An experimental and theoretical study of the catalytic effect of quaternary ammonium salts on the oxidation of hydrocarbons

Article abstract of DOI:10.1016/j.tet.2004.09.060

The enhancement in the autoxidation of ethylbenzene by molecular oxygen in the presence of quaternary ammonium salts (QAS) was investigated from the experimental and theoretical points of view. The primary effect of the addition of QAS to the reaction medium was an increase in ethylbenzene conversion. Quantum chemical calculations, using B3LYP hybrid functional, revealed a weakening of C-O and O-H bonds of the hydroperoxide. These effects favor the formation of ethylbenzenyl and peroxyl radicals, respectively, both of which are involved in the propagation reaction that leads to the formation of hydroperoxide, the desired final product. Analysis of the electronic properties of the structures formed between reactants and catalysts offers a better understanding of the mechanism of ethylbenzene oxidation reactions. The atoms-in-molecules approach enables a rigorous description of atomic charges and bond properties that can be qualitatively related to the experimental data.

Full text of DOI:10.1016/j.tet.2004.09.060

Tetrahedron 60 (2004) 11527–11532
An experimental and theoretical study of the catalytic effect of
quaternary ammonium salts on the oxidation of hydrocarbons
*
L. Barrio, P. P. Toribio, J. M. Campos-Martin and J. L. G. Fierro
†
´
´
Instituto de Catalisis y Petroleoquımica, CSIC, Marie Curie s/n, Cantoblanco, 28049 Madrid, Spain
Received 13 April 2004; revised 2 September 2004; accepted 17 September 2004
Available online 7 October 2004
Abstract—The enhancement in the autoxidation of ethylbenzene by molecular oxygen in the presence of quaternary ammonium salts (QAS)
was investigated from the experimental and theoretical points of view. The primary effect of the addition of QAS to the reaction medium was
an increase in ethylbenzene conversion. Quantum chemical calculations, using B3LYP hybrid functional, revealed a weakening of C–O and
O–H bonds of the hydroperoxide. These effects favor the formation of ethylbenzenyl and peroxyl radicals, respectively, both of which are
involved in the propagation reaction that leads to the formation of hydroperoxide, the desired final product.
q 2004 Published by Elsevier Ltd.
1. Introduction
subsequently decomposes mainly into alkylperoxyl radicals,
or alkyl radicals (Scheme 1b). The alkyl radical species
reacts very quickly in presence of molecular oxygen to
alkylperoxyl radicals; and hence both species could produce
an increase in the selectivity to hydroperoxide product,
because it favors the propagation step.
Autoxidation reactions are acquiring increasing importance
in industrial oxidation processes for several reasons: they
use the most abundant and cheapest oxidizing reagent, and
usually mild temperatures and pressures are required. The
use of molecular oxygen as an oxidant avoids the generation
of pollutants, as occurs with stoichiometric oxidants.
However, hydrocarbon autoxidation occurs by means of a
free radical chain mechanism,1,2 that hinders control of
selectivity. One major drawback of the complex radical
mechanism of these reactions, however, is the lack of a good
understanding of the activity of the catalytic systems
employed in the reactions. Under these circumstances,
quantum chemical calculations are a powerful tool that can
help to explain the catalytic effect observed in these
processes.
The most interesting products of hydrocarbon autoxidation
The susceptibility of any substrate to autoxidation is
determined by the kp/(2kt)1/2 ratio, where kp and kt are,
respectively, the specific rate constants of the propagation
and termination reactions, which determine the length of
the propagating chain and hence the reaction rate (see
Scheme 1a). Alkylperoxyl radicals, which are indeed the
active propagating species in liquid phase, play a pivotal
role within this scheme. Along the reaction, the hydro-
peroxide species formed acts as a radical initiator, which
Keywords: Autoxidation; Ethylbenzene hydroperoxide; Quaternary
ammonium salts; Density-functional calculations; Bond activation.
* Corresponding author. Tel.: C34 915854759; fax: C34 915854760;
e-mail: jlgfierro@icp.csic.es
Scheme 1. (a) Standard autoxidation mechanism. (b) Different radical
products of the decomposition of hydroperoxide.
† http://www.icp.csic.es/eac/index.htm
0040–4020/$ - see front matter q 2004 Published by Elsevier Ltd.
doi:10.1016/j.tet.2004.09.060

11528
L. Barrio et al. / Tetrahedron 60 (2004) 11527–11532
are hydroperoxides. Hydroperoxides are used as oxidizing
agents of olefins, and as important precursors for the
synthesis of phenols and other organic by-products.
Ethylbenzene hydroperoxide is obtained by oxidizing
ethylbenzene with air in a liquid phase, and it is used in
the epoxidation step of the industrial process to obtain
propylene oxide.3
A very broad variety of homogeneous catalytic systems has
been described in the literature for this reaction.4–7 It has
been also reported that quaternary ammonium halides (and
other onium salts, i.e., sulphonium and phosphonium)
enhance the solubility of oxygen in the liquid phase, and
at the same time show catalytic activity in free radical
oxidation reactions, but their role in these processes is not
clear.5,7 In some cases, the explanations invoked to account
for the catalytic activity of these compounds point to the
possibility to produce free radicals, which is related with
the donor effect of the alkyl chains and/or the halide atom
of the salt, whereas in other cases the catalytic activity is
ascribed to oxygen activation.
According to the above, this work was undertaken with the
aim to relate the experimental data on catalytic activity to
quantum chemical calculations of the system involved in the
reaction. The combination of both studies should afford a
better understanding of the factors affecting hydrocarbon
oxidation reactions. This work will focus mainly on the
study of the influence of catalysts on the desired final
product, ethylbenzene hydroperoxide, which also plays an
important role in the initiation and propagation steps. Due to
the high complexity of autoxidation mechanism and the
high number of species present in the reaction medium, one
can take this approach as a preliminary step to understand
the effect of quaternary ammonium salts on reaction
mechanism. The usefulness of computational catalyst relays
on its ability to provide a simple and efficient tool to relate
catalyst structure with its reactivity. Further studies,
involving transition state structures for both initiation and
propagation steps, will be needed to completely identify the
effects of these salts on reaction mechanism. The analysis of
bond properties, by means of atoms-in-molecules method-
ology, offers a descriptor for the stability of the O–O bond,
which accounts for the selectivity of the reaction.
Figure 1. Experimental ethylbenzene conversion during autoxidation with
air at 506 K.
2. Results and discussion
Figure 1 shows the dependence of ethylbenzene conversion
with reaction time in the presence of quaternary ammonium
salt (QAS) catalysts. For the sake of comparison, the time
dependence of ethylbenzene conversion is also included for
the QAS-free reaction. The incorporation of QAS to the
reaction medium led to an increase in ethylbenzene
conversion with respect to the QAS-free reaction. This
increase was much more pronounced with the iodide salt
than with the hydroxide. However, the increase in
ethylbenzene conversion was accompanied by a parallel
increase in the EBHP concentration upon addition of TMAI,
but never in the presence of TMAOH (Fig. 2). The plot of
EBHP selectivity versus ethylbenzene conversion revealed
that TAMOH elicits a drop in EBHP selectivity whereas the
EBHP selectivity profile of the reference experiment did not
Figure 2. EBHP concentration versus reaction time during autoxidation
with air at 506 K.
change to any significant extent in the presence of TMAI
(Fig. 1). This indicates that EBHP selectivity remains
essentially unchanged when the reaction rate is increased,
and hence the hydroperoxide yield is enhanced (Fig. 3).
In order to explain this selectivity pattern, theoretical
calculations of the interaction between the hydroperoxide
and QAS were made. The systems investigated comprise the
individual hydroperoxide, the interaction between EBHP
and QAS salts, and also the interaction between EBHP and
the separate cations and anions (Scheme 2).

L. Barrio et al. / Tetrahedron 60 (2004) 11527–11532
11529
bulk reaction mixture. Study of the properties of these new
species in comparison with when they are absent from the
system should allow us to understand the observed changes
in catalytic performance. The bond distance ant the electron
density at the bond critical point8 (obtained by AIM
methodology) are used to estimate bond strength.
The study focused on the bonds involved in the reaction
mechanism, which were mainly Cd–Og, Og–Ob and Ob–Ha.
As reported in the Introduction, the oxidation mechanism of
ethylbenzene occurs via radical intermediates produced by
homolytic rupture of these bonds. Each bond weakening
observed in the structures was considered to favor radical
formation. In other words, homolytic bond scission is
favored by bond weakening in presence of radicals.
First, we shall address the properties of the neutral EBHP
system and then we shall study the properties of the anionic,
radical and cluster systems. Ethylbenzene hydroperoxide is
a highly polarized system, as can be seen from its charge
distribution (see Table 1). The excess of positive charge on
Ha explains the acidity of the hydroperoxide. One remark-
able fact, well established in the literature, is that the oxygen
atoms of the peroxide group are not electronically
equivalent; that is, the Ob is more electronegative than the
inner one.5 The peroxyl radical system (EBHPr) showed a
dramatic reinforcement of the Og–Ob bond. The opposite
effect was observed for the anionic system (EBHPK), in
which the Og–Ob bond is weakened. This effect implies that
the formation of the anionic system favors the alkoxyl
Figure 3. Selectivity to EBHP versus ethylbenzene conversion.
The energies and molecular properties involved in the
calculations at B3LYP/631G(d,p) level of theory are shown
in Table 1. First, it should be noted that the formation of any
catalyst–hydroperoxide complex proves to be energetically
favorable; that is, the incorporation of QAS or each of its
components leads to the formation of new species in the
Scheme 2. Molecular structures of the isolated systems and the clusters under study.

11530
L. Barrio et al. / Tetrahedron 60 (2004) 11527–11532
Table 1. Results of the calculations: electronic energy (with zero point correction) in kcal/mol, net charges over the atoms in atomic units, bond distance in
angstroms, and electronic density at the bond critical point in atomic units with the 6-31G(d,p) basis set
Structure
ECE0
(kcal/mol)
q Cd
q Og
q Ob
q Ha
Cd–Og bond
Og–Ob bond
˚
Ob–Ha bond
˚
˚
RC–O (A)
r(b.c.p).
rO–O (A)
r(b.c.p).
rO–H (A)
r(b.c.p)
EBHP
0
—
—
K16.29
0.5135
0.6385
0.3994
0.4564
K0.5555
K0.6941
K0.3614
K0.5277
K0.5986
K0.6332
K0.1836
K0.6436
0.5787
—
—
1.436
1.403
1.492
1.463
0.2519
0.2848
0.2245
0.2339
1.461
1.487
1.319
1.458
0.2733
0.2472
0.3917
0.2751
0.971
—
—
0.3664
—
—
EBHPK
EBHPr
EBHPT-
MAC
0.6002
0.972
0.3626
EBHPIK
EBH-
K7.14
K74.68
0.5501
0.6686
K0.5812
K0.6816
K0.6397
K0.6315
0.5965
0.6361
1.425
1.400
0.2618
0.2844
1.450
1.480
0.2752
0.2537
0.990
1.539
0.3360
0.0733
POHK
EBHPT-
MAI
K86.65
0.4884
0.5623
K0.5926
K0.6970
K0.6434
K0.6264
0.5876
0.6339
1.451
1.448
0.2414
0.2616
1.451
1.480
0.2749
0.2547
1.000
1.626
0.3259
0.0584
EBHPT-
MAOH
K176.96
radical, and this radical participates in the reactions to afford
ketone, acid and alcohol products instead hydroperoxide,
implying a reduction in the selectivity to EBHP.
Having analyzed the effect of the cation and the anion
separately, the effect of the whole salt on the EBHP system
was studied. First, the high stabilization energies for both
salts indicate the presence of stronger interactions in these
systems than in the former ones. TMAI produced a
weakening of the Cd–Og and Ob–Ha bond, whereas the
Og–Ob was reinforced (see Table 1). These changes in the
molecular properties favor the reaction mechanism to
the formation of EBHP. The key bond for the selectivity
of the reaction between the oxygen atoms is slightly
reinforced, whereas the formation of the radicals respon-
sible for the reaction is favored. The iodide salt has all the
desired effects on the hydroperoxide system and this
explains its high catalytic activity. TMAOH exerts similar
effects to the hydroxide anion. It weakens or almost breaks
the Ob–Ha bond but the effects on the Cd–Og and Og–Ob
bonds are not so beneficial for the reaction. Since the
presence of the anion favors the formation of alkoxyl
radicals (which yields ketone, acid and alcohol products
instead hydroperoxide), the hydroxide salt decreases
selectivity to EBHP.
Once the systems have been analyzed in isolation, the
effects of the ammonium salts are discussed. The main
effect of the addition of the tetramethylammonium cation is
a weakening of the Cd–Og bond, reflected in an increase in
bond distance and a decrease in electronic density at the
bond critical point. This favors Cd–Og bond scission and,
consequently, the formation of the desired ethylbenzene
radical. Changes in the Og–Ob and Ob–Ha bonds are much
less pronounced. The inclusion of the iodide anion causes an
interaction between the halogen anion and the acid Ha.
The IK attracts the positively charged hydrogen and causes
the Ob–Ha bond to weaken. Accordingly, formation of the
peroxyl radical is favored. The Og–Ob bond remains
practically unchanged, whereas the Cd–Og bond is
reinforced.
The effect of the hydroxide anion is much more significant.
First, the high stabilization energy indicates the presence of
a very strong interaction in the structure. As a strong base,
the OHK reacts with EBHP by catching the Ha to give water
and EBHPK. The data in Table 1 clearly show that the
charge distribution and bond parameters for the oxygen
atoms are quite close to those of the anion. The Ob–Ha bond
distance and electronic density reveal the change in the
nature of the bond, which passes from a covalent interaction
to a hydrogen bond. On this basis, the formation of anionic
species (EBHPK) is strongly favored by OHK. The
hydroperoxo anion favors the formation of the alkoxyl
radical. At the same time, the Cd–Og bond is reinforced and
the Og–Ob bond is weakened. Consequently, the selectivity
of the oxidation reaction is switched to the generation of
undesired products.
In order to check the accuracy of these results a larger basis
set, which included diffuse functions such as 6-31CC
G(d,p), was employed to analyze the salt/EBHP cluster.
These new data are summarized in Table 2. The optimized
geometry of EBHPTMAI and EBHPTMAOH clusters
is shown in Figure 6. If the results obtained with the
6-31GCC(d,p) basis set are compared with that obtained
previously with the 6-31G(d,p) basis set, the following
observations can be made: (i) the stabilization energy for the
cluster EBHPTMAI is quite similar for both basis sets. That
implies that the basis set employed for I atom is large
enough to take into consideration the anionic character of
iodide; (ii) both the bond distances and electronic densities
at bond critical point change slightly with the addition of
Table 2. Results of the calculations: electronic energy (with zero point correction) in kcal/mol, bond distance in angstroms, and electronic density at the bond
critical point in atomic units with the 6-31CCG(d,p) basis set
Structure
ECE0
(kcal/mol)
Cd–Og bond
Og–Ob bond
Ob–Ha bond
˚
RC–O (A)
˚
rO–O (A)
˚
rO–H (A)
r(b.c.p).
r(b.c.p).
r(b.c.p)
EBHP
EBHPTMAI
EBHPTMAOH
0
1.442
1.448
1.430
0.2521
0.2413
0.2599
1.464
1.456
1.474
0.2734
0.2758
0.2592
0.971
0.994
1.525
0.3667
0.3373
0.0734
K86.26
K140.38

L. Barrio et al. / Tetrahedron 60 (2004) 11527–11532
11531
diffuse functions. As it was accomplished previously, the
Og–Ob bond is strengthened whereas the Cd–Og and Ob–Ha
bonds are weakened in this cluster; (iii) with respect to
EBHPTMAOH cluster a drastic increase in stabilization
energy is observed when employing the 6-31CCG(d,p)
basis set. Due to the strong anionic character of EBHP in
this cluster, the addition of diffuse functions is mandatory
for its correct description; (iv), geometrical and bond
parameters also change noticeably with this basis set.
Nevertheless, the global effect of the TMAOH salt over
the EBHP species remains unchanged. The Ob–Ha bond
is almost broken to infer anionic character to EBHP. The
Og–Ob bond is enlarged and the Cd–Og is shortened, as it
was achieved for the 6-31G(d,p) basis set (Fig. 4).
understanding of the mechanism of oxidation reactions.
(i) Quantum chemical calculations, particularly the B3LYP
hybrid functional, proved to be a valuable tool for the study
of the catalytic activity of the ammonium salts on the
oxidation of ethylbenzene. Owing to the considerable
complexity of the radical mechanism for this reaction,
further studies will be necessary in order to gain a deeper
understanding of such process and the effect of catalysts on
reaction mechanisms. Thus, the DFT methodology is a
promising technique for explaining the changes in activity
due to the addition of catalysts. (ii) The strong catalytic
activity and high selectivity to the hydroperoxide of TMAI
is explained in terms of the weakening of Cd–Og and Ob–Ha
bonds. These effects favor the formation of ethylbenzenyl
and peroxyl radicals, respectively, which are involved in the
propagation reaction that leads to hydroperoxide formation.
(iii) The lower selectivity to EBHP reached in the presence
of TMAOH is due to the weakening of Ob–Ha bond,
inducing the formation of the anionic species. The presence
of the anionic species favors the formation of alkoxyl
radicals, which yield ketone, acid and alcohol products
instead the hydroperoxide. Consequently, a reduction of the
selectivity to EBHP is observed.
To sum up, the employment of the 6-31GCC(d,p) basis set
does not change the effect of the catalyst on the EBHP but
reveals the need of diffuse functions when a rigorous
description of anionic character is required.
4. Experimental
A 150 ml cylindrical thermostatable glass vessel provided
with a glass stirrer and a gas inlet system, a type K
thermocouple, and a reflux condenser cooled by water
were used for the experiments. Temperature was
controlled by a heated circulating bath; the stirrer was
powered by a variable-speed engine and the gases were
fed to the reactor through mass flow controllers. In a
typical oxidation with air run, 50 g of ethylbenzene (EB)
containing about 0.4 wt% of ethylbenzene hydroperoxide
(EBHP), kindly provided by Repsol-YPF, and 10K4 M of
quaternary ammonium salt (QAS) were loaded into the
reactor. The temperature of the bath and stirring speed
were set at the desired values, typically 403 K and
1000 rpm, respectively. The QAS compounds selected in
this work were tetramethylammonium iodide (TMAI) and
tetramethylammonium hydroxide (TMAOH), both from
Aldrich Chemie. Then, an air flow of 10 L hK1 was fed.
In order to reduce the reaction time along hydrocarbon
oxidation, a common practice is to add small amounts of
hydroperoxides, which act as radical initiators and
eliminates the induction period typical of liquid phase
oxidations with molecular oxygen. A reference exper-
iment was carried out without the addition of ammonium
salt. Once the temperature had reached a constant value,
the gas flow was set and kept constant during the
experiment. Aliquots of the reaction mixture were taken
at 0, 60, 120 and 180 min of reaction. The total amount
extracted was less than 10 wt% of the total mixture
inside the reactor. The concentration of EBHP was
measured by standard iodometric titration. The remaining
organic compounds were analyzed by GC-FID on a
Hewlett Packard 6890-plus gas chromatograph equipped
with an HP-WAX capillary column. Before GC analysis,
these samples were pre-treated with triphenylphosphine
to decompose the EBHP quantitatively to 1-
phenylethanol.
Figure 4. 3-D structure at B3LYP/6-31GCC(d,p) level for:
(a) EBHPTMAI cluster, (b) EBHPTMAOH cluster.
3. Conclusions
Analysis of the electronic properties of the structures
formed between reactants and catalysts offers a better

11532
L. Barrio et al. / Tetrahedron 60 (2004) 11527–11532
Caution
Supplementary data
As the autoxidation of hydrocarbons with molecular oxygen
entrails some risk, an important safety issue was considered
in all the experiments. The experimental conditions
described in the preceding paragraphs indicate that the
reaction is conducted above the explosion limit of the
reaction mixture. To avoid this hazard two actions were
undertaken: (i) a flow of nitrogen was fed just on the surface
of the liquid, and (ii) the effluent oxygen concentration was
continuously monitored with an oxygen sensor and kept
almost constant at approximately 1 vol%. In any case, the
laboratory glassware was used behind safety screens, and
the volume of reaction was limited to 50 ml solutions.
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/j.tet.2004.09.
060
References and notes
1. Sheldon, R. A.; Kochi, J. K. Metal-catalyzed Oxidations of
Organic Compounds; Academic: New York, 1981.
2. Sheldon, R. A. In New Developments in Selective Oxidation;
Centi, G., Trifiro, F., Eds.; Elsevier: Amsterdam, 1990.
3. Sheldon, R. A. In Applied Homogenous Catalysis with
Organometalic Compounds; Cornils, B., Herrmam, W. A.,
Eds.; VCH: Weinheim, 1996; Chapter 2.4.3.
4.1. Computational methods
All calculations shown were performed with the Gaussian
989 program. The hybrid functional B3LYP was chosen,
since it had been previously shown to describe systems
containing peroxyl radicals or ionic species accurately.10–12
For H, C, N and O (atoms from the first and second rows) a
6-31G (d,p) or a 6-31CCG(d,p) basis set was chosen. For
iodine13 a 6-311GCC(3df) basis set was obtained from the
EMSL14 (the supplementary diffuse and polarization
functions of reference13 were added manually). All
structures presented are full-geometry optimizations at the
B3LYP theoretical level. Frequency analyses were carried
out on each system and it was verified that all were true
minima. Stabilization energies were estimated as the
difference between the isolated system (molecule or ion)
and the structure under study (Scheme 2). All energies
reported include zero-point energy correction. An esti-
mation of the basis set superposition error (BSSE) can be
worked out on the energy difference between the two basis
sets employed. Spin contamination was checked for the
radical analyzed and no significant deviation from the pure
spin state was detected. The atoms-in-molecules15 (AIM)
approach was used, as implemented in the AIMPAC16 suite
of programs, to estimate atomic charges and bond strengths.
´ ´ ´ ´
4. Csanyi, L. J.; Jaky, K.; Kota, Z.; Pali, T. J. Mol. Catal. A:
Chem. 2004, 209, 59.
´
5. Csanyi, L. J.; Jaky, K. Phys. Chem. Chem. Phys. 2001, 3, 2018.
´
´
´
6. Csanyi, L. J.; Jaky, K. J. Mol. Catal. A: Chem. 1997, 120, 125.
7. Pisarenko, L. M.; Kasaikina, O. T. Russ. Chem. Bull. 2002,
51(3), 449.
8. Bader, R. F. W.; Slee, T. S.; Cremer, D.; Kraka, E. J. Am.
Chem. Soc. 1983, 105, 5061.
9. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;
Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T.;
Petersson, G. A.; Montgomery, J. A.; Raghavachari, K.;
Al-Laham, M. A.; Zakrzewski, V. G.; Ortiz, J. V.; Foresman,
J. B.; Peng, C. Y.; Ayala, P. Y.; Chen, W.; Wong, M. W.;
Andres, J. L.; Replogle, E. S.; Gomperts, R.; Martin, R. L.;
Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart,
J. P.; Head-Gordon, M.; Gonzalez, C.; Pople, J. A. Gaussian
98, Revision A.11; Gaussian, Inc.: Pittsburgh, PA, USA, 2001.
10. Rientstra-Kiracof, J. C.; Allen, W. D.; Schaefer, H. F. III
J. Phys. Chem. A 2000, 104, 9823.
11. Wisman, J. L.; Head-Gordon, M. J. Am. Chem. Soc. 2001, 123,
11686.
Cartesian coordinates, thermochemical data for all opti-
mized structures and basis set input for iodine are provided
in the Supplementary information.
12. Lee, J.; Bozzelli, J. W. J. Phys. Chem. A 2003, 107, 3778.
13. Glukhovtsev, M. N.; Pross, A.; McGrath, M. P.; Radom, L.
J. Chem. Phys. 1995, 103, 1878.
14. Basis sets were obtained from the Extensible Computational
Chemistry Environment Basis Set Database, Version 10/21/03,
as developed and distributed by the Molecular Science Com-
puting Facility. http://www.emsl.pnl.gov/forms/basisform.html
15. Bader, R. F. Atoms in Molecules: A Quantum Theory; Oxford
University Press: Oxford, 1994.
Acknowledgements
Thanks are due to Dr. A. Delgado for fruitful discussions.
CTI (CSIC) is gratefully acknowledged for CPU facilities.
LB gratefully acknowledges a fellowship granted by the
European Social Found (I3P-BPG2002 programme). Two
of us (PPTT and JMCM) gratefully acknowledge fellow-
ships granted by Repsol-YPF Company.
16. Konig, F. W. B.; Bader, R. F. W.; Tant, T. J. Comput. Chem.
1982, 13, 317.