Metal-assisted lossen rearrangement

Article abstract of DOI:10.1021/jo300031f

A new reaction mechanism for the Lossen rearrangement of hydroxamic acids catalyzed by basic salts is presented. It is shown that the rearrangement proceeds in metal complexes of deprotonated hydroxamic acids. The deprotonation can occur either at the oxygen atom (observed for the zinc complexes) or at the nitrogen atom (observed for the potassium complexes). Both anionic forms are characterized by infrared multiphoton dissociation spectroscopy. The rearrangements proceed from the reactive N-deprotonated metal hydroxamates and lead to metal carbamates. The mechanism is elucidated by computational chemistry, mass-spectrometric studies, and preparative experiments.

Full text of DOI:10.1021/jo300031f

Article
pubs.acs.org/joc
Metal-assisted Lossen Rearrangement
̌
Lucie Jasíkova, Eva Hanikyrova, Anton Skríba, Juraj Jasík, and Jana Roithova*
̌ ́ ̌ ́ ̌ ́
́
Department of Organic Chemistry, Faculty of Science, Charles University in Prague, Hlavova 2030/8, 12843 Prague 2, Czech
Republic
*
S Supporting Information
ABSTRACT: A new reaction mechanism for the Lossen rearrangement of hydroxamic acids catalyzed by basic salts is presented.
It is shown that the rearrangement proceeds in metal complexes of deprotonated hydroxamic acids. The deprotonation can occur
either at the oxygen atom (observed for the zinc complexes) or at the nitrogen atom (observed for the potassium complexes).
Both anionic forms are characterized by infrared multiphoton dissociation spectroscopy. The rearrangements proceed from the
reactive N-deprotonated metal hydroxamates and lead to metal carbamates. The mechanism is elucidated by computational
chemistry, mass-spectrometric studies, and preparative experiments.
INTRODUCTION
key intermediate, which decomposes to yield the corresponding
■
amine, CO , and isocyanate. The isocyanate returns to the
The Lossen rearrangement is one of three text-book rearrange-
ments (together with the Curtius and Hofmann rearrange-
ments) in which derivatives of carboxylic acids (hydroxamic
acids, azides, and amides, respectively) are transformed via
transient acyl nitrenes to isocyanate intermediates. Isocyanate
intermediates can be either trapped by a nucleophile to form
derivatives of the corresponding carbamic acids or degraded to
the respective amines in the presence of water (Scheme 1). The
classical Lossen rearrangement starts with the formation of a
mixed anhydride from a hydroxamic acid and another acid
derivative such as acetanhydride or tosyl chloride. Heating
with a base then leads to the elimination of acetate or tosylate,
respectively, concomitant with the rearrangement to an
isocyanate.
The mixed anhydride has to be either prepared in a separate
step or it can be generated in situ. The latter procedure has the
disadvantage that the formed isocyanate can react with free
hydroxamic acid to yield the corresponding O-carbamoylhy-
droxamate, which under the heating affords a symmetrically
2
catalytic cycle and reacts with the hydroxamic acid to afford
another molecule of O-carbamoylhydroxamate. There are,
however, many open questions. Thus, why should the O-
carbamoylhydroxamate intermediate decompose to yield the
corresponding amine in the base-catalyzed reaction, while the
decomposition of the isolated O-carbamoylhydroxamate leads
to the formation of the urea product? Is this change in reactivity
an effect of the solvent? Are there some specific effects of the
bases, or is the reaction just dependent on the pK of a
b
1,2
particular base? In this paper, we present an alternative
mechanism, which is not based on the formation of O-
carbamoylhydroxamate, but instead the reaction proceeds
under catalysis of metal ions, which are present in the solution
as counterions of the added base.
RESULTS AND DISCUSSION
We have recently studied the complexation abilities of
hydroxamic acids in the gas phase. In particular, we have
■
9
3,4
substituted derivative of urea. Alternative procedures for the
Lossen rearrangement avoiding this undesired self-dimerization
are based on reactions of hydroxamic acids with dehydrating
addressed the competitive complexation of zinc ions by acetate
and acetohydroxamate. It has been shown that, upon the
collisional activation, acetohydroxamate coordinated to zinc(II)
loses a neutral isocyanate. This finding has inspired us to
investigate a possible correlation between this gas-phase
rearrangement and the mechanism operating in the condensed
phase during the Lossen rearrangements catalyzed by basic
metal salts.
5
,6
agents such as dicyclohexylcarbodiimide (DCC) or carbon-
7
yldiimidazole (DCI).
The simplest approach has been shown recently: The
rearrangement of aromatic hydroxamic acids can be easily
achieved by heating with a base such as K CO . The reaction
mechanism suggested for the base-catalyzed reaction (Scheme
) is based on the analogy with the original mechanism
8
2
3
2
depicted in Scheme 1. Thus, a mixed anhydride of hydroxamic
and carbamic acids, O-carbamoylhydroxamate, is suggested as a
Received: January 10, 2012
Published: February 23, 2012
©
2012 American Chemical Society
2829
dx.doi.org/10.1021/jo300031f | J. Org. Chem. 2012, 77, 2829−2836

The Journal of Organic Chemistry
Article
Scheme 1. Mechanism of the Classical Lossen Rearrangement
Scheme 2. Suggested Mechanism of the Base-catalyzed
Lossen Rearrangement
The reaction is induced by deprotonation of the nitrogen
atom of the hydroxamic acid in most of the designed
procedures (see Scheme 1). For free hydroxamic acids, the
10
site of the deprotonation has been a subject of controversy.
There exists evidence for preferred N-deprotonation, but under
certain conditions and for some hydroxamic acids also O-
deprotonation can prevail. However, in the presence of a metal
cation there exist numerous experimental evidence (crystal
structures) that the hydroxamate is coordinated in the O-
11
deprotonated form.
Accordingly, let us first consider the structure of metal
hydroxamate ions in the gas phase. Structures of ionic metal
complexes in the gas phase can be elucidated by infrared
1
2
multiphoton dissociation (IRMPD) spectroscopy. This
method is based on the investigation of fragmentations induced
by the absorption of IR photons. It is a multiphoton method,
because the absorption of several IR photons is necessary for
the cleavage of (usually covalent) bonds. The IR absorption is
resonantly enhanced at the position of an absorption band of
the investigated ions, which consequently leads to a more
abundant fragmentation. The dependence of the fragmentation
yield on the wavelength of the IR photons hence provides an
experimental (multiphoton) IR spectrum of gaseous ions.
First, we have spectroscopically investigated complex of zinc
with the benzhydroxamate counterion (1-H)¯. Zinc ions in the
Figure 1. IRMPD spectrum of [Zn(1-H)(Phen)]+ (a) and its
comparison with theoretically calculated IR spectra of different
isomers of the complex (b−d). The theoretical spectra are shown as
13
gas phase are preferentially tetracoordinated; therefore, we
+
cannot generate the bare [Zn(1-H)] cations, but instead a
+
complex [Zn(1-H)(1)] is formed. The IRMPD spectrum of
gray bars, the black solid lines show the theoretical spectra folded with
this complex is dominated by two bands at 1520 and 1650 cm−1
−1
a Gaussian function (fwhm = 16 cm ). Notations I3 and I261 refer to
80
+
(see the SI). On the basis of comparison with the theoretical
the relative intensities of the parent complex [Zn(1-H)(Phen)] (m/z
+
3
80) and the fragment [Zn(OH)(Phen)] (m/z 261), respectively.
spectra, these bands are assigned to the carbonyl vibration of
the O-deprotonated and neutral hydroxamic acid, respectively.
The theoretical spectra of [Zn(1-H)(1)] contain overlapping
+
thus corresponds to the occurrence of the Lossen rearragement.
The corresponding IRMPD spectrum shows three main peaks
bands corresponding to the deprotonated and to the neutral
hydroxamic acid and therefore the interpretation is not
straightforward (for details see the SI). We have therefore
rather decided to replace the neutral hydroxamic acid by
phenanthroline ligand (Phen).
−1
at about 1480, 1520, and 1550 cm . Detailed inspection
−1
reveals also two minor features at 1430 and 1310 cm . To
interpret the spectrum, we have explored possible structures of
the complex by DFT calculations.
+
+
Figure 1 shows the IRMPD spectrum of [Zn(1-H)(Phen)] .
The most stable isomer of the [Zn(1-H)(Phen)] complex
The irradiation (or collisional activation, see lower) induces
exclusively elimination of phenylisocyanate (PhNCO), which
bears benzhydroxamic acid deprotonated at the oxygen atom
(A1_Zn, Figure 1b). The corresponding theoretical IR
2
830
dx.doi.org/10.1021/jo300031f | J. Org. Chem. 2012, 77, 2829−2836

The Journal of Organic Chemistry
Article
+
Figure 2. B3LYP/TZVP potential-energy surface for the Lossen rearrangement of within the [Zn(1-H)(Phen)] complex. The energies are given at
0K
0
K relative to E (A1_Zn) = −2826.524477 hartree.
Figure 3. B3LYP/TZVP potential-energy surface for the potassium-assisted Lossen rearrangement. Energies refer to 0 K in the gas phase and are
0K
given relative to E (A1_K) = −1075.552235 hartree; energies in DMF at 298 K can be found in Table 2.
spectrum agrees well with the IRMPD spectrum. The major
at the level of theory chosen. The product complex finally
−1
peak at 1520 cm corresponds to the CO stretch of the
decomposes by the loss of phenylisocyanate, which still results
in exothermicity of 0.09 eV for the whole fragmentation
pathway.
+
carbonyl group (as also found above for [Zn(1-H)(1)] ). The
−1
peak at 1480 cm originates from the CC stretches of the
−1
phenyl ring, and the peak at 1550 cm corresponds to the C−
We have determined the activation energy for the rearrange-
ment in the gas phase by energy-resolved collision induced
dissociation experiment (experimental details and data can be
N stretch of the hydroxamate group. The minor peaks come
−1
from the vibrations of the phenanthroline ring (1430 cm ) and
−1
1
5,16
Fragmentation of the [Zn(1-H)(Phen)]+
benzhydroxamate (1310 cm ).
found in the SI).
The alternative isomer with the hydroxamate deprotonated
at the nitrogen atom (A2_Zn) lies 0.63 eV higher in energy
than the most stable isomer. In accordance, the IR spectrum
complex leads exclusively to the loss of phenylisocyanate.
Analysis of the relative cross section for this fragmentation leads
to a threshold energy of 1.8 ± 0.2 eV, which compares
reasonably well to the theoretically predicted activation energy
of 1.99 eV. The associated transition structure corresponds to
the hydrogen-atom migration from the nitrogen atom to the
(
Figure 1c) does not agree with the experimental IRMPD
+
features of the [Zn(1-H)(Phen)] complex. We have also
considered that we could detect fragmentation of the complex,
which had already undergone the Lossen rearrangement. In
such a case, the zinc cation would be coordinated by
phenanthroline, phenylisocyanate, and a hydroxo ligand
17
oxygen atom and the barrier can be lowered by tunneling.
More probably, however, the experimental threshold is slightly
underestimated, because the studied ions are not thermalized.
We have typically encountered underestimations of threshold
(
B1_Zn). The rearrangement is exothermic and this complex
lies 0.74 eV below the energy of the most stable isomer of zinc
benzhydroxamate (cf. Figure 2). However, the computed
spectrum of this complex (Figure 1d) does not agree with the
18
energies by up to 0.2 eV, which perfectly fits the results here.
We note in passing that we have tried to trap the possible
nitrene intermediate experimentally. To this end, the mass-
selected ions were collided at various collision energies (ranging
from 0 to 5 eV in the center-of-mass framework) with
cyclooctene. Cyclooctene was chosen because it is a very
reactive alkene and hence we expected that the nitrene
intermediate could undergo an addition reaction to the double
bond of cyclooctene. Despite large efforts, we have never
detected any trace of the addition product. This may serve as an
indirect hint that the nitrene is not formed during the
rearrangements and that the reaction proceeds in a single
step as indeed suggested by the DFT calculations.
+
experimental IRMPD spectrum of [Zn(1-H)(Phen)] . We can
hence conclude that we transfer the most stable isomer of the
complex between phenanthroline and zinc benzhydroxamate to
the gas phase and by irradiation with IR photons we promote
the Lossen rearrangement and the concomitant loss of
phenylisocyanate.
The computed potential-energy surface for this rearrange-
ment (Figure 2) reveals that the rate-determining step
corresponds to the migration of the hydrogen atom from the
nitrogen atom to the oxygen atom of benzhydroxamate. The
corresponding transition structure (TS1_Zn) lies 1.99 eV
higher in energy than the initial isomer A1_Zn. The subsequent
rearrangement consists in the cleavage of the N−O bond
associated with the migration of the phenyl group to the
nitrogen atom and leads to the product ion B1_Zn. This step
proceeds via an energy barrier of 0.99 eV (TS2; 1.62 eV above
the energy of A1_Zn) and it is highly exothermic. We have also
In summary, we so far have shown that the Lossen
rearrangement proceeds in the zinc complexes in the gas
phase. Next, the relevance for the base-catalyzed reaction in the
condensed phase will be addressed. Hoshino et al. showed that
8
the most efficient base is potassium carbonate. We have
therefore performed further calculations and experiments for
potassium complexes.
14
searched for a nitrene intermediate, but it is not a minimum
2
831
dx.doi.org/10.1021/jo300031f | J. Org. Chem. 2012, 77, 2829−2836

The Journal of Organic Chemistry
Article
The theoretical calculations show that the potential-energy
surface of the rearrangement of the potassium salt is slightly
different from that found for the zinc complex (Figure 3). The
most stable complex again corresponds to the coordination of
potassium to the benzhydroxamic acid deprotonated at the
oxygen atom (A1_K). However, the alternative isomer
deprotonated at the nitrogen atom (A2_K) is much closer in
energy and it lies only 0.12 eV above A1_K. The hydrogen
rearrangement proceeds via an energy barrier (TS1_K) of 1.70
eV. While the transition structure for the subsequent Lossen
rearrangement is very similar to that found above for the zinc
complex, the reaction sequence does not lead to the formation
of potassium hydroxide with coordinated phenylisocyanate, but
the hydroxyl group rather adds to the carbon atom of
phenylisocyanate and forms N-phenylcarbamate deprotonated
at the nitrogen atom (B2_K). The transition structure for the
Lossen rearrangement (TS2_K) lies 1.60 eV above the energy
of A1_K and the reaction sequence is 1.44 eV exothermic. The
more stable isomer of potassium N-phenylcarbamate deproto-
nated at the oxygen atom (B3_K) is formed via an energy
barrier (TS3_K), which is substantially lower than those in
both preceding steps. Hence, this step is not expected to play a
role in the kinetics of the reaction. The overall reaction is highly
exothermic in that potassium N-phenylcarbamate is 2.28 eV
more stable than potassium benzhydroxamate.
The reaction could alternatively proceed without the
involvement of a metal and hence for the bare anion generated
from hydroxamic acid upon interaction with a base. The N-
deprotonated forms of free hydroxamates are preferred. For the
benzhydroxamate, the N-deprotonated anion is 0.27 eV more
stable than the O-deprotonated anion. The barrier for the
Lossen rearrangement amounts to 1.87 eV. Hence, the reaction
can proceed also for a free hydroxamate, but the complexation
with a metal decreases the energy barrier for the rearrangement
and thus catalyzes the reaction.
In the condensed phase, the height of the first energy barrier
can be significantly decreased in the presence of proton shuttles
19
such water or other protic molecules. The second energy
barrier corresponding to the actual Lossen rearrangement is
very similar for all reactant complexes. Therefore, it is most
likely that the equilibrium between the isomers A1_M and
A2_M or a direct preferential formation of A2_M is decisive for
20
the overall efficiency of the reaction.
The relative stabilities of A1_M and A2_M can be influenced
by solvent. The reactions were performed in DMF (N,N-
dimethylformamide), which we have here accounted for by
PCM (polarized continuum model) calculations of the solvent
effect of DMF (Table 2). It can be seen that the solvent
significantly stabilizes the N-deprotonated complex (A2_M, cf.
Tables 1 and 2) with respect to the O-deprotonated complex.
For the potassium complexes (and also for the sodium
complexes if we consider the Gibbs energies at 298 Kvalues
in parentheses in Table 2), the reactive N-deprotonated
complex becomes the energetically favored one. The energy
barriers associated with the Lossen rearrangement are very
similar for all studied complexes. We have also computationally
explored effects of other solvents. The relative energies are,
however, almost the same for all studied solvents (N,N′-
dimethylformamide, dimethylsulfoxide, water, and ethanol);
details can be found in the SI. For the free anions in DMF, the
N-deprotonated hydroxamate is even more energetically
preferred to the O-deprotonated anion than in the gas phase.
The barrier for the rearrangements, however, stays on the order
of 1.88 eV, which is about 0.3 eV higher than for the metal
complexes.
In the original paper of Hoshino et al. on the base-mediated
8
Lossen rearrangement, it has been reported that the yield of
the amine formed by the Lossen rearrangement exceeds 90% if
K CO is used as the catalyst (2 h in DMF at 90 °C). If
2
3
Na CO or Li CO are used as catalysts under the same
2
3
2
3
conditions, the yields drop to 81% and 12%, respectively.
Hence, we have performed analogous potential-energy surface
calculations with sodium and lithium cations in order to check,
whether the suggested mechanism accounts for these
experimental results (Table 1). The activation energies for
Table 1. Energetics of the Metal-assisted Lossen
Rearrangement in the Gas Phase Catalyzed by Potassium,
a
Sodium, Lithium, and Zinc
A1_M TS1_M A2_M TS2_M
B2_M
[
[
[
[
[
[
[
[
(1-H)K]
0.00
0.00
0.00
0.00
0.15
0.00
0.00
0.00
0.27
1.70
1.69
1.73
1.77
1.83
2.13
1.99
1.96
1.79
0.12
0.03
0.18
0.22
0.00
0.86
0.63
0.59
1.60
1.58
1.62
1.59
1.97
1.21
1.62
1.56
1.87
−1.44
−1.40
−1.01
−0.77
−1.79
−1.75
−0.74
−0.75
−2.16
For all metals investigated here, the reactive form of the
complex with N-deprotonated hydroxamate is most energeti-
cally disfavored for zinc. We have probed the zinc complexes
also by the calculations in DMF. For the simplest model
(1-H)K(DMF)2]
(1-H)Na]
(1-H)Li]
(2-H)K]+
+
complex [(1-H)Zn] , a large energetic disfavor for the reactive
(1-H)Zn]+
(1-H)Zn(Phen)]+
b
N-deprotonated isomer (Erel = 0.86 eV) has been found in the
gas phase (Table 1), which is most probably due to the
coordinative unsaturation of the zinc ion. If we consider
energies of the complex [(1-H)Zn(DMF)2]+ with two
coordinated molecules of DMF, then the energy difference
between the O- and N-deprotonated complexes drops to 0.59
eV. Consideration of the solvent effect leads to the value of 0.37
eV, which nevertheless is still considerably higher than found
for the alkali metals.
We note in passing that the effect of the number of the
coordinated ligands is much smaller for the potassium
complexes than for the zinc complexes. Relative energies
along the potential-energy surface for the rearrangements
b
b
(1-H)Zn(DMF) ]+
2
(
1-H)¯
0.00
b
a
Relative energies are given at 0 K in eV. Product corresponds to
B1_Zn isomer with separated isocyanate and hydroxo ligands
coordinated to the zinc cation.
the initial hydrogen migration slightly rise in the order K < Na
Li, which is reflected also in increasing relative energies of the
<
N-deprotonated forms of the complex. The relative energy of
the transition structure for the subsequent Lossen rearrange-
ment is about the same for all complexes, which means that
with respect to the energy of the reactive N-deprotonated form,
the energy barrier decreases from the potassium complex (1.48
eV) to the sodium complex (1.44 eV) and is smallest for the
lithium complex (1.37).
within the [(1-H)K(DMF) ] complex, where two molecules of
2
DMF are coordinated to the potassium cation, are very similar
to those for the [(1-H)K] complex (cf. the first two entries in
Table 1). The biggest effect is found for the relative stability of
2
832
dx.doi.org/10.1021/jo300031f | J. Org. Chem. 2012, 77, 2829−2836

The Journal of Organic Chemistry
Article
Table 2. Energetics of the Metal-assisted Lossen Rearrangement in DMF Catalyzed by Potassium, Sodium, Lithium, and Zinca,b
A1_M
TS1_M
A2_M
TS2_M
B2_M
[
[
[
[
[
(1-H)K]
0.03 (0.05)
0.00 (0.02)
0.00 (0.00)
0.00 (0.00)
0.00 (0.00)
0.32 (0.35)
1.73 (1.74)
1.73 (1.79)
1.76 (1.75)
1.97 (1.97)
1.93 (1.88)
1.99 (2.00)
0.00 (0.00)
0.01 (0.00)
0.06 (0.04)
0.46 (0.45)
0.37 (0.34)
0.00 (0.00)
1.59 (1.55)
1.59 (1.57)
1.62 (1.58)
1.55 (1.52)
1.58 (1.54)
1.89 (1.88)
−1.72 (−1.76)
−1.47 (−1.48)
−1.34 (−1.36)
−0.95 (−1.04)
−0.76 (−0.90)
−2.27 (−2.28)
(1-H)Na]
(1-H)Li]
(1-H)Zn]+
(1-H)Zn(DMF) ]+
2
(
1-H)¯
a
b
Relative energies are given at 0 K in eV; the values in parentheses correspond to the relative Gibbs energies at 298 K. Solvent effect is included by
PCM calculations.
the O- and N-deprotonated complexes in that the latter is only
0.03 eV higher in energy than the former.
Hence, let us conclude that the N-deprotonated forms of the
metal complexes of hydroxamates are the key intermediates in
the Lossen rearrangements. The microscopic rates of the
Lossen rearrangement will be similar for all complexes as
implied by the similar barrier heights for this step, but the
overall rate will be largest for the potassium complex, because
the population of the key intermediate, that is, the metal
complex of N-deprotonated hydroxamate, will be the largest.
Simple kinetic modeling of this reaction indeed confirms this
trend.
Having suggested the mechanism for the potassium catalyzed
Lossen rearrangement computationally, we wanted to probe the
structure of the key intermediate also experimentally. The
potassium complexes are neutral and hence cannot by sampled
by ESI-MS. The experiments were therefore performed with a
Figure 4. Energy dependent CID of [K(2-H)]+ (solid symbols,
collision gas: Xe). For the determination of the threshold energies, see
the SI.
21−23
charge-tagged molecule;
namely with benzhydroxamic acid
bearing a trimethylamonium group (the charge-tag) in the meta
position of the aromatic ring (2, (3-hydroxycarbamoylphenyl)-
trimethylamonium cation). The introduction of the charge-tag
results in a larger stabilization of the N-deprotonated form of
energies; the potential-energy surface for the other conformers
are analogous and can be found in the SI.
The experimental IRMPD spectrum of the charge-tagged
+
complex [(2-H)K] is shown in Figure 6a. Its comparison with
the complex (A2 _K). It is 0.18 eV more stable than the O-
tag
the theoretical spectra clearly shows that the structure of the
deprotonated form (A1 _K) and hence we should be able to
tag
isolated ions corresponds to the A2 _K isomer (Figure 6b).
tag
study the reactive N-deprotonated isomer in the gas phase.
The collision-induced dissociation of the charge-tagged
complex leads either to the loss of potassium, or to the
elimination of potassium hydroxide. At larger collision energies
also subsequent fragmentations of the tagged phenylisocyanate
are observed, namely a loss of one methyl group or methane
from the trimethylamonium group and a loss of the NCO
group (Figure 4). Evaluation of the collision-energy dependent
Hence, we unequivocally detect the potassium complex with
benzhydroxamate deprotonated at the nitrogen atom. A closer
examination of the experimental spectrum reveals that a weak
−1
band at 1165 cm does not correspond to any peak in the
theoretical spectrum of A2 _K and can be only interpreted, if
tag
we take into account also a minor population of the less stable
isomer A1 _K (Figure 6c). This band hence most probably
tag
corresponds to the N−O stretch of the isomer with O-
deprotonated hydroxamate.
+
CIDs for the K channel and the KOH-loss channel leads to the
same threshold energies (Ethres = 1.7 ± 0.1 eV; the energy
demand is about 0.3 eV lower that predicted by the
computations; see the SI).
Under the conditions of the IRMPD experiment eliminations
+
of K and KOH are observed. The IRMPD spectra obtained, if
these channels are considered separately, are almost identical
−
1
The potential-energy surface for the Lossen rearrangement of
(
see the SI). The only difference is a band at about 1660 cm ,
the potassium complex of the tagged hydroxamate [(2-H)K]+
+
which can be observed only in the elimination of K . This band
can be interpreted as the CO stretching vibration of the
carbamate product (Figure 6d). In agreement, the elimination
of KOH from the carbamate product is not expected.
(Figure 5) is analogous to that of [(1-H)K], only the energy
barriers are slightly higher. Elimination of the potassium cation
is possible for all structures along the reaction coordinate, but
the energy demands for the complexes A1 _K and A2 _K are
tag
tag
Finally, we have probed to which extent the conclusions from
the gas-phase experiments can be transformed into the
synthetic procedures. To this end, the Lossen rearrangement
of p-methylbenzhydroxamic acid 3 was investigated. First, we
+
very large (around 3 eV). Therefore, the elimination of K as
well as KOH can only be observed after the Lossen
rearrangement, which is consistent with the identical energy
thresholds of these fragmentations in the experiment. Note that
two orientations of the hydroxamate group with respect to the
trimethylamonium group are possible (rotation of the
hydroxamate group around the C−C bond). We show only
one variant in Figure 5 and also discuss only one set of
8
have repeated the experiment reported by Hoshino et al. In
agreement, we have obtained quantitative yield of the product,
p-methylaniline, after 2 h heating of p-methylbenzhydroxamic
acid with 1 equiv of K CO in DMF at 90 °C. The same
2
3
experiment with ZnCO did not lead to product formation.
3
2
833
dx.doi.org/10.1021/jo300031f | J. Org. Chem. 2012, 77, 2829−2836

The Journal of Organic Chemistry
Article
Figure 5. B3LYP/TZVP potential-energy surface for the potassium-assisted Lossen rearrangement of the charge-tagged benzhydroxamic acid.
0K
Energies refer to 0 K in the gas phase and are given relative to E (A2 _K) = −1249.156315 hartree. The lowest line shows the fragmentation
tag
+
pathways for the eliminations of K or KOH, respectively.
The reaction mixtures are heterogeneous; therefore it is
difficult to estimate the actual concentration of the catalyst. In
order to make a direct comparison, we have taken soluble metal
acetates as catalysts (Table 3). These salts are less basic and
therefore the deprotonation of the hydroxamic acid is slower or
less efficient. Nevertheless, even though acetic acid is a stronger
acid than hydroxamic acids, the transmetalation between
acetate and hydroxamate occurs due to the chelating
2
4,25
capabilities of the hydroxamic acids.
The reaction of the
hydroxamic acid with potassium acetate in DMSO at 140 °C
gives 57% yield after 2 h. Under the same conditions, the
catalysis by zinc acetate leads only to 5% of the product and
even after 8 h only 18% of the product is formed. This
experiment thus clearly shows that coordination to a metal
leads to the Lossen rearrangement. The small yield in the zinc-
catalyzed reaction agrees well with the suggested mechanism.
To exclude mechanisms involving the formation of a
carbamoylhydroxamate (Scheme 2), we have prepared the
potassium hydroxamate salt in a separate step, dissolved it in
water and refluxed for two hours. This procedure leads to 47%
yield of the product (the remaining part was the unreacted
salt). The use of water as the reaction medium prevents the
formation of carbamoylhydroxamate, and therefore, the
suggested metal-assisted rearrangement remains as the only
possible mechanistic variant for this reaction.
CONCLUSIONS
■
Using a combined approach of several experimental methods
and theory we have demonstrated that the Lossen rearrange-
ment can be efficiently catalyzed by metals. The reaction
sequence starts with the formation of a metal complex of a
deprotonated hydroxamate. The classical metal complexes are
deprotonated at the oxygen atom of the hydroxamates. The
Lossen rearrangement is initiated by a hydrogen migration from
the nitrogen atom to the oxygen atom (Scheme 3). The
nitrogen-deprotonated form of the metal hydroxamate under-
goes the Lossen rearrangement to yield a metal carbamate. The
last step consists in a facile hydrogen migration from oxygen to
nitrogen to yield the final product.
We have characterized complexes between zinc and
benzhydroxamate using IRMPD spectroscopy and proved
that the hydroxamate is deprotonated at the oxygen atom, as
expected. We have also characterized a complex between
potassium and benzhydroxamate bearing a trimethylamonium
group (charge tag) at the aromatic ring. It is unequivocally
shown that in the tagged complex the hydroxamate is
deprotonated at the nitrogen atom. DFT calculations show
that the same situation can be expected also for the nontagged
potassium benzhydroxamate, if the effect of the solvent is taken
into account. The potential-energy surface for the metal-
catalyzed Lossen rearrangement reveals that the reaction
+
Figure 6. IRMPD spectrum of [(2-H)K] (a) and its comparison with
calculated IR spectra of different isomers of the complex (b-d). The
theoretical spectra are shown as gray bars, the black solid lines show
the theoretical spectra folded with a Gaussian function (fwhm =16
cm ). Notations I , I and ΣI refer to the relative intensities of the
fragments K (m/z 39), [(2-H O)] (m/z 177) and the sum of both
−1
39
177
+
+
2
+
mentioned intensities and that of the parent ion [(2-H)K] (m/z 233),
respectively.
2
834
dx.doi.org/10.1021/jo300031f | J. Org. Chem. 2012, 77, 2829−2836

The Journal of Organic Chemistry
Article
Table 3. Lossen Rearrangement of p-Methylbenzhydroxamic Acid (3) under Different Conditions
a
entry
catalyst
K CO
solvent
temperature (°C)
time (h)
yield (%)
3
3
3
3
DMSO
DMSO
DMSO
DMSO
140
140
140
140
100
2
2
2
8
2
quantitative
2
3
CH COOK
57
5
3
Zn(CH COO)
3
2
2
Zn(CH COO)
18
47
3
[
K(3-H)]
H O
2
a1H NMR yields.
3
3
Scheme 3. Mechanism of the Metal-assisted Lossen
Rearrangement
agreement between theory and experiment. The scaling factors were
.980 for the zinc complexes and 0.965 for the potassium complexes.
For the transition structures, also intrinsic reaction coordinates were
0
3
4,35
calculated at the same level of theory.
All minima and transition
states were further reoptimized using the polarized continuum model
3
6
for modeling the effect of solvent and frequency calculations were
again performed in order to control the identity of the stationary
points as well as to obtain thermochemical corrections for energies at 0
K and Gibbs energies at 298 K.
All chemicals were obtained from commercial sources (including 3-
1
(
hydroxycarbamoylphenyl)trimethylamonium iodide). H NMR spec-
1
tra were recorded in a DMSO-d solutions ( H NMR at 300 MHz). 4-
6
Methylbenzhydroxamic acid (3) was prepared according to previously
8
reported procedures. The potassium salt of 4-methylbenzhydroxamic
acid ([K(3-H)]) was prepared by neutralization of the acid 3 with
KOH in methanol. Detailed descriptions of preparation of [K(3-H)]
1
and all H NMR spectra can be found in the Supporting Information.
ASSOCIATED CONTENT
■
*
S
Supporting Information
Data, evaluation, and details about collision-induced dissocia-
tion experiments, IRMPD experiments and their evaluation for
the [(1-H)(1)Zn]+ complex, theoretical results (optimized
geometries and energetics for all calculated structures) and
complete ref 32. This material is available free of charge via the
Internet at http://pubs.acs.org.
efficiency is associated mainly with relative stabilities of O- and
N-deprotonated forms of the hydroxamate. For the potassium
complexes, the largest ratio of the reactive N-deprotonated
complex is expected, which correlates with largest yields of the
product of the Lossen rearrangement observed experimentally.
EXPERIMENTAL AND COMPUTATIONAL DETAILS
AUTHOR INFORMATION
■
■
Gas-phase infrared (IR) spectra of mass-selected ions were recorded
using an ion trap mass spectrometer mounted to a free electron laser at
Corresponding Author
*roithova@natur.cuni.cz
Notes
26
CLIO (Centre Laser Infrarouge Orsay, France). The ions were
generated by electrospray ionization (ESI) of methanolic solutions of a
metal salt and the hydroxamic acid of interest (for the zinc complexes
phenanthroline was added as a coligand). The free electron laser
The authors declare no competing financial interest.
(
FEL) was operated at 44 MeV electron energy which provided light
ACKNOWLEDGMENTS
−1
■
in a 1000−1800 cm range. The relative spectral line width of the
FEL is of about 1% and the precision of the measurement of the
We gratefully acknowledge financial support from the Grant
agency of the Czech Republic (207/11/0338), the European
Research Council (StG ISORI), and the Ministry of Education
of the Czech Republic (MSM0021620857). Philippe Maitre
and the CLIO staff, particularly Vincent Steinmetz, are
gratefully acknowledged for their help and assistance.
−1
wavenumbers with a monochromator is about 1 cm . The ions were
mass-selected and stored in the ion trap. The fragmentation was
induced by 5 to 10 laser macropulses of 8 μs admitted to the ion trap.
The dependence of the fragmentation intensities on the wavelength of
the IR light gives the infrared multiphoton dissociation (IRMPD)
spectra. The reported IRMPD spectra are averages of 2 raw spectra.
Each point in a raw spectrum at a given wavelength is obtained by the
evaluation of 4 mass spectra in that each mass spectrum is an average
of 5 measurements. The IRMPD spectra are not corrected for the
power of the free-electron laser, which slightly changes in dependence
of the wavenumbers (Figure S4 the SI).
REFERENCES
■
(1) Hurd, C. D.; Bauer, L. J. Am. Chem. Soc. 1954, 76, 2791.
(2) Stafford, J. A.; Gonzales, S. S.; Barrett, D. G.; Suh, E. M.;
Feldman, P. L. J. Org. Chem. 1998, 63, 10040.
Density functional theory calculations were performed using the
(3) Stolberg, M. A.; Tweit, R. C.; Steinberg, G. M.; Wagner-Jauregg,
T. J. Am. Chem. Soc. 1955, 77, 765.
(4) Pihuleac, J.; Bauer, L. Synthesis 1989, 61.
(5) (a) Hoare, D. G.; Olson, A.; Koshland, D. E. Jr. J. Am. Chem. Soc.
1968, 90, 1638.
27−30
B3LYP hybrid functional
together with the triple ζ basis set
31
32
TZVP as implemented in the Gaussian 09. All minima and
transition structures were verified by analysis of their Hessian matrixes.
The calculated infrared spectra are scaled according to the best
2
835
dx.doi.org/10.1021/jo300031f | J. Org. Chem. 2012, 77, 2829−2836

The Journal of Organic Chemistry
Article
(
6) Narendra, N.; Chennakrishnareddy, G.; Sureshbabu, V. V. Org.
Biomol. Chem. 2009, 7, 3520.
7) Dube, P.; Nathel, N. F. F.; Vetelino, M.; Couturier, M.; Larrivee
Aboussafy, C.; Pichette, S.; Jorgensen, M. L.; Hardink, M. Org. Lett.
009, 11, 5622.
8) Hoshino, Y.; Okuno, M.; Kawamura, E.; Honda, K.; Inoue, S.
Chem. Commun. 2009, 2281.
(28) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785.
(29) Miehlich, B.; Savin, A.; Stoll, H.; Preuss, H. Chem. Phys. Lett.
1989, 157, 200.
(30) Becke, A. D. J. Chem. Phys. 1993, 98, 5648.
(31) Schaefer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1994, 100,
5829.
(32) Frisch, M. J. et al. Gaussian 09, Revision A.02; Gaussian, Inc.:
Wallingford, CT, 2009.
(33) For the method B3LYP/6-311+G(d,p) (the basis set is closest
to our selection from those that have been systematically studied) a
scaling factor of 0.9688 is suggested in work: Merrick, J. P.; Moran,
D.; Radom, L. J. Phys. Chem. A 2007, 111, 11683 . This scaling,
however, does not lead to an agreement with the experimental spectra..
(34) Gonzalez, C.; Schlegel, H. B. J. Chem. Phys. 1989, 90, 2154.
(35) Gonzalez, C.; Schlegel, H. B. J. Phys. Chem. 1990, 94, 5523.
(36) Tomasi, J.; Mennucci, B.; Cammi, R. Chem. Rev. 2005, 105,
2999.
(
2
(
(
(
9) Duchac
́
k
̌
ova,
́
L.; Roithova,
́
J. Chem.Eur. J. 2009, 15, 13399.
10) (a) Bauer, L.; Exner, O. Angew. Chem., Int. Ed. Engl. 1974, 13,
376. (b) Exner, O.; Hradil, M.; Mollin, J. Collect. Czech. Chem.
Commun. 1993, 58, 1109. (c) Bagno, A.; Comuzzi, C.; Scorrano, G. J.
Am. Chem. Soc. 1994, 116, 916. (d) Senent, M. L.; Nino, A.; Caro, C.
M.; Ibeas, S.; Garcia, B.; Leal, J. M.; Secco, F.; Venturini, M. J. Org.
Chem. 2003, 68, 6535.
(
11) Codd, R. Coord. Chem. Rev. 2008, 252, 1387 and references
therein..
(
12) (a) MacAleese, L.; Maitre, P. Mass Spectrom. Rev. 2007, 26, 583
and references therein. (b) Polfer, N. C.; Oomens, J. J. Mass Spectrom.
Rev. 2009, 28, 468. (c) Roithova,
13) Duchac ova, L.; Schroder, D.; Roithova,
0, 3153.
14) Wentrup, C. Acc. Chem. Res. 2011, 44, 393 and references
therein..
́
J. Chem. Soc. Rev. 2012, 41, 547.
(
́
k
̌
́
̈
́
J. Inorg. Chem. 2011,
5
(
(
15) (a) Armentrout, P. B.; Ervin, K. M.; Rodgers, M. T. J. Chem.
Phys. 2008, 112, 10071. (b) Amicangelo, J. C.; Armentrout, P. B. Int. J.
Mass Spectrom. 2011, 301, 45.
(
16) Narancic, S.; Bach, A.; Chen, P. J. Phys. Chem. A 2007, 111,
006.
17) Albu, T. V.; Lynch, B. J.; Truhlar, D. G.; Goren, A. C.; Hrovat,
D. A.; Borden, W. T.; Moss, R. A. J. Phys. Chem. A 2002, 106, 5323.
18) (a) Duchac ova, L.; Steinmetz, V.; Lemaire, J.; Roithova, J.
Inorg. Chem. 2010, 49, 8897. (b) Rezabel, E.; Duchac ova, L.; Milko,
P.; Holthausen, M. C.; Roithova, J. Inorg. Chem. 2010, 49, 8421.
19) Schroder, D.; Roithova, J; Gruene, P; Schwarz, H; Mayr, H;
Koszinowski, K. J. Phys. Chem. A 2007, 111, 8925.
7
(
(
́
k
̌
́
́
́
k
̌
́
́
(
̈
́
(
20) Kozuch, S.; Shaik, S. Acc. Chem. Res. 2011, 44, 101.
(
21) Examples of ESI-MS studies of reaction mechanisms: (a) Putau,
A.; Koszinowski, K. Organometallics 2011, 30, 4771. (b) Robbins, D.
W.; Hartwig, J. F. Science 2011, 333, 6048. (c) Agrawal, D.; Schroder,
̈
D.; Frech, C. M. Organometallics 2011, 30, 3579. (d) Schwarz, H.
Angew. Chem., Int. Ed. 2011, 50, 10096. (e) Perry, R. H.; Splendore,
M.; Chien, A.; Davis, N. K.; Zare, R. N. Angew. Chem., Int. Ed. 2011,
50, 250. (f) Coelho, F.; Eberlin, M. N. Angew. Chem., Int. Ed. 2011, 50,
5261. (g) Koszinowski, K. J. Am. Chem. Soc. 2010, 132, 6032.
(
h) Wassenaar, J.; Jansen, E.; van Zeist, W. J.; Bickelhaupt, F. M.;
Siegler, M. A.; Spek, A. L.; Reek, J. N. H. Nat. Chem. 2010, 2, 417.
(
(
(
́
i) Roithova, J.; Milko, P. J. Am. Chem. Soc. 2010, 132, 281.
j) Schlangen, M.; Schwarz, H. Chem. Commun. 2010, 46, 1878.
k) Couzijn, E. P. A; Zocher, E.; Bach, A.; Chen., P. Chem.Eur. J.
2
010, 16, 5408. (l) Fedorov, A.; Couzijn, E. P. A; Nagornova, N. S.;
Boyarkin, O. V.; Rizzo, T. R.; Chen, P. J. Am. Chem. Soc. 2010, 132,
3789. (m) Crestoni, M. E.; Fornarini, S.; Lanucara, F.; Warren, J. J.;
1
Mayer, J. M. J. Am. Chem. Soc. 2010, 132, 4336. (n) Amarante, G. W.;
Milagre, H. M. S.; Vaz, B. G.; Ferreira, B. R. V.; Eberlin, M. N.;
Coelho, F. J. Org. Chem. 2009, 74, 3031. (o) Crestoni, M. E.;
Fornarini, S. Inorg. Chem. 2005, 44, 5379.
(
22) Review articles: (a) Eberlin, M. N. Eur. J. Mass Spectrom. 2007,
3, 19. (b) Chen, P. Angew. Chem., Int. Ed. 2003, 42, 2832.
23) Schade, M. A.; Feckenstein, J. E.; Knochel, P.; Koszinowski, K. J.
Org. Chem. 2010, 75, 6848.
24) Vannini, A.; Volpari, C.; Filocamo, G.; Casavola, E. C.; Brunetti,
1
(
(
M.; Renzoni, D.; Chakravarty, P.; Paolini, C.; De Francesco, R.;
Gallinari, P.; Steinkuhler, C.; Di Marco, S. Proc. Natl. Acad. Sci. U.S.A.
2
004, 101, 15064.
25) Dowling, D. P.; Gantt, S. L.; Gattis, S. G.; Fierke, C. A.;
Christianson, D. W. Biochemistry 2008, 47, 13554.
26) Ortega, J. M.; Glotin, F.; Prazeres, R. Infrared Phys. Technol.
006, 49, 133.
27) Vosko, S. H.; Wilk, L.; Nusair, M. Can. J. Phys. 1980, 58, 1200.
(
(
2
(
2
836
dx.doi.org/10.1021/jo300031f | J. Org. Chem. 2012, 77, 2829−2836