An optimised and recoverable tartrate surrogate for sharpless asymmetric epoxidations

Article abstract of DOI:10.1016/j.tetlet.2008.10.069

The tetrahydroxy diester 7d (R = ⁱPr) is almost as effective as diisopropyl tartrate in SAE reactions of (E)-allylic alcohols and can be recovered and re-used following a relatively simple work-up procedure.

Full text of DOI:10.1016/j.tetlet.2008.10.069

Tetrahedron Letters 50 (2009) 35–38
Contents lists available at ScienceDirect
Tetrahedron Letters
journal homepage: www.elsevier.com/locate/tetlet
An optimised and recoverable tartrate surrogate
for sharpless asymmetric epoxidations
*
David W. Knight , Ian R. Morgan
School of Chemistry, Cardiff University, Main College, Park Place, Cardiff, CF10 3AT, UK
a r t i c l e i n f o
a b s t r a c t
The tetrahydroxy diester 7d (R = iPr) is almost as effective as diisopropyl tartrate in SAE reactions of
(E)-allylic alcohols and can be recovered and re-used following a relatively simple work-up procedure.
Ó 2008 Elsevier Ltd. All rights reserved.
Article history:
Received 25 September 2008
Revised 6 October 2008
Accepted 14 October 2008
Available online 18 October 2008
The Sharpless asymmetric epoxidation (SAE) reaction (Scheme
1) is now very much a part of modern synthetic strategy, despite
its seemingly restrictive limitation to (E)-allylic alcohols 1, in terms
of being able to deliver useable enantiomeric enrichments in the
product epoxy-alcohols 2. Due to the imagination of many
synthetic chemists, naturally led by the Sharpless group, this
restriction has been significantly obviated, such that this transfor-
mation now resides amongst the ‘classics’ of synthetic organic
chemistry.1
In its original form, the reaction was not without its drawbacks,
one of which was the need to remove the stoichiometric amount of
tartrate ester, used to form the complex which delivers the oxygen
to the allylic alcohol 1 in an asymmetric manner. This was largely
negated by the introduction of the ‘catalytic’ procedure, the es-
sence of which is the incorporation of molecular sieves to remove
any adventitious water.2 In contrast to many other catalytic proce-
dures,3 little effort has been made to develop a procedure which al-
lows for the recovery of tartrate from an SAE, presumably because
of its ready availability and low cost, as well as its relatively benign
nature.4,5 However, in these more environmentally aware times,
such recoveries have become highly desirable, even essential. In
any event, large-scale SAEs would benefit considerably from such
a procedure, as disposal of impure tartrate by-products would be
inconvenient and expensive; the absence of a tartrate ester hydro-
lysis step would also be an advantage, especially when the synthe-
sis produced an especially sensitive epoxide.
Detailed studies by the Sharpless group have reached the con-
clusion that the key initial moiety in SAEs is the dimeric tartrate
complex 3 (Scheme 2).6 Incorporation of the reacting species, the
allylic alcohol and the hydroperoxide, then lead to the reactive
complex 4 in which the oxygen transfer occurs.
A particularly relevant series of experiments reported by Sharp-
less in support of these ideas was the successful use of a series of
dimeric tartrate surrogates 5, which presumably were incorporated
into complexes 6, but only when the connecting bridge was long
enough (Scheme 3).7
E
E
OiPr
Ti
OiPr
OiPr
O
Ti
PriO
O
O
O
O
PriO
O
O
O
O
R1
OH
Ti
E
E
Ti
E
OiPr
OEt
O
R O
R'
OH
O
O
R
E = CO2Et
O
O
EtO
EtO
[redundant ester ?]
3
4
Scheme 2.
( )n-1
O
R1
R1
O
O
SAE
OR
RO
R2
OH
R2
OH
OH
O
O
OH
CO2R
RO
O
O
O
O
O
R3
R3
Ti
O
Ti
O
( )n
RO2C
O
O
OR
OR
1
2
OH
OH
5
Scheme 1.
O
RO
6
* Corresponding author.
E-mail address: knightdw@cardiff.ac.uk (D. W. Knight).
Scheme 3.
0040-4039/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2008.10.069

36
D. W. Knight, I. R. Morgan / Tetrahedron Letters 50 (2009) 35–38
Throughout, such connecting ester groups in complexes 6, as
0.42 eq. Ti(OiPr)4, 0.21 eq. 7
2.tBuOOH, 4Å MS, CH2Cl2
-20 oC, 15 h.
O
well as those designated ‘E’ in the ‘standard’ complexes 3 and 4
(Scheme 2), appear to be ‘bystanders’, in the sense that they do
not seem to participate directly in the epoxidation process,
although Sharpless has stipulated that all four ester groups are
necessary for the kinetic resolution of racemic allylic alcohols. In
view of this, we reasoned that the simpler diesters 7 could act as
tartrate surrogates for ‘standard’ SAEs. These should have greater
stability than the ‘dimeric’ esters 5, especially with respect to
transesterification7 and also, being tetra-ols, perhaps possess suffi-
ciently unique solubilities, features that in combination might ren-
der them amenable to recovery and recycling.
Ph
OH
Ph
OH
11
12
Scheme 6.
Table 1
SAE of cinnamyl alcohol 11 using ligand 7 (dietheyl ester)
O
Scheme 6
Ph
OH
Ph
OH
OH
OH
CO2Et
OH
OH
CO2R
a) n = 3
b) n = 4
c) n = 5
d) n = 6
e) n = 8
( )n
11
12
EtO2C
( )n
OH
OH
RO2C
7; R = Et
OH
OH
7
n
mol % 7
% ee 12
34
4
5
6
8
15
17
21
21
21
21
12
92
95
96
96
99
Based on the Sharpless results (Scheme 3), it seemed likely that
we would require a connecting chain of at least six atoms [i.e., 7;
n > 3]; conversely, too long a chain might render the necessary
complexes insufficiently stable. Requiring a very rapid asymmetric
synthesis of the diesters 7, we thought that the best option was to
generate the two diol units using the other great Sharpless inven-
tion, the AD-reaction.8 We therefore required a series of bis-unsat-
urated esters 8 (Scheme 4).
Initially, we used a variety of highly (E)-selective Wittig-type
homologations of the corresponding dialdehydes,9 but in all cases,
irritatingly small quantities of (E,Z)- and/or (Z,Z)-isomers proved
refractory to either removal or isomerisation. This problem was
only obviated when we turned to the Grubbs’ metathesis meth-
od,10 the results of which were remarkable. Using only 1 mol % of
the second generation catalyst, reactions between a cycloalkene
D-DIPT
alcohol 12 with around 99% ee1 and because this product is quite
sensitive and hence, literally, something of an ‘acid test’. In these
initial studies, the tartrate surrogates used were the ethyl esters
of structures 7 (see below).
We were delighted to find that our first set of experiments met
with success. The results are summarized in Table 1.13 With the
exception of the smallest diethyl diester ligand [7; n = 3], all gave
excellent levels of enantiomeric enrichment in the product
epoxy-alcohol 12. In all cases, the conversion of cinnamyl alcohol
11 was >95% after 15 h reaction time. We therefore concluded that,
for this epoxidaton at least, the diester ligand should have a mini-
mum value for ‘n’ of 6, when it was almost as effective as the ori-
ginal diisopropyl tartrate [D-DIPT; Table 1]. Of course, the failure of
diester [7; n = 3] to produce high ees lends further weight to the
veracity of the intermediate dimeric complexes 4 (Scheme 2).
Various optimisation experiments using the apparent optimal
eight-methylene ligand [7; n = 6] showed that it worked just as
well at a level of 10 mol % and gave slightly increased ees at
À10 °C. Lower levels of ligand and increased temperatures all gave
lower ees. It was also self-evident that using titanium iso-propox-
ide with the diethyl diester [7; R = Et] was simply illogical, espe-
cially in view of the hoped-for long-term survival of such tartrate
surrogates. A brief study of alternative ester groups showed that
use of the di-iso-propyl ester [7d; R = iPr] made no difference to
the ee value of epoxy-alcohol 12, while the corresponding di-
methyl ester was much less effective, probably because of its much
reduced solubility, an encouraging sign for possible ligand recov-
ery. We therefore chose to focus entirely on the di-iso-propyl ester
of the eight-methylene ligand [7d; R = iPr] and examine SAE reac-
tions of other representative allylic alcohols: (E)-2-hexen-1-ol 13,
prenyl alcohol 14 and geraniol 15. The results are collected in Table
2.
9 (or the corresponding 1,x-diene) and just over 2 equiv of an acry-
late 10 in gently refluxing dichloromethane overnight delivered in
all cases P85% isolated yields of the diesters 8, solely as the (E,E)-
isomers. Further optimisation, using scrupulously clean reagents
and solvent, revealed that catalyst levels as low as 0.13 mol % were
equally effective, a useful finding in view of both catalyst cost and
its complete removal (Scheme 5).11
Guided by literature precedent,8 the final double bis-hydroxyl-
ation was carried out using (DHQD)2PHAL and it delivered excel-
lent yields of the diesters 7, following crystallisation from
chloroform–hexane mixtures. Each had an ee value of >99%,
according to both chiral HPLC and GC analyses.12
In an initial screening of the central idea, we chose to use cin-
namyl alcohol 11 as the substrate both because ‘standard’ SAE con-
ditions (Scheme 6) are known to deliver the corresponding epoxy-
( )n
RO2C
CO2R
( )n
7
8
9
Scheme 4.
When run at À20 °C, SAE of the hexenol 13, as outlined in
Scheme 6 and using ligand 7d as its di-iso-propyl ester, gave essen-
tially the same result, in terms of both chemical and optical yields,
as is obtained using the ‘usual’ SAE conditions with di-iso-propyl
tartrate.1 The more sterically demanding terpenoid alcohols 14
and 15 also gave very similar chemical yields, which lowered
mainly because of product volatility, but ees were some 10% lower
than those obtained using tartrate. The latter two substrates were
( )n
0.13 mol% Grubbs'
Mark II catalyst
9
+
( )n
RO2C
CO2R
CH2Cl2, 40 oC, 16 h
2.
CO2R
10
8
Scheme 5.

D. W. Knight, I. R. Morgan / Tetrahedron Letters 50 (2009) 35–38
37
Table 2
11 as the substrate, indicating that little or no degradation of the
ligand had occurred. Higher yields of ligand recovery were possible
if the various aqueous washings were back-extracted and crystal-
lisation mother liquors processed further. Most likely, on larger
scales, ligand returns would be higher.
Perhaps as expected and in common with the ‘standard’ tartrate
protocols, the ligands 7 gave much poorer results with both (Z)-
allylic and homoallylic alcohols in terms of product ees. However,
these did turn out to be very useful for the kinetic resolution of a
selection of allylic alcohols. Interestingly, the related ligands 5
(Scheme 3) used by Sharpless were relatively ineffective in such
resolutions,7 although this is not an entirely fair statement, as tita-
nium(IV) tetra-t-butoxide rather than iso-propoxide was used; this
subsequently proved to be usually less capable of participating in
this type of kinetic resolution15 (Scheme 7).
SAE of allylic alcohols 13–15 using ligand 7d [R = iPr]
O
OH
13
OH
16 [41%; 88%ee]
Scheme 6
O
OH
OH
14
17 [34%; 70% ee]
O
OH
OH
15
18 [51%; 72% ee]
OH
OH
iPrO2C
CO2 Pr
i
( )6
In our experiments, we used the now-standard ligand [7d;
R = iPr] and 0.6 equiv of cumene hydroperoxide [CHP], under other-
wise normal SAE conditions. Starting with an allylic alcohol 19 un-
der these conditions, if the resolution were to be successful, the
products should be predominantly the epoxy-alcohols 20, together
with a smaller amount of isomers 21 and unreacted starting mate-
rial 22, which should possess high levels of optical activity. In the
event, such resolutions worked very well; the results are collected
in Table 4.
Yields of the recovered allylic alcohols 24, 26, 28, 30, 32 and 34
were routinely in the region 35–45%, with losses usually due to
volatility. The ees of the recovered alcohols13 shown in Table 4
are followed in parenthesis by the related values obtained using
the Sharpless procedure.15 Overall, these results indicate that the
present system is very suited to the kinetic resolution of sterically
unencumbered allylic alcohols, with ees of the recovered alcohols
only becoming lower in the case of the cyclohexyl-substituted sub-
strate 23, wherein it appears that the presence of a bulky substitu-
OH
7d [R = iPr]
OH
not tested using the Sharpless dimeric tartrate surrogates 5
(Scheme 3).7 Possibly, the greater steric bulk of these two alcohols
may interact adversely with the connecting bridge in the dimeric
complex necessary to secure very high ee values (see Schemes 2
and 3).
Having established the viability of ligands 7 in SAE reactions, we
then turned to the central premise, that of their recovery and re-
use. It soon became clear that this was not going to be easily
achieved. Hydrolysis of the titanium complexes without inducing
a similar reaction of the ligand ester groups and their epimerisa-
tion were continual problems, as was formation of intractable
complexes of the ligand 7 and the liberated titanium salts; the poor
solubility of ligands 7 contributed to this. However, using aqueous
citric acid14 to break up such complexes did allow relatively
straightforward product separation. An optimum protocol was as
follows: a completed SAE, maintained at À20 °C, was quenched
using aqueous citric acid, and all organic material was extracted
into ethyl acetate. The combined extracts were washed with 2 M
aqueous potassium carbonate and water, then dried and concen-
trated. The residue was taken up in a minimum of warm chloro-
form, and this solution was added slowly to vigorously stirred
pentane (12 volumes). The precipitated ligand was filtered off,
and the filtrate and washings were evaporated to leave the crude
epoxy-alcohol. The ligand was then taken up into ethyl acetate
and the resulting solution was washed sequentially with aqueous
2 M sodium hydroxide, 2 M hydrochloric acid, aqueous potassium
carbonate, water and brine. Evaporation of the dried solution then
left ligand 7 of sufficient purity that its subsequent use maintained
the expected ee levels of the resulting epoxy-alcohols; it could also
be crystallised from CHCl3–hexane mixtures. The results obtained
from a typical set of runs on a 1–2 mmol scale are collected in Ta-
ble 3; each epoxidation was carried out as summarised in Scheme
6. Yields and ees were comparable throughout the four runs with
similar runs using fresh ligand. Especially encouraging was the
very high ee obtained in the fourth run using cinnamyl alcohol
OH
O
R1
R2
R1
R2
OH
OH
20
Ti(OiPr)4, 7d [R = iPr]
R2
R1
R2
0.6 eq. CHP, CH2Cl2
20 h,-20 oC
OH
19
22
O
R1
21
Scheme 7.
Table 4
Kinetic resolutions of allylic alcohols 19
OH
OH
OH
OH
23
24 [72% (>96)]
25
OH
26 [>99% (89)]
OH
OH
OH
Bu
Bu
Pr
Pr
27
28 [90% (-)]
29
OH
30 [>99 (-)]
Table 3
Ligand recovery using diester 7d [R = iPr]
OH
OH
OH
Allylic
alcohol
Epoxide
yield (%)
Epoxide
ee (%)
Ligand
yield (%)
Ligand mp
13
13
11
11
63
47
50
76
86
84
82
97
82
77
82
81
100–102 oC
100–101 oC
99–101 oC
98–100 oC
31
32 [95% (90)]
33
34 [>99% (-)]

38
D. W. Knight, I. R. Morgan / Tetrahedron Letters 50 (2009) 35–38
Polym. Chem. 2000, 38, 161; Xiang, S.; Zhang, Y.; Xin, Q.; Li, C. Angew. Chem., Int.
Ed. 2002, 41, 821; Dickerson, T. J.; Reed, N. N.; Janda, K. D. Chem. Rev. 2002, 102,
3325.
ent adjacent to the hydroxy group lowers the level of differentia-
tion. However, similar increases in bulk at the opposite end of
the allylic alcohol function [e.g., 33] appear to have little effect.
This is consistent with the proposed transition state geometries
(Schemes 2 and 3), wherein one would expect substituents adja-
cent to the alcohol group to interact more significantly with the
bulk of the connecting methylene chain than similar groups when
attached to the end of the alkene function.
5. Reed, N. N.; Dickerson, T. J.; Boldt, G. E.; Janda, K. D. J. Org. Chem. 2005, 70, 2042
and references cited therein.
6. Woodard, S. S.; Finn, M. G.; Sharpless, K. B. J. Am. Chem. Soc. 1991, 113, 106;
Finn, M. G.; Sharpless, K. B. J. Am. Chem. Soc. 1991, 113, 113. Theoretical
calculations support this conclusion Potvin, P. G.; Bianchet, S. J. Org. Chem.
1992, 57, 6629; Wu, Y.-D.; Lai, D. K. W. J. Am. Chem. Soc. 1995, 117, 11327; Wu,
Y.-D.; Lai, D. K. W. J. Org. Chem. 1995, 60, 673.
7. Carlier, P. R.; Sharpless, K. B. J. Org. Chem. 1989, 54, 4016.
8. Kolb, H. C.; VanNieuwenhze, M. S.; Sharpless, K. B. Chem. Rev. 1994, 94, 2483;
Sharpless, K. B. J. Org. Chem. 1992, 57, 2768.
9. For example Wittig, G.; Geissler, G. Liebigs Ann. Chem. 1953, 580; Maryanoff, B.
E.; Reitz, A. B. Chem. Rev. 1989, 89, 863; Wadsworth, W. S. Org. React. 1977, 25,
73; Boutagy, J.; Thomas, R. Chem. Rev. 1974, 74, 87; Blanchette, M. A.; Choy, W.;
Davis, J. T.; Essenfield, A. P.; Masamune, S.; Roush, W. R.; Sakai, T. Tetrahedron
Lett. 1984, 25, 2183.
10. Chatterjee, A. K.; Grubbs, R. H. Angew. Chem., Int. Ed. 2002, 41, 3171.
11. For a summary of the problems associated with ruthenium complex removal
from metathesis products and of ways to obviate this drawback, see Hong, S.
H.; Grubbs, R. H. Org. Lett. 2007, 9, 1955 and references cited therein.
In conclusion, we contend that the ligand 7d [R = iPr] is a viable
candidate as a recoverable tartrate surrogate in SAE reactions,
especially as it can be prepared in optically pure form in just two
straightforward and scalable steps. Clearly, such an effort would
not be realistic for a ‘one-off’ reaction, but certainly such ligand
recovery could contribute significantly with respect to costs and
in environmental terms in bulk or routine operations, especially
if its recovery were to be further optimised during large-scale
reactions.
12. Ees of the ligands
7 were determined by GC analysis of the derived
trifluoroacetates, which were prepared by heating 10–20 mg of the free
ligand with excess trifluoroacetic anhydride in anhydrous dichloromethane in
a 5 ml screw-top vial at 100 °C for 35 min. Excess reagent, TFA and solvent
were removed from the cooled mixture in a stream of nitrogen. The residue
was analysed using the columns described in Ref. 13, initially heated to 100 °C
for 12 min, then programmed at 15 °C/min up to 110 °C.
Acknowledgements
We are very grateful to Dr. N. C. O. Tomkinson and Professor G. J.
Hutchings for helpful suggestions, to Dr. Robert Jenkins for his ana-
lytical expertise, to the EPSRC Mass Spectrometry Service, Univer-
sity College Swansea, for the provision of high resolution mass
spectrometric data and to the EPSRC for financial support.
13. GC analysis and ee determinations of epoxy-alcohol 12 derived from cinnamyl
alcohol were carried out on an
a-DEX 120 column with an oven temperature of
150 °C and the injection port and detector held at 250 °C. The column was
fitted in a Hewlett Packard 5890 Series II Gas Chromatograph, equipped with
FID and a 6890 series integrator. All samples (0.1–1.0 lL) were subjected to
split injection under a column head pressure of 50 kPa of helium, with a gas
flow rate of 1 ml/min. All other epoxy-alcohols were analysed as their
trifluoroacetates (see Ref. 12), using an Astec Chiraldex GTA column at
between 65 and 95 °C, depending on the analyte and conditions as described
above. The ees of the resolved secondary alcohols (Table 4) were also
determined but by directly using the Chiraldex GTA column at 35–110 °C. In
all cases, ee values quoted are the average of three measurements.
References and notes
1. Katsuki, T.; Martin, V. S. Org. React. 1996, 48, 1 and references cited therein.
2. Hanson, R. M.; Sharpless, K. B. J. Org. Chem. 1986, 51, 1922.
3. For a review of recoverable catalysts for asymmetric synthesis, see Fan, Q.-H.;
Li, Y.-M.; Chan, A. S. C. Chem. Rev. 2002, 102, 3385.
14. Finn, M. G.; Sharpless, K. B.. In Asymmetric Synthesis; Morrison, J. D., Ed.;
Academic Press: New York, 1985; Vol. 5, Chapter 8.
15. McKee, B. H.; Kalantar, T. H.; Sharpless, K. B. J. Org. Chem. 1991, 56, 6966.
4. Exceptions to this are attempts to define heterogeneous systems based on
polymer-bound tartrates. For recent reports, see Sherrington, D. C. Catal. Today
2000, 57, 87; Suresh, P. S.; Srinivasan, M.; Pillau, V. N. R. J. Polym. Sci:. Part A: