Synthesis of the first per(3-deoxy)-cyclooligosaccharide: Hepta(manno-3- deoxy-6-O-t-butyldimethylsilyl)-β-cyclodextrin

Article abstract of DOI:10.1016/S0957-4166(99)00388-2

Reduction of hepta(manno-2,3-anhydro-6-O-t-butyldimethylsilyl)-β- cyclodextrin with lithium triethylborohydride gives hepta(manno-3-deoxy-6-O- t-butyldimethylsilyl)-β-cyclodextrin. This compound plus the hepta(2-O- methyl)- and hepta(2-O-benzyl)-derivativ

Full text of DOI:10.1016/S0957-4166(99)00388-2

Pergamon
Tetrahedron: Asymmetry 10 (1999) 3627–3648
Synthesis of the first per(3-deoxy)-cyclooligosaccharide:
hepta(manno-3-deoxy-6-O-t-butyldimethylsilyl)-β-cyclodextrin
David R. Kelly ∗ and Adel K. Mish’al
Department of Chemistry, Cardiff University, PO Box 912, Cardiff CF1 3TB, Wales, UK
Received 9 August 1999; accepted 1 September 1999
Abstract
Reduction of hepta(manno-2,3-anhydro-6-O-t-butyldimethylsilyl)-β-cyclodextrin with lithium triethylborohy-
dride gives hepta(manno-3-deoxy-6-O-t-butyldimethylsilyl)-β-cyclodextrin. This compound plus the hepta(2-O-
4
methyl)- and hepta(2-O-benzyl)-derivatives all have the C1 conformation. Capillary GC columns manufac-
tured with hepta(manno-2,3-anhydro-, hepta(manno-3-deoxy-2-O-methyl- and hepta(manno-2-O-benzyl-6-O-t-
butyldimethylsilyl)-β-cyclodextrin stationary phases were evaluated for enantio-discrimination with 39 non-polar
racemic analytes. The GC column coated with the benzyl derivative showed enantioselectivity comparable to, and
in some cases superior to, a commercial per(methyl)-β-cyclodextrin column. The other columns showed little or
no enantio-discrimination. A thermodynamics study established a linear enthalpy–entropy compensation effect for
two series of analytes on the commercial permethyl-β-cyclodextrin column, but not for the column coated with the
benzyl derivative. © 1999 Elsevier Science Ltd. All rights reserved.
1. Introduction
Despite the enormous effort devoted to the chemistry of cyclodextrins (CDs), derivatives in which
the stereochemistry has been modified are exceedingly rare. The vast majority have the D-gluco-
configuration and 4C1-conformation of naturally occurring cyclodextrins.1,2 There have been three main
approaches to CDs with novel stereochemistry. The de novo synthesis of CDs from ‘amacrocyclic’
precursors3 has yielded manno-stereoisomers of α-, β- and γ-CDs4 amongst others,5 but the synthetic
routes are long and the yields extremely low (<0.5%). Shorter sequences are possible by cleaving
commercial cyclodextrins, chemically modifying the differentiated ends and then recyclising, but this
approach at best can only give a modified disaccharide moiety.6
The singular or multiple modification of commercially available CDs has the potential to deliver short
sequences to novel compounds if adequate selectivity can be achieved, but it is experimentally very
∗
Corresponding author. E-mail: kellydr@cardiff.ac.uk
0957-4166/99/$ - see front matter © 1999 Elsevier Science Ltd. All rights reserved.
PII: S0957-4166(99)00388-2

3628
D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
demanding.7 For example per(3,6-anhydro)-α- and β-cyclodextrins (1C4) have been prepared in just two
steps by selective 6-O-tosylation and intramolecular nucleophilic displacement.8
During the course of a study of novel cyclodextrin, chiral stationary phases (CSPs) for capillary
column gas chromatography,9 we examined the prospects for modifying the stereochemistry of the 2-
and 3-hydroxy groups of cyclodextrins. There are four possible configurations at the C-2/C-3 centres:
gluco, manno, allo and altro (Scheme 1). In gluco-cyclodextrins the hydroxyl groups at C-2 and C-
3, protrude approximately parallel to the C7 axis and extend the length of the chiral cavity on the
secondary (‘top’) face, which is generally believed to be most important for guest binding. Both the
4
allo and altro stereoisomers have axial substituents at C-3 which occupy the central cavity in the C1
4
conformation. altro-Monosaccharides undergo facile interconversion between C1 and 1C4 conformers.
4
manno-Configuration cyclodextrins in the C1 conformation have a more sparse periphery on the
secondary face (relative to the gluco-compounds), with the C-3 substituents approximately parallel to
the C7 axis and the C-2 substituents perpendicular to the C7 axis.10 These observations are also broadly
applicable to α- and γ-cyclodextrins, but are not applicable to higher homologues which have quite
different stereochemical properties.
Scheme 1. Generalised conformations for cyclodextrins, viewed from the secondary (‘top’) face
Hepta(manno-2,3-anhydro-6-O-t-butyldimethylsilyl)-β-cyclodextrin 2, prepared by D’Souza11,12 and
Coleman,13 offered the opportunity to test these possibilities. Surprisingly, there are only two reports of
its reactions with nucleophiles; epoxide cleavage with water and ammonia gave β-cycloaltrin14 and 3-
amino-3-deoxy-β-cycloaltrin, respectively.15 As with related monosaccharide manno-epoxides16 and β-
cyclodextrin manno-monoepoxide17–19 nucleophilic attack occurs at C-3 to give products of trans-diaxial
ring opening. The 1H NMR spectrum of these compounds only showed a single set of resonances with
3J1,2=4.5 and 5.6 Hz, respectively. These coupling constants are intermediate between those expected
for equatorial, equatorial coupling (1.8 Hz) in the 4C1 conformer and the axial, axial coupling (8.2 Hz)
1
expected for the C4 conformer.20 The conclusion that this indicates an equilibrium between the two
conformers is supported by the X-ray crystal structure of α-cycloaltrin (prepared in the same way as
4
1
β-cycloaltrin) which has alternating rings in C1 and C4 conformations.21 With this information in
hand, we elected to reduce the hepta(manno-epoxide) 2 to the hepta(manno-3-deoxy-alcohol) 3 with the
expectation that the 4C1 conformation would be retained. If the structure of gluco-cyclodextrins 1 can be
drawn (in a cartoon) as a cup, then the structure of the hepta(manno-3-deoxy-alcohol) 3 would have the
aspect ratio of a traditional rimmed soup bowl. Many per(6-deoxy)-CDs22,23 have been prepared, but as
far as we are aware the only other examples of deoxy-CDs in the refereed literature24 are mono-2-deoxy-
α- and β-cyclodextrins.6 A large number of deoxy-cyclodextrins have been claimed in a patent, but no
characterisation data was presented.25
2. Discussion
Carefully dried β-cyclodextrin 1a was silylated as described previously26,27 and the hepta(silyl ether)
1b so formed was tosylated selectively at the 2-position (Scheme 2). Coleman was only able to identify

D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
3629
1
and assign two protons on the pyranoside ring (H-1 and H-2), in the H NMR spectrum of the
hepta(tosylate) 1c.28 With the benefit of a higher field instrument (1H, 400 MHz) and a J-COSY spectrum,
the couplings could be traced unambiguously between H-1, H-2 and H-3. The assignment of H-3 was
further confirmed by coupling to HO-3, which in a single spectrum appeared as a sharp doublet (J 3.0
Hz). In all other spectra, a broad multiplet was observed at a virtually identical chemical shift (δ 3.1).
The assignment of the other signals was made on the basis of the other cross-peaks, but some signals
overlapped and there was no redundant information. H-3 (δ 3.84) and H-6 (δ 3.88) are coupled to a two
proton four line multiplet (δ 3.62), which appears to be the superimposition of a triplet due to H-4 (J 9.4
Hz) and a sharp doublet due to H-60 (J 10.2 Hz). The latter value differs slightly from that seen in H-6
(2J 11.6 Hz) due to the overlap with H-4. The multiplet at δ 3.62 is also coupled to a very broad doublet
signal at δ 3.6 (J 9.1 Hz, H-5), which is due to a strong coupling to H-4, a small coupling to H-6 (J 2.9
3
3
Hz) and a very small coupling to H-60. The coupling constants deduced for JH-5,H-6 and JH-5,H-6 are
0
consistent with the common gg orientation for the methylene-silyl ether moiety.
Scheme 2. Synthesis of per(3-deoxy)-cyclomannins. Reagents and conditions: (i) tBuMe2SiCl (1.15 equiv.), imidazole, DMF,
rt, 3 h (92% yield); (ii) pTsCl (2.8 equiv.), pyr., DMAP, 50°C, 25 h (63% yield); (iii) NaH, THF, 4 h (44% yield); (iv) LiEt3BH
(1 M, in THF, 5.1 equiv.), rt, 2 h, LiEt3BH (1 M, in THF, 2.6 equiv.), reflux, 48 h (40% yield); (v) NaH, MeI, THF, rt, 12 h
(90% yield); (vi) NaH, BnBr, THF, rt, 48 h (59% yield). All products were purified by column chromatography over silica gel.
Equivalents are relative to moles of functional groups.
Treatment of the hepta(tosylate) 1c with sodium hydride in THF effected cyclisation to the
hepta(epoxide) 2. This was sufficiently stable that it could be purified by silica gel chromatography and
stored for months in the freezer. In contrast, D’Souza reported that this compound was unstable at room
temperature and decomposed to the silylated cyclodextrin 1b, although presumably, this would require
an unprecedented nucleophilic attack by water at C-2. We also attempted the direct conversion of the
hepta(silyl ether) 1b to the hepta(epoxide) 2 using sodium hydride and benzenesulfonyl chloride, but
1
were not able to isolate a clean product. The H NMR spectrum of the hepta(epoxide) 2 was virtually
identical to that reported previously, except that the signal for H-5 (δ 3.50), which was reported to be a
multiplet appeared as a broad doublet (J 8.7 Hz) for the reasons noted above for the hepta(tosylate) 1c.
Development of a method for the reduction of the hepta(epoxide) 2 required considerable experimen-
tation. Reduction with excess diisobutylaluminium hydride in toluene gave a complex mixture as did
lithium aluminium hydride. Although in the latter case (in retrospect) traces of the desired hepta(alcohol)
1
3 were evident in the H NMR spectrum. Eventually, it was discovered that addition of two portions
of commercial Super-Hydride® (1 M) in THF to the hepta(epoxide) 2, followed by reflux for 48 hours
gave a 40% yield of the hepta(alcohol) 3a, after repeated column chromatography. This material was
1
clearly homogenous as judged by TLC, 13C NMR and mass spectra, however, the H NMR spectrum
showed line broadening of circa 1.5–2 Hz and the signals were poorly dispersed. In view of this and the
unsuitability of the hepta(alcohol) 3a for use as a GC-CSP we prepared the hepta(2-O-methyl) 3b and
hepta(2-O-benzyl) 3c ethers under standard conditions. The hepta(2-O-methyl) 3b ether was intended
for comparison against commercial permethyl-CDs and per(6-O-t-butyldimethylsilyl-2,3-di-O-methyl)-

3630
D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
CDs, whereas the larger benzyl groups of the hepta(2-O-benzyl) 3c ether should help to fill the sparse
secondary periphery due to the absence of the 3-substituent and be comparable with other recently
reported benzylated CD CSPs.29 Both derivatives showed some line broadening, although this was
somewhat less for the hepta(2-O-benzyl) 3c ether and hence most efforts were devoted to the assignment
of the NMR spectra of this compound.
The 1H NMR signals for the anomeric protons of the three hepta(3-deoxy)-compounds 3a,b,c appeared
as a barely resolved doublet of 2.8 Hz, a broad singlet and a barely resolved doublet of 2.4 Hz,
respectively, with similar chemical shifts (δ 4.62, 4.73 and 4.84, respectively). These values are consistent
with equatorial, equatorial coupling, and hence the 4C1 conformation. The 3JH-1,H-2 coupling could not be
confirmed in the signal for H-2, because this appeared as a broad multiplet, because of the line broadening
and second order effects due to the similar chemical shifts of the two H-3s. No coupling constants could
be extracted from 2H-3, H-4 or H-5 either due to multiple coupling and insufficient resolution or signal
overlap. All of the H-6 signals showed the expected large geminal couplings (10.7–11.2 Hz) and small
or zero vicinal couplings to H-5, consequently the methylene-silyl ether moiety is predominantly in the
gg orientation as deduced for the gluco-compounds 1a–c and the hepta(epoxide) 2.
In the 13C NMR spectrum the signal for C-2 of the hepta(alcohol) 3a appears at δ 68.0, and is shifted
downfield in the corresponding hepta(2-O-methyl) 3b (δ 77.7) and hepta(2-O-benzyl) 3c ethers (δ 76.1).
In contrast the corresponding signals for C-1 and C-3 (3a δ 100.5, 32.9) are shifted slightly upfield (3b
δ 99.3, 31.2; 3c δ 99.2, 31.4).18 The chemical shifts for C-5 and C-6 are virtually identical in all three
compounds.
All compounds were characterised by combustion analysis and FAB-MS using a m-nitrobenzyl
alcohol matrix. No partially functionalised CDs, dimers or multiply charged ions, were observed albeit
that in some cases the ion abundances were low relative to matrix ions. The hepta(tosylate) 1c, the
hepta(epoxide) 2 and the hepta(methyl ether) 3b gave M+Na peaks as is commonly observed for CDs,
however the hepta(2-O-benzyl) ether 3c gave a M+H+Na ion and the hepta(alcohol) 3a gave an ion which
can be rationalised as M−CH4+Na.
The spectroscopic data establishes that the hepta(deoxy)-CDs 3a,b,c have the 4C1 conformation with
the methylene-silyl ether moiety predominantly in the gg orientation. An impression of the structure of
the hepta(2-O-benzyl) ether 3c is shown in Fig. 1. The orientation of the benzyl groups is arbitrary.
3. Capillary column GC
The virtues of using capillary GC to measure enantio-discrimination are not widely appreciated. A
typical 30 metre capillary column only contains about 3 milligrams of stationary phase of which less
than 1 milligram is the enantio-discriminating component. Hence the amount required is rather less than
that employed in NMR and, moreover, given that the CSP is stable, any number of analytes can be tested
without having to separate the guest from the host.
Capillary columns (12 m, 0.25 mm id, 0.20 µm phase thickness) were manufactured using the static
method, with OV-1701 as the diluent (diluent:CD ratio of 85:15) and dichloromethane as the solvent. The
columns are designated as follows (CM=cyclomannin): commercial J & W per(methyl)-CD, permethyl-
CD; hepta(epoxide) 2, h(E)-CM; hepta(2-O-methyl) 3b ether, h(Me)-CM; and hepta(2-O-benzyl) 3c
ether, h(Bn)-CD.
Most GC runs were isothermal and most analytes were run at four temperatures in the range 30–100°C.
Racemic analytes (Scheme 3) which were particularly difficult to separate were also run with the

D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
3631
Figure 1. Structure of the hepta(2-O-benzyl) ether 3c viewed from the secondary face
temperature programme 40°C (5 min)×5°C/min, 100°C (40 min). These are marked P in Table 1. In
a few cases, the same analyte was run at slightly different temperatures on the two columns.
Scheme 3. Analytes
This work was performed as part of a study of CSPs which are enantioselective at low temperature.
To reduce otherwise long retention times, we used comparatively short (12 m) and narrow columns (0.25
mm id). A mass spectrometer was used as the detector and naturally the ion source is at a low vacuum.
The use of narrow columns reduces the length of column which is subjected to an appreciable vacuum
and hence increases the usable length of the column. The 12 metre columns used in the current work
have a resolution equivalent to 65% of that of a 30 metre column.
We were aware of our limitations for manufacturing capillary columns with an even coating of
stationary phase. Consequently it was assumed that the separation factor was solely a function of the
enantio-discrimination of the CSP and that peak shape was a function of this and any inadequacies in the
quality of column coating. Hence our analysis of the enantio-discrimination is based on separation factors
(α) rather than efficiency (theoretical plates). In virtually all cases, peak width increased as the retention
time increased for both the permethyl-CD column and the columns manufactured in house. This effect is
commonly observed and was particularly noticeable with the higher members of the homologous series
of analytes.
The h(E)-CM column showed little or no enantio-discrimination, and the retention times were ge-
nerally shorter than the other columns e.g. bicyclic ketones 12a–d, tR (40°C) 4.00, 7.25, 14.28, and
33.14 min, respectively (cf. Table 1). The possibility that decomposition of the CSP was occurring was

3632
D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
Table 1
Retention times (minutes) and separation factors (α) for racemic analytes on permethyl-CD and
h(Bn)-CM columns
considered, however the lowest possible oven temperatures were used and the retention times of analytes
measured at the beginning of the study were identical to those at the end of the study.
The h(Me)-CM column also had little or no enantio-discrimination, although separation of 1-
phenylpropanol 11b was just about achieved (α=1.02, at 60°C, tR=10.13 and 10.35 min). As with the
h(E)-CM column, retention times were comparable to or shorter than the other columns, e.g. bicyclic
ketones 12a–d, tR (40°C) 5.97, 11.27, 23.03, and 57.74 min, respectively (cf. Table 1).

D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
3633
Despairing of achieving any useful results with these 3-deoxy-manno derivatives, we finally turned to
the H(Bn)-CM column and were rewarded with some excellent results. Table 1 summarises the results
obtained at the lowest usable temperatures with this column and the permethyl-CD column. When one
column is clearly superior to the other, the data is shown in bold.
3.1. Terpenes, 4–10
The performance of the two columns with the terpenes 4–10 are largely comparable. Although the
permethyl-CD column has a better α-factor with α-ionone 6, the peak for the later eluting enantiomer
showed trailing, whereas with the h(Bn)-CM column both peaks had a satisfactory shape and that of the
later eluting enantiomer was marginally sharper. The retention time differences between the columns for
the ketones are remarkable. The α-ionone enantiomers 6 elute on average some 8.5 min later from the
h(Bn)-CM column than from the permethyl-CD dextrin column, whereas with carvone 9 the retention
time is 130% greater, albeit that there is no enantio-discrimination on either column.
3.2. Phenylalkanols, 11a–d
The 1-phenylalkanols are standard test analytes for testing CSPs. The (R)- and (S)-enantiomers
of phenylethanol 11a, phenylpropanol 11b, and phenylbutanol 11c are commercial products and (S)-
phenylpentanol 11d was prepared by yeast reduction of valerophenone 32. Work up and column
chromatography gave the desired enantiomer in 8% yield (66% ee by optical rotation, 45% ee by CSP
GC) which was comparable with the literature values (5–37% yield, 71–55% ee by optical rotation).30
On all the CD columns we have examined which are capable of separating the enantiomers of the
1-phenylalkanols 11a–d, the (R)-enantiomer of the ethanol 11a and propanol 11b homologues, and the
(S)-enantiomers of the butanol 11c and pentanol 11d elute first. This can be rationalised as a preference
for the phenyl group binding into the cavity of the CD for the (S)-enantiomers of the lower homologues31
and the alkyl chain of the (R)-enantiomers of the higher homologues. Whatever the explanation, 1-
phenyl-butanol 11c is generally the most difficult to separate. The permethyl-CD column was incapable
of resolving the enantiomers of the butanol homologue 11c, whereas they were cleanly separated by
the h(Bn)-CM column. The enantiomers of phenylpentanol 11d were separated by the permethyl-CD
column, but the peaks trailed so badly that the peaks were barely distinguishable whereas on the h(Bn)-
CM column, the peak shape was much better and the ∆t was over 5 min. Further aspects of the separation
of phenylalkanols on these columns are discussed in the section on thermodynamics.
The explanation for the reversal in enantioselection in the phenyl-alcohol 11a–d homologous series
is wholly reasonable, however the assignments of the absolute configurations of the higher homologues
are not based on unambiguous methods. The absolute configurations of phenylethanol 11a32 and phenyl-
propanol 11b33 were established by syntheses which can be traced to D-glyceraldehyde, whereas those
of phenylbutanol 11c34 and phenylpentanol 11d were initially assigned on the basis of Freudenberg’s
‘rule of shift’,35 i.e. configurationally related compounds show similar shifts in rotation upon derivati-
sation. The configurations assigned are consistent with those predicted by Brewster’s groups increment
calculations,36 Horeau partial esterification,37 and yeast reduction of the corresponding ketones rationa-
lised using Prelog’s rule.38 The absolute configuration of phenylpentanol 11d is a matter of considerable
importance because it is the product of innumerable reactions in which butyl anions are added to
benzaldehyde in the presence of a chiral auxiliary. Sensing that a revision of this absolute stereochemistry
would prompt reformation of a large body of asymmetric chemistry, we sought a crystalline derivative
with a centre of known stereochemistry whose structure could be determined unambiguously by X-

3634
D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
Figure 2.
ray crystallography. The putative (S)-1-phenyl-1-pentanol 11d, produced by yeast reduction, was reacted
with (R)-α-methoxy-α-trifluoromethylphenylacetyl chloride (MTPA-Cl) in the presence of DMAP. Work
up and repeated column chromatography gave the Mosher’s ester as a yellowish liquid (58% yield).
Crystallisation was attempted using hexane, benzene, ether, petroleum ether and mixtures thereof, but no
crystals could be obtained. Even resort to sonocrystallisation was unsuccessful.39
With this lack of success we also resorted to an indirect method. In the preferred conformation
of MPTA esters of secondary alcohols the trifluoromethyl group, the ester carbonyl and the alkanol
hydrogen are synperiplanar (Fig. 2).40 If the carbonyl group is modelled as a tau bond,41 this is a
fully staggered conformation. The absolute configurations of diastereomeric MTPA esters can usually
1
be assigned from H NMR spectra, by assuming that the phenyl group of the acyl moiety shields the
hydrogens of groups which are in the syn orientation (typically ∆δ 0.05–0.13 ppm).42 In the current
case, the requisite heavily coupled methylene groups in the two enantiomers could not be distinguished.
However the signal for the methoxy group of the major diastereoisomer (δ 3.43) was 0.09 ppm upfield
from that of the minor diastereoisomer (δ 3.52).43,44 This can be attributed to a comparable shielding
effect by the phenyl ring of the alkyl moiety of the ester. Given that the predominant conformation of
the (S)-MTPA-ester in solution is that shown in Fig. 2, only the (S)-enantiomer of phenylpentanol 11d
has the methoxy group syn to the phenyl ring. This correlation has been noted previously for certain
groups of compounds45 and in a forthcoming publication we will show that this is a general and useful
observation.46
3.3. Bicyclo[3.2.0]hept-2-ene-6-ones, 12a–d
The enantiomers of the bicyclic ketones 12a–d were easily distinguished by both columns, the α-
factors for these compounds were the highest obtained in this study and no pair of enantiomers had a
difference in retention time of less than 1 min. The ∆t for the endo-propyl-homologue 12d was almost
14 min on the permethyl-CD column and over 6 min on the h(Bn)-CM column, albeit the peaks had
extensive trails at this temperature. Further aspects of the separation of the bicyclo[3.2.0]hept-2-ene-6-
ones on these columns are discussed in the section on thermodynamics.
3.4. Lactones, 13–15
The h(Bn)-CM column was clearly superior to the permethyl-CD column for the lactones 13–16.
Neither column was capable of separating β-butyrolactone 13, but all the other examples 14, 15a,b were
separated by the h(Bn)-CM column albeit with low, but usable separation factors.
3.5. Alcohols, 16–24
Both columns had comparable performance with each of the analytes in this group, although in all
cases the permethyl-CD column was slightly superior. The separation factors are moderate to low, but

D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
3635
this is mitigated by the short retention times. These compounds are amongst the most volatile used in the
study and the lack of separation is probably due to insufficient residence time in the liquid phase. The
homologous series of 2-alkanols 24a–f was poorly separated with either column and no improvement was
achieved by using a temperature programmed run. Most CD-CSPs give poor results with these analytes.
3.6. Sundries, 25–31
The enantiomers of styrene oxide 25 were almost equally separated on the permethyl-CD (∆t=0.89
min) and h(Bn)-CM columns (∆t=0.76 min), but the retention times on the h(Bn)-CM column were
almost twice as long. This presumably reflects the greater solubility of styrene oxide in the benzylated
CSP. 1,7-Dioxaspiro[5.5]undecane 26 is the female released sexual attraction pheromone of the olive
fly, Bactrocera oleae.47 The enantiomeric excesses of this compound and hydroxylated derivatives have
been determined by GC on a CD-CSP48 and the modes of binding to CDs investigated by NMR.49 The
separation factor and the peak shape were better on the h(Bn)-CM column (α 1.09) than on the permethyl-
CD column (α 1.04), although the retention time was slightly longer. Both columns gave almost identical
and excellent separations (α 1.15) of the enantiomers of 1,2-O-isopropylidene-3-O-methyl glycerol 27,
but none of the trifluoroacetylated amines 28–30 were separated. DL-Valine methyl ester 31 was only
separated with the permethyl-CD column.
The thermal stability of the hepta(2-O-benzyl) ether 3c was assessed by thermogravimetric analysis.
Weight loss commenced slowly at 260°C and the majority of the weight was lost between 325 and
380°C. This indicates that the maximum temperature for the column should be conservatively set to
180–200°C, which is somewhat less than commercial permethyl-CD columns (260/280°C). One possible
interpretation is that the axial benzyloxy substituents undergo thermal elimination more rapidly than the
equatorial methoxy groups present in the permethyl-CD columns.
4. Thermodynamics study
The h(Bn)-CM column is the first cyclodextrin column which does not have gluco-stereochemistry
and moreover benzyl-substituted cyclodextrins are also comparatively rare. It appeared that it would be
worthwhile to undertake a thermodynamic study in order to discover if the mode of discrimination was
fundamentally different to that of other cyclodextrin-CSPs. We considered it plausible that the shallower
cavity plus the more flexible and polarisable rim of the hepta(2-O-benzyl) ether 3c would result in looser
and more disordered binding of analytes relative to permethyl-CD. Consequently enthalpic effects should
play a larger role than entropic effects. As in the prior study, data was acquired for both the new column
and a commercial permethyl-CD column. Two homologous series, the phenyl alkanols 11a–d and the
7-alkyl-bicyclo[3.2.0]hept-2-en-6-ones 12a–d, were investigated so that trends such as enthalpy–entropy
compensation could be distinguished.
Data for the calculation of thermodynamics parameters were acquired by measuring retention times
for each homologous series at a number of temperatures. As in the prior work, mixtures of homologues
were analysed concurrently, so as to minimise errors in the measurement of retention times. The range of
temperatures used was dictated by the need to obtain reproducible retention times without excessive peak
broadening and also to avoid excessively short retention times which are less reproducible. The results
presented in Tables 2 and 3 were calculated using the method of Klug et al.50 which yields ∆∆H and
∆∆GThm (∆∆G at the harmonic mean temperature of the experiments). These parameters were then used

3636
D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
Table 2
Thermodynamic study of 1-phenyl-1-alkanols 11a–d permethyl-CD (Tm 348.16 K, Thm 347.80 K,
Tc 411 K)
to calculate ∆∆S in the usual way. Plots of lnα vs 1/T plots were also made as a check and because they
are easier to interpret visually.
As expected, the α-factors decrease with increases in temperature for all the analytes on all the
columns. On the permethyl-CD column the phenylethanol 11a and phenylpropanol 11b enantiomers
had comparable values for both ∆∆S and ∆∆H as indicated by parallel lines on the lnα vs 1/T plot but
phenylpentanol 11d showed much smaller values for both. On the h(Bn)-CM column the phenylpropanol
11b and phenylpentanol 11c enantiomers had comparable values for both ∆∆S and ∆∆H, although
because of the change in order of elution this represents opposing effects for pairs of enantiomers.
Generally, the absolute values of ∆∆H, ∆∆S are larger on the permethyl-CD column than on the
h(Bn)-CM column. Phenylpentanol 11d and phenylbutanol 11c are the exceptions to this generalisation.
Unfortunately, the span of the α-factors for phenylbutanol 11c on the h(Bn)-CM column is only 0.009
and hence any calculations are ill-conditioned and unreliable. For example, the values of ∆∆H/∆∆S
(based on the maximum uncertainties in the measurement of the retention times) cover a range of almost
100°C. Nevertheless the poor quality of the thermodynamic data, does not detract from the practically
useful result that only the h(Bn)-CM column was able to separate the enantiomers of phenylbutanol 11c.
The large error for bicyclo[3.2.0]hept-2-en-6-one 12a on the h(Bn)-CM column is largely a consequence
of the measurement at 70°C (and the short retention time) and the probable error in ∆∆H/∆∆S is likely
to be circa ꢀ20°C.
The values of ∆∆H and ∆∆S in Tables 2 and 3 have a fairly constant ratio which is conveniently
expressed as ∆∆H/∆∆S. This ratio is also the temperature at which no separation of enantiomers
occurs (see Experimental for explanation). Apparently constant ratios between enthalpy and entropy
are frequently observed from equilibrium (or kinetic) measurements. This phenomenon is variously
known as enthalpy–entropy compensation, the isokinetic effect (for rate studies) or just the compensation
effect.51,52 However demonstration of this effect by plotting ∆∆H against ∆∆S is misleading because the

D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
3637
Table 3
Thermodynamic study of bicyclo[3.2.0]hept-2-en-6-ones 12a–d permethyl-CD (Tm 333.16 K,
Thm 332.56 K, Tc 413 K)
major axis of the error ellipse runs is coincident with the regression line of the data when these parameters
are determined from van’t Hoff (or Arrhenius) plots. If the experimental temperature range is too small
the error ellipse approximates to a straight line (major axis>>>minor axis).53 Consequently, as the errors
in the data increase, the coefficient of regression of the line increases because the data is spread over
a wider span. This problem may be largely avoided by plotting ∆∆H vs ∆∆GThm which enables the
compensation temperature (Tc) to be calculated from the gradient of the regression line (Fig. 3).50,54
The plot for the phenylalkanols on the permethyl-CD column (Fig. 3A) is compromised by the limited
number of data points and clustering of the values for 11a and 11b. The values for phenylpentanol 11d
are plotted as positive differences, although they are actually negative differences. This originates from
the change in elution order of the enantiomers (11a, 11b, α R/S; 11c, 11d α S/R). Plotting the values for
phenylpentanol 11d as negatives increases the relative clustering of the the two higher data points and
unduly weights the regression line relative to the significance of the experimental value. This is also a
feature of the phenylalkanols on the h(Bn)-CM column (Fig. 3B) and it is noteworthy that both ∆∆H
and ∆∆S decrease in sequence with ‘ascent’ of the phenylalkanol homologous series 11a–d on both
columns. The data for the bicyclo[3.2.0]hept-2-en-6-ones 12a–d on permethyl-CD has the highest span
for the separation factors and temperature (Table 3) and consequently the thermodynamic parameters
have the lowest errors in the current study. Fig. 3C shows a high quality regression line for well dispersed
values. The bicyclo[3.2.0]hept-2-en-6-ones 12a–d on the h(Bn)-CM column gave a poor regression line
(Fig. 3D), largely as a consequence of the data for the parent unsubstituted compound 12a and the poor
dispersion of the data points.
Overall, the results provide evidence for the operation of an enthalpy–entropy compensation effect
for the the permethyl-CD column, but not on the h(Bn)-CM column. This is most clearly seen for
the bicyclo[3.2.0]hept-2-en-6-ones 12a–d on permethyl-CD and to a much lesser extent for the phenyl-
alkanols 11a, 11b, 11d on the same column, because of limitations of the data. Both the gradients and

3638
D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
Figure 3. ∆∆H vs ∆∆GThm for 1-phenyl-1-alkanols 11a–d and bicyclo[3.2.0]hept-2-en-6-ones 12a–d
intercepts for these data sets are similar (although the Thm are slightly different) and the compensation
temperatures (Tc 411 and 413 K, respectively), are virtually identical. When the two data sets are
combined (Fig. 4, T 332.56K) the regression line has marginally smaller residuals than either data set
alone. In contrast, the results for the h(Bn)-CM column show poorer quality correlations, albeit based
on poorer quality data (due to shorter spans for α and temperature range, Fig. 3B and 3D). When the
two data sets were combined (T 342.97 K) and replotted in the same way as for the two permethyl-CD
data sets, the residuals were poorer (R=0.72). The results do not wholly exclude an enthalpy–entropy

D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
3639
Figure 4. ∆∆H vs ∆∆GThm for 1-phenyl-1-alkanols 11a, 11b, 11d and bicyclo[3.2.0]hept-2-en-6-ones 12a–d on permethyl-CD
compensation effect for the individual homologous series of analytes on this column, but do not provide
sufficient good quality evidence to support it.
Armstrong has studied the thermodynamics of the separation of trifluoroacetylated alcohols and
amines with a capillary column coated with 2,6-di-O-pentyl-3-O-trifluoroacetyl-β-cyclodextrin. Analytes
were divided into two groups, those with ∆∆S and ∆∆H greater than 18.0 J K−1 mol−1 and 7.5 kJ mol−1,
respectively (Group I), and those with parameters below these values (Group II). Comparable values
were adduced for a homologous γ-cyclodextrin column. It was postulated that the group I analytes are
resolved by differential inclusion of the enantiomers within the cavity of the cyclodextrin, whereas the
group II analytes are resolved by a looser association possibly with the exterior of the cyclodextrin.55 This
rationale parallels our postulated mechanisms for resolution by permethyl-CD and hepta(2-O-benzyl)
ether 3c. Accordingly, if the thermodynamic parameters for the analytes on the permethyl-CD column
have larger absolute values than those for the h(Bn)-CM column, it supports the postulated mechanistic
differrence. As noted above, pairwise comparison of ∆∆S and ∆∆H for individual analytes on the
two columns support this proposal (Tables 2 and 3). All the analytes have larger values of ∆∆S and
∆∆H on the permethyl-CD column than on the h(Bn)-CM column, except for phenylbutanol 11c and
phenylpentanol 11d. This is further supported by the list of values ranked by descending ∆∆S (Table 4)
in which the top of the Table is dominated by separations on the permethyl-CD column. However there
is no clean break in the sequence. Four minor breaks in the ranges can be distinguished but they are
probably not significant.
It has been suggested that smaller absolute values for ∆∆H and ∆∆S are found for analytes with
greater conformational mobility.56 In the current case increased conformational mobility should parallel
ascent of the homologous series. Inspection of Table 4 shows that this trend is followed at best, erratically.
For example, although four of the five highest ranked entries are due to the first or second member of
the homologous series, the highest and lowest ranked entries are from the third member of homologous
series.
In summary, the ratio between the difference in the enthalpy and entropy of binding for the pairs of
enantiomers appears to indicate that energetically favourable binding requires tight binding. In the case
of the two series of analytes 11a–d, 12a–d on the permethyl-CD column this ratio is constant, whereas
for the h(Bn)-column a range of similar values were observed. This is consistant with a range of different
modes of enantio-discrimination.

3640
D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
Table 4
Thermodynamic parameters ranked according to ∆∆S
5. Conclusions
In summary, this data set indicates that neither the gluco-configuration at C-2 and C-3, nor a fully
substituted secondary rim is required for enantio-discrimination by cyclic-oligosaccharides when used
as GC-CSPs. However methyl substitution alone in the 3-deoxy-manno-series is insufficient to confer
enantio-discrimination.
6. Experimental
6.1. General conditions, equipment and software
Purified or dried solvents were freshly distilled under an argon or nitrogen atmosphere from a suitable
drying agent.
All reactions were monitored by thin layer chromatography (TLC), which was run on 0.2 mm Merck
aluminium backed precoated silica gel plates (60 F254) with UV light or ethanolic phosphomolybdic
acid (3%) and heat as developing agents. Merck silica gel 60 (70–230 mesh) was used for flash column
chromatography.
NMR spectra were recorded on a Bruker Advance DPX-400 spectrometer with UXNMR software
at 400 MHz for protons and 100 MHz for carbon-13. Chemical shifts are relative to tetramethylsilane
as an internal standard or the spectrometer reference for the solvent. All integrations for cyclodextrins
are reported as monomeric sugar units and hence the true ratios are seven times those reported. Coupling
constants patterns were analysed using the computer program Multiplet (release NMRUC51, D. R. Kelly,
unpublished work) and are quoted in hertz (Hz). They are reported to 0.1 Hz, but have an uncertainty
of ꢀ0.3 Hz, due to the digital resolution of the FID accumulation and Fourier transformation. Spin
simulations were performed using RACCOON (P. F. Schatz, University of Wisconsin).
Infra red (IR) spectra were recorded on a Perkin–Elmer 1600 Series FTIR spectrophotometer using

D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
3641
sodium chloride cells. Optical rotations were determined at the sodium D line on an Optical Activity
AA-1000 polarimeter at ambient temperatures.
Fast atom bombardment (FAB) mass spectra were recorded at the EPSRC Mass Spectrometry Centre at
Swansea. All of the compounds in this report were above the mass limit for FAB-MS PEG referenced high
resolution mass spectra at the Swansea facility. Ion cluster patterns were calculated using the computer
program HiMass (D. R. Kelly, 1991, unpublished work) and compared visually with the spectra. Ion
clusters with normalised intensity differences of less than 5% from that calculated were deemed identical.
GC–MS was run on an HP5890 gas chromatograph linked to a dual pumped (250 l/s, 50 l/s) Trio-1
mass spectrometer with an EI source at 70 eV. Helium head pressure 8 psi. Injector temperature 200°C,
split 10:1. All commercial capillary columns used for purity evaluation were J & W (30 m×0.32 mm, 0.25
µm). Column evaluation trials were performed with 12 m columns (see below). Spectra (m/e 36–600)
were acquired in 0.9 s, followed by an interscan time of 0.1 s, hence retention times have an uncertainty
of 0.033 min.
Thermogravimetric analysis (TGA) data were measured on Stanton Redcroft, STA-780 instrument,
heating at 10°C/min. Melting points were determined in capillary tubes using a Gallenkamp melting
point apparatus and are uncorrected. Elemental analyses were performed on a Perkin–Elmer 240c.
6.2. Synthesis
6.2.1. 6A,6B,6C,6D,6E,6F,6G-Hepta-O-(tert-butyldimethylsilyl)-2A,2B,2C,2D,2E,2F,2G-hepta-O-(p-tol-
uenesulfonyl)-β-cyclodextrin, 1b
4-(N,N-Dimethylamino)pyridine (6.5 g, 53 mmol) and p-toluenesulfonyl chloride (10.5 g, 55 mmol)
were added with stirring at room temperature to a mixture of hepta(6-O-tert-butyldimethylsilyl)-β-
cyclodextrin 1b (5.4 g, 2.8 mmol) in dry pyridine (250 ml). The mixture was stirred for 24 h at 50°C.
Water (50 ml) was added to the reaction mixture which was concentrated under reduced pressure. The
residue was extracted with ethyl acetate (2×250 ml). The extracts were washed with hydrochloric acid
(2N, 2×125 ml), saturated sodium hydrogen carbonate (75 ml) and water (2×125 ml), and then dried over
sodium sulfate. TLC showed the product at Rf 0.35 (CHCl3:acetone, 95:5). Column chromatography
(CHCl3:acetone, 95:5) gave the title compound 1c as a white powder (5.3 g, 63%). M.p. 172–174°C;
m/z (FAB+, NOBA matrix), 8.5%, M+Na+, found 3036, C133H210O49Si7S7Na: requires 3036. Anal.
C
133H210O49Si7S7: requires C, 52.99, H, 7.02; found: C, 52.71, H, 6.84; [α]20 +62.1 (c 1.11, CHCl3),
D
+57 (c 1.0 CHCl3); δH (400 MHz, CDCl3) 7.73 (2H, d, J 8.3, Ts), 7.30 (2H, d, J 8.3, Ts), 5.25 (1H,
d, J 3.7, H-1), 4.27 (1H, dd, J 9.9, 3.7, H-2), 3.88 (1H, dd, J 11.6, 2.9, H-6), 3.84 (1H, dd, J 9.3, 9.3,
H-3), 3.62 (2H, four lines, ‘quartet-like intensities’, separations 9.4, 8.6, 10.2 Hz, H-4 and H-60), 3.46
(1H, br d, J 9.1, H-5), 3.1 (1H, br or d, J 3.0, 3-OH), 2.43 (3H, s, Ts methyl), 0.85 (9H, C(CH3)3), 0.0
(6H, Si(CH3)2); δH (400 MHz, COSY, CDCl3) 5.25–4.27 (H-1 to H-2), 4.27–3.84 (H-2 to H-3), 3.88 to
3.62 (H-6 to H-60), 3.84–3.62 (H-3 to H-4), 3.62–3.46 (H-4 to H-5); δC (100 MHz, CDCl3) 144.1, 131.9,
128.6, 127.3, 97.8 (C-1), 79.0 (C-2), 78.8 (C-4), 70.6 (C-3), 68.9 (C-5), 60.61 (C-6), 24.8, 20.7, 17.20
[(CH3)3Si], −2.0 and −2.1 [(CH3)2Si].
6.2.2. 6A,6B,6C,6D,6E,6F,6G-Hepta(O-tert-butyldimethylsilyl)-2A ,3A-2B,3B-2C,3C-2D,3D-2E,3E-,2F,3F-
2G,3G-hepta(anhydro)-hepta(manno)-β-cyclodextrin, 2
To a solution of anhydrous hepta(6-O-tert-butyldimethylsilyl-2-O-tosyl)-β-cyclodextrin 1c (4.5 g, 1.5
mmol) in dry THF (300 ml) was added sodium hydride (0.9 g, 60% in mineral oil, 23 mmol). The mixture
was stirred at 60°C for 4 h. Methanol (300 ml) was added slowly to end the reaction. Hydrochloric acid
(2N) was added to pH 7, and the mixture was concentrated under reduced pressure. The residue was

3642
D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
extracted with ethyl acetate (2×400 ml) and then dried over sodium sulfate and concentrated. Column
chromatography (ethyl acetate:dichloromethane, 13:87) gave the title compound 2 as a white solid (1.2
g, 44%). M.p. 139–140°C; m/z (FAB+, NOBA matrix), 5%, M+Na+, found 1831, C84H154O28Si7Na:
requires 1831. Anal. C84H154O28Si7: requires C, 55.78, H, 8.58; found: C, 54.63, H, 8.66. [α]20 +80.3
D
(c 1.0, CHCl3), lit. +83.0 (c 1.0 CHCl3);13 δH (400 MHz, CDCl3) 5.17 (1H, s, H-1), 4.11 (1H, d, J 9.0,
H-4), 3.86 (1H, dd, J 11.6, 3.1, H-60), 3.63 (1H, d, J 11.1, H-6), 3.50 (1H, br. d, J 8.7, H-5), 3.29 (1H,
d, J 3.6, H-2), 3.05 (1H, d, J 3.5, H-3), 0.84 (9H, C(CH3)3), 0.0 (6H, Si(CH3)2); δH (400 MHz, COSY,
CDCl3) 5.17–3.86 (H-1–H-2), 3.86–3.05 (H-2 to H-3), 3.05–4.11 (H-3 to H-4), 4.11–3.50 (H-4 to H-
5), 3.50–3.86 (H-5 to H-6); δC (100 MHz, CDCl3) 97.0 (C-1), 69.9 (C-5), 68.8 (C-4), 62.9 (C-6), 53.8
(C-2), 49.5 (C-3), 26.0, 18.6 [(CH3)3Si], −4.8 and −4.9 [(CH3)2Si]; δH+δC (CDCl3), 5.17–97.0 (H-C-
1), 4.11–68.8 (H-C-4), 3.83 and 3.63–62.9 (2H-C-6), 3.50–69.9 (H-C-5), 3.29–53.8 (H-C-2), 3.05–49.5
(H-C-3).
6.2.3. 6A,6B,6C,6D,6E,6F,6G-Hepta(O-tert-butyldimethylsilyl)-3A ,3B,3C,3D,3E,3F,3G-hepta(deoxy)-
hepta(manno)-β-cyclodextrin, 3a
6.2.3.1. Diisobutylaluminium hydride. Diisobutylaluminium hydride solution (0.7 ml, 1.5 M in tol-
uene) was added slowly at 0°C to a solution of hepta(manno-2,3-anhydro-6-O-t-butyldimethylsilyl)-β-
cyclodextrin 2 (250 mg, 0.14 mmol) in toluene (20 ml) under argon. The solution was stirred at room
temperature for 2 h, then refluxed for 24 h at 60°C. Methanol (10 ml) was added slowly to quench the
reaction. The mixture was concentrated under reduced pressure and redissolved in dichloromethane. The
mixture was extracted with 2N hydrochloric acid (5 ml), saturated sodium hydrogen carbonate (10 ml)
and water (10 ml), and then dried over sodium sulfate. Flash silica gel column chromatography (eluent
1
chloroform:methanol, 96:4 to 80:20) gave four fractions. H NMR spectra indicated that these were
epoxy-alcohols resulting from incomplete reduction. No attempt was made to determine the complete
structures.
6.2.3.2. Lithium aluminium hydride. A solution of hepta(manno-2,3-anhydro-6-O-t-butyldimethyl-
silyl)-β-cyclodextrin 2 (250 mg, 0.14 mmol) in dry THF (10ml) was added slowly to lithium aluminium
hydride (0.38 g, 10 mmol) at room temperature under argon. The solution was stirred for 5 h. Methanol
(10 ml) was added slowly to quench the reaction. The mixture was concentrated under reduced pressure
and redissolved in dichloromethane. The mixture was extracted with hydrochloric acid (2N, 5 ml),
saturated sodium hydrogen carbonate (10 ml) and water (10 ml), and then dried over sodium sulfate.
Column chromatography (eluent chloroform:methanol, 96:4 to 80:20) showed partially reduced product,
plus traces of the desired product 3a δH (400 MHz, CDCl3) 4.60–4.70 (1H, H-1), 4.0–3.66 (7H, broad
peaks, H-2,3,5 and 6), 1.85–2.00 (2H, broad peaks, H-3)
6.2.3.3. Super-Hydride®. Lithium triethylborohydride solution (10 ml, 1.0 M in THF, 10 mmol) was
added slowly to cyclo-α-1,4-hepta[-6-O-tert-butyldimethylsilylmanno-2,3-epoxide] [81] (510 mg, 0.28
mmol) under argon. The solution was stirred at room temperature for 2 h. Lithium triethylborohydride
solution (5 ml, 1.0 M in THF, 5 mmol) was added and the reaction refluxed for 48 h at 60°C. Methanol (10
ml) was added slowly to quench the reaction. The mixture was concentrated under reduced pressure and
redissolved in dichloromethane. The mixture was extracted with hydrochloric acid (2N, 5 ml), saturated
sodium hydrogen carbonate (10 mL) and water (10 ml), and then dried over sodium sulfate. Repeated
flash silica gel column chromatography (eluent chloroform:methanol, gradient 96:4 to 80:20) gave the
title compound 3a as a white solid (0.200 g, 40%). M.p. 152–153°C; m/z (FAB+, NOBA matrix), 4%,

D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
3643
M−CH4+Na+, found 1829, C83H164O28Si7Na: requires 1829. Anal. C84H168O28Si7: requires C, 55.35,
H, 9.30; found: C, 55.22, H, 10.18. We were not able to obtain a value for hydrogen within acceptable
limits; [α]20 +87.6 (c 2.00, CHCl3); δH (400 MHz, CDCl3) 4.62 (1H, d, J 2.8, H-1), 4.00 (1H, m, H-4),
D
3.86 (1H, dd, J 10.9, 3.1, H-6), 3.73 (2H, m, H-2 and H-5), 3.66 (1H, dd, J 10.7, 2, H-60), 1.95 (2H, m, H-
3), 0.85 (9H, C(CH3)3), 0.0 (6H, Si(CH3)2), all peaks were broadened at ambient temperatures; δH (400
MHz, COSY, CDCl3) 4.62–3.73 (H-1 to H-2), 3.73–1.95 (H-2 to H-3), 1.95–4.00 (H-3 to H-4); δC (100
MHz, DEPT, CDCl3) 100.5 (CH, C-1), 74.5 (CH, C-5), 70.4 (CH, C-4), 68.0 (CH, C-2), 62.8 (CH2, C-
6), 32.9 (CH2, C-3), 26.3 (CH3, [(CH3)3CSi]), 18.7 (C, [(CH3)3CSi]), −4.8 and −4.9 (CH3, [(CH3)2Si]);
δH+δC (CDCl3) 4.62–100.5 (H-C-1), 4.00–70.4 (H-C-4), 3.86 and 3.66–62.8 (2H-C-6), 3.73–74.5 (H-
C-5), 3.70–68.0 (H-C-2), 1.95–32.9 (2H-C-3), 1.95–32.9 (2H-C-3), 0.85–26.3 (9H-3C-[(CH3)3CSi]).
6.2.4. 6A,6B,6C,6D,6E,6F,6G-Hepta(O-tert-butyldimethylsilyl)-2A ,2B,2C,2D,2E,2F,2G-hepta(O-methyl)-
3A,3B,3C,3D,3E,3F,3G-hepta(deoxy)-hepta(manno)-β-cyclodextrin, 3b
Sodium hydride (0.4 g, 60% mineral oil) was washed twice with petroleum ether and left to dry
under nitrogen; a solution of hepta(6-O-tert-butyldimethylsilyl-3-deoxy)cyclomaltoheptaose [82a] (0.2
g, 0.11 mmol) in dry THF (3 ml) was added slowly to the neat sodium hydride. Methyl iodide (0.3 ml,
4.7 mmol) was then added dropwise (using a syringe under argon) to the first solution. The reaction
mixture was stirred overnight at room temperature. Methanol was added to decompose the excess of
hydride, the solvents were evaporated, and a solution of the residue in dichloromethane (10 ml) was
washed with water, dried, and concentrated. Column chromatography (4:1 hexane:ethyl acetate) gave
white crystals [82b] (0.19 g, 90%). M.p. 124–125°C; m/z (FAB+, NOBA matrix), 1943 (19%, M+Na+,
C91H182O28Si7Na: requires 1943). Anal. C91H182O28Si7: requires C, 56.90, H, 9.56; found: C, 57.11, H,
10.02. [α]20 +83.8 (c 1.00, CHCl3); δH (400 MHz, CDCl3) 4.73 (1H, br s, H-1), 3.98 (2H, m, H-4 and
D
H-6), 3.65 (1H, d, J 11.2, H-60), 3.54 (1H, m, H-5), 3.29 (3H, s, Me), 3.12 (1H, br s, H-2), 1.86 (2H,
m, H-3), 0.83 (9H, C(CH3)3), 0.0 (6H, Si(CH3)2), all the peaks were broad at ambient temperatures;
δH (400 MHz, COSY, CDCl3) 4.73–3.12 (H-1 to H-2), 3.12–1.86 (H-2 to H-3), 1.86–3.98 (H-3 to H-4),
3.98–3.54 and/or 3.65 (H-4, H-5, 2H-6 correlation, ambiguous); δC (100 MHz, DEPT, CDCl3) 99.3 (CH,
C-1,), 77.7 (CH, C-2), 74.2 (CH, C-4), 71.1 (CH, C-5), 63.2 (CH2, C-6), 57.0 (Me), 31.2 (CH2, C-3),
26.4 (CH3, [(CH3)3CSi]), 18.7 (C, [(CH3)3CSi]), −4.7 and −4.8 [(CH3)2Si].
6.2.5. 2A,2B,2C,2D,2E,2F,2G-Hepta(O-benzyl)-6A,6B,6C,6D,6E,6F,6G-hepta(O-tert-butyldimethylsilyl)-
3A,3B,3C,3D,3E,3F,3G-hepta(deoxy)-hepta(manno)-β-cyclodextrin, 3c
Sodium hydride (0.4 g, 60% mineral oil, 10 mmol) was washed twice with petroleum ether and left
to dry under nitrogen; a solution of hepta(manno-6-O-t-butyldimethylsilyl-3-deoxy)-β-cyclodextrin 3a
(0.2 g, 0.11 mmol) in dry THF (3 ml) was added slowly to the neat sodium hydride. Benzyl bromide (0.4
ml, 1.6 mmol) was then added dropwise. The reaction mixture was stirred for 48 h at room temperature.
Methanol was added to decompose the excess of hydride, the solvents were evaporated under reduced
pressure (75°C, 0.5 mmHg), and a solution of residue in dichloromethane was washed with water,
dried, and concentrated. Column chromatography (eluent hexane:ethyl acetate, 8:1) gave a colourless
oily compound [82c] (0.16 g, 59%). M.p. 48–49°C; m/z (FAB+, NOBA matrix), 2477 (37%, M+H+Na+,
C
133H211O28Si7Na: requires 2477). Anal. C133H210O28Si7: requires C, 65.10, H, 8.63; found: C, 65.14,
H, 9.00. [α]20 +46.9 (c 4.00, CHCl3); δH (400 MHz, CDCl3) 7.39 (5H, Ph), 4.84 (1H, d, J 2.4, H-1),
D
4.60 (1H, d, J 12.0, H-7), 4.50 (1H, d, J 12.0, H-70), 4.06 (2H, m, H-4, H-6), 3.78 (1H, d, J 11.0, H-6),
3.70 (1H, m, H-5), 3.46 (1H, m, H-2), 1.96 (2H, m, H-3), 0.85 (9H, s, C(CH3)3), 0.00 (6H, s, Si(CH3)2);
δH (400 MHz, COSY, CDCl3) 4.84–3.46 (H-1 to H-2), 3.46–1.96 (H-2 to H-3), 1.96–4.06 (H-3 to H-4),
4.06–3.78 and/or 3.70 (H-4, H-5, 2H-6 correlation, ambiguous); δC (100 MHz, DEPT, CDCl3) 139.0 (C,

3644
D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
C-8), 128.7, 127.8 (all CH, C-9, 10 and 11), 99.2 (CH, C-1,), 76.1 (CH, C-2), 74.3 (CH, C-5), 71.4 (CH2,
C-7), 71.2 (CH, C-4), 63.2 (CH2, C-6), 31.4 (CH2, C-3), 26.4 (CH3, [(CH3)3CSi]), 18.7 (C, [(CH3)3CSi]),
−4.6 and −4.8 [(CH3)2Si]; δH+δC (CDCl3), 4.84–99.2 (C-H-1), 4.60 and 4.50–71.4 (2H-C-7), 4.06 and
3.78–63.2 (2H-C-6), 4.06–74.3 (H-C-5), 3.70–71.2 (H-C-4), 3.46–76.1 (H-C-2), 1.96–31.4 (2H-C-3),
26.4–0.85 (9H-3C-[(CH3)3CSi]); TGA onset 260°C, 325°C (25%), 380°C (80%), 450°C (90%).
6.2.6. (S)-1-Phenyl-1-pentanol 11d33
A solution of valerophenone 32 (3 g, 0.019 mol) in ethanol (2 ml) was added to a yeast solution
prepared from sucrose (100 g), D-glucose (100 g) and fresh yeast (200 g, NG & SF company) dissolved
in water (3 l). The reaction mixture was stirred overnight at 32°C. The crude product was filtered through
Celite, and extracted with petroleum ether (40–60). The ether extracts were washed with water (1×200
ml) and dried over sodium sulfate. Flash silica gel column chromatography using petroleum ether and
ether (95:5 to 90:10) as eluant gave (S)-1-phenyl-1-pentanol 11d as a yellowish liquid (0.24 g, 8% yield).
m/z EI+: 164, (6%, M+, C11H16O: requires 164.1201, found 164.1201, 0 ppm, error) 108 (9), 107 (100,
PhCHOH+ or hydroxytropylium), 106 (29), 105 (7), 80 (5), 79 (57), 78 (29), 77 (35), 76 (13), 51 (6), 50
(6), 39 (9); δH (360, CDCl3) 0.87 (3H, t J 7.1 Hz, H-5), 1.35 (4H, m, H-3 and 4), 1.72 (2H, m, H-2),
4.65 (1H, t J 6.7, H-1), 7.29 (5H, m, Ar); δC (100 MHz, DEPT) 143.9 (C, C-6), 127.3, 126.3, 124.9
(3CH, C-7 to C-9), 73.5 (CH, C-1), 37.7 (CH2, C-2), 26.9 (CH2, C-3), 21.8 (CH2, C-4), 13.0 (CH3, C-5);
[α]26 −13.65 (c 13.3, benzene), 79% ee based on the original literature rotation [α]25 −17.17, (c 13.3,
D
D
benzene)57 which has now been revised44 to [α]25 −20.61. On the basis that the original material had
D
83.3% ee, the revised value is 66% ee. Chiral GC–MS measurement, 45% ee; νmax (neat)/cm−1 3362
(br, OH), 2950 (s), 2859 (m), 1708 (w), 1493 (m), 1040 (m); GC tR 28.02 min for the (S)-enantiomer
(72%) and 28.90 min for the (R)-enantiomer (28%), CDX-B (30 m×0.32 mm, 0.25 µm J & W), 110°C
isothermal (40 min).
6.2.7. (1S,13S)-α-Methoxy-α-trifluoromethyl(phenylacetyl)-1-phenyl-1-pentanol 33
A mixture of the 1-phenyl-1-pentanol 11d (0.16 g, 1.0 mmol, prepared by yeast reduction), 4-
dimethylaminopyridine DMAP (a few crystals), and triethylamine (0.5 ml, 5.0 mmol) in dichloromethane
(10 ml) was stirred for 60 min at 0°C. (R)-α-Methoxy-α-trifluoromethyl(phenylacetyl) chloride58 (0.53
g, 2.1 mmol) was added slowly and the mixture was stirred for 24 h. The mixture was diluted with
dichloromethane (30 ml), washed with dilute hydrochloric acid (5 ml), a saturated solution of sodium
hydrogen carbonate (2×5 ml) and water (2×10 ml). It was then dried over sodium sulfate. The crude
residue was purified by flash silica column chromatography using petroleum ether and ether (95:5
to 90:10) as eluant and recolumned using benzene as eluent to give a yellowish liquid (0.22 g, 58%
yield). Crystallisation was attempted using hexane, benzene, ether, petroleum ether and mixtures thereof,
+
but no crystals could be obtained. m/z CI+: (28%, M+NH4 , found 398.1943. C21H27NF3O3: requires
398.1943), 381 [M+1] (3), 252 (36), 198 (29), 164 (100), 147 (84), 108 (22), 91 (70), 46 (12); νmax
(neat)/cm−1 2950 (m), 2859 (m), 1747 (s, C_O), 1453 (w), 1260 (s), 1169 (s), 1018 (m); δH (360, CDCl3)
0.82 (3H, t J 7.0 Hz, H-5), 1.25 (4H, m, H-3 and 4), 1.78 and 1.95 (2H, m, H-2), 3.43 (3H, s, OMe),
1
5.92 (1H, dd J 8.1, 5.8, H-1), 7.32 (10H, m, Ar). This H NMR data is for the major diastereoisomer.
Signals for the minor component were observed at δH 5.86 (1H, dd J 7.8, 6.3), 3.52 (1H, s, OMe), but
all other signals overlapped with those of the major diastereoisomer. The ratio was circa 82:16 (64% de)
from NMR integration.

D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
3645
Table 5
Unretained peak retention times (t0) in minutes at different temperatures
6.3. Manufacture and evaluation of capillary columns
Columns prepared in this work were 12 m in length, 0.25 mm internal diameter and 0.20 µm stationary
film thickness (β=313). The commercial column was permethyl β-cyclodextrin (J & W, 12 m length, 0.32
mm internal diameter, 0.25 µm film thickness) which has the same phase ratio (β=320).
Fused silica tubing was washed with nitric acid (20%) for 3 h (20 ml) to remove reactive cations, water
(1 h, 7 ml), acetone (1 h, 10 ml) and dichloromethane (1 h, 10 ml) and dehydrated at 200°C for 1 h. The
column was deactivated with diphenyltetramethyldisilazane using the static coating method (sealed with
Super glue), installed in the gas chromatograph and heated for 1 h at 300°C under helium. Finally the
column was washed with dichloromethane before coating with the cyclodextrin stationary phases.
The derivatised cyclodextrin (15% w/w) was mixed with the liquid matrix (OV-1701, 85% w/w) diluted
with dichloromethane and statically coated to form a 0.20 µm film thickness using a water bath at 40°C
and a vacuum pump. Columns were ‘conditioned’ in the GC–MS for at least 6 h at 60–80°C before tests
for enantioselection.
Racemic analytes were commercial products or were prepared in these laboratories. Homogeneity
1
was determined by H and/or 13C NMR. The spectra of α-ionone 6 were checked repeatedly, because
it can isomerise to the thermodynamically more stable and achiral isomer, β-ionone. The 7-endo-alkyl-
bicyclo[3.2.0]hept-2-ene-6-ones 12b,c,d contained 10–30% of the 7-exo-isomer which was ignored in
the column evaluations.59
All amines (except DL-valine methyl ester, 32) were trifluoroacetylated prior to analysis. The amino
compound (1 mg) was reacted with trifluoroacetic anhydride (50 µl) in dichloromethane (200 µl) for 60
min at 100°C. Excess reagent was removed by evaporation and the residue dissolved in dichloromethane.
6.4. Retention times and thermodynamic study
Retention times of homologous series were determined simultaneously. GC detection normally requi-
res a solvent delay (3–5 min) in which the filament current is reduced and no peaks can be detected. This
maximises the dynamic range of the data system and prevents carbonisation/destruction of the filament.
In a few cases, analytes had retention times that were less than the normal solvent delay, even at ambient
oven temperatures and the solvent delay was reduced to 1 min or less. Analytes (20 mg) were dissolved
in dichloromethane (1 ml) and 1 µl was injected into the GC–MS.
The unretained peak retention times were determined by injection of methane and are shown in Table 5.
All retention times used in the calculation of thermodynamic data are at least duplicates and are
averaged, although in most cases the retention times differences were less than or equal to the scan
time of the mass spectrometer (0.9 s+0.1 s interscan time). The uncertainties in the retention times were
used to calculate the error ranges for α, ∆∆H and ∆∆S in Tables 2 and 3 using the methods described
below.
The mean and harmonic means of the temperatures used in the thermodynamic study were: 60–90°C,
Tm 348.16 K, Thm 347.80 K; 60–80°C, Tm 343.16 K, Thm 342.97 K; 40–80°C, Tm 333.16 K, Thm 332.56

3646
D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
K; 40–70°C, Tm 328.16 K, Thm 327.78 K. For the four runs, Tm 337.54, Thm 336.94. The difference
between the harmonic mean and the mean in each case is <1 K which is comparable to the accuracy of
the temperature control of the oven; circa ꢀ0.5 K.
The enthalpy difference (∆∆H), entropy difference (∆∆S), free energy difference (∆∆G) and free
energy difference at the harmonic mean temperature (∆∆GThm) for each pair of enantiomers were
calculated in three ways.
(i) Application of the van’t Hoff and Gibbs–Helmholtz equations for the free energy of interaction
(∆∆G=−RTlnα and lnα=−∆∆H/RT+∆∆S/R). A plot of lnα vs 1/T enables ∆∆H to be calculated
from the gradient and ∆∆S from the intercept by multiplying by the gas constant (R).60 All gradients
and intercepts were determined by linear regression (R≥0.99, Cricket Graph).
(ii) Graphically by using the modification of Klug et al. of the Gibbs–Helmoltz equation.50 A plot of
lnα vs (1/T−1/Thm) enables ∆∆H to be calculated from the gradient and ∆∆S from ((gradient/Thm)-
intercept)×R. This method removes bias in the results, due to the temperature at which the
measurements were made.
(iii) ∆∆H and ∆∆GThm were calculated (non-graphically) by adapting the second method of Klug et
al.50 For measurements of lnα at 1,n temperatures T (subscripts indicate range of summations):
X
∆∆GThm = −RThm
(
lnα1,n)/n at Thm
and
X
X
∆∆H = −R( (lnα1,n × (1/T1,n − 1/Thm))/ ((1/T1,n − 1/Thm)2)).
These values were then used to calculate ∆∆S and are reported in Tables 2 and 3.
The ratio of ∆∆H to ∆∆S was used to calculate the temperature at which no separation occurs for each
analyte on each column. This is derived as follows: ∆∆G=∆∆H−T∆∆S. If ∆∆G=0 then ∆∆H=T∆∆S
and hence ∆∆H/∆∆S=T.
For each set of analytes on each column, the compensation temperature (Tc; also designated Tiso or β
by some authors) was calculated from the slope of a graph of ∆∆H vs ∆∆GThm (Fig. 3) as follows:
If a linear enthalpy–entropy compensation exists, ∆∆H/∆∆S=Tc where Tc is a constant for all the
analytes.
Substituting ∆∆S=∆∆H/Tc in the Gibbs–Helmholtz equation (∆∆GThm=∆∆H−Thm∆∆S) gives
∆∆GThm = ∆∆H − Thm∆∆H/Tc
which rearrranges to
∆∆GThm = ∆∆H((Tc − Thm)/Tc)
and hence
∆∆H = ∆∆GThm(Tc/(Tc − Thm)).
Consequently a graph of ∆∆H vs ∆∆GThm for a series of analytes showing linear enthalpy–entropy
compensation has a ∆∆H intercept of zero and a gradient m=(Tc/(Tc−Thm))
Rearrangement gives Tc=Thm/(1−(1/m)).
It should be noted that compensation temperatures for individual analytes are designated as
∆∆H/∆∆S, whereas common compensation temperatures for series of analytes are designated Tc. This
avoids the use of subscripts and avoids the implication that a compensation temperature for a given
analyte necessarily forms part of a series.

D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
3647
Acknowledgements
We gratefully acknowledge the funding of this project by Hikma Pharmaceuticals and an allocation of
instrument time by the EPSRC National Mass Spectrometry Service at the University of Swansea. The
purchase of the GC–MS was funded by an EC grant (ECLAIR 209). We also thank Valerian D’Souza, J.
Fraser Stoddart and Graeme Watson for helpful discussions.
References
1. Recent developments in cyclodextrin chemistry are described in a series of reviews in: Chem. Rev. 1998, 98, 1741–2076;
Comprehensive Supramolecular Chemistry; Atwood, J. L.; Lehn, J.-M., Eds.; Vol. 3, Cyclodextrins, Szejtli, J.; Osa, T.,
Eds.; Elsevier Science: Oxford, 1996; Wenz, G. Angew. Chem., Int. Ed. Engl. 1994, 33, 803–822.
2. Croft, A. P.; Bartsch, R. A. Tetrahedron 1983, 39, 1417–1474.
3. Gattuso, G.; Nepogodiev, S. A.; Stoddart, J. F. Chem. Rev. 1998, 98, 1919–1958.
4. Kuyama, H.; Nukada, T.; Nakahara, Y.; Ogawa, T. Tetrahedron Lett. 1993, 34, 2171–2174.
5. Houdier, S.; Vottero, P. J. A. Angew. Chem., Int. Ed. Engl. 1994, 33, 354–356.
6. Sakairi, N.; Kuzuhara, H. J. Chem. Soc., Chem. Commun. 1992, 510–512; Sakairi, N.; Nishi, N.; Tokura, S.; Kuzuhara, H.
Carb. Res. 1996, 291, 53–62.
7. Khan, A. R.; Forgo, P.; Stine, K. J.; D’Souza, V. T. Chem. Rev. 1998, 98, 1977–1996.
8. Gadelle, A.; Defaye, J. Angew. Chem., Int. Ed. Engl. 1991, 30, 78–79; Ashton, P. R.; Ellwood, P.; Staton, I.; Stoddart, J.
F. J. Org. Chem. 1991, 56, 7274–7280; Yamamura, H.; Yotsuya, T.; Usami, S.; Iwasa, A.; Ono, S.; Tanabe, T.; Iida, D.;
Katsuhara, T.; Kano, K.; Uchida, T.; Araki, S.; Kawai, M. J. Chem. Soc., Perkin Trans. 1 1998, 1299–1303.
9. Reviews: Li, S.; Purdy, W. C. Chem. Rev. 1992, 92, 1457–1470; Juvancz, Z.; Petersson, P. J. Microcol. Sep. 1996, 8,
99–114.
10. Lichtenthaler, F. W.; Immel, S. Tetrahedron: Asymmetry 1994, 5, 2045–2060.
11. Khan, A. R.; Barton, L.; D’Souza, V. T. J. Chem. Soc., Chem. Commun. 1992, 1112–1114.
12. Khan, A. R.; Barton, L.; D’Souza, V. T. J. Org. Chem. 1996, 61, 8301–8303.
13. Zhang, P.; Coleman, A. W. Supramol. Chem. 1993, 2, 255–263.
14. Fujita, F.; Shimada, H.; Ohta, K.; Nogami, Y.; Nasu, K.; Koga, T. Angew. Chem., Int. Ed. Engl. 1995, 34, 1621–1622.
15. Khan, A. R.; Tong, W.; D’Souza, V. T. Supramol. Chem. 1995, 4, 243–246; Groaning, M. D.; D’Souza, V. T. Supramol.
Chem. 1996, 8, 5–8.
16. Pratt, J. W.; Richtmyer, N. K. J. Am. Chem. Soc. 1957, 79, 2597–2600; Prins, D. A. J. Am. Chem. Soc. 1948, 70, 3955–3957.
17. Fujita, K.; Ohta, K.; Ikegami, Y.; Shimada, H.; Tahahara, T.; Nogami, Y.; Koga, T.; Saito, K.; Nakajima, T. Tetrahedron
Lett. 1994, 35, 9577–9580.
18. van Dienst, E.; Snellink, B. H. N.; von Piekartz, I.; Gansey, M. H. B. G.; Venema, F.; Feiters, M. C.; Nolte, R. J. M.;
Engbersen, J. F. J.; Reinhoudt, D. N. J. Org. Chem. 1995, 60, 6537–6545; Yuan, D.-Q.; Ohta, K.; Fujita, K. J. Chem. Soc.,
Chem. Commun. 1996, 821–822.
19. Jiang, T.; Sukumaran, D. K.; Soni, S.-D.; Lawrence, D. S. J. Org. Chem. 1994, 59, 5149–5155.
20. Rao, V. S. R.; Qasba, P. K.; Balaji, P. V.; Chandrasekaran, R. Conformation of Carbohydrates; Harwood Academic
Publishers: Amsterdam, 1998; p. 59.
21. Nogami, Y.; Nasu, K.; Koga, T.; Ohta, K.; Fujita, K.; Immel, S.; Lindner, H. J.; Schmitt, G. E.; Lichtenthaler, F. Angew.
Chem., Int. Ed. Engl. 1997, 36, 1899–1902; Nogami, Y.; Fujita, K.; Ohta, K.; Nasu, K.; Shimada, H.; Shinohara, C.; Koga,
T. J. Inclusion Phenom. Mol. Recognit. Chem. 1996, 25, 57–60.
22. See Ref. 2, 1467–1468 for a compilation of early examples of 6-deoxy-cyclodextrins.
23. Baer, H. H.; Berenguel, A. V.; Shu, Y. Y.; Defaye, J.; Gadelle, A.; Gonzalez, F. S. Carb. Res. 1992, 228, 307–314;
Nishizawa, M.; Imagawa, H.; Kubo, K.; Kan, Y.; Yamada, H. Synlett 1992, 447–448.
24. trans-4-Phenyl-3-buten-2-one and some derivatives complexed with per(3-deoxy-manno)-β-cyclodextrin are reported in
Beilstein (Beilstein registry numbers: 6916689, 6916744, 6916818, 6916997, 6954699), but upon reading the original
paper this was found to be an error; Pitchumani, K.; Velusamy, P.; Sabithamala, S.; Srinivasan, C. Tetrahedron 1994, 50,
7903–7912.
25. Yoshinaga, M. Deoxycyclodextrins and their Manufactures; Jpn. Kokai Tokkyo Koho JP 06,136,004, 17th May 1994; CA
122:163960q.

3648
D. R. Kelly, A. K. Mish’al / Tetrahedron: Asymmetry 10 (1999) 3627–3648
26. Prengel, M. J.; Bruncel, E. Can. J. Chem. 1991, 69, 130–137; Fugedi, P. Carb. Res. 1989, 192, 366–369.
27. N-Tosyl-imidazole has been recommended for this transformation recently; Isac-Garcia, J.; Lopez-Paz, M.; Santoyo-
Gonzalez, F. Carb. Lett. 1998, 3, 109–116.
28. Coleman, A. W.; Zhang, P.; Parrot-Lopez, H.; Ling, C. C.; Miocque, M.; Mascrier, L. Tetrahedron Lett. 1991, 32,
3997–3998.
29. Xiao, D. Q.; Che, B. Q.; Fu, R. N.; Gu, J. L.; Wen, Y. X.; Ling, Y.; Zhang, H. B. Chromatographia 1997, 44, 393–398;
Uccello-Barretta, G.; Cuzzola, A.; Balzano, F.; Menicagli, R.; Salvadori, P. Eur. J. Org. Chem. 1998, 2009–2012.
30. MacLeod, R.; Prosser, H.; Fikentscher, L.; Lanyl, J.; Mosher, H. S. Biochemistry 1964, 3, 838–846.
31. This has been demonstrated by X-ray crystallography for phenylethanol 11a, Harata, K. J. Chem. Soc., Perkin Trans. 2
1990, 799–804.
32. Mislow, K. J. Am. Chem. Soc. 1954, 73, 3954.
33. MacLeod, R.; Welch, F. J.; Mosher, H. S. J. Am. Chem. Soc. 1960, 82, 876–880
34. To further confuse matters, the value of the optical rotation of phenylbutanol 11c was also a matter of dispute at that time,
Mislow, K.; Hamermesh, C. L. J. Am. Chem. Soc. 1955, 77, 1590–1594.
35. Cram, D. J.; McCarty, J. E. J. Am. Chem. Soc. 1957, 79, 2866–2875.
36. Brewster, J. H. J. Am. Chem. Soc. 1959, 81, 5475–5483.
37. Weidmann, R.; Horeau, A. Bull. Chem. Soc. Fr. 1967, 117–124.
38. Davies, H. G.; Green, R. H.; Kelly, D. R.; Roberts, S. M. Biotransformations in Preparative Organic Chemistry; Academic
Press: London, 1989; pp. 99–127; Servi, S. In Biotransformations; Vol. 8a in the series Biotechnology; Kelly, D. R., Vol.
Ed.; Rehm, H.-J.; Reed, G.; Puhler, A.; Stadler, P. J. W., Series Eds.; Wiley-VCH: Weinheim, 1998; pp. 363–390.
39. Kelly, D. R.; Harrison, S.; Jones, S.; Masood, A.; Morgan, J. J. G. Tetrahedron Lett. 1993, 34, 2689–2690.
40. Latypov, S. K.; Seco, J. M.; Riguera, R. J. Org. Chem. 1996, 61, 8569–8577.
41. Vogel, E.; Caravatti, G.; Franck, P.; Aristoff, P.; Moody, C.; Becker, A.-M.; Felix, D.; Eschenmoser, A. Chem. Lett. 1987,
219–222; Wintner, C. E. J. Chem. Ed. 1987, 64, 587–590.
42. Yamaguchi, S. In Asymmetric Synthesis; Morrison, J. D., Ed.; Academic Press: London, 1983; Vol. 1, pp. 125–152.
43. This particular chemical shift difference has been noted previously, but was not exploited. Noyori, R.; Tomino, I.;
Tanimoto, Y. J. Am. Chem. Soc. 1979, 101, 3129–3131.
44. Mazaleyrat, J.-P.; Cram, D. J. J. Am. Chem. Soc. 1981, 103, 4585–4586.
45. Yamaguchi, S.; Kabuto, K. Bull. Chem. Soc. Jpn. 1977, 50, 3033–3038; Furukawa, J.; Iwasaki, S.; Okuda, S. Tetrahedron
Lett. 1983, 24, 5257–5260; Eidenhammer, G.; Hammerschmidt, F. Synthesis 1996, 748–754.
46. Kelly, D. R. Tetrahedron: Asymmetry 1999, 10, 2927–2934.
47. Kelly, D. R. In The Chemistry of Natural Products; Thompson, R. H., Ed.; Blackie: Glasgow; 1993; 2nd ed., pp. 382–441.
48. Fletcher, M. T.; Jacobs, M. F.; Kitching, W.; Krohn, S.; Drew, R. A. I.; Haniotakis, G. E.; Francke, W. J. Chem. Soc.,
Chem. Commun. 1992, 1457–1459.
49. Botsi, A.; Yannakopoulou, K.; Perly, B.; Hadjoudis, E. J. Org. Chem. 1995, 60, 4017–4023.
50. Krug, R. R.; Hunter, W. G.; Grieger, R. A. J. Phys. Chem. 1976, 80, 2335–2341 and 2341–2351.
51. Leffler, J. E. J. Org. Chem. 1955, 20, 1202–1231; Linert, W.; Jameson, R. F. Chem. Soc. Rev. 1989, 18, 477–505; Searle,
M. S.; Westwell, M. S.; Williams, D. H. J. Chem. Soc., Perkin Trans. 2 1995, 141–151.
52. For related work on the ratio of ∆H to ∆S determined by calorimetry, see: Rekharsky, M. V.; Inoue, Y. Chem. Rev. 1998,
98, 1875–1917.
53. In fact plots of ∆∆H vs ∆∆S for each of the four data sets or for the complete data set gave excellent linear regression
lines. In the latter case, ∆∆H=106.65+403.56×∆∆S (R=0.99).
54. We also tried to apply the logα/T vs 1/T plot for the data for the bicyclo[3.2.0]hept-2-en-6-ones 12a–d on the permethyl-
CD column. This gave excellent linear plots, but no unique intersection point. The range of the intersections was similar
to that of the errors reported in Table 2 and hence no additional useful information was produced. Petersen, R. C. J. Org.
Chem. 1964, 29, 3133–3135.
55. Berthod, A.; Li, W.; Armstrong, D. W. Anal. Chem. 1992, 64, 873–879.
56. Reinhardt, R.; Engewald, W.; Gorog, S. J. High Res. Chromatogr. 1995, 18, 259–262.
57. Levene, P. A.; Marker, R. E. J. Biol. Chem. 1932, 97, 379–391.
58. Dale, J. A.; Mosher, H. S. J. Org. Chem. 1969, 34, 2543–2549.
59. Wright, M. A.; Taylor, I. N.; Lenn, M. J.; Kelly, D. R.; Mahdi, J. G.; Knowles, C. J. FEMS Microbiol. Letts. 1994, 116,
67–72.
60. Schurig, V.; Schmalzing, D.; Jung, M. J. Chromatogr. 1991, 552, 43–57.