The molecularity of the Newman-Kwart rearrangement

Article abstract of DOI:10.1021/jo1014382

It was recently reported that the venerable Newman-Kwart rearrangement (1→2) proceeds via mixed first- and second-order kinetics. Prior to this, the rearrangement had been considered to proceed exclusively via an intramolecular O_Ar→S_Ar migration. A new bimolecular pathway, possibly involving an 8-membered cyclic transition state, was proposed to account for reaction rates that increased disproportionately with substrate concentration under microwave heating conditions. We report a reanalysis of the kinetics and molecularity of the rearrangement of N,N-dimethyl O-(p-nitrophenyl)thiocarbamate 1a in N,N-dimethylacetamide solvent. Using HPLC, isotopic labeling (²H, ¹⁸O, ³⁴S), and ESI-ICRMS methods, we show that there is no evidence for a bimolecular pathway en route to 2a, with near-perfect exponential decay in 1a at concentrations ranging from 0.11 to 4.70 M. Instead, it is demonstrated that under the microwave heating conditions, a delayed negative feedback signal to the microwave power balancing loop results in oscillatory reaction overheating. Due to higher tan δ in the solute, the amplitude of this oscillation increases with the concentration of 1a, and this phenomenon best accounts for the kinetic behavior previously misinterpreted as being mixed 'irst- and second-order in nature.

Full text of DOI:10.1021/jo1014382

pubs.acs.org/joc
The Molecularity of the Newman-Kwart Rearrangement
Matthew Burns,† Guy C. Lloyd-Jones,*,† Jonathan D. Moseley,‡ and Joseph S. Renny†
†School of Chemistry, University of Bristol, Cantock’s Close, Bristol BS8 1TS, U.K., and
‡AstraZeneca, Process Research and Development, Avlon Works, Hallen, Bristol BS10 7ZE, U.K.
guy.lloyd-jones@bris.ac.uk
Received July 22, 2010
It was recently reported that the venerable Newman-Kwart rearrangement (1f2) proceeds via
mixed first- and second-order kinetics. Prior to this, the rearrangement had been considered to
proceed exclusively via an intramolecular OArfSAr migration. A new bimolecular pathway, possibly
involving an 8-membered cyclic transition state, was proposed to account for reaction rates that
increased disproportionately with substrate concentration under microwave heating conditions.
We report a reanalysis of the kinetics and molecularity of the rearrangement of N,N-dimethyl
O-(p-nitrophenyl)thiocarbamate 1a in N,N-dimethylacetamide solvent. Using HPLC, isotopic
labeling (2H, 18O, 34S), and ESI-ICRMS methods, we show that there is no evidence for a bimolecular
pathway en route to 2a, with near-perfect exponential decay in 1a at concentrations ranging from
0.11 to 4.70 M. Instead, it is demonstrated that under the microwave heating conditions, a delayed
negative feedback signal to the microwave power balancing loop results in oscillatory reaction
overheating. Due to higher tan δ in the solute, the amplitude of this oscillation increases with the
concentration of 1a, and this phenomenon best accounts for the kinetic behavior previously
misinterpreted as being mixed first- and second-order in nature.
Introduction
intermediates, copper-protein mimics, and precursors to
supramolecular assemblies.4
The thermally induced rearrangement of O-arylthiocarba-
mates (1) was discovered in the 1960s1 and is commonly
referred to as the Newman-Kwart rearrangement2 (NKR).
The ready preparation of N,N-dimethyl O-arylthiocarba-
mates from the corresponding phenol,3 and the facile cleav-
age of the S-arylthiocarbamate rearrangement products (2),
has resulted in the NKR becoming one of the prime routes
to thiophenols, Scheme 1, with applications as diverse as
chiral ligands, agrochemicals, dyestuffs, pharmaceutical
In addition to synthetic utility, the mechanism of the NKR
has also attracted considerable attention.4 Soon after its
discovery,1 independent investigation by Relles et al.5 and
by Miyazaki6 led to the proposal of unimolecular rearrange-
ment (k1) proceeding via a four-membered 1,3-oxathietane
transition state (3, Figure 1); this is fully consistent with
extensive kinetic data and a series of linear free energy rela-
tionship analyses.5,6 Well over three decades later, high-level
DFT studies7 fully supported the proposed mechanism and
(1) (a) Edwards, J. D.; Pianka, M. J. Chem. Soc. 1965, 7338. (b) Kwart,
H.; Evans, E. R. J. Org. Chem. 1966, 31, 410. (c) Newman, M. S.; Karnes,
H. A. J. Org. Chem. 1966, 31, 3980.
(4) Reviews on the NKR: (a) Zonta, C.; Lucchi, O. D.; Volpicelli, R. Top.
Curr. Chem. 2007, 275, 131. (b) Lloyd-Jones, G. C.; Moseley, J. D.; Renny,
J. S. Synthesis 2008, 661.
(2) Zonta et al. (ref 4a) have suggested that the process be referred to as
the Newman-Kwart-Miyazaki rearrangement to reflect the extensive
mechanistic work conducted by Miyazaki (ref 6).
(5) Relles, H. M.; Pizzolato, G. J. Org. Chem. 1968, 33, 2249.
(6) (a) Miyazaki, K. Tetrahedron Lett. 1968, 23, 2793. (b) Kaji, A.; Araki,
Y.; Miyazaki, K. Bull. Chem. Soc. Jpn. 1971, 44, 1393.
(3) Ponaras, A. A.; Zaim, O. Dimethylthiocarboamyl Chloride. In
Encyclopedia of Reagents for Organic Synthesis; Paquette, L., Ed.; John
Wiley: New York, 1995; p 2174.
(7) (a) Albrow, V.; Biswas, K.; Crane, A.; Chaplin, N.; Easun, T.;
Gladiali, S.; Lygo, B.; Woodward, S. Tetrahedron: Asymmetry 2003, 14,
2813. (b) Jacobsen, H.; Donahue, J. P. Can. J. Chem. 2006, 84, 1567.
DOI: 10.1021/jo1014382
r
Published on Web 09/02/2010
J. Org. Chem. 2010, 75, 6347–6353 6347
2010 American Chemical Society

Burns et al.
JOCFeatured Article
SCHEME 1. NKR Route for the Preparation of Thiophenols
from Phenolsa
FIGURE 1. Unimolecular (k1; via 35-7) and bimolecular mecha-
nisms (k2; via 413) for NKR.
aReagents and conditions: (i) DABCO or NaH, ClC(S)NMe2; NMP
or DMF; (ii) NKR, Δ=200-300 °C; (iii) NaOH (aq) or KOMe, Δ;
or LiAlH4; or Na, NH3.
pared to microwave heating; a large data set was then
gathered under microwave heating conditions and analyzed
to determine the effects of substituent, solvent, and concen-
tration on the reaction order.13 In accord with previous
studies,5,6 a strong dependence of reaction rate on the nature
of aryl ring substituent was observed, with only a modest
effect of solvent.12b,c,e However, in contrast to all prior studies
on NKR,4-6 it was reported that as the initial concentration
of 1 was raised, disproportionately higher rates of rearrange-
ment occurred. The kinetics were analyzed by application of
a mixed first-order/second-order model (k1 and k2, Figure 1),
with a cyclic 8-membered transition state (4) tentatively sug-
gested for the second-order process (k2). The second-order
process was proposed to be kinetically significant above
concentrations of ca. 2.0 M, and it is pertinent to note that
all previous studies on the kinetics of the solution-phase
NKR were conducted on relatively dilute solutions of 1
(0.05-0.3 M).1c,5
In light of the fact that the unimolecular nature of the
NKR has become so well-accepted that the reaction has been
used as a “benchmark” first-order process,14 we felt that a
more detailed search for evidence for a bimolecular pathway
is essential. Herein we report on an investigation into the
molecularity of the NKR of N,N-dimethyl O-p-nitro-
phenylthiocarbamate 1a (X = p-NO2) in N,N-dimethyl-
acetamide (DMA) under the reaction conditions for which mixed
first-order/second-order rate analyses had been reported.13
confirmed 1,3-oxathietane 3 as a transition state rather than
an intermediate.
Thermal NKR typically requires4 temperatures of 200-
300 °C to facilitate access to the high energy 1,3-oxathietane
transition state 3. This means that fragile substrates, e.g.
those that decompose, racemize, or undergo other undesired
reactions at these temperatures, cannot readily be used for
NKR.4 A recently developed Pd-catalyzed NKR offers a
partial solution to this problem, allowing rearrangement at
100 °C; however, substrate scope for this process is currently
limited.8
Even for substrates and products that are nominally stable
at the temperatures required for the thermal NKR, the
temperature gradient induced by the external heating source
(oil bath, mantle etc.) can result in extensive “charring” on
the inner surface of the reaction vessel.9 Short exposure
to much higher temperatures (400-700 °C) can provide a
highly effective solution to these problems but requires
specialized equipment.10 However, the most readily appli-
cable method for inducing efficient NKR is the microwave
heating approach, first reported by Villemin in 1996 for
reactions conducted on a graphite support11 and then exten-
sively developed by Moseley and co-workers in 2006 for
solution-phase reactions.12
In 2008, Gilday et al. published a study on the kinetics of
the NKR under microwave heating,13 the principal aim
of which was to demonstrate the utility of the method for
the rapid generation of kinetic data from single time point
experiments. For control experiments employing 0.44 M
solutions in xylene, it was found that identical rates were
obtained under conventional convective heating as com-
Results and Discussion
All of the substrates studied by Gilday et al.13 were
reported to display disproportionately higher rates of
rearrangement at higher concentrations; we thus chose to
focus on one example in detail. The 4-nitro substrate 1a, a
crystalline solid with well-established procedures for purifi-
cation, was selected for its relatively low rearrangement
temperature, meaning that in addition to microwave heating,
conventional heating by immersion in a silicone oil bath at
160 °C could also be conveniently applied; DMA was
selected as the solvent.
(8) Harvey, J. N.; Jover, J.; Lloyd-Jones, G. C.; Moseley, J. D.; Murray,
P. M.; Renny, J. S. Angew. Chem., Int. Ed. 2009, 48, 7612.
(9) In addition, the negative effect of trace impurities in the substrate on
the efficacy of the NKR (see ref 4b for a more detailed discussion) may well
arise from catalytic or autocatalytic charring/decomposition processes.
(10) See, for example, the technique of “direct insertion vapor phase flash
vacuum pyrolysis’’ (DIVPFVP): Higgs, T. C.; Carrano, C. J. Eur. J. Org.
Chem. 2002, 3632.
Reaction Kinetics: Convective Heating. The instantaneous
partitioning between the proposed first- and second-order
processes, Figure 1, is given by k1/k2[1a], this being the
(11) Villemin, D.; Hachemi, M.; Lalaoui, M. Synth. Commun. 1996, 26,
2461.
(12) (a) Moseley, J. D.; Lenden, P.; Lockwood, M.; Ruda, K.; Sherlock,
J.-P.; Thomson, A. D.; Gilday, J. P. Org. Process Res. Dev. 2008, 12, 30.
(b) Moseley, J. D.; Lenden, P.; Thomson, A. D.; Gilday, J. P. Tetrahedron
Lett. 2007, 48, 6084. (c) Moseley, J. D.; Lenden, P. Tetrahedron 2007, 63,
4120. (d) Moseley, J. D.; Lawton, S. J. Chem. Today 2007, 27 (2), 16.
(e) Moseley, J. D.; Sankey, R. F.; Tang, O. N.; Gilday, J. P. Tetrahedron
2006, 62, 4685.
(13) Gilday, J. P.; Lenden, P.; Moseley, J. D.; Cox, B. G. J. Org. Chem.
2008, 73, 3130.
(14) Obermayer, D.; Gutmann, B.; Kappe, C. O. Angew. Chem., Int. Ed.
2009, 48, 8321.
6348 J. Org. Chem. Vol. 75, No. 19, 2010

Burns et al.
JOCFeatured Article
FIGURE 2. Left-hand graph: plots of ln[1a]0/[1a]t = kobst; for NKR of 1a at initial concentrations of 0.4, 0.9, 1.7, and 3.0 M; heating by partial
immersion in an oil-bath at 160 °C. Right-hand graph: the effect of initial concentration of 1a on kobs in DMA; heating in sealed capillary tubes
in a GC oven at 160 °C.
nominal ratio of 2a generated via 3a versus 4a. Using the rate
constants reported by Gilday et al. for the microwave-
induced NKR of 1a in DMA at 160 °C13 (k1 = 16.6 ꢀ
10-5 s-1; k2 = 5.41 ꢀ 10-5 M-1 s-1), the concentration of
1a required for equal partitioning is 3.0 M, and a substantial
deviation from a simple exponential decay in 1a should be
evident as this concentration is approached. We thus con-
ducted NKR on 1a at initial concentrations of 0.11, 0.40,
0.90, 1.71, and 3.0 M in DMA, as well as neat (4.70 M) in
thick-walled boiling tubes, under a nitrogen atmosphere,
using a silicone oil-bath set at 160 °C. Small samples were
removed, via a glass capillary, and rapidly cooled, and then
the conversion (1af2a) was analyzed by HPLC using pre-
determined response factors. In all cases, analysis according
to an integrated second-order rate-law (1/[1a]t - 1/[1a]0 vs t)
gave plots that displayed distinct upward curvature (see the
Supporting Information), whereas very good linear correla-
tions (R2 = 0.9960-0.9999) were obtained by applying a
standard integrated first-order rate-law (ln [1a]0/[1a]t =
Using the published thermodynamic parameters for the
NKR of 1a,13,15 a 2.0 °C change in the reaction temperature
is predicted to result in a change in rate of ca. 20%, and we
suspected that the temperature control under the oil-bath
heating conditions was insufficiently accurate. We thus
conducted a series of NKR in parallel, sealing the solutions
in capillary tubes and then simultanously heating them in
a GC oven that had been precalibrated to 160 °C. This
technique ensured homogeneous heating of all the samples,
and analysis of the data gave a unimolecular rate constant of
kobs = 20.7((1.5) ꢀ 10-5 s-1 at initial concentrations of 1a
ranging from 0.173 to 4.70 M, Figure 2. We therefore find no
evidence for any significant effect of concentration on the
rate of NKR of 1a other than that expected for a standard
first-order relationship: d[1a]/dt = k1[1a], possibly aug-
mented by a very small increase in kobs arising from the
change in the medium12b,c,e from predominantly DMA to
exclusively 1a/2a.
Reaction Kinetics: Microwave Heating. Having found no
evidence for higher molecularity in the NKR of 1a under
convective heating conditions, we then examined whether the
situation was different under microwave heating conditions.
Using the unimolecular and bimolecular rate constants k1
and k2 published by Gilday et al.,13 for NKR at 160 °C, the
concentration-normalized relative initial rates (rnorm) for 0.5,
0.8 and 1.5 M samples of 1a are calculated to be 1.00, 1.09, and
1.28, respectively.17 In other words, over a standard time
interval in the initial phase of reaction, the fractional conver-
sion of 1a (%) is predicted to be about 9% and 28% higher in
0.8 and 1.5 M samples, respectively, as compared to the 0.5 M
sample, by way of the bimolecular process (k2).
kobst), affording an average kobs value of 16.8 ꢀ 10-5 s-1 15
.
Although this value corresponds well with the unimolecular
rate constant (k1 = 16.6 ꢀ 10-5 s-1) reported by Gilday
et al.,13 there was substantial variation between repeated
runs.16
(15) Miyazaki reported a value for k1 at 170 °C of 74.0 ꢀ 10-5 s-1 (ref 6a);
Newman conducted rearrangements at 0.0477 and 0.221 M in Carbowax 400
at 180 °C and found no difference in conversion (<10%) by UV analysis
(ref 1c). Relles conducted rearrangement of the o-tert-butyl analogue of
p-nitro substrate 1a at 0.3 and 1.2 M and found no change in k1 (ref 5).
(16) The major cause of this was probably variation in the degree of
cooling of the nonsubmerged section of the thick-walled reaction vessel by
the air current passing through the fume hood. The cooling caused by this
process was constant within a run, as evidenced by the excellent correlation of
(ln([1a]0/[1a]t) = kobst but varied between runs due to different positions in
the oil bath and different extent of fume hood sash opening, etc. Also of note
was that when NKR was conducted on neat samples of 1a (4.70 M) the
volume of fused 1a was considerably smaller than that of the DMA solutions
used for runs at lower concentrations, meaning that the bulk of the reaction
mixture was well below the surface of the silicone oil bath, leading to
noticeably greater kobs values. See the Supporting Information for further
details.
When we conducted the NKR of 1a under the reported
microwave conditions,18 the appearance of plots of fractional
(17) The initial rate ratio (ri / rii) under conditions i and ii is given
by [{k1[1ai]0 þ k2([1ai]0)2}/{k1[1aii]0 þ k2([1aii]0)2}]; this can then be
concentration-normalized as rnorm = ([1aii]0/[1ai]0)(ri/rii) to reflect the rela-
tive rate acceleration attained by the second-order (bimolecular) process.
(18) See the Supporting Information in ref 13 for details.
J. Org. Chem. Vol. 75, No. 19, 2010 6349

Burns et al.
JOCFeatured Article
SCHEME 2. Preparation of Isotopically Labeled O-Aryl and
S-Aryl Thiocarbamatesa
aReagents and conditions: (i) DCl, D2O, Δ, 82% yield (93% D2, 7%
D1); (ii) DABCO, ClC(S)NMe2, DMF; (iii) 180 °C, N2, 60 min, quant;
FIGURE 3. Apparent concentration dependence on the temporal
fractional conversion for NKR 1af2a under microwave heating
conditions,13 160 °C and 140 °C in DMA for 0.5 M (blue), 0.8 M
(orange), and 1.5 M (red) initial concentrations of 1a. Data points
are single time-point experiments; i.e., they originate from separate
reactions run to the time-point given on the x-axis, cooled, and then
analyzed for conversion by HPLC.
(iv) THF, 18OH2, t-BuOK, 95 °C, 24 h, 98% yield (two samples: 88% 18
O
and 65% 18O); (v) 34S8, Na2S 9H2O, EtOH, reflux, then NaOH, then
3
HCl, 43% yield (52% 34S); (vi) triphosgene, DCM, 0 °C, followed by
(CD3)2NH2Cl, Et3N, THF; 20% yield (>99% atom % D).
SCHEME 3. Lack of Crossover on NKR of 1a and Exchange
between Isotopically Labeled S-Aryl Thiocarbamates under
Thermal Aerobic Conditionsa
conversion of 1a (%) versus time were similar to those
reported byGildayet al.13 inthattheconversionsof1ato2aat
0.8 M/1.5 M initial concentrations were indeed greater than
that at 0.5 M, Figure 3. In fact, the conversions were sig-
nificantly greater: an average of 26% and 44%, respectively,
over the first 1800 s.19
Preparation of Isotopically Labeled Substrates. In contrast
to unimolecular NKR proceeding via 3a, conversion of 1a
into 2a via the 8-membered ring in 4a, irrespective of whether
the latter is a transition state or intermediate, will result in
100% crossover between the aromatic moiety and the thio-
carbamoyl moiety. The molecularity of the NKR was probed
in the early studies of Miyazaki, who coreacted O-arylthio-
carbamates and found no evidence for crossover in the
rearrangement products.6a However, the results were not
strictly definitive as the relative rates of the substrate pairs
that were tested differed by ca. 1 order of magnitude. More-
over, the Miyazaki work involved conventional heating,
under which the NKR was observed to proceed exclusively
via first-order kinetics, as we have found with 1a, vide supra.
We thus designed an isotopic labeling-based probe for cross-
over, which will not suffer any significant difference in the
relative rates of coreacting substrates, the primary kinetic
isotope effects for C-16/18O, C-32/34S cleavage typically
being less than 2% in magnitude. The aryl ring in 1a was
aReagents and conditions: (i) 4.70 M (neat), 180 °C, N2 or air, 30 min,
quant; N.B. the initial sample of [2H2]-1a and [18O]-1a contains
unlabeled 1a which is converted to 2a via the non-crossover pathway
(not shown); (ii) 1.50 M DMA solution microwave heating to 180 °C,
11 min, 61% conversion to 2a; (iii) 4.70 M (neat), 180 °C, dry air,
120 min, ca. 10% equilibration.
of [2H2]-2a and [18O]-2a for the purposes of reference spectra
for the mass spectrometric (MS) analysis. Unexpectedly,
when [2H2]-2a and [18O]-2a were mixed and heated to 180 °C
under (dry) air, exchange occurred to generate ca. 10% [18O,
2H2]-2a and 2a over a 2 h period, Scheme 3, as detected by MS
(vide infra). The exchange process was found to be rather
capricious and did not proceed at all under nitrogen or under
air in the presence of 1a.20
2
labeled with H, via the phenol, and the O-aryl thiocarba-
Refluxing [18O]-2a in water, or heating it to 180 °C with
water in a sealed tube, resulted in no loss of 18O, and thus, the
exchange process (Scheme 3) does not proceed via traces of
water acting catalytically to transfer the carbonyl oxygen
between molecules of [2H2]-2a and [18O]-2a. The requirement
for air suggests a radical chain mechanism for transfer, and
1a is presumably a powerful inhibitor for exchange because
mate, [2H2]-1a, prepared via the standard procedure, Scheme 2.
The thiocarbamoyl unit in 1a was labeled with 18O, via
generation of the 18O-phenol by aromatic nucleophilic sub-
stitution and, again preparation of the O-aryl thiocarba-
mate, [18O]-1a, via the standard procedure, Scheme 2.
The labeled substrates [2H2]-1a and [18O]-1a were
rearranged independently (NKR at 180 °C) to obtain samples
(19) An identical analysis (ref 17) using the rate constants reported in
ref 13 for 140 °C predicts (rnorm) = 1.00, 1.12, and 1.39; under similar
conditions (see ref 23), we found a 75% and 126% average increase in
fractional conversion of 1a after 600, 1200, and 1800 s at 140 °C for
0.8 M/1.5 M initial concentrations of 1a.
(20) The inhibitory effect of 1a on exchange in 2a was powerful: on NKR
of labeled 1a under a dry air atmosphere at 180 °C, no exchange in the
product (2a) was detected until >99% conversion of 1a, after which
exchange-equilibration (as assayed by MS) increased approximately linearly
with time.
6350 J. Org. Chem. Vol. 75, No. 19, 2010

Burns et al.
JOCFeatured Article
SCHEME 4. Possible Radical Chain Mechanism for Exchange
between Isotopically Labeled S-Aryl Thiocarbamates (2) under
Thermal Aerobic Conditions, Together with Inhibition by 1
the thiocarbonyl moiety functions as an efficient radical
trap.21
In an attempt to distinguish whether the exchange process
proceeds via scission of the ArS-C(O) or the Ar-SC(O)
bonds, processes a versus b, Scheme 4, we prepared [34S,
2H6]-2a (Scheme 2), generating the requisite 34S thiophenol
by reaction of p-nitrochlorobenzene with Na234S (52 atom
%; prepared in situ from 34S8, 90 atom %), and the thio-
carbamate [34S,2H6]-2a, via the thiochloroformate. Unfortu-
nately, on mixing with [2H2]-2a and [18O]-2a, the sample of
[34S,2H6]-2a (despite repeated purification) inhibited the
exchange process, presumably because of the presence of
trace impurit(ies) acting as radical trap.
Probing the Molecularity of the NKR via Crossover Experi-
ments. As expected, MS analysis of the product mixture
obtained after conducting the NKR of a 50/50 sample of
[2H2]-1a and [18O]-1a, under the standard thermal conditions
(4.70 M, 180 °C, N2, Scheme 2), confirmed that there was no
detectable crossover product [18O, 2H2]-2a (unlabeled 2a was
present due to the incomplete isotopic substitution in the
substrates), an outcome fully consistent with the reaction
kinetics, vide supra, and with the experiments conducted by
Miyazaki.6a
FIGURE 4. Integrated peak areas in the HR-ESI-ICRMS subspec-
trum of the 231 m/z region (3.6% of the total [2aH]þ ion intensity) of
2a isolated after NKR of a 50/50 mixture of [2H2]-1a and [18O]-1a
(65 atom %) in DMA under microwave heating13 to 180 °C for
10 min. The relative ion intensities for the 34S natural-abundance
isotopologues of the 18O- and 2H2-labeled species ([34S,18O]-2a
[
culated to be 1.00/2.08 by kinetic simulation using the reported
34S,2H2]-2a) versus the crossover product ([18O,2H2]-2a) were cal-
values for k1 and k2;13 see the Supporting Information.
to the isobaric [MH]þ ions arising from [18O,34S]-2a (m/z =
231.0491) and [2H2,34S]-2a (m/z = 231.0574).
Using HR-ESI-ICRMS to resolve and integrate the iso-
baric m/z = 231 ions (3.6% of the total [MH]þ in the
spectrum) in the purified labeled product 2a, Figure 4, we
established that there is no detectable crossover; the net peak
intensity in the range m/z = 231.06 to 231.07, relative to the
sum of the ions arising from [18O,34S]-2a and [2H2,34S]-2a,
corresponds to e1.5% of the intensity that is predicted for the
crossover product [18O,2H2]-2a arising via k2 at an initial
reactant concentration of 1.5 M.
We thus conclude that under both the standard thermal
and the microwave heating conditions, there is no bimolec-
ular pathway for NKR of 1a involving 4a, or if there is, its
contribution to the net process is vanishingly small.
Origin of “Concentration-Enhanced” NKR under Micro-
wave Conditions. It is hard to conceive of a bimolecular
pathway for NKR of 1a that does not cause crossover, and
any direct intermolecular rate-enhancement arising from
change in the reaction medium such as the dielectric or
π-complexation effects should also be detected in the
standard thermal reaction. We thus considered that there
must be a concentration-dependent rate enhancement of
the unimolecular NKR under the microwave conditions,
which may then have been misinterpreted as mixed first
order/second order due to the obtention of sparse data
sets.13
The experimental setup for the microwave-mediated NKR,
as reported by Gilday et al.,13 requires >1.5 mL volumes of
reaction solution, thus placing some economic constraints on
the molarity of the DMA solution of [2H2]/[18O] isotopically
labeled reactant 1a that can be employed. Moreover, given the
capricious exchange that had been observed between iso-
topically labeled samples of 2a, Scheme 3, in the absence of 1a,
we wanted to ensure incomplete conversion of 1a to 2a. Using
2 mL of a 1.5 M solution of a 50/50 mixture of [2H2]-1a and
[18O]-1a, we conducted microwave-mediated NKR at 180 °C
under the published conditions for 675 s. HPLC analysis
indicated 61% conversion. According to the mixed first-
order/second-order rate analysis,13 37%oftheproductshould
have arisen from the bimolecular pathway. Standard resolu-
tion HPLC-MSconfirmedthattherehadbeen no scrambling
in the remaining 39% substrate (1a). We then conducted a
kinetic simulation of the crossover process, again using the
published values of k1 and k2, including natural abundance
2
13C and 34S, along with the H and 18O incorporations, to
predict the total array of isotopomers and isotopologues that
would arise according to the mixed first-order/second-order
rate analysis13 (see the Supporting Information) at 1.5 M
initial concentration and at 61% conversion. From this, we
calculated the predicted ratio of the [MH]þ ion that should
arise from crossover product [18O,2H2]-2a ([m/z = 231.0658)
The procedure reported for the microwave NKR13
involves conducting the reaction in a thick-walled Pyrex
microwave vessel containing a magnetic stirring bead and
sealed with a crimp-top cap. This is then irradiated, with
stirring, in a CEM Discover monomode 300 W microwave
(21) See, for example: Scaiano, J. C.; Ingold, K. U. J. Am. Chem. Soc.
1976, 98, 4727.
J. Org. Chem. Vol. 75, No. 19, 2010 6351

Burns et al.
JOCFeatured Article
FIGURE 5. Temperature profile for NKR (1af2a), as 0.5 M (blue) 0.8 M (orange), and 1.5 M (red) solutions in DMA, under conditions set up
for reaction at 140 °C for 20 min.13,23 Temperature is that recorded by an external IR pyrometer.22,25 A sample with 0 M 1a (i.e., pure DMA)
gave a similar amplitude of oscillation to the 0.5 M sample.
reactor equipped with an IR pyrometer22 and a noninvasive
pressure transducer. Using the standard temperature profile
programming on the instrument,23 the microwave reactor is
set to rapidly attain and hold the desired reaction tempera-
ture,24 for a set period, followed by a rapid-cooling phase
using an ambient temperature forced-air stream around the
outer surface of the reaction vessel. The use of a “balancing
loop”25 to modulate the microwave irradiation power in
response to reaction temperature inherently results in
some oscillation, commonly termed “hunting”,25 about the
temperature set-point. It is important to note that the IR
pyrometer measures the temperature of the outer surface of
the Pyrex microwave vessel, and the use of a thick-walled
Pyrex microwave reaction vessel thus results in a significant
lag-time between a change in temperature of the reacting
contents and the external detection of this temperature. It is
well-known that balancing loops with delayed negative feed-
back signals are prone to a greater amplitude of hunting25
(this being the range between the maxima and minima in
temperature about the set-point), and we thus studied the
power-temperature-time files stored by the CEM Discover
instrument after each run. It soon became evident that the
amplitude of the hunting about the temperature set-point
became larger as the concentration of 1a was increased. This
most likely arises from a change in the dielectric of the
medium, and thus greater loss factor (tan δ)26 and more
efficient conversion of the 2.45 GHz microwave energy into
heat for 1a. Thus, as the medium becomes richer in 1a there
will be a greater temporal discrepancy in negative feedback
and a greater amplitude of oscillatory overheating.
The effect was even more pronounced when the reactions
were run at “140 °C”23 and under these conditions we found
a 75% and 126% average increase in conversion after 600,
1200, and 1800 s, for 0.8 and 1.5 M initial concentrations of
1a, as compared to 0.5 M, Figure 3. For the 1.5 M sample, the
temperature profile, Figure 5, shows that the outer-surface of
the reaction vessel, which being transparent to microwaves is
convectively heated by its contents, reaches well over 170 °C
for at least the first 30 s of reaction, during which period k1
will increase by over an order of magnitude, whereas for the
0.5 M sample, the maximum overheating was just 6 °C.
Moreover, the dependence of the amplitude of oscillatory
overheating on the concentration of 1a/2a is evident for the
entire period of the subsequent 20 min of reaction, during
which the microwave power, while limited in the tempera-
ture program to a maximum value of 30 W, does not exceed
11 W (see the Supporting Information for a full analysis of
temporal power/temperature profiles).
(22) The monitoring of the reaction temperature in microwave reactors
via the sole application of an infrared pyrometer reading from the external
surface of the reactor vessel has been cautioned against, owing to concerns
regarding their sensitivity and the accuracy with which they reflect the
internal reaction temperature; see: (a) Hosseini, M.; Stiasni, N.; Barbieri,
V.; Kappe, C. O. J. Org. Chem. 2007, 72, 1417. (b) Herrero, M. A.; Kremsner,
J. M.; Kappe, C. O. J. Org. Chem. 2008, 73, 36. (c) Obermayer, D.; Kappe,
C. O. Org. Biomol. Chem. 2010, 8, 114. See also ref 25b.
(23) The specific program employed has two stages, with targets for the
temperature, and set limits for the microwave device input power and
reaction vessel pressure: Stage 1: 160 °C, power max = 180 W, ramp time
0 s, hold time 30 s, pressure max = 250 psi. Stage 2: 160 °C, power
max = 30 W, ramp time 0 s, hold time 600-1800 s, pressure max =
250 psi. The magnitude of the set limit(s) for the microwave device input
power will obviously impact on the amplitude of temperature oscillation
that is attainable.
(24) Temperature control in equipment such as this is usually achieved
by a proportional-integral derivative (PID) controller, a control loop feed-
back (“balancing loop”) mechanism that calculates an “error” value as the
difference between a measured process variable (in this case temperature)
and a desired set point (the desired reaction temperature). The controller
attempts to minimize the error by adjusting the process control inputs (in this
case microwave power); if this power adjustment is not rapid or accurately
related or timed to the desired reaction temperature, overheating will occur;
see ref 25.
(25) (a) See: Control Systems, Robotics, and Automation. In Encyclo-
pedia of Life Support Systems (EOLSS);Unbehauen, H., Ed.; Developed
under the Auspices of the UNESCO, Eolss Publishers: Oxford, UK, 2004;
€
[http://www.eolss.net]. See also: (b) Nuchter, M.; Ondruschka, B.; Weiβ, D.;
Beckert, R.; Bonrath, W.; Gum, A. Chem. Eng. Technol. 2005, 28, 871.
(26) Gabriel, C.; Gabriel, S.; Grant, E. H.; Halstead, B. S. J.; Mingos,
D. M. P. Chem. Soc. Rev. 1998, 27, 213.
6352 J. Org. Chem. Vol. 75, No. 19, 2010

Burns et al.
JOCFeatured Article
Conclusions
(0.11-4.70 M) in N,N-dimethylacetamide (DMA). Equal
volumes (2.0 or 4.0 mL) were heated under nitrogen, with
stirring, in thick-walled boiling tubes heated by partial immer-
sion in a silicone oil bath, in sealed glass capillary tubes placed
in the center of a precalibrated GC oven, or in glass sealed
pressure tubes in a CEM Discover monomode 300 W micro-
wave reactor with IR temperature monitoring and non-
invasive pressure transducer. Conversions were determined
by HPLC or LCMS analysis of acetonitrile-diluted samples
and are corrected for relative response factors (RRFs). The
S-arylthiocarbamate 2a was prepared and purified to provide
reference markers for the HPLC or LCMS method and RRF
values.8,12e
NKR Crossover Study Performed under Convective Heating
Conditions. [18O0.88]-1a (0.05 g, 0.22 mmol) and [2H0.97; 1H0.03]-
1a (0.05 g, 0.22 mmol) were fused at 180 °C in a Schlenk tube
under nitrogen for 30 min. A sample was removed from the
reaction mixture via glass capillary, diluted with acetonitrile,
and analyzed by HPLC which indicated that >99% conversion
of 1a to 2a had occurred. This sample was then analyzed by high-
resolution ESI ICRMS (m/z of [MH]þ ion) and found to contain
2a, [1H1]-2a, [18O1]-2a, and [2H2]-2a but no detectable crossover
product [18O1,2H2]-2a (m/z = 231).
In response to the recent report by Gilday et al. that the
solution-phase Newman-Kwart rearrangement (NKR)
proceeds via unimolecular (3) and bimolecular (4) path-
ways,13 we have reinvestigated the kinetics and molecularity of
the NKR of N,N-dimethyl O-p-nitrophenylthiocarbamate 1a.
Under standard thermal conditions, and as expected based on
the extensive prior work of Newman,1a Relles,5 Miyazaki,6
Woodward,7a and Donahue,7b the kinetics of rearrangement
at 160 °C are exclusively first order at concentrations ranging
from 0.11 to 4.70 M. There is also no crossover between
aromatic and thiocarbamoyl moieties during the reaction,
although under aerobic thermal conditions exchange can
occur in the product (2a, Schemes 3 and 4). Under microwave
heating conditions, while there is no evidence for bimolecu-
larity (a crossover experiment with a 1.5 M solution of a 50/50
mixture of [2H2]-1a and [18O]-1a was negative: no [18O,2H2]-2a
could be detected above a baseline threshold of 1.5%), there is a
disproportionate rise in conversion with increasing concentra-
tion of 1a. This was identified as arising from a combination of
two factors: (i) a delayed negative feedback signal of the reac-
tion temperature to the microwave power-control balancing
loop, resulting in oscillatory reaction-overheating, and (ii) a
more efficient microwave to thermal energy conversion (greater
loss-factor, tan δ)26 by 1a as compared to DMA, resulting in a
greater amplitude of oscillatory overheating as the initial
concentration of 1a is raised.
It is instructive to consider the circumstances that gave rise
to the misinterpretation by Gilday et al. that the data is
indicative of simultaneous unimolecular and bimolecular
pathways.13 The major issue arises from the setup of the
microwave experiment. For reasons of safety, the reactions
are conducted in crimp-cap sealed thick-walled Pyrex micro-
wave vessels, which are firmly held in place in the microwave
cavity. This means that it is not convenient to continuously
sample the reaction; instead a series of reactions are con-
ducted, leading to rather sparse temporal-conversion data
sets, a situation that can readily lead to overinterpretation of
data. The use of an IR pyrometer temperature probe, which
measures the external surface of a thick walled reaction
vessel, inherently results in a delayed, and fairly inaccurate,22
negative feedback signal and thus substantial hunting about
the desired temperature set-point, a situation that will be
sensitive to changes in the loss-factor, tan δ,26 as the propor-
tion of solvent/solute is varied.
NKR Crossover Study Performed in a CEM Discover Micro-
wave Reactor. [18O0.65]-1a (0.34 g, 1.50 mmol) and [2H0.97
;
1H0.03]-1a (0.34 g, 1.50 mmol) were dissolved in DMA (2 mL)
to give a solution of 1.50 M total concentration in 1a. A thick-
walled Pyrex microwave vessel equipped with a magnetic
stirring bead was charged with the solution. The vessel was
sealed with a crimp-top cap and irradiated with stirring in a
CEM Discover monomode 300 W microwave reactor with IR
temperature monitoring and noninvasive pressure transducer
at 180 °C for 11 min (stage 1: 180 °C, 180 W, ramp time 0 s, hold
time 75 s, pressure 250 psi; stage 2: 180 °C, 30 W, ramp time 0 s,
hold time 600 s, pressure 250 psi). Once irradiation had ceased
and the sample cooled, the glass microwave reaction vessel was
opened, and a reaction sample removed via glass capillary,
diluted with acetonitrile, and analyzed by HPLC which indi-
cated that 61% conversion of 1a to 2a had occurred. The
remaining reaction mixture was then diluted with a mixture
of ethyl acetate (10 mL) and saturated brine (10 mL). The
organic fraction was separated, washed with brine (2 ꢀ 10 mL),
dried (MgSO4), and concentrated in vacuo. The crude material
was purified by column chromatography on silica gel, eluting
with a 5:1 mixture of toluene/ethyl acetate, to afford 1a
(240 mg, 35%) as an off-white solid and 2a (376 mg, 55%) as
a pale yellow solid. The purified sample of 2a was then
analyzed by high-resolution ESI ICRMS (m/z of [MH]þ ion)
and found to contain 2a (m/z = 227), [1H1]-2a (m/z = 228),
[18O1]-2a (m/z=229), and [2H2]-2a (m/z=229) but no detect-
able crossover product [18O1,2H2]-2a (m/z = 231). For full
details of the mass spectrometric data, see the Supporting
Information.
We thus conclude that microwave reactors can be con-
veniently used to rapidly generate a series of single time point
kinetic data for reactions that require relatively high tem-
peratures; however, as with any kinetic determination, accu-
rate control of temperature is crucial. In this regard, the
current method of choice is probably the elegant approach
developed by Kappe,14 whereby semiconducting silicon-
carbide reaction vessels are heated via ohmic resistance to
microwave-induced electron flow, in turn heating the vessel
contents via a standard convective mechanism, with a fiber-
optic based temperature-power balancing loop.
Acknowledgment. We thank AstraZeneca Global PR&D
for generous funding and Dr. Ross T. Burn (AstraZeneca)
for extensive assistance with the MS analysis. G.C.L.-J. is a
Royal Society Wolfson Research Merit Award holder.
Supporting Information Available: Full experimental details
for kinetic analyses, the preparation of labeled forms of 1a,
HPLC and MS analysis/simulation of isotopomers and iso-
topologues in product mixture 2a, and the analysis of power-
temperature profiles for microwave heating of solutions of 1a in
DMA. This material is available free of charge via the Internet at
http://pubs.acs.org.
Experimental Section
General Methods. For each series of experiments, homo-
geneous solutions of 1a were prepared at the required concentrations
J. Org. Chem. Vol. 75, No. 19, 2010 6353