Mechanistic investigation of a novel vitamin B12-catalyzed carbon-carbon bond forming reaction, the reductive dimerization of arylalkenes

Article abstract of DOI:10.1021/jo0160470

In the presence of catalytic vitamin B₁₂ and a reducing agent such as Ti(III)citrate or Zn, arylalkenes are dimerized with unusual regioselectivity forming a carbon-carbon bond between the benzylic carbons of each coupling partner. Dimerization products were obtained in good to excellent yields for mono- and 1,1-disubstituted alkenes. Dienes containing one aryl alkene underwent intramolecular cyclization in good yields. However, 1,2-disubstituted and trisubstituted alkenes were unreactive. Mechanistic investigations using radical traps suggest the involvement of benzylic radicals, and the lack of diastereoselectivity in the product distribution is consistent with dimerization of two such reactive intermediates. A strong reducing agent is required for the reaction and fulfills two roles. It returns the Co(II) form of the catalyst generated after the reaction to the active Co(I) state, and by removing Co(II) it also prevents the nonproductive recombination of alkyl radicals with cob(II)alamin. The mechanism of the formation of benzylic radicals from arylalkenes and cob(I)alamin poses an interesting problem. The results with a one-electron transfer probe indicate that radical generation is not likely to involve an electron transfer. Several alternative mechanisms are discussed.

Full text of DOI:10.1021/jo0160470

J . Org. Chem. 2002, 67, 837-846
837
Mech a n istic In vestiga tion of a Novel Vita m in B12-Ca ta lyzed
Ca r bon -Ca r bon Bon d F or m in g Rea ction , th e Red u ctive
Dim er iza tion of Ar yla lk en es
J ustin Shey, Chris M. McGinley, Kevin M. McCauley, Anthony S. Dearth,
Brian T. Young, and Wilfred A. van der Donk*
Department of Chemistry, University of Illinois at Urbana-Champaign,
600 S. Mathews Ave., Urbana, Illinois 61801
vddonk@uiuc.edu
Received August 22, 2001
In the presence of catalytic vitamin B12 and a reducing agent such as Ti(III)citrate or Zn, arylalkenes
are dimerized with unusual regioselectivity forming a carbon-carbon bond between the benzylic
carbons of each coupling partner. Dimerization products were obtained in good to excellent yields
for mono- and 1,1-disubstituted alkenes. Dienes containing one aryl alkene underwent intramo-
lecular cyclization in good yields. However, 1,2-disubstituted and trisubstituted alkenes were
unreactive. Mechanistic investigations using radical traps suggest the involvement of benzylic
radicals, and the lack of diastereoselectivity in the product distribution is consistent with
dimerization of two such reactive intermediates. A strong reducing agent is required for the reaction
and fulfills two roles. It returns the Co(II) form of the catalyst generated after the reaction to the
active Co(I) state, and by removing Co(II) it also prevents the nonproductive recombination of alkyl
radicals with cob(II)alamin. The mechanism of the formation of benzylic radicals from arylalkenes
and cob(I)alamin poses an interesting problem. The results with a one-electron transfer probe
indicate that radical generation is not likely to involve an electron transfer. Several alternative
mechanisms are discussed.
In tr od u ction
The properties and reactivity of vitamin B12 derivatives
have been extensively investigated ever since it was
shown to possess a cobalt-carbon bond in the biological
cofactors adenosyl- and methylcobalamin.1-5 These ver-
buffer (pH 8) and polar organic solvents such as ethanol,
methanol, or acetonitrile.12 Perhaps the most interesting
aspect of this transformation involves the mechanism by
which it takes place. We describe in this report the scope
of the reaction and a detailed investigation of its mech-
anism.
satile organometallic complexes are used in a variety of
enzymes to catalyze radical rearrangements and methyl
transfers. Moreover, vitamin B12 has found use in organic
chemistry for carbon-carbon bond formations.6-11 In this
study, we report a previously unknown B12-catalyzed
reaction that forms a new carbon-carbon bond between
two alkenes providing dimerization products 1 (eq 1). It
is noteworthy that the regiochemistry of this coupling is
opposite to that expected for radical chemistry involving
alkenes. The transformation employs catalytic vitamin
B12 and an auxiliary reductant such as Zn or Ti(III)citrate
and can be carried out at 25 °C in a mixture of aqueous
Resu lts a n d Discu ssion
Scop e of B12-Ca ta lyzed Dim er iza tion of Alk en es.
The range of substrates that can be coupled is shown in
Table 1. Monosubstituted arylalkenes gave a 1:1 ratio of
diastereomeric products, whereas the disubstituted sub-
strates produced two quaternary carbon centers in high
yields (entries 4-7). When either vitamin B12 or Ti(III)-
citrate was omitted from the reaction mixture, only
starting alkenes were recovered. Other methods have
been reported for the preparation of these types of
compounds, but they have generally relied on relatively
harsh reaction conditions13-18 or photochemical19 reac-
(1) Lenhert, P. G.; Hodgkin, D. C. Nature 1961, 192, 937-938.
(2) Hodgkin, D. C. Science 1965, 150, 979-988.
(3) Dolphin, D. B12; J ohn Wiley & Sons: New York, 1982.
(4) Kra¨utler, B.; Arigoni, D.; Golding, B. T. Vitamin B12 and B12
Proteins; J ohn Wiley & Sons: New York, 1998.
-
(5) Banerjee, R. The Chemistry and Biochemistry of B12; Wiley:
New York, 1999.
(6) Scheffold, R.; Abrecht, S.; Ruf, H.-R.; Stamouli, P.; Tinembart,
O.; Walder, L.; Weymuth, C. Pure Appl. Chem. 1987, 59, 363-372.
(7) Ogoshi, H.; Kikuchi, Y.; Yamaguchi, T.; Toi, H.; Aoyama, Y.
Organometallics 1987, 6, 2175-2178.
(12) The content of aqueous buffer is determined by the volume of
aqueous Ti(III)citrate added. The reaction rates in these solvents
systems follow the general trend: aqueous buffer > MeOH > EtOH ≈
MeCN.
(13) Brown, W. G.; McClure, D. E. J . Org. Chem. 1970, 35, 2036-
2037 and references therein.
(14) McMurry, J . E.; Silvestri, M. J . Org. Chem. 1975, 40, 2687-
2688.
(15) Prostasiewicz, J .; Mendenhall, G. D. J . Org. Chem. 1985, 50,
3220-3222.
(8) Pattenden, G. Chem. Soc. Rev. 1988, 17, 361-382.
(9) Baldwin, D. A.; Betterton, E. A.; Chemaly, S. M.; Pratt, J . M. J .
Chem. Soc., Dalton Trans. 1985, 1613-1618.
(10) Baldwin, J . E.; Adlington, R. M.; Kang, T. W. Tetrahedron Lett.
1991, 48, 7093-7096.
(11) Branchaud, B. P.; Friestad, G. K. Encyclopedia of Reagents for
Organic Synthesis; Paquette, L., Ed.; Wiley: New York, 1995; pp 5511-
5514.
10.1021/jo0160470 CCC: $22.00 © 2002 American Chemical Society
Published on Web 01/12/2002

838 J . Org. Chem., Vol. 67, No. 3, 2002
Shey et al.
Ta ble 1. Vita m in B12/Ti(III)-Med ia ted Dim er iza tion of
Alk en es
Sch em e 1
alkene
R1
yield (%)a
entry
R2
dimer
reduction product
1
2
3
4
5
6
7
Ph
H
H
H
Me
Me
Me
Ph
50
80
56b
85
80
70
90
4
13
7
7
3
4-MeC6H4
2,5-(Me)2C6H3
Ph
4-ClC6H4
4-FC6H4
Ph
0
5
a
Typical reaction times for complete conversion varied from 8
b
to 12 h. In aqueous tert-butyl alcohol.
plings between alkylhalides and alkenes, or in some cases
alkene polymerizations. In the former, an alkylhalide is
reacted with a cobalt(I) complex to form an organocobalt
intermediate, which in turn serves as a precursor for a
reactive radical via homolytic cleavage of the weak Co-C
bond (e.g. Scheme 1). This radical subsequently initiates
either an inter- or intramolecular addition to an alkene
to generate a product radical that is then trapped by Co-
(II) to form a second organocobalt complex that can be
further elaborated. The structures of both starting ma-
terials and reaction products in Table 1 imply that they
must be formed via pathways that are different from that
in Scheme 1. If the catalyst were to act as a source of
radicals that would add to the arylalkene, either the
regioisomeric product radical 2 or a polymer would be
produced instead of the symmetrical dimerization product
1.
Ch a r t 1
tions of benzylhalides or benzyl alcohols.20 The B12-
catalyzed dimerization, on the other hand, utilizes readily
available styrene derivatives, mild reaction conditions,
and environmentally friendly solvent systems. Several
1,2-disubstituted and trisubstituted alkenes (Chart 1) as
well as alkyl-substituted alkenes that were examined
proved unreactive.
Mech a n istic Con sid er a tion s: Testin g th e In ter -
m ed ia cy of Alk ylcoba la m in s. The transformations in
Table 1 have synthetic utility, but they are more inter-
esting from a pure mechanistic viewpoint. A solution of
Ti(III)citrate (10 equiv) and vitamin B12 displays the
characteristic strong absorption around 385 nm indica-
tive of Co(I),3 suggesting the highly nucleophilic cob(I)-
alamin (B12 in its Co(I) oxidation state) could be the
reactive form of the catalyst. Cob(I)alamins and related
cobalt complexes have been used previously by several
groups for catalytic or stoichiometric carbon-carbon bond
formations.21-28 In general, these reports describe cou-
Cob(I)alamin has been reported to react with styrene
at acidic pH to provide a product that was tentatively
assigned as R-phenethylcobalamin.29 The complex was too
reactive to be isolated, but the authors did not report
formation of dimerization products. Although cob(I)-
alamin reportedly29,30 does not react with styrene under
the slightly basic reaction conditions used here (pH 8.0),
the regiochemistry of the coupling reactions in Table 1
suggests that substituted benzylcobalamins (3) may be
intermediates along the reaction pathway. Cob(I)alamin
is an exceptionally strong nucleophile that readily reacts
with organohalides to form organocobalamins.31,32 We
therefore tested for the intermediacy of alkylcobalamins
by reaction of various benzyl halides under the reaction
conditions used for the coupling of alkenes (Table 2).
Dimerization products were obtained in good to excellent
yields with identical regiochemistry as observed for the
corresponding alkenes (e.g., compare entry 1, Table 1
(16) Bors, D. A.; Kaufman, M. J .; Streitwieser, J . A. J . Am. Chem.
Soc. 1985, 107, 6975-6982.
(17) Popielarz, R.; Arnold, D. R. J . Am. Chem. Soc. 1990, 112, 3068-
3082.
(18) Maslak, P.; Champan, J . W. H. J . Org. Chem. 1990, 55, 6334-
6347.
(19) Pasternak, M.; Morduchowitz, A. Tetrahedron Lett. 1983, 24,
4275-4278.
(20) For methodology starting with aryl ketones and Grignard or
organolithium reagents in the presence of vanadium or tungsten
catalysts, see: (a) Kataoka, Y.; Akiyama, H.; Makihira, I.; Tani, K. J .
Org. Chem. 1996, 61, 6094-6095. (b) Kauffmann, T.; J ordan, J .; Voss,
K.-U. Angew. Chem., Intl. Ed. Engl. 1991, 30, 1138-1139. For other
metal-catalyzed transformations, see also: Billington, D. C. In Com-
prehensive Organic Synthesis; Trost, B. M., Fleming, I., Pattenden, G.,
Eds.; Pergamon Press: New York, 1991; Vol. 3, Chapter 2.1, p 421.
(21) For reviews, see refs 8, 11, and Ali, A.; Gill, G. B.; Pattenden,
G.; Roan, G. A.; Kam, T. S. J . Chem. Soc., Perkin Trans. 1 1996, 1081-
1093.
(26) Hu, C.-M.; Qiu, Y.-L. J . Org. Chem. 1992, 57, 3339-3342.
(27) Inokuchi, T.; Kawafuchi, H.; Aoki, K.; Yoshida, A.; Torii, S. Bull.
Chem. Soc. J pn. 1994, 67, 595-598.
(28) Wayland, B. B.; Poszmik, G.; Mukerjee, S. L.; Fryd, M. J . Am.
Chem. Soc. 1994, 116, 7943-7944.
(29) Schrauzer, G. N.; Holland, R. J . J . Am. Chem. Soc. 1971, 93,
4060-4062.
(22) Okabe, M.; Tada, M. Bull. Chem. Soc. J pn. 1982, 55, 1498-
(30) Cob(I)alamin has been reported not to react with styrene under
neutral or alkaline conditions, but cob(I)aloximes do react with styrene
under these conditions to give R-phenethylcobaloxime. Schrauzer, G.
N., Windgassen, R. J . J . Am. Chem. Soc. 1967, 89, 1999-2007.
(31) Schrauzer, G. N.; Deutsch, E.; Windgassen, R. J . J . Am. Chem.
Soc. 1968, 90, 2441-2442.
1503.
(23) Baldwin, J . E.; Li, C.-S. J . Chem. Soc., Chem. Commun. 1987,
166-168.
(24) Branchaud, B. P.; Detlefsen, W. D. Tetrahedron Lett. 1991, 32,
6273-6276.
(25) Busato, S.; Tinembart, O.; Zhang, Z. D.; Scheffold, R. Tetrahe-
dron 1990, 46, 3155-3166.
(32) Schrauzer, G. N.; Deutsch, E. J . Am. Chem. Soc. 1969, 91,
3341-3350.

Reductive Dimerization of Arylalkenes
J . Org. Chem., Vol. 67, No. 3, 2002 839
Ta ble 2. Vita m in B12/Ti(III)-Med ia ted Dim er iza tion of
Alk ylh a lid es in Aqu eou s Eth a n ol
Sch em e 2
a
b
Reactions were complete within 1 h. Reaction run in aceto-
Given the well-established radical polymerization of
styrene, it appears unlikely at first glance that two
benzylic free radicals 4, present in very low concentra-
tions, would combine to give the observed products in the
presence of excess alkene.47 However, a look at the rate
constants of these competing processes shows that such
a mechanism is feasible. If one takes the known rate
coefficient for the propagation step in homopolymeriza-
tions of bulk styrene, 85 M-1 s-1 at 25 °C,48-51 as a
reasonable approximation for the rate constant for ad-
dition of radical 4 to the alkene (kp), the rate of formation
of products such as 2 from radical 4 would be 85 ×
[styrene][4]. On the other hand, the rate constant for the
competing dimerization of two radicals 4 will be close to
diffusion-controlled (kd ≈ 109 M-1s-1 at room tempera-
ture), giving a rate of 109 × [4]2 for the production of
products 5 from 4. Thus, at the alkene concentrations
used (4 mM initial) the radical concentration needs to
exceed only ∼2 × 10-9 M to provide a >10-fold preference
for dimerization over polymerization. For substituted
styrene derivatives the concentration of radicals can be
even lower as the propagation rate coefficient is even
smaller (e.g., 0.62 M-1 s-1 for R-methylstyrene).52,53
Indeed, the higher yields obtained from R-substituted
styrene derivatives (Table 1) may reflect this higher
nitrile/water.
with entry 2, Table 2).33 These observations suggest that
benzylcobalamins 3 may indeed be common intermedi-
ates in the coupling of alkenes and benzylhalides, al-
though the latter reactions required significantly shorter
reaction times.34
P r ob in g t h e In t er m ed ia cy of Alk yl R a d ica ls.
Alkylcobalamins have weak Co-C bonds, with the bond
strengths decreasing with more bulky alkyl groups.35-39
Benzylcobalamin contains one of the weakest Co-C
bonds (∼23 kcal mol-1) among alkylcobalamins for which
the bond dissociation energies have been determined
quantitatively.36,38-41 Therefore, if alkylcobalamins such
as 3 are formed in the reaction of cob(I)alamin with
arylalkenes, they may produce benzyl radicals 4 that can
dimerize to provide the observed products 5 (Scheme 2).
Such dimerization of radicals is supported by the lack
of diastereoselectivity that was observed in the reac-
tion products in Tables 1 and 2. If the coupling products
were derived directly from alkylcobalamins, some ste-
reoselectivity would be expected by virtue of the chirality
of B12, as was previously found in various synthetic pro-
cesses.7,42-46
(33) Reductive dimerization of benzylhalides using stoichiometric
(PPh3)3CoCl has been reported. Momose, D.-I.; Iguchi, K.; Sugiyama,
T.; Yamada, Y. Chem. Pharm. Bull. 1984, 32, 1840-1853.
(34) This difference in reaction rates implies that the overall rates
cannot be governed by a step after the formation of the common
intermediate. Instead, the kinetics of formation of the intermediates
must be different for the two classes of compounds. This is not
unexpected given the anticipated difference in mechanism of alkylco-
balamin formation from alkenes vs alkylhalides.
(35) Chemaly, S. M.; Pratt, J . M. J . Chem. Soc., Dalton Trans. 1980,
2274-2281.
(45) Anderson, R. J .; Dixon, R. M.; Golding, B. T. J . Organomet.
Chem. 1992, 437, 227-237.
(46) Hisaeda, Y.; Nishioka, T.; Inoue, Y.; Asada, K.; Hayashi, T.
Coord. Chem. Rev. 2000, 198, 21-37.
(47) In this scenario, the observed R,R-dimer products 1 could form
directly by dimerization of benzylic radicals or by rearrangement of
initially formed R,para- and/or R,ortho-semibenzenes. (For a detailed
discussion, see: Langhals, H.; Fischer, H. Chem. Ber. 1978, 111, 543-
553. Skinner, K. J .; Hochster, H. S.; McBride, J . M. J . Am. Chem. Soc.
1974, 96, 4301-4306.) In fact, small but significant amounts (3%) of
the R,p-dimer were observed in entry 4, Table 1. When this reaction
was stopped after a short reaction time (1 h, <40% completion), the
ratio of R,R- to R,p-dimers was 6:1, suggesting R,p-dimers may
rearrange to R,R-dimers in these reactions.
(36) Schrauzer, G. N.; Grate, J . H. J . Am. Chem. Soc. 1981, 103,
541-546.
(37) Kim, S.-H.; Chen, H. L.; Feilchenfeld, N.; Halpern, J . J . Am.
Chem. Soc. 1988, 110, 3120-3126.
(38) Brown, K. L.; Brooks, H. B. Inorg. Chem. 1991, 30, 3420-3430.
(39) Waddington, M. D.; Finke, R. G. J . Am. Chem. Soc. 1993, 115,
4629-4640.
(48) Buback, M.; Gilbert, R. G.; Hutchinson, R. A.; Klumperman,
B.; Kuchta, F. D.; Manders, B. G.; O’Driscoll, K. F.; Russell, G. T.;
Schweer, J . Macromol. Chem. Phys. 1995, 196, 3267-3280.
(49) Beuermann, S.; Buback, M.; Davis, T. P.; Gilbert, R. G.;
Hutchinson, R. A.; Olaj, O. F.; Russell, G. T.; Schweer, J .; Vanherk, A.
M. Macromol. Chem. Phys. 1997, 198, 1545-1560.
(40) Nome, F.; Rezende, M. C.; Sabo´ia, C. M.; da Silva, A. C. Can.
J . Chem. 1987, 65, 2095-2099.
(41) Andruniow, T.; Zgierski, M. Z.; Kozlowski, P. M. J . Am. Chem.
Soc. 2001, 123, 2679-2680.
(50) Solvent effects on the propagation rate coefficient are generally
absent or small. See ref 51.
(42) Maihub, A.; Grate, J . W.; Hui Bi, X.; Schrauzer, G. N. Z.
Naturforsch., B: Anorg. Chem., Org. Chem. 1983, 38B, 643-647.
(43) Davies, A. G.; Golding, B. T.; Hay-Motherwell, R. S.; Mwesigye-
Kibende, S.; Ramakrishna Rao, D. N.; Symons, M. C. R. J . Chem. Soc.,
Chem. Commun. 1988, 378-380.
(51) (a) Morrison, B. R.; Piton, M. C.; Winnik, M. A.; Gilbert, R. G.;
Napper, D. H. Macromolecules 1993, 26, 4368-4372. (b) Zammit, M.
D.; Davis, T. P.; Willett, G. D.; O’Driscoll, K. F. J . Polymer Sci. A:
Polymer Chem. 1997, 35, 2311-2321. (c) Olaj, O. F.; Schno¨ll-Bitai, I.
Monatsh. Chem. 1999, 130, 731-740.
(44) Bonhoˆte, P.; Scheffold, R. Helv. Chim. Acta 1991, 74, 1425-
1444.
(52) Kukulj, D.; Davis, T. P. Macromolecules 1998, 31, 5668-5680.

840 J . Org. Chem., Vol. 67, No. 3, 2002
Shey et al.
Sch em e 3
selectivity for radical dimerization. Furthermore, when
reactions with styrene were performed at higher concen-
trations (100 mM), insoluble material was formed and
the yield of dimers dropped to ∼20-25%, suggesting
effective competition by oligomerization.
To test for the involvement of radicals, dienes 6 and
7, envisioned as potential intramolecular radical traps,
were reacted with B12 in the presence of Ti(III)citrate (eqs
2 and 3).54 Since trisubstituted alkenes are unreactive
57
M-1 s-1 for k-2
,
the equilibrium constant K2 can be
estimated as 2.0 × 10-12 M-1. Therefore, the concentra-
tions of 3 are significantly higher than the concentrations
of radicals 4 for benzylcobalamin. (See below for a
discussion of more hindered alkylcobalamins and the
possible effect of Ti(III)citrate on this equilibrium.) The
rate of product formation from radical 4 in Scheme 3
would thus be k3[3][4], whereas that of Scheme 2 would
be kd[4]2. At present, rate constants for the reaction of
radicals with alkylcobalamins (e.g., k3) are not known,
but as pointed out previously,55 thermodynamically it
should be highly favorable to form a C-C bond at the
expense of a Co-C bond. Thus, this mechanism may be
kinetically competitive with simple free radical dimer-
ization for the formation of the observed products,
although it is a less likely mechanism for more sterically
hindered intermediates 3 and 4. Another (indirect) argu-
ment against this mechanism is the lack of observed
diastereoselectivity, since one might expect some degree
of stereoselectivity in a reaction between a chiral orga-
nometallic compound and an achiral radical. The dimer-
ization of two organic radicals on the other hand is known
to take place nonstereoselectively.13,58
Over com in g th e P er sisten t Ra d ica l Effect. Ti(III)-
citrate, which is present in excess over vitamin B12, plays
an important dual role in the dimerization reaction. The
homolysis of the Co-C bond in benzylcobalamins is
usually a very slow process under anaerobic conditions
(“too slow to be measured accurately”)36 as a result of the
efficient trapping of the benzyl radical by Co(II), thereby
regenerating benzylcobalamin. In the reactions described
here, for every dimerization event two molecules of Co-
(II) are irreversibly formed, and thus without Ti(III)-
citrate, product formation would slow and eventually be
inhibited by build-up of Co(II) that will scavenge organic
radicals, a manifestation of the “persistent radical
effect”.59-61 In previous studies, recombination of alkyl
radicals with cob(II)alamin has been prevented either by
(see above), it was anticipated that the disubstituted
alkene would generate a radical at the carbon R to the
pyrrole that would subsequently react with the trisub-
stituted alkene in 5-exo-trig fashion. Both compounds
provided the expected cyclized products in good yields,
providing further support for the involvement of radicals.
Furthermore, these transformations show synthetic prom-
ise for cyclization under mild conditions in either polar
organic or aqueous solvents. When the reaction in eq 2
was carried out with tert-butyl alcohol as the organic
cosolvent, the reduced cyclization product 8b was not
formed. Unlike ethanol, tert-butyl alcohol does not con-
tain a relatively weak C-H bond. The improved ratio of
8a to the reduction product 8b suggests therefore that
the latter is formed via hydrogen atom abstraction by a
radical intermediate. The retention of the olefin in 8a in
tert-butyl alcohol is important as it provides a handle for
further functionalization increasing the value of the
cyclization reaction.
An alternative dimerization mechanism can be written
on the basis of a recent report by Mosimann and
Kra¨utler.55 This study showed that alkyl radicals can
react with methylcobalamin to form a methylated product
and cob(II)alamin. Similarly, reaction of the radical 4
with alkylcobalamin 3 could yield the observed dimer-
ization product (Scheme 3). The rate constant for ho-
molytic bond dissociation of benzylcobalamin (k2) is ∼2.0
× 10-3 s-1 at 25 °C.36,38,40,56 Taking a value of 1 × 109
(53) It is interesting to compare and contrast the discussion of the
possible involvement of free radicals in this report with an in depth
analysis by Newcomb and Curran of the use of 5-hexenyl halides as
radical probes for the mechanism of nucleophilic substitution (New-
comb, M.; Curran, D. P. Acc. Chem. Res. 1988, 21, 206-214, footnote
52). In that Account, the authors show that diffusion-controlled
coupling of free radicals (R• and Nu•) cannot account for product
formation because the limitations on the maximum radical concentra-
tions imposed by the known rate constants of competing reactions
would give untenably low rates of product formation. Here product
formation by coupling of free radicals may be possible because of the
slow rates of competing reactions such as the propagation rate for
oligomerization.
(56) Blau, R. J .; Espenson, J . H. J . Am. Chem. Soc. 1985, 107, 3530-
3533.
(57) Endicott, J . F.; Ferraudi, G. J . Am. Chem. Soc. 1977, 99, 243-
245.
(58) Greene, F. D.; Berwick, M. A.; Stowell, J . C. J . Am. Chem. Soc.
1970, 92, 867-874.
(59) Daikh, B. E.; Finke, R. G. J . Am. Chem. Soc. 1992, 114, 2938-
(54) Attempts to detect radicals using intermolecular radical traps
such as TEMPO failed as they reacted with Co(I)/Ti(III).
(55) Mosimann, H.; Kra¨utler, B. Angew. Chem., Int. Ed. 2000, 39,
393-395.
2943.
(60) Fischer, H. J . Am. Chem. Soc. 1986, 108, 3925-3927.
(61) Branchaud, B. P.; Yu, G.-X. Organometallics 1993, 12, 4262-
4264.

Reductive Dimerization of Arylalkenes
J . Org. Chem., Vol. 67, No. 3, 2002 841
intercepting the alkyl radical using trapping agents such
as TEMPO,62,63 O2,36 and bis(dimethylglyoximato)co-
balt(II)64,65 or by removing cob(II)alamin via oxidation to
Co(III).61 In the current system, we speculate that the
persistent radical effect is circumvented by reduction of
cob(II)alamin back to the catalytically active Co(I) form.
We experimentally verified this hypothesis by prepar-
ing authentic benzylcobalamin from cob(I)alamin and
benzylbromide and submitting the isolated complex to a
solution containing Ti(III)citrate. Whereas an anaerobic
control reaction showed no appreciable decrease of the
characteristic bands of an alkylcobalamin in the UV-
vis spectrum over a period of 3 h, in the presence of 8
equiv of Ti(III)citrate benzylcobalamin was converted into
cob(I)alamin. These observations can be rationalized
according to Scheme 2 in which Ti(III)citrate intercepts
cob(II)alamin produced in the rate-limiting step (k2), or
they may be interpreted as a bimolecular reaction in
which Ti(III)citrate directly reduces benzylcobalamin.
Such Co-C bond cleavage induced by one-electron reduc-
tion of alkylcobalamins has been extensively studied,66-68
showing that the reduction becomes more facile with
sterically hindered alkyl groups and/or groups that can
stabilize the radical product by delocalization.69 Both
features are present in alkylcobalamins 3, supporting a
potential bimolecular reduction by Ti(III)citrate.
F igu r e 1. Time dependence of the conversion of benzylcobal-
amin to cob(I)alamin by Ti(III)citrate monitored at 500 nm.
Conditions: 0.2 mM benzylcobalamin, 9.7 mM Ti(III)citrate,
50 mM Tris buffer, pH 8.0. The solid line represents a single-
exponential fit using the rate expression for a first-order decay.
Inset shows the linearity of the logarithmic plot over several
half-lives.
A distinction between these two possibilities can be
made on the basis of the predicted kinetic profiles.
According to Scheme 2 the rate of benzylcobalamin
consumption should be independent of Ti(III)citrate
concentration at sufficiently high concentrations of the
latter when homolytic Co-C bond cleavage is fully rate
limiting. On the other hand, a bimolecular mechanism
should be first order in Ti(III)citrate at all concentrations.
Monitoring the reduction of 0.2 mM benzylcobalamin
in the presence of various concentrations of Ti(III)ci-
trate at pH 8.0 resulted in apparent first-order decay of
the absorption bands of benzylcobalamin (eg Figure 1)
and a concomitant increase in the characteristic bands
of cob(I)alamin with an isosbestic point at 424 nm. The
observed rate constants increased nonlinearly with Ti-
(III)citrate concentration, leveling off above 75 equiv to
become zeroth order in Ti(III)citrate (Figure 2). The
observed first-order rate constant at these Ti(III)citrate
concentrations (1.68 ( 0.1 × 10-3 s-1 at 22 °C) is within
experimental error of the rate constant for Co-C bond
cleavage of benzylcobalamin (k2, Scheme 2) determined
in various studies36,38,40,56 (eg 1.79 × 10-3 s-1 at 22 °C).36
Thus, the kinetic data is fully consistent with the
mechanism presented in Scheme 2, indicating that Ti-
(III)citrate “traps” cob(II)alamin once formed.70
F igu r e 2. Dependence of the observed apparent first-order
rate constant on the Ti(III)citrate concentration in the anaero-
bic reductive dealkylation of benzylcobalamin (0.2 mM) in
aqueous buffer, pH 8.0. The observed rate constant approaches
zero order dependence on the concentration of Ti(III)citrate.
Our studies show that reduction of cob(II)alamin
formed during the coupling reaction not only regenerates
the active Co(I) catalyst (Scheme 2) but also prevents the
highly efficient (re)combination of cob(II)alamin with
organic radicals, which would otherwise interfere with
the catalytic cycle. This stands in direct contrast to
cobalt-catalyzed free-radical chain transfer71-73 and cobalt-
mediated living radical polymerization28,74 in which the
cobalt macrocycles are introduced to decrease the steady-
state concentration of free radicals. In other words, the
current process promotes the rate of the “termination”
reaction of combination of two radicals that is prevented
in these polymerization strategies. In comparison with
previously employed reducing agents for the catalytic use
(62) Finke, R. G.; Hay, B. P. Inorg. Chem. 1984, 23, 3041-3043.
(63) Hay, B. P.; Finke, R. G. Polyhedron 1988, 7, 1469-1481.
(64) Halpern, J .; Kim, S. H.; Leung, T. W. J . Am. Chem. Soc. 1984,
106, 8317-8319.
(65) Halpern, J . Polyhedron 1988, 7, 1483-1490.
(66) Lexa, D.; Save´ant, J . M. J . Am. Chem. Soc. 1978, 100, 3220-
3222.
(67) Elliott, C. M.; Hershenhart, E.; Finke, R. G.; Smith, B. L. J .
Am. Chem. Soc. 1981, 103, 5558-5566.
(68) Shepherd, R. E.; Zhang, S.; Dowd, P.; Choi, G.; Wilk, B.; Choi,
S. C. Inorg. Chim. Acta 1990, 174, 249-256.
(69) Zhou, D.-L.; Tinembart, O.; Scheffold, R.; Walder, L. Helv. Chim.
Acta 1990, 73, 2225-2241.
(71) Enikolopyan, N. S.; Smirnov, B. R.; Ponomarev, G. V.; Bel-
govskii, I. M. J . Polymer Sci. 1981, 19, 879-889.
(72) Burczyk, A.; O’Driscoll, K. F.; Rempel, G. L. J . Polym. Sci. 1984,
22, 32255-3262.
(73) Gridnev, A. A. Polym. Sci. U.S.S.R. 1989, 31, 2369-2376.
(74) Matyjaszewski, K.; Patten, T. E.; Xia, J . H. J . Am. Chem. Soc.
1997, 119, 674-680.
(70) This conclusion is valid for benzylcobalamin. More sterically
hindered alkylcobalamins generally have lower reduction potentials.
Therefore, Ti(III)citrate might be a sufficiently strong reductant to
reduce more hindered alkylcobalamins in
a
bimolecular reaction.
Because of the inherent instability of hindered secondary alkylcobal-
amins we could not verify this possibility experimentally.

842 J . Org. Chem., Vol. 67, No. 3, 2002
Shey et al.
Sch em e 4
F igu r e 3. (A) Schematic representation of the steric require-
ments of the corrin ligand in B12 derivatives. Only the side
chains at the â-face are shown for clarity, and the pucker of
the corrin is not shown. The arrows indicate the general
direction of the CR-Câ bond of the axial ligand in primary
alkylcobalamins for which structures have been determined.
(B) Proposed approach of mono- and 1,1-disubstituted alkenes
to a putative hydridocobalamin.
compound 10 was subjected to the reaction conditions.
We have previously shown that trichloroalkenes undergo
vitamin B12-catalyzed dechlorination that is initiated by
electron transfer from Co(I) to the electron-deficient
trichloroalkene.78 The radical anion so produced is un-
stable and eliminates chloride anion, generating a vinyl
radical.79 Compound 10 contains both a trichloroalkene
and a 1,1-disubstituted arylalkene and thus can serve
as a probe for the relative rates of dimerization and
dechlorination. In the event, dimerized product 11 was
detected as the major80 product as a 1:1 mixture of
diastereomers (eq 4), and only traces of dechlorinated
of vitamin B12 in C-C bond forming reactions, such as
NaBH4/NaOH,22,23 Zn,24,26 or electrochemistry,6,25,27,75 Ti-
(III)citrate is an attractive alternative as it is a signifi-
cantly less strong reductant. Ti(III)citrate is not required,
however, for these reactions as similar yields and product
distributions are obtained using Zn dust as the stoichio-
metric reductant.
Mech a n ism of Ra d ica l F or m a tion . The results
discussed so far are consistent with mechanisms for
product formation that involve radicals 4, and the find-
ings in the previous section support formation of these
benzylic radicals from benzylcobalamins. The pathway
for formation of these radicals in the dimerization of
R-substituted styrenes warrants further discussion. Ac-
cording to Schemes 2 and 3, these substrates would form
alkylcobalamins with R-branched tertiary ligands (3, R
) alkyl), which have never been detected before.76 Steri-
cally crowded alkylcobalamins have unfavorable interac-
tions between the axial alkyl ligand and the substituents
on the equatorial corrin (see Figure 3), and therefore
tertiary alkylcobalamins may be so unstable that they
are not viable intermediates. For instance, Scheffold and
co-workers reported preliminary computational studies
that suggested that no low-energy solution is possible for
tert-alkylcobalamins.69 More recent theoretical studies on
Co-C bond dissociation energies of alkylcobalamins
suggest that this bond would be weaker by about 15-20
kcal/mol for tertiary alkyl ligands compared to secondary
and primary alkyl ligands.41 The bond energies calculated
for tertiary alkylcobalamins are so low that again their
intermediacy is questionable.77
products were produced. Whereas this does not prove that
the dimerization reaction does not involve an electron
transfer from Co(I) to the arylalkene, it is suggestive
because the trichloroalkene should be significantly more
electron-deficient. Another piece of indirect evidence
against an outer-sphere electron-transfer mechanism
involves the lack of reactivity of 1,2-di- and trisubstituted
alkenes (Chart 1), which would not be expected to have
significantly different redox potentials compared to mono-
and 1,1-disubstituted alkenes. If an inner-sphere et or
covalent process were responsible for radical generation,
however, the difference in reactivity can be readily
explained by unfavorable steric interactions between the
side chains at the â-face of the corrin ring and the alkene
substituents for 1,2-substituted olefins (see below).
Several mechanisms can be considered for the forma-
tion of benzylcobalamins from cob(I)alamin and styrenes.
As shown in Scheme 4, alkylcobalamins (as well as other
18-electron 6-coordinate organocobalt macrocycles)81-85
Electron transfer (et) from Co(I) to the alkene would
constitute an alternative route to radicals 4 that does not
involve alkylcobalamins. To investigate such pathways
(78) Shey, J .; van der Donk, W. A. J . Am. Chem. Soc. 2000, 122,
12403-12404.
(79) Nonnenberg, C.; van der Donk, W. A.; Zipse, H., submitted for
publication.
(75) Torii, S.; Inokuchi, T.; Yukawa, T. J . Org. Chem. 1985, 50,
5875-5877.
(76) A transient species detected in electrochemical experiments has
been tentatively assigned to tert-butylcobalamin. See ref 69.
(77) For instance, in ref 41 the BDE of tert-butylcobalamin was
computed at 3 kcal/mol.
(80) In addition to 60% R,R,-dimer, about 30% of an R,p-dimer was
produced. See ref 47 for literature precedent.
(81) Ng, F. T. T.; Rempel, G. L.; Mancuso, C.; Halpern, J . Organo-
metallics 1990, 9, 2762-2772.
(82) Gridnev, A. A.; Ittel, S. D.; Fryd, M.; Wayland, B. B. Organo-
metallics 1993, 12, 4871-4880.

Reductive Dimerization of Arylalkenes
J . Org. Chem., Vol. 67, No. 3, 2002 843
Sch em e 5
are known to undergo â-hydride elimination to generate
alkenes. The mechanism of these reactions have been
investigated,9,81,86 and most reports favor homolytic scis-
sion of the Co-C bond to provide a [LCo(II)• •CH2R]
geminate radical pair followed by hydrogen atom ab-
straction from the organic radical by the Co(II)-complex
to give an alkene and a protonated Co(I) complex (Scheme
4).87 For cobaloximes the resulting metal hydride (“hy-
dridocobaloxime”) has a reported pKa of around 10.5.88
Thus, according to the principle of microscopic revers-
ibility, the reverse reaction, the formation of alkylco-
baloximes from cob(I)aloxime and alkenes, can be for-
mulated as a two-step addition of a cobalt-hydride (i.e.
step -b, followed by step -c, Scheme 4). Such hydroco-
baltation has been used to prepare primary alkylco-
baloximes from alkenes.89 A similar route can also be
proposed for the formation of R-substituted benzylcobal-
amins from substituted styrenes. However, the combina-
tion of cob(II)alamin with the tertiary radical formed by
hydrogen atom transfer might be difficult, leading to
either return to the alkene or collapse of the solvent cage
and diffusion of the benzylic radical to give dimers. In
this mechanism, the unstable tertiary alkylcobalamins
are either formed only transiently or not at all. One
problem with this mechanism involves the pKa of “hy-
dridocobalamin”,90 which has been reported to be signifi-
cantly lower, ∼1,91 than that of hydridocobaloxime. Thus,
under the conditions used for the dimerization of alkenes
(pH ) 8) the concentration of protonated cob(I)alamin is
extremely low.
ted.37-39,97 Thus, hydrogen atom transfer from a cobalt-
hydride to the terminal carbon in mono- and 1,1-disub-
stituted olefins is sterically feasible and produces a
resonance stabilized benzyl radical (Figure 3B). On the
other hand, 1,2-di- and trisubstituted alkenes may not
show reactivity (Chart 1) because highly unfavorable
interactions would develop between the alkene and the
â-side chains of the corrin during the approach trajectory.
Deu ter iu m La belin g Stu d ies. An alternative path-
way to alkylcobalamins in the reaction between alkenes
and cob(I)alamin involves nucleophilic attack followed by
or concomitant with protonation by solvent (Scheme 5).
In the former case, the reverse reaction would consist of
an E1cb mechanism, which would be significantly uphill
given the expected low acidity of the methyl protons in
3, and thus the reaction would not be readily reversible.
To test for any reversibility, we performed the dimer-
izations in D2O and terminated the process before
One other indirect piece of support favors the mecha-
nism in Scheme 4. X-ray and NMR analysis of primary
alkylcobalamins has shown that the rotation around the
Co-C bond is restricted because of steric interactions
with the side chains on the â-face of the equatorial corrin
ligand.1,92-96 The preferred orientation of the alkyl ligand
places its â-carbon between rings C and D where the
steric interaction is most favorable (Figure 3). Even alkyl
ligands that are trisubstituted at the â-carbon, such as
a neopentyl group (-CH2CMe3), can be accommoda-
complete conversion of the starting alkene. H and 2H
1
NMR analysis of the recovered alkenes clearly demon-
strated deuterium incorporation at both terminal posi-
tions of the olefin in equal amounts (eqs 5-7). Further-
more, with R-methylstyrene the 2H NMR spectrum
displayed signals for the two terminal vinyl positions as
well as the methyl group (integration ratio 1:1:3). Mass
spectrometric analysis indicated that the recovered al-
kenes contained both mono- and dideuterated molecules,
and as expected, deuterium was also present in the
methyl groups of the products. Interestingly, the level of
deuterium incorporation in the starting alkenes at simi-
lar conversion was dependent on the reductant (eqs 5 and
6), with a larger extent of labeling observed with Zn. In
a separate experiment it was found that qualitatively,
reduction of cob(II)alamin is much faster with Ti(III)-
citrate than with Zn powder under the reaction conditions
used. This difference can explain the higher extent of
labeling in the presence of Zn according to the mechanism
(83) Ng, F. T. T.; Rempel, G. L. J . Am. Chem. Soc. 1982, 104, 621-
623.
(84) Schrauzer, G. N.; Windgassen, R. J . J . Am. Chem. Soc. 1967,
89, 1999-2007.
(85) Nguyen Van Duong, K.; Ahond, A.; Merienne, C.; Gaudemer,
A. J . Organomet. Chem. 1973, 55, 375-82.
(86) Garr, C. D.; Finke, R. G. J . Am. Chem. Soc. 1992, 114, 10440-
10445.
(87) This mechanism stands in contrast to a concerted four-center
process that is preferred for other late transition metal complexes. Such
a concerted pathway has also been proposed for B12 (ref 105), but
several studies have provided experimental evidence against this
mechanism. See refs 9, 81, and 86.
(88) Schrauzer, G. N.; Holland, R. J . J . Am. Chem. Soc. 1971, 93,
1505-1506.
(89) (a) Bhandal, H.; Pattenden, G. J . Chem. Soc., Chem. Commun.
1988, 1110-1112. (b) Howell, A. R.; Pattenden, G. J . Chem. Soc.,
Perkin Trans. 1 1990, 2715-2720.
(90) Hydridocobalamin is a rather ill-defined species altogether. It
has been described as cob(I)alamin protonated on the electron pair in
the dz2 orbital or as a Co(III) hydride. For a discussion of this species,
see: (a) refs 29 and 91. (b) Chemaly, S. M.; Pratt, J . M. J . Chem. Soc.,
Dalton Trans. 1984, 595-599. (c) Aleyunas, Y. W.; Fleming, P. E.;
Finke R. G.; Pagano, T. C.; Marzilli, L. G. J . Am. Chem. Soc. 1991,
113, 3781.
(91) Lexa, D.; Save´ant, J . J . Chem. Soc., Chem. Commun. 1975,
872-874.
(92) Lenhert, P. G. Proc. R. Soc. London, Ser. A 1968, 303, 45-84.
(93) Alcock, N. W.; Dixon, R. M.; Golding, B. T. J . Chem. Soc., Chem.
Commun. 1985, 603-605.
(95) Deuterium incorporation into recovered alkene via hydrogen
atom abstraction by cob(II)alamin from an intermediate radical inside
a solvent cage has precedent in B12 chemistry. Garr and Finke showed
the formation of a â-hydride elimination product from adenosylcobal-
amin (ref 86). For a detailed discussion of radical cage effects in
organotransition metal chemistry, see: Koenig, T. W.; Hay, B. P.;
Finke, R. G. Polyhedron 1988, 7, 1499-1516.
(96) Bax, A.; Marzilli, L. G.; Summers, M. F. J . Am. Chem. Soc.
1987, 109, 566-574.
(97) Mueller, S.; Wolleb, A.; Walder, L.; Keese, R. Helv. Chim. Acta
1990, 73, 1659-1668.
(94) (a) Pagano, T. G.; Marzilli, L. G.; Flocco, M. M.; Tsai, C.; Carrell,
H. L.; Glusker, J . P. J . Am. Chem. Soc. 1991, 113, 531-542. (b)
Wagner, T.; Afshar, C. E.; Carrell, H. L.; Glusker, J . P.; Englert, U.;
Hogenkamp, H. P. C. Inorg. Chem. 1999, 38, 1785-1794.

844 J . Org. Chem., Vol. 67, No. 3, 2002
Shey et al.
reaction (see above), whereas a concerted nucleophilic
addition at the benzylic carbon seems exceedingly dif-
ficult for 1,1-disubstituted alkenes given the steric re-
quirements of the equatorial corrin ligand.
Con clu sion
In summary, we report a new multicomponent reaction
that couples alkenes with defined regiochemistry under
mild conditions in environmentally benign solvent sys-
tems such as ethanol/water. Our mechanistic studies
indicate that organic free radicals are formed in this
reaction. In the current process, the rate of combination
of two radicals is increased by removing the persistent
Co(II) radical with Ti(III) or Zn. This is in contrast to
cobalt-catalyzed free-radical chain transfer and cobalt-
mediated living radical polymerization in which the
cobalt macrocycles function to reduce the steady-state
concentration of free radicals. A disadvantage of the
mechanism of dimerization from a synthetic perspective
is the lack of diastereoselectivity that is inherent to
radical combination.13,58 The intramolecular reaction of
dienes 6 and 7 suggests, however, that the transforma-
tion may find utility for the preparation of macrocycles.
Further work along these lines is in progress.
in Scheme 4. Removal of cob(II)alamin to give cob(I)-
alamin (kred, Scheme 4) is less efficient with Zn and
therefore is less competitive with the return of the radical
intermediate to the starting alkene (kb), leading to more
deuterium incorporation into the alkene. In a semiquan-
titative analysis, comparison of the level of deuterium
incorporation in the recovered alkene and the amounts
of formation of product (one molecule of product is formed
by two cage-escaped radicals) shows that regeneration
of alkene is about 2-fold faster than product formation
in eq 6.95
Exp er im en ta l Section
Gen er a l. All NMR spectra were recorded on Varian U400,
U500, or UI500NB spectrometers. 1H spectra were referenced
to CHCl3 at 7.26 ppm, and 13C spectra were referenced to
CDCl3 at 77.23 ppm unless otherwise indicated. Mass spec-
trometry was carried out by the Mass Spectrometry Laboratory
at the University of Illinois at Urbana-Champaign (UIUC).
Elemental analysis was performed by the Microanalysis
Laboratory at UIUC. Infrared (IR) spectra were taken on a
Mattson Galaxy Series FTIR 5000. GC and gas chromatogra-
phy-mass spectrometry (GC-MS) data were collected on a
Hewlett-Packard 5890A GC equipped with a 25 m HP-1 fused
silica capillary column and a HP 5970 Series mass selective
detector. The anaerobic chamber used was supplied by Coy
Laboratory Products, Inc. The inside atmosphere was main-
tained at 6% H2 and 94% N2 and was deoxygenated with
palladium catalysts. All dimerization reactions were performed
in the Coy anaerobic chamber in vessels covered with alumi-
num foil and initiated in the dark under red light. Solvents
used for dimerizations were deoxygenated under reduced
pressure (10 min) and purged with N2; this was repeated three
times before storage in the anaerobic chamber. When possible,
reactions were monitored by thin-layer chromatography on
Merck silica gel 60 F254 plates. High performance liquid
chromatography (HPLC) was performed on Rainin Dynamax
SD-200 pumps equipped with a Dynamax UV-1 absorbance
detector and a Supelcosil LC-PAH column. Compounds and
solvents were obtained from Fisher, Aldrich, or Chemcycle at
UIUC. Ti(III) chloride (Al-reduced) was purchased from Strem
Chemical Company. Anhydrous MgSO4 was used after workup
for drying organic solutions. UV-vis studies were carried out
using an Olis RSM 2000 instrument. No photolysis of samples
was observed under the conditions of the spectroscopic analy-
sis.
Aside from the mechanisms discussed, one other
pathway could lead to label incorporation into starting
material. Disproportionation of two radicals (eq 8) is well-
known and is likely to take place to some extent.98 For
every deuterated alkene regenerated by this pathway,
one molecule of alkane is formed irreversibly. Hence, this
pathway is relatively unimportant for the deuterium
labeling studies discussed here since significantly larger
amounts of deuterated alkenes than reduced products
were present after partial conversions (eqs 5-7). With
respect to the pathways in Scheme 5, collectively the
deuterium labeling experiments disfavor a stepwise
nucleophilic addition-protonation since the reverse reac-
tion would be very unfavorable, inconsistent with the
high extent of label incorporation via such a reverse
reaction. On the other hand, the results of the deuterium
labeling experiments are consistent with a reversible
process up until the actual dimerization and support
either hydrogen atom transfer from a hydridocobalamin
species (Scheme 4) or a concerted nucleophilic attack and
protonation (Scheme 5). In the latter mechanism, the
reverse reaction that leads to deuterium incorporation
would be an E2 pathway that may be more favorable
than an E1cb mechanism. As a working model, we favor
the hydrogen atom transfer pathway as it can readily
explain the chemoselectivity and regioselectivity of the
Typ ica l P r oced u r e for Dim er iza t ion . A flask covered
with Al foil was charged with 0.01 mmol (10 mol %) of
cyanocobalamin under a nitrogen atmosphere, followed by 7.0
mL of 50 mM aqueous Tris buffer (pH 8.0) and 5.5 mL of a
0.18 M aqueous Ti(III)citrate solution containing tetrabu-
tylammonium hydroxide as phase transfer catalyst.78 The
substrate (0.1 mmol), dissolved in 12.5 mL of EtOH, was added
to the B12 solution. After 1-24 h depending on the substrate,
the reaction mixture was extracted with hexanes (3 × 20 mL),
and the combined organic fractions were washed with water
(98) For instance, radical disproportionation may well be a major
route to the reduced products in Table 1.

Reductive Dimerization of Arylalkenes
J . Org. Chem., Vol. 67, No. 3, 2002 845
(25 mL) and dried (MgSO4), and the solvent was evaporated
under reduced pressure. The crude products were analyzed
by GC-MS and NMR to determine product ratios prior to
purification.
numbering and copies of 1H NMR spectra, see Supporting
Information.
1-(3,3-Dip h en yl-2-p r op en yl)-2-a cetylp yr r ole (13). 3-Bro-
mo-1,1-diphenyl-1-propene was synthesized following litera-
ture procedures.100,101 2-Acetylpyrrole was reacted with 3-bromo-
1,1-diphenyl-1-propene following the procedure of Molander
and Schmitt, except only 1 equiv of the bromide was used.99
After purification by column chromatography using 7% EtOAc
in hexanes (Rf ) 0.26), the desired product was obtained in
Syn th etic Rou te to 1-(3-Meth yl-2-bu ten yl)-2-isop r op e-
n ylp yr r ole (6). For synthetic scheme with compound num-
1
bering and copies of H NMR spectra, see Supporting Infor-
mation.
1-(3-Meth yl-2-bu ten yl)-2-a cetylp yr r ole (12). The proce-
dure of Molander and Schmitt was followed using 2-acetylpyr-
role and prenyl bromide.99 The yield was 76% (clear oil) after
purification by column chromatography eluting with 10%
EtOAc in hexanes (Rf ) 0.39). 1H NMR (499.7 MHz) δ 1.74
(m, 3H), 1.75 (m, 3H), 2.43 (s, 3H), 4.95 (d, J ) 7.0 Hz, 2H),
5.34 (m, 1H), 6.12 (m, 1H), 6.90 (m, 1H), 6.96 (m, 1H). 13C NMR
(125.7 MHz) δ 188.7 (Cq), 136.5 (Cq), 130.4 (Cq), 129.4 (CH),
120.7 (CH), 120.3 (CH), 108.2 (CH), 47.2 (CH2), 27.5 (CH3),
25.8 (CH3), 18.2 (CH3). IR (neat) 1650.2, 1406.8, 739.4, 1330.3,
1085.0, 1234.2, 943.6, 1468.5, 632.9, 1360.34. MS m/z (rel int,
%) FI 177 (100), 178 (13). HRMS (EI) calcd for C11H15NO
177.1154, found 177.1156.
1-(3-Meth yl-2-bu ten yl)-2-isop r op en ylp yr r ole (6). To a
100 mL round-bottom flask was added 1.16 mmol (470 mg)
CH3PPh3I. The flask was gently flame-dried and purged with
Ar. After cooling to room temperature, dry, deoxygenated THF
(4 mL) was added to the flask, which was then immersed into
a -78 °C bath. A solution of n-BuLi in hexanes (0.97 mmol)
was added, and the reaction mixture turned bright yellow.
After 1 h, compound 12 (0.58 mmol) in 4 mL of dry, deoxy-
genated THF was added dropwise via cannula from a 15 mL
conical flask. The reaction was incomplete by TLC after the
addition, so it was allowed to warm to 0 °C. When the reaction
did not proceed any further as judged by TLC, 25 mL of half
saturated NH4Cl was added, and the reaction was extracted
with 3 × 25 mL of ether. The pooled organic layers were
washed with 2 × 25 mL saturated NaCl, dried, and concen-
trated. The compound was purified by silica gel column
chromatography using 5% EtOAc in hexanes (Rf 0.44). Yield
47% (light yellow oil). 1H NMR (499.7 MHz) δ 1.75 (m, 3H),
1.78 (m, 3H), 2.11 (m, 3H), 4.59 (d, J ) 6.6 Hz, 2H), 4.95 (q, J
) 0.8 Hz, 1H), 5.13 (pentet, J ) 1.5 Hz, 1H), 5.35 (m, 1H),
6.16 (m, 2H), 6.69 (m, 1H). 13C NMR (125.7 MHz) δ 136.1 (Cq),
135.3 (Cq), 134.7 (Cq), 122.6 (CH), 121.7 (CH), 112.5 (CH2),
108.3 (CH), 107.5 (CH), 45.8 (CH2), 25.8 (CH3), 24.3 (CH3), 18.1
(CH3). IR (neat) 712.4, 2920.3, 2970.2, 1445.6, 1376.0, 1311.4,
1472.8, 1077.4, 885.0, 1626.7. MS m/z (rel int, %) FI 175 (100),
1
68% yield (viscous yellow oil). H NMR (400.0 MHz) δ 2.46 (s,
3H), 5.02 (d, J ) 6.9 Hz, 2H), 6.13 (dd, J ) 3.0, 2.5 Hz, 1H),
6.29 (t, J ) 7.0 Hz, 1H), 6.78 (dd, J ) 2.4, 1.7 Hz, 1H), 6.98
(dd, J ) 4.0, 1.7 Hz, 1H), 7.24 (m, 7H), 7.40 (m, 3H). 13C NMR
(100.6 MHz) δ 188.7 (Cq), 144.6 (Cq), 141.5 (Cq), 139.2 (Cq),
130.3 (Cq), 129.9 (CH), 129.7 (CH), 128.7 (CH), 128.3 (CH),
127.8 (CH), 127.6 (CH), 124.8 (CH), 120.4 (CH), 108.5 (CH),
48.6 (CH2), 27.5 (CH3). IR (neat) 1649.4, 742.5, 1406.4, 761.9,
1328.1, 1234.2, 1087.8, 1363.0, 1443.8, 1466.8. MS m/z (rel int,
%) FI 301 (100), 302 (24). HRMS (EI) calcd for C21H19NO
301.1467, found 301.1473.
1-(3,3-Dip h en yl-2-p r op en yl)-2-isop r op en ylp yr r ole (7).
The procedure described for the synthesis of 6 was followed
using compound 13. After purification by column chromatog-
raphy eluting with 5% EtOAc in hexanes (Rf ) 0.5 in 10%
EtOAc in hexanes), 7 was obtained in 26% yield (clear oil). 1H
NMR (499.7 MHz) δ 2.06 (dd, J ) 1.6, 0.9 Hz, 3H), 4.69 (d, J
) 6.7 Hz, 2H), 4.86 (q, J ) 0.7 Hz, 1H), 5.04 (quintet, J ) 1.5
Hz, 1H), 6.16-6.19 (m, 2H), 6.20 (t, J ) 6.8 Hz, 1H), 6.71 (dd,
J ) 2.5, 1.9 Hz, 1H), 7.22-7.45 (m, 10H). 13C NMR (125.7
MHz) δ 144.1 (Cq), 141.7 (Cq), 139.0 (Cq), 135.9 (Cq), 134.9
(Cq), 129.9 (CH), 128.7 (CH), 128.4 (CH), 128.0 (CH), 127.9
(CH), 127.7 (CH), 125.5 (CH), 122.8 (CH), 112.8 (CH2), 108.5
(CH), 107.8 (CH), 47.1 (CH2), 24.3 (CH3). IR (neat) 733.6, 702.2,
716.4, 909.5, 759.8, 1444.2, 1493.8, 1312.7, 1248.5, 1472.8. MS
m/z (rel int, %) FI 299 (100), 300 (44). HRMS (EI) calcd for
C
22H21N 299.1674, found 299.1672.
Rea ction of 7 w ith Vita m in B12 a n d Ti(III)citr a te. The
reaction provided exclusively the cyclized product 9 in 80%
yield (Rf ) 0.5 in 10% EtOAc in hexanes). 1H NMR (499.7
MHz) δ 0.97 (s, 3H), 1.19 (s, 3H), 3.51 (t, J ) 10.5 Hz, 1H),
3.65 (ddd, J ) 11.9, 10.5, 7.4 Hz, 1H), 3.77 (dd, J ) 10.5, 7.3
Hz, 1H), 4.07 (d, J ) 11.7 Hz, 1H), 5.69 (dd, J ) 3.5, 1.2 Hz,
1H), 6.15 (t, J ) 3.0 Hz, 1H), 6.42 (dd, J ) 2.6, 1.3 Hz, 1H),
7.17-7.46 (m, 10H). 13C NMR (125.7 MHz) δ 147.7 (Cq), 144.3
(Cq), 143.2 (Cq), 129.1 (CH), 128.8 (CH), 128.5 (CH), 127.7
(CH), 126.94 (CH), 126.89 (CH), 113.1 (CH), 111.2 (CH), 96.5
(CH), 56.5 (CH), 53.0 (CH), 50.7 (CH2), 40.8 (Cq), 28.2 (CH3),
23.7 (CH3). IR (neat) 704.3, 1493.2, 2959.4, 747.1, 1452.3,
2926.3, 1463.9, 738.2, 761.6, 1303.8. MS m/z (rel int, %) FI
301 (100), 302 (47). HRMS (EI) calcd for C22H23N 301.1830,
found 301.1826. The control reaction resulted in recovered
starting material in 95%.
Syn th etic Rou te to 1,1,2-Tr ich lor o-5-p h en yl-1,5-h exa -
d ien e (10). For synthetic scheme with compound numbering
and copies of 1H NMR spectra, see Supporting Information.
4-P h en y-4-p en t en -1-a l Dim et h yl Acet a l (15). Com-
pounds 14 and 16 were prepared following literature proce-
dures.102-104 In a 250 mL round-bottom flask, Ph3PCH3I (1.81
g, 4.5 mmol) and dry THF (54 mL) were combined and cooled
to -78 °C. A 1.6 M solution of butyllithium (0.175 g, 3.44 mL)
in hexane was added dropwise, and the resulting solution was
allowed to stir for 30 min at -78 °C. Compound 14 (0.547 g,
2.6 mmol) in dry THF (54 mL) was added dropwise to the
reaction solution. The solution was allowed to warm to 25 °C
over 6 h. The solution was quenched with saturated NH4Cl/
176 (14). HRMS (EI) calcd for
175.1359.
C12H17N 175.1361, found
Rea ction of 6 w ith Hyd r oxocoba la m in a n d Ti(III)-
citr a te. The reaction using the general conditions described
above provided cyclized products 8a and 8b in a 3:2 ratio (1H
NMR and GC-MS), which could not be separated by chroma-
tography. The combined yield of both products was 70%.
Cyclized product 8a : 1H NMR (499.7 MHz) δ 1.07 (s, 3H), 1.41
(s, 3H), 1.75 (s, 3H), 3.12 (t, J ) 7.9 Hz, 1H), 4.01 (m, 2H),
4.79 (m, 1H), 4.95 (pentet, J ) 1.5 Hz, 1H), 5.73 (dd, J ) 3.6,
1.3 Hz, 1H), 6.19 (t, J ) 2.9 Hz, 1H), 6.53 (dd, J ) 2.6, 1.3 Hz,
1H). 13C NMR (125.7 MHz) δ 143.0 (Cq), 113.6 (CH2), 113.1
(CH), 111.5 (CH), 96.7 (CH), 60.5 (CH), 48.7 (CH2), 41.1 (Cq),
28.6 (CH3), 23.8 (CH3), 23.0 (CH3). HRMS (EI) calcd for
C
12H17N 175.1361, found 175.1362. Cyclized and reduced
product 8b: 1H NMR (499.7 MHz) δ 0.93 (d, J ) 6.6 Hz, 3H),
1.06 (d, J ) 6.9 Hz, 3H), 1.14 (s, 3H), 1.43 (s, 3H), 1.89 (m,
1H), 2.21 (td, J ) 10.0, 7.8 Hz, 1H), 3.60 (t, J ) 10.3 Hz, 1H),
4.03 (dd, J ) 10.3, 7.4 Hz, 1H), 5.70 (dd, J ) 3.4, 1.4 Hz, 1H),
6.17 (t, J ) 2.7 Hz, 1H), 6.48 (dd, J ) 2.5, 1.3 Hz, 1H). 13C
NMR (125.7 MHz) δ 112.8 (CH), 111.2 (CH), 96.2 (CH), 60.8
(CH), 50.1 (CH2), 40.3 (Cq), 28.9 (CH or CH3), 28.7 (CH or CH3),
23.1 (CH3), 22.5 (CH3), 22.4 (CH3). HRMS (EI) calcd for
(100) Simes, B. E.; Rickborn, B.; Flournoy, J . M.; Berlman, I. B. J .
Org. Chem. 1988, 53, 4613-4616.
(101) Davis, M. A.; Herr, F.; Thomas, R. A.; Charest, M. P. J . Med.
Chem. 1967, 10, 627-635.
C
12H19N 177.1518, found 177.1517.
(102) Barnhart, R. W.; Wang, X.; Noheda, P.; Bergens, S. H.;
Whelan, J .; Bosnich, B. J . Am. Chem. Soc. 1994, 116, 1821-1830.
(103) Smith, G. G.; Voorhees, K. J . J . Org. Chem. 1970, 35, 2182-
2185.
(104) Conole, G.; Mears, R. J .; De Silva, H.; Whiting, A. J . Chem.
Soc., Perkin Trans. 1 1995, 1825-1836.
Syn th etic Rou te to 1-(3,3-Dip h en yl-2-p r op en yl)-2-iso-
p r op en ylp yr r ole (7). For synthetic scheme with compound
(99) Molander, G. A.; Schmitt, M. H. J . Org. Chem. 2000, 65, 3767-
3770.

846 J . Org. Chem., Vol. 67, No. 3, 2002
Shey et al.
H2O (1:1; 50 mL) and combined with ether (50 mL). The
organic layer was separated, washed with H2O (100 mL) and
brine (100 mL), and dried over MgSO4. The product was
purified by flash column chromatography eluting with hexane/
ethyl acetate (38:1; Rf ) 0.25) yielding a transparent oil (0.37
g, 70%). Characterization matched the previous literature
data102 for this compound obtained via a different procedure.
anaerobic glovebox, two airtight Starna UV cell were charged
with 1.5 mL of 0.2 mM aqueous benzylcobalamin in 0.05 M
Tris buffer, pH 8.0. To one cell was added 10 µL of 0.175 M
Ti(III)citrate, whereas the second cell served as control. The
UV absorption spectra of both samples were monitored over
60 min, with the cell removed from the spectrophotometer and
stored in the dark between scans.
Gen er a l P r oced u r e for Deu ter iu m La belin g Stu d ies
Usin g Ti(III)citr a te. In an anaerobic chamber, a flask was
charged with vitamin B12 (135 mg, 0.10 mmol), 10 mL of EtOD,
5 mL of Tris buffer in D2O (pD 7.9), and 7.5 mL of 0.26 M
Ti(III)citrate in D2O. The reaction mixture was stirred for 10
min prior to the addition by gastight syringe of styrene or
R-methylstyrene (0.44 mmol). After 20 min, the aqueous
solution was extracted with hexanes (3 × 30 mL), and the
combined organic layers were washed with water (3 × 40 mL).
The organic layer was dried with MgSO4, filtered, and con-
centrated under reduced pressure while maintaining the flask
in an ice bath to avoid loss of volatile products. The deuterium
1,1,1-Tr ich lor o-5-p h en yl-5-h exen -2-ol (17). In a 25 mL
round-bottom flask, trichloroacetic acid (0.18 g, 1.1 mmol) and
sodium trichloroacetate (0.204 g, 1.1 mmol) were combined
with 16 (0.12 g, 0.74 mmol) in DMF (6 mL). After the mixture
stirred for 2 h, hexane/ether (9:1; 100 mL) was added, and the
solution was washed with saturated NaHCO3 (3 × 50 mL) and
brine (100 mL) and dried, yielding a light yellow oil (0.29 g,
1
96%). It was used without further purification. H NMR (400
MHz, CDCl3) δ 1.80 (1 H, m), 2.21 (1 H, m), 2.66 (1 H, m),
2.92 (1 H, m), 4.06 (1 H, dd, J ) 9.98, 1.65 Hz), 5.16 (1 H,
bm), 5.36 (1H, bd), 7.29 (1 H, m), 7.35 (2 H, m), 7.42 (2 H, m).
13C NMR (125 MHz, APT, CDCl3 referenced to 77.00 ppm) δ
29.9 (CH2), 31.4 (CH2), 82.2 (CH), 113.4 (CH2), 126.1 (CH),
127.6 (CH), 128.4 (CH), 140.5 (Cq), 147.1 (Cq). FT-IR (CCl4)
3588, 3055, 2986, 749, 732 cm-1. HRMS (EI) calcd for
C12H13O1Cl3 278.003198, found 278.002871.
2
incorporation was measured using both 1H and H NMR and
MS. Deuterium incorporation at the vinyl positions of styrene
or R-methylstyrene caused chemical shifts in the vinyl protons
in the 1H NMR. Thus, integration of the signals of vinyl
protons in nondeuterated and monodeuterated products in the
crude reaction mixture were used to provide the ratios of these
products. Note that products containing two vinylic deuteriums
will not give vinylic proton resonances and hence cannot be
quantitated this way. Multiply deuterated products were in
fact detected by mass spectrometry, but evaluation of several
available ionization techniques did not result in reproducible
quantitation of product ratios. Therefore, the percentages
deuterium incorporation shown in eqs 5-7 are based only on
NMR data and present an underestimate of the total deute-
rium incorporation. Incorporation into the products and methyl
group of R-methyl styrene was quantitated by 2H NMR before
and after purification. In the case of the methyl group of
R-methyl styrene the deuterium incorporation was integrated
and quantitated in comparison to the integration of the vinylic
deuterium resonances.
Gen er a l P r oced u r e for Deu ter iu m La belin g Stu d ies
Usin g Zin c. A flask containing vitamin B12 (135 mg, 0.10
mmol) and Zn dust (1.25 g, 19.25 mmol) was introduced into
the anaerobic chamber. To this was added 10 mL of EtOD and
10 mL of 0.05 M Tris buffer in D2O. The suspension was stirred
in the dark for 6 h prior to the addition of styrene or
R-methylstyrene (0.87 mmol) via syringe. The reaction was
worked up after 6 h as described for reactions in which Ti-
(III)citrate was used as the reducing agent.
1,1,2-Tr ich lor o-5-p h en yl-1,5-h exa d ien e (10). In a 100
mL round-bottom flask, PPh3 (0.302 g, 1.15 mmol) was added
to dry CCl4 (22 mL) containing 17 (0.29 g, 1.05 mmol). The
system was heated to reflux for 3 d. After the CCl4 was
removed under reduced pressure, the reaction material was
passed through a silica plug eluting with hexane followed by
purification by HPLC. To the light yellow oil was added an
ethanolic NaOH solution (4 mg, 0.1 mmol in 0.5 mL), and the
solution was stirred for 20 h. Ether (50 mL) was added, and
the organic layer was separated, washed with saturated NH4-
Cl (50 mL) and brine (50 mL), and dried with MgSO4. The
product was purified by silica gel column chromatography
eluting with hexane followed by HPLC purification yielding a
light yellow oil. 1H NMR (400 MHz, CDCl3) δ 2.72 (2 H, m),
2.78 (2 H, m), 5.12 (1 H, d, J ) 1.17 Hz), 5.35 (1 H, d, J ) 1.17
Hz), 7.30 (1 H, m), 7.34 (2 H, m), 7.41 (2 H, m). 13C NMR (125
MHz, APT, CDCl3 referenced to 77.00 ppm) δ 32.4 (CH2), 35.3
(CH2), 113.6 (CH2), 117.6 (Cq), 126.1 (CH), 127.7 (CH), 128.4
(CH), 132.2 (CCl2), 140.1 (Cq), 146.1 (Cq). IR (CCl4) 3084, 3055,
2985, 2943, 1629, 1495, 1445, 1265, 896, 725, 707 cm-1. HRMS
(EI) calcd for C12H11Cl3 259.992634, found 259.992677.
Rea ction of 10 w ith Vita m in B12 a n d Ti(III)citr a te. The
general procedure described above was followed and provided
both diastereomers of 11 in 60% yield, which were purified
together using HPLC; Rf 0.21 in hexanes. Diastereomer A: 1H
NMR δ 7.2 (m, 6H), 6.99 (br d, J ) 6.6 Hz, 4H), 2.32 (m, 4H),
2.15 (m, 2H), 1.87 (m, 2H), 1.30 (s, 6H). Diastereomer B: 1H
NMR δ 7.2 (m, 6H), 6.89 (br s, 4H), 2.42 (td, J ) 12.5, 3.2 Hz,
2H), 2.32 (m, 2H), 2.12 (m, 2H), 1.79 (td, J ) 12.6, 4.6 Hz,
2H), 1.35 (s, 6H). HRMS (CI) calcd for C24H23Cl537Cl 522.9901,
found 522.9896.
Ack n ow led gm en t. This work was supported by the
Camille and Henry Dreyfus Foundation (NF-97-078),
the Petroleum Research Fund administered by the
American Chemical Society (33490-G), and the Beck-
man Foundation (Young Investigator Award to WAV).
J .S. was supported by an EPA STAR predoctoral fel-
lowship and W.A.V. is a Cottrell Scholar of the Research
Corporation and an Alfred P. Sloan Fellow.
Syn th esis a n d Red u ction of Ben zylcoba la m in . All
solvents were thoroughly degassed by five freeze-pump-thaw
cycles and stored under argon. Benzylcobalamin was prepared
as described previously,38,105 and was characterized by UV-
visible and 1H NMR spectroscopy. A description of its reduction
at one specific Ti(III)citrate concentration follows. In an
Su p p or tin g In for m a tion Ava ila ble: 1H NMR of com-
pounds 6-13 and 17 and schemes for the synthesis of
substrates 6, 7, and 10. This material is available free of
charge via the Internet at http://pubs.acs.org.
(105) Grate, J . H.; Schrauzer, G. N. J . Am. Chem. Soc. 1979, 101,
4601-4611.
J O0160470