Preparation of fluoroadamantane acids and amines: Impact of bridgehead fluorine substitution on the solution- and solid-state properties of functionalized adamantanes

Article abstract of DOI:10.1021/ja992652x

Functionalized adamantanes are utilized as medicinal therapeutics and a practical route to novel bridgehead fluorinated adamantylamines and acids including fully fluorinated 3,5,7-tri-fluoroadamantane-1-carboxylic acid 1 and 3,5,7-trifluoroadamantane-1-am

Full text of DOI:10.1021/ja992652x

466
J. Am. Chem. Soc. 2000, 122, 466-473
Preparation of Fluoroadamantane Acids and Amines: Impact of
Bridgehead Fluorine Substitution on the Solution- and Solid-State
Properties of Functionalized Adamantanes
V. John Jasys,† Franco Lombardo,† Troy A. Appleton,† Jon Bordner,† Martine Ziliox,‡ and
Robert A. Volkmann*,†
Contribution from the Department of Medicinal Chemistry and Neuroscience, Pfizer Central Research,
Groton, Connecticut 06340, and Bruker Instruments, Inc., Billerica, Massachusetts 01821
ReceiVed July 26, 1999
Abstract: Functionalized adamantanes are utilized as medicinal therapeutics and a practical route to novel
bridgehead fluorinated adamantylamines and acids including fully fluorinated 3,5,7-tri-fluoroadamantane-1-
carboxylic acid 1 and 3,5,7-trifluoroadamantane-1-amine 2 is described. Potassium permanganate-mediated
bridgehead hydroxylation followed by DAST fluorination was utilized to sequentially insert bridgehead fluorine
atoms. 13C NMR chemical shifts of mono-, di-, and trifluoroadamantylamines, acids and carbamates reveal
long-range fluorine-mediated electronic effects within the adamantane framework, most notably by substrates
bearing three bridgehead fluorine atoms. The impact of multiple bridgehead fluorine substitution on structure
(X-ray crystallography) and on physicochemical parameters such as acidity (pKa) and lipophilicity (log P)
which are important for the design of medicinal therapeutics is assessed.
Introduction
influenza infection. Adamantane-containing drugs have shown
efficacy as antiviral, hypoglycemic, antiarrythmic, antidepres-
sant, and antitumor agents, and a number of these compounds
are currently in development.7 Incorporation of the cage-shaped
adamantane nucleus into medicinal agents has provided a viable
approach for designing molecules which can access lipophilic
cell membranes, and this strategy has been particularly useful
in facilitating the penetration of the blood-brain barrier by CNS
therapeutics.8
The benefits of adamantane lipophilicity, however, are often
undermined by functionality’s susceptibility to metabolic oxida-
tive degradation. P-450 hydroxylation of adamantanes is well
documented, with bridgehead oxidation being preferred (3° >
2°).9 The utility of adamantanes in medicinal chemistry has
understandably been hampered by this metabolic liability, and
a strategy which exploits the attractive physicochemical and
pharmacological properties of adamantanes yet manages to
obviate the group’s susceptibility to oxidative degradation is
greatly needed. Isosteric replacement of the three bridgehead
hydrogen atoms with fluorine atoms would provide one such
strategy by removing the molecule’s major avenue of metabolic
degradation while at the same time maintaining the group’s
overall structure. Fluorine incorporation, however, would likely
For over sixty years, adamantane ring-containing compounds
have interested organic chemists.1 The rigid tricyclodecane
skeleton has provided a unique structural template for evaluating
important theoretical concepts in chemistry. Adamantanes have
been particularly useful for probing through-bond and through-
space electronic effects2 and have significantly contributed to
our understanding of hyperconjugation (homo-hyperconjuga-
tion),3 and degenerate rearrangements.4 Furthermore they pro-
vide molecular scaffolds for investigating the transmission of
electronic effects of remote substituents in saturated systems
and its contribution to chemical reactivity5 and to diastereose-
lectivity in addition/elimination reactions.3
The physical and chemical properties of the adamantane
nucleus have largely overshadowed the structure’s contribution
to the discovery of human therapeutics. Adamantadine6 (1-
aminoadamantane), for example, is currently utilized for the
treatment of Parkinson’s disease and for the prophylaxis of
* To whom correspondence should be addressed.
† Pfizer Central Research.
‡ Bruker Instruments, Inc.
(1) (a) Landa, S. Chem. Listy 1933, 27, 415 and 1933, 27, 443. (b) Fort,
R. C., Jr.; von R Schleyer, P. Chem. ReV. 1964, 277-300 and references
therein.
(2) (a) Adcock, W.; Kok, G. B. J. Org. Chem. 1987, 52, 356-364. (b)
Duddeck, H. Tetrahedron 1978, 34, 247-251. (c) Duddeck, H.; Feuerhelm,
H.-T. ibid. 1980, 36, 3009-3015. (d) Duddeck, H.; Hollowood, F.; Karim,
A.; McKervey, M. A. J. Chem. Soc., Perkin Trans. 2 1979, 360-365.
(3) (a) Kaselj, M.; Adcock, J. L.; Luo, H.; Zhang, H.; Li, H.; le Noble,
W. J. J. Am. Chem. Soc. 1995, 117, 7088-7091. (b) Lau, J.; Gonikberg, E.
M.; Hung, J.-t.; le Noble, W. J. ibid. 1995, 117, 11421-11425. (c) Adcock,
W.; Trout, N. A. Chem. ReV. 1999, 99, 1415-1435.
(4) Majerski, Z.; Liggero, S. H.; von R. Schleyer, P.; Wolf, A. P. Chem.
Commun. 1970, 1596-1597.
(7) A number of adamantane-containing compounds are commercial
drugs or are in clinical development. These include: adapalene (antiacne;
multiple sclerosis) Eur. Pat. 199,636, 1989; amantol (antifungal; antibacte-
rial) U.S. Patent 4,288,609, 1981; memantine (peripheral neuropathy; pain;
ocular disease) Eur. Pat. 392,059, 1993; rimantadine (influenza, viral
infections) U.S. Patent 3,352,912, 1967; tromantadine (antiviral) DE
1,941,218, 1972; SR-48692 (anticancer, neuroleptic) Eur. Pat. 699,438, 1996;
HT-90B (antidepressant) WO 93/08182, 1993.
(8) (a) Tsuzuki, N.; Hama, T.; Kawada, M.; Hasui, A.; Konishi, R.;
Shiwa, S.; Ochi, Y.; Futaki, S.; Kitagawa, K. J. Pharm. Sci. 1994, 83, 481-
484. (b) Tsuzuki, N.; Hama, T.; Hibi, T.; Konishi, R.; Futaki, S.; Kitagawa,
K. Biochem. Pharmacol. 1991, 41, R5-R8. (c) Wishnok, J. S. J. Chem.
Educ. 1973, 50, 780-781.
(9) (a) White, R. E.; McCarthy, M.-B.; Egeberg, K. D.; Sligar, S. G.
Arch. Biochem. Biophys. 1984, 228, 493-502. (b) Raag, R.; Poulos, T. L.
Biochemistry 1991, 30, 2674-2684.
(5) Gleicher, G. J.; von R Schleyer, P. J. Am. Chem. Soc. 1967, 89, 582-
593.
(6) (a) Kolocouris, N.; Kolocouris, A.; Foscolos, G. B.; Fytas, G.; Neyts,
J.; Padalko, E.; Balzarini, J.; Snoeck, R.; Andrei, G.; De Clercq, E. J. Med.
Chem. 1996, 39, 3307-3318. (b) Mueller, R.; Auterhoff, H. Arch. Pharm.
1969, 39, 1-6.
10.1021/ja992652x CCC: $19.00 © 2000 American Chemical Society
Published on Web 01/08/2000

Preparation of Fluoroadamantane Acids and Amines
J. Am. Chem. Soc., Vol. 122, No. 3, 2000 467
Scheme 1. General Route to Fluoroadamantanes
carbamate 4 as well as di- and trifluoroadamantane carbamates
in the course of this work. Accordingly, 3 was treated with
diphenylphosphoryl azide15 (TEA, benzyl alcohol 70 °C) to
afford benzyl carbamate 4. Hydrogenolysis (10% Pd/C, HOAc)
cleanly yielded amine 5 as its acetate salt (Figure 1).
Oxidation of mono- and difluoroadamantane carboxylic acids
would be required to access the remaining alcohol targets, and
their preparation suggested a number of viable reagents.
Dimethyldioxirane,16 potassium permanganate,13d ferrous iron-
molecular oxygen,17 and perfluorooxaziridines18 have been
utilized to generate 3-hydroxy-1-adamantanecarboxylic acid
from adamantane carboxylic acid. Treatment of the sodium salt
of 3-fluoroadamantane carboxylic acid 3 with potassium per-
manganate afforded, after removal of unreacted starting material
and esterification (methyl iodide treatment of the product’s
tetrabutylammonium salt), methyl-3-fluoro-5-hydroxyadaman-
tane-1-carboxylate 6.
Diethylaminosulfur trifluoride (DAST)-mediated fluorinations
of alcohols have been utilized for generating, in high yield, a
variety of fluorides, and this reagent was a logical choice for
accessing the desired bridgehead fluorides.12 Treatment of 6 with
DAST in refluxing chloroform followed by saponification of
the crude product afforded 3,5-difluoroadamantane-1-carboxylic
acid 7,19 mp 162-164 °C, in >95% yield. 3,5-Difluoroada-
mantane-1-amine 9 could be smoothly obtained, as described
for the conversion of 3 to 5, as its acetate salt by treatment of
7 with diphenylphosphoryl azide (TEA, benzyl alcohol 70 °C)
and subsequent hydrogenolysis of the resultant benzyl carbamate
8 (10% Pd/C, HOAc).
The same protocol employed for the synthesis of 3,5-
difluoroadamantane-1-carboxylic acid 7 and amine 9 was
adapted for the preparation of the trifluoroadamantane analogues.
Both the potassium permanganate oxidation to generate 10 and
the resultant DAST fluorination required higher temperatures/
longer reaction times, reflecting the presence of the additional
electron-withdrawing fluorine substituents. Nonetheless, 3,5,7-
trifluoroadamantane-1-carboxylic acid 1, mp 198-199 °C,
carbamate 11, mp 91.5-92.0 °C, and amine 2 as its acetate
salt were generated using this procedure.
Spectroscopic Data. 13C and 19F NMR spectroscopy has been
utilized to probe the through-bond and through-space electronic
effects of substituents in structurally rigid organic substrates
and to quantify polar substituent effects in saturated systems.
Substituted fluoroadamantanes including 3-substituted-1-fluoro-
adamantanes,2a,20 perfluoroadamantyl halides,21 fluorodiada-
mantanes,11a mono-, di-, tri-, and tetrafluoroadamantanes2a,11a
were examined in these pioneering studies, and these data, in
concert with reaction rate (solvolyses) and/or equilibria data,
have revealed a rather complicated picture of electronic
transmission modes in saturated systems. Long-range adamantyl
substituent effects have been examined by 13C NMR. While
definite trends are revealed for proximal and distal methylene
carbon resonances, bridgehead carbon resonances have proven
less predictable.11a
effect substrate lipophilicity and, as a result, alter the physico-
chemical and pharmacological properties of the adamantane
core. The magnitude of such an effect is unknown and certainly
would be an important determinant for the utility of these
substrates in drug discovery.
Since a number of important adamantane-containing drugs
can be generated from adamantane-1-carboxylic acid or ada-
mantane-1-amine, fluorine-substituted derivatives of these two
key synthons were selected as synthetic targets for testing the
viability of this overall approach. In addition to possible
medicinal applications, these substrates would likely also be of
interest for assessing bridgehead fluorine-mediated through-bond
and through-space electronic effects, given the structural rigidity
of the adamantane nucleus. Accordingly, we describe herein
the preparation of bridgehead-bearing fluorine-substituted ada-
mantane amines and acids including 3,5,7-trifluoroadamantane-
1-carboxylic acid 1 and 3,5,7-trifluoroadamantane-1-amine 2
and the selected physical properties of these compounds.
Results and Discussion
Fluoroadamantane Synthesis. Three synthetic strategies
were entertained for bridgehead fluorine incorporation: (1)
direct adamantane fluorination,10 (2) halogen (bromine/chlorine/
iodine) insertion followed by fluorine exchange,11 and (3)
sequential bridgehead oxidation-alcohol-to-fluorine conver-
sion.12 We opted for the latter since the technology for
monohydroxylation of bridgehead carbon atoms is well estab-
lished, as is the conversion of tertiary alcohols to fluorides.
While the ability to hydroxylate mono- and difluoroadamantane
carboxylic acids had not been previously demonstrated, the
likelihood for sequential application of this method seemed
reasonable, thereby providing a potential route to di- and
trisubstituted fluoroadamantane acids. These acids, in turn,
would be logical precursors (via Curtius rearrangement) to the
corresponding substituted fluoroadamantylamines. This some-
what lengthy protocol (Scheme 1) would allow the entire set
of mono-, di-, and trifluoro bridgehead adamantylamines (and
carbamates) and acids to be generated, thereby providing
interesting medicinal synthons as well as structures of potential
theoretical interest, capable of probing fluorine-mediated through-
bond and through-space electronic effects on the rigid saturated
adamantane acid and amine core.
The starting 3-fluoroadamantane carboxylic acid 3,2a,13 mp
154-156 °C, was easily obtained from 3-hydroxyadamantane
carboxylic acid. This transformation along with the direct
conversion of 3 to 3-fluoroadamantane-1-amine 5,2a,13d,14a have
been previously described.2a,13d We opted, in addition, to isolate
(15) Huc, I.; Rebek, J., Jr. Tetrahedron Lett. 1994, 35, 1035-1038.
(16) Murray, R. W.; Gu, H. J. Org. Chem. 1995, 60, 5673-5677.
(17) Miura, T.; Shibata, K.; Sawaya, T.; Kimura, M. Chem. Pharm. Bull.
1982, 30, 67-73.
(10) Gal, C.; Rozen, S. Tetrahedron Lett. 1985, 26, 2793-2796.
(11) (a) Olah, G. A.; Shih, J. G.; Krishnamurthy, V. V.; Singh, B. P. J.
Am. Chem. Soc. 1984, 106, 4492-4500. (b) Bhandari, K. S.; Pincock, R.
E. Synthesis 1974, 655-656.
(12) Middleton, W. J. J. Org. Chem. 1975, 40, 574-578.
(13) (a) Stetter, H.; Mayer, J. Chem. Ber. 1962, 95, 667-672. (b) Bolte,
G.; Haas, A. ibid. 1984, 117, 1982-1986. (c) Bolte, G.; Haas, A. J. Fluorine
Chem. 1983, 23, 502. (d) Anderson, G. L.; Burks, W. A.; Harruna, I. I.
Synth. Commun. 1988, 18, 1967-1974.
(14) (a) Kollonitsch, J.; Barash, L.; Doldouras, G. A. J. Am. Chem. Soc.
1970, 92, 7494-7495. (b) Anderson, G. L.; Randolph, B. J.; Harruna, I. I.
Synth. Commun. 1989, 19, 1955-1963.
(18) Sorochinsky, A. E.; Petrenko, A. A.; Soloshonok, V. A.; Resnati,
G. Tetrahedron 1997, 53, 5995-6000.
(19) Sorochinsky, A. E.; Aleksandrov, A. M.; Kukhar, V. P.; Gamaleya,
V. F.; Krasnoshchek, A. P. J. Org. Chem. USSR (Engl. Transl.) 1979, 15,
2240-2244.
(20) Wahl, G. H., Jr.; Peterson, M. R., Jr. J. Am. Chem. Soc. 1970, 92,
7238-7239.
(21) Adcock, J. L.; Luo, H.; Dai, S. Magn. Reson. Chem. 1993, 31, 969-
971.

468 J. Am. Chem. Soc., Vol. 122, No. 3, 2000
Jasys et al.
Figure 1. Fluoroadamantane acid/amine synthesis.
13C NMR chemical shifts were obtained for fluoroadamantyl
carbamates 4, 8, and 11, amines 2, 5, and 9 and acids 1, 3, and
7 and are shown in Table 1. Comparative analysis of 13C NMR
chemical shifts for the fluoroadamantanes reveal interesting
effects resulting from bridgehead fluorine substitution, particu-
larly at the bridgehead carbon atoms. A downfield shift in the
13C NMR was observed for the bridgehead carbon atoms bearing
acid, amine, and carbamate groups with sequential addition of
up to two bridgehead fluorine atoms (carbon 1, Table 1),
consistent with removal of electron density at this center.
Replacement of the last remaining bridgehead hydrogen atom
to generate the trifluoroadamantyl derivatives 1, 2, and 11,
however, results, in all three cases, in an upfield 13C NMR
chemical shift at this center (greater than 2 ppm higher than
the corresponding difluoro derivatives). In addition, similar
results were obtained for the fluorine-bearing bridgehead carbon
atoms (carbon atom 3, Table 1) as an upfield shift (1.2-2.0
ppm) was observed for the fluorine-bearing bridgehead carbon
atoms in trifluoroadamantyl derivatives 1, 2, and 11.
-129.14 ppm while difluoroadamantane 9 was -135.78 ppm
and trifluoroadamantane 2 was at -144.33 ppm.
X-ray Crystallographic Data. Considering the rigid ada-
mantane template, small molecular distortions in the solid-state
structures of mono-, di-, and trifluoroadamantane derivatives
could serve as an indicator of significant inductive, electronic,
and polarization effects caused by the fluorine-containing
bridgehead carbon atoms. While symmetrically substituted
adamantanes often do not form crystals suitable for X-ray
structural determination due to the formation of plastic crystal-
line phases,22 the fact that adamantane-1,3,5,7-tetracarboxylic
acid23 as well as bridgehead halogenated adamantanes24 have
been crystallized was encouraging.
X-ray structures were obtained for mono-, di-, and trifluo-
roadamantane carboxylic acids 3, 7, and 1. Similar to adaman-
tane carboxylic acid,25 di- and trifluoroadamantane acids 7 and
1 crystallize as two molecules in the asymmetric unit joined
together by hydrogen bonding of the carboxylic acid moiety.
Selected adamantane bond lengths and angles are shown in
Figures 2 and 3. A complete summary of bond angles and bond
lengths of these compounds is included with the Supporting
Selected 13C-19F coupling constants are included in Table
1. Sequential addition of bridgehead fluorine atoms results in
an increase in carbon-fluorine coupling. For example, in the
1
(22) a) Fort, R. C., Jr. Adamantane. The Chemistry of Diamond
Molecules; Marcel Dekker: New York, 1976. (b) Rao, C. N. R. In Organic
Solid State Chemistry; Desiraju, G. R., Ed.; Elsevier: Amsterdam, 1987; p
371.
(23) Ermer, O. J. Am. Chem. Soc. 1988, 110, 3747-3754.
(24) Bremer, M.; Gregory, P. S.; von R Schleyer, P. J. Org. Chem. 1989,
54, 3796-3799.
(25) Belanger-Gariepy, F.; Brisse, F.; Harvey, P. D.; Gilson, D. F. R.;
Butler, I. S. Can. J. Chem. 1990, 68, 1163-1169.
carbamate series, JC3F (triflouroadamantane 11) ) 190.29 >
1
1JC3F (difluoroadamantane 8) ) 188.67 > JC3F (monofluoro-
adamantane 4) ) 184.30. A similar trend was obtained in the
amine and acid series. Also sequential addition of fluorine atoms
on the adamantane framework provides a progressive increase
in 19F NMR chemical shifts. For example, the 19F NMR
chemical shift for monofluoroadamantane amine salt 5 was

Preparation of Fluoroadamantane Acids and Amines
J. Am. Chem. Soc., Vol. 122, No. 3, 2000 469
Table 1. 13C NMR Chemical Shifts of the Fluoroadamantanesa-d
a Chemicals shifts are in ppm ((0.05) with respect to external Me4Si in CDCI3 at 20 °C. b d ) double, t ) triplet, q ) quartet, m ) complex
c
multiplet, dd ) double of doublets, dt ) doublet of tripletssthese refer to C-F coupling patterns. 13C-19F coupling constants (in hertz) are given
d
in parentheses. 19F chemical shifts are in ppm ((0.05) with respect to external CFCI3 in CDCI3 at 20 °C. Negative sign indicates upfield shift.
Information. The bridgehead fluorine atom(s) not only influence
the crystal packing within the unit cell but also bond lengths
(and angles) within the adamantane skeleton. Bond lengths of
mono- and difluoroadamantane acids 3 and 7 are well ordered.
Given the pseudoplane of symmetry (neglecting the acid
torsional angle) in monofluoroadamantane 3 (plane through C-1,
C-3, and C-6) and difluoroadamantane 7 (plane through C-1,
C-7, and C-9), bond lengths reflect the symmetry of the
adamantane skeleton, and equivalent adamantyl carbon-carbon
bonds vary by <0.02 Å.
In contrast, the X-ray crystal structure of trifluoroadamantane
acid 1 reveals a disordered structure that crystallizes as two
distinct (and not equivalent) entities, with C-C adamantyl bond
lengths ranging from 1.438 to 1.595 Å (Figure 2) and bond

470 J. Am. Chem. Soc., Vol. 122, No. 3, 2000
Jasys et al.
Figure 4. The aqueous pKa’s were determined by potentiometric
titration. Log P values were determined by potentiometric methods by
comparing the aqueous pKa’s to those obtained with octanol. Compound
assignments are in bold. The subscripts in F0, F1, F2, F3 refer to the
number of adamantane bridgehead fluorine atoms.
Figure 2. Carbon-carbon bond lengths of fluoroadamantanes 3, 7 and
1 were obtained from the X-ray crystallographic data. Di- and
trifluoroadamantane carboxylic acids have two molecules in the
asymmetric unit and bond lengths from both molecules are included.
to hydrogen (i.e., the carboxyl-bearing bridgehead carbon atom)
could be assigned. The solid state 13C NMR spectra supported
the X-ray crystallography data as doubling of resonances were
observed, consistent with inequivalent sites in the crystal
structure. Particularly noteworthy was the chemical shift of the
bridgehead carbon atom bearing the carboxylic acid moiety. The
bridgehead carbon atom for monofluoro 3 was found at 45.705
and 45.963 ppm and difluoro 7 at 46.629 ppm, displaying the
same trend seen in solution NMR. Once again, a significant
upfield shift (43.968 and 41.930 ppm) was observed for the
carboxyl-bearing bridgehead carbon atom in both molecules of
trifluoroadamantane acid 1 found in the asymmetric unit,
demonstrating that the solid-state NMR data mirrors that of
solution NMR, and both, in concert with the X-ray crystal-
lography data, are appropriate for assessing the electronic and
structural effects of bridgehead fluorine substitution.
Physicochemical Parameters of Fluoroadamantanes. The
impact of long-range electronic and structural effects resulting
from fluorine substitution on physicochemical parameters such
as acidity (pKa) and lipophilicity (log Poct), which are important
for the design of medicinal therapeutics, is open to question.
Log Poct and pKa values for the adamantane acids and amines
were obtained by potentiometric titration27 as described in the
Experimental Section. Substrates in aqueous 0.167 M NaCl at
low pH were titrated with 0.5 N NaOH to determine the pKa’s.
The solutions were reacidified, treated with 1-octanol and titrated
again. Log Poct values were calculated from shifts in the pKa
titration curves obtained with 1-octanol.
Figure 3. Selected C-C-C and C-C-F bond angles of fluoroada-
mantanes 3, 7, and 1 were obtained from the X-ray crystallographic
data. Adamantane carbon atoms are labeled as shown in Table 1.
angles ranging from 104.5° to 114.1°. That the X-ray crystal
structure of trifluoroadamantane acid 1 is an aberration appears
not to be the case. X-ray crystal structures of adamantane-1-
carboxylic acid25 (bearing three bridgehead hydrogen atoms)
and 1,3,5 trifluoro-7-iodoadamantane24 show similar distortion
of the adamantane framework at ambient temperature. Moreover,
X-ray studies have shown that a number of adamantane-
containing compounds such as 1- and 2-hydroxyadamantane,
1-adamantanamyne, 1-halogenoadamantanes, and a number of
alkyladamantanes undergo an order-disorder transition, and the
effect of substituent groups on the disordered phase behavior
of these molecules has been investigated.26 The crystal structure
of adamantane-1-carboxylic acid has been determined by X-ray
diffraction at two temperatures, 173 and 280 K. At both
temperatures, the unit cells are triclinic, space group P1, and
each contains two molecules. The low-temperature form is
ordered, while the molecule at room temperature is dis-
ordered.25
Typically, sequential addition of fluorine atoms to aliphatic
molecules results in incremental changes in both pKa and log
P
oct values.28 Bridgehead fluorine incorporation has a significant
Solid State 13C NMR. Since solution NMR data was utilized
to assess the long-range electronic effects exerted by the
adamantane bridgehead fluorine substituents and the structural
data for the three fluoroadamantane acids resides with single-
crystal X-ray crystallographic analyses, solid state 13C NMR
spectra (13 kHz) using regular CP-MAS were obtained for
mono-, di-, and trifluoroadamantane acids 3, 7, and 1 to
complement the solution data. Not all of the carbon atoms in
the solid state spectra were definitively assigned. Nonetheless,
by utilizing dipolar dephasing to suppress carbon atoms with
strong dipolar coupling to protons, those not directly attached
impact on the acidity of the adamantane acids. A linear
relationship exists between fluorine substitution and pKa (Figure
4). Accordingly, trifluoroadamantane carboxylic acid 1 has a
pKa value of 3.24 ( 0.07, while difluoroadamantane acid 7 has
a value of 3.83 ( 0.04 and monofluoro 3 a value of 4.22 (
0.02. This relationship follows suit with the corresponding
adamantylamines with the conjugate acid of trifluoroadamantane
amine 2 having a pKa value of 7.39 ( 0.02 (versus 8.36 (
0.02 for difluoro 9 and 9.45 ( 0.02 for monofluoro 5).
(27) (a) Avdeef, D. J. Pharm Sci. 1993, 82, 183-190. (b) Avdeef, A.;
Barrett, D. A.; Shaw, P. N.; Knaggs, R. D.; Davis, S. S. J. Med. Chem.
1996, 39, 4377-4381.
(28) Chambers, R. D. In Interscience Monographs in Organic Chemistry:
Fluorine in Organic Chemistry; John Wiley and Sons: New York, 1973.
(26) Clark, T.; Mc. O. T.; Knox, H.; Mackle, H.; McKervey, M. A. J.
Chem. Soc., Faraday Trans. 1 1977, 73, 1224-1231.

Preparation of Fluoroadamantane Acids and Amines
J. Am. Chem. Soc., Vol. 122, No. 3, 2000 471
While fluorine addition to aromatic rings generally increases
substrate lipophilicity, addition to saturated aliphatic compounds
results in a decrease in lipophilicity as measured by log Poct. In
the case of the fluoroadamantanes, the relationship between
extent of fluorine substitution and substrate lipophilicity (log
internal C-C-C bond angle at this center would be expected
to provide a downfield shift, as opposed to the 2.48 ppm upfield
shift which was seen in the 13C NMR of 1, 2, and 11.
The fact that the lipophilicity of the fluorinated adamantanes
shows a nonlinear behavior (Figure 4) was also, at first, puzzling
and difficult to rationalize, since the addition of a fluorine atom
in going from difluoro 7 to trifluoro acid 1 was not expected to
have a great impact on the partition of these species between
water and octanol. One explanation that could be given is that
the carboxylic acids might self-associate and form a more
lipophilic “supramolecular” adduct which might partition more
easily into the organic phase, but it seemed unlikely that such
a sudden change would take place in going from 7 to 1.
Furthermore, Schweitzer and Morris29 report that the dimeriza-
tion (association) constant for octanoic acid in 1-hexanol is
essentially zero and Kojima et al.30 report rather similar findings
for propanoic acid, derived from distribution experiments
between water and hydrogen-bonding organic solvents.
The linear free-energy relationship (LFER) developed by
Abraham31 for log Poct has been employed to correlate octanol-
water partition with solvation parameters such as the hydrogen
bond donor and acceptor capability of a given solute. Initially,
we sought to confirm, by an independent method, that the
order of lipophilicity observed by the reverse titration method
was indeed correct. Pagliara et al.32 have shown that a good
correlation between the log Poct and HPLC capacity factors (k′)
can be observed under specified sets of conditions, and Abraham
et al.33 have also confirmed these observations and expanded
the analysis of HPLC stationary and mobile phases to include
correlations between log Palk and HPLC k′ values, via their
solvation parameters.
The method of Pagliara et al. uses either the k′ at 40%
methanol (k′40) or extrapolated values (k′w) at 0% of organic
solvent, on a Supelcosil LC-ABZ column, to correlate the
capacity factors to log Poct values. Even the 40% methanol
mobile phase, in our hands, gave very long retention times for
the adamantane acids. As a result, we resorted to the extrapola-
tion of k′ from a mobile phase containing 65% of methanol
down to 45% methanol, in 5% intervals. Under all of these
conditions, trifluoro acid 1 eluted always between difluoro and
monofluoro acids 3 and 7, and no crossover was observed. A
good linearity was found for the capacity factors of each
compound, and the extrapolation to either mobile phase recom-
mended by Pagliara et al. showed that, indeed, the lipophilicity
was in the same order as found via the reverse titration method.
Incidentally, computed log Poct values for the fluoroadamantane
carboxylic acids, obtained from commonly available fragmental
constant methods, were not predictive of the observed results.
The LFER equation mentioned above31 considers several
parameters such as the hydrogen-bonding (donor and acceptor)
capability, the dipolarity/polarizability, the volume, and the
molar refractivity of the solutes. A retrospective analysis of these
parameters demonstrated, as expected, that all of the values are
either progressively increasing (H-bond acidity, volume) or
decreasing (H-bond basicity, excess molar refractivity), and the
Poct)
27 is not linear and diverges once three bridgehead fluorine
atoms have been added to the adamantane framework. Fluorine
incorporation results in a decrease in substrate lipophilicity for
mono (log Poct 2.55 ( 0.04) and difluoro (log Poct 2.16 ( 0.08)
acids 3 and 7 versus adamantane carboxylic acid (log Poct 3.26
( 0.04) and mono- (log Poct 1.48 ( 0.04) and difluoro (log
Poct 0.98 ( 0.10) amines 5 and 9 versus adamantylamine (log
Poct 2.47 ( 0.04). The decrease in log Poct value for trifluoro-
adamantane amine 2 (log Poct 0.88 ( 0.06), however, is smaller
than anticipated, and in fact an increase in log Poct (2.50 (
0.14) is observed for trifluoroadamantane carboxylic acid 1
(Figure 4).
Discussion
X-ray crystallographic data for adamantane acids 3, 7, and 1
in concert with 13C NMR analysis of the nine adamantane acids,
amines, and carbamates reveal significant structural and elec-
tronic effects mediated by the bridgehead fluorine atoms present
in these molecules. The reason behind the unexpected upfield
13C NMR chemical shift for the bridgehead carbon atom bearing
acid, amine, and carbamate groups upon addition of a third
fluorine atom is subject to debate. It is tempting to correlate
the 13C NMR shifts with the internal C-C-C bond angles
obtained from the X-ray crystallographic data. The addition of
bridgehead fluorine atoms would be expected to effect carbon
hybridization since the electronegative fluorine atom should pull
p character into the C-F bond, thereby increasing the s character
of the C-C bonds at these centers. In addition to through-bond
and through-space inductive effects, one would expect that, as
the internal adamantane C-C-C bond angles at these centers
get larger, the 13C NMR chemical shifts should increase
(downfield shift) because there is more s character in the C-C
bonds.
The X-ray crystallographic data of adamantane acids 3, 7,
and 1 indicate that, at the fluorine-substituted bridgehead carbon
atoms, the average external C-C-F bond angles are less than
109.5° and the internal C-C-C bonds greater than 109.5°
(Figure 3). In the case of monofluoroadamantane carboxylic
acid 3, for example, the average external C-C-F bond angle
is 108.0°, while the average internal C-C-C bond angle at
this center is 110.9°. In the case of difluoroadamantane
carboxylic acid 7, the average external C-C-F bond angle
value is 108.4°, while the average internal C-C-C value at
the two centers is 110.5°. The situation is just the opposite for
the carboxyl-bearing bridgehead carbon atom where the average
value for external C-C-COOH bond angles for monofluoro-
adamantane acid 3 and difluoroadamantane acid 7 is 110.1°,
while the average value for internal C-C-C bond angles is
108.8° for 3 and 7. One would assume that trifluoroadamantane
acid 1 would have three large internal C-C-C bond angles at
the fluorine-bearing carbon atoms, leading to a further decrease
in the internal C-C-C bond angle at the carboxyl-bearing
center. The net result would be an upfield shift in the 13C NMR
for this bridgehead carbon atom. Because of the distortion of
the adamantane nucleus in X-ray crystal of 1, data analysis is
subject to question. However, if one averages bond angles for
equivalent carbon atoms, one finds that the average internal
C-C-C bond angle for the carboxyl-bearing carbon atom is
109.8° (versus 108.8° for 3 and 7). Such an increase in the
(29) Schweitzer, G. K.; Morris, D. K. Anal. Chim. Acta 1969, 45, 65-
70.
(30) Kojima, I.; Yoshida, M.; Tanaka, M. J. Inorg. Nucl. Chem. 1970,
32, 987-995.
(31) Abraham, M. H.; Chadha, H. S.; Whiting, G. S.; Mitchell, R. C. J.
Pharm. Sci. 1994, 83, 1085-1100.
(32) Pagliara, A.; Khamis, E.; Trinh, A.; Carrupt, P.-A.; Tsay, R.-S.;
Testa, B. J. Liq. Chromatogr. 1995, 18, 1721-1745.
(33) Abraham, M. H.; Chadha, H. S.; Leitao, R. A. E.; Mitchell, R. C.;
Lambert, W. J.; Kaliszan, R.; Nasal, A.; Haber, P. J. Chromatogr. A 1997,
766, 35-47.

472 J. Am. Chem. Soc., Vol. 122, No. 3, 2000
Jasys et al.
Table 2. Dipole Momentsa and Dipolarity/PolarizabilitybValue Calculated for Fluoroadamantanes
a All dipole moment computations were performed using the Spartan software package, version 5.1. All values were generated in the gas phase.
MMFF ) Merck Molecular Force Field. AM1 ) Austin Model 1. See Experimental Section. b See refs 33-34.
change is relatively modest, across the series of four adamantane
carboxylic acids.34 In the LFER equation, it turns out, the
coefficient for the hydrogen bond basicity is very large, with a
value of -3.46.31 This factor, a priori, might have influenced
the behavior of these analogues, if a larger difference between
7 and 1, were to be observed in terms of their hydrogen bond
basicity. That is, a significant decrease in the value of this
parameter, between the two analogues considered, and brought
about by the presence of a third fluorine atom would have
resulted in a significant contribution toward a higher log P value.
The estimate for the series of analogues, it was found, called
instead for a fairly modest and linear decrease across the series,
and the hydrogen bond basicity could not be invoked as an
explanation.34 Dipolarity/polarizability, however, does change
significantly across the series and reaches a maximum for
difluoro acid 7 and then a decrease is observed for trifluoro
acid 1. This observation, when coupled with the significant
negative coefficient found for the dipolarity/polarizability term
(-1.05) in the Abraham’s solvation equation,33 points to the
fact that the observed shift in this parameter is the likely cause
of the upward shift in log Poct. Table 2 shows the values of this
parameter for the adamantane carboxylic acids.34
is significant. The impact of three bridgehead fluorine atoms is
particularly noteworthy. X-ray crystallography reveals a highly
dis-symmetric structure for trifluoroadamantane acid 1, which
crystallizes as two distinct (and nonequivalent) entities.
13C NMR chemical shifts for the bridgehead carbon atoms
in the trifluoroadamantylamines, acids, and carbamates are not
well understood. Neither simple inductive and polarization
factors nor internal C-C-C bond angles account for the 13C
NMR chemical shifts of the bridgehead carbon atoms in these
molecules. The NMR data along with the log Poct values
underscore significant through-space effects resulting from the
incorporation of three bridgehead fluorine atoms in the ada-
mantane framework. An increase in log Poct resulting from the
incorporation of a third fluorine atom in the adamantane
framework was not expected, but parallels predicted differences
in the dipole moments of the fluoroadamantane acids.
Methods
Apparatus. The computerized titration instrument (Sirius PCA 101)
was used by Water Technology Associates to perform the pKa and log
P assays, which have been previously described in detail.27
Potentiometric Titrations. The aqueous pKa was determined by
potentiometric titration. If compounds were water insoluble, apparent
pKa’s were found in water/solvent ratios and the aqueous pKa was found
by extrapolating to 0% solvent according to the Yasuda-Shedlovsky
technique.35 The pH-metric technique consisted of two linked titrations.
The first titration is carried out in the aqueous phase, over a pH range
that encompasses the pKa of the drug. An approximate 1 mM solution
of each substrate at a constant ionic strength of 0.167 M NaCl was
titrated from low to high pH. The acid used was 0.5 N HCl and the
base used was 0.5 N NaOH. The acid and base were standardized to
four decimal places using NIST traceable standards. At the completion
of the first titration, 1-octanol (HPLC grade), which was obtained from
the Aldrich Chemical Co. and was water-saturated, was added, and
the reverse titration was performed, returning the two-phase solution
to the starting pH. The log P was determined by comparing the aqueous
pKa to the pKa obtained with 1-octanol.
We also have computed, at different levels of theory, the
dipole moment of the adamantane acids considered in this work,
and we found that in all cases there is a maximum reached for
7 and then a significant decrease for 1, further confirming the
fact that the dipole moment change is responsible for the
observed shift in log Poct. Table 2 shows the results of the
computations, and a clear shift is observed in going from 7 to
1.
Conclusions
In summary, a synthetic route has been developed to access
heretofore unknown trifluoroadamantane acid 1 and amine 2
along with related bridgehead-substituted fluoroadamantane
acids 3 and 7 and amines 5 and 9, possible synthons in medicinal
chemistry. The influence of bridgehead fluorine substitution on
structure (X-ray crystallography) and on physicochemical
parameters such as acidity (pKa) and lipophilicity (log Poct),
which are important for the design of medicinal therapeutics,
RP-HPLC Lipophilicity Determinations. HPLC analysis of the
fluoroadamantane acids was performed on a Spectra-Physics HPLC
system equipped with a P4000 quaternary pump and a AS3000
autosampler using a 4.6 × 150 mm Supelcosil LC-ABZ column (5
µm diameter) kept at ambient temperature, with a flow rate of 1.0 mL/
(34) Abraham, M. H., personal communication. We thank Professor
Abraham for useful comments and an analysis of our log P data.
(35) Avdeef, A.; Comer, J. E. A.; Thomson, S. J. Anal. Chem. 1993, 65,
42-49.

Preparation of Fluoroadamantane Acids and Amines
J. Am. Chem. Soc., Vol. 122, No. 3, 2000 473
min. An Applied Biosystems programmable UV detector was used for
detection at 202 nm. Ratios of MeOH:0.1% TFA in water ranging from
45:55 to 65:35 were used as mobile phases, with 5% increments. The
order of elution was also confirmed, using the 55:45 mobile phase
mixture, by identifying each peak on a HP1100 system equipped with
a Navigator LC/MS benchtop detector. No crossover of the order of
elution was observed, under any of these conditions, for any of the
compounds.
Acknowledgment. The authors gratefully acknowledge
insightful comments by Professor Jack Dunitz and stimulating
discussions with Professor Dan Kemp and Drs. James Blake,
Beryl Dominy, Edward Kleinman, and Jotham Coe. NMR
support was provided by Dr. Walter Massefski and Diane
Rescek. X-ray crystallographic results were obtained by Debra
DeCosta. Log P and pKa determinations were determined by
Dr. Cynthia M. Berger of pIon of Cambridge, Massachusetts.
The authors would like to thank Dr. Amy Antipas and Ms.
Barbara J. Brockhurst for their help with the LC-MS analysis.
Dipole Moment Calculations. The computations were performed
on a Silicon Graphics Indigo-2 workstation using a R4400 200-MHz
processor. All computations were performed using the Merck Molecular
Force Field,36 the semiempirical Austin Model 137 or the ab initio
Hartree-Fock level of theory as implemented in the Spartan software
package, version 5.1.38 A geometry optimization was performed at each
level of theory, followed by a gas-phase computation of the dipole
moments.
Supporting Information Available: Experimental details
and spectral data for all new compounds, crystal and refinement
parameters and tables of atomic coordinates, equivalent isotropic
displacement coefficients, bond lengths, bond angles, anisotropic
displacement coefficients, and H-atom coordinates and isotropic
displacement coefficients for fluoroadamantane carboxylic acids
1, 3, and 7 (PDF). This material is available free of charge via
the Internet at http://pubs.acs.org.
(36) Halgren, T. A. J. Comput. Chem. 1996, 17, 490-519 and following
papers in the same issue.
(37) Dewar, M. J.; Zoebisch, E. G.; Healy, E. F.; Stewart, J. J. P. J. Am.
Chem. Soc. 1985, 107, 3902-3909.
(38) SPARTAN software package, version 5.1, Wavefunction, Inc.:
Irvine, CA.
JA992652X