Reactivity of α-amino-peroxyl radicals and consequences for amine oxidation chemistry

Article abstract of DOI:10.1039/c2cp41306j

A comparative theoretical study is presented on the formation and fate of α-amino-peroxyl radicals, recently proposed as important intermediates in the aerobic oxidation of amines. After radical abstraction of the weakly bonded αH-atom in the amine substrate, the α-amino-alkyl radical reacts irreversibly with O₂, forming the corresponding α-amino-peroxyl radical. HO₂-elimination from various types of α-amino-peroxyl radicals (forming the corresponding imine) and the kinetically competing substrate H-abstraction (forming the α-amino-hydroperoxide) were computationally characterized. Polar solvents were found to reduce the HO ₂-elimination barrier, but increase the barrier for H-abstraction. Depending on the reaction conditions (gas or liquid phase, amine concentration, nature of the solvent, and temperature), either of the two mechanisms is favored. The consequences for aerobic amine oxidation chemistry are discussed.

Full text of DOI:10.1039/c2cp41306j

View Online / Journal Homepage / Table of Contents for this issue
Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Dynamic Article Links
PCCP
Cite this: Phys. Chem. Chem. Phys., 2012, 14, 11002–11007
www.rsc.org/pccp
PAPER
Reactivity of a-amino-peroxyl radicals and consequences for amine
oxidation chemistryw
Martin T. Schumperli, Ceri Hammond and Ive Hermans*
¨
Received 8th March 2012, Accepted 11th June 2012
DOI: 10.1039/c2cp41306j
A comparative theoretical study is presented on the formation and fate of a-amino-peroxyl
radicals, recently proposed as important intermediates in the aerobic oxidation of amines. After
radical abstraction of the weakly bonded aH-atom in the amine substrate, the a-amino-alkyl
ꢀ
radical reacts irreversibly with O2, forming the corresponding a-amino-peroxyl radical. HO2 -
elimination from various types of a-amino-peroxyl radicals (forming the corresponding imine)
and the kinetically competing substrate H-abstraction (forming the a-amino-hydroperoxide) were
ꢀ
computationally characterized. Polar solvents were found to reduce the HO2 -elimination barrier,
but increase the barrier for H-abstraction. Depending on the reaction conditions (gas or liquid
phase, amine concentration, nature of the solvent, and temperature), either of the two
mechanisms is favored. The consequences for aerobic amine oxidation chemistry are discussed.
Introduction
The aerobic oxidation of amines is of considerable technological
importance. From a synthetic point of view, amines are viable
precursors of valuable chemical building blocks such as imines,
nitriles and oximes, amongst others.1 In addition, amines are also
used as absorbers in the post-combustion capture of CO2.2
However, a severe problem with this technology is the sensitivity
of amines to oxidative degradation. Indeed, traces of oxygen in
the flue gas cause deterioration of the amine and reduce the
efficiency of the scrubber. Despite its importance, the chemistry
of aerobic amine oxidations is poorly understood, certainly when
compared to alcohol oxidation.
Scheme 1 Oxidation of cyclohexylamine to cyclohexanone oxime.
(c-C6H10(NH2)OOH). Following its intermediate production,
this hydroperoxide was proposed to dehydrate over the
heterogeneous catalyst (WO3/Al2O3) to the oxime product.
However, this mechanism remains speculative as very little is
known about a-amino-peroxyl radicals in general. Never-
theless, it is clear that this reaction is radical mediated and
achieves an exceptionally high oxime yield of up to 90% which
points to very selective radical chemistry.
Recently, Suzuki et al. investigated the aerobic oxidation of
cyclohexylamine to cyclohexanone oxime using a catalytic
combination of 1,1-diphenyl-2-picrylhydrazyl (DPPH,
a
radical mediator) and WO3/Al2O3.3 This one-step reaction is
of considerable industrial interest as it provides an alternative
route to cyclohexanone oxime, an important polymer building
block. It was proposed that DPPH generates the a-amino-cyclo-
hexyl radical upon abstraction of the aH-atom. This C-centered
radical is, upon O2-addition, converted to the c-C6H10(NH2)OOꢀ
peroxyl radical (Scheme 1).
In this contribution, the chemistry of a-amino-peroxyl
radicals is computationally characterized. In the first part,
their formation via aH-abstraction from the amine and O2
addition is investigated. Subsequently, we look at the fate of
such a-amino-peroxyl radicals. For analogous a-hydroxy-
ꢀ
peroxyl radicals, the unimolecular elimination of HO2 was put
forward earlier as a fast reaction, competing with bimolecular
H-abstraction.4 This reaction proceeds via an H-bonded inter-
mediate, which rapidly eliminates HO2ꢀ via a variational transition
state (Scheme 2).5,6 Depending on the reaction conditions (e.g., T,
The a-amino-cyclohexylperoxyl radical was proposed to abstract
an aH-atom and form the a-amino-cyclohexylhydroperoxide
ꢀ
substrate, etc.), this HO2 -elimination outruns the competing
H-abstraction by the a-hydroxy-peroxyl radical to a-hydroxy
hydroperoxide.
The competition between HO2ꢀ-elimination, forming an
imine, and (a)H-abstraction, forming a-amino-hydroperoxides,
for various types of a-amino-peroxyl radicals (see Scheme 3) is
therefore investigated, in an attempt to verify or dismiss the
ETH Zurich, Department of Chemistry and Applied Bio-Sciences,
Wolfgang-Pauli-Strasse 10, 8093 Zurich, Switzerland.
E-mail: hermans@chem.ethz.ch; Fax: +41 44 6331514;
Tel: +41 44 6334258
w Electronic supplementary information (ESI) available: Cartesian
coordinates of important transition states. See DOI: 10.1039/
c2cp41306j
c
11002 Phys. Chem. Chem. Phys., 2012, 14, 11002–11007
This journal is the Owner Societies 2012

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
View Online
the range of the benchmark techniques, at a much lower
computational cost.
The results in Table 2 show that amines are significantly
more prone to aH-abstraction than iso-electronic alcohols.
Indeed, the computed barriers for aH-abstraction from methanol,
ethanol and iso-propanol by CH3OOꢀ were previously calculated
to be 14.0, 11.8 and 10.3 kcal molꢁ1,4 respectively; on average
more than 3 kcal molꢁ1 higher than for the corresponding
amines. This kinetic difference is a direct consequence of the
lower aC–H bond strengths in amines than in alcohols (Fig. 1).
Notice that the correlation between the abstraction barrier and
the aC–H bond (so-called Evans Polanyi correlation) is slightly
curved and levels off for weaker aC–H bonds.
Scheme 2 HO2ꢀ-elimination from
a general a-hydroxy-peroxyl
radical.
previously hypothesized reaction mechanisms.7 Unimolecular
cleavage of the O–O bond in the peroxyl radical is unimportant
in the relevant temperature range, due to the high bond strength.
Computational methods
The C–H bond strength is inversely proportional to the
stability of the radical product. Fig. 2 correlates the H–CH2X
bond strength with the Mulliken spin density on the C-atom in
the ꢀCH2X radical (X = H, SiH3, F, CH3, OH, NH2,
N(H)CH3 in order of decreasing H–CH2X strength). Various
parameters influence the stability of the ꢀCH2X radical,
including (when possible) p-back-donation of 2p-orbitals of
the substituent X into the half-occupied 2p-orbital on the
C-atom,15 in addition to hyperconjugation and s-effects.
Those combined effects cause the C–H bond strength to
decrease in the order CH3F, CH3OH and CH3NH2, despite
the different substituents being iso-electronic and the electro-
negativity decreasing in the order F 4 O 4 N.
All calculations were carried out using the GAUSSIAN09
software package (Revision A.02).8 Geometries of stationary
points (i.e. stable intermediates and transition states) were
initially optimized at the B3LYP-DFT- or QCISD-level.9
Subsequently, the relative energies (corrected for zero-point
energy – ZPE – differences) were quantitatively refined using
various higher levels of theory, such as G3,10 CBS-QB3,11
CBS-APNO12 and G2M13 (see Table 1). ZPE-corrected relative
energies of the stationary points on the Potential Energy Surface
(PES) are reported at 0 K.
a-Pinene oxidation experiments were performed at 80 1C in
a bubble column reactor at 1 atm O2, as described elsewhere.14
Addition of O2 to the a-amino-alkyl radicals is highly
exothermic, with values for DrH ranging from ꢁ27.2 kcal
molꢁ1 (CBS-APNO) to ꢁ32.5 kcal molꢁ1 (G2Mlarge//DFT)
Results and discussion
Formation of a-amino-peroxyl radicals
ꢀ
for the case of CH2NH2, indicating irreversible formation of
Table 2 compares the adiabatic energy barriers for the
aH-abstraction from different types of amines by the CH3OOꢀ
radical. Earlier work showed that for analogous H-abstraction
reactions from alkanes, alcohols and hydroperoxides, the
DFT//DFT results agree within 0.5 kcal molꢁ1 from benchmark
methods like G3, CBS-QB3, and G2M//DFT.4 However, for the
abstraction from amines, the various benchmark levels of theory
clearly deviate more amongst each other (i.e. nearly 2 kcal molꢁ1).
Given that those high levels of theory are computationally rather
demanding, and that they all have their merits and shortcomings,
we prefer to refer to the DFT//DFT results which agree within
the a-amino-peroxyl radical.
ꢀ
Unimolecular HO2 -elimination to imine
Subsequently, the unimolecular decomposition of the a-amino-
peroxyl radicals to the corresponding imine plus HO2ꢀ (Reaction 1
in Scheme 3) is investigated. Similar to the analogous reaction
of a-hydroxyl-peroxyl radicals (Scheme 2), this reaction is
found (via intrinsic reaction coordinate analysis) to proceed
ꢀ
via an H-bonded complex between the imine and HO2 . The
break up of this H-bonded complex does not, however, control
the overall rate. The rate determining step in the mechanism is
the HO2ꢀ-elimination, and mainly involves breaking of the
C–O bond (see Scheme 4).
Tables 3–5 compare ꢀvarious levels of theory for the decomposi-
ꢀ
ꢀ
tion of OOCH2NH2, OOCHCH3NH2, and OOC(CH3)2NH2,
respectively. One observes an excellent agreement (i.e., within
1 kcal molꢁ1) between the CBS-QB3, CBS-APNO and G3
levels of theory. On the other hand, and in contrast to the
situation with a-hydroxyl-peroxyl radicals,5 G2M//DFT gives
slightly different results. It seems unlikely that this deviation
can be (entirely) attributed to a limited CCSD(T)-basis set,
Scheme 3 Competition between the unimolecular HO2ꢀ-elimination
(1) and the bimolecular H-abstraction (2) for a-amino-peroxyl radicals.
Table 1 Summary and details of the levels of theory utilised
Method
Energy calculation
Geometry optimization and ZPE
G2Msmall/DFT E[UCCSD(T)/6-31G(d)] + {E[UMP2/6-311++G(3df,3pd)] ꢁ E[UMP2/6-31G(d)]}
G2Mlarge//DFT E[UCCSD(T)/aug-cc-pVDZ] + {E[UMP2/aug-cc-VTZ] ꢁ E[UMP2/aug-cc-pVDZ]}
UB3LYP/6-311+G(df,pd)
UB3LYP/aug-cc-pVTZ
G2M//QCISD E[UCCSD(T)/6-311G(df,pd)] + {E[UMP2/6-311++G(3df,3pd)] ꢁ E[UMP2/6-311G(df,pd)]} UQCISD/6-311G(d,p)
DFT//DFT
E[UB3LYP/6-311++G(3df,3pd)]
UB3LYP/6-31G(d,p)
c
This journal is the Owner Societies 2012
Phys. Chem. Chem. Phys., 2012, 14, 11002–11007 11003

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
View Online
Table 2 Adiabatic energy barriers (ZPE-corrected; kcal molꢁ1) for the aH-abstraction by CH3OOꢀ radicals from various types of amines
Aminea
CBS-QB3
G3
G2Msmall//DFT
G2Mlarge//DFT
DFT//DFT
R1 = R2 = R3 = H
8.2
5.9
4.1
8.3
12.8
8.0
6.1
9.8
12.3
11.1
9.2
12.6
10.5
9.0
9.9
8.5
7.7
9.2
R1 = R3 = H; R2 = CH3
R1 = R2 = CH3; R3 = H
R1 = R2 = H; R3 = CH3
10.0
10.6
a
aC–H bond strengths of the various amines are 89.2, 87.6, 86.0 and 88.5 kcal molꢁ1, respectively, at the B3LYP/6-311++G(d,p) level of theory.
Fig. 1 Correlation between the computed H-abstraction barrier
Scheme 4 Geometry of the transition state for HO2ꢀ-elimination
(DFT//DFT level) and the aC–H bond strength (B3LYP/
from ꢀOOCH2NH2 at various levels of theory (viz., CBS-QB3, CBS-
6-311++G(d,p) level) for various amines (m) and alcohols (K).
APNO, G3, QCISD/6-311G(d,p) and B3LYP/aug-cc-pVTZ).
doublet structure to neglect multi-reference character. Similar
to the situation with a-hydroxy-peroxyl radicals, the DFT//
DFT results seem to generate inconsistent errors when com-
pared to the benchmark levels. Functionals which include
long-range interactions, such as CAM-B3LYP and wB97,
did not, unfortunately, lead to more consistent results.
ꢀ
The endothermicity of HO2 -elimination from the a-amino-
peroxyl is higher than for the corresponding a-hydroxyl-peroxyl
cases, in line with the lower stability of 4CQN– bonds
compared to 4CQO bonds. In addition, the activation
barriers for the a-amino-peroxyl radicals are found to be
nearly 6 kcal molꢁ1 higher than for the corresponding
Fig. 2 Computed H–CH2X bond strength (B3LYP/6-311++G(d,p)
level) vs. the Mulliken spin density on the C-atom in the ꢀCH2X
a-hydroxyl-peroxyl radicals. These observations could suggest
ꢀ
radical (see the text).
a slightly later HO2 -elimination transition state for the
amino-systems. This is confirmed upon comparison of the
TS structures in Scheme 5: in the case of CH2(NH2)OOꢀ,
given the close agreement between G2Msmall//DFT and
G2Mlarge//DFT. More likely, this deviation is due to a slight
difference in starting geometry (density-based vs. wave-func-
tion based methods) for the single-point energy calculations
(see Scheme 4). Indeed, the G2M//QCISD results are more in
line with the CBS-QB3, CBS-APNO and G3 results. Spin
contamination of the wavefunction was found to be negligible
(hhS2ii = 0.75 ꢂ 0.03 before spin annihilation). In addition, we
performed the T1-diagnostic at the UCCSD(T)/6-311++G(df,pd)//
QCISD/6-311G(d,p) level of theory for tꢀhe transition state of
the Cꢃ ꢃ ꢃO bond is significantly more elongated, compared
to the CH2(OH)OOꢀ. Nevertheless, the ratio of partition
functions (TS over reactant) barely changes, leading to similar
pre-exponential factors in the rate constants.
Subsequeꢀntly, the effect of the R3 substituent (i.e., N-alkyl)
on the HO2 -elimination is investigated (Table 6). Apparently
substituting a methyl group for an H-atom lowers the
endothermicity and the reaction barrier, in line with the higher
stability of substituted imines.
ꢀ
In conclusion, HO2 -elimination from a-amino-peroxyl
ꢀ
the OOCH2NH2 - CH2QNH + HO2 reaction. The value
radicals is slightly more endothermic than for a-hydroxyl-peroxyl
radicals, and proceeds with a higher barrier. Alkyl-substituents
obtained (i.e. 0.032) is sufficiently small for an open-shell
c
11004 Phys. Chem. Chem. Phys., 2012, 14, 11002–11007
This journal is the Owner Societies 2012

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
View Online
Table 3 Relative energies of the characteristic stationary points on the Potential Energy Surface (PES; ZPE-corrected) for the unimolecular
ꢀ
decomposition reaction ꢀOOCH2NH2 2NH+HO2
Stationary point
CBS-QB3
CBS-APNO
G3
0.0
17.9
7.6
G2Msmall//DFT
G2Mlarge//DFT
G2M//QCISD
DFT//DFT
ꢀOOCH2NH2
TS
Complex
0.0
18.6
8.5
0.0
17.4
8.4
0.0
21.9
9.5
0.0
21.4
10.1
20.6
0.0
18.5
12.8
20.9
0.0
18.6
9.8
ꢀ
CH2NH + HO2
16.7
16.8
16.0
19.9
19.1
Table 7 Adiabatic energy barriers (ZPE-corrected; kcal molꢁ1) for
the aH-abstraction from different amines at the DFT//DFT level of
theory
Table 4 Relative energies of the characteristic stationary points on the
Potential Energy Surface (PES; ZPE-corrected) for the unimolecular
ꢀ
decomposition reaction ꢀOOCHCH3NH2 " CH3CHNH+HO2
CBS- CBS-
QB3 APNO G3 DFT
G2Msmall// G2Mlarge// DFT//
DFT
Stationary point
DFT
ꢀOOCHCH3NH2
TS
Complex
0.0 0.0
17.7 16.5
7.8 7.6
17.1 16.8
0.0 0.0
17.1 20.7
7.2 8.7
0.0
20.7
8.3
0.0
16.2
6.7
Amine
DFT//DFT
CH3CHNH +
ꢀ
16.5 19.7
20.2
16.8
R1 = R2 = R3 = H
12.0
10.6
7.2
HO2
R1 = R3 = H; R2 = CH3
R1 = R2 = CH3; R3 = H
R1 = R2 = H; R3 = CH3
11.1
Table 5 Relative energies of the characteristic stationary points on the
Potential Energy Surface (PES; ZPE-corrected) for the unimolecular
ꢀ
decomposition reaction ꢀOOC(CH3)2NH2 " (CH3)2CNH+HO2
CBS-
QB3 G3 DFT
G2Msmall// G2Mlarge// DFT//
DFT
Stationary point
DFT
ꢀOOCHCH3NH2
TS
Complex
0.0
0.0 0.0
0.0
19.2
7.9
0.0
12.3
3.3
15.8 17.5 18.9
7.3 6.6 7.4
ꢀ
CH3CHNH + HO2 18.2 17.6 20.2
20.8
15.2
Scheme 5 CBS-APNO optimized TSs for HO2ꢀ-elimination from
CH2(OH)OOꢀ (left) and CH2(NH2)OOꢀ (right); distances in A.
Fig. 3 Computed barriers for HO2ꢀ-elimination (D; average of high
levels of theory in Tables 3–6) and H-abstraction (J; DFT//DFT
level) by a-amino-peroxyl radicals vs. aC–H bond strength (B3LYP/
6-311++G(d,p)-level) for various amines.
Table 6 Relative energies of the characteristic stationary points on
the Potential Energy Surface (PES; ZPE-corrected) for the unimoleꢀ-
cular decomposition reaction ꢀOOCH2NHCH3 " CH2NCH3+HO2
Solvent effects and quantifying the imine/hydroperoxide
branching fraction
CBS-
QB3 G3 DFT
G2Msmall// G2Mlarge// DFT//
DFT
Stationary point
DFT
All calculations reported so far were modelled in a solvent-free
environment (viz., vacuum). As amine oxidations are typically
performed in the liquid phase, it remains to be investigated if a
solvent could significantly affect the barrier of either of the
two reactions. Unfortunately, taking into account all
specific solvent interactions is very challenging and certainly
beyond routine. The polarisable continuum model (PCM) as
implemented in the Gaussian09 software package8 provides
ꢀOOCH2NHCH3
TS
Complex
0.0
17.2
4.5
0.0
17.7 20.0
3.7 5.6
13.0 16.9
0.0
0.0
21.2
6.2
0.0
14.6
5.3
ꢀ
CH2NCH3 + HO2 13.7
17.5
15.0
on the aC-atom or the N-atom reduce both the energy barrier
and the endothermicity of the imine formation.
however
a
convenient and good approximation.16 This
H-abstractions: formation of a-amino-hydroperoxides
method creates a solute cavity around the species via a set of
ꢀ
overlapping spheres. Fig. 4 shows the result of such calcula-
ꢀ
by the ꢀOOC(CH3)2NH2 radical at the DFT//DFT level of
theory. A list containing all solvents and their dielectric
constant e is given in the ESI.w
Competing with the HO2 -elimination is the abstraction of
tions for the barriers of HO2 -elimination and H-abstraction
H-atoms from the amine substraꢀte by the a-amino-peroxyl radical
(Table 7). Compared to the HO2 -elimination reaction, the barrier
of this aH-abstraction reaction is more sensitive to the aC–H bond
strength in the corresponding amine substrate (see Fig. 3).
c
This journal is the Owner Societies 2012
Phys. Chem. Chem. Phys., 2012, 14, 11002–11007 11005