Unexpected radical mechanism in a [4+1] cycloaddition reaction

Article abstract of DOI:10.1039/d1nj00660f

9,10-Phenanthrenequinone monoimines (2,7-R-PQI, R = tBu, H, Br, NO2, 1a-d) undergo a [4+1] cycloaddition reaction with triphenylphosphine to give 2,3-dihydro-2,2,2-triphenylphenanthro[9,10-d]-1,3,2λ5-oxazaphospholes (3a-d). During the reaction, highly colored radicals are formed as intermediates, which were characterized by EPR and UV-vis spectroscopy. The formation rate and the rate of decay of these radicals were determined kinetically. These radicals exhibit high persistency and under inert conditions can be handled conveniently. Based on detailed kinetic and spectroscopic studies and DFT calculations, a plausible mechanism has been proposed. This journal is

Full text of DOI:10.1039/d1nj00660f

NJC
View Article Online
View Journal
COMMUNICATION
Unexpected radical mechanism in a [4+1]
cycloaddition reaction†
Cite this: DOI: 10.1039/d1nj00660f
b
d
Istvan Bors,a Mihaly Purgel,
Peter Pal Feher,c Tamas Varga,
Gabor Speier,a
´
´
´
´
´
´
´
Laszlo Korecze and Jozsef Kaizer
*
a
Received 8th February 2021,
Accepted 15th April 2021
´
´
´
DOI: 10.1039/d1nj00660f
rsc.li/njc
9,10-Phenanthrenequinone monoimines (2,7-R-PQI, R = tBu, H, Br,
During the synthesis of 3, we have observed that the color of
NO2, 1a–d) undergo a [4+1] cycloaddition reaction with triphenyl- the solution mixtures turns to deep red, which fades at the end
phosphine to give 2,3-dihydro-2,2,2-triphenylphenanthro[9,10-d]- of the reaction to give pale yellow end products (Fig. 1).
1,3,2k5-oxazaphospholes (3a–d). During the reaction, highly colored The presence of radical species was suspected and a more
radicals are formed as intermediates, which were characterized by detailed investigation of the reaction was undertaken.
EPR and UV-vis spectroscopy. The formation rate and the rate of
decay of these radicals were determined kinetically. These radicals
exhibit high persistency and under inert conditions can be handled
conveniently. Based on detailed kinetic and spectroscopic studies
and DFT calculations, a plausible mechanism has been proposed.
Although electrocyclic reactions, in general, attract intense
interest, [4+1] cycloaddition (or cheletropic) reactions are some-
what outside of this scope.1 Trivalent phosphorus compounds2
readily undergo additions with 1,4-dioxa-1,3-dienes (o-quinones),
and in these cases a concerted reaction mechanism has been
assumed.3 It was observed that o-quinone monoimines (1a–d)
readily react with triphenylphosphine in a cheletropic reaction to
2,3-dihydro-2,2,2-triphenylphenanthro
[9,10-d]-1,3,2-l5-oxaza-
phospholes (3a–d) (eqn (1)),4 which serve as good bioinspired
organocatalysts for the oxidation of thiophenol, cysteine, and
glutathione to their disulfides,5 triphenylphosphine to triphenyl-
phosphine oxide,6 3,5-di-tert-butylcatechol to the corresponding
o-quinone and 2-aminophenol to 2-aminophenoxazine-3-one by
molecular oxygen.7
a Research Group of Bioorganic and Biocoordination Chemistry, University of
´
Pannonia, Veszprem H-8200, Hungary. E-mail: kaizer@almos.uni-pannon.hu
b Department of Physical Chemistry, University of Debrecen, Debrecen H-4032,
Hungary
c Institute of Organic Chemistry, Research Centre for Natural Sciences,
Budapest H-1117, Hungary
d
´
Department of Process Engineering, University of Pannonia, Veszprem H-8200,
Hungary
e Institute of Materials and Environmental Chemistry, Research Centre for Natural
Fig. 1 Color (A) and visible spectral change during the reaction of 9,10-
phenanthrenequinone monoimine 1b (0.5 mM) and triphenylphosphine
(0.5 mM) (B). Inset: Time course of formation and decay of the radical
intermediate from 1b and triphenylphosphine in 10 mL acetonitrile at 50 1C
under Ar (C).
Sciences, Budapest H-1117, Hungary
† Electronic supplementary information (ESI) available: Additional spectroscopic
and kinetic data, synthetic procedures and characterisation. See DOI: 10.1039/
d1nj00660f
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021
New J. Chem.

View Article Online
Communication
NJC
Mixing 0.5 mM solutions of 9,10-phenanthrenequinone using EPR software to identify the radicals.9 The EPR spectra can
monoimine 1b and triphenylphosphine (0.5 mM) in acetonitrile be simulated by assuming one phosphorus and one nitrogen
at 50 1C under argon gives in ca. 0.5–1 h intense red solutions atom as well as eight protons for the 1b and six protons for the
(Fig. 1A).
disubstituted 1a,c,d derived radicals. In addition, two different
The formation and decay of the red compound were followed 13C couplings were taken into account in the simulations, but
by UV-vis spectroscopy. In addition to the absorption of 1b unfortunately the determination of the exact number of atoms
[lmax (e) 216 (22 101), 263 (29 335), 313 (3706), 395 belonging to each coupling was not allowed by the relatively
(1778 molÀ1 L cmÀ1) nm], new absorption bands at 524 and broad linewidth. The comparison of hyperfine coupling helps to
486 nm appeared with time (Fig. 1B). These peaks reached assign the couplings in positions 2 and 7. We assume that
maxima after 5 h and almost disappeared after 24 h (Fig. 1C). radical 2bꢀ or 3bꢀ (or both) is the source of the obtained EPR
A similar spectrum was observed from the reaction of 3b and signal. Different sets of hyperfine coupling constants can provide
TEMPO resulting in the oxazaphospholyl radicals (2bꢀ or 3bꢀ), simulations of almost the same quality due to the relatively low
which can be used for the calculation of the radical concen- resolution. Nitrogen and two almost equivalent protons have a
tration using a calibration curve (Fig. S1, ESI† [lmax (e) 486 very similar effect on the spectra so their hyperfine coupling
(3154 molÀ1 L cmÀ1) nm]). After two days the solution was pale constants are interchangeable if the resolution is not good
yellow. That means that the formation of the radical species is enough. The agreement between the DFT calculations and the
faster than its reaction to the end product and that is responsible EPR parameters determined from the spectra is adequate and
for the enrichment of the proposed oxazaphospholyl radical with only the nitrogen coupling is overestimated by the DFT
1b based on the stoichiometry of the reaction above (Fig. 1C). calculation. DFT geometry optimization for free radicals, both
DFT mechanistic study and TDDFT calculations predicted stable closed and open ring, always resulted in the open ring geometry
radical pairs (RP1 and RP2) consisting of 2bꢀ and PQIHꢀ, whose contrary to closed-shell systems. Hence, we identify the radical
spectrum showed good agreement with the experimental one; as open ring 2bꢀ. The unique situation that only radical 2bꢀ
see below and the ESI.† However, the EPR spectra of the radicals could be seen in the EPR spectra suggests that the radical species
formed in the case of compounds 1a–d are almost perfectly (PQIHꢀ) formed during the reaction are planar and probably
symmetrical (Fig. 2 and Fig. S2–S4, ESI†) suggesting that the stacked in solution and so undergo intermolecular interactions to
observed spectra cannot be signals of the RP. In the case of weak render them EPR-silent.10 Isoelectronic 9,10-phenanthrenequinone
coupling or no coupling, the spectrum would be a superimposed semiquinone radical anions tend to interact in this manner gen-
spectra of the two different radicals with different g values, erally and the separation of nearly parallel aligned semiquinone
which cannot give a symmetric spectrum. The case of strong ligands in [Cu(9,10-PSQ)(PPh3)]2 of 2.8 Å suggests p–p interactions
coupling is more complicated but, based on similar symmetry between the highly delocalized p-system rings.10 The cyclic
considerations, the case of the radical pair is almost ruled out. It voltammograms (CVs) display reversible oxidation waves in the
is worth noting that almost identical kinetic runs were observed case of 1,3,2-oxazaphospholes (3a–d) and irreversible waves (1a–d)
following the reaction by parallel UV-vis and EPR measurements in the case of imines.
(Fig. S5, ESI†). Density functional theory (DFT) computations
The substituted 9,10-phenanthrenequinone monoimines
were performed for radicals originating from compounds 1a–d (2,7-dinitro (1d), 2,7-dibromo (1c), and 2,7-di-tert-butyl (1a))
using the ORCA package.8 Using the results of DFT calculations show a decreasing reaction rate in the listed order, while the
as a starting point, simulations of the spectra were carried out by CVs of the quinone monoimines 1a–d exhibited an increasing
reduction potential E0 in CH3CN, 1a = À0.955, 1b = À0.885,
pc
1c = À0.698 and 1d = À0.558 V (50 mV sÀ1 scan rate, Fc/Fc+
internal standard, 0.1 M) (Fig. S6, ESI†). This is in agreement
with the assumption that with a higher oxidation potential of
the quinone imines (1) a faster reaction rate is observed
(Table 1).
Table
1
The redox potentials of the imines (1a–d) and 1,3,2-
oxazaphospholes (3a–d)a
E0 (1a–d) (V) E01/2(3a–d) (V) 104 Vin (MsÀ1
)
1/2
b
Entry
R
1
2
3
4
1a 2a tBu À0.955
0.032
0.083
0.213
0.351
1.28
1.45
—
1b 2b
H
À0.885
À0.698
1c 2c Br
1d 2d NO2 À0.558
18
a
In MeCN, tert-butylammonium perchlorate as a supporting electro-
lyte, scan speed 50 mV sÀ1, referenced to Ag/AgCl, glassy carbon and
platinum electrodes. [2,7-R-PQI] = 0.25 mM, [PPh3] = 10 mM in 4 mL
Fig. 2 The X band EPR spectrum from 1b and triphenylphosphine (mea-
sured – red and simulated – blue). Parameters: g = 2.0032, aP = 6.90, aN
1.10, aH1–7 = 2.35, 2.26, 2.11, 0.85, 0.79, 0.73, 0.67, a13C 1.5336 and 2.987G.
b
=
CH3CN at 25 1C under Ar.
New J. Chem.
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021

View Article Online
NJC
Communication
is B12 at R.T. ([1b] = 1 mM; [PPh3] = 10 mM). A comparison of the
reaction rates measured with H-PQI and D-PQI shows a noticeable
H/D kinetic isotope effect (KIE = 1.8), demonstrating that proton
transfer and/or disruption of hydrogen bonding is involved in the
rate-determining step.11
Density functional theory (DFT) calculations were carried
out to understand the formation of the radical species (ESI†).
ꢀ
We found that neither the reaction 2 PQI - PQIHꢀ + PQI
ÀH
À
(DGr = 107.5 kJ molÀ1) nor the reaction PQI + PPh3 - PQIꢀ
+
PPh3 (DGr = 193.3 kJ molÀ1) was favored thermodynamically;
therefore, we expected direct complex formation (Fig. S9, ESI†)
by the reaction [PQI + PPh3]solvent cage - PQI–PPh3 (1), whose
activation barrier was 77.2 kJ molÀ1 and which resulted in a
stable intermediate I1 (À8.3 kJ molÀ1). This form could interact
with a second PQI resulting in the experimentally observed
stable 2bꢀ and PQIHꢀ through stable radical pairs (Fig. S10,
ESI†) in a solvent cage (2), which is thermodynamically favored
(DGr = À23.3 kJ molÀ1). After the separation of the radicals, the
resting PQIHꢀ could react with another PQIHꢀ predicting an
antiferromagnetic interaction. Taking into account the reaction
between forms having stacking interactions (3), the energetics
was exergonic (DGr = À11.4 kJ molÀ1). From I1 both the 2b
(À15.3 kJ molÀ1) and 3b (À26.4 kJ molÀ1) products could form
directly by internal hydrogen transfer or cyclization (4–5), where
+
ꢀ
Fig. 3 Reactions of 2,7-R-PQI derivatives with PPh3 in CH3CN under Ar.
(A) Vin versus [PPh3]0 for the formation of 3bꢀ at 25 1C ([1b] = 1 mM). (B) Vin
versus [1b]0 for the formation of 3bꢀ at 25 1C ([PPh3] = 10 mM). (C) Eyring
plot for the formation of 3bꢀ. (D) Vin versus [PPh3]0 for the formation of 3dꢀ
at 25 1C ([1d] = 1 mM). (B) Vin versus [1d]0 for the formation of 3dꢀ at 25 1C
([PPh3] = 10 mM). (C) Eyring plot for the formation of 3dꢀ.
the latter one had a significantly lower barrier by 25.3 kJ molÀ1
.
The formation of the final products 2b and 3b could also occur
by hydrogen transfers of the PQI derivatives; however, it was
ꢀ
N-Methyl- and N-phenyl-9,10-phenanthrenquinone monoi- found that the reaction 2bꢀ + PQI - 2b + PQI
was strongly
ÀH
mine did not react with triphenylphosphine under the same endergonic (DGr = 126.6 kJ molÀ1) while the quenching by
conditions, indicating that the hydrogen on the N-atom seems PQIHꢀ and PQIH2 was less unfavored with DGr = 16.3 and
to have a special role in this reaction. Phenanthrenequinone 15.7 kJ molÀ1 (6–7). Note that these values are lower by
monoxime reacted with triphenylphosphine in acetonitrile 11.1 kJ molÀ1 in the case of 3b (8–9); however, the latter one
under argon at room temperature and yielded triphenylpho- could be formed indirectly through I1. The formation of 2b or
sphine oxide (proved by IR n(P–16O) = 1191 cmÀ1
23 ppm and UV-vis spectroscopy (Fig. S7, ESI†)) and 9,10- PQIH2. Comparing the energies of the predicted intermediates,
phenanthrene quinone monoimine (n(N–H) = 3200 cmÀ1
the type of atom transfer was analyzed. We proposed SET-PT
,
31P-NMR 3b is depending on which side of 2bꢀ interacts with PQIHꢀ or
)
(Fig. S8, ESI†). Detailed kinetic studies on the reaction of 1b (single electron transfer-proton transfer) in the case of
(0.25–1 mM) with triphenylphosphine (2.5–20 mM) were quenching from PQIHꢀ and SPLET (sequential proton loss
carried out in CH3CN at 15–45 1C, and the formation (decay) of electron transfer) in the case of quenching from PQIH2.
the radical was followed as an increase (decrease) in absorbance at
The radical complex formation of the expected open-form
486 nm (Fig. 3 and Tables S1 and S2, ESI†). Under these oxazaphospholyl radical (PQI ꢀ + PPh3 - 2bꢀ) was significantly
ÀH
conditions, the reaction rate is directly proportional to the more preferred both kinetically (D‡G = 44.0 kJ molÀ1) and
concentration of PQI (1b) and shows Michaelis–Menten type thermodynamically (DGr = À147.7 kJ molÀ1); however, the
saturation kinetics with [PPh3]0 (Vmax = 1.54(5) Â 10À3 MsÀ1
,
,
formation of PQI
was very unfavored, see above. We
ꢀ
ÀH
KM = 10 mM, kc = 1.54(6) sÀ1 at 25 1C) with DH‡ = 33(3) kJ molÀ1
proposed that previously I1 formed, which is due to the relatively
DS‡ = À193(9) J molÀ1 KÀ1, and DG‡ = 91(5) kJ molÀ1, establishing low activation barrier.
intermediate complex (adduct) formation in an associative-type
These results predict slow formation of the non-radical
bimolecular reaction (Fig. 3A–C). A similar kinetic feature was species 2b and 3b and the long-term existence of 2bꢀ.
observed for 1d with Vmax = 1.95(1) Â 10À2 MsÀ1, KM = 11 mM, and
In summary, we proposed a mechanism according to the
following reactions:
kc = 19(2) sÀ1 at 25 1C with DH‡ = 14(1) kJ molÀ1, DS‡
=
À238(4) J molÀ1
K
À1, and DG‡ = 84(2) kJ molÀ1 (Fig. 3D–F).
[PQI + PPh3]solvent cage - I1
(1)
The activation enthalpy of 14(1) kJ molÀ1 and the calculated
Gibbs energy of 84(2) kJ molÀ1 observed for 1d are smaller than
those observed for 1b, which is consistent with the higher
reactivity of 1d with PPh3 (krel(1d/1b) = 15 at 25 1C). The relative
rate based on the formation and decay process (Vform/Vdec) of 2bꢀ
DGr = À8.3 kJ molÀ1
I1 + PQI - [RP1]solvent cage - 2bꢀ + PQIHꢀ
(2)
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021
New J. Chem.

View Article Online
Communication
NJC
both 2bꢀ and PQIHꢀ had a similarity to the experimental
spectrum but species in which PQI, PQIHꢀ, and PQIH2
bonded to 2bꢀ in different orientations showed better
agreement and it was the same in the case of TEMPO as well
(see Fig. S11, ESI†).
Conclusions
Based on detailed kinetic and spectroscopic studies and DFT
calculations,
(Fig. 4).
a plausible mechanism has been proposed
The 9,10-phenanthrenequinone monoimine (1b) forms a
charge-transfer complex in a fast preequilibrium process with
triphenilphosphine (Michaelis–Menten kinetics) followed by
the formation of a stable complex (I1). The end products can
be formed directly from I1 but thermodynamically the
formation of the oxazaphospholyl radical is more favored. The
hydrogen transfer from I1 to the 9,10-phenanthrenequinone
Fig. 4 Proposed mechanism of 1,3,2-oxazaphosphole formation.
DGr = À23.3 kJ molÀ1
monoimine yields radical-pair
1 (RP1), probably via an
electron-transfer-proton-transfer mechanism (ET-PT) based on
the stoichiometry and the small KIE value.11 Radicals of this type
seem to be very stable due to the high delocalization of
the unpaired electron on the large aromatic system and/or the
H-bonds. The end product formation can be interpreted by slow
H-atom abstraction and P–O bond formation via SET-PT (single
electron transfer-proton transfer) in the case of quenching
from PQIHꢀ and SPLET (sequential proton loss electron transfer)
based on the DFT calculations. Since electron-withdrawing
substituents on the imine enhance and electron-releasing sub-
stituents decrease the overall reaction rate the heterodiene acts
as a Lewis acid in a normal electron-demand cycloaddition
reaction. The present study allowed us to gain more insight
into the mechanism of the cheletropic reaction of monoimines
with PPh3 via unexpected radical intermediates resulting in 2,3-
dihydro-2,2,2-triphenylphenanthro[9,10-d]-1,3,2l5 oxazaphosp-
holes, molecules that can be applied as flavoprotenin mimics
and highly efficient organocatalysts for various oxidation
processes.
[PQIHꢀÁ Á ÁPQIHꢀ]stacking - [PQIÁ Á ÁPQIH2]stacking
(3)
(4)
(5)
(6)
(7)
(8)
(9)
DGr = À11.4 kJ molÀ1
I1 - 2b
DGr = À7.0 kJ molÀ1
I1 - 3b
DGr = À11.8 kJ molÀ1
2bꢀ + PQIHꢀ - 2b + PQI
DGr = 16.4 kJ molÀ1
2bꢀ + PQIH2 - 2b + PQIHꢀ
Conflicts of interest
DGr = 15.7 kJ molÀ1
There are no conflicts to declare.
2bꢀ + PQIHꢀ - 3b + PQI
DGr = 5.3 kJ molÀ1
Acknowledgements
2bꢀ + PQIH2 - 3b + PQIHꢀ
Financial support from the Hungarian National Research
Fund (OTKA K108489 to JK), GINOP-2.3.2-15-2016-00049 (JK)
and TKP2020-IKA-07 (JK) is gratefully acknowledged. This
work was partially supported by the European Union and the
European Social Fund through project Supercomputer, the
DGr = 4.6 kJ molÀ1
´
We simulated UV-vis spectra to see which species had the National Virtual Lab, grant no.: TAMOP-4.2.2.C-11/1/KONV-
best correlation to the experimental spectrum. The spectra of 2012-0010 (MP).
New J. Chem.
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021

View Article Online
NJC
Communication
´
7 G. Szekely, N. Bagi, J. Kaizer and G. Speier, New J. Chem.,
Notes and references
2015, 39, 5908.
1 W. L. Mock, Cheletropic Reactions in Organic Chemistry,
Pericyclic Reactions, ed. A. P. Marchand and R. E. Lehr,
Academic Press, New York, 1977, vol. 35, ch. 2, pp. 141–179.
2 (a) F. Ramirez, Acc. Chem. Res., 1968, 1, 168; (b) F. Ramirez,
C. P. Smith, J. F. Pilot and A. S. Gulati, J. Org. Chem., 1968,
33, 3787; (c) D. Hellwinkel, W. Blaicher, W. Krapp and
W. S. Sheldrick, Chem. Ber., 1980, 113, 1406.
8 (a) F. Neese, ORCA an ab initio, density functional and
semiempirical program package, Version 2.6, University of
Bonn, Bonn, Germany, 2008; (b) F. Neese, J. Chem. Phys.,
2001, 115, 11080–11096; (c) F. Neese, J. Phys. Chem. A,
2001, 105, 4290–4299; (d) F. Neese, J. Chem. Phys., 2003,
118, 3939–3948; (e) F. Neese, J. Chem. Phys., 2005,
122, 034107.
3 (a) A. A. Kutyrev and V. V. Moskva, Russ. Chem. Rev., 1987,
9 A. Rockenbauer and L. Korecz, Appl. Magn. Reson., 1996, 10,
29–43.
´
56, 1798; (b) E. Balogh-Hergovich and G. Speier, Phosphorus,
Sulfur Silicon Relat. Elem., 1990, 48, 223; (c) M. M. Sidky and
M. F. Zayed, Tetrahedron Lett., 1971, 12, 2313; (d) M. M. Sidky,
M. F. Zayed, A. A. El-kateb and I. T. Hennawy, Phosphorus,
Sulfur Silicon Relat. Elem., 1981, 9, 343.
10 (a) G. Speier, New J. Chem., 1994, 18, 143; (b) G. Nocton,
W. W. Lukens, C. H. Booth, S. S. Rozenel, S. A. Medling,
L. Maron and R. A. Andersen, J. Am. Chem. Soc., 2014, 136,
8626–8641; (c) J. M. Wittman, R. Hayoun, W. Kaminsky,
M. K. Coggins and J. M. Mayer, J. Am. Chem. Soc., 2013, 135,
12956–12959.
´
¨
´ ´
4 G. Speier, Z. Tyeklar, V. Fu¨lop and L. Parkanyi, Chem. Ber.,
1988, 121, 1685.
5 N. Bagi, J. Kaizer and G. Speier, RSC Adv., 2015, 5, 45983.
6 I. Bors, J. Kaizer and G. Speier, RSC Adv., 2014, 4, 16928.
¨
11 R. Tyburski, T. Liu, S. D. Glover and L. Hammarstrom, J. Am.
Chem. Soc., 2021, 143, 560–576.
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2021
New J. Chem.