Imidazoline synthesis: Mechanistic investigations show that Fe catalysts promote a new multicomponent redox reaction

Article abstract of DOI:10.1039/d1dt00919b

Multicomponent reactions are attracting strong interest because they contribute to develop more efficient synthetic chemistry. Understanding their mechanism at the molecular level is thus an important issue to optimize their operation. The development of

Full text of DOI:10.1039/d1dt00919b

Dalton
Transactions
View Article Online
View Journal
PAPER
Imidazoline synthesis: mechanistic investigations
show that Fe catalysts promote a new
multicomponent redox reaction†
Cite this: DOI: 10.1039/d1dt00919b
Guillaume Coin,a,b Patrick Dubourdeaux,a Pierre-Alain Bayle,c Colette Lebrun,d
a
Pascale Maldivi*d and Jean-Marc Latour
*
Multicomponent reactions are attracting strong interest because they contribute to develop more
efficient synthetic chemistry. Understanding their mechanism at the molecular level is thus an important
issue to optimize their operation. The development of integrated experimental and theoretical approaches
has very recently emerged as most powerful to achieve this goal. In the wake of our recent investigation
of amidine synthesis, we used this approach to explore how an Fe-catalyzed aziridination can lead to an
imidazoline when run in acetonitrile. We report that the synthesis of imidazoline by combination of
styrene, acetonitrile, an iron catalyst and a nitrene precursor occurs along a new kind of multicomponent
reaction. The formation of imidazoline results from acetonitrile interception of a benzyl radical styrene
aziridination intermediate within Fe coordination sphere, as opposed to classical nucleophilic opening of
the aziridine by a Lewis acid. Comparison of this mechanism to that of amidine formation allows a
rationalization of the modes of intermediates trapping by acetonitrile according to the oxidation state Fe
active species. The molecular understanding of these processes may help to design other multicompo-
nent reactions.
Received 20th March 2021,
Accepted 20th April 2021
DOI: 10.1039/d1dt00919b
rsc.li/dalton
amides). Another method that is comparatively less developed
Introduction
consists in the nucleophilic opening of an aziridine by a Lewis
Imidazolines are a class of compounds of strong interest in acid (L. A., i.e. BF3·OEt2,15 Bi(OTf)3,16 TiF4 or HOTf,18,19
pharmacy1,2 and medicine3,4 which is continuously enhanced Scheme 1 and Scheme S1†) and trapping of the resulting cat-
by new developments, as for example for imaging5,6 and anti- ionic intermediate by a nitrile. A further development of these
bacterial7 purposes. Similarly, their established industrial synthetic methods has been the introduction of multicompo-
significance8,9 has been recently reinforced by new appli- nent reactions (MCR) based on a carbonyl/amine/isonitrile
cations as ligands for enantioselective catalysis10 as well as cor- combination20,21 or the association of two imines, CO and an
rosion additives.11 As a consequence of this continuous and acid chloride22,23 or the addition of an imine to an aziridine.24
17
renewed importance, a constant effort for developing new syn-
We recently developed integrated experimental25–28 and
thetic methods of imidazoline is pursued.8,12–14 Many imida- theoretical26–29 studies of styrene aziridination through Fe-
zoline syntheses have been proposed starting from either dini- catalyzed nitrene transfer (Scheme 2a). These studies even-
trogen derivatives (i.e. diamines, amidines) or unsaturated tually revealed that Fe-catalyzed aziridination is governed by
nitrogen compounds (i.e. imines, cyanides, isocyanides, the electron affinity of the Fe-imide active species.28 This con-
clusion was based on the study of mononuclear (1a) and
dinuclear (1b) Fe complexes of aminophenol ligands26,27
(Fig. 1) and mononuclear complexes of nitrogen ligands.28
aUniv. Grenoble Alpes, CEA, CNRS, IRIG – LCBM/pmb, F-38000 Grenoble, France.
E-mail: jean-marc.latour@cea.fr
bUniv. Grenoble Alpes, CNRS, UMR 5250, DCM, F-38000 Grenoble, France
cUniv. Grenoble Alpes, CEA, IRIG, MEM, F-38000 Grenoble, France
dUniv. Grenoble Alpes, CEA, CNRS, IRIG – SyMMES, F-38000 Grenoble, France.
E-mail: pascale.maldivi@cea.fr
†Electronic supplementary information (ESI) available: Experimental details
(synthesis and catalysis procedures, imidazoline characterization data, 1H and
15N NMR data) and computational details (energetic profiles, geometrical fea-
tures and spin densities, charges and cartesian coordinates for key intermedi-
ates. See DOI: 10.1039/d1dt00919b
Scheme 1 Imidazoline synthesis through aziridine opening.14
This journal is © The Royal Society of Chemistry 2021
Dalton Trans.

View Article Online
Paper
Dalton Transactions
tion. The reaction was carried out in the presence of a catalytic
amount of various Lewis acids in acetonitrile under mild con-
ditions (Table 1).
In the absence of L. A., no reaction was observed (Table 1,
entry 1). 3 is the only product formed in presence of all Lewis
Acids. It appeared that tris-chlorido complexes and cuprous
salts were much less active than triflate salts of tripositive
cations (Table 1, entries 2–4 vs. 5–9), the most active in the
series being Al(OTf)3 (Table 1, entry 10). This yield is compar-
able to literature results owing to the mild conditions used
(small L. A. molecular ratio and low temperature).14 It is note-
worthy that in the same reaction conditions a higher yield of
iminothiazolidine is obtained (87% vs. 70%),31 which is con-
sistent with the higher nucleophilicity of isothiocyanate vs.
acetonitrile.
In a second series of experiments, the aziridination of
styrene was run in acetonitrile using the diiron(III,II) catalyst
1b30 (Fig. 1) in the presence of indium or aluminum triflates
(Table 2). For the sake of comparison, the reaction was run
Scheme 2 Styrene aziridination and its coupling to aziridine opening (R
and R’ = aryl or alkyl).28,31
Table 1 Efficiency of various Lewis acid for the ring-opening cycliza-
tion of 2a
Fig. 1 Structures of the catalysts used in this study.27,30
Entry Lewis acid 2 (%) 3 (%) Entry Lewis acid 2 (%) 3 (%)
1
2
3
4
5
None
InCl3
Cu(OTf)2
AlCl3
Yb(OTf)3
100
81
75
66
57
0
6
7
8
9
Sn(OTf)2
Fe(OTf)3
Sc(OTf)3
In(OTf)3
Al(OTf)3
50
47
44
43
30
50
53
56
57
70
19
25
34
43
In addition, we reported a one-pot synthesis of 2-iminothia-
zolidines where the aziridination run in dichloromethane
(DCM) is coupled to aziridine opening by a Lewis acid and
trapping of the resulting cation by an isothiocyanate
(Scheme 2b) in a Domino Ring Opening Cyclization (DROC).31
In the course of these studies, we observed that when the aziri-
dination was run in acetonitrile instead of DCM, imidazoline
was formed in small ratio (ca. 15%, Scheme 2c). Moreover, we
observed that imidazoline ratio can be increased up to ca. 68%
by the one-pot DROC procedure that we used for the synthesis
of iminothiazolidine31 (Scheme 2d). This prompted us to study
the mechanism of the initial formation of imidazoline in
absence of L. A. (Scheme 2c) and we show that it follows a
different and original MCR mechanistic pathway.
10
a Yields determined by 1H NMR, using mesitylene as an internal stan-
dard. The uncertainty in these determinations is 3%.
Table 2 Aziridine
aziridinationa
vs.
imidazoline
formation
during
styrene
Entry
L. A.
2 (%)
3 (%)
Conv. (%)
1
None
85
28
31
23
32
15
72
69
77
68
98
62
73
58
73
2
In(OTf)3
In(OTf)3
Al(OTf)3
Al(OTf)3
Results and discussion
Ring-opening of aziridine
3b
4
5b
In a first series of experiments, we screened various L. A. for
the aziridine ring-opening with acetonitrile as nucleophile.
2-Phenyl-N-tosylaziridine 2 was chosen as representative sub-
strate31 and gave 2-(methyl)-4-phenyl-1-tosyl-4,5-dihydro-1H-
imidazole 3 through aziridine opening and acetonitrile inser-
a Reaction conditions : 1b/L. A./PhINTs/styrene molar ratio = 0.05/0.15/
1/10, CH3CN, 18 h, 25 °C. Conversion and 2/3 distribution yields deter-
mined by 1H NMR, using mesitylene as an internal standard. The
uncertainty in these determinations is 3%. b In(OTf)3 and Al(OTf)3
were added after 5 h.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2021

View Article Online
Dalton Transactions
Paper
initially in absence of L. A. and a distribution ratio 2/3 of 85/15
was noted as reported.26 When In(OTf)3 was used (Table 2,
entry 2), this distribution was reversed to 28/72, which shows
that catalytic imidazoline formation occurs in these con-
ditions. However, the conversion into the 2+3 mixture was
reduced as tosylamine and other products were formed. This
is likely caused by the fact that the Fe imine species active in
aziridination is sensitive to Lewis acids32,33 and can thus be
deactivated. In addition, 2 may be opened without formation
of 3 if either acetonitrile attack or cyclization is not effective.
When Al(OTf)3 was used (Table 2, entry 4), similar results were
obtained: distribution 2/3 ratio 23/77 with a diminished
overall conversion (58%). It follows that L. A. increase for-
mation of imidazoline but lower the overall conversion.
To try and counter this diminished conversion, we then per-
formed the reaction in the “1-pot 2-steps” conditions: the aziri-
dination was run in absence of M(OTf)3 which was added
when all PhI = NTs was consumed. In these conditions
(Table 2, entries 3 and 5) similar distribution 2/3 ratios ca. 31/
69 were obtained but a better conversion was achieved (73 vs.
ca. 60%), giving a 50% overall yield of 3.
The imidazoline formed in the latter reaction conditions
can therefore have two distinct origins: (i) the first reaction
run in absence of Al(OTf)3 (Scheme 2c) and (ii) the opening of
the aziridine in its presence (Scheme 1). The mechanism of
aziridine opening by a L. A. and subsequent nucleophilic
attack is well documented (Scheme S1†). By contrast, the
mechanism of imidazoline formation occurring in parallel to
the aziridine reaction (Scheme 2c) is not known. The possi-
bility that it may be similar to the classical mechanism of
Scheme S1† was suggested by the observation that aziridine
opening is catalyzed by ferric triflate.31 As a matter of fact, at
the end of the aziridination reaction, the catalyst is under
the form of a diferric complex. The limited formation of
imidazoline could therefore be due to a too limited reaction
time. To evaluate this hypothesis, we repeated the aziridina-
tion reaction (in absence of L. A.) in the same conditions but a
reaction time extended from 4 to 24 hours. This had no effect
on the imidazoline content of the final reaction mixture
(Table S1†). This observation suggests that imidazoline is
not formed in parallel to aziridination (Scheme 2c) through
aziridine opening by the aziridination catalyst but by
a different and unknown mechanism, which prompted us to
investigate it.
Scheme 3 Structures of the two possible imidazoline isomers: 3a
(2-methyl-3-tosyl-4-phenyl-1,3-imidazoline) and 3b (2-methyl-3-tosyl-
5-phenyl-1,3-imidazoline).
gate it, we resorted to 15N NMR analysis of the reaction
mixture at the end of catalysis. The catalytic reaction was run
with 1b in usual overall conditions (room temperature, cata-
lyst/nitrene precursor/styrene: 0.05/1/10, acetonitrile). The iso-
topic content of the nitrene precursor was varied by using
either PhI = 14NTs or the combination PhI(OAc)2/Ts-15NH2. A
complete 1H and 13C NMR study of the isolated imidazoline
led us to assign the three protons of the cycle and the methyl
group at δ = 4.96 (Ha), 4.17 (Hc), 3.45 (Hb) and 2.32 (CH3) ppm,
respectively (Fig. S7†). Fig. S8† presents the 15N NMR spectrum
of the catalytic solution after separation of the catalyst. Three
signals were detected at δ = 85.2, 93.8 and 148.6 ppm which
are assigned to the NTs groups of the aziridine 2, tosylamine
and imidazoline 3, respectively. The 1H-15N 2D spectrum of
the catalytic solution obtained with PhI(OAc)2/Ts-15NH2
(Fig. 2) exhibits strong correlations of the imidazoline 15NTs
with the signals assigned to CH3 at 2.32 ppm and Hc at
4.17 ppm. Consistently, strong 15N couplings are detected for
both resonances in the 1H NMR spectra (Fig. 3). This indicates
that the 15NTs is bound to styrene Cβ (isomer 3b). This obser-
vation is consistent with the mechanism proposed for aziridi-
nation by these catalysts26 which implies an initial attack of
styrene Cβ onto the imido nitrogen. The full characterization
of 3 allowed us to investigate its mechanism of formation.
Identification of the imidazoline isomer
As a preamble to the mechanistic study, it was necessary to
identify which imidazoline isomer was formed in parallel to
the aziridination reaction. As a matter of fact, the imidazoline
chemical nature was deduced from mass spectrometry (Fig. S1
and S2†) and 1H and 13C NMR (Fig. S3–S7†) analyses.
However, neither of these techniques could discriminate easily
between the two possible isomers which differ by the linkage
of the NTs group either to Cα or to Cβ of styrene (Scheme 3).
Knowledge of this position will be mechanistically insight-
Fig. 2 1H – 15N 2D NMR spectrum of the final catalytic solution con-
ful since it will inform on the site of nitrene attack. To investi- taining a mixture of 2, 3 and TsNH2 (after elimination of catalyst 1b).
This journal is © The Royal Society of Chemistry 2021
Dalton Trans.

View Article Online
Paper
Dalton Transactions
Table 4 Influence of THF on imidazoline formationa
Entry DCM MeCN THF Conv.b (%) 2b (%)
3b (%) 4b (%)
1
2
3
4
5
6
50c
—
—
—
—
—
40
90
80
70
60
50
10c
10
20
30
40
50
95
93
82
98
98
98
92
3
0
35 (90)d 4 (10)d 61
20 (87)
16 (88)
12 (86)
11 (86)
3 (13)
2 (12)
2 (14)
2 (14)
77
81
86
87
Fig. 3 14N vs. 15N 1H NMR spectra of HA, HB, HC and CH3 protons (see
Scheme 3 for assignments).
a Reaction conditions: 1b/PhI = NTs/styrene molar ratio = 0.05/1/10,
25 °C, 5 h. b Yield based on 1H NMR analysis of the reaction mixture,
using mesitylene as internal standard. The uncertainty in these deter-
minations is 3%. c MeCN was added after 5 h and the reaction
mixture was stirred for another 5 h. d In parentheses aziridine and imi-
dazoline distribution.
Catalytic experiments
In a third series of experiments (Table 3), we investigated
whether imidazoline formation and nitrene transfer are strictly
linked by changing the reaction medium. In pure DCM (entry
7) only aziridine was formed of course. Working in pure MeCN
(entry 1) gave a 2/3 ratio of 84/16. This ratio was not affected
when the MeCN content was halved (entry 2). By contrast,
adding the same amount of MeCN after the aziridination reac-
tion has subsided (entry 8) decreased imidazoline formation
four times. No further effect was observed upon addition of a
MeCN/THF mixture in the same conditions (Table 4, entry 1).
This shows that most imidazoline formation is linked to
nitrene transfer. Then we decreased the MeCN content in
DCM systematically from 25% to 0.1% (entries 3–6). When
MeCN content was 11%, the 2/3 ratio was not affected (entries
3 and 4). By contrast, when it reached 1%, only 3% imidazo-
line were formed and traces were barely detected at 0.1%
MeCN in DCM. This shows that full formation of imidazoline
in the present conditions (styrene 10 eq. vs. PhI = NTs)
requires that the MeCN content be as high as 10 eq. vs.
styrene.
Our study of the mechanism of amidine formation through
MeCN insertion revealed that in similar conditions THF abol-
ished amidine formation by blocking acetonitrile binding to
1b, thereby demonstrating that MeCN binds to the Fe cata-
lyst.34 To investigate whether MeCN binding is required to
form imidazoline, we performed a fourth series of experiments
(Table 4) in which the reaction medium was MeCN/THF mix-
tures of variable compositions. IWe showed earlier that THF
can be α-aminated to 2-tosylaminotetrahydrofuran
4
(Scheme S2†) in the same conditions using 1b as catalyst.35 In
the present series of experiments, we anticipated that THF
binding to Fe would block acetonitrile binding and potentially
affect the formation of 3 and thereby the 2/3 ratio. Table 4,
entries 2–6 show that progressive inclusion of THF in the reac-
tion mixture from 10 up to 50% leads to increased formation
of 4 from 61 up to 87% at the expense of 2 and 3. But the com-
position of the 2/3 mixture (88
2/12
2) is essentially
Table 3 Influence of solvent composition on imidazoline formationa
unaffected. The retention of 2/3 ratio in presence of THF indi-
cates that 3 is formed without acetonitrile binding to Fe. The
diminished production of the 2/3 mixture in presence of THF
confirms that both 2 and 3 are formed during the nitrene
transfer.
Finally, in a fifth series of experiments, we studied the
nitrile influence on imidazoline/aziridine distribution in the
aziridination reaction (Table S2†). Catalytic experiments were
thus run in usual conditions using various nitriles R–CuN,
RvtBu, Me and Ph. The same distribution was obtained, and
the absence of inductive effects goes against the possibility
that the nitrile reacts with a cationic intermediate, thereby con-
trasting the classical mechanism (Scheme S1†).
Entry
DCM
MeCN
Conv.b (%)
2b (%)
3b (%)
1
2
3
4
5
6
7
8c
—
50
75
89
100
50
25
11
1
0.1
—
50
100
100
94
93
91
91
87
98
84
85
88
90
16
15
13
10
99
97
3
99.9
100
50
>99
100
94
Traces
4
a Reaction conditions: 1b/PhI = NTs/styrene molar ratio = 0.05/1/10,
25 °C, 5 h. b Yield based on 1H NMR analysis of the reaction mixture,
using mesitylene as internal standard. The uncertainty in these deter-
minations is 3%. c MeCN was added after 5 h and the reaction
mixture was stirred for another 5 h.
DFT calculations
In order to uncover this different mechanism, we turned to
DFT modelling. As a matter of fact, our recent investigations of
Dalton Trans.
This journal is © The Royal Society of Chemistry 2021

View Article Online
Dalton Transactions
Paper
Fe-catalyzed aziridination revealed that DFT calculation of of this metallacycle from the imido FeV intermediate is ≥20 kcal
reaction profiles was instrumental to decipher the aziridina- mol−1 higher than the low energy pathways leading to direct
tion mechanism as they gave rise to a remarkable consistency aziridination.26 Our present calculations (Fig. S9 and Table S3†)
with experiments.26–28 Furthermore, using the same approach, reveal that the formation of the metallacycle from an imido FeIV
we analyzed very recently the far more complex mechanism of intermediate also requires a high activation energy (≥20 kcal
amidination34 combining nitrene transfer and nitrile inser- mol−1). An analogous result had been obtained from an imido
tion, which is especially relevant to the present purpose. Such FeIV intermediate by Jenkins and coll.38 for Fe macrocyclic cata-
reactivity calculations using the hybrid B3LYP functional are lysts. (ii) In both FeIV or FeV active models, the resulting metalla-
impracticable for catalyst 1b because of the size of the system cycle bears a radical character on the N atom issued from aceto-
combined to the multiplicity of spin states of both Fe.35,36 nitrile that should thus attack styrene Cβ carbon which is the
Therefore, they were performed on the mononuclear catalyst prefered site of attack in the aziridination.26 But, from the
1a. This is justified by the fact, as abovementioned, that both present NMR experiments, we know that this Cβ atom is bound
catalysts 1a and 1b give overall similar reactivity for aziridine to the NTs group in the formed imidazoline.
vs. imidazoline formation.26 Indeed other relevant mono-
Attack of an aziridination intermediate. The second possi-
nuclear models keeping only the FeIVNTs side proved to be bility is therefore to proceed with the first step of aziridination
difficult to design, mainly because the two iron atoms share to the first intermediate bearing a radical on styrene Cα atom26
three ligands: the phenolate oxygen and two carboxylates and consider that acetonitrile attacks this radical before the
which play a critical role in balancing the charges of the closure step of the aziridine (Scheme 4B). Several active
complex. In addition, the dinuclear catalyst 1b was shown to species can be considered depending on the Fe oxidation state
operate on either a FeIV 36 or a FeV 35 active species. This was and potential acetonitrile binding. We showed earlier that
taken into account in the computations by considering both acetonitrile does not bind significantly to FeIV = NTs26 but that
FeIV(vNTs) and FeV(vNTs) active species derived from 1a as it does to FeV = NTs34 owing to the higher acidity of the metal.
in our investigation of amidine formation.34 Such high-valent Therefore, three active species must be considered, i.e.
Fe-imido species are well known to exhibit some radical char- [FeIV(vNTs)], [FeV(vNTs)] and [FeV(vNTs)(CH3CN)]. Previous
acter on the nitrogen atom,26,29,35,36 but for the sake of clarity calculations have shown that these active species have widely
we prefer this notation based on Fe oxidation state. different electron affinities. [FeV = NTs] has a very high electron
Accordingly, we used the same computational approach as affinity (143 kcal mol−1 35
developed in these latter mechanistic studies.34 tion potential of styrene (139 kcal mol−1).28 As a consequence,
)
which excedes the calculated ioniza-
Two main possibilities may be envisaged (Scheme 4) for the an easy electron transfer from styrene would occur to form a
insertion of acetonitrile: (i) acetonitrile attacks the Fe-bound cationic intermediate. Our experimental studies (Table 3) have
NTs through its Csp carbon to generate a metallacycle (diaza- failed to evidence an electro-inductive influence in imidazoline
ferracyclobutane) as a new nitrene able to attack the olefin, or formation. This rules out the potential acetonitrile attack of a
(ii) acetonitrile attacks through its nitrogen the Cα carbon of cationic intermediate, and hence, the intervention of FeV
=
the first aziridination intermediate.26,28
NTs in imidazoline formation. Moreover, this intermediate
Formation of a metallacycle. The formation of a metallacycle should be very short-lived, the overall process leading to the
intermediate was considered first (Scheme 4A). Such a process aziridine without barrier.
was indeed demonstrated in the case of copper compounds.37
In addition, our experiments using acetonitrile/THF solvent
However, it can be ruled out in the present case on two mixtures showed that THF did not change the aziridine/imida-
different grounds. (i) Indeed, we showed in our previous study zoline ratio, thereby indicating that acetonitrile binding did not
on amidine formation34 that the activation energy for formation take part in the process. This excludes the involvement of the
adduct [FeV = NTs(CH3CN)] as active species. From these obser-
vations, it appears that imidazoline formation is linked only to
aziridination catalyzed by the [FeIV = NTs] active species.
This reaction is sketched in Scheme 5 together with the
molecular structure of the model [FeIV = NTs] active species.
The corresponding thermodynamic profile leading to 3 from
the FeIII intermediate Int0 is represented in Fig. 4. Only the
quintet state is shown here as we have observed that on the
triplet surface a much higher activation barrier is required to
yield the first intermediate.26
The formation of imidazoline from Int0 involves a two-step
process. A first step leads to intermediate Int1 (Fig. 4, Tables
S4 and S5†) resulting from the bonding of acetonitrile N to
styrene Cα carbon. From this intermediate, a cyclization could
readily occur to form imidazoline weakly coordinated by one
tosyl oxygen to the FeII ion. To gain more information on the
Scheme 4 Mechanistic possibilities depicted for an FeIV active species
through (A) metallacycle formation and (B) benzyl radical interception
by acetonitrile.
This journal is © The Royal Society of Chemistry 2021
Dalton Trans.

View Article Online
Paper
Dalton Transactions
isolated acetonitrile, showing that the triple bond is still
present while a new Cα–N bond at 1.44 Å is formed. This
points to a tetravalent nitrogen atom which indeed exhibits a
positive partial charge. It is worth noting that in this process
acetonitrile attack induces the reduction of the Fe center by
the benzyl radical. Finally the cyclization leads to imidazoline
with a C–N (from initial acetonitrile) bond length of 1.28 Å
consistent with a double bond, and a Cα–N bond length of
1.47 Å consistent with a C–N single bond. Dissociation of the
imidazoline from the FeII complex (not shown on Fig. 4)
brings a further stabilization by a very weak enthalpic value of
−1.8 kcal mol−1. The cyclization process from Int 1 does not
require a high activation enthalpy. By contrast, an enthalpic
barrier of 16.6 kcal mol−1 is associated to acetonitrile addition
to the benzyl radical which is clearly the determining step.
This enthalpy value is higher than the ones involved in aziridi-
nation processes, all being less than 10 kcal mol−1 at the same
computational level.26 However, being the solvent, acetonitrile
is in huge excess (1.4 104 eq. vs. 1b) which renders the imida-
zoline pathway competitive to aziridine formation on statistical
criteria. These opposite factors are reflected in the aziridine vs.
imidazoline ratio (ca. 6) strongly in favor of aziridine.
Scheme 5 Top: model active species (FeIV) derived from 1a; bottom:
proposed pathway for aziridination and imidazoline formation.
General mechanistic considerations
To summarize, our mechanistic studies have revealed that imi-
dazoline can be accessed through an unprecedented multi-
component redox reaction where free acetonitrile traps an
intermediate benzylic radical. Acetonitrile insertion into a
product of nitrene transfer has been described only in a very
limited number of cases: allylic amination catalyzed by Mn or
Fe porphyrins,39 and aliphatic aminations catalyzed by
copper40 and iron34 complexes. In all cases these reactions led
to an amidine derivative. A similar process involving aceto-
nitrile insertion competing with aziridination to form an imi-
dazoline had never been described.
As far as mechanisms are concerned, it is clear that aceto-
nitrile insertion through formation of metallacyclobutane
(Scheme 4) is possible but it is generally unfavorable with
respect to other pathways.34,38 Albeit acetonitrile insertions to
form imidazoline and amidine bear overall resemblance, their
mechanisms depart in many respects as highlighted in
Scheme 6. The first difference resides in the oxidation state of
Fig. 4 Two-step reaction pathway for imidazoline cyclization from
acetonitrile attack on the aziridination intermediate. Calculations with
B3LYP-D3/BS2-COSMO//B3LYP-D3-BS1. Enthalpies in kcal mol−1. Some
selected bond distances (Å) are shown.
actual electronic features of the key species, we analyzed their
Mulliken charges and spin densities (Table S6†). The initial
species Int0, which is the intermediate in the aziridination
process,26 is a high spin FeIII complex antiferromagnetically
coupled to the radical on styrene Cα. The acetonitrile attack
yields Int1, in which the spin densities clearly point to a
reduction of the Fe ion to a high spin FeII state. No other
radical character can be detected. The C–N (from acetonitrile)
bond length in intermediate Int1 is identical (1.15 Å) to that in
Scheme 6 Comparison of the mechanisms of formation of imidazoline (top) and amidine (bottom).
Dalton Trans.
This journal is © The Royal Society of Chemistry 2021

View Article Online
Dalton Transactions
Paper
the catalyst: as shown in the present work, imidazoline for-
mation is catalyzed by the FeIV species, whereas amidine for-
Acknowledgements
mation is catalyzed by an acetonitrile-bound FeV catalyst.34 The G. C. and J.-M. L. thank the French National Agency for
second difference stays in the nature of the intermediate Research (ANR) “programme Labex” (Labex ARCANE and
attacked by acetonitrile: a radical for imidazoline vs. a cation for CBH-EURGS (ANR-17-EURE-0003) for funding. This work was
amidine. The last difference is to be found in the nature of the performed thanks to HPC resources from GENCI (IDRIS and
attacking acetonitrile molecule which comes from the solvent in CINES, grants A0060807648 and A0080807648).
the case of imidazoline, whereas it is bound to Fe in the case of
amidine. There is a strong intrinsic consistency among these fea-
tures: aliphatic amination requires a more powerful oxidation
catalyst than aziridination, i.e. an FeV species as opposed to an
References
FeIV one. This FeV species performs a H• abstraction generating a
benzyl radical and an FeIV intermediate, powerful enough to
further oxidize the radical to a cation. This further oxidation is
not possible by the intermediate FeIII resulting from the aziridi-
nation first step. In addition, this FeV catalyst is acidic enough
for acetonitrile to bind it, whereas it cannot bind the FeIV
catalyst.26,27 Consequently, the acetonitrile molecule attacking
the benzyl cation comes from Fe coordination sphere, whereas
that attacking the benzyl radical comes from the solvent.
Noteworthy, in both cases, all these events occur within or
close to the Fe coordination sphere. As a further comparison
between the two processes, the extent of acetonitrile insertion
deserves to be commented. Indeed, ca. 15% of imidazoline is
formed (vs. 85% aziridine) as opposed to ca. 88% of amidine
(vs. 12% of amine). This limited imidazoline formation can be
explained by the fact that acetonitrile insertion competes with
closure of the aziridine ring which generally implies a small
activation barrier.28,38 By contrast, in the case of amidine for-
1 M. Krasavin, Eur. J. Med. Chem., 2015, 97, 525–537.
2 M. Krasavin, Chem. Heterocycl. Compd., 2017, 53, 240–255.
3 J. A. Lowry and J. T. Brown, Clin. Toxicol., 2014, 52, 454–
469.
4 F. Saczewski, A. Kornicka and L. Balewski, Expert Opin.
Ther. Pat., 2016, 26, 1031–1048.
5 Q. Xu, C. H. Heo, G. Kim, H. W. Lee, H. M. Kim and
J. Yoon, Angew. Chem., Int. Ed., 2015, 54, 4890–4894.
6 P. Zhang, H. Wang, Y. Hong, M. Yu, R. Zeng, Y. Long and
J. Chen, Biosens. Bioelectron., 2018, 99, 318–324.
7 M. S. Faillace, A. P. Silva, A. L. Alves Borges Leal, L. M. da
Costa, H. M. Barreto and W. J. Pelaez, ChemMedChem,
2020, 15, 851–861.
8 H. Liu and D.-M. Du, Adv. Synth. Catal., 2009, 351, 489–
519.
9 R. Tyagi, V. K. Tyagi and S. K. Pandey, J. Oleo Sci., 2007, 56,
211–222.
mation, acetonitrile insertion competes with the radical 10 S. Nakamura, K. Hyodo, M. Nakamura, D. Nakane and
rebound process which requires a substantial activation energy
(>15 kcal mol−1).34,41
H. Masuda, Chem. – Eur. J., 2013, 19, 7304–7309.
11 K. Zhang, B. Xu, W. Yang, X. Yin, Y. Liu and Y. Chen,
Corros. Sci., 2015, 90, 284–295.
12 R. D. Crouch, Tetrahedron, 2009, 65, 2387–2397.
13 B. H. Rotstein, S. Zaretsky, V. Rai and A. K. Yudin, Chem.
Rev., 2014, 114, 8323–8359.
14 M. S. Al Mehedi and J. J. Tepe, Adv. Synth. Catal., 2020, 362,
4189–4225.
15 S. Gandhi, A. Bisai, B. A. B. Prasad and V. K. Singh, J. Org.
Chem., 2007, 72, 2133–2142.
16 X. Li, X. Yang, H. Chang, Y. Li, B. Ni and W. Wei,
Eur. J. Org. Chem., 2011, 3122–3125.
17 T. Ishikawa, M. Yoshiki, T. Tanaka, K. Ogata, Y. Yamada
and T. Hanamoto, Synthesis, 2016, 48, 1322–1330.
18 R. Li, H. Jiang, W.-Y. Liu, P.-M. Gu and X.-Q. Li, Chin.
Chem. Lett., 2014, 25, 583–585.
Conclusion
Combined experimental and computational studies of these
solvent insertion reactions allow to reach a detailed molecular
understanding of these multicomponent reactions and to
delineate the main energetic factors orienting the selectivity
toward insertion. Moreover, by coupling this process to classi-
cal aziridine opening by a nucleophile in “one pot two steps”
DROC process, it is possible to enhance imidazoline for-
mation. This could pave the way to developing this approach
to address solvent incorporation as a synthetic tool.
19 Q. Zuo, Z. Shi, F. Zhan, Z. Wang, J.-S. Lin and Y. Jiang,
Tetrahedron Lett., 2019, 151576.
Author contributions
20 N. Elders, R. F. Schmitz, F. J. J. de Kanter, E. Ruijter,
M. B. Groen and R. V. A. Orru, J. Org. Chem., 2007, 72,
6135–6142.
21 C. T. Martha, A. Heemskerk, J.-C. Hoogendoorn, N. Elders,
W. M. A. Niessen, R. V. A. Orru and H. Irth, Chem. – Eur. J.,
2009, 15, 7368–7375.
G.C., P.D., P.-A.B. and C.L. performed and analyzed experiments.
P.M. designed, performed and analyzed computations. G.C., P.
M. and J.-M.L. designed the work and wrote the manuscript.
Conflicts of interest
22 K. Worrall, B. Xu, S. Bontemps and B. A. Arndtsen, J. Org.
Chem., 2011, 76, 170–180.
There are no conflicts to declare.
This journal is © The Royal Society of Chemistry 2021
Dalton Trans.

View Article Online
Paper
Dalton Transactions
23 S. Bontemps, J. S. Quesnel, K. Worrall and B. A. Arndtsen, 33 X. Lu, Y.-M. Lee, M. S. Seo and W. Nam, Chem. Commun.,
Angew. Chem., Int. Ed., 2011, 50, 8948–8951. 2020, 56, 11207–11210.
24 M. R. Kuszpit, W. D. Wulff and J. J. Tepe, J. Org. Chem., 34 G. Coin, P. Dubourdeaux, F. Avenier, R. Patra, L. Castro,
2011, 76, 2913–2919.
25 F. Avenier and J.-M. Latour, Chem. Commun., 2004, 1544–
1545.
26 R. Patra, G. Coin, L. Castro, P. Dubourdeaux,
M. Clémancey, J. Pécaut, C. Lebrun, P. Maldivi and
J.-M. Latour, Catal. Sci. Technol., 2017, 7, 4388–4400.
27 G. Coin, R. Patra, M. Clémancey, P. Dubourdeaux, J. Pécaut,
C. Lebrun, L. Castro, P. Maldivi, S. Chardon-Noblat and
J.-M. Latour, ChemCatChem, 2019, 11, 5296–5299.
28 G. Coin, R. Patra, S. Rana, J. P. Biswas, P. Dubourdeaux,
C. Lebrun, P.-A. Bayle, J. Pécaut, G. Blondin, P. Maldivi and
J.-M. Latour, ACS Catal., 2021, 11, 2253–2266.
35 E. Gouré, D. Senthilnathan, G. Coin, F. Albrieux, F. Avenier,
P. Dubourdeaux, C. Lebrun, P. Maldivi and J.-M. Latour,
Angew. Chem., Int. Ed., 2017, 56, 4305–4309.
36 E. Gouré, F. Avenier, P. Dubourdeaux, O. Sénèque,
F. Albrieux, C. Lebrun, M. Clémancey, P. Maldivi and
J.-M. Latour, Angew. Chem., Int. Ed., 2014, 53, 1580–1584.
37 A. Bakhoda, Q. Jiang, J. Bertke, T. R. Cundari and
T. H. Warren, Angew. Chem., Int. Ed., 2017, 56, 6426–6430.
M. Clemancey, S. P. de Visser, D. Maiti, P. Maldivi and 38 S. B. Isbill, P. P. Chandrachud, J. L. Kern, D. M. Jenkins
J.-M. Latour, ACS Catal., 2020, 10, 10010–10020.
29 R. Patra and P. Maldivi, J. Mol. Model., 2016, 22, 278.
30 S. Chardon-Noblat, O. Horner, B. Chabut, F. Avenier,
and S. Roy, ACS Catal., 2019, 9, 6223–6233.
39 D. A. Evans, M. M. Faul and M. T. Bilodeau, J. Am. Chem.
Soc., 1994, 116, 2742–2753.
N. Debaecker, P. Jones, J. Pécaut, L. Dubois, C. Jeandey, 40 V. Bagchi, P. Paraskevopoulou, P. Das, L. Chi, Q. Wang,
J.-L. Oddou, A. Deronzier and J.-M. Latour, Inorg. Chem.,
2004, 43, 1638–1648.
31 G. Coin, O. de Ferrier de Montal, P. Dubourdeaux and
J.-M. Latour, Eur. J. Org. Chem., 2021, 443–448.
32 T. Devi, Y.-M. Lee, W. Nam and S. Fukuzumi, Coord. Chem.
Rev., 2020, 410, 213219.
A. Choudhury, J. S. Mathieson, L. Cronin, D. B. Pardue,
T. R. Cundari, G. Mitrikas, Y. Sanakis and P. Stavropoulos,
J. Am. Chem. Soc., 2014, 136, 11362–11381.
41 S. Kumar, A. S. Faponle, P. Barman, A. K. Vardhaman,
C. V. Sastri, D. Kumar and S. P. de Visser, J. Am. Chem. Soc.,
2014, 136, 17102–17115.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2021