Aminomethylation of Michael acceptors: Complementary radical and polar approaches mediated by dialkylzincs

Article abstract of DOI:10.1002/chem.201102366

Phtalimidomethyl iodide and substituted maleimidomethyl iodide were used as radical precursors in dialkylzinc-mediated radical addition to diethyl fumarate. The reactions led stereoselectively to functionalized pyrrolizidines. The radical mechanism was supported by spin-trapping experiments and rationalized by theoretical calculations. Radical additions, on the one hand, and carbozincation followed by transmetalation with copper(I), on the other, were shown to be complementary methods to achieve the formal aminomethylation of activated unsaturated compounds. Copyright

Full text of DOI:10.1002/chem.201102366

FULL PAPER
DOI: 10.1002/chem.201102366
Aminomethylation of Michael Acceptors: Complementary Radical and Polar
Approaches Mediated by Dialkylzincs
Julien Maury, Dominique Mouysset, Laurence Feray,* Sylvain R. A. Marque,
Didier Siri, and Michꢀle P. Bertrand*[a]
Abstract: Phtalimidomethyl iodide and substituted maleimidomethyl iodide were
used as radical precursors in dialkylzinc-mediated radical addition to diethyl fuma-
rate. The reactions led stereoselectively to functionalized pyrrolizidines. The radi-
cal mechanism was supported by spin-trapping experiments and rationalized by
theoretical calculations. Radical additions, on the one hand, and carbozincation
followed by transmetalation with copper(I), on the other, were shown to be com-
plementary methods to achieve the formal aminomethylation of activated unsatu-
rated compounds.
Keywords: aminomethylation · dia-
lkylzincs · ESR experiments · radi-
cal reactions · theoretical calcula-
tions · zinc
Introduction
The growing demand for environmentally friendly processes
has stimulated the development of tin-free radical method-
ologies.[1] In this context, attention has focused on the use of
dialkylzincs, not only as simple sources of radicals in atom
transfer reactions,[2] but also as chain transfer agents.[3,4,5] An
interesting feature of these reagents, which can be compared
to alkylboron derivatives, is the possibilities they offer to
generate nucleophilic species from an initial radical elemen-
tary step, and thus to perform radical polar crossover reac-
tions. In this regard, dialkylzincs enable the use of a wider
range of radical acceptors than alkylboranes.[1e–g,4b,6]
In the course of our studies, we recently described the di-
alkylzinc-mediated addition of functionalized radicals of
type 2, chosen as synthetic equivalents of hydroxymethyl
radicals, to activated double bonds (Scheme 1).[7,8] Therein,
we report the results obtained with phtalimidomethyl radical
4, selected as a synthetic equivalent of the aminomethyl rad-
ical. The use of phtalimidomethyl radical 4 for the hydro-
aminomethylation of alkenes has been investigated by Zard
et al. by using the xanthate transfer methodology. It must be
noted that these reactions concerned electron-rich accept-
ors.[9]
Scheme 1. The formation of radicals 2 and 4.
Very few examples of additions of radical 4 to electron-
poor radical acceptors have been reported in the litera-
ture.[10] In addition, the related organometallic reagents, i.e.,
mixed Zn/Cu species, were shown to have little affinity for
electrophiles, such as enones or nitro olefins.[11]
Results and Discussion
Phtalimidomethyl iodide 3 was prepared in three steps from
phtalimide by adapting a known procedure.[12] Dialkylzinc-
mediated addition of radical 4 was first tested on diethyl fu-
marate in the presence of dialkylzincs (2 or 3 equiv;
Table 1). Air was introduced into the reaction mixture
through a syringe pump at room temperature. The use of at
least 5 equivalents of 3 was necessary to completely avoid
the competitive addition of methyl or ethyl radicals formed
from the mediator. In the cases shown in Table 1, entries 1
and 5, the reaction led to the formation of a single isomer of
tricyclic compound 5. The structure of this compound was
assigned from X-ray analysis. Yields were determined from
[a] J. Maury, Dr. D. Mouysset, Dr. L. Feray, Prof. S. R. A. Marque,
Prof. D. Siri, Prof. M. P. Bertrand
Institut de Chimie Radicalaire, UMR 7273
Facultꢀ des Sciences et Techniques de Saint-Jꢀrꢁme
Avenue Normandie-Niemen, Aix-Marseille Universitꢀ
13397 Marseille Cedex 20 (France)
1
the H NMR spectrum of the crude mixture in the presence
E-mail: laurence.feray@univ-amu.fr
of pentamethylbenzene as the internal reference. Purifica-
tion of 5 by chromatography on silica gel led to its partial
dehydration and compounds 5 and 6 were isolated in 54 and
michele.bertrand@univ-amu.fr
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201102366.
Chem. Eur. J. 2012, 18, 3241 – 3247
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3241

Table 1. Dialkylzinc-mediated radical addition to diethyl fumarate.
Entry
ZnR2 ([equiv])
Workup
5 [%]
6 [%]
1
2
3
4
5
ZnMe2 (3)
ZnMe2 (3)
ZnMe2 (3)
ZnMe2 (3)
ZnEt2 (2)
sat. NH4Cl
sat. NH4Cl
ClSiMe3
Ac2O
sat. NH4Cl
95[a]
54[b]
–
–
38[b]
92[b]
94[b]
–
–
96[a]
1
[a] Yield was determined by H NMR spectroscopy with pentamethylben-
zene as the internal reference. [b] Isolated yield after purification on
silica gel.
38% yields, respectively (Table 1, entry 2).[13] These results
can be compared to the data reported by Mariano et al. for
the photoinitiated addition of N-(silylmethyl)phtalimide to
diethyl fumarate, which led to a mixture of two silylated dia-
stereomeric adducts in a 2:1 ratio.[10a] When the crude reac-
tion mixture was treated with trimethylsilyl chloride or
acetic anhydride, only compound 6 was isolated in high
yield (Table 1, entries 3 and 4).
Mono-activated olefins were also tested as radical accept-
ors. However, they were much less reactive than diethyl fu-
marate. The results obtained with ethylacrylate, acryloni-
trile, and 3-(prop-2-enoyl)-1,3-oxazolidin-2-one are reported
in Scheme 2. Whereas the first acceptor in the series led to
the expected tricyclic product, albeit in only an average
yield, the reaction of diethylzinc and 3 with acrylonitrile and
the enoyloxazolidinone led to acyclic adducts 8 and 9.
Scheme 3. Mechanism for the stereoselective formation of 5.
give a zinc enolate (step c). The cyclization in step d gives 5
after hydrolysis of the zinc alkoxide (step e). The stereo-
chemical outcome of the reaction can be explained if the in-
termediate zinc enolate adopts a chair-like configuration in
the transition state in which the ethoxycarbonyl substituent
lies in a pseudo-equatorial position (Scheme 3).[15]
The formation of radical 4 was supported by ESR experi-
ments performed with phtalimidomethyl iodide
3 and
Me2Zn or Et2Zn in the presence of 2,4,6-tri-tert-butylnitroso-
benzene (TTBNB), which was used as the radical trap. As
an example, at room temperature, an undegassed solution of
diethylzinc in dichloromethane, containing compound 3 and
TTBNB, led to the signals of four radical adducts. As ex-
pected, two nitroxyl and two anilino radicals were detected
from the competitive trapping of ethyl and phtalimidometh-
yl radicals (Scheme 4). The corresponding spectra are shown
in Figure 1. Authentic ESR spectra of phtalimidomethyl rad-
ical adducts NP and AP were obtained by irradiating iodide
3 and TTBNB in the presence of (Bu3Sn)2 (the nitroxyl and
anilino radicals were formed in a 44:56 ratio, Figure 2). It is
noteworthy that the spectrum of the anilino radical adduct
could be recorded when 3 and TTBNB were mixed in the
absence of light (Figure 3).[16]
Scheme 2. i) ZnEt2 (2 equiv), 3 (5 equiv), CH2Cl2, RT, 18 h; then Ac2O;
then NH4Cl.
The mechanism, proposed for the formation of compound
5, is depicted in Scheme 3. Dialkylzincs react with oxygen to
give alkyl radicals.[4g] The phtalimidomethyl radical 4,
formed through iodine atom transfer (step a),[14] adds to the
activated double bond (step b). The resulting a-carbethoxy
radical evolves through homolytic substitution at zinc to
The formation of a mixed dialkylzinc species through a
formal zinc iodine exchange could not be excluded at this
stage. Theoretical calculations enabled us to exclude homo-
lytic substitution at zinc involving spin density at the oxygen
3242
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 3241 – 3247

Aminomethylation of Michael Acceptors
FULL PAPER
alkylzinc), compound 11 was
isolated in 43% yield (Table 2,
entry 2). The formation of 11
can be explained through the
reaction of radical
4
with
oxygen (Scheme 5, path c). The
resulting peroxyl radical can be
trapped by the dialkylzinc re-
agent to give a peroxyzinc com-
pound. The reduction of the
latter by the excess R2Zn leads
to alcohol 11 after hydrolysis.
When the reaction was con-
ducted with diethylzinc in an
undegassed medium, products
10, 11, and 12 were isolated
C
Scheme 4. Trapping of Et and 4 by TTBNB, leading to the formation of nitroxyl (N) and anilino (A) radicals.
atom due to the extremely low value of the latter (Figure 4).
Nevertheless, zinc iodine exchange could be favored by the
possibility to coordinate zinc to the oxygen atom of the car-
bonyl group (step f, Scheme 3). This possibility, leading to a
chelated mixed dialkylzinc, has already been suggested by
Charette et al.[17] Although this exchange can be envisaged
in the case of diethylzinc, it is excluded in the case of dime-
thylzinc owing to the bond strength of the Zn–Me bond.[4g]
Control experiments were conducted in the absence of di-
ethyl fumarate. The results are listed in Table 2. No evolu-
tion of 3 was observed when the reaction was conducted in
the presence of dimethylzinc (Table 2, entry 1). In the pres-
ence of oxygen (in a ratio of 0.5:1 with respect to the di-
after 2 h, in 39, 14, and 22% yields, respectively (Table 2,
entry 3). In all likelihood, compound 12 results from the re-
arrangement of the intermediate zinc alkoxide.[18]
Two pathways are proposed in Scheme 5 to explain the
formation of N-methyl phtalimide 10. As discussed before,
radical 4 can be trapped reversibly by diethylzinc to give or-
ganozinc intermediate I, which can be hydrolyzed upon
workup to give compound 10 (path a). Deuteration experi-
ments resulting from workup with ND4Cl in deuterated
water led to traces of deuterated 10. This means that 10
cannot be formed through path a. Therefore, we suggest
that hydrogen atom transfer to radical 4 from any hydrogen
C
&
*
Figure 1. Trapping of Et and
4 by TTBNB. Ratio: NP ( )/AP ( )/
&
*
NEt (\)/AEt (ꢁ)=25:52:16:8. Et2Zn (1m in hexane, 0.1 mmol),
3 (0.5 mmol), and (tBu)3C6H2N=O (0.25m in CH2Cl2, 0.05 mmol) in un-
degassed CH2Cl2 at RT.
Figure 2. Trapping of
(Bu3Sn)2 (0.05 mmol),
CH2Cl2, 0.05 mmol) in undegassed CH2Cl2 at RT.
4
3
by TTBNB. Ratio: NP
( )/AP ( )=44:56.
(0.5 mmol), and (tBu)3C6H2N=O (0.25m in
Chem. Eur. J. 2012, 18, 3241 – 3247
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3243

L. Feray, M. P. Bertrand et al.
Scheme 5. Mechanism for the formation of 10, 11, and 12.
pare Table 2, entries 3 and 4). Surprisingly, no deuteration
was detected when quenching the reaction with ND4Cl/
D2O.[19] This remains puzzling, although the involvement of
an organocopper species formed from intermediate I was in-
directly demonstrated by the experiments depicted in
Scheme 6.
&
Figure 3. Spontaneous formation of AP ( ).
3
(0.5 mmol) and
(tBu)3C6H2N=O (0.25m in CH2Cl2, 0.05 mmol) in undegassed CH2Cl2 at
RT.
Figure 4. Spin density and NBO charges for radical 4 (calculated at the
UB3LYP/6-311+ +GACHTUNGTRENNUNG(3df,3pd)//UB3LYP/6-31+GCAHTUNGTRENN(UGN d,p) level of theory).
Scheme 6. i) ZnEt2 (1.2 equiv), CuI (1 equiv), CH2Cl2, RT, 2 h; ii) allyl
bromide (9 equiv), RT, 4 h; iii) NH4Cl; iv) diethyl fumarate (6 equiv), RT,
4 h; v) ethylacrylate (6 equiv), RT, 4 h.
Table 2. Phtalimidomethyl iodide in the presence of dialkylzincs.
The organocopper reagent was formed through the reac-
tion between iodide 3 (1 equiv) and diethylzinc (1.2 equiv)
followed by transmetalation with copper(I) (1 equiv). It was
trapped with allylbromide and ethylacrylate to give com-
pounds 13 and 15 in 79 and 52% yields, respectively. When
diethyl fumarate was used as the trap, adduct 14 was isolat-
ed in 71% yield.[20] It should be noted that none of the tricy-
clic compounds 5 or 6 were detected in this case because,
unlike the zinc enolate, the intermediate copper enolate was
not able to undergo cyclization (step d, Scheme 3).
Entry
ZnR2
Additive
10
[%]
11
[%]
12
[%]
Recovered 3
[%]
[a]
[a]
1
2
3
4
ZnMe2
ZnMe2
ZnEt2
ZnEt2
–
air
–
–
–
39
31
–
–
–
22
2
97
51
–
43
14
18
[b]
[b]
CuI (2 equiv)
–
[a] Reaction time: 18 h. [b] Reaction time: 2 h.
Under these conditions, formal aminomethylation can be
performed in good yield through the organometallic path-
way. These reactions, like the following ones, lead to b- and
g-amino esters and diesters.
atom donor present in the reaction mixture (path b) is a
possible route to compound 10.
Attempts to trap I with CuI to form an organocopper spe-
cies led to the result reported in Table 2, entry 4. The
amount of oxidation products 11 and 12 was reduced (com-
Finally, we decided to test activated acetylenic derivatives
as either radical or polar acceptors to explore the scope and
limitations of these processes. The comparative behavior of
3244
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 3241 – 3247

Aminomethylation of Michael Acceptors
FULL PAPER
acceptors 16 and 18 is highly informative (Scheme 7). It is
worth noting that no adduct of a preformed organocopper
derivative was detected under conditions C starting from
Figure 5. Relative energies of the frontier orbitals of alkynes 16 and 18
and radical 4.
Scheme 7. Method A: i) 3 (5 equiv), ZnMe2 (3 equiv), R1CCR2 (1 equiv),
air, CH2Cl2, RT, 18 h; ii) NH4Cl. Method B: i) 3 (5 equiv), ZnMe2
(3 equiv), R1CCR2 (1 equiv), air, CH2Cl2, RT, 18 h; ii) CuI (2 equiv),
allylBr (9 equiv) or I2 (6 equiv); iii) NH4Cl. Method C: i) 3 (1 equiv),
ZnEt2 (1.2 equiv), CuI (1 equiv), undegassed CH2Cl2, RT, 2 h; ii) R1CCR2
(2–6 equiv), 4 h; iii) NH4Cl.
16 are 2.777 and 3.558 eV, respectively. Both interactions are
weak and provide little driving force for the addition of 4 to
16. In comparison, the energy gaps between the SOMO of
radical 4 and the HOMO and LUMO of alkyne 18 are 1.593
and 3.285 eV, respectively. This means that the SOMO–
HOMO interaction should govern the reactivity and acceler-
ate the addition of 4 to 18. It must be noted that, with
regard to this acceptor, radical 4 behaves as an electrophilic
species. Moreover, in agreement with the observed regiose-
lectivity, the atomic orbital coefficients of the carbon atoms
bearing the phenyl and carbethoxy groups are 0.47 and 0.53,
respectively, in the HOMO.
Calculations of the relative stability of radicals 22 and 23
(selected as models for the two intermediate radicals that
could be formed from radical addition to 18) were per-
formed to determine which radical adduct should be favored
by enthalpic factors (Figure 6). Radical 22 is more stable
compound 18. On the other hand, the organocopper reagent
readily added to 16 to give 17 in an 81% yield.[21] Radical 4
exhibited a reversal in behavior in this reaction. Under con-
ditions A, the phtalimidomethyl radical did not add to 16.[22]
This contrasts with the ready addition of simple alkyl radi-
cals.[4h] Conversely, its addition to 18 proceeded readily and
led to 19 in 84% yield. Only one isomer resulting from the
stereoselective alkylzinc group transfer, controlled by chela-
tion of zinc to the carbethoxy group,[4h] was detected. The
formation of a vinylzinc intermediate was confirmed by the
stereoselective allylation and iodination reactions leading to
20a and 20b under conditions B.
Proposing
a rationale for these observations is not
straightforward. The addition of radical 4 to either 16 or 18
should be exothermic. Due to the importance of orbital fac-
tors in such reactions, the energies of the frontier orbitals of
all reactants were calculated at the B3LYP/6-311+ +G-
A
ACHTUNGTRENNUNG
AHCTUNGTRENNUNG
tional frequencies were calculated at the same level of
theory to ensure that the obtained geometries were
minima). To obtain more accurate energy values, single-
point calculations were performed on the (U)B3LYP/6-31+
Figure 6. Enthalpies (a.u.) at 298 K of radicals 22 and 23 calculated by
the G3ACTHNUGRTENUNG(MP2)–RAD method.
GACHTUNGTRENNUNG(d,p) geometries at the (U)B3LYP/6-311+ +GACHUTNGTRENN(GUN 3df,3pd)
level of theory at 298 K. Natural bond orbital (NBO) calcu-
lations were used to visualize molecular orbitals.
The energy diagram shown in Figure 5 enables qualitative
comments on how orbital interactions influence both reac-
tivity and regioselectivity. The energy gaps between the
SOMO of radical 4 and the HOMO and LUMO of alkyne
than radical 23 by 29.5 kJmolÀ1. Therefore, both frontier-or-
bital interactions and enthalpic factors favor the regioselec-
tive addition of radical 4 to the sp carbon atom bearing the
carbethoxy group in 18.
The succinimidomethyl radical was shown to be slightly
less reactive than the phtalimidomethyl radical.[9b] Since the
Chem. Eur. J. 2012, 18, 3241 – 3247
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3245

L. Feray, M. P. Bertrand et al.
809–822; d) K. C. Majumdar, P. Debnath, Tetrahedron 2008, 64,
9799–9820; e) C. Ollivier, P. Renaud, Chem. Rev. 2001, 101, 3415–
3434; f) V. Darmency, P. Renaud, Top. Curr. Chem. 2006, 263, 71–
106; g) A. P. Schaffner, P. Renaud, Eur. J. Org. Chem. 2004, 2291–
2298; h) B. Quiclet-Sire, S. Zard, Chem. Eur. J. 2006, 12, 6002–6016;
i) A. Studer, Chem. Soc. Rev. 2004, 33, 267–273; j) P. A. Baguley,
J. C. Walton, Angew. Chem. 1998, 110, 3272–3283; Angew. Chem.
Int. Ed. 1998, 37, 3072–3082.
structure of maleimidomethyl iodide 24 is more closely re-
lated to that of phtalimidomethyl iodide than succinimido-
methyl iodide, the reactivity of 24 was investigated. This
radical precursor would lead to pyrrolizidine 25 when react-
ing with dimethylzinc and diethyl fumarate by following the
same pathway as phtalimidomethyl iodide. The result of this
reaction is reported in Scheme 8. It must be noted that
iodide 24 is unstable under the reaction conditions and the
excess could not be recovered. The tetrasubstituted double
bond in product 25 does not interfere as a potential radical
acceptor in the process. The iodinated pyrrolizidine 25, ob-
tained as a single isomer,[23] contains synthetic handles for
further transformations.
[2] L. Feray, M. P. Bertrand, Eur. J. Org. Chem. 2008, 3164–3170.
[3] For reviews, see: a) S. Bazin, L. Feray, M. P. Bertrand, Chimia 2006,
60, 260–265; b) T. Akindele, K.-I. Yamada, K. Tomioka, Acc. Chem.
Res. 2009, 42, 345–355.
[4] a) M. P. Bertrand, L. Feray, R. Nouguier, P. Perfetti, Synlett 1999,
1148–1150; b) M. P. Bertrand, L. Feray, R. Nouguier, P. Perfetti, J.
Org. Chem. 1999, 64, 9189–9193; c) M. P. Bertrand, S. Coantic, L.
Feray, R. Nouguier, P. Perfetti, Tetrahedron 2000, 56, 3951–3961;
d) S. Bazin, L. Feray, D. Siri, J.-V. Naubron, M. P. Bertrand, Chem.
Commun. 2002, 2506–2507; e) F. Cougnon, S. Bazin, L. Feray, M. P.
Bertrand, Tetrahedron 2007, 63, 11959–11964; f) S. Bazin, L. Feray,
N. Vanthuyne, D. Siri, M. P. Bertrand, Tetrahedron 2007, 63, 77–85;
g) J. Maury, L. Feray, S. Bazin, J.-L. Clꢀment, S. R. A. Marque, D.
Siri, M. P. Bertrand, Chem. Eur. J. 2011, 17, 1586–1595; h) J. Maury,
L. Feray, M. P. Bertrand, Org. Lett. 2011, 13, 1884–1887.
[5] For recent contributions from other groups in this field, see: a) K.-I.
Yamada, M. Nakano, M. Maekawa, T. Akindele, K. Tomioka, Org.
Lett. 2008, 10, 3805–3808; b) K.-I. Yamada, H. Umeki, M. Maeka-
wa, Y. Yamamoto, T. Akindele, M. Nakano, K. Tomioka, Tetrahe-
dron 2008, 64, 7258–7265; c) K.-I. Yamada, M. Maekawa, T. Akin-
dele, Y. Yamamoto, M. Nakano, K. Tomioka, Tetrahedron 2009, 65,
903–908; d) A. Pꢀrez-Luna, C. Botuha, F. Ferreira, F. Chemla,
Chem. Eur. J. 2008, 14, 8784–8788; e) S. Giboulot, A. Pꢀrez-Luna,
C. Botuha, F. Ferreira, F. Chemla, Tetrahedron Lett. 2008, 49, 3963–
3966; f) F. Chemla, F. Dulong, F. Ferreira, M. P. Nꢃllen, A. Pꢀrez-
Luna, Synthesis 2011, 1347–1360; g) T. Cohen, H. Gibney, R.
Ivanov, E. A.-H. Yeh, I. Marek, D. P. Curran, J. Am. Chem. Soc.
2007, 129, 15405–15409; h) Z. Chen, Y.-X. Zhang, Y. An, X.-L.
Song, Y.-H. Wang, L.-L. Zhu, L. Guo, Eur. J. Org. Chem. 2009,
5146–5152.
[6] J.-Y. Liu, Y.-J. Jang, W.-W. Lin, J.-T. Liu, C.-F. Yao, J. Org. Chem.
2003, 68, 4030–4038.
[7] J. Maury, L. Feray, P. Perfetti, M. P. Bertrand, Org. Lett. 2010, 12,
3590–3593.
[8] For triethylborane- and dimethylzinc-mediated radical additions of 1
to tosylimines derived from benzaldehyde see: Ref. [5a].
[9] a) B. Quiclet-Sire, G. Revol, S. Z. Zard, Org. Lett. 2009, 11, 3554–
3557; b) B. Quiclet-Sire, S. Z. Zard, Org. Lett. 2008, 10, 3279–3282.
[10] a) U. C. Yoon, D. U. Kim, C. W. Lee, Y. S. Choi, Y.-J. Lee, H. L.
Ammon, P. S. Mariano, J. Am. Chem. Soc. 1995, 117, 2698–2710;
b) E. F. Campbell, A. K. Park, W. A. Kinney, R. W. Fengl, L. S. Lie-
beskind, J. Org. Chem. 1995, 60, 1470–1472; c) J.-M. Receveur, J.
Guiramand, M. Rꢀcasens, M.-L. Roumestant, P. Viallefont, J. Marti-
nez, Bioorg. Med. Chem. Lett. 1998, 8, 127–132.
[11] P. Knochel, T.-S. Chou, C. Jubert, D. Rajagopal, J. Org. Chem. 1993,
58, 588–599.
[12] a) R. P. Iyer, D. Yu, N.-H. Ho, S. Agrawal, Synth. Commun. 1995,
25, 2739–2749; b) A. B. Charette, A. Beauchemin, S. Francoeur, J.
Am. Chem. Soc. 2001, 123, 8139–8140. See: Supporting Informa-
tion.
Scheme 8. The formation of functionalized pyrrolizidine 25.
Conclusion
Phtalimidomethyl iodide was shown to be a readily available
precursor for the phtalimidomethyl radical upon reaction
with dialkylzincs under tin-free aerobic conditions. Evidence
for the formation of the phtalimidomethyl radical was ob-
tained from spin-trapping experiments. The use of dialkyl-
zincs as radical initiators and chain-transfer agents enables
the stereoselective formation of functionalized pyrrolizidine
derivatives by reaction with diethyl fumarate. Carbozinca-
tion followed by transmetalation with copper(I) and radical
chemistry provide complementary simple pathways for the
formal aminomethylation of activated unsaturated com-
pounds. A rationale based on enthalpic and orbital factors is
proposed to explain the reactivity of the phtalimidomethyl
radical with activated alkynes. The results described show
again how the reactivity of dialkylzinc compounds can be
tuned depending on synthetic goals.
Acknowledgements
[13] Typical experiment: 2-Iodomethylisoindole-1,3-dione (3; 5 equiv,
876 mg, 3.05 mmol) was added to a solution of diethyl fumarate
(100 mL, 0.61 mmol) in CH2Cl2 (0.3m) at room temperature, under
an argon atmosphere. Dimethylzinc (3 equiv, 1.31 mL, 1m in hep-
tane) was then added and air was introduced in the reaction mixture
through a syringe pump (40 mL, 2 h). After 18 h at room tempera-
ture, the reaction was quenched with saturated NH4Cl. The layers
were separated and the aqueous layer was extracted twice with
CH2Cl2. The combined organic layers were dried (MgSO4), filtered,
and concentrated in vacuo. After purification by flash column chro-
matography on silica gel (pentane/EtOAc, from 100:0 to 40:60),
The authors gratefully acknowledge the ANR (D2R2; ANR-09-JCJC-
0081) for financial support, the CRCMM for computing facilities,
M. Christophe Chendo for ESI mass analyses, and Dr Luc Lewandowski
for helpful discussions.
[1] For reviews, see: a) C. Chatgilialoglu, Chem. Eur. J. 2008, 14, 2310–
2320; b) C. Chatgilialoglu, V. I. Timokhin, Adv. Organomet. Chem.
2008, 57, 117–181; c) S. Kim, S. Kim, Bull. Chem. Soc. Jpn. 2007, 80,
3246
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 3241 – 3247

Aminomethylation of Michael Acceptors
FULL PAPER
diethyl 9b-hydroxy-5-oxo-2,3,5,9b-tetrahydro-1H-pyrroloACTHNUGRTNEUNG[2,1-a]iso-
indole-1,2-dicarboxylate (5) and diethyl 5-oxo-2,5-dihydro-3H-
pyrrolo
(110 mg, 0.33 mmol) and 38% (72 mg, 0.23 mmol) yields, respective-
ly. Diethyl 9b-hydroxy-5-oxo-2,3,5,9b-tetrahydro-1H-pyrrolo[2,1-a]-
ACHTUNGTRENNUNG[2,1-a]isoindole-1,2-dicarboxylate (6) were isolated in 54
AHCTUNGTRENNUNG
isoindole-1,2-dicarboxylate (5): 1H NMR (400 MHz, CDCl3): d=
1.21 (3H, t, J=7.0 Hz), 1.35 (3H, t, J=7.0 Hz), 3.05 (1H, d, J=
10.8 Hz), 3.62 (2H, AB part of ABX), 4.10 (2H, q, J=7.3 Hz),
4.15–4.37 (3H, m), 4.42 (1H, brs), 7.43 (1H, t, J=7.5 Hz), 7.49
(1H, d, J=7.3 Hz), 7.56 (1H, t, J=7.3 Hz), 7.75 ppm (1H, d, J=
7.8 Hz); 13C NMR (75 MHz, CDCl3): d=14.1 (CH3), 14.2 (CH3),
42.6 (CH2), 48.2 (CH), 53.2 (CH), 61.5 (CH2), 61.6 (CH2), 95.4 (C),
123.5 (=CH), 124.1 (=CH), 130.1 (=CH), 131.1 (=C), 133.0 (=CH),
145.6 (=C), 168.5 (C=O), 169.3 (C=O), 171.9 ppm (C=O); HRMS
(ESI): m/z calcd for C17H19NO6: 334.1285 [M+H]+; found: 334.1288.
[17] A. B. Charette, A. Beauchemin, J.-F. Marcoux, J. Am. Chem. Soc.
1998, 120, 5114–5115.
[18] The absence of compound 12 in Table 2, entry 2 is probably a result
of the slow rate of homolytic substitution at dimethylzinc.
[19] This transmetalation was assumed to involve the formation of radi-
cals, see: P. Knochel, Synlett 1995, 393–403. This was indirectly con-
firmed by ESR experiments performed in the presence of 3, diethyl-
zinc, CuI, and TTBNB. The same spectrum as in Figure 1 was ob-
served for these reactions.
[20] In all cases, oxidation products 11 and 12, formed through radical
pathways, are only detected in trace amounts. Under pure radical
conditions, no radical allylation could be observed with allylbro-
mide.
Diethyl 5-oxo-2,5-dihydro-3H-pyrroloACTHNUGTRNEUNG[2,1-a]isoindole-1,2-dicarboxy-
late (6): 1H NMR (400 MHz, CDCl3): d=1.25 (3H, t, J=7.0 Hz),
1.33 (3H, t, J=7.0 Hz), 4.02 (1H, dd, J=11.8, 5.3 Hz), 4.15–4.24
(2H, AB part of ABX3), 4.17 (1H, superimposed pseudo t, J=
11.0 Hz), 4.24–4.36 (2H, AB part of ABX3), 4.42 (1H, dd, J=10.8,
5.3 Hz), 7.53–7.65 (2H, m), 7.74–7.84 (1H, d, J=6.8 Hz), 8.51 ppm
(1H, d, J=6.8 Hz); 13C NMR (75 MHz, CDCl3): d=14.2 (CH3), 14.4
(CH3), 44.3 (CH2), 51.7 (CH), 60.9 (CH2), 61.7 (CH2), 108.5 (=C),
123.5 (=CH), 127.1 (=CH), 129.5 (=C), 131.9 (=CH), 132.3 (=CH),
135.9 (=C), 150.8 (=C), 163.4 (C=O), 163.8 (C=O), 171.8 ppm
(C=O); HRMS (ESI): m/z calcd for C17H18NO5: 316.1179 [M+H]+;
found: 316.1178.
[21] The product 21, resulting from two subsequent Michael additions,
was isolated in a 15% yield, see the Supporting Information.
[14] The CH bond dissociation energy in N-methylphtalimide calculated
by use of the G3ACHTUNGTRENNUNG
(MP2)-RAD method is 95 kcalmolÀ1. Radical 4 is
stabilized by 7.8 kcalmolÀ1 with respect to the methyl radical.
[15] Chelates have already been proposed to explain the diasteroselectiv-
ity of diethylzinc-mediated radical polar cascade reactions leading
to trisubstituted lactones, see: a) S. Bazin, L. Feray, N. Vanthuyne,
M. P. Bertrand, Tetrahedron 2005, 61, 4261–4274; b) Ref. [7]; for
their involvement in the syn-selective aldol condensation of zinc
enolates, see also: c) S. Lai, C. K. Zercher, J. P. Jasinski, S. N. Reid,
R. J. Staples, Org. Lett. 2001, 3, 4169–4171.
[16] We assume the mechanism below to explain this result. Trace
amounts of nucleophilic substitution of iodide 3 might lead to an
N-iodo alkoxyamine that is a potential precursor for spontaneous
formation of anilino radical AP.
[22] Pivaloylmethyl and a-THF radicals also failed to add to 16.
[23] Stereochemistry was assigned by using the similarity between the
coupling constants observed in the proton NMR spectra of both
products 5 (J1-2 =10.8) and 25 (J1-2 =11.2).
Received: July 29, 2011
Revised: November 14, 2011
Published online: February 10, 2012
Chem. Eur. J. 2012, 18, 3241 – 3247
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
3247