Highly efficient heterogeneous polymer-supported Sharpless alkene epoxidation catalysts

Article abstract of DOI:10.1016/S0957-4166(98)00137-2

More widespread, and cost effective, use of soluble transition metal complex asymmetric catalysts would be achieved if highly efficient heterogeneous analogues could be developed. We now report the synthesis of branched/crosslinked poly(tartrate ester)s and their use as heterogeneous ligands for the complexation of Ti(O(i)Pr)₄, and use of the resulting insoluble catalysts in the asymmetric epoxidation of a range of trans- allylic alcohols employing t-butylhydroperoxide (tBHP) as the oxidant. Isolated yields of epoxides of up to 80%, with enantiomeric excesses up to 98% have been achieved. Removal and recovery of the polymer catalyst is by simple filtration at the end of reactions. The influence of the ratio of polymer-ligand-titanium, and the polymer backbone molecular architecture on the activity and enantioselectivity of the catalysts has been assessed. Typically polymer branching ratios of 3-11% and a polymer-ligand:titanium ratio of 2:1 yield the optimum results.

Full text of DOI:10.1016/S0957-4166(98)00137-2

TETRAHEDRON:
ASYMMETRY
Pergamon
Tetrahedron: Asymmetry 9 (1998) 1563–1575
Highly efficient heterogeneous polymer-supported Sharpless
alkene epoxidation catalysts
J. K. Karjalainen,a O. E. O. Hormi a and D. C. Sherrington b,∗
aDepartment of Chemistry, University of Oulu, PL 400, SF-90571, Oulu, Finland
bDepartment of Pure and Applied Chemistry, University of Strathclyde, 295 Cathedral Street, Glasgow G1 1XL, UK
Received 27 February 1998; accepted 1 April 1998
Abstract
More widespread, and cost effective, use of soluble transition metal complex asymmetric catalysts would
be achieved if highly efficient heterogeneous analogues could be developed. We now report the synthesis of
i
branched/crosslinked poly(tartrate ester)s and their use as heterogeneous ligands for the complexation of Ti(O Pr)4,
and use of the resulting insoluble catalysts in the asymmetric epoxidation of a range of trans-allylic alcohols
employing t-butylhydroperoxide(tBHP) as the oxidant. Isolated yields of epoxides of up to 80%, with enantiomeric
excesses up to 98% have been achieved. Removal and recovery of the polymer catalyst is by simple filtration at
the end of reactions. The influence of the ratio of polymer-ligand–titanium, and the polymer backbone molecular
architecture on the activity and enantioselectivity of the catalysts has been assessed. Typically polymer branching
ratios of 3–11% and a polymer-ligand:titanium ratio of 2:1 yield the optimum results. © 1998 Elsevier Science
Ltd. All rights reserved.
1. Introduction
The numerous advantages of polymer-supported systems have been well recognized1 but unfortunately
in the case of asymmetric catalysts the supported systems often suffer from a significant drop in
enantioselectivity compared to the corresponding low molecular weight species.2 Recently attention
has focussed on polymer-supported asymmetric alkene oxidation catalysts.3 Some approaches to the
production of a heterogeneous Sharpless-type epoxidation catalyst have also been published4 but the
level of asymmetric induction achieved was generally only modest. We have previously reported on the
synthesis of a group of linear soluble poly(tartrate ester) ligands and their use as homogeneous catalysts
with titanium tetraisopropoxide and tert-butyl hydroperoxide (tBHP) as the oxidant in epoxidising trans-
hex-2-en-1-ol in high chemical yield and with good enantioselectivity.5 Although the enantiomeric
∗
Corresponding author. E-mail: M.P.A.Smith@Strath.ac.uk
0957-4166/98/$19.00 © 1998 Elsevier Science Ltd. All rights reserved.
PII: S0957-4166(98)00137-2

1564
J. K. Karjalainen et al. / Tetrahedron: Asymmetry 9 (1998) 1563–1575
excesses are in general a little lower than those achieved using monomeric tartrate esters, they are still
significantly better than those from previously reported polymer-supported systems4 as well as those
from some homogeneous Sharpless-type modified systems.6 We believe therefore that our results are
important not only in methodological (and potentially technological) terms, but also scientifically since
they offer the prospect of new information regarding the structure of the active species and its mechanistic
role, and provide design criteria for new catalysts.
We have now synthesized a number of insoluble branched and/or crosslinked poly(tartrate ester)s and
have studied their use as heterogeneous optically active ligands in asymmetric epoxidations of a number
of trans-allylic alcohols with titanium tetraisopropoxide and tBHP. We have investigated the effect of the
ratio of polymer-ligand to titanium on the enantioselectivity and yields, and the influence of the polymer
backbone molecular architecture on asymmetric induction. Furthermore we also present our preliminary
ideas explaining the various polymer effects we have observed and will report in more detail on these
later, along with molecular modelling studies.
2. Results and discussion
The engineering of processes utilising structurally well-defined soluble metal complex catalysts7 can
be difficult and costly. Attachment of such species to linear macromolecules which remain homogeneous
improves the situation somewhat, but the use of insoluble polymer-supports as the basis for heterogeneous
catalysts really transforms the prospects for exploitation of such metal complexes in catalysis.8,9
In recent publications we have described the synthesis of linear optically active polyesters which
were obtained by bulk or phase transfer catalysed polycondensation of diacids with diols.5 We have
now developed the bulk polymerisation further. In general, it is difficult to prepare linear polyesters
by melt polycondensation having pendant functional groups, such as hydroxyl groups, on a main-
chain since the pendant groups tend to participate in the reaction leading to crosslinking.10 Indeed,
1
we were surprised in our earlier studies not to detect any branching/crosslinking but the H NMR
spectra of these products confirmed the absence of branching/crosslinking.11 However we have now
synthesized some branched/crosslinked optically active polyesters using an elevated reaction temperature
and prolonged reaction time. These gel-type polyester ligands have from 3% up to a 15% degree of
branching/crosslinking but they all swell enough in the catalysis solvent (CH2Cl2) to form a complex with
titanium tetraisopropoxide, thus allowing them to be used heterogeneously in asymmetric epoxidation
reactions.12
2.1. Synthesis of branched/crosslinked optically active poly(tartrate ester)s
The branched/crosslinked C8-alkyl group-containing poly(tartrate ester) gels 3 were synthesised from
L-(+)-tartaric acid 1, 20 mol% excess of 1,8-octanediol 2 and 3 wt% p-toluene sulfonic acid as a catalyst,
as shown in Scheme 1.13 The C8-diol was chosen because the linear poly(tartrate ester) derived from
this had earlier proven to be the most enantioselective ligand, probably because of easy access of Ti to
the tartrate groups,14 and because this polymer can most easily adopt the conformation for formation
of the active metal complex.15 The polymerisation conditions used were anaerobic and more forcing
than in earlier preparations5 and this seems to be the origin of the branching/crosslinking. Elemental
microanalysis of 3 typically yielded C, 55.3; H, 8.5; and the theoretical values for a linear version of
3 are C, 55.4; H, 7.75. The FTIR spectrum of 3 displayed the expected bands 1743 (C_O) and 3500
(OH) cm−1. The exact structure and architecture of 3 varied with different batches. Most were soluble

J. K. Karjalainen et al. / Tetrahedron: Asymmetry 9 (1998) 1563–1575
1565
either in hot pyridine, hot THF or hot Me2SO and thus are probably branched rather than crosslinked.16
All the polymers which were soluble in hot THF, displayed a positive optical rotation with [α]D25=+9
typically. This value compares with one of +17 for the earlier linear version of 3.5,17 The presence of
1
a CH2OH signal in the H NMR spectra of these soluble polymers confirmed the presence of these
end groups, but the intensity was generally too high to be accounted for by a linear polymer structure.
The occurrence of significant chain branching however is consistent with this observation. Indeed the
1
molecular structure of the polymers was deduced from both H and 13C NMR spectra, and the degree
of branching/crosslinking was calculated by integration of approriate signals in the former spectra.18 In
the 13C-{1H}-spectra a diagnostic methine (CH–OH) signal from the linear part of the polymer chain
was found at δ 72 ppm. Small but clear signals from the branched/crosslinked segments of the polymer
chain were also found at δ 80 ppm (–OOC–CH(CH–OH)–OCO–; in repeat unit y in Scheme 1). The
proton coupled carbon spectrum shows the small signal at δ 80 ppm to be a doublet hence confirming the
structure assignment. The degree of branching varied from 3 to 15% of all segments as determined by
integration of the appropriate characteristic signals in the 1H spectra18 and those samples which proved
to be insoluble in hot solvents were assumed to be crosslinked with a level of crosslinking ꢀ15%. The
effect of the polymer backbone architecture (branching/crosslinking degree) on asymmetric induction in
epoxidation reactions was investigated.
Scheme 1. Reagents and conditions: toluene p-sulfonic acid (3 mass%), 20 mol% excess of diol 2, ca. 130°C, 4 days
2.2. Asymmetric epoxidation of trans-hex-2-en-1-ol using heterogeneous poly(tartrate ester) ligands 3
(Scheme 2)
Polymer-ligand 3 did not form CH2Cl2 soluble complexes with Ti(OiPr)4. Inevitably between 70 and
100% of the polymer was recovered from reaction mixtures by simple filtration. Recovered polymer
was washed thoroughly with CH2Cl2 and dried under vacuum at room temperature and IR spectroscopic
analysis showed the polymer to have the same characteristic bands at 3400–3500 cm−1 and 1750 cm−1
as before the reaction, indicating that in principle the polymer-ligand might be reused.19 The results of
epoxidations of trans-hex-2-en-1-ol are summarised in Table 1. Enantiomeric excesses were determined
1
as derived acetates with chiral shift reagents or as Mosher esters using H NMR.12 Some non-GC-
detectable side products were apparent from a shortfall (∼20–30%) in the mass balance so dodecane
was used an internal GC standard to determine the actual yields. It seems most likely that these side-
products are formed via oligomeratisation of the epoxy alcohol products since they did not emerge from
the GC column employed.20 More often than not heterogeneously catalyzed processes are more complex
than homogeneously catalysed ones since the catalyst is not uniformly distributed thoughout the reaction
medium and a distribution (spatial and chemical) of active sites are generally found in heterogeneous
catalysts.21

1566
J. K. Karjalainen et al. / Tetrahedron: Asymmetry 9 (1998) 1563–1575
Scheme 2. Reagents and conditions: poly(tartrate ester), Ti(OPri)4, ButOOH (2 equiv.), 4 Å molecular sieves, CH2Cl2
Table 1
Epoxidation of trans-hex-2-en-1-ol with tBHP by L-(+)-polyester 3 and Ti(OPri)4
Epoxidation of trans-hex-2-en-1-ol using polymer 3 gave similarly high chemical yields to those with
soluble C8-poly(tartrate ester)5 but surprisingly a better enantioselectivity was obtained (Table 1, entry
3, 87% compared to 70% ee earlier). The influence of the ligand:Ti ratio was investigated and also a
series of experiments using different substrate:catalyst ratios was performed. The enantioselectivities
varied a lot with different ligand:Ti ratios as well as depending strongly on the polymer architecture.
In Table 1, the reactions shown as entries 9 and 10 were performed using the same polymer-ligand
with a degree of branching or crosslinking ꢀ15%. This polyester is presumably highly crosslinked
rather than branched since it did not dissolve in any hot solvents. The asymmetric induction observed
is only modest, 26–38% ee, even though the polymer swelled well in CH2Cl2. The reactions shown as
entries 2 and 3 were performed using the polymer-ligand with ∼3% of branching/crosslinking units. The
best result was obtained using a ligand:Ti ratio of 2:1 (87% ee, entry 3). The use of a large excess of
ligand and a lower reaction temperature retarded the reaction (entry 4) although this polymer-ligand also
has a branching/crosslinking ratio of ∼6%. Generally therefore it is quite obvious that different levels
of branching/crosslinking have a significant effect on the degree of asymmetric induction achieved.

J. K. Karjalainen et al. / Tetrahedron: Asymmetry 9 (1998) 1563–1575
1567
Table 2
Epoxidation of trans-2-undecen-1-ol with tBHP by L-(+)-polyester 3 and Ti(OPri)4
The reactions shown as entries 5–8 in Table 1 were performed using polymer-ligands with ∼9–11%
branching/crosslinking. These results are quite consistent and show that the enantioselectivity varies only
according to the ligand:Ti ratio used. Furthermore it is clear that a polymer with branching/crosslinking
levels of up to 11% can be used quite successfully in asymmetric epoxidation by an appropriate choice
of the ligand:Ti ratio.
The complexity of these Ti–polymer-ligand systems makes it difficult to define the optimium polymer-
ligand structure which in turn will allow the the best ligand:Ti ratio to be defined. However surprisingly,
the findings so far suggest that the most efficient heterogeneous polymer–titanium complex has a
2:1 structure instead of the Sharpless low molecular weight 2:2 dimer. In principle, the reactivity or
effectiveness of a particular functional group or ligand on a polymer can depend upon whether it is
located at or near a crosslinking point or somewhere on the chain well removed from such a point.14
Some of the tartrate residues in the 2:1 structures may not therefore play a functional role, and the real
structure may indeed be an analogue of the 2:2 soluble dimer.
2.3. Asymmetric epoxidation of other allylic alcohols using poly(tartrate ester) 3 (Scheme 2)
With the successful epoxidation of trans-hex-2-en-1-ol using polymer-ligand 3 we were encouraged to
perform a series of epoxidation reactions of trans-allylic alcohols using this heterogeneous system. The
results are summarized in Tables 2–4.
In order to investigate the effect of reaction temperature, some reactions were carried out at room
temperature, 0, −20 and −30°C. A mixture of unidentified compounds was generally obtained when
each reaction was performed at room temperature, although typically ∼20% of the target epoxy alcohol
was also present. Enantioselectivities were in general ∼10–20% ee when the reactions were carried out
at 0°C (Table 1, entry 9 and Table 2, entry 4, Table 4, entry 9) but also lowering the reaction temperature
much below −20°C did not bring any advantage (Table 1, entry 4).22
The results of the epoxidation of trans-2-undecen-1-ol in Table 2 show the same effect of the polymer
architecture on asymmetric induction as observed before with trans-hex-2-en-1-ol. The reactions shown
as entries 4–6 (Table 2) were carried out using the highly crosslinked (ꢀ15%) polyester, and the

1568
J. K. Karjalainen et al. / Tetrahedron: Asymmetry 9 (1998) 1563–1575
Table 3
Epoxidation of trans-cinnamyl alcohol with tBHP by L-(+)-polyester 3 and Ti(OPri)4
Table 4
Epoxidation of trans-geraniol with tBHP by L-(+)-polyester 3 and Ti(OPri)4
enantioselectivities are very low compared to those obtained using the polymer with only 3–6% of
branched/crosslinked units (entries 2 and 3). In the latter cases the enantioselectivities are excellent
(88–98% ee) and are combined with reasonable non-optimised isolated yields (40–50%). Even using
a stoichiometric amount of polymer-ligand relative to substrate (entry 2) the polymer could be easily
filtered off from the reaction mixture. These results show clearly the synthetic potential of these
heterogeneous asymmetric epoxidations.
With the sensitive substrate trans-cinnamyl alcohol some difficulties were observed. The results

J. K. Karjalainen et al. / Tetrahedron: Asymmetry 9 (1998) 1563–1575
1569
in Table 3 show that the chemical yields drop significantly. Titanium isopropoxide is a potential
polymerization catalyst23 and these low yields may arise from epoxide polymerization.
Nevertheless the effect of the polymer architecture is seen again in the data in Table 3. The best
enantioselectivity was obtained using the 6% branched/crosslinked polymer-ligand with a ligand:Ti ratio
of 2:1 (89% ee, entry 3). The polymer with a lower degree of branching/crosslinking (3%) did not
give a better enantioselectivity even using a ligand:Ti ratio of 1.18:1 (entry 2). This again indicates the
importance of an appropriate ligand:Ti ratio.
Several ligand:Ti ratios with different polymer batches were studied in the case of trans-geraniol as the
substrate (Scheme 3). The results in Table 4 show again that the best polymer-ligand:Ti ratio is 2:1 using
the polymer with a degree of branching/crosslinking of 6% (entry 2). The level of asymmetric induction
drops significantly (∼30%, entries 2 and 6) when the branching/crosslinking is increased to ꢀ15%. The
highest enantioselectivity obtained is 72% (entry 2) which is somewhat lower than that achieved with
other substrates but higher than that reported with an earlier polymer-supported catalyst.4 This poorer
performance with this substrate is not too surprising however since Sharpless et al. have also observed
that using the monomeric tartrate with this substrate the enantioselectivity is reduced (Table 4, entry 1).24
Scheme 3. Reagents and conditions: poly(tartrate ester), Ti(OPri)4, ButOOH (2 equiv.), 4 Å molecular sieves, CH2Cl2
Overall the results of experiments with four allylic alcohols show that the optimium asymmetric
induction is obtained with heterogeneous polymer-ligands with a degree of branching/crosslinking of
3–6%, when used in a polymer-ligand:Ti ratio of 2:1.
2.4. Effect of leached titanium on asymmetric induction
The influence of leached Ti on enantioselectivity has also been investigated. Typically when polymer-
ligand 3 was removed by filtration from the reaction solution with a polymer-ligand:Ti ratio of 2:1
recoveries were ∼90–100%. Atomic absorption spectroscopic analysis indicates that generally about
80% of Ti was bound to the polymer-ligand and the remaining 20% is uncomplexed or complexed with
tartrate monomer or oligomer in a solution.25 To examine the activity of this 20% ‘free’ titanium, a control
epoxidation reaction was carried out. A Ti–polymer-ligand catalyst was prepared using the standard
procedure but the Ti–polymer complex was filtered off before adding the substrate; trans-hex-2-en-1-
ol. The reaction proceeded slowly (GC analysis) and after 10.5 hours stirring at −20°C and storing in
a freezer for over 2 weeks the yield of epoxide was 23%.26 The epoxide was isolated by Kugelrohr
distillation to determine the enantioselectivity of the reaction which was very low (4% ee). This data
shows that the catalytic activity of the free Ti is very low and that this component is probably not tartrate
complexed. This is confirmed by the low % ee seen which again suggests an absence of any optically
active monomer or oligomer in solution. Polymer 3 therefore appears not to contain any significant
CH2Cl2 soluble component.

1570
J. K. Karjalainen et al. / Tetrahedron: Asymmetry 9 (1998) 1563–1575
3. Experimental
3.1. General
1
The H NMR and 13C NMR spectra were obtained on Bruker AM200 MHz or Bruker DPX400
MHz spectrometers. Chemical shifts (δ) are given in ppm relative to TMS. Gas chromatographic
(GC) analysis were carried out on a Carlo Erba GCHR 5300 Mega Series employing a SGE BPX5
column or a Perkin–Elmer 8500 equipped with FID detector employing a 25 m column packed with
fused silica (phase layer 0.25 µm). FTIR spectra were measured on a Bruker IFS 66 or Nicolet
Impact 400d instruments, wavenumbers are in cm−1. Enantiomeric ratios measured by HPLC were
determined using a Daicel Chiralcel OB column (flow rate=0.5 ml/min, hexane:isopropanol=97.5:2.5).
High resolution GC/MS analysis were performed by the Mass Spectrometry Laboratory of the University
of Oulu, Finland. Optical rotations were determined on a Perkin–Elmer WM250 or a Perkin–Elmer
243 B polarimeter using a 1 cm3 capacity, 1 dm path length, quartz cell. Elemental analyses were
performed by the Microanalytical Laboratory of the University of Strathclyde, Scotland or by the
Microanalytical Laboratory of the University of Oulu, Finland. Metal leaching was determined by atomic
absorption spectrophotometry (AAS) by the Microanalytical Laboratory of the University of Oulu,
Finland. Molecular weights were measured by gel permeation chromatography (polystyrene standards
in THF) by RAPRA, UK. Flash chromatography was performed using Merck silica gel 60 (230–400
mesh) with diethyl ether/petrol ether (40°C–60°C) in various proportions as eluent.
3.2. Materials
Pre-activated and powdered 4 Å molecular sieves were available from the Aldrich Chemical Co.
and were used as received. The dichloromethane used did not contain methanol and therefore was not
distilled but was stored over CaCl2. Aqueous 70% tert-butyl hydroperoxide (tBHP) was obtained from
the Aldrich Chemical Co. trans-Cinnamyl alcohol was distilled under vacuum. All other reagents were
used as obtained from Aldrich Chemical Co. or from Fluka Chemica-Biochemica (trans-2-undecenol).
Reagents handled by syringe were measured by weight or by volume.
3.3. Anhydrous tert-butyl hydroperoxide (tBHP) in iso-octane27
To a 250 ml separating funnel was added a 70% tBHP/30% H2O solution (65 ml) and reagent grade iso-
octane (80 ml). The flask was swirled (not shaken) to avoid the formation of an emulsion. The aqueous
layer was separated and the organic layer was transferred to a 250 ml round-bottomed flask equipped
with a Dean Stark trap and a reflux condenser. A few boiling chips were added and the solution was
refluxed. After approximately one hour of refluxing, a drying tube filled with anhydrous CaCl2 was fitted
and refluxing was continued for a further two hours. After the tBHP/iso-octane solution had been allowed
to cool, it was transferred to a brown glass bottle and was stored over 4 Å molecular sieves.
3.4. Molarity determination of anhydrous tBHP/iso-octane26
A 0.1 M aqueous sodium thiosulphate solution was prepared by dissolving Na2S2O3·5H2O (12.16 g)
in de-ionised water (500 ml). To a 250 ml conical flask, isopropyl alcohol (25 ml) and glacial acetic acid
(2 ml) were added. A 20% solution of NaI was prepared in 100 ml of warm isopropyl alcohol and 10
ml of this solution was added to the conical flask. A quantity (0.5 ml) of the anhydrous tBHP iso-octane

J. K. Karjalainen et al. / Tetrahedron: Asymmetry 9 (1998) 1563–1575
1571
solution was then added and the mixture was heated to boiling and was allowed to boil for 40 seconds.
After dilution with 100 ml of de-ionised water, the warm solution was rapidly titrated with the 0.1 M
thiosulphate solution until the yellow iodine colour disappeared. The titration was repeated three times.
The concentration was calculated from Eq. 1:
(molarity of titrant) ∗ (volume of titrant added)
Concentration =
(1)
(volume of tBHP solution) ∗ 2
3.5. General preparation and NMR analysis of Mosher esters12
This method was suitable for the determination of the enantiomeric excess (ee) of all trans-
disubstituted epoxy alcohols in particular it is the only suitable method for trans-epoxycinnamyl alcohol.
The reactions were run on a 0.15 mmol scale. A mixture of 18 mg (0.15 mmol, 1.0 equiv.) of 4-
(dimethylamino)pyridine (DMAP) and 100 µl of triethylamine in 10.5 ml of CH2Cl2 was treated with the
substrate. Immediately, 30 µl of neat (+)-α-methoxy-α-(trifluoromethyl)phenylacetyl chloride (MTPA
chloride) was added. The solution became warm and turned orange. Reactions were quenched by addition
of 3-(dimethylamino)propylamine (40–60 µl) and the residue was passed though a plug of silica gel in
order to remove polar impurities (20% EtOAc/hexane). 1H NMR analysis in C6D6 at 400 MHz focussed
on the terminal methylene protons. These protons typically were observed as a diastereomeric pair of
AB doublets (or doublets of doublets) around δ 4.0–4.3 ppm. The pair was compared by integration to
determine the enantiomeric excess.
3.6. General preparation and NMR shift analysis of acetates12
This method was suitable for nearly all trans-disubstituted epoxy alcohols and for 2,3-epoxygeraniol.
Acetates were prepared as described above for Mosher esters, using acetyl chloride (0.30 mmol
1
scale). H NMR analysis involved sequential treatment of a solution of about 10 mg of an acetate
in 0.5 ml of C6D6 with 0.05–0.1 mol equiv. portions of a solution of 0.14 g of europium(III) tris[3-
(heptafluoropropylhydroxymethylene)-d-camphorate], Eu(hfc)3, in 2 ml of C6D6 and observation of the
acetate CH3.
3.7. Absolute configurations
Absolute configurations were determined by comparison of the observed rotation by polarimetry with
the literature value.
3.8. AAS-measurements
The leaching of titanium from the polymer-support was determined by AAS analysis. After filtration
the polymer-ligand was washed throughly with dichloromethane and the sample from the filtrate (known
reaction volume) was taken for AAS-measurements. The results are expressed as mg titanium/L.
3.9. General preparation of branched/crosslinked optically active polyesters13 (3, Scheme 1)
An oven-dried 100 ml three-necked round-bottomed flask equipped with a magnetic stir bar, nitrogen
inlet and bubbler was charged with L-(+)-tartaric acid (10 g, 0.067 mol), 20 mol% excess of 1,8-octane

1572
J. K. Karjalainen et al. / Tetrahedron: Asymmetry 9 (1998) 1563–1575
diol (11.67 g, 0.080 mol) and ∼3 wt% p-toluene sulphonic acid (0.6 g) (excess of diol was employed
to account for possible loss of this component in the N2 stream at high temperature). The temperature
was raised to between 140 and 150°C to achieve a homogeneous mixture and then it was stirred at
120–130°C for between 3 and 4 days under the nitrogen. Water and unreacted diol were removed by
distillation under high vacuum at the end of reaction. The resulting solid was swelled in ethyl acetate by
refluxing and then it was poured into n-hexane. The solvents were decanted and the polymer obtained
was dried under high vacuum at +40°C for 6 hours and at room temperature for 2 days to yield 16.6 g,
95% of C8-poly(tartrate ester) 3. [α]D25=+9 (c 1.6, THF). FTIR (KBr) 3450, 2932, 2857, 1743, 1466,
1
1395, 1128, 1091, 1011, 956. Anal. found: C, 55.3; H, 8.5. H NMR (400 MHz, DMSO-d6, 70°C) δ
5.75 (br s), 5.41 (d, J=3.15 Hz), 4.62 (d, J=2.83 Hz), 4.37 (s, 2H), 4.09 (t, J=6.49 Hz, 4H), 1.58 (m, 4-
5H), 1.30 (m, 9H). 13C NMR (400 MHz, DMSO-d6, 70°C) δ 172.42 (s), 171.28 (s), 171.03 (s), 80.41 (d,
JCH=142.04 Hz), 73.45, 72.45, 72.19 (d, JCH=146.24 Hz), 64.47 (t, JCH=147.2 Hz), 28.32 (t, JCH=125.64
Hz), 28.04 (t, JCH=125.92 Hz), 25.07 (t, J=125.66 Hz). Note: different batches of 3 analysed (1H NMR)
to show different levels of branching/crosslinking.
3.10. General procedure for the catalytic asymmetric epoxidation using heterogeneous polymer-ligand12
The literature procedure for the epoxidation of allylic alcohols was followed with modifications. An
oven-dried three-necked round-bottomed flask equipped with a magnetic stir bar, nitrogen inlet, septum
and bubbler was charged with 4 Å powdered, activated molecular sieves (3 g), polymer-ligand 3 (see
Tables 1–4 for substrate:Ti:ligand ratios) and dry CH2Cl2. The flask was cooled to −20°C and Ti(O-i-Pr)4
(5 mmol, via syringe) was added sequentially with stirring. The reaction mixture was stirred at −20°C
and after about one hour tBHP (0.2 mmol in iso-octane) was added with syringe at a moderate rate. The
resulting mixture was stirred at −20°C for at least 1 hour. Substrate (0.1 mol, dissolved in dry CH2Cl2)
was added dropwise with a syringe, being careful to maintain the reaction temperature between −15°C
and −20°C. The mixture was stirred for an additional 3–12 hours between −15°C and −20°C. The
reaction mixture was stored in a freezer for various periods before work-up. The reaction was monitored
by gas chromatography (GC) using dodecane as a GC internal standard. The polymer was filtered off the
reaction mixture and washed thoroughly with CH2Cl2. The recovered solution was then quenched with a
30% aqueous solution of sodium hydroxide saturated with sodium chloride. After 10 wt% of diethyl ether
was added, the cold bath was removed and the stirred mixture was allowed to warm to 10°C. Stirring was
maintained for an additional 10 min at 10°C, whereupon MgSO4 and Celite were added. After a final 15
min of stirring, the mixture was allowed to settle and the solution was filtered through a pad of Celite and
washed with diethyl ether. Excess tBHP was removed by azeotropic distillation with toluene. The crude
product was purified by flash-chromatography (eluent: petrol ether (40°C–60°C):diethyl ether, 50%) or
by Kugelrohr-distillation and the enantioselectivity of the epoxide was calculated from the 400 MHz
NMR spectrum using chiral Eu(hfc)3-reagent (as acetates) or as Mosher esters.
3.11. (2S-trans)-3-Propyloxiranemethanol (4) from heterogeneous catalyses
The epoxidation was performed as described in the general procedure, in this case 40 ml of CH2Cl2
with 0.3 g of powdered, activated 4 Å molecular sieves, 2.6 g (0.01 mol) of the hetereogeneous C8-
poly(tartrate ester) 3, 1.42 g (0.005 mol) of Ti(OPri)4, 12.5 ml of a 3.2 M solution of tBHP in iso-
octane and 2 g (0.02 mol) of trans-hex-2-en-1-ol (dissolved in 10 ml CH2Cl2). The reaction mixture
was stirred at −20°C for 6 h and placed in a freezer overnight. The polymer-ligand was filtered off and
washed throughly with CH2Cl2. The work-up was then performed and crude the product was purified by

J. K. Karjalainen et al. / Tetrahedron: Asymmetry 9 (1998) 1563–1575
1573
flash-chromatography (eluent: petrol ether (40°C–60°C):diethyl ether, 50%) to give a colourless oil (4)
1
(1.22 g, 53% yield, ∼100% purity, 87% ee by H NMR analysis of the derived acetate with Eu(hfc)3):
[α]25D=−38 (c 2.01, CHCl3), GC analysis before work-up: 78% epoxide. FTIR (CHCl3) 2977, 2952,
1717, 1457, 1237, 1196, 933, 881, 806, 796. 1H NMR (200 MHz, C6D6) δ 3.72 (ddd, J=3, 6, 12 Hz, 1H),
3.46 (ddd, J=5, 7, 12 Hz, 1H), 2.79–2.85 (m, 1H), 2.70–2.74 (m, 1H), 2.33 (t, J=6 Hz, 1H), 1.34–1.45
(m, 4H), 0.93 (t, J=7 Hz, 3H). MS (EI, m/z, relative intensity) 99 (0.8, M−17)+, 81 (7), 73 (20), 61 (8),
55 (100), 43 (86). HRMS (EI) calcd for C6H12O2 116.0837, found 116.0882.
3.12. (2S-trans)-3-Octyloxiranemethanol (5) from heterogeneous catalyses
The epoxidation was performed as described in the general procedure using in this case 35 ml of
CH2Cl2 with 0.3 g of powdered, activated 4 Å molecular sieves, 1.56 g (0.0059 mol) of the heterogeneous
C8-poly(tartrate ester) 3, 0.85 g (0.003 mol) of Ti(OPri)4, 7.5 ml of 3.2 M solution of tBHP in iso-octane
and 1 g (0.0059 mol) of trans-undec-2-en-1-ol (dissolved in 5 ml CH2Cl2). The reaction mixture was
stirred at −20°C for 6 h and placed in a freezer overnight. The polymer-ligand was filtered off and
washed throughly with CH2Cl2. The work-up was then performed and the crude product was purified
by crystallization from petrol ether to yield 0.55 g of a white solid (5) (50% yield, 88% ee, by 1H NMR
analysis of the Mosher ester, GC analysis after 6 h: 87% epoxide). [α]25D=−22 (c 1.33, CHCl3). FTIR
1
(CHCl3) 2975, 2949, 2879, 2867, 2844, 1701, 1653, 1559, 1507, 1237, 1196, 933, 879, 836, 817. H
NMR (200 MHz, CDCl3) δ 3.88 (dd, J=2, 13 Hz, 1H), 3.59 (ddd, J=5, 7, 10 Hz, 1H), 2.87–2.96 (m, 2H),
2.68 (br s), 1.24–1.55 (m, 14H), 0.85 (t, J=3 Hz). MS (EI, m/z, relative intensity) 109 (1, M−77)+, 95
(14), 83 (51), 69 (100), 55 (94), 41 (99), 29 (28). HRMS (EI) calcd for C11H22O2 186.162, found (M+1)
187.1700.
3.13. (2S-trans)-3-Phenyloxiranemethanol (6) from heterogeneous catalyses
The epoxidation was performed as described in the general procedure using in this case 40 ml
of CH2Cl2 with 0.3 g of powdered, activated 4 Å molecular sieves, 1.95 g (0.0075 mol) of the
hetereogeneous C8-poly(tartrate ester) 3, 1.07 g (0.00375 mol) of Ti(OPri)4, 9.4 ml of 3.2 M solution
of tBHP in iso-octane and 2 g (0.015 mol) of trans-2-cinnamyl alcohol (dissolved in 10 ml CH2Cl2)
at −20°C for 6 h and the reaction mixture was placed in a freezer overnight. The polymer-ligand was
filtered off and washed throughly with CH2Cl2. The work-up was then performed and the crude product
1
was purified by Kugelrohr-distillation to yield a light yellow solid 0.85 g (6) (38% yield, 89% ee H
NMR analysis of the Mosher ester): [α]25D=−36 (c 2.37, CHCl3). FTIR (CHCl3) 3035, 3007, 2976,
1
2399, 1734, 1717, 1684, 1616, 1489, 1457, 1296, 1237, 1196, 1093, 1074, 1057, 977, 938, 879. H
NMR (200 MHz, CDCl3) δ 7.26–7.34 (m, 5H), 4.06 (ddd, J=2, 5, 13 Hz, 1H), 3.94 (d, J=2 Hz, 1H), 3.81
(ddd, J=4, 8, 12 Hz, 1H), 3.21–3.26 (m, 1H), 1.97 (dd, J=5, 6 Hz, 1H). MS (EI, m/z, relative intensity)
149 (2, M−1)+, 132 (17), 121 (28), 107 (78), 91 (100), 79 (56), 63 (27), 51 (31), 43 (15). HRMS (EI)
calcd for C9H10O2 150.0681, found 150.0697.
3.14. (2S-trans)-3-Methyl-3-(4-methyl-3-pentenyl)oxiranemethanol (8) from heterogeneous catalyses
The epoxidation was performed as described in the general procedure using in this case 30 ml of
CH2Cl2 with 0.3 g of powdered, activated 4 Å molecular sieves, 1.69 g (0.0065 mol) of heterogeneous C8-
poly(tartrate ester) 3, 0.46 g (0.00163 mol) of Ti(OPri)4, 8.1 ml of 3.2 M solution of tBHP in iso-octane
and of trans-geraniol 7 (dissolved in 10 ml CH2Cl2) at −20°C for 7.5 h and the reaction mixture was

1574
J. K. Karjalainen et al. / Tetrahedron: Asymmetry 9 (1998) 1563–1575
placed in a freezer overnight. The polymer-ligand was filtered off and washed throughly with CH2Cl2.
The work-up was then performed and the crude product was purified by flash-chromatography (eluent:
petrol ether (40°C–60°C):diethyl ether, 50%) to give a colourless oil 1.48 g (8) (66% yield, 72% ee 1H
NMR analysis of the derived acetate with Eu(hfc)3): [α]25D=−4 (c 3.6, CHCl3). FTIR (CHCl3) 3035,
1
3007, 2896, 1701, 1684, 1653, 1237, 1196, 938, 879, 835, 816. H NMR (200 MHz, C6D6) δ 5.23 (tt,
J=1.4, 7 Hz, 1H), 3.61–3.69 (m, 2H), 2.98 (dd, J=5, 6 Hz, 1H), 2.16 (q, J=7 Hz, 2H), 1.98 (t, J=6 Hz,
1H), 1.77 (s, 3H), 1.64 (s, 3H), 1.49–1.72 (m, 2-3H), 1.19 (s, 3H). MS (EI, m/z, relative intensity) 149
(1, M−21)+, 121 (4), 109 (70), 95 (25), 82 (46), 69 (100), 61 (36), 55 (45), 41 (97). Found HRMS (EI)
calcd for C10H18O2 170.1307, found 170.1331.
4. Conclusions
Branched or crosslinked poly(tartrate ester)s have been prepared by p-toluene sulfonic acid catalysed
condensation of L-(+)-tartaric acid with 1,8-octanediol. Different batches of the polymer have been
1
shown by H NMR analysis to have different levels of branching (3% to ꢀ15%). In the range 3–11%
the polymers can be solubilised in hot solvents and so are branched, whereas polymers with apparent
branching (ꢀ15%) are insoluble, cannot be analysed quantitatively, and are probably crosslinked rather
than branched. All of the species can be employed as heterogeneous ligands for the complexation
of Ti(OiPr)4 and the insoluble complexes formed have been used as catalysts in the asymmetric
epoxidation of trans-hex-2-en-1-ol, trans-2-undecen-1-ol, trans-cinnamyl alcohol and trans-geraniol.
Polymer catalysts can be removed by simple filtration at the end of reactions, and epoxide isolated yields
up to 80% and enantiomeric excesses up to 98% have been achieved. The ratio of polymer-ligand:Ti and
the molecular architecture of the polymer backbone influence the effectiveness of the catalysts. Optimum
results are obtained typically with a polymer-ligand:Ti ratio of 2:1, and with polymers having a branching
ratio of between 3 and 11%.
Currently the effectiveness of recycling of recovered polymer-ligands is being assessed, in combination
with a molecular modelling study aimed at providing a structural basis for our observed results.
Acknowledgements
JKK acknowledges financial support from the Neste Oy Foundation, Finland which made this work
possible.
References
1. (a) Sherrington, D. C. Polymer-supported Synthesis In Chemistry of Waste Minimisation; Ed. J. H. Clark: Blackie
Publishers, 1995; Chap. 6, pp. 141–196; (b) Hodge, P. Polymer-supported Asymmetric Organic Synthesis In Innovation
and Perspectives in Solid Phase Synthesis; Ed. R. Epton: SPCC UK, 1990; pp. 273–292; (c) Shuttleworth, S. J.; Allin S.
M.; Sharma, P. K. Synthesis 1997, 1217; (d) Bolm, C.; Gerlach, A. Eur. J. Org. Chem. 1998, 21.
2. (a) Pini, D.; Petri, A.; Nardi, A.; Rosini, C.; Salvadori, P. Tetrahedron Lett. 1991, 32, 5175–5178; (b) De, B. B.; Lohray, B.
B.; Sivaram, S.; Dhal, P. K. Tetrahedron: Asymm. 1995, 6, 2105–2108; (c) Minutolo, F.; Pini, D; Salvadori, P. Tetrahedron
Lett. 1996, 37, 3375–3378.
3. See Ref. 2. (a) Kim, B. M.; Sharpless, K. B. Tetrahedron Lett. 1990, 31, 3003–3006; (b) Jacobsen, E. N.; Zhang, W.; Muci,
J. R.; Echer, J. R.; Deng L. J. Am. Chem. Soc. 1991, 113, 7063–7064; (c) Lohray, B. B.; Thomas, A.; Chittari, P.; Ahuja,
J. R.; Dhal, P. K. Tetrahedron Lett. 1992, 33, 5453–5456; (d) Pini, D.; Petri, A.; Salvadori, P. Tetrahedron: Asymm. 1993,

J. K. Karjalainen et al. / Tetrahedron: Asymmetry 9 (1998) 1563–1575
1575
4, 2351–2354; (e) Itsuno, S.; Koizumi, T.; Okumura, C.; Ito, K. Synthesis 1995, 150–152; (f) Han, H.; Janda, K. D. J. Am.
Chem. Soc. 1996, 118, 7632–7633.
4. (a) Farrall, M. J. F.; Alexis, M.; Trecarten, M. Nouv. J. de Chimie 1983, 7, 449–451; (b) Choudary, B. M.; Valli, V. L. K.;
Prasad, A. D. J. Chem. Soc., Chem. Comm. 1990, 1186–1187.
5. Canali, L.; Karjalainen, J. K.; Hormi, O. E. O.; Sherrington, D. C. J. Chem. Soc., Chem. Comm. 1997, 123–124;
Kaajalainen, J. K.; Hormi, O. E. O.; Sherrington, D. C. Molecules, 1998, 3, 51.
6. (a) Adam, W.; Kolb, M. N. Tetrahedron: Asymm. 1997, 8, 1131–1142; (b) According to the literature, several asymmetric
ligands other than tartrates have been tested. See Table in: Finn, M. G.; Sharpless, K. B. In Asymmetric Synthesis; Ed.
Morrison, J. D., Academic Press, Orlando, FL, Vol. 5, 1985; pp. 293–299.
7. Gates, B. C. Catalytic Chemistry; John Wiley & Sons, Inc., New York, 1992, pp. 15–19.
8. Hartley, F. R. Supported Metal Complexes; D. Reidel Publishing Company, Holland, 1985, pp. 1–9.
9. Grubbs, R. H.; Sweet, E. M.; Phisanbut, S. Catalysis in Organic Synthesis 1976, Ed. Rylander, P. N. and Greenfield, H.,
Academic Press, Inc., New York, 1976, pp. 153–160.
10. Stevens, M. P. Polymer Chemistry an Introduction; 2nd Ed., Oxford University Press, 1990; pp. 392–417.
11. Polymer samples were dissolved in hot Me2SO. The 1H and 13C NMR spectra were obtained on a 400 MHz spectrometer
at 343 K and showed no evidence of any characteristic branching/crosslinking methine signals in the 1H NMR spectra.
12. The Sharpless catalytic method for asymmetric epoxidation was followed with slight modifications. Gao, Y.; Hanson, R.
M.; Klunder, J. M.; Ko, S. Y.; Masamune, H.; Sharpless, K. B. J. Am. Chem. Soc. 1987, 109, 5765–5780.
13. Seebach, D.; Kalinowski, H.-O.; Langer, G.; Grass, G.; Wilka, E. M. Organic Synthesis, 1983, 61, 24.
14. Maud, J. M. Solid Supports and Catalysts in Organic Synthesis, Ed. Smith K., Ellis Horwood limited, England, 1992, pp.
40–58.
15. (a) Finn, M. G.; Sharpless, K. B. In Asymmetric Synthesis; Ed. Morrison, J. D., Academic Press, Orlando, FL, Vol. 5, 1985;
pp. 247–301; (b) Finn, M. G.; Sharpless, K. B. J. Am. Chem. Soc. 1991, 113, 113–126.
16. (a) Elias, H.-G. An Introduction to Polymer Science, VCH, Weinheim, 1997, pp. 16–23; (b) Stevens, M. P. Polymer
Chemistry an Introduction; 2nd Ed., Oxford University Press, 1990; pp. 7–12, 98–100.
17. [α]D25=+9 (c 1.6, THF). Optical rotation did not depend on the degree of branching/crosslinking. Some polymer batches
were not soluble in hot THF and thus it was assumed that these polymers had degrees of branching/crosslinking ꢀ15%
i.e. very much higher than THF soluble polymers (3–15%).
18. Polymer samples were dissolved in hot Me2SO. The 1H NMR spectra were obtained on a 400 MHz NMR spectrometer
at 343 K. An example of the calculation of the branching/crosslinking degree by integration is as follows. 1H NMR
spectrum of poly(tartrate ester) 4: δ 5.41 (d, intensity=0.185H) +4.61 (d, 0.164H) [–OOC–CH(OCO–)–CH(OH)COO–,
–OOC–CH(OCO–)–CH(OH)COO–, Scheme 1, unit y]/ 4.37 (s, 2.001H) [–OOC–CH(OH)–CH(OH)–COO–, Scheme 1,
unit x]+y*100%=14.8%.
19. Development of a reusable heterogeneous epoxidation catalyst is under investigation in our laboratory.
20. Stevens, M. P. Polymer Chemistry an Introduction; 2nd Ed., Oxford University Press, 1990; pp. 360–368 and references
therein.
21. Augustine, R. L. Heterogeneous Catalysis for the Synthetic Chemist; Marcel Dekker, Inc., New York, 1996; pp. 153–160.
22. The reaction rate was slower and enantioselectivity was not any better than at −20°C.
23. See Ref. 20. The metal alkoxide and metal alkyl catalysts are generally much more reactive in the presence of water or
alcohols both of which are formed during the epoxidation reaction.
24. Enantioselectivity of geraniol oxide was 91% using L-(+)-diethyl tartrate as a ligand. See Ref. 12.
25. The polymer was washed throughly with CH2Cl2 and a sample was taken from a known volume for AAS-analysis.
26. An existing catalyzed process is improved by addition of a specific ligand, which leads to a faster, ligand-accelerated
reaction. Both homogeneous and heterogeneous catalysts are known to exhibit this behavior. Berrisford, D. J.; Bolm, C.;
Sharpless, K. B. Angew. Chem. Int. Ed. Engl. 1995, 34, 1059–1070.
27. Hill, J. G.; Rossiter, B. E.; Sharpless, K. B. J. Org. Chem. 1983, 48, 3607–3608.