Photo-Arbuzov rearrangements of dimethyl benzyl and dimethyl p-acetylbenzyl phosphite

Article abstract of DOI:10.1021/ja982440k

The direct ultraviolet irradiation of dimethyl benzyl phosphite (1) and dimethyl p-acetylbenzyl phosphite (8) was investigated in acetonitrile, cyclohexane, and benzene. Phosphite 1 gives predominantly the photo-Arbuzov product, dimethyl benzylphosphonate (2), in 67-81% accountability yields, based of phosphite consumed, along with minor amounts of bibenzyl (20) and dimethyl phosphite (10). The quantum yield for formation of 2 in cyclohexane, φp, is 0.43. By contrast, irradiation of phosphite 8 yields only 7-13% of photo-Arbuzov phosphonate (9) but relatively large amounts of radical diffusion products: dimethyl phosphite (10) the p-acetylbenzyl radical dimer (11); and p-acetyltoluene (12). Evidently 8, closely related to acetophenone, reacts predominantly via the triplet excited estate to generate long-lived, triplet, free-radical pairs (6 and 7a). In benzene, further products (15, 16, 17a and 17b) are identified that result from addition of the phosphinoyl radical (6) to benzene to give cyclohexadienyl radical 14, followed by combination and disproportionation reactions with radical 7a. (Total product quantum yields in benzene (Σφ_i) = 0.47.) In benzene, accountabilities of radical 6 from photolysis of 8 as high as 56% are encountered along with up to 92% accountabilities of p-acetylbenzyl (7a) radicals. Addition of radical scavengers PhSH, PhCH₂Br, and TEMPO in the three solvents establishes the cage yield of 9 as 3-5%. The products of radical trapping provide further proof of the radical-pair nature of the photolysis of phosphite 8, including a 95% accountability of 6 with PhCH₂Br in benzene. It is proposed that the CH₂-O scission of triplet 8 must occur concertedly with partial phosphoryl (P=O) bond formation. The trapping of radicals 6 and 7b from irradiation of phosphite 1 as the benzene adducts 22 and 23, analogous structurally to those (16 and 17) from phosphite 8, supports the postulation that photoisomerization of 1 to 2 proceeds via short-lived, presumably singlet, free-radical pairs.

Full text of DOI:10.1021/ja982440k

J. Am. Chem. Soc. 1999, 121, 2085-2096
2085
Photo-Arbuzov Rearrangements of Dimethyl Benzyl and Dimethyl
p-Acetylbenzyl Phosphite
Srinivasan Ganapathy, B. B. V. Soma Sekhar, S. Matthew Cairns, K. Akutagawa, and
Wesley G. Bentrude*
Contribution from the Department of Chemistry, UniVersity of Utah, Salt Lake City, Utah 84112
ReceiVed July 10, 1998
Abstract: The direct ultraviolet irradiation of dimethyl benzyl phosphite (1) and dimethyl p-acetylbenzyl
phosphite (8) was investigated in acetonitrile, cyclohexane, and benzene. Phosphite 1 gives predominantly the
photo-Arbuzov product, dimethyl benzylphosphonate (2), in 67-81% accountability yields, based of phosphite
consumed, along with minor amounts of bibenzyl (20) and dimethyl phosphite (10). The quantum yield for
formation of 2 in cyclohexane, φP, is 0.43. By contrast, irradiation of phosphite 8 yields only 7-13% of
photo-Arbuzov phosphonate (9) but relatively large amounts of radical diffusion products: dimethyl phosphite
(10) the p-acetylbenzyl radical dimer (11); and p-acetyltoluene (12). Evidently 8, closely related to acetophenone,
reacts predominantly via the triplet excited estate to generate long-lived, triplet, free-radical pairs (6 and 7a).
In benzene, further products (15, 16, 17a and 17b) are identified that result from addition of the phosphinoyl
radical (6) to benzene to give cyclohexadienyl radical 14, followed by combination and disproportionation
reactions with radical 7a. (Total product quantum yields in benzene (Σφi) ) 0.47.) In benzene, accountabilities
of radical 6 from photolysis of 8 as high as 56% are encountered along with up to 92% accountabilities of
p-acetylbenzyl (7a) radicals. Addition of radical scavengers PhSH, PhCH2Br, and TEMPO in the three solvents
establishes the cage yield of 9 as 3-5%. The products of radical trapping provide further proof of the radical-
pair nature of the photolysis of phosphite 8, including a 95% accountability of 6 with PhCH2Br in benzene.
It is proposed that the CH2-O scission of triplet 8 must occur concertedly with partial phosphoryl (PdO)
bond formation. The trapping of radicals 6 and 7b from irradiation of phosphite 1 as the benzene adducts 22
and 23, analogous structurally to those (16 and 17) from phosphite 8, supports the postulation that
photoisomerization of 1 to 2 proceeds via short-lived, presumably singlet, free-radical pairs.
We have published preliminary reports1,2 of the rearrange-
ment, on direct irradiation with ultraviolet light, of dimethyl
benzyl phosphite (1) to the isomeric benzylphosphonate 2.
studies of the potential formation of radical pairs (6 and 7) in
the photorearrangements of phosphites 1 and 8 and the 1-naph-
thylmethyl analogue of 1 have been published.
This rearrangement is formally a 1,2-sigmatropic shift. It can
be termed a photo-Arbuzov rearrangement. By use of phosphites
3 and 4, we were able to establish the stereochemistry of the
process both at phosphorus (retention)2 and at the migratory
carbon (primarily retention).1 This new reaction has been
successfully applied to the preparation of acyclic nucleoside-
based phosphonates, 5.3,4 Moreover, CIDNP5 and CIDEP6
In this paper we report detailed product studies of the
photorearrangements of 1 and 8 in cyclohexane, acetonitrile,
and benzene, along with quantum yields for product formation
(1) Cairns, S. M.; Bentrude, W. G. Tetrahedron Lett. 1989, 30, 1025.
(2) (a) Bentrude, W. G.; Lee, S.-G.; Akutagawa, K.; Ye, W.; Charbonnel,
Y.; Omelanczuk, J. Phosphorus Sulfur 1987, 30, 105. (b) Omelanzcuk, J.;
Sopchik, A. E.; Lee, S.-G.; Akutagawa, K.; Cairns, S. M.; Bentrude, W. G.
J. Am. Chem. Soc. 1988, 110, 6908.
(4) Mullah, K. B.; Bentrude, W. G. Nucleosides Nucleotides 1994, 13,
127.
(5) Koptyug, I. V.; Sluggett, G. W.; Ghatlia, N. D.; Landis, M. S.; Turro,
N. J.; Ganapathy, S.; Bentrude, W. G. J. Phys. Chem. 1996, 100, 14581.
(6) Koptyug, I. V.; Ghatlia, N. D.; Sluggett, G. W.; Turro, N. J.;
Ganapathy, S.; Bentrude, W. G. J. Am. Chem. Soc. 1995, 117, 9486.
(3) Bentrude, W. G.; Mullah, K. B. J. Org. Chem. 1991, 56, 7218.
10.1021/ja982440k CCC: $18.00 © 1999 American Chemical Society
Published on Web 02/26/1999

2086 J. Am. Chem. Soc., Vol. 121, No. 10, 1999
Ganapathy et al.
Table 1. Products of Photoreaction of 8 (0.016-0.030 M)
Scheme 1a
product accountability, %a
solvent
% conv. of 8
9
10 11b 12 15 16 17a 17b
CH3CN
14
25
45
65
18
25
41
55
11
28
53
77
7.5 7.8 18 13
7.2 5.4 17 9.4
7.3 6.8 16 7.5
7.5 3.5 15 7.0
7.1 9.4 13 9.8
6.9 8.6 12 9.4
7.0 6.6 12 8.3
7.0 8.8 13 7.4
CH3CN
CH3CN
CH3CN
cyclohexane
cyclohexane
cyclohexane
cyclohexane
benzene
13
10
9.3
9.3
c
c
c
c
38 14 16 8.6 8.9 9.6
a Equations are numbered in parentheses.
benzene
32 8.7 13 6.8 7.6 8.3
30 7.7 12 5.5 6.6 7.4
30 7.9 13 4.7 5.7 6.8
benzene
Scheme 2
benzene
a Based on consumed 8. b Yield of 11 doubled to account for the
stoichiometry of its formation from 8. c Not observed.
Table 2. Products of Ultraviolet Light Irradiation of 1
(0.010-0.012 M)
product accountability, %a
solvent
CH3CN
% conv. of 1
2
20b
15
14
26
59
83
14
35
62
80
14
30
46
60
81
80
68
67
71
68
64
65
62
54
57
55
12
8.4
6.2
6.8
6.2
5.6
5.6
5.8
8.6
6.0
6.0
5.8
cyclohexane
C6H6
1.4
1.5
1.5
1.5
in benzene. The low yield of phosphonate 9 from 8 and the
major effects on product distribution of added free-radical traps
PhSH, PhCH2Br and TEMPO are defined. The photolysis of 8
evidently proceeds primarily, and perhaps totally, via the
relatively long-lived triplet radical pair 6 and 7a (Ar )
p-acetylphenyl) that can be diverted by scavengers to phosphite
10, phosphorobromidate 13, and p-acetyltoluene (12) and
establish a cage yield of phosphate 9 of 3-5%. In addition the
chemistry of the phosphinoyl radical-benzene adduct, the
cyclohexadienyl radical 14, in the presence of the relatively
stable p-acetylbenzyl radical is defined (Scheme 2).
The contrasting product distributions found for phosphites 1
and 8 argue for the primarily singlet nature of the photorear-
rangement of 1. Thus, the photorearrangement of 1 gives
phosphonate 2, as the primary product, and also generates small
amounts of free-radical-derived products, for example bibenzyl
(20) and phosphite 10. This process is most simply interpreted
as proceeding through short-lived, singlet radical pair 6 and
7b (Ar ) Ph) that largely combine in the solvent cage to form
phosphonate 2. The formation of radical pairs in the photo-
chemistry of 1 and 8 finds broader relevance in the context of
the extensive recent studies of the formation of ion and radical
pairs on photolysis of arylmethyl and diaryl methyl derivatives,
including halides and esters.7
a At g 65% conversion of 1 CH3CN and cyclohexane, 4-8% of 10
was observed. b Yield of the dimer (20) was doubled to account for
the stoichiometry of its formation from 1.
countability yields of those products at various conversions of
phosphite 8 are recorded in Table 1. The yield of phosphonate
9 (the photo-Arbuzov product) is dramatically reduced from the
phosphonate yields found in previous work on the photorear-
rangements of arylmethyl phosphites including a preliminary
study of phosphite 1 (see subsequent discussion, and Table 2)
1 and 2. Clearly, radical pair 6 and 7a (Ar ) p-acetylphenyl) is
formed on photolysis of 8 and undergoes predominant diffu-
sional separation as shown (Scheme 1, eq 3). Modest yields of
products generated from radicals 6 and 7a are detectable by
GC analysis: dimethyl phosphite 10 (eq 4), dimer 11 (eq 5),
and p-acetyltoluene 12 (eq 6). Based on consumed 8, the overall
accountability of phosphinoyl radical 6 as products 9 and 10 is
11-17%. Products 9, 11 and 12 account for 27-39% of the
p-acetyl radicals potentially generated.
In addition, numerous very small identified peaks are seen
on GC analysis of the photolyzates generated in cyclohexane
and acetonitrile. They amount, however, to only 15-20% of
the total product peak area, including identified products. 31P
NMR spectra of photolyzates also display numerous unidentified
resonances. Products with known chemical shifts8 that could
result from dimerization of phosphinoyl radical 6 were not
evident: (MeO)2P(O)-(O)P(OMe)2, δ +8; (MeO)2P(O)-OP-
(OMe)2, δ 130 and -10. Photolyzate from irradiation of 8 in
acetonitrile shows a very broad peak under the resonance for
phosphonate 9 suggestive of polymer formation or a myriad of
phosphonates in small quantities. The possibility remains as well
that some dimerization of radicals 6 and 7b takes place by attack
on the phenyl ring of 7b.
Results
Photoreaction of 8. Irradiation (450 W medium-pressure UV
lamp) through a uranium filter (λ > 320 nm) of 0.016-0.030
M deoxygenated solutions of 8 in the solvents cyclohexane and
acetonitrile yielded phosphonate 9 in minor amounts (7-8%),
along with 10-12 (gas chromatography (GC) analysis). Ac-
(7) For key reviews and representative papers on the formation of radical
and/or ion pairs on photolysis of arylmethyl and diarylmethyl derivatives,
including halides and esters, see: (a) Pincock, J. A. Acc. Chem. Res. 1997,
30, 43. (b) Nevill, S. M.; Pincock, J. A. Can. J. Chem. 1997, 75, 232. (c)
Lipson, M.; Deniz, A. A.; Peters, K. S. J. Am. Chem. Soc. 1996, 118, 2992.
(d) Das, P. K. Chem. ReV. 1993, 93, 119. (e) Kropp, P. J. Acc. Chem. Res.
1984, 17, 4967. (f) Cristol, S. J.; Bindel, T. H. Org. Photochem. 1983, 6,
327. (g) Slocum, G. H.; Schuster, G. B. J. Org. Chem. 1984, 49, 2177.
(8) Levin, Ya. A.; Il’yasov; A. V. Goldfarb, E. I.; Vorkunova, E. I. Org.
Magn. Res. 1973, 5, 487.

Photo-ArbuzoV Rearrangements
J. Am. Chem. Soc., Vol. 121, No. 10, 1999 2087
Table 3. Effects of Scavengers PhSH and TEMPO on the
Photoreaction of 8 (0.014-0.018 M) at 15-20% Conversion of 8
Cyclohexadiene 16, a single diastereoisomer, was isolated by
high-performance liquid chromatography (HPLC), but 17a and
17b were isolated only as aromatized product 18. The structures
product accountability %a
1
of 16, 17a, 17b, and 18 were assigned by LRMS, HRMS, H,
solvent
CH3CN
CH3CN
CH3CN
CH3CN
cyclohexane PhSH
cyclohexane PhSH
cyclohexane PhSH
cyclohexane PhSH
C6H6
C6H6
C6H6
scavenger scavenger/8
9
10 11b 12 15
13C, and 31P NMR spectroscopy (see Experimental).
PhSH
PhSH
PhSH
PhSH
0
9.7
9
22 10
Thus, in addition to the molecular ion, the mass spectrum of
16 shows major peaks corresponding to the loss of p-acetyl-
benzyl (base peak), the loss of (MeO)2P(O), and the formation
of the (MeO)2P(O) cation. Evidence for the 1,3-cyclohexadiene
ring is seen in the NMR spectra of 16, which display four
distinctly different olefinic proton resonances (δ 5.42-5.82) and
four corresponding 13C resonances (δ 119.5-128.9) that exhibit
phosphorus-carbon couplings (5.8, 6.2, 12.4, and 12.8 Hz). The
allylic proton at the ring carbon attached to phosphorus gives a
broad multiplet at δ 2.62 which contains a predictably large
0.25
0.5
1.0
0
2.7 35
2.7 37
2.7 45
7
6
33
42
47
0
8.3
6
18
8
0
7.5
0.24
0.29
0.98
0
0.47
1.1
0
4.8 27
4.3 31
4.1 41
9.0
3.9 11
3.4 12
8.3
5.2
4.6
4.3
36
42
47
0
PhSH
PhSH
PhSH
0
31 10
13
23
20
4
0
51
53
cyclohexane TEMPO
cyclohexane TEMPO
cyclohexane TEMPO
cyclohexane TEMPO
9.4 12
9.6
0.09
0.25
2.0
3.4
3.0
0
8
4
2
6.7
2.0
0
2
(29.6 Hz) JPH coupling.10 Its allylic neighbor proton (δ 3.04),
located at the point of attachment of the p-acetylbenzyl group,
displays a multiplet that includes a 21.7-Hz coupling to
phosphorus (3JPH) in addition to 7.2- and 8.3-Hz couplings to
the adjacent diastereotopic benzyl protons. The allylic carbons
display assignable 133.6 (1JPC) and 4.1 Hz (2JPC) phosphorus-
carbon coupling constants.11 Significantly, the benzylic carbon
displays a three-bond 24.0-Hz coupling to phosphorus which,
along with the presence of the four vinylic 1H and 13C
resonances noted earlier, shows that ArCH2 and (MeO)2P(O)
are attached 1,2 rather than 1,3 on the cyclohexadiene ring. The
methoxy groups of 16 are diastereotopic in both the proton and
carbon spectra, reflecting the presence of the adjacent stereo-
genic allylic carbon. The benzylic hydrogens of 16 are also
diastereotopic and generate an AMX spectrum with a geminal
proton-proton coupling (2JHH ) -13.4 Hz) along with the
proton-proton couplings to the adjacent allylic ring proton (3JPH
a Based on consumed 8. b Yield of 11 doubled to account for the
stoichiometry of its formation from 8.
Although phosphonate 9 is generated in 9-13% yields in
benzene, certain of the photoproducts of 8 (Table 1) are very
different from those formed in cyclohexane and acetonitrile.
This results from the addition of phosphinoyl radical 6 to
benzene to form the cyclohexadienyl adduct 14 (Scheme 2), a
well-established process.9 The p-acetyltoluene (12) formed (8-
14%) almost certainly stems from the disproportionation reaction
of p-acetylbenzyl radical (7a) with 14.
Thus, when the solvent is C6D6, deuterated (CH2D) 12 is
formed exclusively (GC/MS). Notably, dimethyl phenylphos-
phonate (15) is generated in the same step and accounts for
12-16% of the phosphinoyl radicals (6) potentially generated
from 8. Cross-combination of p-acetyl radical 7a with radical
14 competes ably with disproportionation and gives two products
in combined yields of 17-26%: 16, as a single isomer, and 17
as diastereomers (17a and 17b), whose specific cis or trans
geometries have not been assigned. Together, photo-Arbuzov
product 9, phosphonate 15, and radical combination products
16 and 17 account for 39-56% of the phosphinoyl radicals (6)
potentially formed in benzene. Interestingly, no phosphite 10
was detected in benzene, perhaps because of the rapid formation
of 14.
Photolysis of 8 in C6D6 results in incorporation of deuterium,
in accordance with Scheme 2, into the molecular ions and
fragment ions (GC/MS) of not only product 12, but also 15,
16, 17a and 17b. In benzene p-acetylbenzyl radicals (7a) not
trapped as 12, 16, and 17 combine to yield dimer 11 (Scheme
1, eq 5) in increased accountability yields (30-38%) over those
generated (12-18%) in cyclohexane and acetonitrile. The
accountability of p-acetylbenzyl radicals in benzene is 38-52%,
when based solely on products 11 and 12, but totals 65-92%
when photo-ArbuzoV phosphonate 9 and radical combination
products 16 and 17 are included. (In accounting for p-
acetylbenzyl radicals, the accountability of 11 recorded in Tables
1 and 3-5 is twice the actual yield to reflect the stoichiometry
of its formation.)
Products 17a and 17b, formed randomly in roughly equal
amounts (7-10% each at 11 and 28% conversions, Table 1),
are prone to aromatization to 18 on standing at room temperature
in solution and on isolation by HPLC. Nonetheless, 31P NMR
spectra and both low and high-resolution mass spectra of 16,
17a, and 17b (GC/MS) were obtained on the photolyzate.
1
) 7.2, 8.3 Hz) seen in the H NMR spectrum of that proton.
The individual isomers 17a and 17b show the same major
GC/MS peaks as 16 except for the absence of a mass
corresponding to loss of (MeO)2P(O). The 31P chemical shifts
for 17a and 17b at δ 25.97 and 26.30 are consistent with their
allylphosphonate structures.12 Unlike 16, 17, and 17b, the
aromatized phosphonate 18 does not display significant peaks
in its mass spectrum from scission of bonds to the p-acetylbenzyl
or (MeO)2P(O) groups, but instead gives a base peak at m/z )
303 from loss of methyl and no other assignable peaks of relative
intensity greater than 2% of the base peak. The 31P NMR
resonance of 18 at δ 23.7 is representative of phenylphospho-
1
nates such as 15.13 The H and 13C NMR spectra of 18 affirm
its highly symmetrical nature and the presence of two p-
substituted benzene rings (see Experimental Section).
The single diastereomer of 16 is predicted to be the trans
isomer, formed in a transition state for coupling which avoids
steric interactions between the two large groups. Consistent with
this idea is the large value (24.0 Hz) noted earlier for the three-
bond coupling between the benzylic carbon of 16 and phos-
phorus that appears to require a very large C-C-C-P dihedral
angle.11 The simultaneously large 3JPH value (21.7 Hz) we assign
to the proton at the point of ring attachment of the arylmethyl,
however, argues for an antiperiplanar arrangement of phosphorus
and hydrogen12 attainable only with the cis isomer (Dreiding
models).
(10) Benezra, C. J. Am. Chem. Soc. 1973, 95, 6890.
(11) Thiem, J.; Meyer, B. Org. Magn. Reson. 1978, 11, 50. Neeser, J.-
R.; Tronchet, J. M. J.; Charollais, E. J. Can. J. Chem 1983, 61, 2112.
(12) See for 31P of (MeO)2P(O)CHPhdCH2. Bentrude, W. G.; Dockery,
K. P.; Ganapathy, S.; Lee, S.-G.; Tabet, M.; Wu, Y.-W.; Cambron, R. T.;
Harris, J. M. J. Am. Chem. Soc. 1996, 118, 6192.
(9) Griller, D.; Marriott, P. R.; Nonhebel, D. C.; Perkins, M. J.; Wong,
P. C. J. Am. Chem. Soc. 1981, 103, 7761.
(13) Rozinor, V. G.; Pensionerova, G. A.; Glukhikh, V. I.; Grechkin, E.
F. J. Gen. Chem. USSR 1976, 46, 1840 (English).

2088 J. Am. Chem. Soc., Vol. 121, No. 10, 1999
Ganapathy et al.
Table 5. Effects of Added PhCH2Br as Scavenger on the
Table 4. Effects of Added PhCH2Br as Scavenger on the
Photoreaction of 8 (0.015 M) at 15-18% Conversion in CH3CN
and Cyclohexane
Photoreaction of 8 (0.015 M) at 15-18% Conversion in C6H6
solvent PhCH2Br/8 11 12 13 15 16 17a 17b 20 21 22 23
12 38 12 0 12 8.9 8.9 9.3 0 0
9
product
C6H6
C6H6
C6H6
C6H6
C6H6
C6H6
0
3.7
7.4
20
30
51
0
0
accountability %a,b
10 38 0 38 4.5 2.6 2.7 2.9 13 23 1.1 2.7d
solvent
CH3CN
CH3CN
CH3CN
CH3CN
CH3CN
CH3CN
cyclohexane
cyclohexane
cyclohexane
cyclohexane
cyclohexane
PhCH2Br/8
2
9
11 12 13 20 21
7.0 36 0 43
5.3 32 0 83
4.5 34 0 84
4.5 34 0 95
0
0
0
0
1.7 1.9 1.9 17 26 1.1 2.5
c
c
c
c
c
c
c
c
c
22 26 c
20 24 c
24 26 c
c
c
c
0
0.5
2
7.6 17 10
0
0
0
3.8
3.5
3.9
3.6
1.7
16 42
8.7 30
7.1 36
4.0 32
3.5 32
7.2 16 8.6
17 32 0.5 61 18 21
11 30
7.5 24
5.0 22
0
0
0
0
0
67 19 24
68 19 22
69 26 28
76 28 23
79 32 23
0
8
a Based on consumed 8. b Yields of 11 and 20 doubled to reflect
stoichiometry of formation from 8. c <1% yield. d Isomers 23a and 23b
are present in approximately equal amounts (31P NMR).
15
23
0
0.52
4.1
19
46
4.1
2.6
2.0
1.9
tions of free radical scavengers PhSH (0.0035-0.017 M), benzyl
bromide (0.007-0.70 M), and TEMPO (0.0016-0.034 M).
Thiophenol. Even at low concentrations (PhSH/8 ) 0.3-
1.1), thiophenol diverts p-acetylbenzyl radicals (7a) to p-
acetyltoluene (12) (highest 12 accountability yields: 47%,
acetonitrile; 53%, benzene) to the exclusion of p,p′-diacetyl-
bibenzyl (11) and radical combination products 16 and 17
(Scheme 1, eq 6, PhSH as H-donor). The same effect is seen in
cyclohexane even at PhSH/8 ) 0.29 (yield of 12, 42%).
Phosphonate 9 accountabilities are leveled out in all three
solvents at 3-4% at 0.5-1.0 ratios of PhSH/8. Phosphinoyl
radical 6 is increasingly trapped (eq 4, PhSH as H-donor) as
dimethyl phosphite (10) as the amount of added PhSH increases
(maximum yields in acetonitrile and cyclohexane: 45 and 41%,
respectively). The reasonably good yields of 10 and p-
acetyltoluene (12) in acetonitrile and cyclohexane found with
PhSH added are especially significant, as the accountabilities
of radicals 6 and 7a (Ar ) p-MeCOC6H4) are low in these
solvents in the absence of a scavenger. The yield of 10 in
benzene, however, at a 1.1 ratio of PhSH/8 is seen to be only
12%. Moreover, phenylphosphonate 15 (Scheme 2) is formed
in increased amounts on PhSH addition and persists in 20%
yield at 1.1 PhSH/8 (Table 3). Even at PhSH/8 ) 5.7 (0.09 M
PhSH concentration; data not given in Table 3), a 10% yield of
15 was measured.
Under photolysis conditions, PhSH gives rise to PhSSPh and
side products 29a, 29b, and 29c, typical of the known14 thermal
Arbuzov-like reaction of PhSSPh with phosphites. Products 29a
and 29b also could result from combination of phenylthiyl
radicals with radicals 6 and 7a. (GC, GC/MS, and 31P NMR
evidence; see Supporting Information for structures 29a-29c,
their independent preparation, and characterization). The reaction
of PhSSPh with 8 was confirmed in acetonitrile by a dark-
reaction control and precluded the use of high concentrations
of PhSH, even at low photochemical conversions of 8. Thus,
the 10% yield of phosphonate at 0.09 M PhSH, given in the
previous paragraph, is somewhat low. Indeed, the less-than-
quantitative accountabilities of radicals 6 and 7a of Table 3 are
in part a result of this side reaction of 8 with PhSSPh or
scavenging by phenylthiyl radical.
0
0
0
69 24 23
71 22 21
84 24 23
a Based on consumed 8. b Yield of 11 and 20 doubled to account
for stoichiometry of formation from 8.
Photoreaction of 1. Thoroughly deoxygenated 0.010-0.012
M solutions of phosphite 1 in cyclohexane, benzene, and
acetonitrile were irradiated through quartz with 254 nm UV
light. Identified products formed are recorded in Table 2 as a
function of phosphite consumed (GC analysis). Accountabilities
of reacted phosphite in terms of product phosphonate 2 (eq 1)
demonstrate a small decrease with increased phosphite conver-
sion. However, irradiation of 2 shows it to be photostable over
time periods comparable to those required for rearrangement
of 1 to 2. Notable are the good accountability yields of 2 (e.g.,
67-81% at 14-83% conversion in acetonitrile) along with
modest accountabilities of benzyl radicals (6-12%) as bibenzyl
(20), eq 7. This is in marked contrast to the radical diffusion
•
PhCH CH Ph
2PhCH2 f
(7)
2
2
20
products that dominate the photochemistry of phosphite 8 (Table
1). Unfortunately, toluene (19), potentially formed in minor
amounts (eq 8), eluted with solvent under the GC conditions
• PhCH H
+H
8
(8)
2
19
and could not be assayed. Phosphite 10 becomes detectable at
conversions of 65% and above in 4-8% yields in acetonitrile
and cyclohexane, but not in benzene. These accountabilities of
10 are somewhat less than those seen from photolysis of 8 even
at low conversions of 8 (Table 1).
Although 20-30% of the phosphorus of phosphite 1 remains
unaccounted for in acetonitrile and cyclohexane, the 31P NMR
spectra of photolyzate solutions run at high signal-to-noise
showed the absence of any but very small peaks in addition to
6. GC analysis showed minute peaks at long retention times. It
was noted earlier that a large number of unidentified products
are generated in small quantities in these solvents on photolysis
of 8. From 8 approximately 95% of the initial pairs diffuse from
the solvent cage, and only a few percent recombine randomly
to form 9. In acetonitrile and cyclohexane, only bibenzyl (20)
and dimethyl phosphite (10) are identifiable. The failure to
account for the missing radicals 6 and 7b from 1, therefore, is
not surprising. The possibility of formation of polymers or other
high-boiling side products from both 1 and 8 remains.
A possible photochemical side-product of phosphonate 9 in
the presence of PhSH is the alcohol p-MeCH(OH)C6H4CH2P-
(O)(OMe)2, potentially formed on reduction of the carbonyl
functionality. Although GC/MS evidence for the generation of
this product was obtained on extended irradiation of phosphonate
9 with PhSH, it was not formed (GC) under the reaction condi-
tions used for the photorearrangements of 8 with PhSH added.
Benzyl Bromide. When added in sufficient amounts (PhCH2-
Br/8 g 2 in CH3CN and cyclohexane; PhCH2Br/8 g 7.4,
benzene) to the photoreactions of 8 in all three solvents, benzyl
Scavenger Studies. Recorded in Tables 3-5 are representa-
tive data showing the effects on the distribution and yields of
products from photolysis of 8 of adding a range of concentra-
(14) Harvey, R. G.; Jacobson, H. J.; Jensen, E. V. J. Am. Chem. Soc.
1963, 85, 1618.

Photo-ArbuzoV Rearrangements
J. Am. Chem. Soc., Vol. 121, No. 10, 1999 2089
bromide diverts phosphinoyl radicals (6) that escape geminate
recombination with 7a (Ar ) p-acetylphenyl) to phosphoro-
bromidate 13 via a known reaction (eq 9)15 in very good
of 16, 17a and 17b (see Experimental Section). Unlike the cis/
trans isomeric 17a and 17b, only a single peak in the GC was
observed for the benzyl counterparts 23a and 23b that do,
however, display different 31P chemical resonances (δ 27.72,
27.68) in approximately equal intensities. (The available Sup-
porting Information gives an expanded version of Table 5 at
nine ratios of PhCH2Br/8.)
accountabilities: 69-84% in cyclohexane and acetonitrile; 43-
95% in benzene (Tables 4 and 5). (Photoirradiations were run
with light from a 450 W medium-pressure UV lamp filtered
through a uranium glass filter to remove light of wavelengths
shorter than 320 nm and preclude any photolysis of benzyl
bromide, even at high concentrations.) Even at low concentra-
tions of PhCH2Br, dimethyl phosphite (10), normally seen in
cyclohexane and acetonitrile, is absent. In benzene at concentra-
tions greater than 0.1 M (PhCH2Br/8 g 7.4), PhCH2Br reduces
the yield of PhP(O)(OMe)2 (15) to zero. Indeed, even at modest
PhCH2Br/8 ratios, the accountability of phosphinoyl radical 6
as phosphorobromidate 13 is greatly improved over that found
when PhSH is used to divert 6 to phosphite 10 (Scheme 1, eq
4). (See the yields of (MeO)2P(O)H (10) given in Table 3: 41
and 45% in cyclohexane and acetonitrile, respectively; PhSH/8
) 1.0). Interestingly, low concentrations of added benzyl
bromide in cyclohexane and acetonitrile increase the yield of
phosphonate 9 from 7 to 8% to 16-17%.
Comparisons of the two scavengers show that both PhSH
and PhCH2Br in sufficient amounts reduce the yield of phos-
phonate 9 to 3-5%, which presumably corresponds to the yield
from geminate radical pair combination. Interestingly, the
required amount of PhCH2Br (PhCH2Br/8 ) 23-51) in all three
solvents is much greater than the quantity of PhSH (PhSH/8 )
0.5-1.0) needed to prevent the out-of-cage formation of
phosphonate 9. By contrast, phenylphosphonate, (15) undoubt-
edly formed via reaction of cage-escape phosphinoyl radicals
(6) with benzene (Scheme 2), disappears at lower benzyl
bromide concentrations (PhCH2Br/8 ) 7.4) than are required
to minimize the yield of phosphonate 9 at 4-5% (PhCH2Br/8
) 30). Indeed, as noted earlier, a 10% yield of phosphonate 15
persists at PhSH concentrations (PhSH/8 ) 5.7) much above
those required to reduce the yield of 9 to the 3-5% range. These
differences between the two scavengers are addressed in the
Discussion section.
TEMPO. The addition of TEMPO in amounts similar to
those for added thiophenol (TEMPO/8 ) 0.09-2.0) in cyclo-
hexane causes the yield of 9 to level out at 4-5%, providing a
third measure of the cage yield of 9 (Table 3). Concentrations
of TEMPO were kept very low so as to minimize the “anti-
scavenger” effect noted by Turro et al.16 Simultaneously, the
yields of (MeO)2P(O)H (10) and p-acetyl toluene (12) drop to
zero. Radical coupling product 25 was isolated from photolysis
of 8 in the presence of added TEMPO in acetonitrile solution
and characterized by HRMS, and by comparison of its spectral
properties to those previously reported for the benzyl analogue.17
However, the formation of dimer 11 at the ratio TEMPO/8 )
2.0 is not totally eliminated, though its accountability yield
reduced from 12% to 2%. The photolysis of the benzyl analogue
of trapping product 25 (formed by radical coupling) has been
reported.17b Photolysis of 25 would yield p-acetylbenzyl radicals
(7a) and provide a secondary pathway for reformation of dimer
11 under TEMPO-trapping conditions.
Under benzyl bromide-scavenging conditions in cyclohexane
and acetonitrile, the yield of p-acetyl toluene is reduced to zero
at relatively low concentrations of the bromide (PhCH2Br/8 g
0.5). p-Acetylbenzyl radicals couple to form increased quantities
of dimer 11 and also scavenge benzyl radicals, generated by
reaction 9, to form the cross-coupling product 21. The account-
p-MeCOC6H4CH2CH2C6H5
21
ability of p-acetylbenzyl radicals, taking into account the
stoichiometry of product formation, is increased even at PhCH2-
Br/8 ) 0.5 and is in the range of 52-82% over the range of
PhCH2Br concentrations used (Table 4).
The benzyl radicals (7b) formed in reaction 9 lead, in
cyclohexane and acetonitrile, to 2-4% of the radical coupling
product dimethyl benzylphosphonate (2, eq 10), bibenzyl (20,
6 + 7b f 2
(10)
eq 7), and cross-coupling product 21. Bibenzyl (20) and cross-
dimer 21, but not 2, are formed in benzene (Table 5). In benzene
benzyl radicals also trap intermediate 14 to give the benzyl
analogues (22, 23a, and 23b) of 16, 17a, and 17b in 1-2%
yields each. (Product 24 not formed under GC conditions.)
Accountabilities of p-acetylbenzyl radicals are 46-62% in
benzene. At relatively high concentrations of benzyl bromide
(PhCH2Br/8 g 51, Table 5), the only remaining phosphorus-
containing material in benzene is phosphonate 9 (4-5%), which
is accompanied in cyclohexane and acetonitrile by 2-4% of
benzylphosphonate 2.
Furthermore, a peak at δ 6.7 in the 31P NMR spectrum (CD3-
CN) of the crude photolyzate is assigned to 26. This is consistent
The structures of 22, 23a, and 23b were confirmed by the
with the literature value for 27 (δ31P ) 4.9, acetone-d6).18
Attempts to isolate 26 by liquid chromatography failed. The
(16) Step, E. N.; Buchachenko, A. L.; Turrow, N. J. J. Am. Chem. Soc.
1994, 116, 5426.
(17) (a) Johnston, L. J.; Tencer, M.; Scaiano, J. C. J. Org. Chem. 1986,
51, 2806. (b) Korolenko, E. C.; Cozens, F. L.; Scaiano, J. C. J. Phys. Chem.
1995, 99, 14123.
(18) Busfield, W. K.; Grice, I. D.; Jenkins, I. D. Aust. J. Chem. 1995,
48, 625.
close similarity of their GC/MS fragmentation patterns to those
(15) Anpo, M.; Sutcliffe, R.; Ingold, K. U. J. Am. Chem. Soc. 1983,
105, 3580.

2090 J. Am. Chem. Soc., Vol. 121, No. 10, 1999
Ganapathy et al.
thermal instability of 26 precluded its study by GC/MS.
Convincing evidence for the formation of 26 came from
photolysis of di-tert-butyl peroxide in a solution of TEMPO in
dimethyl phosphite (10). Observed in the 31P NMR spectrum
of the crude products was a very predominant peak with
chemical shift (δ 31P ) 6.1, C6D6 ) close to that seen from the
8/TEMPO photolysis. Undoubtedly, radical 6 is formed by tert-
butoxy radical abstraction of hydrogen from 10 and is then
trapped by TEMPO to form 26. Indeed, 27 was previously
prepared by the same approach.18
Phosphite 1. Both PhSH and benzyl bromide effectively
scavenge radicals from photolysis of 8 that escape geminate
combination (Tables 3-5). However, the necessary use of UV
light of relatively short wavelengths (254 nm) for the photo-
rearrangement of 1 results in the photolysis of added benzyl
bromide.7 Indeed, a control irradiation of an acetonitrile solution
of benzyl bromide in the presence of (MeO)3P generated major
amounts of phosphorobromidate 13, presumably from reaction
of bromine atoms with (MeO)3P and subsequent loss of methyl
radical by â scission. Use of tert-BuBr, which is an effective
scavenger15 of phosphinoyl radical 6 formed on direct irradiation
(>300 nm) of dimethyl 1-naphthylmethyl phosphite,19 also is
precluded by the appreciable UV absorption and reaction of
tert-BuBr at 254 nm.
state. (Indeed, its UV spectrum is closely similar to that of
acetophenone.) The quantum yields for product formation
indicate that photoreaction occurs quite efficiently (ΣΦi ) 0.47).
The low cage yield of phosphonate 9 (3-5%) is consistent with
the initial formation of triplet radical pairs (6 and 7a, Ar )
p-acetylphenyl), most of which undergo diffusion rather than
combination. A fraction of them (4-8%) undergo nongeminate
recombination as random free pairs to raise the total account-
ability yield of phosphonate 9 to 7-13%. As will be illustrated
with examples later in the paper, the low percentage random
recombination of 6 and 7a, in which only one is a relatively
stable species, is quite normal.
The cage-escape radicals 6 and 7a (Ar ) p-acetylphenyl) from
photolysis of 8 are trapped by radical scavengers, PhSH and
benzyl bromide (eqs 4, 6, 9), to yield phosphorobromidate 13,
phosphite 10, p-acetyltoluene (12) and cross-combination
product 21. With PhCH2Br as a scavenger in cyclohexane and
acetonitrile (Table 4), the sum of the yields of dimer 11 and
cross coupling product 21, if stoichiometry is considered,
account for 45-66% of the p-acetylbenzyl radical potentially
formed from 8 (Table 4). Including phosphonate 9, p-acetyl
radical accountabilities are 52-82%. (The increase in yield of
phosphonate 9 in acetonitrile and cyclohexane upon addition
of benzyl bromide in relatively low concentrations (PhCH2Br/8
≈ 0.5) is not readily rationalized.) Similarly, in the presence of
PhSH as a scavenger, p-acetyltoluene accountabilities as high
as 53% (benzene) attest to the large numbers of radical pairs
generated by irradiation of phosphite 8. Added TEMPO yields
scavenging products 25 and 26 from interception of p-acetyl
benzyl (7a) and phosphinoyl (6) radicals, respectively. Indeed,
the good yields of radical trapping products with PhCH2Br and
PhSH as scavengers confirm that the reaction involves formation
of relatiVely long-liVed, presumably triplet radical pairs (6 and
7a), and that geminate recombination to form phosphonate 9
is a minor pathway. All three radical scaVengers point to a
cage yield of phosphonate 9 of 3-5%.
The apparent 3-5% geminate recombination of initial pairs
6 and 7a (Ar ) p-acetylphenyl) to give phosphonate 9
potentially could include some product from rapid combination
of singlet pairs formed from the singlet excited state of 8 in
competition with singlet-triplet intersystem crossing. Ben-
zophenone undergoes singlet-triplet intersystem crossing (kISC
) 1010 s-1) about ten times more rapidly than does acetophe-
none,22 to which phosphite 8 is closely related. A study of the
cage yield of the p-benzoylbenzylphosphonate, from irradiation
of the benzophenone analogue of 8, could be informative.
Benzene solvent itself is a scavenger of phosphinoyl radical
6, as shown by the products formed (Scheme 2). This results in
even higher accountabilities of radicals from the presumed triplet
pair 6 and 7a (Ar ) p-acetylphenyl) in benzene compared to
cyclohexane and acetonitrile, as seen in Table 1 and noted in
the Results section. The trapping of key radicals from the
photolysis of 8 in benzene will be discussed in more detail in
later paragraphs.
Unfortunately, phenyl mercaptan also is an inappropriate
scavenger of radicals from 1 as it displays strong ultraviolet
absorption at 254 nm. Likewise, TEMPO absorbs strongly at
this wavelength.
Benzene as Radical Scavenger in the Photorearrangement
of 1. Since solvent benzene was shown to be a very effective
free-radical scavenger in the photolysis of 8 (Scheme 1), we
looked for phosphonate 15 and products 22 and 23 on direct
irradiation of 1 in benzene. Indeed, the formation (Table 2) of
1-2% of phenylphosphonate 15 was confirmed by GC/MS and
quantitated by GC. Furthermore, the generation of 22, 23a, and
23b in total amounts 1-2% (GC detection and quantitation)
was indicated by 31P spectroscopy (δ31P ) 30.79, 27.72, 27.68)
of the crude photolyzate. Products 23a and 23b were seen in
near-equal amounts as estimated from their 31P NMR resonances
at δ 27.22 and 27.68. (See Experimental Section and formation
of these products on photolysis of 8 in the presence of benzyl
bromide discussed earlier.)
Quantum Yields for Photolysis of 1 and 8. The quantum
yields for the photoreactions of 1 (cyclohexane) and 8 (benzene),
at 5-10% conversion of phosphite, were determined by
irradiation at 266 nm (phosphite 1) or 335 nm (phosphite 8)
with light from the high-pressure UV lamp of a PTI Quantacount
electronic actinometer.20 The quantum yield for formation of
phosphonate 2 (ΦP) is 0.43. For 8 ΦP is 0.074, while the
quantum yields for both 11 and 15 formation are 0.11.21 Taken
together the total quantum yield for the three radical combination
products 16, 17a, and 17b is 0.18. (The quantum yields for
formation of 11 are based on its chemical yield and not doubled
to account for the stoichiometry of p-acetylbenzyl radical
dimerization to give 11.) The total quantum yield (ΣΦi) for all
products accounted for from the photoirradiation of 8 is 0.47.
It should be clearly noted that not one of the three scavengers
(benzyl bromide, PhSH, or solvent benzene) that give identifi-
able, quantitated products by itself gives maximum account-
abilities of both radicals 6 and 7a. However, use of all three
allows accountabilities of greater than 90% to be obtained for
both radicals.
Discussion
Photolysis of Phosphite 8. Phosphite 8, an acetophenone
derivative, is ideally set up for reaction via its triplet excited
Scheme 2 is totally consistent with previously reported
CIDEP6 and 31P CIDNP5 work on the photolysis of phosphite
(19) Ganapathy, S.; Soma Sekhar, B. B. V. Unpublished results from
this laboratory.
(20) The quantum yield determinations were described previously.6
(21) Previously reported quantum yields for formation of 11 and 156
have been corrected to reflect repeated determinations of product yields.
(22) Turro, N. J. Modern Molecular Photochemistry; Benjamin/Cum-
mings Publishing: Menlo Park, CA, 1971; Chapter 8.

Photo-ArbuzoV Rearrangements
J. Am. Chem. Soc., Vol. 121, No. 10, 1999 2091
8. Thus, the predominance of initial triplet radical pairs was
revealed by the E*/A (enhanced emission) doublet pattern of
the CIDEP spectrum for the phosphinoyl radical 6 from direct
irradiation of 8. For the p-acetylbenzyl radical, appropriate net
emissive polarization was seen, on which was superimposed
an E/A pattern.6 Furthermore, 31P CIDNP polarizations for
products 10, 13, and 15-17 conformed to those expected from
the reactions of initially formed triplet radicals.5 A classic field
effect on the sign of the polarization of (MeO)2P(O)Br (13),
formed from radical 6 following its escape from the initial
solvent cage, also was noted.6
the products 15, 22, and 23 from addition of phosphinoyl radical
6 to benzene (Scheme 1 with radical 7b in place of 7a).
However, we could not find a triplet quencher with suitable
energetics and optical properties that was not also a radical trap.
The unsuitability of PhSH and the tert-butyl and benzyl
bromides as scavengers of radical intermediates from photolysis
of 1 is particularly disappointing as their use presumably would
have further established the presence of cage-free radicals from
irradiation of 1. Thus, in results as yet unpublished,19 the direct
irradiation of 0.01-M solutions of a related phosphite, dimethyl
1-naphthylmethyl phosphite, was seen to yield the corresponding
1-naphthylmethylphosphonate in yields comparable to those for
2 seen in the present study. As with phosphite 1, no CIDEP or
31P CIDNP phenomena were obserVed.23 Scavengers give strong
evidence for the formation of radical pairs. With the 1-naphthyl
phosphite,19 addition of tert-BuBr in increasing quantities
reduced the accountability yields of phosphonate (15-17%
conversions) from 71% to 61% in cyclohexane and from 74 to
69% in benzene. These reductions are accompanied by formation
of 10-15% of (MeO)2P(O)Br (13). Added PhSH increases
yields of 10 and 1-methylnaphthalene and reduces to zero the
yield of the 1-naphthylmethyl radical dimer.
Photolysis of Phosphite 1. By contrast to 8, however, the
product distribution from photo reaction of 1 reflects the
predominant formation of the photo-Arbuzov product, phos-
phonate 2, in high quantum efficiency (ΦP ) 0.43). Nonetheless,
a small percentage yield (3-6%) of cage escape product
bibenzyl (20) from dimerization of radical 7b (Ar ) Ph), eq 7,
is noted. This corresponds to a 6-12% accountability (Table
2) of potentially formed benzyl radicals (7b). The 4-8% yield
of dimethyl phosphite (10), detectable at higher conversions of
1 (see Results), probably arises from phosphinoyl radical 6 (eq
4). Suggestive of the formation of cage-free phosphinoyl radical
6 and its trapping by solvent benzene is the reduced yield of
phosphonate 2 in benzene at 14% conversion of 1, compared
to its yield in the other two solvents. Both 31P NMR and GC/
MS confirmatory evidence for cage-free 6 is seen in the
formation in benzene of phenylphosphonate 15 (1-2%), along
with small amounts of products 22 and 23 (Scheme 2 with
Finally, the possibility that phosphonate 2 arises via a chain
reaction involving reaction of benzyl radicals with phosphite 1
via phosphoranyl radical 28 (eq 11) must be addressed. Contrary
•
•
PhCH2 (7b) in place of p-MeCOC6H4CH2 (7a)).
On the basis of the relatively minor amount of radical
diffusion that accompanies the photorearrangement of 1 to 2,
this process is most simply understood in terms of reaction via
the singlet excited state of 1 to generate singlet radical pairs 6
and 7b (Ar ) Ph). These pairs are relatively short-lived and,
therefore, primarily undergo geminate combination to give
photo-Arbuzov rearrangement product 2 [(1 f (6, 7b) f 2].
The failure to observe CIDEP or 31CIDNP signals on direct
irradiation of 123 also is in accord with the postulation that the
radical pairs involved are very short-lived. This is in contrast
to the important CIDEP and 31P CIDNP effects seen from the
predominantly triplet pairs formed on direct irradiation of 8 or
triplet-sensitized photoreaction of the 1-naphthylmethyl analogue
of 1.5,6,23
Further evidence for the formation of radical pairs in the direct
photolysis of 1 and related phosphites comes from research from
this laboratory on 3 and the analogue of phosphite 3 with a
methyl substituent at the ring carbon next to oxygen.24 The
photo-Arbuzov rearrangement of essentially a single enantiomer
of the diastereomer with ring methyl and 1-phenylethoxy ring
substituents oriented in cis fashion was carried out. X-ray
crystallograpy and 31P NMR spectroscopy demonstrated that
approximately 20% of the photo-Arbuzov product phosphonate
was formed with inverted stereochemistry at the carbon stereo-
genic center of the 1-phenylethyl bonded to phosphorus.
evidence comes from the quantum yield for formation of 2 of
less than one. Moreover, a crossover study, reported in our
earlier communication,1 excluded such a mechanism. Further-
more, in a recent paper25 we reported the failure of benzyl
radicals, formed on â scission of phosphoranyl radical [Et-
(PhCH2O)P(OMe)2]•, to react with PhCH2OP(OMe)2. Benzyl
radicals bond too weakly to phosphites (eq 11a) to generate
28.
Scavenging of Radical Pairs from 8 in Benzene by
PhCH2Br. In benzene at lower conversions of 8 in the absence
of scavengers, as high as 56% of the phosphinoyl radicals (6)
potentially formed are accounted for in products 9, 15, 16, 17a,
and 17b (Table 5). Likewise, if corrections are made for reaction
stoichiometry, products 9, 11, 12, 16, 17a, and 17b account for
up to 92% of theoretically formed radical 7a (Ar ) p-
acetylphenyl). Clearly, benzene as solVent is a highly effectiVe
trap for radicals from 8.
The product and radical scavenger studies in benzene are in
general accord with and support the reactions of Scheme 1 and
merit discussion in more detail. The absence of p-acetyltoluene
(12) at a benzyl bromide/8 ratio of 3.7/1 is accompanied by a
large reduction, from 12 to 4.5%, in the yield of phenylphos-
phonate 15. The latter product presumably is formed in the
disproportionation reaction of 14 with 7a that also generates
12 (Scheme 1). These results are consistent with much reduced
formation of 14 when radical 6 is diverted by increasing amounts
of PhCH2Br to form 13. Further addition of benzyl bromide
(benzyl bromide/8 ) 7.4, Table 5) brings the yield of phos-
phonate 15 to zero. At that concentration of benzyl bromide,
the accountability of radical 6 as the phosphorobromidate (13)
is 43%. Though phosphonate 15 is no longer detected at a
PhCH2Br/8 ratio of 7.4/1, the total yields of coupling products
A combination of concerted and free-radical pair pathways
for the formation of 2 cannot be ruled out. However, there is
no evidence for it. There remains as well the possibility that a
very limited portion of the excited singlet of 1 undergoes
intersystem crossing to the molecular triplet, followed by
generation of triplet radical pairs that are at least in part
responsible for the radical diffusion product bibenzyl (20) and
(23) Sluggett, G. W.; Landis, M. S.; Turro, N. J. Unpublished results.
(24) Bhanthumnavin, W.; Arif, A.; Bentrude, W. G. J. Org. Chem. 1998,
63, 7753.
(25) Dockery, K. P.; Bentrude, W. G. J. Am. Chem. Soc. 1997, 119,
1388.

2092 J. Am. Chem. Soc., Vol. 121, No. 10, 1999
Ganapathy et al.
16, 17a and 17b, which also must arise from reaction of
p-acetylbenzyl radical with 14, still are 5-6% (Table 5). This
is consistent with the dominance of coupling products 16, 17a,
and 17b (27% total) compared to 15 (16%) at 11% conversion
of 8 in the absence of PhCH2Br (Table 1). Nonetheless, these
products essentially disappear (<1% of each) at PhCH2Br/8
ratios above 9.7/1, at which point the phosphonate 9 yield (5.3%)
is close to the fully minimized 4.5% yield of 9 encountered at
30/1 and higher ratios of PhCH2Br/8 (Table 5). At high PhCH2-
Br/8 ratios (20/1-51/1), the accountability of 6, as measured
by formation of 13, rises to 83-95%. The ratio of percentage
yields of 11/21/20 at these ratios of PhCH2Br/8 is about 16/
26/12 (not corrected for stoichiometry of formation of 11 and
20), reasonably close to the ratio expected for random combina-
tion of benzyl and p-acetylbenzyl radicals present in nearly equal
concentrations.
is 1.5 × 107 M-1 s-1 at 23 ( 2 °C,26 while that for removal of
bromine atom from benzyl bromide is smaller, 5.9 × 106 M-1
s-1 27
However, with PhSH the reduction in the yield of
.
phosphonate 9 from combination of cage-free radicals 6 and
7a (Ar ) p-acetylphenyl) is the result of the simultaneous
trapping of both 6 and 7a. Indeed, at a PhSH/8 ratio of only
(0.25) in cyclohexane and acetonitrile, the yields of 11, formed
by dimerization cage-free p-acetylbenzyl radicals (7a), are
greatly reduced with concomitant large increases in amounts
of the trapping products 10 and 12. The effect reaches a
maximum at a PhSH/8 ratio of 1.0 when all p-acetylbenzyl
radicals are diverted to 12. Indeed, it is the simultaneous remoVal
of both the phosphinoyl (10) and p-acetylbenzyl (7a) radicals
that is primarily responsible for the much greater efficiency of
PhSH, compared to PhCH2Br, in reduction of the yield of 9 in
cyclohexane and acetonitrile to the 3-5% leVel.
In benzene similar effects on the yields of phosphonate 9
are seen (Table 3), except that the scaVenging of 6 by PhSH to
form phosphite 10 at PhSH/1 ) 1 is less efficient than in
cyclohexane and acetonitrile. This is revealed by the yield of
10 of only 12% and the remaining 20% yield of phenylphos-
phonate 15, both of which incorporate the (MeO)2P(O)• radical
(6), even though the yield of phosphonate 9 has been reduced
to 3-4%. Thus, it would appear that the rate of abstraction of
hydrogen from PhSH by phosphinoyl radical 6 is slower than
its rate of addition to benzene to form 14 and the products
derived from it. Unfortunately, the rate constant for abstraction
of hydrogen from PhSH by (MeO)2P(O)• is not known. As noted
in the previous paragraph, that for hydrogen abstraction by Ph2P-
(O)• is 1.5 × 107 M-1 s-1 at 23 ( 2 °C.26 This is only about
2.5 times greater than the rate constant (5.9 × 106 M-1 s-1) for
abstraction of bromine from PhCH2Br by Ph2P(O)•. If the
relative rate constants for (EtO)2P(O)• and Ph2P(O)• for reaction
with PhSH and PhCH2Br are similar, then for (EtO)2P(O)• (and
presumably (MeO)2P(O)•) the rate constant for abstraction from
PhSH should be about 2.5 times the known rate constant15 for
reaction of (EtO)2P(O)• with PhCH2Br (1.2 × 106 M-1 s-1) or
3 × 106 M-1 s-1. The pseudo first-order rate constants for
formation of adduct 14 and abstraction by radical 6 of hydrogen
from PhSH, at the concentrations of PhSH (0.015 M) and
benzene (≈ 13 M) used, are calculated to be approximately
equal ((4-5) × 104s-1). Thus, the continued formation of
phosphonate 15 via adduct 14 at PhSH/8 ) 1.0 ([PhSH] ) 0.015
M) and above, in competition with the trapping of phosphinoyl
radical 6 by its conversion to phosphite 10, is not surprising.
All p-acetylbenzyl radicals (7a) are scavenged as p-acetyltoluene
(12) at PhSH/8 ) 1 and below to turn off the formation of
phosphonate 9 by random encounter of radicals 6 and 7a (Ar
) p-acetylphenyl). This, along with the formation of 14, reduces
the yield of 9 to 3%.
The concentrations of benzyl bromide scavenger required to
reduce the yield of phosphonate 9 to the 3-5% level (apparent
cage yield), by trapping phosphinoyl radical 6, are close to what
might be predicted. Thus, the known rate constant for addition
of (EtO)2P(O)• to benzene is 2.9 × 103 M-1 s-1 at 25 °C9 and
for reaction with benzyl bromide is 1.2 × 106 M-1 s-1 at
ambient temperature.15 Both rate constants are presumably very
close to those for radical 6 (eq 9 and formation of 14, Scheme
2). Therefore, at 0.015 M concentration of benzyl bromide
(PhCH2Br/8 )1) in solvent benzene (benzene ≈ 13 M), the
addition of (C2H5O)2P(O)• to benzene can then be predicted to
be about two times as rapid as abstraction of bromine from
benzyl bromide. At a PhCH2Br/8 ratio of (4-8)/1 (Table 4),
the rate of trapping of 6 by benzyl bromide (eq 9) should be
2-4 times faster than its rate of addition to benzene. The
accountability yield of phosphonate 9 is reduced to 7% at
PhCH2Br/8 ) 8. A 20/1 ratio is required to decrease the yield
of 9 to 5.3. Ratios greater than 25/1 bring the yield of 9 into
the 3-5% range, in which all 6 that escape the solvent cage
have been scavenged. At PhCH2Br/8 ) 25, the calculated ratio
for the rate of bromine abstraction to that for addition to benzene
is 12.
In keeping with Scheme 1, the formation of benzyl radicals
in the PhCH2Br trapping experiments in benzene (eq 9) results
in the generation (eq 7) of bibenzyl, 20. Benzyl radicals also
participate in the reactions of Scheme 2 to form the benzyl
analogues of adducts 16, 17a, and 17b and the products 22 and
23 (Table 5). As the PhCH2Br/8 ratio is increased gradually
form 0.5 to 7.4, the bibenzyl accountability yield (corrected for
stoichiometry) increases accordingly from 4% to 34% (Table
5). In keeping with Scheme 1, the yields of cross dimer 21 rise
from 10 to 26%. Perhaps surprisingly, the products 22 and 23
from benzyl radical trapping of 14 remain constant at 3-4%
total yield up to a PhCH2Br/8 ratio of 7.4 (Table 5).
Adduct 14 may indeed be trapped by disproportionation with
phenylthiyl radicals to give 15 directly and perhaps also by
abstraction of hydrogen from PhSH to form the cyclohexadiene
product which is subsequently aromatized to 15. The high
efficiency of modest concentrations of PhSH in benzene in
reducing the formation of phosphonate 9 results primarily from
its trapping of the p-acetylbenzyl radical (7a) and only
secondarily from its interception of radical 6. In this way the
inefficiency of PhSH in reducing the yield of phenylphosphonate
15 is readily understood. If in fact the formation of 14 is
Scavenging by PhSH of Radical Pairs from Photolysis of
8 in Benzene and Other Solvents. The amount of thiophenol
required to scavenge all of the cage escape radicals 6 and 7a
(Ar ) p-MeCOC6H4) in acetonitrile and cyclohexane and reduce
the yield of phosphonate 9 to 3-5% is much less than the
quantity of benzyl bromide required to haVe the same effect
(Tables 3 and 4). This great a difference would not be expected
if the primary reaction responsible for reduction of the yield of
9 is the trapping by PhSH of phosphinoyl radical 6 (eq 4).
Indeed, on the basis of model reactions, PhSH should be the
better trap for 6, but only marginally so. Thus, for the related
radical Ph2P(O)•, the rate constant for reaction with thiophenol
(26) Sluggett, G. W.; Turro, C.; George, M. W.; Koptyug, I. V.; Turro,
N. J. J. Am. Chem. Soc. 1995, 117, 5148.
(27) Sluggett, G. W.; McGarry, P. F.; Koptyug, I. V.; Turro, N. J. J.
Am. Chem. Soc. 1996, 118, 7367.

Photo-ArbuzoV Rearrangements
J. Am. Chem. Soc., Vol. 121, No. 10, 1999 2093
reVersible, the puzzling increase in yield of 15 encountered on
addition of PhSH is also explained.
yield of phosphonate 9 to vary with solvent polarity also speaks
against ion pair intermediates from 8. Studies of the effects of
ring substitutents on the direct photolysis of 1, and the failure
to find products of ion pairs in the presence of added MeOH,
will be reported in a subsequent publication on photolyses of
ring-substituted benzyl and 1-naphthylmethyl phosphites.34
The photolysis of 1 via presumed singlet free radical pairs
(6,7b) is characterized by 60-80% cage combination. This
assumes that random radicals formed by diffusion from the
solvent cage undergo very little return, as was confirmed for
those from 8. It is significant that the radical pair 6,7b is formed
without an intervening molecule such as nitrogen or carbon
dioxide that is commonly formed,33b for example, on thermolysis
of azo compounds, R-NdN-R, and tert-butyl peresters, RCO2-
OBu-tert. Thus, the pair 6, 7b is similar to the so-called
proximate radical pairs35 postulated in the thermal rearrange-
ments of the chiral ketenimine Ph2CdCdNCHPhMe to Ph2C-
(CN)-CHPhMe. The singlet radical pairs formed evidently
couple very rapidly, as they give high yields of product and
greatly increased retention of configuration compared to their
azo counterparts, for example PhCH2-NdN-CHPhMe,36 and
MePhCh-NdN-CHPhMe.37 The radical centers in the singlet
pair 6,7b combine with retention of configuration at phosphorus.
Energetics of Radical Pair Formation. Benzyl excited
singlets clearly possess enough energy to cleave the benzylic
C-O bond in 1 to form an intermediate radical pair (6 and 7b,
Ar ) C6H5). Thus, the singlet energy for benzyl alcohol, a
reasonable model for 1, is 107 kcal/mol in nonpolar solvents.28
The bond dissociation energy for the carbon-oxygen bond of
benzyl alcohol (77 kcal/mol)29 represents a likely value for that
of the simple cleaVage of the benzyl-oxygen bond of PhCH2-
OP(OMe)2 (1). However, the cleavage of the C-O bond in 8
evidently occurs via the triplet state with an energy presumably
only about that for triplet acetophenone, 72 kcal/mol.30 The
formation of product 2 from 1 is at least 25 kcal/mol
exothermic.31 This is a result of the formation of the phosphoryl
π bond since the phosphorus-carbon bond formed in 2 is
weaker than phosphorus-oxygen bonds broken in 1. It seems
likely, therefore, that carbon-oxygen bond scission in excited
triplet 8 is assisted by partial phosphorus oxygen π bond
formation in the transition state for generation of phosphinoyl
radical 6. (That is, the benzylic bond dissociation energy is less
than the 77 kcal/mol value29 for PhCH2OH). The previously
reported generation of triplet pair 6, 7 (Ar ) 1-naphthyl) from
the triplet excited state of dimethyl 1-naphthylmethyl phosphite,6
which is presumably close in energy to that of 1-methylnaph-
thalene (59.6 kcal/mol),32 most assuredly is assisted by rehy-
bridization about phosphorus in the transition state for C-O
cleavage (results to be published).
Comparisons with Related Radical-Pair Forming Reac-
tions. The formation and reactions of geminate free-radical pairs
by thermal and photochemical means has been studied exten-
sively.33 As noted in the Introduction, there is considerable
current interest7 in the mechanisms of photolysis of arylmethyl
and diarylmethyl systems, ArCH2-Z and Ar2CH-Z, for example
where Z is a halogen.7c,f,g This applies as well to esters, ArCH2-
OCOR, with, e.g., Ar ) phenyl and 1-naphthyl and R ) CH3
or Bu-tert.7a,b Photolysis of the halides and esters gives products
of both radical pairs and ion pairs, depending on solvent and
substituents in the aryl rings.7 Unlike phosphite 8, the esters
cannot be caused to react from the triplet excited state.7a
However, the photolyses of the halides can be induced by
interaction with triplet sensitizers.7a,d,f,g For the esters it is
proposed that on direct irradiation, radical pairs are formed
initially, followed by electron transfer to form ion pairs in
competition with decarboxylation.7a,b Diphenylmethyl chloride
photolysis evidently initially forms both ion and radical pairs.7c
The studies of 1 and 8 reported here have not directly
addressed the possibility of formation of ion pairs. Nevertheless,
the similar amounts of phosphorobromidate 13 generated on
trapping of radical 6 in polar and nonpolar solvents argues
against the formation of ion pairs following intersytem crossing
of the radical pairs (6,7a) generated from 8. The failure of the
Photocleavage of the C-O bond, oriented as shown below to
optimize the anomeric interaction of the phosphorus lone pair
with the O-C antibonding orbital (anomeric effect), would
require the radicals so formed to undergo very little motion prior
to coupling.
The photo-Arbuzov rearrangement of 1 is also mechanistically
reminiscent of the thermal Stevens rearrangement (eq 11), found
to proceed largely intramolecularly and at least in part via radical
pairs.38 Furthermore, the photo-Fries rearrangements of phenyl
and 1- and 2-naphthyl acetates give CIDNP phenomena as
evidence for radical pair intermediates that undergo largely
geminate recombination.39
The inefficiency of coupling of random phosphinoyl (6) and
p-acetylbenzyl (7a) radicals that results in the low yields of
phosphonate 9 found in the photolysis of 8 is not surprising on
consideration of other radical pair systems. Thus, the ether sec-
BuOBu-tert is not formed from the thermolysis of sec-BuCO2-
OBu-tert which generates the geminate and then random pair
(28) Murov, S. L.; Carmichael, I.; Hug, G. L. Handbook of Photochem-
istry; Marcel-Dekker: New York, 1993.
(29) Benson, S. W. Thermochemical Kinetics. Methods for the Estimation
of Thermochemical Data and Rate Parameters, Wiley: New York, 1968.
(30) Goshal, S. K.; Sarkar, S. K.; Kastha, G. S. Bull. Chem. Soc. Jpn.
1981, 54, 3635.
(34) Banthumnavin, W. Unpublished results from this laboratory.
(35) Lee, K.-W.; Horowitz, N.; Ware, J.; Singer, L. A. J. Am. Chem.
Soc. 1977, 99, 2622.
(36) Kopecky, K. R.; Gillan, T. Can. J. Chem. 1969, 47, 2371.
(37) Greene, F. D.; Berwick, M. A.; Stowell, J. C. J. Am. Chem. Soc.
1970, 92, 867.
(31) Bhattacharya, A. K.; Thyagarajan, G. Chem ReV. 1981, 81, 415.
Brill, T. S.; Landon, S. J. Ibid. 1984, 84, 577. Lewis, E. S.; Colle, K. S. J.
Org. Chem. 1981, 46, 4369.
(32) Berlman, I. B. Handbook of Fluorescence Spectra of Aromatic
Molecules; Academic Press: Second Edition: New York, 1971; p 473.
(33) (a) Curran, D. P.; Porter, N. A.; Geise, B. Stereochemistry of Radical
Ractions; VCH: Weinheim; Ch. 6 (b) Porter, N. A.; Krebs, P. J. Topics in
Stereochemistry 1988, 18, 97. (c) Gibian, M. J.; Corley, R. C. Chem. ReV.
1973, 73, 441.
(38) Dolling, U. H.; Closs, G. L.; Cohen, A. H.; Ollis, W. D. J. Chem.
Soc., Chem. Commun. 1975, 545. Ollis, W. D.; Rey, M.; Sutherland, I. O.;
Closs, G. L. J. Chem. Soc., Chem. Commun. 1975, 543.
(39) See the following and refs. therein. Gritsan, N. P.; Tsentalovich,
Y. P.; Yurkovskaya, A. V.; Sagdeev, R. Z. J. Phys. Chem. 1996, 100, 4448.
Cui, C.; Wang, X.; Weiss, R. G. J. Org. Chem. 1996, 61, 1962. Arai, T.;
Tobita, S.; Shizuka, H. J. Am. Chem. Soc. 1995, 117, 3968. Andrew, D.;
Des Islet, B. T.; Margaritis, A.; Weedon, A. C. J. Am. Chem. Soc. 1995,
117, 6132.

2094 J. Am. Chem. Soc., Vol. 121, No. 10, 1999
Ganapathy et al.
sec-BuO•/tert-Bu•, initially separated by a CO2 molecule.40
Furthermore, the Norrish Type I photoreaction of optically active
PhCOCHPhMe, to form triplet geminate pairs,41 evidently
features only about 4% recombination of random, cage-free
benzoyl and 2-phenylethyl radicals. Higher yields of random
combination are typically found only with more stable radicals,33
as noted in the accountabilities of bibenzyl (20) and p-
acetylbenzyl radical dimer (11).
F254 plates. Column chromatography was conducted on Merck silica
gel (230-400 mesh) from EM Science. Microanalyses were performed
by Atlantic Microlab, Inc., Norcross, GA. High-performance liquid
chromatographic (HPLC) separations of products were performed under
isocratic conditions with a Waters 590 solvent delivery system, equipped
with an ISCO V4 UV Absorbance detector, using a 10 mm ID
semipreparative, or a 21.4 mm ID preparative Dynamax HPLC column
(100 Å spherical microsorb SiO2 packing, 5 mm particle size, Rainin
Instrument Co, Inc.).
Potential Effect of Large Phosphorus Hyperfine Splitting
Constant of Radical 6. The phosphinoyl radical (MeO)2P(O)•
6 features a phosphorus hyperfine splitting constant of ap-
proximately 700 G.5,6 A potential consequence could be a rapid
hyperfine-induced intersystem crossing of the radical pairs
formed from 1 and 8. However, a consequent large increase in
combination of geminate pair 6,7a to give large amounts of
phosphonate 9 clearly is not observed. This is despite the strong
field-dependent effect of the large hyperfine splitting on the
identity of the triplet pair formed.5
Conclusions. The photoreactions of phosphites 1 and 8
proceed with high efficiencies as shown by the quantum yields
of product formation. The products of direct irradiation of
phosphite 8 are accounted for by the proposal that 8 reacts
primarily through its triplet excited state to generate relatively
long-lived triplet radicals pairs (6 and 7a, Ar ) p-acetylphenyl).
Consequently, the photo-Arbuzov product, phosphonate 9, is
formed in small amounts along with major quantities of radical
diffusion products that are trapped by radical scavengers (solvent
benzene, PhSH, PhCH2Br, and TEMPO). The apparent cage
yield of 9 is 3-5%. The contrasting, relatively high yield of
photo-Arbuzov phosphonate 2, formed on direct irradiation of
1, is indicative of the reaction of 1 largely via the singlet excited
state. The formation of small amounts of both radical products
and those from scavenging by benzene of radicals from 1
suggest that its photolysis, like that of 8, proceeds via radical
pairs that are singlet, not triplet, in nature and largely recombine.
However, the possibility that phosphite 1 reacts in part following
intersystem crossing of its singlet excited state to the triplet,
which generates triplet radical pairs, cannot be ruled out.
NMR spectra were recorded on Varian NMR spectrometers (models
1
XL 300B, Unity 300, and VXR 500). H and 13C chemical shifts are
reported in δ ppm downfield from internal TMS. Coupling constants
are in Hertz (Hz) and, if not otherwise noted, are proton-proton
couplings. Phosphoric acid (85%) was used as external reference for
the 31P spectra. Mass spectra were recorded on a Finnigan MAT 95
mass spectrometer (EI, 70 eV) mode. GC/MS was run on a Hewlett-
Packard Model 5890A series II gas chromatograph coupled to a
Hewlett-Packard 5971A mass selective detector (MSD) operated in the
EI mode. Inlet column was a DB-1 capillary (1% methyl silicone, J &
W Scientific), size 30 M X 0.25 µm. HRMS spectra were obtained in
the EI or CI (DB-210 capillary column) mode on a Finnigan Mat 95
High-Resolution Gas Chromatograpy/Mass Spectrometer with a Finni-
gan MAT ICIS II operating system. Masses are recorded in atomic
mass units (m/z). GC analyses were performed on an HP 5890 series
II gas chromatograph in the FID mode on a DB-1 (1% methyl silicone,
J & W Scientific) capillary column (30 M X 0.25 µm). Quantitative
GC analyses utilized tri-n-butyl phosphate as an internal standard against
which all peaks were calibrated for sensitivity (response factor), except
as otherwise noted. Ultraviolet spectra were obtained on a Hewlett-
Packard 8452A diode array instrument. Wavelengths are reported in
nm with extinction coefficients ꢀ in M-1cm-1. UV data for 1 (CH3-
CN): λmax(ꢀ) 252 (151), 258 (185), 264 (145); λ(ꢀ) 280 (1.5), 300 (0.8).
UV data for 8 (CH3CN): λmax(ꢀ) 248 (15,780), 280 (1,242), 316 (71);
λ (ꢀ) 340 (38).
Quantum Yields. Determination of quantum yields for the photo-
reaction of 8 and other phosphites by use of a Quantacount Instrument,
manufactured by Photon Technology, International, was described
earlier.6
Photolysis of Benzyl Dimethyl Phosphite (1) in Cyclohexane and
Acetonitrile. Stock solutions containing 1 (0.010-0.012 M) and tri-
n-butyl phosphate (0.003 M) were prepared in argon-saturated solvents
in a glovebag filled with argon. Portions (2 × 5 mL) of the solution
were transferred into two quartz tubes and capped with an airtight
septum. The solutions were purged with a slow stream of argon (10
min) and photolyzed at 24-26 °C at 254 nm in a Rayonet reactor. The
reaction was monitored over time by GLC analysis. Products were
identified by co-injection with authentic compounds and by GC/MS
analysis. Product accountability yields, utilizing predetermined response
factors, were based on phosphite converted. Pertinent data are given
in Table 2.
Photolysis of Benzyl Dimethyl Phosphite (1) in Benzene. In
addition to the products formed in cyclohexane and acetonitrile, minor
amounts of phosphonate 15 as well as 22, 23a, and 23b are generated.
The latter three products were identified and quantitated as noted below
in connection with the photolysis of phosphite 8 in benzene with added
benzyl bromide trap.
Photolysis of p-Acetylbenzyl Dimethyl Phosphite (8). Stock
solutions containing 8 (0.016-0.030 M) and internal standard (0.003
M) were prepared in the respective argon-saturated solvents in an argon
glovebag. Portions (2 × 9 mL) of the solution were transferred into
two Pyrex tubes, septum capped, argon purged (10 min), placed in a
water bath at about 26 °C, and irradiated with light from a 450-W
Hanovia medium-pressure mercury vapor lamp filtered through a
uranium filter sleeve. The progress of the reaction was monitored by
GLC analysis. Products were identified by co-injection with authentic
samples and by GC/MS analysis. Accountability yields and other
pertinent data appear in Table 1.
Experimental Section
Materials. Bibenzyl (20) (Aldrich), 4,4′-diacetylbibenzyl (11) (Trans
World Chemicals), p-acetyltoluene (12) (Aldrich), N-bromosuccinimide
(Janssen), PhSSPh (Aldrich), trimethyl phosphite (Aldrich), sodium
benzenesulfinate acid (Aldrich), TEMPO (Aldrich), and di-tert-butyl
peroxide (Aldrich) were used as received. Solvents were dried and argon
saturated prior to use in photolysis. Benzene (Photorex reagent, Baker)
was distilled from sodium. Acetonitrile and cyclohexane (Mallinckrodt,
SpectrAR) were distilled from CaH2 before use in the photochemical
work. Thiophenol (PhSH, Aldrich) was distilled from calcium sulfate.
Benzyl bromide (Aldrich), dimethyl phosphite (10) (Aldrich), benzyl
alcohol (Aldrich), and dimethyl phenylphosphonate (15) (TCI) were
distilled before use. Benzyl dimethyl phosphite (1),25 dimethyl phos-
phorobromidate (13),42 and dimethyl N,N-diethylaminophosphoramid-
ite43 were prepared by literature procedures.
General Information. Melting points (Thomas-Hoover melting point
apparatus) are uncorrected. Unless stated otherwise, distillations of
products were performed using a short path apparatus. Thin-layer
chromatography (TLC) analyses were performed on EM silica gel 60
(40) Koenig. T.; Owens, J. J. Am. Chem. Soc. 1974, 96, 4052. Koenig,
T.; Owens, J. J. Am. Chem. Soc. 1973, 95, 8485.
(41) Step, E. N.; Buchachenko, A. L.; Turro, N. J. J. Am. Chem. Soc.
1992, 57, 7018. For an earlier study of this reaction, see: Lewis, F. D.;
Magyar, J. G. J. Am. Chem. Soc. 1973, 95, 5973. See also ref 16 for the
photolysis of optically active MePhCHCOCHPhMe.
(42) Goldwhite, H.; Saunders: B. C. J. Chem. Soc. 1955, 3564.
(43) Arbuzov, B. A.; Yarmukhametova, D. K. Dokl. Akad. Nauk. USSR
1955, 101, 675.
General Procedure for the Photolysis of 8 in the Presence of
Radical Traps. Solutions (5 mL) containing 8 (0.014 M-0.018M) and
internal standard (0.003 M), with varying amounts of added radical
traps, were prepared as described above for photolyses of 8 and

Photo-ArbuzoV Rearrangements
J. Am. Chem. Soc., Vol. 121, No. 10, 1999 2095
irradiated with a light from a 450-W Hanovia medium-pressure mercury
vapor lamp filtered through a uranium glass filter sleeve in a water
bath at room temperature to 16-18% conversions of 8. Reactions were
monitored by GC during irradiation (450-W Hanovia medium-pressure
lamp, uranium filter). Pertinent results are given in Tables 3-5.
(51); 278 (M+, 1), 277 (2), 188 (9), 187 (100), 156 (2), 155 (21), 110
(8), 109 (95), 92 (13), 91 (49).
Preparation of Dimethyl Benzylphosphonate (2). Benzyl bromide
(4.0 g, 23.4 mmol) and trimethyl phosphite (17.4 g, 140 mmol) were
refluxed under argon at 110 oC for 2 h. The reaction mixture was
concentrated in vacuo and dried under high vacuum. The crude residue
was purified by distillation (bp 70 °C, 0.5 mmHg) to furnish 2 (3.04 g,
Preparative Photolysis of p-Acetylbenzyl Dimethyl Phosphite (8).
A solution of phosphite 8 (0.125 g, 0.52 mmol) in deoxygenated, argon-
saturated benzene (100 mL) was irradiated as described previously (450
W medium-pressure Hanovia, uranium glass filter). Progress of the
reaction was monitored by GC analysis and the irradiation was
discontinued (80 min) after 90% consumption of 8. 31P spectrum (C6D6)
of the photolyzate displayed signals at δ 29.97, 27.83, 26.30, 25.97
and 21.64. The peaks at δ 27.83 and 21.64 correspond to 9 and 15,
respectively. The GC and GC/MS analysis of the photolyzate showed
three unidentifiable peaks with M+ 320 showing the nearly identical
pattern. The peak at δ 29.97 was assigned to the adduct 16, which was
isolated, and the spectral data are given below. However, the remaining
unidentifiable 31P NMR peaks at 26.30 and 25.97 can be assigned to
17a and 17b, although we have no evidence as to which is the trans
isomer and which is the cis form. The photolyzate was concentrated
and flash chromatographed on a silica gel column, eluting with 4%
methanol:dichloromethane (monitored by GC). The polar products were
collected and further purified by repeated HPLC chromatographic
separations to furnish 16 (10.8 mg, 6.5%) and 18 (14.7 mg, 8.9%).
During the isolation procedure, the unstable cyclohexadiene products
17a and 17b were aromatized to 18. Spectral data for 16: 31P NMR
(CDCl3, 121 MHz): δ 31.69.1H NMR (C6D6, 500 MHz): δ 7.70 (d, 2
H, J ) 8.0 Hz), 6.91 (d, 2 H, J ) 8.0 Hz), 5.79 (m, 1 H), 5.68 (m, 1
H), 5.59 (m, 1 H), 5.45 (m, 1 H), 3.31 (d, 3 H, 3JPOCH ) 10.7 Hz), 3.29
(d, 3 H, 3JPOCH ) 10.7 Hz), 3.04 (ddddd, 1 H, JHH ) 2.4. 5.8, 7.2, 8.3
1
15.2 mmol, 65%). 31P NMR (CDCl3, 121 MHz): δ 29.44. H NMR
3
(CDCl3, 300 MHz): δ 7.23 (br s, 5 H), 3.59 (d, 6 H, JPOCH ) 11.0
2
Hz), 3.09 (d, 2 H, JPH ) 21.7 Hz). 13C NMR (CDCl3, 75 MHz): δ
2
3
4
131.16 (d, JPC ) 9.6 Hz), 129.65 (d, JPC ) 6.6 Hz), 128.59 (d, JPC
) 3.0 Hz), 126.94 (5JPC ) 3.5 Hz), 52.83 (2JPOC ) 7.1 Hz), 32.76 (1JPC
) 138.0 Hz). MS (EI) m/z (rel intensity): 200 (M+, 100), 104 (51), 91
(100). HRMS (EI) m/z (M+): calcd 200.0602, obsd 200.0678.
Preparation of p-Acetylbenzyl Bromide. p-Acetyltoluene (12) (83.0
g, 618 mmol), N-bromosuccinimide (132.0 g, 743 mmol), and benzoyl
peroxide (250 mg) were dissolved in 1000 mL of benzene. The solution
was purged with nitrogen for 10 min and stoppered. The solution was
stirred vigorously during irradiation for 6 h with light from a 450-W
Hanovia medium-pressure mercury lamp. The photolyzate was filtered,
and the filtrate was washed with water. The organic layer was washed
with saturated NaHCO3 solution, dried over anhydrous Na2SO4, and
concentrated. To the concentrate anhydrous K2CO3 (100 mg) was added.
Distillation (bp 115 °C, 0.5 mmHg) afforded 63.2 g (48%) of the desired
product, mp 42-43 °C (lit. 38-39 oC).44 1H NMR (CDCl3, 300
MHz): δ 7.94 (d, 2 H, J ) 8.3 Hz), 7.45 (d, 2 H, J ) 8.3 Hz), 4.48
(s, 2 H), 2.57 (s, 3 H). 13C NMR (CDCl3, 75 MHz): 197.49, 142.87,
136.89, 129.33, 128.91, 32.31, 26.79.
Preparation of p-Acetylbenzyl Alcohol. p-Acetylbenzyl bromide
(25.0 g, 117 mmol) and CaCO3 (63.0 g, 630 mmol) were added to
dioxane-water (1:1, 720 mL). This reaction mixture was stirred at
reflux for 15 h and then reduced in volume under vacuum. To the stirred
mixture of the concentrate in CH2Cl2 (500 mL) was slowly added dilute
aqueous HCl until the solid material was dissolved in the water layer.
The separated organic layer was washed with Na2CO3, dried over Na2-
SO4, and concentrated to a viscous oil. Crystallization of the concentrate
from ether:pentane at -20 °C provided fine white crystals of the desired
3
2
Hz; JPH ) 21.7 Hz), 2.62 (dddd, 1 H, JHH ) 1.5, 2.7, 5.6 Hz; JPH
)
29.6 Hz) 2.54 (dd, 1 H, JHH ) 7.2, 13.4 Hz), 2.46 (dd, 1 H, JHH ) 8.3
Hz, 13.4 Hz), 2.07 (s, 3 H). 13C NMR (CD2Cl2, 125.8 MHz): δ 197.44,
144.55, 135.80, 129.74, 128.91 (d, 3JPC ) 5.8 Hz), 128.57, 125.90 (d,
2JPC )12.8 Hz), 122.89 (d, 4JPC ) 6.2 Hz), 119.20 (d, 3JPC ) 12.4 Hz),
53.02 (d, 2JPOC ) 7.0 Hz), 52.86 (d, 2JPOC ) 7.0 Hz), 39.80 (d, 3JPC
)
24.0 Hz), 36.32 (d, 1JPC ) 133.6 Hz), 33.90 (d, 2JPC ) 4.1 Hz), 26.54.
MS (EI) m/z (rel intensity): 320 (M+, 1), 319 (0.41), 318 (2.02), 211
(19.2), 187 (100), 165 (2.87), 155 (13.2), 109 (44.3). HRMS (EI) m/z
(M+): calcd 320.1177, obsd 320.1156. For 18: 31P NMR (CD2Cl2,
121 MHz): δ 23.66. 1H NMR (CD2Cl2, 500 MHz): δ 7.88 (d, 2 H, J
o
o
product (12.0 g, 60%), mp 53 C (lit. 54 C).45 1H NMR (CDCl3, 300
MHz): δ 7.90 (d, 2 H, J ) 8.3 Hz), 7.42 (d, 2 H, J ) 8.5 Hz), 4.74
(s, 2 H), 3.04-2.73 (br s, 1 H), 2.57 (s, 3 H). 13C NMR (CDCl3, 75
MHz): δ 198.41, 146.61, 136.28, 128.73, 126.73, 64.57, 26.80.
Preparation of p-Acetylbenzyl Dimethyl Phosphite (8). A solution
of p-acetylbenzyl alcohol (2.75 g, 18.3 mmol) and 1-H-tetrazole (0.75
g, 10.5 mmol) in freshly distilled acetonitrile was stirred under argon.
Neat dimethyl N,N-diethylaminophosphoramidite (9.0 g, 54.5 mmol)
was added dropwise at room temperature over a 5-min period. The
reaction was stirred for 18 h at room temperature. Solvent removal
under vacuum gave a solid material that was dissolved in dry pentane
(75 mL), stirred for 5 min, and then allowed to settle. The supernatant
liquid was cannulated under argon into a flask by Schlenk techniques.
Solvent was removed from the stirred solution under vacuum. The crude
residue was transferred to a molecular distillation apparatus in an argon
glovebag. Distillation (oil bath temp. 75 °C, 0.025 mmHg) provided 8
(3.81 g, 86%) in 98% purity (GC). 31P NMR (CDCl3, 121 MHz): δ
3
) 8.2 Hz), 7.70 (dd, 2 H, JHH ) 8.2 Hz, JPH ) 13.0 Hz), 7.32 (dd, 2
H, JHH ) 8.2 Hz, JPH ) 4.0 Hz), 7.29 (d, 2 H, J ) 8.2 Hz), 4.09 (s, 2
H), 3.70 (d, 6 H, 3JPOCH ) 10.1 Hz), 2.55 (s, 3 H). 13C NMR (CD2Cl2,
4
125.8 MHz): δ 197.87, 146.20, 145.81 (d, JPC ) 2.9 Hz), 136.13,
132.67 (d,2JPC ) 10.5 Hz), 129.67, 129.66 (d, 3JPC ) 15.8 Hz), 129.14,
1
2
125.82 (d, JPC ) 189.9 Hz), 53.07 (d, JPOC ) 5.7 Hz), 42.31, 26.96.
MS (EI) m/z (rel intensity): 318 (M+, 33), 303 (100), 276 (2.5), 209
(2.7), 165 (14.4), 109 (2.1). HRMS (EI) m/z (M+): calcd 318.1021,
obsd 318.1018. For 17a or 17b: 31P NMR (C6D6, 121 MHz): δ 26.30
or 25.97. GC/MS (EI) m/z (rel intensity) (shorter GC retention time):
320 (M+, 1), 318 (2), 187 (100), 155 (19), 133 (5), 127 (17), 109 (70).
GC-HRMS (CI) m/z (M++ H): calcd for C17H22O4P 321.1256, found
321.1257. For 17a or 17b: 31P NMR (C6D6, 121 MHz): δ 26.30 or
25.97. GC/MS (EI) m/z (rel intensity) (longer GC retention time): 320
(M+, 1), 319 (2), 187 (100), 155 (20), 133 (5), 127 (14), 109 (68).
GC-HRMS (CI) m/z (M++ H): calcd for C17H22O4P 321.1256, obsd
321.1254.
1
141.25. H NMR (CDCl3, 300 MHz): δ 7.95 (d, 2 H, J ) 8.3 Hz),
3
7.45 (d, 2 H, J ) 8.3 Hz), 4.92 (d, 2 H, JPOCH ) 7.8 Hz), 3.53 (d, 6
3
H, JPOCH ) 10.9 Hz), 2.60 (s, 3 H). 13C NMR (CDCl3, 75 MHz): δ
3
197.92, 143.84 (d, JPC ) 4.5 Hz), 136.60, 128.71, 127.30, 63.37 (d,
2
2JPOC ) 11.1 Hz), 49.59 (d, JPOC ) 11.1 Hz), 26.85. HRMS (EI) m/z
Products in the Photolysis of 8 in Benzene with PhCH2Br added.
For photolyses in benzene with added PhCH2Br/8 in the range 0.5-
7.4, the crude photolyzate was examined by 31P NMR spectroscopy.
In addition to the products observed on photolysis of 8 in benzene in
the absence of benzyl bromide, a product peak at δ 28.8 corresponding
to benzylphosphonate 2 was noted and also identified by GC. In addition
new resonances at δ 30.79, 27.72, and 27.68, analogous to those for
16, 17a, and 17b, reported above, were noted and assigned to the
structures 22, 23a, and 23b. Unlike 17a and 17b, the cis and trans
isomers, 23a and 23b, were not separable by GC. The amounts of 22
and 23a/23b were quantitated using the sensitivity factor for 16. For
22 or 23a/23b GC/MS (EI) m/z (rel intensity): 278 (M+, 1), 277 (1),
188 (9), 187 (100), 156 (2), 155 (21), 110 (4), 109 (93), 92 (6), 91
(M+): calcd 242.0708, obsd 242.0709. Anal. Calcd for C11H15O4P: C,
54.54; H, 6.25. Found: C, 54.20; H, 6.17.
Preparation of Dimethyl p-Acetylbenzylphosphonate (9). p-
Acetylbenzyl bromide (9.0 g, 42.2 mmol) and trimethyl phosphite (36.0
o
g, 290 mmol) were refluxed under argon at 110 C for 2 h. Solvent
removal yielded a solid residue that was dried under high vacuum.
Recrystallization from methanol:ether afforded pale yellow crystals of
9 (6.31 g, 62%), mp 64-66 °C. 31P NMR (CDCl3, 121 MHz): δ 28.22.
1H NMR (CDCl3, 300 MHz): δ 7.88 (d, 2 H, J ) 7.6 Hz), 7.36 (d, 2
(44) Jarvis, B. B.; Saukaitis, J. C. J. Am. Chem. Soc. 1973, 95, 7708.
(45) Schmidt, L.; Swoboda, W.; Wichtl, M. Monatsh. 1952, 83, 185.
Smith, J. G.; Dibble, P. W.; Sandborn, R. E. J. Org. Chem. 1986, 51, 3762.

2096 J. Am. Chem. Soc., Vol. 121, No. 10, 1999
Ganapathy et al.
H, J ) 8.1 Hz), 3.65 (d, 6 H, 3JPOCH ) 10.9 Hz), 3.19 (d, 2 H, 2JPH
)
and checked by 1H NMR. Further purification of this fraction with 1%
ethyl acetate:hexane as an eluent gave 25 (93 mg, 59%). 1H NMR
(CDCl3, 300 MHz): δ 7.92 (d, 2 H, J ) 8.4 Hz), 7.43 (d, 2 H, J ) 8.2
Hz), 4.88 (s, 2 H), 2.58 (s, 3 H), 1.80-1.28 (br m, 6 H), 1.22 (s, 6 H),
1.15 (s, 6 H). 13C NMR (CDCl3, 75 MHz): δ 197.81, 143.95, 136.14,
128.43, 127.08, 78.19, 60.12, 39.75, 33.10, 26.67, 20.34, 17.14. MS
(EI) m/z (rel intensity): 289 (M+, 1), 156 (100). HRMS (EI) m/z
(M+): calcd 289.2042, obsd 289.2067.
Photoreduction of 9 by PhSH. In conditions analogous to those
for photoreaction of 8, a solution of 0.02 M phosphonate 9, 0.04 M
PhSH, and 0.0031 M tri-n-butyl phosphate (internal standard) was
irradiated. At 32% conversion of 9 (approximately 9 h), three major
GC peaks were detected. One peak had a molecular ion mass (GC/
MS) at m/z ) 244, consistent with the product of reduction of the
carbonyl of 9 to form p-MeCH(OH)C6H4CH2P(O)(OMe)2. However,
the latter product was not observed during the photorearrangements of
phosphite 8.
22. 2 Hz), 2.55 (s, 3 H). 13C NMR (CDCl3, 75 MHz): δ 197.72, 137.09
2
5
3
(d, JPC ) 9.1 Hz), 135.95 (d, JPC ) 3.5 Hz), 130.04 (d, JPC ) 6.0
Hz), 128.77 (4JPC ) 3.0 Hz), 53.09 (2JPOC ) 7.1 Hz), 33.12 (1JPC
)
137.5 Hz), 26.72. MS (EI) m/z (rel intensity): 242 (M+, 19), 227 (100).
HRMS (EI) m/z (M+): calcd 242.0708, obsd 242.0707. Anal. Calcd
for C11H14O4P: C, 54.54; H, 6.25. Found: C, 54.60; H, 6.23.
Photolysis of 8 with TEMPO as a Trap. Phosphite 8 (45.0 mg,
0.185 mmol) and TEMPO (30.0 mg, 0.192 mmol) were dissolved in
1.2 mL of degassed CD3CN and irradiated in a flame-sealed quartz
NMR tube to approximately 30% consumption (31P NMR) of the
phosphite under conditions such as those described previously for the
photolysis of 8. In addition to the peaks normally observed, the 31P
NMR spectrum displayed a predominant peak at δ 6.67 that was
assigned to 26, formed by TEMPO trapping the phosphinoyl radical.
Photolysis of Dimethyl Phosphite/di-tert-Butyl Peroxide/TEMPO
Solution. A solution of dimethyl phosphite (59.9 mg, 0.544 mmol),
TEMPO (70.9 mg, 0.454 mmol) and di-tert-butylperoxide (0.08 mL,
63.6 mg, 0.435 mmol) in a 5-mL dry Pyrex tube was capped with
airtight septum, argon purged for 5 min, and then irradiated with light
from a 450-W Hanovia medium-pressure mercury vapor lamp. The
progress of the reaction was monitored by 31P NMR (C6D6), which
displayed a predominant peak at δ 6.07, corresponding to 26. Attempted
purification of the photolyzate on silica gel column chromatography
failed to isolate compound 26.
Acknowledgment. This research was generously supported
by grants from the National Science Foundation and the National
Institutes of Health. We thank Professor Cheves Walling for
suggesting phosphite 8 as a photochemical source of triplet
radical pairs.
Supporting Information Available: Experimental details
for the preparation of 21, products of dark reaction of 8 with
PhSSPh to give 29a-29c, and an expanded version of Table 5
that includes individual yields of 16, 17a, 17b, 22, 23a, and
23b (PDF). This material is available free of charge via the
Internet at http://pubs.acs.org.
Isolation of 25 from the Photolysis of 8 in the Presence of
TEMPO. A solution of phosphite 8 (132 mg, 0.545 mmol) and TEMPO
(88.0 mg, 0.563 mmol) in acetonitrile (30 mL prepared using the
glovebag techniques described above) was irradiated as before (450
W Hanovia lamp, uranium glass filter). At 70% conversion of 8 (GC),
irradiation was discontinued. The product solution was concentrated
and then subjected to silica gel column chromatography. Elution with
the 5% ethyl acetate:hexane, first fraction, was collected, concentrated,
JA982440K