Generalized synthesis and physical properties of dialkoxy disulfides

Article abstract of DOI:10.1021/jo050574s

A substrate study was undertaken in order to probe the scope of S ₂Cl₂ coupling of alcohols to form dialkoxy disulfides. Compounds 1b and If are new; along with 1a, 1c, 1h, and 1j, all of the title compounds are fully characterized, and the yields of 1a and 1c have been optimized from previously reported syntheses. The effect of the R-substituent about the OSSO moiety has been carefully probed as yields vary. A substrate and a solvent study of the coalescence behavior of this class was carried out. The origin of the inherently large barrier to rotation and the resultant thermal decomposition pathway is discussed. Both phenomena are shown to be solvent independent; hindered rotation is substrate independent. The decomposition of 1a is ca. 7 kcal/mol higher than the barrier to rotation about the S-S bond. The combined evidence suggests acyclic unsymmetric homolytic cleavage of the dialkoxy disulfide.

Full text of DOI:10.1021/jo050574s

Generalized Synthesis and Physical Properties of Dialkoxy
Disulfides
Eli Zysman-Colman and David N. Harpp*
Department of Chemistry, McGill University, Montreal, Quebec, Canada H3A 2K6
david.harpp@mcgill.ca
Received March 21, 2005
A substrate study was undertaken in order to probe the scope of S2Cl2 coupling of alcohols to form
dialkoxy disulfides. Compounds 1b and 1f are new; along with 1a, 1c, 1h, and 1j, all of the title
compounds are fully characterized, and the yields of 1a and 1c have been optimized from previously
reported syntheses. The effect of the R-substituent about the OSSO moiety has been carefully probed
as yields vary. A substrate and a solvent study of the coalescence behavior of this class was carried
out. The origin of the inherently large barrier to rotation and the resultant thermal decomposition
pathway is discussed. Both phenomena are shown to be solvent independent; hindered rotation is
substrate independent. The decomposition of 1a is ca. 7 kcal/mol higher than the barrier to rotation
about the S-S bond. The combined evidence suggests acyclic unsymmetric homolytic cleavage of
the dialkoxy disulfide.
Molecules of the form ROSSOR have been known for
over a century.1 Their connectivity was partially eluci-
dated when Thompson and co-workers2,3 were able to
synthesize and characterize this unbranched polychal-
cogenide known as a dialkoxy disulfide 1. Since then, the
work conducted generally has been for a specific subset
of compounds4-14 or for theoretical investigations.15-18
Although compounds of form 2 have been character-
ized,19-22 the only known systems with this thionosulfite
moiety are cyclic.20-22 Compounds containing the OSSO
fragment 1 possibly isomerize into their valence-bond
isomeric thionosulfite form (OS(dS)O 2).23
(1) Lengfeld, F. Ber. 1895, 28, 449.
(2) Thompson, Q. E.; Crutchfield, M. M.; Dietrich, M. W.; Pierron,
E. J. Org. Chem. 1965, 30, 2692.
(3) Thompson, Q. E. Quart. Rep. Sulfur Chem. 1970, 5, 245.
(4) Seel, F.; Gombler, W.; Budenz, R. Liebigs Ann. Chem. 1970,
735, 1.
(15) Gleiter, R.; Hyla-Kryspin, I.; Schmidt, H.; Steudel, R. Chem.
Ber. 1993, 126, 2363.
(16) Koritsanszky, T.; Buschmann, J.; Schmidt, H.; Steudel, R. J.
Phys. Chem. 1994, 98, 5416.
(5) Kobayashi, M.; Minato, H.; Shimada, K. Int. J. Sulfur Chem., A
1971, 1, 105.
(6) Motoki, S.; Kagami, H. J. Org. Chem. 1977, 42, 4139.
(7) Motoki, S.; Kagami, H.; Hanzawa, T.; Suzuki, N.; Yamaguchi,
S.; Saito, M. Bull. Chem. Soc. Jpn. 1980, 53, 3658.
(8) Ritzel, R.; Wenschuh, E. Sulfur Lett. 1986, 4, 161.
(9) Tardif, S. L.; Williams, C. R.; Harpp, D. N. J. Am. Chem. Soc.
1995, 117, 9067.
(17) Steudel, R.; Schmidt, H.; Baumeister, E.; Oberhammer, H.;
Koritsanszky, T. J. Phys. Chem. 1995, 99, 8987.
(18) Snyder, J. P.; Nevins, N.; Tardif, S. L.; Harpp, D. N. J. Am.
Chem. Soc. 1997, 119, 12685.
(19) Thompson, Q. E.; Crutchfield, M. M.; Dietrich, M. W. J. Am.
Chem. Soc. 1964, 86, 3891.
(10) Borghi, R.; Lunazzi, L.; Placucci, G.; Cerioni, G.; Foresti, E.;
Plumitallo, A. J. Org. Chem. 1997, 62, 4924.
(20) Harpp, D. N.; Steliou, K.; Cheer, C. J. J. Chem. Soc., Chem.
Commun. 1980, 825.
(11) Cerioni, G.; Plumitallo, A. Org. Magn. Res. 1998, 36, 461.
(12) Smith, J. D.; O’Hair, R. A. J. Phosphorus, Sulfur Silicon Relat.
Elements 1997, 126, 257.
(21) Zysman-Colman, E.; Abrams, C. B.; Harpp, D. N. J. Org. Chem.
2003, 68, 7059.
(13) Braverman, S.; Pechenick, T. Tetrahedron Lett. 2002, 43, 499.
(14) Roeschenthaler, G. V. Phosphorus Sulfur 1978, 4, 373.
(22) Tanaka, S.; Sugihara, Y.; Sakamoto, A.; Ishii, A.; Nakayama,
J. J. Am. Chem. Soc. 2003, 125, 9024.
10.1021/jo050574s CCC: $30.25 © 2005 American Chemical Society
Published on Web 06/18/2005
5964
J. Org. Chem. 2005, 70, 5964-5973

Synthesis and Physical Properties of Dialkoxy Disulfides
SCHEME 1. Representation of the Original
Concerted Mechanism
SCHEME 2
mides,36,40,41 sulfenamides,42 acrylonitriles (DMAAN),43
and carbamates44,45 are due in part to resonance-induced
double-bond character in these systems.46 In biphe-
nyls47,48 and related compounds,49-52 however, they are
due to steric interactions about either an sp2-sp2 or an
sp2-sp3 carbon-carbon bond.53,54
TABLE 1. S-S Torsional Barrier of Some Related
Polychalcogens
compd
barriera
ref
Herein, we report a generalized method for the syn-
thesis of dialkoxy disulfides. We examine the diastero-
topic coalescence phenomenon as a function of the
substituent as well as the solvent. We also investigate
the nature of the thermal decomposition pathway. We
used the relatively stable bis(p-nitrobenzyloxy) disulfide
1a as a representative dialkoxy disulfide in most of these
studies.
MeS-SMe
6.8
14.5
18.4
55
56
10
MeOS-SN(Me)2
EtOS-SOEt
a In kcal/mol.
Most intriguingly, Thompson2 observed that upon
heating, 1 decomposed to afford the corresponding alde-
hyde, alcohol, and elemental sulfur. He hypothesized a
six-membered cyclic transition state, similar to that
shown in Scheme 1, to account for product formation,
though to our knowledge there is no experimental
evidence to support this mechanistic conclusion. We have
shown that diatomic sulfur can indeed be trapped by
dienes in good yield (61-79%) in what was similarly
suggested to be a thermal pseudopericyclic reaction; in
the absence of a diene trap, S2 condenses to the more
stable S8 allotrope.9
One of the defining and highly unusual properties of
1 is the exceptionally short S-S bond (ca. 1.91 Å)16,18,24
suggestive of a S-S rotational barrier that is much
higher (ca. 18 kcal/mol)4,10 than that of a normal disulfide
(ca. 7 kcal/mol). Restricted rotation in 1 appears to arise
entirely from electronic modulation of the S-S σ-bond.
Indeed, the degree of this electronic effect manifests itself
through electron-withdrawing elements immediately ad-
jacent to the S-S bond (Table 1). Restricted rotation
about single bonds25 is not usually influenced solely
through stereoelectronic interactions. For instance, well-
documented high torsional barriers in amides,26-39 thioa-
Results and Discussion
Synthesis. As part of our wider interests in the
physical properties of dialkoxy disulfides, we synthesized
and characterized several dialkoxy disulfides (derived
from the corresponding alcohols 3) according to a modi-
fication of the procedure used by Thompson2 (Scheme 2,
Table 2). Dilute conditions and freshly distilled sulfur
monochloride (S2Cl2) are key in attaining high yields and
purity. This is due to the photolytic instability of S2Cl2
over a wide range of wavelengths.57 Our current synthetic
method is effective, with addition times reduced by ca.
90% from earlier preparative methods.9 Longer reaction
(39) Duffy, E. M.; Severance, D. L.; Jorgensen, W. L. J. Am. Chem.
Soc. 1992, 114, 7535.
(40) Laidig, K. E.; Cameron, L. M. J. Am. Chem. Soc. 1996, 118,
1737.
(41) Wilson, N. K. J. Phys. Chem. 1971, 75, 1067.
(42) Raban, M.; Yamamoto, G. J. Am. Chem. Soc. 1979, 101, 5890.
(43) Rablen, P. R.; Miller, D. A.; Bullock, V. R.; Hutchinson, P. H.;
Gorman, J. A. J. Am. Chem. Soc. 1999, 121, 218.
(44) Cox, C.; Lectka, T. J. Org. Chem. 1998, 63, 2426.
(45) Rablen, P. R. J. Org. Chem. 2000, 65, 7930.
(46) Some have argued that the high observed barrier in amides is
due to the inherent preference for nitrogen planarity, resulting in its
effective increase in electronegativity and thus better ability to stabilize
itself. See for instance: Wiberg, K. B.; Breneman, C. M. J. Am. Chem.
Soc. 1992, 114, 831.
(47) Meyer, W. L.; Meyer, R. B. J. Am. Chem. Soc. 1963, 85, 2170.
(48) Grein, F. J. Phys. Chem. A. 2002, 106, 3823.
(49) Gust, D.; Mislow, K. J. Am. Chem. Soc. 1973, 95, 1535.
(50) Mislow, K.; Glass, M. A. W.; Hopps, H. B.; Simon, E.; Wahl, G.
H., Jr. J. Am. Chem. Soc. 1963, 86, 1710.
(51) Casarini, D.; Rosini, C.; Grilli, S.; Lunazzi, L.; Mazzanti, A. J.
Org. Chem. 2003, 68, 1815.
(52) Casarini, D.; Lunazzi, L.; Mazzanti, A. J. Org. Chem. 1997, 62,
3315.
(53) Sterically induced high torsional barriers are not limited to
rotation about carbon-carbon bonds. For instance, for rotation about
the sp2-sp2 carbon-nitrogen bond in N-aryl-tetrahydropyrimidines,
see: Beatriz Garcia, M.; Grilli, S.; Lunazzi, L.; Mazzanti, A.; Orelli, L.
R. J. Org. Chem. 2001, 66, 6679.
(54) For another example of a sterically induced high barrier to
rotation, i.e., rotation about the sp2-carbon-phosphorus bond in
secondary phosphine oxides, see: Gasparrini, F.; Lunazzi, L.; Mazzanti,
A.; Pierini, M.; Pietrusiewicz, K. M.; Villani, C. J. Am. Chem. Soc. 2000,
122, 4776.
(23) Harpp, D. N.; Nevins, N.; Snyder, J. P.; Zysman-Colman, E.
Manuscript in preparation.
(24) Baumeister, E.; Oberhammer, H.; Schmidt, H.; Steudel, R.
Heteroatom Chem. 1991, 2, 633.
(25) For a review on rotation about single bonds, see: Sternhell, S.
Dynamic Nuclear Magnetic Resonance Spectroscopy; Jackman, L. M.,
Cotton, F. A., Eds.; Academic Press: New York, 1975; Vol. 6.
(26) Schnur, D. M.; Yuh, Y. H.; Dalton, D. R. J. Org. Chem. 1989,
54, 3779.
(27) Li, P.; Chen, X. G.; Shulin, E.; Asher, S. A. J. Am. Chem. Soc.
1997, 119, 1116.
(28) Bushweller, C. H.; O’Neil, J. W.; Halford, M. H.; Bissett, F. H.
J. Am. Chem. Soc. 1971, 93, 1471.
(29) Scherer, G.; Kramer, M. L.; Schutkoxski, M.; Reimer, U.; Fisher,
G. J. Am. Chem. Soc. 1998, 120, 5568.
(30) Suarez, C.; Tafazzoli, M.; True, N. S.; Gerrard, S.; Lemaster,
C. B.; Lemaster, C. L. J. Phys. Chem. 1995, 99, 8170.
(31) Suarez, C.; Lemaster, C. B.; Lemaster, C. L.; Tafazzoli, M.; True,
N. S. J. Phys. Chem. 1990, 94, 6679.
(32) Stewart, W. E.; Siddall, T. H., III. Chem. Rev. 1970, 70, 517.
(33) Ross, B. D.; True, N. S.; Decker, D. L. J. Phys. Chem. 1983, 87,
89.
(34) Drakenberg, T.; Dahlqvist, K.-I.; Forse´n, S. J. Phys. Chem.
1972, 76, 2178.
(55) Hubbard, W. N.; Douslin, D. R.; McCullough, J. P.; Scott, D.
W.; Todd, S. S.; Messerly, J. F.; Hossenlopp, I. A.; George, A.;
Waddington, G. J. Am. Chem. Soc. 1958, 80, 3547.
(56) Cerioni, G.; Cremonini, M. A.; Lunazzi, L.; Placucci, G.; Plu-
mitallo, A. J. Org. Chem. 1998, 63, 3933.
(57) Einfeld, T. S.; Maul, C.; Gericke, K.-H. J. Chem. Phys. 2002,
117, 4214 and references therein.
(35) Wiberg, K. B.; Rablen, P. R.; Rush, D. J.; Keith, T. A. J. Am.
Chem. Soc. 1995, 117, 4261.
(36) Taha, A. N.; Neugebauer-Crawford, S. M.; True, N. S. J. Phys.
Chem. A 1998, 102, 1425.
(37) Taha, A. N.; True, N. S. J. Phys. Chem. A 2000, 104, 2985.
(38) Gao, J. J. Am. Chem. Soc. 1993, 115, 2930.
J. Org. Chem, Vol. 70, No. 15, 2005 5965

Zysman-Colman and Harpp
TABLE 2. Yields of Dialkoxy Disulfides
TABLE 3. Chemical Shift and Coupling Constant Data
for Related ROSSORa
entry
compd
JAB
νz
δ∆ν/JAB
R
1
2
3
4
1a
2d
1b
1c
12.40
11.25
11.50
11.25
4.94
4.84
4.82
4.77
3.84
4.67
4.81
4.54
p-NO2-Bn
Bn
p-tert-butyl-Bn
p-MeO-Bn
a JAB in Hz; νz in ppm.
Evaluation of Rate Parameters. Although Thomp-
son first concluded that the barrier to rotation was close
to that of disulfides (Ea ) 8.6 ( 1.7 kcal/mol),2 such a
low value would require an unexpectedly62 large negative
∆Sq. Subsequent work has shown that the reported value
(8.6 kcal/mol) is erroneous.4
Seel4 demonstrated that such a barrier for MeOSSOMe
1m was much higher (∆Gq ) 17.8 ( 0.1 kcal/mol).
Lunazzi and co-workers10 determined the thermodynamic
properties for 1a in perchloroethene (C2Cl4) at 105 °C
(∆Gq ) 19.0 ( 0.2 kcal/mol, ∆Hq ) 20 ( 1 kcal/mol,
∆Sq ) 2 ( 5 eu).
We were first interested in determining whether the
barrier height could be modulated by altering the sub-
stituent R group of a dialkoxy disulfide. A substituent
study (Table 3) of compounds 1a-d reveals a modest but
measurable electronic effect on the benzyl proton signals.
As the para-substituent is altered from an electron-
donating to an electron-withdrawing group, the coupling
constant, JAB, increases and the chemical shift of the AB-
quartet, νz, moves downfield. These two parameters do
not change uniformly as noted in their relative ratio
(which is a marker of the magnetic environment of the
diastereotopic benzyl protons).
Studies of the mutual-site exchange kinetics in DMSO-
d6 provide the torsional barrier about the S-S bond in
1a-c. Complete line-shape analysis (LSA) of the exchange-
broadened benzyl signals was accomplished by fitting the
experimental data with the computer program WIND-
NMR63 (Table 4).64 Standard activation parameters were
obtained from the linear least-squares fit of the experi-
mental rate data to both the Eyring and Arrhenius
equations assuming a transmission coefficient of unity.
The Pearson regression factor (R2) for these fits was
greater than 0.98 as shown in Figure 1. The error limits
in Table 4 assume only random errors. It is unclear why
1a decomposed in previous studies9,18 though the
cleavage of dialkoxy disulfides is both acid-2,5 and base-
sensitive.2,65-67
a Complex mixture of products. b Only starting material de-
tected.
times were sometimes necessary (5 h instead of 3 h), in
particular for 1b-d, wherein the para-substituted group
was not electron-withdrawing. Initial purification at-
tempts with 1c resulted in lower yields, as this product
was unstable to chromatographic conditions.9 When R )
trityl, as in 1g, only starting alcohol was obtained.
This series represents a varied substrate study in the
synthesis of dialkoxy disulfides, although Thompson
prepared several aliphatic examples in his original
paper.2 The coupling of p-cresol to form dialkoxy disulfide
1k and 4-N,N-dimethylaminobenzyl alcohol to form di-
alkoxy disulfide 1e proved to be unsuccessful. Interest-
ingly, only one aromatic dialkoxy disulfide has ever been
reported (R ) 2,2′-diaminophenoxy).58,59 Here, there are
strong electron-donating groups in the ortho positions
that apparently increase the nucleophilicity of the phe-
nolic oxygens enough to drive the reaction forward. It is
unclear why 1e could not be prepared given that dialkoxy
disulfides with electron-donating groups such as 1c can
be synthesized and are stable. During the reaction with
1k, the mixture turns bright green. This is in contrast
to the yellow solutions that are usually observed. Direct
electrophilic substitution of sulfur monochloride is not
unprecedented,58-60 and the benzene ring may act as a
competing nucleophile in this particular case.
Compounds 1a-d, 1f, and 1h-i were conveniently
stored at -10 °C for months with only minor decomposi-
tion. Compound 1j decomposed to a brown solid upon
reduced-pressure solvent removal but could be stored in
CH2Cl2 for weeks. Braverman and co-workers also ob-
served the same solution stability.61 All the dialkoxy
disulfides synthesized possess a sweet, fruity aroma.
It should be noted that the synthesis of 1a, 1c, 1d, and
1h were all optimized. The preparation of 1b and 1f had
previously not been reported. Compound 1j was only
recently reported.61
Clearly the electronic effect shown in Table 3 did not
affect the barrier to rotation (see ∆Gq298, Table 4). Thus,
the high barrier in dialkoxy disulfides is a function not
of the electronics of the R group but intrinsically that of
the OS-SO moiety.
Compound 1a is organic-soluble, crystalline, stable,
and nicely suited for the evaluation of the influence of
solvent polarity on the torsional barrier. The MP2/6-
(62) Fraser, R. R.; Boussard, G.; Saunders: J. K.; Lambert, J. B. J.
Am. Chem. Soc. 1971, 93, 3822.
(58) Su, Y. Z.; Ma, W. S.; Gong, K. C. Mater. Res. Soc. Symp. Proc.
2000, 575, 403.
(63) WINDNMR, 7.1.5 ed.: Reich, H. J. J. Chem. Educ. Software
1996.
(59) Su, Y. Z.; Gong, K. C. Mater. Res. Soc. Symp. Proc. 2000, 600,
215.
(64) For an excellent reference on DNMR spectroscopy, see: Sand-
stro¨m, J. Dynamic NMR Spectroscopy; Academic Press: Toronto, 1982.
(65) Meuwsen, A.; Gebhardt, H. Ber. 1936, 69, 937.
(66) Meuwsen, A.; Gebhardt, H. Ber. 1935, 68, 1011.
(67) Meuwsen, A. Ber. 1936, 69, 935.
(60) Yamomoto, K.; Jikei, M.; Miyatake, K.; Katoh, J.; Nishide, H.;
Tsuchida, E. Macromolecules 1994, 27, 4312.
(61) Braverman, S.; Pechenick, T.; Gottlieb, H. E. Tetrahedron Lett.
2003, 44, 777.
5966 J. Org. Chem., Vol. 70, No. 15, 2005

Synthesis and Physical Properties of Dialkoxy Disulfides
TABLE 4. Substrate Study of the Activation Parameters for 1a-c
activation parametersa
b
∆Hq
(kcal/mol)
∆Sq
(eu)
∆Gq
∆Gq
EA
(kcal/mol)
log A
(s-1
298
(kcal/mol)
Tc
compd
(kcal/mol)
)
Tc (K) ( 0.5
1a
1b
1c
12.9 ( 0.9
16.6 ( 1.0
13.7 ( 0.6
-14.9 ( 2.7
-4.9 ( 2.9
-12.4 ( 1.8
17.4 ( 1.2
18.1 ( 1.3
17.4 ( 0.8
18.3 ( 1.3
18.4 ( 1.4
18.2 ( 0.9
13.6 ( 0.9
17.2 ( 1.0
14.3 ( 0.6
23.1 ( 1.4
28.0 ( 1.5
24.2 ( 0.9
360.7
362.7
362.7
a Errors, as given here, represent a 68% confidence interval in the least squares deviation calculation. b ∆Gq determined at the Tc for
each compound.
FIGURE 1. Eyring plot of the rotation about the S-S bond
in 1a in DMSO-d6.
311G(3d)68-70 dipole moments of the optimized gauche
ground state and the trans transition state for
MeOSSOMe 1m (2.4 and 0 D, respectively) suggest a
significant difference potentially responsive to a substan-
tial variation in solvent polarity. Examination of the
torsional potential of 1a in different solvents comple-
ments the work by Lunazzi10 and co-workers, who evalu-
ated simple dialkoxy disulfides revealing little or no
influence on barrier height with respect to the origin of
the R group.
As shown in Figure 2, the two doublets of the AB
system eventually coalesce into a single line since fast
rotation about the S-S creates a dynamic plane of
symmetry that makes the benzyl protons enantiotopic.
The Pearson regression factor (R2) for the linear fits to
FIGURE 2. Temperature dependence of the benzyl CH2 signal
the Eyring and Arrhenius equations ranged from 0.96
of 1a (500 MHz in DMF-d7). Superimposed on each spectrum
to 0.99, indicating that the simulations are in good
is the display of each computer simulation obtained with the
agreement with the obtained spectra. The free energies
of activation scaled to a common temperature (298 K)
for six solvents with empirical solvent polarity parameter
(ET) values71 ranging from 33 to 45 exhibit no apparent
rate constants (in s-1) indicated. Above each spectrum is a
difference spectrum indicating the goodness of fit of the
simulation.
medium effect (Table 5). The spread of ∆Gq
is a
298
the dielectric constant (Figures 3 and 4). The lack of
solvent correlation with the Onsager reaction field model
is at first surprising, as energy differences74 in gauche
and trans conformers and the rotational barriers75 of 1,2-
dichloroethane are well correlated. This lack in correla-
tion may be due to the limited choice of solvents, most of
which are aromatic; solvents must be relatively high-
boiling for this study.
diminutive 0.7 kcal/mol that is essentially flat with
respect to solvent polarity as illustrated by plots of
activation free energy against the empirical solvent
polarity parameter, ET, and the Onsager72,73 dielectric
constant function, defined as (ꢀ - 1)/(2ꢀ + 1), where ꢀ is
(68) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;
Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T.; Petersson,
G. A.; Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.;
Zakrzewski, V. G.; Ortiz, J. V.; Foresman, J. B.; Cioslowski, J.;
Stefanov, B. B.; Nanayakkara, A.; Challacombe, M.; Peng, C. Y.; Ayala,
P. Y.; Chen, W.; Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts,
R.; Martin, R. L.; Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.;
Stewart, J. P.; Head-Gordon, M.; Gonzalez, C.; Pople, J. A. Gaussian
94, revision D.4; Gaussian, Inc.: Pittsburgh, PA, 1995.
(69) Møller, C.; Plesset, M. S. Phys. Rev. 1934, 46, 618.
(70) Head-Gordon, M.; Pople, J. A.; Frisch, M. J. Chem. Phys. Lett.
1988, 153, 503.
π
kc )
∆ν2 + 6J
(1)
(2)
2
x
AB
x
2
kB
kc
∆Gq ) RT ln
- ln
( )
c
( )
[
]
h
Tc
(71) Reichardt, C. Chem. Rev. 1994, 94, 2319.
(72) Onsager, L. J. Am. Chem. Soc. 1936, 58, 1486.
(74) Wiberg, K. B.; Murcko, M. A. J. Phys. Chem. 1987, 91, 3616.
(75) Wong, M. W.; Frisch, M. J.; Wiberg, K. B. J. Am. Chem. Soc.
1991, 113, 4776.
(73) The Handbook of Chemistry and Physics, 3rd ed.; CRC Press:
Boca Raton, FL, 2001.
J. Org. Chem, Vol. 70, No. 15, 2005 5967

Zysman-Colman and Harpp
TABLE 5. NMR-Derived Activation Parameters for Coalescence of the Diastereotopic Methylene Protons of 1a
Obtained by Line Shape Analysis; S-S Bond Rotation, 298 Ka
c
∆Hq
(kcal/mol)
∆Sq
(eu)
∆Gq
Ea
(kcal/mol) (kcal/mol)
log A
(s-1
ET(30)b
JAB
(Hz)
∆νd
298
solvent
)
kc (s-1) ( 10 (kcal/mol)
(Hz) Tc (K) ( 0.5
e
C2Cl4
20.0 ( 1
2 ( 5
32.1
-12.5 37.5
-12.5 75.0
-13.0 54.5
-12.3 69.7
-12.7 67.6
-12.5 45.8
-13.0 39.9
378.2
383.2
365.1
378.6
394.7
381.3
360.7
p-xylene
pyridine
toluene
8.6 ( 0.5 -28.1 ( 1.4 17.0 ( 0.7
9.3 ( 0.5 16.4 ( 0.7
87
103
127
145
92
33.1
33.3
33.9
39.1
43.8
45.0
15.4 ( 1.3
-7.7 ( 4.0 17.7 ( 1.8 16.1 ( 1.3 26.7 ( 2.0
13.1 ( 0.5 -15.4 ( 1.4 17.7 ( 0.6 13.7 ( 0.5 22.8 ( 0.7
chlorobenzene 10.4 ( 0.7 -23.5 ( 2.1 17.5 ( 1.0 11.1 ( 0.8 18.8 ( 1.1
DMF
11.1 ( 0.7 -21.5 ( 2.1 17.5 ( 1.0 11.7 ( 0.7 19.7 ( 1.1
12.9 ( 0.9 -14.9 ( 2.7 17.4 ( 1.2 13.6 ( 0.9 23.1 ( 1.4
DMSO
68
a Errors, as given here, are assumed to be only random and represent a 68% confidence interval in the least squares deviation calculation.
b From ref.71 c Two bond couplings assigned a negative value consistent with the general rule for geminal couplings. d Chemical shift
difference at slow exchange (500 MHz at 23 °C) except for C2Cl4 (300 MHz at 22 °C). e From ref,10 ∆Gq determined at 105 °C.
tanes, and 2-acetoxy-3-halobutanes,78 which all exhibit
a detectable solvent dependence on the barrier to rota-
tion, rotation in 1a appears to be indifferent to medium
influences (Scheme 3).
The solvent effects observed for amide rotation79
depend on a reduction in dipole moment during the
dynamic process.80 For amides, the decrease amounts to
ca. 0.2 or 1.8 D, depending on the nature of the torsional
transition state (DMA, Figure 5).30,31,35,37,39
A polar solvent preferentially stabilizes the charge
separation in the more polar ground state relative to the
less polar transition states;34 the lack of a solvent effect
in carbamates has been attributed in part to the presence
of increased charge separation in the transition state44
and the relatively smaller dipole moments of carbamates
as compared to analogous amides.45 As implied above for
MeOSSOMe 1m, the dipole moment difference for di-
alkoxy disulfides is significant. To evaluate the situation
for 1a, the X-ray10 structure and the corresponding trans
transition state were optimized with the MM3* force
field81 and subsequently subjected to single-point calcula-
tions with density functional theory using the B3LYP/
6-311G* method.82-84 While the transition state model
is essentially nonpolar (µcalcd ) 0.03 D), the enantiomeric
ground states are estimated to sustain a substantial
dipole moment (µcalcd ) 5.9 D). Since the ground state-
transition state difference is of the same order of mag-
nitude as that for amides, a solvent dependence is
expected. It should be noted, however, that for non-
hydrogen-bonding solvents similar to those used in the
FIGURE 3. Relationship between the observed rotational
barriers of 1a and the Dimroth-Reichardt solvent polarity
parameter, ET.71
FIGURE 4. Relationship between the observed rotational
barriers of 1a and the Onsager function (ꢀ - 1)/(2ꢀ + 1), where
ꢀ is the dielectric constant.73
Rate constants kc and free energies of activation ∆Gq
at the coalescence temperature Tc were also calculated
using the Gutowski-Holm76 approximation equations 1
and 2 following Raban and co-workers.77 For each sol-
vent, the coupling constant, JAB, the chemical shift
difference, ∆ν, and the full-width-at-half-height (fwhh)
were obtained directly from the frequency separation
of the appropriate peaks in the slow exchange region
(kc , ∆ν), in the present cases at least 70 °C below each
individual Tc.
(78) ∆Gq varies by ca. 0.2 kcal/mol over two solvents (CH2CHCl and
CBrF3) for a series of 2-alkoxy-3-halobutanes and 2-acetoxy-3-halobu-
tanes. See: Wang, C. Y.; Bushweller, C. H. J. Am. Chem. Soc. 1977,
99, 313.
(79) ∆Gq can increase by as much as 3 kcal/mol upon changing from
an apolar solvent to water. Drakenberg, T.; Dahlqvist, K.-I.; Forse´n,
S. J. Phys. Chem. 1972, 76, 2178.
(80) Some have attributed the observed solvent effect in the rotation
about the C-N amide bond to internal pressure and the increased
volume requirements of the transition state structure in comparison
to the ground-state structure. See for instance: Suarez, C.; Lemaster,
C. B.; Lemaster, C. L.; Tafazzoli, M.; True, N. S. J. Phys. Chem. 1990,
94, 6679. It should be noted that in the case of dialkoxy disulfides, the
transition state does not demand greater steric requirements compared
to the ground state. Thus, any increase in the barrier in the liquid
phase compared to the gas-phase cannot be due to this reasoning.
(81) Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.;
Lipton, M.; Caufield, C.; Chang, G.; Hendrickson, T.; Still, W. C. J.
Comput. Chem. 1990, 4, 440. We have modified the MM3* force field
to include new atom types and parameters to better handle the OSSO
and OS(S)O functionalities.
The LSA free energies from Table 5 recalculated at the
corresponding coalescence temperatures (Table 6) are
within 0.4 kcal/mol on average from those derived from
Tc, indicating that both free energy assessment methods
provide comparable values. In this context, the ∆Gq
Tc
values span the slightly larger range of 1.5 kcal/mol, but
once again show no correlation with solvent polarity.
Unlike amides,34,35,37 acrylonitriles,43 2-alkoxy-3-halobu-
(82) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
(83) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785.
(84) Miehlich, B.; Savin, A.; Stoll, H.; Preuss, H. Chem. Phys. Lett.
1989, 157, 200.
(76) Gutowsky, H. S.; Holm, C. H. J. Chem. Phys. 1943, 25, 1228.
(77) Kost, D.; Carlson, E. H.; Raban, M. Chem. Commun. 1971, 656.
5968 J. Org. Chem., Vol. 70, No. 15, 2005

Synthesis and Physical Properties of Dialkoxy Disulfides
TABLE 6. Comparison of S-S Torsion Barriers for 1a Derived from Complete Line Shape Analysis (LSA) and the Tc
Method of Equations 1 and 2 at Tc
LSAa
Tc methodb
discrepancy
c
c
∆Gq
∆Gq
∆Gq
(kcal/mol)
ET(30)d
(kcal/mol)
Tc
Tc
solvent
(kcal/mol)
kc (s-1) ( 60
(kcal/mol)
kc(s-1) ( 61
e
C2Cl4
19.0 ( 0.2
19.4 ( 0.7
18.2 ( 2.0
18.9 ( 0.7
19.7 ( 1.1
19.3 ( 1.1
18.3 ( 1.3
32.1
33.1
33.3
33.9
39.1
43.8
45.0
p-xylene
pyridine
toluene
chlorobenzene
DMF
DMSO
180
140
169
165
122
113
18.7 ( 0.3
17.9 ( 0.3
18.5 ( 0.3
19.3 ( 0.3
18.9 ( 0.4
17.9 ( 0.4
-93
-37
-41
-20
-30
-45
0.7 ( 0.8
0.3 ( 2.0
0.4 ( 0.8
0.4 ( 1.2
0.4 ( 1.2
0.4 ( 1.4
a ∆Gq was determined using LSA. b Comparison of ∆Gq was made at Tc for each of the solvents. c ∆Gq determined at the Tc of 1a in
each respective solvent. d From ref.71 e From ref,10 ∆Gq determined at 105 °C.
mantyl)methane88 7 and related alcohols89,90 and hydro-
carbons.89
Unimolecular isomerization of carbamates in protic
solvents also display a significant negative ∆Sq.44 In this
FIGURE 5. Gas-phase ground- and transition-state dipoles
case, the negative ∆Sq observed may be associated with
(D) for amide and carbamate isomerization. (a) Dipole mo-
stronger solvation of the more polar transition state (TS2
in Figure 5) as compared to the ground state, resulting
in it being the more favored transition state in water.
ments calculated at HF/6-31++G**. Here, rotation through
TS1 is preferred by 3.5 kcal/mol.37 (b) Dipole moments calcu-
lated at HF/6-31G*. Here, TS1 is to be more stable than TS2
by 4.1 kcal/mol.39 (c) Dipole moments calculated at HF/6-
311G**. Here, TS1 is more stable than TS2 by 0.6 kcal/mol.44
SCHEME 3. Interconversion of Enantiomers of 1
The racemization of N-benzenesulfonyl-N-carboxymethyl-
2,4-dimethyl-6-nitroanaline 8 in 26 different solvents
consistently displays a large negative entropy of activa-
tion.91 Systems where torsional motion is severely re-
stricted are inherently entropically unfavorable.92,93 This
is the case with 1a. A possible interpretation of our
observed large ∆Sq is simply a reflection of the degree of
rigidity afforded by the OSSO functionality.
present work, the solvent-induced activation free energy
variation for amides is relatively small (0.5-1.5 kcal/mol).
If the NMR measurements incorporate an error of ( 0.5
kcal/mol, such an effect would therefore not be observed.
While the random errors for 1a are only on the order of
( 0.3 kcal/mol, complementary errors due in part to
temperature control and acquisition procedures appear
to have raised the accumulated errors beyond the thresh-
old where a small medium effect can be observed.
Our torsional barrier values of ∆Gq are between 18 and
19 kcal/mol and are consistent with previous work18 and
the efforts of Lunazzi10 with both small and large R
The ∆Hq values are smaller than those cited by
Lunazzi,10 whereas the ∆Sq values are much larger and
negative. Though rotation about single bonds is usually
characterized by ∆Sq values of ca. 0 eu,85-87 large negative
∆Sq values of the same order of magnitude as our values
are not unprecedented as with anti-o-tolyldi(1-ada-
(87) Though not strictly rotation about a single bond, due to
resonance stabilization afforded by the benzene ring, ∆Sq ranges
between
0 and -3 eu over three aprotic solvents for p-trifluo-
romethanesulfonylbenzyltriflone carbanion. See: Gromova, M.; Be´guin,
C. G.; Goumont, R.; Faucher, N.; Tordeux, M.; Terrier, F. Magn. Reson.
Chem. 2000, 38, 655.
(88) Lomas, J. S.; Anderson, J. E. J. Org. Chem. 1995, 60, 3246.
(89) Lomas, J. S.; Bru-Caodeville, V. J. Chem. Soc., Perkin. Trans.
2 1994, 459.
(90) Lomas, J. S.; Dubois, J. E. Tetrahedron 1981, 37, 2273.
(91) McKelvey, D. R.; Kraig, R. P.; Zimmerman, K.; Ault, A.; Perfetti,
R. J. Org. Chem. 1973, 38, 3610.
(85) Experimentally determined ∆Sq for the rotation about the
carbon-nitrogen single bond in tert-butylamine is 1.3 eu. See: Bush-
weller, C. H.; O’Neil, J. W.; Bilofsky, H. S. J. Am. Chem. Soc. 1970,
92, 6349.
(92) For a review on methodology that gives estimates and trends
concerning torsional entropy about single bonds, see: Mammen, M.;
Shakhnovich, E. I.; Whitesides, G. M. J. Org. Chem. 1998, 63, 3168
and references therein.
(93) For a commentary on the definition of torsional entropy, see:
Ercolani, G. J. Org. Chem. 1999, 64, 3350.
(86) Experimentally determined ∆Sq values ranged from -1 to -5
eu for a series of polychalogenated ethane derivatives. See: Brunelle,
J. A.; Letendre, L. J.; Weltin, E. E.; Brown, J. H.; Bushweller, C. H. J.
Phys. Chem. 1992, 96, 9225.
J. Org. Chem, Vol. 70, No. 15, 2005 5969

Zysman-Colman and Harpp
TABLE 7. Photolytic and Thermolytic Activity Experiments for 1a
conditions
1H NMR product yield distribution (%)d
entryb
dark
x
UV
air
N2
solid state
x
CDCl3
temp (°C)c
1a
3a
4a
5a
1
2
x
x
27
27
27
27
27
27
27
60
50
60
50
86
87
89
74
83
85
88
21
0
10
9
8
0
1
0
1
1
0
0
1
0
1
0
4
3
x
x
x
x
x
x
x
x
x
x
x
x
x
x
3
x
x
x
x
x
3
4a
5a
6
x
x
x
x
21
14
10
9
45
85
15
15
3
2
x
x
x
x
4
7
3
8
10
0
9
10
11
x
x
81
85
3
0
a Silica added. b Reaction time was 20-26 h. c No change in relative integrations at 37 °C. d 1H NMR yields ( 5%.
substituents. Both the absolute values and the 1 kcal/
mol range are entirely compatible with the data of Tables
4-6. Clearly, neither substituent size, electronics, nor
medium effects significantly perturb the energetics of the
S-S rotation barrier. The magnitude of the S-S rotation
barrier for dialkoxy disulfides is of interest, as it is
related to the barrier of thermal cleavage outlined next.
Evaluation of the Thermal and Photolytic Stabil-
ity of Dialkoxy Disulfides. Although acid- and base-
catalyzed2,5,65-67 decomposition of dialkoxy disulfides has
been briefly investigated, only one report exists on their
photochemistry.94 The authors were able to detect the
presence of radicals both at high temperatures as well
as under photolytic conditions. We initially examined the
photolytic decomposition of 1a both in the solid state as
well as in solution (CDCl3), in air, and under an inert
atmosphere. In addition, we probed the effect of silica
on the decomposition mechanism after observing that the
corresponding sulfite byproduct 5a is produced during
flash chromatography. The results are summarized in
Table 7.
After ca. 24 h at 27 °C, no matter whether in solution
(entries 2-4) or not (entries 6, 7) or under an inert
atmosphere (entries 2, 6) or not (entries 3, 7), little
change was detected in the product ratios of 1a as
compared to a control sample (entry 1) that was left at
room-temperature exposed to the atmosphere (but shielded
from light). The addition of silica catalyzed a slight
decomposition to the corresponding alcohol 3a while
exposed to air (entry 4) or under an inert atmosphere
(entry 5). However, when 1a was heated to temperatures
greater than 37 °C under an inert atmosphere while in
the solid state (entry 9), it decomposed completely to form
the alcohol 3a and trace amounts of other unidentified
decomposition products.
decomposition observed in the solid state at elevated
temperatures did not translate to the solution state
(entries 10, 11). In fact, decomposition in solution was
only observed at much higher temperatures (ca. 100-
140 °C). It should be noted that in all thermolysis
experiments, there was consistently detected a much
greater amount of alcohol 3a than aldehyde 4a. Such a
product distribution inequality argues against a con-
certed mechanism; these results suggest that these
compounds are much more UV stable than previously
reported.94 No rearrangement products such as those
observed by Braverman13,61 were observed because the
p-nitrobenzyl group is not prone to undergo such [2,3]-
sigmatropic rearrangements. In addition, no isomeriza-
tion to the thionosulfite was observed.
Decomposition to the corresponding alcohol and alde-
hyde in a non 1:1 ratio (vide infra) was also observed at
elevated temperatures in all the solvents used in the
determination of the rotational barrier. Qualitative ob-
servations revealed that 1a was much less stable and
decomposed much more readily in pyridine relative to the
other solvents tested. This is most likely due to the
nucleophilic nitrogen of the solvent catalyzing the pro-
cess.6,12 We undertook a quantitative solvent study in
order to elucidate this decomposition pathway. The rate
constant of decomposition of 1a, over a temperature
range of 40 °C, was extrapolated from a series of 1H NMR
spectra taken at regular intervals, each containing an
internal standard.
We chose three solvents that span a large polarity
range. Upon least-squares fitting of the resulting data,
we determined that the rate of decomposition of the
reaction was first order in 1a, rate ) kd[1a] (e.g., Figure
6).95
From these data, through the use of Eyring transition
state theory, the following activation parameters were
obtained (Table 8). It should be noted that although the
individual decomposition kinetics were linear, the cor-
responding Eyring plots yielded relatively poor linear fits
(R2 ranged from 0.47 to 0.67), which is reflected by the
Upon heating 1a in the solid state (exposed to the
atmosphere (entry 8), the corresponding aldehyde 4a as
well as the sulfite 5a were observed as decomposition
products. Except for this particular case, only traces of
sulfite 5a and no sulfoxylate 6a were ever detected. The
(94) Borghi, R.; Lunazzi, L.; Placucci, G.; Cerioni, G.; Plumitallo,
A. J. Org. Chem. 1996, 61, 3327.
(95) Average R2 value for the linear fits was ca. 0.97.
5970 J. Org. Chem., Vol. 70, No. 15, 2005

Synthesis and Physical Properties of Dialkoxy Disulfides
isomerization of 6a. Interestingly, room-temperature
photolysis of di-tert-butyloxy disulfide 1o in the presence
of DMPO (4-N,N-dimethylpyridine-N-oxide) yielded an
ESR spectrum103 characteristic of the presence of the
t-butoxyl radical (t-BuO•). Photolysis of 1d in the presence
of C60-fullerene yielded a single adduct, indicating ben-
zyloxyl radical formation.
FIGURE 6. First-order rate plot of the decomposition of 1a
at 105.7 °C in DMSO-d6. Errors as shown here represent a
68% confidence interval in the linear fit of the data.
The Lunazzi group also trapped an alkoxyl radical
under moderately elevated temperatures (50-70 °C).
Their thermolysis of 1d yielded a spectrum similar to that
of its photolysis; in addition, formation of the correspond-
ing benzyl adduct was detected.
TABLE 8. Activation Parameters for the Decomposition
of 1aa
∆Hq
∆Sq
∆Gq
ET(30)b
298
solvent
p-xylene
chlorobenzene
DMSO
(kcal/mol)
(eu)
(kcal/mol) (kcal/mol)
To account for our results, we propose the following
unsymmetric diradical decomposition mechanism, Scheme
4. This pathway is consistent not only with the radical
trapping experiments summarized above but also with
the decomposition kinetics and product decomposition
ratios obtained during the thermally induced homolysis
of dialkoxy disulfides (vide supra, Tables 7 and 8).
Diradical pathways have also been implicated in the
photolysis104-106 and thermolysis of cinnamyl-4-nitroben-
zene sulfenate and related compounds to their corre-
sponding sulfoxides.107-109 It should be noted that the
concerted mechanism originally proposed by Thompson
(Scheme 1) is not consistent with the data.
10 ( 4
11 ( 6
13 ( 5
-46 ( 10
-45 ( 15
-38 ( 12
24 ( 5
24 ( 7
25 ( 6
33.1
39.1
45.0
a Errors, as given here, represent a 68% confidence interval in
the least squares deviation calculation. b From ref.71
large errors in Table 8. The large ∆Sq values indicate a
highly ordered and associative mechanism. The negative
entropies are also comparable with other processes that
have nonpolar transition states, though these examples
relate to concerted reactions.96-102 While more positive
entropies of activation were expected (vide infra), nega-
tive ∆Sq values are not unusual in certain parallel
reactions.
A nonpolar transition state is suggested, as there
seems to be little solvent dependency (∆Gq ranges by ca.
1 kcal/mol over the three solvents). Lunazzi94 demon-
strated the trapping of aliphatic sulfenyl and sulfonyl
radicals in the photolysis (-20 °C) of the corresponding
dialkoxy disulfide EtOSSOEt 1n. The latter of these two
radicals resulted from oxidation of the former, ostensibly
in the presence of the radical trap. The Lunazzi group
attributed the formation of the sulfenyl radical through
S-S scission. GC-MS analysis of the products derived
from the photolysis of 1n in benzene resulted in the
corresponding sulfite (EtOS(O)OEt 5n) and sulfoxylate
(EtOSOEt 6n), as well as elemental sulfur.
It is reasonable that the benzylic radicals that the
Lunazzi group observed might arise from C-O scission
to form highly chalcogenated radical RCH2OSSO• fol-
lowed by SO loss to afford the sulfenyl radical 14a. This
mechanism is somewhat supported through the HRMS
of decomposition products of bis(p-MeO-benzyloxy) di-
sulfide 1c wherein SO was lost (HRMS of a second
decomposition product wherein SO2 was lost was also
observed). Indeed, the loss of SO was noted in the MS of
all of the dialkoxy disulfides. Of significance is that no
decomposition products of the form 1a-SO or 1a-SO2
1
were ever detected by H NMR during either photolysis
or thermolysis experiments as compared with authentic
samples such as sulfide 10a or sulfoxide 11a; related
sulfone 12a was also never detected. Though we did not
detect any benzyl-derived products resulting from C-O
cleavage such as p-nitrobibenzyl, we nevertheless ob-
served benzyl radical cations by EI MS. C-O homolysis
in 1a might have been expected given the origin of the
para nitro group. The presence of this electron-withdraw-
ing group effectively raises the electronegativity of the
Such a product distribution could be envisioned to
occur from the coupling of alkoxyl radicals with the
sulfur-centered radicals; these alkoxyl radicals could
originate from S-O scission. There were only trace
amounts of the analogous sulfite 5a and no sulfoxylate
6a or sulfinate 9a in the photolysis of 1a; the latter
compound likely resulted from the room-temperature
(103) It should be noted that the sensitivity of ESR spectroscopy is
much higher than that of 1H NMR. Thus, radical concentrations of
10-6 can easily be detected by this method. The ESR photolytic method
thus produces radicals in such small concentrations as to not contribute
to the decomposition of dialkoxy disulfides as detected by 1H NMR
spectroscopy.
(96) Snyder, J. P.; Harpp, D. N. J. Am. Chem. Soc. 1976, 98, 7821.
(97) Tardif, S.; Harpp, D. N. J. Org. Chem. 2000, 65, 4791.
(98) Jorgensen, W. L.; Blake, J. F. J. Am. Chem. Soc. 1991, 113,
7430.
(99) Ruiz-Lo´pez, M. F.; Assfeld, X.; Garc´ıa, J. I.; Mayoral, J. A.;
Salvatella, L. J. Am. Chem. Soc. 1993, 115, 8780.
(100) Doorakian, G. A.; Freedman, H. H. J. Am. Chem. Soc. 1968,
90, 3582.
(104) Guo, Y.; Jenks, W. S. J. Org. Chem. 1995, 60, 5480.
(105) Guo, Y.; Jenks, W. S. J. Org. Chem. 1997, 62, 857.
(106) Pasto, D. J.; Cottard, F. Tetrahedron Lett. 1994, 35, 4303.
(107) Amaudrut, J.; Wiest, O. J. Am. Chem. Soc. 2000, 122, 3367.
(108) Miller, E. G.; Rayner, D. R.; Mislow, K. J. Am. Chem. Soc.
1966, 88, 3139.
(101) von E. Doering, W.; Toscano, V. G.; Beaseley, G. H. Tetrahe-
dron 1971, 27, 5299.
(102) Schuler, F. W.; Murphy, G. W. J. Am. Chem. Soc. 1950, 72,
3155.
(109) Amaudrut, J.; Pasto, D. J.; Wiest, O. J. Org. Chem. 1998, 63,
6061.
J. Org. Chem, Vol. 70, No. 15, 2005 5971

Zysman-Colman and Harpp
SCHEME 4. Thermally Induced Radical Decomposition Mechanism of 1aa
a Disp. ) disproportionation; diff. ) diffusion.
benzylic carbon, thus decreasing the electronegativity
difference and thus the BDE between it and the adjoining
oxygen atom.110
FIGURE 7. Ratio of the integration of p-nitrobenzyl alcohol
3a to p-nitrobenzaldehyde 4a per mole of integratable protons.
oxides114-117 and hyponitrites118-122 have been shown to
decompose by analogous mechanisms; some have sug-
gested symmetric homolytic cleavage as the first step,123,124
but it is only the asymmetric cleavage of tetroxides that
leads to product formation. Formation of minor amounts
of sulfite 5a is outlined in Scheme 3B.125 The presence
of only small amounts of sulfite 5a suggests that in our
system, the propensity for in situ oxidation of radical
species such as 14a is low.
In our scheme, alkoxyl radicals (13a) would arise from
initial caged S-O bond scission; a second caged S-O
bond cleavage is also possible. Pasto111 has shown in a
related H-S-S-O-H system that S-S cleavage is only
slightly more favored than S-O cleavage. The preference
for S-S cleavage over S-O cleavage was shown to be
highly dependent on the stabilities of the radicals formed.
Additionally, for nondiaryl disulfides, C-S homolysis is
slightly more favorable than symmetric S-S bond break-
age.112 The presence of the benzyl group, containing
conjugated π-electrons, exerts a bond weakening effect
upon the S-O bond.113
The observed product ratios during the high-temper-
ature decomposition kinetics experiments are shown in
Figure 7 and further support a stepwise decomposition
Finally, the cage could then disproportionate (via a
remote hydrogen abstraction106 of either of the alkoxyl
radicals 13a or the solvent by the other alkoxyl radical)
to yield the alcohol 3a that was the major product
observed in our thermolysis experiments. If the source
of the hydrogen atom abstraction is another alkoxyl
radical or if â-scission of 13a occurs, then this would
afford observed aldehyde 4a. Diatomic sulfur is the final
disproportionation product, which would immediately
concatenate to form the more stable S8 allotrope in the
absence of a diene trap. This stepwise type of decomposi-
tion mechanism is not unprecedented. Indeed tetr-
(114) Russell, G. A. J. Am. Chem. Soc. 1957, 79, 3871.
(115) Fessenden, R. W. J. Chem. Phys. 1968, 48, 3725.
(116) Ingold, K. U.; Adamic, K.; Howard, J. A. Can. J. Chem. 1969,
47, 3803.
(117) Francisco, J. S.; Williams, I. H. Int. J. Chem. Kinetics 1988,
20, 455.
(118) Neumann, W. P.; Lind, H. Chem. Ber. 1968, 101, 2837.
(119) Mendenhall, G. D.; Ogle, C. A.; Martin, S. W.; Dziobak, M. P.;
Urban, M. W. J. Org. Chem. 1983, 48, 3728.
(120) Mendenhall, G. D.; Quinga, E. M. Y. Int. J. Chem. Kinetics
1985, 17, 1187.
(121) Ingold, K. U.; Paul, T.; Young, M. J.; Doiron, L. J. Am. Chem.
Soc. 1997, 119, 12364.
(122) Ingold, K. U.; Konya, K. G.; Paul, T.; Lin, S.; Lusztyk, J. J.
Am. Chem. Soc. 2000, 122, 7518.
(123) Howard, J. A.; Bennet, J. E. Can. J. Chem. 1972, 50, 2374.
(124) Minato, T.; Yamabe, S.; Fujimoto, H.; Fukui, K. Bull. Chem.
Soc. Jpn. 1978, 51, 682.
(110) Nau, W. M. J. Phys. Org. Chem. 1997, 10, 445.
(111) Pasto, D. J. J. Mol. Struct. 1998, 446, 75.
(112) Benassi, R.; Taddei, F. J. Comput. Chem. 2000, 21, 1405.
(113) Benson, S. W. Chem. Rev. 1978, 78, 23.
(125) Formation of sulfenyl radical 13 via homolysis of the S-S bond
cannot be ruled out.
5972 J. Org. Chem., Vol. 70, No. 15, 2005

Synthesis and Physical Properties of Dialkoxy Disulfides
SVP) level. The bond dissociation energy for FS-SF is
61 ( 4 kcal/mol,113 which is slightly less than that
reported for MeS-SMe. Thus, it is also reasonable to
conclude that the BDE(ROS-SOR) would approximate
this value. The ∆Hf° for t-BuO• ) -22.8 kcal/mol;133
this is a good estimate for other alkoxyl radicals. Thus,
in order for BDE(ROSS-OR) to be less than BDE-
(ROS-SOR), the contribution of ∆Hf°(ROSS•) + ∆Hf°-
(RO•) - 2 ∆Hf°(ROS•) has to be negative in magnitude.
Given a large negative value for ∆Hf°(RO•) and a small
positive value for ∆Hf°(ROS•), as well as rationales in the
stabilization of ROSS• (vide supra), it is reasonable to
expect that the S-O bond would be the weaker bond and
thus more prone to homolysis than the S-S bond in
ROSSOR 1.
It appears that the two observable processes (decom-
position and coalescence) are each solvent-insensitive.
Moreover, the decomposition phenomenon is ca. 6-8 kcal/
mol higher. This is in accordance with our qualitative
observations; if the decomposition phenomenon were
more energetically favored, then one would not be able
to observe coalescence.
FIGURE 8. Resonance form of ROS• vs ROSS•.
mechanism. The average 3a:4a (ROH:RCHO) ratio is ca.
2:1 for chlorobenzene and p-xylene. There is consistently
considerably more alcohol formed in DMSO. The non 1:1
ratio of decomposition products can usually120,121 be
ascribed to alkoxyl radicals 13a diffusing from the cage
and subsequently abstracting hydrogens from the envi-
ronment. In hydroxylic solvents, a 1,2-H shift from carbon
to oxygen has also been suggested as a viable possibil-
ity;126 this shift is solvent-assisted and would occur due
to increased radical stabilization (benzylic radical forma-
tion).127 The relative excess of alcohol observed in DMSO
may be due to the presence of water in the solvent, which
might promote the fast 1,2-H shift.122 It should be noted
that invoking a concerted mechanism (Scheme 1) implies
a 1:1 ratio of ROH:RCHO.
To our knowledge, the S-O and S-S bond dissociation
energies for dialkoxy disulfides have not been reported.
Estimating them is problematic due to the formation of
highly chalcogenated radicals. Nevertheless, the BDEs
are related by equation 3.
Conclusion
We have optimized a generalized synthetic procedure
for the synthesis of acyclic dialkoxy disulfides. We have
shown that this synthesis can be effectively applied with
a variety of substrates. We carried out a substituent and
solvent study on their ability to influence the S-S
torsional barrier. Our data indicates that there exists no
significant substituent or solvent effect. The latter point
may seem incongruent given our calculated dipoles for
the ground and trans transition states, but the observed
∆∆Gq has been rationalized as being too small to detect
within experimental error (vide supra). In addition, we
propose a mechanism to account for the observed thermal
decomposition of 1a. Our experimentally determined
activation parameters for this process indicate that 1a
decomposes under first-order kinetics and that there also
does not seem to exist any appreciable solvent effect.
Decomposition appears to proceed via initial S-O bond
homolysis and is ca. 6-8 kcal/mol more thermally
demanding than overcoming the internal S-S rotation
barrier. To our knowledge, this represents the first study
to account for the mechanism of decomposition of this
highly unusual class of compounds.
BDE(ROSS-OR) ) ∆Hf°(ROSS•) + ∆Hf°(RO•) -
2∆Hf°(ROS•) - BDE(ROS-SOR) (3)
In peroxides (RO-OR), the lone pair repulsion between
oxygen atoms is believed to be responsible for the
weakening of the O-O bond.128 In disulfides, the lone
pairs are more diffuse, and therefore the S-S bond is
stronger. The BDE of sulfenates (RO-SR) is intermediate
between peroxides and disulfides. This is evidenced by
their relative BDEs: BDE(MeO-OMe) ) 38 ( 2 kcal/
mol,129 BDE(MeS-OMe) ) 53 ( 10 kcal/mol,107 BDE-
(MeS-SMe) ) 65 ( 1 kcal/mol.130 Analogously, we would
expect S-O bond cleavage to be easier than S-S bond
cleavage in dialkoxy disulfides 1. Additionally, the formed
ROSS• radical is able to be better stabilized through a
three-electron π-system with enhanced hyperconjugation
compared to ROS•,131 which would result from S-S
scission. This is qualitatively shown in Figure 8. As
oxygen is more electronegative than sulfur, dipolar
structure A in Figure 8 is likely to be a poor contributor
to the overall stability of the sulfenyl radical. Sulfur is
also more polarizable than oxygen, so charge-separated
species as in Figure 8B would be seen as more viable
resonance contributors. In fact, quantitatively, Gregory
and Jenks132 have shown the ROS• radical to be quite
unstable: ∆Hf° ) 1.4 kcal/mol for MeOS• at the G2(MP2,-
Acknowledgment. We thank the Natural Sciences
and Engineering Research Council of Canada (NSERC)
for financial support of this work. We are indebted to
Dr. S. Bohle (McGill University) for helpful suggestions
concerning the decomposition mechanism. We are es-
pecially grateful for the insightful contributions of Dr.
K. U. Ingold (NRC) concerning analogous decomposition
mechanisms in tetroxides and hyponitrites. E.Z.-C.
would like to thank FQRNT for a personal scholarship.
(126) Gilbert, B. C.; Holmes, R. G. G.; Laue, H. A. H.; Norman, R.
O. C. J. Chem. Soc., Perkin. Trans. 2 1976, 1047.
(127) Elford, P. E.; Roberts, B. P. J. Chem. Soc., Perkin. Trans. 2
1996, 2247.
Supporting Information Available: Experimental pro-
cedures and characterization for selected compounds and
experimental procedures for dynamic NMR and thermal
decomposition experiments. This material is available free of
charge via the Internet at http://pubs.acs.org.
(128) Bach, R. D.; Ayala, P. Y.; Schlegel, H. B. J. Am. Chem. Soc.
1996, 118, 12758.
(129) McMillen, D. F.; Golden, D. M. Annu. Rev. Phys. Chem. 1982,
33, 493.
(130) Nicovich, J. M.; Kreuter, K. D.; van Dijk, C. A.; Wine, P. H. J.
Phys. Chem. 1992, 96, 2518.
JO050574S
(131) The three electrons would be placed in a new set of π and π*
orbitals.
(132) Gregory, D. D.; Jenks, W. S. J. Org. Chem. 1998, 63, 3859.
(133) Nangia, P. S.; Benson, S. W. J. Phys. Chem. 1979, 83, 1138.
J. Org. Chem, Vol. 70, No. 15, 2005 5973