Convenient syntheses of halo-dibenz[b,f]azepines and carbamazepine analogues via N-arylindoles

Article abstract of DOI:10.1039/c3ob41252k

The dibenz[b,f]azepine heterocyclic system and related molecules with a single 10,11-bond are important templates for well-prescribed drug molecules, notably carbamazepine (anticonvulsant), clomipramine and imipramine (antidepressants). We synthesised a range of halogenated carbamazepine analogues, in connection with metabolic and immunological studies, as probes for structure-metabolism and hypersensitive effects and have published on their metabolic behaviour. While a number of synthetic routes to such analogues are possible, we naturally sought short and efficient methods for our target compounds. In the following report we present an effective two-step synthesis of a range of dibenz[b,f]azepines from appropriate indoles via N-arylation, then acid-catalysed rearrangement, with a critical analysis of other approaches. We showed earlier that this route was effective for fluoro analogues and here present a broader review of its scope. The 5-(carboxamido) side chain of carbamazepine may be added in various ways, affording overall a convenient access to drug molecules.

Full text of DOI:10.1039/c3ob41252k

Organic &
Biomolecular Chemistry
View Article Online
View Journal | View Issue
PAPER
Convenient syntheses of halo-dibenz[b,f]azepines and
carbamazepine analogues via N-arylindoles†‡
Cite this: Org. Biomol. Chem., 2013, 11,
8426
Emma-Claire Elliott,§a James L. Maggs,b B. Kevin Park,b Paul M. O’Neilla and
Andrew V. Stachulski*a
The dibenz[b,f]azepine heterocyclic system and related molecules with a single 10,11-bond are important
templates for well-prescribed drug molecules, notably carbamazepine (anticonvulsant), clomipramine
and imipramine (antidepressants). We synthesised a range of halogenated carbamazepine analogues, in
connection with metabolic and immunological studies, as probes for structure-metabolism and hypersen-
sitive effects and have published on their metabolic behaviour. While a number of synthetic routes to
such analogues are possible, we naturally sought short and efficient methods for our target compounds.
In the following report we present an effective two-step synthesis of a range of dibenz[b,f]azepines from
appropriate indoles via N-arylation, then acid-catalysed rearrangement, with a critical analysis of other
approaches. We showed earlier that this route was effective for fluoro analogues and here present a
broader review of its scope. The 5-(carboxamido) side chain of carbamazepine may be added in various
ways, affording overall a convenient access to drug molecules.
Received 18th June 2013,
Accepted 22nd October 2013
DOI: 10.1039/c3ob41252k
www.rsc.org/obc
appropriate halo-dibenzo[b, f]azepine.8,9 Summarising our
results, both the 2-chloro and the 2,8-difluoro analogue of 3
Introduction
The tricyclic heterocycles dibenzo[b,f]azepine 1 and its 10,11- effectively block ring hydroxylation in microsomal prep-
dihydro version 21 are major structural elements for a number arations or hepatocytes; very recently we have found that the
of important CNS-active drugs, such as carbamazepine (CBZ) halogenated derivatives are still capable of stimulating T cells,
3,2 imipramine 4,3 clomipramine 54 and opipramol 6.5
though to a lesser extent than 3 itself.9a,b
CBZ has a complex metabolic profile;6 especially, cytochrome
P450-mediated oxygenation of the benzene rings of 3 is known
to generate protein-reactive electrophilic species7a,b and more
recently the 10,11-epoxide was also shown to form protein and
The syntheses proved to be a demanding challenge. In
Scheme 1, four possible routes to the dibenzo[b, f]azepine
(DBA) intermediates are summarized. We initially employed
route (1), via an N-aryl isatin, to synthesise (2-fluoro)dibenzo-
[b, f]azepine,8 but despite its scientific interest and generally
good yields this route is long. The 9-acridinemethanol to
dibenz[b, f]azepine rearrangement step in this sequence has
been used by others.10 Route (2), employing controlled halo-
genation of 10,11-dihydrodibenz[b, f]azepine 2 (‘iminodiben-
zyl’, IDB) is also long, requiring N-protection and deprotection
in addition to radical halogenation and elimination steps: we
employed this route to prepare mono- and di-chloro and
bromo analogues of 3.9 For both bromo and chloro analogues,
separation of mono- and di-halogenated species from initial
glutathione adducts.7c In connection with
a programme
seeking to identify analogues less prone to metabolic acti-
vation, we synthesized a set of halogenated derivatives via the
aDepartment of Chemistry, University of Liverpool, Liverpool L69 7ZD, UK.
E-mail: stachuls@liv.ac.uk; Fax: +44 (0)151 794 3588; Tel: +44 (0)151 794 3482
bMRC Centre for Drug Safety Science, Institute of Translational Medicine, University
of Liverpool, Liverpool L69 3GE, UK
†This paper is dedicated with affection to Drs. Tom Gilchrist and Richard Storr,
two great colleagues and leading exponents of heterocyclic chemistry.
‡Electronic supplementary information (ESI)
available. See DOI:
10.1039/c3ob41252k
§Current address: Unilever Research and Development Laboratories, Port Sun-
light, Wirral, UK.
8426 | Org. Biomol. Chem., 2013, 11, 8426–8434
This journal is © The Royal Society of Chemistry 2013

View Article Online
Organic & Biomolecular Chemistry
Paper
Scheme 1 Synthetic routes to dibenz[b,f] azepines: (1) via isatins; (2) from IDB; (3) via styrenes; (4) via stilbenes.
reaction of 2 is necessary and overhalogenation must be
avoided.11
Tsvelikhovsky and Buchwald recently devised a synthesis
from an o-bromostyrene and an o-chloroaniline, route (3), pub-
lished while our studies were in progress.12a,b The key to this
Scheme 2 Synthesis of dibenz[b,f]azepines via N-aryl indoles.
was ligand 7, which diverted the bracketed intermediate
toward the desired [6,7,6] ring system and away from other
possible ring closures that would generate acridines or
indoles. Although a most elegant route, this does require in
general the availability of trisubstituted aromatics for both
fragments, which could be difficult of access, depending on X1
and X2. Furthermore, the authors later published a correc-
tion12b in which they stated that the yield of dibenzazepine
from 2-chloroaniline and 2-bromostyrene, whether by the one
or two-step procedure, was significantly lower than reported
earlier (70% rather than 99%), and that selectivity was incom-
plete. This leaves a clear concern that more highly substituted
examples may also give lower yields and selectivity. Similarly,
route (4) of Liang et al. requires the synthesis of mono/
disubstituted o/o′-dibromo-Z-stilbenes and applies only
to N-alkyl or N-aryl amines, though the ring closure is
impressively efficient.13
efficient syntheses of N-aryl indoles are now available (v.i.),
and the rearrangement step was discovered and evaluated for a
number of analogues by Tokmakov and Grandberg.15 Although
there are restrictions on the substitution patterns compatible
with these conditions, the attractions of a two-step synthesis
are obvious. We now report in detail on our findings, includ-
ing especially a discussion of the scope of the rearrangement
step and substituent effects, with finally a survey of methods
for addition of the carboxamide group present in 3.
Regarding the general availability of mono- and disubsti-
tuted IDBs [we applied route (2) for Br/Cl analogues only],
Jorgensen et al. prepared a number of examples, including
fluoro and methoxy analogues, in seven linear steps via appro-
priate o-nitrostilbenes; for our purpose, the 10,11-double bond
would still need to be reintroduced.14 This and other older
routes are summarised by Kricka and Ledwith.1
Results and discussion
N-Arylindole synthesis
In our studies, we employed both routes (1) and (2)8,9 but
also extensively studied a much shorter and attractive route,
shown in Scheme 2, namely the acid-catalysed (polyphospho-
ric acid, PPA) rearrangement of N-aryl indoles. A number of
We studied various conditions for the N-arylation reaction,
first optimising conditions for N-phenyl indole itself; initially
employing the palladium-catalysed route of Watanabe et al.16
and Buchwald et al.17
This journal is © The Royal Society of Chemistry 2013
Org. Biomol. Chem., 2013, 11, 8426–8434 | 8427

View Article Online
Paper
Organic & Biomolecular Chemistry
Bulky, electron-rich ligands, e.g. PBut3, are favoured for this satisfactory yields of 24 (59%; Table 1, footnote a, method B)
chemistry16 as they allow reaction at lower temperatures and using 16 or 45% using PhBr. However, we had to work at a
greatly reduce dehalogenation as a side reaction. Buchwald much higher temperature than that given, and product iso-
et al. preferred the ligand 7 noted above, and related o,o′- lation proved difficult. Instead the method of Ma and Cai, who
disubstituted biphenyls, with aryl bromides, chlorides or also studied functionalized examples, proved to be robust and
triflates as the donors.17
high-yielding; use of L-proline (or, less effectively, sarcosine) as
a ligand for Cu is key to this procedure.19 Working in DMSO at
80–100 °C, but mostly at 80–90 °C, we obtained yields from
62–92% for a variety of functionalized examples, and 97% for
the parent 23. Only for the free phenol 20 did we fail to obtain
any product. Regarding halogenated examples, we had earlier
used this method successfully for a range of fluoro analogues
(Table 1, products 24–30).8 Methoxy arenes20,21 are also excel-
lent substrates, and it is particularly noteworthy that Br and Cl
substitution20,22 is fully compatible, yielding products 35–40;
no self-condensation or polymerisation was observed, even for
the previously unreported dibromo compound 40. Finally, we
note that this method was effective on a multigram scale: this
was essential for optimization of the following cyclisation step.
For simplicity, we employed PBut3 and used the more reactive
aryl iodides (Table 1; indoles 8–15, aryl iodides 16–22, pro-
ducts 23–40). From indole 8 and iodobenzene 16, we obtained
an excellent yield of 94% on a 1 g scale using the conditions
described (footnote a, method A1), and the yield was practi-
cally as good using bromobenzene. Reaction at 85 °C was poss-
ible but in our hands gave lower yields than those described.
Similarly, using 4-fluoroiodobenzene 17, good to very good
yields were obtained from 8; however, the reaction did not
scale-up well as can be seen from footnotes d, g (5 g scale).
The highly pyrophoric nature of PBut3 is a significant concern
for larger scale work; moreover, overreaction was observed
when excess of 16 was employed, possibly from the formation
of N,C(3)-bisaryl products.
We therefore switched to Ullmann conditions employing
copper catalysts, which are also well described for this trans-
Cyclisation reaction
formation. Older versions used Cu/CuI mixtures, frequently We first prepared a sample of 1 from 23 using the conditions
with neat reagents, at temperatures of up to 200 °C.15 By using of Tokmakov and Grandberg,15 then applied this method for
a combination of CuI/Cs2CO3 in DMF, or an appropriate ligand the halo analogues. A number of experimental details were sig-
for copper, lower temperatures may be used. We initially used nificant. Commercial PPA is not anhydrous, but we deoxy-
the method of Chandrasekhar et al.,18 with a recyclable PEG genated the reagent by passage of dry argon before use and
medium and ethylene diamine as ligand, and obtained performed the reactions under argon as the products are
8428 | Org. Biomol. Chem., 2013, 11, 8426–8434
This journal is © The Royal Society of Chemistry 2013

View Article Online
Organic & Biomolecular Chemistry
Paper
Table 1 N-arylation of indoles using CuI and L-proline; yields quoted apply to 1–5 g scale reactions
Reference if
known compound
Indole
Aryl iodide
Temp (°C)/time (h)
90/24
Product no.
Yield (%)
97
Yield (%) (other methods)a
96 (A1),b,c 94 (A2)
8
16
19b
23
Fluorinated examples
8
17
18
16
16
16
17
18
100/24
100/28
100/24
100/24
100/24
100/24
100/24
8
8
8
8
8
8
8
24
25
26
27
28
29
30
81
78
70
76
83
75
85
88 (A1),d 59 (B)e, f
73 (A1)g
8
9
10
11
10
11
Chloro, bromo and alkoxy examples
8
19
16
16
20
21
16
21
22
16
22
90/36
90/36
90/36
90/36
80/24
80/24
80/24
90/36
90/36
90/36
19a
20
21
—
15
—
—
22
20
—
31
32
33
34
35
36
37
38
39
40
68
71
62
12
13
8
h
—
8
87
92
85
79
83
51
14
14
8
15
15
a Other conditions: A1, Pd(OAc)2 (1 mol%), PBut3, o-xylene, 110 °C; A2, as A1, using the corresponding aryl bromide; B. CuI (5 mol%), (CH2NH2)2,
K2CO3, PEG-400, 160 °C. b Yield on a 1 g scale; this dropped to 22% on a 5 g scale. c A 53% yield was obtained at 85 °C. d A 22% yield was
obtained on a 5 g scale. e No useful product isolated at 80 °C. f A 45% yield was obtained using 4-bromofluorobenzene. g A 12% yield was
obtained on a 5 g scale. h No product isolated.
oxygen-sensitive. In earlier experiments23 we had conducted indole 25 gave rise to both 1- and 3-fluoro-DBAs 42, 43; simi-
the rearrangement at 150 °C, but this was found to give signifi- larly, difluoroindole 30 gave two DBA products 45 and 46.8
cant amounts of 9-methylacridine by-products. This side-reac- Moreover, the yields (of combined products where appropriate)
tion was not commented on by Tokmakov and Grandberg,15 were generally good (apart from the 4-fluoro analogue 26),
but they did generally work at significantly lower temperatures, reflecting, we believe, the minimal steric requirement of fluo-
viz. 75–120 °C, and generally at 90–100 °C. In the present rine and its ability to stabilise adjacent carbonium ions by
studies we worked at 100 °C as closely as possible: this gene- 2p-lone pair donation. By contrast, strongly inductive electron-
rally minimized the formation of 9-methylacridines, although withdrawing groups NO2 and CF3 on the N-aryl group gave no
traces (<5%) were still seen by NMR, especially the N-methyl desired product at all.15 Less surprisingly perhaps, methoxy
group, e.g. δH (CDCl3) 2.98 for 2,7-difluoro-9-methylacridine. substituents also lead to reasonable yields, as shown for NAIs
Scheme 3 presents feasible mechanisms for the cyclisation 31–33, Table 2; here the 5-methoxy precursor 31, not previously
step (a) and the by-product formation (b), and Table 2 sum- studied, afforded a better yield of 47 than the 4′-methoxy precursor
marises our results for products 1 and 41–51.
32 (cf. below). The 3-methoxy product 48 was also prepared, in
10% yield, by Lucini et al.24 These analogues are potentially
useful precursors of the 2- and 3-hydroxy metabolites of 3.9a
As stated above, we were particularly interested in halo ana-
logues, and in addition to the fluoro compounds8 we con-
sidered the possibility of obtaining bromo- and chloro-DBAs in
this way, with interesting results. In the earlier work,15
2-chloro-DBA 49 was obtained from 35 in 25% yield; we isolated
a somewhat better yield of 41%. Here, however, because of the
pseudo-symmetry of the product, 36 should be an equally
good precursor and indeed we obtained a very satisfactory
yield of 49 (61%) from 36. Thus the undoubted electron-with-
drawing nature of Cl is not a barrier to successful reaction,
In the earlier work, only one successful example with a m-sub-
stituted N-aryl was recorded, viz. N-(m-tolyl)indole, and this
gave only (3-methyl)DBA.15 We found that N-(3′-fluorophenyl)-
This journal is © The Royal Society of Chemistry 2013
Org. Biomol. Chem., 2013, 11, 8426–8434 | 8429

View Article Online
Paper
Organic & Biomolecular Chemistry
Scheme 3 Mechanistic schemes. (a) N-Aryl indole to DBA rearrangement, after ref. 15. (b) Suggested mechanism for 9-methylacridine by-product formation,
based on ref. 1. (c) Proposed dehalogenation mechanism.
especially if the substituent resides in the indole ring of the
substrate. The main limitation is actually dehalogenation: in
fact only trace amounts of 1 were observed from 36, but the
5,4′-dichloro substrate 37 afforded a separable mixture of 2,8-
dichloro-DBA 50 (32%) and 2-chloro-DBA 49 (15%).
Table 2 Acid-catalysed cyclisation of N-aryl indoles to dibenz[b,f]azepinesa
We suggest that protonation of the DBA products in the
strongly acidic medium is the likely cause of loss of Cl. Where
possible, protonation will occur on the non-chlorinated ring;
when both rings are chlorinated [as Scheme 3c], protonation
allows subsequent dechlorination as a side reaction. Neverthe-
less, the yields of both 49 and 50 are now at least competitive
with those obtained via chlorination of 2,9 and only two steps
as opposed to five are needed.
However, bromine was not compatible with the reaction
conditions. From the reaction of 5-bromo-(N-phenyl)indole 39,
only a 5% yield of 51 was isolable and milder conditions
(65 °C) were necessary. Nevertheless, even this modest yield
was useful for a reference sample, as it proved very difficult to
obtain pure 2-bromo-DBA 51 via the halogenation route from
2, as noted earlier.9 This is probably owing to radical-induced
rearrangement during the introduction of the 10,11-double
bond.
Reference if known
compound
N-Aryl indole
Product(s)b
Yield (%)
67
23
1
15
Fluorinated examples
24
25
26
27
28
29
30
41
42, 43
42
41
43
44
45, 46
40
18, 22
24
47
48
66
16, 35
8
8
8
8
8
8
8
Chloro, bromo and alkoxy examples
31
32
33
35
36
37
39
47
47
48
49
49
50
51
25
37
22
41
60
32e
5f
15c
15
24d
9
9
9
9
Carboxamidation
a Reaction carried out at 100 °C and for 36–72 h, until judged complete
by TLC, unless stated. b In cases where X2 is a meta-substituent, two
regioisomers are produced: see text. c An 8% yield was reported.
d A 10% yield was reported. e Also 49 (15%) was produced. f Reaction
carried out at 65 °C.
A number of methods for adding the carboxamide unit to
dibenz[b, f]azepines, generating carbamazepine analogues,
have been investigated. Many of these have appeared in the
8430 | Org. Biomol. Chem., 2013, 11, 8426–8434
This journal is © The Royal Society of Chemistry 2013

View Article Online
Organic & Biomolecular Chemistry
Paper
Table 3 Carboxamidation of dibenzazepines using phosgene equivalents
Dibenzazepine
Solvent
PhMe
Temperature (°C)
Time (h) prior to NH3 addition
8
Product
Yield (%)
79
Diphosgenea
1
10–15
3
Triphosgene
1
1
1
THF
PhMe
PhMe
10–15
6
6
6
3
3
3
27
92
—
10–15
10–110b
Triphosgene with halogenated DBAs
50
PhMe
PhMe
60
10–15
12
6
55
56
12
24
54
a Using 0.5 eq. 52 or 0.33 eq. 53 with an equivalent amount of Et3N unless stated. b No Et3N added.
patent literature, particularly after carbamazepine 3 itself came
off patent. We showed earlier9 that by varying the traditional
procedure using an alkali metal isocyanate,25 in particular
employing TFA as a stronger acid catalyst,26 a range of haloge-
nated analogues of 3 was readily accessible. Nevertheless, we
screened a number of other reagents, and this led to some
interesting chemistry. In particular, we found good alternative
methods for the synthesis of dibromo and dichloro analogues,
involving the generation of an activated N–CvO·X inter-
mediate and its reaction, possibly following isolation, with an
appropriate nucleophile.
As 53 was the superior reagent, it was evaluated for the halo
analogues of 1 also. However, only the dichloro 50 and
dibromo 54 compounds gave >10% yields of the corresponding
CBZ analogues 55 and 56, as shown. Here again, solubility is
key: dibromo-CBZ 56 is the least soluble of the series and pre-
cipitates in modest yield. It may be also that the intermediate
ring-halogenated N-carbonyl chlorides are more susceptible to
aqueous hydrolysis than the unsubstituted variant.
Substituted organic isocyanates were more promising,
but here it was important to be able to isolate the N-acyl (or
sulfonyl) urea intermediate. Thus on reaction of 2,8-dibromo-
DBA 54 with N-(chlorosulfonyl)isocyanate30 in CH2Cl2, the pre-
sumed N-chlorosulfonyl intermediate was filtered off and
Considering first phosgene and equivalents, phosgene itself is reacted directly with dilute aq. NaHCO3. We obtained a low
used industrially,27 but in view of its high toxicity we explored yield (ca. 10%) of the CBZ analogue, but preparative TLC was
both (trichloromethoxy)carbonyl chloride28 (diphosgene; equi- necessary to purify the product.23
valent to two molecules of phosgene) 52 and bis-(trichloro-
The best alternative to the acid-catalysed alkali metal iso-
methyl)carbonate29 (triphosgene; equivalent to three cyanate reaction was the use of (trichloroacetyl)isocyanate
molecules of phosgene) 53. In either case, we studied first DBA (Table 4).31 For the symmetrical substrates (viz. 1, 50 and 54)
1 itself, reacting it with either 52 or 53 in an inert solvent, then the initial reaction was fast and the N-acyl ureas were obtained
adding aqueous ammonia (Table 3). Addition of triethylamine in high yield by simple filtration. The difficulty, however, was
was essential to liberate phosgene from 52/53 at a reasonable the lack of selectivity in the hydrolysis step, Table 4. Thus, in
rate: the ensuing N-COCl species was sufficiently stable to be the unsubstituted case, treatment of the N-acylurea derived
monitored by TLC (and indeed is now commercially available). from 1 with either dilute acid or base lead to mixtures of 1 and
A very good yield of 3 was obtained from both reagents, with 3. Relatively best was 5% aq. Na2CO3, leading to 3 as the main
53 slightly superior, and toluene was greatly superior to THF product. Similarly, for the dibromo and dichloro examples, the
as a solvent. The outcome reflects the solubility and stability of N-acylureas could be isolated in good yields from 50 and 54
both the presumed intermediate (viz. N-carbonyl chloride) and and afforded 50–60% of 55 and 56 on mild basic hydrolysis.
the final product; aqueous hydrolysis, regenerating 1, is a The method was not suitable for mono-halo DBAs.
major side reaction. If the final product, here 3, precipitates it
is effectively protected from further reaction.
Although the N-acylureas are moderately stable for short
periods at room temperature, they are readily hydrolysed, even
This journal is © The Royal Society of Chemistry 2013
Org. Biomol. Chem., 2013, 11, 8426–8434 | 8431

View Article Online
Paper
Organic & Biomolecular Chemistry
Table 4 Carboxamidation of dibenzazepines using N-(trichloroacetyl)isocyanatea
DBA
Temp. (°C)
Time (h)
6
Yield of N-acylurea (%)
Hydrolysis conditions
Product ratio A : B
Desired product
Yield (%)
56
1
20–0
89
0.1 M HCl
10 : 90
12 : 88
57 : 33
34 : 56
60 : 40
60 : 40
0.1 M NaOH
5% aq. Na2CO3
5% aq. K2CO3
5% aq. Na2CO3
5% aq. Na2CO3
3
50
54
20–0
20–0
24
8
75
78
55
56
45
47
a Using toluene as solvent in all cases. DBA = dibenz[b,f]azepine.
during NMR acquisition to some extent. Nevertheless, it
was possible to obtain satisfactory H NMR spectra and HRMS
Experimental
1
for the dichloro example, and this data was adjudged Organic extracts were finally washed with saturated brine and
sufficient since both final products 55 and 56 gave identical dried over anhydrous Na2SO4 prior to rotary evaporation at
spectroscopic data to material obtained via the NaNCO/TFA <30 °C. Moisture sensitive reactions were carried out in anhy-
method.9
drous organic solvents (purchased from Sigma-Aldrich) under
a N2 or Ar atmosphere. Reactions were monitored by analytical
thin-layer chromatography using Merck Kieselgel 60 F254 silica
plates, and were viewed under UV or by staining with anisalde-
hyde, vanillin, KMnO4, iodine, or bromocresol green. Prepara-
tive flash column chromatography was performed on either
VWR Prolabo silica gel or Sigma-Aldrich silica gel (particle size
40–63 Å). Melting points were recorded using a Bibby-Sterlin
Stuart SMP3 melting point apparatus and are uncorrected.
High resolution mass spectrometry for the N-aryl indoles was
performed by the EPSRC National Mass Spectrometry Service,
Swansea University. Other mass spectra were obtained in
either electrospray mode (ES) with a Micromass LCT or chemi-
cal ionization (CI) mode with a Micromass Trio 1000 using
ammonia; two final high-resolution mass spectra (compounds
47, 48) were obtained in CI mode using an Agilent QTOF 7200
instrument in this Department. Elemental analyses were
Conclusions
There are a number of possible syntheses of dibenz[b, f]aze-
pines from available starting materials. We developed very suc-
cessfully a simple two-step synthesis, starting from readily
available indole derivatives. N-Arylation of the indoles was
effected in good to excellent yield using the procedure of
Ma et al.,19a,b employing CuI with proline as ligand. The con-
ditions were fully compatible with other halogens. Rearrange-
ment of the N-aryl indoles to dibenz[b, f]azepines was effected
using strong acid catalysis (PPA, 100 °C), in yields from
22–66%. We developed the original conditions of Tokmakov
and Grandberg:15 a significant by-product was a 9-methyl acri-
dine but this could be minimized. Both fluoro and chloro sub-
stituents were compatible with the conditions, though with
the latter the site of Cl substitution (viz. on the indole ring or
in the N-aryl substituent) was an important factor. With an
N-(3-fluoro)phenyl substituent, both 1- and 3-fluoro products
resulted. The simplicity of this two-step sequence compensates
for moderate yields in the rearrangement step. Finally, attach-
ment of the carboxamide substituent in carbamazepine 1 and
its 2,8-dibromo and dichloro analogues can be effected using
1
performed by Mr Steve Apter, University of Liverpool. H and
13C NMR spectra were obtained using a Bruker Avance or a
Bruker DPX 400 instrument operating at 400 and 101 MHz,
respectively; chemical shifts are reported in ppm (δ) relative to
Me4Si. Coupling constants (J) are reported in Hz.
General method for N-arylindole synthesis
various reagents, including triphosgene and (trichloroacetyl)- A mixture of the appropriate indole (2.2 mmol), K2CO3
isocyanate, with following ammonolysis or hydrolysis (5.0 mmol), CuI (0.1 mmol) and L-proline (0.2 mmol) was
a
step. However, the combination of KNCO or NaNCO and TFA stirred in anhydrous DMSO (4 mL) and heated to 100 °C. After
is most versatile and gives access to monohalo analogues 10 min the appropriate iodobenzene (2.0 mmol) was added
also.9
dropwise over 20 min, then the mixture was stirred at this
8432 | Org. Biomol. Chem., 2013, 11, 8426–8434
This journal is © The Royal Society of Chemistry 2013

View Article Online
Organic & Biomolecular Chemistry
Paper
temperature for 24 h. The mixture was cooled and partitioned complete. The product was filtered, washed with water and
between EtOAc (3 × 50 mL) and water; the combined organic dried to afford 3, 55 or 56.
extracts were evaporated to dryness and the product isolated
by chromatography, eluting with EtOAc–hexane mixtures.
Via N-(Trichloro)acetyl urea intermediates. The DBA 1, 50 or
54 was dissolved in toluene (2 mL per mmol) and trichloroace-
Spectroscopic and analytical data, including photocopy tyl isocyanate (1.2 eq.) was added by syringe under the solvent
NMR spectra, for compounds 24–30 have been given earlier.8 level. On stirring, a yellow or orange coloration was initially
Data for compounds 23, 31–33, 35, 36 and 38–40 are placed in observed, turning colourless, then the reaction was left for
the ESI.‡ All yields are in Table 1.
24 h. Precipitation was induced if necessary by addition of
hexane, then the intermediate was filtered off, washed with
ice-cold water and dried. The N-(trichloroacetyl)intermediate
was then reacted under appropriate hydrolysis conditions
(Table 4); the crude solid was filtered off, dissolved as far as
possible in CH2Cl2 and filtered through Celite, then the filtrate
was washed with satd. aq. NaHCO3, brine, dried and evapor-
ated. Crystallization from toluene, or chromatography eluting
with EtOAc–hexane mixtures (20–50% EtOAc), then afforded
pure product 3, 55 or 56.
5-Chloro-N-(4-chloro)phenylindole 37
M.p. 66–67 °C. Found: C, 64.1; H, 3.5; N, 5.2; m/z 262.0183.
C14H9Cl2N requires C, 64.2; H, 3.5; N, 5.3%; C14H10NCl2 (MH+)
requires m/z, 262.0185; δH 6.62 (1 H, dd, J = 3.3 and 0.8 Hz,
3-H), 7.17 (1 H, dd, J = 8.8 and 2.0 Hz, 6-H), 7.30 (1 H, d, J =
3.3 Hz, 2-H), 7.37–7.42 and 7.46–7.51 (4 H, approx. dd, 2′-, 3′-,
5′- and 6′-H), 7.40 (1 H, d, J = 8.7 Hz, 7-H) and 7.64 (1 H, d, J =
1.8 Hz, 4-H); δC 104.0, 111.7, 121.0, 123.2, 125.9, 126.6, 129.4,
130.3, 130.8, 132.8, 134.6 and 138.3.
The N-(trichloroacetyl)intermediates derived from 1 and 54
were sufficiently stable on isolation to allow characterization:
N-(2,2,2-trichloroacetyl)-5H-dibenz[b, f]azepine: white powder.
General method for dibenz[b, f]azepine synthesis
Found: m/z, 402.9784.
C
17H1135Cl3N2O2Na requires m/z,
Polyphosphoric acid (1 mL per 100 mg N-arylindole) was
purged with Ar and heated to 100 °C for 30 min. The appropri-
ate N-arylindole was then added via a syringe and the reaction
was stirred at 100 °C, generally for 36–72 h (see Table 2). Once
it was judged complete by TLC, the reaction was cooled to
ambient temperature, poured cautiously onto ice-cold NaHCO3
aq. and stirred vigorously for 1 h, then extracted with CH2Cl2
(2 × 100 mL). The combined organic extracts were washed with
water and evaporated; the resulting crude product was purified
by chromatography, eluting with 10% EtOAc–hexane.
402.9784; δH [(CD3)2SO] 6.05 (2H, s, 10-H + 11-H), 6.52 (2 H, d,
J = 8.4 Hz), 6.88–6.98 (2 H, m), 7.06–7.21 (4 H, m) and 10.85
(1 H, br s, NH); δC [(CD3)2SO] 93.1, 113.6, 121.0, 131.1, 131.9,
132.2, 132.7, 148.4 and 162.9; m/z (ES +ve mode) 403 (MNa+,
100%). N-(2,2,2-Trichloroacetyl)-2,8-dibromo-5H-dibenz[b, f]
azepine:
white
powder.
Found:
m/z,
558.8021,
C17H979Br235Cl3N2O2Na requires m/z, 558.7994; δH [(CD3)2SO]
7.08(2 H, s, 10-H + 11-H), 7.35–7.60 (6 H, m) and 10.46 (I H, br
s, NH); δC [(CD3)2SO] 91.8, 119.1, 121.9, 127.1, 129.3, 129.3,
129.5, 130.4, 132.1, 133.8, 135.1, 138.2, 140.5, 149.5, 150.3,
158.4, and 162.9; m/z (ES +ve mode) 559(MNa+, 23%) and 561
(100%).
Spectroscopic data for compounds 41–46,8 49,9 509 and 519
have been given earlier; since we now report the synthesis of
49, 50 and 51 by the N-aryl indole rearrangement route, NMR
data and photocopy spectra for these compounds and for com-
pound 48 are included in the ESI.‡
We give NMR data for 55 and 56, as prepared by the
trichloroacetyl isocyanate route, together with photocopy NMR
spectra in the ESI‡ to show identity with material obtained
earlier9a using the alkali metal isocyanate route.
2-Methoxy-5H-dibenz[b, f]azepine 4715
This was prepared from either N-aryl indole 31 (25% yield) or
32 (37% yield), Table 2. Found: m/z, 224.1075. C15H14NO
requires m/z, 224.1070 (MH+); δH [(CD3)2SO] 3.63 (3 H, s,
CH3O), 6.13, 6.18 (2 H, ABq, 10-H + 11-H), 6.41 (1 H, d, J =
2.1 Hz, 6-H), 6.59 (1 H, d, J = 2.6 Hz, 9-H), 6.62 (1 H, d, J =
7.6 Hz, 4-H), 6.69 (1 H, t, J = 7.2 Hz, 8-H), 6.74–6.79 (2 H, m,
1-H + 3-H) and 6.97 (1 H, t, J = 7.2 Hz, 7-H); δC [(CD3)2SO] 55.1,
114.4, 115.3, 118.9, 120.0, 121.6, 128.9, 129.5, 130.4, 131.6,
132.7, 134.1, 142.4, 150.3, and 154.7; m/z (CI) 224 (MH+,
100%).
Acknowledgements
We are grateful to the EPSRC for funding (DTA maintenance
grant to EE).
Notes and references
1 L. J. Kricka and A. Ledwith, Chem. Rev., 1974, 74, 101–123.
2 (a) F. Albani, R. Riva and A. Baruzzi, Pharmacopsychiatry,
1995, 28, 235–244; (b) A. Sidebottom and S. Maxwell,
J. Clin. Pharm. Ther., 1995, 20, 31–35.
3 (a) M. R. Mavissakalian and J. M. Perel, Am. J. Psychiatry,
1995, 152, 673–682; (b) D. H. Barlow, J. M. Gorman,
M. K. Shear and S. W. Woods, J. Am. Med. Assoc., 2000, 283,
2529–2536.
General procedure for carboxamidation reactions
Use of phosgene equivalents. A solution or suspension of
the appropriate DBA 1, 50 or 54 (1 mmol) in toluene (2 mL) at
10 °C was treated with 52 (0.5 equivalents) or 53 (0.33 equivalents)
and triethylamine (0.5 or 0.33 eq. respectively), then stirred
until starting material had fully reacted. Aqueous NH3 was
then added and stirring continued until precipitation was
This journal is © The Royal Society of Chemistry 2013
Org. Biomol. Chem., 2013, 11, 8426–8434 | 8433

View Article Online
Paper
Organic & Biomolecular Chemistry
4 (a) P. Thoren, M. Asberg, B. Cronholm, L. Joenestedt and
L. Traskman, Arch. Gen. Psychiatry, 1980, 37, 1281–1285;
T. Yamamoto and Y. Koie, Tetrahedron Lett., 1998, 39, 617–
620.
(b) D. McTavish and P. Benfield, Drugs, 1990, 39, 136– 17 D. W. Old, M. C. Harris and S. L. Buchwald, Org. Lett.,
153. 2000, 2, 1403–1406.
5 T. S. Rao, J. A. Cler, S. J. Mick, D. M. Ragan, 18 S. Chandrasekhar, S. S. Sultana, S. R. Yaragorla and
T. H. Lanthorn, P. C. Contreras, S. Iyengar and P. L. Wood,
Neuropharmacology, 1990, 29, 1199–1204.
6 (a) K. Lertratanangkoon and M. G. Horning, Drug Metab.
Dispos., 1982, 10, 1–10; (b) J. L. Maggs, M. Pirmohamed,
N. R. Kitteringham and B. K. Park, Drug Metab. Dispos.,
1997, 25, 275–280.
N. R. Reddy, Synthesis, 2006, 839–842.
19 (a) D. Ma and Q. Cai, Synlett, 2004, 128–130; (b) H. Zhang,
Q. Cai and D. Ma, J. Org. Chem., 2005, 70, 5164–5173;
(c) A related procedure uses CuI without a ligand, in DMF
at 120 °C: L. Zhu, P. Guo, G. Li, J. Lan, R. Xie and J. You,
J. Org. Chem., 2007, 62, 8535–8538.
7 (a) S. Madden, J. L. Maggs and B. K. Park, Drug Metab. 20 R. K. Rao, A. B. Naidu, E. A. Jaseer and G. Sekar, Tetrahe-
Dispos., 1996, 24, 469–479; (b) R. E. Pearce, J. P. Uetrecht dron, 2009, 65, 4619–4624.
and J. S. Leeder, Drug Metab. Dispos., 2005, 33, 1819–1826; 21 G. Tarzia, A. Duranti, G. Gatti, G. Piersanti, A. Tontini,
(c) H. Z. Bu, P. Kang, A. J. Deese, P. Zhao and W. F. Pool,
Drug Metab. Dispos., 2005, 33, 1920–1924.
S. Rivara, A. Lodola, P. V. Plazzi, M. Mor, S. Kathuria and
D. Piomelli, ChemMedChem, 2006, 1, 130–139.
8 E. Elliott, E. R. Bowkett, J. L. Maggs, J. Bacsa, B. K. Park, 22 W. Deng, Y. F. Wang, C. Zhang, L. Liu and Q. X. Guo, Chin.
S. L. Regan, P. M. O’Neill and A. V. Stachulski, Org. Lett.,
2011, 13, 5592–5595.
Chem. Lett., 2006, 17, 313–316.
23 E. R. Bowkett, unpublished observations; cf. ref. 8.
9 (a) E. Elliott, S. L. Regan, J. L. Maggs, E. R. Bowkett, 24 V. Lucini, M. Pannacci, F. Scaglione, F. Fraschini, S. Rivara,
L. Parry, D. P. Williams, B. K. Park and A. V. Stachulski,
J. Med. Chem., 2012, 55, 9773–9784; (b) J. Farrell,
M. Lichtenfels, A. Sullivan, E. Elliott, A. Alfirevic,
M. Mor, F. Bordi, P. V. Plazzi, G. Spadoni, A. Bedini,
G. Piersanti and G. Tarzia, J. Med. Chem., 2004, 47, 4202–
4212.
A. V. Stachulski, M. Pirmohamed, D. J. Naisbitt and 25 R. Eckardt and H.-J. Jaensch, Process for the preparation of
B. K. Park, J. Allergy Clin. Immunol., 2013, 132, 493–495.
10 E.g., (a) P. N. Craig, B. M. Lester, A. J. Saggiomo, C. Kaiser
5H-dibenz[b, f]azepine-5-carboxamide from iminostilbene
and alkali cyanates, DE 4307181, 10th Nov. 1994.
and C. L. Zirkle, J. Org. Chem., 1961, 26, 135–138; 26 G. J. Durant, Improved synthesis of N,N-diarylureas, Chem.
(b) W. Schindler and H. Blattner, Helv. Chim. Acta, 1961, 44,
753–763.
Ind., 1965, 1428–1429. These conditions were originally
employed for N,N-diarylamines.
11 Cf. K. Smith, D. M. James, A. G. Mistra, M. R. Bye and 27 W. Raml and G. Eichberger, Preparation of carbamazepine
D. J. Faulkner, Tetrahedron, 1992, 48, 7479–7488; up to four
Br atoms may be introduced into 2 by acid-catalysed
bromination.
by the phosgenation and amidation of 5H-dibenz[b, f]
azepine, AT01174B, 1996.
28 (a) L. Cotarca and H. Eckert, Phosgenations – A Handbook,
Wiley-VCH, Weinheim, 2004; (b) G. Skorna and I. Ugi,
Angew. Chem., Int. Ed. Engl., 1977, 16, 259–260.
12 (a) D. Tsvelikhovsky and S. L. Buchwald, J. Am. Chem. Soc.,
2010, 132, 14048–14051; (b) D. Tsvelikhovsky and
S. L. Buchwald, J. Am. Chem. Soc., 2012, 134, 16917– 29 H. Eckert and B. Forster, Angew. Chem., Int. Ed. Engl., 1987,
16917. 26, 894–895.
13 X. Zhang, Y. Yang and Y. Liang, Tetrahedron Lett., 2012, 53, 30 (a) Cf. N. Ghiasi, A. A. Oskouie and H. Saeidian, Carbohydr.
6406–6408.
Res., 2012, 348, 47–54; (b) D. Che, N. Corelli-Rennie,
B. R. Guntoori and J. Faught, Process for the preparation of
oxcarbazepine and related intermediates, US 20050282797,
2005.
14 T. K. Jorgensen, K. E. Andersen, J. Lau, P. Madsen and
P. O. Huusfeldt, J. Heterocycl. Chem., 1999, 36, 57–64.
15 G. P. Tokmakov and I. I. Grandberg, Tetrahedron, 1995, 51,
2091–2098.
16 (a) M. Watanabe, M. Nishiyama, T. Yamamoto and Y. Koie,
Tetrahedron Lett., 2000, 41, 481–483; (b) M. Nishiyama,
31 This reagent has generally been used for carbamoylation of
alcohols, e.g. T. M. Tagmose and M. Bols, Chem.–Eur. J.,
1997, 3, 453–462.
8434 | Org. Biomol. Chem., 2013, 11, 8426–8434
This journal is © The Royal Society of Chemistry 2013