Competing intermolecular and intramolecular hydride transfers in the ionic hydrogenation of (2-alkoxyphenyl)di(1-adamantyl)-methanols

Article abstract of DOI:10.1039/a903303c

ortho-Lithiation of anisole followed by reaction with di(1-adamantyl) ketone gives syn-(2-anisyl)di(1-adamantyl)-methanol with the C-OH proton intramolecularly hydrogen-bonded to the methoxy group. Reaction of the alcohol with trifluoroacetic acid (TFA) i

Full text of DOI:10.1039/a903303c

View Article Online / Journal Homepage / Table of Contents for this issue
Competing intermolecular and intramolecular hydride transfers in
the ionic hydrogenation of (2-alkoxyphenyl)di(1-adamantyl)-
methanols†
John S. Lomas*a and Jacqueline Vaissermannb
a Institut de Topologie et de Dynamique des Systèmes, Université de Paris 7,
associé au C.N.R.S., 1 rue Guy de la Brosse, 75005 Paris, France
b Laboratoire de Chimie des Métaux de Transition, Université de Paris 6, associé au C.N.R.S.,
Case 42, 4 place Jussieu, 75252 Paris Cedex 05, France
Received (in Cambridge) 26th April 1999, Accepted 15th June 1999
ortho-Lithiation of anisole followed by reaction with di(1-adamantyl) ketone gives syn-(2-anisyl)di(1-adamantyl)-
methanol with the C–OH proton intramolecularly hydrogen-bonded to the methoxy group. Reaction of the alcohol
with trifluoroacetic acid (TFA) in dichloromethane leads to a trifluoroacetate and a substituted phenol. These are
formed via a carboxonium ion, resulting from an intramolecular 1,5-hydride shift of the initially formed carbocation.
Ionic hydrogenation of the alcohol with TFA and a hydrosilane in dichloromethane results in the expected anti and
syn deoxygenation products as well as the trifluoroacetate and the phenol. anti-(2-Anisyl)diadamantylmethane
(benzylic hydrogen remote from methoxy, confirmed by a single crystal X-ray diffraction study) is formed directly
from the carbocation while the syn isomer results from reduction of the carboxonium ion. Reduction of the
carbonium ion is more hydrosilane-selective than that of the carboxonium ion. Kinetic isotope effects (kH/kD) on the
reaction of the carbocation at room temperature average 1.50 for triethylsilane and dimethylphenylsilane. Analogous
reaction of the (2-ethoxyphenyl) derivative gives only syn-(2-ethoxyphenyl)diadamantylmethane and the substituted
phenol. Kinetic isotope effects on the reduction of the corresponding carboxonium ion average 1.34 for the same
hydrosilanes.
investigation of hydrogen bonding and possible hydride trans-
fer in alcohols bearing an alkoxy group. In this paper the
synthesis and reactions of (2-anisyl)di(1-adamantyl)methanol,
1S-Me, and, more briefly, of its (2-ethoxyphenyl) analogue,
Introduction
The chemistry of compounds containing 1-adamantyl groups1
is often characterized by the rigidity and integrity of these
groups. Because of their rigidity, adamantyl groups are steric-
ally more demanding than tert-butyl, for example, and tend to
increase rotation barriers in molecules where conformational
isomerism occurs, making it possible to isolate rotamers as dis-
OR
tinct species, stable under normal experimental conditions.2–5
1S-Me
1S-Et
Furthermore, they have a marked tendency to conserve their
structural identity regardless of what is happening at the
carbon atom to which they are attached. For this reason the
tri(1-adamantyl)methyl radical6 and cation7 are persistent,
and o-tolyldi(1-adamantyl)methyl cations are relatively stable
species which can be studied at room temperature.8,9 Other
aryldi(1-adamantyl)methyl cations can be observed at low
temperaures in superacid media.10 Aryl- or heteroaryldi-
(1-adamantyl)methyl derivatives owe much of their unusual
chemistry to the fact that the adamantyl groups tend to be
located on either side of the ring plane, though not usually
symmetrically, and that a fourth substituent to the methyl
carbon or, in the case of carbocations, the vacant orbital, is
necessarily close to the ring plane. This favours intramolecular
hydrogen bonding in (2-pyridyl)diadamantylmethanols11 and
hydride transfer from a suitable ortho-alkyl substituent to the
aryldiadamantylmethyl cation.12
Ad
OH
Ad
1S-Et, are described. First we shall consider their structures
and then the reactivity of the corresponding carbocations.
When carbocations are generated from alcohols in the presence
of various hydride donors, such as hydrosilanes, they are
reduced in a reaction generally known as “ionic hydrogenation”
to give products of alcohol deoxygenation.14 This has proved to
be an interesting means of investigating the stereochemistry of
intermolecular hydride transfer in conformationally locked
systems.15
Results and discussion
Alcohol synthesis and structure
ortho-Lithiation of anisole16 by means of n-BuLi–TMEDA
in diethyl ether or THF, followed by reaction with di(1-ada-
mantyl) ketone, gives the corresponding (2-anisyl)diadamantyl-
Aromatic ethers such as anisole are lithiated at the ortho
position by exchange with n-butyllithium.13 It occurred to us
that this might be an interesting starting point for the further
1
methanol, 1S-Me, as a single isomer in good yield. H NMR
(δH of OH = 6.58 ppm in chloroform) and IR spectra (νOH 3499
cmϪ1) of the product show clearly that the alcohol proton is
intramolecularly hydrogen-bonded and, therefore, that it is the
syn rotamer with the OH group in proximity to the alkoxy sub-
stituent. The position of the 1H NMR signal varies by less than
† Details of all KINAL calculations (Tables 5S and 6S) are available as
supplementary data. For direct electronic access see http://www.rsc.org/
suppdata/p2/1999/1639, otherwise available from BLDSC (SUPPL.
NO. 57589, pp. 2) or the RSC Library. See Instructions for Authors
available via the RSC web page (http://www.rsc.org/authors).
J. Chem. Soc., Perkin Trans. 2, 1999, 1639–1648
1639

View Article Online
0.002 ppm when the concentration is reduced 1000-fold from
0.13 M, is almost independent of the solvent, and has a low
temperature coefficient (Ϫ1.6 ppb ЊCϪ1) as in the 2-pyridyl
analogues.11 The (2-ethoxyphenyl) derivative, 1S-Et, prepared
in the same way as 1S-Me, shows very similar NMR and IR
characteristics, with the NMR shift of the alcohol proton
slightly greater (6.79 ppm in chloroform) and the IR νOH
slightly lower (3480 cmϪ1). The differences between the IR
maxima of 1S-Me and 1S-Et and those of free OH groups are
much more modest than the corresponding differences in the
1H NMR signals of the OH protons, which means that these
alcohols would not fit the previously established NMR–IR
correlation, essentially based on OH ؒ ؒ ؒ N hydrogen bonds in
pyridyl and thiazoyl derivatives.11
OCH2OCOCF3
OH
H
OCH2OH
Ad
Ad
H
Ad
H
Ad
Ad
Ad
2
4
3
acetate (about 90%) and a minor product, associated with 13C
NMR signals indicating the presence of a CH2 group, a
benzylic CH, three (of presumably 4) aromatic CH groups and
two quaternary aromatic carbons. These data are compatible
with a substituted phenoxymethanol, 3, but, unfortunately, no
comparable data are available in the literature and, moreover,
the product could not be isolated.
Treatment of the alcohol with TFA in dichloromethane
followed by conventional aqueous work-up gives the trifluoro-
acetate, 2, and a phenol, 4, in the same proportions as 2 and 3 in
the NMR experiment. The two products are separable by gas
chromatography on a capillary column and give appropriate
ion trap detector (ITD) spectra, the molecular ion being, how-
ever, very weak even with chemical ionization. The trifluoro-
acetate is converted to the same phenol upon chromatography,
being hydrolysed by the acidic silica gel. In chloroform the
13C NMR shifts of the benzylic carbons in 2 and 4 are 56.1 and
57.2 ppm, respectively. These values are to be compared with
values of about 60 ppm for a benzylic group with the hydrogen
oriented towards various modified ortho ethyl and isopropyl
groups, whereas the opposite orientation is associated with a
shift of 74 ppm.12
Molecular mechanics calculations on the steric energies of
the rotamers which would be produced by simple deoxygen-
ation of the alcohol 1S-Me make the syn more stable than the
anti isomer by 2.1 kcal molϪ1, while the syn phenol is estimated
to be 2.4 kcal molϪ1 more stable than the anti rotamer, and
similar differences can be assumed for the phenoxymethanol
and trifluoroacetate rotamers. The NMR spectra and calcu-
lation establish therefore that compounds 2, 3 and 4 are the syn
rotamers. Identification of these materials as the anti isomers
would conflict with the NMR data and also require a totally
implausible rotation to a less stable isomer.
A reasonable interpretation of these results is that formation
of the (2-anisyl)diadamantylmethyl cation, 5, is followed by
1,5-hydride transfer from the adjacent methoxy group to the
syn face of the cation with the formation of an oxygen-
stabilized carboxonium ion, 6.23 This latter is then attacked
by trifluoroacetate ion to give trifluoroacetate, 2, or by water,
either adventitious or formed at the same time as the carbo-
cation, to give the phenoxymethanol, 3 (Scheme 1). Analogous
results have been reported for phenyldiadamantylmethanols
with 2-ethyl or 2-isopropyl substituents, where carbocation
formation is followed by 1,4-hydride transfer, to give various
products with a modified ortho substituent.12 1,5-Hydride shifts
are more common than 1,4-, probably because it is easier to
attain the appropriately short Cϩ ؒ ؒ ؒ H distance in the larger
system and to satisfy the stereoelectronic requirements of the
transition state.24 In the present case, the overall reaction is
qualitatively slower than that of the ethyl or isopropyl deriv-
atives,12 but this is related to the rate at which the respective
carbocations are formed and has no bearing on the relative
rates of 1,4- and 1,5-hydride shifts.
These results are consistent with previous work which shows
that relatively strong hydrogen bonding occurs in α,α-disubsti-
tuted o-methoxybenzyl alcohols.17 In the di(tert-butyl) deriv-
ative the hydrogen bond formation enthalpy, estimated by
studying the temperature dependence of the bonded and free
IR absorptions at 3508 and 3616 cmϪ1, amounts to 1.4 kcal
molϪ1.‡ The νOH of 1S-Me and 1S-Et at 3499 and 3480 cmϪ1
,
respectively, implies slightly stronger hydrogen bonds; we were
unable to locate the absorption of free OH.
Molecular mechanics [MM2(85)] calculations18 on the steric
energies of the 2-anisyl alcohol, 1-Me, make the syn isomer 1.8
kcal molϪ1 more stable than the anti. This is a much smaller
difference than for the corresponding 2-methyl (7.0 kcal
molϪ1 3b or 2-ethyl (70.912 Ϫ 63.7 = 7.2 kcal molϪ1) derivatives,
)
which suggests that the effective size of the methoxy group is
less than that of either. The molecular mechanics calculation
does not take into account hydrogen bonding, which means
that the real energy difference between the rotamers must be,
according to previous work,17 greater by probably slightly more
than 1.4 kcal molϪ1. An attempt to determine the structure of
the alcohol by X-ray diffraction was not totally successful.19
There is one disordered dichloromethane molecule for two 1S-
Me molecules (confirmed by elemental analysis) and the struc-
ture failed to converge to an acceptable R value. All other atoms
were located either directly or from the Fourier difference map.
The H ؒ ؒ ؒ O and O ؒ ؒ ؒ O distances are 1.47 and 2.55 Å, respect-
ively, and the O–H ؒ ؒ ؒ O angle is about 168Њ. These are fairly
typical values for this type of hydrogen bond.20 All other
features are typical of aryl- and heteroaryldiadamantylmethyl
derivatives previously studied.4,5,12
Trifluoroacetylation of syn-(2-anisyl)di(1-adamantyl)methanol,
1S-Me
The alcohol was treated with TFA (4% v/v) in deuteriated
dichloromethane in an NMR tube at 25 ЊC. The addition of
TFA causes no colouration but the immediate disappearance of
the proton signal at 6.58 ppm and a small downfield displace-
ment of the aromatic and methyl group proton signals. The
first-order rate constant, kexp, based on the progressive dis-
appearance of the methyl signal is about 7 × 10Ϫ4 sϪ1, this value
being little affected by trifluoroacetic anhydride (TFAA), added
to the medium in order to scavenge adventitious water or that
produced by reaction of the alcohol. The methyl signal is
replaced by CH2 and benzylic CH signals at 6.00 ppm and 2.84
ppm, respectively. The two groups were further identified by 13
C
NMR (corresponding signals at 89.9 and 56.5 ppm, respect-
ively). The values for the proton and carbon shifts of the CH2
group are closely similar to those reported for aryloxymethyl
benzoates21 and alkoxymethyl acetates,22 respectively, and
clearly indicate that the methoxy group is trifluoroacetylated at
the same time as the alcohol is deoxygenated, to give a tri-
fluoroacetate, 2. Inspection of the proton spectrum taken after
about 10 half-lives reveals complete conversion to trifluoro-
Aryloxymethyl cations do not appear to have been described
previously but alkoxymethyl cations have long been proposed
as intermediates in the solvolysis of acetals, alkoxymethyl esters
and chloromethyl ethers,25,26 and have been directly observed
by NMR at low temperature.23b They are also formed by 1,6-
and higher order hydride shifts in the gas-phase, but not
the solution-phase, reactions of CH3O(CH2)nSCH2
ϩ 26
.
A 1,5-
hydride shift from an ethoxy group to a carbocation has also
been reported.27
‡ 1 cal = 4.184 J.
1640
J. Chem. Soc., Perkin Trans. 2, 1999, 1639–1648

View Article Online
Table 1 X-Ray crystallographic data for anti-(2-anisyl)diadamantyl-
methane, 7A-Me. Comparison with MMP2(85)-calculated geometry
OMe
OH
7A-Mea
MMP2(85)
Ad
Bond lengths/Å
Ad
C(1)–C(10)
C(10)–C(101)
C(10)–C(201)
C(1)–O(1)
1.539(6)
1.589(5)
1.585(5)
1.376(5)
1.414(5)
1.539
1.576
1.579
1.371
1.414
1S-Me
OCH2OCOCF3
O(1)–C(7)
Ad
H
Bond angles/Њ
Ad
C(1)–C(10)–C(101)
C(1)–C(10)–C(201)
C(2)–C(1)–C(10)
C(2)–O(1)–C(7)
C(1)–C(2)–O(1)
110.7(3)
121.5(3)
126.1(3)
117.2(3)
119.4(3)
112.4
123.9
123.1
117.3
119.1
O+CH2
2
OMe
Ad
+
Ad
H
Ad
Ad
OCH2OH
5
6
Torsion angles/Њ
C(2)–C(1)–C(10)–C(101)
C(2)–C(1)–C(10)–C(201)
C(1)–C(10)–C(101)–C(102)
C(1)–C(10)–C(101)–C(108)
C(1)–C(10)–C(101)–C(109)
C(1)–C(10)–C(201)–C(202)
C(1)–C(10)–C(201)–C(208)
C(1)–C(10)–C(201)–C(209)
C(10)–C(1)–C(2)–O(1)
Ϫ79
63
Ϫ171
64
Ϫ55
139
24
Ϫ100
Ϫ5
Ϫ83
66
Ϫ165
70
Ϫ48
134
18
Ϫ104
0
Ad
H
Ad
3
Scheme 1
Reaction of 1S-Et under the same conditions as for 1S-Me
gives the phenol, 4, and acetaldehyde directly. At no time is
there an NMR signal corresponding to a trifluoroacetate.
C(1)–C(2)–O(1)C–(7)
170
159
a Means of values for two independent molecules.
Competing trifluoroacetylation and ionic hydrogenation
syn-(2-Anisyl)di(1-adamantyl)methanol. (i) General aspects.
A much-used method for alcohol deoxygenation, known as
ionic hydrogenation, involves hydrosilane reduction of the car-
bocation generated in TFA–dichloromethane.14 When applied
to 1S-Me this procedure gives a mixture of four materials, in
proportions depending on the nature of the hydrosilane and
1
its concentration. The H and 13C NMR spectra of the crude
mixture obtained with triethylsilane (TES) (TES:1S-Me = 4)
indicate that it consists of two (2-anisyl)diadamantylmethanes,
in a ratio of about 4:1, a trifluoroacetate and a phenol. The last
two compounds are the same, 2 and 4, respectively, as those
already identified in the reaction of the alcohol with TFA alone.
Chromatographic separation of the products leads to isolation
of the anisyldiadamantylmethane mixture, 7-Me, and of
(2-hydroxyphenyl)diadamantylmethane, 4, in good overall
yield. Abandoning the first fraction for several days at room
temperature yielded crystals of the minor isomer. A single crys-
tal X-ray diffraction study showed it to have the anti conform-
ation, that in which the benzylic hydrogen is remote from the
methoxy group.§
Fig.
1 CAMERON diagram for anti-(2-anisyl)di(1-adamantyl-
methane), 7A-Me, showing 30% probability displacement ellipsoids.
The geometry (Table 1, Fig. 1) is typical of aryl- and hetero-
aryldiadamantylmethyl derivatives,4,5,12 and is in good agree-
ment with that calculated by molecular mechanics.
Hydrogen atoms have been omitted for clarity.
The isomeric (2-anisyl)diadamantylmethanes, 7A-Me and
7S-Me, are associated with benzylic carbon 13C NMR shifts of
72.3 and 55.3 ppm, respectively. This finding confirms that the
differences observed for other aryldiadamantylmethanes12,15
apply also in the present series.
The carboxonium ion 6 is, therefore, either attacked by
nucleophiles to give 2 and 3 or is reduced by hydride transfer
from the hydrosilane to give syn-(2-anisyl)diadamantyl-
methane, 7S-Me. A competing pathway is intermolecular
hydride transfer from the hydrosilane to the initially formed
OR
OR
Ad
Ad
H
H
Ad
Ad
7S-R
7A-R
carbocation, 5, leading to anti-(2-anisyl)diadamantylmethane,
7A-Me (Scheme 1). However, we cannot at this stage exclude
the possibility that all or part of the syn isomer is also the result
of direct hydride transfer from the hydrosilane. It should be
remarked that in the analogous reaction of (2-alkylphenyl)-
diadamantylmethanols intermolecular hydride transfer does
not occur,12 except when sodium borohydride is used.15 The fact
that 7A-Me is formed at all indicates that the intermolecular/
intramolecular rate ratio is higher in the present system.
§ In the past3,5,12 we have systematically used the same conformational
descriptor, anti or syn, for an alcohol and the alkane obtained by
removal of the oxygen atom, despite the fact that OH and H do not
have the same priority with respect to carbon. This practice, though
incorrect, has the advantage that analogous structures bear the same
descriptor, and we shall continue to employ it here.
J. Chem. Soc., Perkin Trans. 2, 1999, 1639–1648
1641

View Article Online
Table 2 Reduction of (2-anisyl)di(1-adamantyl)methanol, 1S-Me, by TFA–hydrosilane in dichloromethane; initial [ROH] = 0.0183–0.0192; initial
[TFA] = 0.48–0.50; estimated [R3SiH] at one half-life.a Rate constants for reaction of hydrosilanes with 4-anisylphenylcarbenium ion at Ϫ70 ЊC
(ref. 28)
R3SiH
[R3SiH]
7a-Me
7S-Me
2
4
%7A/(100 Ϫ %7A)
%7S/(%2 ϩ %4)
k2/dm3 molϪ1 sϪ1
None
NHS
DPS
TPS
TIPS
TES
—
—
12
9
0
0
8
27
21
—
47
45
37
28
41
34
9
91
36
38
49
58
40
31
54
9
5
8
15
14
11
9
—
—
—
0.070
0.071
0.072
0.073
0.071
0.070
0.072
0.138
0.098
0.0
1.14
0.99
0.58
0.40
0.82
0.86
0.13
0.048
1.20
8.27
36.7
124
149
457
0.0
0.087
0.361
0.259
DMPS
TTMSS
17
a Data are rounded to the nearest integer %; sum = 100 1%. Ratios calculated before rounding.
Heating the isolated mixture of anisyldiadamantylmethanes
k5
k1
k0
5
2
6
1S
for 8 h at 240 ЊC results in complete conversion to the syn iso-
mer, 7S-Me, the half-life being about 1.5 h at this temperature;
this corresponds to an activation energy of around 40 kcal
molϪ1.Previouswork8 hasshownthatanti-(2-tolyl)diadamantyl-
methane, 8A-H, rotates much more slowly than the correspond-
k3
k2
k5'
k4
7A
7S
3
4
Scheme 2
addition of TFAA has little effect upon the rate of reaction of
1S-Me suggests that the first step is not reversible, i.e. that the
carbocation is reduced or converted to carboxonium ion as
soon as it is formed, and that kexp = k0.
Me
If direct reaction of 5 to give 7S-Me is neglected, this scheme
suggests that the ratio of anti-anisyldiadamantylmethane to all
other products, %7A/(100 Ϫ %7A) (the tag “-Me” is omitted
from all ratios), will be given by k2[R3SiH]/k1, and the ratio of
syn-anisyldiadamantylmethane to 2 and 4, %7S/(%2 ϩ %4), by
k4[R3SiH]/(k5 ϩ k5Ј), where [R3SiH] is the hydrosilane concen-
tration, assumed to be in sufficiently large excess for it to be
treated as constant.
DMPS and TES were chosen for further study. Changing the
DMPS concentration leads to significant variations in the
product composition (Table 3). If the initial molar ratio of
hydrosilane to alcohol is raised from 4 to 16 there is a regular
increase in the yield of anti-anisyldiadamantylmethane, 7A-
Me, and a decrease in 2 and 4, with that of 7S-Me going
through a maximum.
8A-H
Ad
X
8A-OH
Ad
ing alcohol, 8A-OH, the barrier difference being about 5 kcal
molϪ1
That of anti-(2-anisyl)diadamantylmethanol is
.
expected therefore to be of the order of 35 kcal molϪ1, which
means that if it were formed it would be isolated as such.
The fact that only the syn alcohol can be isolated suggests
that the stereochemistry of the addition is controlled by
attractive interactions between the incipient alcohol oxygen
and the methoxy group. The difference between the rotation
barriers of ortho-tolyl- (ca. 45 kcal molϪ1 at 250 ЊC)8 and
(2-anisyl)-diadamantylmethanes is consistent with the MM
calculations, which indicate that methoxy is less sterically
demanding than methyl or ethyl.
Inspection of the resulting values of k2/k1 and k4/(k5 ϩ k5Ј)
obtained by dividing the product ratios by the hydrosilane con-
centration at one half-life indicates that Scheme 2 is valid for
the anti isomer but perhaps not for the syn; values of k2/k1 show
only random variations, averaging 5.08 0.17 MϪ1. Analogous
experiments carried out with TES give similar results (Table 3)
(ii) Hydride donor product dependence. A number of hydro-
silanes covering a wide range of reactivity28 and bearing sub-
stituents of very different sizes were chosen as hydride donors.
Varying the hydride donor (R3SiH:1S-Me = 4) has substantial
effects upon the product ratios (Table 2). n-Hexylsilane (NHS),
TES and diphenylsilane (DPS) give similarly small yields of 7A-
Me, with 2 and 7S-Me predominating, while in the presence
of dimethylphenylsilane (DMPS) the yield of anti-anisyl-
diadamantylmethane, 7A-Me, is almost as great as that of the
syn isomer, 7S-Me. More sterically hindered hydrosilanes such
as triphenylsilane (TPS) and tri(isopropyl)silane (TIPS) give no
7A-Me and modest yields of 7S-Me, with trifluoroacetate 2
and phenol 4 accounting for about two-thirds of the reaction
products. In the case of tris(trimethylsilyl)silane (TTMSS),
anti-anisyldiadamantylmethane is formed in preference to the
syn isomer.
but with a significantly smaller value for k2/k1, 1.18 0.03 MϪ1
.
An alternative, rather more rigorous method of calculating k2/
k1 is to solve the kinetic differential equations by means of a
kinetic analysis programme such as KINAL29 and to use the
yields of 7A-Me and 7S-Me as targets (see Experimental
section), starting with the calculated concentrations of alcohol
and hydrosilane. This avoids the approximation involved in
assuming that [R3SiH] is constant and leads to k2/k1 values of
5.04 0.15 and 1.17 0.02 MϪ1 for DMPS and TES, respect-
ively, making the ratio of k2 values (assuming k1 constant) for
DMPS with respect to TES about 4.3.†
These variations in product selectivity clearly depend on dif-
ferences between the rates of intermolecular hydride transfer
and those of intramolecular hydride transfer and nucleophilic
attack by trifluoroacetate ion or water on the rearranged cation.
The simple kinetic scheme shown in Scheme 2 is an attempt to
rationalize these results; the TFA concentration is assumed to
be constant and is included in the appropriate rate constants.
The question arises as to whether the rate constant for tri-
fluoroacetylation, kexp, corresponds to k0 or k1. In fact, the
carbocation could be in pre-equilibrium with the alcohol,
whereupon the observed rate constant would be Kk1, where K
is a (very small) equilibrium constant. However, the fact that
Values of k2/k1 calculated from the single-concentration
experiments for the other hydrosilanes range from zero (TPS
and TIPS) to 0.36 MϪ1 (TTMSS) and follow no obvious trend
in steric or polar effects. As has been observed in other work on
the ionic hydrogenation of aryldi(1-adamantyl)methanols,15
there is no correlation of the relative rates of reduction of the
carbocation with the rates of hydride transfer from these
hydrosilanes to diarylcarbenium ions, which are a reflection of
almost pure polar effects upon the stability of the incipient
silylium ion.28 The two bulky hydrosilanes, TPS and TIPS, of
moderate reactivity do not react at all with the carbocation,
whereas the bulky but highly reactive TTMSS does. Steric
1642
J. Chem. Soc., Perkin Trans. 2, 1999, 1639–1648

View Article Online
Table 3 Hydrosilane concentration and isotope effects on the reduction of (2-anisyl)di(1-adamantyl)methanol, 1S-Me, by TFA–hydrosilane in
dichloromethane; initial [ROH] = 0.0183–0.0192; initial [TFA] = 0.48–0.50; estimated [R3SiL] at one half-lifea
R3SiL
[R3SiL]
7A-Me
7S-Me
2
4
%7A/(100 Ϫ %7A)
%7S/(%2 ϩ %4)
TES-h
TES-h
TES-h
TES-h
TES-d
TES-d
TES-d
DMPS-h
DMPS-h
DMPS-h
DMPS-h
DMPS-d
DMPS-d
DMPS-d
0.071
0.144
0.216
0.286
0.145
0.216
0.286
0.070
0.143
0.215
0.285
0.144
0.216
0.286
8
14
21
25
10
15
19
27
43
52
59
31
42
48
41
50
53
53
49
53
55
34
37
34
31
38
35
33
40
28
21
18
33
25
21
31
16
11
9
11
8
6
5
9
7
5
9
4
3
2
6
4
3
0.087
0.167
0.258
0.333
0.112
0.179
0.233
0.361
0.757
1.062
1.410
0.458
0.721
0.916
0.817
1.390
1.970
2.363
1.180
1.706
2.164
0.863
1.802
2.382
2.879
1.224
1.450
1.757
25
20
16
a Data are rounded to the nearest integer %; sum = 100 1%. Ratios calculated before rounding.
effects would appear to be more important in the present
reaction than in Mayr’s work.28
have been proposed for hydride transfers from hydrosilanes to
carbocations.
The values of k4/(k5 ϩ k5Ј), whether they be obtained simply
from the %7S/(%2 ϩ %4) ratios or by the KINAL treatment
above, tend to decrease as [R3SiH] increases, by about 20–30%
for both hydrosilanes as [R3SiH]:[1S-Me] goes from 4 to 16,
average values being 11.7 1.2 and 9.7 1.4 MϪ1 for DMPS
and TES, respectively. If an average value is taken for k2/k1 and
the direct route from 5 to 7S-Me by hydride transfer to the syn
face of the carbonium ion is included as only insignificantly
small, random values of k3/k1 are obtained by KINAL opti-
mization. Comparison of %7S/(%2 ϩ %4) ratios for like values
of [R3SiH] indicates that the reactivities of NHS, DPS, DMPS
and TES with the carboxonium ion are very similar. The more
bulky hydrosilanes, TPS and TIPS give significant amounts
of the syn isomer, with %7S/(%2 ϩ %4) ratios not much
less than for the other hydrosilanes. TTMSS is exceptional in
that both %7A/(100 Ϫ %7A) and %7S/(%2 ϩ %4) are small, the
latter being the smaller (Table 3).
The isotope effect study (vide infra) confirms the kinetic
analysis and shows that syn-anisyldiadamantylmethane is
formed from the H-shifted cation and not directly from the
initially formed cation. However, its rate of formation does not
appear to be a simple function of the hydrosilane concentration
and/or the rates of formation of 2 and 3 are to some extent
[R3SiH]-dependent. The ratio of trifluoroacetate to phenol,
%2/%4, shows no significant variation with the DMPS or TES
concentration, being 3.7 0.3 for the 8 data. Values for the
other hydrosilanes range from 3.3 to 6.8. All these values are
surprisingly smaller than in the simple reaction of 1S-Me with
TFA, where it is about 9, and the phenol yield is in some cases
higher than when no hydrosilane is present. A logical interpret-
ation would be that hydrosilane is involved in the formation of
the phenol, but this makes little sense chemically.
The KINAL-Simplex treatment can be extended to include
the variation of the water concentration due to its formation
from the alcohol and reaction with 6 and possibly with the
silylium ion.30 However, this necessitates the separation of k5
and k5Ј, and involves the explicit introduction of k0, k1 and a
rate constant for reaction of silylium ion with water. Rate con-
stant k0 is known experimentally, the other two and k5Ј can be
set arbitrarily, whereupon k2, k4 and k5 can be optimized. In this
way relative rate constants are calculated; they are not very
sensitive to the arbitrary values. As far as k2/k1 is concerned this
treatment gives similar results to the above, 4.98 0.15 and
1.15 0.02 MϪ1 for DMPS and TES, respectively. TES shows
the same trend as before for k4 but not DMPS; as expected,
similar values of k5 are found for the two hydrosilanes.
R3SiH ϩ ϩCRЈ3
[R3SiH ϩCRЈ3]
encounter
complex
{R3SiϩH ؒ ؒ CRЈ3}
free radical pair
R3Siϩ ϩ HCRЈ3 (1)
R3Siϩ ϩ HCRЈ3 (2)
R3SiH ϩ ϩCRЈ3
[R3Si ؒ ؒ ؒ ؒ H ؒ ؒ ؒ ؒ CRЈ3]ϩ
transition state
Arguments in favour of the SET mechanism advanced by
Chojnowski et al.31 have been countered by recent work on
isotope effects28 and ab initio calculations,32 which strongly
support the SHT process; theory indicates an early, linear tran-
sition state for the reaction of CH3ϩ with SiH4. The recent sug-
gestion that the transition structure of the rate-determining
step (in the hydride transfer reaction from arylsilanes to carbo-
cations) is close to the intermediate silyl cations33 is based on
what seems to be an erroneous interpretation of Yukawa–
Tsuno rϩ values.
In the ionic hydrogenation of 1S-Me, replacing DMPS by the
silicon-deuteriated compound, Me2PhSiD, has relatively small
effects on the product composition, the most notable change
being a decrease in the amount of anti-(2-anisyl)diadamantyl-
methane, 7A-Me, and an increase in trifluoroacetate, 2 (Table
3). NMR study of the product mixture shows no benzylic CH
proton signal for this compound and a very weak triplet in the
13C spectrum. The syn isomer, 7S-Me, on the other hand has
normal CH signals but the methyl signals are replaced by trip-
lets slightly upfield from the usual values: 55.4 ppm (J = 22 Hz)
for carbon and 3.77 ppm (J = 1.5 Hz) for proton. The spectra of
the trifluoroacetate 2 and phenol 4 are perfectly normal. These
results are consistent with the mechanism outlined above, and
establish that syn-(2-anisyl)diadamantylmethane, 7S-Me, is
formed by reduction of a H-shifted carbocation, i.e. the carb-
oxonium ion, 6.
Calculation of the kinetic isotope effects (KIE) from the
product composition, individual values of the %7A/(100 Ϫ
%7A) ratio for each DMPS concentration being used, leads to a
small value for the KIE on k2, 1.55 0.09 at room temperature
(20 2 ЊC). This is similar to values reported by Chojnowski31
at 25 ЊC and considered by him to be secondary isotope effects
supporting the SET mechanism. However, Mayr has shown
that small isotope effects are consistent with rate-determining
Si–H cleavage.28 It is more difficult to say anything about the
effects on the reduction of the carboxonium ion, since the
kinetic analysis is less satisfactory. Point-by-point comparisons
(iii) Isotope effects. Single electron transfer [SET, eqn. (1)]
and synchronous hydride transfer [SHT, eqn. (2)] mechanisms
J. Chem. Soc., Perkin Trans. 2, 1999, 1639–1648
1643

View Article Online
Table 4 Reduction of (2-ethoxyphenyl)di(1-adamantyl)methanol, 1S-Et, by TFA–hydrosilane in dichloromethane; initial [ROH] 0.0183–0.0192;
initial [TFA] 0.48–0.50; estimated [R3SiL] at one half-life.a Rate constants for reaction of hydrosilanes with 4-anisylphenylcarbenium ion at Ϫ70 ЊC
(ref. 28)
R3SiL
[R3SiL]
4
7S-Et
%7S-Et/%4
k2/dm3 molϪ1 sϪ1
KIE
None
NHS-h
DPS-h
TPS-h
TIPS-h
TES-h
TES-d
TES-h
TES-d
TES-h
DMPS-h
DMPS-d
DMPS-h
DMPS-d
DMPS-h
DMPS-d
DMPS-h
TTMSS-h
—
100
97
85
78
60
27
31
15
19
12
33
39
17
22
12
16
10
61
0
3
0.076
0.075
0.073
0.072
0.069
0.070
0.143
0.143
0.214
0.070
0.070
0.143
0.143
0.215
0.215
0.285
0.072
0.032
0.179
0.283
0.658
2.76
2.27
5.56
4.24
7.44
2.07
1.54
5.00
3.65
7.52
5.21
9.26
0.63
0.048
1.20
8.27
36.7
124
15
22
40
73
69
85
81
88
67
61
83
78
88
84
90
39
1.22
1.31
149
457
1.35
1.37
1.44
a Data are rounded to the nearest integer %; sum = 100 1%. Ratios calculated before rounding.
give a similar value, 1.58 0.10, which must correspond essen-
tially to the KIE on k4. The value for the isotope effect on k2
when TES is replaced by Et3SiD is 1.45 0.03, which is in fair
agreement with the value for DMPS, but the effect on k4 is
unexpectedly small, 1.14 0.05. We have no explanation for
this difference nor for such a small value. Treatment of the data
by KINAL in its reduced or more elaborate form does not give
significantly different results.
and this must enhance its stability. Conversely, there is no
evidence for nucleophilic attack by trifluoroacetate ion nor
for formation of a 1-phenoxyethanol analogue of 3. Instead,
the carboxonium ion eliminates acetaldehyde, presumably by
reaction with water, to give phenol 4 directly.
Variation of the nature and the concentration of the hydro-
silane (Table 4) gives results differing qualitatively from those
for 1S-Me. With a four-fold ratio of hydrosilane to alcohol, the
three least reactive hydrosilanes, according to Mayr’s classifi-
cation,28 NHS, DPS and TPS, give the smallest amounts of 7S-
Et while the most reactive, TTMSS, gives similar amounts of
7S-Et and 4, as does TIPS. TES and DMPS favour 7S-Et
relative to 4. Leaving aside TTMSS, which gives far too little
deoxygenation product, there is a rough correlation of the %7S-
Et/%4 ratio with Mayr’s rate constants28 for reaction with a
diarylcarbonium ion, TES and DMPS, however, being reversed.
The slope of the correlation is, nevertheless, small (1.5–2%) and
suggests that if the polar effects of the hydrosilane substituents
do affect k4 (Scheme 2) they are still of minor importance. What
is remarkable is that such a correlation should exist at all for
1S-Et when the corresponding rate constant for 1S-Me appears
to be controlled primarily by steric effects.
syn-(2-Ethoxyphenyl)di(1-adamantyl)methanol. Reaction of
alcohol 1S-Et with TFA in the presence of a hydrosilane is
much simpler than that of 1S-Me in that it gives only the
phenol, 4, and syn-(2-ethoxyphenyl)diadamantylmethane, 7S-
Et (Scheme 3). Using a 16-fold excess of the most reactive
OEt
Ad
OH
Ad
1S-Et
By analogy with Scheme 2 we can reasonably expect that the
ratio of 7S-Et to 4 will be given by k4[R3SiH]/k5Ј, in which
case %7S-Et/%4 should be a linear function of [R3SiH], with the
assumptions stated above. In the case of DMPS a point-by-
OEt
Ad
H
point evaluation of k4/k5Ј gives an average of 33.1 2.6 MϪ1
.
Ad
Again, a better method of calculating k4/k5Ј is by means of
KINAL29 using the yields of 7S-Et as targets. This gives an
average value of 33.2 2.5 MϪ1, in good agreement with the
first estimate. The same procedure applied to the three data for
normal TES gives a slightly higher value, 38.0 2.8 MϪ1. As
with 1S-Me, TES and DMPS react at similar rates with the
carboxonium ion. If k4/k5Ј is calculated from the %7S-Me/%4
ratios for reaction of 1S-Me, there is considerable scatter but no
well defined trend in the values, giving 44.1 5.7 and 54.5 4.9
7S-Et
O+CH-CH3
OEt
Ad
+
Ad
H
Ad
Ad
OH
6'
5'
Ad
H
Ad
M
Ϫ1 for TES and DMPS, respectively. These values are not very
4
different from those obtained for 1S-Et, which shows that the
relative rates of reduction and reaction with water are similar
for the two carboxonium ions. The major difference is that 6Ј
does not react with trifluoroacetate.
Kinetic isotope effects on k4 should be more reliable in this
system, given that there are only two products rather than four
in the previous case. Calculating for each pair of experiments
at the same hydrosilane:alcohol ratio (Table 4) gives values
slightly higher for DMPS (average 1.39 0.05) than for TES
Scheme 3
hydrosilane, DMPS, serves only to reduce the phenol yield. This
shows clearly that the relative rate of intramolecular hydride
shift to intermolecular hydride transfer from a hydrosilane is
much greater for the ethoxy group than for methoxy. This is
perfectly reasonable insofar as the primary carboxonium ion 6Ј
bears a methyl substituent on the formally charged carbon,
1644
J. Chem. Soc., Perkin Trans. 2, 1999, 1639–1648

View Article Online
(1.26 0.07, 2 points only). However, given the uncertainty on
the product ratios, even here the individual KIE values carry an
uncertainty of about 0.1. Neither the apparent variation of
the KIE with increasing DMPS concentration nor the differ-
ence between TES and DMPS can be considered as significant.
temperature under argon was added a solution of n-butyl-
lithium in hexane (1.6 M, 3 cm3, 4.8 mmol). After 1 h a solution
of di(1-adamantyl) ketone (0.153 g, 0.5 mmol) in diethyl ether
(20 cm3) was added in about 10 min. The reaction mixture was
stirred at room temperature overnight, then quenched with
water and the organic material extracted with hexane, washed
with water, dried (MgSO4) and purified on silica gel to give the
required alcohol which crystallized with one molecule of
dichloromethane for two molecules of alcohol (0.206 g,
89%), 1S-Me: mp 202 ЊC (dichloromethane–hexane); νOH/cmϪ1
(CCl4) 3499; δC (chloroform) 29.4 (6 CH), 37.1 (6 CH2), 39.2
(6 CH2), 46.4 (2 Cq), 57.5 (CH3), 87.2 (OH), 114.3 (C3), 119.4
(C5), 126.9 (C4), 131.2 (C6), 131.3 (C1) and 159.0 (C2);
δH (chloroform) 1.60–2.0 (br m, Ad), 3.87 (CH3), 6.58 (OH,
constant for 0.00013–0.13 M), 6.94 (H5, J 1.4, 7.2 and 8.1), 6.95
(H3, J 0.3, 1.4 and 8.3), 7.20 (H4, J 1.7, 7.2 and 8.3) and
7.57 (H6, J 0.3, 1.7 and 8.1); δH(OH) 6.74, 6.53 and 6.80 ppm
in C6D6, DMSO-d6 and C5D5N, respectively; ∆δ/δT = Ϫ1.60
0.04 ppb ЊCϪ1 in the last solvent (Found: C, 76.5; H, 8.9; Cl,
7.2. C28H38O2ؒ0.5CH2Cl2 requires C, 76.23; H, 8.75; Cl,
7.89%).
Conclusion
Intramolecularly hydrogen-bonded alcohols are obtained by
ortho-lithiation of alkoxybenzenes followed by reaction with
di(1-adamantyl) ketone. When a carbocation is formed from
the methoxy derivative, 1S-Me, in TFA–dichloromethane, it
is rapidly converted to a carboxonium ion by intramolecular
1,5-hydride transfer and thence to a trifluoroacetate and a
phenoxymethanol derivative. If a hydrosilane is added in order
to trap either or both of the cations by intermolecular hydride
transfer, trifluoroacetate formation remains of major import-
ance. The product data show that intra- and intermolecular
hydride transfer proceed at similar rates. The variation of the
product distribution with the nature and the concentration of
the hydrosilane can be understood in terms of a partition of the
carbocation and the carboxonium ion between pathways which
are [R3SiH]-dependent or not. The anti- and syn-(2-anisyl)-
diadamantylmethanes, 7A-Me and 7S-Me, are formed
exclusively by reduction of the carbocation and the carboxo-
nium ion, respectively. Kinetic deuterium isotope effects on the
reaction of the carbocation are normal, but the behaviour of
the carboxonium ion appears inconsistent. Certain aspects of
the reactivity of the carboxonium ion are anomalous; this may
be due to the relative imprecision of the data or, possibly,
the effects of organosilane species arising from reactions of
the silylium ion. Intramolecular hydride transfer is much
faster for the (2-ethoxyphenyl) analogue, 1S-Et, and neither tri-
fluoroacetate nor anti-(2-ethoxyphenyl)diadamantylmethane is
formed.
An alternative procedure was to run the lithiation in THF (10
cm3), adding the ketone (0.30 g, 1.0 mmol) in THF (15 cm3)
after about 25 min. The reaction was complete in less than 1 h,
and chromatography on alumina gave the alcohol (free of
dichloromethane) in 91% yield, mp 202.5 ЊC.
syn-(2-Ethoxyphenyl)di(1-adamantyl)methanol, 1S-Et. As for
1S-Me, reaction of phenetole with n-butyllithium in THF in the
presence of TMEDA, followed by reaction with diadamantyl
ketone, gave the required alcohol (73%): mp 194 ЊC (hexane);
νOH/cmϪ1 (CCl4) 3480; δC (chloroform) 15.1 (CH3), 29.4 (6 CH),
37.2 (6 CH2), 39.3 (6 CH2), 46.4 (2 Cq), 66.7 (CH2), 87.4 (OH),
115.3 (C3), 119.4 (C5), 126.9 (C4), 131.4 (C6), 131.5 (C1) and
158.3 (C2); δH (chloroform) 1.46 (CH3, J 7.0), 1.60–2.0 (br m,
Ad), 4.11 (CH2, J 7.0), 6.79 (OH), 6.92 (H5, J 1.4, 7.2 and 8.1),
6.93 (H3, J 0.3, 1.4 and 8.3), 7.17 (H4, J 1.7, 7.2 and 8.3) and
7.57 (H6, J 0.3, 1.7 and 8.1); δH(OH) 6.99, 6.72 and 7.05 ppm
in C6D6, DMSO-d6 and C5D5N, respectively; ∆δ/δT = Ϫ1.59
0.02 ppb ЊCϪ1 in the last solvent (Found: C, 82.7; H, 9.8.
C29H40O2 requires C, 82.81; H, 9.58%).
Experimental
General methods
NMR measurements were performed on a Bruker AS 200
FT instrument operating at 200 MHz (proton) or 50 MHz
(carbon). Chemical shifts are given in ppm and J values in Hz.
Measurements were made in hexadeuteriobenzene, deuterio-
chloroform, pentadeuteriopyridine, hexadeuteriodimethyl
sulfoxide or dichlorodideuteriomethane (reference values:
δH = 7.16, 7.26, 8.71, 2.50 and 5.32 ppm for 1H; δC = 128.0, 77.0,
149.9, 39.5 and 53.8 ppm). Carbon and hydrogen shifts of
the aromatic system are numbered: C2, C3, etc. Generally, the
proton signals were assigned on the basis of shifts, coupling
constants34 and spectrum simulation by the gNMR program
(Cherwell Scientific).35 The corresponding 13C signals were
identified by heteronuclear correlation experiments. IR spectra
were measured in carbon tetrachloride on a Nicolet Magna 860
FTIR spectrometer with 1 cmϪ1 resolution. GC/MS measure-
ments were performed on a CP-Sil 5 capillary column coupled
to a Finnigan MAT ITD 800B Ion Trap Detector with chemical
ionization (isobutane). Gas chromatography was performed on
a 30 cm 10% SE30 on Chrompack column. Column chrom-
atography was performed on silica gel 60 (Merck) in light
petroleum (boiling range 35–60 ЊC)–dichloromethane mixtures
or on alumina (Merck, Brockmann III) in light petroleum–
diethyl ether mixtures. Melting points were determined in capil-
lary glass tubes on a Mettler FP5 instrument with a heating rate
Reactions with TFA
syn-(2-Anisyl)di(1-adamantyl)methanol, 1S-Me. NMR study.
To the alcohol (12 mg, 0.03 mmol) and deuteriated dichloro-
methane (0.5 cm3) in an NMR tube was added TFA (0.02 cm3).
The mixture was briefly shaken and the tube replaced in the
1
NMR apparatus at 25 ЊC. The H spectrum was recorded at
intervals over a period of about 1 h, corresponding to about 3
half-lives of the reaction. Signals at 6.00 and 2.86 ppm, shown
by correlation with the 13C spectrum to correspond to CH2 and
benzylic CH groups, respectively, appeared progressively at the
expense of the methyl signal (3.94 ppm) of the alcohol. These
new signals were attributed to a trifluoroacetate, 2. An “infin-
ity” sample taken after ca. 10 half-lives showed 90% conversion
to the trifluoroacetate, 2, and about 10% of a material revealed
by small signals at 5.46 (CH2) and 2.41 (CH) ppm, correlated
with 13C signals at 67.7 and 58.8 ppm, respectively. Only three
aromatic CH carbons could be located (117.8, 128.7 and 133.3
ppm) and two quaternary carbons (129.7 and 153.4 ppm). This
product was tentatively identified as syn-2-[di(1-adamantyl)-
methyl]phenoxymethanol, 3.
A first-order rate constant was estimated by integration of
the CH2 and CH3 signals, the solvent CHDCl2 peak being
used as an internal standard. Plotting log (CH3)t or log
[(CH2)∞ Ϫ (CH2)t] vs. time (t) gave a good straight line over
two-half lives corresponding to a first-order rate constant of
7.15 0.14 × 10Ϫ4 sϪ1 (4 runs). Prior addition of trifluoroacetic
anhydride (0.02 cm3) to the reaction mixture gave an auto-
of 3 ЊC minϪ1
.
Alcohol synthesis
syn-(2-Anisyl)di(1-adamantyl)methanol, 1S-Me. To a solution
of anisole (0.54 cm3, 0.54 g, 5 mmol) and TMEDA (0.75 cm3,
5 mmol) in sodium–dry diethyl ether (15 cm3) stirred at room
J. Chem. Soc., Perkin Trans. 2, 1999, 1639–1648
1645

View Article Online
catalytic reaction with an inital rate constant (one half-life) of
7.2 0.5 sϪ1
anti-(2-Anisyl)di(1-adamantyl)methane, 7A-Me: mp 148 ЊC
.
(crystallized from a mixture of 7A-Me and 7S-Me on standing
for several days, pentane-washed); δC (chloroform) 29.6 (6 CH),
37.2 (6 CH2), 38.9 (2 Cq), 43.5 (6 CH2), 53.8 (CH3), 72.3 (CH),
110.7 (C3), 119.3 (C5), 126.9 (C4), 131.1 (C1), 136.6 (C6) and
157.9 (C2); δH (chloroform) 1.5–2.0 (br m, Ad), 2.05 (CH), 3.80
(CH3), 6.81 (H5, J 1.2, 7.3 and 7.4), 6.87 (H3, J 0.6, 1.2 and
8.2), 6.97 (H6, J 0.6, 1.8 and 7.4) and 7.19 (H4, J 1.8, 7.3 and
8.2) (Found: C, 86.2; H, 9.9. C28H38O requires C, 86.10; H,
9.81%).
syn-(2-Anisyl)di(1-adamantyl)methane, 7S-Me: mp 200 ЊC
(crystallized from another mixture of 7A-Me and 7S-Me,
pentane-washed); δC (chloroform) 29.3 (6 CH), 37.1 (6 CH2),
39.1 (2 Cq), 42.9 (6 CH2), 55.3 (CH), 55.7 (CH3), 110.4 (C3),
118.7 (C5), 125.9 (C4), 131.1 (C6), 131.4 (C1) and 158.5 (C2);
δH (chloroform) 1.5–2.0 (br m, Ad), 2.97 (CH), 3.80 (CH3),
6.86 (H3, J 0.5, 1.1 and 8.1), 6.89 (H5, J 1.1, 7.4 and 7.7),
7.17 (H4, J 1.7, 7.4 and 8.1) and 7.35 (H6, J 0.5, 1.7 and
7.7) (Found: C, 86.1; H, 9.6. C28H38O requires C, 86.10; H,
9.81%).
Aqueous work-up. Alcohol 1S-Me (100 mg, 0.25 mmol) was
treated with TFA (0.5 cm3) in dichloromethane (10 cm3) at
room temperature for 3 h and then quenched in a mixture of
water and pentane. The organic layer was twice washed with
water, then dried (MgSO4) and the solvents evaporated. NMR
analysis of the residue indicated a 9:1 mixture of trifluoro-
acetate, 2, and a phenol, 4. Chromatography on silica gave the
phenol (85 mg; 82%). Attempts to crystallize the trifluoro-
acetate were unsuccessful.
syn-[2-(α-Trifluoroacetylanisyl)]di(1-adamantyl)methane, 2:
δC (chloroform) 29.2 (6 CH), 37.0 (6 CH2), 39.1 (2 Cq), 42.9 (6
CH2), 56.1 (CH), 89.4 (CH2), 113.4 (C3), 114.3 (CF3, J 286 Hz),
121.9 (C5), 126.5 (C4), 131.9 (C6), 132.6 (C1), 155.3 (C2) and
156.6 (C᎐O, J 43 Hz); δH (chloroform) 1.5–2.0 (br m, Ad), 2.84
᎐
(CH), 6.00 (CH2), 7.04 (H3, J 0.4, 1.2 and 8.2), 7.06 (H5, J 1.2,
7.4 and 7.8), 7.20 (H4, J 1.7, 7.4 and 8.2 ) and 7.42 (H6, J 0.4,
1.7 and 7.8); m/z (ITD) 501 (M Ϫ 1, 4%), 389, 387, 366, 253,
251, 136, 135 (100%), 107, 93, 79.
syn-(2-Hydroxyphenyl)di(1-adamantyl)methane, 4: mp
223 ЊC (dichloromethane–hexane); νOH/cmϪ1 (CCl4) 3609;
δC (chloroform) 29.2 (6 CH), 37.0 (6 CH2), 39.4 (2 Cq), 43.0
(6 CH2), 57.2 (CH), 115.0 (C3), 119.1 (C5), 126.1 (C4), 128.6
(C1), 131.4 (C6) and 154.1 (C2); δH (chloroform) 1.5–2.0 (br m,
Ad), 2.57 (CH), 4.64 (OH), 6.76 (H3, J 0.1, 1.3 and 8.0), 6.86
(H5, J 1.3, 7.3 and 7.8), 7.06 (H4, J 1.7, 7.3 and 8.0) and 7.33
(H6, J 0.1, 1.7 and 7.8); m/z (ITD) 376 (M, 7%), 375, 240, 239,
136, 135 (100%), 107, 93, 79 (Found: C, 79.2; H, 8.8; Cl, 8.0.
C27H36Oؒ0.5CH2Cl2 requires C, 78.82; H, 8.90; Cl, 8.46%).
Small-scale experiments. To a stirred solution of alcohol
1S-Me (40 mg, 0.1 mmol) and a hydrosilane (0.15–2.4 mmol)
in dichloromethane (5 cm3) at room temperature (20 2 ЊC)
was added TFA (0.2 cm3). After 3 h water and pentane were
added, the pentane extract washed twice with water, then dried
(MgSO4) and the solvent evaporated under reduced pressure.
The residue was taken up in deuteriochloroform for 13C NMR
analysis, the aromatic and benzylic CH signals being used to
estimate the relative yields ( 2%, i.e. x% means x 2% except
for the smallest values where the uncertainty is 1%) of the
four products identified above. No significant amount of any
other material was detected. The hydrosilane concentrations
were calculated on the assumption that dichloromethane, TFA
and hydrosilane volumes are additive; values given in Tables are
for one half-life of the alcohol, based on the calculated uptake
of hydrosilane to give 7A-Me and 7S-Me.
Reactions with TFA
syn-(2-Ethoxyphenyl)di(1-adamantyl)methanol, 1S-Et. NMR
study. As for 1S-Me, to the alcohol (12 mg, 0.03 mmol) and
deuteriated dichloromethane (0.5 cm3) in an NMR tube was
added TFA (0.02 cm3). The mixture was briefly shaken and the
1
tube replaced in the NMR apparatus at 25 ЊC. The H NMR
syn-(2-Ethoxyphenyl)di(1-adamantyl)methanol, 1S-Et. Pre-
parative experiment. Treatment of alcohol 1S-Et (114 mg, 0. 27
mmol) in dichloromethane (10 cm3) with TES (0.15 cm3, 0.7
mmol) and TFA (0.5 cm3) at room temperature for 1 h gave syn-
(2-ethoxyphenyl)di(1-adamantyl)methane, 7S-Et, together with
a smaller amount (ca. 40%) of phenol, 4. No other products
were detected. Chromatographic separation gave 7S-Et (60 mg,
55%) and 4 (29 mg, 28%).
spectrum recorded over about 30 min showed progressive dis-
appearance of the ethyl group signals at 1.44 and 4.10 ppm and
modifications in the aromatic part of the spectrum. A well
defined doublet (J = 3.0 Hz) at 2.34 ppm coupled with a quad-
ruplet at 9.75 ppm was attributed to the formation of acetalde-
hyde. The 13C NMR spectrum of the mixture, taken 5–15 hours
after mixing, was identical with that of phenol, 4, under the
same conditions.
syn-(2-Ethoxyphenyl)di(1-adamantyl)methane, 7S-Et: mp
171 ЊC (hexane); δC (chloroform) 15.2 (CH3), 29.3 (6 CH), 37.1
(6 CH2), 39.0 (2 Cq), 42.9 (6 CH2), 55.2 (CH), 63.6 (CH2), 111.2
(C3), 118.4 (C5), 125.8 (C4), 131.1 (C6), 131.5 (C1) and 157.8
(C2); δH (chloroform) 1.43 (CH3, J 7.0), 1.5–2.0 (br m, Ad), 3.03
(CH), 4.01 (CH2, J 7.0), 6.84 (H3, J 0.4, 1.2 and 8.1), 6.87 (H5,
J 1.2, 7.3 and 7.7), 7.14 (H4, J 1.7, 7.3 and 8.1) and 7.35 (H6,
J 0.4, 1.7 and 7.7) (Found: C, 86.0; H, 9.9. C29H40O requires C,
86.08; H, 9.96%).
Isotope effects. Triethylsilane-d and dimethylphenylsilane-d
were prepared by the reduction of the corresponding silyl chlor-
ides by lithium aluminium deuteride in diethyl ether at reflux.36
Small-scale experiments were carried out as described above.
The only important changes in the NMR spectra were as
follows: for anti-(2-anisyl)diadamantylmethane, 7A-Me, loss
of benzylic proton signal and replacement of corresponding
carbon signal by a weak triplet at 71.7 ppm (J = 18 Hz); for syn-
(2-anisyl)diadamantylmethane, 7S-Me, replacement of methyl
signals by triplets slightly upfield from the usual values: 55.4
ppm (J = 22 Hz) for carbon and 3.77 ppm (J = 1.5 Hz) for
proton; 7S-Et, replacement of CH2 carbon signals by triplet
slightly upfield from the usual value: 63.3 ppm (J = 22 Hz). The
CHD proton signal is a slightly broadened quartet at 3.98
ppm (J = 6.9 Hz) and the CH3 signal a doublet at 1.41 ppm
(J = 6.9 Hz).
Ionic hydrogenation
syn-(2-Anisyl)di(1-adamantyl)methanol, 1S-Me. Preparative
experiments. (i) Treatment of alcohol 1S-Me (150 mg, 0.37
mmol) in dichloromethane (15 cm3) with TES (0.2 cm3, 1.25
mmol) and TFA (0.75 cm3) at room temperature for 1 h gave a
4:4:1:1 four-component mixture. The major components were
identified as syn-(2-anisyl)di(1-adamantyl)methane, 7S-Me,
and the trifluoroacetate, 2. The minor components, identified
and characterized in subsequent experiments, proved to be anti-
(2-anisyl)di(1-adamantyl)methane, 7A-Me, and the phenol, 4.
Separation of the less polar from the more polar materials by
chromatography on silica gel gave a mixture of 7A-Me and
7S-Me (74 mg, 49%) and the phenol, 4 (68 mg, 47%).
(ii) Treatment of alcohol 1S-Me (100 mg, 0.25 mmol) in
dichloromethane (10 cm3) with TES (1 cm3, 6.3 mmol) and TFA
(0.25 cm3) at room temperature for 40 h gave a mixture of 7S-
Me, 7A-Me and trifluoroacetate, 2, in a ratio of 4:4:1 with
traces of the phenol. Chromatographic separation gave a mix-
ture of 7A-Me and 7S-Me (78 mg, 81%), followed by phenol, 4
(9 mg, 10%). Heating the mixture of 7A-Me and 7S-Me in
chloroform for 8 h at 240 ЊC resulted in complete conversion to
7S-Me, confirming that this is the more stable, syn isomer. The
half-life was estimated to be 1.5 h at this temperature.
1646
J. Chem. Soc., Perkin Trans. 2, 1999, 1639–1648

View Article Online
5 J. S. Lomas and J. Vaissermann, J. Chem. Soc., Perkin Trans. 2, 1998,
1777.
Molecular mechanics calculations
Steric energies and geometries were calculated with Allinger’s
MM2(85) force field.18 The alcohol and deoxygenation product
rotamers give the following steric energies: 1S-Me, 63.17; 1A-
Me, 64.98; 7S-Me, 50.79; 7A-Me, 52.90 kcal molϪ1. To estimate
the steric energies of the phenol rotamers it was necessary to
introduce parameters for a type 2–6–21 bond angle; it was
taken as equivalent to type 1–6–21; values were 47.04 (4S)
and 49.46 (4A) kcal molϪ1. Inclusion of the phenolic oxygen in
the π-electron system reduced these values by about 0.1 kcal
6 A. G. Davies and A. G. Neville, J. Chem. Soc., Perkin Trans. 2, 1991,
2021.
7 G. A. Olah, G. K. S. Prakash and R. Krishnamurti, J. Am. Chem.
Soc., 1990, 112, 6422.
8 J. S. Lomas and J. E. Anderson, J. Org. Chem., 1995, 60, 3246.
9 J. S. Lomas, J. Chem. Soc., Perkin Trans. 2, 1996, 2601.
10 G. A. Olah, M. D. Heagy and G. K. S. Prakash, J. Org. Chem., 1993,
58, 4851; M. D. Heagy, G. A. Olah, G. K. S. Prakash and J. S.
Lomas, J. Org. Chem., 1995, 60, 7355.
11 J. S. Lomas, A. Adenier, C. Cordier and J. C. Lacroix, J. Chem. Soc.,
Perkin Trans. 2, 1998, 2647.
molϪ1
.
12 J. S. Lomas and J. Vaissermann, J. Chem. Soc., Perkin Trans. 2, 1996,
1831.
13 R. A. Rennels, A. J. Maliakal and D. B. Collum, J. Am. Chem. Soc.,
1998, 120, 421 and references therein.
Kinetic modelling
In KINAL29 the solution of kinetic differential equations is
based on a fourth-order semi-implicit Runge–Kutta method. In
our hands it is associated with a Simplex routine, making it
possible to optimize a number of rate constants by matching
experimental yields (“targets”) with those calculated. Since the
number of targets is almost invariably smaller than the number
of rate constants in the system, certain rate constants are given
arbitrary values. Normally, the relative rate constants are
insensitive to these values. For calculations involving water it
was assumed that the initial concentration of water in dichloro-
methane was the upper limit indicated for this reagent (ACS
grade, ≤0.2%).
14 D. N. Kursanov, Z. N. Parnes and N. M. Loim, Synthesis, 1974, 633;
I. Fleming, Organic Silicon Chemistry in Comprehensive Organic
Chemistry, ed. D. Barton and W. D. Ollis, Pergamon Press, Oxford,
1979, vol. 3, p. 541; Y. Nagai, Org. Prep. Proced. Int., 1980, 12, 13;
W. P. Weber, Silicon Reagents for Organic Synthesis, Springer-
Verlag, Berlin-Heidelberg, 1983, p. 273; E. W. Colvin, Silicon
Reagents in Organic Synthesis, Academic Press, London, 1988;
E. Keinan, Pure Appl. Chem., 1989, 61, 1737; G. L. Larson
in The Chemistry of Organosilicon Compounds, ed. S. Patai and
Z. Rappoport, Wiley Interscience, Chichester, 1989, Part 1, ch. 11,
p. 776.
15 J. S. Lomas and J. Vaissermann, J. Chem. Soc., Perkin Trans. 2, 1997,
2589.
16 H. W. Gschwend and H. R. Rodriguez, Org. React. (N.Y.), 1979, 26,
1.
X-Ray crystallography
17 M. Ito and M. Hirota, Bull. Soc. Chem. Jpn., 1981, 54, 2093.
18 N. L. Allinger, Quantum Chemistry Program Exchange, Program
MMP2(85), Indiana University, USA.
anti-(2-Anisyl)di(1-adamantylmethane), 7A-Me: C28H38O.
Crystal data.¶ M = 390.6. Monoclinic a = 10.945(8), b =
25.069(11), c = 16.530(10) Å, β = 104.71(5), V = 4387(5) Å3 (by
least squares refinement on diffractometer angles for 25 auto-
matically centred reflections, λ = 0.71069 Å), space group P21/n,
Z = 8, Dx = 1.18 g cmϪ3. The asymmetric unit contains two
independent molecules. Colourless prismatic crystals, µ(Mo-
19 J. Vaissermann, unpublished results.
20 R. Taylor and O. Kennard, Acc. Chem. Res., 1984, 17, 320.
21 T. Ohishi, J. Yamada, Y. Inui, T. Sakaguchi and M. Yamashita,
J. Org. Chem., 1994, 59, 7521.
22 K. Pihlaja and A. Lampi, Acta. Chem. Scand., Ser. B, 1986, 40, 196.
23 (a) H. Perst, Oxonium Ions in Organic Chemistry, Verlag Chemie,
Weinheim, 1971; (b) G. A. Olah, Angew. Chem., Int. Ed. Engl., 1993,
32, 767 and references therein.
Kα) = 0.64 cmϪ1
.
24 J. L. Fry and G. J. Karabatsos, in Carbonium Ions, ed. G. A. Olah
and P. v. R. Schleyer, Wiley-Interscience, New York, vol. II, 1970,
ch. 14.
25 R. A. McClelland, Can. J. Chem., 1975, 53, 2763.
26 J. K. Pau, J. K. Kim and M. C. Caserio, J. Am. Chem. Soc., 1978,
100, 3838, 4242.
27 U. R. Ghatak, B. Sanyal, S. Ghosh, M. Sarkar, M. S. Raju and
E. Wenkert, J. Org. Chem., 1980, 45, 1081.
28 H. Mayr, N. Basso and G. Hagen, J. Am. Chem. Soc., 1992, 114,
3060.
Data collection and processing. Enraf-Nonius CAD4 diffract-
ometer, ω/2θ mode with ω scan width = 0.8 ϩ 0.345 tan θ,
graphite-monochromated Mo-Kα radiation. 8342 reflections
measured (1 ≤ θ ≤ 25Њ), 7699 unique, giving 3523 with I > 3σ(I).
Structure analysis and refinement. Full-matrix least-squares
refinement with all non-hydrogen atoms anisotropic; hydrogens
located from Fourier difference map with one, overall, refined
isotropic thermal parameter (524 refinable parameters). No
absorption correction. Final R and Rw (Chebyshev series)
values are 0.051 and 0.064. Program used is the PC version of
CRYSTALS37 for refinements and CAMERON38 for views.
29 T. Turányi, Comput. Chem., 1990, 14, 253. T. Turányi, New. J.
Chem., 1990, 14, 795.
30 For recent contributions to the silylium ion debate, see: S. H.
Strauss, Chemtracts: Inorg. Chem., 1993, 5, 119; J. B. Lambert,
S. Zhang and S. M. Ciro, Organometallics, 1994, 13, 2430; J. B.
Lambert and S. Zhang, Science, 1994, 263, 984; G. A. Olah,
G. Rasul, X. Li, H. A. Buchholz, G. Sandford and G. K. S. Prakash,
Science, 1994, 263, 983; L. Pauling, Science, 1994, 263, 983; C. A.
Reed and Z. Xie, Science, 1994, 263, 985; J. B. Lambert, L. Kania
and S. Zhang, Chem. Rev., 1995, 95, 1191; G. A. Olah, G. Rasul,
X. Y. Li and G. K. S. Prakash, Bull. Soc. Chim. Fr., 1995, 132, 569;
Z. Xie, R. Bau, A. Benesi and C. A. Reed, Organometallics, 1995, 14,
3933; L. Olsson, C. H. Ottosson and D. Cremer, J. Am. Chem. Soc.,
1995, 117, 7460; M. Arshadi, D. Johnels, U. Edlund, C. H. Ottosson
and D. Cremer, J. Am. Chem. Soc., 1996, 118, 5120; Z. Xie,
J. Manning, R. W. Reed, R. Mathur, P. D. W. Boyd, A. Benesi and
C. A. Reed, J. Am. Chem. Soc., 1996, 118, 2922; J. B. Lambert and
Y. Zhao, Angew. Chem., Int. Ed. Engl., 1997, 36, 400; H. U.
Steinberger, T. Müller, N. Auner, C. Maerker and P. v. R. Schleyer,
Angew. Chem., Int. Ed. Engl., 1997, 36, 626; P. v. R. Schleyer,
Science, 1997, 275, 39; J. Belzner, Angew. Chem., Int. Ed. Engl.,
1997, 36, 1277; C. A. Reed, Acc. Chem. Res., 1998, 31, 325.
31 J. Chojnowski, W. Fortuniak and W. Stanczyk, J. Am. Chem. Soc.,
1987, 109, 7776; J. Chojnowski and W. Stanczyk, Adv. Organomet.
Chem., 1990, 30, 243.
Acknowledgements
We are indebted to Ms Claude Charvy for the GC-MS studies,
to Dr A. Adenier for IR measurements and to Drs M. Hedaya-
tullah and J-M. El Hage Chahine for helpful discussions. We
thank Dr T. Turányi for access to KINAL and P. Levoir for help
in programming the Simplex version.
¶ CCDC reference number 188/171. See http://www.rsc.org/suppdata/
p2/1999/1639 for crystallographic files in .cif format.
References
1 R. C. Fort, Adamantane: the Chemistry of Diamond Molecules,
Marcel Dekker, New York, 1976.
2 J. S. Lomas and J. E. Dubois, Tetrahedron, 1981, 37, 2273.
3 (a) J. S. Lomas, P. K. Luong and J. E. Dubois, J. Org. Chem., 1977,
42, 3394; (b) J. S. Lomas and V. Bru-Capdeville, J. Chem. Soc.,
Perkin Trans. 2, 1994, 459.
4 J. S. Lomas and J. Vaissermann, Bull. Soc. Chim. Fr., 1996, 133, 25;
J. Vaissermann and J. S. Lomas, Acta Crystallogr., Sect. C, 1997, 53,
1341.
32 Y. Apeloig, O. Merin-Aharoni, D. Danovich, A. Ioffe and S. Shaik,
Isr. J. Chem., 1993, 33, 387.
33 H. Tashiro, K. Kikukawa, K. Ikenaga, N. Shimizu and M. Mishima,
J. Chem. Soc., Perkin Trans. 2, 1998, 2435.
J. Chem. Soc., Perkin Trans. 2, 1999, 1639–1648
1647

View Article Online
34 E. Pretsch, W. Simon, J. Seibl and T. Clerc, Tables of Spectral Data
for Structure Determination of Organic Compounds, Springer-
Verlag, Berlin-Heidelberg, 2nd Edn., 1989.
35 gNMR from Cherwell Scientific Publishing Limited, The Magdalen
Centre, Oxford Science Park, Oxford, UK OX4 4GA.
36 M. P. Doyle, C. C. McOsker and C. T. West, J. Org. Chem., 1976, 41,
1393.
User Guide, Chemical Crystallography Laboratory, University of
Oxford, 1988.
38 J. Pearce, D. J. Watkin and D. P. Prout, CAMERON. A Program
for Plotting Molecular Structures, Chemical Crystallography
Laboratory, University of Oxford, 1992.
Paper 9/03303C
37 D. J. Watkin, J. R. Carruthers and P. W. Bettridge, CRYSTALS.
1648
J. Chem. Soc., Perkin Trans. 2, 1999, 1639–1648