Highly efficient Narasaka-Heck cyclizations mediated by P(3,5-(CF 3)2C6H3)3: Facile access to N-heterobicyclic scaffolds

Article abstract of DOI:10.1002/anie.201107511

N-heterobicyclic scaffolds: Highly efficient palladium-catalyzed cyclizations of oxime esters with cyclic alkenes were used as a general entry to perhydroindole and related scaffolds. The chemistry is reliant upon the use of P(3,5-(CF₃)₂

Full text of DOI:10.1002/anie.201107511

Angewandte
Chemie
DOI: 10.1002/anie.201107511
Palladium-Catalyzed Cyclization
Highly Efficient Narasaka–Heck Cyclizations Mediated by
P(3,5-(CF3)2C6H3)3: Facile Access to N-Heterobicyclic Scaffolds**
Adele Faulkner and John F. Bower*
In recent years, the drive for more potent and selective drugs
has seen medicinal chemistry move away from “flat” com-
pounds and become increasingly interested in evaluating
chiral “3D” drug candidates.[1] The complexities associated
with accessing such molecules require the development of
more efficient and versatile entries to chiral scaffolds. In this
regard, and given the importance of nitrogen,[2] new methods
3
À
for the efficient and controlled formation of C(sp ) N bonds,
specifically those embedded within N-heterocyclic systems,
are likely to be of great interest to the pharmaceutical sector.
With these considerations in mind, we initiated explora-
tions into the Pd-catalyzed Heck-type cyclization of oxime
esters[3] with cyclic alkenes. In this scenario the mechanistic
requirements of syn-imino-palladation and syn-b-hydride
3
À
elimination would be expected to mandate C(sp ) N bond
formation (Scheme 1a). Importantly, the oxime ester sub-
strates are readily accessible and further derivatization of the
cyclization products is possible by manipulation of either the
imine or alkene functionalities.
The Pd-catalyzed cyclization of oxime esters with alkenes
was first reported by Narasaka, et al. and provides an entry to
pyrroles via the catalytic generation and trapping of an imino-
PdII intermediate (Scheme 1b).[3] Applications of the “Nar-
asaka–Heck” method to other heteroaromatic classes[4] and
cascade reactions[5] have been reported and the scope of the
oxime activating group has been examined.[6] The catalytic
Scheme 1. a) Proposed Pd-catalyzed approach to N-heterobicyclic scaf-
folds. b) The Narasaka–Heck pyrrole synthesis.
generation and trapping of metalated imines has also been
exploited in other contexts.[7–9]
In general, methods that are reliant upon this process
2
À
produce C(sp ) N bonds and are therefore not applicable to
oxime ester, with both hydrolysis and Beckmann rearrange-
ment often competing.[3,10] This lack of generality has
precluded the widespread use of the reaction in synthesis.
Herein we show that efficient and general 5-exo cyclizations
involving cyclic alkenes are achievable by employing
P(3,5-(CF3)2C6H3)3 as the ligand. The net result is a facile
the synthesis of chiral N-heterocycles.[3–5] Extension to the
formation of C(sp ) N bonds is an attractive possibility but,
3
À
cascade processes aside,[5a,b] general methods that achieve this
have not been reported. Of particular relevance to this study
is a single example by Fꢀrstner, et al., who showed that
cyclization of oxime ester 1 provided 2 in 54% yield.[10]
However, high catalyst loadings and a symmetrical diene
substrate were required for useful conversion. Indeed,
reported conditions for the Narasaka–Heck process are
highly sensitive to the electronic and steric demands of the
and asymmetric entry to perhydroindole and related scaffolds
3
À
by formation of a key C(sp ) N bond. This work sets the stage
for the development of other general entries to chiral
N-heterocyclic architectures.
Our initial studies focused upon the cyclization of penta-
fluorobenzoyl oxime ester 3a to imine 4a.[11–13] Under
conditions related to those employed by Fꢀrstner, et al.,[10a]
adduct 4a was accessible in only 29% yield and hydrolysis of
the oxime ester to the corresponding ketone and, to a lesser
extent, oxime predominated (Table 1, entry 1). Improve-
ments in yield were observed when the ligand was changed
to xantphos (entry 2) and, using this as a basis, we were able to
achieve a 58% yield of 4a after the optimization of other
parameters (entry 3). At this stage, a more thorough ligand
screen was performed and we found that P(p-FC6H4)3 allowed
[*] A. Faulkner, Dr. J. F. Bower
School of Chemistry, University of Bristol
Bristol, BS8 1TS (UK)
E-mail: john.bower@bris.ac.uk
[**] We thank the Royal Society for a University Research Fellowship (to
J.F.B.) and Dr. Mairi Haddow (University of Bristol) for X-ray
crystallography.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201107511.
Angew. Chem. Int. Ed. 2012, 51, 1675 –1679
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1675

.
Angewandte
Communications
Table 1: Selected optimization results for the cyclization of 3a to 4a.[a]
Table 2: Scope of the oxime ester.[a]
3a: R1 =Ph, R2 =H
3b: R1 =4-Py, R2 =H
3d: R1 =nBu, R2 =H 3g: R1 =c-Pr, R2 =H
Entry Pd-source (X mol%) Ligand (Y mol%)
T [8C] Yield %
3e: R1 =iPr, R2 =H
3h: R1 =Ph, R2 =Me
3i: R1 =H, R2 =H
1
2
Pd(OAc)2 (10)
Pd(OAc)2 (10)
[Pd2(dba)3] (5)
[Pd2(dba)3] (5)
[Pd2(dba)3] (5)
[Pd2(dba)3] (5)
[Pd2(dba)3] (5)
[Pd2(dba)3] (5)
[Pd2(dba)3] (2.5)
[Pd2(dba)3] (5)
P(o-tolyl)3 (10)
xantphos (10)
xantphos (10)
P(2-furyl)3 (20)
P(C6F5)3 (20)
120
120
100
100
100
100
100
29
52
58
38
10
76
85
90
93
29
3c: R1 =2-Npth, R2 =H 3 f: R1 =tBu, R2 =H
3
4
5
6
7
P(p-FC6H4)3 (20)
P(p-CF3C6H4)3 (20)
4a 93% yield (608C)
4b 74% yield (608C) 4c 81% yield (808C)
8
P(3,5-(CF3)2C6H3)3 (20) 100
9
10
P(3,5-(CF3)2C6H3)3 (10)
P(3,5-(CF3)2C6H3)3 (20)
60
20
[a] dba=dibenzylideneacetone, xantphos=4,5-bis(diphenylphosphino)-
4d 74% yield (1208C) 4e 85% yield (1208C) 4 f 56% yield (1208C)
9,9-dimethylxanthene.
the generation of 4a in 76% yield (entry 6). Further
sequential improvements were observed upon switching to
P(p-CF3C6H4)3 (85%; entry 7) and then P(3,5-(CF3)2C6H3)3
(90%; entry 8). Using this ligand, we were able to decrease
both the catalyst loading and the reaction temperature such
that 4a became accessible in 93% yield using 2.5 mol%
[Pd2(dba)3] at 608C (entry 9). Indeed, under these conditions,
significant conversion to 4a was observed even at room
temperature (entry 10). It should be recognised that other
electron deficient phosphines, such as P(C6F5)3 and P(2-furyl)3
(entries 4 and 5), were not effective for this transformation.[14]
The relative stereochemistry of 4a was determined by NOE
analysis, according to the method of Butts et al.[15] and was
later confirmed by X-ray crystallography (see below).
4g 77% yield (1208C) 4h 89% yield (808C) 4i not observed
[a] Reaction temperatures are specified in parentheses. Npth=naphthyl,
Py=pyridyl.
to 4a and, in several cases, were confirmed by NOE analysis
(see the Supporting Information).
Access to higher substitution patterns is facilitated by
increasing the substitution at C2 (Table 3). Oxime esters 5a–c
Table 3: Cyclization of C2-disubstituted substrates.
Having established suitable conditions on model substrate
3a, we investigated the scope of the process with respect to
the oxime ester (Table 2). The reaction conditions tolerate
both electron deficient and electron rich aryl groups and both
4b and 4c were formed in good yield. Cyclizations involving
alkyl-substituted substrates were more challenging. For
example, cyclization of 3d was not efficient at 608C and
significant hydrolysis to the corresponding ketone (40%) was
observed. However, increasing the reaction temperature to
1208C resulted in an efficient and fast cyclization to 4d with
only minimal quantities (< 15%) of ketone formed; similar
trends were seen with cyclopropyl variant 3g.[16] Isopropyl
substrate 3e, which is sterically less susceptible to hydrolysis,
cyclized in excellent yield at 1208C, and even the hindered
tert-butyl variant 3 f underwent cyclization in moderate yield.
The reaction also proceeds when the oxime ester is
a-trisubstituted (for example, 3h to 4h); this demonstrates
that cyclization does not require prior tautomerization to a
palladated enamine. As expected,[17] the method does not
tolerate aldoxime ester substrates, and the cyclization of 3i
did not afford target 4i. In this case, Beckmann rearrange-
ment predominated to deliver the corresponding nitrile; this
byproduct was also observed, to a lesser extent (28%), in the
cyclization of 3 f to 4 f.[17] The stereochemical outcomes of all
of the cyclizations shown in Table 2 were assigned by analogy
5a, R1 =Ph, R2 =Me
5b, R1 =Ph, R2 =Bn
5c, R1 =Me, R2 =Et
6a 88% yield[a]
(>19:1 d.r.)
6b 76% yield
(>19:1 d.r.)
6c 55% yield
(10:1 d.r.)
[a] 6a was obtained in 88% yield and 7:2 d.r. when the reaction was run
at 608C.
were synthesized as 1:1 mixtures of diastereomers at C2, but
this lack of selectivity was inconsequential, as cyclization at
1208C delivered heterocyclic adducts 6a–c in moderate to
excellent yield and in high diastereomeric purity. When the
cyclization of 5a to 6a was performed at 608C, lower
diastereoselectivity (7:2 vs. > 19:1) was observed. Resubject-
ing this mixture to the reaction conditions at 1208C led to the
isolation of 6a in greater than 19:1 d.r. and this supports base-
mediated equilibration to the major diastereomer after
1676
www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 1675 –1679

Angewandte
Chemie
cyclization. The relative stereochemistry of 6a–c was con-
firmed by NOE analysis (see the Supporting Information).
This method is also efficient for 5-exo cyclizations onto
other cyclic alkenes (Scheme 2). For example, cyclization of
Scheme 2. Cyclizations involving five- and seven-membered-ring
alkenes. For standard conditions, see Table 2.
7a, which involves a five-membered-ring alkene, generated
8a in 82% yield and as a single diastereomer, the stereo-
chemistry of which was determined by NOE analysis (see the
Supporting Information). Cyclization of 7b delivered adduct
8b in high yield and as a 5:1 mixture of diastereomers; the
imperfect selectivity observed here may be a consequence of
the less conformationally rigid seven-membered-ring cyclic
alkene.
We were interested in evaluating whether the process
could tolerate aryl bromides and chlorides, as these have an
obvious sensitivity to the Pd-catalyzed conditions but provide
a potentially useful handle for further manipulation. We
therefore synthesized substrates 9a and 9b, which possess
relatively activated aryl halides (Scheme 3). Cyclization of
Scheme 4. Asymmetric synthesis of (À)-4a and downstream manipu-
lations of the imine and alkene functional groups. For standard
conditions, see Table 2. dba=dibenzylideneacetone, NMO=N-methyl-
morpholine N-oxide, Ts=tosyl.
Table 2; 608C) provided (À)-4a in high yield and without
erosion of enantiopurity (as determined by chiral HPLC
analysis of 12 and (À)-4a using the corresponding racemates
as standards; see the Supporting Information).
To demonstrate potential manipulations of the imine and
alkene functionalities, we reduced (À)-4a with NaBH4 and,
upon N-tosylation, sulfonamide 13 was obtained in high
diastereoselectivity. That the reduction occurred preferen-
tially from the least hindered imine face was confirmed by
X-ray crystallographic analysis of the major diastereomer of
13.[20] This also served to confirm both the relative stereo-
chemical outcome of the Pd-catalyzed cyclization and the
absolute stereochemical outcome of the asymmetric Tsuji–
Trost process. Upjohn dihydroxylation[21] of 13 then occurred
from the convex face to provide diol 14 as a single
diastereomer and in high yield. Thus, compound 14 is
accessible in over 40% yield from 11 with control over both
the absolute and relative stereochemistry of all five stereo-
centers.
The origins of the high efficiency of P(3,5-(CF3)2C6H3)3 in
the reactions presented herein remain to be rigorously
determined. However, beneficial effects upon both the life-
time and reactivity of the putative imino-PdII intermediate
can be envisaged. Electron deficient phosphine ligands may
enhance the stability of this species by rendering the Pd center
more electron-deficient. This, in turn, would be expected to
lead to greater s donation from the imine moiety and the
resulting stronger imino-Pd bond should be less susceptible to
protodepalladation (which would lead to ketone hydrolysis
products).[22] Furthermore, a more electron-deficient imino-
PdII intermediate may enhance the rate of iminopalladation.
Recent computational and experimental studies by Hartwig
Scheme 3. Cyclization of substrates containing aryl halides. For stan-
dard conditions, see Table 2.
chloride 9a was reasonably efficient and adduct 10a was
formed in 63% yield. However, the reaction was much slower
than, for example, the cyclization of 3a to 4a and so required
higher temperatures for reasonable rates. This observation, in
conjunction with the lower yield of 10a (vs. 4a), suggests
competitive Pd insertion into the aryl chloride bond. Cycliza-
tion of bromide 9b was not efficient and 10b was isolated in
only 20% yield.[18]
As alluded to in the introduction, oxime ester substrates
are highly flexible because they are synthesized by established
carbonyl a-functionalization strategies. As such, the method-
ology presented herein is readily translated to asymmetric
synthesis. To demonstrate this, we have prepared adduct 4a in
enantioenriched form (Scheme 4). Accordingly, decarboxyla-
tive Tsuji–Trost cyclohexenylation of 11, following the
method of Tunge and Burger,[19] delivered ketone 12 in
good yield and with high levels of enantioenrichment (93%
ee). Advancement to oxime ester 3a was then readily
achieved and cyclization under standard conditions (see
À
on the migratory insertion of alkenes into related Pd NPh2
Angew. Chem. Int. Ed. 2012, 51, 1675 –1679
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1677

.
Angewandte
Communications
13, 5394; b) S. Liu, L. S. Liebeskind, J. Am. Chem. Soc. 2008, 130,
6918; c) S. Liu, Y. Yu, L. S. Liebeskind, Org. Lett. 2007, 9, 1947.
bonds have shown that electron-deficient phosphine ligands
accelerate this process.[23]
In summary, we present highly efficient palladium-cata-
lyzed cyclizations of oxime esters with cyclic alkenes as the
basis of a general entry to perhydroindole and related
À
[8] Pd-catalyzed intramolecular aryl C H amination: Y. Tan, J. F.
Hartwig, J. Am. Chem. Soc. 2010, 132, 3676. X-ray crystallog-
raphy was used to confirm the formation and geometry of the
imino-PdII intermediate.
scaffolds. The chemistry is reliant upon the use of
[9] Pd-catalyzed decarboxylative nitrene generation: K. Okamoto,
T. Oda, S. Kohigashi, K. Ohe, Angew. Chem. 2011, 123, 11672;
Angew. Chem. Int. Ed. 2011, 40, 11470
[10] a) A. Fꢀrstner, K. Radkowski, H. Peters, Angew. Chem. 2005,
117, 2837; Angew. Chem. Int. Ed. 2005, 44, 2777; b) A. Fꢀrstner,
K. Radkowski, H. Peters, G. Seidel, C. Wirtz, R. Mynott, C. W.
Lehmann, Chem. Eur. J. 2007, 13, 1929. The sensitivity and
3
À
P(3,5-(CF3)2C6H3)3 for the key C(sp ) N bond-forming
process and this facilitates cyclizations with enhanced levels
of efficiency across a range of sterically and electronically
distinct substrates. The rigid N-heterobicyclic products arising
from these studies can easily be accessed in enantioenriched
form and are readily manipulated through both the imine and
alkene functionalities. Future studies will focus upon the
development of enhanced catalyst systems and applications to
the synthesis of other chiral N-heterocyclic classes.
volatility of cyclization product 2 were noted as complicating
3
À
factors in its isolation. For the observation of C(sp ) N bond
formation as a minor byproduct under Narasaka– Heck con-
ditions, see reference [3a].
[11] The oxime esters employed in this study were synthesized in high
yield from the corresponding carbonyl compounds. These, in
turn, were prepared by alkylation of the corresponding b-keto
ester or hydrazone with the appropriate cyclic allylic bromides;
details are provided in the Supporting Information. Pentafluor-
obenzoyl esters were chosen, as they have been established to be
largely resistant to Beckmann rearrangement (see refer-
ence [3]). We have also evaluated other oxime ester derivatives
(for example, benzoyl, pivaloyl, and picalinoyl), but these were
less effective; full details will be reported in due course.
[12] Oxime ester starting materials were usually obtained as a
mixture of geometric isomers. Previous studies have established
that this is inconsequential, as interconversion occurs at the stage
of the imino-PdII intermediate (see reference [3]). Unambiguous
assignment of oxime ester geometry by NMR is not readily
achievable.
[13] Reactions were monitored by TLC until full consumption of
starting material occurred. Under optimized conditions, pro-
longed reaction times do not adversely affect product yield.
[14] In the case of P(2-furyl)3, complete consumption of 3a occurred,
but significant quantities of ketone hydrolysis product were
observed. When P(C6F5)3 was used, the conversion of 3a was
low. These observations point to a potentially narrow steric and/
or electronic window for effective ligands for these specific
processes.
Received: October 25, 2011
Published online: January 9, 2012
Keywords: asymmetric synthesis · homogeneous catalysis ·
.
Narasaka–Heck reactions · N-heterocycles
[1] For an insightful discussion, see: F. Lovering, J. Bikker, C.
Humblet, J. Med. Chem. 2009, 52, 6752; see also: W. P. Walters, J.
Green, J. R. Weiss, M. A. Murcko, J. Med. Chem. 2011, 54, 6405
[2] Over 95% of the 40 best-selling brand name drugs contain
nitrogen. Of these, approximately three-quarters contain N-
heterocycles; see: http://cbc.arizona.edu/njardarson/group/sites/
default/files/Top%20200%20Brand-
name%20Drugs%20by%20US%20Retail%20Sale
s%20in%202010sm_0.pdf.
[3] a) H. Tsutsui, K. Narasaka, Chem. Lett. 1999, 28, 45; b) H.
Tsutsui, M. Kitamura, K. Narasaka, Bull. Chem. Soc. Jpn. 2002,
75, 1451; For reviews, see: c) M. Kitamura, K. Narasaka, Chem.
Rec. 2002, 2, 268; d) K. Narasaka, M. Kitamura, Eur. J. Org.
Chem. 2005, 4505.
[4] Imidazoles: a) S. Zaman, K. Mitsuru, A. D. Abell, Org. Lett.
2005, 7, 609; Azaazulenes: b) M. Kitamura, S. Chiba, O. Saku, K.
Narasaka, Chem. Lett. 2002, 31, 606; c) S. Chiba, M. Kitamura,
O. Saku, K. Narasaka, Bull. Chem. Soc. Jpn. 2004, 77, 785;
Isoquinolines and pyridines: d) H. Tsutsui, K. Narasaka, Chem.
Lett. 2001, 30, 526; e) M. Kitamura, D. Kudo, K. Narasaka,
ARKIVOC 2006, (iii), 148; f) J.-L. Zhu, Y.-L. Su, Y.-H. Chan, I.-
C. Chen, C.-C. Liao, Heterocycles 2009, 78, 369; phenanthri-
dines: g) T. Gerfaud, L. Neuville, J. Zhu, Angew. Chem. 2009,
121, 580; Angew. Chem. Int. Ed. 2009, 48, 572; h) Pyrroles: J.
Ichikawa, R. Nadano, N. Ito, Chem. Commun. 2006, 4425; see
also reference [3].
[5] Heck cascades: a) M. Kitamura, S. Zaman, K. Narasaka, Synlett
2001, 974; b) S. Zaman, M. Kitamura, K. Narasaka, Bull. Chem.
Soc. Jpn. 2003, 76, 1055; Reductive and Suzuki cascades: c) M.
Kitamura, Y. Moriyasu, T. Okauchi, Synlett 2011, 643; d) Car-
bonylative cascades: S. Zaman, M. Kitamura, A. D. Abell, Aust.
J. Chem. 2007, 60, 624; see also: e) T. Nishimura, Y. Nishiguchi,
Y. Maeda, S. Uemura, J. Org. Chem. 2004, 69, 5342; f) T.
Nishimura, S. Uemura, J. Am. Chem. Soc. 2000, 122, 12049.
[6] Hydrazonium salts: M. Kitamura, H. Yanagisawa, M. Yamane,
K. Narasaka, Heterocycles 2005, 65, 273; diethylphosphinylox-
ime esters: J. L. Zhu, Y. H. Chan, Synlett 2008, 1250; for other
activating groups, see references [3a,b] and [5e].
[15] a) C. P. Butts, C. R. Jones, E. C. Towers, J. L. Flynn, L. Appleby,
N. J. Barron, Org. Biomol. Chem. 2011, 9, 177; b) C. P. Butts,
C. R. Jones, J. N. Harvey, Chem. Commun. 2011, 47, 1193; details
are given in the Supporting Information.
[16] At lower temperatures, the rate of hydrolysis is competitive with
the rate of cyclization in these cases. Our observations are that
aryl-substituted oxime esters undergo cyclization at faster rates
than alkyl-substituted variants. After oxidative addition, there is
a requirement for the imino-PdII moiety to be oriented towards
the alkene. Geometric interconversion to this configuration is
likely enhanced by higher reaction temperatures and bulkier R1
substituents. The comparatively low steric bulk of the alkyl
substituent of 3d may explain the inefficiencies observed at
lower temperatures for this substrate; this will be a focus of
future studies. It is unclear whether hydrolysis to the corre-
sponding ketone occurs directly from the oxime ester or via the
presumed imino-PdII intermediate. Reactions run in the absence
of palladium result in minimal oxime ester hydrolysis, but these
conditions do not exactly mimic a catalytically active system.
The use of drying agents (for example, molecular sieves or
Na2SO4) was only marginally beneficial. Taken together, these
observations suggest that the ketone is formed by protodepalla-
dation to the corresponding imine and subsequent hydrolysis
upon workup.
[7] Cu-catalyzed intermolecular cross-couplings: a) Z.-H. Ren, Z.-
Y. Zhang, B.-Q. Yang, Y.-Y. Wang, Z.-H. Guan, Org. Lett. 2011,
1678
www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 1675 –1679

Angewandte
Chemie
[17] For a review of the Beckmann rearrangement, see: R. E.
Gawley, Org. React. 1988, 35, 1; for similar observations see,
for example, reference [5b]. Compound 3i is relatively unstable
and undergoes normal Beckmann rearrangement (to the corre-
sponding amide) slowly at room temperature. Compound 3 f is
stable to the reaction conditions in the absence of palladium; this
implicates a Pd-mediated Beckmann rearrangement pathway in
this case.
[18] In these reactions, dehalogenated cyclization products (such as
4a) are not observed. The mass balance of the reactions is low,
and we suspect this is due to polymerization of aryl-PdII
intermediates via Heck-type pathways.
[19] E. C. Burger, J. A. Tunge, Org. Lett. 2004, 6, 4113.
[20] CCDC 847234 (13) contains the supplementary crystallographic
data for this paper. These data can be obtained free of charge
from The Cambridge Crystallographic Data Centre via www.
ccdc.cam.ac.uk/data_request/cif.
[21] V. VanRheenen, R. C. Kelly, D. Y. Cha, Tetrahedron Lett. 1976,
17, 1973.
[22] This suggestion does not preclude hydrolysis by other pathways.
[23] P. S. Hanley, J. F. Hartwig, J. Am. Chem. Soc. 2011, 133, 15661;
see also: J. D. Neukom, N. S. Perch, J. P. Wolfe, Organometallics
2011, 30, 1269, and references therein.
Angew. Chem. Int. Ed. 2012, 51, 1675 –1679
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1679