Alkyl radical generation in water under ambient conditions - A new look at the Guareschi reaction of 1897

Article abstract of DOI:10.1002/anie.200700632

(Chemical Equation Presented) Chemical and biochemical significance of a long-forgotten 19th century observation: The hydrocarbon production from quaternary glutarimides on neutralization in water is the consequence of formation of alkyl radicals, which means that it is possible to generate alkyl radicals under very mild conditions. Oxygen trapping competes with hydrogen abstraction (see scheme).

Full text of DOI:10.1002/anie.200700632

Angewandte
Chemie
DOI: 10.1002/anie.200700632
Aqueous Radicals
Alkyl Radical Generation in Water under Ambient Conditions—
A New Look at the Guareschi Reaction of 1897**
Bao Nguyen, Katya Chernous, Daniel Endlar, Barbara Odell, Michela Piacenti, John M. Brown,*
Alexander S. Dorofeev, and Alexander V. Burasov
In memory of Arthur J. Birch
In a series of papers initially published between 1897 and
1901, I. Guareschi (Turin) and his colleagues reported the
synthesis of 4,4’-dialkyl-3,5-dicyanoglutarimides by the con-
densation of ketones other than acetone with ethyl cyano-
acetate in ethanolic ammonia.[1,2] The reaction has frequently
been used subsequently as a route to functional quaternary
centers,[3] and because glutarimides have shown potential as
drugs.[4] In Guareschiꢀs work, these otherwise unremarkable
compounds demonstrated a remarkable property. The imide
1a, on neutralization with alkaline ammonia or Mg(OH)2,
gave pyridone 2 with evolution of ethane [Eq. (1)]. Likewise,
procedure,[1] and both species characterized by 1H and
13C NMR spectroscopy.[5] In [D6]DMSO, the imide 1a exists
predominantly as a mixture of the two syn-CN diastereomers,
one of which was further characterized by X-ray analysis. Salt
3a exists in D2O mainly as two diastereomers with the enolate
structure drawn in the formula. A minor ( ꢀ 10%) more
symmetrical component is also present. Under these con-
ditions, NH and CHCN protons were completely exchanged
for deuterons. On dissolution of salt 3a in D2O buffered to
pH 7.2, 8.0, and 9.1, respectively, a slow, clearly visible
evolution of C2H5D (EI MS: m/z 31.0535; calcd 31.0532)
was observed, and formation of the aromatic product 2
occurred concurrently. The ethyl homologue of 2 was not
detected. After reaction in H2O (pH 8 buffer, 408C, 4h,
sealed vial) ethane was detected in the head-space with
< 1:1000 parts of butane. The fastest reaction occurred at
pH 8.0; in all three cases both ethane evolution and product
formation ceased before the reaction was complete. In
addition to the ethane, a significant amount of EtOD
(ꢁ20%) was consistently observed when the reaction was
carried out in D2O.
=
hydrocarbons RH (R = Et, Pr, nBu, n-C6H13, Me2C CH-
CH2CH2, n-C9H19) were formed starting with glutarimides
derived from ketones RCOMe. This aspect of Guareschiꢀs
work appears to have been at best neglected, at worst
forgotten.
We began by verifying a simple case. Imide 1a was
isolated as the ammonium salt 3a according to the original
[*] B. Nguyen, K. Chernous, D. Endlar, Dr. B. Odell, M. Piacenti,
Dr. J. M. Brown
Chemical Research Laboratory
Oxford University
In order to elucidate the mechanism of alkane elimina-
tion, a series of analogues was prepared. In some cases it
proved convenient to isolate the imide (1c, 1d, 1g, 1h), and in
others the ammonium salt was more easily accessible (3a, 3b,
3e, 3 f, 3i). They could be interconverted (NH3(g) or P2O5 in
vacuum) and were used as prepared, normally in deuterated
buffer solution. The benzylimide salt 3b was subjected to the
conditions described above. Formation of product 2 occurred
over several days, but benzyl alcohol was the main product
along with some benzaldehyde; C7H7D was not detected after
extraction into CDCl3. Reaction in 10%-enriched 17OH2
demonstrated that the oxygen atom of the benzyl alcohol
did not arise from water (m/z 109/108 = 0.06, calculated for
the absence of 17O incorporation: 0.07). Further, an Ar-
degassed solution of 3b showed little decomposition after
several days; after bubbling through this solution O2 for 5 min
Mansfield Road, Oxford OX13TA (UK)
Fax: (+44)1865-285-002
E-mail: john.brown@chem.ox.ac.uk
Homepage: http://www.chem.ox.ac.uk/researchguide/jbrown.html
A. S. Dorofeev, A. V. Burasov
Institute of Organic Chemistry
Leninsky Prospect 47, 119991 Moscow (Russia)
[**] We thank the EU for award of a Marie Curie Host Fellowship
(HPMT-CT-2001-00317) for supporting the exchange of M.P. from
the University of Florence, 2005. B.N. acknowledges the award of
Clarendon and ORS Scholarships from Oxford. We thank Prof. Tito
Scaiano (Ottawa) and Prof. Allen Hill (Oxford) for very useful
discussions. A.S.D. is indebted to the Russian Science Support
Foundation for a postgraduate scholarship in 2007. For X-ray
structures we thank Dr. Andrew Cowley and Dr. Amber Thompson.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2007, 46, 7655 –7658
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
7655

Communications
the reaction proceeded normally.[6] All of this suggested an
from a comparison with the rate constants for cyclization of
the 5-hexenyl radical (k1 = 1 ” 105 sꢂ1) and for ring-opening of
the cyclopropylmethyl radical (k1 = 1.3 ” 108 sꢂ1).[9]
alkyl radical intermediate. Cyclic voltammetry experiments in
H2O/0.1m NaClO4 revealed an irreversible one-electron
oxidation at 0.80 V, pH 6.89. A similar peak was observed in
CH3CN. In H2O, the corresponding peak at 0.75 V broadened
in more basic solution and at pH 9.64, the highest value used,
a second broad oxidation peak centered at 0.50 V was seen.
With the substituted-benzyl derivatives 3c and 3d, com-
parable rates of product formation were observed at 258C.
This process corresponded to a first-order rate constant (k1)
for the decomposition of 3c of 2.2 ” 10ꢂ6 sꢂ1 and for that of 3d
of 2.7 ” 10ꢂ6 sꢂ1. The relative reactivities were confirmed by an
internal competition experiment. An induction period was
The iPr derivative 3g was prepared by a two-step synthesis
via the Knoevenagel product, following Cope et al.[10] As
expected, alkyl cleavage was much faster for 3g than for the
nPr analogue 3h. iPrOOH (5; d(CHO) = 4.20 ppm) was the
major oxygenated product, confirmed by in situ NMR
comparison with authenic 5,[11] which reacted slowly with
NaI to give iPrOH. In contrast to the benzylglutarimide 3b,
derivative 3g is reactive in degassed solutions under standard
conditions, indicating mechanistic differences between the
routes to hydrocarbon and alcohol product.
One intriguing claim in Guareschiꢀs original work is that
imide salt 3i derived from hexan-3-one gave exclusively
ethane on aromatization in water.[1] We observed, however,
the formation of both [D1]ethane and [D1]propane (EI MS),
with a stronger signal from the former. Monitoring by NMR
showed that ethanol and propanol were both formed in a ratio
of 69:31. In accordance, the pyridones 2 with nPr and Et
instead of Me as substituent at position 4were formed in a
ratio of 70:30. Hydrocarbon formation was the minor process
(RD:ROH ca. 1:2). The observed selectivity towards ethyl
cleavage may originate from the different hydrophobicity and
differential solvation of ethyl and propyl radicals, since their
gas-phase stabilities are the same.[12]
observed, but absent when Ba2+ rather than NH4 was the
+
cation. When decomposition of 3c was taken to 80%
completion, the product was a mixture of 4-methoxybenzal-
dehyde (45%) and 4-methoxybenzyl alcohol (55%). This is
consistent with benzyl radical generation, and its capture by
dissolved oxygen. Hence the products arise via a benzylic
hydroperoxide.
Imide salt 3e was prepared from oct-7-en-2-one in 34%
yield.[7a] Reaction in buffered D2O resulted in slow formation
of a 47:53 mixture of hex-5-en-1-ol and cyclopentylmethanol
together with pyridone 2 (Scheme 1). Formation of the
The decomposition of glutarimides in buffered D2O is
suppressed by the stable spin trap 4-hydroxy-TEMPO 6. In its
absence, the iPr species 3g reacted completely after 21 h at
608C. In the presence of one equivalent of 6, 27% decom-
position was observed under the same conditions. Similar
observations were made using the hexenyl derivative 3e.
Reactions run in the presence of azobisisobutyronitrile
(AIBN; added in [D6]DMSO) showed an enhanced rate of
decomposition, however.
A mechanistic postulate consistent with these observa-
tions is shown in Scheme 2. In path A reaction is initiated by
D abstraction, most likely by oxygen, and a stabilized radical
7 is the precursor of the pyridone 2 and R’C. An alternative
path B involves an electron-transfer step from imide 3 to
oxygen to generate first the corresponding anion radical, but
seems less likely because of the endothermicity of the step.[13]
Scheme 1. Radical rearrangements in the fragmentation of imides 3.
alcohols was verified by NMR comparison with authentic
samples. Extraction into CDCl3 confirmed the absence of
both cyclohexanol and C6 hydrocarbons. The alcohol ratio in
the product was consistent, both over time and between
experiments.
Compound 3 f was prepared from the corresponding
ketone in 41% yield.[7b] Monitoring its decomposition by
NMR spectroscopy in buffered D2O showed that but-3-en-1-
ol was formed, but in lower amount than pyridone 2. On
extraction with CDCl3 the remaining C4 product was found to
be [4-2H]but-1-ene (Scheme 1; d(4-CH2) = 0.95 ppm, JH,D
=
1.2 Hz). No additional products were detectable. For compar-
ison, SN1 hydrolysis of cyclopropylmethyl mesylate (4) at
pH 8.0 gave a mixture of cyclobutanol and cyclopropylme-
thanol together with traces of but-3-en-1-ol (49.5:48.5:2).[8]
These results demonstrate the intermediacy of an alkyl radical
in the Guareschi reaction and indicate that it possesses a
lifetime of > 1 ms under the reaction conditions. This follows
Scheme 2. Suggested pathways for the radical fragmentation of 3 in
D2O.
7656
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 7655 –7658

Angewandte
Chemie
Once formed, the alkyl radical R’C can engage in a chain
reaction by further D abstraction, in competition with being
trapped by dissolved oxygen.[14] In one of the early papers on
ethyl radical generation in water, a switch from radical
recombination to radical–oxygen chemistry was observed in
the presence of O2.[14a] Reaction of an alkyl radical with O2 in
water is diffusion-controlled, at k = 2 ” 109 Lmolꢂ1 sꢂ1 for
ethyl. This is consistent with the radical lifetime of several
microseconds observed herein, at air under ambient con-
pathways in the thermal fragmentation of 8, but favored
radical pathways in the fragmentation of the related com-
pound 9.[16] Guareschiꢀs chemistry is important for its unique-
ness: ambient conditions, near-neutral aqueous solution,
ditions ([O2] = 2.5 ” 10ꢂ4 [14b]).
m
The feasibility of the proposed pathway was further tested
through DFT calculations at the B3LYP/6-31G(d,p) level.
The results are summarized in Scheme 3, and presented more
fully in the Supporting Information. They indicate that the
dependency on an unusual intervention of molecular
oxygen. There is both chemical and biochemical significance
in the ability to generate alkyl radicals under such mild
conditions.[18]
Received: February 11, 2007
Revised: July 15, 2007
Published online: September 4, 2007
ꢂ
Keywords: alkyl radicals · C C cleavage · glutarimides ·
radical trapping · water chemistry
.
[1] The work described was initially published in Atti. R. Accad.
Torino 32–35, 1897–1899; a useful, and more accessible prØcis of
Guareschiꢀs work appeared in Chem. Zentralblatt II 1901, 577–
582; this journal is not available in electronic format but the
parallel J. Chem. Soc. Abstracts (free Web access) is; see
particularly the successive summaries of work by E. Grande, E.
Quenda, A. Pasquali, L. Sabbatini, I. Guareschi, J. Chem. Soc.
Abstr. 1898, 74, 272 – 274.
Scheme 3. Enthalpy changes in key steps of the proposed fragmenta-
tion mechanism, computed at the B3LYP/6-31G(d,p) level. All the data
on individual species were corrected to include zero-point energies,
without taking thermal contributions into account.
[2] a) I. Guareschi, Gazz. Chim. Ital. 1918, 48ii, 83 – 98; b) I.
Guareschi, Gazz. Chim. Ital. 1919, 49i, 124– 133.
[3] For example: a) U. Schoen, J. Antel, R. Brückner, J. Messinger,
R. Franke, A. Gruska, J. Med. Chem. 1998, 41, 318 – 331; b) G. J.
Handley, E. R. Nelson, T. C. Somers, Aust. J. Chem. 1960, 13,
127 – 144; c) S. M. McElvain, D. H. Clemens, J. Am. Chem. Soc.
1958, 80, 3915 – 3923; d) A. J. Birch, R. Robinson, J. Chem. Soc.
1942, 488 – 497; G. A. R. Kon, J. Thorpe, J. Chem. Soc. 1919, 115,
686 – 704.
key step of a H/D abstraction from C3 of anion 3g by CiPr in
the radical chain is enthalpically favored by nearly 20 kcal
molꢂ1, and by 1.7 kcalmolꢂ1 with COOiPr (Scheme 3a). The
corresponding abstraction from CH3CN by CiPr is shown for
comparison in Scheme 3b. The resulting radical 7 bears a
resemblance to the stabilized long-lived radicals character-
ized by Scaianoꢀs group. These react slowly with O2 once
formed, and in our case would give the alternative fragmen-
tation of radical 7 time to occur.[15] This subsequent fragmen-
tation (Scheme 3c) is also facilitated by the aromaticity of the
pyridindionyl anion 2, which provides a thermodynamic
[4] Example: “Martindale, the complete drug reference, 2006”
(http://www.medicinescomplete.com).
[5] The 13C NMR spectrum of imide 1a in [D6]DMSO shows two
symmetrical diastereomers in a 1:1 ratio. For the ammonium salt
3a in D2O, two unsymmetrical diastereomers (diastereotopic
CH2 groups) with traces of a third symmetrical isomer were
observed. For full details see the Supporting Information.
[6] The solubility of oxygen in water under ambient conditions is
1.28 mm: J. Livingston, R. Morgan, A. H. Richardson, J. Phys.
Chem. 1930, 34, 2356 – 2366.
ꢂ
driving force for the ensuing C C cleavage step, also entropi-
cally favored. For the alternative direct formation of the
ꢂ
radical at C4by homolytic C C cleavage, DH0 is 51 kcal
[7] a) R. Sedrani et al., J. Am. Chem. Soc. 2003, 125, 3849 – 3859;
b) J. Christoffers, T. Kauf, T. Werner, M. Roessle, Eur. J. Org.
Chem. 2006, 2601 – 2608, and references therein.
[8] See: a) Z. Majerski, S. Borcic, D. E. Sunko, Tetrahedron 1969, 25,
301 – 313; b) M. Nikoletic, S. Borcic, D. E. Sunko, Tetrahedron
1967, 23, 649 – 660.
[9] D. Griller, K. U. Ingold, Acc. Chem. Res. 1980, 13, 317.
[10] A. C. Cope, C. M. Hofmann, C. Wyckoff, E. J. Hardenbergh, J.
Am. Chem. Soc. 1941, 63, 3452 – 3456.
molꢂ1. Finally, the competing pathway of oxygen trapping of
the alkyl radical is shown in Scheme 3d.
There are several scattered examples in which a hetero-
cyclic aromatic compound is formed by elimination of an
alkyl group, generally under forcing conditions.[16,17] Coherent
interpretation of these processes has previously been lacking;
for example, Elderfield and co-workers opted for carbanion
Angew. Chem. Int. Ed. 2007, 46, 7655 –7658
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7657

Communications
[11] C. Walling, S. A. Buckler, J. Am. Chem. Soc. 1955, 77, 6032 –
d) J. C. Scaiano, A. Martin, G. P. A. Yap, K. U. Ingold, Org. Lett.
2000, 2, 899 – 901.
6038.
[16] R. C. Elderfield, E. C. McClenachan, J. Am. Chem. Soc. 1960,
82, 1982 – 1988, and earlier papers.
[12] S. Gronert, J. Org. Chem. 2006, 71, 1209 – 1219, and references
therein.
[17] Selected examples: a) K. N. Zelenin, I. V. Ukraintsev, V. V.
Alekseev, Chem. Heterocycl. Compd. 1998, 34, 329 – 333; b) S. D.
Larsen, E. Martinborough, Tetrahedron Lett. 1989, 30, 4625 –
4628; c) A. P. Reddy, V. Veeranagaiah, Indian J. Chem. Sect. B
1989, 28, 272 – 273; d) K. Schulze, C. Richter, R. Ludwig,
Tetrahedron Lett. 1989, 30, 2369 – 2370; e) P. K. Dubey, C. V.
Ratnam, Indian J. Chem. Sect. B 1980, 19, 863 – 865; f) R.
Huisgen, R. Grashey, H. Gotthardt, R. Schmidt, Angew. Chem.
1962, 74, 29; Angew. Chem. Int. Ed. 1962, 1, 48 – 49.
[18] For full experimental details see the Supporting Information.
CCDC 653648–653653 contain the supplementary crystallo-
graphic data for this paper. These data can be obtained free of
charge from The Cambridge Crystallographic Data Centre via
www.ccdc.cam.ac.uk/data_request/cif.
[13] Electron transfer from the radical anion to O2 is endothermic,
given the standard potentials derived from CV study (see above)
and for the half-reaction O2 + e!O2Cꢂ (ꢂ0.13 V: J. Petlicki,
T. G. M. van de Ven, J. Chem. Soc. Faraday Trans. 1998, 94,
2763 – 2767).
[14] a) B. Hickel, J. Phys. Chem. 1975, 79, 1054– 1059; b) M. A.
Marchaj, D. G. Kelley, A. Bakac, J. H. Espenson, J. Phys. Chem.
1991, 95, 4440 – 4441.
[15] See for example: a) E. Font-Sanchis, C. Aliaga, R. Cornejo, J. C.
Scaiano, Org. Lett. 2003, 5, 1515 – 1518; b) E. Font-Sanchis, C.
Aliaga, E. V. Bejan, R. Cornejo J. C. Scaiano, J. Org. Chem.
2003, 68, 3199 – 3204; c) E. Font-Sanchis, C. Aliaga, K. S.
Focsaneanu, J. C. Scaiano, Chem. Commun. 2002, 1576 – 1577;
7658
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 7655 –7658