Asymmetric allylic substitution catalyzed by C1-symmetrical complexes of molybdenum: Structural requirements of the ligand and the stereochemical course of the reaction

Article abstract of DOI:10.1002/chem.200501574

Application of new chiral ligands (R)-(-)-12a and (S)-(+)-12c (VALDY), derived from amino acids, to the title reaction, involving cinnamyl (linear) and isocinnamyl (branched) type substrates (4 and 5 → 6), led to excellent regio- and enantioselectivities (>30:1, ≤98% ee), showing that ligands with a single chiral center are capable of high asymmetric induction. The structural requirements of the ligand and the mechanism are discussed. The application of single enantiomers of deuterium-labeled substrates (both linear 38c and branched 37c) and analysis of the products (41-43) by ²{ ¹H) NMR spectroscopy in a chiral liquid crystal matrix allowed the stereochemical pathways of the reaction to be distinguished. With ligand (S)-(+)-12c, the matched enantiomer of branched substrate was found to be (S)-5, which was converted into (R)-6 with very high regio- and stereoselectivity via a process that involves net retention of stereochemistry. The mismatched enantiomer of the branched substrate was found to be (R)-5, which was also converted into (R)-6, that is, with apparent net inversion, but at a lower rate and with lower overall enantioselectivity. This latter feature, which may be termed a "memory effect", reduced the global enantioselectivity in the reaction of the racemic substrate (±)-5. The stereochemical pathway of the mismatched manifold has been shown also to be one of net retention, the apparent inversion occurring through equilibration via an Mo-allyl intermediate prior to nucleophilic attack. Incomplete equilibration leads to the memory effect and thus to lower enantioselectivity. Analysis of the mismatched manifold over the course of the reaction revealed that the memory effect is progressively attenuated with the nascent global selectivity increasing substantially as the reaction proceeds. The origin of this effect is suggested to be the depletion of CO sources in the reaction mixture, which attenuates turnover rate and thus facilitates greater equilibrium. The linear substrate was also converted into the branched product with net syn stereochemistry, as shown by isotopic labeling. An analogous process operates in the generation of small quantities of linear product from branched substrate.

Full text of DOI:10.1002/chem.200501574

DOI: 10.1002/chem.200501574
Asymmetric Allylic Substitution Catalyzedby C1-Symmetrical Complexes of
Molybdenum: Structural Requirements of the Ligand and the Stereochemical
Course of the Reaction
Andrei V. Malkov,*[a] Laure Gouriou,[b] Guy C. Lloyd-Jones,*[b] Ivo Stary,[a, c]
´
Vratislav Langer,[d] Paul Spoor,[e] Victoria Vinader,[f] andPavel Kocˇovsky*[a]
´
Dedicated to Professor Richard Heck on the occasion of his 75th birthday
Abstract: Application of new chiral li-
gands (R)-(ꢀ)-12a and (S)-(+)-12c
(VALDY), derived from amino acids,
to the title reaction, involving cinnamyl
(linear) and isocinnamyl (branched)
type substrates (4 and 5 ! 6), led to
excellent regio- and enantioselectivities
(>30:1, ꢁ98% ee), showing that li-
gands with a single chiral center are ca-
pable of high asymmetric induction.
The structural requirements of the
ligand and the mechanism are dis-
cussed. The application of single enan-
tiomers of deuterium-labeled substrates
(both linear 38c and branched 37c)
and analysis of the products (41–43) by
2H{1H} NMR spectroscopy in a chiral
liquid crystal matrix allowed the ster-
eochemical pathways of the reaction to
be distinguished. With ligand (S)-(+)-
12c, the matched enantiomer of
branched substrate was found to be
(S)-5, which was converted into (R)-6
with very high regio- and stereoselec-
tivity via a process that involves net re-
tention of stereochemistry. The mis-
matched enantiomer of the branched
substrate was found to be (R)-5, which
was also converted into (R)-6, that is,
with apparent net inversion, but at a
lower rate and with lower overall enan-
tioselectivity. This latter feature, which
may be termed a “memory effect”, re-
duced the global enantioselectivity in
the reaction of the racemic substrate
(Æ)-5. The stereochemical pathway of
the mismatched manifold has been
shown also to be one of net retention,
the apparent inversion occurring
through equilibration via an Mo–allyl
intermediate prior to nucleophilic
attack. Incomplete equilibration leads
to the memory effect and thus to lower
enantioselectivity. Analysis of the mis-
matched manifold over the course of
the reaction revealed that the memory
effect is progressively attenuated with
the nascent global selectivity increasing
substantially as the reaction proceeds.
The origin of this effect is suggested to
be the depletion of CO sources in the
reaction mixture, which attenuates
turnover rate and thus facilitates great-
er equilibrium. The linear substrate
was also converted into the branched
product with net syn stereochemistry,
as shown by isotopic labeling. An anal-
ogous process operates in the genera-
tion of small quantities of linear prod-
uct from branched substrate.
Keywords: allylic substitution
·
asymmetric catalysis · chiral ligands ·
deuterium labeling · memory effect ·
molybdenum
[d] Dr. V. Langer
´
ˇ
´
[a] Dr. A. V. Malkov, Dr. I. Stary, Prof. Dr. P. Kocovsky
Department of Chemistry, WestCHEM, Joseph Black Building
University of Glasgow, Glasgow G12 8QQ (UK)
Fax : (+44)141-330-4888
Environmental Inorganic Chemistry
Department of Chemical and Biological Engineering
Chalmers University of Technology, 41296 Gçteborg (Sweden)
E-mail: amalkov@chem.gla.ac.uk
[e] P. Spoor
pavelk@chem.gla.ac.uk
Department of Chemistry, University of Leicester
Leicester LE1 7RH (UK)
[b] Dr. L. Gouriou, Prof. Dr. G. C. Lloyd-Jones
The Bristol Centre for Organometallic Catalysis, School of Chemistry
University of Bristol, Cantockꢀs Close, Bristol BS8 1TS (UK)
Fax : (+44)117-929-8611
[f] Dr. V. Vinader
Medicines Research Centre, GlaxoSmithKline PLC
Research and Development, Stevenage, Herts SG1 2NY (UK)
E-mail: guy.lloyd-jones@bris.ac.uk
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author: Crystallographic
analysis with fully labeled ORTEP diagram for (R)-(ꢀ)-12a, atomic
coordinates, selected bond lengths and angles, anisotropic displace-
ment parameters, and hydrogen coordinates.
´
[c] Dr. I. Stary
Institute of Organic Chemistry and Biochemistry
Academy of Sciences of the Czech Republic
16610 Prague 6 (Czech Republic)
6910
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 6910 – 6929

FULL PAPER
Introduction
classical inv–inv pathway could not be excluded in less con-
strained systems such as the cinnamyl-type carbonates 4 and
5. Interestingly, the stoichiometric reaction of the allylic sys-
tems, involving the isolation of the h3 complex, is known to
proceed via ret–inv.[5]
The transition-metal-catalyzed allylic substitution reaction,
starting with an allylic ester 1 (X = OCOR), involves an in-
itial formation of the intermediate h3-complex 2 (Scheme 1).
The subsequent nucleophilic attack at the latter intermedi-
ate affords substitution product 3 with a concomitant release
of the metal in its active form.[1] While Pd0-catalyzed substi-
tution with stabilized C nucleophiles proceeds via double in-
version (inv–inv),[1,2] rather little is known about its Mo0-cat-
alyzed counterpart.[3] In the case of bicyclic allylic systems,
we have shown that a double retention pathway (ret–ret) is
operative.[4] However, the stereochemical outcome in this
particular case may be dependent on the ligand(s), solvent,
and the combination of the reacting partners. Therefore, the
Scheme 1.
While this work was in progress,[6–9] Trost reported on the
first examples of high asymmetric induction in Mo0-cata-
lyzed allylic substitution employing cinnamyl carbonates 4
and 5 and their aromatic and heteroaromatic counterparts
(Scheme 2): with malonate-type nucleophiles and bis-a-pico-
linic amide 8 as the chiral ligand (see below), he attained ex-
cellent regio- and enantioselectivities (Table 1, entries 1 and
2).[10,11]
Abstract in Czech: NovØ chirµlní ligandy (R)-(ꢀ)-12a a (S)-
(+)-12c (VALDY), kterØ jsou odvozenØ od aminokyselin,
byly pouzity pri allylovØ substituci katalyzovanØ molybdenem
ˇ
ˇ
ˇ
ˇ
´
´
v prípade cinnamylovych (lineµrních) a isocinnamylovych
ˇ
´
˚
(rozvetvenych) substrµtu (4 a 5 ! 6). Byla pozorovµna
vysokµ regioselektivita a enantioselektivita (>30:1, ꢁ98%
ˇ
ˇ
ee), coz ukazuje, ze jedinØ centrum chirality je schopno zaji-
stit vysokou míru asymetrickØ indukce. Jsou diskutovµny me-
ˇ
ˇ
chanismus reakce a strukturní pozadavky na ligandy. Pouzití
´
˚
˚
ˇ
´
jednotlivych enantiomeru substrµtu znacenych deuteriem (jak
ˇ
´
´
˚
lineµrních 38c tak rozvetvenych 37c) a analyza produktu
2
(41–43) pomocí H{1H} NMR spektroskopie v chirµlní ka-
ˇ
ˇ
ˇ
ˇ
ˇ
palne krystalickØ fµzi umoznily rozlisit reakcní kanµly lisící
ˇ
ˇ
se z hlediska stereochemie. V prípade ligandu (S)-(+)-12c je
ˇ
ˇ
´
´
´
pri reakci s rozvetvenym substrµtem souhlasnym partnerem
Scheme 2.
ˇ
enantiomer (S)-5, ktery byl preveden na (R)-6 s vysokou re-
ˇ ˇ
a stereoselektivitou. V uvedenØm prípade
gioselektivitou
dochµzí k celkovØ retenci konfigurace. Pri reakci rozvetvenØ-
ˇ
ˇ
´
Shortly afterwards, Pfaltz designed analogous ligands with
oxazoline units (e.g., 11) in place of Trostꢀs a-picolinic
amide moieties.[12] As a follow-up, Moberg has demonstrated
a further improvement of the reactivity of the Mo com-
plexes of 8 by microwave heating.[13] Most recently, Hughes
has illustrated the capability of Mo·8 complex in the kinetic
resolution of (Æ)-5.[14]
ho substrµtu je nesouhlasnym partnerem enantiomer (R)-5,
´
ˇˇ
ˇ
ktery byl rovnez preveden na (R)-6. Reakce probíhµ s celko-
vou inverzí konfigurace, pomaleji a s nizsí enantioselektivi-
ˇˇ
´
˚ ˇ
´
ˇ
´
tou. Pozorovany jev muze byt nazvµn “pametꢀovym efektem”
ˇ
ˇ
a projevuje se snízením celkovØ enantioselektivity pri konver-
´
zi racemickØho substrµtu (Æ)-5. Reakce mezi nesouhlasnymi
partnery probíhµ s celkovou retencí konfigurace, kdy v
ˇ
ˇ
The Trost/Moberg and Pfaltz ligands 8 and 11 and their
analogues, which facilitate the asymmetric Mo0-catalyzed al-
lylic substitution, are limited to one C2-symmetrical scaffold,
namely trans-1,2-diaminocyclohexane.[10,12,13] It can be hy-
pothesized that the requisite chiral environment about the
metal center might be secured just by one chiral center
(rather than two) in the ligand, as in 12a–d (see below),
which may then be as effective as ligands 8 and 11. This ap-
proach would take advantage of the chiral pool of amino
acids as starting materials, most of which are now available
in both enantiomeric forms at a comparable cost. The proof
of this concept has been demonstrated in our preliminary
communications[8] and further confirmed by the most recent
work of Trost, Hughes, and Krska.[15] Herein, we present an
orchestration of our original work and discuss mechanistic
dílcím kroku dochµzí k inverzi prostrednictvím ekvilibrace
h –(allyl)Mo intermediµtu jeste pred nukleofilním atakem.
Pokud tato ekvilibrace není fflplnµ, uplatní se pametꢀovy
1
ˇ ˇ
ˇ
ˇ
´
´
ˇˇ
ˇ
´
efekt, ktery pak vede k nizsí enantioselektivite. Z analyzy
˚ ˇ
´
˚
´
ˇ
ˇ
prubehu reakce nesouhlasnych partneru vyplyvµ, ze je pame-
´
ˇ
tꢀovy efekt postupne utlumovµn a nascentní enantioselektivita
se zvysuje tak, jak reakce postupuje. Puvod tohoto efektu
spocívµ pravdepodobne v postupnØm zµniku zdroje CO v
reakcní smesi, coz se projeví snízením reakcní rychlosti a
tudíz usnadnením ekvilibrace. Lineµrní substrµt rovnez
podlØhµ premene na rozvetveny produkt s celkovou syn ste-
reochemií, jak bylo ukµzµno pomocí izotopovØho znacení.
ˇ
˚
ˇ
ˇ
ˇ
ˇ
ˇ
ˇ
ˇ
ˇ
ˇ
ˇ
ˇˇ
ˇ
ˇ ˇ
ˇ
´
ˇ
´
ˇ
ˇ
ˇ
Analogicky proces se uplatnuje pri vzniku malØho mnozství
ˇ
lineµrního produktu z rozvetvenØho substrµtu.
Chem. Eur. J. 2006, 12, 6910 – 6929
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6911

ˇ
´
P. Koc ovsky, A. V. Malkov, G. C. Lloyd-Jones et al.
Table 1. Mo0-Catalyzed allylic substitution (Scheme 2).[a]
type substrates is a stereospecif-
Entry
Substrate
Ligand
(R,R)-8[d]
(R,R)-8[d]
R (ligand)
t [h]
23:13
3:113
4
4
4
4
4
12
12
72
24
24
48
24
48
Ratio[b] 6:7
Yield [%]
ee 6 [%][c] (configuration)
ic process (net retention), with
the linear and both enantiomers
of the branched isomers of sub-
strate all following the same
stereochemical pathway; and c)
that p–s–p equilibration, and
not Mo–Mo transfer, is the
mechanism by which stereo-
chemical convergence is ach-
ieved, thereby facilitating asym-
metric induction through chiral
ligand control.
1
2
3
4
5
6
7
8
4
5
4
4
5
4
5
4
5
4
4
4
4
4
4
4
G
1,2-cHex
1,2-cHex
Ph
Ph
Ph
PhCH2
PhCH2
iPr
iPr
tBu
88
70
99 (
92 (
S)[f]
ACHTREUNG
S)[f]
(R)-12a[d]
(R)-12a[e]
(R)-12a[d]
(S)-12b[d]
(S)-12b[d]
(S)-12c[d]
(S)-12c[d]
(S)-12d[e]
(R)-13[d]
(R)-14[e]
(R)-15[d]
(R)-16[e]
(S)-17[d]
8:1
8:1
6392 (
S)
65
72
69
68
68
59
64
31
51
0
92 (S)
88 (S)
89 (R)[g]
74 (R)[g]
98 (R)[g]
97 (R)[g]
59 (R)[g,h]
78 (S)
89 (S)
–
12:1
13:1
13:1
32:1
38:1
13:1
12:1
9:1
9
10
11
12
13
14
15
16
Ph
Ph
–
1:1
2:1
29
47
10 (S)
27 (R)
R)
Ph
(S)-18[d]
31:1
57
5 (
Results andDiscussion
[a] Conditions: THF, 608C, cat. 7–10 mol%. [b] Determined from the 1H NMR spectra of the product mix-
tures. [c] Determined by chiral HPLC. [d] The catalyst was generated from [(EtCN)3Mo(CO)3]. [e] The cata-
Although the mode of coordi-
nation of 8 and 11 to Mo in the
lyst was generated from [(C7H8)Mo(CO)3]. [f] Refs. [10a,15]. [g] Note that the ligand has the opposite absolute
1
configuration to 12a. [h] Determined by H NMR with [D]-Eu
A
active catalyst has not been
firmly established,[10,15,16] there
is a growing body of evidence that a tridentate, anionic, fac-
binding mode is involved. By analogy, it can be argued that
the R group in ligands 12 could act as an anchor, presuma-
bly occupying an “equatorial” position in the cyclic complex,
thereby mimicking the rigid scaffold of 8. We reasoned that
the ligand performance may be tuned by varying the bulk of
the anchor R with the goal of finding an optimal architec-
ture of the whole framework. Therefore, we synthesized a
set of ligands 12a–d, derived from vicinal diamines, originat-
ing from amino acids with the varied side-chain R (phenyl-
glycine, Phe, Val, and tLeu) and the analogues 13–21.
Synthesis of ligands 12a–d: Ligand (R)-12a (Scheme 3) was
readily prepared by conversion of methyl phenylglycinate
(R)-22a into amide (R)-23a (aq. NH3, toluene; 70%),[17] fol-
lowed by reduction (LiAlH4, THF; 45%),[18] and transfor-
mation of the resulting diamine (R)-24a into the desired bis-
amide (R)-(ꢀ)-12a [a-picolinic acid, (PhO)3P, pyridine,
1008C, overnight; 62%].[19,20] The remaining members of the
series, that is, (S)-12b–d, were synthesized in a similar fash-
ion. Thus, the methyl ester of (S)-phenylalanine (S)-22b was
converted into amide (S)-23b (aq. NH3, toluene; 74%);[17]
its reduction (BH3·THF, THF, 708C, 5 h) furnished diamine
(S)-24b (76%). The final acylation [a-picolinic acid,
(PhO)3P, pyridine, 1008C, overnight][19] produced the desired
bisamide (S)-(+)-12b (55%). An analogous sequence, start-
ing with the commercially available valine amide (S)-23c
and proceeding through diamine (S)-24c (LiAlH4, THF;
63%) afforded on final acylation [a-picolinic acid, (PhO)3P,
pyridine, 1008C, overnight] the isopropyl analogue (S)-(+)-
12c (84%). The tert-butyl ligand (S)-(+)-12d was prepared
from the dihydrochloride of diamine (S)-24d[21] [2.0 equiv a-
picolinic acid, 1-hydroxybenzotriazole (2.0 equiv), N-methyl-
morpholine (4.1 equiv), CH2Cl2, 08C, 5 min, then N-[3-(di-
issues in light of our earlier and current results and of the
mechanistic pictures proposed by others. A key outcome of
this study is a) the full account of the structure–enantiose-
lectivity relationship for the ligands coordinated to Mo; b)
the demonstration that Mo-catalyzed allylation of cinnamyl-
methylamino)propyl]-N’-ethylcarbodiimide
hydrochloride
(2.2 equiv), 08C to RT, overnight, 44%].[22]
6912
www.chemeurj.org
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 6910 – 6929

Asymmetric Allylic Substitution
FULL PAPER
Scheme 3. Py = a-pyridyl.
Scheme 4. Ar = Ph or 4-C6H4NO2.
The other required ligands were synthesized as follows
(Scheme 4). Phenylglycinol (R)-25 was acylated with a-pico-
linic acid chloride (Et3N, CH2Cl2, RT, 2 h) and the resulting
hydroxyamide 26 (69%) was converted into amino amide 27
by the Mitsunobu reaction (phthalimide, DEAD, Ph3P, RT,
overnight; 51%), followed by hydrazinolysis (N2H4·H2O,
DMF, RT, overnight; 54%). Amine 27 thus obtained was
acylated with PhCOCl (Et3N, CH2Cl2, RT, 2 h) to afford the
desired amide 13 (80%). In the synthesis of 14, the straight-
forward route from 25 via N-benzoylation, followed by the
Mitsunobu-type amination, could not be used since the ben-
zamide derived from 25 underwent the 5(O)p,n–exo-trig cyc-
lization[23] under the Mitsunobu conditions to produce the
corresponding oxazoline. Therefore, the nitrogen in 25 was
first protected by Boc group [(Boc)2O, CH2Cl2, 08C, 12 h]
and the resulting N-protected amino alcohol 28 (96%) was
submitted to the Mitsunobu reaction (phthalimide, DEAD,
Ph3P, 0 8C to RT, 18 h), followed by hydrazinolysis (N2H4,
EtOH, 808C, 5 h) that afforded the monoprotected diamine
29 (48% overall). Acylation of 29 using the mixed anhy-
dride method (PyCO2H, ClCO2Me, THF; 94%), followed
by deprotection (CF3CO2H, RT, 2 h), furnished amino
amide 31 (86%); benzoylation (PhCOCl, Et3N, THF, 08C,
2 h) afforded the required diamide 14 (49%). The “invert-
ed” ligand 15 was prepared in two steps from methyl phe-
nylglycinate 22a via N-acylation (PyCOCl, Et3N, CH2Cl2,
RT, 2 h) to give (R)-18 (89%), followed by aminolysis of the
ester group (PyCH2NH2, NH4Cl, 858C, 2 h; 78%). Amide
ester 16 was obtained via acylation of (R)-phenylglycinol
(R)-25 [PyCOCl (2 equiv), Et3N, CH2Cl2, RT, 2 h; 67%). Fi-
nally, acylation of (S)-(ꢀ)-a-methylbenzylamine under the
same conditions (PyCOCl, Et3N, CH2Cl2, RT, 2 h) afforded
(S)-17 (72%).
Scheme 5. a) 2.1 equiv PyCO2H, 2.1 equiv (PhO)3P, pyridine, 1008C, 24 h;
76%; b) 1.1 equiv NaH, DMF, 408C, 45 min, then 1.1 equiv CH3I, RT,
2 h, 37%; c) 2.2 equiv NaH, DMF, 408C, 45 min, then 2.2 equiv CH3I,
RT, overnight, 37%.
0
Allylic substitution catalyzedby complexes of Mo with li-
gands 12a–d: Cinnamyl and isocinnamyl carbonates 4 and 5
were employed in conjunction with malonate nucleophile
NaCH(CO2Me)2 to probe the efficiency of ligands 12a–d
G
(Scheme 2, Table 1). The catalyst was generated in situ from
[(EtCN)3Mo(CO)3][24] or [(C7H8)Mo(CO)3][25] and the ligand
in THF.[26] The reactions with NaCH
(CO2Me)2, carried out
U
in THF at 608C, proved to be regio- and enantioselective in
favor of the branched product 6 with good yields (entries 3
and 4).[27] The benzyl ligand (S)-(+)-12b (entries 6 and 7)
turned out to exhibit rather lower enantioselectivity (74–
89% ee)[28] than the phenyl derivative (R)-(ꢀ)-12a (compare
entries 3–5 with 6 and 7). By contrast, the isopropyl ligand
(S)-(+)-12c (VALDY)[29] gave much improved results (en-
tries 8 and 9),[30] whereas the tert-butyl ligand (S)-(+)-12d
performed worse than any other member of this series
(entry 10). Some small differences were observed between
the regioisomeric substrates 4 and 5 (e.g., compare entries 3
vs 5 and 8 vs 9) and identical results were obtained with the
In order to further elucidate the role of the nature of the
amide groups in the ligand, we prepared the nonchiral
ligand 19 (Scheme 5) and its mono- and bismethylated ana-
logues 20 and 21. The parent diamide 19 was obtained by
acylation of ethylene diamine 32; selective mono- and bis-
N-methylation afforded 20 (37%) and 21 (37%), respective-
ly.
catalyst
generated
from
[(EtCN)3Mo(CO)3]
and
[26]
[(C7H8)Mo(CO)3] (compare entries 3and 4).
Chem. Eur. J. 2006, 12, 6910 – 6929
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6913

ˇ
´
P. Koc ovsky, A. V. Malkov, G. C. Lloyd-Jones et al.
Structural features of the ligands relevant to catalysis: Li-
gands 8, 11, and 12 can offer a maximum of four ligating
atoms, namely the pyridine nitrogens[31] and either the
amidic carbonyl groups or nitrogen atoms. Trost originally
proposed a bidentate coordination of ligand 8 to molybde-
num with trans-configuration of the ligating nitrogens about
the metal center (A, see below).[10] Pfaltz has reported an X-
dentate coordination as an essential feature in controlling
the level of asymmetric induction. The enhanced reaction
rate attained with bispicolinic amides 12 suggests that the
additional sp2 nitrogen increases the reactivity of the resting
state of the catalyst, but has a minor effect on the stereose-
lectivity.
The role of the NH in the amide groups was assessed with
the aid of the nonchiral ligand 19 and its mono- and bisme-
thylated analogues 20 and 21. Molybdenum complexes of li-
gands 19–21 were generated in situ in the same way as those
of 12a–d and employed as catalysts for the reaction of 4
with NaCH(CO2Me)2 under the same conditions
G
(Scheme 2). Ligand 19 proved to react with a similar effi-
ciency as its chiral counterparts 12a–d, giving a 5:1 ratio of
6 and 7 (58% in 8.5 h).[34–37] Reduced catalytic activity was
observed with the monomethylated ligand 20 (6/7 3.5:1;
42% in 41 h); bismethyl analogue 21 failed to promote the
reaction. These results clearly demonstrate that at least one
but preferentially two secondary amide functions (-CO-NH-)
are crucial for the reaction to occur with a reasonable
rate.[38] Similar observations were subsequently reported by
Trost and Hughes,[15b] who employed the N,N’-dimethyl
ligand analogue 10.
The deprotonation of a secondary amide unit in 8 was
demonstrated by Trost, Hughes and Krska using NMR spec-
troscopy.[15] The behavior of 19–21 sheds light on the ques-
tion whether mono- or bisdeprotonation is required: The
N,N-bismethyl ligand 21, that cannot be deprotonated, is
inert, whereas 20, that can be mono-deprotonated reacts
similarly to 19 (though more slowly). Hence, these observa-
tions can be regarded as indirect evidence in favor of mono-
deprotonation as the decisive factor in the reactivity of
these bisamidic ligands, which is consistent with the 15N
NMR study.[15,39]
ray structure of the complex obtained from ligand 11 and
[(EtCN)3Mo(CO)3], which shows a tridentate coordination
B (with two sp2 nitrogens and one carbonyl group involved
in the coordination).[16] Most recently, Krska, Hughes and
Trost have reported on the X-ray structure of the Mo–h3-
complex C (Ar = Ph), in which the binding to the deproto-
nated amidic nitrogen has been confirmed by solution 15N
NMR spectroscopy.[15] However, external nucleophilic attack
on C would give rise to the “wrong” enantiomer of the
product, which illustrates the difficulties associated with the
mechanistic issues of this reaction.[32]
Ligand 13, lacking one pyridine nitrogen atom, exhibited
high enantioselectivity in the Mo-catalyzed reaction
(Table 1, entry 11) but the conversion was lower.[33] Interest-
ingly, monopyridine ligand 14 (with the pyridine moiety
remote from the chiral center) exhibited higher reactivity
and selectivity than its positional isomer 13 (entry 12). Simi-
larly, Trost and Hughes[15b] have demonstrated good reactivi-
ty of the monopyridine derivative 9. By contrast, ligands 15
and 16 turned out to be inferior (entries 13and 14). Thus,
15 (a positional isomer of 12a) failed to bring about the re-
action, while 16 (an ester/amide) was non-selective and gave
a practically racemic product in low conversion. Interesting-
ly, the truncated analogue 17, which can be regarded as
“semi-12a”, was found to catalyze the reaction (entry 15)
with better selectivity than 16 and 18. Finally, ligand 18
proved to be the most reactive in the series (57% yield in
3h; entry 16), albeit giving a racemic product. These experi-
ments show that 1) the original structural characteristics of
the Trost–Moberg and Pfaltz ligands 8 and 11, namely the
two rigid amide groups, are essential for the reaction to
occur and that 2) one chiral center in the scaffold is suffi-
cient to induce high levels of enantioselectivity. Noteworthy
is the enhanced reactivity of these Mo catalysts, as com-
pared to the previously studied bipyridine and phenanthro-
line complexes.[31]
Using scalemic and racemic samples of ligand (R)-(ꢀ)-
12a, we tested whether there is a linear or nonlinear rela-
tionship between the enantiomeric excess of the ligand and
that of (S)-6, the product from the asymmetric Mo-catalyzed
alkylation reaction.[40] As is evident in Figure 1, there is a
small positive deviation from linearity resulting in a modest
amplification of the ee of (S)-6 (Figure 1, large data points).
The simple shape of this curve, with maximum amplification
(ca. 6% ee) when the ligand is of 40–50% ee, suggests that
the nonlinear effect arises from equilibrium of “Mo(L)”
with “(Mo)n(L)2”, where n=1 (doubly ligated) or n=2 (di-
meric species). Given the crystallographic evidence from
Krska et al. of a [Mo(CO)2L
(allyl)] intermediate[15b] and the
A
above study, which strongly suggests that only ligands that
can operate in a tridentate monoanionic coordination mode
generate an active and selective species, it seems unlikely
that complexes of the type “(Mo)n(L)2” would be involved
in turnover. Nonetheless, such species could act as reservoirs
of active “Mo(L)” species and if the homochiral form of
“(Mo)n(L)2” is less stable than the heterochiral, then a posi-
tive nonlinear relationship would arise. Using the Kagan
“reservoir” model[40e] [employing Equations (1), (2) and (3)
in ref. [40a], with the limitation that (Mo)n(L)2 is catalytical-
The comparison of the bispicolinic ligands 12 with the mo-
nopicolinic analogues 13 and 14 shows that the catalysts
generated from the latter ligands react more slowly but
retain high enantioselectivities. This behavior indicates a tri-
6914
www.chemeurj.org
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 6910 – 6929

Asymmetric Allylic Substitution
FULL PAPER
Scheme 6.
Figure 1. Non-linear relationship between the enantiomeric excess of
ligand (R)-12a and branched product (S)-6 obtained on asymmetric Mo-
through the h1-complex 33 (with a nonchiral allylic moiety)
or via transfer of the allyl unit between Mo centers, as has
been identified (in restricted cases) in Pd–allyl chemis-
try.[2b,41]
*
catalyzed allylic alkylation of 4. : data points from experiments; a:
linear correlation; +: calculated relationship based on the Kagan “reser-
voir model” for ML2 with g=0 and K=0.06.[39a]
Interestingly, experimental evidence accumulated for the
related W0-catalyzed allylic substitution demonstrates that
isolable h3 complexes are not involved in the productive
part of the catalytic cycle.[3h] As an alternative, one might
consider h1-type intermediates as the active component of
the catalytic cycle. If a similar mechanistic picture were to
apply to Mo, then 4 and both enantiomers of (Æ)-5 would
generate the same h1-complex 33[42] and the facial selectivity
of the subsequent nucleophilic attack would be dictated by
the chirality of the ligand. In such a manner, the enantio-
and regioselectivity of the reaction should be independent
of the identity of the substrate, that is there should be no
“memory effect”.[43] As can be seen in Table 1, for certain li-
gands, this is not the case with the linear and branched sub-
strates giving different enantiomeric excesses and regioselec-
tivities. Hughes et al. have reported on memory effects in
the analogous reactions involving “Mo-8”, with an efficient
kinetic resolution accompanying the reaction of the racemic
branched substrate (Æ)-5.[14] To test for kinetic resolution
during the reactions catalyzed by the complexes bearing the
C1-symmetric type ligands described herein, we monitored
the ee of 5 against conversion (c), starting with racemic sub-
strate and employing an internal standard for chiral GC
analysis. Nonlinear regression of the relationship of c versus
ee for a series of data obtained with ligand (S)-12c
(VALDY) yielded a selectivity factor, s, of 2 in favor of the
reaction of (S)-5 versus (R)-5. The selectivity was substan-
tially lower than that observed with ligand 8.[14] An analo-
gous experiment with ligand (R)-12a yielded an s value of 3
[in favor of (R)-5, consistent with the use of the opposite
configuration]. These experiments suggest that the slower
reacting enantiomers are mismatched in chirality with the
ly inactive and thus g=0], the data may be fit by nonlinear
regression to yield an equilibrium constant K=0.06 between
Mo(L) and (Mo)n(L)2 (Figure 1, small data points). In an
analogous series of experiments with ligand (R)-14, which
bears just one picolinic amide group, no deviation from line-
arity was observed.
Determination of the net stereochemical pathways leading
to asymmetric induction: Racemic isocinnamyl carbonate
(Æ)-5 proved to give similar but not identical regio- and
enantioselectivities as its nonchiral counterpart 4 (Table 1,
entries 1–3, 5–9). This behavior is in sharp contrast to W0-
catalyzed allylation, where asymmetric induction was only
achieved with linear substrates, whereas the branched race-
mic substrates gave racemic product.[6] Indeed, it was later
demonstrated that such reactions proceed with essentially
perfect stereospecificity to give products of net retention.[3h]
The observation that under the asymmetric Mo-catalyzed
conditions, both enantiomers of 5 are converted into the
same enantiomer of product, demonstrates that one enan-
tiomer reacts with overall retention, while the other with in-
version. This analysis indicates that either i) isomerization
must occur in the case of the mismatched pair (of 5 and the
chiral catalyst), either by prior inversion of 5 or at the stage
of an intermediate, or ii) the enantiomers of 5 react with a
Mo complex in an enantiodivergent manner, such that one
reacts with inversion and the other with retention, to yield
the same intermediate (Scheme 6). In the former case, iso-
merization can be conjectured to occur either via diastereo-
facial interconversion of the h3-complexes 34A and 34B
(which are diastereoisomeric by virtue of planar chirality)
Chem. Eur. J. 2006, 12, 6910 – 6929
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6915

ˇ
´
P. Koc ovsky, A. V. Malkov, G. C. Lloyd-Jones et al.
2
1
catalyst and, if a memory effect is operating, will be process-
ed with lower selectivity.[44]
(95% H by H NMR) was obtained from cinnamaldehyde
40 by reduction (93%). Derivatization of alcohol 38a with
enantiopure (R)-(+)-a-methyl benzylisocyanate afforded
two diastereoisomers of 39 (58%). These were used as a
0% de 1H NMR standard for analysis of enantioenriched
samples of 38a after conversion into 39, in which the dia-
stereotopic protons at C1 display suitable dispersion
[4.71 ppm in the C1-(R) diastereoisotopomer and 4.67 ppm
in the C1-(S) diastereoisotopomer].
In order to address the above mechanistic issues, in partic-
ular to elucidate the net stereochemical pathways and
modes of stereochemical convergence (Scheme 6), we de-
signed a stereospecific isotopic labeling strategy that would
probe for p–s–p equilibration in Mo–allyl intermediates. To
study the pathways involving the branched substrate 5, we
deployed enantiomerically enriched samples of monodeuter-
ated branched methyl carbonate 37c where the deuterium
atom is located in the cis position at the allylic terminus
distal from the phenyl group (Scheme 7). To study the reac-
The same methodology was employed in the synthesis of
enantiomerically enriched substrates 37c and 38c. Thus, the
starting racemic alcohol (Æ)-35 was resolved into enantiom-
ers by Todaꢀs method,[47] to obtain (R)-35 (>95% ee) and
(S)-35 (>95% ee), whose conversion into the cis-deuteriat-
ed enantiomeric acetates (R)-37b and (S)-37b was carried
out in analogy to the racemic counterpart (Scheme 7).[48]
The PdII-catalyzed rearrangement of the latter enantiomers
proved to occur stereospecifically,[46] affording the terminal
acetates (R)-38b (ꢂ95% ee) and (S)-38b (ꢂ95%). The
high enantiomeric excess of these latter samples was con-
firmed by 1H NMR analysis of the corresponding carba-
mates (39), prepared from alcohols (R)- and (S)-38a, which
in turn were obtained by methanolysis of the enantiomeric
acetates 38b. Alcohols (R)- and (S)-38a were then conver-
ted into the corresponding carbonates (R)- and (S)-38c.
With stereospecifically labeled substrates 37c and 38c in
hand, we then developed a robust methodology for the anal-
ysis of the products arising from 37c/38c in the Mo-cata-
lyzed allylic substitution, that is, 41, 42, and 43 (Scheme 8),
Scheme 7. a) D2O, K2CO3, RT, 1 h; b) i) DIBAL-H, CH2Cl2, RT, 10 min;
ii) [Cp2Zr(H)Cl], 08C, 10 min; c) Ac2O, DMAP, Et3N, CH2Cl2, 258C, 3h;
d) ClCO2Me, pyr, CH2Cl2, reflux, 18 h; e) 2.5 mol% [(PhCN)2PdCl2],
CHCl3, 258C, 3.5 h; f) K2CO3, MeOH, 258C, 6 h; g) (R)-(+)-PhC(Me)N-
CO, DMAP, toluene, reflux overnight; h) NaBD4, MeOH, 08C, 10 min;
i) (EtO)2P(O)Cl, pyr, CH2Cl2, RT, 2 h.
tion of the achiral linear substrate 4, we deployed enantiom-
ers of the linear methyl carbonate 38c in which the deuteri-
um label at the allylic methylene introduces a stereocenter.
Synthesis of the isotopically labeled substrates was carried
out as follows (Scheme 7). In the racemic series, the label
was introduced in the first step by base-catalyzed H/D ex-
change of terminal acetylenic proton of 35, which afforded
the labeled propargyl alcohol 36 (69%; 98% 2H by
1H NMR). Hydroxyl deprotonation of the latter product
with DIBAL-H, followed by hydrozirconation of the result-
ing propargylic alkoxide with Schwartzꢀs reagent,[45] afforded
the deuterio alcohol 37a (79%; 98% [2H1] by 1H NMR)
with >99% Z stereoselectivity as revealed by 1H NMR
spectroscopy [the level of the nondeuterated material was
greater than the level of the (E)-deuterated product]. The
latter alcohol was converted into the corresponding acetate
37b (72%) and carbonate 37c (67%). Palladium(ii)-cata-
lyzed rearrangement[46] of allylic acetate 37b afforded the
linear isomer 38b (71%), in which no Z isomer was detect-
Scheme 8. E = CO2Me.
which form a six-component mixture. The analysis was fa-
cilitated by use of a chiral liquid crystal matrix (CLCM)
consisting of
a solution of polybenzyl-l-glutamate in
CH2Cl2[49] in combination with chiral HPLC.
The anisotropy induced through partial ordering in the
liquid crystal matrix causes quadrupolar coupling (DjnQj)
to be manifested in the H{1H} NMR spectrum. By using a
2
chiral matrix of the appropriate concentration and viscosity,
the six components can, in principle, be resolved. In the
event, the enantiomers of 41 and of 43 were both reasonably
well resolved with average DDjnQj of 20 and 33 Hz, respec-
tively. Although the technique did not resolve the enantiom-
ers of 42, the quadrupolar splitting of (Æ)-42 was found to
be much smaller (DjnQj=49 Hz) than that of 41 (DjnQj
ca. 525 and 544 Hz) and of 43 (DjnQj of ca. 693and
726 Hz) and thus in a clear window of the 2H{1H}} NMR
spectrum. By knowledge of the global enantiomeric excess
1
ed by H NMR spectroscopy. The latter acetate 38b under-
went methanolysis to afford alcohol 38a (70%) that was
converted into the corresponding carbonate 38c (63%) and
phosphate 38d (26%). Alternatively, racemic alcohol 38a
6916
www.chemeurj.org
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 6910 – 6929

Asymmetric Allylic Substitution
FULL PAPER
(eeg) of branched products (41 and 42), as measured by
chiral HPLC, one may then deduce the enantiomeric ratio
of (S)-42 and (R)-42. Assignments of each of the five re-
solved components [that is, (R)-41, (S)-41, 42, (R)-43 and
(S)-43] were made on the basis of reference mixtures
(Figure 2). Although the quadrupolar couplings did vary be-
tween experiments (Æ4%) it was found that the changes
were proportional across the five resolvable components
and therefore spectra can be normalized to a basis set (see
Experimental Section for full details). Thus, regioselective
>90% ee samples of branched isomers (R)-41 and (R)-42,
together with some linear isomer (Æ)-43 (Figure 2, spectrum
c). A regioselective, Pd-catalyzed alkylation of (Æ)-38b af-
forded racemic linear (Æ)-43 (Figure 2, spectrum d) and the
analogous stereospecific Pd-catalyzed reaction of (S)-38b
(95% ee) furnished enantiomerically enriched linear product
(S)-43 (95% ee) (Figure 2, spectrum e).
Having established an assay (2H{1H} NMR in CLCM/
chiral HPLC) for all six components of the anticipated prod-
uct mixture, we tested the individual Mo-catalyzed reactions
2
W-catalyzed alkylation ([W(CO)
A
of the branched and linear H-labeled carbonates (R)-37c,
branched carbonate (Æ)-37c gave cis-(Æ)-41 with essentially
no trace of trans-(Æ)-42 (Figure 2, spectrum a). An analo-
gous reaction, employing linear carbonate (Æ)-38c, pre-
pared via simple reduction of cinnamaldehyde (40,
Scheme 7), produced a mixture of (Æ)-41 and (Æ)-42, there-
by distinguishing 41 (enantiomers resolved) from 42
(Figure 2, spectrum b). To assign 41, an asymmetric, W-cata-
(S)-37c, (R)-38c and (S)-38c (all ca. 95% ee or greater)
with NaCHE2. The ligands (R)-(ꢀ)-12a and (S)-(+)-12c
gave complementary results in terms of stereochemical out-
come and “memory effect” (see below). Consequently, the
discussion below is restricted to the case of (S)-(+)-12c
(VALDY) which, being the more enantioselective ligand,
was explored in greater detail. The experiments were con-
lyzed alkylation ([W(CO)
A
ducted using [Mo(CO)
3(h6-C7H8)] as the molybdenum
A
line)[6] of phosphate (Æ)-38d was performed, which gave
source and, in contrast to the reactions reported in Table 1,
the procatalyst solution [(S)-(+)-12c + the Mo source] was
not heated before addition of substrate and NaCHE2 in
order to avoid decomposition.[50] After complete consump-
tion of the substrate, followed by work-up and purification,
the alkylation product mixture (41/42/43) was analyzed by
2
chiral HPLC and H{1H}} NMR using the CLCM method.
The resulting spectra a–d are given in Figure 3.
Considering first the product mixtures obtained from the
linear substrate 38c (see spectra a and b in Figure 3), the
minor product in both cases is, as expected, the linear
isomer 43 (Scheme 9), with the branched products 41/42 ob-
Figure 2. 2H{1H} NMR spectra (61.4 MHz) of reference samples of mono-
deuterated allylic alkylation products 41, 42, and 43 in a chiral liquid
crystal matrix (CLCM) consisting of a ca. 5% w/w solution of poly-g-
*
*
benzyl-l-glutamate in CH2Cl2 at 238C (*: CDHCl2, : CDCl3, : (S)-43,
&: (R)-43). Differential partial ordering effects results in differential
quadrupolar splittings (DjnQj) see experimental section for full details.
Spectrum a, racemic cis-deuterated branched isomer [(Æ)-41]; spectrum
b, racemic 1:1 mixture of cis-[(Æ)-41] and trans-[(Æ)-42] deuterated
branched isomers; spectrum c, enantiomerically enriched (>90% ee)
samples of a 1:1 mixture of cis-[(R)-41] and trans-[(R)-42] deuterated
branched isomers, together with a small amount of the linear isomer (Æ)-
43. Spectrum d, racemic sample of linear isomer [(Æ)-43]; spectrum e,
enantiomerically enriched (>90% ee) sample of linear isomers (S)-43.
Figure 3. 2H{1H} NMR spectra (61.4 MHz) in
a chiral liquid crystal
matrix (CLCM) of samples of monodeuterated allylic alkylation products
41, 42, and 43 obtained from “Mo-(S)-12c”-catalyzed allylic alkylation of
enantiomerically enriched (>95% ee) samples of (S)-38c (spectrum a);
(R)-38c (spectrum b); (R)-37c (spectrum c); and (S)-37c (spectrum d).
For product assignments and conditions, see Figure 2.
Chem. Eur. J. 2006, 12, 6910 – 6929
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6917

ˇ
´
P. Koc ovsky, A. V. Malkov, G. C. Lloyd-Jones et al.
with concomitant allylic trans-
position, or in other words, with
overall net retention of allylic
stereochemistry (Scheme 9).
The
reactions
of
the
branched substrates also pro-
ceeded, predominantly, by ste-
reospecific processes. Z-Config-
ured (R)-37c gives 41/42 in low
global eeg [74% (R), HPLC) to-
gether with ca. 8% of the linear
isomer (S)-43, with no (R)-43
detected (Figure 3, spectrum c).
The stereochemistry of the gen-
eration of the linear isomer (S)-
43 from branched (R)-37c thus
mirrors the reaction of linear
(S)-38c in that the nucleophile
is delivered syn to the leaving
group, in this case with concom-
itant allylic transposition (over-
all “net retention”), as depicted
Scheme 9. Outcome from reactions of (S)-38c; (R)-38c; (R)-37c and (S)-37c with NaCHE2 catalyzed by Mo/
2
(S)-(+)-12c (see text for full details) according to analysis by chiral HPLC and H{1H} NMR (61.4 MHz) spec-
troscopy in a chiral liquid crystal matrix (CLCM; see Figure 3). Enantiomeric excesses of 41 and 43 are de-
rived directly from NMR analysis. Enantiomeric excesses of 42 are deduced from the ratio (NMR)/global ee
(HPLC) of the branched isomers 41/42.
tained in 90–93% global enantioselectivity (eeg, established
by HPLC). However, it is immediately evident that the reac-
tions are stereospecific since different sets of products are
obtained from (R)-38c versus (S)-38c (spectra b and a, re-
spectively). Using the reference spectra (Figure 2, d and e)
it can be deduced that the reactions giving the linear prod-
ucts proceed with net retention at the allylic carbon. Thus
(S)-38c gives (S)-43 (Figure 3,a) and (R)-38c gives (R)-43
(Figure 3, b). Whether the reactions proceed with complete
stereofidelity is hard to establish as the low proportion of
the linear isomer in the mixtures (7Æ1%) precludes detec-
tion of 43 generated with net inversion of stereochemistry
unless it contributes more than about 15% of the linear
product and thus greater than about 1% of the overall prod-
uct mixture. The major product in both cases is the
branched isomer (41/42). Again, a stereospecific outcome is
evident as enantiomeric substrates give enantiodivergent
double bond geometry in the major component. Thus, (S)-
38c gives the E-configured isotopomer (R)-42 as the major
product and in >95% ee, (Figure 3, a) with a trace of Z-
configured (R)-41 (the ratio 41/42 is ca. 1:22). It should be
noted that the configuration and low ee value [40% (R)] of
the minor product 41 arises from the non-enantiopure
nature of the substrate [ca. 3% of (R)-38c]. With the oppo-
site enantiomer of substrate (R)-38c (Figure 3, b) the
double bond geometry of the major product is reversed and
Z-configured (R)-41 (>95% ee) is obtained together with a
small proportion of E-configured (S)-42 (>95% ee). As
noted above, the configuration and ee of the minor product
(41/42 is ca. 14:1) is dependent on the enantiopurity of the
substrate ((R)-38c), which in this case was higher (<1%
(S)-38c) than its enantiomer (S)-38c. These outcomes dem-
onstrate that the major (branched) isomer of product from
the reaction of the linear substrate (compare 4) is one in
which the nucleophile is delivered syn to the leaving group
in Scheme 9. The major branched product isomer is E-con-
figured 42 [87% ee (R)] where the double bond geometry
has been transposed. The minor branched isomer is Z-con-
figured 41 (ratio 41:42 ca. 1:11) and is rich in the S enan-
tiomer (40% ee) and thus of opposite configuration, lower-
ing the global ee to 74%. The low enantio- and regioselec-
tivity of the reaction confirms that the slower reacting (R)-5
(see above) is indeed the mismatched enantiomer.
The reaction of enantiomeric (S)-37c gives the highest se-
lectivity of all four substrates, with the branched products
41/42 obtained in 95% eeg [(R), HPLC; Figure 3, spectrum
d). The linear product 43 represents only 1.4% of the mix-
ture and is of the opposite configuration [(R)-43] to that ob-
tained from mismatched (R)-37c. The dominant species
from the reaction is the Z-configured (R)-41 generated in
>95% ee (Scheme 9), with only a trace (1.3%) of E-config-
ured (S)-42 in which double bond geometry has been trans-
posed.
From the above analysis it is clear that the dominant ste-
reochemical pathways associated with the reaction of both
of the enantiomers of the branched substrate 5 as well as
the linear substrate 4 are related and are stereospecific. The
reaction of the matched branched substrate (S)-5 [(S)-37c]
proceeds with net retention whilst the mismatched branched
substrate (R)-5 [(R)-37c] proceeds with apparent inversion
(Scheme 9). The deuterium labeling reveals that both pro-
cesses proceed with overall net retention, which involves an
isomerization process at the Mo–allyl stage, resulting in
double bond geometry being transposed in the major prod-
uct from the mismatched substrate. The memory effect
arises from this isomerization being incomplete and this
then reduces the net global enantiomeric excess. Also of
note is that the deployment of (R)-37c reveals that the reac-
tion of the mismatched substrate (R)-5 does not proceed
with complete overall net retention since 8.5% of the
6918
www.chemeurj.org
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 6910 – 6929

Asymmetric Allylic Substitution
FULL PAPER
branched product arises from net inversion, a significantly
higher level than the minor enantiomer (S)-37c present in
the 95% ee substrate (i.e., ca. 2.5%). Experiments in which
the unreacted substrate (R)-37c was recovered and analyzed
prior to complete conversion indicate no cis–trans scram-
2
bling of the H label. Furthermore, the ee versus c relation-
ship observed in the kinetic resolution of (Æ)-5 (see above)
correlates well with the theoretical curve generated by s =
1.8
=
ln
G
U
U
G
there is little or no inversion of the mismatched substrate
during reaction. Control experiments confirmed that neither
4 nor 5 react with NaCHE2 in THF at 608C over a period of
18 h. To test whether a noncomplexed pre-catalyst, derived
from [Mo(CO)
3(h6-C7H8)], might be responsible for a slow
background reaction, we tested the reaction of 4 and of 5
with NaCHE2 (E=CO2Me; 2 equiv) in the presence of
10 mol% of [Mo(CO)
3(h6-C7H8)] in THF at 608C over a
G
period of 24 h. Only the linear substrate 4 reacted (24%
conversion) and gave exclusively the linear product 7. It
therefore seems likely that with the mismatched substrate, a
small proportion (ca. 5%) of the reaction proceeds via a net
inversion process, analogous to stoichiometric examples,[5]
and potentially arising from Mo–Mo transfer with inversion
as a side reaction.
Having delineated the net stereochemical outcome over
the complete reaction for the matched and the mismatched
manifolds, we then returned to using racemic substrate so
that we could study the reaction of interest, that is, how the
Mo catalyst “Mo-(S)-12c” processes racemic (Æ)-5 to give
(R)-6 in high enantioselectivity. By again employing the ster-
eospecific labeling strategy, but using racemic Z-configured
(Æ)-37c, we were able to track matched and mismatched
manifolds simultaneously. The reaction was sampled at vari-
ous intervals and the conversion and ee of the remaining
37c was determined by chiral GC. This information then al-
lowed the conversions of matched (S)-37c and mismatched
(R)-37c to be deduced. A simple analysis of the alkene
Figure 4. Graph depicting the evolution of a “Mo-(S)-12c”-catalyzed al-
lylic alkylation of a racemic sample of cis monodeuterated branched al-
lylic carbonate 37c with conversion (by GC analysis with internal stand-
ard; x axis). The various mol fractions of the branched product isomers
[(R)-41, (S)-41, and (R)-42] together with their global enantiomeric
excess (eeg) are plotted on the y axis. The quantities of (S)-42 and linear
43, generated in the reaction, are negligible. Ratios of (R)-41, (S)-41, and
(R)-42 are deduced from the ee of and conversion of substrate (37c, by
GC) and the 1H NMR spectrum of the product mixture (using reactions
of the single enantiomers of substrate, Figure 3, as reference).
1
region of the H NMR spectrum of the crude product mix-
sion, the global enantioselectivity for branched products (41/
42) is eeg = 57% (R); in stark contrast, from 46–100% con-
version the nascent enantioselectivity is about 96%, with
global enantioselectivity at 100% conversion of eeg = 80%
(R). The magnitude of the memory effect in the mismatched
manifold is controlled by the rate of equilibration of the
Mo-intermediates relative to their rate of attack by nucleo-
phile. During the reaction, factors which increase the former
or decrease the latter will decrease memory and facilitate
greater global enantiomeric excess. Clearly, the nucleophile
concentration will drop during reaction. However, the use
of two equivalents of NaCHE2 suggests that the cessation of
the memory effect is not related to this factor (at 30% con-
version [Nu]t/[Nu]0 is 0.85). A co-product of the reaction is
MeONa. This is known to be required to deprotonate the
amido functionality, thereby engendering a selective and
active molybdate type intermediate,[15b] and would not be
present in the first stages of reaction. However, methoxide
generated in excess of the catalyst stoichiometry should be
protonated by the malonate C-H group of the nascent prod-
2
ture yields the E/Z ratio of the H label in branched prod-
ucts, and thus the ratio of E-configured 42 to Z-configured
41. Since the reaction of matched (S)-37c gives, over the
whole course of reaction, essentially a single branched iso-
topomer [ca. 98.7% (R)-41], the ratio of (S)-41/(R)-42 aris-
ing from the mismatched substrate (R)-37c is readily de-
rived. A plot of conversion against mol fraction product
(Figure 4) reveals that in the initial phases of reaction
(<30% conversion) the mismatched substrate (R)-37c is
converted predominantly into Z-configured (S)-41 (and
matched substrate into (R)-41) and hence reactions pro-
ceeds with a low global eeg which is slightly augmented by
the kinetic resolution (s = ca. 2) favoring the matched sub-
strate and generating (R)-41.
However, after about 30% conversion, the generation of
Z-configured (S)-41 essentially ceased and the memory
effect in the mismatched manifold thus disappeared. The
enantiomeric excess of the nascent product from this point
on is remarkably high. For example, from 0–46% conver-
Chem. Eur. J. 2006, 12, 6910 – 6929
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6919

ˇ
´
P. Koc ovsky, A. V. Malkov, G. C. Lloyd-Jones et al.
ucts (6/7, or in this case 41/42/43). Furthermore, in control
experiments, deliberate addition of catalytic quantities of
MeONa at the start of reaction did not result in increased
equilibration and greater global enantiomeric excess.
least one of the amido groups must be secondary such that
deprotonation to generate an anionic ligand, and thus mo-
lybdate species, is possible. All of the above suggests that
the ligands that are highly effective, such as VALDY 12c,
coordinate in a tridentate mode. This is fully consistent with
the X-ray crystal structures and NMR-derived solution
structures for Mo complexes bearing ligands 8[15] and 11,[16]
which are in many ways analogous to those described
herein. The detection of a nonlinear effect (Figure 1) indi-
cates that two, or more, ligands may be accommodated at
the Mo center. However, in view of the intermediacy of al-
lylic complexes and the tridentate ligand requirements, com-
plexes bearing two or more ligands are likely to be in equili-
brium with the catalytic cycle rather than on it and may thus
act as reservoirs.
Compared with analogous transformations involving Pd,
where there has been extensive mechanistic investigation
and the processes by which stereochemical convergence of
chiral but racemic substrates are known in intimate detail,
very little is known about asymmetric Mo-catalyzed allyla-
tions. The design and deployment of a stereospecific isotopic
labeling technique, in which racemic and enantiomerically
enriched substrates 37 and 38 are reacted and then the prod-
ucts 41, 42, and 43 analyzed by 2H{1H} NMR in a chiral
liquid crystal matrix (CLCM), has allowed three important
mechanistic features to be deduced. Firstly, the reactions are
stereospecific and proceed with net retention of stereochem-
istry of the linear and both enantiomers of the branched
substrates. These results demonstrate that the mismatched
and matched branched substrates do not undergo enantiodi-
vergent reactions on generation of the Mo–allyl intermedi-
ate and that there is no significant transfer of the allyl group
(inv) between Mo centers, as exemplified for a retention
based mechanism in Scheme 10.
Krska et al.[15b] have identified that a CO source, such as
[Mo(CO)6], is an essential component of the catalytic
milieu. Indeed, the isolated Mo allyl complex {[(8)Mo(h3-
PhC3H4)(CO)2], where 8 is deprotonated at N} only reacts
with stoichiometric NaCHE2 in the presence of 2 equiv CO
(provided as either [Mo(CO)6] or as an atmosphere of
CO(g)) and this process generates
6 (95% ee) and
[(8)Mo(CO)4]Na. The latter complex was demonstrated to
be an active carrier for catalysis. In the present system
(Figure 4) the CO source must be co-generated during the
formation of the active catalyst from the substrate, ligand,
and Mo precursor. As the reaction is conducted at 608C, the
CO sources required for turnover, for example, CO(g) or
[Mo(CO)6], will be slowly purged from reaction; the former
by out-gassing, the latter by precipitation or sublimation.
Reduced levels of CO source would lead to slower turnover
with decreased rate of nucleophilic attack. Such a phenom-
enon would facilitate greater equilibration and thus reduced
memory effect.
Conclusion
C1-symmetric bispicolinamide ligands that bear a single ster-
eogenic center in the linking 1,2-diaminoethane framework
between two amide groups can, with the correct choice of
substituent, facilitate Mo-catalyzed allylic alkylation of
branched and linear cinnamyl-type substrates (cf 4 and 5)
with very high asymmetric induction and regioselectivity for
the branched isomer of product (cf. 6 and 7). Such ligands
are readily derived from a-amino acids, for example, valinol
(cf. 12c), which are often available in both enantiomeric
forms, thereby allowing access to either enantiomer of prod-
ucts of type 6, where the Ar ring can be varied. Of the li-
gands studied, where R = Ph, Bn, iPr, and tBu, the valinol-
derived ligand VALDY 12c (R = iPr) emerged as the most
selective, while the closely related tert-leucine-derived
ligand 12d (R = tBu) proved significantly poorer. This out-
come mirrors the relationship between the amino-acid pre-
cursor and the activity/enantio- and regioselectivity ob-
served in the analogous reactions catalyzed by W complexes
bearing diphenylphosphinoaryl oxazoline ligands, where an
iPr substituent was found to be optimum and a tBu substitu-
ent to give rather poor results.[6]
Scheme 10.
A systematic study of the structural components of li-
gands of type 12 demonstrates that only one of the amide
groups need be picolinic in nature, the other can be a simple
amide such as benzamide (cf. 13, 14). However, in such
cases the stereogenic center is best located distal to the pico-
linamide on the diaminoethane linker and catalytic activity
is reduced somewhat. Ligands not possessing two amide
groups are less active and only poorly selective (cf. 16, 17).
As has also been reported by Trost and Hughes et al.,[15a] at
Secondly, the reactions of the linear and both enantiomers
of the branched substrates [4 and (Æ)-5] all generate the
same dominant regioisomer and enantiomer of product
(Scheme 11), therefore the mechanism facilitates stereo-
chemical and regiochemical convergence through equilibra-
tion of intermediates, for example, of type 34. However, in-
complete equilibration is evident (Figure 4), which then
leads to a memory effect. This memory effect reduces enan-
6920
www.chemeurj.org
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 6910 – 6929

Asymmetric Allylic Substitution
FULL PAPER
clusion that an inv–inv sequence[32] would lead to the wrong
sense of asymmetric induction. The alternative ret–ret mech-
anism, in which reductive elimination after attack of the nu-
cleophile at the Mo center, or syn-attack of a “s enyl” inter-
mediate at the benzylic carbon, can be envisaged and would
be consistent with our earlier demonstration of a ret–ret
mechanism in a related Mo-catalyzed allylation reaction.[4]
Although we have been unable to isolate intermediates
from the present system, by employing the labeling techni-
ques we have described, it should now be possible to find a
suitable system whereby intermediates can be observed and
identified and the ret–ret versus inv–inv issue addressed on
the basis of NMR analysis.[32]
Moreover, a key outcome of the study reported herein, is
that irrespective of whether the anti/inv–inv sequence or the
syn/ret–ret sequence is operative, we have confirmed that di-
asteroisomeric intermediates are able to interconvert via a
p–s–p mechanism, involving diastereofacial exchange of the
Ph–allyl moiety. Without this process, asymmetric catalysis
of the allylic alkylation of cinnamyl substrates by “Mo(12)”
would be limited to the linear substrates 4, with kinetic reso-
lution the only opportunity for asymmetric induction with
the branched substrates 5. This latter situation is exactly
what was observed with the analogous reactions catalyzed
by W complexes,[3h,7] which are thus much more limited in
scope.
Scheme 11.
tio- and regioselectivity in the manifold arising from mis-
matched branched substrate (R)-5, but is attenuated as reac-
tion proceeds. The same manifold is also partially accessed
by the linear substrate through imperfect enantiofacial selec-
tivity on generation of the Mo–allyl intermediates (see
dashed lines in Scheme 11). Thirdly, the involvement of the
terminal allylic alkene carbon in the reaction of the
branched substrates confirms that allyl Mo intermediates, in
which the C3-allylic carbon is Mo-bound, must be involved
in the generation of branched product from branched sub-
strate. It is tempting to suggest that the regioselectivity
arises from attack of an h1-Mo allyl species, which is bound
through the C3-allylic carbon (cf. 33 in Schemes 6 and 11).
The asymmetric induction would arise then through diaster-
eo-facial selectivity in an SN2’-like reaction. However, if the
allylic moiety were bound trans to the anionic amido moiety
of the ligand, as would be consistent with the data of Krska
and Hughes et al.[15] (see C), then there should be no
memory effect since all precursors should lead to the same
intermediate 33, in which the allyl unit itself is achiral.[51] It
therefore appears more likely that h3-Mo allyl intermediates
are involved. Of course, the equilibration of such intermedi-
ates may well involve h1-bound Mo species, as outlined in
Scheme 11, and the regioselectivity of the attack of the nu-
cleophile on h3 intermediates will undoubtedly be influenced
by the complex electronic and steric parameters associated
with pseudo-octahedral geometry, as well as the possibility
of distorted h3-binding modes.[42,52]
Experimental Section
General methods: Melting points were determined on a Kofler block and
are uncorrected. The NMR spectra were recorded in CDCl3, 1H at 250,
270 or 400 MHz and 13C at 62.9 or 100.6 MHz with CDCl3 (d 7.26, 1H; d
77.0, 13C) as internal standard; 2D techniques were used to establish the
structures and to assign the signals. 2H NMR measurements were per-
formed on a 400 MHz (1H) spectrometer equipped with a selective 5 mm
deuterium probe operating at 60 MHz. Conventional 2H NMR spectra
were run in CH2Cl2, using CDCl3 (ca. 1%) as internal reference.
2H NMR (60 MHz, CH2Cl2, 228C): d (CDHCl2)=5.32 ppm, d (CDCl3)=
2
7.30 ppm. H NMR spectra run in a chiral liquid crystal matrix were cali-
brated against the natural abundance CDHCl2 doublet centered at d
5.32 ppm. The deuterium content was determined by 1H NMR and con-
firmed by 13C NMR. The IR spectra were recorded for a solution in
chloroform between NaCl plates unless otherwise stated. The EI and/or
CI mass spectra were measured on a dual sector mass spectrometer using
direct inlet and the lowest temperature enabling evaporation. The GC-
MS analysis was performed with RSL-150 column (25 m”0.25 mm). The
X-ray data were collected at 183K on an Siemens Smart CCD diffrac-
tometer equipped with LT-2 low-temperature device and using MoKa ra-
diation (l=0.71069 Š, graphite monochromator). Full sphere of recipro-
cal sphere was scanned by 0.38 steps in w with a crystal-to-detector dis-
tance of 3.97 cm. Data were processed using SMART and SAINT soft-
ware (Siemens AXS, Madison, Wisconsin, 1995) and empirically correct-
ed for absorption and other effects using SADABS [G. M. Sheldrick,
University of Gçttingen (Germany), 1996]. The structure was solved by
direct methods and refined by full-matrix least-square technique using
program suite SHELXTL [SHELXTL, version 5.10, Bruker AXS Inc.,
Madison, Wisconsin, 1997]. Allylic substitution reactions were performed
under an atmosphere of dry argon in oven-dried glassware at least twice
evacuated and filled with argon. Solvents and solutions were transferred
by syringe–septum and cannula techniques. All solvents for the reactions
were of reagent grade and were dried and distilled immediately before
The net retention of stereochemistry observed in the
linear/matched branched manifolds and the apparent inver-
sion in the mismatched manifold, is indicative that either an
anti/inv–inv sequence or a syn/ret–ret sequence is exclusively
operative across all manifolds. There are clear analogies
with reactions catalyzed by ligand 8, where crystallographic
and NMR data of [8·Mo(h3-PhC3H4)(CO)2] lead to the con-
Chem. Eur. J. 2006, 12, 6910 – 6929
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6921

ˇ
´
P. Koc ovsky, A. V. Malkov, G. C. Lloyd-Jones et al.
use as follows: diethyl ether from lithium aluminum hydride, tetrahydro-
furan (THF) from sodium/benzophenone; dichloromethane from calcium
hydride or alternatively were obtained freshly from an Anhydrous Tech-
nologies drying train. Where appropriate, tetrahydrofuran and dichloro-
methane were de-gassed (freeze–thaw cycles) and then saturated with ni-
trogen prior to use. All reagents were purchased at highest commercial
quality and used without further purification, unless otherwise stated.
[(EtCN)3Mo(CO)3] and [(h6-C7H8)Mo(CO)3] were prepared according to
CCDC-291520 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.
(S)-(+)-1,2-Bis(2-pyridinylcarboxamido)-3-phenylpropane
[(S)-(+)-
(12b)]: (S)-Phenylalanine amide (S)-23b was obtained by aminolysis of
hydrochloride of (R)-(ꢀ)-phenylalanine methyl ester (S)-22b (2.00 g,
74%). 1H NMR: d = 1.40 (brs, 2H, 2-NH2), 2.73(dd, J=13.7, 9.6 Hz,
1H, 3-CHH), 3.28 (dd, J=13.7, 4.1 Hz, 1H, 3-CHH), 3.62 (dd, J=9.6,
4.1 Hz, 1H, 2-CH), 5.65 (brs, 2H, 1-NH2), 7.28 (m, 5H, Ph); MS (ES):
the literature procedures;[24] additionally
a
sample of [(h6-
m/z: 187 [M+Na]+, 165 [M+H]+, in agreement with literature data.[17]
A
C7H8)Mo(CO)3] was purchased from Strem Chemical Co. D2O (>99.5%
2H) was purchased from Cambridge isotope laboratories. NaB[D4] was
purchased from Sigma-Aldrich; PBLG (polybenzyl-l-glutamate) (DP=
1m solution of borane·THF complex in THF (50 mL) was added drop-
wise to a suspension of the latter amide (S)-23b (1.72 g, 10.5 mmol) in
THF (25 mL) at 108C. The mixture was stirred for 1 h at room tempera-
ture and than heated at reflux (708C) for 5 h. The solution was cooled
with ice, methanol (15 mL) was then added, and the mixture was stirred
overnight. The solvent was removed in vacuo; 6m HCl (150 mL) was
added, the mixture was heated at reflux (1108C) for 4 h, and then evapo-
rated to dryness in vacuo. The residue was extracted with methanol (2”
50 mL); the methanolic solution was evaporated, the residue was treated
with 1m aqueous KOH (150 mL), and then extracted with CH2Cl2 (4”
100 mL). The combined organic extracts were dried over K2CO3 and the
solvent was evaporated to give the corresponding diamine (S)-24b
(1.20 g, 76%) as a yellow oil. [a]2D0 = +8.2 (c = 2.0, CHCl3); 1H NMR:
d = 1.14 (brs, 4H, 1-NH2 + 2-NH2), 2.36 (m, 2H, 3-CH2), 2.64 (m, 2H,
1-CH2), 2.79 (m, 1H, 2-CH), 7.15 (m, 5H, Ph); MS (ES): m/z: 151
[M+H]+, in agreement with the literature data.[54] A solution of the
latter diamine (S)-24b (1.20 g, 8 mmol) in pyridine (12 mL) was added to
a solution of a-picolinic acid (1.97 g, 16 mmol) and (PhO)3P (4.96 g,
16 mmol) in pyridine (20 mL) at 808C and the mixture was heated at
1008C overnight. The cooled solution was carefully poured into water
(30 mL) and the resulting mixture was extracted with CH2Cl2 (2”40 mL).
After the organic solvent was dried over MgSO4 and removal of solvent
in vacuo, the crude product was purified by chromatography on a silica
gel column (15”2.5 cm) with ethyl acetate as an eluent to afford (S)-(+)-
12b (1.59 g, 55%) as pale yellow solid. M.p. 117–1188C; [a]2D0 = +1.3( c
= 2.1, CHCl3); 1H NMR: d = 2.99 (dd, J=14.0, 7.1 Hz, 1H, 3-CHH),
3.11 (dd, J=14.0, 6.7 Hz, 1H, 3-CHH), 3.73 (m, 2H, 1-CH2), 4.61 (m,
1H, 2-CH), 7.20–7.38 (m, 7H, Ph + Py + Py’), 7.77 (tt, J=7.8, 1.6 Hz,
2H, Py + Py’), 8.14 (dd, J=7.4, 6.7 Hz, 2H, Py +Py’), 8.40, 8.42 (2”brs,
2”1H, 2”NH), 8.50 (m, 2H, Py + Py’); 13C NMR: d = 39.2 (CH2), 43.0
(CH2), 51.8 (CH), 122.5, 122.6 (Py, Py’, 2”CH), 126.5 (Py, Py’, 2”CH),
127.0 (Ph, CH), 129.0 (Ph, 2”CH), 129.7 (Ph, 2”CH), 137.6, 137.6 (Py,
Py’, 2”CH), 137.9 (Ph, C), 148.5, 148.6 (Py, Py’, 2”CH), 150.06, 150.09
(Py, Py’, 2”C), 165.0, 165.5 (2”CO); IR: n˜ = 3375m, 2930m, 1663s,
1595w, 1575w, 1465m, 1430 cmꢀ1 m; MS (ES): m/z: 383 [M+Na]+, 3 61
[M+H]+; HRMS (FAB): m/z: calcd for C21H21N4O2: 361.16645; found:
361.16653.
564) was purchased from Sigma. [W(CO)
3(h6-C7H8)] was synthesized ac-
E
cording to a modified literature procedure.[53] All allylic carbonates are
known compounds[3c] and were prepared by stirring the corresponding al-
lylic alcohols with methyl chloroformate in pyridine followed by a stand-
ard workup. Yields are given for isolated product showing one spot on a
TLC plate and no impurities detectable in the NMR spectrum. The iden-
tity of the products prepared by different methods was checked by com-
parison of their NMR, IR, and MS data and by the TLC behavior. Enan-
tiopurity of the products was determined by HPLC on Diacel Chiralpak
AD, Chiracel OJ, or Chiralcel OD-H using a hexane/2-propanol mixture
as an eluent, or by 1H NMR with [D]-Eu
(hfc)3 (for the ratios, see the in-
N
dividual experiments). Chiral GC was carried out with capillary columns
FS-HYDRODEX b-3P.
(1R)-(ꢀ)-1,2-Bis[(2-pyridinylcarboxamido]-1-phenylethane
[(R)-(ꢀ)-
(12a)]: Phenylglycine amide (R)-(ꢀ)-23a was obtained by aminolysis of
hydrochloride of (R)-(ꢀ)-phenylglycine methyl ester 22a (2.30 g, 70%).
1H NMR (CD3OD): d = 4.70 (s, 1H, CH), 5.06 (s, 4H, 2”NH2), 7.53–
7.70 (m, 5H, Ph); MS (ES): m/z: 173[ M+Na]+, 151 [M+H]+, in agree-
ment with the literature data.[17] It was reduced with LiAlH4 in THF at
1
reflux for 18 h to afford diamine 24a (620 mg, 45%). H NMR: d = 1.37
(brs, 4H, 2”NH2), 2.81 (dd, J=12.6, 7.1 Hz, 1H, 2-CHH), 2.92 (dd, J=
12.6, 5.3Hz, 1H, 2-CH H), 3.89 (dd, J=7.1, 5.3 Hz, 1H, 1-CH), 7.23–7.34
(m, 5H, Ph); MS (ES): m/z: 13 7 M[ +H]+, in agreement with the litera-
ture data.[18]
A solution of diamine 24a (500 mg, 4.1 mmol) in pyridine (8 mL) was
added to a solution of a-picolinic acid (1.0 g, 8.13mmol) and (PhO) 3P
(2.52 g, 8.13mmol) in pyridine (20 mL) at 80 8C and the mixture was
heated at 1008C overnight. The cooled solution was carefully poured into
water (30 mL) and the resulting mixture was extracted with CH2Cl2 (2”
40 mL). After drying over MgSO4 and removal of the solvent, the crude
product was purified by chromatography on a silica gel column (15”
2.5 cm) with ethyl acetate as an eluent to give (R)-(ꢀ)-12a (880 mg,
62%) as a white solid. M.p. 166–1688C; [a]2D0 = ꢀ22.2 (c = 1.4, CHCl3);
1H NMR: d = 4.00 (m, 2H, CH2), 5.44 (dd, J=13.8, 7.8 Hz, 1H, CH),
7.26–7.49 (m, 7H, Ph + Py + Py’), 7.80 (tt, J=7.8, 1.5 Hz, 2H, Py +
Py’), 8.16 (t, J=7.1 Hz, 2H, Py + Py’), 8.38 (brt, J=4.5 Hz, 1H, NH),
8.50 (d, J=4.8 Hz, 1H, Py), 8.57 (d, J=4.6 Hz, 1H, Py’), 8.80 (brd, J=
8.0 Hz, 1H, NH); 13C NMR: d = 44.9 (CH2), 54.6 (CH), 122.7 (Py, Py’,
CH), 126.6 (Py, Py’, CH), 127.1 (Ph, 2”CH), 128.3(Ph, CH), 129.3(Ph,
2”CH), 137.6 (Py, Py’, CH), 140.0 (Ph, C), 148.5, 148.6 (Py, Py’, CH),
150.1 (Py, Py’, C), 164.9 (2”CO); IR (neat, KBr): n˜ = 3380m, 2990m,
1675s, 1590m, 1570m, 1505s, 1465m, 1430 cmꢀ1 m; MS (ES): m/z: 3 69
[M+Na]+, 347 [M+H]+; HRMS (FAB): m/z: calcd for C20H19N4O2:
347.15080; found: 347.15082.
(S)-(+)-1,2-Bis(2-pyridinylcarboxamido)-3-methylbutane [(S)-(+)-(12c)]:
The title compound was prepared in the same way as described above for
(R)-(ꢀ)-12a. (S)-Valinamide hydrochloride (S)-23c (2.50 g, 16.4 mmol)
was reduced with LiAlH4 (1.90 g, 50 mmol) in THF (50 mL) to give dia-
mine (S)-15c (1.05 g, 63%) as a yellow oil. [55] A solution of the latter dia-
mine (S)-24c (1.05 g, 10.3mmol) in pyridine (10 mL) was treated with a-
picolinic acid (2.58 g, 21.0 mmol), (PhO)3P (6.52 g, 21.0 mmol) in pyridine
(30 mL) and the reaction mixture was stirred at 1008C overnight to
afford (S)-(+)-12c as white crystals (2.70 g, 84%). M.p. 94–958C
(MeOH); [a]2D0 = +353.0 (c = 2.1, MeOH); 1H NMR (CD3OD): d =
1.24, 1.27 (2”d, J=7.0 Hz, 2”3H, 2”Me), 2.20 (m, 1H, 3-CH), 3.80 (dd,
J=13.5, 9.4 Hz, 1H, 3-CHH), 3.96 (dd, J=13.7, 4.1 Hz, 1H, 1-CHH),
4.40 (m, 1H, 2-CH), 7.65–7.75 (m, 2H, Py + Py’), 8.05–8.22 (m, 4H, Py
+ Py’), 8.72, 8.83(2”m, 2”1H, Py + Py’); 13C NMR d 19.1, 20.4 (2”
Me), 32.3 (3-CH), 42.7 (CH2), 55.8 (2-CH), 123.4, 123.6 (Py, Py’, 2”CH),
128.1, 128.2 (Py, Py’, 2”CH), 139.1, 139.2 (Py, Py’, 2”CH), 150.07, 151.1
(Py, Py’, 2”CH), 151.2 (Py, Py’, 2”C), 165.0 (2”CO); IR: n˜ = 3380m,
2960m, 1660s, 1592w, 1572w, 1460w, 1430 cmꢀ1 w; MS (ES): m/z: 3 3 5
[M+Na]+, 313 [M+H]+; HRMS (FAB): m/z: calcd for C17H21N4O2:
313.16645; found: 313.16649.
Crystal data for (R)-(ꢀ)-12a: C20H18N4O2, M=346.38; colorless crystals
were obtained from
a CH2Cl2 solution; orthorhombic, space group
P212121; a=5.8672(1), b=16.0079(1), c=18.134(1) Š, V=1703.14(3) Š3,
Z=4, 1calcd =1.351 gcmꢀ3, m=0.090 mmꢀ1. A total of 24002 reflections
were measured, 3539 of them unique (Rint =0.0495), with 2847 having I
> 2s(I). All 3539 reflections were used in the structure refinement based
on F2 by full-matrix least-squares techniques with hydrogen atoms calcu-
lated in theoretical positions, riding during refinement on the respective
pivot atom (253parameters). Final RF =0.040, Rw =0.095 on F2 for ob-
served data. The estimated error in bond lengths is 0.002 Š.
6922
www.chemeurj.org
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 6910 – 6929

Asymmetric Allylic Substitution
FULL PAPER
(S)-(+)-1,2-Bis(2-pyridinylcarboxamido)-3,3-dimethylbutane
[(S)-(+)-
1485m, 1460m, 1435 cmꢀ1 m; MS (FAB): m/z (%): 368 (8) [M+Na]+, 3 46
(56) [M+H]+, 307 (17), 289 (11), 211 (13), 154 (100), 136 (69), 105 (48),
102 (55), 89 (20), 77 (22); HRMS (FAB): m/z: calcd for C21H20N3O2:
346.1555; found: 346.1554.
(12d)]: N-Methylmorpholine (630 mL, 5.730 mmol, 4.1 equiv) was added
to a suspension of dihydrochloride of (S)-3,3-dimethyl-1,2-butanediamine
24d[21] (264 mg, 1.396 mmol, 1.0 equiv), a-picolinic acid (347 mg,
2.819 mmol, 2.0 equiv), and 1-hydroxybenzotriazole (380 mg, 2.812 mmol,
2.0 equiv) in dry CH2Cl2 (20 mL) under nitrogen. The mixture was
cooled to 08C and a solution of N-[3-(dimethylamino)propyl]-N’-ethylcar-
bodiimide hydrochloride (594 mg, 3.098 mmol, 2.2 equiv) in dry CH2Cl2
(20 mL) was added over a period of 3min. The mixture was stirred at
08C to room temperature for 16 h to obtain a clear solution. The solvent
was evaporated in vacuo; the residue was purified by flash chromatogra-
phy on a silica gel column (15”2.5 cm) with an ethyl acetate/methanol
96:4 to afford pure (S)-(+)-12d as an oil that slowly solidified (200 mg,
44%). [a]2D0 = +84 (c = 0.61, CH2Cl2); 1H NMR: d = 1.10 [s, 9H,
(CH3)3C], 3.60 (ddd, J=13.8, 10.8, 6.2 Hz, 1H, one H of CHCH2), 3.87
(ddd, J=13.8, 5.4, 3.3 Hz, 1H, one H of CHCH2), 4.22 (dt, J=10.6, 10.6,
3.2 Hz, 1H, CHCH2), 7.33 (ddd, J=7.6, 4.8, 1.2 Hz, 1H, arom.), 7.40
(ddd, J=7.6, 4.8, 1.2 Hz, 1H, arom.), 7.75 (dt, J=7.7, 7.7, 1.7 Hz, 1H,
arom.), 7.80 (dt, J=7.7, 7.7, 1.7 Hz, 1H, arom.), 8.08 (dt, J=7.8, 0.9,
0.9 Hz, 1H, arom.), 8.14 (dt, J=7.8, 0.9, 0.9 Hz, 1H, arom.), 8.21 (brd,
J=10.6 Hz, 1H, NH), 8.31 (m, 1H, NH), 8.47 (dq, J=4.8, 0.8, 0.8,
0.8 Hz, 1H, arom.), 8.56 (dq, J=4.8, 0.8, 0.8, 0.8 Hz, 1H, arom.);
13C NMR: d = 27.0 (CH3)3C, 34.8 (CH3)3C, 41.0 (CH2), 58.0 (CH), 122.4
(CH), 122.7 (CH), 126.3 (CH), 126.5 (CH), 137.4 (CH), 137.6 (CH),
148.5 (CH), 148.5 (CH), 150.0 (s, C), 150.2 (s, C), 165.4 (s, 2”C=O); IR:
(R)-(ꢀ)-1-Benzamido-2-(2-pyridinylcarboxamido)-1-phenylethane [(R)-
(ꢀ)-(14)]: A solution of benzoyl chloride (0.45 mL, 3.8 mmol) in THF
(3mL) was added dropwise at 0 8C to a solution of crude amine (R)-31
(640 mg, 2.65 mmol) in THF (20 mL) and triethylamine (560 mg,
5.5 mmol). The cooling bath was removed and the mixture was stirred at
room temperature for 2 h and then quenched with water (20 mL). The
mixture was extracted with CH2Cl2 (3”20 mL); the organic layer was
washed successively with water (20 mL), satd aq NaHCO3 (20 mL) and
again water (20 mL) and subsequently dried over MgSO4. The solvent
was removed in vacuo and the oily residue was purified by chromatogra-
phy on a silica gel column (15”2.5 cm) with ethyl acetate/methanol 96:4
to afford pure (R)-(ꢀ)-14 as white microcrystals (448 mg, 48%). M.p.
180–1828C (MeOH); [a]D20 = ꢀ40.3( c = 1.1, CHCl3); 1H NMR: d =
3.71 (ddd, J=14.3, 5.9, 3.5 Hz, 1H, CHH), 3.96 (ddd, J=14.3, 9.0, 7.6 Hz,
1H, CHH), 5.25 (m, 1H, CH), 7.18–7.42 (m, 9H, Ph + Ph’ + Py), 7.76–
7.83(m, 3H, Ph ’ + Py), 8.13–8.16 (m, 2H, Py + NH), 8.41 (brs, 1H,
NH), 8.45 (dm, J=4.7 Hz, 1H, Py); 13C NMR: d = 45.4 (CH2), 57.3
(CH), 122.7 (CH), 126.8 (CH), 127.0 (CH), 127.6 (CH), 128.0 (CH),
128.9 (CH), 129.2 (CH), 131.8 (CH), 134.1 (C), 137.8 (CH), 140.3 (C),
148.7 (CH), 149.4 (C), 167.2 (CO), 167.3(CO); IR: n˜ = 1664 cmꢀ1 s;
HRMS (FAB): m/z: calcd for C21H20N3O2: 346.1555; found: 346.1554; el-
emental analysis calcd (%) for C21H19N3O2: C 73.03, H 5.54, N 12.17;
found: C 72.84, H 5.49, N 11.68.
n˜
= 3376w, 3019vw, 3006m, 2967m, 1670s, 1592w, 1570m, 1528s,
1465m, 1434m, 1370w, 1291vw, 1240w, 1160vw, 1088vw, 1041vw, 998w,
908w, 818 cmꢀ1 vw; HRMS (EI): m/z: calcd for C18H22N4O2: 326.17423;
found: 326.17428.
(R)-(ꢀ)-2-(2-Pyridinylcarboxamido)-2-phenyl-(2-pyridinylmethyl)aceta-
mide [(R)-(ꢀ)-(15)]: A mixture of (R)-2-(pyridine-2-carbamido)-2-phe-
nylacetic acid methyl ester (R)-18 (350 mg, 1.3 mmol), 2-(aminomethyl)-
pyridine (700 mg, 6.5 mmol) and ammonium chloride (27 mg, 0.5 mmol)
was heated at 858C for 2 h. The mixture was cooled to room temperature
and treated with CH2Cl2 (10 mL) to dissolve the solid; the solution was
washed with water (2”15 mL) and the organic layer was dried over
MgSO4. The solvent was removed in vacuo and the residue was recrystal-
lized by addition of a hexane/ethyl acetate 3:1 (5 mL). The white crystals
were collected by filtration, washed with Et2O, and dried in vacuo to
(R)-(ꢀ)-1-(2-Pyridinylcarboxamido)-2-benzamido-1-phenylethane [(R)-
(ꢀ)-(13)]: A solution of diethyl diazoacetate (1.44 g, 8.25 mmol) in THF
(20 mL) was added dropwise to a suspension of hydroxyamide (R)-(ꢀ)-
26 (2.0 g, 8.25 mmol), phthalimide (1.21 g, 8.25 mmol) and triphenylphos-
phine (2.16 g, 8.25 mmol) in THF (40 mL) and the mixture was stirred at
room temperature overnight. The mixture was then concentrated in
vacuo and Et2O (20 mL) was added to the residue to form a white pre-
cipitate. The precipitate was collected by filtration, washed with Et2O,
and dried in vacuo to afford crude phthalimido derivative (1.60 g, 51%),
afford (R)-(ꢀ)-15 (350 mg, 78%). M.p. 125–1278C (decomp); [a]D20
=
1
which was used in the next step without further purification. H NMR: d
ꢀ0.2 (c = 1.8, CHCl3); 1H NMR: d = 4.51 (dd, J=16.3, 5.0 Hz, 1H,
CHH), 4.63(dd, J=16.3, 5.3 Hz, 1H, CHH), 5.79 (d, J=7.3Hz, 1H,
CH), 7.15–7.62 (m, 9H, Ph + Py + Py’+NH), 7.79 (td, J=7.6, 1.6 Hz,
1H, Py), 8.10 (dt, J=7.8, 1.0 Hz, 1H, Py), 8.45 (dm, J=5.0 Hz, 1H, Py),
8.58 (dm, J=4.8 Hz, 1H, Py), 8.70 (brd, J=7.3Hz, 1H, NH); IR: n˜ =
3365m, 2980m, 1665s, 1593m, 1570m, 1490s, 1460m, 1430 cmꢀ1 m; MS
(ES): m/z: 369 [M+Na]+, 347 [M+H]+; HRMS (FAB): m/z: calcd for
C20H19N4O2: 347.15080; found: 347.15083.
= 4.26 (dd, J=14.0, 4.1 Hz, 1H, CHH), 4.41 (dd, J=14.0, 9.9 Hz, 1H,
CHH), 5.79 (td, J=9.4, 4.1 Hz, 1H, CH), 7.44–8.01 (m, 11H, Ph + Py +
Ar), 8.22 (d, J=7.8 Hz, 1H, Py), 8.82 (dm, J=4.8 Hz, 1H, Py), 9.04 (brd,
J=8.2 Hz, 1H, NH); MS (ES): m/z: 394 [M+Na]+, 372 [M+H]+. The
latter phthalimido derivative (1.58 g, 4.25 mmol) was dissolved in DMF
(10 mL), hydrazine hydrate (250 mg, 5 mmol) was added, and the mixture
was stirred at room temperature overnight. The mixture was then diluted
with water (30 mL), extracted with CH2Cl2 (4”20 mL), the organic ex-
tracts were dried over MgSO4, and the solvent was removed in vacuo.
The residue was passed through a short silica gel column (5”2 cm) with
a mixture of chloroform/methanol 9:1 to afford crude amine (R)-27
(550 mg, 54%), which was used in the next step without further purifica-
tion. 1H NMR: d = 3.32 (m, 2H, CH2), 5.38 (dt, J=7.6, 5.3Hz, 1H,
CH), 7.42–7.72 (m, 6H, Ph + Py), 8.02 (td, J=7.6, 1.6 Hz, 1H, Py), 8.38
(d, J=7.8 Hz, 1H, Py), 8.78 (dm, J=4.6 Hz, 1H, Py), 8.95 (brd, J=
7.5 Hz, 1H, NH); MS (ES): m/z: 242 [M+H]+. The crude amine 27
(275 mg, 1.14 mmol) was dissolved in CH2Cl2 (20 mL) and triethylamine
(555 mg, 5.5 mmol) was added, followed by a dropwise addition of a solu-
tion of benzoyl chloride (280 mg, 2 mmol) in CH2Cl2 (3mL) at 0 8C. The
mixture was stirred at room temperature for 2 h and then water (20 mL)
was added. The organic layer was separated and washed successively
with water (20 mL), satd aq NaHCO3 (20 mL) and water (20 mL), and
dried over MgSO4. The solvent was removed in vacuo and the oily resi-
due was recrystallized on addition of Et2O (3mL). The resulting solid
was recrystallized from a mixture of hexane/ethyl acetate to give (R)-
(ꢀ)-13 as white crystals (313 mg, 80%). M.p. 183–1848C (CH2Cl2/Et2O);
[a]2D0 = ꢀ60.0 (c = 1.5, CHCl3); 1H NMR: d = 3.98 (m, 2H, CH2), 5.44
(m, 1H, CH), 7.27–7.49 (m, 10H, Ph + Ph’ + Py + NH), 7.77–7.87 (m,
3H, Ph’ + Py), 8.18 (d, J=7.8 Hz, 1H, Py), 8.55 (dm, J=4.8 Hz, 1H,
Py), 8.73(brd, J=7.6 Hz, 1H, NH); IR: n˜ = 3360m, 2960m, 1658s,
(R)-(ꢀ)-1-(2-Pyridinylcarboxamido)-2-(2-pyridinylcarboxy)-1-phenyl-
ethane [(R)-(ꢀ)-(16)]: a-Picolinic acid (1.23g, 10 mmol) was heated at
reflux in SOCl2 (10 mL, 137 mmol) at 858C for 2 h. The volatiles were re-
moved in vacuo, and the residue was dissolved in CH2Cl2 (10 mL). The
resulting solution was added dropwise to a stirred solution of (R)-(ꢀ)-
phenylglycinol (R)-(ꢀ)-25 (0.41 g, 3.0 mmol) and Et3N (2.02 g, 20 mmol)
in CH2Cl2 (10 mL) at 08C. The mixture was stirred at room temperature
for 2 h and then water (20 mL) was added. The organic layer was sepa-
rated and washed successively with water (20 mL), satd aq NaHCO3
(20 mL), and again water (20 mL), and dried over MgSO4. The solvent
was removed in vacuo; Et2O (5 mL) was added to the oily residue that
slowly solidified. The solid was recrystallized from hexane/benzene to
give (R)-(ꢀ)-16 (0.70 g, 67%) as white crystals. M.p. 154–1558C; [a]D20
=
ꢀ4.3( c = 1.9, CHCl3); 1H NMR: d = 4.75 (dd, J=11.5, 5.0 Hz, 1H,
CHH), 4.84 (dd, J=11.5, 7.6 Hz, 1H, CHH), 5.70 (td, J=8.0, 5.0 Hz, 1H,
CH), 7.27–7.52 (m, 7H, Ph + Py + Py’), 7.79 (m, 2H, Py + Py’), 8.04
(d, J=7.8 Hz, 1H, Py), 8.15 (d, J=7.8 Hz, 1H, Py’), 8.56 (d, J=4.8 Hz,
1H, Py), 8.74 (m, 2H, Py’
+ = 53.1 (CH), 67.9
NH); 13C NMR: d
(CH2), 122.7 (Py, CH), 125.7 (Py’, CH), 126.7 (Py, CH), 127.29 (Py’, CH),
127.33 (Ph, 2”CH), 128.5 (Ph, CH), 129.3 (Ph, 2”CH), 137.4 (Py, CH),
137.7 (Py’, CH), 138.6 (Ph, C), 148.0, 150.0 (Py, Py’, C), 148.6, 150.4 (Py,
Py’, CH), 164.5, 165.3(2”CO); IR: n˜ = 3375m, 2985m, 1735s (ester),
Chem. Eur. J. 2006, 12, 6910 – 6929
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6923

ˇ
´
P. Koc ovsky, A. V. Malkov, G. C. Lloyd-Jones et al.
1675s (amide), 1585m, 1570m, 1500s, 1465m, 1430 cmꢀ1 m; MS (ES):
m/z: 370 [M+Na]+, 348 [M+H]+; HRMS (FAB): m/z: calcd for
C20H18N3O3: 348.13482; found: 348.13483.
(R)-29: A solution of diethyl azodicarboxylate (1.20 g, 6.89 mmol) in
THF (20 mL) was added dropwise to a suspension of crude hydroxy-
amide (R)-(ꢀ)-28 (1.5 g, 6.79 mmol), phthalimide (1.00 g, 6.79 mmol),
and triphenylphosphine (1.78 g, 6.79 mmol) in THF (40 mL). The mixture
was stirred at room temperature overnight and was then concentrated in
vacuo. Et2O (20 mL) was added to the residue to form a white precipi-
tate. The precipitate was collected by filtration, washed with Et2O, and
dried in vacuo to afford crude phthalimido derivative, which was directly
used in the next step without further purification. 1H NMR: d = 1.27 (s,
9H, tBu), 3.95 (m, 2H, CH2), 5.13(m, 1H, PhC H), 5.32 (brs, 1H, NH),
7.28–7.47 (m, 5H, Ph), 7.73(m, 2H, Ar), 7.88 (m, 2H, Ar).
(R)-(ꢀ)-2-(2-Pyridinylcarboxamido)-2-phenylethane [(S)-(ꢀ)-(17)]:
A
solution of a-picolinic acid chloride [prepared from a-picolinic acid
(1.23g, 10 mmol) and SOCl 2 (10 mL, 137 mmol) as described for the syn-
thesis of 16] in CH2Cl2 (5 mL) was added dropwise to a stirred solution
of (S)-(ꢀ)-a-methylbenzylamine (0.48 g, 4 mmol) and Et3N (2.02 g,
20 mmol) in CH2Cl2 (20 mL) at 08C. The mixture was stirred at room
temperature for 2 h and then water (20 mL) was added. The organic
layer was separated and washed successively with water (20 mL), satd aq
NaHCO3 (20 mL) and again water (20 mL), and dried over MgSO4. The
solvent was removed in vacuum and Et2O (5 mL) was added to the oily
residue that slowly solidified. The solid material was recrystallized from
hexane/benzene to give (S)-(ꢀ)-17 as white crystals (0.65 g, 72%). M.p.
54–558C; [a]2D0 = ꢀ1.76 (c = 1.8, CHCl3); 1H NMR: d = 1.62 (d, J=
6.9 Hz, 3H, Me), 5.32 (m, 1H, CH), 7.22–7.43 (m, 6H, Ph + Py), 7.82
(td, J=7.8, 1.6 Hz, 1H, Py), 8.19 (dt, J=7.8, 0.9 Hz, 1H, Py), 8.32 (brd,
J=6.0 Hz, 1H, NH), 8.53(dm, J=4.6 Hz, 1H, Py); IR: n˜ = 3380m,
2980m, 1670s, 1595m, 1570m, 1510s, 1465m, 1450m, 1430 cmꢀ1 m; MS
(ES): m/z: 249 [M+Na]+, in agreement with the literature data.[56]
The latter phthalimido derivative was dissolved in ethanol (30 mL), hy-
drazine hydrate (2 mL, 41 mmol) was added, and the mixture was heated
at reflux (808C) for 5 h. The mixture was then cooled, diluted with water
(30 mL), and extracted with CH2Cl2 (4”20 mL). The organic extracts
were dried over MgSO4 and the solvent was removed in vacuo. The resi-
due was passed through a short silica gel column (5”2 cm) with dichloro-
methane/methanol 9:1 to afford crude amine (R)-29 (0.77 g, 48% per
two steps), which was used in the next step without further purification.
1H NMR: d
= 1.44 (s, 9H, tBu), 3.01 (m, 2H, CH2), 4.67 (m, 1H,
PhCH), 5.33 (brs, 1H, NH), 7.26–7.48 (m, 5H, Ph).
(R)-31: Methyl chloroformate (0.38 mL, 4.95 mmol) was added dropwise
via syringe to a solution of picolinic acid (610 mg, 4.95 mmol) and Et3N
(0.50 g, 4.95 mmol) in THF (20 mL) at 08C. The mixture was stirred for
30 min at that temperature, while white precipitate formed during this
time. The precipitate was removed by filtration under nitrogen and the
filtrate was added dropwise to a solution of the crude amine (R)-29
(0.77 g, 3.27 mmol) and Et3N (0.50 g, 4.95 mmol) in THF (20 mL) at 08C.
The mixture was allowed to warm to room temperature and left stirring
overnight. Water (30 mL) was added to the mixture, and the product was
taken up into CH2Cl2 (3”20 mL). The organic extracts were dried over
MgSO4 and the solvent was removed in vacuo. The residue was passed
through a short silica gel column (5”2 cm) with ethyl acetate/methanol
9:1 to afford crude (R)-30 as a white solid (1.05 g, 94%) which was im-
mediately used in the next step.
(R)-(ꢀ)-2-(2-Pyridinylcarboxamido)-2-phenylacetic acid methyl ester
[(R)-(ꢀ)-(18)]: The title compound was prepared by the same procedure
as 16 using hydrochloride of (R)-(ꢀ)-phenylglycine methyl ester (R)-(ꢀ)-
22a (800 mg, 3.97 mmol). The crude product was purified by chromatog-
raphy on silica gel with hexane/ethyl acetate 3:1 to give (R)-(ꢀ)-18
(960 mg, 89%) as a white solid. M.p. 61–638C; [a]2D0 = ꢀ76.5 (c = 2.2,
CHCl3); 1H NMR: d = 3.77 (s, 3H, OMe), 5.78 (d, J=7.6 Hz, 1H, CH),
7.27–7.50 (m, 6H, Ph + Py), 7.82 (td, J=7.8, 1.6 Hz, 1H, Py), 8.16 (d,
J=7.8 Hz, 1H, Py), 8.58 (d, J=4.8 Hz, 1H, Py), 8.70 (brd, J=6.9 Hz,
1H, NH); IR: n˜ = 3380m, 3000m, 2975m, 1742s (ester), 1675s (amide),
1591m, 1570m, 1496s, 1465m, 1430 cmꢀ1 m; MS (ES): m/z: 293[ M+Na]+
, 271 [M+H]+; HRMS (FAB): m/z: calcd for C15H15N2O3: 271.10827;
found: 271.10823.
(R)-(ꢀ)-2-(2-Pyridinylcarboxamido)-2-phenylethan-1-ol [(R)-(ꢀ)-(26)]:
A solution of a-picolinic acid chloride [prepared from a-picolinic acid
(1.847 g, 15 mmol) and SOCl2 (10 mL, 137 mmol) as described for the
preparation of 16] in CH2Cl2 (5 mL) and the resulting solution was added
The product was dissolved in TFA (2 mL) at room temperature and stir-
red for 2 h. Volatiles were removed in vacuo to give crude (R)-31 as a
yellow oil (640 mg, 86%), which was immediately used in the next step.
1,2-Bis(2-pyridinylcarboxamido)ethane
(19):
(PhO)3P
(5.50 mL,
dropwise to
a
stirred solution of (R)-(ꢀ)-phenylglycinol 25 (2.06 g,
20.99 mmol, 2.1 equiv) was added to
a solution of a-picolinic acid
15 mmol) and Et3N (4.04 g, 40 mmol) in CH2Cl2 (20 mL) at 08C. The
mixture was stirred at room temperature for 2 h and then water (20 mL)
was added. The organic layer was separated and washed successively
with water (20 mL), satd aq NaHCO3 (20 mL), and again water (20 mL),
and dried over MgSO4. The solvent was removed in vacuo and Et2O
(5 mL) was added to the oily residue that slowly solidified. The solid was
recrystallized from hexane/benzene to give (R)-(ꢀ)-26 as white crystals
(2.5 g, 69%). M.p. 115–1168C (decomp); [a]2D0 = ꢀ5.2 (c = 2.6, CHCl3);
1H NMR: d = 3.07 (t, J=6.2 Hz, 1H, OH), 3.99 (t, J=6.0 Hz, 2H, CH2),
5.44 (dt, J=7.6, 5.3Hz, 1H, CH), 7.26–7.44 (m, 6H, Ph + Py), 7.83(td,
J=7.8, 1.6 Hz, 1H, Py), 8.17 (dt, J=7.8, 1.1 Hz, 1H, Py), 8.53(dm, J=
4.6 Hz, 1H, Py), 8.70 (brd, J=6.9 Hz, 1H, NH); IR: n˜ = 3600w, 3380m,
2940m, 1669s, 1595m, 1575m, 1500m, 1460m, 1430 cmꢀ1 m; MS (ES): m/
z: 265 [M+Na]+, 243[ M+H]+; HRMS (FAB): m/z: calcd for
C14H15N2O2: 243.11335; found: 243.11330.
(2.585 g, 21.00 mmol, 2.1 equiv) in dry pyridine (35 mL) under nitrogen.
The solution was heated to 858C and freshly distilled 1,2-ethylenedia-
mine (670 mL, 10.02 mmol, 1.0 equiv) was added dropwise over a period
of 5 min. The reaction mixture was stirred at 1008C for 24 h and allowed
to stand at RT overnight. A white precipitate was separated by suction,
washed with water (3”20 mL), and dried at RT overnight to obtain 19 as
white crystals (2.051 g, 76%). 1H NMR (CDCl3): d = 3.75–3.77 (m, 4H,
2”CH2), 7.42 (ddd, J=7.6, 4.8, 1.2 Hz, 2H, arom.), 7.84 (ddd, J=7.7, 7.7,
1.7 Hz, 2H, arom.), 8.20 (d, J=7.8 Hz, 2H, arom.), 8.42 (m, 2H, 2”NH),
1
8.56 (brd, J=4.1 Hz, 2H, arom.); H NMR ([D8]THF): d = 3.60–3.62 (m,
4H, 2”CH2), 7.42 (ddd, J=7.6, 4.7, 1.2 Hz, 2H, arom.), 7.85 (ddd, J=
7.7, 7.7, 1.7 Hz, 2H, arom.), 8.12 (dt, J=7.8, 1.0, 1.0 Hz, 2H, 2H, arom.),
8.53(dq, J=4.7, 0.9, 0.9, 0.9 Hz, 2H, arom.), 8.67 (m, 2H, 2”NH);
13C NMR (CDCl3): d = 39.9 (2”CH2), 122.6 (2”CH-arom), 126.6 (2”
CH-arom), 137.7 (2”CH-arom), 148.6 (2”CH-arom), 150.1 (2”C-2),
165.4 (2”C=O); 13C NMR ([D8]THF): d = 40.2 (2”CH2), 122.8 (2”CH-
arom), 126.6 (2”CH-arom), 137.8 (2”CH-arom), 148.9 (2”CH-arom),
151.7 (2”C-arom), 165.0 (2”C=O); IR: n˜ = 3392w, 3060vw, 3018m,
1672s, 1592w, 1571w, 1527s, 1465w, 1435w, 1363vw, 1288w, 1240w,
1161w, 998w, 905 cmꢀ1 vw.
(R)-(ꢀ)-28: Di-tert-butyl dicarbonate [(Boc)2O] (3.27 g, 15 mmol) was
added in several portions to a stirred solution of (R)-(ꢀ)-phenylglycinol
25 (2.06 g, 15 mmol) and Et3N (4.04 g, 40 mmol) in CH2Cl2 (20 mL) at
08C. The mixture was allowed to warm to room temperature and stirred
for 12 h. After addition of water (20 mL), the organic layer was separated
and washed successively with water (20 mL), satd aq NaHCO3 (20 mL),
and again water (20 mL), and dried over MgSO4. The solvent was re-
moved in vacuo to give (R)-(ꢀ)-28 as an oily residue that slowly solidi-
fied on standing (3.41 g, 96%). The product thus obtained was used in
the next step without further purification. 1H NMR: d = 1.36 (s, 9H,
tBu), 2.27 (brs, 1H, OH), 3.77 (m, 2H, CH2), 4.71 (m, 1H, PhCH), 5.17
(brs, 1H, NH), 7.10–7.31 (m, 5H, Ph) in accordance with the litera-
ture.[57]
N-Methyl-N-{2-[(pyridinyl-2-carboxamido]ethyl}-2-pyridinecarboxamide
(20): A solution of diamide 19 (200 mg, 0.740 mmol, 1.0 equiv) in dry
N,N-dimethylformamide (7 mL) was added dropwise to NaH (33 mg,
60% suspension, 0.825 mmol, 1.1 equiv) in dry N,N-dimethylformamide
(5 mL) under nitrogen. The mixture was stirred at 408C for 45 min and
cooled to room temperature; a white suspension was formed. Methyl
iodide (50 mL, 0.803mmol, 1.1 equiv) was added and the mixture was stir-
red at room temperature for 2 h, during which period an almost clear sol-
6924
www.chemeurj.org
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 6910 – 6929

Asymmetric Allylic Substitution
FULL PAPER
ution was formed. The solvent was removed in vacuo using rotary evapo-
rator and the residue was partitioned between CH2Cl2 and water. The or-
ganic layer was separated, washed with water and dried over anhydrous
MgSO4. The crude product was repeatedly purified from unreacted 19
and the bismethylated product 21 by flash chromatography on a silica gel
column (15”2.5 cm) with ethyl acetate/methanol 98:2 ! 80:20, which af-
forded pure monomethyl derivative 20 as an oil that slowly solidified
3-2H]; MS (EI): m/z (%): 133 (100) [M]+, 116 (30), 105 (40), 89 (10), 84
(20), 80 (10), 77 (60), 74 (11), 66 (8), 63(20), 54 (44).
(Æ)-(Z)-3-[2H1]-1-Phenylprop-2-en-1-ol [(Æ)-(37a)]: DIBAL-H (1m in
hexane, 2.62 mL, 2.62 mmol) was added slowly to a solution of 3-[2H1]-1-
phenylprop-2-yn-1-ol [(Æ)-36] (314 mg, 2.36 mmol, 98% 2H) in dichloro-
methane (4 mL) in a Schlenk tube under nitrogen. The mixture was stir-
red at room temperature for 10 min before adding to a slurry of Schwartz
reagent (676 mg, 2.62 mmol) in anhydrous dichloromethane (20 mL)
under nitrogen at 08C. After 10 min the mixture became pale orange in
color. The reaction mixture was quenched by dropwise addition of satd
aq NaHCO3, filtered through a plug of silica gel (2.5”1 cm); the solvent
was evaporated to afford (Æ)-37a as a colorless oil (253mg, 79%; 98%
2H by 1H NMR) which was not further purified. Note that the product
was contaminated with 1–5% of 3-deuterio-1-phenylpropanol (an over-
reduction product), which did not need to be separated for the subse-
quent synthetic steps. The Z/E ratio was 99.76:0.24 as shown by
1H NMR. 1H NMR (270 MHz, CDCl3, 228C, TMS): d = 7.42–7.24 (m,
5H, arom. H), 6.05 (m, 1H, 2-H), 5.26–5.16 (m, 2H, 3-H, 1-H); 13C NMR
(100 MHz, 218C, TMS): d = 142.6 (arom C), 140.2 (C-2), 128.7, 128.5,
1
(77 mg, 37%). H NMR (two stereoisomers in a 2:1 ratio): d = 3.08 (s, 1/
3of 3H, CH 3), 3.11 (s, 2/3 of 3H, CH3), 3.68–3.80 (m, 4H, 2”CH2), 7.26
(ddd, J=7.6, 4.9, 1.2 Hz, 1/3of 1H, arom.), 7.30 (ddd, J=7.3, 4.9, 1.4 Hz,
2/3 of 1H, arom.), 7.32–7.37 (m, 1H, arom.), 7.50 (d, J=7.8 Hz, 1/3of
1H, arom.), 7.63(d, J=7.7 Hz, 2/3of 1H, arom.), 7.65–7.72 (m, 1H,
arom.), 7.74–7.80 (m, 1H, arom.), 8.09 (d, J=7.8 Hz, 2/3of 1H, arom.),
8.12 (d, J=7.8 Hz, 1/3of 1H, arom.), 8.43(m, 1/3of 1H, NH), 8.47 (d,
J=4.2 Hz, 2/3of 1H, arom.), 8.52 (d, J=4.7 Hz, 2/3of 1H, arom.), 8.63
(d, J=4.5 Hz, 2/3of 1H, arom.), 9.49 (m, 2/3of 1H, NH); 13C NMR (two
stereoisomers in a 2:1 ratio): d = 33.9 (CH3), 37.8 (CH2), 38.1 (CH3),
38.3 (CH2), 47.9 (CH2), 49.6 (CH2), 122.4 (CH-arom.), 122.6 (CH-arom.),
123.9 (CH-arom.), 124.8 (CH-arom.), 125.0 (CH-arom.), 125.1 (CH-
arom.), 126.4 (CH-arom.), 126.5 (CH-arom.), 137.3 (CH-arom.), 137.6
(CH-arom.), 137.6 (CH-arom.), 148.0 (CH-arom.), 148.5 (CH-arom.),
148.7 (CH-arom.), 148.7 (CH-arom.), 150.6 (C-2), 154.0 (C-2), 165.7 (C=
O), 169.4 (C=O) and other signals of a minor isomer overlapped; IR: n˜
= 3689w, 3387w, 3018vs, 1669m, 1632s, 1590w, 1569m, 1528s, 1465w,
1434w, 1403w, 1289w, 1238w, 932 cmꢀ1 vw; HRMS (EI): m/z: calcd for
C15H16N4O2: 284.12739; found: 284.12733.
1
2
126.3, (arom CH), 114.8 (t, J
(C,2H)=23.9 Hz; C-3), 75.9 (C-1); H NMR
A
(46 MHz, CH2Cl2, 238C, CDCl3): d = 5.54 (brs, 3-2H); MS (EI): m/z
(%): 135 (90) [M]+, 149 (5), 119 (15), 116 (100), 105 (30), 92 (27), 84
(57), 77 (35), 71 (5), 63 (9), 56 (11),52 (6).
(Æ)-(Z)-3-[2H1]-1-Phenylprop-2-enyl acetate [(Æ)-(37b)]: A solution of
(Z)-3-[2H1]-1-phenylprop-2-en-1-ol (Æ)-37a (185 mg, 1.37 mmol, 98%
2H), Et3N (284 mg, 2.80 mmol), and DMAP (5.6 mg, 0.046 mmol) in di-
chloromethane (2 mL) at 08C was slowly treated with a solution of Ac2O
(202 mg, 1.98 mmol) in dichloromethane (0.1 mL) and then allowed to
warm to 258C over a period of 3h. After quenching with an ice-cold
50% aqueous solution of NH4Cl (5 mL), the mixture was extracted with
dichloromethane (3”5 mL). The combined organic extracts were washed
with satd aq NH4Cl (25 mL), satd aq NaCl and satd aq NaHCO3, before
being dried (MgSO4) and then concentrated in vacuo. Chromatography
on silica gel (50 g) with hexane/ethyl acetate 4:1 afforded (Æ)-37b as a
colorless oil (174 mg, 72%; 98% 2H by 1H NMR). 1H NMR (400 MHz,
N-Methyl-N-{2-[methyl(2-pyridinyl)carboxamido]ethyl}-2-pyridinecar-
boxamide (21): A solution of diamide 20 (200 mg, 0.740 mmol, 100%) in
dry DMF (7 mL) was added dropwise to NaH (65 mg, 60% suspension,
1.625 mmol, 2.2 equiv) in dry DMF (5 mL) under nitrogen. The mixture
was stirred at 408C for 45 min and cooled to room temperature; a white
suspension was formed. Methyl iodide (100 mL, 1.606 mmol, 2.2 equiv)
was added and the mixture was stirred at room temperature overnight,
during which period an almost clear solution was formed. The solvent
was removed in vacuo using rotary evaporator and the residue was parti-
tioned between CH2Cl2 and water. The organic layer was separated,
washed with water, and dried over anhydrous MgSO4. The crude product
was repeatedly purified from unreacted 19 and the mono-methylated
product 20 by flash chromatography on a silica gel column (15”2.5 cm)
with ethyl acetate/methanol 80:20, which afforded pure bismethylated de-
rivative 21 as an oil that slowly solidified (82 mg, 37%). 1H NMR (three
stereoisomers in a 4:3:3 ratio): d = 2.84 (s, 3/10 of 6H, CH3), 2.94 (s, 2/
10 of 6H, CH3), 3.11 (s, 3/10 of 6H, CH3), 3.19 (s, 2/10 of 6H, CH3),
3.67–3.82 (m, 4H, 2”CH2), 7.25–7.30 (m, 2H, arom.), 7.54 (t, J=8.8 Hz,
1H, arom.), 7.62–7.77 (m, 3H, arom.), 8.50 (dt, J=14.5, 14.5, 4.3Hz, 2H,
arom.); 13C NMR (three stereoisomers in a 4:3:3 ratio): d = 34.6 (CH3),
35.0 (CH3), 38.1 (CH3), 38.3 (CH3), 45.6 (CH2), 47.6 (CH2), 48.4 (CH2),
49.7 (CH2), 123.6 (CH-arom.), 124.1 (CH-arom.), 124.5 (CH-arom.),
124.7 (CH-arom.), 124.7 (CH-arom.), 124.9 (CH-arom.), 125.0 (CH-
arom.), 125.1 (CH-arom.), 137.3 (CH-arom.), 137.4 (CH-arom.), 137.5
(CH-arom.), 137.6 (CH-arom.), 148.1 (CH-arom.), 148.3 (CH-arom.),
148.5 (CH-arom.), 148.9 (CH-arom.), 154.3(C-arom.), 154.5 (C-arom.),
154.6 (C-arom.), 155.0 (C-arom.), 169.0 (C=O), 169.1 (C=O), 169.4 (C=
O), 169.7 (C=O); IR: n˜ = 3018s, 1632vs, 1589w, 1568m, 1495m, 1463w,
1443w, 1425w, 1406m, 1359vw, 1289w, 1240w, 1150w, 1079w, 1046w,
996w, 965vw, 809 cmꢀ1 vw; HRMS (EI): m/z: calcd for C16H18N4O2:
298.14301; found: 298.14298.
CDCl3, 228C, TMS): d = 7.49–7.28 (m, 5H, arom H), 6.26 (d, 3J
ACHTREUNG
5.8 Hz, 1H; 1-H), 6.02 (m, 1H; 2-H), 5.23(d, 3J
ACHTREUNG
H), 2.10 (s, 3H; CH3); 13C NMR: d = (CDCl3, 70 MHz, 218C, TMS): d
= 167.0 (C=O), 138.9 (arom C), 136.2 (C-2), 128.5, 128.1, 127.1 (arom
CH), 116.6 [t, 1J(C,2H)=23.8 Hz, C-3], 75.8 (C-1), 21.2 (CH3); 2H NMR
G
(46 MHz, CH2Cl2, 238C, CDCl3): d = 5.18 (brs, 3-2H); MS (EI): m/z
(%): 177 (1) [M]+, 143 (6), 132 (32), 116 (23), 107 (30), 91 (10), 84 (62),
79 (52), 56 (42).
(Æ)-(Z)-3-[2H1]-1-Phenylprop-2-enyl methyl carbonate [(Æ)-(37c)]:
Methyl chloroformate (1.25 mL, 17 mmol) was added dropwise to a solu-
tion of (Z)-3-[2H1]-1-phenylprop-2-en-1-ol (Æ)-37a (1.00 g, 7.41 mmol;
98% 2H) and pyridine (5 mL, 70 mmol) in dichloromethane (10 mL) at
08C and the reaction mixture was heated under reflux for 18 h. After
quenching with 50% satd aq NH4Cl (10 mL), the mixture was extracted
with dichloromethane (3”15 mL). The combined organic extracts were
washed with satd aq NaCl and then dried (MgSO4). Evaporation of the
solvent under reduced pressure afforded an oil that was purified by chro-
matography on silica gel (125 g) with hexane/ethyl acetate 6:1, and then
distilled (4 mbar, Kugelrohr oven, T=1208C) to afford (Æ)-37c as a col-
orless oil (960 mg, 67%; 98% 2H/H by 1H NMR). 1H NMR (400 MHz,
3
(Æ)-3-[2H1]-1-Phenylprop-2-yn-1-ol [(Æ)-(36)]: 1-Phenylprop-2-yn-1-ol
(35) (1 g, 7.57 mmol) was added to a mixture of D2O (20 mL) and K2CO3
(1.04 g, 7.57 mmol) under nitrogen. After 1 h at room temperature the
product was extracted with dichloromethane (3”10 mL). The combined
extracts were dried (MgSO4), and evaporated to afford a yellow oil that
was purified by distillation (36 mbar, Kugelrohr oven at 1408C) to afford
(Æ)-36 as a yellow oil (0.70 g, 69%; 98% 2H by 1H NMR). 1H NMR
(270 MHz, CDCl3, 228C, TMS): d = 7.59–7.49 (m, 2H, CHortho), 7.43–
7.29 (m, 3H, arom. H), 5.44 (s, 1H, 1-H), 2.16 (brs, 1H, OH); 13C NMR
(100 MHz, 218C, TMS): d = 140.0 (arom. C), 128.7, 128.5, 126.6 (arom.
CDCl3, 228C, TMS): d = 7.40–7.29 (m, 5H, arom. H), 6.09 (d, J
A
6.3Hz, 1H, 1-H), 6.03(brm, 1H; 2-H), 5.26 (d, 3J
(H,H)=10.3Hz, 1H, 3-
G
H), 3.78 (s, 3H, CH3); 13C NMR (100 MHz, 218C, TMS): d = 155.0 (C=
O), 138.3 (arom C), 135.7 (C-2), 128.6, 128.4, 127.0 (arom CH), 116.9 [t,
1J(C,2H)=25.0 Hz, C-3], 80.1 (C-1), 54.8 (CH3); 2H NMR (60 MHz,
G
CH2Cl2, 238C, CDCl3): d = 5.35 (brs, 3-2H); MS (EI): m/z (%): 193(22)
[M]+, 149 (40), 134 (21), 118 (100), 106 (54), 92 (37), 84 (78), 77 (54), 63
(19). Samples of (S)-(ꢀ)-37c (>95% ee) and (R)-(+)-37c (>95% ee)
were prepared analogously from the corresponding enantiomerically en-
riched alcohols 37a and their ee was determined by HPLC on Chiral OJ
column employing a mixture of hexane and 2-propanol (93:7) as eluent
with UV detection at 220 nm (0.5 mLminꢀ1; tS =24.5 min, tR =27.2 min).
CH), 83.0 [t, 2J
N
(C,2H)=38.4 Hz, C-3],
Chem. Eur. J. 2006, 12, 6910 – 6929
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6925

ˇ
´
P. Koc ovsky, A. V. Malkov, G. C. Lloyd-Jones et al.
Absolute configurations were assigned by comparison of HPLC and opti-
1H, 3-H), 6.30 (dd, 3J
H), 4.14 (dq, 3J(H,H)=7.2, 3J
(H,H)=7.2 Hz, 6H, 2”CH3); 13C NMR (100 MHz, 218C, TMS): d
155.9 (C=O), 133.8 (arom CH), 128.5 (C-3), 128.0, 126.5, 123.4 (arom
ACHTREUNG
cal rotation data from 37c with unlabelled 5.[3h]
G
ACHTREUNG
(Æ)-(E)-1-[2H1]-3-Phenylprop-2-en-1-ol (Æ)-(38a): Method A: NaBD4
(501 mg, 12.0 mmol) was added to a solution of (E)-3-phenyl-prop-2-enyl
aldehyde 40 (1.52 g, 11.5 mmol) in methanol (30 mL) at 08C. After
10 min, the reaction was quenched with water and the mixture was dilut-
ed with dichloromethane (30 mL). The layers were separated and the
aqueous layer was extracted with dichloromethane (20”3mL) and dried
(MgSO4). Evaporation of the solvent afforded (Æ)-38a as yellow crystals
(1.44 g, 93%, 95% 2H).
N
=
CH), 123.3 (C-2), 67.5 [t, 1J(C,2H)=21.8 Hz, C-1], 63.7 (CH2), 16.2
U
(CH3); 2H NMR (46 MHz, CHCl3, 238C, CDCl3): d = 4.64 (brs, 1-2H);
MS (EI): m/z (%): 271 (45) [M]+, 155 (38), 134 (5), 127 (28), 116 (100),
99 (35), 92 (17), 86 (5), 81 (14), 65 (6).
1-(1R,1’R)-(E)-1-[2H1]-3-Phenylprop-2-en-1-yl 1’-phenylethyl carbamate
[(R,R)-(39)] and1-(1 S,1’R)-(E)-1-[2H1]-3-phenylprop-2-en-1-yl 1’-phenyl-
ethyl carbamate [(S,R)-(39)]: (R)-(+)-a-methyl benzylisocyanate (55 mg,
0.370 mmol) was added to a solution of racemic (E)-1-[2H1]-3-phenyl-
prop-2-en-1-ol (Æ)-38a (50 mg, 0.370 mmol) and 4-(N,N-dimethylamino)-
pyridine (5 mg, 10%) in toluene (3mL) and the mixture was heated
under reflux overnight. A white solid was obtained on removal of the sol-
vent and washing of the residue with n-pentane. Recrystallization from
hexane/benzene gave pure carbamate 39 (60 mg, 58%). M.p. 74–768C;
Method B: A solution of (E)-1-[2H1]-3-phenyl-prop-2-enyl acetate (Æ)-
2
38b (125 mg, 0.706 mmol, 98% H) in methanol (3mL) was treated with
K2CO3 (33 mg, 0.24 mmol) and the resulting suspension was stirred at
258C for 6 h. After the volatiles were removed in vacuo, the yellow resi-
due was dissolved in Et2O and the combined organic phases were dried
(MgSO4) and evaporated. Chromatography on silica gel (50 g) with
hexane/ethyl acetate 4:1 afforded (Æ)-38a as a colorless oil (67 mg, 70%;
98% 2H by 1H NMR). 1H NMR (270 MHz, CDCl3, 228C, TMS): d =
1H NMR (400 MHz, CDCl3, 228C, TMS):[58]
arom H), 6.62 (d, 3J(H,H)=15.6 Hz, 1H, 3-H), 6.27 (dd, 3J
5.9 Hz, 1H, 2-H), 5.06 (brs, 1H, NH or CH-N), 4.92–4.86 (brm, 1H, CH-
N or NH), 4.71 [d, 3J(H,H)=5.9 Hz, 1H, (R)-1-H], 4.67 [d, 3J
(H,H)=
5.9 Hz, 1H, (S)-1-H], 1.48 (d, 3J(H,H)=6.3Hz, CH 3); 13C NMR
d
=
7.41–7.21 (m, 10H,
(H,H)=15.1,
7.43–7.20 (m, 5H, arom. H), 6.62 (dd, 3J
1H; 3-H), 6.36 (dd, 3J
(H,H)=15.8, 5.6 Hz, 1H; 2-H), 4.31 (brm, 1H; 1-
H); 13C NMR (100 MHz, 218C, TMS): d = 135.8, 134.1, 128.6, 128.2,
(H,H)=15.8, 4J
(H,H)=1.3Hz,
E
ACHTREUNG
ACHTREUNG
A
ACHTREUNG
126.7, 124.8 (arom. C, arom. CH, C-3, C-2), 45.2 (t, 1J(C,2H)=23Hz; C-
U
AHCTREUNG
1); 2H NMR (46 MHz, CH2Cl2, 238C, CDCl3): d = 4.27 (brs, 1-2H); MS
(EI): m/z (%): 135 (85) [M]+, 118 (35), 116 (60), 106 (45), 103 (22), 92
(100), 89 (7), 86 (62), 84 (93), 80 (15), 78 (75), 65 (12), 63 (20), 56 (17),
52 (15).
(100 MHz, 218C, TMS): d = 155.5 (C=O), 143.5 (arom C), 136.3 (arom
C), 133.8 (C-3), 128.6, 128.7, 128.5, 127.3, 126.5, 125.9 (arom CH), 123.8
(C-2), 65.1 (t, 1J(C,2H)=23.7 Hz, C-1), 50.6 (CH-N), 22.4 (CH3);
N
2H NMR (60 MHz, CH2Cl2, 238C, CDCl3): d
=
4.63(brs, 1- 2H); IR
(KBr): n˜ = 3200–3500 (broad), 1700 cmꢀ1; MS (EI): m/z (%): 281 (100)
[M]+, 117 (82), 105 (100), 91 (31), 84 (44), 77 (31), 65 (6); elemental anal-
ysis calcd (%) for C18H19NO2: C 76.84, H 6.81; found: C 76.59, H 6.87.
(Æ)-(E)-1-[2H1]-3-Phenylprop-2-enyl acetate [(Æ)-(38b)]: Method A:
This derivative was prepared in an identical manner to (Æ)-37b but start-
2
ing from (Æ)-38a as a colorless oil (80%, 95% H).
Method B: A solution of (Æ)-(Z)-3-[2H1]-1-phenylprop-2-enyl acetate
(Æ)-37b (104 mg, 0.59 mmol; 98% 2H) in chloroform (10 mL) was treat-
ed with [(CH3CN)2PdCl2] (3.9 mg, 0.015 mmol, 2.5 mol%). After stirring
at 258C for 3.5 h, the chloroform was removed in vacuo and the crude
product was purified by chromatography on silica gel (50 g) with 4:1
General procedure for the preparative allylic substitution catalyzed by
molybdenum(0):
A
mixture of [(EtCN)3Mo(CO)3(EtCN)3] (34 mg,
A
0.1 mmol) and a ligand (0.15 mmol) was dissolved in THF (3mL). The
solution, that instantaneously turned deep red, was heated with stirring
at 608C for 40 min. The solution was cooled to room temperature and
then a solution of the corresponding sodiomalonate (2.0 mmol) in THF
(2 mL), generated from dimethyl malonate (or dimethyl methylmalonate)
and NaH, and a solution of allylic carbonate (1.0–1.3mmol) in THF
(1 mL) were successively added. Usually, the addition of the reactants
was accompanied by a change of color to orange or yellow-brown. The
mixture was stirred at 608C until the reaction was complete (as evi-
denced by TLC), then diluted with Et2O (20 mL), and washed successive-
ly with 5% aqueous NaHCO3 and water. The organic phase was dried
with MgSO4 and the solvent was evaporated under reduced pressure. The
crude product was purified by flash chromatography on silica gel (15”
2 cm) with hexane/ethyl acetate 9:1. Enantiomeric purity of the products
6 was determined by chiral HPLC using Chiralcel OD-H column (equip-
ped with a silica gel guard column) and hexane/2-propanol 99.5:0.5; UV
detection at 220 nm. The retention times were as follows: tS =17.4, tR =
18.7 min. Alternatively, the enantiomeric purity of 6 was determined by
2
hexane/AcOEt to afford (Æ)-38b as a colorless oil (74 mg, 71%; 98% H
by 1H NMR). 1H NMR (400 MHz, CDCl3, 228C, TMS): d = 7.44–7.22
(m, 5H; arom H), 6.64 (d, 3J(H,H)=15.7 Hz, 1H; 3-H), 6.27 (dd, 3J-
A
(H,H)=15.7, 6.2 Hz, 1H; 2-H), 4.71 (brm, 1H; 1-H), 2.11 (s, 3H; CH3);
13C NMR (100 MHz, 218C, TMS): d = 170.7 (C=O), 136.2 (arom C),
1
134.2 (C-3), 128.5, 128.0, 126.5 (arom CH), 123.1 C-2), 64.7 (t, J
A
23.0 Hz, C-1), 20.9 (CH3); 2H NMR (46 MHz, CH2Cl2, 238C, CDCl3) d=
4.70 (brs, 1-2H), MS (EI): m/z (%): 177 (25) [M]+, 135 (44), 116 (100),
106 (49), 92 (37), 84 (89), 77 (47), 65 (10), 60 (19); no Z isomer was de-
1
tected by H NMR.
(Æ)-(E)-1-[2H1]-3-Phenylprop-2-enyl methyl carbonate [(Æ)-(38c)]: This
compound was prepared in an identical manner to (Æ)-37a but starting
from (Æ)-38a. (Æ)-38c was obtained as a colorless oil (63%; 98% 2H by
1H NMR). 1H NMR (270 MHz, CDCl3, 228C, TMS): d = 7.43–7.22 (m,
5H, arom H), 6.69 (d, 3J
N
N
1H NMR with [D]-Eu
(hfc)3.
N
15.9 Hz, 6.6 Hz, 1H, 2-H), 4.78 (brm, 1H; 1-H), 3.82 (s, 3H, CH3);
13C NMR (68 MHz, 218C, TMS): d = 155.7 (C=O), 136.1 (arom CH),
134.9 (C-3), 128.6, 128.2, 126.7 (arom CH), 122.3 (C-2), 68.0 (t, 1J-
Typical procedure for the asymmetric molybdenum(0)-catalyzed allylic
substitution with deuterium-labeled substrates: A mixture of [Mo(CO)3-
(h6-C7H8)] (7.1 mg, 26 mmol, 10 mol%) and (ꢀ)-(2S)-N,N’-3-methyl-1,2-
A
CDCl3): d = 4.76 (brs, 1-2H); MS (EI): m/z (%): 193(39) [ M]+, 149
(37), 132 (27), 121 (10), 116 (72), 106 (37), 92 (24), 84 (100), 77 (49), 63
(19).
diaminobutylbis(2-pyridine-carboxamide) (S)-12c (12.2 mg, 39 mmol,
15 mol%) was dissolved in tetrahydrofuran (0.5 mL) to form a deep-red
solution and sodium dimethyl malonate (0.53mmol) in tetrahydrofuran
(1 mL), generated from dimethyl malonate and NaH (60%), and a solu-
tion of the (Z)-deuterated branched allylic methyl carbonate (Æ)-37c
(50 mg, 0.26 mmol) in tetrahydrofuran (1 mL) were successively added.
The mixture was stirred at 608C overnight (20 h), then diluted with Et2O
(20 mL), quenched with 5% aq NaHCO3, and the organic phase was
washed with water. The aqueous phase was extracted with dichlorome-
thane and the combined extracts were dried (MgSO4) and evaporated.
The crude product was purified by flash chromatography on silica gel
(50 g) with hexane/ethyl acetate 6:1. The enantiomeric purity of the prod-
uct thus obtained was found to be 88% ee by chiral HPLC using a Chiral
OJ column with hexane/2-propanol 93:7; UV detection at 220 nm,
0.5 mLminꢀ1 (tS =31, tR =36 min).
(Æ)-(E)-1-[2H1]-3-Phenylprop-2-enyl diethyl phosphate [(Æ)-(38d)]:
Chlorodiethyl phosphate (1.35 mL, 7.8 mmol) was added to a solution of
(E)-1-[2H1]-3-phenylprop-2-en-1-ol (Æ)-38a (1.00 g, 7.4 mmol) and pyri-
dine (0.67 mL) in dichloromethane (10 mL) at 08C over a period of
5 min and the resulting white slurry was stirred at room temperature for
2 h. The reaction mixture was diluted in Et2O (10 mL) and washed suc-
cessively with a 10% aq HCl solution (5”3mL), satd aq NaHCO 3 (5”
3mL) and satd aq brine (5 mL). The organic layer was dried (MgSO 4);
the solvent was evaporated to give a crude oily product that was purified
by chromatography on silica gel (125 g) with hexane/ethyl acetate 2:1 to
afford (Æ)-38d as a brown oil (0.53g, 26%). 1H NMR (270 MHz, CDCl3,
3
228C, TMS): d = 7.43–7.23 (m, 5H, arom H), 6.68 (d, J
A
6926
www.chemeurj.org
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 6910 – 6929

Asymmetric Allylic Substitution
FULL PAPER
Kinetic resolution studies: Samples were taken from the reaction mixture
at regular intervals. Each sample was quenched with water, diluted with
hexane, then filtered through a plug of silica gel and dried (MgSO4). For
each sample the enantiomeric excess and the conversion of substrate was
determined by GC (b-column) using (Æ)-methyl 3-[2H1]-1-phenylpropyl
carbonate as the internal standard. The cis/trans ratios for the branched
product were determined by 1H NMR spectroscopy. The s value was de-
termined by non-linear regression. GC conditions: Capillary columns FS-
HYDRODEX b-3P, temperature 1208C, col. flow: 0.9 mLminꢀ1 (tS =
15.34, tR =15.93min; t for internal standard were 14.47 and 14.94 min).
chloromethane (4”3mL). The combined extracts were dried (MgSO 4)
and then filtered through a short plug of silica gel. Evaporation of the
solvent afforded an oil that was purified by chromatography on silica gel
(100 g) with hexane/ethyl acetate 12:1 and then placed in vacuo
(0.1 mbar) to remove traces of dimethyl malonate, which gave (Æ)-41 as
2
colorless oil (71 mg, 79%; 98% H).
Method C (racemic synthesis of a mixture of E/Z isomers 41/42): These
compounds were prepared in an identical manner to Method B, but em-
ploying (Æ)-37c to give a 24:1 mixture of (Æ)-41/42 (1:1) and (Æ)-43 as
2
a colorless oil (60%; 95% H).
Sample preparation for chiral liquidcrystal matrix deuterium NMR : The
preparation of the sample is very important as it can affect the quality
and reproducibility of the NMR spectra. The samples were prepared by
the following procedure: PBLG (85–88 mg) was weighed directly into a
5 mm o.d. NMR tube and a solution of the product (16–25 mg) in di-
chloromethane (400–500 mg) was added. All samples were of the same
length (3.3 cm) and were centrifuged in both directions (” 3000 r.p.m).
The extent of centrifugation is also an important factor in obtaining good
quality NMR spectra. The NMR tube was centrifuged for 1 h in each di-
rection and then 30 min in each direction and then analyzed within 1 h.
All NMR spectra were obtained on an Eclipse 400 NMR spectrometer at
23Æ18C. The NMR tube was not spun in the magnet. The absolute
values of the quadrupolar coupling vary from experiment to experiment
(by up to ca. 4%). Therefore, control experiments were conducted in
which the following four parameters were varied within limits that could
reasonably be expected under experimental conditions: i) concentration
of PBLG, ii) degree of polymerization of the PBLG, iii) analyte concen-
tration, and iv) temperature. These experiments demonstrated that the
DjnQj values for the five resolvable components [(R)- and (S)-41, 42,
and (R)- and (S)-43] varied in a directly proportional manner and also
that chemical shift anisotropy was negligible. Consequently, DjnQj
values can be normalized to allow confident assignment. The following
Palladium(0)-catalyzed allylic substitution:
A
mixture of [Pd(h-
G
C3H5)MeCN)2]OTf (3.4 mg, 9.0 mmol) and DPPF (5.0 mg, 9.0 mmol) was
dissolved in tetrahydrofuran (1 mL) and the solution was stirred under
nitrogen at 258C for 15 min to afford a brown solution. A solution of
sodium dimethyl malonate (1.42 mmol) in tetrahydrofuran (4 mL), gener-
ated from dimethyl malonate (218 mg, 1.42 mmol) and NaH (60%)
(57 mg, 1.42 mmol), and a solution of (E)-1-[2H1]-3-phenylprop-2-enyl
acetate (Æ)-38b (63mg, 0.36 mmol) in tetrahydrofuran (4 mL) were suc-
cessively added. The mixture was stirred at 258C for 12 h, then diluted
with Et2O (20 mL), and washed successively with 5% aqNaHCO3 and
water. The organic phase was dried (MgSO4) and the crude product was
purified by flash chromatography on silica gel (50 g) with hexane/ethyl
acetate 4:1 to afford (Æ)-43 as a colorless oil (72 mg, 80%; 98% 2H).
The regioselectivity for 43 over 41/42 was greater than 95%. An identical
procedure using (S)-38b (>95% ee) gave (S)-43 (50%, 98% 2H, >95%
ee). The following is the routine (isotropic phase) NMR data for the
three racemic 2H-labeled compounds (obtained from Pd, W or Mo-cata-
lyzed reactions).
(Æ)-41: 1H NMR (400 MHz, CDCl3, 228C, TMS): d = 7.33–7.19 (m, 5H,
3
arom H), 5.99 (brm, 1H, 2-H), 5.07 (d, J
A
(dd, 3J
A
A
3.74 (s, 3H, CH3), 3.49 (s, 3H; CH3).
values act as
a
basis set: (R)-41, DjnQj=525 Hz; (S)-41, DjnQj=
1
(Æ)-42: H NMR (400 MHz, CDCl3, 228C, TMS): d = 7.33–7.19 (m, 5H,
544 Hz; (R and S)-42, DjnQj=49 Hz; (R)-43, DjnQj=726 Hz; and (S)-
3
arom H), 5.98 (brdd, J
17.1 Hz, 1H, 1-H), 4.11 (dd, J
A
G
43, DjnQj=693Hz.
3
3
A
Tungsten(0)-catalyzedallylic substitution
AHCTREUNG
Dimethyl 3-[2H1]-1-phenyl-2-butene-4,4-dicarboxylate and dimethyl
(Æ)-41/42: 13C NMR (75 MHz, 218C, TMS): d = 168.1 (C=O), 167.7, (C=
1-[2H1]-3-phenyl-1-butene-4,4-dicarboxylate
O), 140.1 (arom CH), 138.1 (C-2), 128.6, 127.8, 127.1 (arom CH), 116.5
(t, 1J(C,2H)=21.2 Hz, C-1), 57.3(C-3), 52.6 (CH 3), 52.4 (CH3), 49.7 (C-
A
Method A (asymmetric synthesis): An orange-red solution of [W(CO)3-
4); 2H NMR (46 MHz, CH2Cl2, 238C, CDCl3): d = 5.06 (brs, 1-2H); MS
(EI): m/z (%): 249 (2) [M]+, 207 (8), 190 (50), 183(15), 158 (10), 130
(30), 118 (59), 105 (35), 92 (9), 84 (100), 77 (12), 59 (11).
(h6-C7H8)] (24.4 mg, 0.068 mmol, 10 mol%) and (S)-4,5-dihydro-2-[2’-(di-
phenylphosphino)phenyl]-4-isopropyloxazole
(44.1 mg,
0.118 mmol,
17 mol%) in degassed, nitrogen-saturated anhydrous tetrahydrofuran
(60 mL), was heated under nitrogen to 608C for 15 min, resulting in a ho-
mogeneous brown-black solution. After cooling to 258C, solid sodium di-
methyl malonate (271 mg, 1.76 mmol) was added and the suspension was
stirred vigorously at 608C for 10 min. The resulting gray-black solution
was cooled to 258C and (E)-1-[2H1]-3-phenylprop-2-enyl diethyl phos-
phate (Æ)-38d (183mg, 0.678 mmol; 95% 2H) was added by micro sy-
ringe. After heating to 608C for 18 h, the deep red homogeneous solution
was quenched with 50% satd aq NH4Cl (50 mL) and the mixture was ex-
tracted with dichloromethane (50”3mL). The combined extracts were
dried (MgSO4) and then filtered through a short plug of silica gel. Evapo-
ration of the solvent afforded an oil that was purified by chromatography
on silica gel (100 g) with hexane/ethyl acetate 9:1 and then placed under
vacuum (0.1 mbar) to remove traces of dimethyl malonate. This proce-
dure afforded a 4:1 mixture of (R)-41/42 (E/Z 1:1) and (Æ)-43 as a color-
less oil (91 mg, 54%; 90% ee, 95% 2H).
1
(Æ)-43: H NMR (400 MHz, CDCl3, 228C, TMS): d = 7.38–7.18 (m, 5H,
arom. H), 6.47 (d, 3J
C
N
U
AHCTREUNG
(arom CH), 125.3(C-2), 52.5 (2”CH 3), 51.6 (C-4), 31.9 [t, 1J(C,2H)=
G
20.0 Hz, C-3]; 2H NMR (60 MHz, CH2Cl2, 238C, CDCl3): d = 2.75 (brs,
3-2H); MS (EI): m/z (%): 249 (32) [M]+, 215 (11), 205 (24), 189 (32), 158
(17), 129 (91), 118 (100), 105 (21), 92 (21), 84 (61), 77 (19), 59 (31).
Acknowledgements
We thank the EPSRC for grants No. GR/H 92067, GR/K07140, and GR/
NO5208 and EPSRC and GlaxoSmithKline for a CASE award to P.S. We
also thank Dr. R. Colman and Lancaster Synthesis for funding (L.G.).
G.C.L.-J. thanks the Royal Society of Chemistry for award of the Hickin-
bottom Fellowship. We also thank Dr. Carl A. Busacca, Boehringer In-
gelheim, CT, for a donation of the diamine required for the synthesis of
(S)-12d.
Method B [racemic synthesis of (Z)-isomer 41]: An orange-red solution
of [W(CO)3
(h6-C7H8)] (14.4 mg, 0.04 mmol, 11 mol%) and 2,2’-bipyridine
G
(6.4 mg, 0.04 mmol, 11 mol%) in anhydrous tetrahydrofuran (4.5 mL)
was heated under nitrogen to 608C for 15 min, resulting in a homogene-
ous brown-black solution. After cooling to 258C, solid sodium dimethyl
malonate (140 mg, 0.91 mmol) was added and the suspension was stirred
vigorously at 608C for 10 min. The resulting gray-black solution was
cooled to 258C and combined with (Z)-1-[2H1]-3-phenyl-prop-2-enyl
2
methyl carbonate (Æ)-37c (70 mg, 0.36 mmol, 98% H). After heating at
[1] For reviews, see: a) L. S. Hegedus, Transition Metals in the Synthesis
of Complex Organic Molecules, University Science Books, Mill
Valley, CA, 1994; b) C. G. Frost, J. Howarth, J. M. J. Williams, Tetra-
608C for 18 h, the deep red homogeneous solution was quenched with
50% satd aq NH4Cl (10 mL) and the mixture was extracted with di-
Chem. Eur. J. 2006, 12, 6910 – 6929
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6927

ˇ
´
P. Koc ovsky, A. V. Malkov, G. C. Lloyd-Jones et al.
hedron: Asymmetry 1992, 3, 1089; c) B. M. Trost, D. L. Van Vrank-
en, Chem. Rev. 1996, 96, 395; d) B. M. Trost, Acc. Chem. Res. 1996,
29, 355.
[19] For the method, see ref. [11b].
[20] The structure of (R)-(ꢀ)-12a was confirmed by single crystal X-ray
analysis (see the Supporting Information for details).
[21] C. A. Busacca, D. Grossbach, E. Spinelli, Tetrahedron: Asymmetry
2000, 11, 1907.
[2] The ret–inv mechanism has also been occasionally observed: a) I.
´
ˇ
´
´
Stary, P. Kocovsky, J. Am. Chem. Soc. 1989, 111, 4981; b) I. Stary, J.
ˇ
ˇ
´
Zajꢂcek, P. Kocovsky, Tetrahedron 1992, 48, 7229; c) H. Kurosawa, S.
Ogoshi, Y. Kawasaki, S. Murai, M. Miyoshi, I. Ikeda, J. Am. Chem.
Soc. 1990, 112, 2813; d) H. Kurosawa, H. Kajimaru, S. Ogoshi, H.
Yoneda, K. Miki, N. Kasai, S. Murai, I. Ikeda, J. Am. Chem. Soc.
1992, 114, 8417; e) M. E. Krafft, A. M. Wilson, Z. Fu, M. J. Procter,
[22] In this case, the PyCO2H/
[23] For the notation, see: a) P. Kocovsky, I. Stieborovꢃ, J. Chem. Soc.
N
Perkin Trans. 1 1987, 1969; b) P. Kocovsky, M. Pour, J. Org. Chem.
1990, 55, 5580.
[24] a) G. J. Kubas, Inorg. Chem. 1983, 22, 692; b) G. J. Kubas, L. S. van
der Sluys, Inorg. Synth. 1990, 28, 29; c) G. J. Kubas, Inorg. Synth.
1991, 27, 4.
ˇ
O. Dasse, J. Org. Chem. 1998, 63, 1748; f) C. N. Farthing, P. Kocov-
sky, J. Am. Chem. Soc. 1998, 120, 6661.
´
[3] a) B. M. Trost, M. Lautens, J. Am. Chem. Soc. 1982, 104, 5543;
b) B. M. Trost, M. Lautens, J. Am. Chem. Soc. 1983, 105, 3343;
c) B. M. Trost, M. Lautens, Organometallics 1983, 2, 1687; d) B. M.
Trost, M. Lautens, B. Peterson, Tetrahedron Lett. 1983, 24, 4525;
e) B. M. Trost, M. Lautens, J. Am. Chem. Soc. 1987, 109, 1469;
f) B. M. Trost, M. Lautens, Tetrahedron 1987, 43, 4817; g) B. M.
Trost, C. A. Merlic, J. Am. Chem. Soc. 1990, 112, 9590. For a mecha-
nistic study involving W0, see: h) J. Lehmann, G. C. Lloyd-Jones, Tet-
rahedron 1995, 51, 8863, and references therein.
[25] C7H8 = cycloheptatriene; a) for the complex preparation, see: F. A.
Cotton, J. McCleverty, J. E. White, Inorg. Synth. 1990, 28, 45; b) For
the numerous benefits of the analogous [(C7H8)W(CO)3] complex as
the catalyst for allylic substitution, see ref. [3h].
[26] a) We favor [(C7H8)Mo(CO)3] as it is more air-stable than
[(EtCN)3Mo(CO)3] and, according to our experience, somewhat
easier to prepare in a pure and defined form. Thus, while the prepa-
ration of the former complex is straightforward (reflux for 8 h, fol-
lowed by Soxhlet extraction),[3h] the latter complex is usually conta-
minated by [(EtCN)2Mo(CO)4] so that prolonged reflux (up to 3d)
and repeated recrystallization is often required to obtain the pure
species. The complex generated in situ from the latter contaminant
and ligand 12a is very slow in the catalytic reaction, as revealed by
control experiments {our results, given in Table 1, were obtained
with pure [(EtCN)3Mo(CO)3]}. b) This behavior seems to suggest
that a tridentate coordination of the metal by 12 is required to gen-
erate an active catalyst.
ˇ
´
ˇ
´
[4] D. Dvorꢃk, I. Stary, P. Kocovsky, J. Am. Chem. Soc. 1995, 117, 6130.
[5] a) J. W. Faller, D. Linebarrier, Organometallics 1988, 7, 1670;
b) Y. D. Ward, L. A. Villanueva, G. D. Allred, L. S. Liebeskind, J.
Am. Chem. Soc. 1996, 118, 897; c) M. E. Krafft, M. J. Procter, K. A.
Abboud, Organometallics 1999, 18, 1122.
[6] For the first successful asymmetric induction in the related, W0-cata-
lyzed allylic substitution and high level of enantioselection attained
with a W0-phosphinooxazoline catalyst, see: G. C. Lloyd-Jones, A.
Pfaltz, A. Angew. Chem. 1995, 99, 53 4;Angew. Chem. Int. Ed.
Engl. 1995, 34, 462.
[7] a) For an earlier observation by us of an asymmetric induction with
Mo–bisoxazoline complexes, see ref. [4]. Since this experiment re-
quired a high catalyst loading, it was only cited in a footnote.
b) Chiral a,a’-bipyridines exhibited low asymmetric induction: A. V.
Malkov, I. R. Baxendale, M. Bella, V. Langer, J. Fawcett, D. R.
[27] With NaCMe(CO2Me)2, the reaction proved to be much slower and
N
the selectivity lower, but still giving mainly the branched product.[8b]
[28] This outcome corresponds with the difference in A values: 2.8 (Ph)
and 1.68 (PhCH2) kcalmolꢀ1: E. L. Eliel, S. H. Wilen, L. N. Mander,
Stereochemistry of Organic Compounds, Wiley, New York, 1994,
p. 697.
[29] Unfortunately, the A value for iPr (2.21 kcalmolꢀ1) was measured at
278C, whereas those for Ph and PhCH2 were obtained at ꢀ100 and
ꢀ718C,[28] respectively, so that direct comparison is not straightfor-
ward.
[30] High selectivities were also observed for the 2-thienyl and 1-cyclo-
hexenyl analogues of 4 and 5;[8b] these results are in the same range
as the those reported by Trost.[10]
ˇ
´
Russel, D. J. Mansfield, M. Valko, P. Kocovsky, Organometallics
2001, 20, 673.
ˇ
´
[8] For preliminary communications of this work, see: a) P. Kocovsky,
ˇ
ˇ
A. V. Malkov, S. Vyskocil, G. C. Lloyd-Jones, Pure Appl. Chem.
ˇ
´
1999, 71, 1425; b) A. V. Malkov, P. Spoor, V. Vinader, P. Kocovsky,
Tetrahedron Lett. 2001, 42, 509.
[9] For a recent review, see: O. Belda, C. Moberg, Acc. Chem. Res.
2004, 37, 159.
[10] a) B. M. Trost, I. Hachiya, J. Am. Chem. Soc. 1998, 120, 1104;
b) B. M. Trost, S. Hildbrand, K. Dogra, J. Am. Chem. Soc. 1999, 121,
10416.
[31] Note that, for instance, a,a’-bipyridine and phenanthroline-type li-
gands are particularly suitable for Mo0- and W0-catalyzed allylic sub-
stitution. For selected examples, see the following: a) H. Frisell, B.
Škermark, Organometallics 1995, 14, 561; b) L. Eriksson, M. P. T.
Sjçgren, B. Škermark, Acta Crystallogr. Sect.
C 1996, 52, 77;
[11] For the original preparation of ligand 8, see: a) M. Mulqi, F. S. Ste-
phens, R. S. Vagg, Inorg. Chim. Acta 1981, 53, L91; for its previous
use in asymmetric catalysis, see: b) H. Adolfsson, C. Moberg, Tetra-
hedron: Asymmetry 1995, 6, 2023; c) C. Moberg, H. Adolfsson, K.
Wꢄrnmark, Acta Chim. Scand. 1996, 50, 195 and references therein.
[12] F. Glorius, A. Pfaltz, Org. Lett. 1999, 1, 141.
[13] a) O. Belda, N.-F. Kaiser, U. Bremberg, M. Larhed, A. Hallberg, C.
Moberg, J. Org. Chem. 2000, 65, 5868; b) N.-F. Kaiser, U. Bremberg,
M. Larhed, C. Moberg, A. Hallberg, Angew. Chem. 2000, 112, 3741;
Angew. Chem. Int. Ed. 2000, 39, 3596.
c) M. P. T. Sjçgren, H. Frisell, B. Škemark, P.-O. Norrby, L. Eriks-
son, A. Vitagliano, Organometallics 1997, 16, 942.
[32] Subsequent work involving use of enantiomerically enriched
(>90% ee) deuterated substrates and analysis of the solution-phase
structure and reactivity of deuterated samples of C by NMR, has
led to the conclusion that C is attacked “internally”, that is, via a re-
tention pathway, to give the major product enantiomer. With the
matched enantiomer of branched substrate the reaction would
therefore proceed with overall net retention, via a two-fold reten-
tion mechanism [a) G. C. Lloyd-Jones, S. W. Krska, D. L. Hughes, L.
Gouriou, V. D. Bonnet, K. Jack, Y. Sun, R. A. Reamer, J. Am.
Chem. Soc. 2004, 126, 702; b) D. L. Hughes, G. C. Lloyd-Jones, S. W.
Krska, L. Gouriou, V. D. Bonnet, K. Jack, Y. Sun, R. A. Reamer,
Proc. Natl. Acad. Sci. USA 2004, 101, 5379], in consonance with our
earlier observation of the ret–ret pathway.[4] This process could be
envisaged to occur by coordination of the nucleophile to the Mo,
followed by reductive elimination, or by an SN2’-like process involv-
ing a slipped h3 complex or a full h1 complex. The coordinative satu-
ration at the Mo in C suggests that the former is unlikely. The inter-
mediacy of a full h1 complex would not account for a memory effect
(see below). For a computational approach to the structure of mo-
lybdenum p-complexes, see the following: J. A. R. Luft, Z. X. Yu,
[14] D. L. Hughes, M. Palucki, N. Yasuda, R. A. Reamer, P. J. Reider, J.
Org. Chem. 2002, 67, 2762.
[15] a) B. M. Trost, K. Dogra, I. Hachiya, T. Emura, D. L. Hughes, S.
Krska, R. A. Reamer, M. Palucki, N. Yasuda, P. J. Reider, Angew.
Chem. 2002, 114, 2009; Angew. Chem. Int. Ed. 2002, 41, 1929;
b) S. W. Krska, D. L. Hughes, R. A. Reamer, D. J. Mathre, Y. Sun,
B. M. Trost, J. Am. Chem. Soc. 2002, 124, 12656.
[16] F. Glorius, M. Neuburger, A. Pfaltz, Helv. Chim. Acta 2001, 84,
3178.
[17] D. J. Ager, I. Parkash, Synth. Commun. 1996, 26, 3865.
[18] Yu. N. Belokon, L. K. Pritula, V. I. Tararov, V. I. Bakhmutov, Yu. T.
Struchkov, J. Chem. Soc. Dalton Trans. 1990, 1867.
6928
www.chemeurj.org
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2006, 12, 6910 – 6929

Asymmetric Allylic Substitution
FULL PAPER
ˇ
ˇ ˇ
´
Lloyd-Jones, S. Vyskocil, P. Kocovsky, J. Organomet. Chem. 2003,
D. L. Hughes, G. C. Lloyd-Jones, S. W. Krska, K. N. Houk, Tetrahe-
dron: Asymmetry 2006, 17, 716.
ˇ
´
687, 525; l) P. Kocovsky, J. Organomet. Chem. 2003, 687, 256.
[44] However, it should be noted that there is no direct correlation be-
tween the magnitude of the various kinetic resolution factors and
the overall degree of asymmetric induction with 12a (s=3), 12c (s=
2), and 8 (s=10).
[33] a) This result, first reported by us in a preliminary communicatio-
n,[8a] has now been confirmed by Trost and Hughes,[15b] who reported
77% ee (S), 43% yield, and 9:1 regioisomer ratio. The minor varia-
tion can be tentatively attributed to the slightly different reaction
conditions and different Mo precatalayst employed, that is,
[(EtCN)3Mo(CO)3] and [(C7H8)Mo(CO)3], respectively. b) The reac-
tion catalyzed by Mo·14 mirrors the result reported by Trost and
Hughes[15b] (90% ee, 72% yield, and 8:1 regioisomer ratio).
[34] The reaction of cinnamyl carbonate 4 with sodium dimethyl malo-
nate in the presence of [Mo]/19 proved to proceed with ordinary ki-
netics, namely with no induction period and no appreciable decom-
position or isomerization of products; the reaction was complete
within ca 5 h.
[35] The regioselectivities of the catalytic and stoichiometric (in [Mo]/19)
reaction turned out to differ marginally (83:17 vs 90:10); the catalyt-
ic version was faster.
[36] Cinnamyl carbonate 4 and the corresponding bromide provided
almost identical results (i.e., regioselectivity and yields) in stoichio-
metric (in [Mo]/19) reactions, (90:10, 36% and 89:11, 39%, respec-
tively). The catalytic reactions cannot be compared since cinnamyl
bromide undergoes a non-catalytic reaction, which provides solely
the linear product.
[45] J. Schwartz, J. A. Labinger, Angew. Chem. 1976, 88, 402; Angew.
Chem. Int. Ed. Engl. 1976, 15, 333.
[46] a) L. E. Overman, F. M. Knoll, Tetrahedron Lett. 1979, 20, 321. For
ˇ
´
an overview of Pd-catalyzed rearrangements, see: b) P. Kocovsky, I.
´
Stary, Rearrangements of Allylpalladium and Related Derivatives in
Handbook of Organopalladium Chemistry for Organic Synthesis,
Vol. II (Ed.: E. Negishi), Wiley, New York, 2002, p. 2011; c) H. Na-
kamura, Y. Yamamoto, Rearrangement Reactions Catalyzed by Pal-
ladium in Handbook of Organopalladium Chemistry for Organic
Synthesis, Vol. II (Ed.: E. Negishi), Wiley, New York, 2002, p. 2919.
[47] a) F. Toda, Y. Tohi, J. Chem. Soc. Chem. Commun. 1993, 1238;
b) this procedure involved enantioselective co-crystallization of 35
with enantiomerically pure a,a,a’,a’-tetraphenyl-2,2-dimethyl-1,3-di-
oxolan-4,5-dimethanol (TADDOL) from an aqueous-surfactant sol-
ution, followed by liberation of the free 35 in 85–95% ee by distilla-
tion. Repeated recycling using both enantiomers of TADDOL af-
forded both enantiomers of 35 in high ee and in reasonable yields.
[48] The ee values were determined by chiral GC or HPLC; the absolute
configurations of the corresponding carbonates 37c was determined
by optical rotation, which then confirms the absolute configuration
of the precursors and other products, e.g., the acetate.
[37] 1-Phenyl-but-1-en-3-yl acetate does not react with sodium dimethyl
malonate under molybdenum catalysis, showing that molybdenum
requires primary allylic esters.
[38] For a dramatic, beneficial effect on N-methylation of a-picolinic
amide ligands (related to these) in the copper-catalyzed Et2Zn con-
[49] This methodology was pioneered by Courtieu. For an excellent over-
view of the technique, see: M. Sarfati, P. Lesot, D. Merlet, J. Cour-
tieu, Chem. Commun. 2000, 2069.
ˇ
jugate addition to enones, see: A. V. Malkov, J. B. Hand, P. Kocov-
´
sky, Chem. Commun. 2003, 1948.
[50] The regio- and enantioselectivities obtained with the D-labeled sub-
strates are slightly lower than those obtained with the unlabelled
substrates 4 and 5. Experiments reported below, employing racemic
37c, demonstrate how selectivity in the mismatched manifold
changes with conversion. A possible origin of this effect is the pres-
ence of greater quantities of a CO source, for example, [Mo(CO)6],
in the initial stages of the reaction due to the pro-catalyst formation
under ambient temperature conditions. For a full discussion, see the
main text.
[51] It is important to note that the memory effect in the reaction of the
mismatched enantiomer of branched substrate is not unique to the
C1-symmetric bispicolinamide ligands described herein, but is also
prevalent in reactions catalyzed by C2-symmetric bispicolinamide li-
gands such as 8.
[39] The good reactivity of ligand 18 lends additional credence to mono-
deprotonation. Here, in contrast to the other ligands, the three ligat-
ing centers must take a meridional arrangement (mer) and that
could explain the lack of enantioselectivity, which is determined by
the turn in the ligand forming facial (fac) coordination of the three
ligating centers around metal.
[40] For the concept and leading references to non-linear relationships in
catalysis see: a) D. Guillaneux, S.-H. Zhao, O. Samuel, D. Rainford,
H. B. Kagan, J. Am. Chem. Soc. 1994, 116, 9430; b) C. Puchot, O.
Samuel, E. Dunach, S. Zhao, C. Agami, H. B. Kagan, J. Am. Chem.
Soc. 1986, 108, 2353; c) C. Girard, H. B. Kagan, Angew. Chem. 1998,
110, 3088; Angew. Chem. Int. Ed. 1998, 37, 2922; d) H. B. Kagan,
Synlett, 2001, 888; e) H. B. Kagan, Adv. Synth. Catal. 2001, 343, 227.
[41] K. L. Granberg, J.-E. Bꢄckvall, J. Am. Chem. Soc. 1992, 114, 6858.
[42] A distorted h3 geometry (toward h1) of the intermediate complex
may also be considered. For a discussion of the h1/h2 coordination in
the case of allylic complexes of Rh, see: a) D. N. Lawson, J. A.
Osborn, G. Wilkinson, J. Chem. Soc. 1966, 1733; b) P. A. Evans, J. D.
Nelson, J. Am. Chem. Soc. 1998, 120, 5581.
[43] For leading references to “memory effects” in Pd-catalyzed allylic
alkylations see: a) J. C. Fiaud, J. L. Malleron, Tetrahedron Lett. 1981,
22, 1399; b) B. M. Trost, N. R. Schmuff, Tetrahedron Lett. 1981, 22,
2999; c) B. M. Trost, R. C. Bunt, J. Am. Chem. Soc. 1996, 118, 23 5;
d) U. Burckhardt, M. Baumann, A. Togni, Tetrahedron: Asymmetry
1997, 8, 155; e) T. Hayashi, M. Kawatsura, Y. Uozumi, J. Am. Chem.
Soc. 1998, 120, 1681; f) G. C. Lloyd-Jones, S. C. Stephen, Chem. Eur.
J. 1998, 4, 2539; g) A. J. Blacker, M. L. Clarke, M. S. Loft, J. M. J.
Williams, Org. Lett. 1999, 1, 1969; h) J. M. Longmire, B. Wang, X.
Zhang, Tetrahedron Lett. 2000, 41, 5435; i) G. C. Lloyd-Jones, S. C.
[52] For isomerizatin of (p-allyl)palladium complexes via h1 intermedi-
ate, see: B. M. Trost, F. D. Toste, J. Am. Chem. Soc. 1999, 121, 4545.
[53] G. C. Lloyd-Jones, A. Pfaltz, Z. Naturforsch. 1995, 50b, 361.
[54] H. Brunner, R. Kroiss, M. Schmidt, H. Schonenberg, Eur. J. Med.
Chem. 1986, 21, 333.
[55] C. Betschart, L. S. Hegedus, J. Am. Chem. Soc. 1992, 114, 5010.
[56] H. Brunner, B. Nuber, M. Prommesberger, J. Organomet. Chem.
1996, 523, 179.
[57] G. C. Cox, L. M. Harwood, Tetrahedron: Asymmetry 1994, 5, 1669.
[58] Carbamate 39 is formed as a 1:1 mixture of diastereoisomers. How-
ever, since the chiral center at C-1 in the cinnamyl moiety is “low
grade”, only the diastereotopic H nuclei adjacent to that center dis-
play significant magnetic nonequivalence due to diastereotopicity.
Thus the 1H, 13C, and 2H signals are reported as though 39 were a
single compound, except for the H at the chiral center.
ˇ
ˇ
ˇ
´
ˇ
Stephen, M. Murray, C. P. Butts, S. Vyskocil, P. Kocovsky, Chem.
Eur. J. 2000, 6, 4348; j) I. J. S. Fairlamb, G. C. Lloyd-Jones, S. Vysko-
cil, P. Kocovsky, Chem. Eur. J. 2002, 8, 4443; k) L. Gouriou, G. C.
Received: December 15, 2005
Published online: June 29, 2006
ˇ
ˇ
´
Chem. Eur. J. 2006, 12, 6910 – 6929
ꢁ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
6929