Opposite regiospecific ring opening of 2-(cyanomethyl)aziridines by hydrogen bromide and benzyl bromide: Experimental study and theoretical rationalization

Article abstract of DOI:10.1021/jo100687q

(Figure presented) Ring opening of 1-arylmethyl-2-(cyanomethyl)aziridines with HBr afforded 3-(arylmethyl)amino-4-bromobutyronitriles via regiospecific ring opening at the unsubstituted aziridine carbon. Previous experimental and theoretical reports show treatment of the same compounds with benzyl bromide to furnish 4-amino-3-bromobutanenitriles through ring opening at the substituted aziridine carbon. To gain insights into the regioselective preference with HBr, reaction paths have been analyzed with computational methods. The effect of solvation was taken into account by the use of explicit solvent molecules. Geometries were determined at the B3LYP/6-31++G(d,p) level of theory, and a Grimme-type correction term was included for long-range dispersion interactions; relative energies were refined with the meta-hybrid MPW1B95 functional. Activation barriers confirm preference for ring opening at the unsubstituted ring carbon for HBr. HBr versus benzyl bromide ring opening was analyzed through comparison of the electronic structure of corresponding aziridinium intermediates. Although the electrostatic picture fails to explain the opposite regiospecific nature of the reaction, frontier molecular orbital analysis of LUMOs and nucleophilic Fukui functions show a clear preference of attack for the substituted aziridine carbon in the benzyl bromide case and for the unsubstituted aziridine carbon in the HBr case, successfully rationalizing the experimentally observed regioselectivity.

Full text of DOI:10.1021/jo100687q

pubs.acs.org/joc
Opposite Regiospecific Ring Opening of 2-(Cyanomethyl)aziridines by
Hydrogen Bromide and Benzyl Bromide: Experimental Study and
Theoretical Rationalization
Saron Catak,†,§ Matthias D’hooghe,‡ Toon Verstraelen,†,§ Karen Hemelsoet,†,§
Andries Van Nieuwenhove,‡ Hyun-Joon Ha, Michel Waroquier,†,§ Norbert De Kimpe,*,‡
and Veronique Van Speybroeck*,†,§
†Center for Molecular Modeling, Ghent University, Technologiepark 903, B-9052 Zwijnaarde, Belgium,
‡Department of Organic Chemistry, Faculty of Bioscience Engineering, Ghent University, Coupure Links
653, B-9000 Ghent, Belgium, §QCMM-Alliance, Ghent-Brussels, Belgium, and Department of Chemistry
and Protein Research Center for Bio-Industry, Hankuk University of Foreign Studies, Yongin 449-791, Korea
veronique.vanspeybroeck@ugent.be; norbert.dekimpe@UGent.be
Received April 14, 2010
Ring opening of 1-arylmethyl-2-(cyanomethyl)aziridines with HBr afforded 3-(arylmethyl)amino-4-
bromobutyronitriles via regiospecific ring opening at the unsubstituted aziridine carbon. Previous
experimental and theoretical reports show treatment of the same compounds with benzyl bromide to
furnish 4-amino-3-bromobutanenitriles through ring opening at the substituted aziridine carbon. To
gain insights into the regioselective preference with HBr, reaction paths have been analyzed with
computational methods. The effect of solvation was taken into account by the use of explicit solvent
molecules. Geometries were determined at the B3LYP/6-31þþG(d,p) level of theory, and a Grimme-
type correction term was included for long-range dispersion interactions; relative energies were
refined with the meta-hybrid MPW1B95 functional. Activation barriers confirm preference for ring
opening at the unsubstituted ring carbon for HBr. HBr versus benzyl bromide ring opening was
analyzed through comparison of the electronic structure of corresponding aziridinium intermediates.
Although the electrostatic picture fails to explain the opposite regiospecific nature of the reaction,
frontier molecular orbital analysis of LUMOs and nucleophilic Fukui functions show a clear
preference of attack for the substituted aziridine carbon in the benzyl bromide case and for the
unsubstituted aziridine carbon in the HBr case, successfully rationalizing the experimentally
observed regioselectivity.
4530 J. Org. Chem. 2010, 75, 4530–4541
Published on Web 05/27/2010
DOI: 10.1021/jo100687q
2010 American Chemical Society
r

Catak et al.
JOCArticle
Introduction
SCHEME 1. Ring Opening of Aziridinium Salts
The aziridine moiety represents one of the most valuable
three-membered ring systems in synthetic organic chemistry,1-11
and regio-controlled ring-opening reactions of C-substituted
aziridines constitute useful tools in organic synthesis for the
preparation of a large variety of functionalized nitrogen-contain-
ing target compounds.
Ring opening of activated aziridines, i.e., aziridines bearing an
electron-withdrawing group on the nitrogen, has been studied
intensively in the literature.4 Nonactivated aziridines, however,
have to be activated prior to ring opening because of the
presence of an electron-donating substituent on the nitrogen
and have been evaluated to a limited extent up to now. Never-
theless, the reactivity and applications of nonactivated aziridines
are different and often complementary as compared to activated
aziridines and epoxides, providing interesting opportunities for
the selective synthesis of a variety of valuable amines. The most
common approach for the activation of nonactivated aziridines
involves the formation of highly electrophilic aziridinium inter-
mediates through N-alkylation or complexation with a Lewis
acid, which then can easily be opened by different types of
nucleophiles. In that respect, the ring opening of aziridinium
salts by halides constitutes a convenient approach toward
β-halo amines, which are useful building blocks in organic
chemistry.12-15 If 2-substituted aziridines are used for the
synthesis of the corresponding β-halo amines, the issue of
regioselectivity in the ring opening of the intermediate aziridi-
nium salts becomes important. As depicted in Scheme 1, ring
opening can occur at the unhindered (path a) or the hindered
aziridine carbon atom (path b), leading either to primary
halides (path a) or to secondary halides (path b).
2-(cyanomethyl)aziridines with acid chlorides, affording
mainly N-(2-chloro-3-cyanopropyl)amides through ring open-
ing at the substituted position.19 However, also the opposite
regioselectivity has been described sporadically, e.g., upon reac-
tion of a methyl aziridine-2-carboxylate with acetyl chloride.20
Furthermore, the reaction of alkyl aziridine-2-carboxylates
with hydrogen halide (hydrogen chloride21 or hydrogen bro-
mide22) has been evaluated, resulting in ring opening at the more
hindered carbon atom. If 2-(trifluoromethyl)aziridines are used
instead, N-(1-halomethyl-2,2,2-trifluoroethyl)amines are obtained
via ring opening at the unsubstituted aziridine carbon atom.23,24
A different class of substrates, i.e., 2-(bromomethyl)-,25 2-(aryl-
oxymethyl)-,26 2-(alkanoyloxymethyl)-,27 2-(cyanomethyl)-,28
and 2-(cyanoethyl)aziridines,29,30 has been used extensively in
ring-opening reactions with arylmethyl bromides in acetonitrile.
In all cases, the intermediate aziridinium salts were opened
regiospecifically at the substituted carbon atom, and these experi-
mental results have been rationalized on the basis of DFT-based
calculations.28,31-33 In analogy, the ring opening of 2-alkyl-sub-
stituted aziridinium salts by chloride at the more hindered position
has been reported recently, thus affording the thermodynamic
products.34 Alternative regiochemistry in the ring opening of
aziridines with HBr, which was later rectified by the present
authors,35 have appeared in literature.
In this respect, the ring opening of 2-acyl16,17 and 2-arylaziri-
dinium salts,18 obtained through reaction of the starting azir-
idines with acid chlorides, has been studied in the literature,
pointing to a preferential ring opening at the substituted aziri-
dine carbon atom. This effect can be explained considering
the high electrophilicity of the substituted aziridine carbon
atom in 2-acyl- and 2-arylaziridines. A less pronounced yet
similar regioselectivity was observed in the ring opening of
The use of aziridinium ions for regio- and stereoselec-
tive ring-opening reactions remains a challenging topic in
(19) D’hooghe, M.; Vervisch, K.; Van Nieuwenhove, A.; De Kimpe, N.
Tetrahedron Lett. 2007, 48, 1771–1774.
(20) Lee, K.-D.; Suh, J.-M.; Park, J.-H.; Ha, H.-J.; Choi, H. G.; Park,
C. S.; Chang, J. W.; Lee, W. K.; Dong, Y.; Yun, H. Tetrahedron 2001, 57,
8267–8276.
€
(1) Lindstrom, U. M.; Somfai, P. Synthesis 1998, 1998, 109–117.
(21) Gnecco, D.; Orea F., L.; Galindo, A.; Enriquez, R. G.; Toscano,
R. A.; Reynolds, W. F. Molecules 2000, 5, 998.
(22) Sulser, U.; Widmer, J.; Goeth, H. Helv. Chim. Acta 1977, 60, 1676.
(23) Katagiri, T.; Takahashi, M.; Fujiwara, Y.; Ihara, H.; Uneyama, K.
J. Org. Chem. 1999, 64, 7323–7329.
(2) Zwanenburg, B.; ten Holte, P. Top. Curr. Chem. 2001, 216, 93–124.
(3) Sweeney, J. B. Chem. Soc. Rev. 2002, 31, 247–258.
(4) Hu, X. E. Tetrahedron 2004, 60, 2701–2743.
(5) Tanner, D. Angew. Chem., Int. Ed. 1994, 33, 599–619.
(6) Osborn, H. M. I.; Sweeney, J. Tetrahedron: Asymmetry 1997, 8, 1693–
(24) Karimova, N.; Teplenicheva, Y.; Kolomiets, A.; Fokin, A. Russ.
Chem. Bull. 1997, 46, 1136–1139.
1715.
(7) McCoull, W. M.; Davis, F. A. Synthesis 2000, 1347–1365.
(8) Watson, I. D. G.; Yu, L.; Yudin, A. K. Acc. Chem. Res. 2006, 39, 194–
(25) D’hooghe, M.; Van Brabandt, W.; De Kimpe, N. J. Org. Chem. 2004,
69, 2703–2710.
206.
(26) D’hooghe, M.; Waterinckx, A.; Vanlangendonck, T.; De Kimpe, N.
Tetrahedron 2006, 62, 2295.
(9) Fantauzzi, S.; Gallo, E.; Caselli, A.; Piangiolino, C.; Ragaini, F.; Re,
N.; Cenini, S. Chem.;Eur. J. 2009, 15, 1241–1251.
(10) Tsang, Derek S.; Yang, S.; Alphonse, F.-A.; Yudin, A. K. Chem.;
Eur. J. 2008, 14, 886–894.
(11) Singh, G. S.; D’hooghe, M.; De Kimpe, N. Chem. Rev. 2007, 107,
2080–2135.
(27) D’hooghe, M.; De Kimpe, N. Synlett 2006, 2089.
(28) D’hooghe, M.; Van Speybroeck, V.; Van Nieuwenhove, A.;
Waroquier, M.; De Kimpe, N. J. Org. Chem. 2007, 72, 4733–4740.
(29) D’hooghe, M.; Vervisch, K.; De Kimpe, N. J. Org. Chem. 2007, 72,
7329–7332.
€
(30) Vervisch, K.; D’hooghe, M.; Tornroos, K. W.; De Kimpe, N.
(12) Fang, Y.-Q.; Karisch, R.; Lautens, M. J. Org. Chem. 2007, 72, 1341–
1346.
(13) Couty, F.; David, O.; Larmanjat, B.; Marrot, J. J. Org. Chem. 2007,
72, 1058–1061.
Org. Biomol. Chem. 2009, 7, 3271–3279.
(31) D’hooghe, M.; Van Speybroeck, V.; Waroquier, M.; De Kimpe, N.
Chem. Commun. 2006, 1554–1556.
(14) May, B. C. H.; Zorn, J. A.; Witkop, J.; Sherrill, J.; Wallace, A. C.;
Legname, G.; Prusiner, S. B.; Cohen, F. E. J. Med. Chem. 2006, 50, 65–73.
(15) D’hooghe, M.; De Kimpe, N. Tetrahedron 2006, 62, 513–535.
(16) Kim, Y.; Ha, H.-J.; Yun, H.; Lee, B. K.; Lee, W. K. Tetrahedron
2006, 62, 8844–8849.
(32) Yun, S. Y.; Catak, S.; Lee, W. K.; D’hooghe, M.; De Kimpe, N.;
Van Speybroeck, V.; Waroquier, M.; Kim, Y.; Ha, H.-J. Chem. Commun.
2009, 2508–2510.
(33) Catak, S.; D’hooghe, M.; De Kimpe, N.; Waroquier, M.; Van
Speybroeck, V. J. Org. Chem. 2010, 75, 885–896.
(34) Sivaprakasham, M.; Couty, F.; Evano, G.; Srinivas, B.; Sridhar, R.;
Rao, K. R. Arkivoc 2007, x, 71–93.
(17) Sim, T. B.; Kang, S. H.; Lee, K. S.; Lee, W. K.; Yun, H.; Dong, Y.;
Ha, H.-J. J. Org. Chem. 2002, 68, 104–108.
(18) Savoia, D.; Alvaro, G.; Di Fabio, R.; Gualandi, A.; Fiorelli, C.
J. Org. Chem. 2006, 71, 9373–9381.
(35) D’hooghe, M.; Hofkens, A.; De Kimpe, N. Tetrahedron Lett. 2003,
44, 1137–1139.
J. Org. Chem. Vol. 75, No. 13, 2010 4531

Catak et al.
JOCArticle
SCHEME 2. Ring Opening of 1-Arylmethyl-2-(cyanomethyl)aziridines by HBr and Benzyl Bromide
organic chemistry, and a thorough insight into the role of the
different factors can result in a common and general use of azi-
ridinium salts as versatile substrates in organic synthesis. In
recent years, many efforts have been devoted to the develop-
ment of new methods for the biocatalytic conversion of amino-
nitriles into the corresponding amino acids,36-40 and as a
consequence, the search for novel types of functionalized amino-
nitrile derivatives has gained much interest and has become
an important challenge in organic synthesis.41,42 In that res-
pect, the use of 2-(cyanomethyl)aziridines has been studied to
a limited extent and might provide a convenient entry into a
variety of novel aminonitrile derivatives.
In the present paper, the ring opening of 1-arylmethyl-2-
(cyanomethyl)aziridines with HBr in acetonitrile or dichloro-
methane is described, affording 3-(arylmethyl)amino-4-bromo-
butyronitriles via regiospecific ring opening at the unsubstituted
carbon atom. It is clear that the regioselectivity associated with the
ring opening of C-substituted aziridinium salts is a complex issue,
in which the nature of the substrate, the nucleophile, and the sol-
vent can be of importance. Theoretical methods used in this study
are ideally suited to discriminate between the various effects
mentioned, and when combined with experimental data, this
approach can lead to an in-depth understanding of the factors
determining the reaction outcome. Moreover, understanding the
underlying factors for the regioselective preference is of high
interest, since this knowledge may eventually be useful in the selec-
tivesynthesisofrelevantnitrogen-containingtarget compounds.
1-arylmethyl-2-(bromomethyl)aziridines43-46 upon treat-
ment with 1 equiv of potassium cyanide in DMSO and heat-
ing at 60-70 °C for 3 h.47 The search for new pathways
toward aminonitriles as precursors of the corresponding
amino acids is an important challenge in organic synthesis.
The combination of an aziridine moiety and a cyano group in
aziridines 1 enables the preparation of a variety of functio-
nalized aminonitriles through ring-opening reactions of the
constrained ring.
As reported before, treatment of 1-arylmethyl-2-(cyano-
methyl)aziridines 1 with benzyl bromide in acetonitrile afforded
4-amino-3-bromobutanenitriles 2 through a regiospecific ring
opening of the intermediate 2-(cyanomethyl)aziridinium salts
4 by bromide at the more hindered aziridine carbon atom
(Scheme 2), which was further validated by means of DFT
calculations.28,31 Long reaction times at room temperature
were required in order to avoid subsequent dehydrohalo-
genation at elevated temperatures. This method offers a
convenient approach toward β-substituted γ-aminonitriles
upon nucleophilic displacement of the bromo atom in
β-bromonitriles 2.
Surprisingly, treatment of the same substrates 1 with 1.2
equiv of hydrogen bromide (as a 33% solution in acetic acid) in
acetonitrile or dichloromethane as a solvent for one hour at
room temperature resulted exclusively in 3-N-(arylmethyl)-
amino-4-bromobutanenitriles 3 in good yields (Scheme 2).
Due to the high intrinsic reactivity of β-bromoamines 3, these
compounds are not suited for long time preservation, and
should therefore be used for further elaboration shortly after
preparation. These observations can be explained through the
formation of intermediate aziridinium ions 5 upon N-proton-
ation of the aziridine ring, followed by regiospecific ring
Results and Discussion
Experimental Results. 1-Arylmethyl-2-(cyanomethyl)azi-
ridines 1 can be prepared easily and in high yields from
€ €
(36) Fitz, M.; Lundell, K.; Lindroos, M.; Fulop, F.; Kanerva, L. T.
Tetrahedron: Asymmetry 2005, 16, 3690.
(37) Winkler, M.; Martinkova, L.; Knall, A. C.; Krahulec, S.; Klempier,
N. Tetrahedron 2005, 61, 4249.
(38) Preiml,M.;Hillmayer, K.;Klempier, N.Tetrahedron Lett. 2003, 44, 5057.
(39) Preiml, M.; Honig, H.; Klempier, N. J. Mol. Catal. B: Enzym. 2004,
29, 115.
(43) De Smaele, D.; Bogaert, P.; De Kimpe, N. J. Chem. Soc., Chem.
Commun. 1994, 1221–1222.
(44) De Kimpe, N.; De Smaele, D.; Sakonyi, Z. J. Org. Chem. 1997, 62,
2448–2452.
(45) D’hooghe, M.; Waterinckx, A.; De Kimpe, N. J. Org. Chem. 2004,
70, 227–232.
(40) Sugai, T.; Yamazaki, T.; Yokoyama, M.; Ohta, H. Biosci., Biotech-
nol., Biochem. 1997, 61, 1419.
(46) D’hooghe, M.; Rottiers, M.; Jolie, R.; De Kimpe, N. Synlett 2005,
931.
(41) Groger, H. Chem. Rev. 2003, 103, 2795–2828.
(42) Enders, D.; Shilvock, J. P. Chem. Soc. Rev. 2000, 29, 359.
(47) D’hooghe, M.; Mangelinckx, S.; Persyn, E.; Van Brabandt, W.;
De Kimpe, N. J. Org. Chem. 2006, 71, 4232–4236.
4532 J. Org. Chem. Vol. 75, No. 13, 2010

Catak et al.
JOCArticle
opening by bromide at the unsubstituted aziridine carbon atom
(Scheme 2). Obviously, the structural difference between 1,1-
dialkylaziridinium salts 4 and 1-protio-1-alkylaziridinium salts
5 has a profound influence on the reactivity toward bromide. It
is likely that the presence of a single benzyl substituent on the
aziridinium nitrogen in intermediates 5 changes the charge
distribution and the electrostatic potential in the aziridine ring
compared to the 1,1-dialkylaziridinium ions 4; however, it is
also likely that there is a significant difference in the frontier
orbital picture; these effects will be further investigated with
computational techniques.
In summary, 1-arylmethyl-2-(cyanomethyl)aziridines 1
can be easily transformed into γ-amino-β-bromobutaneni-
triles 2 or β-amino-γ-bromobutanenitriles 3, depending on
the reagent used. The complementarities of both approaches,
i.e., treatment with benzyl bromide in CH3CN or hydrogen
bromide in CH3CN or CH2Cl2, consolidates the versatility of
2-(cyanomethyl)aziridines 1 toward the synthesis of different
types of aminonitriles.
Computational Results. The current study aims to eluci-
date the regiospecific preference in the ring opening of
1-arylmethyl-2-(cyanomethyl)aziridines 1 by HBr and com-
paratively analyze the opposite regioselectivity observed
with benzyl bromide. It is difficult to directly compare these
two reaction mechanisms, since solvation is different; how-
ever, the nature of the aziridinium intermediates will be
studied in detail in terms of electronic structure.
The initial step of the reaction is proposed to be the
formation of aziridinium ion 5, which then undergoes
attack by the bromide ion at the unsubstituted (less hindered)
aziridine carbon (Scheme 2), resulting in 3-N-(arylmethyl)-
amino-4-bromobutanenitriles 3, exclusively. This is in con-
trast with results attained from ring opening by benzyl bro-
mide, which yields 4-amino-3-bromobutanenitriles2,28,31 sug-
gesting bromide attack at the substituted (more hindered)
aziridine carbon. The latter case has been previously theore-
tically investigated31 and will serve as a basis in the effort to
understand the difference in regioselectivity observed in these
two reactions.
SCHEME 3. Atom Labeling and Schematic Representation of
1-trans and 1-cis
All DFT calculations were carried out with the Gaussian 03
program package.50 Stationary points were fully optimized with
the B3LYP/6-31þþG(d,p) level of theory,51,52 which is known
to produce reliable geometries for aziridine compounds, and
characterized as minima or first-order saddle points via fre-
quency calculations. Intrinsic reaction coordinate (IRC)53 cal-
culations followed by full geometry optimizations were used to
verify reactant complexes (ion-dipole complexes) and products
reached by each transition state. Free energies of activation
(ΔGq) were calculated as the difference of free energies between
transition states and reactive conformers. Relative energies were
further refined using the meta-hybrid GGA MPW1B95.48 Fur-
thermore, a computationally feasible method, which consists
of adding an empirical -C6R-6 correction on DFT energies to
account for dispersive interactions was adopted. The so-called
DFT-D approach, shown to provide high accuracy in a variety
of simulations,54 was implemented utilizing the ORCA 2.6.35
software package.55
Invertomers for 1-Arylmethyl-2-(cyanomethyl)aziridines.
A conformational search taking into account the most pertinent
degrees of freedom was performed on both invertomers, 1-trans
and 1-cis (Scheme 3), of 1-arylmethyl-2-(cyanomethyl)aziridines
1 by means of a systematic relaxed-potential energy surface
(PES) scan at the B3LYP/6-31þþG(d,p) level of theory, in or-
der to identify the most stable conformer for each isomer. Simul-
taneous rotation around dihedrals CN-C4-C2-N and C2-
N-C5-Ph has revealed six significant conformers for the cis and
nine for the trans invertomer.
Relative free energies at different levels of theory (Table 1)
show a range of approximately 5 kJ/mol energy difference
among conformers for each invertomer at all levels of theory,
excluding conformers 1-trans-g, 1-trans-h, and1-trans-i in which
the aromatic group is directed toward the aziridine ring. Calcu-
lations show a consistent difference in stability between 1-trans
and 1-cis, favoring the trans invertomer (17.4, 12.7, and 14.7 kJ/
mol for B3LYP, B3LYP-D, and MPW1B95, respectively) as
seen in the difference between the most stable conformers of
each invertomer, namely, 1-trans-d and 1-cis-d.
Computational Methodology. An in-depth level of theory
study on the ring opening of aziridines with benzyl bromide was
recently performed.33 It was shown that the MPW1B9548 level
of theory successfully reproduced SCS-MP249 results and ade-
quately described the reaction barriers and product stabilities,
provided solvation effects were taken into account via effec-
tively solvating the bromide with explicit solvent molecules. The
same approach is adopted in the present study.
(48) Zhao, Y.; Truhlar, D. G. J. Phys. Chem. A 2004, 108, 6908–6918.
(49) Gerenkamp, M.; Grimme, S. Chem. Phys. Lett. 2004, 392, 229–235.
(50) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Montgomery, Jr., J. A.; Vreven, T.; Kudin, K. N.;
Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci,
B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada,
M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima,
T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.;
Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.;
Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.;
Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.;
Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels,
A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.;Raghavachari,
K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski,
J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin,
R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.;
Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.;
Gonzalez, C.; Pople, J. A. Gaussian, Rev. C.02; Gaussian, Inc.: Wallingford,
CT, 2004.
Pyramidal inversion at the nitrogen center of amines is known
to occur at a fast rate, preventing a permanent chiral center on
the nitrogen atom and hence not allowing the isolation of inver-
tomers (inversion isomers). However, there are literature studies
in which invertomers have been isolated by slowing down the
inversion rate through the use of large central atoms (P and As)
or by the inclusion of a heteroatom in a small ring, as seen in
certain aziridines.56 Increased angle strain in the aziridine ring is
(51) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785–789.
(52) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652.
(53) Fukui, K. Acc. Chem. Res. 2002, 14, 363–368.
(54) Grimme, S. J. Comput. Chem. 2004, 25, 1463–1473.
(55) ORCA, 2.6.35 ed.; http://www.thch.uni-bonn.de/tc/orca/.
(56) Padwa, A.; Battisti, A. J. Org. Chem. 2002, 36, 230–231.
J. Org. Chem. Vol. 75, No. 13, 2010 4533

Catak et al.
JOCArticle
TABLE 1. Relative Free Energies (kJ/mol) for Stable Conformers of Aziridine Invertomers 1-trans and 1-cisa-c
aFree energies listed in brackets are relative to 1-trans-d. bB3LYP/6-31þþG(d,p) geometries. cThermal free energy corrections from B3LYP/
6-31þþG(d,p) calculations at 1 atm and 298 K.
TABLE 2. Free Energies of Activation (ΔGq) for Nitrogen Inversion, Free
Energy Change for cis-trans Isomerization (ΔG°), and Invertomer Ratiosa,b
invertomer
ratio
ΔG°trans-cis (trans:cis)
ΔGq
ΔGq
transfcis
cisftrans
B3LYP/
6-31þþG(d,p)
67.8
71.5
66.7
50.5
58.8
52.0
17.4
12.7
14.7
10000:8
B3LYP-D/
10000:63
10000:27
6-31þþG(d,p)
MPW1B95/
6-31þþG(d,p)
aThermal free energy corrections from B3LYP/6-31þþG(d,p) calcu-
lations at 1 atm and 298 K. bEnergies in kJ/mol.
Calculated barriers are in line with earlier reports on experi-
mental and theoretical aziridine inversion barriers,57-59 which
were shown to range from 50 to 90 kJ/mol (79 kJ/mol for the
parent, unsubstituted aziridine) depending on ring substituents,
in particular on the identity of the N-substituent, where elec-
tron-withdrawing groups were shown to destabilize the transi-
tion state by inductive effects and therefore increase the barrier
to inversion. It is generally expected that resolution of a single
invertomer by slow inversion is possible if the barrier to nitrogen
inversion is higher than 100 kJ/mol (at room temperature).
However, in the present study, free energy change (ΔG°) for
trans-cis isomerization is significantly high (14.7 kJ/mol). Consi-
dering the relatively low barrier, rapid inversion at room tem-
perature is expected to lead to thermodynamic equilibration;
it is highly likely that the trans invertomer (1-trans) will be
overpopulated. This is in line with previous experimental
findings, where the trans invertomer was shown to be pre-
ferred in the crystalline structure.60 Table 2 shows forward
FIGURE 1. Free energy profile (kJ/mol, MPW1B95/6-31þþG(d,p))
for nitrogen inversion in 1-arylmethyl-2-(cyanomethyl)aziridines.
known to lead to an increased barrier for nitrogen inversion, in
some cases high enough for the isolation of separate inverto-
mers. For this reason, pyramidal inversion at the ring nitrogen
of 1-arylmethyl-2-(cyanomethyl)aziridines 1 was computation-
ally investigated, in order to verify the possibility of invertomer
resolution (Scheme 3).
The free energy barrier (ΔGq) for nitrogen inversion at
1-arylmethyl-2-(cyanomethyl)aziridines 1 and the stan-
dard free energy change (ΔG°) for isomerization of inver-
tomers 1-trans and 1-cis are shown in Figure 1. The trigo-
nal pyramidal nature of the nitrogen center is depicted in
the inversion transition state (TS-inversion). The free
energy barrier for inversion (MPW1B95) from trans to
cis (ΔGqtransfcis) is shown to be 66.7 kJ/mol, whereas the
(57) Alvernhe, G. M.; Ennakoua, C. M.; Lacombe, S. M.; Laurent, A. J.
J. Org. Chem. 2002, 46, 4938–4948.
(58) Boggs, G. R.; Gerig, J. T. J. Org. Chem. 2002, 34, 1484–1486.
(59) Andose, J. D.; Lehn, J. M.; Mislow, K.; Wagner, J. J. Am. Chem. Soc.
2002, 92, 4050–4056.
reverse barrier for inversion from cis to trans (ΔGq
)
cisftrans
(60) Lee, K.-D.; Suh, J.-M.; Park, J.-H.; Ha, H.-J.; Choi, W. G.; Park,
C. S.; Lee, W. K.; Dong, Y.; Yun, H. Tetrahedron 2001, 57, 8267–8276.
is 52.0 kJ/mol.
4534 J. Org. Chem. Vol. 75, No. 13, 2010

Catak et al.
JOCArticle
strong hydrogen bonds with charged species; therefore, the
bromide ion is not expected to be so tightly bound to the
nitrogen proton in solution. In order to get more insight into
the effect of solvation, discrete solvent molecules were added
to the model as described in the next section.
Effect of Solvation: A Discrete Solvent Model. Nucleo-
philic substitution reactions are known to be influenced by
the solvent environment.62 Previous theoretical studies on
regioselective ring opening of aziridinium ions with bromide
have shown that the role of the solvent cannot be underesti-
mated.28,31,33 Gas-phase calculations, in which a bare halide ion
attacks the electrophilic aziridinium ion, are unrealistic and
incapable of representing the real system at hand. Accordingly,
previous studies on modeling the bromide-induced ring opening
of aziridinium ions have effectively made use of explicit solvent
molecules.28,31,33
Simulations of reactions in organic solvents are considered
one of the most challenging tasks in the field of molecular mode-
ling. In many theoretical studies, the solvent is either neglected or
simulated by means of continuum solvation models.63-65 In
such a dielectric model, the solvent is modeled as a continuous
medium, usually assumed homogeneous and isotropic, charac-
terized solely by a scalar, static dielectric constant. In some cases,
these results give the correct reactive behavior, provided there
are no essential explicit solvent interactions with the solute.
However, in many cases, specific solvent interactions are in play
in the first solvation shell. Ideally, the reactive species could be
simulated by means of molecular dynamics calculations in a
solvent box, and the number of solvent molecules in the first
solvation shell can be considered to determine the coordination
number at the site of interest. However, this approach is com-
putationally very expensive and is not routinely applied. An-
other alternative is to include explicit solvent interactions by
placing discrete solvent molecules around the chemically active
species.28,31,33,66-70 The number of explicit solvent molecules
that should be incorporated in this “discrete solvent model” is
often not clear.71 This number is generally determined by the
value at which the coordination solvation energy converges.
This methodology was followed by the present authors in a
recent study, which successfully pointed to the correct regio-
selective outcome.28,33 Additionally, the supermolecule can be
placed in a continuum with a fixed dielectric constant to account
for bulk solvation; this is referred to as a mixed implicit/explicit
solvent model.71-73 A comparative study of continuum,
FIGURE 2. Intramolecular H-bonding in aziridinium ion 5-trans-f.
(B3LYP/6-31þþG(d,p)).
and reverse inversion barriers and invertomer stabilities at
different levels of theory. Calculated invertomer ratios show a
clear preference for the trans invertomer, 10000:27 (trans:cis) at
the MPW1B95/6-31þþG(d,p) level of theory, and therefore
the cis invertomeris not expectedtohave a significant presence
in the reaction mixture.
Nucleophilic Ring Opening of 1-Arylmethyl-2-(cyanomethyl)-
aziridines with HBr. Despite aziridine’s inherent reacti-
vity due to the ring strain energy (111.7 kJ/mol),61 non-
activated aziridines are generally not prone to nucleophilic
ring-opening reactions, without prior activation, which can
be achieved by protonation, N-substitution with electron-
withdrawing groups, or quaternization. The increased s char-
acter of the nitrogen center renders aziridines less basic than
acyclic aliphatic amines; however, they are still expected to be
easily protonated to yield intermediary aziridinium ions and
hence become activated in acidic medium. The intermediate
aziridinium ions 5 that form upon protonation of the parent
aziridines, 1-arylmethyl-2-(cyanomethyl)aziridines 1, are much
less stable and therefore more susceptible to ring opening than
their unactivated aziridine counterparts.
Aziridinium conformers 5 corresponding to the parent azir-
idine 1 were generated for all structures listed in Table 1; trans
conformers were energetically more favorable at all levels of
theory. Henceforth, only the trans invertomers will be analyzed
in terms of relative stabilities and specific interactions in the
corresponding aziridinium ion. Once protonated, the aziridini-
um moiety is further capable of forming intramolecular H-bond-
ing interactions as depicted in 5-trans-f (Figure 2), causing a
strong preference for certain trans conformers. In the presence of
a polar protic solvent (acetic acid), intramolecular interactions
are likely to be less significant, and therefore the extra stabiliza-
tion observed in conformers with intramolecular H-bonds is
most likely an artifact of gas-phase calculations. Intramolecular
H-bonds do not occur in the cis case.
(62) Vayner, G.; Houk, K. N.; Jorgensen, W. L.; Brauman, J. I. J. Am.
Chem. Soc. 2004, 126, 9054–9058.
(63) Cramer, C. J.; Truhlar, D. G. In Solvent Effects and Chemical
Reactivity; Kluwer: Dordrecht, 1996; p 1-80.
(64) Tomasi, J.; Mennucci, B.; Cammi, R. Chem. Rev. 2005, 105, 2999–
3094.
(65) Takano, Y.; Houk, K. N. J. Chem. Theory Comput. 2004, 1, 70–77.
(66) Catak, S.; Monard, G.; Aviyente, V.; Ruiz-Lopez, M. F. J. Phys.
Chem. A 2006, 110, 8354–8365.
(67) Van Speybroeck, V.; Moonen, K.; Hemelsoet, K.; Stevens, C. V.;
Waroquier, M. J. Am. Chem. Soc. 2006, 128, 8468–8478.
(68) Catak, S.; Monard, G.; Aviyente, V.; Ruiz-Lopez, M. F. J. Phys.
Chem. A 2008, 112, 8752–8761.
(69) Catak, S.; Monard, G.; Aviyente, V.; Ruiz-Lopez, M. F. J. Phys.
Chem. A 2009, 113, 1111–1120.
(70) De Sterck, B.; Vaneerdeweg, R.; Du Prez, F.; Waroquier, M.; Van
Speybroeck, V. Macromolecules 2009, 43, 827–836.
(71) da Silva, E. F.; Svendsen, H. F.; Merz, K. M. J. Phys. Chem. A 2009,
113, 6404–6409.
Similarly, the reactive complex that forms between the
aziridinium ion and bromide displays a strong interaction
between the bare halide ion and the nitrogen proton. How-
ever, in reality the bromide ion will be solvated by solvent
molecules, since acetic acid molecules are capable of forming
(72) Kelly, C. P.; Cramer, C. J.; Truhlar, D. G. J. Phys. Chem. A 2006,
110, 2493–2499.
(73) Pliego, J. R.; Riveros, J. M. J. Phys. Chem. A 2001, 105, 7241–7247.
(61) Bach, R. D.; Dmitrenko, O. J. Org. Chem. 2002, 67, 3884–3896.
J. Org. Chem. Vol. 75, No. 13, 2010 4535

Catak et al.
JOCArticle
SCHEME 4. Nucleophilic Attack Modes on Aziridinium Ions 5
and Aziridinium Conformers
Bromide Attack on Aziridinium Ions. In light of these results,
the nucleophilic ring opening of aziridinium ions 5with bromide
was further investigated with three explicit acetic acid molecules
coordinating the halide ion. Bulk solvation was not taken into
account since the dielectric constant as well as the solvent radius
is difficult to estimate for the solvent mixture at hand. More-
over, the problems reported by Warshel74 invoke caution when
using a combined implicit/explicit solvent model.
Reaction pathways for nucleophilic attack at the unsubsti-
tuted (path a) and substituted (path b) aziridine ring carbons
were modeled. Scheme 4 depicts modes of nucleophilic attack
on aziridinium ion 5. Needless to say, bromide attack will take
place in an SN2 fashion through a backside attack, where the
leaving group is the ring nitrogen. Although 1-trans-a and 1-
trans-d were previously shown to be two of the best conformers
for the aziridine (Table 1), backside attack of the nucleophile is
not favorable in the corresponding aziridinium conformers
(5-trans-a and 5-trans-d), since the cyanide group is in the attack
trajectory. These conformers would only be able to accommo-
date a frontside attack, which would clearly have a higher
barrier. Therefore, nucleophilic attacks were modeled using
aziridinium ion 5-trans-b, which corresponds to one of the most
energetically favorable aziridine conformers, 1-trans-b.
Transition state geometries for the solvent-assisted ring open-
ing are depicted in Figure 4, where the bromide ion is shown to
H-bond with three acetic acid molecules. Critical distances
indicate “product-like” transition states, where the nitrogen-
aziridine carbon distances are quite large (1.875 and 1.909 A,
for TS-a-3AcOH and TS-b-3AcOH, respectively) compared to
the reactant-complexes (∼1.5 A). For comparative purposes,
these transition states were also modeled without explicit solvent
molecules; critical distances showed drastic differences from the
solvated case, where nitrogen-aziridine carbon distances were
FIGURE 3. Coordination solvation energies (CSE) for the aziridi-
nium-bromide complex (MPW1B95/6-31þG*).
explicit, and mixed solvation models performed by Warshel
illustrated that mixed implicit/explicit solvation gave unre-
liable results with increasing number of explicit solvent
molecules and was highly influenced by their orientation/
alignment.74
The present study involves a solvent mixture composed of
acetic acid and acetonitrile, and thus the initial set up for the
cluster model is more complicated and the incorporation of
explicit solvent molecules is not straightforward. To investi-
gate this issue, the reactive complex between the intermediate
aziridinium 5 and bromide was coordinated by an increasing
number of explicit acetic acid and acetonitrile molecules, the
corresponding coordination solvation energies (CSE) were
calculated. At each stage various possibilities were explored,
in order to investigate the preference for coordination with
acetic acid or acetonitrile, and a thorough conformational
search was conducted to locate the most stable points on the
PES. As expected the bare bromide ion coordinates very
strongly with surrounding solvent molecules. The CSE of
acetic acid and acetonitrile, as the first coordinating solvent
molecule, were calculated to be -74.0 and -22.0 kJ/mol,
respectively. The preference for coordination with the polar
protic acetic acid over the polar aprotic acetonitrile is under-
standable. Figure 3 depicts CSEs with respect to a varying
number of solvent molecules. Coordination is favorable
up to four acetic acid molecules; from that point on the
CSE converges and further addition of explicit solvent
molecules does not cause any appreciable stabilization. The
first three acetic acid molecules form hydrogen bonds with
the bromide anion, whereas the fourth hydrogen bonds to
neighboring solvent molecules. The fifth acetic acid does not
form any explicit hydrogen bonds. In the present study, a
supermolecule model consisting of the substrate, the bro-
mide anion, and three acetic acid molecules was used. This
choice is motivated by the fact that only three acetic acid
molecules have explicit contacts with the bromide anion.
Moreover, free energy of solvation is expected to converge
earlier as a result of the entropic penalty of adding more
solvent molecules.
FIGURE 4. Transition state geometries (B3LYP/6-31þG(d,p)) for
the solvent-assisted ring opening of aziridinium ions by bromide:
unhindered (TS-a-3AcOH) and hindered (TS-b-3AcOH) pathway.
Some critical distances (A) are given.
(74) Kamerlin, S. C. L.; Haranczyk, M.; Warshel, A. ChemPhysChem
2009, 10, 1125–1134.
4536 J. Org. Chem. Vol. 75, No. 13, 2010

Catak et al.
JOCArticle
TABLE 3. Free Energies of Activation (ΔGq) and Free Energies of
Reaction (ΔGrxn) for the Non-solvated and Solvated Bromide-Induced
Ring Opening of 5-trans-ba,b
SCHEME 5. Aziridinium Ions 4 and 5
ΔGq
ΔGrxn
path a path b path a path b
nonsolvated B3LYP/6-31þþG(d,p)
B3LYP-D/6-31þþG(d,p)
11.8
15.4
7.5 -142.3 -135.0
9.2 -136.9 -131.2
14.1 -135.2 -128.8
MPW1B95/6-31þþG(d,p) 15.3
solvated
B3LYP/6-31þþG(d,p)
20.0
21.5
45.8
43.6
55.3
-33.7
-27.4
-28.7
-7.2
-4.1
-4.1
B3LYP-D/6-31þþG(d,p)
by bromide. In this section, the effect of N-substituents on the
aziridinium electronic structure will be investigated. The corre-
lation between regioselective preference in ring-opening reac-
tions and aziridinium N-substituent identity will be explored.
For this purpose, the electronic structures of aziridinium ions 4
and 5 have been subjected to a thorough analysis from an elec-
trostatic and a frontier molecular orbital point of view.
Electrostatic Analysis. Population analysis is a conven-
tional tool to gain insight into the electronic structure of
molecular systems.75-80 This technique is appealing to chemists,
since it gives a simple depiction of net atomic charges based on
the more complex density matrix representation of the electronic
many-body system. However, results from a population anal-
ysis should be carefully interpreted. There is no unique defini-
tion of the atomic charge from the quantum-mechanical view-
point, i.e., the atom is not a quantum-mechanical observable.81
Several schemes (mathematical definitions) have been devel-
oped over the past decades to derive atomic charges from
electronic structure computations. Each scheme has its strengths
and weaknesses, which must be carefully taken into account in
the interpretation of partial charges.
MPW1B95/6-31þþG(d,p) 33.0
aThermal free energy corrections from B3LYP/6-31þþG(d,p) calcu-
lations at 1 atm and 298 K. bEnergies in kJ/mol.
approximately 1.6 A and bromide-aziridine carbon distances
were 2.8 A. This is a clear indication that in the absence of explicit
solvent molecules, the reaction profile is significantly different
and transition states are “reactant-like” in nature. The difference
in the solvated versus nonsolvated potential energy surfaces is
merely an artifact of gas-phase calculations and can be attributed
to the extremely reactive nature of the bare bromide ion, whereas
in the solvated case the bromide ion is stabilized by solvent
molecules and is considerably less reactive. This is also reflected in
the corresponding barrier heights (Table 3); nonsolvated barriers
are significantly smaller in comparison to the solvated ones. This
indicates that the solvated bromide ion is less reactive because of a
considerable amount of stabilization, whereas the bare bromide
is extremely active as shown in the exceptionally small barriers for
both pathways. Additionally, reaction energies show dramatic
differences between the solvated and nonsolvated cases; the non-
solvated case is unrealistically exothermic as a result of the stabi-
lity difference between the bare bromide ion in the reactant and
the stabilized bromide in the product.
In the nonsolvated case, the barrier heights are practically
indistinguishable. This is most likely because of the considerably
long bromide-aziridine carbon distances (approximately 2.8 A)
previously mentioned. At this distance, the bromide ion is not
affected by the difference in steric hindrance between the two
aziridine carbon atoms and does not discriminate between them.
In the solvated case, however, calculations reveal a clear prefe-
rence for path a (Scheme 1) at all levels of theory, confirming the
experimentally observed regioselectivity. The solvated case
shows a clear kinetic preference for path a, as barriers for the
reverse reaction are less than 60 kJ/mol for each pathway. This
would suggest the possibility of thermodynamic equilibration.
Incidentally, path a also leads to the thermodynamically more
stable product. However, this argument does not hold for the
nonsolvated case, which has reverse barrier heights that are more
than 150 kJ/mol, preventing any possibility of thermodynamic
equilibration. These conclusions clearly indicate the necessity to
solvate halide ions in order to get a realistic overview of the PES,
in line with results obtained in previous theoretical studies on
bromide-induced aziridine ring-opening reactions.28,31,33
HBr- versus Benzyl Bromide-Induced Ring Opening of
Aziridines. The current study has presented experimental and
theoretical evidence suggesting that the ring opening of azir-
idines (1) with HBr proceeds via the unsubstituted aziridine
carbon, whereas previous experimental and theoretical reports
have shown that treatment of the same compounds with benzyl
bromide leads to ring opening at the substituted aziridine car-
bon.28,31 The main difference between these two reactions, apart
from the difference in solvent, is the nature of the intermediate
aziridinium ion (Scheme 5) that undergoes nucleophilic attack
MPA (Mulliken Population Analysis)82 charges are regularly
reported to be very sensitive to the choice of basis set used; large
basis sets are known to reflect unphysical trends.83-86 CHELPG
(CHarges from the ELectrostatic Potential)87 charges are known
to be unnecessarily sensitive to the choice of the grid points, the
orientation of the molecule, and conformational changes.88 The
NPA (Natural Population Analysis)83 scheme is clearly superior
to MPA and ESP-fitted charges.89 Its robustness toward large
basis sets has been tested extensively.85,90 However, a remaining
drawback of NPA charges is the poor reproduction of the ESP
(75) Hedegard,E. D.;Bendix,J.;Sauer, S. P. A.THEOCHEM 2009, 913, 1–7.
(76) Parks, J. M.; Guo, H.; Momany, C.; Liang, L.; Miller, S. M.;
Summers, A. O.; Smith, J. C. J. Am. Chem. Soc. 2009, 131, 13278–13285.
€
€
(77) Schwobel, J.; Ebert, R.-U.; Kuhne, R.; Schuurmann, G. J. Comput.
€€
Chem. 2009, 30, 1454–1464.
(78) Cheshmedzhieva,D.;Ilieva, S.;Hadjieva, B.;Trayanova,T.; Galabov,
B. Mol. Phys. 2009, 107, 1187–1192.
(79) Kaur, D.; Kohli, R. Int. J. Quantum Chem. 2008, 108, 119–134.
(80) Mora, J. R.; Tosta, M.; Domınguez, R. M.; Herize, A.; Barroso, J.;
´
ꢀ
Cordova, T.; Chuchani, G. J. Phys. Org. Chem. 2007, 20, 1021–1031.
(81) Parr, R. G.; Ayers, P. W.; Nalewajski, R. F. J. Phys. Chem. A 2005,
109, 3957–3959.
(82) Mulliken, R. S. J. Chem. Phys. 1955, 23, 1833–1840.
(83) Reed, A. E.; Weinstock, R. B.; Weinhold, F. J. Chem. Phys. 1985, 83,
735–746.
(84) Guerra, C. F.; Handgraaf, J.-W.; Baerends, E. J.; Bickelhaupt, F. M.
J. Comput. Chem. 2004, 25, 189–210.
(85) Martin, F.; Zipse, H. J. Comput. Chem. 2005, 26, 97–105.
(86) Bultinck, P.; Ayers, P. W.; Fias, S.; Tiels, K.; Van Alsenoy, C. Chem.
Phys. Lett. 2007, 444, 205–208.
(87) Breneman, C. M.; Wiberg, K. B. J. Comput. Chem. 1990, 11, 361–373.
(88) Francl, M. M.; Chirlian, L. E. In Reviews in Computational Chem-
istry; Lipkowitz, K. B., Cundari, T. R., Boyd, D. B., Eds.; Wiley-VCH:
New York, 2007; pp 1-31.
(89) Yerushalmi, R.; Scherz, A.; Baldridge, K. K. J. Am. Chem. Soc. 2004,
126, 5897–5905.
ꢀ
(90) Pacios, L. F.; Gomez, P. C.; Galvez, O. J. Comput. Chem. 2006, 27,
ꢀ
1650–1661.
J. Org. Chem. Vol. 75, No. 13, 2010 4537

Catak et al.
JOCArticle
TABLE 4. Mulliken, CHELPG, NPA, and Hirshfeld-I Charges (MPW1B95/6-31þþG(d,p)) for Aziridine Carbons C2 and C3 on Conformers of
Aziridinium Ions 5-trans
MPA
CHELPG
NPA
Hirshfeld-I
C2
C3
C2
C3
C2
C3
C2
C3
5-trans-a
5-trans-b
5-trans-c
5-trans-d
5-trans-e
5-trans-f
5-trans-g
5-trans-h
5-trans-i
Mean
-0.297
0.095
0.227
-0.343
-0.019
0.079
-0.403
-0.138
0.122
-0.054
0.187
-0.188
-0.245
-0.308
-0.177
-0.202
-0.187
-0.108
-0.130
-0.266
-0.208
0.055
0.137
0.145
0.111
-0.092
-0.047
-0.123
0.080
0.064
0.032
0.025
0.104
-0.233
-0.247
-0.209
-0.066
-0.106
-0.034
-0.105
-0.083
-0.052
-0.131
0.083
-0.049
-0.048
-0.050
-0.045
-0.043
-0.046
-0.052
-0.051
-0.053
-0.047
0.004
-0.230
-0.230
-0.237
-0.234
-0.238
-0.240
-0.234
-0.235
-0.241
-0.235
0.003
0.100
0.101
0.101
0.107
0.109
0.108
0.108
0.109
0.109
0.107
0.004
-0.196
-0.190
-0.206
-0.195
-0.196
-0.205
-0.192
-0.188
-0.203
-0.197
0.005
Std Dev
around the molecule compared to ESP-fitted charges.91,92
Charges obtained with the Hirshfeld-I (Iterative Hirshfeld)86
method are comparable to NPA charges in terms of robust-
ness with respect to the choice of basis set86 and do have the
additional advantage of reproducing the ESP around the
molecule.91,92 In light of this, Hirshfeld-I charges are con-
sidered the most reliable.
attributed to the difference in charge distribution in the respec-
tive aziridinium ions. The difference in charges on C2 and C3 in
aziridinium 4 is on average 0.330, whereas this difference is
reduced to 0.304 in aziridinium 5. There is no significant change
in the bond-dipole, and it is clear that the regioselectivity has no
electrostatic driving force. Note that considering a single con-
former could be misleading when nonrubost population
schemes such as MPA or CHELPG are used.
Various conformers of the aziridinium ions provide an ideal
case study for a comparative investigation of different popula-
tion analysis schemes. As the aim is to rationalize the opposite
preference for nucleophilic attack at the aziridine carbons of
aziridinium ions 4 and 5-trans, electrostatic properties of these
reactive sites have been investigated. Four population analysis
schemes mentioned above, namely, MPA, CHELPG, NPA,
and Hirshfeld-I, are used for comparison. Although, the first
three schemes are regularly used,75-80 the last method is a recent
development, and is not yet extensively tested. Atomic charges
for aziridine carbons C2 and C3 are listed for aziridinium con-
formers of 5-trans and 4 in Tables 4 and 5, respectively. While
MPA and CHELPG charges significantly fluctuate from one
conformer to the other, NPA and Hirshfeld-I charges are more
robust with respect to changes in geometry, as reflected in their
respective standard deviations, and will therefore be further used
to explain the electrostatic interactions between the aziridinium
and bromide ions. The geometry dependence of the MPA
charges in Tables 4 and 5 is mainly due to the dependence of
the (atom-centered) basis functions on the geometry, which in
turn affects MPA charges. These results confirm once again that
MPA charges are not reliable.
C3 is shown to bear a slightly larger negative charge (average
Hirshfeld-I = -0.260) in aziridinium 4 when compared to its
counterpart (average Hirshfeld-I = -0.197) in 5, whereas C2
charges do not deviate too much among the two aziridinium
ions, although C2 in 5 is slightly more positive (average Hirsh-
feld-I=þ0.107). This is in line with the expectation that the pos-
itive charge on the quaternary nitrogen is better stabilized
through electron donation from two benzyl groups (as in 4)
rather than one (as in 5). In turn, this causes less electron with-
drawal from the ring carbons, resulting in less positive charges
on C2 and C3 for aziridinium ion 4. The identity of the N-
substituents is shown to have an effect on the charge distribution
in the aziridine ring. Although these results provide insight,
the experimentally observed regioselective preference cannot be
From the overall charge analysis viewpoint, it is tempting
to conclude that C2 is always the preferred atom to undergo a
nucleophilic attack by the bromide ion. However, it is
important to realize that the electrostatic interaction be-
tween two molecules can never be ascribed to the net charge
of a few atoms in each molecule. Figure 5 depicts the ESPs on
the van der Waals surface of conformers 4g and 5-trans-b. It
is clear that the electrostatic potential in the vicinity of C2
and C3 is nearly equal in both cases. The value of the
electrostatic potential is not solely based on net charges on
C2 and C3 but also on atoms in close proximity. Therefore,
conclusions drawn from a population analysis should always
be compared and verified by analyzing the ESP.
Frontier Molecular Orbital Analysis. Results drawn from
the previous section clearly indicate that experimental regios-
electivities cannot be explained on the basis of pure electrostatic
considerations. In the attempt to rationalize the observed diffe-
rence in nucleophilic attack, a secondary approach, analysis of
frontier molecular orbitals, is applied. For this purpose, DFT-
based reactivity indicators are used.93-95 In particular, local
Fukui functions f(r), which can be used to describe orbital-
controlled reactions, were analyzed.96,97 Although, a recent study
on other local descriptors, namely, local softness and hardness,
indicates that their interpretation is not always straightforward,98
reactivity descriptors have been successfully used to elucidate
reaction pathways for a variety of reactions.67,99-101
(93) Parr, R. G.; Yang, W. Density-Functional Theory of Atoms and
Molecules; Oxford University Press: New York, 1989.
(94) Geerlings, P.; De Proft, F.; Langenaeker, W. Chem. Rev. 2003, 103,
1793–1874.
(95) Chermette, H. J. Comput. Chem. 1999, 20, 129–154.
(96) Parr, R. G.; Yang, W. J. Am. Chem. Soc. 1984, 106, 4049–4050.
(97) Ayers, P. W.; Levy, M. Theor. Chim. Acta 2000, 103, 353–360.
(98) Torrent-Sucarrat, M.; De Proft, F.; Geerlings, P.; Ayers, Paul W.
Chem.;Eur. J. 2008, 14, 8652–8660.
(99) De Witte, B.; Van Langenhove, H.; Hemelsoet, K.; Demeestere, K.;
De Wispelaere, P.; Van Speybroeck, V.; Dewulf, J. Chemosphere 2009, 76,
683–689.
(91) Van Damme, S.; Bultinck, P.; Fias, S. J. Chem. Theory Comput. 2009,
5, 334–340.
(92) Verstraelen, T.; Van Speybroeck, V.; Waroquier, M. J. Chem. Phys.
2009, 131, 044127-19.
(100) Hemelsoet, K.; Van Speybroeck, V.; Marin, G. B.; De Proft, F.;
Geerlings, P.; Waroquier, M. J. Phys. Chem. A 2004, 108, 7281–7290.
(101) Rozeboom, M. D.; Tegmo-Larsson, I. M.; Houk, K. N. J. Org.
Chem. 2002, 46, 2338–2345.
4538 J. Org. Chem. Vol. 75, No. 13, 2010

Catak et al.
JOCArticle
TABLE 5. Conformers of Aziridinium Ions 4 and Mulliken, CHELPG, NPA, and Hirshfeld-I Charges (MPW1B95/6-31þþG(d,p)) for Aziridine
Carbons C2 and C3
MPA
CHELPG
NPA
Hirshfeld-I
C2
C3
C2
C3
C2
C3
C2
C3
4a
4b
4c
4d
4e
4f
4g
4h
4i
4j
4k
4l
mean
std dev
0.082
0.038
0.325
-0.206
0.280
0.009
0.324
-0.199
0.143
0.437
0.335
0.032
0.133
0.211
-0.095
-0.287
-0.150
0.040
-0.199
-0.254
-0.177
0.076
-0.113
-0.545
-0.109
-0.076
-0.157
0.161
0.044
0.076
0.116
0.024
0.109
0.059
0.116
0.036
0.043
0.105
0.042
-0.026
0.062
0.044
-0.223
-0.160
-0.218
-0.109
-0.269
-0.144
-0.198
-0.115
-0.114
-0.134
-0.192
-0.098
-0.164
0.055
-0.040
-0.049
-0.039
-0.045
-0.037
-0.047
-0.034
-0.043
-0.038
-0.047
-0.039
-0.044
-0.042
0.005
-0.226
-0.225
-0.225
-0.230
-0.229
-0.227
-0.231
-0.234
-0.227
-0.228
-0.228
-0.233
-0.229
0.003
0.066
0.064
0.066
0.071
0.067
0.065
0.072
0.074
0.075
0.071
0.072
0.078
0.070
0.005
-0.261
-0.255
-0.261
-0.261
-0.260
-0.253
-0.262
-0.260
-0.262
-0.261
-0.264
-0.263
-0.260
0.003
The three-dimensional Fukui function is usually approxi-
mated using a finite difference methodology, resulting in three
different Fukui functions, fþ(r), f-(r), and f 0(r), that corres-
pond to nucleophilic, electrophilic, and radical attacks, respec-
tively. For the nucleophilic ring opening of aziridinium ions, the
following three-dimensional Fukui function was utilized:
values of the Fukui function, which provide a value for the
local indicator at the position of the atomic center, are
more often applied.105 The atom-based Fukui function is
defined as:
fkþ ¼ qkðN þ 1Þ - qkðNÞ
f þðrÞ ꢀ FN þ 1ðrÞ - FNðrÞ ꢀ FLUMOðrÞ
where qk(N) is the electron population on the kth atom of a
molecule with N electrons. The resulting values, which are
condensed-to-atoms, are inevitably influenced by popula-
tion analysis methods.106-109 However, DFT-based indica-
tors have shown little dependency on the DFT functional
The final approximation results from the idea that the
Fukui function generalizes the frontier molecular orbital
(FMO) concept developed by Fukui.102-104 This approx-
imation is valid whenever the orbital relaxation effects are
small, which is usually the case. Practical use of three-
dimensional indicators is tedious, and therefore condensed
(105) Yang, W.; Mortier, W. J. J. Am. Chem. Soc. 2002, 108, 5708–5711.
(106) De Proft, F.; Martin, J. M. L.; Geerlings, P. Chem. Phys. Lett. 1996,
230, 393–401.
(107) De Proft, F.; Martin, J. M. L.; Geerlings, P. Chem. Phys. Lett. 1996,
256, 400–408.
(108) Gilardoni, F.; Weber, J.; Chermette, H.; Ward, T. R. J. Phys. Chem.
A 1998, 102, 3607–3613.
(109) Thanikaivelan, P.; Padmanabhan, J.; Subramanian, V.; Ramasami,
T. Theor. Chim. Acta 2002, 107, 326–335.
(102) Fukui, K. Theory of Orientation and Sterioselection; Springer-
Verlag: New York, 1973.
(103) Fukui, K. Science 1982, 218, 747–754.
(104) Fukui, K.; Yonezawa, T.; Shingu, H. J. Chem. Phys. 1952, 20, 722–
725.
J. Org. Chem. Vol. 75, No. 13, 2010 4539

Catak et al.
JOCArticle
and 5-trans, respectively. These results are in line with the
experimentally observed regioselectivities, albeit with one ex-
ception (5-trans-c with preference for the substituted carbon
atom).
Fukui functions and LUMOs for aziridinium ions 4g and
5-trans-e are shown in Figure 6. Both properties show that
aziridinium ion 4g and 5-trans-e have a clearly opposite
preference, i.e., nucleophilic attack at the substituted versus
the unsubstituted aziridine carbons, respectively. These re-
sults indicate that the experimentally observed regio prefer-
ence can be rationalized in terms of frontier orbital concepts.
In terms of Fukui functions condensed-to-atoms, the type of
population analysis method used dramatically influences the
results obtained, as in the case of atomic charge analysis (see
Supporting Information for a detailed list of results). For
aziridinium ion 5-trans, the Hirshfeld-I method shows for all
conformers the largest f þk values for the C3 atom, in agreement
with the experimentally observed regioselectivity. In case of ion
4, the differences between the f þk values for the C2 and C3 atoms
are much smaller, and no definite conclusions can be drawn.
FIGURE 5. MPW1B95 electrostatic potential mapped on the elec-
tron-density surface [isovalue 0.0004] for aziridinium ions 4g and 5-
trans-b. Color code: [þ0.100 au; þ0.212 au] where red corresponds
to electron-rich and blue corresponds to electron-poor regions. The
choice of a positive range is motivated by the cationic nature of the
aziridinium ions.
Conclusions
The ring opening of 1-arylmethyl-2-(cyanomethyl)aziri-
dines with HBr (33% acetic acid solution) in acetonitrile or
dichloromethane was shown to afford 3-(arylmethyl)amino-
4-bromobutanenitriles via regiospecific ring opening at the
unsubstituted aziridine carbon. The aziridine ring system
and nucleophilic ring opening of the corresponding aziridi-
nium ion obtained through N-protonation at both aziridine
ring carbons has been studied by means of computational
methods in order to rationalize the experimentally observed
regioselectivity. The investigation of the PES with explicit
solvent molecules (acetic acid) solvating the bromide ion has
revealed a clear preference of attack at the unhindered
aziridine carbon, indicating kinetic control. HBr versus
benzyl bromide ring opening was analyzed through compar-
ison of the electronic structure of corresponding aziridinium
intermediates. Although the electrostatic picture indicates
that charge distribution and electrostatic potential sur-
faces of aziridinium ions is influenced by the difference in N-
substituents, it fails to explain the opposite regiospecific
nature of the reaction. However, frontier molecular orbital
analysis of LUMOs and nucleophilic three-dimensional
Fukui functions show a clear preference of attack for the sub-
stituted aziridine carbon in aziridinium 4 and for the unsub-
stituted aziridine carbon in aziridinium 5, successfully ratio-
nalizing the experimentally observed regioselectivity. This
shows that the outcome of aziridinium ring-opening reac-
tions is dictated by orbital control rather than electrostatics.
FIGURE 6. (a) Nucleophilic Fukui functions [isovalue 0.003 au]
and (b) LUMOs [isovalue 0.03 au] for aziridinium ions 4g and
5-trans-e.
and/or the basis set utilized, qualitative trends are shown to
be unaltered.110,111
In the present work, nucleophilic Fukui functions as well
as LUMO orbitals were calculated for all conformers of 4
and 5-trans (see Supporting Information for a complete list
of figures). The main focus is on the difference in the reac-
tivity between aziridine carbons C2 and C3 in each aziridi-
nium ion. The largest (positive) value of the Fukui function
indicates the most reactive site, while FMO suggests that a
site where the LUMO is localized is a good electrophilic site,
susceptible to nucleophilic attack. For some conformers no
clear preference for either of the aziridine carbons was ob-
served; however, the overall picture of the nucleophilic Fukui
functions shows a significant preference for the substituted
(C2) and unsubstituted (C3) carbon atoms, for structures 4
Experimental Section
Synthesis of 3-N-(Arylmethyl)amino-4-bromobutanenitriles 3.
To a stirred solution of 2-(cyanomethyl)aziridine 1 (16 mmol) in
dichloromethane (15 mL) was added a hydrogen bromide solution
(4.65 g, 1.2 equiv, 33% in acetic acid) at 0 °C. The resulting solution
was further stirred at room temperature for 1 h, after which the
reaction mixture was poured into a saturated KHCO3 solution
(40 mL). Extraction with dichloromethane (3 ꢁ 25 mL), drying
(MgSO4), and removal of the solvent afforded 3-N-(arylmethyl)-
amino-4-bromobutanenitrile 3, which was purified by means of
filtration through a pad of silica gel (hexane/EtOAc 2:1). The
initially orange oil turned black upon prolonged preservation.
(110) Hemelsoet, K.; Lesthaeghe, D.; Van Speybroeck, V.; Waroquier,
M. J. Phys. Chem. C 2007, 111, 3028–3037.
(111) Hemelsoet, K.; Lesthaeghe, D.; Van Speybroeck, V.; Waroquier,
M. Chem. Phys. Lett. 2006, 419, 10–15.
4540 J. Org. Chem. Vol. 75, No. 13, 2010

Catak et al.
JOCArticle
4-Bromo-3-[N-(phenylmethyl)amino]butanenitrile 3a. Orange
oil, yield 87%. Rf = 0.39 (hexane/EtOAc 2:1). 1HNMR(300MHz,
CDCl3): δ 2.62 and 2.68 (2H, 2ꢁdꢁd, J=16.9, 6.7, 5.9 Hz); 3.11-
3.18 (1H, m); 3.60 (2H, ∼d, J = 4.7 Hz); 3.82 and 3.88 (2H, 2 ꢁ d,
J = 13.2 Hz); 7.26-7.44 (4H, m). 13C NMR (75 MHz): δ 22.4
(CH2); 35.3 (CH2); 50.7 (CH2); 53.7 (CH); 117.3 (C); 127.5 (CH);
128.1 (CH); 128.6 (CH); 138.9 (C). IR (NaCl, cm-1): νΝH = 3324,
m/z (%) 287/89/91 (Mþ þ 1, 38); 207/9 (100). Anal. Calcd for
C11H12BrClN2: C 45.94, H 4.21, N 9.74. Found: C 45.78, H 4.06,
N 9.91.
4-Bromo-3-{N-[(4-methoxyphenyl)methyl]amino}butanenitrile
3d. Orange oil, yield 66%. Rf =0.49 (hexane/EtOAc 2:1). H
1
NMR (300 MHz, CDCl3): δ 2.61 and 2.66 (2H, 2 ꢁ d ꢁ d, J=
16.7, 6.5, 5.9 Hz); 3.11-3.18 (1H, m); 3.52-3.68 (2H, m); 3.72-
ν
Anal. Calcd for C11H13BrN2: C 52.19, H 5.18, N 11.07. Found: C
CN=2250. MS (70 eV): m/z (%) 253/5 (Mþþ1, 30); 173 (100).
3.84 (2H, m); 3.80 (3H, s); 6.80-6.92 and 7.22-7.30 (4H, m). 13
C
NMR (75 MHz, ref=CDCl3): δ 22.5 (CH2); 35.4 (CH2); 50.2
(CH2); 53.7 (CH); 55.4 (CH3); 114.1 (CH); 117.4 (C); 129.4 (CH);
130.9 (C); 159.1 (C). IR (NaCl, cm-1): νΝH=3346, νCN=2250.
MS (70 eV): m/z (%) no Mþ; 203(Mþ - Br, 100). Anal. Calcdfor
C12H15BrN2O: C 50.90, H 5.34, N 9.89. Found: C 50.74, H 5.55,
N 9.72.
51.94, H 4.93, N 10.86.
4-Bromo-3-{N-[(4-methylphenyl)methyl]amino}butanenitrile
1
3b. Orange oil, yield 51%. Rf = 0.40 (hexane/EtOAc 2:1). H
NMR (300 MHz, CDCl3): δ 2.34 (3H, s); 2.60 and 2.66 (2H, 2 ꢁ
d ꢁ d, J = 17.0, 6.7, 5.8 Hz); 3.11-3.18 (1H, m); 3.59 (2H, ∼d,
J = 4.7 Hz); 3.77 and 3.83 (2H, 2 ꢁ d, J = 13.2 Hz); 7.14-7.48
(4H, m). 13C NMR (75 MHz, ref = CDCl3): δ 21.2 (CH3); 22.5
(CH2); 35.5 (CH2); 50.6 (CH2); 53.7 (CH); 117.4 (C); 128.1 (CH);
129.4 (CH); 135.9 (C); 137.3 (C). IR (NaCl, cm-1): νNH = 3341,
Acknowledgment. This work was supported by the Fund
for Scientific Research Flanders (FWO-Vlaanderen) and the
Research Board of the Ghent University (BOF-GOA).
Computational resources and services used in this work were
provided by Ghent University.
ν
Anal. Calcd for C12H15BrN2: C 53.95, H 5.66, N 10.49. Found:
CN = 2250. MS (70 eV): m/z (%) 267/9 (Mþ þ 1, 55), 187 (100).
C 53.68, H 5.49, N 10.33.
4-Bromo-3-{N-[(4-chlorophenyl)methyl]amino}butanenitrile
1
3c. Orange oil, yield 82%. Rf = 0.45 (hexane/EtOAc 2:1). H
Supporting Information Available: Cartesian coordinates
(B3LYP/6-31þþG**), imaginary and low frequency modes,
and absolute energies (MPW1B95/6-31þþG(d,p)) of transition
states; condensed-to-atoms atomic Fukui functions for aziridi-
nium ions 4 and 5; nucleophilic Fukui functions and LUMOs for
aziridinium ions 4 and 5-trans; full ref 50. This material is
available free of charge via the Internet at http://pubs.acs.org.
NMR (300 MHz, CDCl3): δ 2.61 and 2.66 (2H, 2 ꢁ d ꢁ d, J =
16.7, 6.5, 5.9 Hz); 3.10-3.17 (1H, m); 3.59 (2H, ∼d, J=4.4 Hz);
3.78 and 3.85 (2H, 2 ꢁ d, J=13.5 Hz); 7.28-7.35 (4H, m). 13
C
NMR (75 MHz, ref=CDCl3): δ 22.6 (CH2); 35.3 (CH2); 50.1
(CH2); 53.8 (CH); 117.3 (C); 128.9 (CH); 129.5 (CH); 133.4 (C);
137.5 (C). IR (NaCl, cm-1): νΝH=3340, νCN=2250. MS (70 eV):
J. Org. Chem. Vol. 75, No. 13, 2010 4541