Methyl transfer from CH3CoIIIPc to thiophenoxides revisited: Remote substituent effect on the rates

Article abstract of DOI:10.1021/ic0503378

A two-step mechanism of the reaction of CH₃Co^IIIPc (Pc = dianion of phthalocyanine) with thiophenoxides in DMA has been confirmed, and the visible spectrum of the inactive transient, CH₃Co ^IIIPc(SAr)^-, has been determined. Rapid rates for ligation of CH₃Co^IIIPc, yielding CH₃Co ^IIIPc(S-C₆H₄-X)^-, are virtually independent of X; this step proceeds probably by an I_d mechanism. Kinetic data for the follow-up methyl-transfer step yield second-order rate constants and stability constants for CH₃Co^IIIPc(S-C ₆H₄-X)^- consistent with those estimated from concentration dependence of the amplitude of the ligand-exchange step. Cyclic voltammetry provides first reduction potential for CH₃Co ^IIIPc(DMA) of -1.42 V vs Fc⁺/Fc, which makes an OSET mechanism unlikely. Homolytic decay of CH₃Co^IIIPc(SAr) ^- has also been ruled out. All of the kinetic data, including Hammett's p = -2.3 ± 0.1, N-donor inhibition, and alkyl group effect, Me > Et, indicate that the reaction is a normal S_N² methyl transfer, only very fast. Methyl transfer to aliphatic thiolates is also rapid and follows the same S_N² mechanism. Exceptional methyl-transfer reactivity of the phthalocyanine model sharply contrasting with the inertness of methylcobaloxime is explained.

Full text of DOI:10.1021/ic0503378

Inorg. Chem. 2005, 44, 5483−5494
Methyl Transfer from CH3CoIIIPc to Thiophenoxides Revisited: Remote
Substituent Effect on the Rates
Wlodzimierz Galezowski*
Department of Chemistry, Adam Mickiewicz UniVersity, Grunwaldzka 6, 60-780 Poznan, Poland
Received March 4, 2005
A two-step mechanism of the reaction of CH3CoIIIPc (Pc
has been confirmed, and the visible spectrum of the inactive transient, CH3CoIIIPc(SAr)-, has been determined.
Rapid rates for ligation of CH3CoIIIPc, yielding CH3CoIIIPc(S X)-, are virtually independent of X; this step
C6H4
proceeds probably by an Id mechanism. Kinetic data for the follow-up methyl-transfer step yield second-order rate
constants and stability constants for CH3CoIIIPc(S X)- consistent with those estimated from concentration
C6H4
dependence of the amplitude of the ligand-exchange step. Cyclic voltammetry provides first reduction potential for
CH3CoIIIPc(DMA) of 1.42 V vs Fc+/Fc, which makes an OSET mechanism unlikely. Homolytic decay of
CH3CoIIIPc(SAr)- has also been ruled out. All of the kinetic data, including Hammett’s F ) −2.3
0.1, N-donor
) dianion of phthalocyanine) with thiophenoxides in DMA
−
−
−
−
−
±
inhibition, and alkyl group effect, Me > Et, indicate that the reaction is a normal SN2 methyl transfer, only very fast.
Methyl transfer to aliphatic thiolates is also rapid and follows the same SN2 mechanism. Exceptional methyl-
transfer reactivity of the phthalocyanine model sharply contrasting with the inertness of methylcobaloxime is explained.
Introduction
al.,6 claimed virtually complete inertness of the Co-C bond
toward nucleophilic attack. Isolation of thiolate adducts with
methylcobaloxime was possible with virtually no loss of the
cobalt-bonded methyl group.6
Thiolates are the most common methyl acceptors in
enzymatic methyl transfers. Methionine synthase catalyzed
methyl transfer from methylcobalamin (MeCbl)1 to ho-
mocysteine has been the most studied enzymatic methyl-
transfer reaction for decades.2 This reaction has also been
referred to as “difficult”, since in the absence of the protein,
it proceeds extremely slowly. As shown by Norris and Pratt,
methylcobinamide (MeCbi+) is a substantially more effective
methyl donor than MeCbl.3 However, most of the enzymatic
enhancement occurs in the homocysteine-binding module of
the enzyme, and mechanisms other than the initially proposed
SN2 have also been recently discussed.2 Methyl transfer from
the leading MeCbl model, methylcobaloxime, to thiolates
has been controversial. While Schrauzer and Stadlauber
reported some kinetic data,4 especially for the BF2-containing
methylcobaloxime, Brown and Kallen5 and later Marzilli et
Recent studies on axial base inhibition of methyl transfer
from methylcobalt(III) phthalocyanine (CH3CoIIIPc) suggest
that any trans axial ligation results in alkyl-transfer inactivity
of the organocobalt(III) substrate.7 (Examples of heterolytic
Co-C bond cleavage in six-coordinate complexes such as
8
AdoCbl(CN)- cannot be classified as alkyl (formally
cation) transfers.) Alkanethiolates do not coordinate to the
cobalt center in methylcorrinoids.3,9,10 This certainly makes
the demethylation of MeCbi+ easier. In contrast, methyl-
cobaloximes and methylated Costa-type models show very
strong demand for six-coordination.11 These electron-
(5) Brown, K. L.; Kallen, R. G. J. Am. Chem. Soc. 1972, 94, 1894.
(6) Polson, S. M.; Hansen, L.; Marzilli, L. G. Inorg. Chem. 1997, 36,
307.
(7) Galezowski, W. Inorg. Chem. 2005, 44, 1530.
(8) Brasch, N. E.; Haupt, R. J. Inorg. Chem. 2000, 39, 5469.
(9) Hogenkamp, H. P. C.; Bratt, G. T.; Kotchevar, A. T. Biochemistry
1987, 26, 4723.
(10) (a) Garr, C. D.; Sirovatka, J. M.; Finke, R. G. J. Am. Chem. Soc.
1996, 118, 11142. (b) Garr, C. D.; Sirovatka, J. M.; Finke, R. G. Inorg.
Chem. 1996, 35, 5912.
* E-mail: wlodgal@amu.edu.pl
(1) Abbreviations: Pc, dianion of phthalocyanine; MeCbl, methylcobal-
amin; MeCbi+, methylcobinamide; Ado, adenosyl; DMA, dimethyl-
acetamide; DMSO, dimethyl sulfoxide; DBU, 1,8-diazabicyclo[5.4.0]-
undec-7-ene; TBAP, tetrabutylammonium perchlorate; TPP, dianion
of 5,10,15,20-tetraphenylporphyrin.
(2) Matthews, R. G. Acc. Chem. Res. 2001, 34, 681.
(3) Norris, P. R.; Pratt, J. M. BioFactors 1996, 5, 240.
(4) Schrauzer, G. N.; Stadlauber, E. A. Bioinorg. Chem. 1974, 3, 353.
(11) Marzilli, L. G.; Summers, M. F.; Bresciani-Pahor, N.; Zangrando, E.;
Charland, J.-P.; Randaccio, L. J. Am. Chem. Soc. 1985, 107, 6880
and references therein.
10.1021/ic0503378 CCC: $30.25
Published on Web 06/25/2005
© 2005 American Chemical Society
Inorganic Chemistry, Vol. 44, No. 15, 2005 5483

Galezowski
Scheme 1. Methyl Transfer from CH3CoIIIPc to Thiolates (SN2
been reported yet, is of interest, as well as possible changes
in this spectrum with varying electron-donation properties
of the thiolate. Remote substituent effect on the reaction
spectra and the rates should provide a deeper insight into
the mechanism, and perhaps some of the mechanistic
alternatives to SN2 can be ruled out. It seems worthwhile to
study whether the alkyl group effect, Me > Et, and N-donor
inhibition observed for the dealkylations of RCoIIIPc with
variety of nucleophiles7 will also hold for the thiolate
reaction.
Mechanism)a
The methyl-transfer reactions from CH3CoIIIPc to thiolates
bear no immediate relevance to biological systems, where
the equatorial system is different, but they are probably the
most rapid nonenzymatic methyl transfers from methyl-
cobalt(III) known. It is of importance to gain more confi-
dence that these fast reactions, in fact as fast as enzymatic
methyl transfer from MeCbl to homocysteine,17 really are
SN2 processes.
a In this scheme, no distinction is made between CH3CoIIIPc, possibly
the actual reactive species, and inactive CH3CoIIIPc(DMA). The two forms
exist in rapid equilibrium.
Experimental Section
+
deficient complexes are not as good CH3 donors as
expected. An intermediate strength Lewis acid, CH3CoIII-
Pc,7 showed, in turn, an outstanding methyl-transfer ability
toward benzenethiolate,12 despite postulated extensive liga-
tion of the substrate (Scheme 1). The methyl transfer is so
fast that the thiophenoxide adduct cannot be isolated and
conventional ligand binding studies are not viable.
In accordance with other binding nucleophiles,7 atypical
(for SN2) saturation-type concentration dependencies of the
rate constants are observed in the thiophenoxide reaction.
Similar saturation-type kinetics caused by strong affinity of
thiolates for the cobalt center in phthalocyanine complexes
has also been seen in the reactions of Co(II) tetrasulfoph-
thalocyanine with thiolates in water13 and DMF.14 This is
likely to be observed no matter if the adduct or ligand-free
substrate is the reactive species.
In view of the inertness of MeCbi+ and methylcobaloxime
toward alkanethiolates, one may also suspect that fast methyl-
transfer rates between CH3CoIIIPc and benzenethiolate result
from some specific features of aromatic thiolates. Displace-
ment reactions promoted by aromatic thianions are usually
faster than those of aliphatic thiolates of similar basicity,
which used to be explained either by favorable interaction
of the aromatic ring with the substrate or by the fact that
aromatic thiolates are “softer”.15 As thiolates are not only
excellent nucleophiles but also good electron donors, mech-
anisms involving outer- or inner-sphere electron transfer
should also be considered.16
Rapid rates for demethylation of the phthalocyanine model
may seem to present a wild exception among extremely slow
thiolate reactions of MeB12 and related complexes. Therefore,
a more detailed study of this system seems warranted. The
spectrum of the transient CH3CoIIIPc(SPh)-, which has not
Materials. CH3CoIIIPc and CH3CH2CoIIIPc were prepared as
described earlier,7 and their solutions were handled in dim light.
All of the chemicals were purchased from Aldrich. Dimethyl-
acetamide (DMA) was purified and stored as described elsewhere.12
Liquid thiols were freshly vacuum distilled and kept under argon.
4-Nitrobenzenethiol was sublimed under reduced pressure, which
substantially improved its purity. In conformity with the earlier
study,12 sodium benzenethiolates were used. The salts were prepared
as described earlier for sodium thiophenoxide.12 N-Methylimidazole
(N-MeIm) (double distilled) was used without further purification.
Rate Measurements. The rates were determined spectrophoto-
metrically in excess thiolate. All of the stopped-flow runs were
performed on an Applied Photophysics SX.17 MV instrument under
anaerobic conditions. Before measurements, the circulating liquid,
in which the drive syringes were immersed, was deoxygenated with
argon, then a portion of sodium thiosulfate/Tris buffer was added.
An aqueous solution of the same mixture was passed through the
instrument before measurements to remove possible traces of
oxidants and followed by rinsing with deoxygenated DMA. Extreme
care was exercised to avoid air oxidation of dilute thiolate solutions,
which were always freshly prepared under argon and anaerobically
transferred to the drive syringes. Effectiveness of these measures
was tested by using thiophenoxide dissolved in sodium borohydride
(0.0005 M) solution, which should eliminate disulfides. The rates
were similar to those obtained in the absence of borohydride. CoIPc-
was formed regardless of the presence of borohydride. When a
0.0005 M sodium borohydride solution is combined with CH3CoIII-
Pc in a stopped flow, a relatively slow demethylation is observed
which could, in part, result from photochemical cleavage of the
Co-C bond. (Solutions of concentrated mixtures of CH3CoIIIPc
and CoIPc- in DMSO-d6 in the presence of sodium borohydride
are stable in the dark for a prolonged period of time.)
All of the kinetic measurements were performed under pseudo-
first-order conditions in excess nucleophile and/or ligand, if
applicable. The concentration of the organocobalt complex was ∼1
× 10-5 M. In regular kinetic measurements, the rates were measured
mainly at two wavelengths, 660 nm (substrate decay) and 700 nm
(the absorption band of the CoIPc- product). The rates for the slower
second (alkyl transfer) step measured at 660 nm were larger than
(12) Galezowski, W.; Lewis, E. S. J. Phys. Org. Chem. 1994, 7, 90.
(13) Leung, P. S.; Hoffmann, M. R. J. Phys. Chem. 1989, 93, 434.
(14) Tyapochkin, E. M.; Kozliak, E. I. J. Porphyrins Phthalocyanines 2001,
5, 405.
(15) Freter, R.; Pohl, E. R.; Wilson, J. M.; Hupe, D. J. J. Org. Chem. 1979,
44, 1771.
(16) Lewis, E. S. J. Am. Chem. Soc. 1989, 111, 7576 and references therein.
(17) Bandarian, V.; Matthews, R. G. Biochemistry 2001, 40, 5056.
5484 Inorganic Chemistry, Vol. 44, No. 15, 2005

Methyl Transfer from CH3CoIIIPc to Thiophenoxides
Table 1. Yields of Light Hydrocarbon Products of the Reactions of
those determined at 700 nm, unless the illuminating light intensity
had been appropriately reduced. Photolability of CH3CoIIIPc has
been noted earlier. In dilute solutions of basic thiophenoxide, double
exponential curves were observed. At [thiolate] > 0.002 M, the
first rapid step was beyond the stopped-flow range. (See examples
of kinetic curves and end-point plots in Supporting Information).
The observed rate constants were found from nonlinear least-squares
fits using the Applied Photophysics software. Double-exponential
fits were not necessary, because the time scales for the two
processes, rapid ligand binding and slower alkyl transfer, were so
different.
RCoIIIPc with Thiophenoxides and Bromide in DMA
yield (%)
R
nucleophile
PhSNa
CH4
CH2dCH2
CH3CH3
CH3
CH3CH2
CH3
CH3CH2
CH3
0.07
0.28
1.8
0.6
0.5
0.34
0.04
0.2
0.2
0.08
0.03
PhSNa
traces
0.003
traces
0.04
2,6-Cl2C6H3SNa
2,6-Cl2C6H3SNa
Bu4NBra
a 0.03 M, 24 h, at room temperature.
The formation constant, K, was determined from the concentra-
tion dependence of the end-point, A, of the fast step (absorbance
drop at 660 nm) as a ratio intercept/slope of the linear dependence
of 1/(Ao - A) vs 1/[thiolate], where Ao is the absorbance of CH3-
CoIIIPc with no thiolate present (see Figure S6, Supporting
Information). Point-by-point reaction spectra were obtained from
kinetic runs. The wavelength was changed every 2 nm in the region
of the absorption bands, especially around the sharp Q-band (about
660 nm). Reproducibility of kinetic runs acquired on the Applied
Photophysics instrument was fairly good as witnessed by the
transient spectra determined for a reaction time as short as 10 ms.
Absorbance of thiolates obscured the spectral region below 360
nm. Red solutions of 4-nitrobenzenethiolate did not allow for the
determination of the reaction spectra at wavelengths <600 nm. The
700 nm band of CoIPc- formed in the 4-nitrobenzenethiolate
reaction disappeared in a slow process, which was, however,
considerably faster in concentrated solutions of the nucleophile and
would complicate determination of the rate for methyl transfer at
concentrations >0.02 M. A similar slow process was also observed
in concentrated solutions of 3,4-dichlorothiophenoxide, yielding a
mixed spectrum of CoIPc- and probably CoIIPc(3,4-Cl2C6H3S)
Conductances of sodium salts of thiophenoxides were measured
as described elsewhere,7 with the exclusion of air.
Product Studies. The reactions of CH3CoIIIPc or CH3CH2CoIII-
Pc with thiophenoxides always gave CoIPc- and thioethers, in good
yield, within seconds. The thioether products were identified either
by GC as described earlier for thioanisole12 or by 1H NMR spectra
in a DMSO-d6 solution. For selected thiolates, the gas phase over
solution was GC analyzed for light hydrocarbons as described
elsewhere.7 An example of such an assay is given below, and the
results are collected in Table 1.
CH3CH2CoIIIPc with PhSNa in DMA. A solution of CH3CH2-
CoIIIPc (12 mg, 0.02 mmol) and PhSNa (60 mg, 0.45 mmol) in 10
mL of DMA was sealed in a vial under argon in the dark. The
solution turned yellow-green. GC analysis of the gas phase7 gave
ethylene (1.8%) and ethane (0.2%).
Results and Discussion
Cyclic Voltammograms. Cyclic voltammograms of CH3-
CoIIIPc in DMA are similar to those measured in DMF.7 The
only difference of interest is the behavior of the first
reduction peak at -1.42 V, (Epc - Epa)/2, vs Fc+/Fc couple
(Figure 1), which at a scan rate of 0.1 V/s is irreversible but
at about 10 V/s (20 ( 1 °C) reaches quasi-reversibility (Epc
- Epa > 60 mV). A new oxidation peak, the potential of
which agrees with that for the oxidation of CoIPc-, is
observed on reverse sweep only when CH3CoIIIPc has been
reduced. This peak is most pronounced at intermediate scan
rates. Analysis of voltammograms measured at varied scan
rates20 gives a rate constant of 2.8 ( 0.6 s-1 at 20 ( 1 °C
for unimolecular Co-C bond cleavage in [CH3CoPc]-,
yielding CoIPc- and presumably a methyl radical.21 At
ambient temperatures and scan rates up to 50 V/s, the first
reduction of CH3CH2CoIIIPc still remains irreversible. No
convincing explanation for the larger stability of [CH3CoPc]-
in DMA than in DMF7 is offered.
(λ
) 672 nm and a 530 nm charge-transfer band); the latter
max
was identical with the spectrum of the product of CoIIPc with 3,4-
Cl2C6H3SNa in air. These slow secondary reactions did not interfere
with the determination of the methyl transfer rates and were not
further studied.
In pseudo-first-order conditions, there is always a possibility that
some other nucleophiles exist in solution, either impurities or
possible solvolysis products, but none of them could compete with
thiolates, especially in view of the high yields of thioethers. Pseudo-
first-order conditions are favorable here, with potential impurities
less reactive than thiolates. An important point is that in neutral
solution, RCoIIIPc complexes are stable in the presence of disulfides
or thiols at least for an hour. Disulfides, even in trace quantities,
according to Hogenkamp,18 can react with cob(I)alamin to give the
Co(II) complex and thiolate ion. In the phthalocyanine system, in
excess of strongly basic thiolates, this could not be observed since
CoIIPc is readily reduced by them. With less basic 4-NO2-
thiophenoxide, a slow follow-up oxidation is possible (see above).
The major concern was the possible loss of thiolate. The control
runs with borohydride showed that oxidation was not a substantial
problem. Occasionally, DBU was added to thiophenoxide solutions
to revert possible protolysis of the thiolate by traces of water, but
this had little effect on the rates.
The increased stability of [CH3CoPc]- compared with one-
electron reduced methylcorrinoids, which give reversible
voltammograms only at low temperatures and high sweep
(19) Lever, A. B. P.; Milaeva, E. R.; Speier, G. In Phthalocyanines,
Properties and Applications; Leznoff, C. C., Lever A. B. P., Eds.;
Wiley-VCH: New York, 1993; Vol 3, p 8.
Kinetics of cyanide binding by CH3CoIIIPc was measured by
monitoring the rise in absorbance at 508 nm. Teterabutylammonium
cyanide was used.
Cyclic voltammetry was carried out as described elsewhere.7
The potentials were internally referenced to Fc+/Fc couple (+0.47
V vs SCE in DMA).19
(20) Heinze, J. Angew. Chem., Int. Ed. Engl. 1984, 23, 831.
(21) Interestingly, a rate constant of 3 ( 1 s-1 was measured for a
unimolecular decomposition of transient species formed in the reaction
of Co(II) tetrasulfophthalocyanine, CoII(TSPc)4-, with methyl radicals
by Sorek, Y.; Cohen, H.; Meyerstein, D. J. Chem. Soc., Faraday Trans.
I 1989, 85, 1169. This rate constant was tentatively ascribed to the
homolytic decay of [CH3CoIII(TSPc)]4-, which in view of infinite
stability of CH3CoIIIPc is extremely unlikely. However, if one-electron-
reduced [CH3CoIII(TSPc)]4- was the transient, the rate would be
consistent with the findings of the present work.
(18) Hogenkamp, H. P. C.; Bratt, G. T.; Sun, S. Biochemistry 1985, 24,
6428.
Inorganic Chemistry, Vol. 44, No. 15, 2005 5485

Galezowski
Scheme 2. Thermodynamic Cycle Showing the Relationship between
Co-C Bond Dissociation Free Energies in an Alkylcobalt(III) Complex
and Corresponding Anion Radical
anion radical. The shift in reduction potentials between
MeCbl and cob(II)alamin, 0.75 V (Save´ant’s data), is larger
than that of ∼0.6 V for the phthalocyanine or TPP model, a
difference that corresponds to a noticeable 3.5 kcal/mol
lowering of the free energy barrier, which would be,
however, canceled by the larger ∆GBDq value for homolysis
of MeCbl27 than CH3Co(TPP)28 or CH3CoPc. Perhaps the
above analysis fails to explain the lability of the MeCbl anion
radical because of crude assumptions, uncertainty in redox
potentials measured in different solvents, or differences in
trapping conditions27 and trans ligation.
One important stabilizing factor may be the delocalization
of the extra charge in the one-electron-reduced species. We
could speculate that large, conjugated equatorial ligands such
as phthalocyanine should diminish excess electron density
in the axial system more efficiently than corrin. However,
recently, Brunold’s DFT calculations29 cast doubt on the
belief27,30 that the lowest unoccupied molecular orbital
(LUMO) in alkylcorrinoids has a substantial Co-C anti-
bonding character. The fact that the LUMO has negligible
contribution from Co atomic orbitals as well as those of the
axial alkyl group agrees with the tendency of CH3Co(TPP)
to form a ligand-centered anion radical rather than the
Co(II) complex.24 Nevertheless, possible correlation with the
size of the equatorial conjugated arrangement is still worth
consideration, because activation energy of the Co-C bond
cleavage in the anion radical may be correlated to an
intramolecular charge transfer from the ligand-centered π
orbital, occupied by the odd electron, to a higher energy
Co-C antibonding orbital. Since, in systems with enlarged,
planar, conjugated systems energy of the former orbital will
be lowered, relative to that of the latter, faster decay of one-
electron reduced alkylcorrinoids than their phthalocyanine
or TPP congeners could be accounted for.
Figure 1. Cyclic voltammograms for CH3CoIIIPc in DMA. Potentials vs
Fc+/Fc couple measured at a platinum electrode, [CH3CoIIIPc] ) 2 × 10-4
M and [TBAP] ) 0.2 M.
rates,22 seems to be consistent with electron deficiency of
the phthalocyanine model. The first reduction potentials for
MeCbl are 0.65,22 0.55, or 0.29 V (base-off MeCbl)23 more
negative than that for CH3CoPc. However, no simple
correlation with the reduction potentials exists, even among
porphyrinic models. For instance, the anion radical of CH3-
Co(TPP) is quite stable in CH2Cl2 at room temperature, scan
25
rate 0.1 V/s,24 but not in DMSO; it is more stable than
[CH3CoPc]-, characterized by a ∼0.4 V less negative redox
potential. Even though it is intuitively appealing that Co-C
bonds in easier reduced complexes will be more stable, the
reduction potential for RCoIII(chel) itself cannot be directly
related to Co-C bond cleavage. As follows from a thermo-
dynamic cycle depicted in Scheme 2, the lowering of the
free energy of Co-C bond dissociation in the anion radical,
∆GBD′, relative to ∆GBD for the parent complex, is propor-
tional to the gap in the reduction potentials (eq 1, where F
is the Faraday constant).
∆GBD′ ) ∆GBD + F(Eo
- - Eo
)
Co(II)/Co(I)
(1)
Reaction Spectra. The reaction spectra shown in Figure
2 are representative of all of the strongly basic thiophenoxides
(4-CH3O, 4-CH3, H, and 4-Cl derivatives). Within mil-
liseconds, an equilibrium between the substrate and the
transient is established (Scheme 1). The amplitude of the
RCo(III)/[RCo]
q 26
If we replaced ∆GBD′ and ∆GBD with ∆GBDq′ and ∆GBD
,
respectively, which for the two analogous endoergic pro-
cesses may not be an unreasonable approximation, eq 1
would not predict any conspicuous lability of the MeCbl
(27) Martin, B. D.; Finke, R. G. J. Am. Chem. Soc. 1992, 114, 585.
(28) Fukuzumi, S.; Miyamoto, K.; Suenobu, T.; Van Caemelbecke, E.;
Kadish, K. M. J. Am. Chem. Soc. 1998, 120, 2880.
(29) Stich, T. A.; Brooks, A. J.; Buan, N. R.; Brunold, T. C. J. Am. Chem.
Soc. 2003, 125, 5897.
(30) (a) Rubinson, K. A.; Itabashi, E.; Mark, H. B. Inorg. Chem. 1982, 21,
3571. (b) Rubinson, K. A.; Parekh, H. V.; Itabashi, E.; Mark, H. B.
Inorg. Chem. 1983, 22, 458.
(22) Lexa, D.; Save´ant, J.-M. J. Am. Chem. Soc. 1978, 100, 3220.
(23) Kim, M.-H.; Birke, R. L. J. Electroanal. Chem. 1983, 144, 331.
(24) Kadish, K. M.; Han, B. C.; Endo, A. Inorg. Chem. 1991, 30, 4502.
(25) Fauvet, M. P.; Gaudemer, A.; Bouckly, P.; Devynck, J. J. Organomet.
Chem. 1976, 120, 439.
(26) In fact, we assume here only that ∆GBDq - ∆GBD ) ∆GBDq′ - ∆GBD′,
that is, equal free energy barriers for reverse reactions.
5486 Inorganic Chemistry, Vol. 44, No. 15, 2005

Methyl Transfer from CH3CoIIIPc to Thiophenoxides
Figure 4. Representative concentration dependencies of kobs for the methyl-
transfer step of the reactions of CH3CoIIIPc with thiophenoxides in DMA
(25 °C): 2,6-dichloro- (triangles), 4-chloro- (squares), and 3,4-dichloro-
thiophenoxide (dots).
Figure 2. Reaction spectra of CH3CoIIIPc with 4-CH3C6H4SNa in DMA
25 °C for 0, 0.0125, 0.1, 0.2, 0.5, 1, and 5 s, [4-CH3C6H4SNa] ) 0.005 M
and [CH3CoIIIPc] ) 1 × 10-5 M.
Scheme 3. Outer-Sphere Electron-Transfer Mechanism for
Demethylation of CH3CoIIIPc by Thiolates
Q-band in the spectrum of CH3CoIIIPc(2,6-Cl2) must be
similar to that of CH3CoIIIPc(DMA), a situation previously
observed in DMA with N-donor axial bases.7 It seems
plausible that with strongly electron-withdrawing substituents
on the phenyl ring, the spectra of the adducts are “normal”,
that is, similar to those of the other CH3CoIIIPc(L) complexes.
However, as the substituent becomes increasingly electron
donating the Q-band loses its maximum intensity and
broadens, possibly because of a charge transfer within the
complex. Finally, the transient spectra characterized by broad
Q-bands might be those of [CH3CoPc]- anion radical
produced via OSET (Scheme 3). This, however, in view of
large differences in redox potentials (>0.9 V) between
thiophenoxides33 and RCoIIIPc appears unlikely.
Concentration dependencies of the rates for alkyl-transfer
reactions exhibit saturation (Figure 4). A trivial explanation
for this would be an ion association of sodium thiophen-
oxides, provided the ion pairs were inactive or at least far
less reactive than free ions. Ion association of sodium
thiophenoxide in four solvents, including polar aprotic DMF,
has been studied by Westaway et al.34 Different reactivity
of free thiophenoxide ions and solvent-separated ion pairs
toward n-butyl chloride was found with free ions only 10-
20% more reactive than solvent-separated ion pairs. Such a
moderate difference would not account for the strong
curvatures seen in Figure 4. Furthermore, in DMF, which is
Figure 3. Reaction spectra of CH3CoIIIPc with 2,6-Cl2C6H3SNa in DMA
25 °C for 0 (thick line), 0.0125, 0.25, 0.5, 1, 2, and 5 s, [2,6-Cl2C6H3SNa]
) 0.005 M and [CH3CoIIIPc] ) 1 × 10-5 M.
spectral change accompanying the initial fast reaction is
thianion concentration dependent, and the rates are first order
in the thianion (see Tables S1-S5 in Supporting Informa-
tion). Visible spectra of the transient species do not vary
significantly as the remote substituent is changed from the
electron-donating 4-CH3O to the moderately electron-
withdrawing 4-Cl. Both the Q-band at 660 nm and the ∼600
nm overtone are very broad (Figure 2). There is no marked
metal to (equatorial) ligand charge-transfer band, associated
with the CoI oxidation state, in the spectra of the transient
species. These spectra are, however, unlike those of the other
RCoIIIPc(L) complexes.7 As a matter of fact, they resemble
the spectrum of CoIIPc(DMA), only the Q-band appears less
intense at its maximum and broader. The spectra of the
follow-up process exhibit three clean isosbestic points, thus
indicating there are no further mechanistic complications.
Regardless of the basicity of the thiolate, the final spectrum
is always that of CoIPc-.
In accordance with our earlier observations,12 the 684 nm
isosbestic point is different from that for CH3CoIIIPc and
CoIPc- at 678 nm (Figure 2). Also, two other isosbestic
points do not include the spectrum of the substrate. By
contrast, with the less basic and more sterically hindered 2,6-
dichlorothiophenoxide, the initial change in 660 nm absor-
bance is small (Figure 3). The Q-band intensity initially even
slightly rises with a cosmetic red shift. The spectrum is not
(31) The substituted thiophenoxides will be hereafter labeled by bold
designations of the substituents.
(32) The [CH3CoIIIPc(2,6-Cl2)-]/[CH3CoIIIPc] ratio, calculated for [2,6-
Cl2] ) 0.005 M using the K value of Table 2, is 0.41.
(33) (a) Andrieux, C. P.; Hapiot, P.; Pinson, J.; Save´ant, J.-M. J. Am. Chem.
Soc. 1993, 115, 7783. (b) Larsen, A. G.; Holm, A. H.; Roberson, M.;
Daasbjerg, K. J. Am. Chem. Soc. 2001, 123, 1723.
31
exactly that of the 2,6-Cl2 complex, since at 0.005 M
thiolate CH3CoIIIPc(DMA) prevails,32 but it is clear that the
(34) Westaway, K. C.; Lai, Z. G. Can. J. Chem. 1988, 66, 1263.
Inorganic Chemistry, Vol. 44, No. 15, 2005 5487

Galezowski
Table 2. Kinetic Data for the Reaction of CH3CoIIIPc with Thiophenoxides in DMA (25 °C)
methyl transfer
ligand binding
b
f
h
substituent
k2, M-1 s-1 a
K, M-1 a
k2/K, s-1
10-6k1, M-1 s-1
K, M-1 g
k-1, s-1
4-NO2
2,6-Cl2
3,4-Cl2
4-Cl
81 ( 2
255 ( 4
750 ( 40
3100 ( 200
20000 ( 1000
17900c
70 ( 10e
77 ( 7
2100 ( 200
2600 ( 100
10100 ( 700
9400c
1.1 ( 0.1e
3.7 ( 0.4
0.36 ( 0.01
1.19 ( 0.03
1.97 ( 0.03
1.91 ( 0.02c
1.89 ( 0.02
1.06 ( 0.03
1.2 ( 0.2
1700 ( 300
2400 ( 100
700 ( 100
330 ( 20
65 ( 6
0.79 ( 0.04
0.84 ( 0.08
H
13000 ( 2000
4-Me
4-MeO
43000 ( 5000
23000 ( 3000
0.79 ( 0.03
0.9 ( 0.1
27000 ( 3000
53000 ( 9000
29 ( 1
17 ( 2
d
d
a Determined by eq 2. b Saturated kobs c Results from ref 12. d Not determined, because kinetic saturation was achieved at too low a concentration of the
.
nucleophile. e These values from a nonlinear least-squares fit. f The rate constant for presumed thiolate binding to Co. g Stability constant for CH3CoIIIPc(SR)-,
estimated from concentration dependence of a pre-methyl-transfer drop in 660 nm absorbance. h Rate constant for thiolate liberation from CH3CoIIIPc(SR)-,
calculated from k1 and K.
in many respects similar to DMA, ion association was visible
for concentrations above 7 × 10-3 M. An interaction that
weak could not disturb this kinetic study even if free ions
were considerably more reactive than ion pairs. Nevertheless,
conductivity of sodium thiophenoxide and sodium 4-chlo-
rothiophenoxide in DMA was studied over the concentration
ranges used in the stopped-flow runs, and no indication of
ion association was detected. With the two selected thiophe-
noxides, the kobs values are close to saturation above 1 ×
10-3 M, while the conductances grow almost linearly at least
1
up to 1 × 10-2 M. The Kohlrausch plots of Λ vs c / gave
2
Λï values of 42.4 ( 0.3 and 43.7 ( 0.6 Ω-1 cm2 mol-1 at
25 °C for C6H5SNa and 4-Cl-C6H4SNa, respectively, and
the slopes close to the theoretical values (see the plots in
Supporting Information).
Because of the instability of dilute thiophenoxide solutions,
saturated kobs values (k2/K in Table 2) are clearly the most
accurately determined kinetic parameters for the reactions
promoted by the most basic thiophenoxides. The observed
concentration dependence of the rates (Figure 4) is consistent
with the previously proposed mechanism involving a prior
equilibrium leading into a cul-de-sac (Scheme 1). The
second-order rate constant for methyl transfer, k2, and
formation constant for the inert CH3CoIIIPc(SR) complex,
K, can be found from a double reciprocal plot,7 eq 2 (Figure
5B).
Figure 5. Kinetics of CH3CoIIIPc with 4-NO2C6H4SNa in DMA (25 °C)
(A) and a double-reciprocal plot, eq 2 (B). Error bars are at 2σ.
reaction is consistent with significant steric hindrance to
ligand binding by the cobalt center.7 Furthermore, if the
reaction indeed were to follow an OSET mechanism, the
fast rates for assumed uphill electron-transfer reactions, k1
in Scheme 3 and in Table 2, would be unreasonably
insensitive to electronic factors. It is worth nothing that the
rate for decay of the electrochemically generated [CH3CoPc]-
is (incidentally) close to the k2/K values. The difference in
redox potentials between CH3CoIIIPc/[CH3CoPc]- and RS•/
RS- couples,33 >0.9 V, is too large for an OSET mechanism,
especially with the electron-withdrawing substituents onto
thiophenoxide.
Given the versatile chemistry of thiolates, the possibility
that the demethylation follows a mechanism in which the
thiolato adduct is the reactive species (Scheme 4) should also
be considered, although with other nucleophiles the six-
coordinate species proved to be inactive.7 The rate-limiting
homolytic cleavage of the Co-C bond in CH3CoPc(SR)-
would be seemingly consistent with observed concentration
dependencies of the rates. The rate constant for the homolytic
fission step would again be represented by the k2/K values
1
kobs
1
K
k2
)
+
(2)
k2[RS-]
The relatively small variance of the saturated kobs (the k2/K
column in Table 2) is conspicuous. While it could raise
suspicions that mechanisms other than SN2 are involved, it
can also be explained by almost proportional changes of k2
and K. Besides, the apparent invariance cannot be readily
accommodated into other mechanisms as well. An OSET
mechanism (Scheme 3) predicts that the demethylation rates
for all of the thiolates be equal at infinite thiolate concentra-
tion (saturation) and similar to the rate of [CH3CoPc]- decay
obtained from cyclic voltammetry. It would also explain the
atypical transient spectra, which are similar over a range of
various substituents. However, a 10-fold change in k2/K
between 2,6-Cl2 and 3,4-Cl2 reactions does not agree with
this mechanism, whereas the large value for the 2,6-Cl2
5488 Inorganic Chemistry, Vol. 44, No. 15, 2005

Methyl Transfer from CH3CoIIIPc to Thiophenoxides
is less plausible because CoIIPc(SR)- is unstable. Dispro-
portionation of free ethyl radicals is unlikely, since it would
give approximately equimolar amounts of ethane and eth-
ylene.
Scheme 4. Homolytic Cleavage of CH3CoPc(SR)-
Methane in a few percent yield could indicate a competi-
tion from H donors present in the solvent for methyl radicals
formed by the mechanisms of Schemes 3 or 4 but ethylene
as the main gaseous byproduct of the CH3CoIIIPc reactions,40
and surprisingly low yields of methane, whereas thermolysis
in DMA gives methane in >90% yield,38 is incompatible
with this proposal. Faced with a methane/ethane ratio of 50
observed on the thermolysis,38 sharply contrasting with the
ratio of 0.015 for the thiophenoxide assay (Table 1), we are
forced to state that, in the latter case, ethane could not be
formed by the self-trapping of methyl radicals. The results
of product studies, especially for methyl system, do not
support a significant contribution of homolytic mechanisms
to the overall rate of disappearance of the organocobalt
substrate. The distribution of hydrocarbon products is also
incompatible with chain mechanisms initiated by the reaction
of Scheme 4, where hydrocarbons would be formed either
on self-trapping chain termination or on scavenging of alkyl
radicals by H donors. Other reasons to refute chain mech-
anisms have been given earlier.12 Now, it can be also argued
that, because of the presence of H donors in DMA, the
chains, if any, would not be long, as witnessed by the high
methane/ethane ratios obtained on the thermolysis of CH3-
CoPc. Because of the large size of the phthalocyanine planar
of Table 2. The Co-C bond rupture is probably not fast
unless charge transfer to the antibonding σ* orbital along
this bond occurs,27 which would depend, in turn, on the
electron donation properties of substituents onto the trans
axial ligand. Strong acceptor groups such as NO2 are known
for their significant bond strengthening/radical destabilizing
effect.35 Hence, similar k2/K values for 4-NO2 and 4-MeO
reactions are at odds with the homolytic mechanism.
Small but detectable amounts of light hydrocarbons (Table
1) indicate that some mechanisms other than SN2 , either
concurrent or follow up, must be operating. It is noteworthy
that somewhat larger hydrocarbon yields were found in
product studies with other nucleophiles such as phosphines
or thiocyanate.7 Thus, with this respect, thiolate reactions,
in which transients of atypical spectra are observed, are no
different from dealkylations promoted by other nucleo-
philes,7,36 which are usually not suspected of diversified
chemistry. In the CH3CH2CoIIIPc system, if ethylene is
formed by a concerted â-elimination, the concurrent reaction
can be neglected, since the organocobalt substrate decay
would follow the main SN2 path in at least 98%. The ethylene
could also be formed by a cage reaction within the
•
system, cage recombination of CH3 and thiyl radicals
(Scheme 4) is unlikely.12 Hence, large yields of thioethers
formed in a moderately trapping solvent DMA are in support
of a simple SN2 mechanism.
In our original paper,12 considerations of a typical SN2
behavior of the reactions of CoIPc- with alkyl halides, in
terms of the microreversibility principle, led us to indirect
evidence for SN2 mechanism of the demethylation of CH3-
CoPc. It was also argued that the ratio of PhS-/I- with CH3-
CoPc is similar to those observed with other methyl donors
(in aprotic solvents). An important point is that, contrary to
the notion that methylcobalt complexes are inert, CH3CoIII-
Pc is a reasonably good methyl donor in aprotic solvents,
merely an order of magnitude slower than CH3I. The CH3I/
CH3CoPc second-order rate constant ratio for methyl transfer
to PPh3 is 1.2 × 10-2/4.8 × 10-4 ≈ 25 (in propylene
carbonate41 and DMA,7 respectively), while for the iodide
reaction in DMA the ratio is equal 11.8/0.492 ≈ 24.42 Rapid
demethylation of CH3I by thiophenoxide at 43 °C in
sulfolane43 is three times faster than the reaction of CH3-
CoIIIPc with thiophenoxide in DMA (Table 2). The rate for
the latter reaction clearly falls into the range of expected
values, especially if the fact that typical methyl transfers in
sulfolane are an order of magnitude slower than those in
•
CH3CH2 ,CoIIPc(SR)- geminate radical pair,37 which is the
possible intermediate in the mechanism of Scheme 4. If we
assume that the presence of a trans-thiolato ligand does not
qualitatively change the nature of the homolytic process, and
disregard the 40-50 °C difference in temperature ranges
between the thermolysis and thiolate reaction studies, then
judging by 21% and ∼70% yields of ethylene found on
thermolysis of CH3CH2CoIIIPc in DMA38 in the presence and
in the absence of radical trap TEMPO, respectively, the upper
limit of the homolytic process in the unsubstituted thiophe-
noxide reaction can be roughly estimated at 9 or 3%. The
former estimate (1.8% of Table 1 multiplied by 100 divided
by 21) is based on the assumption that there is no
recombination of free ethyl radicals with the cobalt center;
that is, they are all intercepted by H-donors, thiyl radicals,
or trace disulfide impurities or they dimerize.39 The latter
value is obtained when we allow recombination to the extent
observed in thermolysis (with no trap added). This estimate
(35) (a) Bordwell, F. G.; Cheng, J.-P. J. Am. Chem. Soc. 1991, 113, 1736.
(b) Bordwell, F. G.; Zhang, X.-M.; Satish, A. V.; Cheng, J.-P. J. Am.
Chem. Soc. 1994, 116, 6605. (c) Chandra, A. K.; Nam, P.-C.; Nguyen,
M. T. J. Phys. Chem. A 2003, 107, 9182.
(36) Note that some of these nucleophiles, especially triphenylphosphine
and bromide, are very weakly bound, in contrast to thiophenoxides.
(37) Halpern, J. Acc. Chem. Res. 1982, 15, 238.
(40) Note that ethylene appears to be the main hydrocarbon byproduct of
the reaction of CH3CoIIIPc with bromide (Table 1).
(41) McCortney, B. A.; Jacobson, B. M.; Vreeke, M.; Lewis, E. S. J. Am.
Chem. Soc. 1990, 112, 3354.
(42) Galezowski, W.; Ibrahim, P. N.; Lewis, E. S. J. Am. Chem. Soc. 1993,
115, 8660.
(38) Galezowski, W.; Kubicki, M. Unpublished work.
(39) The yields of butane, the dimerization product, are negligible.
(43) Lewis, E. S.; Vanderpool, S. H. J. Am. Chem. Soc. 1978, 100, 6422.
Inorganic Chemistry, Vol. 44, No. 15, 2005 5489

Galezowski
S1-S5, Supporting Information). Rough equilibrium con-
stants determined from the amplitudes are consistent with
the values obtained from deconvolution of kobs for the methyl-
transfer step via eq 2. All of this is in agreement with a
limiting dissociative mechanism of ligand exchange in CH3-
CoIIIPc(DMA). However, rapid thiophenoxide reactions are
slower than aromatic amine ligation by CH3CoIIIPc in DMA.7
A search for other ligands that bind to cobalt at a measurable
rate led to determination of the rate constant of 261 000 (
6000 M-1 s-1 for possibly the strongest binding cyanide
ligand, K ) 2.1 × 106 M-1 42 (see Figure S7, Supporting
Information).
Figure 6. Hammett plot for methyl transfer from CH3CoIIIPc to thio-
phenoxides in DMA (25 °C). The σ value for 4-NO2 is modified as ex-
plained in the text.
For a purely dissociative ligand exchange in CH3CoPc-
(DMA) proceeding via the five-coordinate CH3CoIIIPc in-
termediate, a steady-state approach yields the familiar eq 3
DMA is accounted for. Hence, rapid rates of methyl transfer
from a good methyl donor, CH3CoPc, to an excellent
nucleophile, such as thiophenoxide, could be predicted, and
they are in agreement with the normal SN2 mechanism.
k-SkL[L] + kSk-L[solv]
kobs
)
(3)
kS[solv] + kL[L]
After due correction of the σ- value for the 4-NO2 group,44
the rate constants, k2, fit the Hammett equation adequately
(Figure 6) yielding F ) -2.3 ( 0.1. The kinetics F values
are usually compared with those for the equilibrium, Feq,
which in the present case are not accessible. A rare example
of Feq ) -3.8 (or -6.6 after far extrapolation to 25 °C) for
a methyl transfer to arenethiolates has been determined in
protic EtOH at elevated temperatures.45 The reaction constant
for protonation of thiophenoxides in an aprotic solvent, a
process in which the formal negative charge on sulfur is also
entirely reduced, is not likely to be much different from that
for methyl-transfer equilibria. The F value determined from
pKa values for thiophenols in polar aprotic DMSO is -4.84.46
There is a reason to believe that the F value in DMA will
not be much different, since a close correspondence exists
between the pKa values, including those for thiophenols, in
DMSO and DMF.47 Given that, a 50% negative charge
outflow from sulfur in the transition state could be concluded.
The F ) -2.3, similar to the values determined for the
reaction of butyl chloride with thiolates in DMSO46 and
DMF,34 or other thiophenoxide promoted displacement
reactions,43,45,48 is clearly consistent with the SN2 mechanism.
where k-S and kS refer to desolvation/solvation of CH3CoIII-
Pc and kL and k-L refer to ligation of (five-coordinate) RCoIII-
Pc by L. Saturation predicted by eq 3 is not reached in dilute
solutions of incoming ligand, in which the stopped-flow
studies are viable. Then, in the denominator, kS[solv] must
be much larger than kL[L], a condition that in view of
[solvent] . [L] is likely to be fulfilled. The slopes of kobs vs
[L] can be interpreted as a combination of three rate
constants, k-S, kS[DMA], and kL. The ratio k-S/(kS[DMA])
represents a fraction of the five-coordinate species in DMA,
which has been estimated to be of the order of 0.01.7
Consequently, kL would be ∼108 M-1 s-1, which is still an
order of magnitude smaller than the value predicted by the
Smoluchowski equation for a diffusion-controlled reaction
in DMA. Given the equation was derived for spherical
reagents, perhaps ligation by thiolates could be diffusion
controlled, which would account for the rates nearly equal
for all of the thiophenoxides studied. On the other hand, a
preassociation mechanism would allow for the differences
in the rates between pyridine, N-MeIm, thiophenoxide, and
cyanide ligation. The association constant for the spectator
mechanism would be the largest for incoming ligands that
can favorably interact with the phthalocyanine macrocycle.
This would explain the relatively slow cyanide ligation
compared to aromatic amines. Pressure effect on the rates
would be helpful for establishing the mechanism with a
reasonable confidence. Whatever the mechanism for the
ligand exchange, this study underlines the amazing kinetic
lability of the electron-deficient CH3CoIIIPc model.7
Lability of CH3CoIIIPc(SR)-. The rates for the initial fast
step can be measured by stopped flow, albeit with difficulty,
in dilute thiophenoxide solutions. A striking feature of the
rough second-order rate constants for thiophenoxide binding,
k1, is that they are invariant within experimental error (Table
2), while the amplitude of the initial drop in the Q-band
intensity decreases markedly from 4-MeO to 3,4-Cl2 (Tables
The order of the rates for dissociation of CH3CoIIIPc(SR)-,
k-1, calculated from k1 and K (Table 2), is consistent with a
limiting dissociative or Id mechanism. The rate for thiolate
ligand liberation from CH3CoIIIPc(4-CH3O-C6H4-S-) is
only about 140 times larger than k-1 ) 0.124 s-1 for
dissociation of possibly the most stable cyanide complex.49
Faster rates for dissociation of pyridine than for thiolate
(44) Previous LFER studies of the reaction of butyl chloride with
thiophenoxides in DMSO (ref 46) clearly showed that the σ- for the
4-NO2 substituent, which was determined in hydroxylic solvents, is
unsuitable for correlations in aprotic solvents. It appeared that an
effective substituent constant of 1.0, that is, the intermediate between
σ and σ-, would place the 4-NO2 point on the line determined by the
data for non-R+ substituents. The same seems to be true in the present
case.
(45) Lewis, E. S.; Kukes, S. J. Am. Chem. Soc. 1979, 101, 417.
(46) Bordwell, F. G.; Hughes, D. L. J. Org. Chem. 1982, 47, 3224.
(47) Maran, F.; Celadon, D.; Severin, M. G.; Vianello, E. J. Am. Chem.
Soc. 1991, 113, 9320.
(49) Calculated from the stability constant for CH3CoIIIPc(CN)- from ref
42 and the rate constant for cyanide binding by cobalt.
(48) Westaway, K. C.; Ali, S. F. Can. J. Chem. 1978, 57, 1089.
5490 Inorganic Chemistry, Vol. 44, No. 15, 2005

Methyl Transfer from CH3CoIIIPc to Thiophenoxides
Table 3. Kinetic Data for Methyl Transfer from CH3CoIIIPc to
2,6-Dichlorothiophenoxide in DMA
[2,6-Cl2-C6H3S-],
temp, °C
M
k
obs, s-1
k2, M-1 s-1 a K, M-1
15
20
20
20
20
20
20
25
30
35
0.002-0.02
0.002
0.003
0.004
0.006
0.01
0.02
0.00125-0.1
0.002-0.02
0.002-0.02
0.198-0.840
0.247 ( 0.006
0.365 ( 0.005
0.479 ( 0.006
0.64 ( 0.01
0.851 ( 0.009
1.275 ( 0.009
0.283-2.91
0.856-3.78
1.26-6.16
115 ( 3
79 ( 8
139 ( 4
255 ( 4
490 ( 20
730 ( 20
77 ( 7
69 ( 7
65 ( 10
63 ( 7
Figure 7. Dependence of equilibrium constants for ligation of X-py (b)
and thiophenoxides (O) by CH3CoIIIPc in DMA (25 °C) upon basicity of
the ligand. The K values for X-py from ref 7, pKa values for thiophenols
from ref 46, and pKa of 4-bromothiophenol was adopted for 4-Cl. For pKa
values for X-pyH+, see ref 51.
a ∆Hq ) 16.4 ( 1.3 kcal/mol, ∆Sq ) 7.6 ( 4.2 cal mol-1 K-1
.
small. A larger value, âeq ) 0.38, for coordination of aliphatic
amines to methylcobaloxime, than âeq (X-py) ) 0.21 and
âeq (RS-) ) 0.18, was used as an argument for a strong back
π bonding of thiolates, stronger than that of X-py. Accord-
ingly, by the criteria adopted by Brown et al.,50 the reverse
would be true for the phthalocyanine model: X-py ligands
would exhibit significantly stronger π bonding to CH3CoIII-
Pc than that of thiophenoxides. It is not clear, though,
whether a qualitative change in π-bonding properties occurs
between methylcobaloxime and CH3CoIIIPc or the compari-
son is not reliable because of the different types of thiols
and solvents used.
complexes are in agreement with the behavior of analogous
methylcobaloxime complexes.50
A reasonable comparison of thiolate and N-donor binding
by CH3CoIIIPc(L) is possible only when ligands of similar
basicity are compared. It is customary to use aqueous pKa
values for protonated ligands, but in the present case, it would
be inappropriate especially for thiophenoxides, of which the
basicity changes with remote substitution much faster in
aprotic solvent, such as in DMSO rather than in water.46 In
Figure 7,the basicities of protonated ligands are either those
determined in DMSO (thiolates) or those determined in
acetonitrile and recalculated (pyridines). It can be roughly
estimated that, in DMSO, the basicity of 4-chlorothiophe-
noxide should be similar to that of 4-Me2N-py.51 The
corresponding ligand association constants of these two
ligands with CH3CoPc in DMA are not much different, 2600
( 100 (Table 2) and 1840 ( 70 M-1,7 respectively. However,
the response to changes in ligand basicity, represented by
âeq ) d(log K)/d(pKa), is significantly stronger for thio-
phenoxide than X-py ligands, 0.48 ( 0.02 and 0.33,52
respectively (Figure 7). Hence, in the low pKa range
thiophenoxides are weaker binding ligands than X-py,
whereas in the high pKa range, where alkanethiolates belong,
the reverse would be true if only pyridines basic enough
existed. This is contrary to the methylcobaloxime model,
which shows significantly stronger affinity for binding by
thiolates than X-py over the whole range of practical pKa
values, and the âeq values for both types of complexes are
A reasonable prediction can be made, using trends shown
in Figure 7, that strongly basic alkanethiolates (pKa ∼17 in
DMSO) will coordinate to CH3CoPc stronger than to
pyridine, and such strong binding is indeed observed (vide
infra). These two ligands have been invoked here, because
in aqueous solution a change in the order of ligand affinity
for coordination to the Co center is observed on going from
methylcobaloxime, RS- > X-py, to MeCbi+, where pyridine
binding is measurable but thiolate coordination has not been
observed.10 The behavior of CH3CoIIIPc, alkanethiolate >
py, is consistent with its electron-deficient character, which
positions this model closer to methylcobaloxime than to
MeCbi+.7 The fact that the formation constants for
CH3CoIIIPc(SR)-, determined indirectly from kinetic data,
are comparable to those for the pyridine adduct stands in
support of the mechanistic assignments of this work. This is
of some importance for a system in which thiolate binding
cannot be directly studied because of rapid demethylation.
Alkyl Group and Axial Base Effects. It has been
demonstrated that the alkyl group effect Me > Et is one
characteristic feature of SN2 alkyl transfer from CH3CoIIIPc
to nucleophiles.7 In the present work, the 2,6-Cl2 reaction,
which can be most conveniently studied, has been scruti-
nized. The data of Tables 3 and 4 demonstrate that, in
accordance with alkyl transfers to other nucleophiles, CH3-
CoIIIPc is considerably faster than CH3CH2CoIIIPc. The Me/
Et second-order rate constants ratio of ca. 10 is similar to
those determined by Norris and Pratt for slow alkyl transfer
from MeCbi+ to alkanethiolates.3 The ∆Sq values for both
methyl and ethyl transfer are moderately positive:7 8 ( 4
and 2 ( 3 cal mol-1 K-1, respectively, which seems to be
the rule for methyl transfers from CH3CoIIIPc to anionic
(50) Brown, K. L.; Chernoff, D.; Keljo, D. J.; Kallen, R. G. J. Am. Chem.
Soc. 1972, 94, 6697.
(51) This estimate is based on the following: (i) pKa values for X-pyH+
in acetonitrile from Kaljurand, I.; Rodima, T.; Leito, I.; Koppel, I.
A.; Schwezinger, R. J. Org. Chem. 2000, 65, 6202 and Tables of Rate
and Equilibrium Constants of Heterolytic Organic Reactions; Palm.
V., Ed.; VINITI: Moscow-Tartu, Russia, 1975-1985. (ii) Assumed
parallel shift of pKa values between acetonitrile, DMSO, and DMA,
the pKa shift -8.94 ) pKa(pyH+)DMSO - pKa(pyH+)MeCN. (iii) The
pKa value for pyridine in DMSO from Izutsu, K. Acid-Base
Dissociation Constants in Dipolar Aprotic SolVents; IUPAC Chemical
Data; Series No. 35; Blackwell Scientific: Oxford, U.K., 1990. (iv)
pKa values for thiols in DMSO from ref 46.
(52) This slope is very similar to that of 0.30 ( 0.03, determined in toluene
using four X-py ligands (ref 7). Similar sensitivity of X-py binding to
methylcobaloxime in DMSO or DMSO/H2O to changes in X can be
seen in Tables 21 and 50 in Bresciani-Pahor, N.; Forcolin, M.; Marzilli,
L. G.; Randaccio, L.; Summers, M. F.; Toscano, P. J. Coord. Chem.
ReV. 1985, 63, 1.
Inorganic Chemistry, Vol. 44, No. 15, 2005 5491

Galezowski
Table 4. Kinetic Data for Ethyl Transfer from CH3CH2CoIIIPc to
2,6-Dichlorothiophenoxide in DMA
where KN-MeIm, the formation constant of CH3CoPc(N-
MeIm), 2000 ( 20 M-1,7 appears satisfactory. Similar
inhibition was also observed for the unsubstituted thiophe-
noxide reaction. The N-MeIm inhibition, while consistent
with mechanisms of Schemes 1 and 4, allows us to rule out
the OSET mechanism, since N-MeIm has little impact on
the reduction potential.7
[2,6-Cl2-C6H3S-],
temp, °C
M
k
obs, s-1
k2, M-1 s-1 a K, M-1
6.0 ( 0.2 46 ( 10
15
20
25
25
25
25
25
25
35
40
0.002-0.02
0.002-0.02
0.002
0.003
0.004
0.006
0.01
0.02
0.0111-0.061
0.0224-0.107 12.4 ( 0.5 73 ( 12
0.0358
0.0504
0.0641
0.0846
0.118
0.199
0.080-0.495
0.129-0.880
Methyl Transfer to Aliphatic Thiolates. As is apparent
from Table 2, the more powerful a thiolate nucleophile, the
less accurate the k2 value. The value for the 4-MeO reaction
could not be determined with any reasonable accuracy. On
the basis of the results of Bordwell’s study of thiolate attack
on butyl chloride, where strongly basic aliphatic thiolates
(pKa ∼17 in DMSO)46 were faster than thiophenoxide (pKa
) 10.3 in DMSO),46 a prediction could be made that, in the
present case, determination of k2 for reactions with aliphatic
thiolates (similar to homocysteine thiolate) would not be
possible. There is another hurdle, an extreme instability of
basic aliphatic thiolates in DMA. Nevertheless, it seems
important to demonstrate that methyl transfers from CH3-
CoIIIPc to alkanethiolates, similar to homocysteine thiolate,
are rapid. A DMA solution of the organocobalt substrate is
stable in the presence of 0.3 M t-BuSH or n-BuSH in the
spectrophotometric cell under argon. Subsequent injection
of an equimolar amount of Bu4NOH results in the instan-
taneous formation of CoIPc-. Stopped-flow runs of the
n-BuS- reaction were not attempted because of the extreme
instability of the thiolate in DMA. For the t-BuS- reaction,
two exponential kinetic curves, similar to those of basic
thiophenoxide reactions, are observed in the stopped flow.
Rough second-order rate constants for the preequilibrium
process are within experimental error indistinguishable from
the k1 values listed in Table 2. The initial drop in 660 nm
absorbance is as large as that observed with basic thiophe-
noxides. Hence, although complete reaction spectra have not
been obtained, it is plausible that they are similar to those
of Figure 2. The kobs values for the methyl transfer step of
2.2 ( 0.1 s-1 were observed for [t-BuS-] > 0.0008 M at 25
°C. There is no compelling reason to believe that there is
any qualitative difference between the reactions of basic
thiophenoxides and the methyl transfer to the considerably
more basic t-BuS-. They are likely to follow the mechanism
of Scheme 1.
Is the “Extremely Reactive” CH3CoIIIPc Consistent
with Inert Methylcobaloxime? In our work, we have been
trying to show that Co(I) nucleophiles, although very
powerful, are consistent with other ordinary nucleophiles.42
Conversely, CH3CoIII complexes should be consistent with
other methyl donors. In fact, in polar aprotic solvents, CoIPc-
is 2 orders of magnitude faster a nucleophile than I-, while
CH3CoPc is approximately a 10-fold slower methyl donor
than CH3I. All of the CH3CoIII complexes known are
considerably weaker methyl donors than methyl iodide; they
are also weaker than CH3CoIIIPc. Methyl transfers from CH3-
CoIIIPc to a variety of nucleophiles, not only thiolates, are
“unusually fast”.7 Hence, there is nothing special about the
fast methyl transfer from CH3CoIIIPc specifically to thiolates.
Rather, the phthalocyanine model appears uncharacteristically
20.0 ( 0.4 63 ( 6
0.002-0.02
0.002-0.02
44 ( 2
71 ( 5
51 ( 14
50 ( 20
a ∆H‡ ) 16.4 ( 1.0 kcal/mol, ∆S‡ ) 2.4 ( 3.2 cal mol-1 K-1
.
Table 5. N-Methylimidazole Inhibition of Methyl Transfer from
CH3CoIIIPc to 2,6-Dichlorothiophenoxide in DMA (25 °C)
kobs s-1
kcalcd s-1 a
[2,6-Cl2-C6H3S-],
[N-MeIm] )
[N-MeIm] )
0.0125 M
[N-MeIm] )
0.0125 M
M
0
0.002
0.003
0.004
0.006
0.01
0.44 ( 0.1
0.63 ( 0.1
0.81 ( 0.1
1.08 ( 0.2
1.53 ( 0.2
0.023 ( 0.001
0.036 ( 0.002
0.045 ( 0.002
0.059 ( 0.001
0.092 ( 0.002
0.020
0.029
0.039
0.058
0.095
a Calculated using eq 4, KN-MeIm ) 2000 ( 20 M-1 from ref 7.
nucleophiles in coordinating polar solvents.7 Even though
2,6-Cl2 is the weakest of the thiolate nucleophiles24,28 studied,
the ∆Hq ) 16 ( 1 kcal/mol appears to be the lowest among
methyl transfers from CH3CoIIIPc to anionic nucleophiles in
DMA determined so far.7 The activation parameters for
methyl and ethyl transfer are indistinguishable within
experimental error. Experimental error in the K values is
larger yet, not allowing any reasonable determination of
reaction parameters for 2,6-Cl2 binding by cobalt. It tran-
spires, however, that -∆H must be very small. The alkyl
group effect is not restricted to weak thiolate nucleophiles
since a similar Me/Et rate constant ratio of 10 ( 2 has been
determined also for the thiophenoxide reaction (data for the
ethyl derivative, k2 ) 900 ( 100 M-1 s-1, K ) 1900 ( 300
M-1).
Homolytic cleavage of the Co-C bond is usually faster
for ethyl than the methyl σ-bonded group. This was observed
for electrochemically generated anions and cations of
σ-bonded alkyl porphyrins. Also in this study, [CH3CoPc]-
was found to be more stable than [CH3CH2CoPc]- (the first
reduction of CH3CH2CoPc is still irreversible when that of
CH3CoPc reaches quasi-reversibility). Faster rates for the
methyl derivative are against any mechanism in which the
homolytic fission is rate limiting. With attention to the
mechanisms of Schemes 3 and 4, larger k2/K values for
methyl than ethyl transfer to 2,6-Cl2 are noteworthy.
Another familiar feature of methyl transfers from Co(III),
axial base inhibition,7 is also found for thiolate reactions as
can be seen in Table 5. Given all sorts of experimental error,
the fit of experimental data to eq 4,7
k2[thiolate]
kcalcd
)
(4)
1 + K[thiolate] + KN-MeIm[N-MeIm]
5492 Inorganic Chemistry, Vol. 44, No. 15, 2005

Methyl Transfer from CH3CoIIIPc to Thiophenoxides
reactive. This reactivity stems from the “electron-poor”
nature of CH3CoIIIPc combined with a considerable lability
of trans axial ligands compared with other electron-deficient
alkylcobalt complexes.7 The association constant of pyridine
with methylcobaloxime in DMSO is about 200 times larger
than that obtained with CH3CoIIIPc.7 The increased demand
of methylcobaloxime for six coordination11 alone makes CH3-
drive for demethylation of CH3CoIIIPc.42 The unique proper-
ties of the phthalocyanine model should result in larger
reactivity of the five-coordinate CH3CoIIIPc species than that
of the hypothetical five-coordinate form of methylco-
baloxime. Of note, the order of trans ligand binding proper-
ties, methylcobaloxime > CH3CoIIIPc, is contrary to what
could be predicted from electrochemical data. It is striking
that the methylcobaloxime model is characterized by both
slow methyl transfer and slow trans ligand exchange. A
simple rationale behind it would be that these different
reactions have a common intermediate, the five-coordinate
methylcobaloxime species, which, in contrast to CH3CoIII-
Pc, is extremely unstable.
A consistent, although simplified, description of alkyl
(formally cation) transfer from cobalt(III) is possible: some
alkylcobalt(III) complexes are inactive because of their
“electron-rich” nature; and others have an “electron-deficient”
equatorial ligand but heavy trans axial ligation reduces their
methyl donation power. Finally, in some models, the
increased electron-deficient character of the equatorial system
is only partially compensated by an enhanced demand for
hexacoordination. Methyl transfer from such complexes to
thiolates should be measurable in nonhydroxylic, possibly
weak binding, solvents in the absence of strong axial bases.
MeCbi+ has been shown to persist in a significant
percentage as a five-coordinate complex in aqueous solu-
tion.57 Accordingly, it reacts with alkanethiolates slowly but
measurably.3 The reaction should be reasonably fast in a
nonhydroxylic, weakly binding, polar solvent.
+
CoPc about 102-fold more reactive a CH3 donor than
methylcobaloxime (that is, if we assume for a moment equal
reactivity of the reactive five-coordinate forms). Since the
R carbon in the 16-electron five-coordinate species is
considerably more electrophilic than that in 18-electron six-
coordinate complexes, the latter complex appears inactive.7
One should also keep in mind that nucleophilic power of
thiolates in aprotic solvents such as DMA is significantly
increased compared with aqueous or alcoholic solution where
demethylations of methylcobaloximes and MeCbi+ have been
studied (or attempted). If we allow a factor of 104 for this
effect,53 then a methyl transfer reaction from CH3CoPc in
DMA will be overall a million times faster than that from
methylcobaloxime in methanolic solution (seconds instead
of a month). Thus, the apparent inertness of electron-poor
methylcobaloxime or Costa-type models toward thiolates can
nearly be reconciled with the reactive phthalocyanine model.
The conclusion made by Brown,5 Marzilli, and co-workers6
that thiolate ligation is the primary process observed upon
addition of a thiolate to methylcobaloxime, and that the
Co-C bond remains intact, would also be true for CH3CoIII-
Pc if only the observation time was sufficiently short (∼10
ms). It is noteworthy that methyl transfer from methylco-
baloxime to thiolates was lately attempted in the presence
of very strong trans axial ligands, Me2PhP or Me3P.6 This
must have further reduced the already low population of the
reactive species. Furthermore, because of the kinetic satura-
tion such as observed in this work, the rate for demethylation
of methylcobaloxime or a Costa-type model could not be
significantly increased by raising the concentration of the
nucleophile.
Conclusions
Our original claim of normal SN2 mechanism of methyl
transfer from Co to thiophenoxide is confirmed and even
extended to alkanethiolates, despite distracting atypical
transient spectra and the coincidence of the rates for methyl
transfer to thiolates and the rate for homolysis of [CH3CoPc]-.
As demonstrated a decade ago and confirmed by this study,
nonenzymatic SN2 methyl transfers from cobalt to thiolates
can be as fast as enzymatic systems demand. A prediction
based on Bordwell and Hughes’ work46 yields a second-order
rate constant for the reaction of CH3CoIIIPc with butanethi-
olate in DMA on the order of 106 M-1 s-1, compared with
135 000 M-1 s-1 for the analogous reaction of methionine
synthase fragments.17 The increased methyl donation power
of a RCo(III) complex implies a weaker nucleophilic ability
of the corresponding Co(I) complex. Consequently, the
phthalocyanine system probably could not replace the
corrinoid because CoIPc- is too weak a nucleophile to attack
a nitrogen-bonded methyl group.
The electron-poor character of the phthalocyanine model
is evident from the small negative reduction potential for
CH3CoIIIPc, -0.95 V vs SCE in DMA or -0.98 V in DMF,7
compared to -1.35 V for methylcobaloxime in DMF,54 and
a small negative oxidation potential for CoIPc- 19 of -0.37
V in DMF55 compared with -1.17 V for the cobaloxime
Co(II)/Co(I) couple.56 The phthalocyanine model has excep-
tional ability to stabilize negative charge in the transition
state and in the product of demethylation. This shows in low,
for a cobalt supernucleophile, thermodynamic nucleophilicity
of CoIPc-, and consequently, in an enhanced thermodynamic
The high reactivity of a five-coordinate electron-poor
complex seems to suggest one possible mode of activation
of inert MeCbl. Another lesson, not new, is that chemical
precedent studies in aqueous solution are somewhat mislead-
ing because in a hydroxylic solvent, the nucleophilic power
of a nucleophile is drastically lowered compared with aprotic
(53) See, for instance, a 4 order-of-magnitude faster reaction of n-BuBr
with thiophenoxide in DMF than in MeOH in Table 4. Alexander,
R.; Ko, E. C. F.; Parker, A. J.; Bronxton, T. J. J. Am. Chem. Soc.
1968, 90, 5049.
(54) Costa, G.; Puxeddu, A.; Tavagnacco, C. J. Organomet. Chem. 1985,
296, 161.
(55) Lever, A. B. P.; Milaeva, E. R.; Speier, G. In Phthalocyanines,
Properties and Applications; Leznoff, C. C., Lever A. B. P., Eds.;
Wiley-VCH: New York, 1993; Vol. 3, p 47.
(56) Elliot, C. M.; Hershenhart, E.; Finke, R. G.; Smith, B. L. J. Am. Chem.
Soc. 1981, 103, 5558.
(57) Hamza, M. S. A.; Zou, X.; Brown, K. L.; van Eldik, R. Eur. J. Inorg.
Chem. 2003, 268 and references therein.
Inorganic Chemistry, Vol. 44, No. 15, 2005 5493

Galezowski
Supporting Information Available: Figures showing concen-
tration dependencies of conductances of C6H5SNa and 4-ClC6H4-
SNa in DMA at 25 °C and corresponding Kohlrausch plots. Figures
showing typical kinetic curves and corresponding end-point plots.
Tables of detailed kinetic data for methyl transfer from CH3CoIII-
Pc to sodium thiophenoxides in DMA (25 °C). Figure showing
concentration dependence of the rates for methyl transfer to
4-CH3C6H4S- rates and the double reciprocal plot. An example
for determination of K from the amplitudes of the fast step. Figure
showing the concentration dependence of the rates for cyanide
binding by CH3CoIIIPc in DMA (25 °C). This material is available
free of charge via the Internet at http://pubs.acs.org.
solvents, which probably better mimic a hydrophobic,
enzymatic environment. Nucleophilic participation by cobalt,
which plays an important role in the phthalocyanine model
and makes the methyl transfer effectively slower, is not that
important for the MeCbi+ reactions with thiolates, where
linear concentration dependencies of the rates are observed.
Methylcobaloxime, in turn, exhibits a very strong demand
for six-coordination and with this respect should be regarded
as a MeCbl antimodel.58
Acknowledgment. This work was supported by the
Polish State Committee for Scientific Research Grant 7 T09A
07421.
IC0503378
(58) Doll, K. M.; Finke, R. G. Inorg. Chem. 2004, 43, 2611.
5494 Inorganic Chemistry, Vol. 44, No. 15, 2005