Synthesis of Citronellal by RhI-Catalysed Asymmetric Isomerization of N,N-Diethyl-Substituted Geranyl- and Nerylamines or Geraniol and Nerol in the Presence of Chiral Diphosphino Ligands, under Homogeneous and Supported Conditions

Article abstract of DOI:10.1002/1522-2675(20010131)84:1<230::AID-HLCA230>3.0.CO;2-V

For the asymmetric isomerization of geranyl- or neryldiethylamine (( E)- or(Z)-1, resp.) and allyl alcohols geraniol or nerol ((E)- or (Z)-2, resp.) to citronellal (4) in the presence of a [Rh^I(ligand)cycloocta-1,5-diene)]^- catalyst, the atropic ligands 5-11 are compared under homogeneous and polymer-supported conditions with the non-C2-symmetrical diphosphino ferrocene ligands 12-16. The ^tBu-josiphos ligand 13 or daniphos ligand 19, available in both antipodal series, already catalyse the reaction of (E)-1 at 20 deg (97 percent e.e.) and favourably compare with the binap ligand 5 (see Table 1). Silica-gel- or polymer-supported diphosphino ligands usually afford similar selectivity as compared to the corresponding ligands applied under homogeneous conditions, but are generally less reactive. In this context, a polymer-supported ligand of interest is the polymer-anchored binap (R)-6, in terms of reactivity, selectivity, and recoverability, with a turnover of more than 14400.

Full text of DOI:10.1002/1522-2675(20010131)84:1<230::AID-HLCA230>3.0.CO;2-V

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
230
Helvetica Chimica Acta ± Vol. 84 (2001)
Synthesis of Citronellal by RhI-Catalysed Asymmetric Isomerization
of N,N-Diethyl-Substituted Geranyl- and Nerylamines or Geraniol and Nerol
in the Presence of Chiral Diphosphino Ligands, under Homogeneous
and Supported Conditions1)
by Christian Chapuis*, Michel Barthe, and Jean-Yves de Saint Laumer
Firmenich SA, Corporate R&D Division, P.O. Box 239, CH-1211 Geneva 8
For the asymmetric isomerization of geranyl- or neryldiethylamine ((E)- or (Z)-1, resp.) and allyl alcohols
‡
geraniol or nerol ((E)- or (Z)-2, resp.) to citronellal (4) in the presence of a [RhI(ligand)cycloocta-1,5-diene)]
catalyst, the atropic ligands 5 ± 11 are compared under homogeneous and polymer-supported conditions with the
t
non-C2-symmetrical diphosphino ferrocene ligands 12 ± 16. The Bu-josiphos ligand 13 or daniphos ligand 19,
available in both antipodal series, already catalyse the reaction of (E)-1 at 208 (97% e.e.) and favourably
compare with the binap ligand 5 (see Table 1). Silica-gel- or polymer-supported diphosphino ligands usually
afford similar selectivity as compared to the corresponding ligands applied under homogeneous conditions, but
are generally less reactive. In this context, a polymer-supported ligand of interest is the polymer-anchored binap
(R)-6, in terms of reactivity, selectivity, and recoverability, with a turnover of more than 14400.
Introduction. ± Isomerization of diethylgeranylamine ((E)-1) for the production of
practically optically pure (‡)-citronellal ((R)-4; 98 ± 99% e.e.) in the presence of ( )-
(S)-binap ((S)-5)/RhI is an exceptional industrial process [1] (binap ˆ [1,1'-binaph-
thalene]-2,2'-diylbis[diphenylphosphine], leading to key intermediates for the fra-
grance and flavour industry. The ene cyclization of (R)-4, affording ( )-isopulegol [2],
is certainly the best industrial process [3] for the preparation of ( )-menthol [4], as
well as for the synthesis of the insect repellent (‡)-cis-p-menthane-3,8-diol [5]. This
methodology has also been successfully extended to the preparation of (‡)-7-hydroxy-
or (‡)-7-methoxycitronellal [6]. The acidic cyclization of the ( )-antipode (S)-4 to
(‡)-(1R,6S)-trans-dihydrocyclocitral [7]2) is also fundamental for the synthesis of (‡)-
norlimbanol [10]3), as well as of optically pure dihydroionone [7] and dihydrodamas-
cone analogues [12a]. ( )-Citronellal ((S)-4) has also been used as a potential starting
material for the synthesis of ( )-(2S,4R)-rose oxide [12b,c].
For the RhI-catalysed isomerization process (E)- or (Z)-1 ! 4, several variants of
binap ligands have been tested and patented by Takasago [13], Hoffmann La Roche
[14], and Bayer [15], but the technical know-how of the Japanese company, which is
reluctant to offer both enantiomers of 4 in bulk quantities, has maintained its
monopoly. The efficiency of the binap-containing catalyst was earlier attributed to the
Presented at the ꢀ13th International Conference on Organic Synthesisꢁ, 1-5.07.2000, Warsaw, Poland.
For the racemic version, see [8]. See also [9] for a report that ignores the priority of [7][10][11].
Norlimbanol is the trivial name for the racemic form of 1-[(1R,6S)-2,2,6-trimethylcyclohexyl]hexan-3-ol,
an extremely strong woody fragrance manufactured by Firmenich SA in both racemic and optically active
form [10].
1
)
)
)
2
3

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Helvetica Chimica Acta ± Vol. 84 (2001)
231
high Lewis acidity of the central metal atom conferred by the fully aromatic structure of
the diphosphino ligand [1d]. Finally, this methodology is also advantageous when
compared to the asymmetric hydrogenation of citral, which necessitates a tedious
distillative fractionation to obtain the stereoisomerically pure substrates4), geranic acid
[17]5) or geraniol [19]. These last substrates require, in addition, a supplementary
reductive or oxidative step. We now wish to present the first example of the use of a
chiral diphosphino ligand, neither C2-symmetric nor atropic, which enables the RhI-
catalysed preparation of 4 in both antipodal forms from (E)- or (Z)-1 (or from (E)- or
(Z)-2) with similarly high enantioselectivity, under homogeneous or silica-gel- or
polymer-supported conditions (Scheme).
Scheme
i) [RhI (diphosphino ligand) (cod)]CF3SO3, THF, 668. ii) AcOH, H2O, Et2O.
Results. ± First of all, we compared the activity and selectivity of the (‡)-(R)-binap
ligand (R)-5 with its commercially available polymer-supported analogue (R)-66),
‡
using triflate as the counter ion of Rh ([RhI(ligand)(cod)]CF3SO3; cod ˆ cycloocta-
1,5-diene), an anion not mentioned in the Japanese patent [3e,f]. Under these
conditions, the ligand (R)-6 exhibited very similar reactivity and selectivity, even at
0.25 mol-% of the catalyst (i.e. of the ligand) for both diethylgeranylamine ((E)-1) [21]
and diethylnerylamine ((Z)-1) [22]7) when compared to (R)-5 used under homoge-
neous conditions (see Table 1, Entries 1 and 2). The ratio (R)-6/(E)-1 could even be
4
)
By analogy with [16], a 95 :5 (E)/(Z) mixture of geranial was more selectively hydrogenated at
90 atm for 16 h in toluene, in the presence of 1 mol-% of (R,R)-chiraphos (ˆ(2R,3R)-(‡)-2,3-
bis(diphenylphosphino)butane ˆ [(1R,2R)-1,2-dimethylethane-1,2-diyl]bis[diphenylphosphine]; Aldrich)
chelated with 0.25 mol-% of [Rh4(CO)12] to afford quantitatively (S)-4 with 83% e.e., while under the
same conditions, a 7:93 (E)/(Z) mixture of neral delivered quantitatively (R)-4 with 84% e.e. The
practically similar reactivities of the (E)- and (Z)-isomers did not allow an efficient dynamic hydrogenation
of citral under (E)/(Z) isomerization conditions.
For an asymmetric 1,4-reduction of ethyl geranate, see [18].
The ligand (R)-6 is not included in the general structure claimed in [3e] and has only been previously
reported for asymmetric hydrogenation and aldol or Mannich reactions [20].
5
6
)
)
7
)
For a detailed mechanistic study of this isomerization, see [1d][23].

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
232
Helvetica Chimica Acta ± Vol. 84 (2001)
decreased to 0.1 mol-% without affecting either the conversion or the enantioselec-
tivity8). Thus, hydrolysis of the transient enamine (S)-3 and bulb-to-bulb distillation
afforded (S)-4 in 98% e.e. and 95% isolated yield. The great advantage of a polymer-
supported ligand is its straightforward physical separation from the substrate. Using a
0.25 mol-% ratio, we could thus successfully recycle (R)-6 37 times by simple
decantation/filtration techniques, without any loss of chiral and chemical efficiency.
The isomerization of allyl alcohols was earlier reported [25] to proceed with 37%
e.e. for geraniol ((E)-2) in the presence of [RhI(binap)(cod)]ClO4 [23]. We were thus
surprised to reach 60% enantioselectivity for the same ligand, but with triflate as
counterion. The selectivity decreased to 50% e.e. for nerol ((Z)-2), accompanied by a
lower reactivity (88 ± 90%) as compared to the complete conversion observed for the
corresponding amines (E)- and (Z)-1 (Table 1, Entries 1 and 2). The reactivity still
decreased in the case of the polymer-supported binap (R)-6, to 46 and 72% conversion
for (E)- and (Z)-2, respectively; however, the selectivity remained identical with that
obtained under homogeneous conditions. The fact that traces of (E)/(Z) isomerization
(3 ± 5%) were observed after analysis of the unreacted allyl alcohols 2 partially
accounts for the lower selectivity observed with these substrates.
8
)
The enantioselectivity was determined directly by chiral capillary GC analysis (50% 2,3-di-O-acetyl-6-O-
[(tert-butyl)dimethylsilyl]-b-cyclodextrin (Brechbuehler SA), 25 m, 0.25 mm) of the enamine (608, 1.58/
min, 1008 for 10 min, 70 KPa iso, 0.92 ml/min, 25.2 cm/s He tR 32.96 min for (S)-3 and 33.33min for (R)-3)
or of citronellal (1008 for 36 min, 158/min, 1508 for 5 min, 75 KPa for 36 min, 20 KPa/min, 170 KPa for
3 min, 0.8 ml/min, 24.9 cm/s He; tR 33.1 min for (S)-4 and 34.1 min for (R)-4) [24]. Standard error Æ1%.

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Helvetica Chimica Acta ± Vol. 84 (2001)
233
Table 1. Screening of the Substrates Diethylgeranylamine ((E)-1), Diethylnerylamine ((Z)-1), Geraniol ((E)-2,
and Nerol ((Z)-2) in the Presence of Diphosphino Ligands 5 ± 16 (1.0 mol-%) and [RhI(cod)2]CF3SO3
(1.0 mol-%) in Refluxing THF: Conversion [%] to 4 after 20 h, Chiroptical Property of 4 as well as Its e.e.
Entry
Ligand
Conversion (sign of aD of 4; e.e. of 4)
from (E)-1
from (Z)-1
from (E)-2
from (Z)-2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
(R)-5
(R)-6
(S)-7
(S)-8
(S)-9
(S)-10
(S)-11
(R,S)-12
(R,S)-13
(R,S)-14
(R,S)-15
(R,S)-16
(R,S)-17
(R,S)-18
(R,S)-19
99 (( ); 97)a)
100 (( ); 98)a)
100 ((‡); 97)a)
95 ((‡); 91)
100 ((‡); 95)a)
100 ((‡); 98)a)
100 (( ); 97)a)
15 (( ); 91)
99 (( ); 95)
40 (( ); 95)
88 (( ); 60)
46 (( ); 61)
75 ((‡); 44)
90 ((‡); 51)
72 ((‡); 50)
85 (( ); 32)
95 ((‡); 89)
74 ((‡); 51)
78 (( ); 37)
79 ((‡); 88)a)
48 ((‡); 96)a)
99 (( ); 78)a)
99 (( ); 92)a)
81 (( ); 83)
19 (( ); 96)a)
100 ((‡); 86)a)
100 ((‡); 97)a)
68 ((‡); 88)
62 ((‡); 9)
85 (( ); 22)
63 (( ); 21)
72 (( ); 20)
62 (( ); 4)
70 (( ); 16)
96 ((‡); 31)
54 ((‡); 20)
95 ((‡); 21)
19 ((‡); 11)
81 (( ); 96)a)
54 ((‡); 78)
99 ((‡); 96)
30 (( ); 81)a)
44 (( ); 76)
99 (( ); 94)a)
99 (( ); 96)a)
a) In the presence of 0.25 mol-% of catalyst.
We then turned our attention towards the known (S)-MeO-biphep ligand (S)-7
[14], since its MeO groups are potential anchors for the attachment to polymeric
supports. First of all, (S)-7 was quantitatively deprotected (BBr3, CH2Cl2 (99%) [26])
to afford, without racemization, the new diphosphinobiphenyldiol (S)-8, prior to
refunctionalization, either to the trimethylsilyl(Me3Si)-protected ether (S)-9 (1-
(trimethylsilyl)-1H-imidazole and CH2Cl2 [27], or Me3SiCl, Et3N, and CH2Cl2 [28]
(99%)), or to the TentaGel ꢂ-supported [29] ester (S)-10 (TentaGel ꢂ S-COOH
dicyclohexylcarbodiimide (DCC), N,N-dimethylpyridin-4-amine (DMAP), CH2Cl2
[30] (77%)) or ether (S)-11 (BuLi, THF, TentaGel ꢂ S-Br (72%)).
The MeO-biphep ligand (S)-7 was as efficient as binap for the isomerization of the
allylamines, and slightly less reactive and selective for that of the allyl alcohols (Table 1,
Entry 3), while under homogeneous conditions, the deprotected diphosphinobiphenyl-
diol (S)-8 and its trimethylsilyl ether (S)-9 gave lower conversions and selectivities
(89 ± 95% e.e.; Entries 4 and 5). This may be tentatively explained by modification of
either the metal bite angle of the chelate and/or the C(2) C(1) C(1') C(2') dihedral
angle of the diphosphino ligand, the importance of which was earlier demonstrated in
the case of asymmetric hydrogenations [31]. With (E)-1, the TentaGel ꢂ-supported
ester (S)-10 gave a similar selectivity (88% e.e.) as (S)-9, while its ether analogue (S)-
11 was much more selective, yielding 96% e.e., as the MeO-biphep (S)-7 (Entries 6
and 7); however, at an identical concentration, (S)-10 and (S)-11 were much less
reactive than the analogous ligands (S)-7 and (S)-9 used under homogeneous
conditions.
Since the results earlier reported with diphosphinoferrocenes were far from
promising (25% e.e. [1a]), we were surprised and gratified to observe that this class of
ligands emerged positively from a screening of more than 60 different diphosphino