Isomerization of triphenylmethoxyl: The wieland free-radical rearrangement revisited a century later

Article abstract of DOI:10.1002/anie.201001008

(Figure Presented) Old wisdom teaches a new lesson : The rate constant for the isomerization of thermally generated triphenylmethoxyl to Ph ₂(PhO)C has been determined both experimentally and by DFT calculations (see scheme), and the results vindicated Wieland's 1911 observations and conclusions-in contrast to photochemically based findings.

Full text of DOI:10.1002/anie.201001008

Communications
DOI: 10.1002/anie.201001008
Wieland Rearrangement
Isomerization of Triphenylmethoxyl: The Wieland Free-Radical
Rearrangement Revisited a Century Later**
Gino A. DiLabio, K. U. Ingold,* Shuqiong Lin, Grzegorz Litwinienko, Olga Mozenson,
Peter Mulder, and Thomas T. Tidwell
In 1911, Wieland[1] decomposed 20 g (38.6 mmol) of (Ph3CO)2
(1) under CO2 in boiling xylene for 10 minutes, and separated
a crystalline product (13–15 g) by the addition of 70 mL of
absolute alcohol and concentration of the crude product. This
product was identified as (Ph2(PhO)C)2 (4, 65–75% yield;
Scheme 1). Vacuum distillation (11 mm) of the residue
Scheme 2. Formation of oxaspiro intermediates, 6.
with these calculations, cumyloxyl radicals, PhC(Me)2OC, para
substituted with a 2,2-diphenylcyclopropyl reporter group,
Scheme 1. Rearrangement of triphenylmethoxyl (2).
have been demonstrated to be in equilibrium with
spiro radicals 6.[12]
Rate constants (10ꢀ6 ꢁ k2 sꢀ1) measured at room temper-
yielded 2.3 g of a yellow oil from which benzophenone
(almost 2 g) and phenol (0.2 g) were separated. Further
heating gave a substantial, but not quantifiable, amount of
Ph3COH. Wielandꢀs interpretation was that triphenyl-
methoxyl radicals (Ph3COC, 2) had been formed and had
isomerized to Ph2(PhO)CC radicals (3) which then coupled
(Scheme 1). This was the first clearly demonstrated, and
explicitly shown, free-radical rearrangement—a priority often
overlooked.
ature by laser flash photolysis (LFP) in CH3CN were 2.5,[6]
2.8,[7,8] and 3.2 sꢀ1,[2] and are nicely bracketed by the result of
density functional theory (DFT) calculations (10ꢀ6 ꢁ k2 sꢀ1
=
0.93[9] and 7.9 sꢀ1[10]). Experiment[8] and theory[9,10] agree that
k2 depends on the nature of the para substituent. In addition,
k2 decreases as solvents become more polar.[8] In contrast, the
rate of b scission of the cumyloxyl radical increases in more
polar solvents (Scheme 3).[13]
The rate constants and mechanisms of isomerization of
triphenylmethoxyl (2)[2–4] and the analogous isomerizations of
Ph2C(Me)OC (5),[2–10] and related radicals (Scheme 2),[7–12]
have received considerable attention. Claims that discrete
spiro intermediates (6) had been identified in the re-
arrangements of 2[3] and 5[2,3] have been disproven.[4,6]
However, computational studies on the rearrangement of
5[9,10] (and PhCH2OC)[11] do indicate stepwise processes with
spiro radicals (6) as intermediates (Scheme 2). Consistent
Scheme 3. b Scission of the cumyloxyl radical.
The only previous experimental study of the kinetics of
the rearrangement of 2 to 3 was carried out by Schuster
et al.,[2] who relied on the photolysis of Me3COOCPh3 (7) to
generate 2 (Scheme 4). Picosecond LFP (266 nm) of 7 in
[*] Dr. K. U. Ingold, S. Lin, O. Mozenson
National Research Council
100 Sussex Drive, Ottawa, ON, K1A 0R6 (Canada)
Dr. G. A. DiLabio
National Institute for Nanotechnology, Edmonton (Canada)
Prof. G. Litwinienko
Faculty of Chemistry, University of Warsaw (Poland)
Scheme 4. Photolysis of peroxide 7.
Dr. P. Mulder
Leiden Institute of Chemistry, Leiden University (Netherlands)
Prof. T. T. Tidwell
Department of Chemistry, University of Toronto (Canada)
CH3CN gave a broad absorption signal (lmax = 545 nm)
attributed to 3 that appeared within the 17 ps width of the
laser pulse and persisted for at least 6 ns.[2] The authors[2]
concluded that k1 exceeds 5 ꢁ 1010 sꢀ1. Furthermore, photol-
ysis of 7 at 254 nm in iPrOH saturated with N2 gave dimer 4,
benzophenone, and phenol, but no (< 0.5%) Ph3COH.[2]
Assuming that k3 is approximately equal to the Me3COC +
[**] We thank Malgosia Daroszewka for technical support, and Nanyan
Fu and Annette Allen for analytical samples of 1 and 2. We are
extremely grateful to Hꢀlꢁne Lꢀtourneau for assistance with the
cover image.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201001008.
5982
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 5982 –5985

Angewandte
Chemie
Me2CHOH rate constant (1.8 ꢁ 106 mꢀ1 sꢀ1)[14] the lower limit
for k1 in neat iPrOH (13m) is ꢁ (1.8 ꢁ 106 ꢁ 13)/0.005 = (2.3 ꢁ
107) ꢁ 200 = 5 ꢁ 109 sꢀ1, a result consistent with the LFP
experiment.
1,4-C6H8 rate constant, that is, 4 ꢁ 107 mꢀ1 sꢀ1, see the Support-
295K
ing Information, the mean value of k1
(derived from
several experiments, see Supporting Information) was 1.4 ꢁ
108 sꢀ1.
The reported[2] absence of Ph3COH[15] and presence of
benzophenone led us to suspect (see the Supporting Infor-
mation) that UV photolysis of compound 7 did not give 2 in its
ground state and to doubt that k1 is larger than 5 ꢁ 1010 sꢀ1. We
therefore redetermined k1 using 2 generated thermally in the
presence of hydrogen-atom donating solvents.
We next repeated Wielandꢀs experiment. As the purity
and water content of his “xylene” are unknown, 1 was
decomposed at 411 K in: 1) 1.2 mL of dry m-xylene at reflux,
2) as in 1) but water saturated, and 3) as in 1) but using
EtOH[1] in the work-up procedure. After 1 (0.099 mmol) had
completely decomposed (10 min, as evident by HPLC),
solutions were diluted to 10 mL with CH3CN (1 and 2) or
EtOH (3) and analyzed by GC. Ph3COH was present in low,
but equal, yields in these three experiments (see the
Supporting Information). As each molecule of 1 gives two
Ph3COC (2), the yields of alcohol indicated that 1.34% of the
triphenylmethoxyl (2) has abstracted hydrogen from the
xylene, Scheme 6, and that the other 98.66% must have
The most convenient, room temperature, thermal source
of 2 seemed likely to be hyponitrite 8 (Scheme 5).[16]
Scheme 5. Hyponitrite route to triphenylmethoxyl (2).
Scheme 6. Wieland’s experiment (formation of Ph3COH).
Thermolysis of 8 in CH2Cl2 in air at 295 K (k4295K = 1.1 ꢁ
10ꢀ4 sꢀ1, see the Supporting Information) with periodic
analyses by HPLC (see Table S1 and Figure S1 in the
Supporting Information) and (after complete decomposition
of compound 8) analyses by GC and GC/MS, showed that the
main products were 4 (by comparison with authentic 4, which
is unstable in solution, see the Supporting Information),
phenol, benzophenone, and a minor amount of PhCO2Ph.[17]
Samples collected after reaction times of 67 ꢁ 103 seconds and
486 ꢁ 103 seconds contained ꢁ 9.5% of 1,[18] which is stable at
295 K. Identified species (unchanged 8 and its products)
accounted for approximately 100% of the phenyl groups for
up to 11 ꢁ 103 seconds (see the Supporting Information).
In the absence of air, 8 was thermolysed in solvent
mixtures containing CH2Cl2 (to solubilize 8) and the hydrogen
atom donor, 1,4-cyclohexadiene. HPLC analyses showing the
loss of 8 and formation of products are presented for 1,4-
C6H8/CH2Cl2 (80:20, v/v), in the Supporting Information. The
main product was Ph3COH, which would not have been
detected if k1 were 5 ꢁ 1010 sꢀ1 or larger. Other lesser products
(see the Supporting Information) included Ph2(PhO)CH, two
compounds resulting from combinations with 1,4-C6H8, minor
amounts of 2 (at short reaction times), and (after 624 ꢁ 103 s)
2.6% of 1. Triphenylmethanol was also formed with lower 1,4-
C6H8/CH2Cl2 ratios (see the Supporting Information). The
yield of freely diffusing 2 will be 2 ꢁ [(80ꢀ8t)ꢀ1t] or, after
complete decomposition of compound 8, 2 ꢁ (80ꢀ1final). Based
on the 2.6% yield of 1, the minor correction for in-cage
combination of geminate 2 will be assumed to be a constant
2.5% of decomposed 8. Thus, the yield of free 2 during the
reaction is 2 ꢁ 0.975 ꢁ (80ꢀ8t), with a final yield of 2 ꢁ 0.975 ꢁ
80. It is only some fraction of 2, f = Ph3COH/[1.95ꢁ(80ꢀ8t)],
that can form Ph3COH (Scheme 5), while the remaining, 1-f,
fraction of these radicals will isomerize and form other
products. This competition yields k1 via Equation (1). Again
assuming that k5 will be essentially equal to the Me3COC +
isomerized. A value for k6411K = 4 ꢁ 106 mꢀ1 sꢀ1 was estimated
from room temperature kinetic data (see the Supporting
Information) which was combined with the Ph3COH yield
and the molarity of neat m-xylene at 411 K (7.1m) to give
k1411K = (4 ꢁ 106 ꢁ 7.1)/0.0134 = 2.1 ꢁ 109 sꢀ1.
Combination of k1295K = 1.4 ꢁ 108 sꢀ1 with k1411K = 2.1 ꢁ
109 sꢀ1 would yield a two point Arrhenius plot, Ea1 = 5.6 kcal
molꢀ1, log(A1sꢀ1ꢀ1) = 12.3 (for a discussion of the “expected”
value of A1 see the Supporting Information). We also applied
DFT[19] to 2. Phenyl migration was again found to proceed via
a spiro intermediate, 6 (Figure 1a), that lies in a shallow
energy minimum. Relative free enthalpies (at 298 K in
kcalmolꢀ1) along the reaction coordinate corresponding to
the isomerization of 2 are: reactant, 0.0; transition state (TS)
#1 (O approach to C1 of Ph), 5.7; intermediate 6, 5.2; TS #2
(opening of the 3-membered ring), 5.5; final product, ꢀ20.8
(see Figure S3 in the Supporting Information). More inter-
295K
estingly, k1
was computed to be 2.0 ꢁ 108 sꢀ1, with log
(A1sꢀ1) = 12.9 and Ea1 = 6.2 kcalmolꢀ1 calculated over the
Figure 1. a) View of the spiro intermediate 6, predicted by DFT for the
isomerization of 2. b) Isomerization of 2, Arrhenius plot calculated by
&
DFT ( ) and two experimental rate constants (*).
Angew. Chem. Int. Ed. 2010, 49, 5982 –5985
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5983

Communications
temperature range 295–411 K. These results are in very good
agreement with experiment, see Figure 1b.
been unambiguously detected, even by picosecond LFP,
because the broad absorption of Ph2(PhO)CC will differ only
marginally from the broad absorption of Ph3COC.[24]
Wielandꢀs identification of the first free-radical rearrange-
ment did not rely on modern instrumentation but on careful
product analyses by classical methods and inspired chemical
insight. We take this opportunity to pay homage to this
outstanding organic chemist.[26,27]
Schuster et al.[2] reported that Ph3COH was not produced
(< 0.5%) during photolysis of 7 at 254 nm under N2 in neat
iPrOH, a result that is inconsistent with k1295K = 1.4 ꢁ 108 sꢀ1
(taking k3[iPrOH] = 2.3 ꢁ 107 sꢀ1, see above). Complete ther-
mal decomposition of 8 (3.30 mmol) in iPrOH (6.55m in
CH2Cl2, the latter required for solubility) at 295 K gave
0.682 mmol of Ph3COH, a yield of 10.6% (based on two
molecules of 2 formed per 8, see above). This translates to
k1295K = 1.05 ꢁ 107 sꢀ1 which is only about 8% of the value in
1,4-C6H8/CH2Cl2. As the rate constants used for the two
competing hydrogen-abstraction reactions come from the
same technique (and source),[14] the rate of isomerization of 2
must be solvent dependent.[20]
Received: February 17, 2010
Keywords: peroxide · radical reactions · reaction kinetics ·
.
rearrangements
All the results presented above prove that UV photolysis
of Me3COOCPh3 (7) does not yield significant amounts of
triphenylmethoxyl (2) in its ground state. Instead, photolysis
must yield Ph3COC in an excited state, [Ph3COC]*, and, before
this “cools” to its ground state, it undergoes unimolecular
decomposition with k > 5 ꢁ 109 sꢀ1 (see above, not necessarily
with k > 5 ꢁ 1010 sꢀ1, see below). Benzophenone formation[2]
implies that some “hot” Ph3COC undergoes b scission
(Scheme 7).[21]
[1] H. Wieland, Ber. Dtsch. Chem. Ges. 1911, 44, 2550.
[2] D. E. Falvey, B. S. Khambatta, G. B. Schuster, J. Phys. Chem.
1990, 94, 1056.
[3] L. Grossi, S. Strazzari, J. Org. Chem. 2000, 65, 2748.
[4] K. U. Ingold, M. G. Smeu, A. DiLabio, J. Org. Chem. 2006, 71,
9906.
[5] J. A. Howard, K. U. Ingold, Can. J. Chem. 1969, 47, 3797.
[6] J. T. Banks, J. C. Scaiano, J. Phys. Chem. 1995, 99, 3527.
[7] C. S. Aureliano Antunes, M. Bietti, O. Ercolani, O. Lanzalunga,
M. Salmone, J. Org. Chem. 2005, 70, 3884.
[8] M. Bietti, M. Salamone, J. Org. Chem. 2005, 70, 10603.
[9] M. Bietti, G. Ercolani, M. Salamone, J. Org. Chem. 2007, 72,
4515.
[10] M. Smeu, G. A. DiLabio, J. Org. Chem. 2007, 72, 4520.
[11] G. da Silva, J. W. Bozzelli, J. Phys. Chem. A 2009, 113, 6979.
[12] M. Salamone, M. Bietti, A. Calcagni, G. Gente, Org. Lett. 2009,
11, 2453.
[13] D. V. Avila, C. E. Brown, K. U. Ingold, J. Lusztyk, J. Am. Chem.
Soc. 1993, 115, 466.
Scheme 7. Excited state radical formation.
[14] H. Paul, R. D. Small, Jr., J. C. Scaiano, J. Am. Chem. Soc. 1978,
100, 4520.
Production of excited Ph3COC by UV photolysis of
peroxide 7 is plausible because the Ph3C group will absorb
[15] Reference [2] misses Wielandꢀs[1] identification of Ph3COH and
states: “this radical (Ph3COC) rearranges to the phenoxydiphe-
nylmethyl radical to the exclusion of hydrogen atom abstraction
reactions” with a citation to Wieland!.
ꢀ
> 99% of the UV energy and the O O bond < 1%, as was
recognized by Schuster et al.,[2] who nevertheless presumed a
“rapid intramolecular energy transfer” followed by “oxygen–
oxygen bond rupture”. A difference in the behavior of
thermally and photochemically produced radicals was first
reported by Bevington and Lewis[22] who showed that the
[16] M. A. Spielman, J. Am. Chem. Soc. 1935, 57, 1117.
[17] In the presence of oxygen, benzophenone and phenyl benzoate
will be formed by the reaction cascade:
Ph2(PhO)CC
+
O2!!Ph2(PhO)COC!Ph2CO
+
PhOC
Ph2(PhO)COC!(PhO)2(Ph)CC!O2!!(PhO)2(Ph)COC!
thermal decomposition of benzoyl peroxide (labeled with 14
C
PhC(O)(OPh) + PhOC.
in the carboxylate positions) yielded benzoyloxyl radicals
exclusively (Scheme 8, f = 0), whereas photolysis gave both
[18] Similarly, the thermolysis of cumyl hyponitrite yielded 3–7%
dicumyl peroxide.[13]
[19] F. A. Hamprecht, A. J. Cohen, D. J. Tozer, N. C. Handy, J. Chem.
Phys. 1998, 109, 6264. For additional details see the Supporting
Information.
[20] k2 only varies from a high value corresponding of 4.8 ꢁ 106 sꢀ1 in
CCl4 to a low value of 1.5 ꢁ 106 sꢀ1 in CF3CH2OH.[8] These kinetic
solvent effects on the isomerizations of 5 and 2 arise because the
alkoxyl radicals are more strongly solvated in polar and hydro-
gen-bond donor solvents than the transition states leading to
isomerization.
Scheme 8. Benzoyl peroxide homolysis.
PhCO2C and PhC (f = 0.29). The PhCO2C that survived photo-
generation behaved in the same way as the thermally
generated PhCO2C, and presumably came from the
PhC(O)O half of the molecule that did not absorb the
incident photon—a presumption consistent with the signifi-
cantly smaller PhCO2/PhC ratio in photolyzed tert-butyl
perbenzoate.[23] Ironically, even if Schuster et al.[2] had gen-
erated 2 in the ground state it is improbable that it could have
[21] The photolysis in 2-propanol yielded[2] “the product normally
assumed to be the [Ph2(PhO)CC] dimer” possibly via an in-cage
collapse of the (still “hot”) products, see Scheme 7 i.e.,
[Ph2CO PhC]*cage!Ph2(PhO)CC.
[22] J. C. Bevington, T. D. Lewis, Trans. Faraday Soc. 1958, 54, 1340;
see also: J. Chateauneuf, J. Lusztyk, K. U. Ingold, J. Am. Chem.
Soc. 1988, 110, 2877.
[23] T. Koenig, J. A. Hoobler, Tetrahedron Lett. 1972, 13, 1803.
5984 www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 5982 –5985

Angewandte
Chemie
[24] All ArC(R1,R2)OC radicals have absorptions in the visible
region,[25] e.g., (C6H5)2(CH3)COC has a broad absorption width
at lmax = 535 nm.[6,7]
[25] D. V. Avila, J. Lusztyk, K. U. Ingold, J. Am. Chem. Soc. 1992,
114, 6576; D. V. Avila, K. U. Ingold, A. A. Di Nardo, F. Zerbetto,
M. Z. Zgierski, J. Lusztyk, J. Am. Chem. Soc. 1995, 117, 2711.
[26] Wieland biographies: B. Witkop, Angew. Chem. 1977, 89, 575;
Angew. Chem. Int. Ed. Engl. 1977, 16, 559; E. Vaupel, Angew.
Chem. 2007, 119, 9314; Angew. Chem. Int. Ed. 2007, 46, 9154.
Neither article cites reference [1].
[27] For a review of radical isomerizations involving aryl migrations,
see: A. Studer, M. Bossart, Tetrahedron 2001, 57, 9649.
Angew. Chem. Int. Ed. 2010, 49, 5982 –5985
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5985