Rapid and selective spiro-cyclisations of O-centred radicals onto aromatic acceptors

Article abstract of DOI:10.1039/c3sc50500f

Substituted benzyloxycarbonyloxyl radicals were generated by sensitised photolyses of benzyl oxime carbonates. EPR spectroscopy showed they ring closed exclusively by spiro-cyclisation onto the ipso-C-atoms of the aromatic rings. β-Scission of the alkoxycarbonyloxyls to CO₂ and benzyloxyl radicals increasingly competed and became dominant above 270 K. The first rate parameters for spiro-cyclisations of O-centred radicals onto aromatics were obtained by the steady-state kinetic EPR method. Pentafluoro-substitution of the ring substantially reduced the spiro-cyclisation rate. Activation barriers of the spiro-cyclisations were computed by DFT to be about half those of the alternative ortho-cyclisations. Consideration of the TS structures suggested the preference for spiro- over ortho-cyclisation resulted from better overlap of the oxyl SOMO with the aromatic π-system during spiro closure.

Full text of DOI:10.1039/c3sc50500f

Chemical Science
View Article Online
View Journal | View Issue
EDGE ARTICLE
Rapid and selective spiro-cyclisations of O-centred
radicals onto aromatic acceptors†
Cite this: Chem. Sci., 2013, 4, 2028
Roy T. McBurney,‡ Annika Eisenschmidt,b Alexandra M. Z. Slawinb
a
*
b
*
and John C. Walton
Substituted benzyloxycarbonyloxyl radicals were generated by sensitised photolyses of benzyl oxime
carbonates. EPR spectroscopy showed they ring closed exclusively by spiro-cyclisation onto the ipso-C-
atoms of the aromatic rings. b-Scission of the alkoxycarbonyloxyls to CO2 and benzyloxyl radicals
increasingly competed and became dominant above 270 K. The first rate parameters for spiro-
cyclisations of O-centred radicals onto aromatics were obtained by the steady-state kinetic EPR method.
Pentafluoro-substitution of the ring substantially reduced the spiro-cyclisation rate. Activation barriers of
the spiro-cyclisations were computed by DFT to be about half those of the alternative ortho-cyclisations.
Consideration of the TS structures suggested the preference for spiro- over ortho-cyclisation resulted
from better overlap of the oxyl SOMO with the aromatic p-system during spiro closure.
Received 21st February 2013
Accepted 14th March 2013
DOI: 10.1039/c3sc50500f
www.rsc.org/chemicalscience
Not surprisingly therefore, most reported ring closures onto
aromatics are of the second ortho-type. In the great majority of
1 Introduction
Ring closures onto aromatic rings by free radicals necessarily
involve disruption of the 6p-electron system. For aryl species
with 4-member radical arms 1 the two main options are: (i) ipso-
attack (Ar1-5 closure) with formation of a spiro-cyclohexadienyl
intermediate 2, or (ii) ortho-attack (Ar1-6 closure) with formation
of a bicyclo-cyclohexadienyl intermediate 4 (Scheme 1). Spiro-
cyclisations involve formation of strained quaternary C-atoms
and have no straightforward reaction channel for return to
aromaticity. They can be reversible, depending on the archi-
tecture of the chain and the extent of strain in the spiro-radical.
Furthermore, if the chain is unsymmetrical as in 2, then ring-
opening may occur in the alternative mode to generate radical 3.
This amounts to a 1,4-aryl group migration. On the other hand
ortho-cyclisations can easily be followed by return to aromaticity
either by transfer out of the labile H-atom, or by transfer of an
electron to a suitable sink with generation of the corresponding
carbo-cation, followed by proton loss.
such processes the radical is C-centred; oen an aryl radical as
in the Pschorr and related reactions.1
Spiro-centres are important structural motifs in many
natural products and pharmaceuticals,2 so considerable effort
has gone into developing radical-based spiro-cyclisations.3,4
Aromatics5 and hetero-aromatics6 were found to be suitable
radical spiro-acceptors. Kinetic or other data to help predict the
efficiency of a novel aromatic ring closure, or to help decide
which mode would be favoured, is very sparse. Rate constants
were estimated for cyclisations of 4-phenylbutyl7 and 4-naph-
thylbutyl8 and the spiro-cyclisation of the latter was shown to be
reversible at 350 K by deuterium scrambling experiments.8,9
However, other evidence indicated these rate constants were
probably over-estimates.10 The paucity of information is even
more marked for spiro-cyclisations of O-centred radicals for
which spiro-intermediates have only been suggested in a few
instances with aroyloxyls.11
aUniversity of St. Andrews, EaStCHEM School of Chemistry, St. Andrews, Fife, KY16
9ST, UK
bUniversity of St. Andrews, EaStCHEM School of Chemistry, St. Andrews, Fife, KY16
9ST, UK. E-mail: jcw@st-and.ac.uk; Fax: +44 (0)1334 463808; Tel: +44 (0)1334
463864
† Electronic supplementary information (ESI) available: General procedures:
preparation and characterization data of oxime carbonates. Sample EPR spectra
and kinetic data. X-ray structural data for 6d, 6f, 6g and 7a. 1H and 13C NMR
spectra for novel compounds. DFT computed GS and TS structures. CCDC
923531–923534. For ESI and crystallographic data in CIF or other electronic
format, see DOI: 10.1039/c3sc50500f
Scheme 1 Ring closure options for radicals with 4-member arms onto aromatic
‡ Current address: School of Chemistry, University of Glasgow, Glasgow G12 8QQ,
UK. E-mail: roy.mcburney@glasgow.ac.uk.
rings.
2028 | Chem. Sci., 2013, 4, 2028–2035
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Edge Article
Chemical Science
We discovered recently that oxime carbonates ArC(R)]N–
OC(O)OR0 are clean and convenient precursors for iminyl and O-
centered radicals.12 Their weak N–O bonds selectively cleave on
UV photolysis, particularly when sensitized with 4-methox-
yacetophenone (MAP), thus facilitating investigations of the
behavior of both iminyl ArC(R)]N_ and alkoxycarbonyloxyl
radicals R0OC(O)O_. A distinct advantage of these precursors is
that they enable the iminyl radical intermediates to be directly
monitored by EPR spectroscopy. Although alkoxycarbonyloxyl
radicals are EPR silent,12,13 the radicals derived from them can
be characterized by EPR, thus providing mechanistic insight
into the evolution of their chemistry. We have now prepared a
set of model oxime carbonates with the aim of studying
competition between Ar1-5 and Ar1-6 ring closures of the
released radicals and of obtaining kinetic information. The test
collection was built around a functionalized benzyloxy core
linked to either a benzaldehyde or an acetophenone oxime
(Scheme 2). In 6a–d and 7a,b appropriate substituents modied
the aromatic character of the benzyl unit. In 6f–i potential
radical leaving groups were incorporated into the benzyl
4-substituents. The precursors with R1 ¼ H were particularly
valuable for EPR spectroscopy because the PhC(H)]N_ radicals
released on photolysis have very large (ꢀ80 G) hyperne split-
tings (hfs) from their iminyl H-atoms. Thus the signals from
this iminyl are situated in the wings of the EPR spectrum where
they do not obscure other radicals but provide a valuable
reference. This paper reports our study of the chemistry of
O-centred alkoxycarbonyloxyl radicals by means of solution EPR
spectroscopy, product analyses and DFT computations. This
enabled us to obtain the rst kinetic data for spiro-cyclisations
of O-centred radicals. The results allowed us to place spiro-cyc-
lisations in context with cyclisations in general and to assess the
viability of this reaction channel in relation to others.
2 Results and discussion
Preparation of precursor oxime carbonates
For the benzyloxy oxime carbonates 6 we employed our new
synthetic method, which avoids lab use of phosgene, and is
useful when the chloroformate of the alcohol is not commer-
cially available.12b The requisite alcohols were reacted with
carbonyldiimidazole to afford intermediates of type 5. Reaction
of these with an oxime and sodium hydride then afforded the
oxime carbonates in satisfactory yields (Scheme 2).
Oxime carbonates 6d,f,g and 7a were obtained in crystalline
form and the X-ray crystallographic structure of 6f is shown in
Fig. 1.14 The iminyl OC(O)O units adopted extended all trans
conformations with nearly planar ArC(R)N–OC(O)O sections.
These near planar structures will promote p–p stacking of the
oxime moieties with the MAP photosensitizer and hence facil-
itate energy transfer. The N–O bond lengths [1.441(3), 1.472(7),
˚
1.434(8) and 1.47(1) A in 6d, 6f, 6g and 7a respectively] were
comparable to that of the analogous bond of an oxime oxalate
15
16
˚
˚
amide (1.453 A), and of dioxime oxalates (1.442 A). Both
these classes of compound undergo N–O bond scission on UV
irradiation, so similar behavior for 6 and 7 was expected.
Ring closures of alkoxycarbonyloxyl radicals
The photolytic reactions of oxime carbonates 6a–d,f–i and 7a,b
were investigated in solution by 9 GHz EPR spectroscopy. Dea-
erated samples of each oxime carbonate, plus 1 equiv. of MAP,
in t-BuPh or cyclopropane solvent, were irradiated with unl-
tered UV light from a medium pressure Hg lamp directly in
the spectrometer resonant cavity. The spectrum obtained from
6c (Fig. 2a) showed an overlapping mixture of PhC(H)]N_
[g-factor ¼ 2.0030, a(1N) ¼ 9.8, a(1H) ¼ 80.1 G; similar to those
reported]17 and a second cyclohexadienyl type radical. Similar
spectra, consisting of an iminyl radical and a cyclohexadienyl
radical, were obtained from the oxime carbonates 6a–d,f–i in
the temperature range 210 to 290 K. EPR parameters are in
Table 1 and the spectrum from 6f is also shown in Fig. 2b.
Photo-dissociations of 6 were expected to lead to iminyl
radicals (Im) together with alkoxycarbonyloxyls
8 (see
Scheme 3). The EPR parameters (Table 1) demonstrated
unambiguously that the cyclohexadienyl radicals generated
from 6 were the corresponding 1,3-dioxaspiro[4,5]decadien-2-
onyls 9 and not the dihydro-4H-benzo[1,3]dioxin-2-onyls 11. The
DFT computed EPR hfs of the radicals (Table 1) provided sup-
porting evidence. It follows that type 8 phenalkoxycarbonyloxyl
Fig. 1 X-Ray crystal structure of benzaldehyde O-4-fluorobenzyloxycarbonyl
˚
oxime 6f. [Selected bond lengths and dihedral angles: O10–N11, 1.472(7) A,
Scheme 2 Benzyloxy oxime carbonates studied.
O10–N11–C12–C13 ꢁ178.63(46)ꢂ.]
This journal is ª The Royal Society of Chemistry 2013
Chem. Sci., 2013, 4, 2028–2035 | 2029

View Article Online
Chemical Science
Edge Article
Fig. 2 EPR spectra during photolyses of benzyloxy oxime carbonates. (a) 4-
Methyl precursor 6c at 215 K in t-BuPh showing Im and 9c. (b) 4-Fluoro-precursor
6f at 200 K in cyclopropane showing Im and 9f. Experimental spectra in black and
simulations in red.
radicals ring closed selectively in spiro-mode (5-exo) onto the
ipso C(1) atoms of their aromatic rings.
Scheme 3 Ring closures and dissociations of benzyloxycarbonyloxyl radicals.
The concentration ratios [9]/[Im] were determined from the
EPR spectra obtained from oxime carbonates 6c,d. The data was
rather scattered, because of the low S/N, but in each case [9]/[Im]
was ꢀ1.0 at T ¼ 220 K and dropped to near zero by 290 K. The
best quality data was obtained with the 4-t-Bu-precursor 6d and
is shown in Fig. 3. Photolyses of precursors 6 will yield equal
amounts of Im and 8 so the observation that [9]/[Im] ¼
1.0 indicated that spiro-cyclisations of 8c,d were very rapid at
T < 220 K.
The reduction in [9]/[Im] to higher temperatures can be
accounted for by increasing competition from dissociation of 8
to benzyloxyl radicals 10 and CO2. No new radicals were detec-
ted at higher temperatures in the spectra from 6c,d, as expected
with this decarboxylation process, because alkoxyls 10 are
known to be EPR “silent”. No radicals ArOC(O)OCH2_ were
detected from any of the precursors at any temperature. It
follows that rearrangements by 1,4-Ar migrations with produc-
tion of radicals of type 3 did not compete.
The data from precursor 6c was of lower quality but the drop
off in signal from spiro-radical 9c occurred in the same
temperature window. This is the same temperature region as
was previously observed for the unsubstituted radicals from
6a,b. We conclude that the rate parameters for CO2 loss from
ArCH2OC(O)O_ are about the same for H, Me and t-Bu substit-
uents in the 4-position. This is in accord with expectation
because 4-substituents are comparatively remote from the
breaking O–C(O) bond and separated from it by an sp3 C-atom.
The main products from UV photolyses of 6a–d,f with MAP at
1
rt were shown by H NMR and GC-MS analyses to be the cor-
responding benzyl alcohols R2C6H4CH2OH, dimer Im2, ImH
and the corresponding aldehyde or ketone from ImH hydro-
lysis. In the case of the benzaldehyde-derived oximes 6a and
6c,d,f, nitriles R2C6H4CN were also obtained via the electro-
cyclic CO2 loss process described previously.12 At rt the EPR data
showed loss of CO2 predominates and the observation of
Table 1 Experimental and computed EPR hfs for 1,3-dioxaspiro[4,5]-decadien-2-onyl radicals 9a
Precursor
Radical (R2)
Temp.b/K or DFT
g-Factor
a(H2,6
)
a(H3,5
)
a(R4)
a(Hg)
6a
6a
6c
6d
6e
6f
9a (H)
9a (H)
9c (Me)
9d (t-Bu)
9e (OMe)
9f (F)
230
DFT
205
215
DFT
230
DFT
230
DFT
235
DFT
2.0023
—
2.0020
2.0023
—
2.0025
—
2.0026
—
9.0
ꢁ9.4
8.7
2.7
3.4
2.3
2.5
3.1
2.6
3.2
2.0
2.4
2.1
2.4
13.0(1H)
ꢁ13.5(1H)
14.6(3H)
—
ꢁ0.5(3H)
32.8(1F)
24.1(1F)
2.0(2H)
2.9(2H)
1.3(1H)
1.2(1H)
0.6(2H)
ꢁ0.8(2H)
—
0.8(1H)
ꢁ0.8(2H)
—
8.6
ꢁ9.1
9.0
6f
9f (F)
ꢁ9.4
8.2
ꢁ0.8, ꢁ0.1
6h
6h
6i
9h (OCH2Ph)
9h (OCH2Ph)
9i (OCHPh2)
9i (OCHPh2)
—
ꢁ8.5
ꢁ0.8, ꢁ0.3
—
ꢁ0.6, ꢁ0.1
2.0026
—
8.2
6i
ꢁ8.5
a
In t-BuPh solution; hfs in Gauss. b DFT with the UB3LYP/6-311+G(2d,p) method; note that the signs of hfs cannot be obtained from isotropic EPR
spectra.
2030 | Chem. Sci., 2013, 4, 2028–2035
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Edge Article
Chemical Science
only the iminyl PhCH]N_ [see Fig. 4(c)]. It follows that spiro-
cyclisation of C6F5CH2OC(O)O_ was slower than termination
reactions in this temperature range. The EPR spectrum
obtained from irradiation of 7a in n-pentane at 290 K showed a
mixture of Im (53%), pent-2-yl (25%) and pent-3-yl (22%)
[Fig. 4(d) and (e)]. It follows that H-abstraction from the
n-pentane was more rapid than spiro-cyclisation.18 We infer that
the spiro-cyclisation of alkoxycarbonyloxyl radicals onto C6F5
was materially slower than onto phenyl itself or to phenyl with
electron-releasing substituents.
The slower ring closure rate of 8F5 suggested to us that the
cyclisations of the other alkoxycarbonyloxyls might be observ-
able by EPR spectroscopy at lower temperatures. Accordingly
photolyses of precursors 6a,d and f were examined in cyclo-
propane solvent and the [spiro]/[Im] ratios are plotted in
Fig. 3.19 We interpret the curved [spiro]/[Im] plots as follows. In
the temperature range 160–200 K, where [spiro]/[Im] increased,
spiro-cyclisation increased in importance converting more 8 to 9
until at about 200 K where [9] reached parity with [Im]. Above
this temperature CO2 loss from 8 supervened and hence the
[9]/[Im] ratio passed through a maximum and then declined.
The mechanism before the onset of decarboxylation (<200 K) will
be as shown in Scheme 3, with the addition of self-termination
and cross-termination processes for Im, 8 and 9 radicals. Making
the steady-state approximation it can easily be shown that:20
Fig. 3 Variation in relative yields [spiro]/[Im] with temperature. Black circles:
data for unsubstituted 6a in cyclopropane. Red circles: data for 4-t-Bu compound
6d (filled in cyclopropane, open in t-BuPh). Blue diamonds: data for 4-F
compound 6f.
R2C6H4CH2OH, R2C6H4CN, and the iminyl-derived products, is
in good accord with this.
The pentauoro-precursors 7a,b gave rise to signicantly
different behaviour. Sensitized photolyses of 7a (and 7b) in t-
BuPh yielded spectra showing the corresponding iminyl radical
and a second species with EPR parameters as follows: g ¼
2.0026, a(1H) ¼ 33.5, a(1H) ¼ 13.1, a(1H) ¼ 9.3, a(1H) ¼ 8.1,
a(1H) ¼ 2.7 G at 210 K (Fig. 4a and b). We attribute this spec-
trum to the meta-adduct radical from addition of the
C6F5CH2OC(O)O_ radical (8F5) to the t-BuPh solvent. The
[adduct]/[Im] ratio was found to be 0.9 at 225 K showing that
virtually all the C6F5CH2OC(O)O_ radicals added to solvent.
Hence spiro-cyclisation of 8F5 must have been signicantly
slower than this intermolecular addition and slower than for
radicals 8a–d,f–i.
ksc/2kt ¼ [9]/[8] ꢃ {kꢁsc/2kt + [8] + [9] + [Im]}
(1)
where all the terminations were assumed to be diffusion
controlled and have the same rate constant 2kt, and k is the
ꢁsc
rate constant for reverse ring-opening of the spiro-radicals 9.
Equi-molar quantities of Im and 8 will be formed in the initial
photochemical bond ssion and hence [Im] ¼ [8] + [9]. Thus eqn
(1) can be written as:
EPR spectra from MAP sensitized photolyses of 7a in cyclo-
propane solvent in the temperature range 150–235 K showed
ksc/2kt ¼ [9]/{[Im] ꢁ [9]} ꢃ {kꢁsc/2kt + 2[Im]}
(2)
In principle the concentration [Im] could be varied, for
example by altering the intensity of UV irradiation, and the
corresponding [9] determined by EPR. Then both ksc/2kt and
k
ꢁsc/2kt could be obtained by appropriate solution of eqn (2). In
practice the S/N in the EPR experiments was too small to allow
application of this method. If the reverse ring-opening of spiro-
radicals 9 is signicantly slower than cyclisation then: kꢁsc/2kt
ꢄ 2[Im] and eqn (1) simplies to:
ksc/2kt ¼ 2[9][Im]/{[Im] ꢁ [9]}
(3)
Rate constants ksc were derived for the unsubstituted, 4-t-Bu-
and 4-uoro-radicals 8a,c and f by use of eqn (3) with the radical
concentrations determined from the EPR measurements in the
T < 200 K region. Since all the radicals are small, terminations will
bediffusion controlled, so thewell-established 2kt rateparameters
for t-Bu_ radicals derived by Fischer and Schuh21 [log At
¼
Fig. 4 EPR spectra from UV photolyses of pentafluoro-precursor 7a in different
solvents. (a) UV photolysis of 7a in t-BuPh at 225 K showing Im and 8F5 radicals.
(b) Simulation of (a). (c) UV photolysis of 7a in cyclopropane at 155 K showing
only Im. (d) UV photolysis of 7a in n-pentane at 290 K. (e) Simulation of (d)
including Im, pent-2-yl and pent-3-yl radicals.
11.63 Mꢁ1 sꢁ1, Et ¼ 2.25 kcal molꢁ1], corrected for changes in
solvent viscosity as described previously,22 were used for 2kt
without introducing serious errors. Fig. 5 shows the resulting
Arrhenius plots.
This journal is ª The Royal Society of Chemistry 2013
Chem. Sci., 2013, 4, 2028–2035 | 2031

View Article Online
Chemical Science
Edge Article
Table 2 Rate data for 5-exo- and spiro-cyclisations of C- and O-centered radicals
a
log(Ac
/
Ec/
kc
Radical
Mode sꢁ1
)
kcal molꢁ1 (300 K)/sꢁ1 Ref.
5-Exo 10.4
6.85
2.3 ꢃ 105
ꢀ5 ꢃ 106
23
5-Exo [10.5]
ꢀ5.2
12b
<5 ꢃ 104
Spiro
7 and 10
8 and 10
This work
This work
This work
(323 K)
<104
(353 K)
Spiro
Spiro [10.5]
Spiro [10.5]
Spiro [10.5]
7.2 ꢅ 1.1
7.7 ꢅ 0.9
8.3 ꢅ 2.1
2 ꢃ 105
8 ꢃ 104
3 ꢃ 104
Fig. 5 Arrhenius plots of spiro-cyclisation rate data for benzyloxycarbonyloxyls.
Black circles for unsubstituted radical 8a; red circles for 4-t-Bu-radical 8d; red open
circles for 8d calculated with eqn (4) including reverse ring opening; blue dia-
monds for 4-F-radical 8f; green square: estimate for penta-F-radical 8F5 from
precursor 7a.
The logarithms of the pre-exponential factors derived from
linear regression of the Arrhenius plots of ksc for the three
radicals 8a, 8d and 8f were 10.3 ꢅ 1.3, 12.0 ꢅ 1.1 and 10.8 ꢅ 1.8.
These are in the expected range for radical cyclisations in
solution. However, the temperature range was too short and the
data too scattered for these to be reliable. Most radical cyclisa-
tions have log(Ac/sꢁ1) close to 10.5,23 so, for consistency in
comparing the data, this value was assumed for all the spiro-
cyclisation data in Table 2. From the absence of any detectable
spiro-radical in the spectrum from the pentauoro-precursor 7a
at 235 K we estimate that [spiro]/[Im] < 0.2 and hence, utilizing
the measured [Im] in the analogue of eqn (3), ksc for
C6F5CH2OC(O)O_(8F5) was estimated to be <0.8 sꢁ1 at 235 K. By
assuming log(Asc/sꢁ1) ¼ 10.5 for 8F5 the limiting rate data listed
in Table 2 was derived and is included in Fig. 5.
These are the rst measurements of rate data for spiro-cyc-
lisations onto aromatic rings by O-centred radicals. In fact,
published kinetic data for spiro-cyclisations (or ortho-cyclisa-
tions even of C-centred radicals onto aromatic rings) is
extremely sparse (Table 2). A note of caution needs to be
sounded in that the spiro-cyclisation of 4-naphthylbutyl was
shown to be reversible at about 350 K by deuterium scrambling
experiments.7,8 If the spiro-cyclisations of benzylox-
ycarbonyloxyls 8 were also appreciably reversible the measured
ksc values would be subject to error.
Spiro [10.5]
>11.3
<0.2 ꢃ 103 This work
a
Values in parenthesis assumed.
experiments were carried out at about 100 K lower than the
deuterium scrambling study with 4-naphthylbutyl. This also
militates against participation by reverse spiro-ring opening in
our system. We conclude that k
than ksc for all 8 and that the rate data in Table 2 is reasonably
must be signicantly less
ꢁsc
reliable.
The data in Table 2 demonstrate that the rate constant for
the archetype 8a is nearly an order of magnitude greater than
the estimated rate constants for spiro-cyclisations of the C-
centred analogues 4-phenylbutyl7 and 4-naphthylbutyl.8 The
faster rate of spiro-cyclisation of benzyloxycarbonyloxyl 8a is as
expected by analogy with the larger rate constant for 5-exo-cyc-
lisation of allyloxycarbonyloxyl in comparison with hex-5-enyl9b
(see Table 2). Similarly, alkoxycarbonyloxyl radicals add to
alkenes more rapidly than do C-centred radicals.13,24 The data
showed that the Esc and ksc values for spiro-cyclisation of 8d with
the 4-t-Bu substituent differed little from those of unsubstituted
8a. The EPR observations with precursor 6c having a 4-Me
substituent suggested radical 8c underwent spiro-cyclisation at
least as fast as 8a. The 4-uoro substituent had a small inhibi-
tory effect on spiro-cyclisation but ve uorine substituents
reduced the rate and increased the activation energy
signicantly.
In the special case of the rate constants for spiro-cyclisation
and reverse ring opening being equal then eqn (2) simplies
to (4):
ksc/2kt ¼ 2[9][Im]/{[Im] ꢁ 2[9]}
(4)
In fact for all the data, except the two lowest temperature
Compounds 6g–j were prepared to test the idea that the
points for 8d, we found that 2[9] > [Im] so that eqn (4) leads to spiro-radicals 8g–j might dissociate to release a new radical X_
impossible ksc rate constants. This indicates that k is prob- with formation of 1,3-dioxaspiro[4,5]deca-6,9-diene-2,8-dione
ꢁsc
ably smaller than ksc in the T range of the EPR experiments. The 12 (Scheme 3). This had potential to be a preparatively useful
two low temperature points for 8d, calculated from (4), are process. EPR spectra from the sulfonate ester 6g showed only
shown in Fig. 5 and do not deviate greatly from the line the corresponding iminyl (Im) and iminoxyl PhC(Me)NO_in the
obtained by neglecting reverse ring opening. The EPR range 225 to 285 K. No MeSO2_ radicals were detected.25
2032 | Chem. Sci., 2013, 4, 2028–2035
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Edge Article
Chemical Science
Table 3 DFT computed activation energies (DE‡298) and reaction enthalpies
The benzyloxy-(6h) and diphenylmethyloxy-precursors (6i)
gave good spectra of the corresponding spiro-radicals 9h and 9i at
220 K (EPR parameters in Table 1) together with Im. The spiro-
radical spectra of 9i and 9j weakened to ꢀ280 K and above this
only Im was detectable; neither the benzyl nor the diphe-
nylmethyl radical appeared. The EPR data suggested that the
alkoxycarbonyloxyls 8h,i dissociated to 10h,i and CO2 at these
higher temperatures thus precluding spiro-cyclisation followed
by dissociation to 12 and X_. GC-MS analysis of the products from
photolysis of 6i supported this conclusion. The major products
were found to be 4-BnOC6H4CHO, 4-BnOC6H4CH2OH and Im2
from CO2 loss and termination reactions of 10i and Im. Precursor
6j, which might have released the much more persistent Ph3C_
radical, was unstable and could not be obtained pure.
(DH298)
in kcal molꢁ1 for benzyloxycarbonyloxyl radicals with the UB3LYP
functional
Subs.
Spiro
Spiro
Ortho
Ortho
Spiro
Struct.
Methoda
DE‡298
DH298
DE‡298
DH298
E‡sc expt.
F5 8F5
F5 8F5
4-F 8f
4-F 8f
8a
A
B
A
B
A
B
A
A
A
7.70
7.74
5.38
5.83
5.89
6.35
4.96
4.49
3.07
0.59
0.72
0.38
13.72
13.58
12.65
13.02
12.01
12.35
12.0
ꢁ2.13
ꢁ2.25
6.35
6.45
5.74
5.74
ꢁ1.0
5.3
>11.2
>11.2
8.3
8.3
7.2
7.2
—
7.7
—
0.16
ꢁ0.45
ꢁ0.03
ꢁ0.6
ꢁ1.0
ꢁ2.5
8a
4-Me 8c
4-t-Bu 8d
4-MeO 8e
11.2
10.0
5.1
a
A ¼ UB3LYP/6-311+G(2d,p)//UB3LYP/6-311+G(2d,p); B ¼ UB3LYP/cc-
pVTZ//UB3LYP/cc-pVTZ.
QM computations
To shed further light on the competition between spiro- and
ortho-cyclisation for the benzyloxycarbonyloxyl radicals 8, QM
computations were carried out with several methods.26 Our rst
task was to test some levels of theory using the experimental
data for cyclisation of 8a as the benchmark. As mentioned
before (Table 2), the spiro-cyclisation of 8a exhibited a value for
Esc of 7.2 ꢅ 1.1 kcal molꢁ1, assuming log(Asc/sꢁ1) ¼ 10.5. We
therefore took this value as a good approximation of the real
value and used it to validate our theoretical approach. Full
details are given in the ESI.† Becke's 3-parameter hybrid
exchange potential (B3) with the Perdew–Wang (PW91)27
gradient-corrected correlation functional, UB3PW91, and the
6-311+G(2d,p) basis set was deemed to be unsatisfactory
because it gave DE‡298 values for spiro-cyclisations decreasing as
one and then ve F-atoms were introduced to the ring; contrary
to experiment. The Møller–Plesset perturbation theory MP2
method, with a 6-311+G(2d,p) basis set and geometries derived
from UB3LYP/6-311+G(2d,p) computations, was also dis-
counted because the DE‡298 values for spiro-cyclisations greatly
exceeded experiment (18.2 kcal molꢁ1 for 8a and 17.5 kcal
molꢁ1 for 8f). M06 hybrid meta functionals were reported to
have broad applicability and M06-2X was found to be good for
main group elements.28 However, DE2‡98 values computed for
spiro-cyclisations with M06-2X (also with UB3LYP/6-311+G(2d,p)
geometries) were much too small (2.9 for 8a and 2.2 for 8f). Our
best results were obtained with the standard B3LYP functional
and the 6-311+(2d,p) or the correlation consistent polarized
triple zeta cc-pvtz basis sets (Table 3). For the series of radicals
8a,c–f and 8F5 (the pentauoro-analogue) both the spiro- and
the endo-ring closure processes were explored and the transition
structures leading to the ve- and six-membered species were
computed. The derived activation parameters [DE‡298] and
reaction enthalpies [DH298], corrected to 298 K for thermal
effects are listed in Table 3.
indicate small exothermicities. Slight endothermicities were
found for the F-substituted radicals 8f and 8F5. This implies
that reverse spiro-ring opening reactions might be important
and is somewhat at odds with experiment (see above). The
computed DE‡298 values for spiro-cyclisations show a decreasing
trend from the electron-withdrawing to the electron-releasing
substituents. The absolute magnitudes of the computed DE2‡98
values were all smaller than found by experiment (Table 3). The
B3LYP functional is known to underestimate barrier heights29
(on average by about 4 kcal molꢁ1) so our results conform with
this same tendency.
The UB3LYP/cc-pvtz computed energetics of the four
competing reaction channels for the archetype radical 8a are
illustrated in Fig. 6. The ortho-cyclisation step to 11 is endo-
thermic and has a comparatively high activation energy so does
not compete in the accessible temperature range. A 1,4-phenyl
shi would produce primary radical 3a (analogous to 3 in
Scheme 1). Although rearrangement to 3a is only slightly
endothermic, the required C–C bond ssion in the spiro-inter-
mediate 9a has a high activation energy and so 1,4-phenyl shi
For all six benzyloxycarbonyloxyls, 8a,c–f and 8F5, the
computed DE‡298 values were smaller (by nearly a factor of two)
for spiro- than for ortho-cyclisation. Thus, irrespective of
substituents, spiro-cyclisation was predicted to be favoured in
agreement with experiment. The spiro-cyclisation reaction
Fig.
6
UB3LYP/cc-pvtz computed reaction pathways for the benzylox-
enthalpies (DH298) for the archetype 8a and 8c–e radicals ycarbonyloxyl radical 8a.
This journal is ª The Royal Society of Chemistry 2013
Chem. Sci., 2013, 4, 2028–2035 | 2033

View Article Online
Chemical Science
Edge Article
were not very sensitive to substituents in the 4-positions of their
aryl rings, although pentauoro-substitution did reduce the
rate. The spiro-rate constant for the archetype unsubstituted
alkoxycarbonyloxyl 8a was greater than that of the analogous
C-centred radical, 4-phenylbutyl. This is consistent with the
known high rates of alkoxycarbonyloxyl addition and abstrac-
tion reactions. As expected, however, the rate constant for spiro-
cyclisation of 8a was less than the rate constant for 5-exo-cycli-
sation of the archetype hex-5-enyl radical. Activation barriers of
the spiro-cyclisations were computed by DFT to be about half
those of the alternative ortho-cyclisations. The DFT computa-
tions also predicted lower barriers and faster rates for alkox-
ycarbonyloxyls with electron-releasing 4-substituents. However,
in agreement with previous work, the B3LYP functional
underestimated the reaction barriers. DFT found the spiro-cyc-
lisations to be close to thermoneutral. Consideration of the TS
structures suggested the preference for spiro- over ortho-cycli-
sation resulted from better overlap of the oxyl SOMO with the
aromatic p-system during spiro closure.
Fig. 7 DFT computed TS structures and SOMOs for spiro-(left, TSsp) and ortho-
(right, TSor) cyclisations of radical 8a.
cannot compete with C–O bond ssion. b-Scission with loss of
CO2 is the most exothermic process but the activation energy of
this step is comparatively high. This can partly be accounted for
by the fact that the RO–C(O)O bond is strengthened by delo-
calization of the unpaired electron onto the RO moiety. The end
result is that the dominant process at low temperatures is spiro-
cyclisation, which is practically thermoneutral and has a lower
activation energy than dissociation to CO2. The latter process
can only dominate at higher temperatures.
Why is spiro-cyclisation favoured over ortho-cyclisation for
this range of benzyloxycarbonyloxyl radicals? The UB3LYP/6-
311+D(2d,p) computed transition state structures, with their
frontier SOMOs, are displayed in Fig. 7 for the two cyclisation
modes and help to explain this.
The orbital containing the unpaired electron is a p-orbital
in the plane of the_OC]O system (see TSsp). The geometry of the
spiro approach to the ring is almost directly from above
enabling this p-orbital to overlap very efficiently with the ring
Acknowledgements
We thank the EPSRC (grant EP/I003479/1) & EaStChem for
funding, the EPSRC UK National Electron Paramagnetic Reso-
nance Service at the University of Manchester and the EPSRC
National Mass Spectrometry Service, Swansea. We are grateful
to Dr Herbert Fruchtl for help with the computing.
Notes and references
˚
p-system. Thus the TS is early [r(O/C) ¼ 1.929 A, :C/O–C ¼
103.5ꢂ], the ring maintains much of its aromaticity, keeping the
activation energy low. The geometry of ortho approach to the
ring tilts the _OC]O p-orbital on its side such that overlap
with the ring p-system is poorer (see ꢂTSor). The TS occurs later
1 (a) G. Pratsch and M. R. Heinrich, Radicals in Synthesis III,
Top. Curr. Chem., 2012, 320, 33; (b) K. K. Laali and
M. Shokouhimehr, Curr. Org. Synth., 2009, 6, 193.
2 See for example: (a) T. Cottin, C. Pyrkotis, C. I. Stathakis,
I. Mavridis, I. A. Katsoulis, P. Anastasopoulou, G. Kythreoti,
A. L. Zografos, V. R. Nahmias, A. Papakyriakou and
D. Vourloumis, ChemBioChem, 2011, 12, 71; (b)
S. Beaumont, E. A. Ilardi, N. D. C. Tappin and A. Zakarian,
Eur. J. Org. Chem., 2010, 5743; (c) B. M. Trost and
M. K. Brennan, Synthesis, 2009, 3003; (d) S. A. A. El Bialy,
H. Braun and L. F. Tietze, Synthesis, 2004, 2249.
3 (a) D. H. Hey and A. R. Todd, J. Chem. Soc. C, 1967, 1518; (b)
D. H. Hey, G. H. Jones and M. J. Perkins, J. Chem. Soc. C,
1971, 116; (c) D. H. Hey, G. H. Jones and M. J. Perkins,
J. Chem. Soc., Perkin Trans. 1, 1972, 113.
4 (a) S. Guidott, R. Leardini, D. Nanni, P. Pareschi and
G. Zanardi, Tetrahedron Lett., 1995, 36, 451; (b) D. Crich
and J.-T. Hwang, J. Org. Chem., 1998, 63, 2765; (c)
M. Koreeda, Y. Wang and L. Zhang, Org. Lett., 2002, 4,
3329; (d) A. K. Ganguly, C. H. Wang, T. M. Chan, Y. H. Ing
and A. V. Buevich, Tetrahedron Lett., 2004, 45, 883;
(e) F. G.-L. Turiso and D. P. Curran, Org. Lett., 2005, 7, 151;
(f) K. C. Majumdar and S. Alam, Org. Lett., 2006, 8, 4059;
(g) K. C. Majumdar, P. Debnath, A. K. Pal,
S. K. Chattopadhyay and A. Biswas, Can. J. Chem., 2007, 85,
˚
[r(O/C) ¼ 1.885 A, :C/O–C ¼ 111.8 ] with more disruption of
the 6p ring system and hence a higher activation energy.
3 Conclusions
Benzyl oxime carbonates are easily made and handled and have
long shelf lives. We have found them to be useful precursors for
the photochemical generation of iminyl and alkoxycarbonyloxyl
radicals. The latter radicals ring close exclusively by spiro-cyc-
lisation onto the ipso-C-atom of the aromatic rings. The spiro-
cyclohexadienyl radicals formed in this way were observable by
EPR spectroscopy at temperatures below 270 K. In this way it
was shown that b-scission of the alkoxycarbonyloxyls to CO2 and
BnO_radicals increasingly competes with spiro-cyclisation above
about 210 K and becomes dominant above 270 K. Because of
this the main products obtained from rt photolyses were
derived from the BnO_ and iminyl radicals.
A steady state kinetic equation was derived from the mech-
anism and enabled rate and Arrhenius parameters for the spiro-
cyclisations to be estimated by the steady-state kinetic EPR
method. Evidence from concentration measurements showed
that the reverse spiro-ring opening was not important at the low
temperatures of the EPR experiments. The spiro-rate constants
445;
(h)
C.
V.
Stevens,
E.
Van
Meenen,
K. G. R. Masschelein, Y. Eeckhout, W. Hooghe, B. D'Hondt,
2034 | Chem. Sci., 2013, 4, 2028–2035
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Edge Article
Chemical Science
ꢂ
ꢂ
˚
˚
V. N. Nemykin and V. V. Zhdankin, Tetrahedron Lett., 2007,
48, 7108; (i) T. Lanza, R. Leardini, M. Minozzi, D. Nanni,
P. Spagnolo and G. Zanardi, Angew. Chem., Int. Ed., 2008,
47, 9439; (j) G. Rousseau, F. Robert and Y. Landais, Chem.–
Eur. J., 2009, 15, 11160; (k) S. Van der Jeught, N. De Vos,
K. Masschelein, I. Ghiviriga and C. V. Stevens, Eur. J. Org.
Chem., 2010, 5444; (l) Z. Tang, J. Mayrargue and M. Alami,
J. Heterocycl. Chem., 2011, 48, 1238.
A, c ¼ 18.99(3) A, a ¼ 98.61(6) , b ¼ 91.40(8) , g ¼
3
ꢂ
˚
91.03(3) , U ¼ 843(2) A , T ¼ 173 K, space group P-1 (#2),
Z ¼ 2, 11643 reections measured, 2905 unique, (Rint
¼
0.1733) which were used in all calculations. The nal
wR(F2) was 0.4028 (all data). Crystal data for 7a: CCDC
923534, C15H8F5NO3, M ¼ 345.23, colourless platelet,
˚
0.120 ꢃ 0.030 ꢃ 0.010 mm, monoclinic, a ¼ 7.89(3) A, b ¼
3
ꢂ
˚
˚
˚
5.60(2) A, c ¼ 16.51(6) A, b ¼ 102.65(7) , U ¼ 712(5) A ,
T ¼ 173 K, space group P21 (#4), Z ¼ 2, 7300 reections
measured, 2479 unique, (Rint ¼ 0.1756) which were used in
all calculations. The nal wR(F2) was 0.4300 (all data).
5 (a) W. R. Bowman, C. F. Bridge and P. Brookes, J. Chem. Soc.,
Perkin Trans. 1, 2000, 1; (b) H. Ohno, H. Iwasaki, T. Eguchi
and T. Tanaka, Chem. Commun., 2004, 2228.
6 (a) P. J. Parsons, M. Penverne and I. L. Pinto, Synlett, 1994, 15 E. M. Scanlan, A. M. Z. Slawin and J. C. Walton, Org. Biomol.
721; (b) D. C. Harrowven, B. J. Sutton and S. Coulton, Chem., 2004, 2, 716.
Tetrahedron Lett., 2001, 42, 9061; (c) C. Escolano and 16 F. Portela-Cubillo, J. Lymer, E. M. Scanlan, J. S. Scott and
K. Jones, Tetrahedron Lett., 2000, 41, 8951. J. C. Walton, Tetrahedron, 2008, 64, 11908.
7 J. K. Kochi and R. D. Gilliom, J. Am. Chem. Soc., 1964, 86, 17 A. R. Forrester and F. A. Neugebauer in Landolt–Bornstein,
¨
5251.
Magnetic Properties of Free Radicals, ed. H. Fischer and
F. A. Neugebauer, Springer-Verlag, Berlin, 1979, vol. 9c1,
pp. 115–121.
8 (a) M. Julia, Pure Appl. Chem., 1974, 40, 553; (b) J. C. Chottard
and M. Julia, Tetrahedron, 1972, 28, 5615.
9 (a) J. A. Howard and K. U. Ingold, Can. J. Chem., 1969, 47, 18 Note, however, that at 290
K
decarboxylation of
3797; (b) C. Chatgilialoglu, K. U. Ingold and J. C. Scaiano,
J. Am. Chem. Soc., 1981, 103, 7739.
C6F5CH2OC(O)O_ would be signicant and therefore some
of the H-abstraction may be due to C6F5CH2O_ radicals.
10 A. L. J. Beckwith and K. U. Ingold, in Rearrangements in 19 Experiments could only be conducted down to 180 K with
Ground and Excited States, ed. P. de Mayo, Academic Press,
the 4-F-precursor 6f because it was poorly soluble and
1980, ch. 4, p. 161.
crystallised out of the cyclopropane solvent.
11 P. M. Brown, P. S. Dewar, A. R. Forrester and R. H. Thomson, 20 See ESI for details of the derivation of eqn (1).†
J. Chem. Soc., Perkin Trans. 1, 1972, 2842.
12 (a) R. T. McBurney, A. M. Z. Slawin, L. A. Smart, Y. Yu and
J. C. Walton, Chem. Commun., 2011, 47, 7974; (b)
21 (a) H.-H. Schuh and H. Fischer, Helv. Chim. Acta, 1978, 61,
2130; (b) H.-H. Schuh and H. Fischer, Int. J. Chem. Kinet.,
1976, 8, 341.
R. T. McBurney, A. D. Harper, A. M. Z. Slawin and 22 (a) B. Maillard and J. C. Walton, J. Chem. Soc., Perkin Trans. 2,
J. C. Walton, Chem. Sci., 2012, 3, 3436.
1985, 443; (b) A. F. Bella, L. V. Jackson and J. C. Walton,
13 D. J. Edge and J. K. Kochi, J. Am. Chem. Soc., 1973, 95, 2635.
J. Chem. Soc., Perkin Trans. 2, 2002, 1839.
¨
14 Crystal data for 6d: CCDC 923531, C19H21NO3, M ¼ 311.38, 23 See: (a) A. L. J. Beckwith and S. Brumby, in Landolt–Bornstein,
colourless prism, 0.120 ꢃ 0.030 ꢃ 0.030 mm, triclinic,
Radical Reaction Rates in Liquids, ed. H. Fischer, Springer
Verlag, Berlin, 1994, vol. II18a, p. 171; (b) A. L. J. Beckwith,
˚
˚
˚
a ¼ 6.229(3) A, b ¼ 11.758(7) A, c ¼ 12.009(7) A, c ¼
ꢂ
ꢂ
ꢂ
˚
¨
12.009(7) A, a ¼ 88.61(4) , b ¼ 77.64(6) , g ¼ 83.80(4) ,
in Landolt–Bornstein, Radical Reaction Rates in Liquids, ed.
3
˚
U ¼ 854.1(8) A , T ¼ 173 K, space group P-1 (#2), Z ¼ 2,
H. Fischer, Springer Verlag, Berlin, 1984, vol. II13a, p. 252.
8939 reections measured, 3049 unique, (Rint ¼ 0.1235) 24 J. Chateauneuf, J. Lusztyk, B. Maillard and K. U. Ingold, J.
which were used in all calculations. The nal wR(F2) was
Am. Chem. Soc., 1988, 110, 6727 and references cited therein.
0.0964 (all data). Crystal data for 6f: CCDC 923532, 25 The iminoxyl was almost certainly generated from trace
C
15H12FNO3,
M
¼
273.26, colourless prism, 0.240
ꢃ
impurities of the oxime in the starting material and was
detected because of its persistence.
26 M. J. Frisch, et al., Gaussian 09, Revision A.02, Gaussian,
Inc., Wallingford CT, 2009, see ESI† for full citation.
˚
0.100 ꢃ 0.020 mm, monoclinic, a ¼ 14.698(2) A, b ¼
˚
˚
4.5311(5) A,
c
¼
19.448(3) A,
b
¼
97.309(4)ꢂ,
U
¼
3
˚
1284.7(3) A , T ¼ 173 K, space group P21/n (#14), Z ¼ 4,
15766 reections measured, 2261 unique, (Rint ¼ 0.0465) 27 J. P. Perdew, K. Burke and Y. Wang, Phys. Rev. B: Condens.
which were used in all calculations. The nal wR(F2) was Matter, 1996, 54, 16533.
0.3223 (all data). Crystal data for 6g: CCDC 923533, 28 Y. Zhao and D. G. Truhlar, Acc. Chem. Res., 2008, 41, 157.
´
´
C
17H17NO6S, M ¼ 363.38, colourless platelet, 0.100 ꢃ 29 Y. Zhao, N. Gonzalez-Garcıa and D. G. Truhlar, J. Phys. Chem.
˚
0.100 ꢃ 0.010 mm, triclinic, a ¼ 5.707(7) A, b ¼ 7.877(11)
A, 2005, 109, 2012.
This journal is ª The Royal Society of Chemistry 2013
Chem. Sci., 2013, 4, 2028–2035 | 2035