Electrostatic and electrophilic catalysis in the reductive cleavage of alkyl aryl ethers. The influence of ion pairing on the regioselectivity

Article abstract of DOI:10.1021/jo991059v

Electrophilic and electrostatic catalysis have been identified as distinct contributions that affect the reactivity of radical anions in the reductive cleavage of alkyl aryl ethers. Two modes of mesolytic scission of these radical anions are possible: homolytic (dealkylation, a thermodynamically favored but kinetically forbidden process) and heterolytic (dealkoxylation). From our studies (alkali metal reductions, electrochemical studies, use of substrates with a preformed positive charge in certain positions of their structure) it can be concluded that the heterolytic scission is very much dependent on the electrophilic assistance by the counterion and it is only observed in contact ionic pairs with unsaturated cations (electrophilic catalysis). On the other hand, the homolytic scission is observed in solvent-separated ionic pairs, and it is especially efficient when the pair has a controlled topology with a tetralkylammonium cation (saturated cation) near the oxygen atom. The effect of the cation has, in this case, electrostatic origin (electrostatic catalysis), probably lowering the barrier of the intramolecular π*-σ* electron transfer process and thus reducing the kinetic control of the reaction in such a way that the thermodynamically more favorable process is produced.

Full text of DOI:10.1021/jo991059v

322
J . Org. Chem. 2000, 65, 322-331
Electr osta tic a n d Electr op h ilic Ca ta lysis in th e Red u ctive
Clea va ge of Alk yl Ar yl Eth er s. Th e In flu en ce of Ion P a ir in g on th e
Regioselectivity
Francisco Casado,† Luisa Pisano,‡ Maria Farriol,† Iluminada Gallardo,† J ordi Marquet,*,† and
Giovanni Melloni*,‡
Departament de Quı´mica, Universitat Auto`noma de Barcelona, 08193 Bellaterra, Barcelona, Spain,
Dipartimento de Chimica, Universita` di Sassari, Via Vienna 2, I-07100, Sassari, Italy
Received J uly 1, 1999
Electrophilic and electrostatic catalysis have been identified as distinct contributions that affect
the reactivity of radical anions in the reductive cleavage of alkyl aryl ethers. Two modes of mesolytic
scission of these radical anions are possible: homolytic (dealkylation, a thermodynamically favored
but kinetically forbidden process) and heterolytic (dealkoxylation). From our studies (alkali metal
reductions, electrochemical studies, use of substrates with a preformed positive charge in certain
positions of their structure) it can be concluded that the heterolytic scission is very much dependent
on the electrophilic assistance by the counterion and it is only observed in contact ionic pairs with
unsaturated cations (electrophilic catalysis). On the other hand, the homolytic scission is observed
in solvent-separated ionic pairs, and it is especially efficient when the pair has a controlled topology
with a tetralkylammonium cation (saturated cation) near the oxygen atom. The effect of the cation
has, in this case, electrostatic origin (electrostatic catalysis), probably lowering the barrier of the
intramolecular π*-σ* electron transfer process and thus reducing the kinetic control of the reaction
in such a way that the thermodynamically more favorable process is produced.
The electron distribution in reactive intermediates
Unimolecular fragmentation of radical ions to yield
radicals and ions (mesolytic cleavages) constitutes the
elementary step of many electron-transfer-initiated pro-
cesses.7 The rates of these reactions are usually signifi-
cantly faster than those observed for the homolytic
cleavage of the same bonds in neutral substrates.8,9
Alkali metals are known to induce cleavage of the C-O
bonds of aryl alkyl ethers under aprotic conditions in
various solvents.10 Cleavage of the alkyl-oxygen bond
with formation of phenols (dealkylation) is most often
observed; cleavage of the phenyl-oxygen bond (dealkoxy-
lation) is achieved only in particular cases,11 especially
in the presence of potassium and in solvents of low12 or
very low13 polarity. The first reaction step leads to radical
anions, ROAr.-, known since 1968 from ESR studies.14
Dianions were also discussed in the past as intermedi-
largely determines the outcome of a chemical or photo-
chemical reaction. In the case of charged intermediates,
however, the electron distribution is heavily influenced,
among other factors, by the interaction with the coun-
terion. As to the effect of the counterion on the outcome
of a chemical process (“metal ion catalysis”, “electrophilic
catalysis”, etc.),1 it has been traditionally attributed2 to
Lewis acid complexation, ignoring the important associ-
ated electrostatic effect. Only recently has this electro-
static effect been recognized as responsible for the lowest
energy conformation of the radical anion/cation pair in
alkyl aryl ethers,3 the acceleration of the electrocyclic
reactions by metal cation complexation,4 the strong
association of cations to aryl rings even in aqueous
solution,5 and the acceleration of the catalytic acylation
step by acetylcholinesterase.6
(7) (a) Savea´nt, J .-M. Tetrahedron, 1994, 50, 10117. (b) Maslak, P.
Top. Curr. Chem. 1993, 168, 1. (c) Save´ant, J .-M. Acc. Chem. Res. 1993,
26, 455.
* To whom correspondence should be addressed. For J .M.: phone,
34-3-5811029; fax, 34-3-5811265; e-mail, iqor3@cc.uab.es.
† Universitat Auto`noma de Barcelona.
(8) Chanon, M.; Rajzmann, M.; Chanon, F. Tetrahedron 1990, 46,
6193.
(9) (a) Maslak, P.; Narvaez, J . N.; Vallombroso, T. M., J r. J . Am.
Chem. Soc. 1995, 117, 12373. (b) Maslak, P.; Vallombroso, T. M.;
Chapman, W. H., J r.; Narvaez, J . N. Angew. Chem., Int. Ed. Engl.
1994, 33, 73. (c) Maslak, P.; Narvaez, J . N. Angew. Chem., Int. Ed.
Engl. 1990, 29, 283.
(10) (a) Azzena, U.; Denurra, T.; Fenude, E.; Melloni, G.; Rassu, G.
Synthesis, 1989, 28. (b) Thornton, T. A.; Woolsley. N. F.; Bartak, D. E.
J . Am. Chem. Soc. 1986, 108, 6497. (c) Koppang, M. D.; Woolsley N.
F.; Bartak, D. E. J . Am. Chem. Soc. 1984, 106, 2799. (d) Patel, K. M.;
Baltisberger, R. J .; Stenberg, V. I.; Woolsley, N. F. J . Org. Chem. 1982,
47, 4250. (f) Bhatt, M. V.; Kulkarni, S. A. Synthesis 1983, 249.
(11) Maercker, A. Angew. Chem., Int. Ed. Engl. 1987, 26, 972 and
references therein.
‡ Universita` di Sassari.
(1) Loupy, A.; Tchoubar, B. Salt Effects in Organic and Organome-
tallic Chemistry; VCH: New York, 1992.
(2) (a) Pregel, M. J .; Buncel, E. J . Org. Chem. 1991, 56, 5583. (b)
Dunn, E. J .; Moir, R. Y.; Buncel, E.; Purdon, J . G.; Bannard, R. A. B.
Can. J . Chem. 1990, 68, 1837. (c) Breslow, R.; Guo, T. J . Am. Chem.
Soc. 1988, 110, 5613. (d) Casaschi, A.; Desimoni, G.; Faita, G.;
Invernizzi, A. G.; Lanati, S.; Righetti, P. P. J . Am. Chem. Soc. 1993,
115, 8002.
(3) Lazana, M. C. R. L. R.; Franco, M. L. T. M. B.; Herold, B. J . J .
Am. Chem. Soc. 1989, 111, 8640.
(4) (a) J iao, H.; Schleyer, P. v. R Angew. Chem., Int. Ed. Engl. 1993,
32, 1760. (b) J iao, H.; Schleyer, P. v. R. J . Chem. Soc., Faraday Trans.
1994, 90, 1559. (c) J iao, H.; Schleyer, P. v. R. J . Am. Chem. Soc. 1995,
117, 11529.
(12) Azzena, U.; Denurra, T.; Melloni, G.; Piroddi, A. M. J . Org.
Chem. 1990, 55, 5386.
(13) (a) Azzena, U.; Denurra, T.; Melloni, G.; Fenude, E.; Rassu, G.
J . Org. Chem. 1992, 57, 1444. (b) Azzena, U.; Denurra, T.; Melloni,
G.; Fenude, E. Gazz. Chim. Ital. 1996, 126, 141.
(5) (a) Dougherty, D. A. Science 1996, 271, 163. (b) Ma, J . C.;
Dougherty, D. A. Chem. Rev. 1997, 97, 1303.
(6) Wlodek, S. T.; Antosiewicz, J .; Briggs, J . M. J . Am. Chem. Soc.
1997, 119, 8159.
(14) Bowers K. W. In Radical Ions; Kaiser, E. T., Kevan, L., Eds.;
Wiley: New York, 1968, p 232.
10.1021/jo991059v CCC: $19.00 © 2000 American Chemical Society
Published on Web 12/29/1999

Reductive Cleavage of Alkyl Aryl Ethers
J . Org. Chem., Vol. 65, No. 2, 2000 323
Sch em e 1
mode, Scheme 1A). In Scheme 1 the two thermodynami-
cally allowed fragmentation processes for alkyl aryl ether
radical anions are described. The alkyl-O bond cleavage
corresponds to a “homolytic” cleavage (Scheme 1C) that
does not regioconservate the spin density and should
therefore show an intrinsic kinetic barrier. On the other
hand, the less favorable (from a thermodynamic point of
view) aryl-O bond cleavage corresponds to a “heterolytic”
cleavage (Scheme 1D), and therefore no extra kinetic
barrier should exist in this case. This situation is
reflected in the behavior of alkyl nitrophenyl ether radical
anions, which are inert toward any fragmentation.20
Recently, some of us predicted that the alteration of
the “normal” electron distribution of a charged interme-
diate through electrostatic interactions (“topologically
controlled Coulombic interactions”, TCCI)20a,b could lead
to previously unknown processes. Following this idea, the
previously unknown reductive cleavage of alkyl nitro-
phenyl ethers20 and the change of regioselectivity in the
reductive fragmentation of other ethers have been
achieved.21 On the basis of theoretical calculations,20b we
have proposed that a positive noncoordinating charge
(aminium radical cation or ammonium cation) placed in
the neighborhood of the alkyl-oxygen bond reduces the
intrinsic barrier for the intramolecular electron transfer
π*-σ* by stabilizing the σ* state through electrostatic
interactions. It seems that electrostatic effects allows the
restriction imposed by the “spin regioconservation effect”
to be overcome so that “homolytic” mesolytic cleavages
become competitive under conditions where only the
“heterolytic” cleavages are normally observed.
ates,15 although more recent literature10-13 showed that,
in most cases, this is an unnecessary hypothesis.
Remarkably, till recently, the role assigned to the
metal in metal-promoted processes was merely that of
acting as the electron source; no attention was given to
its role in the rate-determining cleavage step. In par-
ticular, counterion-free radical anions have been usually
invoked as key intermediates in the alkali metal-
promoted reductive cleavage of ethers,8,11 alkyl halides,16
and sulfides.17 Some years ago, Herold and co-workers3
suggested that the position of the counterion (together
with the relative orientation of the alkyl ether bond with
respect to the benzene ring plane) was a main factor
governing the regioselectivity of the radical anion frag-
mentation in alkyl aryl ethers. In particular, they
proposed that dealkylation occurred when the counter-
cation was close to the oxygen atom, but that dealkoxy-
lation was the main process when the countercation
moved to a position over the aromatic ring. On the other
hand, Save´ant and co-workers have recently demon-
strated the effect of solvation and ion pairing on the rates
of cleavage of radical anions of aryl halides.18
The “bond activation” achieved by electron transfer has
thermodynamic origin; however, in aryl derivatives such
as alkyl aryl ethers, the bond fragmentation requires that
the excess of the electron density is transferred from the
π* system of the aryl ring to the area between the two
atoms of the scissile bond. Such an electron density
redistribution is responsible for the barrier observed in
these processes. Considering the electron apportionment
to the fragments, two modes of mesolytic scission are
possible: the homolytic mode, where the charge density
is largely localized on the same set of atoms before and
after the scission (Scheme 1A), and the heterolytic mode,
where the spin density is similarly “regioconserved”
(Scheme 1B). Both cleavages show distinct properties,
and in this context Guthrie and Maslak proposed the
“spin regioconservation principle” from studies on benzyl
aryl ethers.19 These authors showed the existence of an
intrinsic kinetic barrier for cleavage processes that do
not regioconservate the spin density (homolytic cleavage
On the other hand, it has been also recently demon-
strated that in some particular “heterolytic” mesolytic
cleavage of ethers, lithium plays a key role in promoting
the cleavage by acting as an internal Lewis acid in
assisting the departure of the leaving group.22
We present here studies that indicate that the proper-
ties of the countercation (as a Lewis acid or just electro-
static) and the type and topology of ion pairs have a
decisive influence on the evolution of radical anions, and
we wish to introduce electrostatic catalysis as a distinct
concept from the more commonly considered electrophilic
catalysis when dealing with the effects of cations on
chemical reactions. Indeed, our results indicate that
“homolytic” mesolytic cleavages are mainly dependent on
electrostatic catalysis, whereas “heterolytic” mesolytic
cleavages show a strong dependence on electrophilic
catalysis (Lewis acids). This study has been carried out
with a single type of substrate, alkyl 2,6-diphenylphenyl
ethers (Scheme 2). These substrates have a convenient
reduction potential that allows both chemical (alkali
metals) and electrochemical (cyclic voltametry and pre-
parative electrolysis) studies to be carried out and
compared. In addition, the presence of the phenyl rings
in the 2 and 6 positions places the O-Me bond out of
the plane of the central phenyl ring, thus eliminating the
possible ambiguity derived from the existence of several
conformers, which could be an additional factor contrib-
(15) (a) Testaferri, L.; Tiecco, M.; Tingoli, M.; Chianelli, D.; Mon-
tanucci, M. Tetrahedron 1982, 38, 3687. (b) Itoh, M.; Yoshida, S.; Ando,
T.; Miyaura, N. Chem. Lett. 1976, 271. (c) Eisch, J . J . J . Org. Chem.
1963, 28, 707. (d) Birch, A. J . J . Chem. Soc. 1947, 102. (e) Birch, A. J .
J . Chem. Soc. 1944, 430.
(16) Garst, J . F. Acc. Chem. Res. 1971, 4, 400.
(17) (a) Screttas, C. G.; Screttas, M. M. J . Org. Chem. 1978, 43, 1064.
(b) Cohen, T.; Bupathy, M. Acc. Chem. Res. 1989, 22, 152. (c) Severin,
M. G.; Farnia, G.; Vianello, E.; Arevalo, M. C. J . Electroanal. Chem.
1988, 251, 369. (d) Severin, M. G.; Arevalo, M. C.; Farnia, G.; Vianello,
E. J . Phys. Chem. 1987, 91, 466. (e) Arevalo, M. C.; Farnia, G.; Severin,
M. G.; Vianello, E. J . Electroanal. Chem. 1987, 220, 201. (f) Griggio,
L.; Severin, M. G. J . Electroanal. Chem. 1987, 223, 185.
(18) Andrieux, C. P.; Robert, M.; Save´ant, J .-M. J . Am. Chem. Soc.
1995, 117, 9340.
(20) (a) Cayo´n, E.; Marquet, J .; Lluch, J . M.; Martin, X. J . Am. Chem.
Soc. 1991, 113, 8970. (b) Marquet, J .; Cayo´n, E.; Mart´ın, X.; Casado,
F.; Gallardo, I.; Moreno, M.; Lluch, J . M. J . Org. Chem. 1995, 60, 3814.
(c) Gonzalez-Blanco, R.; Bourdelande, J . L.; Marquet, J . J . Org. Chem.
1997, 62, 6910.
(21) Azzena, U.; Casado, F.; Fois, P.; Gallardo, I.; Pisano, L.;
Marquet, J .; Melloni, G. J . Chem. Soc., Perkin Trans. 2 1996, 2563.
(22) Saa´, J . M.; Ballester, P.; Deya´, P. M.; Capo´, M.; Garc´ıas, X. J .
Org. Chem. 1996, 61, 1035.
(19) (a) Maslak, P.; Guthrie, R. D. J . Am. Chem. Soc. 1986, 108,
2628. (b) Maslak, P.; Guthrie, R. D. J . Am. Chem. Soc. 1986, 108, 2637.

324 J . Org. Chem., Vol. 65, No. 2, 2000
Casado et al.
Sch em e 2
cleavage at the dianion level would lead to an aryl
σ-anion and no effect of the isotopically labeled hydrogen
atom source would be expected in the outcome of the
reaction. When experiment 1 of Table 1 was repeated in
THF-d8, 23% yield of deuterium-labeled m-terphenyl (2)
was obtained (1H NMR analysis). This result is at
variance with what it is reported for the reduction of
anisole with potassium in isooctane,13a where the hydro-
gen atoms come from the departing methoxyde group and
no hydrogen atoms from the solvent are incorporated. To
test this second possibility, the same reaction was
repeated using 2,6-diphenylphenyl methyl-d3 ether. In
this case, a 10% yield of deuterium-labeled m-terphenyl
was obtained (1H NMR analysis). These results strongly
support the hypothesis that the cleavage happens at the
radical anion level.
uting to the kinetic barriers in the intramolecular
electron-transfer step.
Resu lts
In experiment 6, the corresponding reaction of N-meth-
yl-N-[2-(2,6-diphenylphenoxy)ethyl]piperidinium tetraflu-
oroborate (1d ) is described, where a very significant
amount (41%) of dealkylation product is obtained. The
obvious interpretation is that the short TCCI chain is
not able to reproduce the geometry of the tight ionic pair
but interferes with the lithium cation to afford the
reaction of a solvent-separated-like ionic pair responsible
for the dealkylation process.
Experiments 7-12 (Table 1), carried out under homo-
geneous conditions (catalytic amounts of naphthalene),
show exactly the same trend of the reactions carried out
under heterogeneous conditions, although with a lower
regioselectivity, being possible in this case by using a
short TCCI chain to achieve complete dealkylation
(experiment 10). The smooth change in regioselectivity
observed in these experiments (experiments 10-12)
suggests that the same mechanism is operating in all the
cases, since an alternative Grob type cleavage mecha-
nism, which would lead to ethene and N-methylpiperi-
dine from the radical anion of substrate 1d , can be ruled
out since this would only operate in a substrate with a
two-methylene unit linkage. In addition, evidence for the
formation of products of disproportionation of an alkyl
radical was obtained by careful examination of the
fraction soluble in water of the reaction mixture of
experiment 10. 1H NMR analysis of this fraction showed
the presence of the N-ethyl-N-methylpiperidinium (5) and
N-ethenyl-N-methylpiperidinium (6) cations in the ratio
ca. 1:1 (Scheme 3).
2. Stu d ies on th e Electr och em ica lly P r om oted
Red u ctive Clea va ge of Alk yl 2,6-Dip h en ylp h en yl
Eth er s. Cyclic voltammetry curves of substrates 1a and
1d (short TCCI alkyl chain) in DMF at two different
sweep rates are shown in Figure 1. Substrate 1a shows
a reversible (one electron) reduction wave (E° ) -2.54 V
vs SCE) in both cases (Figure 1A,B); on the other hand,
substrate 1d shows an irreversible (two electrons) reduc-
tion wave at a slow sweep rate (0.1 V s-1, Figure 1C) that
changes to a slightly reversible wave (one electron) at a
faster sweep rate (Figure 1D). In Figure 2 the variation
of ip/cv1/2 (the current function) with respect to log v is
shown for substrate 1d , demonstrating the change from
a reversible one-electron process at fast sweep rate to a
two-electron irreversible process when slower sweep rates
are used.
1. Stu d ies on th e Lith iu m -P r om oted Red u ctive
Clea va ge of Alk yl 2,6-Dip h en ylp h en yl Eth er s (1). In
Table 1 the reductive cleavage of 2,6-diphenylanisole (1a )
with lithium in ethereal solvents and under different
conditions is described. Formation of major amounts of
m-terphenyl (2) (dealkoxylation) was observed in all the
cases. The reaction was completely regioselective in THF
(experiment 1), but a significant amount of 2,6-diphe-
nylphenol (3) (dealkylation) was obtained when a more
solvating solvent, such as DME (which favors the forma-
tion of solvent separated ionic pairs), was used (experi-
ment 2). This trend was confirmed by carrying out the
reaction in the presence of 12-crown-4 (experiment 3),
which led to 43% of dealkylation still with complete
conversion of the starting material. Related results have
been reported for the reductive cleavage of several alkyl
aryl ethers with alkali metals in polar aprotic solvents
such as HMPA,15a where only dealkylation products were
obtained.
A similar situation was found when the ionic strength
of the THF solution was increased by using lithium
tetrafluoroborate in what seems to be a nonspecific salt
effect (experiment 4). However, experiment 5 shows a
very interesting specific salt effect by the tetraalkylam-
monium cation. Thus, the reaction carried out (experi-
ment 5) in the presence of tetrabutylammonium tetraflu-
oroborate (same concentration as that of lithium tetra-
fluoroborate used in experiment 4) was significantly
slower and, remarkably, afforded only the dealkoxylation
product. It is well-known23,24 that tetraalkylammonium
cations in THF displace lithium cations from contact ionic
pairs with large, delocalized anions, which is the case of
the radical anion of 2,6-diphenylanisole. The result
described in experiment 5, when compared with those of
experiments 1-4, demonstrates on one side the depen-
dence of the dealkoxylation reaction on the presence of
the alkali cation and on the other side, the reluctance of
the radical anion of 2,6-diphenylanisole to undergo
dealkylation at the contact ionic pair stage.
To test the hypothesis that the aryl-O cleavage under
our conditions happens at the radical anion level, we
carried out our reactions in the presence of isotopically
labeled hydrogen atom sources. The cleavage at the
radical anion level leads to an aryl σ-radical that we
would expect to be at least partially trapped by the
isotopically labeled hydrogen atom source, despite its
easy reduction by the alkali metal. On the other hand,
The reduction peak potential for the N-methyl-N-[2-
(2,6-diphenylphenoxy)ethyl]piperidinium tetrafluorobo-
rate (1d ) was measured as a function of sweep rate and
concentration in DMF to determine the molecular order
(23) Smid, J . Angew. Chem., Int. Ed. Engl. 1972, 11, 112.
(24) J ensen, S.; Parker, V. D. J . Am. Chem. Soc. 1975, 97, 5211.

Reductive Cleavage of Alkyl Aryl Ethers
J . Org. Chem., Vol. 65, No. 2, 2000 325
Ta ble 1. Lith iu m -P r om oted Red u ctive Clea va ge of Alk yl 2,6-Dip h en ylp h en yl Eth er s (Sch em e 2)
products
expt
subst
R
conditionsa
conversion (%)
2 (%)
3 (%)
1
2
3
1a
1a
1a
CH3
CH3
CH3
THF, 24 h
DME, 24 h
THF, 24 h
100
100
100
100
81
57
19
43
12-crown-4
4
5
1a
1a
CH3
CH3
THF, 24 h
LiBF4 (25 equiv)
THF, 24 h
100
43
85
43
15
TBATFB (25 equiv)
THF, 24 h
THF, 24 h
6
1d
1a
(CH2)2N+(CH3)C5H10
CH3
100
100
56
70
41
30
7b
naphthalene (10%)
THF, 24 h
naphthalene (10%)
THF, 24 h
naphthalene (10%)
THF, 24 h
naphthalene (10%)
THF, 24 h
8
9
1b
1c
1d
1e
1f
(CH2)5CH3
100
100
100
100
100
95
72
5
28
(CH2)2NC5H10
10b
11b
12
(CH2)2N+(CH3)C5H10
(CH2)3N+(CH3)C5H10
(CH2)4N+(CH3)C5H10
100
80
20
95
naphthalene (10%)
THF, 24 h
5
naphthalene (10%)
a
b
Li, 2.5 equiv, rt. Previously reported in ref 20.
F igu r e 2. Variation of the current function with the sweep
rate at different concentrations for N-[2-(2,6-diphenylphenoxy)-
ethyl]-N-methylpiperidinium tetrafluoroborate (1d ).
centration range of 1.01-7.08 mM. These data are
consistent with a first-order chemical reaction following
electron transfer in a sequential ECE type mechanism.25
In Figure 3, the corresponding cyclic voltammetries in
THF are reported. In this solvent, substrate 1a shows a
less reversible behavior than in DMF (Figure 3A,B).
However, in both cases a rather slow one-electron trans-
fer was observed. In any case, it seems that the chemical
process associated with the electron transfer corresponds
to reduction of the solvent or of the supporting electrolyte
rather than to any productive reaction (cleavage or
reduction) from the substrate (see below, Table 3). With
substrate 1d (short TCCI chain), a significant dependence
of the current function with the sweep rate was observed.
Indeed, in a particular experiment in THF ([1d ] ) 3.01
mM), the value of ip/cv1/2 varied from 98.2 for v ) 0.1 V
s-1 (Figure 3C) to 58.9 for v ) 10 V s-1 (Figure 3D),
indicating the operation of a two-electron process when
the sweep rate is slow enough that changes to a one-
electron process at faster sweep rates. We assign the two-
F igu r e 1. Cyclic voltammograms in DMF (0.1 M TBATFB,
13 °C): (A) 2,6-diphenylanisole (1a , 6.40 mM) at 0.1 V s-1 and
(B) at 10 V s-1; (C) N-[2-(2,6-diphenylphenoxy)ethyl]-N-meth-
ylpiperidinium tetrafluoroborate (1d , 2.99 mM) at 0.1 V s-1
and (D) at 10 V s-1
.
Sch em e 3
for the disappearance of the radical anion. At relatively
slow scan rates (i.e., 0.1-2.0 V s-1) the peak potential
(Epc) shifted in the positive direction with decreasing scan
rate (v) so that dEpc/d(log v) ) -29 mV. The peak
potential was independent of concentration over a con-
(25) Bard, A. J .; Faulkner, L. R. Electrochemical Methods; Wiley:
New York, 1980; Chapter 11.

326 J . Org. Chem., Vol. 65, No. 2, 2000
Casado et al.
no significant amount of dealkoxylation product was
produced in any case.
No significant effect was observed when the electrolysis
of 1a was carried out (experiment 3) in THF in the
presence of 2.8 equiv of lithium tetrafluoroborate, which
is poorly soluble in THF. When the electrolysis in DMF
was performed using lithium tetrafluoroborate as sup-
porting electrolyte, the result was even worse than when
a tetralkylammonium salt was used (experiment 4). In
the first case, the result confirms that the lithium cation
is not able to displace the tetrabutylammonium cation
from the contact ionic pair with the aromatic radical
anion in THF, and in the second the result probably
means that, at the very low potential used, the lithium
cation is partially reduced, consuming electricity in a
nonproductive way. We have been unable to achieve
electrochemically the conditions needed for the dealkoxy-
lation process to be observed in a significant proportion.
This confirms the crucial role played by the alkali cation,
probably in a contact ionic pair, in these processes.
Experiments 5 and 6 in Table 3 correspond to reactions
using N-methyl-N-[2-(2,6-diphenylphenoxy)ethyl]piperi-
dinium tetrafluoroborate (1d ) as starting material in
THF and DMF, respectively. The reactions are now much
more efficient, giving rise to 2,6-diphenylphenol (3) in
very good yields (88% and 91%, respectively). No dealkox-
ylation product (m-terphenyl) could be detected. These
results support the interpretation given for the corre-
sponding reactions carried out under chemical conditions
(experiments 6, 10, and 11 in Table 1). Thus, the short
TCCI chain is not able to reproduce the geometry of the
tight ion pair, but the internal ammonium cation com-
petes efficiently with the external cations, precluding the
formation of the contact ion pair with the external
tetraakylammonium cation in the reaction carried out
in THF. The high efficiency in the dealkylation process,
observed in both reactions (experiments 5 and 6), can be
explained by considering the intervention of “intramo-
lecular solvent-separated-like ion pairs” of a very par-
ticular topology. Thus, in these intramolecular pairs, the
positive charge must be located, as an average, much
closer to the ether linkage than it is in an intermolecular
solvent-separated ionic pair (experiment 2 and 4), and
therefore the associated electrostatic effect on the ether
bonds will be much more intense (see the Discussion).
The necessity of two electrons (mechanism ECE) to drive
the reaction to completion was strongly supported by the
result of experiment 7 in Table 3. Thus, when only 1.1
F/mol was consumed, the yield of the reaction dropped
to half.
F igu r e 3. Cyclic voltammograms in THF (0.6 M TBATFB,
13 °C): (A) 2,6-diphenylanisole (1a , 2.10 mM) at 0.1 V s-1 and
(B) at 10 V s-1; (C) N-[2-(2,6-diphenylphenoxy)-ethyl]-N-
methylpiperidinium tetrafluoroborate (1d , 5.03 mM) at 0.1 V
s-1 and (D) at 10 V s-1
.
Ta ble 2. Lim itin g Lifetim es for th e Ra d ica l An ion s of
Su bstr a tes 1a a n d 1d a t 13 °C, Mea su r ed by Cyclic
Volta m etr y
entry
substrate
solvent
lifetime (τ) (ms)
1
2
3
4
1a
1a
1d
1d
DMF
THF
DMF
THF
>200
50
1
0.2
electron process to the cleavage of the radical anion of
the substrate 1d in an ECE type mechanism also in THF.
From the study of the variation of the cyclic voltam-
mograms with the sweep rate, the radical anions limiting
lifetimes indicated in Table 2 have been calculated. The
introduction of the TCCI alkyl chain (substrate 1d )
significantly reduces the lifetime of the radical anion
intermediate in all the cases.
In Table 3, the results of the preparative electrolysis
at controlled potential (-2.6 V vs SCE) of substrates 1a
and 1d in THF and DMF are reported (for comparison
purposes, a charge consumption of 2.1 F was used in all
except one of the cases of Table 3). The electrolysis of
2,6-diphenylanisole (1a ) in THF (experiment 1) gives rise
only to a trace of m-terphenyl (dealkoxylation). A trace
amount of Birch reduction products could also be de-
tected, but interestingly, no 2,6-diphenylphenol (dealky-
lation) was produced. This result (low Faradaic yield)
confirms the reluctance of the radical anion of 2,6-
diphenylanisole to undergo dealkylation at the contact
ionic pair stage when a noncoordinating tetralkylammo-
nium cation is the counterion. On the other hand, the
corresponding electrolysis of 1a in DMF afforded a 29%
yield of 2,6-diphenylphenol (dealkylation), using the same
applied potential and after the same current consump-
tion. Traces of Birch reduction products could be detected
also in this case by GC/MS analysis. Thus, the use of a
solvent with better cation-solvating properties that would
favor the presence of solvent-separated ionic pairs allows
the intervention of the mechanistic pathway leading to
dealkylation, but still with a low Faradaic yield. It is
interesting to notice that in the absence of alkali metal
Investigation of the fate of the alkyl chain after
cleavage in these electrochemical experiments (experi-
ments 5 and 6) was handicapped by the presence of a
large excess of the tetraalkylammonium salt used as
supporting electrolyte. Generally, the fraction of these
reaction crudes soluble in water showed the presence of
a complex mixture that was not studied further. In
addition, 1H NMR analysis of the mixture in experiment
6 showed the presence of a triplet at δ 1.0 that was
attributed (comparison with an authentical sample) to
the presence of N-ethyl-N-methylpiperidinium tetrafluo-
roborate (5) in the mixture. This is additional evidence
for the occurrence of an ECE mechanism in these
electrochemical reactions.
The results described in experiments 8 and 9 of Table
3, corresponding to the reactions of substrates 1e (three

Reductive Cleavage of Alkyl Aryl Ethers
J . Org. Chem., Vol. 65, No. 2, 2000 327
Ta ble 3. Electr och em ica lly P r om oted Red u ctive Clea va ge of Alk yl 2,6-Dip h en ylp h en yl Eth er s (Sch em e 2)
products
expt
subst
conditionsa
2 (%)
3 (%)
1
2
3
1a
1a
1a
CH3
CH3
CH3
THF, TBATFB, 2.1 F
DMF, TBATFB, 2.1 F
THF, TBATFB
trace
29
trace
LiTFB (x equiv), 2.1 F
DMF, LiTFB, 2.1 F
THF, TBATFB, 2.1 F
DMF, TBATFB, 2.1 F
DMF, TBATFB, 1.1 F
DMF, TBATFB, 2.1 F
DMF, TBATFB, 2.1 F
4
5
6
7
8
9
1a
1d
1d
1d
1e
1f
CH3
<5
91
88
43
46b
59
(CH2)2N+(CH3)C5H10
(CH2)2N+(CH3)C5H10
(CH2)2N+(CH3)C5H10
(CH2)3N+(CH3)C5H10
(CH2)4N+(CH3)C5H10
a
b
Controlled potential: -2.6 V relative to SCE at 13 °C. Allyl 2,6-diphenylphenyl ether was also detected in this case.
Ta ble 4. Th er m od yn a m ics of Ca r bon -Oxygen Bon d s
F r a gm en ta tion Estim a ted by Th er m och em ica l Cyclesa
value
parameter (kcal mol-1
)
method of calculation
(∆G°)1a
(∆G°)1b
(∆G°)2
(∆G°)3a
(∆G°)3b
(∆G°)4a
(∆G°)4b
59
93
53
-7
-7
-1
33
estimated as described in the text
estimated as described in the text
(∆G°)2 ) -F(E°)2, where (E°)2 ) -2.30 V
(∆G°)3a ) -F(E°)a, where (E°)a ) 0.32 V4
(∆G°)3b ) -F(E°)b, where (E°)a ) 0.30 V14
(∆G°)4a ) (∆G°)1a + (∆G°)3a - (∆G°)2
(∆G°)4b ) (∆G°)1b + (∆G°)3b - (∆G°)2
a
Refer to Figure 4 for definition of thermodynamic quantities.
F is the Faraday constant (23.06 kcal mol-1 V-1). In this particular
table the potentials are reported relative to NHE.
F igu r e 4. Thermodynamic cycles for calculation of free energy
for C-O bond fragmentation in alkyl aryl ethers radical
anions.
possible: (a) alkyl-O bond (homolytic fission) and (b)
aryl-O bond (heterolytic fission). Three free energies are
required for estimation of (∆G°)4 (Table 4): the free
energy for C-O bond homolysis in the neutral substrate,
(∆G°)1; the free energy for electrochemical reduction of
the neutral substrate, (∆G°)2; and the free energy for
electrochemical reduction of one of the products of bond
fission, (∆G°)3. Table 4 provides a tabular summary of
how each thermodynamic parameter was obtained or
estimated. First, an estimate for the free energy of
homolysis (∆G°)1 was arrived at by taking the literature
bond dissociation energies for anisole (Me-O,27 64 kcal/
mol; Ph-O,28 98 kcal/mol) and using an entropy value
which is typical29 for C-O bond homolysis (∆S = 17 cal
mol-1K-1; since both processes should show a similar
entropy, the exact value is not very important for
comparisons). These values lead to an estimate of (∆G°)1a
= 59 kcal/mol for the alkyl-O fission and (∆G°)1b = 93
kcal/mol for the aryl-O fission. (∆G°)2 was calculated
from the experimentally measured E° value (-2.54 V for
the reduction of 2,6-diphenylanisole, relative to SCE, and
-2.30 V, relative to NHE), and (∆G°)3a and (∆G°)3b were
calculated from the reported standard potentials for the
oxidation of 2,6-diphenylphenolate (E° ) 0.08 V, relative
to SCE; 0.32 relative to NHE)30 and methoxide anion (E°
) 0.06 V, relative to SCE; 0.30 relative to NHE),31
respectively, both in acetonitrile. The values obtained for
(∆G°)4 indicate that demethylation of the radical anion
of 2,6-diphenylanisole is a slightly downhill process
[(∆G°)4a =-1 kcal/mol], whereas the alternative demethox-
ylation process (in the absence of any assistance by the
methylene units TCCI chain) and 1f (four methylene
units TCCI chain), respectively, indicate that the effect
of the positive charge is smoothly reduced when its
average distance to the ether bond increases. In the
reaction carried out with substrate 1e, another product,
tentatively assigned to allyl 2,6-diphenylphenyl ether,
was detected (it was not isolated from the mixture), in
addition to 46% of phenol 3. This constituted a significant
difference from the other reactions reported in the Table
3. Considering the very low potential used in our reac-
tions and the possibility that the allyl ether was produced
by an elimination type reaction elicited by an electro-
generated base, we tested the stability of substrate 1e
in the presence of a very strong base such as potassium
tert-butoxide in THF. Under these conditions 1e gave rise
to 13% of 2,6-diphenylphenol (3). In any case, the
electrochemical reaction (experiment 8, Table 3) shows
a significantly higher proportion of phenol, indicating
that even though the elimination is in this particular case
a competitive reaction, unimolecular cleavage must still
be considered the main process from the radical anion.
Discu ssion
Before going into kinetic arguments, consideration of
the thermodynamics for C-O bond fragmentation in the
radical anions of our substrates provides a starting point
to understand the regioselectivity preferences observed
in our reactions. Figure 4 outlines the thermodynamic
cycles that can be applied to calculate the standard free
energy for C-O bond fragmentation, (∆G°)4, from the
radical anion, and Table 4 provides a tabular summary
of the necessary equations and the calculated thermo-
dynamic quantities.26 Two modes of fragmentation are
(27) Suryan, M. M.; Kafafi, S. A.; Stein, S. E. J . Am. Chem. Soc.
1989, 111, 1423.
(28) Handbook of Chemistry and Physics, 66th ed.; CRC Press: Boca
Raton, 1985-1986; F-193.
(29) Andersen, M. L.; Long, W.; Wayner, D. D. M. J . Am. Chem.
Soc. 1997, 119, 6590.
(30) Hapiot, P.; Pinson, J .; Yousfi, N. New J . Chem. 1992, 16, 877.
(31) Eberson L. Acta Chem. Scand. 1984, B38, 439.
(26) Wayner, D. D. M.; Parker, V. D. Acc. Chem. Res. 1993, 26, 287.

328 J . Org. Chem., Vol. 65, No. 2, 2000
Casado et al.
As indicated, our preparative results show exclusive
demethoxylation with lithium as reducing agent in THF
(Table 1) or practically no reaction (trace of demethoxy-
lation) under electrochemical conditions in the same
solvent (experiment 1, Table 3). Cyclic voltametry experi-
ments (Figure 1, Table 2) confirm the stability of the
radical anion of 2,6-diphenylanisole toward fragmenta-
tion in THF, probably with a slow evolution through
reduction of the solvent or of the supporting electrolyte.
Interestingly enough, when the reaction with lithium was
carried out by trying to mimic the conditions of the
electrochemical experiments (the presence of tetrabutyl-
ammonium tetrafluoroborate, experiment 5, Table 1), the
reaction slowed, but demethoxylation was still the ex-
clusive fragmentation process. This result points out that
the alkali cation are responsible for the success of the
demethoxylation reaction and confirms the inertia of the
radical anion to undergo demethylation, although this
is the thermodynamically preferred process.
F igu r e 5. Thermodynamic cycles for calculation of free energy
for C-O bond fragmentation in alkyl aryl ethers dianions.
metal) is strongly uphill [(∆G°)4b = +32 kcal/mol), thus
showing that the system, when no external influences
are present, has a significant thermodynamic driving
force in favor of demethylation.
From the point of view of the initial spin delocalization,
the vast majority of aniomesolytic reactions result in
departure of a negatively charged fragment. It has been
postulated19 that the fragmentation process is controlled
by factors other than product stability, and a mode of
cleavage requiring a charge transfer across the scissile
bond is kinetically preferred (the “spin regioconservation
principle”).
Even though dianions15 have not been considered lately
as possible intermediates in these types of reactions, and
our electrochemical results indicate that dealkylation
must be produced from the radical anion, the hypothetical
intervention of a dianion as intermediate in the dealkox-
ylation processes that we observe in the presence of alkali
metals cannot be completely ruled out in the light of our
preparative results (Table 1). Figure 5 outlines the
thermodynamic cycles that can be applied to calculate
the standard free energy for C-O bond fragmentation
(∆G°)7 from the dianion. The second reduction wave of
2,6-diphenylphenyl ethers is very close to the wall and
no reliable measurements can be carried out on it. In any
case, considering (∆G°)5 is common to the two cycles (A
and B) and that (∆G°)6a = (∆G°)6b, an important differ-
ence (>30 kcal mol-1) in the thermodynamic driving force
favoring demethylation would also exist in this case
[(∆G°)7a . (∆G°)7b]. In the absence of external influences,
the higher free energy of the dianion (reactivity), when
compared to the radical anion, could justify a lower
selectivity in the reaction but never the complete change
of regioselectity observed when an alkali metal is the
reducing agent. Therefore, it seems that there is no need
to postulate a dianion as an intermediate and that the
alkali metal must play a specific role on the radical anion
other than simply acting as a reducing agent. To be more
comfortable with this interpretation, the radical anion
hypothesis was tested by carrying out experiments in the
presence of isotopically labeled hydrogen atom sources.
Deuterium incorporation into m-terphenyl in the reac-
tions in the presence of alkali metals, carried out in THF-
d8 or using 2,6-diphenylphenyl methyl-d3 ether as a
substrate, strongly supports the operativity of the 2,6-
diphenylphenyl radical as intermediate and therefore the
occurrence of the cleavage at the radical anion level.
Dissociating π-radical anions can be divided into two
different structural groups,33-35 depending on whether
the possibility of overlap between the π-network and the
σ* of the scissile bond exists (like in benzylic halides) or
it is precluded (like in aryl halides). In the radical anions
of aryl alkyl ethers, such as 2,6-diphenylanisole, the two
different situations coexist, as they correspond to the two
different C-O bonds. Thus, the cleavage of the aryl-O
bond, orthogonal to the π-system bearing the unpaired
electron, must involve, to occur, a discrete intramolecular
π*-σ* electron-transfer step (if the σ* state is a local
minimum)36 or at least a large intramolecular π*-σ*
electron-transfer component (if the σ* state is a frag-
mentative one). As to the alkyl-O bond, the possible
direct overlap between the orbitals of interest should
increase as the reaction progresses toward a resonance-
stabilized anion or radical. The transition state for the
cleavage of this type of bond may be described in terms
of two different electronic configurations.35,37 One of these
configurations places significant negative charge on the
departing fragment and has been compared with an
elimination reaction. In our case, however, where the
polarization of the bond is opposite to that of benzylic
halides, this configuration can be neglected. The second
configuration may be represented as a σ* radical anion.
The fragmentation is kinetically controlled by the ability
of the system to delocalize the charge across the scissile
bond. For radical anions where this tendency is reinforced
by preexisting polarization of the scissile bond and by
thermodynamic stability of the product to be formed, the
reaction is facile. However, if such a charge shift leads
to extensive “counterthermodynamic” charge accumula-
One extra possibility to consider is that the dealkoxy-
lation took place in the dimeric form of radical anion32
and that the counterion play their role in shifting the
monomer-dimer equilibrium. The absence of concentra-
tion effects in the reactions described in Table 1 allows
this possibility to be discarded. Therefore, an aniome-
solytic cleavage of the radical anion is the best mecha-
nistic hypothesis also for the dealkoxylation process.
(33) Dressler, R.; Allan, M.; Haselbach, E. Chimia 1985, 39, 385.
(34) Symons, M. C. R.; Bowman, W. R. J . Chem. Soc., Perkin Trans.
2 1988, 584.
(35) Maslak, P.; Narvaez, J . N.; Kula, J .; Malinski, D. S. J . Org.
Chem. 1990, 55, 4550.
(32) Koppang, M. D.; Woolsley, N. F.; Bartak, D. E. J . Am. Chem.
Soc. 1985, 107, 4692.
(36) Pierini, A. B.; Duca J r., J . S. J . Chem. Soc., Perkin 2 1995, 1821.
(37) Guthrie, R. D.; Shi, B. J . Am. Chem. Soc. 1990, 112, 3136.

Reductive Cleavage of Alkyl Aryl Ethers
J . Org. Chem., Vol. 65, No. 2, 2000 329
vide infra)20b and afterward acting as a handle (Lewis
acid) to which the leaving group RO- adheres prior to
detachment. Hence, we interpret the result of experiment
1 in Table 1 as a joint electrostatic plus electrophilic
catalysis by the countercation in a contact ionic pair.
When solvents with stronger solvating properties are
used, the demethylation process starts to take place.
Indeed, the electrochemical reduction of 2,6-diphenylani-
sole in DMF leads in a poorly efficient process to 2,6-
diphenylphenol (experiment 2, Table 3). In the same
direction, when the chemical reductions are carried out
in DME, in the presence of crown ethers, or in the
presence of naphthalene (experiments 2, 3, 7, 8, and 9,
Table 1), significant amounts of demethylation product
are obtained. Maercker10 had already noticed that solvent-
separated ionic pairs favored the dealkylation process,
and Testaferri et al.14a had reported major dealkylation
in several reductive cleavages of alkyl aryl ethers carried
out with sodium in a strongly cation-solvating solvent
such as HMPA. From the results reported in Tables 1
and 3, it is evident that electrostatic interactions in the
solvent-separated ionic pairs are responsible for the
operation of the dealkylation pathway when it is opera-
tive. Thus, on the basis of theoretical calculations,20b we
have proposed that a positive noncoordinating charge in
the neighborhood of the ether bond reduces the barrier
for the intramolecular electron transfer π*-σ* by stabi-
lizing the σ* state through attractive electrostatic inter-
actions. This is just the situation in the solvent-separated
ionic pairs, where coordination with the ether oxygen
bond is not possible. Both intramolecular π*-σ* electron
transfers should show electrostatic catalysis in the
solvent-separated ionic pairs C and D (SSIP-C, SSIP-D,
Figure 6), in equilibrium with the corresponding contact
ionic pairs A and B, dealkylation probably being the
favored reaction from the SSIP since electrostatic cataly-
sis reduces the kinetic control and dealkylation is the
thermodynamically preferred process.
F igu r e 6. Ionic pairs involved in the reductive cleavage of
alkyl 2,6-diphenylphenyl ethers.
tion on the leaving fragment, as in our case, an alterna-
tive mode of scission becomes dominant.35 This mode is
similar to the one described for aryl-O bonds and would
involve a large component of π*-σ* electron transfer
followed by dissociation of the three-electron bond.
Therefore, even though the two types of bonds involved
in our study are very different, their mode of fragmenta-
tion must be similar, involving in both cases a π*-σ*
electron-transfer step or component in the transition
state.
Experiments 5 and 6 in Table 3 confirm the importance
of the electrostatic interactions in the outcome of the
fragmentation reaction. Cyclic voltammetry experiments
show also a significant reduction in the lifetime of the
radical anion of these substrates. In these cases, the SSIP
must have a particular topology, with the positive charge
concentrated in the neighborhood of the alkyl ether bond
(SSIP-like-E, Figure 6). Therefore, the electrostatic in-
teraction is now much more directional, the effect on the
barrier of the π*-σ* (alkyl bond) electron transfer being
much larger than on the barrier of the π*-σ*(aryl bond)
electron transfer, thus justifying the very efficient dealky-
lation process observed in these cases. A related situation
is found in the chemical reduction (experiments 6 and
10, Table 1), the lower selectivity observed in experiment
6 probably being due to the fact that in this case a
competition between two productive ionic pairs (CIP-B
leading to dealkoxylation, and SSIP-E leading to dealky-
lation) can be envisaged. A parallel competition in the
electrochemical experiments would involve a productive
ionic pair, SSIP-E, and a nonproductive one, CIP-A, thus
justifying the very high selectivity observed in these
experiments (Table 3). The specific effect of the TCCI
smoothly disappears upon increasing its length (experi-
ments 10-12 in Table 1 and experiment 6, 8, and 9 in
Table 3).
Our results confirm that the mesolytic cleavage of the
radical anion of 2,6-diphenylanisole is a kinetically
controlled process, forbidden in the absence of external
influences (experiment 1, Table 3), probably due to the
unfavorable π*-σ* electron-transfer energetics for the
two alternatives (aryl-O or alkyl-O). The results re-
ported in Tables 1 and 3 can be explained by considering
different types of ionic pairs and the distinct electrophilic
or electrostatic effect of the counterion in each case.
In Figure 6, the different ionic pairs likely involved in
our reactions are described. The starting point would be
contact ionic pair A (CIP, Figure 6), which would cor-
respond to the case described in the experiment 1 of Table
3. It is well-known that tetraalkylammonium cations
form contact ion pairs with delocalized anions in solvents
of low polarity such as THF.23,24 The electrochemical
results suggest that this ionic pair is stable toward
fragmentation, probably due to the electrostatic stabiliza-
tion of the π* state by the noncoordinating counterion in
the tight ionic pair. However, when the counterion is a
coordinating alkali cation, such as lithium, regioselective
dealkoxylation is observed (experiment 1, Table 1). In this
case a contact ion pair such as B (Figure 6) would be
involved, where the lithium cation coordinated to the
oxygen atom of the ether bond would be probably acting
in two ways, first allowing the population of the σ* state
by electrostatic stabilization of this state (reducing the
barrier for the π*-σ* intramolecular electron transfer,
In conclusion, the first multiapproach (chemical, elec-
trochemical, TCCI20b) to the reductive fragmentation of

330 J . Org. Chem., Vol. 65, No. 2, 2000
Casado et al.
N, 3.92. Found: C, 83.72; H, 7.50; N, 3.85. The N-methylpi-
peridinium tetrafluoroborates 1e and 1f were prepared by the
method described in ref 20b, through the corresponding
amines, as white solids, and purified by recrystallization from
Me2CO/Et2O.
N-[2-(2,6-d ip h en ylp h en oxy)p r op yl]p ip er id in e was pu-
rified by bulb-to-bulb distillation at 1 Torr (bath temperature,
230 °C): 1H NMR (CDCl3) δ 1.21-1.43 (m, 8H), 1.83 (br t, J )
7.8 Hz, 2H), 2.03 (br s, 4H), 3.16 (t, J ) 6.3 Hz, 2H), 7.15 (dd,
J ) 7.2 Hz, J ) 6.6 Hz, 1H), 7.23-7.40 (m, 8H), 7.52, 7.59 (m,
4H); 13C NMR (CDCl3) δ 24.33, 25.79, 27.10, 54.18, 55.82,
71.69, 124.07, 126.95, 127.96, 129.45, 130.11, 136.02, 138.70,
153.87. Anal. Calcd for C26H29NO: C, 84.06; H, 7.87; N, 3.77.
Found: C, 83.90; H, 7.92; N, 3.65.
N-[-3-(2,6-Dip h en ylp h en oxy)p r op yl]-N-m et h ylp ip er i-
d in iu m tetr a flu or obor a te (1e): mp 179-182 °C; IR (KBr)
3048, 2939, 1106, 1068, 1044, 761, 705 cm-1; 1H NMR (DMSO-
d6) δ 1.36-1.80 (m, 8H), 2.77 (s, 3H), 2.77-2.84 (m, 2H), 2.95-
3.05 (m, 2H), 3.11-3.50 (m, 4H), 7.20-7.60 (m, 13H); 13C NMR
(DMSO-d6) δ 19.3, 20.6, 22.0, 46.9, 60.1, 60.3, 69.8, 124.9,
127.6, 128.5, 129.3, 130.4, 135.4, 138.2, 153.1. Anal. Calcd for
alkyl aryl ethers has provided insight into the mecha-
nistic details that govern the regioselectivity of these
reactions and has allowed the proposal of electrostatic
and electrophilic catalysis as distinct contributions that
affect the reactivity of the intermediate radical anions.
Both fragmentations (alkyl-O and aryl-O) must involve
an intramolecular π*-σ* electron-transfer step, or at
least a large component of π*-σ* electron transfer in the
transition state. The fragmentation of alkyl aryl ethers
can thus be analyzed in terms of the factors that alter
the π*-σ* electron-transfer step for both processes and
the evolution of the σ* state. Thus, contact ion pairs (CIP)
seem to evolve in a very different way, depending on the
coordinating properties of the cation. CIP of noncoordi-
nating cations (tetralkylammonium) are stable while CIP
of cations with coordinating abilities (lithium) give rise
to dealkoxylation in a process under kinetic control,
catalyzed by the electrostatic and the electrophilic prop-
erties of the cation. On the other hand, the cases in which
solvent-separated ionic pairs (SSIP) are involved give rise
to dealkylation. Comparisons with experiments carried
out with TCCI compounds, with well-defined electrostatic
interactions in the ionic pair, have led to the identifica-
tion of electrostatic interactions in the SSIP as the main
responsible for the observed major dealkylation. This
electrostatic catalysis would act by lowering the barrier
of the intramolecular π*-σ* electron transfer processes,
reducing the kinetic control of the reaction (in its absence
the alkyl-O bond cleavage is a forbidden process) in such
a way that the thermodynamically more stable product
is produced, overcoming the spin regioconservation prin-
ciple.
C
27H32NOBF4: C, 68.51; H, 6.81; N, 2.96. Found: C, 68.09; H,
7.03; N, 2.88.
N-[2-(2,6-Dip h en ylp h en oxy)bu tyl]p ip er id in e was puri-
fied by bulb-to-bulb distillation at 1 Torr (bath temperature,
230 °C): 1H NMR (CDCl3) δ 1.08-1.19 (m, 4H), 1.32-1.45 (m,
2H), 1.45-1.59 (m, 4H) 1.88-2.02 (m, 2H), 2.19 (s, 4H) 3.12-
3.22 (m, 2H), 7.17-7.22 (m, 1H), 7.28-7.36 (m, 4H), 7.36-
7.46 (m, 4H), 7.59-7.66 (m, 4H); 13C NMR (CDCl3) δ 23.14,
24.43, 25.89, 27.90, 54.27, 58.78, 72.92, 123.99, 126.92, 127.92,
129.34, 129.44, 130.11, 136.02, 138.75, 154.02. Anal. Calcd for
C27H31NO: C, 84.11; H, 8.10, N, 3.63. Found: C, 83.95; H, 7.90;
N, 3.58.
N-[-4-(2,6-Diph en ylp h en oxy)bu tyl]-N-m eth ylp ip er id in -
iu m tetr a flu or obor a te (1f): mp 230-231.5 °C; IR (KBr)
3045, 2948, 1063, 760, 740 cm-1; 1H NMR (DMSO-d6) δ 0.98-
1.28 (m, 4H), 1.37-1.82 (m, 6H), 2.80 (s, 3H), 2.92-3.15 (m,
6H), 3.21 (t, J ) 5 Hz, 2H), 7.23-7.61 (m, 13H); 13C NMR
(DMSO-d6) δ 18.1, 19.3, 20.8, 26.2, 47.1, 60.1, 62.1, 72.2, 124.7,
127.4, 128.4, 129.4, 130.4, 135.7, 138.5, 153.5.
Gen er a l P r oced u r e for th e Lith iu m -P r om oted Red u c-
tive Clea va ge of Com p ou n d s (1a -f) Descr ibed in Ta ble
1. Li metal [50 mg atom, 1.15 g of a 30% dispersion in mineral
oil (Aldrich),39 2.5 equiv] was placed under Ar in a two-necked
flask equipped with reflux condenser and magnetic stirrer,
washed with anhydrous THF (3 × 10 cm3), and suspended in
anhydrous THF (30 cm3). To this suspension, cooled to 25 °C
if necessary, the appropriate substrate (1a -f) (20 mmol) was
added at once without solvent, and the mixture was stirred
at 25 °C for 24 h. The mixture was cooled to 0 °C and quenched
by slow dropwise addition of H2O (10 cm3) (Ca u tion ! Str on gly
exoth er m ic!). After 1 h of stirring at room temperature the
mixture was neutralized with aqueous HCl, cooled, and
extracted with Et2O (3 × 30 cm3); the organic layer was then
separated and dried (CaCl2) and the solvent evaporated. The
ratio between the reaction products (Table 1) was determined
on the crude reaction mixture by 1H NMR; this ratio was
confirmed, and the yields were determined by separation of
the reaction products on flash chromatography, with mixtures
of hexane and EtOAc as eluent. In the case of the N-
methylpiperidinium tetrafluoroborates (1d -f), the aqueous
layer after the Et2O extraction was evaporated to dryness to
afford an untractable sticky residue; any attempt to purify this
mixture failed.
It is interesting to notice that the regioselectivity of
the mesolytic fragmentation can be completely reversed
by playing with appropriately designed intermediate ionic
pairs CIP-B and SSIP-E (Figure 6) and the associated
kinetic or thermodynamic control for the reaction.
Exp er im en ta l Section
1
Gen er a l Con sid er a tion s. H NMR were recorded at 250
or 400 MHz and the 13C NMR at 62.5 or 100 MHz for solutions
in CDCl3 and DMSO-d6 with tetramethylsilane (TMS as an
internal standard). Flash chromatography was performed on
silica gel (ICN Silica 32-63, 60 Å). Elemental analyses were
performed at the Microanalytical Laboratory of the Diparti-
mento di Chimica, Universita` di Sassari, or at the “Servei
d’Analisi Qu´ımica de la Universitat Auto`noma de Barcelona”.
Syn th esis of Sta r tin g Ma ter ia ls a n d Rea ction P r od -
u cts. m-Terphenyl (2) and 2,6-diphenylphenol (3) were pur-
chased from Aldrich. 2,6-Diphenylanisole (1a ) was prepared
according to a literature method.38 2,6-Diphenylphenyl hexyl
ether (1b) and N-(-2-(2,6-diphenylphenoxy)ethyl)-N-meth-
ylpiperidinium tetrafluoroborate (1d ) were prepared as previ-
ously described.21
N-[2-(2,6-Dip h en ylp h en oxy)eth yl]p ip er id in e (1c) was
prepared by the reaction of the sodium salt of 3 (prepared by
treatment of 3 with excess NaH in THF) with N-(2-chloroet-
hyl)piperidine (purchased from Aldrich) in THF under reflux
for 15 h and purified by bulb-to-bulb distillation at 1 Torr (bath
temperature, 210 °C): 1H NMR (CDCl3) δ 1.21-1.42 (m, 6H),
1.90-2.02 (m, 4H), 2.11 (t, J ) 6.0 Hz, 2H), 3.32 (t, J ) 6.0
Hz, 2H), 7.19-7.27 (m, 1H), 7.29-7.37 (m, 4H), 7.38-7.46 (m,
4H), 7.58-7.65 (m, 4H); 13C NMR (CDCl3) δ 24.16, 25.72, 54.26,
58.37, 70.18, 124.19, 125.05, 128.01, 129.51, 130.22, 136.10,
138.67, 154.02. Anal. Calcd for C25H27NO: C, 83.99; H, 7.61;
Electr och em ica l Mea su r em en ts. The electrochemical cell
and measurement procedures for CV have been described
previously.40 All potentials are reported vs an aqueous satu-
rated calomel electrode (except when indicated, i.e., Table 4);
glassy carbon was used as working electrode. The limiting
(39) Lithium metal tends to accumulate in the upper layer of
commercially available dispersions; drawing a sample without homog-
enizing the dispersion with a spatula can lead to stoichiometric errors.
(40) Andrieux, C. P.; Larumbe, D.; Gallardo, I. Electroanal. Chem.
1991, 304, 241.
(38) Luttringhaus, A.; Saaf, G. v. Liebigs Ann. Chem. 1939, 542,
241.

Reductive Cleavage of Alkyl Aryl Ethers
J . Org. Chem., Vol. 65, No. 2, 2000 331
lifetimes for the radical anions were calculated using k ) RT/
Fv as described in ref 25. Electrolysis reactions were carried
out by using a P.A.R. 273 A potentiostat.
indicated the presence of traces of m-terphenyl (2) in experi-
ments 1 and 3, and 2,6-diphenylphenol (3) as a single product
in experiments 2, 4-9. The presence of a large excess of
supporting electrolyte in the aqueous layer precluded any
detailed study on the fate of the alkyl chain after the cleavage
when compounds (1d -f) were used as starting material.
Gen er a l P r oced u r e for th e Electr och em ica lly P r o-
m oted Red u ctive Clea va ge of Com p ou n d s 1a ,d -f De-
scr ibed in Ta ble 3. A solution of starting material 1a ,d -f,
(0.64 mmol) in 150 mL of THF or DMF (0.1 M Et4NBF4 as
supporting electrolite except in experiment 4 where LiNBF4
was used) was electrolyzed using -2.6 V as the applied
potential with glassy carbon as electrode and under inert (N2)
atmosphere. After 2.1 F were consumed, the reaction was
quenched by addition of 30 mL of 1 M HCl. The reaction crude
was extracted between ether/water and the organic layer
washed several times with water. The organic layer was dried
and evaporated. Analysis of the residue by GC and 1H NMR
Ack n ow led gm en t. Financial support from DGES
(MEC of Spain) through project PB96-1145, from MURST
(40% and 60% Funds), and from “Generalitat de Cata-
lunya” through the project 1997SGR00414 and a grant
for F.C. is gratefully acknowledged.
J O991059V