of water,
-
and the anions were exchanged with ClO4 loaded on a Dowex X3
anion exchange column. The solution was concentrated and the
complex slowly recrystallized. The solid was filtered and dried under
vacuum to give a (-)589-[Co2(1a)3](ClO4)6 as a red solid. Yield: 10.0
mg (21%). [R]589 ) -1022(4)°, [R]578 ) - 1085(4)°, [R]546 ) -1681-
(5)° (T ) 25 °C, acetonitrile). The second crop of crystals gives 18.9
mg of complex (54%) which is a nonracemic mixture of ∆ and Λ
isomers. Other analyses (1H-NMR and IR) correspond to the racemic
complex.
Preparation of [Co2(1b)3](ClO4)6‚8H2O. A 40 mg sample of [Co2-
(1b)3](ClO4)4‚4H2O‚MeOH (19.8 µmol) was dissolved in 5 mL of
acetonitrile. A 5 mL portion of bromine was added and the solution
refluxed for 12 h. The excess bromine was evaporated under reduced
pressure and 10 mL of diethyl ether added to the solution. A red-
orange precipitate formed which was filtered and suspended in 20 mL
of acetonitrile containing 300 mL of LiClO4. The mixture was refluxed
for 1 h, filtered hot, and slowly cooled to room temperature. The solid
obtained was filtered and dried under vacuum to give [Co2(1b)3]-
(ClO4)6‚8H2O as a red-orange solid. Yield: 29.6 mg (65%). Anal.
Calcd for Co2C93H90N18Cl6O24‚8H2O: C, 48.18; H, 4.61; N, 10.87.
Found: C, 48.05; H, 4.42; N, 10.69. 1H-NMR in CD3CN: δ 1.71 (t,
3
3J ) 6.9 Hz); 8.25 (d, 1H, J ) 8.0 Hz); 8.50 (s, 1H). 13C-NMR in
CDCl3: δ 18.52, 32.67 (primary C); 109.86, 119.93, 122.52, 123.15,
124.33, 137.41, 149.05 (tertiary C); 133.65, 137.39, 142.61, 148.04,
150.63 (quaternary C). EI-MS: m/z 223 (M•+). Anal. Calcd for
C14H13N3: C, 75.29; N, 18.82; H, 5.88. Found : C, 74.43; N, 18.53;
H, 5.64.
Preparation of Complexes. Caution. Perchlorate salts with
organic solvents are potentially explosive and should be handled with
care.36
3
18H, J ) 7.4 Hz, br), 2.30 (s, 18H), 3.62 (s, 6H, br), 4.42 (s, 6H),
Preparation of [Co2(1b)3](ClO4)4‚4H2O‚MeOH. A 200 mg sample
of ligand 1b (0.41 mmol) was dissolved in 20 mL of CH2Cl2 and a
solution of 106 mg (0.28 mmol) of Co(ClO4)2‚6H2O in 20 mL of CH3-
CN added. The solution, which turns orange instantaneously, was
evaporated to dryness and dried under vacuum for 1 h. The solid was
dissolved in a minimum of acetonitrile and methanol slowly diffused
on it. The complex was obtained as orange crystals which were filtered.
Concentration of the mother liquor afforded a second crop. Yield: 211
mg (78%). Anal. Calcd for Co2C93H90N18Cl4O16‚4H2O‚MeOH: Co,
5.67; C, 54.29; H, 4.94, N, 12.12. Found: Co, 5.71; C, 54.56; H, 5.00;
N, 12.19. 1H-NMR in CD3CN, all signals are broad: δ -41.9 (s, 6H),
-3.86 (s, 6H), 0.52 (s, 18H), 1.87 (s, 6H), 8.64 (s, 18H), 11.94 (s,
6H), 14.67 (s, 6H), 19.61 (s, 6H), 33.2 (s, 6H), 60.6 (s, 6H), 80.8 (s,
4.95 (q, 12H, 3J ) 7.4 Hz, br), 7.02 (s, 6H, br), 7.20 (d, 6H, br), 7.84
(d, 6H), 8.22 (d, 6H, br), 8.40 (d, 6H). 13C-NMR in CD3CN: δ 15.5,
19.5 (primary C); 41.4, 43.1 (secondary C); 113.9, 115.0, 127.7, 129.8,
144.9, 154.0 (tertiary C); 135.7, 139.4, 141.0, 143.8, 144.5, 150.9
(quaternary C.). IR (cm-1): 3120, 3080, 2983, 2940 (νCHaro); 1612,
1586 (νCdC, νCdN); 1465, 1383 (δCH , δCH ); 1088, 622 (νClO ); 789,
3
2
4
718, 692 (δCHaro.). ES/MS (CH3CN): m/z ) 262.8 [Co2(1b)3]6+, 335.4
[Co2(1b)3(ClO4)]5+, 444.0 [Co2(1b)3(ClO4)2]4+, 625.4 [Co2(1b)3-
(ClO4)3]3+, 987.6 [Co2(1b)3(ClO4)4]2+
.
Preparation of [Co(2)3](ClO4)2‚MeOH. A 223 mg sample of 2 (1
mmol) was dissolved in 5 mL of CH2Cl2/CH3CN (1/1) and 5 mL of
CH3CN containing 122 mg (0.33 mmol) of Co(ClO4)2‚6H2O added.
The orange solution was stirred for 5 min and evaporated to dryness.
The solid was dissolved in 2.5 mL of acetonitrile and methanol slowly
diffused to give, after filtration, orange crystals of [Co(2)3]-
(ClO4)2‚MeOH. Yield: 306 mg (96%). Anal. Calcd for
CoC42H39N9Cl2O8‚CH3OH: C, 53.80; N, 4.52; H, 13.13. Found: C,
53.39, N, 4.42; H, 13.27. 1H-NMR in CD3CN (22 °C, all signals
singlets): fac isomer, δ -0.2, 3.6, 12.6, 35.0, 79.9 (single protons),
1.6, 17.2 (methyl protons); mer isomer, δ -4.1, -2.6, 0.3, 5.0, 7.5,
10.8, 13.4, 14.2, 17.7, 34.9, 47.8, 49.1, 51.8, 64.2, 91.2 (single protons),
-8.8, -6.8, 4.5, 14.9, 28.6. 30.3 (methyl protons); eight protons
appeared as very broad signals and could not be attributed to mer or
fac isomers, δ -50.2, -26, -21.0, 5.9, 21.0, 48.0, 73.2, 107.8.
Spectroscopic and Analytical Measurements. UV-vis spectra
were recorded at 22 °C in acetonitrile or dichloromethane on a Perkin-
Elmer Lambda 5 spectrophotometer with quartz cells of 1 and 0.1 cm
path length. The spectrophotometric titrations were recorded on a
computer driven Perkin-Elmer Lambda 5 as previously described.13
Infrared spectra were recorded on a Perkin-Elmer IR 883 as KBr pellets
with wavenumbers in inverse centimeters. EI-MS spectra (70 eV) were
recorded with VG 7070E and Finnigan 4000 spectrometers. Elemental
analyses were performed by Dr. H. Eder of the Microchemical
Laboratory of the University of Geneva. Metal contents were
determined by atomic absorption (Pye Unicam SP9) after acidic
oxidative mineralization of the complexes.
6H). IR (cm-1): 3078, 2979 (νCH ); 1609, 1585, 1539 (νCdC, νCdN);
aro
1485, 1465 (δCH , δCH ); 1093, 623 (νClO ). ES/MS (CH3CN): m/z )
3
2
4
394.4 [Co2(1b)3]4+, 559.0 [Co2(1b)3(ClO4)]3+
.
Preparation of [Co2(1a)3](ClO4)6‚6H2O. A 250 mg sample of [Co2-
(1a)3](ClO4)4 (0.127 mmol) was dissolved in 100 mL of CH3CN, and
420 µL of 30% H2O2 in water (0.38 mmol) and 230 µL of 7% HClO4
in water were added with 2 mg of [Cp2Fe](BF4) as catalyst. The
solution was heated to 40 °C for 4 h with addition of 2 mg of [Cp2-
Fe](BF4) each hour. The solution was cooled to room temperature and
concentrated, and a saturated solution of LiClO4 in water added until
the solid began to precipitate. The solution was cooled to 4 °C, and
the red-orange solid was filtered and dissolved in a minimum of
acetonitrile. Methanol was slowly diffused to give red crystals which
were filtered and dried (10-2 Torr, 4 h. 60 °C). The mother liquor
was concentrated to give a second crop. Yield: 243 mg (88%). Anal.
Calcd for Co2C87H78N18Cl6O24‚6H2O: Co, 5.36; C, 47.53; H, 4.13, N,
11.47. Found: Co, 5.57; C, 47.41; H, 3.96; N, 11.48. ES/MS: m/z )
946 [Co2(1a)3(ClO4)4]2+; 597.2 [Co2(1a)3(ClO4)3]3+; 423.0 [Co2(1a)3-
(ClO4)2]4+; 318.6 [Co2(1a)3(ClO4)]5+; 249.0 [Co2(1a)3]6+ 1H-NMR in
.
CD3CN: δ 2.16 (s, 18H), 3.71 (s, 6H), 4.29 (s, 6H), 4.52 (s, 18H),
7.03 (s, 6H), 7.20 (s, 6H), 7.81 (d, 6H), 8.23 (d, 6H), 8.55 (d, 6H).
13C-NMR in CD3CN: δ 15.6, 19.5 (primary C); 43.2 (secondary C);
113.8, 115.0, 127.7, 129.7, 144.8, 154.1 (tertiary C); 135.7, 139.4, 141.0,
143.7, 150.9, 152.2 (quaternary C). IR (cm-1): 3040, 3127 (νCHaro),
ES-MS spectra were recorded on a Finnigan-Mat SSQ 7000
spectrometer at the Mass Spectrometry Laboratory of the University
of Geneva. For a typical kinetic experiment followed by ES-MS, equal
1612, 1586, 1546 (νCdC, νCdN), 1485, 1417 (δCH
δCH ), 1090, 622
3,
2
(νClO ).
4
Preparation of (-)589-[Co2(1a)3](ClO4)6. A 47 mg sample of [Co2-
(1a)3](ClO4)6‚6H2O was dissolved in 10 mL of acetonitrile. A 5 mL
portion of water was added followed by addition of 5 mL of a saturated
solution of Na2[(+)589-(Sb2(C4O6H2)2)]‚5H2O21 in water. The solution
was concentrated until the beginning of crystallization. A 2 mL sample
of acetonitrile was added and the solution heated to 80 °C to dissolve
the solid. The solution was filtered hot and slowly cooled to room
volumes of each solution of complexes or of ligands, of about 10-5
M
concentration, were mixed at time zero. Spectra were then recorded
at regular intervals, taking as time t the time of detection of the species.
Relative intensities of the peaks corresponding to [Co2(1a)3-x(1b)x]4+
(x ) 0-3) were measured and weighted by the sum of the total intensity
of these signals to give the relative percentages of each detected species.
CD spectra were recorded on a Jasco 710 spectrometer using
thermostated quartz cells of 0.5, 1, and 2 cm path length. The resolution
(36) Wolsey, W. C. J. Chem. Educ. 1978, 55, A355.

2496 J. Am. Chem. Soc., Vol. 119, No. 10, 1997
Charbonnie`re et al.
was 1 nm, the sensitivity 0.020°, and the time response 1s. For a typical
kinetic experiment followed by CD, spectra were recorded at regular
intervals for a total of at least 4 half-lifes. ∆θ values as a function of
time are collected for 10 wavelengths showing maximum variations in
the intensity of the signal. These wavelengths were collected in an
interval of 50 nm. The evolution of ∆θ as a function of t for each
wavelength was fitted to a decreasing exponential model with a
nonlinear least squares algorithm. The fitting gives the rate of decrease
at the 10 wavelengths, and the average value affords the racemization
rate of the complex and its standard deviation.
accumulated over a total spectral width of 45 450 Hz. Transversal
relaxation times, T2, were obtained by deconvolution of peaks into
Lorentzian curves using the program Anaspec39 (Bruker ARX 400). T2
values were then corrected for field inhomogeneity and line broadening
by subtraction of the width at half-height of the TMS signal. The
analysis of the experimental data was accomplished by a nonlinear least
squares program fitting the values of the desired parameters.40 Reported
errors are 1 standard deviation.
Acknowledgment. We thank Mr. Werner Kloeti for running
the ESMS spectra and Dr. Markus Albrecht (University of
Karlsruhe) for supplying coordinates of an X-ray structure. This
research was supported by the Swiss National Science Founda-
tion.
1H-NMR spectra and 13C-NMR spectra for characterization at room
temperature and pressure were recorded on a Varian Gemini 300 with
tetramethylsilane (TMS) as the inner reference. Chemical shifts are
given in parts per million, and abbreviations are as follows: s, singlet;
d, doublet; t, triplet; q, quadruplet; m, multiplet; br, broad. Variable
temperature and pressure 1H-NMR spectra were recorded on a Bruker
ARX 400 spectrometer, with solutions 10-2 M in [Co(2)3]2+ in
acetonitrile and with TMS as the shift and line width reference. The
ambient pressure measurements were followed in a commercial
thermostated probe, and the temperature was found to be constant within
(0.2 K as measured by a substitution technique.37 Variable pressure
measurements were made up to 200 MPa using a high-pressure probe.38
1
Supporting Information Available: Relaxation of the H
NMR signal as a function of temperature and pressure (Tables
S1 and S2) (2 pages). See any current masthead page for
ordering and Internet accesss instructions.
JA963693L
1
The variable temperature (variable pressure) H-NMR spectra were
obtained using a 90° pulse length of 6 µs (17 µs) in the quadrature
detection mode, with 64K data points resulting from 64 (256) scans
(38) Merbach, A. E.; Frey, U.; Helm, L.; Roulet, R. In AdVanced
Applications of NMR to Organometallic Chemistry; Gielen, M., Willem,
R., Wrackmeyer, B., Eds.; John Wiley & Sons: Chichester, 1996; p 199.
(39) Helm, L. Program Library, University of Lausanne, 1996.
(40) Program Scientist, Version 2.0, MicroMath Inc., 1995.
(37) Amman, C.; Meier, P.; Merbach, A. E. J. Magn. Reson. 1982, 46,
319.