Characterization of reactive intermediates generated during photolysis of 4-acetoxy-4-aryl-2,5-cyclohexadienones: Oxenium ions and aryloxy radicals

Article abstract of DOI:10.1021/ja805336d

Aryloxenium ions 1 are reactive intermediates that are isoelectronic with the better known arylcarbenium and arylnitrenium ions. They are proposed to be involved in synthetically and industrially useful oxidation reactions of phenols. However, mechanistic studies of these intermediates are limited. Until recently, the lifetimes of these intermediates in solution and their reactivity patterns were unknown. Previously, the quinol esters 2 have been used to generate 1, which were indirectly detected by azide ion trapping to generate azide adducts 4 at the expense of quinols 3, during hydrolysis reactions in the dark. Laser flash photolysis (LFP) of 2b in the presence of O₂ in aqueous solution leads to two reactive intermediates with γ_max 360 and 460 nm, respectively, while in pure CH₃CN only one species with λ_max 350 nm is produced. The intermediate with Amax 460 nm was previously identified as lb based on direct observation of its decomposition kinetics in the presence of N₃^-, comparison to azide ion trapping results from the hydrolysis reactions, and photolysis reaction products (3b). The agreement between the calculated (B3LYP/6-31G(d)) and observed time-resolved resonance Raman (TR³) spectra of 1b further confirms its identity. The second intermediate with λ _max 360 nm (350 nm in CH₃CN) has been characterized as the radical 5b, based on its photolytic generation in the less polar CH ₃CN and on isolated photolysis reaction products (6b and 7b). Only the radical intermediate 5b is generated by photolysis in CH₃CN, so its UV-vis spectrum, reaction products, and decay kinetics can be investigated in this solvent without interference from 1b. In addition, the radical 5a was generated by LFP of 2a and was identified by comparison to a published UV-vis spectrum of authentic 5a obtained under similar conditions. The similarity of the UV-vis spectra of 5a and 5b, their reaction products, and the kinetics of their decay confirm the assigned structures. The lifetime of 1b in aqueous solution at room temperature is 170 ns. This intermediate decays with first-order kinetics. The radical intermediate 5b decomposes in a biphasic manner, with lifetimes of 12 and 75 μs. The decay processes of 5a and 5b were successfully modeled with a kinetic scheme that included reversible formation of a dimer. The scheme is similar to the kinetic models applied to describe the decay of other aryloxy radicals.

Full text of DOI:10.1021/ja805336d

Published on Web 10/31/2008
Characterization of Reactive Intermediates Generated During
Photolysis of 4-Acetoxy-4-aryl-2,5-cyclohexadienones:
Oxenium Ions and Aryloxy Radicals
Yue-Ting Wang,† Kyoung Joo Jin,† Samuel H. Leopold,† Jin Wang,‡ Huo-Lei Peng,‡
Matthew S. Platz,‡ Jiadan Xue,§ David Lee Phillips,§ Stephen A. Glover,| and
Michael Novak*,†
Department of Chemistry and Biochemistry, Miami UniVersity, Oxford, Ohio 45056, Department
of Chemistry, The Ohio State UniVersity, 100 West 18th AVenue, Columbus, Ohio 43210,
Department of Chemistry, UniVersity of Hong Kong, Hong Kong, P.R. China, and School of
Biological, Biomedical, and Molecular Sciences, DiVision of Chemistry, UniVersity of New
England, Armidale, 2351, New South Wales, Australia
Received July 10, 2008; E-mail: novakm@muohio.edu
Abstract: Aryloxenium ions 1 are reactive intermediates that are isoelectronic with the better known
arylcarbenium and arylnitrenium ions. They are proposed to be involved in synthetically and industrially
useful oxidation reactions of phenols. However, mechanistic studies of these intermediates are limited.
Until recently, the lifetimes of these intermediates in solution and their reactivity patterns were unknown.
Previously, the quinol esters 2 have been used to generate 1, which were indirectly detected by azide ion
trapping to generate azide adducts 4 at the expense of quinols 3, during hydrolysis reactions in the dark.
Laser flash photolysis (LFP) of 2b in the presence of O2 in aqueous solution leads to two reactive
intermediates with λmax 360 and 460 nm, respectively, while in pure CH3CN only one species with λmax 350
nm is produced. The intermediate with λmax 460 nm was previously identified as 1b based on direct
observation of its decomposition kinetics in the presence of N3-, comparison to azide ion trapping results
from the hydrolysis reactions, and photolysis reaction products (3b). The agreement between the calculated
(B3LYP/6-31G(d)) and observed time-resolved resonance Raman (TR3) spectra of 1b further confirms its
identity. The second intermediate with λmax 360 nm (350 nm in CH3CN) has been characterized as the
radical 5b, based on its photolytic generation in the less polar CH3CN and on isolated photolysis reaction
products (6b and 7b). Only the radical intermediate 5b is generated by photolysis in CH3CN, so its UV-vis
spectrum, reaction products, and decay kinetics can be investigated in this solvent without interference
from 1b. In addition, the radical 5a was generated by LFP of 2a and was identified by comparison to a
published UV-vis spectrum of authentic 5a obtained under similar conditions. The similarity of the UV-vis
spectra of 5a and 5b, their reaction products, and the kinetics of their decay confirm the assigned structures.
The lifetime of 1b in aqueous solution at room temperature is 170 ns. This intermediate decays with first-
order kinetics. The radical intermediate 5b decomposes in a biphasic manner, with lifetimes of 12 and 75
µs. The decay processes of 5a and 5b were successfully modeled with a kinetic scheme that included
reversible formation of a dimer. The scheme is similar to the kinetic models applied to describe the decay
of other aryloxy radicals.
Introduction
responsible for the generation of commercially useful polymers,
such as poly(2,6-dimethyl-1,4-phenylene oxide).6,7 However,
until recently mechanistic studies of 1 were fairly limited and
Aryloxenium ions 1 are oxygen-centered cations in one of
their canonical structures and are isoelectronic with well-known
arylnitrenium and arylcarbenium ions. They have been proposed
to be the reactive intermediates responsible for synthetically and
industrially useful electrochemical and chemical oxidations of
phenols.1-5 For example, it has been proposed that they are
(2) Rieker, A.; Beisswenger, R.; Regier, K. Tetrahedron 1991, 47, 645–
654.
(3) Pelter, A.; Ward, R. S. Tetrahedron 2001, 57, 273–282.
(4) (a) Rieker, A.; Speiser, B.; Straub, H. Dechema-Monographien 1992,
125, 777–782. (b) Dimroth, K.; Umbach, W.; Thomas, H. Chem. Ber
1967, 100, 132–141.
† Miami University.
(5) Rodrigues, J. A. R.; Abramovitch, R. A.; de Sousa, J. D. F.; Leiva,
G. C. J. Org. Chem. 2004, 69, 2920–2928.
‡ The Ohio State University.
§ University of Hong Kong.
(6) (a) Baesjou, P. J.; Driessen, W. L.; Challa, G.; Reedjik, J. J. Am. Chem.
Soc. 1997, 119, 12590–12594. (b) Driessen, W. L.; Baesjou, P. J.;
Bol, J. E.; Kooijman, H.; Spek, A. L.; Reedjik, J. Inorg. Chim. Acta
2001, 324, 16–20.
| University of New England.
(1) (a) Swenton, J. S.; Carpenter, K.; Chen, Y.; Kerns, M. L.; Morrow,
G. W. J. Org. Chem. 1993, 58, 3308–3316. (b) Swenton, J. S.;
Callinan, A.; Chen, Y.; Rohde, J. L.; Kerns, M. L.; Morrow, G. L. J.
Org. Chem. 1996, 61, 1267–1274.
(7) Kobayashi, S.; Higashimura, H. Prog. Polym. Sci. 2003, 28, 1015–
1048.
9
10.1021/ja805336d CCC: $40.75
2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 16021–16030 16021

A R T I C L E S
Wang et al.
little was known about the generation, reactivity, and selectivity
of these species, or even whether they have actually been
produced in the reactions in which they have been invoked.1,3,5
Prior to our investigations only a few highly delocalized
examples of 1, such as 1d and 1e, were observed or isolated.4
More recently, the very long-lived oxenium ion derived from
R-tocopherol has been investigated.8 Unfortunately, these
unusually stabilized species provided little understanding of
the reactions of this class of transient ions. Furthermore, the
previously existing mechanistic studies of these species were
not consistent with each other.9-13 Discrepancies existed
concerning the regiochemistry of reactions of purported 1
generated from different sources and the possible involvement
of triplet ions.2,9,11,13 In most cases unequivocal evidence for
the existence of the cations was not presented.2,9-12
Figure 1. Transient absorbance spectra obtained after 266 nm excitation
of 2b in O2-saturated pH 7.1 phosphate buffer (A) or in O2-saturated CH3CN
(B). Key for A: red, 20 ns after flash; green, 120 ns; blue, 220 ns. Key for
B: red, 20 ns after flash; blue, 1 µs; green, 100 µs; magenta, 200 µs. All
spectra recorded over a 20 ns window.
Scheme 1. Photolysis of 2a and 2b
Several years ago we initiated a study of the chemistry of
aryloxenium ions. We were able to indirectly detect the
generation of 4′-substituted-4-biphenylyloxenium ions 1a-c
-
from catalyzed or uncatalyzed solvolysis of 2a-c through N3
-
trapping, common ion effects, and 18O-labeling studies in
18O-H2O.14-16 More recently, we succeeded in directly observ-
ing the generation and decomposition of 1b by laser flash
photolysis (LFP) in aqueous solution.17 This transient ion with
λmax 460 nm was characterized initially by the kinetics of its
-
reaction with N3-. The second-order reaction of 1b with N3
is apparently diffusion limited, and the rate constant ratio kaz/ks
(8) (a) Williams, L. L.; Webster, R. D. J. Am. Chem. Soc. 2004, 126,
12441–12450. (b) Lee, S. B.; Lin, C. Y.; Gill, P. M.; Webster, R. D.
J. Org. Chem. 2005, 70, 10466–10473.
(9) (a) Abramovitch, R. A.; Inbasekaran, M.; Kato, S. J. Am. Chem. Soc.
1973, 95, 5428–5430. (b) Abramovitch, R. A.; Alvernhe, G.; In-
basekaran, M. N. Tetrahedron Lett. 1977, 1113–1116. (c) Abramovitch,
R. A.; Inbasekaran, M. N. J. Chem. Soc., Chem. Commun 1978, 149–
150. (d) Abramovitch, R. A.; Alvernhe, G.; Bartnik, R.; Dassanayake,
N. L.; Inbasekaran, M. N.; Kato, S. J. Am. Chem. Soc. 1981, 103,
4558–4565.
(Scheme 1) measured directly from the kinetics of decomposi-
tion of 1b is equivalent, within experimental error, to the same
ratio calculated from the yields of 4b and 3b during the
hydrolysis of 2b in N3--containing aqueous buffers. A second
longer-lived transient intermediate with λmax 360 nm was not
identified.17
In this paper, further characterization of 1b and the identifica-
tion of the other observed reactive species are presented. For
1b, a nanosecond time-resolved resonance Raman (ns-TR3)
spectrum has been recorded and compared with vibrational
transitions calculated by DFT methods (B3LYP/6-31G(d)).
These results confirm our assignment of the 460 nm absorbance
band to 1b and strongly indicate that the properties of 1b are
predominately those expected for its 1-oxo-2,5-cyclohexadienyl
carbenium ion resonance structure as previously indicated by
(10) Li, Y.; Abramovitch, R. A.; Houk, K. N. J. Org. Chem. 1989, 54,
2911–2914.
(11) (a) Endo, Y.; Shudo, K.; Okamoto, T. J. Am. Chem. Soc. 1977, 99,
7721–7723. (b) Shudo, K.; Orihara, Y.; Ohta, T.; Okamoto, T. J. Am.
Chem. Soc. 1981, 103, 943–944. (c) Endo, Y.; Shudo, K.; Okamoto,
T. J. Am. Chem. Soc. 1982, 104, 6393–6397.
(12) Uto, K.; Miyazawa, E.; Ito, K.; Sakamoto, T.; Kikugawa, Y.
Heterocycles 1998, 48, 2593–2600.
(13) Hegarty, A. F.; Keogh, J. P. J. Chem. Soc., Perkin Trans. 2 2001,
758–762.
(14) Novak, M.; Glover, S. A. J. Am. Chem. Soc. 2004, 126, 7748–7749.
(15) Novak, M.; Glover, S. A. J. Am. Chem. Soc. 2005, 126, 8090–8097.
(16) Novak, M.; Poturalski, M. J.; Johnson, W. L.; Jones, M. P.; Wang,
Y.; Glover, S. A. J. Org. Chem. 2006, 71, 3778–3785.
(17) Wang, Y.-T.; Wang, J.; Platz, M. S.; Novak, M. J. Am. Chem. Soc.
2007, 129, 14566–14567.
9
16022 J. AM. CHEM. SOC. VOL. 130, NO. 47, 2008

Oxenium Ions and Aryloxy Radicals
A R T I C L E S
Table 1. Rate Constants Obtained from LFP Experimentsa
reaction
conditions
rate constant (units),
wavelength monitored
intermediate
value
1b
1b
5b
5b
5b
5b
5b
5b
5b
5b
5b
5b
5a
5a
5 vol% CH3CN/H2O, O2 sat
5 vol% CH3CN/H2O, O2 sat
5 vol% CH3CN/H2O, O2 sat
5 vol% CH3CN/H2O, O2 sat
CH3CN, O2 sat
CH3CN, O2 sat
CH3CN, O2 sat
CH3CN, O2 sat
CH3CN, Ar sat
CH3CN, Ar sat
CH3CN, Ar sat
CH3CN, Ar sat
CH3CN, O2 sat
kaz (M-1 s-1), 460 nm
ks (s-1), 460 nm
(6.6 ( 0.2) × 109b
(5.8 ( 0.4) × 106b
(8.06 ( 0.08) × 104
(1.29 ( 0.02) × 104
(11.0 ( 0.7) × 104
(2.35 ( 0.07) × 104
(13.2 ( 0.6) × 104
(2.82 ( 0.32) × 104
(7.18 ( 0.57) × 104
(1.18 ( 0.09) × 104
(7.83 ( 0.30) × 104
(1.19 ( 0.11) × 104
(5.56 ( 0.48) × 104
(6.85 ( 1.70) × 103
k1AQO2 (s-1), 360 nm
k2AQO2 (s-1), 360 nm
k1ACNO2 (s-1), 350 nm
k2ACNO2 (s-1), 350 nm
k1ACNO2 (s-1), 575 nm
k2ACNO2 (s-1), 575 nm
k1ACNAr (s-1), 350 nm
k2ACNAr (s-1), 350 nm
k1ACNAr (s-1), 575 nm
k2ACNAr (s-1), 575 nm
k1ACNO2 (s-1), 340 nm
k2ACNO2 (s-1), 340 nm
CH3CN, O2 sat
a All rate constants were collected at 22 °C under the conditions described in the Table. Rate constants and error limits for 5a and 5b are averages
and standard deviations of three to five independent measurements. Rate constants for 1b were determined as described in ref 17. b Determined from the
slope (kaz) or intercept (ks) of a plot of kobs vs [N3-] at six [N3-] at 0-4 mM (see ref 17).
the experimental properties and calculated structures of 1b and
related ions.14-18 The other long-lived intermediate generated
along with 1b under the same conditions has been identified as
radical 5b. This is the only reactive intermediate generated by
LFP of 2b in CH3CN. We have also detected the generation of
5a during LFP of 2a in both pH 7.1 aqueous buffer and CH3CN.
The identity of 5a can be confirmed by comparison of its
UV-vis spectrum to that of authentic 5a.19,20 Photolysis reaction
products (6 and 7) derived from both intermediates 5a and 5b
have been isolated and characterized for both quinol esters 2a
and 2b. The kinetics of the decomposition of 5a and 5b have
been successfully modeled by a kinetic scheme that includes
reversible formation of a dimer. The scheme is similar to the
kinetic models applied to describe the decay of other aryloxy
radicals.21-28
× 109 M-1 s-1 (Table 1), and its lifetime is insensitive to the
presence of O2.17 Further confirmation of the structural assign-
ment was obtained from the equivalence of the kaz/ks ratio
measured from LFP of 2b ((1.14 ( 0.09) × 103 M-1) with the
same ratio measured from the N3--trapping product study of
2b in identical aqueous solutions at 30 °C of (1.0 ( 0.2) × 103
product of 2b is the quinol 3b.17 This is the expected product
of the reaction of 1b with the aqueous solvent and is the only
detected product of the hydrolysis of 2b.16
M-1 16
. Steady state photolysis confirmed that a major photolysis
Time-Resolved Resonance Raman Characterization of the
Oxenium Ion. Nanosecond time-resolved resonance Raman (ns-
TR3) spectra were obtained at room temperature in phosphate
buffered 20 vol% CH3CN/H2O and in CH3CN on freshly
prepared 2 mM solutions of 2b. A pump laser pulse of 266
nm, identical to that used for the LFP studies, was used to
generate transient species. Laser pulses of 354.7 and 435.7 nm
were used to probe the transients. These two probe pulses are
within the absorption bands for the 360 and 460 nm species
observed by LFP. TR3 spectra were recorded on freshly prepared
buffered aqueous solutions of 2b, but no signal was observed
10 min after mixing 2b with buffered water. This is consistent
with the known rapid decomposition of 2b in water.16 Figure 2
shows the 435.7 nm TR3 spectra obtained in 20 vol% CH3CN/
H2O (A) and in CH3CN (B) 10 ns after flash photolysis of 2b.
The scaled difference spectrum (diff ) (A - B)) was generated
to remove the band υ11 contribution of (B) from (A). The
vibrational frequencies obtained from the difference spectrum
are tabulated in Table 2 with the calculated frequencies obtained
from the fully optimized structure of 1b at the B3LYP/6-31G(d)
level of theory. The tabulated frequencies for spectra (A) and
(B) are available in the Supporting Information (Table S1).
The assigned vibrational modes in the 1100-1700 cm-1
range include all of the most intense predicted IR (IR1-IR4)
and Raman (R1-R7) active modes from the frequency calcula-
tions (the vibrational modes are shown in Figure S1 in the
Supporting Information). Agreement between the calculated and
observed frequencies is generally good (< (25 cm-1 for 8 of
11 assigned bands), but some modes, notably R1 and IR3, have
discrepancies between calculated and observed frequencies of
ca. 35 cm-1. Figure 3 shows a simulation of the observed Raman
spectrum obtained by scaling the calculated IR and Raman
intensities independently of each other to match the intensities
of the observed spectrum.
Results and Discussion
Laser flash photolysis of 2b (2.5 × 10-4 M) at 266 nm in
O2-saturated pH 7.1, 0.02 M phosphate buffer (5 vol% CH3CN/
H2O, µ ) 0.5 (NaClO4), 22 °C) generates two transient
absorption bands that can be detected within 20 ns after the
flash. These bands with λmax 460 and 360 nm, respectively,
decay at significantly different rates (Figure 1A) and therefore
must be associated with two different species. The intermediate
with λmax 460 nm decays in a first-order manner with a lifetime
of (170 ( 10) ns. This species was identified as the oxenium
-
ion 1b because it reacts rapidly with N3 with an apparently
diffusion-limited second-order rate constant, kaz, of (6.6 ( 0.2)
(18) Glover, S. A.; Novak, M. Can. J. Chem. 2005, 83, 1372–1381.
(19) Das, T. M.; Neta, P. J. Phys. Chem. A 1998, 102, 7081–7085.
(20) Das, P. K.; Encinas, M. V.; Steenken, S.; Scaiano, J. C. J. Am. Chem.
Soc. 1981, 103, 4162–4166.
(21) Land, E. J.; Porter, G. Trans. Faraday Soc. 1963, 59, 2016–2026.
(22) Dobson, G.; Grossweiner, L. I. Trans. Faraday. Soc. 1965, 61, 708–
714.
(23) Weiner, S. A. J. Am. Chem. Soc. 1972, 94, 581–584.
(24) Mahoney, L. R.; Weiner, S. A. J. Am. Chem. Soc. 1972, 94, 585–
590.
(25) Weiner, S. A.; Mahoney, L. R. J. Am. Chem. Soc. 1972, 94, 5029–
5033.
(26) Parnell, R. D.; Russell, K. E. J. Chem. Soc., Perkin Trans. 2 1974,
161–164.
(27) Kuzmin, V. A.; Khudjakov, I. V.; Levin, P. P., Jr.; Emanuel, N. M.;
De Jonge, C. R. H. I.; Hageman, H. J.; Biemond, M. E. F.; van der
Maeden, F. P. B.; Mijs, W. J. J. Chem. Soc., Perkin Trans. 2 1979,
1540–15444.
(28) Becker, H.-D. J. Org. Chem. 1965, 30, 982–989.
9
J. AM. CHEM. SOC. VOL. 130, NO. 47, 2008 16023

A R T I C L E S
Wang et al.
Figure 3. Observed and simulated vibrational spectrum of 1b. The
simulated spectrum was obtained by independently scaling the calculated
intensities of the IR active bands (blue) and the Raman active bands (red).
Figure 2. 435.7 nm resonance Raman spectra obtained 10 ns after 266
nm photolysis of 2b in mixed aqueous solution (A) and pure CH3CN (B).
The bottom curve (diff ) A - B) is obtained by subtracting the scaled (B)
from (A), removing the band ν11 contribution of (B) from (A). Star symbols
mark solvent-subtraction artifacts, stray-light, and ambient-light artifacts.
Table 2. Comparison of the Experimental and Calculated
Vibrational Frequencies for 1b
transient resonance
Raman frequency
B3LYP/6-31G(d)
calcd value (cm-1
vibrational mode description
)
shift (cm-1
)
817
969
R7
ArCH bend
1167
1184
1212
1226
1272
1308
1328
1340
1373
1391
1411
1426
1438
1452
1495
1500
1542
1596
1616
1672
1164
1188
IR4 Ar′CH bend, CsCH3 stretch
R6
Ar′CH bend, CsCH3 stretch
1241
Figure 4. Calculated (B3LYP/6-31G(d)) bond lengths (Å), Mulliken
charges (charges on H summed into heavy atoms), and dominant resonance
structures for 1b and the corresponding nitrenium ion, 8b.
IR3 ArsAr′ stretch
R5 CH3 bend
IR2 ArCH bend, ArsAr′stretch
1290
1357
1418
at this level of theory. At the HF/6-31G* level of theory the
cation is planar. There is good evidence for strong bond
alternation throughout the biphenyl moiety, and the C-O bond
length of 1.222 Å is typical of a carbonyl. This suggests the
oxenium ion has significant cyclohexadienyl character in the
phenyl rings and carbonyl character in the carbon-oxygen bond.
This is similar to the significant cyclohexadienyl character in
the phenyl rings and imine character in the carbon-nitrogen
bond of the analogous 4-biphenylylnitrenium ion previously
studied by TR3 spectroscopy and DFT calculations.29 The
highest frequency Raman band in the gas phase, corresponding
to a C-O stretch, is also typical of a carbonyl (1672 cm-1 after
scaling). The observed Raman shift is somewhat lower at 1635
cm-1, but still within the range of a carbonyl with significant
π-delocalization. We note that hydrogen bonding of water at
the carbonyl bond has been previously observed to noticeably
downshift the CdO stretching Raman band for aromatic
carbonyls like para-methoxyacetophenone from ∼1695 cm-1
in cyclohexane to ∼1672 cm-1 in a 50% H2O/50% CH3CN
(v/v) solvent.30 This is quite similar to the downshift in the
vibrational frequency observed for the analogous CdO Raman
R4
R3
Ar′ angle deformation
ArCdC asym. stretch
1512
1525
1568
1593
1635
IR1 Ar′CdC sym. stretch
R2
R1
ArCdC sym stretch
CO stretch
TR3 spectra taken in mixed aqueous solvent and in CH3CN
with the 354.7 nm probe pulse are different from those taken
with the 435.7 nm probe (Figures S2 and S3 in the Supporting
Information). The spectrum recorded in the aqueous solvent
resembles spectrum (A) in Figure 2, but the relative peak
intensities are different and the Raman shifts are somewhat
different. Since we know that the absorbance at 360 nm observed
in the LFP experiments is predominately not due to 1b, and 1b
is not generated by LFP in CH3CN (below), these results are
not surprising. There does appear to be some absorbance due
to 1b at 360 nm, so it is possible that the aqueous solution TR3
spectrum taken with the 354.7 nm probe contains some
contributions from 1b.
(29) Zhu, P.; Ong, S. Y.; Chan, P. Y.; Poon, Y. F.; Leung, K. H.; Phillips,
D. L. Chem.sEur. J. 2001, 7, 4928–4936.
The optimized geometry of 1b is depicted in Figure 4 (the
optimized Z-matrix is given in Table S2 in the Supporting
Information). There is a small inter-ring dihedral angle of 14°
(30) Chan, W. S.; Ma, C.; Kwok, W. M.; Phillips, D. L. J. Phys. Chem. A
2005, 109, 3454–3469.
9
16024 J. AM. CHEM. SOC. VOL. 130, NO. 47, 2008

Oxenium Ions and Aryloxy Radicals
A R T I C L E S
band in the spectrum of the oxenium ion (from 1672 cm-1 for
the DFT gas phase calculation to the experimental 1635 cm-1
seen in the aqueous solution where hydrogen bonding of the
carbonyl is very likely significant). Thus, the hydrogen bonding
of water to the C-O bond most likely accounts for the larger
difference in the Raman vibrational frequencies observed
between the gas phase DFT predicted frequency for the CdO
stretch Raman band and that actually observed in the experi-
mental TR3 spectrum for the oxenium ion. Mulliken atomic
charges confirm delocalization of the positive charge into the
4-para-tolyl substituent. Group charges on the substituent ring
account for 0.52 of the total charge, and charge is concentrated
upon both C-4 (+0.12) and C-4′ (+0.2). The carbonyl bears a
charge of only +0.07. The dipole moment of 4.45 D is directed
along the longitudinal axis of the molecule, as expected. In
valence bond terms, resonance structures II and III, rather than
I (Figure 4), are dominant. We have found similar results for
1a from calculations at the HF/6-31G* and pBP/DN*//HF/6-
31G* levels of theory.18 The agreement between the calculated
and observed vibrational spectrum for 1b indicates that the
calculated properties of the cation do reproduce those of the
actual cation.
The computed properties of the analogous nitrenium ion 8b
(Table S3 in the Supporting Information) at the B3LYP/6-
31G(d) level are similar to 1b. Less charge is localized into the
4-para-tolyl substituent (+0.45) though Mulliken charges at the
C-4 and C-4′ positions are similar. There is a slightly larger
inter-ring dihedral angle of 16.7°. This and the longer inter-
ring bond indicate less resonance through to the 4-para-tolyl
substituent. Bond alternation is also somewhat reduced through-
out the structure, and nearly twice as much positive charge
resides on the CNH of 8b (0.112) than on the CO of 1b. The
dipole moment of 1.479 D is considerably modified by the NH
bond. The component of the dipole moment along the C-N
bond, corrected for the N-H bond contribution, is estimated to
be 1.22 D, considerably less than the dipole moment of 1b.
Similar differences in calculated properties have been noted for
other aryloxenium ions, including 1a, and their analogous
arylnitrenium ions at the HF/6-31G* and BP/DN*//HF/6-31G*
levels of theory.18
It is initially surprising that the IR active bands (IR1-4)
appear as significant bands in the TR3 spectrum of 1b. Figure
4 shows that the strong IR bands are also weakly allowed in
the calculated Raman spectrum due to the low symmetry of
1b. However, the observed intensities of the TR3 spectrum are
quite different from the calculated Raman intensities. The
intensities of resonance Raman spectra can be very different
from those observed or calculated in Raman spectra obtained
under nonresonant conditions.31 The probe Raman excitation
wavelength is in resonance with an electronic transition of the
species being probed. The bands in the resonance Raman
spectrum associated with vibrational modes of molecular
components that undergo a significant change in structure in
the excited state relative to the ground state will be enhanced.31
For example, selective probe wavelength-dependent resonance
Raman enhancement of bands assigned to vibrational modes
of the benzoin chromophore in benzoin diethyl phosphate has
been demonstrated.31a In the present example, the Raman bands
R1 and R2 and the ΙR band IR1 are CdO, ArCdC, and Ar′CdC
stretch modes and are fairly intense bands in the TR3 spectrum.
Figure 5. HOMO-1 (π, top), nonbonding (πY, middle), and LUMO (π,
bottom) orbitals for 1b calculated at the B3LYP/6-31G(d) level.
The HOMO-1 (π), nonbonding (πY), and LUMO (π) orbitals
are depicted in Figure 5. The π-π* and n-π* transitions lead
to changes in bond order of the ArCdC and Ar′CdC as well
as the CdO bonds and hence the lengths of these bonds in the
excited state. This change in excited-state structure would be
expected, as observed, to enhance vibrational modes associated
with changes in these bond lengths.
Characterization of the Aryloxy Radical. The decay of the
transient absorbance at 360 nm in Figure 1A occurs in a biphasic
manner. The absorbance vs time data at t g 1 µs (to avoid
complications from absorbance of 1b at that wavelength) can
be fit to an equation containing two first-order rate constants
O2
(k1AQ and k2AQO2) that are not dependent on [N3-]. All rate
constants are reported in Table 1.The lifetimes of these decay
processes (12 and 75 µs) are significantly longer than that of
1b.17 Assignment of the 360 nm absorbance band to a radical
intermediate 5b was based on the LFP generation of the same
intermediate in CH3CN, comparison of the observed UV-vis
spectra attributed to 5a and 5b with previously published spectra
for 5a,19,20 isolated photolysis reaction products of 2a and 2b
that are attributable to 5a and 5b, and modeling of the decay
kinetics of 5a and 5b.
LFP of 2b in O2-saturated CH3CN (Figure 1B) apparently
generates only one transient. The absorbance observed at 460
nm in aqueous solution that was assigned to 1b has disappeared.
The strong transient absorbance band at shorter wavelengths
remains, but λmax has shifted to ca. 350 nm. A much weaker
band at longer wavelengths (>460 nm) is also observed. Both
of these bands disappear with biphasic first-order kinetics and
with equivalent rate constants (Table 1). This indicates that both
absorbance bands are associated with the same intermediate.
O2
O2
The magnitudes of these rate constants (k1ACN and k2ACN
)
are similar to those observed for the decay of the 360 nm band
in O2-saturated aqueous solution indicating that this absorbance
band is due to a neutral intermediate not strongly affected by
solvation that is equivalent to the intermediate with λmax 360
nm detected in water. The lack of generation of 1b in the
CH3CN solutions is consistent with solvation-based destabiliza-
(31) (a) Chan, W. S.; Ma, C.; Kwok, W. M.; Zuo, P.; Phillips, D. L. J.
Phys, Chem. A 2004, 108, 4047–4058. (b) Du, Y.; Ma, C.; Kwok,
W. M.; Xue, J.; Phillips, D. L. J. Org. Chem. 2007, 72, 7148–7156.
9
J. AM. CHEM. SOC. VOL. 130, NO. 47, 2008 16025

A R T I C L E S
Wang et al.
Scheme 2. Photolysis Product Yields
Figure 6. Transient absorbance spectra obtained 20 ns after 266 nm
excitation of 2a in O2-saturated pH 7.1 phosphate buffer (blue) or in O2-
saturated CH3CN (red). Spectra recorded over a 20 ns window.
tion of the transition state leading to the cation in CH3CN under
conditions in which a competing homolytic photolysis to
generate 5b is not strongly affected by the solvent. In Ar-
saturated CH3CN the band at 350 nm and the weak long
wavelength band are still observed. In addition, a prominent
shoulder at 400 nm that disappears within 10 µs is detected
(Figure S4 in the Supporting Information). This rapidly decaying
transient absorbance does not interfere with the decay of the
two major absorbance bands. Monitoring of the kinetics at 350
and 575 nm shows that the intermediate disappears in a biphasic
is concluded that the transient spectra in Figure 6 are due to 5a
and that the similar transient absorbance observed in Figure 1
is due to 5b. The kinetics of decay of 5a was examined in less
detail than that of 5b, but the transient absorbance at 340 nm
in O2-saturated CH3CN decays in a biphasic manner very similar
to that observed for 5b. Rate constants are shown in Table 1.
Photolysis Products. Reaction products from steady-state
photolysis of 2a and 2b are summarized in Scheme 2. The
quinols (3) and the phenols (6) are known compounds.14,16 The
new peroxy compounds (7) were fully characterized. All
compounds were isolated for identification, but yields are based
on HPLC response factors. All reactions except the aqueous
solution photolysis of 2b were performed with an initial ester
concentration of 2.5 mM. The aqueous solution photolysis of
2b was performed with an initial ester concentration of 1 ×
10-4 M.17 Both 3 and 7 are significantly photoreactive under
our conditions. To minimize this effect, product yields for the
reactions run at 2.5 mM were obtained at early reaction times
corresponding to ca. 15-30% photoreaction. The aqueous
solution photolysis of 2b at lower concentration led to ca. 95%
conversion after 45 s of irradiation. In this case the observed
yields of 3b and 7b were corrected for the observed photolytic
decomposition of the authentic materials under the reaction
conditions.17 Other minor products were detected by HPLC of
photolysis mixtures. Most were not characterized, but one minor
product of 2a with a long HPLC retention time has an m/e of
337 by negative ion mass spectrometry. This mass corresponds
to (M - 1)- for a dimer of 5a. Since this material can be
detected by negative ion mass spectrometry, it is likely to be a
phenol, but it was not isolated since it is ,1% of the photolysis
product mixture of 2a in CH3CN. The peroxy product 7a is not
produced in N2-saturated CH3CN, and experiments in CD3CN
demonstrate that the methyl group in 7a is derived from the
acyl methyl of 2a, not from the solvent. The quinol product 3a
or 3b can be detected from the steady-state photolysis of 2a or
2b in aqueous solution, but these products cannot be detected
during photolysis in CH3CN. The quinols are also generated as
the exclusive hydrolysis products of 2a and 2b in the absence
of other nucleophiles.14,16 The yields of 3a and 3b shown in
Scheme 2 were either obtained under conditions in which the
Ar
manner with rate constants (k1ACN and k2ACNAr). The magni-
tudes of these rate constants are somewhat smaller than those
observed in the presence of O2.
Results of LFP of 2a in O2-saturated pH 7.1 phosphate buffer
or in O2-saturated CH3CN are shown in Figure 6. No transient
that can be assigned to 1a is detected in the aqueous solution.
Although steady state photolysis product studies indicate that
some 1a is formed (below), it is not surprising that it was not
detected in the ns-LFP experiments because the lifetime of 1a
estimated from an N3--trapping product study performed during
hydrolysis of 2a under the same aqueous conditions is only 12
ns.14 Monitoring absorbance vs time at 450 nm (near λmax for
1b) at early reaction times reveals an apparent fluorescence that
decays within 15 ns after the flash. After that, slow decay of
the signal occurs on a longer 10-100 µs time scale. The
fluorescence signal would obscure any absorbance due to 1a at
short reaction times within the estimated lifetime of 1a. The
-
LFP results confirm our previous conclusions based on N3
-
trapping of hydrolysis reactions of oxenium ion precursors that
the aqueous solution lifetimes of 4′-substituted-4-biphenyly-
loxenium ions are highly dependent on the nature of the 4′-
substituent.16 These results are also in agreement with the
calculations that show significant delocalization of the positive
charge into the distal rings of these cations.16-18
Although 1a was not detected, another transient with λmax
350 nm was detected in aqueous solution. In CH3CN a transient
is generated with λmax 340 nm. This solvent shift is equivalent
to that observed for 5b, although in both solvents the transient
absorption maximum for 5b is shifted to a longer wavelength
by ca. 10 nm. In both solvents a weaker absorbance band,
apparently associated with the same transient, is observed at λ
> 460 nm. The spectrum observed in aqueous solution is
indistinguishable from that previously observed during pulse
radiolysis of N2O-saturated aqueous solutions of 6a that was
attributed to 5a.19 A very similar absorbance spectrum attributed
to 5a was observed in 1:2 benzene/di-tert-butyl peroxide.20 It
9
16026 J. AM. CHEM. SOC. VOL. 130, NO. 47, 2008

Oxenium Ions and Aryloxy Radicals
A R T I C L E S
Scheme 3. Possible Mechanism for the Formation of 7
Scheme 4. Kinetic for the Decay of 5
hydrolysis is too slow to compete with the photolysis reaction
(3a) or corrected for the background hydrolysis reaction (3b).17
Both 6 and 7 are detected during photolysis of 2a and 2b in
O2-saturated aqueous solution. This is consistent with observa-
tion of 5a and 5b during LFP of 2a and 2b in water.
4 has been used to analyze these cases.21,26 The dimer 9 was
not isolated, but its structure is assumed to be analogous to the
isolatable but highly reactive dimer obtained from DDQ
oxidation of 2,6-di-tert-butyl-4-methylphenol.28 This scheme
was applied to the decay of the absorbance of 5b at 360 nm in
O2-saturated 5% CH3CN/H2O or at 350 nm in both O2- and
Ar-saturated CH3CN and to the decay of the absorbance of 5a
at 340 nm in O2-saturated CH3CN. The scheme was utilized
because of the biphasic nature of the decay of the absorbance
under these conditions and because the negligible yields of
dimeric products observed in these photolyses suggest that 9
does not have an efficient pathway to lead directly to stable
dimeric products. The monomeric products 6 and 7 are clearly
obtained from 5 in processes that are not kinetically first-order,
but a pseudo-first-order rate constant, k3, can be used to describe
these processes under reaction conditions in which species other
than 5 are held at constant concentration. Since the initial
concentrations of 5a and 5b in the LFP experiments are e3 ×
10-5 M (see below) this may be the case.
The quinols 3a and 3b have been shown by solvent 18O-
labeling studies and N3--trapping experiments to be the products
of the attack of H2O on the oxenium ions 1a and 1b that are
generated during hydrolysis of 2a and 2b.14,16 The lack of
generation of 3b in CH3CN is consistent with the LFP
observations that show 1b is not detected in CH3CN, although
it is generated in aqueous solution. Although 1a was not directly
detected by LFP, the generation of 3a in the aqueous solution
photolysis experiments indicates that this short-lived intermedi-
ate is generated by photolysis. The greater yield of 3b compared
to 3a in the aqueous photolysis experiments may be due to
transition state stabilization for photolytic generation of the more
stable cation. The products 6 and 7 have not been detected
during hydrolysis of 2a or 2b and are clearly generated from
the intermediates produced by photolysis in CH3CN. The
phenols 6 are the expected products of hydrogen abstraction of
5. A mechanism for generation of the peroxy compounds 7 that
is consistent with the available data is shown in Scheme 3.
Simple alkyl radicals, including a methyl radical, react with O2
to form alkyl peroxyl radicals with rate constants of ca. 109
M-1 s-1, while aryloxy radicals are unreactive with O2.32,33
Primary alkyl peroxyl radicals are relatively unreactive to
hydrogen abstraction, double bond addition, and electron
transfer.32 Self-reaction occurs near the diffusion limit, but at
low peroxyl radical concentration trapping by other radicals,
such as 5, should be competitive.32
Absorbance vs time data were fit to the mechanism of Scheme
4 using a Powell fitting algorithm implemented in the KINTE-
CUS kinetics simulation program.34 The three rate constants
k1/ꢀ, k2, k3 and the initial (t ) 0) absorbance were used as
variable parameters to optimize the fit. Examples of fits are
shown in Figure 7, and fitted rate constants are gathered in
Table 3.
Evaluation of the rate constant k1 requires knowledge of the
extinction coefficient, ꢀ, for 5a or 5b at its short wavelength
λmax in the 340-360 nm range. The previously measured
extinction coefficient for 5a at the long wavelength absorbance
band (502 nm) in 1:2 benzene/di-tert-butyl peroxide of 2580
M-1 cm-1 provides an estimate of ꢀ at 340 nm for 5a of 1.9 ×
104 M-1cm-1 based on the observed absorbance ratio at λmax
of the two bands for 5a in CH3CN.20 The same extinction
coefficient was assumed for 5b at its short wavelength λmax in
both CH3CN and the aqueous solution. The error limit for ꢀ
may exceed (50%, but it provides a reasonable estimate of the
initial concentration of 5a or 5b of ca. (1-3) × 10-5 M in all
the LFP experiments. This is 3%-10% of the initial concentra-
tions of 2a or 2b present in solution. Estimates for k1 are
provided in Table 3. The values are 5-10-fold less than the
expected diffusion limit. Previously measured dimerization rate
constants of aryloxy radicals near room temperature have varied
from ca. 107 M-1 s-1 to the approximate diffusion controlled
limit of (5-6) × 109 M-1 s-1 in solvents as diverse as benzene,
Kinetic Modeling of the Decay of the Aryloxy Radicals. The
biphasic decay kinetics observed for 5a and 5b can be fit by a
double exponential rate equation, but this fit does not provide
a mechanistic understanding of the decay processes for these
radicals. It is known that aryloxy radicals are relatively long-
lived species in solution that decay primarily by dimerization
in the absence of radical traps.21-27 In many cases in which
there is a 4-alkyl or 4-aryl substituent, including 5a, the
dimerization is reversible and biphasic decay kinetics can be
observed.21,24,26 A mechanism similar to that shown in Scheme
(32) (a) Neta, P.; Huie, R. E.; Ross, A. B. J. Phys. Chem. Ref. Data 1990,
19, 413–513. (b) Marchaj, A.; Kelley, D. G.; Bakac, A.; Espenson,
J. H. J. Phys. Chem. 1991, 95, 4440–4441.
(33) (a) von Sontag, C.; Schuchmann, H.-P. Angew. Chem., Int. Ed. Engl.
1991, 30, 1229–1253. (b) Hunter, E. P. L.; Desrosiers, M. F.; Simic,
M. G. Free Radical Biol. Med. 1989, 6, 581–585. (c) Berho, F.;
Lesclaux, R. Chem. Phys. Lett. 1997, 279, 289–296. (d) Platz, J.;
Nielson, O. J.; Wallington, T. J.; Ball, J. C.; Hurley, M. D.; Straccia,
A. M.; Schneider, W. F.; Sehested, J. J. Phys. Chem. A 1998, 102,
7964–7974.
(34) Ianni, J. C. Kintecus, Windows Version 3.95, 2008, http://www.
kintecus.comm/.
9
J. AM. CHEM. SOC. VOL. 130, NO. 47, 2008 16027

A R T I C L E S
Wang et al.
CCl4.21,26 For a more closely related radical, 2,4-dicyclohexyl-
4-phenylphenoxyl, Kd is 2.5 × 104 M-1 at 20 °C in propanol.27
Conclusions
While hydrolysis of the quinol esters 2 with electron-donating
substituents on the distal ring generates the aryloxenium ions 1
exclusively based on the quantitative yield of the quinols 3 and
results of N3--trapping, common ion effects, and 18O-labeling
studies,14-16 photolysis of these esters in aqueous solution
generates both 1 and aryloxy radicals 5. Product yields from
photolysis of 2a and 2b suggest that photolytic generation of 1
is more favorable for the ester with the more electron-donating
substituent, 2b. Only 1b was detected directly after nanosecond
LFP at 266 nm because the lifetime of 1a is too short for
detection. The ion 1b with λmax 460 nm was characterized by
the N3--dependence of its decomposition kinetics, by compari-
son of kaz/ks measured directly with the same ratio measured
by N3--trapping of the hydrolysis reaction, and by the com-
parison of the TR3 spectrum of the intermediate with calculated
vibrational transitions at the B3LYP/6-31G(d) level of theory.
The agreement between calculated and observed transitions
indicates that the calculated structure is an accurate representa-
tion of the cation. That structure indicates that 1b should largely
be considered as a 4-(para-tolyl)-1-oxo-2,5-cyclohexadienyl-
carbenium ion (resonance structure II of Figure 4) with
significant charge delocalization into the substituent ring. The
strong dependence of cation lifetimes on the substituent in the
distal ring and the nature of the hydrolysis reaction products,
3, are consistent with this interpretation.
Figure 7. Kinetic fits for absorbance of 5b at 350 nm vs time in (A) O2-
saturated CH3CN and (B) Ar-saturated CH3CN. Blue lines: the fit to a
biphasic first-order rate equation (Table 1). Green lines: the fit to the kinetic
mechanism of Scheme 4 (Table 3). ∼90% of the data points were removed
for clarity, but all data points were used for the fits.
CCl4, and water.21-25,27 The dimerization rate constant of 5a
The excellent agreement of the direct measurement of the
in benzene at 30 °C is reported to be 3.4 × 107 M-1 s-1 23
.
-
lifetime of 1b with the indirect measurements made by N3
-
Dimerization rate constants for several other aryloxy radicals
in benzene are about an order of magnitude smaller than those
for the same radicals in water, so our values for k1 appear to be
in reasonable agreement with the limited available data.23
The difference between k1 values in aqueous solution and in
CH3CN for 5b should not be over interpreted considering our
assumptions concerning the magnitude of ꢀ. Comparisons of
the rate constants for 5b in O2-saturated CH3CN with those in
Ar-saturated CH3CN show that the most significant difference
is in k3. The two rate constants that describe the dimerization
equilibrium (k1 and k2) are quite similar under the two reaction
conditions, and the dimerization equilibrium constant, Kd ) k1/
k2, is invariant at 3.1 × 104 M-1 under both reaction conditions,
but k3 in O2-saturated CH3CN is 2.4-fold larger than that in
Ar-saturated CH3CN. The relatively small effect of O2 on the
decay kinetics of 5b is consistent with the reported lack of
reactivity of aryloxy radicals with O2.33 The observed increase
in k3 in O2-saturated CH3CN is consistent with the proposed
role of O2 in the formation of 7b (Scheme 3). The dimerization
equilibrium for 5a in O2-saturated CH3CN appears to be very
similar to that of 5b with a Kd of 2.4 × 104 M-1, but k3 is
somewhat smaller for 5a in O2-saturated CH3CN than k3 for
5b under the same reaction conditions. We do not understand
the reasons for the substituent effect on k3. Speculation on the
nature of the effect would not be warranted at this time, since
the decay processes represented by k3 are quite complicated and
we have not identified all of the products generated from the
decay of 5a and 5b. The dimerization equilibrium constants
measured here for 5a and 5b appear to be in the range of Kd
measured for other aryloxy radicals.21,26,27 For example, Kd for
2,6-di-tert-butyl-4-methylphenoxyl has been measured at room
temperature as 6 × 104 M-1 in benzene and 1 × 106 M-1 in
trapping of the hydrolysis reaction of 2b provides confidence
in the indirect lifetime estimates we have made for several other
4′-substituted biphenylyloxenium ions (MeO, H, Br).16,17 Ren
and McClelland have measured the lifetimes of the correspond-
ing nitrenium ions under similar conditions (20 vol% CH3CN/
H2O, 20 °C).35 A plot of log ks for the biphenylyloxenium ions
vs log ks for the corresponding biphenylylnitrenium ions (Figure
8) shows that relative substituent effects on the stabilities of
these ions are nearly identical (slope ≈ 1.0), although the
nitrenium ions are ca. 30-fold more stable than the corresponding
oxenium ions within the 3 orders of magnitude range of cation
stabilities that can be compared. This is surprising given the
results of the DFT calculations that show somewhat more charge
delocalization into the distal rings of the biphenylyloxenium
ions. This suggests that the oxenium ions would be more
sensitive to substituent effects in the distal ring, which would
be manifested by a slope > 1.0. The calculated differences are
not large though, and compensating factors such as differential
solvation of the more polar oxenium ions could mask the
expected effect.
Although 1b is only generated during photolysis of 2b in
water, the aryloxy radical 5b can be detected after LFP in both
aqueous solution and CH3CN. The unsubstituted radical 5a can
also be detected after LFP of 1a in both solvents. The radicals
appear to be the exclusive initial photolysis products in CH3CN,
based on UV-vis spectra taken after LFP and on isolated
product yields. These radicals were characterized by their
reaction products, by comparison with the published UV-vis
spectrum of 5a, and by modeling of their decay kinetics.
(35) Ren, D.; McClelland, R. A. Can. J. Chem. 1998, 76, 78–84.
9
16028 J. AM. CHEM. SOC. VOL. 130, NO. 47, 2008

Oxenium Ions and Aryloxy Radicals
A R T I C L E S
Table 3. Rate Constants Derived from Kinetic Modeling to Scheme 4
intermediate, reaction conditionsa
k1/ꢀ (cm s-1
)
k1 (M-1 s-1
)
k2 (s-1
)
k3 (s-1
)
5b, 5% CH3CN/H2O, O2 sat
5b, CH3CN, O2 sat
5b, CH3CN, Ar sat
(9.9 ( 0.5) × 104
(4.1 ( 0.2) × 104
(3.0 ( 0.1) × 104
(2.4 ( 0.2) × 104
(1.9 ( 0.9) × 109
(7.8 ( 3.9) × 108
(5.7 ( 2.9) × 108
(4.6 ( 2.3) × 108
(1.7 ( 0.1) × 104
(2.5 ( 0.1) × 104
(1.8 ( 0.2) × 104
(1.9 ( 0.3) × 104
(2.8 ( 0.2) × 104
(5.2 ( 0.2) × 104
(2.2 ( 0.1) × 104
(1.1 ( 0.3) × 104
5a, CH3CN, O2 sat
a Reaction conditions are the same as those described in Table 1. Average rate constants were obtained from fits of three to five independent runs
under each set of reaction conditions. The error limits for k1/ꢀ, k2, and k3 are the standard deviations of the average values. The error limit for k1 also
includes an estimate of the error limit in the value of ꢀ.
time spans that ranged from 1.25 to 250 µs depending on quenching
rates of individual intermediates. Kinetics measurements for 2a were
made at 340 nm in CH3CN. Kinetics measurements at all
wavelengths were averaged three to five times. Between each
measurement, the sample was remixed to avoid depletion of
reactants within the volume of the cuvette exposed to the flash.
Kinetics traces were fit to either a standard first-order rate equation
containing a single exponential (460 nm) or a rate equation
containing a double exponential (340, 350, 360, 575 nm). All data
at t g 10 ns were used for the first-order fits at 460 nm; delays of
1 to 7 µs were used for the data taken at the other wavelengths.
The delays were necessitated by fast decaying transients that
disappeared at early reaction times.
Fitting of the absorbance vs time data taken at 360 nm in aqueous
solution and at 340 and 350 nm in CH3CN to the mechanism of
Scheme 4 was performed with the Kintecus software package.34
A
small background absorbance that remained after decay of the
transient of ca. 0.005-0.02 AU (<5% of the initial absorbance)
was subtracted from the data before curve fitting. The four
parameters k1/ꢀ, k2, k3, and the initial absorbance were used as
adjustable parameters to optimize the fit. Approximately 1.2 × 104
data points were used in each individual fit. The Powell fitting
algorithm was employed. Adjusted r2 values for individual fits were
in the range 0.85-0.91 with estimated standard deviations of the
model of 0.005-0.015 AU. Pearson’s r coefficient for each
individual fit was in the range 0.90-0.96, and the Shapiro-Wilk
W-test for Gaussian normality gave W ) 0.96-0.98 with P )
0.14-0.33. Standard deviations of the individual fitting parameters
were estimated from averages of 3 to 5 independent kinetic runs
under each reaction condition.
Steady State Photolysis Experiments. The detailed procedures
for steady state photolysis in aqueous solutions in a Rayonet
photochemical reactor in a jacketed quartz vessel kept at 15 °C
have been published.17 The following modifications were made for
photolysis in CH3CN. UVC lamps that have emission (ca. 90% of
total lamp energy) in the range 235-280 nm with a sharp maximum
at 254 nm were used as the UV source. The desired amount of 2a
or 2b was dissolved in ca. 1 mL of CH3CN. This solution was
transferred into 100 mL of O2 or N2 saturated CH3CN in the
photolysis reactor by disposable pipet. Reaction solutions were
generated with initial concentrations of 2.5 × 10-3 M. Slow O2 or
N2 bubbling was continued throughout the photolysis. The pho-
tolysis was initiated after the addition of 2a or 2b was complete,
and the reaction solution was fully mixed. Irradiation times were
variable, ranging from 3 to 15 min. The progress of photolysis as
a function of irradiation time was monitored by HPLC. The HPLC
conditions for monitoring the photolysis of 2b have been pub-
lished.17 The HPLC conditions for 2a were slightly modified (C-8
reversed phase column, 60/40 MeOH/H2O eluent, 1 mL/min,
monitored by UV absorbance at 221 and 238 nm).
Figure 8. Plot of log ks for substituted biphenylyl oxenium ions (O) vs
log ks for the corresponding nitrenium ions (N).
Although the esters 2 with electron-donating substituents on
the distal ring generate trappable oxenium ions during hydroly-
sis, those with electron-withdrawing substituents (R ) CN, NO2)
do not.16 Instead, these esters react directly with nucleophiles
-
such as N3 via kinetically bimolecular processes that appear
to be SN2′ in nature.16 The photochemistry of these electron-
poor esters has not been reported. Although they are unlikely
to lead to detectable singlet oxenium ions, they could be sources
of triplet oxenium ions and/or aryloxy radicals.18 We have begun
to explore the photochemistry of these esters with an emphasis
on their triplet sensitized photochemistry.
Experimental Section
Laser Flash Photolysis and Kinetics. Laser flash photolysis
(LFP) was carried out using a Nd:YAG laser (266 nm, ca. 5 ns
pulse) with a 1 cm light path. All reactions were performed at
ambient temperature (22 °C). Solutions of 2a and 2b were prepared
in O2 or Ar saturated pH 7.1, 0.02 M phosphate buffer (5 vol%
CH3CN-H2O, µ ) 0.5(NaClO4)) or in O2 or Ar saturated CH3CN
by injecting 15 µL of a ca. 0.04 M stock solution of 2a or 2b into
3 mL of solution, so that the initial concentrations were ca. 2.5 ×
10-4 M. These solutions had an absorbance of ca. 0.50 to 0.75 at
266 nm with a 1 cm light path. Since both 2a and 2b undergo
slow hydrolysis under these conditions, all solutions were used
promptly after mixing, and experiments were completed within 5
min after mixing. Concentrations of N3- were in the range from 0
to 6 mM.
Transient absorbance spectra were monitored in the range
280-540 nm. Initial spectra were obtained with a delay of 20 ns
after the laser pulse. Subsequent spectra were obtained at times
dictated by the quenching of transient bands generated by LFP of
2b at 360 and 460 nm in aqueous solution and at 350 and 575 nm
in CH3CN, and for 2a at 350 nm in aqueous solution and 340 nm
in CH3CN. All transient spectra were collected over a 20 ns time
window. Kinetics measurements for 2b were made at 360 and 460
nm in aqueous solutions and at 350 and 575 nm in CH3CN, for
Product Studies. The quinols 3a and 3b and the phenols 6a
and 6b were identified by direct comparison to authentic samples.14,16
Quantification of all products was obtained by calibration of HPLC
peak areas with those of known concentrations of the photolysis
products. A general procedure for isolation of photolysis products
7a and 7b is as follows. When the yield of these products reached
a maximum, based on HPLC analysis (typically after 9-12 min of
irradiation at initial concentrations of 1a or 1b of 2.5 × 10-3 M),
9
J. AM. CHEM. SOC. VOL. 130, NO. 47, 2008 16029

A R T I C L E S
Wang et al.
CH3CN was removed from the reaction mixture by room temper-
ature rotary evaporation under vacuum. An initial separation of
photoproducts was performed by radial chromatography on silica
gel using 60/40 hexanes/EtOAc. The fraction containing 7 was
evaporated to dryness under vacuum, and the residue was redis-
solved in 10 mL of CH2Cl2 and washed with cold 1 M aqueous
NaOH (2 × 5 mL) and cold saturated aqueous NaHCO3 (2 × 5
mL). The CH2Cl2 solution was dried over Na2SO4 before the solvent
was evaporated. The final purification of 7 was carried out by radial
chromatography on silica gel with a different solvent system: 97.5/
2.5 CH2Cl2/CH3CN. Purified products were subjected to HPLC
analysis before final characterization.
through a depolarizer and the entrance slit of a 0.5 m spectrograph
that dispersed the light onto a liquid nitrogen cooled CCD detector.
The Raman signal was accumulated for 30 s before being read out
to an interfaced PC computer, and ∼10 to 20 of these Raman spectra
were averaged to increase the signal-to-noise ratio of the spectrum
obtained. The TR3 spectra were determined by subtracting the
pump-probe spectrum at negative 100 ns from the pump-probe
spectra acquired at positive delay times so as to remove the solvent
and precursor Raman bands. The known vibrational wavenumbers
of the mixed solvent CH3CN Raman bands were employed to
calibrate the wavenumbers of the TR3 spectra to an absolute
accuracy of ∼ (3 cm-1 (and a relative accuracy of (1 to 2 cm-1
from scan to scan).
4-(Methylperoxy)-4-phenylcyclohexa-2,5-dienone (7a). Clear
oil at room temperature. IR (thin film) 3060 (w), 2965 (w), 1668,
1630, 1449, 1390, 1165, 1063, 1000, 855, 752, 696 cm-1; 1H NMR
(500 MHz, CD2Cl2) δ 3.91 (3H, s), 6.31(2H, d, J ) 10.5 Hz), 7.03
(2H, d, J ) 10 Hz), 7.38 (5H, m); 13C NMR (125.8 MHz, CD2Cl2)
δ 64.44, 81.35, 126.27, 129.26, 129.33, 129.73, 137.13, 148.50,
185.60; LC/MS (ESI, positive) m/e 192 (M + Na - CH3O2)+, 217
(M + H)+, 239 (M + Na)+; High-resolution MS (ES, positive),
C13H12O3Na (M + Na) requires 239.0684, found 239.0694.
4-(Methylperoxy)-4-(4′-methylphenyl)cyclohexa-2,5-dienone
(7b). Clear oil at room temperature. IR (thin film) 3030 (w), 2963
(w), 1668, 1630, 1390, 1165, 1065, 1005, 958, 856, 815 cm-1; 1H
NMR (500 MHz, CD2Cl2) δ 2.34 (3H, s), 3.90 (3H, s), 6.30 (2H,
d, J ) 10 Hz), 7.03 (2H, d, J ) 10.5 Hz), 7.18 (2H, d, J ) 8.0
Hz), 7.29 (2H, d, J ) 8.5 Hz); 13C NMR (125.8 MHz, CD2Cl2) δ
21.19, 64.41, 81.25, 126.18, 129.57, 129.90, 134.02, 139.56, 148.73,
185.63; LC/MS (ESI, positive) m/e 206 (M + Na - CH3O2)+, 253
(M + Na)+, 269 (M + K)+; High-resolution MS (ES, positive),
C14H14O3Na (M + Na) requires 253.0841, found 253.0817.
Nanosecond Time-Resolved Resonance Raman (ns-TR3)
Experiments. Stock solutions of 10 mM 2b were prepared in
CH3CN and kept cool in an ice-water bath. 0.015 M Na2HPO4/
0.005 M NaH2PO4 buffered distilled water and pure CH3CN were
also kept in the ice-water mixture. In each run of the experiment,
4 mL of 2b stock solution were mixed with 16 mL of buffered
water or CH3CN to make a fresh 20 mL sample solution, which
was immediately subjected to the TR3 experiment. The experiment
was accomplished within 2 min after the solution had been made
to avoid any noticeable sample decomposition or photoproduct
accumulation.
Density Functional Calculations. Density functional calcula-
tions on 1b and 8b were carried out using the Gaussian 03 suite of
programs.38 Geometry optimization of the ground-state oxenium
ion was executed at the B3LYP/6-31G(d) level, and its verification
as a minimum and derivation of IR and Raman vibrational
frequencies and normal modes of vibration were executed using a
harmonic frequency analysis, which afforded all real frequencies.
Vibrational frequencies for comparison with experimental data were
scaled by a factor of 0.9614 according to Scott and Radom.39
For comparative purposes, computed Raman intensities were
weighted by a factor of 40 relative to the IR intensities to emulate
relative experimental intensities of the most intense IR and Raman
active bands in the resonance Raman spectrum. The combined,
weighted theoretical spectrum was then scaled to approximate
absolute resonance Raman intensities in the experimental spectrum
which was first subjected to a biphasic baseline correction about
the Rayleigh frequency.
Acknowledgment. We thank the Donors of the American
Chemical Society Petroleum Research Fund (Grant No. 43176-AC4)
for support of this work. Y.-T.W. thanks the Department of
Chemistry and Biochemistry at Miami University for a Dissertation
Fellowship, and K.J.J. thanks Miami University for an Undergradu-
ate Summer Scholars award. The support of the NSF in Columbus
and OSU Center for Chemical and Biophysical Dynamics is greatly
appreciated. J.W. thanks the OSU Graduate School for a Presidential
Fellowship. This work was supported in part by a grant from the
Research Grants Council (RGC) of Hong Kong (HKU 7040/06P)
to D.L.P. D.L.P. thanks the Croucher Foundation for the award of
a Croucher Foundation Senior Research Fellowship (2006-07) and
the University of Hong Kong for an Outstanding Researcher Award
(2006).
A brief description of the apparatus for the ns-TR3 experiment
is given here. A more detailed description is published elsewhere.36,37
Two Nd:YAG nanosecond pulsed lasers in conjunction with
hydrogen Raman shifters were used to produce the different laser
wavelengths employed in the ns-TR3 experiments. The fourth
harmonic of a Nd:YAG laser (266 nm) was employed as the pump
pulse. The third harmonic of a Nd:YAG laser (354.7 nm) and the
first anti-Stokes Raman shifted line of the 532 nm second harmonic
(435.7 nm) were employed as the probe laser pulses. Two Nd:
YAG lasers here were electronically synchronized to each other
by a pulse delay generator used to control the relative timing of
two lasers. The relative timing of these two lasers was monitored
using a fast photodiode and a 500 MHz oscilloscope. The jitter
between the pump and probe laser pulses was observed to be less
than 5 ns. The laser beams were lightly focused onto a flowing
liquid stream of the sample using a near collinear backscattering
geometry. The Raman scattered light was collected using a
backscattering geometry and reflective optics that imaged the light
Supporting Information Available: Table S1, containing all
vibrational frequencies obtained during the TR3 experiments;
Figure S1, containing the vibrational modes calculated for 1b;
Figure S2, containing the 354.7 nm TR3 spectra obtained after
flash photolysis of 2b; Figure S3, containing a comparison of
the TR3 spectra obtained using 354.7 and 435.7 nm probe lasers,
in mixed aqueous solvent (20% CH3CN 80% buffered water)
and CH3CN; Figure S4, containing the transient absorbance
spectra recorded after 266 nm LFP of an Ar-saturated CH3CN
solution of 2b; Tables S-2 and S-3, containing the Z-matrices
for the optimized structures of 1b and 8b at the B3LYP/6-
31G(d) level of theory; 13C NMR spectra for 7a and 7b, an
HMQC spectrum for 7a, and complete ref 38. This material is
available free of charge via the Internet at http://pubs.acs.org.
(36) (a) Li, Y.-L.; Leung, K. H.; Phillips, D. L. J. Phys. Chem. A 2001,
105, 10621–10625. (b) Li, Y. L.; Chen, D. M.; Phillips, D. L. J. Org.
Chem. 2002, 67, 4228–4235.
(37) (a) Chan, P. Y.; Kwok, W. M.; Lam, S. K.; Chiu, P.; Phillips, D. L.
J. Am. Chem. Soc. 2005, 127, 8246–8247. (b) Xue, J.; Guo, Z.; Chan,
P. Y.; Chu, L. M.; But, T. Y. S.; Phillips, D. L. J. Phys. Chem. A
2007, 111, 1441–1451. (c) Xue, J.; Chan, P. Y.; Du, Y.; Guo, Z.;
Chung, C. W. Y.; Toy, P. H.; Phillips, D. L. J. Phys. Chem. B 2007,
111, 12676–12684.
JA805336D
(38) Frisch, M. J. et al. Gaussian 03, revision E.01; Gaussian, Inc.:
Wallingford, CT, 2005.
(39) Scott, A. P.; Radom, L. J. Phys. Chem. 1996, 100, 16502.
9
16030 J. AM. CHEM. SOC. VOL. 130, NO. 47, 2008