Studies on the Toxic Oil Syndrome: proposal of a mechanism for the thermal conversion of 3-N-phenylamino-1,2-propanediol esters into anilides under deodorisation conditions

Article abstract of DOI:10.1016/j.tet.2008.09.102

A study on the reaction mechanism for the conversion of title esters, the species recognised as toxic biomarkers of the oil batches responsible for the Toxic Oil Syndrome, into the corresponding anilides under the thermal conditions of an oil deodorisatio

Full text of DOI:10.1016/j.tet.2008.09.102

Tetrahedron 65 (2009) 418–426
Contents lists available at ScienceDirect
Tetrahedron
journal homepage: www.elsevier.com/locate/tet
Studies on the Toxic Oil Syndrome: proposal of a mechanism for the thermal
conversion of 3-N-phenylamino-1,2-propanediol esters into anilides under
deodorisation conditions
a
b
a,
*
´
Jordi Escabros , Ramon Crehuet , Angel Messeguer
a Department of Chemical and Biomolecular Nanotechnology, Institut de Qu´ımica Avançada de Catalunya, CSIC, Barcelona, Spain
b Department of Biological Chemistry and Molecular Modeling, Institut de Qu´ımica Avançada de Catalunya, CSIC, Barcelona, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
A study on the reaction mechanism for the conversion of title esters, the species recognised as toxic
biomarkers of the oil batches responsible for the Toxic Oil Syndrome, into the corresponding anilides
under the thermal conditions of an oil deodorisation process was performed using experimental and
computational techniques. The results obtained suggest a reaction course that includes two basic steps:
an intramolecular process involving the reaction of the amine group of the diester derivative with the
secondary ester of the same compound, followed by the attack of an aniline molecule to form the
(E)-isomer of an imidic acid, which would finally tautomerise to give the final anilide.
Ó 2008 Elsevier Ltd. All rights reserved.
Received 29 July 2008
Received in revised form
22 September 2008
Accepted 29 September 2008
Available online 10 October 2008
Keywords:
Toxic Oil Syndrome
Deodorisation
Anilides
3-N-Phenylamino-1,2-propanediol esters
Computational studies
Reaction mechanism
1. Introduction
considered better toxic oils biomarkers.10 These PAP derivatives
were not detected in the oil batches distributed in Catalonia, which
Toxic Oil Syndrome (TOS) was an epidemic disease that
appeared in central Spain in 1981, causing over 400 deaths and
affecting more than 20,000 people.1,2 Epidemiological data
revealed that the disease was associated to the consumption of
rapeseed oil denatured with aniline, illegally refined at the ITH oil
refinery in Seville, mixed with others oils and sold as a grocery in an
itinerant sale.3 Several aniline derivatives were eventually detected
in oil batches (Scheme 1). Fatty acid anilides were first identified (in
concentrations among 500–2000 mg/L) and characterised in the
suspected toxic oils.3 However, their association with TOS was not
clearly established. These fatty acid anilides were also present in
adulterated oils distributed in other areas (Catalonia) with no
toxicity reports among consumers4 and they were established as
adulterated toxic oil biomarkers.3–5 Noticeably, aniline derivatives
different from anilides were only found in toxic oil batches.6 These
compounds, identified as 3-(N-phenylamino)propane-1,2-diol
(PAP) and its mono-, di- and triacyl derivatives6–10 (mPAP, dPAP
and tPAP, respectively, Scheme 1), have been subsequently
suggests that something unique occurred to the aniline-denatured
oils in the ITH refinery of Seville that produced the toxic oils. Un-
fortunately, the efforts carried out to identify the compounds that
caused the intoxication have not been conclusive. In any case, the
available data point to PAP esters or their metabolism products as
Scheme 1. Compounds associated with TOS: anilides as biomarkers of the adulterated
oils and PAP derivatives as biomarkers of the toxic oil. R1CO, R2CO, and R3CO are fatty
acyl residues.
* Corresponding author. Tel.: þ34 934006121.
E-mail address: angel.messeguer@iiqab.csic.es (A. Messeguer).
0040-4020/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tet.2008.09.102

J. Escabro´s et al. / Tetrahedron 65 (2009) 418–426
419
the most probable species responsible for eliciting the toxic
effects.11–17
It is accepted that the agents that caused the TOS were gener-
ated during the refining of the denatured oil, in particular during
the final deodorisation process. However, the chemical routes ac-
counting for the generation of the fatty acid anilides and the dif-
ferent PAP derivatives had not been clarified. In the deodorisation
step the oils are subjected to steam-distillation under vacuum to
eliminate stinking compounds. During this process, abnormally
high temperatures can be reached in parts of the tank, thus
favouring the generation of undesired products. It is now accepted
that deviations from the standard deodorisation conditions must
have occurred causing the formation of PAP derivatives.18 These
authors observed that PAP esters were formed only when the oil
was heated at 300 ꢀC but not at 250 ꢀC. Interestingly, PAP esters
formed at 300 ꢀC were lost during the processing at that
temperature.
With these antecedents, we reproduced at a laboratory scale the
deodorisation process that would have been carried out at the ITH
refinery. Specifically, the production of refined aniline-denatured
rapeseed oils with a composition of aniline derivatives similar to
that found in the toxic oil batches was attempted.14 This study fo-
cused on the influence of different parameters on the formation of
model anilide and PAP derivatives from oleic acid (OA, OPAP and
OOPAP, respectively), the interactions between any two of these
variables and the occurrence of multivariable interactions. Of
particular interest was the interaction observed between OOPAP
and OA since these compounds have been associated to the toxic
response. Specifically, we observed that the diester derivatives
were unstable under the deodorisation conditions giving rise to the
generation of the corresponding anilides.
Scheme 2. Different reaction courses postulated for the nucleophilic attack of aniline
on triglycerides to generate anilides or PAP diesters during the rapeseed oil model
deodorisation process. R1CO, R2CO, and R3CO are fatty acyl residues.
The second path involves a gem-diol intermediate: the first step
consists in the addition of water followed by proton transfer to the
carboxylic oxygen to give the gem-diol; then, this intermediate
decomposes and water is released. Both mechanisms can be
catalysed by water molecules that assist in the proton transfer step.
Water helps by generating less strained five- and six-membered
rings transition states.
The study of the decomposition of PAP esters resulted into a set
of reaction mechanisms analogous to those described above. The
peculiarity is that the functional groups involved are in the same
molecule, which means that all processes are intramolecular.
Another fundamental difference with aminolysis reactions is that
the reactivity of PAP derivatives takes place in oil, a non-polar,
non-protic solvent, very different from water or amines. These
intramolecular transformations and an additional bimolecular
mechanism will be discussed in detail to explain the formation of
anilides from PAP derivatives.
The present contribution reports the study of the thermal
instability of PAP derivatives by using both experimental and
theoretical approaches. This study was addressed to propose
a mechanism of such a transformation and its potential relevance to
account for the toxicity effects elicited by the original oil batches.
2.1. Thermal stability studies on PAP derivatives
2. Results and discussion
Although rapeseed oil is formed by a complex mixture of fatty
acids and glycerides, oleic acid is the most abundant component.
For this reason the oleanilide (OA), and the mono and dioleyl
ester of PAP (O(1)PAP and OO(1,2)PAP, respectively) were se-
lected to carry out part of our studies. Analogous PAP derivatives
(HA, H(1)PAP and HH(1,2)PAP), containing a shorter fatty acyl
residue (hexanoyl) to find out whether the length of the acid
chain could influence on the thermal stability of these com-
pounds, were also selected (Scheme 3). Finally, compounds 1–3
lacking one of the hydroxy moieties of PAP derivatives completed
the set studied.
The thermogravimetry analysis profiles of some of the above
compounds are shown in Figure 1. While anilide OA and PAP were
thermostable until 300 ꢀC, compounds like monoesters O(1)PAP
and H(1)PAP, as well as diesters OO(1,2)PAP and HH(1,2)PAP
showed two slopes (more clearly differentiated for O(1)PAP and
HH(1,2)PAP), thus indicating that the thermal decomposition of the
compounds took place in two stages. The first one coincides with
the split of the molecule to give the corresponding anilide and the
second one with the evaporation of the generated by-product.
Similar results were observed for compounds 1–3 (Fig.1 shows only
the profile for 3).
Our previous studies on the generation and interconversion of
anilides and PAP derivatives during the deodorisation process
showed a complex picture of transformations. Triglycerides present
in the rapeseed oil containing 2% aniline reacted with this amine to
give either PAP esters or anilides depending upon where the nu-
cleophilic attack of aniline takes place (Scheme S1, Supplementary
data). While these anilides are essentially stable under the high
temperatures maintained during the deodorisation process, PAP
esters were not, being then converted mostly into the corre-
sponding anilides (cf. Scheme 2).14
This decomposition reaction should bear resemblances with the
formation of amides by aminolysis of esters, a model reaction for
the synthesis of peptides from amino acids in pre-biotic condi-
tions.19 A similar reaction is also the hydrolysis of amides, of
interest when studying proteolytic enzymes.20 Mechanism studies
still disagree the intermediates involved in both reactions, but it is
accepted that different paths are possible. Of special concern is the
formation of the zwitterion form that arises after the initial attack
of the nucleophilic species to the carboxy moiety. The results are
controversial because the existence of this zwitterionic in-
termediate has been proved by some authors,21,22 but questioned
by others.20 In any case, this zwitterion species should be quite
unstable and would not compete with the major reaction paths. For
the hydrolysis of amides, these main pathways include a proton
rearrangement accompanying the C–Owater bond formation. Two
reaction mechanisms have been postulated for this transformation.
The first one is a concerted bond formation and amine elimination.
In another set of experiments, monoesters O(1)PAP and
H(1)PAP, diester HH(1,2)PAP, and aminoester 3 were subjected to
different constant temperatures (200, 220, 240, 260 and 280 ꢀC)
during a given period of time, and the final residue was analysed by
HPLC. In all cases the respective anilides from oleic and hexanoic
acid acids were detected in the assays conducted at 260 and 280 ꢀC.
Taken together, these results confirmed the observations from the

420
J. Escabro´s et al. / Tetrahedron 65 (2009) 418–426
Scheme 3. Set of PAP derivatives and related compounds selected to analyse their thermal stability by thermogravimetry and calorimetry. HA: hexaneanilide; OA: oleanilide.
In PAP esters fatty acyl moieties RCO are referred to hexanoyl (H) or oleyl (O) residues. Compounds 1–3 bear hexanoyl residues.
deodorisation process: PAP mono and diesters are thermally un-
stable giving rise to the generation of the corresponding anilides
and this instability is not dependent on the length of the fatty acid
chain. On the other hand, it was also observed that compound 3, in
spite of bearing a simplified structure than PAP derivatives, showed
the same thermal behaviour, which suggested a mechanistically
related pathway to account for the anilide formation. At this point
we concluded that these experimental data required the support
from theoretical studies.
model and the structural simplifications carried out in comparison
with the real system.
Different reaction mechanisms for the decomposition of 4 have
been found and some of them explain the formation of the detected
product 5. The first set of mechanisms leads to the formation of an
anilide 6 with steps analogous to the aminolysis of esters. A second
possible mechanism is a direct path from 4 to 5, avoiding the anilide
formation (cf. Scheme 4a). Finally, anilide 6 can suffer a nucleophilic
attack from a solvent molecule to give also 5 through a bimolecular
mechanism (Scheme 4b). We will see that only the energetics of the
later path is in agreement with the experimental results.
2.2. Computational study of the thermal stability of PAP
derivatives
2.3. Anilide 6 formation and its decomposition: TS1–TS7
(Scheme 4a and Fig. 2)
The formation of anilides from PAP esters or from 3 but not from
PAP itself deserves further analysis. In particular, we explored
a reaction mechanism that could explain the formation of OA or HA
from pure PAP esters samples under the high temperature
conditions. We built a model system (see Section 4.3) where the
anilides OA and HA are modelled as 5 (Scheme 4a) and the starting
ester is substituted by 4. Although precise kinetic data for the for-
mation of anilide 5 (or OA, HA) were not available, its experimental
appearance in time scale of minutes at around 530 K gives an order
of magnitude for the reaction time of 4. At a temperature of 535 K,
reaction barriers of 35, 40 and 45 kcal/mol give reaction times of
0.005, 0.55, 60 h, respectively (see Section 4.3). At 555 K these rates
are reduced to 0.0015, 0.13 and 12 h, respectively. These results
suggest that barriers larger than 45 kcal/mol cannot explain the
formation of 5 in the experimental measured times and those lower
than 35 kcal/mol would result in 4 being too unstable even to be
detected, in disagreement with thermogravimetry results. There-
fore the reaction energy barriers that are expected lie in the range
of 35–45 kcal/mol. This range can vary a few kcal/mol owing to the
limitations of the quantum chemistry method used, the solvation
A nucleophilic attack of the amine nitrogen to the carbonyl
group is a reasonable first step in the decomposition of 4. The
product of this reaction would seem to be a cyclic zwitterion. All
attempts to locate the zwitterion were unsuccessful. This is in
agreement with previous work showing that the zwitterion for the
aminolysis of esters is highly unstable; even in water its existence is
not established. 19,20 The oil used in our studies is much less polar
than water; therefore, in such media, zwitterions and ionic species
will be destabilised. Interestingly, two paths that are equivalent to
the anilide formation in water were found, TS1 and TS2þTS3.
Anilide 6 is an intermediate of our complete reaction mechanism; it
is 5.6 kcal/mol more stable than aminoester 4 and therefore it could
be detected experimentally as N-acylated PAP derivative. In fact, N-
acylated derivatives were detected in the model deodorisation
experiments with doped rapeseed oil.14 The direct path that leads
to anilide 6 is TS1. This transition state is 58.3 kcal/mol above from
4. This transition state has a peculiar geometry, with two five-
membered rings, corresponding to the rearrangement of the acyl
%
%
%
%
100
100
100
100
-
-
80
60
40
20
0
80
60
40
20
80
80
60
40
20
0
60
40
20
0
OA
PAP
H(1)PAP
O(1)PAP
100 200 300 400 500
0
100 200 300 400 500 °C
100 200 300 400 500 °C
100
200
%
300
°C
°C
%
%
100
100
100
80
60
40
20
0
80
60
40
20
0
80
60
40
20
HH(1,2)PAP
OO(1,2)PAP
3
0
100
200
300
°C
100 200 300 400 500 600 °C
100
200
300
400 °C
Figure 1. Thermogravimetric analysis of OA, PAP, H(1)PAP, HH(1,2)PAP, O(1)PAP, OO(1,2)PAP and compound 3.

J. Escabro´s et al. / Tetrahedron 65 (2009) 418–426
421
and the proton in opposite directions (Fig. 2). The energy of this
barrier is too high to explain the formation of 6. Its zwitterionic
character makes its solvation energy larger than that for reactants
(see Table 1); however, solvation effects do not contribute
substantially to its stabilisation.
A two-step process, equivalent to the gem-diol formation in the
hydrolysis of amides was then sought. In our case, the gem-diol
corresponds to the cyclic hemiacetal 7 (Scheme 4a). TS2 results
from the nucleophilic attack of the amine onto the carboxy moiety,
but instead of giving rise to the zwitterion, a proton is concertedly
rearranged in a four-membered TS to the carbonyl oxygen (Fig. 2).
This TS step has a barrier of 42.6 kcal/mol; therefore intermediate 7
can be formed under the deodorisation experimental conditions.
This intermediate is less stable than the reactants and its concen-
tration will remain very small and probably beyond the sensitivity
of the experimental detection methods. The formation of species 6
from 7 takes place through TS3. This TS has an energy higher than
TS2. In this step, the C–O bond is broken while the proton is
transferred back, again in a four-membered ring (Fig. 2). The overall
path TS2þTS3 has a much lower energy barrier than TS1.
The presence of proton donor–acceptor molecules can act as
catalyst and release the strain of both TS2 and TS3 four-membered
rings. This is a process that has been thoroughly studied with
water20,23–25 but is beyond the scope of the present work. In our
system, a different molecule should play the role of water. It is
conceivable that any PAP derivative can act as a proton donor–ac-
ceptor, as well as the carboxylic group of a fatty acid. Even if this
catalytic effect takes place, it will not change the picture of the
global reaction path. Indeed, the formation of 6 is the most
favourable path; consequently, if catalytic effects accelerate its
formation, it will remain the main intermediate along the
decomposition path of aminoester 4.
The barriers for the formation of anilide 6 are higher than those
previously reported for the aminolysis of carboxylic acids.20,22 This
difference may be due to: first, the carbon atom in the ester group
of 4 is less electrophilic than in a carboxylic group because the alkyl
Scheme 4. Different intramolecular (a) and bimolecular (b) pathways proposed for the
formation of anilide 5 from the model aminoester 4.
Figure 2. Transition state geometries TS1–TS8 for the reaction mechanisms described in Scheme 4. Some relevant distances are depicted in angstrom.

422
J. Escabro´s et al. / Tetrahedron 65 (2009) 418–426
Table 1
Energies, Gibbs free energies and solvation energies (kcal/mol) for the transition
states, intermediates and products depicted in Scheme 4a
D
E/b2
D
G
DDG (solv)
4
0
0
0 (ꢁ3.6)
ꢁ1.2
ꢁ0.4
ꢁ1.0
1.5
0.8
0.2
0.0
5þacetone
ꢁ8.5
22.2
3.4
ꢁ5.8
11.1
0.8
ꢁ4.3
28.0
8.2
ꢁ5.6
17.8
4.9
5þepoxide
5þenol
6
7
E-8þacetone
Z-8þacetone
0.0
4.3
TS1
56.6
41.6
41.8
79.6
52.9
80.1
75.1
66.8
71.9
31.8
58.3
42.6
45.2
77.6
47.5
74.2
69.9
64.3
69.6
33.2
ꢁ4.1
TS2
TS3
TS4
TS5
TS6
TS7
TS8
TS9
0.7
1.1
ꢁ1.0
2.1
1.3
1.6
0.8
ꢁ1.4
TS10þacetone
0.4
substituent on the oxygen is an electron donor; second, the aniline
in 4 is less nucleophilic than an amine; and last, in our study the
reaction takes place in a single molecule and thus, strain features
become more involved.
Postulation of intermediate 7 makes possible other reaction
paths. One of them is a concerted cleavage of 7 to give acetone and
an imidic acid (Z-8) in its Z configuration, which is a tautomer of
anilide 5. The structure TS4 corresponds to this concerted path
(Fig. 2). This path is energetically very expensive and will not
contribute to the reactivity of the system. The rearrangement of the
formal hydride in the three-membered ring is probably the cause of
such a high barrier.
Figure 3. Transition state geometries TS9, TS10, and TSbi for the reaction mechanisms
described in Scheme 4. Some relevant distances are depicted in angstrom.
also possible, affording the same products as TS8. TS9 is a complex
TS, where five bonds are being formed or cleaved: C–H, C–O and C–
N bonds are cleaved while new C–H and C–N bonds are formed.
Interestingly, the nucleophilic attack precedes all other reorgan-
isation processes. The new C–Nanilide bond is already formed at the
TS, at a distance of 1.47 Å, and it is the separation of the anilide
fragment 5 that brings a high barrier (69.9 kcal/mol). This high
barriers are expected from the large number of bonds being cleaved
or formed.26 In general, a step-wise mechanism will have lower
barriers and this is what is found when comparing TS9 with the
path through TS8. Therefore the mechanism though TS9 is
kinetically irrelevant.
The same products Z-8 can be reached from anilide 6 through
TS5. The barrier for this path is 47.5 kcal/mol. A similar TS, where
the hydrogen is transferred to the nitrogen is TS6 (Fig. 2). This TS
avoids the three-membered ring strain and leads to an enol instead
of acetone. Even so, its energy barrier is much higher (74.2 kcal/
mol) than for TS5, which suggests that it is not the ring strain but
the electronic redistribution the cause of such high barriers. If, in-
stead of transferring the hydrogen attached to the carbon, it is the H
of the hydroxyl that is transferred (TS7, Fig. 2), anilide 5 and the
corresponding epoxy derivative are formed. At the transition state,
the epoxide formation is concomitant with the C–N bond cleavage.
As expected, the large reorganisation is accompanied by a rather
high barrier for TS7. The last TS found for the decomposition of 6 is
TS8 (Fig. 2). For this path an energy barrier of 64.3 kcal/mol was
computed. Again, it is a different combination of hydrogen transfer
and bond cleavage, analogous to TS5, but transferring the hydrogen
to the nitrogen instead of to the anilide oxygen. The high energy of
these transition states combined with the fact that their geometry
show a proton that is advanced in the reaction progress suggest
that it is not the proton transfer that generates the energy barrier,
but the cleavage of the C–N bond and the reorganisation of these
electrons from the sigma to the pi system.
2.5. Bimolecular mechanism
At this point, it seemed that all paths that could lead to the
anilide 5 had barriers that cannot explain the fast decomposition of
the starting aminoester
4 at the temperature range of the
deodorisation process. All barriers appeared to be high because
they involved strained transition states or they needed to cleave the
C–N bond and reallocate the electrons to the
p system. In other
words, the amine being a good leaving group, there is no
nucleophile capable to displace it directly from the carbon atom at
which it is attached. It can be argued that a solvent molecule can
release the strain of many of the TSs and thus catalyse the reaction.
But there are two problems with this conjecture. First, none of the
species present in the media is as good as a donor–acceptor as
water. Even water does not usually lower the barriers from about
65 kcal/mol to 45 kcal/mol, the value that would be required for
many of the transition states discussed above. Secondly, in many
mechanisms it is not the proton transfer that is contributing more
to the barrier but the cleavage of the C–N bond. Thus, it is difficult to
envisage how a solvent molecule (in our case one of the reactants)
could act on the reaction as an efficient catalyst.
Taken together, after exploring exhaustively the reactivity of
intermediate 6, none of the paths can explain the generation of 5.
The reaction barriers involved are too high to be surmounted
within a reasonable period of time at the working temperature
range of 535–555 K reached during the deodorisation process of
the model of toxic rapeseed oil.
2.4. One-step path: TS9 (Fig. 3)
As an alternative to the mechanism that gives the anilide
intermediate, a concerted reaction path operating through TS9 is
As an alternative path, the elimination of an anilide 5 from
intermediate 6 can be triggered by the nucleophilic attack at the

J. Escabro´s et al. / Tetrahedron 65 (2009) 418–426
423
carbon atom in a SN2 type reaction. Aniline has been selected as the
most probable nucleophile present in the reaction medium, but
other aniline derivatives could play the same role. The transition
state TSbi (Scheme 4b and Fig. 3) corresponds to this process. The
IRC from this transition state shows that, well after the saddle point,
a proton transfer takes place form the protonated aniline to the
imidate oxygen of the leaving group. Thus the product is not the
anilide 5 but its tautomer, the imidic acid 8. This acid can exist in
both Z and E configurations. In this reaction, the isomer formed has
the hydroxyl group in E conformation with respect to the nitrogen
lone pair (E-8). As discussed below, this conformation makes
possible a fast conversion to the anilide 5. The energy barrier for
this step is 41.3 kcal/mol.
Scheme 5. Proposal for the evolution of an intermediate species via a five- or six-
membered ring in the formation of anilides from PAP derivatives. In the calculations,
R1¼R2¼CH3.
Figure S1 illustrates the free energy profile. As expected, the
conformational space for the six-membered chain is longer than
the one for five-membered chain, which lowers its free energy in
comparison with that of the five-membered one. This effect is,
however, very small: integration of the free energy for the two
basins gives only 0.3 kcal/mol of extra conformational flexibility for
the six-membered ring. On the other hand, the transition state for
the five-membered ring is higher in 1 kcal/mol, counteracting the
effect of the entropy. The ratio in rate constants is exp(DDGz/RT),
where DDGz represents the difference in activation energies. In
our case, this difference is that arising from the barrier (1 kcal/
mol) minus the difference arising from the larger conformational
space (0.3 kcal/mol). The result gives a rate constant for the
six-membered process that is 1.9 times higher than for the five-
membered, at 535 K. This is in good agreement with the experi-
mental results in our group,14 and it is a result of opposite entropy
and enthalpy effects.
2.6. Conversion of imidic acid to anilide
Several reaction paths discussed above yield the imidic acid 8
instead of anilide 5. The conversion between these species is
possible via a proton tautomerism. This mechanism has been
studied27,28 and can be catalysed both by acids and bases. A direct
mechanism is possible if the nitrogen lone pair and the oxygen are
in the cis orientation, as the result of TSbi. Conversely, transition
states TS4 and TS5 lead to the trans configuration (Z-8). A trans–cis
conversion has a high barrier because of the double bond character
of the N–C bond. The kinetically relevant path is through TSbi,
which gives E-8. In this molecule, the proton and the lone pair are
facing the same direction and a direct path is then possible, via
TS10. As can be seen in Figure 3, the proton is transferred along the
plane while the electrons rearrange in the
p system. The barrier for
this process with respect to E-8 is 28.9 kcal/mol and it has an
exothermicity of ꢁ8.6 kcal/mol. This means that under the
deodorisation experimental conditions the imidic acid E-8 will
rearrange to the anilide 5 without the need of the participation of
a solvent molecule. This tautomerism will probably be much faster
in a protic solvent. This is because the participation of a protic
solvent molecule can reduce the barrier by releasing the tension of
the transition state. This is a process less likely to happen in the
solvent used in this work, but in any case it would result in a faster
equilibration of the tautomerism. The unimolecular mechanism
reported has energy barriers in agreement with other tautomeric
processes involving imidic acids, such as those occurring in
formamide, with a (potential energy) barrier of 33.6 kcal/mol29 and
formohydroxamic acid, with a (potential energy) barrier of around
35 kcal/mol.30
3. Conclusions
The results obtained herein have established new aspects of the
chemical behaviour of several aniline derivatives during the model
deodorisation process designed to imitate what occurred with the
rapeseed oil batches that caused TOS intoxication. The thermal
instability of PAP diesters due to the high temperatures achieved
during the deodorisation process led to their conversion into the
corresponding anilides. Therefore, PAP derivatives had to be con-
sidered as a new source of anilides not taken into account in
previous studies.
A reaction mechanism to explain the thermal conversion of PAP
derivatives in anilides has been established. It has been supported
by theoretical studies and experimental results. This mechanism
involves the combination of an intramolecular and intermolecular
process. Figure 4 depicts the overall suggested reaction mechanism.
First, from the PAP derivative, an intermediate amide 6 is formed.
This intermediate is the result of the intramolecular reaction of the
amino group of PAP derivative with the secondary ester of the same
2.7. Six-membered ring versus five-membered ring
Because PAP diesters have two carboxylic groups susceptible of
nucleophilic attack, two possible products can be formed. In one
case it results in 7 through TS2. The other attack would result in
a six-membered ring equivalent to 7 through TS20 (Scheme 5). Five-
and six-membered rings have similar stability and the product
distribution depends on the peculiarities of each reaction.31 Pre-
vious experimental work in our group detected a ratio 2:1 of
product yield resulting from the attack to the closer carboxylic
carbon (TS2) with respect to the more distant one (TS20).14
Sampling techniques to assess the flexibility of the reactant are
needed to predict the product ratio, and this was done by Umbrella
Sampling, as described in Section 4.3. Only TS2 was compared be-
cause this is the initial step that decides the branching of the re-
action mechanism, i.e., whether the attack gives rise to a five- or
a six-membered ring (Scheme 5). It can be assumed that all the
other steps for the six-membered ring will have very similar bar-
riers as those described previously. The model used for the
six-membered ring is depicted in Scheme 5.
ΔG / kcal mol-1
45.2
TS3
42.6
TS2
35.7
aniline
33.2
TSbi
TS10
17.8
+9
7
8.3
0.0
-0.2
E-8 + 9
-5.6
4
5 + 9
6
Figure 4. Overall reaction profile for the formation of 5 from 4. Only the kinetically
relevant mechanism is depicted. For other non-contributing paths, see Scheme 4 and
Tables 1 and 2.

424
J. Escabro´s et al. / Tetrahedron 65 (2009) 418–426
compound. Once formed, this intermediate reacts with the aniline
present in the reaction medium to form the E isomer of an imidic
acid E-8, which finally tautomerises to render the expected anilide.
The formation of anilides through this mechanism explains the
concomitant formation of oligomeric by-products that could
remain in the insoluble resinous residues generated in the deo-
dorisation experiments with model oil batches. This fraction was
investigated in model oils without satisfactory results regarding its
compound composition.14
Taken overall, the formation of anilides and PAP derivatives
from aniline and triglycerides during the deodorisation process
can be explained by a complex set of chemical conversions
(Fig. S2, Supplementary data). In this context, the mechanistic
studies described herein confirm a new path for the generation
of anilides from the thermal decomposition of PAP esters. In-
terestingly, the profile shown in Figure S2 might explain the
differences of toxicity observed between the Catalonian and
Seville refined oils. The absence of PAP derivatives in Catalonian
oils could be due to potential high temperatures reached during
the deodorisation, which could cause the progressive
decomposition of PAP esters present in the adulterated oil to
render the corresponding anilides and by-products. These
extreme thermal conditions could have been present until the
complete elimination of free aniline, thus preventing the
generation new of PAP derivatives from the reaction of the amine
with triglycerides. In contrast, in the Seville oils, the deodor-
isation could have been carried out under not so extreme con-
ditions, which conditioned that PAP esters could remain in some
extent in the final refined oil.
4.2. Synthesis of 3-(phenylamino)propyl hexanoate (1)
4.2.1. 3-(Phenylamino)propanol
Aniline (400 ml, 4.38 mmol) was added dropwise to 3-cloro-1-
propanol (410 mg, 4.33 mmol). When the addition was completed,
the mixturewas stirred under reflux for 11 h (HPLC monitoring). The
crude reaction mixture was purified by flash chromatography (9:1
Et2O/CH2Cl2) to give the expected compound as a colourless oil
(80 mg,12%).1H NMR
d
7.17 (t, J¼7 Hz, 2H, CHAr), 6.71 (t, J¼7.5 Hz,1H,
CHAr), 6.62 (d, J¼7.5 Hz, 2H, CHAr), 3.75 (t, J¼6 Hz, 2H, CH2OH), 3.23
(t, J¼6.5 Hz, 2H, CH2NHPh),1.83 (q, J¼6 Hz, 2H, CH2CH2OH).13C NMR
d
142.2 (C-1 aniline),129.2 (C-3, C-5 aniline),117.6 (C-4 aniline),113.1
(C-2, C-6 aniline), 61.3 (CH2OH), 41.8 (CH2NHPh), 31.7 (CH2CH2OH).
HRMS: m/z 152.1075; m/z for (MþH)þ C9H14NO requires 152.1066.
4.2.2. 3-(Phenylamino)propyl hexanoate (1)
A mixture of 3-(phenylamino)propanol (39 mg, 0.25 mmol),
oleic acid (328 mg, 0.28 mmol), DCC (61 mg, 0.29 mmol) and DMAP
(5 mg, 0.04 mmol) in CH2Cl2 (2 mL) was allowed to react at room
temperature. When the reaction was completed (3 h, TLC moni-
toring), the solvent was evaporated to dryness and the residue
obtained was dissolved in hexane to induce precipitation of the
urea derivative. Purification of the crude reaction mixture by flash
chromatography (silica gel, 8:1 hexane/EtOAc) afforded the
expected ester 1 as a colourless oil (46 mg, 72% yield). 1H NMR
d 7.17
(t, J¼7.5 Hz, 2H, CHAr), 6.70 (t, J¼7.5 Hz, 1H, CHAr), 6.61 (d, J¼8.5 Hz,
2H, CHAr), 4.20 (t, J¼6.5 Hz, 2H, CH2OCO), 3.77 (1H, NH), 3.22 (t,
J¼6.5 Hz, 2H, CH2NH), 2.32 (t, J¼7.5 Hz, 2H, CH2CO), 1.94 (q,
J¼6.5 Hz, 2H, CH2CH2CO), 1.63 (q, J¼7.5 Hz, 2H, CH2CH2CH2CO),
1.34–1.29 (4H, CH2), 0.90 (t, J¼7 Hz, 3H, CH3). 13C NMR
d 173.9 (CO),
148.0 (C-1 aniline), 129.3 (C-3, C-5 aniline), 117.4 (C-4 aniline), 112.8
(C-2, C-6 aniline), 62.1 (CH2OCO), 40.7 (CH2NHPh), 34.3 (CH2CO),
31.7 (CH2CH2OH), 28.6 (CH2), 24.7 (CH2), 22.3 (CH2), 13.9 (CH3).
HPLC-MS: 250.1 (MþH)þ. EIMS m/z (%): 249.13 (50) molecular ion,
132.04þ(20), 106.04 (100), 77.03 (30). HRMS: m/z 250.1807; m/z for
(MþH) C15H24NO2 requires 250.1826.
4. Experimental section
4.1. General
Aniline, oleic acid, hexanoic acid, N,N-(dimethylamino)pyridine
(DMAP) and N,N-dicyclohexylcarbodiimide (DCC) were from
Aldrich, Germany. Solvents were of HPLC grade and purchased from
Merck, Germany. The synthesis of PAP, OPAP, OOPAP and OA has
been described elsewhere.32 Anilide from hexanoic acid was
available at the laboratory. The NMR spectra were recorded on
a Varian Inova 500 (499.08 MHz for 1H and 125.67 MHz for 13C).
Spectra were taken in neutralised CDCl3 solutions unless otherwise
4.2.3. N-Hexanoyl-3-(phenylamino)propanol (2)
Hexanoyl chloride (95 ml, 0.60 mmol) and triethylamine (45 ml,
0.32 mmol) were added to a solution of 3-(phenylamino)propanol
(52 mg, 0.34 mmol) in CH2Cl2 (1 mL) and the mixture was stirred
for 1 h at 25 ꢀC (HPLC monitoring). After the elimination of the
solvent under vacuum, the residue was purified by semi-pre-
parative HPLC eluting with 15:85 H2O/acetonitrile to yield the
indicated. Chemical shifts (d) are given in parts per million relative
to tetramethylsilane (1H, 0.0 ppm) or CDCl3 (13C, 77.0 ppm). The LC/
MS analyses were carried out with a Hewlett–Packard Series 1100
LC/MSD or 1090LC/1100MSD apparatus equipped with an electro-
spray source working in positive ion mode. The GC/MS analyses
were carried out with a ThermoFinnigan Trace GC/MS apparatus
equipped with an electron ionisation detector (70 eV). Analytical
HPLC samples were run on an HP series 1100 system equipped with
expected compound (32 mg, 40% yield). 1H NMR
d
7.44 (t, J¼9.5 Hz,
2H, CHAr), 7.38 (t, J¼7.5 Hz, 1H, CHAr), 7.14 (d, J¼4.5 Hz, 2H, CHAr),
3.87 (t, J¼6.0 Hz, 2H, CH2OH), 3.65 (2H, CH2NCO), 2.05 (t, J¼7.5 Hz,
2H, CH2CO), 1.66 (qu, J¼5.5 Hz, 2H, CH2CH2CON), 1.57 (qu, J¼7.5 Hz,
2H, CH2CH2OH),1.25–1.14 (4H, CH2CH2CH2CH2CON), 0.83 (t, J¼7 Hz,
3H, CH3). 13C NMR
d 174.8 (CO), 142.0 (C-1 aniline), 129.9 (C-3, C-5
aniline), 128.1 (C-4 aniline), 128.0 (C-2, C-6 aniline), 58.1 (CH2OH),
45.6 (CH2NH), 34.1 (CH2CON), 31.3 (CH2CH2OH), 29.9 (CH2), 25.2
(CH2), 22.3 (CH2), 13.9 (CH3). HPLC-MS: 521.5 (MþMþNa)þ, 272.2
(MþNa)þ, 250.2 (MþH)þ. HRMS: m/z 250.1807; m/z for (MþH)þ
C15H24NO2 requires 250.1822.
a direct-phase column (Nucleosil 100-5CN, 5
Scharlau, Spain). Elution conditions: hexane/2-propanol mixtures
at 1 mL/min. UV detection was set at
¼245 and 270 nm. A Biotage
system (Dyax Corp., Cambridge, MA) equipped with silica gel Kpsil
(32–63 m, 60 Å) columns was employed for flash chromatography
purifications. The preparative HPLC purifications were performed
by semi-preparative HPLC using a Kromasil C8 (25ꢂ2 cm, 5 m)
column, CH3CN/H2O mixtures containing 0.1% TFA as mobile phases
and a flow rate of 5 mL/min. UV detection was set at
mm, 150ꢂ4 mm from
l
m
4.2.4. 2-(Phenylamino)ethyl hexanoate (3)
m
A mixture of 2-(phenylamino)ethanol (200 mg, 1.46 mmol),
hexanoic acid (180 mg, 1.55 mmol), DCC (332 mg, 1.61 mmol) and
DMAP (16 mg, 0.13 mmol) in CH2Cl2 (4 mL) was allowed to react at
room temperature. When the reaction was complete (3 h, HPLC
monitoring), working-up and purification of the crude reaction
mixture as described above for 1 afforded pure 5 (273 mg, 80%
l¼220 nm.
High resolution mass spectra (HRMS-FAB) were carried out at the
Mass spectrometry Service of the University of Santiago de Com-
postela (Spain). The elemental analyses were performed at the
IQAC Microanalysis Service. Thermogravimetry studies were per-
formed with a Mettler Toledo TG50 apparatus and calorimetry
studies with a DSC821 apparatus.
yield). 1H NMR
d
7.19 (t, J¼7.5 Hz, 2H, CHAr), 6.73 (t, J¼7.5 Hz, 1H,
CHAr), 6.64 (d, J¼7.5 Hz, 2H, CHAr), 4.29 (t, J¼5.5 Hz, 2H, CHHOR),
3.89 (s, 1H, CNHAr), 3.40 (q, J¼5.5 Hz, 2H, CHHN), 2.32 (t, J¼7.5 Hz,

J. Escabro´s et al. / Tetrahedron 65 (2009) 418–426
425
2H, COCH2), 1.60–1.64 (2H, COCH2CH2), 1.40–1.20 (4H, CH2), 0.88 (t,
Potentials of mean force were calculated with the Umbrella
Sampling method combined with the Weighted Histogram Analysis
Method.37,38 Due to the high computational cost of the molecular
dynamics simulations needed for this technique, the semi-empir-
ical Hamiltonian AM1 was used. This should be reliable enough
because only energy differences for the same mechanism were
compared. These calculations were performed with the DYNAMO
suite of subroutines.39,40 As a reaction coordinate (rc) to build the
free energy profile or potential of mean force, the following com-
bination of distances was used rc¼d(C–N)þd(H–O)ꢁd(H–N).
J¼6.5 Hz, 3H, CH3). 13C NMR
d 173.9 (CO), 147.6 (CHAr), 129.3
(2CHAr), 117.8 (CHAr), 112.9 (2CHAr), 62.8 (CH2OCOR), 42.9 (CH2NH),
34.1 (CH2), 31.3 (CH2), 24.6 (CH2), 22.3 (CH2), 13.9 (CH3). Anal. Calcd
for C14H21NO2: C, 71.46%; H, 8.99%; N, 5.95%. Found: C, 71.49%; H,
9.15%; N, 6.13%.
4.3. Computational details
The aim of the computational work is to derive a mechanism for
the formation of anilides HA and OA from PAP derivatives such as
m(2)PAP or PAP analogues such as 3 (see Section 2). Because the
length of substituents was not relevant for the formation of the
anilides, aminoester 4 was selected as model PAP surrogate, and R1
and R2 were chosen as methyl groups.
Acknowledgements
This work was supported by FISAT-WHO Grants. We thank Josep
Carilla for the thermogravimetry measurements, J.M. Anglada for
fruitful discussions and J. Luque for helpful hints in solvation cal-
All calculations were done based on the Density Functional
Theory, using the hybrid functional B3LYP.33 Stationary points were
located and characterised with frequency calculations. We checked
that minima had all positive frequencies and that Transition States
(TSs) had only one imaginary frequency. To prove that each TS
connected with the reactants and products, the Intrinsic Reaction
Coordinate (IRC) was followed. The basis set used for the optimi-
sation was 6-31G(d), which we will refer as b1. To get more reliable
energies, single point calculations on the optimised geometries
with 6-311þG(d,p) were carried out. This triple-zeta basis set will
be referred as b2. Enthalpy and free energy corrections to the b2
basis set were calculated with the thermochemistry results at
B3LYP/b1 at 533 K. This seems reasonable since the energy results
did not differ significantly with both basis sets (see Tables 1 and 2,
and Tables S1 and S2), thus indicating that potential energy
surfaces described with the different basis are very similar. All these
calculations were performed with the PC GAMESS program.34
´
culations. R.C. would like to thank the Ramon y Cajal program of the
Spanish government and MEC for financial support (Grant
CTQ2006-01345/BQU). This research has been partly performed
using the CESCA resources.
Supplementary data
Schemes S1 and S2 containing more details of reaction mecha-
nisms; Figure S1, depicting the free energy profile for the bi-
molecular mechanism; Tables S1 and S2, complementing Tables 1
and 2, are provided. Supplementary data associated with this
article can be found in the online version, at doi:10.1016/
j.tet.2008.09.102.
References and notes
1. Posada de la Paz, M.; Philen, R. M.; Abaitua Borda, I. Epidemiol. Rev. 2001, 23,
231–247.
Table 2
Energies, Gibbs free energies and solvation energies (kcal/mol) for the transition
states, intermediates and products depicted in Scheme 4b
2. Diggle, G. E. Int. J. Clin. Pract. 2001, 55, 371–375.
3. Pestan˜a, A.; Mun˜oz, E. Nature 1982, 298, 608.
4. Ventura Dı´az, L. M. Grasas y Aceites 1982, 33, 73–78.
5. Gradjean, P.; Tarkowski, S. Toxic Oil Syndrome. Mass Food Poisoning in Spain.
WHO Regional Office for Europe: Copenhagen, Denmark, 1984, pp 3–16.
DE/b2
D
G
DDG (solv)
6þaniline
TSbi
E-8þ9
0
49.7
12.0
0
41.3
13.9
0 (ꢁ5.4)
0.7
ꢁ1.1
´
´
6. Vazquez-Roncero, A.; Janer del Valle, C.; Maestro Duran, R.; Graciani Constante,
E. Lancet 1983, ii, 1024–1025.
7. Va´zquez-Roncero, A.; Maestro Dura´n, R.; Gutie´rrez, R. Grasas y Aceites 1984, 35,
15–21.
8. Va´zquez Roncero, A.; Go´mez Go´ mez, R. Grasas y Aceites 1987, 38, 87–92.
The environment effects were included with a polarisable
continuum model, PCM.35 Because the olive oil or the oleic acid had
not been parametrised, we used diethylether as solvent, which has
similar dielectric properties. As results are not significantly changed
by the solvent, its replacement is an acceptable approximation.
Solvation effects were calculated at 300 K because the parameters
for diethylether are not available at 500 K (where it is no longer
a liquid). Again, because solvent effects are small and the energy
differences between mechanisms large, the temperature changes
should not be significant. PCM calculations were performed with
Gaussian 98.36 Throughout the results, when discussing energy
barriers and exothermicity data, we will refer to free energy values
in vacuum unless otherwise stated.
´
9. Posada de la Paz, M.; Philen, R. M.; Schurz, H.; Hill, R. H. J.; Gimenez Ribota, O.;
Go´mez de la Ca´mara, A.; Kilbourne, E. M.; Abaitua Borda, I. Epidemiology 1999,
10, 130–134.
10. Schurz, H. H.; Hill, R. H.; Posada de la Paz, M.; Philen, R. M.; Borda, I. A.; Bailey,
S. L.; Needham, L. L. Chem. Res. Toxicol. 1996, 9, 1001–1006.
´
´
11. Ladona, M. G.; Bujons, J.; Messeguer, A.; Ampurdanes, C.; Morato, A.; Corbella, J.
Chem. Res. Toxicol. 1999, 12, 1127–1137.
12. Bujons, J.; Ladona, M. G.; Messeguer, A.; Morato,
´
A.; Ampurdane´s, C. Chem. Res.
Toxicol. 2001, 14, 1097–1106.
13. Morato´ , A.; Martı´nez-Cabot, A.; Escabro´s, J.; Bujons, J.; Messeguer, A. Chem. Res.
Toxicol. 2004, 17, 889–895.
14. Morato´ , A.; Escabro´s, J.; Manich, A.; Reig, N.; Castan˜o, Y.; Abian, J.; Messeguer, A.
Chem. Res. Toxicol. 2005, 18, 665–674.
15. Martı´nez-Cabot, A.; Morato´, A.; Messeguer, A. Chem. Res. Toxicol. 2005, 18, 1721–
1728.
16. Martinez-Cabot, A.; Messeguer, A. Chem. Res. Toxicol. 2007, 20, 1556–1562.
17. Reig, N.; Calaf, R. E.; Castan˜o, Y.; Prieto, R.; Messeguer, A.; Morato´, A.; Escabro´s, J.;
Transition State Theory (TST) was used to calculate rate
constants. Harmonic partition functions were used and no trans-
lational or rotational effects were included because they are only
meaningful in gas-phase. The reaction time was defined as the
inverse of the rate constant, k(T). TST expression of the reaction rate
constant is expressed as Eq. 1, where h is the Planck constant, T the
´
Gelpı, E.; Abian, J. J. Mass Spectrom. 2007, 42, 527–541.
18. Ruı´z-Me´ndez, M. V.; Posada de la Paz, M.; Abian, J.; Calaf, R. E.; Blount, B.;
´
Castro-Molero, N.; Philen, R.; Gelpı, E. Food Chem. Toxicol. 2001, 39, 91–96.
19. Chalmet, S.; Harb, W.; Ruiz-Lo´pez, M. F. J. Phys. Chem. 2001, 105, 11574–11581.
20. Gorb, L.; Asensio, A.; Tun˜o´ n, I.; Ruiz-Lo´pez, M. F. Chem.dEur. J. 2005, 11, 6743–
6753.
21. Cascella, M.; Raugei, S.; Carloni, P. J. Phys. Chem. B 2004, 108, 369–375.
22. Zahn, D. Eur. J. Org. Chem. 2004, 69, 4020–4023.
23. Crehuet, R.; Anglada, J. M.; Bofill, J. M. Chem.dEur. J. 2001, 7, 2227–2235.
24. Anglada, J. M.; Aplincourt, P.; Bofill, J. M.; Cremer, D. ChemPhysChem 2002, 2,
215–221.
25. Aplincourt, P.; Anglada, J. M. J. Phys. Chem. 2003, 107, 5812–5820.
26. Dewar, M. J. S. J. Am. Chem. Soc. 1984, 106, 209–219.
27. Eriksson, M. A.; Hard, T.; Nilsson, L. Biophys. J. 1995, 69, 329–339.
temperature, kB the Boltzmann constant, R the gas constant and DG
the activation free energy.
ꢀ
ꢁ
kBT
h
D
G
RT
kðTÞ ¼
exp
(1)

426
J. Escabro´s et al. / Tetrahedron 65 (2009) 418–426
28. Ciarkowski, J.; Chen, F. M. F.; Benoiton, N. L. J. Comput.-Aided Mol. Des. 1991, 5,
599–616.
29. Adamo, C.; Cossi, M.; Barone, V. J. Comput. Chem. 1997, 18, 1993–2001.
30. Hwa, D.-H.; Ho, J.-J. J. Phys. Chem. 1998, 102, 3582–3586.
31. Carey, F. A.; Sundberg, R. J. Advanced Organic Chemistry: Structure and Mecha-
nisms. Plenum: New York, NY USA, 2000.
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.; Ragha-
vachari, 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 03. Revision C.02. Gaussian: Wallingford, CT, USA.
37. Kumar, S.; Rosenberg, J. M.; Bouzida, D.; Swendsen, R. H.; Kollman, P. A.
J. Comput. Chem. 1992, 13, 1011–1021.
38. Roux, B. Comput. Phys. Commun. 1995, 91, 275–282.
39. Field, M. J.; Ce´line, M. A.; Proust-De, B. F.; Thomas, M. A. J. Comput. Chem. 2000,
21, 1088–1100.
40. Field, M. J. A Practical Introduction to the Simulation of Molecular Systems;
Cambridge University Press, Cambridge, UK, 1999.
´
32. Ferrer, M.; Galceran, M.; Sanchez-Baeza, F.; Casas, J.; Messeguer, A. Liebigs Ann.
Chem. 1993, 507–511.
33. Becke, A. M. J. Chem. Phys. 1993, 98, 5648–5652.
34. Granovsky, A. A. PC GAMESS version 7.0, http://classic.chem.msu.su/gran/
gamess/index.html.
35. Tomasi, J.; Mennucci, B.; Cammi, R. Chem. Rev. 2005, 105, 2999–3094.
36. 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.;