3-Fluoropiperidines and N-methyl-3-fluoropiperidinium salts: The persistence of axial fluorine

Article abstract of DOI:10.1002/chem.200400835

It has previously been shown that the fluorine atom in N-protonated 3-fluoropiperidine salts in water strongly prefers the axial orientation in the six-membered ring chairs. In the present work we examine the proposition that the N-methyl salts are equall

Full text of DOI:10.1002/chem.200400835

FULL PAPER
3-Fluoropiperidines and N-Methyl-3-fluoropiperidinium Salts: The
Persistence of Axial Fluorine
Aiming Sun,[a] David C. Lankin,*[b] Kenneth Hardcastle,[a] and James P. Snyder*[a]
Abstract: It has previously been shown
that the fluorine atom in N-protonated
3-fluoropiperidine salts in water strong-
ly prefers the axial orientation in the
six-membered ring chairs. In the pres-
ent work we examine the proposition
that the N-methyl salts are equally dis-
posed to present axial fluorine. Initial-
ly, we explored this point by comparing
the structures of the corresponding
salts. These were evaluated by one-
and two-dimensional NMR spectrosco-
py in [D6]DMSO to fully corroborate
the axial disposition of the fluorine in
each of the compounds. X-ray crystal
structure determinations were likewise
performed for the diphenyl-3-fluoro
sible for the axial-F conformation in
the parent protonated fluoropiperidini-
um compounds carries over to doubly
alkylated salts, we show that it extends
to molecular orientation in the packing
of the unit cells in the solid state as
well. Finally, using the computational
methods that successfully motivated
our synthesis and structural work, we
have made predictions for a number of
new structures and re-examined some
parallel results reported by the Eliel
+
+
NH2 and NMe2 systems to substanti-
ate axial-F. Comparison of the X-ray
structures of the fluorinated and un-
+
+
NH2+, NHMe+, and NMe2 salts by
fluorinated NMe2 salts reveals that
means of density functional theory
(DFT), ab initio, and MMFF force
field calculations with and without
aqueous solvation models. The predic-
tions unambiguously pointed to axial
fluorine for all salts investigated, in-
cluding those with simultaneous axial F
and (N)Me. The calculations were fol-
lowed by synthesis of the correspond-
ing series of 4,4-diphenylpiperidinium
the fluorine resides axial in spite of
substantial steric compression. While
the charge-dipole phenomenon respon-
À
group in the early 1970s. Although C
À
F···H N hydrogen bonds are reported
to be weak and few in number, the
CF···HN charge-dipole orienting effect
Keywords:
charge–dipole interactions
formational analysis
3-fluoropiperidines
·
is
a powerful directing force that
·
con-
matches the hydrogen-bond in both its
energetic contribution and conforma-
tional consequences.
·
density
functional calculations · fluorine ·
molecular modeling
Introduction
The dynamic properties of substituted cyclohexanes repre-
sent one of the cornerstones of conformational analysis.
Mono-substituted analogues universally prefer a substitutent
to occupy the equatorial orientation, although the energy
gap between equatorial and axial conformations can typical-
ly range from 0.5 to 5.0 kcalmolÀ1 depending on the nature
of the substituent and the medium.[1] Occasionally claims
are made to the contrary,[2,3] but they correspond either to
an organomercurial system[3] or to error.[4] Ordinarily, cyclo-
hexanes with multiple substitutents and ring-atom replace-
ment by heteroatoms behave in a similar manner.[5,6] One
unusual exception is the family of N-protonated 3-fluoro-
and 3,5-difluoropiperidines. In water these compounds exist
almost exclusively with the fluorine atom(s) occupying an
axial and diaxial orientation, respectively.[7,8] The source of
the ring inversion found for six-membered rings has been at-
[a] Dr. A. Sun, Dr. K. Hardcastle, Dr. J. P. Snyder
Department of Chemistry, Emory University
Atlanta, GA 30322 (USA)
Fax : (+1)404-727-6586
E-mail: snyder@heisenbug.chem.emory.edu
[b] Dr. D. C. Lankin+
Pharmacia Corporation
4901 Searle Parkway, Skokie, IL 60077 (USA)
E-mail: lankindc@uic.edu
[+] New address: Medicinal Chemistry and Pharmacognosy College of
Pharmacy, University of Illinois at Chicago College of Pharmacy, Chi-
cago, IL, 60612–7231 (USA), Fax: (+1)312-996-7107
Supporting information (1D and 2D NMR spectra for 6; 13C data for
model compounds; supplementary calculations for Scheme 1, and
ESP map for structure 26) for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
Chem. Eur. J. 2005, 11, 1579 – 1591
DOI: 10.1002/chem.200400835
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1579

tributed to a four-center dipole–dipole interaction between
parallel CF and NH bonds as in 1a. Recently, we reported
the operation of this effect in a molecule possessing a freely
rotating CH2F moiety.[9]
these earlier predictions.[7,8] Table 1 displays typical results
for the protonated 3-fluoropiperidine using the 6–
1
311G(d,p) basis set supplemented with DFT (Becke3LYP),
MP2 correlation or a continuum solvation model (H2O). In
all cases, the axial fluorine conformer 1a is estimated to
show a population above 99% as observed. By comparison,
fluorocyclohexane in solution sustains an equatorial/axial
ratio in the range of 65:35 depending on the solvent.[10] Sub-
sequent investigations involving molecular modeling of the
conformational profile of an acyclic variant possessing the
NH/FC charge–dipole interaction was likewise substantiated
in all details by extensive NMR experiments.[9]
Accordingly, we considered a similar strategy to evaluate
the hypothesis that a cationic axial N-methyl group in piper-
idine 2 is sufficiently polar to induce fluorine at C-3 to
prefer an axial orientation. The intuitive basis for this ex-
pectation resides in the computational fact that N-methyl
groups with a formal charge on nitrogen redistribute that
charge primarily to the hydrogen atoms of the methyl group
when evaluated by quantum-chemical methods.[11,12] We sur-
mised that the residual charge on hydrogen in the axial
methyl group of 2 would provide sufficient electrostatic pull-
ing power to induce the partially negatively charged fluorine
to occupy a position of closest proximity; that is, F-axial 2a.
The equatorial form 2b sacrifices the electrostatic attraction
In the present work, we examine the proposition that re-
placement of (NH2)+ with (NMe2)+ provides chair piperi-
dines in which the axial fluorine survives in spite of appa-
rent steric congestion afforded by axial methyl and the short
C-N ring bonds. The study is supported by results of density
functional theory (DFT) predictions, chemical synthesis, and
detailed one- and two-dimensional NMR spectroscopy and
X-ray crystallographic analysis.
Results and Discussion
Preliminary predictions: That the fluorine atom(s) in proto-
nated 3-fluoro- and 3,5-difluoropiperidines prefer to occupy
an axial orientation was first predicted from the results of
molecular modeling. Both molecular mechanics (MM) and
quantum-mechanical methods (QM) unambiguously fore-
cast that the population of the F-axial conformer would
strongly dominate the equilibrium between the interconvert-
ing chair piperidines in both the gas phase and aqueous so-
lution. Follow-up synthesis and NMR evaluation of the equi-
librium positions of a number of analogues fully confirmed
Table 1. Absolute and relative energies for axial and equatorial conformers of piperidines and salts derived from DFT, MP2, and MMFF calculations
with and without a continuum solvation model.[a]
E(abs) [au] and DE(rel) [kcalmolÀ1] (pop%)
MMFF/H2O
basis 1[b]
DErel (%)
basis 2[c]
DErel (%)
basis 3[d]
DErel (%)
DErel (%)[e]
1a
1b
2a
2b
4a
4b
5a
5b
À351.615 82
À351.607 28
À430.256 49
À430.250 05
À813.602 10
À813.594 89
À852.920 02
À852.912 80
À892.227 64
À892.221 73
À852.528 97
À852.531 96
À390.558 52
À390.559 98
À390.940 79
À390.932 32
À390.935 64
À390.928 61
0.0 (99.99)
5.4 (0.01)
0.0 (99.9)
4.0 (0.1)
0.0 (99.95)
4.5 (0.05)
0.0 (99.95)
4.5 (0.05)
0.0 (99.8)
3.7 (0.2)
À351.720 43
À351.715 50
À430.341 10
À430.337 56
À813.692 74
À813.691 97
À852.985 04
À852.980 72
À892.297 33
À892.291 77
À852.534 34
À852.535 46
À390.561 52
À390.562 18
À391.010 65
À391.006 31
À391.007 61
À391.002 37
0.0 (99.5)
3.1 (0.5)
0.0 (97.6)
2.2 (2.4)
0.0 (100)
9.8 (0.0)
0.0 (99.3)
2.9 (0.7)
0.0 (99.7)
3.5 (0.3)
0.7 (23.5)
0.0 (76.5)
0.4 (33.0)
0.0 (67.0)
0.0 (95.1)
2.7 (1.0)
1.9 (3.9)
5.2 (0.02)
À350.735 09
À350.729 97
À429.103 65
À429.100 28
0.0 (99.6)
3.2 (0.4)
0.0 (97.2)
2.1 (2.8)
0.0 (99.6)
3.3 (0.4)
0.0 (99.5)
3.1 (0.5)
0.0 (99.8)
3.8 (0.2)
0.0 (99.7)
3.4 (0.3)
0.0 (99.7)
3.4 (0.3)
0.3 (37.6)
0.0 (62.4)
0.1 (45.8)
0.0 (54.2)
0.0 (94.4)
3.6 (0.2)
1.0 (5.4)
4.9 (0.02)
6a
6b
11a
11b
18a
18b
19a
19b
19c
19d
1.8 (4.6)
0.0 (95.4)
0.91 (17.7)
0.0 (82.3)
0.0 (99.5)
5.3 (0.01)
3.2 (0.5)
À389.446 22
À389.446 96
À389.897 28
À389.892 69
À389.894 17
À389.889 30
0.5 (30.1)
0.0 (69.9)
0.0 (96.0)
2.9 (0.7)
2.0 (3.3)
5.1 (0.02)
7.6 (0.0)
[a] Compounds 4a/4b, 5a/5b, and 6a/6b are conformers of compounds 12, 13, and 14, respectively. [b] Becke3LYP/6–311G(d,p)//Becke3LYP/6–
311G(d,p) except for the diphenyl analogues 4–6 and 11 which employ Becke3LYP/6–31G*//MMFF/GABA/H2O. [c] Becke3LYP/6–311G(d,p)/PCM//
Becke3LYP/6–311G(d,p)/PCM except for diphenyl analogues 4–6 and 11 which employ Becke3LYP/6–31G*/PCM//MMFF/GABA/H2O. [d] MP2/6–
311G(d,p)//Becke3LYP/6–311G(d,p). [e] MMFF/GBSA/H2O.
1580
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1579 – 1591

Persistence of Axial Fluorine
FULL PAPER
by placing the oppositely charg-
ed centers at a much greater
distance.
Such attraction is, of course,
offset by the close approach of
two axial substituents in the six-
membered ring. The cyclohex-
ane A-value for methyl is sub-
stantial
(1.74 kcalmolÀ1),
whereas that for fluorine is con-
siderably less (0.25–0.42 kcal-
molÀ1) but still positive.[10] The
values together signal a significant steric effect between ax-
ially disposed F and Me and imply that axial fluorine in 3,3-
dimethylfluorocyclohexane, 3, will be disfavored relative to
the equilibrium position in fluorocyclohexane. NMR meas-
urements for 3 in CS2 at À858C reveal an 0.64 kcalmolÀ1
free energy difference between axial and equatorial con-
formers corresponding to a population of 85% for equatori-
al 3b.[13] While the halogen atoms (Cl, Br, and I) in the
with the density functional Becke3LYP/6–311G(d,p) proto-
col both in the gas phase and in “water” using the polarized
continuum model of Tomasi and co-workers[18] and reevalu-
ated for energy using MP2 correlation. Similar to the results
for 1, the F-axial conformation is predicted to dominate at
> 99%.
Structures 4–6 are too large for full DFT or ab initio opti-
mization. Therefore, they were optimized with the MMFF
force field[19] coupled to a continuum water model and em-
ployed in subsequent single point DFT calculations (Beck-
e3LYP/6–31G*). Again both protio and methyl analogues
are posited to present axial-F (4a–6a) at the >99% popula-
tion level (Table 1). Interestingly, the MMFF/GBSA/H2O
force field provides a very similar account of energies and
populations for 1 and 2. These molecular mechanics and
quantum-chemical calculations motivated the synthesis and
spectroscopic measurements described below.
other 3,3-dimethylhalocyclohexanes are sufficiently bulky to
reveal no trace of the axial isomer by NMR spectroscopy,
the F···Me steric effect is surprisingly diminutive. Relative to
fluorocyclohexane, the presence of an axial methyl group in-
creases the conformational free energy by only 0.34 kcal-
molÀ1 and decreases the axial population by just 15–25%.
This can be understood by a comparison of the van der
Waals radii. The fluorine vdw radius (1.47 ꢁ)[14] is only 23%
larger than that of hydrogen
Experimental justification for using the MMFF structures
in this context is documented in Table 2 and constitutes
comparison with the corresponding X-ray structures to be
+
discussed below. For the NH2 system 4a, the (N)H···F dis-
tances determined by MMFF and X-ray structural analysis
À
À
are identical (2.60 ꢁ). Similarly, the C N and C F bond
(1.20 ꢁ), whereas that for chlor-
[b]
Table 2. Selected distances and angles for pipiridines and salts[a] derived from Becke3LYP/6–311G**
MMFF/GABA/H2O optimization[c] and X-ray crystallography.[d]
and
ine (1.75 ꢁ) is expanded by
46%. Nevertheless, the confor-
mational equilibrium in 3 over-
whelmingly favors the equatori-
Distances [ꢁ]
(N)H···F (XC)H···F
q [8]
(X)CCF
f [8]
f [8][e]
RX···CF
q [8]
MeXC
f [8]
À
À
XC CF
MeX CC
À
1a
1b
2a
2b
3a
3b
2.311
3.993
–
–
–
–
–
104.8
105.4
108.5
104.6
109.5
108.3
106.9
106.4
108.4
109.1
107.5
107.9
108.3
106.9
108.9
108.2
60.2
175.3
70.8
174.5
71.3
175.8
60.6
59.2
-176.8
67.9
67.5
177.9
65.0
65.6
1.7
127.4
0.2
60.2
175.3
70.8
174.5
71.3
175.8
–
1.7
127.4
0.2
125.8
À1.3
124.1
–
al conformer. Since the C N
bonds in the piperidinium salt 2
2.220
4.073
2.359
4.177
–
are approximately 0.06 ꢁ short-
er than their C C counterparts
125.8
À1.3
À
in 3,[15,16,17] the 1,3 diaxial steric
component of the total molecu-
lar energy for the former is ex-
pected to drive the equilibrium
even further in the direction of
the equatorial form.
124.1
À7.2
4a/12[d]
4a’
2.602
2.468
–
À3.8
–
–
–
–
4b’
–
–
–
–
–
–
–
–
4.046
2.223
2.260
4.078
–
–
–
–
120.8
À4.4
6a/14[d]
6a’
111.2
111.6
111.2
110.8
110.3
111.7
112.0
À74.6
À75.0
À74.6
177.5
178.3
176.4
172.9
À3.9
6b’
117.5
À124.5
À122.1
À111.4
À15.4
11[d]
11’
With these considerations as
background, we performed cal-
culations for 2a and 2b as well
as for the 4,4-diphenyl ana-
logues 4 and 5. Specifically, the
18a
18b
69.1
178.0
[a] Compounds 4a/4b and 6a/6b are conformers of compounds 12 and 14, respectively. [b] Basis set 2, Table 1.
[c] Structures marked with a prime (e.g. 4a’) are force-field optimized. [d] Structures determined by X-ray
À
À
À
À
conformers of 2 were optimized crystallography. [e] Improper torsions: Me X···C F or H X···C F, where X=N, C.
Chem. Eur. J. 2005, 11, 1579 – 1591
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1581

D. C. Lankin, J. P. Snyder et al.
Scheme 1. Synthesis of 10–14. a) TMSCl, Et3N, DMF; b) Selectfluor, CH3CN; c) TfOH, C6H6; d) CH3I, NaHCO3; e) HCl (g), CHCl3; f) CH3I, KOH, ace-
tone; g) TfOH, C6H6; h) CH3I, CH2Cl2.
lengths are within 0.01 ꢁ, whereas bond angles around the
NMR (1H/13C/gCOSY/gNOESY/gHSQC/gHMBC) investi-
gations in [D6]DMSO. The results presented in Tables 3–6
À
C F bond are within one degree. The cationic NHMe ana-
1
logue 5a likewise shows excellent geometric agreement be-
summarize the H and 13C chemical shift assignments, the
1
tween modeled and crystallographic structures (Table 2).
1H,1H, and H,19F coupling constants, and the 19F chemical
shifts, respectively. The assignment strategy for this series of
compounds involved: 1) complete assignment of the protons
for each of the piperidines 12–14 using gCOSY, which pro-
vides information about proton coupling networks, and
gNOESY (Figure 1), which affords information about spatial
relationships (stereochemistry) via the nuclear Overhauser
effect; 2) use of the one-bond C,H correlation experiment
(gHSQC) for assignment of the protonated carbon atoms in
each of the piperidines, and 3) use of gHMBC for obtaining
H,C correlations two and three bonds away. Careful evalua-
tion of all of the measured coupling constants (Table 5) and
stereochemical information derived from the gNOESY ex-
Synthesis: Although we have analyzed the problem of 3-flu-
oropiperidine conformation in terms of a minimally substi-
tuted ring, we have elected to utilize the 4,4-diphenyl series
in the present investigation. Not only are these compounds
easily synthesized, but interpretation of 2D NMR spectra
and production of crystals for X-ray analysis are greatly sim-
plified. Preparation of the target 3-fluoro-4,4-diphenylpiperi-
dines 10–14 was initiated from the commercially available
Boc-protected piperidone 7 (Scheme 1). Using literature
procedures of the Terlings Park group,[20] fluorinated inter-
mediate 9 was prepared in two steps in 85 and 70% yields,
respectively. Fluorinated diphenylpiperidine 10 was obtained
as a colorless oil (85%) by TfOH-catalyzed double electro-
philic addition to benzene using the procedure developed by
Klumpp et al.[21] Protonation and methylation provided salts
12–14 and the neutral piperidine 11. Compounds 11, 12, and
14 were crystallized to provide samples suitable for X-ray
crystallographic analysis (see below). The N-methyl proto-
nated salt 13 was not isolated but generated by protonation
in situ for analysis by NMR spectroscopy.
Table 3. 1H chemical shift assignments for diphenylpiperidinium salts 12,
13, and 14 and free base 11; [D6]DMSO, 258C, ppm.[a]
+
+
Proton NH2 (12) NHMe+ (13) NMe2 (14) NMe (11)
[b]
H-2ax
H-2eq
H-3
H-5ax
H-5eq
H-6ax
H-6eq
H-2’,6’
H-3’,5’
H-4’
3.145 dd
3.607 t
6.287 br.d 6.436 br.d
2.513 t
3.056 d
2.699 t
3.263 d
7.405 dd
7.303 t
3.357 dd
3.909 t
3.598 dd
4.074 t
6.426 br.d
2.588 t
3.196 d
3.100 t
3.676 d
7.427 dd
7.313 t
[b]
5.861 d
[b]
2.526 t
3.217 d
2.865 t
3.494 d
7.416 dd
7.306 t
7.164 tt
7.502 dd
7.379 t
7.243 tt
2.800 s
[b]
[b]
[b]
To compare fluorinated and unfluorinated salts, piperidine
16 was generated directly from commercially available 4-pi-
peridone 15 by the same procedure used for the preparation
of 10. Methylation provided the iodide salt 17, which was
subjected to X-ray crystallographic analysis.
7.365 dd
7.249 t
7.175 tt
7.399 dd
7.269 t
7.164 tt
7.504 dd
7.371 t
7.108 tt
H-2’’,6’’
H-3’’,5’’
H-4’’
7.423 dd
7.367 t
7.234 tt
7.248 tt
7.126 tt
Determination of the structure and conformation in
[D6]DMSO solution by NMR spectroscopy: The solution
structures of piperidinium salts 12 and 14 as well as the pro-
tonated form (CF3COOD) of the free base 11, namely 13,
were confirmed by extensive one- and two-dimensional
N-Me
–-
3.255 s (eq) 2.092 s
3.123 s (ax)
[a] All assignments were confirmed by 2D NMR (gCOSY, gNOESY,
gHSQC, gHMBC) experiments. [b] Shifts from 2.02–2.90 ppm broadened
presumably by proton exchange.
1582
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1579 – 1591

Persistence of Axial Fluorine
FULL PAPER
Table 4. 13C chemical shift assignments and nJC,F for diphenylpiperidinium
salts 12, 13, and 14 and free base 11; [D6]DMSO, 258C, ppm.[a]
tion of scalar J-coupling and spatial relationships attributed
to the phenyl ring protons was crucial for derivation of an
+
+
NHMe+ (13) NMe2 (14)
NMe (11)
1
Carbon NH2 (12)
unequivocal and self-consistent set of assignments for all H
C-2
C-3
C-4
43.0
52.8
60.3
55.1
and 13C signals in the diphenylpiperidine series.
2JC,F =21.0
88.0
2JC,F =20.7
89.1
2JC,F =19.3
89.4
2JC,F =20.2
90.9
An interesting feature of the 13C chemical shifts for 11
and 13 relates to calculated charge distribution. As noted
above, while the positive charge on ammonium salts is gen-
erally assigned intuitively to the more electronegative nitro-
gen atom, quantum-chemical charge calculations routinely
show it to have preferentially migrated to the hydrogen
atom of (N)H or to the protons of an N-methyl group
((NC)H3) leaving both N and C partially negatively charged
(see Figure 3 and S6 in the Supporting Information).[11]
13C NMR chemical shifts support this idea. For example, the
(N)Me 13C chemical shifts for neutral 11 and cationic 13 are
found at d=45.3 and and 42.5 ppm, respectively; the methyl
carbon atom of the cation is more shielded (Dd=2.8 ppm)
than the corresponding carbon atom in the neutral species .
A similar relationship exists for carbon atoms C-2 and C-6
in the neutral and cationic systems yielding an average 13C
shielding of 1.9 ppm for the three carbon atoms bound to
the N atom in the protonated salt (cf. Table 4 and Figure S6
in the Supporting Information). The NBO[11,12] atomic charg-
es at carbon bonded to nitrogen for Becke3LYP/6–
311G(d,p) optimized structures of 18a and the correspond-
ing protonated cation 19a are nearly equivalent with values
of À0.35, À0.22, À0.18 and À0.34, À0.20, À0.16, respectively
(see Figure S6 in the Supporting Information). While the
differences in the two sets of charges do not translate con-
1JC,F =174.5 1JC,F =173.6
1JC,F =175.8
46.5
1JC,F =173.8
47.4
46.8
46.5
2JC,F =19.8
25.3
2JC,F =20.2
26.3
2JC,F =19.9
23.9
2JC,F =17.9
29.8
C-5
C-6
39.7
50.4
58.6
51.0
C-1’
C2’,6’
C3’,5’
C-4’
C-1’’
144.3
125.8
128.4
126.2
144.5
126.4
128.6
126.4
143.8
126.5
128.6
126.5
144.2 br
126.8
128.1
125.7
144.4 br
141.5
141.3
141.2
3JC,F =6.7
126.5
3JC,F =7.0
126.7
3JC,F =7.2
126.6
C2’’,6’’
127.1
C-3’’,5’’ 129.1
129.4
129.4
128.5
C-4’’
N-Me
126.8
–
127.0
42.5 (eq)
127.1
56.6 (eq)
51.1 (ax), JC,F =6.5
126.0
45.3 (eq)
[a] All assignments were confirmed by 2D NMR (gCOSY, gNOESY,
gHSQC, gHMBC) experiments.
Table 5. 1 H,1H, and 1H,19F coupling constants for diphenylpiperidinium
salts 12, 13, and 14; [D6]DMSO, 258C, Hz.
+
+
J(H,H) or J(H,F)
NH2 (12)
NHMe+ (13)
NMe2 (14)
2J(H,H)
2J2ax,2eq
À14.4
À16.1
À13.0
À13.8
À15.0
À11.3
À14.5
À16.3
À13.6
2J5ax,5eq
2J6ax,6eq
3J(H,H)
3J5ax,6ax
12.6
12.5
12.4
aromatic rings
3J2’,3’ =3J6’,5’
3J4’,3’ =3J4’,5’
4J2’,4’ =4J6’,4’
3J2’’,3’’ =3J6’’,5’’
3J4’’,3’’ =3J4’’,5’’
4J2’’,4’’ =4J6’’,4’’
8.3
8.3
1.6
8.3
8.3
1.6
8.2
8.2
1.6
8.2
8.2
1.6
8.2
8.2
1.6
8.2
8.2
1.6
veniently into 13C chemical shifts, they do suggest that, if
charges are a significant factor, then the shifts for the am-
monium cation should be similar to those for the neutral
amine as observed.
3
2J(H,F) or J(H,F)
2JH-3,F
42.9
40.0
10.4
10.7
40.5
40.5
11.8
11.8
42.8
39.3
10.7
10.7
3JH-2ax,F
3JH-2eq,F
3JH-2eq,F
To put this analysis in perspective, we note that 13C chemi-
cal shifts are not fully attributable to atomic charge accumu-
lation. Unlike the diamagnetic shielding characterizing
proton chemical shifts (sd), 13C shifts are dominated by para-
magnetic shielding (sp). The latter is composed of three con-
tributions as expressed by the Ramsey–Karplus–Pople equa-
tion (sp ꢀ(1/DE)<1/r3 >(ꢀQij)).[22] The DE and ꢀQij terms
correspond to electronic excitation to low-lying electronic
states and p-bonding, respectively. Since saturated carbon,
amines, and ammonium salts possess neither characteristic,
the contributions are expected to be relatively small. The <
1/r3 > contribution is a radial distribution term reflecting the
average distance of 2p electrons from the nucleus. It paral-
lels the inductive effect at protons, and given the electron-
withdrawing effect of nitrogen, most likely makes the largest
contribution in the present cases to the chemical shifts of
C(N). Consequently, we regard it as plausible that the rela-
tive deshielding of the latter centers for 11 relative to 13
Table 6. 19F chemical shifts for diphenylpiperidinium salts 12, 13, and 14;
[D6]DMSO, 258C, ppm.[a,b]
+
+
NH2 (12)
NHMe+ (13)
NMe2 (14)
À193.8
À192.7
À190.1
[a] 19
F chemical shifts are expressed in ppm relative to external
CF3COOH (TFA) (d=À76.4 ppm) relative to CFCl3 (d=0.00 ppm).
[b] The 19F chemical shift for the free base of the N-Me compound in
[D6]DMSO was observed at d=À189.6 ppm.
periments provides a complete solution structure for each of
the fluoropiperidines. The 1H NMR spectra (400 MHz) of
piperidine salts 12–14 exhibited well-resolved proton signals
associated with both the piperidine ring and the two dis-
tinctly different phenyl rings (axial and equatorial). Resolu-
Chem. Eur. J. 2005, 11, 1579 – 1591
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1583

D. C. Lankin, J. P. Snyder et al.
proton at C-2 for each of the
salts (Table 5), 2) a large vicinal
3JH,H coupling constant (12.4-
12.6 Hz) between axial protons
at C-5 and C-6, and 3) the sub-
stantial coupling (6.7–7.2 Hz,
Table 4) between the fluorine
atom and the ipso-carbon atom
C-1’’ of the axial phenyl ring.
With respect to 1), the maxi-
3
mum observed vicinal JH,F for
an F-C-C-H dihedral angle of
1808 (47–48 Hz) occurs in cyclo-
hexanes.[24,25] The somewhat re-
duced values observed for pi-
peridinium salts 12–14 (40.5–
42.9 Hz), the largest thus far
3
observed for this type of JH,F
coupling, result from attenua-
tion by the electronegative
charged ring nitrogen. Nonethe-
less, they correspond to a fully
anti F-C-C-H dihedral angle as
Figure 1. Scalar and cross relaxation pathways from NMR analysis of piperidines 12, 13, and 14; a) 3J(H,F);
b) J(C,F); c) piperidine ring NOE cross-peaks; d) NOE cross-peaks associated with the 4,4-phenyl rings.
3
owes its origin in large measure to the sizeable negative
charge at nitrogen-bonded carbon in the ammonium cation.
The influence of the fluorine atom on the proton spectra
of piperidines 12–14 provides a valuable marker for assess-
ing structure and conformation. Characterized by 19F spin–
spin coupling (vicinal) to each of the equatorial and axial
protons at C-2 of the piperidine ring and by spin-spin cou-
pling to the proton at C-3 (geminal coupling, Figure 1a), the
spectra are completely compatible with observations report-
ed earlier for 3-fluoropiperidine hydrochloride 1a in D2O
solution.[7]
The fluorine atom also exhibits spin-spin coupling to
carbon atoms in the immediate “local neighborhood” proxi-
mal to the fluorine (Figure 1b). Fluorine coupling to C-3
(1JC,F), C-2, and C-4 (2JC,F) and the ipso-carbon of the axial
phenyl ring C-1’’ (3JC,F) was observed for piperidines 12–14.
The latter assignments were rendered unequivocal on two
counts. First, the proton signals originating from each of the
phenyl rings were clearly observable, and second the axial
and equatorial nature of each phenyl ring could be evaluat-
ed from a combined analysis of the gCOSY, gNOESY,
gHSQC, and gHMBC two-dimensional NMR data. For the
dimethylpiperidinium compound 14, a long-range 13C,19F
coupling from the fluorine atom to the axial methyl group
was also observed (6.5 Hz). This type of coupling, mediated
by means of a through-space F···H interaction, has been ob-
served in fluorinated steroid systems between an axial
(beta) fluorine at C-6 and the angular methyl group at C-
19.[23]
predicted by a carefully parameterized Karplus relationship
3
for JH,F for couplings.[24] Thus, couplings 1)–3) in
[D6]DMSO are consistent with a conformationally biased pi-
peridinium system in which the equatorial conformers are
populated to such a minimal extent that only the axial con-
formers make a significant contribution to the NMR spectra.
They likewise conform in detail to the computational predic-
tions that 12–14 exist as the axial conformers 4a–6a, respec-
tively, in polar solution (Table 1).
The axial and equatorial methyl groups bonded to the ni-
trogen atom in piperidine 14 are clearly differeniated by the
gNOESY measurements. The axial methyl shows NOE
cross peaks to the equatorial protons at C-2 and C-6, the
axial proton at C-5, and the equatorial methyl group attach-
ed to the nitrogen atom. The latter displays NOE cross
peaks to both axial and equatorial protons attached to C-2
and C-6 and to the axial N-methyl group. In protonated 11,
namely 13, the methyl group bonded to the nitrogen atom
exhibits NOE cross peaks only to the axial and equatorial
protons at C-2 and C-6. No cross peaks were observed from
the N-methyl group to the axial proton at C-5 strongly sug-
gesting that protonation of 11 to give 13 occurs in an axial
manner. This point is elaborated below in connection with
calculations of the relative energies for the four isomers
19a–19c.
1
The H NMR spectrum of the neutral piperidine 11 was
also investigated as a preliminary to generating its protonat-
ed form 13. In both CDCl3 and [D6]acetone (400 MHz,
258C) although the aromatic and N-Me protons are sharp,
the remaining ring protons (including the 2-H(CF)) are
somewhat broad indicative of dynamic broadening. Raising
the temperature of the sample to 508C in the NMR spec-
trometer causes sharpening of the ring proton signals. Spec-
ulating that the latter was due to residual HI from N-meth-
The solution conformational profile for all three piperidi-
nium salts 12–14 is consistent with an essentially exclusive
axial fluorine atom located at C-3. This is clearly indicated
3
by: 1) the large (40.5–42.9 Hz) vicinal JH,F coupling constant
between the fluorine atom attached to C-3 and the axial
1584
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1579 – 1591

Persistence of Axial Fluorine
FULL PAPER
ylation and/or HCl from deuterated chloroform, we stirred
the CDCl3 NMR sample with basic alumina, evaporated it
to dryness, and took up the residue in C6D6. The resulting
1H NMR spectrum (258C) proved to be sharp, well-resolved
and nearly identical to the CDCl3 spectrum recorded at
508C. Slow equilibration between neutral 11 and a low con-
centration of its protonated form, 13, appears to be respon-
sible for the line broadening.
local intramolecular dipole–dipole interactions nonetheless
operate. In the present case, the atomic charge calculations
depicted in Figure 3a and Figure S6 in the Supporting Infor-
À
À
mation suggests that the C F and N H bond dipoles are of
similar magnitude. A more graphic illustration compares the
electrostatic potential energy surface of conformational iso-
mers 1a and 1b in Figure 4. The F-axial form clearly brings
two charge-complementary surfaces together.
X-ray crystallography and charge distribution: Single-crystal
X-ray structure determination has been carried out for pi-
peridine 11 and piperidinium salts 12, 14, and 17. The results
are in complete accord with the present and previous[7,8]
NMR measurements, but for the first time structural details
for the 3-fluoropiperidinium salts are available. Selected in-
ternal variables for 12 are depicted in Figure 2a. Most strik-
Figure 4. Chair conformers of piperidinium salt
1 (Becke3LYP/6–
311G(d,p) optimized) complemented by HF/6–31G(d,p) electrostatic po-
tential surfaces; blue corresponds to regions of strong positive charge,
while red reflects strong negative charge; a) and c) equatorial 1b; b) and
d) axial 1a.
Within the Becke3LYP/6–311G(d,p)/PCM solvation
model, the axial isomer 1a is less solvated than equatorial
1b by 3.6 kcalmolÀ1, a number slightly higher than that pre-
viously derived by AMSOL calculations.[7] In water, al-
though the equatorial form is computed to be more stabi-
lized by solvation, the axial form persists with an advantage
of at least 2.5 kcalmolÀ1 based on the lack of observation of
the equatorial form by NMR spectroscopy or X-ray crystal-
lography. Accordingly, the intramolecular NH···FC interac-
tion is suggested to amount to >4–6 kcalmolÀ1, a quantity
generally associated with another more familiar type of non-
covalent interaction; namely, a reasonably strong hydrogen
bond.
Figure 2. Selected internal variables from the X-ray structures of piperidi-
nium salts; a) 12 (4a), the NH···FC local geometry; improper torsion
f(NH···FC)=6.98; b) 14 (6a), (N)CH···F local geometry; improper tor-
sion f((N)CH···FC)=40.68; and c) 17, local geometry around N+Me2.
ing is the geometry around the NH···FC dipoles. The H···F
separation at 2.60 ꢁ is clearly outside values ascribed to the
À
À
usual hydrogen-bond, as are the N H···F and H···F C
angles. The latter and the improper NH···FC torsion
A similar analysis for the fluorodimethylpiperidinium
salts 2a and 2b likewise indicates the equatorial isomer to
be favored by solvation in the PCM aqueous continuum
model; DDGsolv(eqÀax)=À2.0 kcalmolÀ1. Nonetheless only
F-axial 6a/14 is observed both in solution and in the solid
state. For 1a, with the (N)H···F(C) separation at the van der
Waals boundary (2.60 versus 2.67 ꢁ, the sum of F (1.47) and
H (1.20) radii),[14] the axial halogen is a balance between
coulombic and solvation forces. For dimethyl compound 2a,
the situation appears to be a bit more complicated, since
1,3-diaxial repulsion effects would appear to be involved as
well. Figure 2b illustrates that for 6a/14 the H···F separation
of 2.22 ꢁ is 0.45 ꢁ below the sum of the van der Waals
radii, as is the C···F distance of 2.91 ꢁ (vdW sum=3.17 ꢁ;
Dr=0.26 ꢁ). The crowding is manifested by subtle but coor-
dinated atom movements in the molecule. Comparison with
the unfluorinated dimethyl salt 17 provides a measure of the
effect. Figure 2b and c illustrate that the NMe2 group and
À
(f(NH···FC) À6.98) do reflect, however, that the N H and
À
C F bonds are nearly parallel and planar. The geometry is
precisely that required for a strong dipole–dipole association
accompanied by a charge–dipole component as illustrated
by structure 1a and Figure 3a. We note that although a
dipole moment cannot be assigned to a charged molecule,
Figure 3. Mulliken charges obtained at the Becke3LYP/6–311G(d,p)
level. a) 1a; b) 2a.
Chem. Eur. J. 2005, 11, 1579 – 1591
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1585

D. C. Lankin, J. P. Snyder et al.
À
the C F bond in 6a/14 have retreated from one another.
Thus, the Me-N-Me angle is diminished, whereas the C-N-
Me and Me-N-centroid angles are enlarged in 17 relative to
those in 14 reflecting the backward movement of the axial
N-methyl group. Similarly, fluorine backpedaling in 6a/14 is
indicated by the slight reduction in the C4-C-F bond angle
relative to that in 12 (Figure 2a). Accompanying these angu-
À
À
lar deformations, slight stretching of the N Me and C F
bonds is observed in 6a/14. Although each individual struc-
tural variation in the latter is small, the cumulative changes
make it clear that the compacted F and Me atoms are at-
tempting to minimize 1,3-diaxial steric repulsion. It is impor-
tant to add that the relative errors in the bond lengths and
angles, ~0.008 ꢁ and 0.58, respectively, are well below the
deformations noted above.
Figure 5. Chair CH···FC interactions in the X-ray crystal structure of 11;
H···F separations are given in ꢁ.
Quaternary salts 4a/12 and 6a/14 evidence intermolecular
contacts very similar to those observed intramolecularly for
6a/14. The latter, for example, presents a pair of CH bonds
adjacent to the nitrogen center in a chelating type arrange-
ment with fluorine as shown in Figure 6a. The H···F separa-
Remarkably, the combination of solvation favoring the F-
equatorial conformation and the repulsive elements inherent
in the F-axial conformer are insufficient to favor the F-equa-
torial form for 6a/14 in either solution or the solid state.
The short intramolecular distances (Figure 2b) and comple-
mentary electrostatic forces shown in Figure 3b appear to be
the source of conformational preference. While calculated
partial charges are well known to arise as a consequence of
the partitioning scheme,[26] the depicted values nonetheless
reflect the distribution expected for a significant Me···F at-
traction as well as charge polarization in the methyl group
to maximize the interaction.[11] A very similar set of partial
charges obtain for the MMFF optimized structure 4a evalu-
ated as a single point at the Becke3LYP/6–31G* level
(Table 1). Within the PCM aqueous continuum model, the
charges shown in Figure 3 are attenuated by 5–10% for 4a,
but otherwise present the identical picture. A somewhat dif-
ferent distribution would arise by including explicit water
molecules in the calculation, although it will certainly offer
the same qualitative description.
Finally, we have also obtained the X-ray crystal structure
of the neutral fluorinated piperidine 11. Surprisingly, the
compound displays an axial fluorine. Given that both NMR
and the conformational analysis (Table 1) suggest that the
two conformers exist side by side in a dynamic equilibrium,
the molecule must enjoy stabilizing interactions with neigh-
boring molecules in the solid state. The unit cell analysis
presented below provides a clear basis for this viewpoint.
Interaction within the unit cells: Examination of the unit
cells for the piperidine crystal structures reveals a number
of interesting features. For compound 11, the rings present
as dimers as illustrated in Figure 5. Reminiscent of the
methyl–fluorine interaction in 4a/12 (Figure 2b and Fig-
Figure 6. Intermolecular CH···FC interactions in the X-ray crystal struc-
tures of 6a/14 (a) and 4a/12 (b); H···F separations are given in ꢁ.
tions are well below the van der Waals radii, and the 2.3–
2.4 ꢁ separations are only 0.08–0.18 ꢁ longer than the cor-
responding intramolecular distances. It is safe to conclude
that the forces between molecules in the crystal are not
benign, but provide association energies of the same order
of magnitude as that sustained by NMe and F in the same
molecule. Surprisingly, protonated 4a/12 does not involve
NH···FC association in the crystal, but shows NCH···FC en-
counters instead (Figure 6b). Apparently, the diaxial intra-
molecular forces are sufficiently strong to damp cross-mole-
+
cule interactions at the NH2 center in the crystal. Again
the intermolecular distances are identical to those observed
in 6a/14 with presumably the same consequences for crystal
stability.
Further possible examples of the charge–dipole effect: The
predominance of axial fluorine in 6/14 makes it clear that
the CF···MeN+ interaction is significant. This raises the
question as to whether there is an effective competition be-
tween the latter and the CF···HN+ interaction in 5/13. X-ray
and NMR analyses for the compound allow the conclusion
that the piperidine ring sustains a trans-diaxial interaction
between fluorine and the C2 vicinal proton. However, no in-
formation has been forthcoming regarding the simultaneous
axial disposition of both the N-methyl group and the fluo-
rine atom in 5/13, that is, 19d.
À
ure 3b), the fluorine is associated with C H bonds adjacent
to centers bearing electronegative atoms at distances just
below the van der Waals sum (2.52–2.54 ꢁ versus the vdW
sum of 2.67 ꢁ[14]). It is difficult to distinguish whether the in-
teractions in free amine 11 are energetically productive, or
simply allowable. Nevertheless, the atomic associations are
entirely consistent with observations described above and
below.
1586
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1579 – 1591

Persistence of Axial Fluorine
FULL PAPER
The issue has been examined by evaluating the four con-
formers 19a–19d with DFT/H2O supplemented by single
point MP2/H2O calculations (Table 1). The form with F-
posed above the ring plane. It rotates one of its methyl
groups to provide maximal interaction with only one of the
two axial F atoms. Once again electrostatic F···H interaction
appears responsible for the predicted outcome. As is well-
known, calculated atomic charge distribution alternates
down a chain of atoms. In 26, the 6–31G(d,p) natural
(NBO) charges[11] for the chain N-C(ax-tBu)-C-H varies as
follows: À0.32, +0.16, À0.71, +0.24 (+0.26, +0.27). The
corresponding Mulliken charges on the protons of the CH3
group near the axial F atoms are +0.14, +0.15, and +0.19.
As a result, the coulombic interactions between the tBu
group and the axial F atom (À0.42 (NBO), À0.39 (Mul-
liken)) in 26 are of the same order of magnitude as for axial
6a/14 (cf. Figure 3). Figure S5 in the Supporting Information
provides an illustration of the tBu methyl group···ax-F at-
traction similar to that shown in Figure 4 and Figure 7.[27]
Thus, in spite of the severe axial crowding, a significant pop-
axial and methyl group equatorial, 19a, is the global mini-
mum with a predicted population of 95–96% in agreement
with the X-ray structure for 5a. However, the axial–axial
conformer 19d is posited to participate in the equilibrium at
298 K with a population of 3–4%. The A-value for methyl-
cyclohexane (1.75 kcalmolÀ1)[10] corresponds to the same
population for the axial-methyl group conformation. The
À
relatively short N C bond lengths in the ring most certainly
occasion steric repulsion as described for 6a/14 (Figure 2).
Yet the double axial system 19d compensates electrostatical-
ly such that the energy penalty amounts to only that of a
single axial methyl in cyclohexane. In an effort to obtain ex-
perimental verification of the presence of the conformation
represented by 19d, the gNOESY spectrum of 13 was care-
fully examined for a cross peak between the N-methyl
group and the axial proton at C-5. If such a cross-peak is
present, it proved too weak to observe. Finally, the F-equa-
torial isomers 19b and 19c are estimated to contribute
<1% to the dynamic equilibrium.
Figure 7. Chair conformers of dimethylpiperidinium salt 2 (Becke3LYP/
6–311G(d,p) optimized) complemented by HF/6–31G(d,p) electrostatic
potential surfaces; blue corresponds to regions of strong positive charge,
while red reflects strong negative charge; a) and c) equatorial 2b; b) and
d) axial 2a.
Three additional structures, 25–27, appeal to us as inter-
esting synthetic targets that would extend the principles de-
scribed in this work. Both 25 and 26 can, in principle, exist
as triaxial and axial-di-equatorial conformers in solution.
ulation of the diaxial fluorine conformation is predicted.
Structure 27 can sustain three chair piperidine conforma-
tions for the fluorine pairs: ax–ax, eq—ax, and eq–eq. These
are prognosticated to equili-
brate with populations of >99,
<0.01, and <0.01% (298 K),
respectively.
Finally, while this work and
its recent precursors[7,8,9] focus
exclusively on the conforma-
tional properties of fluoropiper-
Using the solvent-based methods described above (e.g.
idines, it is not the first effort to take advantage of electro-
static effects to tip the balance in favor of axial orientation
in six-membered rings. Substitution of cyclohexanol and its
benzoate with electronegative groups at C-4 elicits eq/ax
ratios from 49/51 to 26/74 at low temperatures.[28] Piperidine
salts with RO, Br, and F at C-4 provide a similar spread for
eq/ax conformers.[29,30] Dimethylpiperidinium ions with OH
and OAc at C-3 were reported to give eq/ax ratios of 49/52
and 34/66, respectively, in perdeuteromethanol.[28] These
Becke3LYP/6–31G*/PCM/H2O//MMFF/GABA/H2O),
we
predict that the triaxial forms will predominate with predict-
ed populations of >99 and 54% (298 K), respectively. The
nearly 50:50 population for axial and equatorial isomers of
di-tert-butyl compound 26 is surprising. Like 6a/14, the axial
isomer is crowded with predicted F···H distances ranging
from 2.1–2.3 ꢁ (cf. Figure 2). However, the axial tert-butyl
group for the optimized structure is not symmetrically dis-
Chem. Eur. J. 2005, 11, 1579 – 1591
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1587

D. C. Lankin, J. P. Snyder et al.
ratios correspond to a relative stabilization of the axial con-
former by 0.1–0.8 kcalmolÀ1. Even more striking is the posi-
tion of the acid-catalyzed chemical equilibrium in substitut-
ed dioxolanes 28 reported in a series of publications by the
Eliel group.[31]
Summary and Conclusions
Predictions based on density functional theory that 3-fluoro-
piperidine salts exist exclusively as the F-axial conformers
have been substantiated by the synthesis of a series of 4,4-
diphenyl analogues, detailed analysis of the conformation by
NMR spectroscopy in D2O and X-ray crystallography. Of
particular importance is the finding that the N,N-dimethyl
Of particular relevance to the present work are the charg-
+
ed substituents X=Me2S+, NH3+, NHMe2+, and NMe3 all
of which were shown to exist as 28a. The latter was deter-
+
salt 14, like its less encumbered NH2 and NHMe+ counter-
parts, was predicted and then established to likewise display
axial-fluorine in solution and the solid state in spite of con-
siderable diaxial steric compression. Short range CH···FC
charge–dipole interactions mediated by positive charge ac-
cumulation at hydrogen atoms of the N-methyl groups are
sufficient to counterbalance the latter. An overall energetic
advantage for axial fluorine is estimated to be 4–6 kcal-
molÀ1, very similar to that of a strong hydrogen bond. The
intramolecular phenomenon appears to be more general as
illustrated by its intermolecular operation in the solid state
for 11, 12, and 14 as well. The surprising appearance of axial
fluorine in neutral 11 in the crystal is attributable to this
source.
Three additional points are noteworthy. First, the use of
the MMFF force field to optimize molecular geometries fol-
lowed by single-point large basis set DFT and/or MP2 calcu-
lations provides energy differences that are equivalent to
using a full quantum-chemical treatment throughout (cf.
Table 1). This strategy should prove useful in evaluating
larger structures that push the limits of computational re-
sources. Second, the MMFF/DFT strategy has been applied
to the analysis of a series of previously reported dioxolanes
28 that exhibit unusual conformational profiles. The charge–
dipole effect in both its geometric and energetic manifesta-
tions carries over to such cyclic structures as it does to acy-
clic systems.[9] The phenomenon is certainly general and
likely to express itself broadly across biologically active
small molecules[32,33,34] and protein systems alike.[35] In this
context we have predicted that a series of highly crowded
structures (25–27) should also conform to the F-axial para-
digm.
Finally, while the conformational analysis of any number
of substituted cyclohexanes and piperidines yields a mixture
of rapidly equilibrating conformers,[1,27,28,29] the 3-fluoropi-
peridine salts are satisfyingly “clean”. Yet the energy differ-
ence is certainly no more than 2–3 kcalmolÀ1, a quantity at-
tributable to a small but productive charge attenuated
dipole–dipole interaction. It is conceivable that the over-
whelming axial bias in 3-fluorinated piperidines can be used
productively in organic synthesis. The substitution might be
used to guide the regiochemistry of a synthetic reaction or
to control product stereochemistry in the spirit of a chiral
auxiliary. Subsequent reductive removal of fluorine would
restore the unhalogenated heterocycle.
mined to be more stable than 28b by >2.0 kcalmolÀ1 corre-
sponding to an axial population of >97% at 298 K in vari-
ous polar carboxylic acid solvents. To examine the consisten-
cy of these results with our own, we have modeled the con-
+
+
figurational diastereomers of 28 for X=NH3 and NMe3
with the Becke3LYP/6311G(d,p)/PCM//MMFF/GABA/H2O
protocol. Both ammonium substituents are predicted to
favor the axial form 28a by 4–6 kcalmolÀ1 in agreement
with observation. The optimized structures are fully compat-
ible with those derived for the piperidines (Figure 8). Specif-
Figure 8. Selected internal variables from MMFF/GBSA/H2O optimiza-
+
tion of Eliel and Alcudiaꢂs axial dioxanes; a) 28a (X=NH3
) the
NH···OC local geometry; the improper torsion f(NH···OC)=5.68; b) 28a
+
(X=NMe3
) the (NC)H···OC local geometry; the improper torsion
f((N)CH···OC)=62.68.
ically, for 28a (X=NH3+) the calculated NH···O distance of
2.60 ꢁ is very similar to that derived for the NH···F separa-
tions in 1a, 4b/12, and 5a/13 (2.60 ꢁ, Figure 2) Likewise,
the closest (NCH2)H···O distance in 28a (X=NMe3+) is
2.61 ꢁ, slightly longer than the (NCH2)H···F separation in
14 and 19c (2.22 ꢁ). Equally important are the dipole orien-
tations. As with the piperidines discussed above, the NR
and CO bonds do not satisfy the classical hydrogen bonding
geometries, instead conforming to a favorably oriented
charge–dipole or dipole–dipole interaction. In 12/4a and
28a (X=NH3+) the improper torsions f(NH···XC) are 6.9
and 5.68, respectively, whereas the bond angles around the
NH and XC centers (X=F and O) are similar (cf. Figure 2
and Figure 8). A comparison of the N-methylated systems
14/6a and 28a (X=NMe3+) demonstrates a similar corre-
À
spondence. In this pair of comparisons, the two parallel C
O bonds in the dioxane rings of 28 plays the same role as
À
the single C F bond in the fluoropiperidines.
1588
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1579 – 1591

Persistence of Axial Fluorine
FULL PAPER
(m, 1H), 3.57 (d, J=14 Hz, 1H), 3.78 (t, J=11 Hz, 1H), 6.21–6.28 (m,
1H), 7.25–7.49 ppm (m, 10H); MS (FAB+): m/z: 256.3; X-ray crystal
structure.
Experimental Section
General: Melting points were determined by using a Thomas–Hoover ca-
pillary melting point apparatus and are reported without correction.
Mass spectrometric analysis was provided by the Emory University Mass
Spectrometry Center. Routine proton and carbon NMR spectra mea-
sured during synthesis were obtained on Varian Inova-400 (400 MHz) or
Varian Inova-600 (600 MHz) spectrometers. Solvent for NMR was deu-
teriochloroform with residual chloroform (d=7.26 ppm for proton and
d=77.7 ppm for carbon) taken as internal reference and reported in
parts per million (ppm). TLC and preparative thin-layer chromatography
(PTLC) were performed on precoated, glass-backed plates (silica gel 60
N-Methyl-4,4-diphenyl-3-fluoropiperidine·TFA (13): Piperidine 11 was
taken up in [D6]DMSO and its proton and 13C NMR spectra recorded.
Subsequently, two drops of trifluoroacetic acid (TFA) were added to the
NMR tube and the corresponding spectra of the salt were recorded. See
Table 3 and Table 4 for the spectra. No attempt was made to isolate the
compound beyond this exercise.
N,N-Dimethyl-3-fluoro-4,4-diphenylpiperidinium iodide (14): Compound
10 (79 mg, 0.31 mmol) was treated with potassium hydroxide (119 mg,
2.12 mmol) in acetone. The reaction mixture was refluxed for 20 min,
then iodomethane (1 mL) was added dropwise to the mixture. After re-
fluxing for 2 h, the reaction mixture was filtered and the filtrate was
evaporated to furnish a white solid; recrystallization from MeOH/diethyl
ether delivered 14 as a white solid (90 mg, 71%): m.p. 278—2808C;
1H NMR (600 MHz, D2O): d=3.04–3.17 (m, 2H), 3.12 (s, 3H), 3.33 (s,
3H), 3.34–3.37 (m, 2H), 3.68 (d, J=13 Hz, 1H), 3.83 (dd, J=15, 39 Hz,
1H), 4.08 (t, J=12 Hz, 1H), 6.29 (m, 1H), 6.27–7.50 ppm (m, 10H); MS
(FAB+): m/z: 284.4 [MÀI]+; elemental analysis calcd (%) for C19H23FNI:
C 55.47, H 5.60, N 3.41; found: C 54.89, H 5.62, N 3.36; X-ray crystal
structure.
F
254; 0.25 mm thickness) from EM Science and were visualized by UV
lamp. Chromatography was performed with silica gel (230–400 mesh
ASTM) or neutral alumina (80–200 mesh) from EM Science using the
“flash” method.[36] Elemental analyses were performed by Atlantic Mi-
crolab Inc. Norcross, Georgia. All solvents and other reagents were pur-
chased from Aldrich Chemical Co., Milwaukee. The reagents were used
as received. All reactions were performed under anhydrous nitrogen at-
mosphere in oven-dried glassware.
1-tert-Butoxycarbonyl-1,2,3,6-tetrahydro-4-(trimethylsilyloxy)pyridine
(8):[20] To a stirred solution of 7 (995 mg, 5 mmol) in dry DMF (5 mL)
was added TMSCl (0.76 mL, 6 mmol) and then dry Et3N (1.68 mL,
12 mmol). The mixture was stirred at 808C overnight (about 20 h), dilut-
ed with hexane, and washed with NaHCO3 (3ꢃ5 mL). The organic layer
was washed with brine and dried over Na2SO4. Purification by chroma-
tography using hexane/EtOAc (9:1) as eluent gave 8 (1.15 g, 85%) as a
colorless oil: 1H NMR (400 MHz, CDCl3): d=0.20 (s, 9H), 1.45 (s, 9H),
2.15 (m, 2H), 3.52 (t, J=5.7 Hz, 2H), 3.85 (m, 2H), 4.78 ppm (m, 1H).
1-tert-Butoxycarbonyl-3-fluoro-4-piperidone (9):[20] Selectfluor reagent
(3.9 g, 11 mmol) was added to a stirred solution of 8 (2.71 g, 10 mmol) in
dry MeCN (100 mL) at 08C, warmed to room temperature for about
1.5 h, then poured into EtOAc (20 mL), washed with brine, and dried
over Na2SO4. The product was first purified by chromatography using
hexane/EtOAc (2:1) as eluent, then rechromatographed on alumina
using MeOH/EtOAc (5:95) as eluent to give 9 (493 mg, 70%) as a color-
N,N-Dimethyl-4,4-diphenylpiperidinium iodide (17): 1-Methyl-4-piperi-
done (15) (8.0 mmol, 0.9 g) was dissolved in dry benzene (5 mL), treated
with TfOH (5 mL), and stirred at room temperature for 3 h. The reaction
mixture was then poured into ice- water, neutralized with NaHCO3 (s),
and extracted with CHCl3 (3ꢃ10 mL). The combined organic phase was
washed with brine, dried with MgSO4, and concentrated in vacuo to give
diphenylpiperidine 16 (2.0 g) as a white solid (quantitative yield), which
was used for the next step without further purification. Compound 17
was prepared by refluxing a solution of 16 (251 mg, 1.0 mmol) in CH2Cl2
(5 mL) and CH3I (1.0 mL) for 1 h. After removal of the solvent, the resi-
due was recrystallized from MeOH/diethyl ether to obtain a light yellow
solid 17[37] (350 mg, 89%): m.p. 302–3038C; MS (FAB+): m/z: 266.4
[MÀI]+; X-ray crystal structure.
NMR spectroscopy: For the detailed structural work on piperidines 11,
12, 13, and 14 the NMR data was collected on a Varian INOVA-400
high-resolution NMR spectrometer equipped with a 5-mm Nalorac 4N
Plus probe. The 1H, 13C, and 19F observation frequencies were 400, 100,
and 376 MHz, respectively. Samples were dissolved in [D6]DMSO
(Isotec) and run in 5-mm NMR tubes (Wilmad 535) at 258C. The
carbon-13 multiplicity was deduced from the gradient-enhanced-HSQC
(gHSQC) heterocorrelation data, the APT (attached proton test), and/or
the DEPT-135 experiments. All other two-dimensional experiments
(gCOSY, gNOESY, gHMQC) were the gradient-enhanced versions[38] as
provided in the Varian VNMR 6.1B version of the software.
1
less oil: H NMR (400 MHz, CDCl3): d=1.50 (s, 9H), 2.48–2.60 (m, 2H),
3.20–3.27 (m, 2H), 4.18 (m, 1H), 4.44 (m, 1H), 4.84 ppm (m, 1H).
4,4-Diphenyl-3-fluoro-piperidine (10): TfOH (0.5 mL) was added drop-
wise to a stirred solution of 9 (10 mg, 46 mmol) in dry benzene (0.5 mL).
The solution was stirred at room temperature for 2.5 h, then poured into
CHCl3 (5 mL), washed with saturated NaHCO3, dried over MgSO4, con-
centrated in vacuo, and purified by chromatography with EtOAc/MeOH/
Et3N (3:1:0.01) as eluent. Compound 10 was obtained as a colorless solid
(10.0 mg, 85%): m.p. 65–678C; 1H NMR (600 MHz, CDCl3): d=1.88 (m,
1H), 2.51 (m, 2H), 2.78–2.82 (m, 1H), 3.00–3.04 (m, 1H), 3.14 (dd, J=
14, 35 Hz, 1H), 3.24–3.26 (m, 1H), 5.48 (m, 1H), 7.14–7.35 ppm (m,
10H); 13C NMR (600 MHz, CDCl3): d=32.57, 42.43, 46.89, 46.90, 47.02,
47.04, 90.92, 92.09, 126.21, 126.46, 127.01, 127.30, 128.43, 128.90, 143.86,
143.90, 145.99 ppm; MS (FAB+): m/z: 256.2.
Crystal structure analysis: The best available crystals of 11, 12, 14, and 17
were coated with Paratone N oil, suspended in small fiber loops, and
placed in a cooled nitrogen gas stream at 100 K on a Bruker D8 SMART
1000 CCD sealed tube diffractometer with graphite-monochromated
CuKa (1.54178 ꢁ) radiation. Data were measured by using a series of
combinations of phi and omega scans with 10 s frame exposures and 0.38
frame widths, except for 11 that had 30 s frame times. Data collection, in-
dexing, and initial cell refinements were all carried out using SMART[39]
software. Frame integration and final cell refinements were done by
using SAINT[40] software. All four data sets were corrected for absorption
using the program SADABS.[41]
N-Methyl-4,4-diphenyl-3-fluoro-piperidine (11): To a solution of 4,4-di-
phenyl-3-fluoropiperidine 10 (20 mg, 80 mmol) in dry dichloromethane
(2 mL) was added iodomethane (11 mg, 80 mmol) and saturated NaHCO3
(20 mL). The reaction was monitored by TLC, which showed that starting
material had disappeared after 20 min. The mixture was quenched with
aqueous NaHCO3 and extracted with dichloromethane (2ꢃ5 mL). The
combined organic phases were washed with brine, dried over anhydrous
Na2SO4, and purified by chromatography with EtOAc/MeOH (3:1) as
eluent to give compound 11 as a colorless solid (10 mg, 47%): 1H NMR
(600 MHz, CDCl3): d=2.28 (s, 3H), 2.40 (m, 1H), 2.48–2.50 (m, 1H),
2.66 (m, 1H), 2.73–2.81 (m, 2H), 2.95 (m, 1H), 5.52 (m, 1H), 7.15–
7.37 ppm (m, 10H); X-ray crystal structure.
The structures were solved by using direct methods and difference Fouri-
er techniques (SHELXTL, V5.10).[42] The hydrogen atoms were located
in difference Fourier maps or positioned by using the HFIX command in
SHELXTL, and were included in the final cycles of least-squares with
isotropic Uij values. All non-hydrogen atoms were refined anisotropically
in 11 and 17, but in 12 and 14 only the F, Cl, N and I, F atoms, respective-
ly, were anisotropically refined. Scattering factors and anomalous disper-
sion corrections are taken from the International Tables for X-ray Crys-
tallography.[43] Structure solution, refinement, graphics and generation of
publication materials were performed by using SHELXTL, V5.10 soft-
ware.
4,4-Diphenyl-3-fluoro-piperidine·HCl (12): Piperidine 10 (57 mg,
0.24 mmol) dissolved in CHCl3 (5 mL) was treated with hydrochloride
gas for 2 h to form the hydrochloride salt. Recrystallization from MeOH/
diethyl ether gave 12 as a white solid (58 mg, 83%): m.p. 184–1858C;
1H NMR (600 MHz, CDCl3): d=2.73 (br t, 1H), 3.07–3.12 (m, 2H), 3.50
Chem. Eur. J. 2005, 11, 1579 – 1591
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1589

D. C. Lankin, J. P. Snyder et al.
CCDC-247306–CCDC-247309 contain the supplementary crystallograph-
ic data for compounds 11, 12, 14 and 17, respectively, in this paper. These
data can be obtained free of charge via www.ccdc.cam.ac.uk/conts/retrie-
ving.html (or from the Cambridge Crystallographic Data Centre, 12
Union Road, Cambridge CB2 1EZ, UK; fax: (+44) 1223–336–033; or e-
mail: deposit@ccdc.cam.ac.uk).
[14] A. Bondi, J. Phys. Chem. 1964, 68, 441–451.
À
[15] The C C bond lengths in cyclohexane by electron diffraction are
1.536 ꢁ,[16] while a combined ED and microwave study for piperi-
[17]
À
dine found the C N bond separations to be 1.472 ꢁ.
[16] O. Bastiansen, L. Fernholt, H. M. Seip, H. Kambara, K. Kuchitsu, J.
Mol. Struct. 1973, 18, 163–168, and references therein.
[17] G. Gunderson, D. W. H. Rankin, Acta Chem. Scand. Ser. A 1983, 37,
865–874.
Computational aspects Molecular mechanics calculations were carried
out with MacroModel 6.5.[44] Ab initio and DFT calculations were per-
formed with the Gaussian 98 series of programs.[45] All structures were
geometry-optimized either with the solvent-enhanced MMFF/GBSA/
H2O force field (MacroModel 6.5), the DFT method Becke3LYP/6–
311G(d,p) or the latter coupled to the Tomasi continuum solvation
model for water[18] (PCM), that is, Becke3LYP/6–311G(d,p)/PCM.
Single-point calculations were performed at three levels of theory: Beck-
e3LYP/6 31G*//MMFF/GBSA/H2O, Becke3LYP/6–311G(d,p)//MMFF/
GBSA/H2O, or MP2/6–311G(d,p)/PCM//Becke3LYP/6–311G(d,p)/PCM.
The HF/6–31G* electrostatic surfaces of Figure 4 and Figure 5 were gen-
erated within Spartan 02[46] from the MMFF-optimized structures. Con-
former populations were calculated by using a Boltzmann distribution at
298 K (or other temperature when specified). Unit cell analysis leading
to Figure 6 and Figure 7 were performed with CCDꢂs Mercury pro-
gram.[47]
[18] a) S. Miertus, E. Scrocco, J. Tomasi, Chem. Phys. 1981, 55, 117–129;
b) M. Cossi, V. Barone, R. Cammi, J. Tomasi, Chem. Phys. Lett.
1996, 255, 327–335; c) M. T. Cances, V. Mennucci, J. Tomasi, J.
Chem. Phys. 1997, 107, 3032–3041; d) V. Barone, M. Cossi, J.
Tomasi, J. Chem. Phys. 1997, 107, 3210–3221; e) V. Barone, M.
Cossi, J. Tomasi, J. Comput. Chem. 1998, 19, 404–417; as imple-
mented in Gaussian 98.[46]
[19] a) T. A. Halgren, J. Comput. Chem. 1996, 17, 490–519; b) T. A.
Halgren, J. Comput. Chem. 1996, 17, 616–641; c) T. A. Halgren,
R. B. Nachbar, J. Comput. Chem. 1996, 17, 587–615; d) T. A. Halg-
ren, J. Comput. Chem. 1999, 20, 730–748.
[20] M. B. van Niel, I. Collins, M. S. Beer, H. B. Broughton, S. K. F.
Cheng, S. C. Goodacre, A. Heald, K. L. Locker, A. M. MacLeod, D.
Morrison, C. R. Moyes, D. OꢂConnor, A. Pike, M. Rowley, M. G. N.
Russell, B. Sohal, J. A. Stanton, S. Thomas, H. Verrier, A. P. Watt,
J. L. Castro, J. Med. Chem. 1999, 42, 2087–2104.
[21] D. A. Klumpp, M. Garza, A. Jones, S. Mendoza, J. Org. Chem. 1999,
64, 6702–6705.
[22] J. B. Lambert, H. F. Shurvell, D. A. Lightner, R. G. Cooks, Organic
Structural Spectroscopy, Prentice Hall, New Jersey, 1998, pp. 47–48.
[23] a) H. L. Holland, Tetrahedron Lett. 1978, 19, 881–882; b) J. T.
Nelson, L. J. Kurz, Magn. Reson. Chem. 1993, 31, 203–213.
[24] J.-P. Berthelot, J.-C. Jacquesy, J. Levisalles, Bull. Soc. Chim. Fr. 1971,
1896–1902.
Acknowledgements
A.S. and J.P.S. are grateful to Professor Dennis Liotta (Emory Universi-
ty) for generous encouragement and support. We are also appreciative to
the NIH for providing support for an X-ray diffractometer (NIH S10-
RR13673).
[25] C. Thibaudeau, J. Plavec, J. Chattopadhyaya, J. Org. Chem. 1998, 63,
4967–4984.
[26] S. M. Bachrach in Reviews in Computational Chemistry, Vol.5
(Eds.: K. B. Lipkowitz, D. B. Boyd), Wiley-VCH, New York, 1994,
pp. 171–268.
[27] The charges that correspond to the electrostatic potential surfaces
(ESP) of Figure 5 and S5 are somewhat different from Natural/NBO
and Mulliken charges. Specifically, negative charge on nitrogen is re-
[1] E. L. Eliel, S. H. Wilen, Stereochemistry of Organic Compounds,
Wiley, New York, 1994.
[2] F.-A. Kang, C.-L. Yin, J. Am. Chem. Soc. 1997, 119, 8562–8563.
[3] a) F. A. L. Anet, J. Krane, W. Kitching, D. Doddrell, D. Praeger, Tet-
rahedron Lett. 1974, 15, 3255–3258; b) P. F. Barron, D. Doddrell, W.
J. Kitching, J. Organomet. Chem. 1977, 6, 361–383; c) K. Kwetkat,
W. J. Kitching, J. Chem. Soc. Chem. Commun. 1994, 345–347.
[4] a) B. Cornett, M. Davis, S. Wu, N. Nevins, J. P. Snyder, J. Am. Chem.
Soc. 1998, 120, 12145–12146; b) M. C. Davis, B. Cornett, S. Wu, J. P.
Snyder, J. Am. Chem. Soc. 1999, 121, 11864–11870.
[5] E. L. Eliel, S. H. Wilen, Stereochemistry of Organic Compounds,
Wiley, New York, 1994, pp. 686–726, 740–754.
[6] Conformational Behavior of Six-Membered Rings, (Ed.: E. Juaristi),
VCH Publishers, 1995.
À
distributed to hydrogen atoms on C H bonds, leading to a positive
charge at nitrogen for 2a and 26. Thus, the figures illustrate a posi-
tive charge distribution (“blue”) in the vicinity nitrogen. Nonethe-
less, the ESP charges at F and the spatially closest methyl groups
are very similar to the NBO and Mulliken charges: 2a (C)H–-F, +
0.22 (H) and À0.32 (F); 26 (C)H–-F, +0.09, +0.16 (H) and À0.31,
À0.32 (F). Consequently, in the two figures, the green-red interface
corresponds to the (N)H–-F(C) interactions.
[28] a) R. D. Stolow, T. W. Giants, J. D. Roberts, Tetrahedron Lett. 1968,
9, 5777–5780; b) R. D. Stolow, T. Groom, P. D. McMaster, Tetrahe-
dron Lett. 1968, 9, 5781–5784.
[29] Y. Terui, K. Tori, J. Chem. Soc. Perkin Trans. 2 1973, 127–133.
[30] R. J. Abraham, C. J. Medforth, P. E. Smith, J. Comput.-Aided Molec.
Des. 1991, 5, 205–212.
[7] D. C. Lankin, N. S. Chandrakumar, S. N. Rao, D. P. Spangler, J. P.
Snyder, J. Am. Chem. Soc. 1993, 115, 3356–3357.
[8] J. P. Snyder, N. S. Chandrakumar, H. Sato, D. C. Lankin, J. Am.
Chem. Soc. 2000, 122, 544–545.
[9] D. C. Lankin, G. L. Grunewald, F. A. Romero, I. Y. Oren, J. P.
Snyder, Org. Lett. 2002, 4, 3557–3560.
[31] a) E. L. Eliel, H. D. Banks, J. Am. Chem. Soc. 1972, 94, 171–176;
b) E. L. Eliel, H. D. Banks, J. Am. Chem. Soc. 1972, 94, 8587–8589;
c) E. L. Eliel, F. Alcudia, J. Am. Chem. Soc. 1974, 96, 1939–1941.
[32] a) Biomedical Aspects of Fluorine Chemistry, (Eds.: R. Filler, Y. Ko-
bayasi), Elsevier, New York, 1982; b) Biomedical Frontiers of Fluo-
rine Chemistry, (Eds.: I. Ojima, J. R. McCarthy, J. T. Welch), ACS
Symposium Series No. 639, Washington D. C., 1996; c) D. OꢂHagan,
H. S. Rzepa, Chem. Commun. 1997, 645–652; d) B. K. Park, N. R.
Kitteringham, P. M. OꢂNeill, Annu. Rev. Pharmacol. Toxicol. 2001,
41, 443–470.
[33] a2-Adrenoreceptor ligands: a) G. L. Grunewald, V. H. Dahanukar,
R. K. Jalluri, K. R. Criscione, J. Med. Chem. 1999, 42, 1982–1990;
b) G. L. Grunewald, T. M. Caldwell, Q. Li, K. R. Criscione, J. Med.
Chem. 1999, 42, 3588–3601; c) G. L. Grunewald, T. M. Caldwell, Q.
Li, K. R. Criscione, J. Med. Chem. 2001, 44, 2849–2856;
[10] a) C. H. Bushweller in Conformational Behavior of Six-Membered
Rings, (Ed.: E. Juaristi), VCH Publishers, 1995, Ch 2, p. 25; b) E. L.
Eliel, S. H. Wilen, Stereochemistry of Organic Compounds, Wiley,
New York, 1994, p. 696.
[11] For example, Becke3LYP/6–311G(d,p) optimization of (CH3)4N+
leads to the following Mulliken charge distribution: N À0.32, C
À0.20, H +0.18. Within the NBO framework,[12,25] the charges for
the same structure are: Becke3LYP, N À0.31, C À0.35, H +0.23;
MP2, N À0.29, C À0.32, H +0.21.
[12] E. D. Glendening, A. E. Reed, J. E. Carpenter, F. Weinhold, NBO
Version 3.1 as implemented in Gaussian 98;[45] A. E. Reed, L. A.
Curtiss, F. Weinhold, Chem. Rev. 1988, 88, 899–926; cf. http://
www.chem.wisc.edu/~nbo5/
[13] D. S. Balley, J. A. Walder, J. B. Lambert, J. Am. Chem. Soc. 1972, 94,
177–180.
1590
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2005, 11, 1579 – 1591

Persistence of Axial Fluorine
FULL PAPER
d) G. L.Grunewald, F. A. Romero, K. R. Criscione, J. Med. Chem.
2004, ASAP, DOI: 10.1021/jm049594x.
[40] SAINT Version 6.02, 1999, Bruker AXS, Inc., Analytical X-ray Sys-
tems, 5465 East Cheryl Parkway, Madison WI 53711–5373.
[41] SADABS, 1996, George Sheldrick, University of Gçttingen.
[42] SHELXTL V5.10, 1997, Bruker AXS, Inc., Analytical X-ray Sys-
tems, 5465 East Cheryl Parkway, Madison WI 53711–5373.
[43] International Tables for X-ray Crystallography, Volume C. (Ed.:
A. J. C. Wilson), Kynoch, Academic Publishers, Dordrecht, 1992,
Tables 6.1.1.4 (pp. 500–502) and 4.2.6.8 (pp. 219–222).
[44] F. Mohamadi, N. G. J. Richards, W. C. Guida, R. Liscamp, M.
Lipton, C. Caufield, G. Chang, T. J. Hendrickson, W. C. Still, J.
Comput. Chem. 1991, 12, 1110–1117; cf. http://www.schrodinger.-
com/Products/macromodel.html
[34] Serotonin-subtype receptor ligands: a) See reference [20]; b) D. J.
Hallett, U. Gerhard, S. C. Goodacre, L. Hitzel, T. J. Sparey, S.
Thomas, M. Rowley, J. Org. Chem. 2000, 65, 4984–4993; c) M.
Rowley, D. J. Hallett, S. Goodacre, C. Moyes, J. Crawforth, T. J.
Sparey, S. Patel, R. Marwood, S. Patel, S. Thomas, L. Hitzel, D.
OꢂConnor, N. Szeto, J. L. Castro, P. H. Hutson, A. M. MacLeod, J.
Med. Chem. 2001, 44, 1603–1614.
[35] a) H. S. Duewel, E. Daub, V. Robinson, J. F. Honek, Biochemistry
2001, 40, 13167–13176; b) J. Feeney, J. E. McCormick, C. J. Dauer,
B. Birdsall, C. M. Moody, B. A. Starkmann, D. W. Young, P. Francis,
R. H. Havlin, W. D. Arnold, E. Oldfield, J. Am. Chem. Soc. 1996,
118, 870–876.
[45] M. J. Frisch, et al,; Gaussian 98 (Revision A.1), Gaussian, Inc, Pitts-
burgh PA, 1998; cf. http://www.gaussian.com/
[36] W. Still, M. Kahn, A. Mitra, J. Org. Chem. 1978, 43, 2923–2925.
[37] N. Sperber, M. Sherlock, D. Papa, J. Am. Chem. Soc. 1953, 75,
1122–1124.
[38] a) T. Parella, Magn. Reson. Chem. 1998, 36, 467–495; b) W. F. Rey-
nolds, R. G. Enriquez, J. Nat. Prod. 2002, 65, 221–244.
[46] Spartan ’02, Wavefunction, Inc.; Irvine, CA; cf. http://www.wavefun.-
com
[47] Mercury software; CSD version 5.24 (November 2002); cf. http://
www.ccdc.cam. ac.uk
[39] SMART Version 5.55, 2000, Bruker AXS, Inc., Analytical X-ray
Systems, 5465 East Cheryl Parkway, Madison WI 53711–5373.
Received: August 13, 2004
Published online: January 20, 2005
Chem. Eur. J. 2005, 11, 1579 – 1591
www.chemeurj.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1591