Preparation and reactions of 2,3,4,6-tetrafluoropyridine and its derivatives

Article abstract of DOI:10.1016/S0022-1139(99)00189-X

A reliable route to 2,3,4,6-tetrafluoropyridine has been established starting from the readily available 3,5-dichlorotrifluoropyridine by halogen exchange under controlled conditions to give 3-chlorotetrafluoropyridine and its subsequent hydrodechlorination using hydrogen over palladium on alumina at 250-270°C. The formation and reactions of the 3-lithio derivative have been studied with the aim of obtaining 3,4-disubstituted trifluoropyridines. Routes to such materials have been developed and their conversion to deazapurine derivatives as potential substrates for the generation of anti-sense nucleosides are reported.

Full text of DOI:10.1016/S0022-1139(99)00189-X

Journal of Fluorine Chemistry 101 (2000) 45±60
Preparation and reactions of 2,3,4,6-tetra¯uoropyridine and its derivatives
Paul L. Coe*, Anthony J. Rees
School of Chemistry, The University of Birmingham, Edgbaston, Birmingham B15 2TT, UK
Received 16 April 1999; accepted 3 August 1999
Abstract
A reliable route to 2,3,4,6-tetra¯uoropyridine has been established starting from the readily available 3,5-dichlorotri¯uoropyridine by
halogen exchange under controlled conditions to give 3-chlorotetra¯uoropyridine and its subsequent hydrodechlorination using hydrogen
over palladium on alumina at 250±2708C. The formation and reactions of the 3-lithio derivative have been studied with the aim of obtaining
3,4-disubstituted tri¯uoropyridines. Routes to such materials have been developed and their conversion to deazapurine derivatives as
potential substrates for the generation of anti-sense nucleosides are reported. # 2000 Elsevier Science S.A. All rights reserved.
Keywords: Exchange reactions; Hydrodechlorination; 2,3,4,6-tetra¯uoropyridine; Poly¯uoropyridyl lithium reagents; Deazapurine derivatives
1. Introduction
been shown that some amine nucleophiles will give a
signi®cant amount of product with ortho substitution on
reaction with poly¯uorocarboxylic acids [4]. Thus, it
seemed likely to us that if we could prepare tetra¯uoroni-
cotinic acid 6 (see Scheme 3) it could react in the 4-position
with ammonia to give us a potentially useful intermediate.
This reaction should be very likely to succeed as the 4-
position in poly¯uoropyridines is known to be the most
active towards nucleophilic substitution and together with
the added activating effect of the carboxyl group be even
more likely to lead to the desired product. Tetra¯uoronico-
tinic acid 6 has been prepared previously by Chambers [5]
from 2,3,4,6-tetra¯uoropyridine 7 which in turn was pre-
pared by the hydrogenation of 3-chloro-tetra¯uoropyridine
8. This work has not been pursued, possibly due to the
relative lack of available starting material. The attraction of
this route led us to reinvestigate both the formation of 3-
chloro-tetra¯uoropyridine 8 the starting material in the
Chambers' study and its hydrogenation to 2,3,4,6-tetra¯uor-
opyridine 7. We also proposed to make a more detailed study
of the reactions of 7.
Earlier work concerned with the preparation of poly¯uor-
opyridines [6,7] was of two types. Firstly, uncatalysed
reactions of ¯uoride ion with pentachloropyridine in dipolar
aprotic solvents such as sulpholane at 190±2008C, or sec-
ondly by reaction of pentachloropyridine and potassium
¯uoride in an autoclave without solvent at 400±4808C. In
the reaction in solvents both groups reported that the prin-
cipal product obtained was 3,5-dichlorotri¯uoropyridine 9
with only relatively small amounts of 3-chlorotetra¯uoro-
pyridine 8 and only trace amounts of penta¯uoropyridine 10
We have, for some time, been interested in the use of
¯uorine containing deazapurines as precursors of potential
anti-viral agents and as compounds for studies in anti-sense
nucleosides. There is some evidence that replacement of a
ring nitrogen in ribovirin derivatives with a CF-group has a
bene®cial effect on antiviral activity [1]. We decided that
poly¯uoropyridines were potential starting materials for this
project, particularly if it were possible to obtain 3,4- dis-
ubstituted compounds. There has been only one signi®cant
paper concerned with the preparation of these compounds
when the 3-substituent is not a halogen. Chambers [2]
showed that amination of 4-nitro-tetra¯uoropyridine 1
affords signi®cant amounts of the 3-amino-4-nitrocom-
pound 2, this is the same pattern of substitution that was
observed in a similar reaction with penta¯uoronitrobenzene
[3]. Reduction of 2 affords 3,4-diamino-tri¯uoropyridine 3,
which would be a possible precursor in our deazpurine
synthesis. However, the original reaction was complicated
by the formation of signi®cant quantities of 4-aminotetra-
¯uoropyridine 4, formed by aminodenitration and 2-amino-
4-nitrotri¯uoropyridine 5 as shown in Scheme 1. This
mixture of products was dif®cult to separate, which when
allied with the somewhat hazardous oxidation of amino to
nitro groups to obtain the starting material, did not suggest
this was a viable route to the desired diamine 3. We thus
decided to look for a more feasible route to our target. It has
*Corresponding author. Tel.: ‡44-121-414-4360; fax: ‡44-121-414-
4403.
0022-1139/00/$ ± see front matter # 2000 Elsevier Science S.A. All rights reserved.
PII: S 0 0 2 2 - 1 1 3 9 ( 9 9 ) 0 0 1 8 9 - X

46
P.L. Coe, A.J. Rees / Journal of Fluorine Chemistry 101 (2000) 45±60
Scheme 1.
being formed. In autoclave reactions it was found that at ca.
4508C the main products were the 3-chloro-tetra¯uoropyr-
idine 8 and penta¯uoropyridine 10. When the temperature
was raised to 4808C the major component (80±90%) was
penta¯uoropyridine 10. Although the latter reaction show
excellent results for the formation of penta¯uoropyridine 10
it suffers from the drawback of requiring somewhat specia-
list equipment and can only be used on a relatively limited
batch size. We required a route which, without the use of
specialist equipment, would lead to good yields of the 3-
chlorotetra¯uoropyridine as the major product. For the next
step in our proposed route we required good yields of
2,3,4,6-tetra¯uoropyridine 7. The preparation of this com-
pound on a relatively small-scale has been reported by
Chambers [5] and we decided to re-investigate the hydro-
genation route. Having obtained reasonable quantities of the
desired compound 7 the aim of the project was then to study
its reactions in the hope of ®nding a route to the required
deazapurine.
Thin layer chromatography was performed on TLC plas-
tic sheets silica gel 60 F254, pre-coated with a layer thickness
of 0.2 mm from Merck, Art. 5735. Gas chromatographic
analysis was carried out using a Philips PYE Unicam, Series
304 chromatograph with a 50-m CD-SIL-CB 19 column.
The data were registered by a JCL 600 chromatography data
system.
2.1. Fluorination of 3,5-dichlorotrifluoropyridine 9 with
potassium fluoride: a typical procedure
Sufolane (previously distilled in vacuo) (300 g), spray-
dried potassium ¯uoride (re-dried in a vacuum oven at 608C
for 12 h) (44 g, 0.759 mol) and tetraphenyl phosphonium
bromide (10 g, 0.024 mol) were heated together in vacuo
until ca. 30 cm3 of distillate were obtained. The ¯ask was
allowed to cool to below 508 C when 9 (105 g, 0.505 mol)
was added. The mixture was magnetically stirred at 2008 C
for 7 h under total re¯ux. After this time the mixture was
fractionally distilled from the solvent to yield a mixture of
poly¯uoropyridines (77.8 g) boiling between 808C and
1608C. GC analysis of the mixture showed three peaks
attributable, by comparison with a known mixture, to penta-
¯uorpyridine 10 (28.7%) [6,7], 3-chlorotetra¯uoropyridine
8 (65%) [6,7] and unreacted starting material 9 (6.3%).
A mixture from ®ve similar runs (590 g) was then re-
fractionated to yield (i) penta¯uoropyridine 10 (25.5 g) b.p.
83±848C cited 848C [6], (ii) mixed fractions of 8 and 10
(24.9 g) b.p. 85±1188C (iii) 3-chlorotetra¯uoropyridine 8
(189 g) b.p. 119±1208C cited 1198C [6] and a 1 : 1 mixture
of 8 and 9 (162 g), the latter was recycled in further
experiments. All the pure products were identi®ed by
comparison of their spectra with authentic samples.
2. Experimental
1H NMR (300 MHz) and 13C NMR spectra (75 MHz)
were measured on a Bruker AC 300 NMR spectrometer
1
unless stated otherwise. H NMR spectra (400 MHz) were
measured on a Bruker AMX 400 NMR spectrometer. 19F
NMR spectra were carried out either on a Jeol NMR
spectrometer, type FX 90 Q (84.26 MHz) or on a Bruker
AC 300 NMR spectrometer (282.4 MHz); tetramethylsilane
and ¯uorotrichloromethane were used as internal refer-
ences. The following abbreviations are used: s ˆ singlet,
d ˆ doublet, t ˆ triplet, q ˆ quartet, quin ˆ quintet, sext ˆ
sextet, m ˆ multiplet, dd ˆ doublet of doublets, dt ˆ doub-
let of triplets, ``quin'' ˆ pseudo quintet, etc. J values are
given in Hz. The mass spectra (CI-MS/EI-MS) were mea-
sured on a VG-Prospec-triple focusing mass spectrometer.
For GC-MS analysis, a Carlo Erba, 8000 series GC was used
with a 50-m column, BPX 5 (helium carrier gas, 70 eV,
electron impact).
2.2. Preparation of tetraphenylphosphonium bromide 11
Triphenylphosphine (52.4 g, 0.20 mol), nickel bromide
(36.46 g, 0.11 mol), bromobenzene (37.5 g, 0.24 mol) in
benzonitrile (25 cm3) were heated together under re¯ux

P.L. Coe, A.J. Rees / Journal of Fluorine Chemistry 101 (2000) 45±60
47
using a Dean and Stark trap. When no more water was
observed in the trap, heating was continued for 3 h under
total re¯ux. The mixture was cooled and then the product
was steam distilled from the residual solids. The organic
layer of the steam distillate, which solidi®ed on standing,
was dissolved in dichloromethane (400 cm3), the aqueous
layer was extracted with dichloromethane (2 Â 25 cm3), the
combined extracts and the original solution were dried
(MgSO4) and the solvent removed to leave a slush. This
latter was washed with ether (300 cm3) to leave a crystalline
solid which was ®ltered off and the residue washed with
ether several times to leave, after drying in vacuo, pure
tetraphenylphosphonium bromide (71.8 g, 85.6%) m.p.
290±2958C cited 2878C [9]; (Found: C, 68.4; H, 4.7%
C24H20BrP requires C, 68.7; 4.8%) ꢀH (acetone d6) 7.54
(m, 2H, 2-H), 7.73 (m, 2H, 3-H), 7.83 (m, 1H, 4-H); ꢀC
117.4 (d, 1JPC 89.5, 1-C), 130.9 (d, 2JPC 12.9, 2-C), 134.3 (d,
reaction was stirred for a further 60 min at 788C when the
reaction was allowed to warm to 558C and carbon dioxide
(dried by passing down a calcium chloride tube) was
bubbled into the solution. A white precipitate formed
immediately and the temperature began to rise. After two
hours of passing carbon dioxide through the solution all the
solvent had evaporated to leave a white paste. Hydrochloric
acid (2M, 40 cm3) and ether (40 cm3) were added to the
reaction ¯ask until all solids had dissolved. The ether layer
was separated, washed with brine (30 cm3), dried (MgSO4),
®ltered and the solvent removed on a rotary evaporator to
leave, after recrystallisation from hexane, 2,4,5,6-tetra¯uor-
onicotinic acid 6 (1.29 g, 6.56 mmol, 89.1%), m.p. 120±
1218C (cited [5] 121±1228C); ꢀF (acetone d6) 65.9 (1F, m,
4
3
2-F), 81.4 (1F, m, 6-F), 112.5 (ddd, 1F, JFF 23.4, JFF
19.1, 4JFF 10.6, 4-F), 167.0 (ddd, 1F, 5JFF 23.5, 3JFF 22.2,
3JFF 19.2, 5-F); ꢀC (acetone d6) 105.8 (m, 3-C), 134.4 (dddd,
1JCF 257.9, 2JCF 28.0, 2JCF 15.0, 4JCF 8.5, 5-C), 151.5 (dddd,
‡
4
3JP;3-C 10.3, 3-C), 135.8(d, JPC 2.8, 4-C); m/z 339 (M
2
3
3
[Ph4P]), 262, 185, 108.
1JCF 245.0, JCF 18.3, JCF 12.9, JCF 7.5, 6-C), 154.2
1
3
3
4
(dddd, JCF 249.3, JCF 16.2, JCF 9.7, JCF 3.2, 2-C),
1
2
3
3
3
2.3. Hydrogenolysis of 3-chlorotetrafluoropyridine 8
160.1 (dddd, JCF 71.8, JCF 10.8, JC, or JCF 7.6, JCF
‡
‡
or 3JCF 6.5, 4-C), 175.0 (s, CO2H); m/z 195 (M ), 178 (M ±
‡
‡
‡
In a typical experiment 8 (19.7 g) was dripped at a rate of
0.22 g/min onto the column head of a vertically mounted
quartz column (2.5 cm  15 cm heated length) packed with
0.5% palladium on alumina pellets (ex Johnson Matthey)
held in place by plugs of quartz wool and fused alumina.
The column was heated at 2758 C and a stream of hydrogen
at 50 cm3/min was passed through during the addition of 8.
The products were collected in a trap cooled in liquid air. At
the end of the addition the gas ¯ow was continued for
30 min when the products (16.7 g) were collected from the
trap. GC analysis showed the presence of 2,3,4,6-tetra¯uor-
opyridine 7 (76%) and 3-chlorotetra¯uoropyridine 8 (24%).
No products from hydrode¯uorination were observed. The
products from several runs (128 g) were fractionated
through a Spaltrohr column to yield (i) 2,3,4,6-tetra¯uoro-
pyridine 7 (65.4 g) b.p. 97±988C (cited [5] 89±908C); ꢀH
OH), 167 (M ±CO), 151 (M ±CO2), 150 (M ±CO2H or
‡
M ±NCF).
(ii) A repeat of this experiment (on a smaller scale) was
also carried out in hexane with a yield of 0.24 g of tetra-
¯uoronicotinic acid 6 (65%).
(iii) Using butyllithium and solid carbon dioxide. A
repeat of the experiment in which the carbon dioxide was
added as dry ice worked well giving good yields (in excess
of 70% as standard) on a small-scale (less than 1 g of
2,3,4,6-tetra¯uoropyridine). Increasing the scale of the
reaction up to 7.50 g of 2,3,4,6-tetra¯uoropyridine failed
and a yield of only 10% of 6 was obtained along with a large
amount of polymeric material.
2.5. Using lithium di-isopropylamide (LDA) and solid
carbon dioxide
5
3
4
(acetone d6) 7.29 (dddd, 1H, JHF 1.5, J HF 2.5, JHF 4.0,
3JHF 6.0, 5-H); ꢀF (acetone d6) 67.6 (bs, 1F, 6-F), 83.6
(bs, 1F, 2-F), 113.0 (ddd, 1F, JFF 9.2, JFF 33.6, JFF 18.3, 4-
F), 169 9 (m, 1F, 3-F); ꢀC (acetone d6) 98.3 (dddd, 2JFC
43.1, 2JFC 21.2, 4JFC 6.6, 4JFC 2.1, 5-C), 134.0 (dddd, 1JFC
255.4, 2JFC 28.5, 2JFC 14.2, 4JFC 8.5, 3-C), 150.6 (dm, 1JFC
2,3,4,6-Tetra¯uoropyridine (0.2 cm3, 0.30 g, 1.99 mmol)
was added to a solution of LDA in THF (25 cm3) at 788C
and the mixture stirred for 40 min. The cooling bath
was removed and crushed (solid) carbon dioxide was
added to the solution. A white precipitate formed as the
temperature began to rise towards room temperature. After
4 h water (5 cm3) was added and the reaction mixture
poured into 2M hydrochloric acid (20 cm3) and ether
(20 cm3), the ether layer was separated and washed with
saturated sodium bicarbonate solution until the aqueous
layer remained basic to pH paper. The ether layer was
removed and discarded. The aqueous layer was acidi®ed
with 2M hydrochloric acid and extracted with ether
(3 Â 50 cm3). The combined ether layers were washed
with brine (30 cm3) and dried (MgSO4), ®ltered and the
solvent removed to leave 2,4,5,6-tetra¯uoronicotinic acid
6 (0.11 g, 0.56 mmol, 28.1%).
1
1
260.8, 2-C), 155.9 (dm, JFC 244.3, 6-C), 161.0 (dm, JFC
‡
‡
‡
263.4, 4-C); m/z 151 (M ), 132 (M ±F), 120 (M ±CF), 106
‡
‡
‡
‡
(M ±NCF), 93 (M ±NCF±CH), 87 (M ±NCF±F), 82 (M ±
‡
CF±F±F), 75 (M ±NCF±CF).
2.4. Preparation of 2,4,5,6-tetrafluoronicotinic acid 6
(i) Using butyllithium and gaseous carbon dioxide. To
2,3,4,6-tetra¯uoropyridine (1.00 g, 6.62 mmol) and dry
ether (100 cm3) cooled to 788C was added butyllithium
in hexane (1.6M, 4.20 cm3, 6.72 mmol) over 20 min so that
the internal temperature did not rise above 658C. The

48
P.L. Coe, A.J. Rees / Journal of Fluorine Chemistry 101 (2000) 45±60
2.6. Preparation of ethyl tetrafluoroisonicotinate 25
NCF), 101 ((CH3)2SiCCF) and (ii) 2-butyl-3,4,6-tri¯uoro-5-
trimethylsilylpyridine 18 (0.18 g) (n.c.) b.p. 95±988C/
20 mmHg; (Found: C, 55.9; H, 6.8% C12H18F3NSi requires
C, 55.2; H, 6.9%); ꢀH (acetone d6) 0.40 (s, 9H, Si(CH3)3),
2,3,5,6-Tetra¯uoropyridine (5.00 g, 33.11 mmol) and dry
THF (60 cm3) were cooled to 788 C and butyllithium
(1.6M, 20.6 cm3, 32.96 mmol) in hexane was added over
30 min and the mixture stirred for a further 30 min. Freshly
distilled ethyl chloroformate (4.00 g, 36.87 mmol) in ether
(5 cm3) was added to the red solution at 788C and the
reaction stirred at this temperature for 1 h. The solution was
allowed to warm to room temperature and then stirred for
18 h. Water (5 cm3), hydrochloric acid (2M, 60 cm3) and
ether (60 cm3) were added successively. The organic layer
was separated, washed with brine (30 cm3), dried (MgSO4),
®ltered and most of the volatile components removed on a
rotary evaporator to leave a clear liquid. This was then
carefully distilled at room temperature and then residue re-
distilled under vacuum to give ethyl tetra¯uoroisonicotinate
25 (4.64 g, 20.80 mmol, 63%) (n.c.) b.p. 67±698C/9 mmHg;
(Found: C, 42.8; H, 2.18% C8H5F4NO2 requires C, 43.1; H,
2.3%); ꢀH (acetone d6) 1.40 (t, 3H3JHH 0.7 Hz, 3H,
3
0.95 (t, JHH 7.5 Hz, 3H, CH2CH2CH2CH3), 1.40 (sextet,
3
3
2H, JHH 7.5, CH2CH2CH2CH3), 1.69 (quintet, 2H, JHH
3
7.5, CH2CH2CH2CH3), 2.75 (dt, 2H, JHH 7.5, JHF 2.5,
CH2CH2CH2CH3); ꢀF (acetone d6) 56.0 (m, 1F, 6-F),
3
110.7 (t, 1F, JFF ˆ 4, JFF 18.3, 4-F), 157.0 (m, 1F,
‡
‡
‡
3-F); m/z 261 (M ), 260 (M ±H), 246 (M ±CH3), 232
(M ±CH3CH2), 219 (M ±CH3CH2CH2), 218 (M ±
‡
‡
‡
‡
‡
CH3CH2±CH3), 204 (M ±CH3CH2CH2CH2), 203 (M ±
‡
CH3CH2CH2±CH3), 177 (M ±CH3CH2CH2±CH3±C), 163
‡
(M ±CH3CH2CH2±CH3±CN).
(ii) Using LDA. 2,3,4,6-Tetra¯uoropyridine (0.2 cm3,
0.30 g, 1.99 mmol) was added at
788 C to LDA
(2.22 mmol in ether 15 cm3) and the mixture stirred for
40 min. Trimethylsilyl chloride (0.26 cm3, 0.22 g,
2.03 mmol) was added and the reaction stirred for 4 h at
708 C. The solution was warmed to room temperature and
stirred for 30 min. Water (30 cm3) and ether (30 cm3) were
added, the ether layer was separated, washed with brine
(30 cm3), dried (MgSO4), ®ltered and the solvent removed
to leave a clear liquid (0.18 g). Two components were
present in a ratio of 65 : 35. These were identi®ed as
2,3,4,6-tetra¯uoro-5-trimethylsilylpyridine 17 (65%) and
2-(N, N-diisopropylamino)-3,4,6-tri¯uoro-5-trimethylsilyl-
pyridine 17a (35%). Separation by column chromatography
(silica, hexane/ethyl acetate) gave (i) 17 and (ii) 2-(N,N-
diisopropylamino)-3,4,6-tri¯uoro-5-trimethylsilylpyridine
17a; ꢀF (acetone d6) 56.9 (bs, 1F, 6-F), 112.9 (m, 1F, 4-
F), 163.8 (m, 1F, 3-F).
3
CH2CH3), 4.52 (q, 2H, JHH 0.7, CH2CH3); ꢀF (acetone
d6) 90.6 (m, 2F, 2-F, 6-F), 141.7 (m, 2F, 3-F, 5-F); m/z
‡
‡
‡
223 (M ), 195 (M ±CH2CH2), 178 (M ±NCF or EtO), 150
‡
‡
‡
(M ±CO2Et), 131 (M ±CO2Et±F), 105 (M ±CO2Et±NCF).
2.7. Preparation of 2,3,4,6-tetrafluoro-5-
trimethylsilylpyridine 17
(i) Using butyllithium. To dry ether (12 cm3) and 2,3,4,6-
tetra¯uoropyridine (0.6 cm3, 0.90 g, 5.96 mmol), cooled to
below 708C butyllithium in hexanes (1.6M, 3.73 cm3,
5.97 mmol) was added over 30 min so that the internal
temperature did not rise above 558C. The reaction was
stirred for a further 40 min at 708C. An excess of tri-
methylsilyl chloride (0.90 cm3, 0.77 g, 7.10 mmol) was
added and the reaction stirred for 2 h at 708C. The solution
warmed to room temperature and was stirred a further
30 min. Water (30 cm3) and ether (30 cm3) were added,
the ether layer separated, washed with brine (30 cm3), dried
(MgSO4), ®ltered and the solvent removed on a rotary
evaporator to leave a clear liquid (0.93 g). By 19F NMR
analysis, two components were present in a ratio of 70 : 30.
With the help of GCMS analysis they were identi®ed as
2,3,4,6-tetra¯uoro-5-trimethylsilylpyridine 17 and 2-butyl-
3,4,6-tri¯uoro-5-trimethylsilylpyridine 18. The mixture
was separated by column chromatography (silica, hexane/
ethyl acetate 4 : 1) to give (i) 2,3,4,6-tetra¯uoro-5-
trimethylsilylpyridine 17 (0.52 g) (n.c.) b.p. 71±738C/
20 mmHg; (Found: C, 42.8; H, 4.3%,C8H9F4NSi requires
C, 43.0; H, 4.1%); ꢀH (acetone d6) 0.1 (s, 9H, Si(CH3)3); ꢀF
(acetone d6) 57.1 (bs, 1F, 6-F), 87.1 (bm, 1F, 2-F),
104.9 (m, 1F, 4-F), 170.8 (q, 1F, JFF 21.4, 3-F); ꢀC
(acetone d6) 2.0 (s, CH3), 108.1 (m, C-5), 133.7 (dm, 1JFC
2.8. Preparation of 2,3,4,6-tetrafluoro-5-
tributylstannylpyridine 19
Butyllithium (1.6M, 1.24 cm3, 1.98 mmol) in hexanes
was added to 2,3,4,6-tetra¯uoropyridine (0.2 cm3, 0.30 g,
1.99 mmol) in dry ether (12 cm3) at 788C over 20 min at
such a rate that the internal temperature did not rise above
658C. The reaction was stirred for a further 90 min at
708C when tributylstannyl chloride (0.54 cm3, 0.65 g,
2.00 mmol) was added and the reaction stirred for 2 h at
708C. The solution warmed to room temperature and was
stirred for 30 min, water (30 cm3) and ether (30 cm3) were
then added. The ether layer was separated, washed with
brine (30 cm3), dried over MgSO4, ®ltered and the solvent
removed to leave a clear liquid. By 19F NMR analysis, only
one component was present and was identi®ed as 2,3,4,6-
tetra¯uoro-5-tributylstannylpyridine (0.64 g, 73.3%) (n.c.)
b.p. 1258C/20 mmHg; (Found: C, 46.2; H, 6.3%
C17H27F4NSn requires C, 46.4; H 6.2%); ꢀH 0.880 (t, 9H,
3
3JHH 7, CH3), 1.04 (t, 6H, JHH 8, CH2), 1.33 (sextet, 6H,
1
1
3
256.6, 3-F), 151.6 (dm, JFC 241.6, 2-C), 151.8 (dm, JFC
‡
3JHH 7, CH2), 1.52 (q, 6H, JHH 7, CH2); ꢀF (acetone d6)
55.3 (m, 1F, 6-F), 89.0 (m, 1F, 2-F), 100.2 (m, 1F, 4-F),
1
238.1, 6-C), 158.8 (dm, JFC 240.3, 4-C); m/z 223 (M ),
‡
‡
‡
‡
208 (M ±CH3), 148 (M ±(CH3)2±NCF), 133 (M ±(CH3)3±
170.8 (m, 3-F); m/z 440 (M ), 383, 326, 269.

P.L. Coe, A.J. Rees / Journal of Fluorine Chemistry 101 (2000) 45±60
49
2.9. Preparation of 2,3,4,6-tetrafluoro-5-iodopyridine 20
hydroxynicotinaldehyde 16 (0.28 g) (n.c.) b.p. 908C/
2 mmHg; (Found: C, 40.4; H, 1.3% C6H2F3NO2 requires
C, 40.7; H, 1.1%); ꢀH (acetone d6) 10.20 (d, JH,2-F 1.8 Hz,
CHO); ꢀF (acetone d6) 75.2 (bm, 1F, 2-F), 80.4 (bm, 1F,
Butyllithium in hexanes (1.6M, 1.24 cm3, 1.98 mmol)
was added over 10 min to 2,3,4,6-tetra¯uoropyridine
(0.2 cm3, 0.30 g, 1.99 mmol) in dry ether (20 cm3) at
788C so that the internal temperature did not rise above
708C. The reaction was stirred for a further 30 min at
788C. Iodine (0.75 g, 2.95 mmol) was added and the
reaction stirred for 30 min at 788C. The solution allowed
to warm to room temperature over 1 h and stirred for a
further 1 h when water (5 cm3) was added. Aqueous sodium
thiosulphate was added, the ether layer was separated,
washed with sodium bicarbonate (20 cm3), brine (20 cm3)
and dried (MgSO4). The solution was ®ltered and most of
the ether removed, distillation using a Kugelrohr distillation
apparatus to give 2,3,4,6-tetra¯uoro-5-iodopyridine 20
(0.29 g) (n.c.) b.p. 143±1458C; (Found: C, 21.4; N, 4.9%
C5F4IN requires C, 21.7; N, 5.1%); ꢀF (acetone d6), 58.9
‡
‡
6-F), 170.0 (1F, m, 1F, 5-F); m/z 177 (M ), 176 (M ±H),
‡
‡
‡
148 (M ±CHO), 131 (M ±CHO±OH), 120 (M ±CHO±
‡
CO), 119 (M ±CHO±HO±C).
2.11. Attempted preparation (and subsequent oxidation) of
tetrafluoro-5-nitrosopyridine
Nitrosyl chloride (pre-prepared from sodium nitrite and
conc. HCl) was bubbled through tetra¯uoro-3-lithiopyridine
prepared as above from 7 (0.30 g) at 788C. The reaction
mixture was slowly warmed to 608C and kept at this
temperature for 2 h. The solution was allowed to warm to
room temperature over 2 h during the whole of this time a
constant supply of nitrosyl chloride being passed through
the solution. The reaction was stirred at room temperature
for a further 2 h and then dry air was drawn through the
mixture for several hours. Immediately after this oxidation
was stopped 19F NMR analysis of the mixture was carried
out which showed that there were three types of aromatic
ring in the mixture, from either two or three compounds.
The peaks were at 71.7 (bm), 72.7 (bm), 75.7 (bm),
78.4 (bm), 80.0 (bm), 80.7 (bm), 114.9 (bm),
119.1 (m), 163.0 (m), 163.1 (m), 166.1 (m). In
the light of later results, shown below, the products are
suggested to be bis-(2,4,5,6-tetra¯uoro-3-pyridyl)amine 21,
bis-(2,4,5,6-tetra¯uoro-3-pyridyl)hydroxylamine 22 and
bis-(2,4,5,6-tetra¯uoro-3-pyridyl)nitroxide 23.
(bm, 1F, 6-F), 88.0 (bm, 1F, 2-F), 96.6 (dt, 1F,
3
4JFF ˆ 3JFF 21.4, JFF 12.2, 4-F),
165.3 (q, 1F,
5JFF ˆ 3JFF ˆ JFF 21, 3-F); ꢀC (acetone d6) 62.9 (dddd,
2JCF, 49.3, 2JCF 26.8, 3JCF 7.5, 4JCF 4.3, 5-C), 133.3 (dddd,
1JCF 259.5, 2JCF 29.0, 2JCF 16.1, 4JCF 8, 3-C), 150.8 (dddd,
1JCF 241.0, 2JCF 17.2, 3JCF 13.9, 4JCF 7.5, 2-C), 153.3 (dddd,
1JCF 234.9, 3JCF 15.0, 3JCF 9.6, 4JCF 3.2, 6-C), 161.1 (dddd,
3
2
3
3
1JCF 257.4, JCF 10.7, JCF 8.6, JCF 6.4, 4-C); m/z 277
‡
‡
‡
‡
‡
‡
(M ), 150 (M ±I), 127 (I ), 105 (M ±I±NCF), 100 (M ±I±
‡
‡
CF±F), 93 (M ±CI±NCF), 86 (M ±I±F±NCF), 74 (M ±I±
NCF±CF); Accurate m/z 276.9013262 (calc. for C5F4IN
276.901164).
2.10. Preparation of tetrafluoronicotinaldehyde 15 and
2,5,6-trifluoro-4-hydroxynicotinaldehyde 16
2.12. Attempted preparation (and subsequent amination)
of tetrafluoro-5-nitrosopyridine
2,3,4,6-Tetra¯uoropyridine (0.2 cm3, 0.30 g, 1.99 mmol)
in dry ether (20 cm3) was cooled to below 788C, butyl-
lithium in hexane (1.6M, 1.24 cm3, 1.98 mmol) was added
over 10 min so that the internal temperature did not rise
above 758C. The reaction was stirred for a further 30 min
Tetra¯uoro-3-lithiopyridine was reacted as above with
nitrosyl chloride but instead of the oxidation step 0.880
ammonia was added to the reaction and the solution turned
black immediately. The reaction was stirred for a further
18 h before being poured into concentrated hydrochloric
acid (60 cm3). Ether (30 cm3) was added, the ether layer
separated, dried (MgSO4), and after ®ltration the solvents
were removed by rotary evaporation. The residue (0.15 g)
was a black semi-solid. 19F analysis of this showed the
presence of two compounds which were readily separated
by column chromatography (silica, hexane/ethyl acetate
4 : 1) as (i) bis-(2,4,5,6-tetra¯uoro-3-pyridyl)amine 21
and (ii) N-(40-amino-20,50,60-tri¯uoro-30-pyridyl)-3-amino-
tetra¯uoropyridine 24 (see below for characterisation) in
about equal amounts.
at
788C. N-Methylformanilide (0.25 cm3, 0.27 g,
2.000 mmol) was added and the reaction stirred for 2 h at
788C when the solution allowed to warm to room tem-
perature over 2 h. The reaction mixture was poured into
saturated ammonium chloride solution, the ether layer was
separated, combined with extracts of the aqueous layer
(2 Â 25 cm3), dried (MgSO4). The ether was removed by
rotary evaporation to leave a slightly coloured liquid
(0.39 g), which crystallised on standing. Recrystallisation
from hexane/benzene 10 : 1 afforded tetra¯uoronicotinald-
hyde 15 (0.3 g) m.p. 59±618C (cited [16] 59±608C); ꢀH
(acetone d6) 10.18 (t, 4JHF ˆ 4JHF 1.1 Hz, CHO); ꢀF (acet-
one d6) 71.2 (bm, 1F, 2-F), 76.9 (bm, 1F, 6-F), 118.4
2.13. Attempted preparation (and subsequent reduction) of
tetrafluoro-5-nitrosopyridine
‡
‡
(m, 1F, 4-F), 166.8 (m, 1F, 5-F); m/z 179 (M ), 178 (M ±
‡
‡
H), 150 (M ±CHO), 100 (M ±CHO±CF±F).
In a repeat of this experiment but using a non-acidic
work-up we obtained, as the sole product, 2,5,6-tri¯uoro-4-
Tetra¯uoro-3-lithiopyridine (from 7 1.8 g) was reacted as
above with nitrosyl chloride. At the end of the addition of

50
P.L. Coe, A.J. Rees / Journal of Fluorine Chemistry 101 (2000) 45±60
the nitrosyl chloride the solvents were removed by rotary
evaporation, water (30 cm3) and zinc powder (8.50 g) were
then added followed by the dropwise addition of concen-
trated hydrochloric acid (60 cm3). The ®nal colour of the
solution (after all the zinc had been used up) was pale
yellow. The ether layer was separated, washed with water
(30 cm3) and dried (MgSO4). After ®ltration most of the
solvent was removed on a rotary evaporator to leave a
residue (2 g) which was separated by column chromatogra-
phy (silica, dichloromethane) to give bis-(2,4,5,6-tetra-
¯uoro-3-pyridyl)amine 21 (1.20 g, 63%) (n.c.) m.p. 86±
87oC as a yellow solid; (Found: C, 38.3; H, 0.28%
C10HF8N3 requires C, 38.1; H, 0.32%); ꢀF (acetone d6)
80.1 (bm, 1F 6-F), 93.2 (bm, 1F 2-F), 127.1 (m, 1F, 4-
F), 165.9 (m, 1F, 5-F); ꢀC (acetone d6) 115.0 (m, 3-C),
C, 37.71; H, 2.11; N, 21.99%); ꢀH (acetone d6) 6.05 (bs, 2H,
NH2), 7.50 and 7.75 (broad s, 2H, CONH2 (protons are not
equivalent)); ꢀF (acetone d6) 94.3 (dd, 1F, 3JFF 24.4, 5JFF
30.5, 6-F), 144.2 (dd, 1F, 4JFF 6.1, 5JFF 33.6, 3-F), 160.3
3
4
(dd, 1F, JFF 27.5, JFF 6.1, 5-F); ꢀC (acetone d6) 126.7 (t,
2JFC ˆ 2JFC 19.5, 4-C), 131.9 (dd, 1JFC 247.2, 2JFC 32.5, 5-
1
3
C), 140.0 (dd, JFC 252.2, JFC 4.9, 3-C), 143.9 (t,
1
3JFC ˆ 2JFC 16.2 H, 2-C), 145.8 (dd, JFC 229.0 Hz, JFC
2
‡
13.8, C-2), 160.8 (s, CONH2); m/z 191 (M‡), 148 (M ±
‡
‡
CONH), 121 (M ±CONH±CNH), 101 (M ±CONH±
CNH2±F), 75 (M ±CONH±CNH2±NCF).
‡
Two other compounds were obtained in small amounts
but could not be fully characterised. On the basis of their 19F
NMR spectra we believe these compounds to be ethyl-2-
amino-3,5,6-tri¯uoroisonicotinate 28 ꢀF (acetone d6) 94.3
1
1
4
5
133.6 (dm, JCF 258.3, 5-C), 144.1 (dm, JCF 239, 6-C),
1
(m, 1F, 6-F), 144.3 (dd, 1F, JFF 6.1, JFF 30.5, 3-F),
1
3
4
147.5 (dm, JCF 241.0, 2-C), 153.1 (dm, JFF 260.5, 4-C);
‡
162.0 (dd, 1F, JFF 24.4, JFF, 6.1, 5-F) and 2,3,5,6-
tetra¯uoroisonicotinamide 29 ꢀF (acetone d6) 90.4 (m,
1F, 2-F), 141.7 (m, 1F, 3-F).
‡
‡
‡
m/z 315 (M ), 296 (M ±F), 295 (M ±HF), 277 (M ±F±F),
‡
‡
‡
276 (M ±HF±F), 269 (M ±NCF±H), 250 (M ±NCF±HF),
‡
‡
‡
245 (M ±CF±HF±F), 231 (M ±NCF±HF±F), 119 (M ±
‡
C5F4N±NH±CF), 100 (M ±C5F4N±NH±CF±F).
2.16. Amination of tetrafluoronicotinic acid
2.14. Reaction of 2,4,5,6-tetrafluoro-3-pyridyllithium with
nitrosonium tetrafluoroborate
2,4,5,6-Tetra¯uoronicotinic acid (2.56 g, 13.13 mmol)
and 0.880 ammonia (30 cm3) were stirred together for
15 h. Ether (60 cm3) was added followed by 5M hydro-
chloric acid until the aqueous layer was acidic. The ether
layer was separated, washed with saturated sodium bicar-
bonate solution (40 cm3), the basic layer removed and re-
acidi®ed and ether (40 cm3) was added. The ether was
separated, washed with water (25 cm3), brine (25 cm3)
and dried (MgSO4). The solvent was removed on a rotary
evaporator to leave 4-amino-2,5,6-tri¯uoronicotinic acid 26
(1.87 g, 71.9 %) (n.c.), m.p. 216±2188C; (Found: C, 37.4; H,
1.82; N 14.9%. C6H3F3N2O2 requires C, 37.52; H, 1.57; N,
14.6 %). ꢀF (acetone d6) (at 400 MHz) 60.5 (dd, 1F, 5JFF
23.4, 4JFF 8.3, 2-F), 90.6 (dd, 1F, 3JFF 20.8, 4JFF 9.9, 6-F),
170.8 (dd, 1F, 5JFF 23.7, 3JFF 20.9, 2-F); ꢀC (acetone d6)
94.8 (dm, 2JFC 30.8, 3-C), 131.5 (ddd, 1JFC 243.3, 2JFC 27.4,
5JFC 5.8, 5-C), 150.2 (ddd, 1JFC 237.3, 2JFC 21.1, 3JFC 13.1,
6-C), 152.5 (m, 4-C), 157.3 (dm, 1JFC 247.6, C-2), 166.8 (d,
Tetra¯uoro-3-lithiopyridine was prepared as above. The
reaction was stirred for a further 30 min at 78 8C when
nitrosonium tetra¯uoroborate (0.94 g, 8.034 mmol) was
added and the reaction stirred for a further 2 h at 788C.
The solution was warmed to room temperature and the
reaction stirred for 30 min. Hydrochloric acid (2M,
20 cm3) was then slowly added. The ether layer was sepa-
rated, dried (MgSO4) and the solvent removed on a rotary
evaporator to leave a yellow liquid (0.40 g) which crystal-
lised on standing. The 19F NMR spectrum of the crude
product showed it to be very largely 21.
2.15. Amination of ethyl tetrafluoroisonicotinic acid
Ethyl2,3,5,6-tetra¯uoroisonicotinate(3.12 g,13.99 mmol)
was dissolved in ether (200 cm3) in a round bottomed ¯ask
and ammonia gas was bubbled through for 6 h. After 1 h the
solution became yellow but even after 6 h no precipitate of
ammonium ¯uoride was seen. 19F NMR analysis showed
that only ethyl 2,3,5,6-tetra¯uoroisonicotinate remained.
The ether was mostly distilled off using an Oldershaw
column and the residue from this experiment and 0.880
ammonia (20 cm3) were stirred at room temperature for
18 h. The solution was acidi®ed and ether (40 cm3) was
added. The ether layer was washed with water (25 cm3),
brine (25 cm3), dried (MgSO4) and the solvent removed on a
rotary evaporator to yield a yellow solid (2.6 g). Column
chromatography (hexane-ethyl acetate, 1 : 2) on a small
sample (0.5 g) of the product gave (i) 2-amino-3,5,6-tri-
¯uoroisonicotinamide 27 (0.38 g) (n.c.) m.p. 216±218oC;
(Found: C, 37.31; H, 2.36; N 21.60. C6H3F3N2O2 requires
‡
‡
3JFC 7.6, CO2H); m/z 192 (M ), 174 (M ±OH±H), 147
‡
‡
(M ±CO2H or M ±NCF).
2.17. Preparation of 4-aminotrifluoronicotinic acid from
2,3,4,6-tetrafluoropyridinein a ``one pot'' reaction
Tetra¯uoro-3-lithiopyridine from 7 (6 g) as above in ether
(100 cm3) was stirred for 40 min at 788C and then allowed
to warm to 558C, carbon dioxide (dried by passing down a
calcium chloride tube) was bubbled into the solution for 3 h
when all the solvent had evaporated. 0.880. Ammonia
(60 cm3) was then carefully added, the mixture was stirred
for 15 h, 6M hydrochloric acid was then very carefully
added until the solution was acidic. A white solid precipi-
tated as more hydrochloric acid was added. The solution
was left to cool for 30 min and ether (80 cm3) was added.

P.L. Coe, A.J. Rees / Journal of Fluorine Chemistry 101 (2000) 45±60
51
The ether layer was separated and washed with saturated
sodium bicarbonate solution (40 cm3), the basic aqueous
layer was removed and re-acidi®ed, ether (80 cm3) was
added. The ether layer was separated, washed with water
(40 cm3), brine (40 cm3) and dried (MgSO4). The solvent
was removed on a rotary evaporator to leave 4-amino-2,5,6-
tri¯uoronicotinic acid 31 (6.30 g, 32.81 mmol, 82.6%).
38.1; H 1.1; N 22.2%); ꢀH (DMSO) 11.8 (s, D2O exchange-
able, N±H), 12.4 (s, D2O exchangeable, N±H); ꢀF (acetone
d6) 92.0 (dd, 1F, JFF 27.5, JFF 14.2, 6-F), 99.0 (dd, 1F,
JFF 22.6, JFF 14.2, 2-F), 166.6 (dd, 1F, JFF 27.5, JFF 22.6,
3-F); ꢀC (acetone d6) 112.2 (dm, 2JFC 34.5, 5-C), 128.6 (ddd,
1JFC 249.2, 2JFC 32.6, 5JFC 7.1, 3-C), 130.0 (m, 4-C), 137.9
1
2
(dd, JFC 231.7, JFC, 15.0, coupling between 2-C and 6-F
must be too small to be seen), 147.6 (ddd, 1JFC 231.0, 3JFC
‡
14.5, 4JFC 14.5 Hz, C-6), 155.2 (s, 8-C); m/z 189 (M ), 170
2.18. Amination of bis-(2,4,5,6-tetrafluoro-3-
pyridyl)amine
‡
‡
‡
‡
(M ±F), 161 (M ±CO), 142 (M ±FCO), 134 (M ±CO±
‡
CHN), 107 (M ±CO±CHN±CHN
To bis-(2,4,5,6-tetra¯uoro-3-pyridyl)amine (1.00 g,
3.175 mmol) was added 880 ammonia (10 cm3). The reac-
tion mixture, which went black immediately, was then
stirred for 20 h before being poured into concentrated
hydrochloric acid (200 cm3). Ether (50 cm3) was added,
the ether layer separated, dried (MgSO4) and the solvents
removed by rotary evaporation. The residue (0.80 g) was a
black semi-solid. A sample of this solid (0.5 g) was sepa-
rated by column chromatography (silica, hexane/ethylace-
tate 4 : 1) to give (i) unreacted starting material 0.2 g) and
(ii) N-(2,4,5,6-tetra¯uoro-3-pyridyl)-N-(4-amino-2,5,6-tri-
¯uoro-3-pyridyl)amine 29 (0.23 g) (n.c.) m.p. 106±108oC;
(Found: C, 38.2; H, 0.8; N, 17.8% C4H3F7N4 requires C,
38.5; H, 0.97; N, 17.9%); ꢀF 83.5 (bm, 20-F), 83.6 (bm,
2-F), 95.1 (bq, 60-F), 99.3 (bm, 6-F), 133.1 (m, 4-F),
167.1(q, 3-F) and 168.2 (t, 30-F) ppm.
2.21. Reaction of 2,4,5,6-tetrafluoro-3-pyridyllithium with
benzaldehyde
Tetra¯uoro-3-lithiopyridine (from 7 0.30 g) in dry ether
(20 cm3) was reacted at 788C with freshly distilled ben-
zaldehyde (0.21 cm3, 0.22 g, 2.07 mmol) and the reaction
stirred for 30 min at 788C. The solution was allowed to
warm to room temperature over one hour and was then
stirred at room temperature for a further 1 h. Sulphuric acid
(4M, 30 cm3) was added the ether layer was separated,
washed with water (20 cm3), brine (20 cm3) and dried
(MgSO4) and the ether removed by rotary evaporation to
leave a yellow oil which was distilled in vacuo using a
Kugelrohr apparatus to give phenyl-(2,4,5,6-tetra¯uoro-3-
pyridyl)methanol 33 (0.36 g, 1.40 mmol, 70.7%) (n.c.) b.p.
120±125oC/0.1 mmHg; (Found: C, 56.1; H, 2.6% C12H7F4N
requires C, 56.0; H, 2.7%), ꢀH (acetone d6) 6.2 (1H, s, CH±
OH), 7.1±7.5 (5H, m, Ar); ꢀF (acetone d6) 71.4 (bm, 1F, 2-
F), 87.6 (bm, 1F, 6-F), 117.0 (m, 1F, 4-F), 168.0 (m,
1F, 3-F); ꢀC (acetone d6) 66.6 (s, CH±OH), 126.3 (s, 20-C,
60-C), 128.3 (s, 40-C), 129.1 (s, 30-C, 50-C), 142.8 (s, 10-C)
(signals for 2-C, 3-C, 4-C, 5-C and 6-C are all too small to be
2.19. Attempted direct nitration of 2,3,4,6-
tetrafluoropyridine
Fuming (90%) nitric acid (0.93 cm3, 13.26 mmol) was
added slowly to sulfolane (4 cm3) at 08C, boron tri¯uoride/
dihydrate (0.84 cm3, 1.38 g, 13.27 mmol) was then added
dropwise. 2,3,4,6-Tetra¯uoropyridine (1.00 g, 6.62 mmol)
was added and the stirred mixture heated to, and maintained
at 758C for 2 h. After cooling, the mixture was poured onto
crushed ice and the organic material was extracted with
dichloromethane. 19F NMR analysis of this dichloro-
methane layer showed only the presence of the starting
material.
‡
interpretable due to extensive C-F coupling); m/z 256 (M ±
‡
‡
‡
H), 236 (M ±H±HF), 180 (M ±Ph), 164 (M ±Ph±HF), 150
‡
‡
‡
(C5F4N ), 149 (M ±Ph±CF), 130 (M ±Ph±NCF), 107
‡
‡
(M ±C5F4N), 77 (Ph ).
2.22. Preparation of 1,1-bis-(2,4,5,6-tetrafluoro-3-
pyridyl)ethyl acetate 36
2.20. Curtius reactions of 4-amino-2,5,6-trifluoronicotinic
acid using diphenyl phosphorazidate (DPPA)
Acetyl chloride (0.28 cm3, 0.31 g, 3.974 mmol) was
added at ±78oC to tetra¯uoro-3-lithiopyridine prepared as
above from 7 (0.6 g) and the reaction stirred for 30 min at
708C, the solution was then allowed to warm to room
temperature over 1 h, at about 208C the solution became
cloudy, and the reaction was stirred at room temperature for
a further hour. 4M Hydrochloric acid (30 cm3) was added,
the ether layer separated, washed with sodium bicarbonate
(20 cm3), brine (20 cm3) and dried (MgSO4). The ether was
removed by rotary evaporation to leave a pale yellow
viscous liquid shown after distillation in vacuo to be 1,1-
bis-(2,4,5,6-tetra¯uoro-3-pyridyl)ethyl acetate 36 (0.62 g,
1.606 mmol, 80.8%) (n.c.) b.p. 1758C 4.5 mmHg; (Found:
C, 43.43; H, 1.62; N, 7.04%. C14H6F8N2O2 requires C,
4-Aminotri¯uoronicotinic acid 26 (1.00 g, 5.21 mmol) in
t-butanol (20 cm3), triethylamine (0.3 cm3, 0.53 g,
5.24 mmol) and DPPA (1.12 cm3, 1.43 g, 5.20 mmol) were
stirred together under re¯ux for 4 h. The solvents were
evaporated, the residue dissolved in ether (30 cm3) the
solution was washed with water (25 cm3) and then brine
(25 cm3). The ether layer was dried (MgSO4) and evapo-
rated to give a yellow residue. Column chromatography
(silica, hexane-ethyl acetate, 1 : 1) gave 3-deaza-2,3,6-tri-
¯uoro-8-hydroxypurine (0.17 g, 19.3%) m.p. 239±2418C.
(Found: C, 37.9; H, 0.9; N, 21.9% C6H2F3N3O requires C,

52
P.L. Coe, A.J. Rees / Journal of Fluorine Chemistry 101 (2000) 45±60
43.54; H, 1.57; N, 7.25%); ꢀH (acetone d6) 2.14 (s, 3H,
OCOCH3), 2.36 (q, 3H, JHF 2.0, CCH3); ꢀF (acetone d6)
65.0 (bm, 1F, 2-F), 85.8 (bm, 1F, 6-F), 112.4 (m, 1F, 4-
F), 166.3 (m, 1F, 5-F); ꢀC (acetone d6) 21.3 (s, 9-C), 25.4
(q, JCF ˆ JCF 3.8, C-10), 78.0 (s, 7-C), 113.7 (m, 3-C), 134.4
(dddd, 1JCF 257.1, 2JCF 28.4, 2JCF 16.2, 4JCF 7, 5-C), 149.0
2.0); ꢀF (acetone d6) 64.3 (bm, 1F, 2-F), 89.7 (bm, 1F, 6-
F), 112.6 (m, 1F, 4-F), 168.2 (m, 1F, 3-F); ꢀC (acetone
d6) 30.7 (t, JCF) 3.8, CH3), 44.5 (s, C±OH), 119.4 (m, 3-C),
134.1 (dddd, 1JCF 255.7, 2JCF 28.5, 2JCF 16.9, 4JCF 8.0, 5-C),
1
1
148.1 (dm, JCF 239.2, 6-C), 152.7 (dm, JCF 247.3, 2-C),
‡
‡
159.2 (dm, 1JCF 265.8, 4-C); m/z 194 (M ±CH3), 178 (M ±
‡ ‡
1
1
(dm, JCF 177.9, 6-C), 152.2 (dm, JCF 178.6, 2-C), 158.6
‡
CH2±OH), 174 (M ±CH3±HF), 164 (M ±CH3±O±CH2) or
‡ ‡ ‡
(M ±CH3±CF), 150 (C5F4N ), 146 (M ±CH3±OH±CF),
1
(dm, JCF 268.2, 4-C), 169.3 (s, C-8); m/z 386 (M ), 327
‡
‡
‡
‡
(M ±CO2Me), 326 (M ±HCO2Me via McLafferty rearran-
132 (M ±CH3±OH±NCF), 59 (Me2C±OH ). It was impos-
sible to obtain a boiling point of 34 as it decomposed on
heating even at low temperature in vacuo to the correspond-
ing alkene.
‡
‡
gement), 307 (M ±HCO2Me±F), 287 (M ±HCO2Me±F±
HF), 178 (C5F4NCO), 150 (C5F4N); Accurate m/z
386.031304 (calc. for C14H6F8N2O2 386.030154).
2.23. Preparation of 1,1-bis-(2,4,5,6-tetrafluoro-3-
pyridyl)-2,2,2-trifluoroethanol 37
2.25. Dehydration of 2-(2,3,4,6-tetrafluoropyridyl)
propan-2-ol 35
Tetra¯uoro-3-lithiopyridine from 7 (0.3 g) was stirred for
30 min at 708C when tri¯uoroacetic anhydride (0.28 cm3,
0.42 g, 2.000 mmol) was added and the reaction stirred for
30 min at 708C. The solution was warmed to room
temperature over 1 h and stirred for a further 1 h. Sulphuric
acid (4M, 30 cm3) was added, the ether layer was separated
washed with water (40 cm3), sodium bicarbonate (20 cm3)
and dried (MgSO4). The ether was removed by rotary
evaporation to leave a clear liquid which after distillation
afforded 1,1-bis-(2,4,5,6-tetra¯uoro-3-pyridyl)-2,2,2-tri-
¯uoroethanol 37(0.45 g) (n.c.) b.p. 156oC/2 mmHg; (Found:
C, 36.3; H, 0.4% C12HF11N2 requires C, 36.2; H, 0.3%); ꢀF
(acetone d6) 62.9 (bm, 2F, 2-F), 77.8 (q, 3F, JFF and JFF
15.3, CF3), 83.4 (bm, 2F, 6-F), 109.8 (m, 2F, 4-F),
Alcohol 34 (0.3 g) was heated at 180oC with phosphoric
oxide (0.5 g) in a small distillation apparatus. The product
which distilled off at ca. 160oC was re-distilled to yield 2-
(2,3,4,6-tetra¯uoropyridyl) prop-2-ene 35 (0.2 g) (n.c.) b.p.
160±165oC; (Found: C, 50.2; H, 2.52% C8H5F4N requires
C, 50.27; H, 2.64%), ꢀH (acetone d6) 2.10 (dd, 3H, 3JHH 1.5,
3JHH 1.0, CH3), 5.28 (distorted, 1H, ˆCH2), 5.58 (q, 1H,
3JHH ˆ 1JHH, 1.5, ˆCH2); ꢀF (acetone d6) 71.5 (bm, 1F, 6-
F), 88.7 (bm, 1F, 2-F), 117.2 (m, 1F, 4-F), 168.3 (m,
1F, 5-F); ꢀC (acetone d6) 22.5 (s, CH3), 122.2 (s, ˆCH2), (s,
C±CH3), 112 (m, 3-C), 131.8 (s, ˆC(Ar)Me), 135 ((d)m, 5-
C), 149 ((d)m, 6-C), 153 ((d)m, 6-C), 158 ((d)m, 4-C); m/z
‡
‡
‡
‡
191 (M ), 190 (M ±H), 176 (M ±CH3), 172 (M ±F), 164
‡
‡
‡
(M ±C2H3), 156 (M ±CH3±HF), 106 (M ±C3H4±NCF).
2
165.5 (m, 2F, 5-F); ꢀC (acetone d6) 75.6 (q, JCF 36.6,
1
C±OH), 109.8 (m, 3-C), 124.4 (q, JCF 285.8, CF3), 134.2
(dddd, 1JCF 258.2, 2JCF 28.2, 2JCF 15.9, 4JCF 7.9, 5-C), 150.4
3. Results and discussion
1
1
(dm, JCF 244.5, 6-C), 152.2 (dm, JCF 245.0, 2-C), 159.3
‡
‡
1
It seemed clear from the above discussion that any
desirable route to 3-chlorotetra¯uoropyridine 8 should be
solvent based and take place at a reasonable temperature.
Since the early work of Banks and Chambers there have
been a number of developments in so-called ``Halex''
chemistry and particularly in the use of quaternary salts
as catalysts. The area has been subject to a number of
reviews [8]. We chose ®rst to investigate the use of tetra-
phenylphosphonium bromide 11 as a catalyst. This com-
pound is commercially available but is very expensive. We
have developed a cheap synthesis of 11 based on the existing
literature preparation [9] involving the metal catalysed
reaction between bromobenzene and triphenylphosphine
in benzonitrile. A number of salts worked well and nickel
bromide was found to be the best. In the course of our
work we found that the desired salt was very soluble in
ether whereas the remaining byproducts and unchanged
triphenylphosphine are very insoluble in ether thus mak-
ing puri®cation very simple. Using this modi®cation to
the procedure we were able to obtain very pure 11 in
multi-gram quantities in a matter of hours in 86% isolated
yield.
(dm, JCF 270.3, 4-C); m/z 398 (M ), 381 (M ±OH), 379
‡
‡
‡
‡
(M ±F), 362 (M ±OH±F), 359 (M ±HF±F), 329 (M ±
CF3), 309 (CF3±HF), 293 (CF3±F±OH), 281 (CF3±CF±
‡
‡
OH), 262 (CF3±CF±OH±F), 248 (M ±C5F4N) or (M ±
‡
CF3±NCF±OH±F), 236 (M ±CF3±NCF±OH±F), 231
‡
‡
(M ±CF3±CF±CF±OH±F), 178 (M ±C5F4N±CF3±H), 150
(C5F4N).
2.24. Preparation of 2-(2,4,5,6-tetrafluoro-3-pyridyl)
propan-2-ol 34
Acetone (1.00 cm3, 0.79 g, 13.62 mmol) was added to a
solution of tetra¯uoro-3-lithiopyridine from 7 (0.3 g) and
the reaction stirred for 30 min at 788C. The solution was
warmed to room temperature over 1 h and stirred for a
further 1 h. Sulphuric acid (4M, 30 cm3) was added, the
mixture separated and the ether layer washed with water
(20 cm3), brine (20 cm3) and dried (MgSO4). The ether
removed by rotary evaporation to leave a clear liquid which
was identi®ed as 2-(2,4,5,6-tetra¯uoro-3-pyridyl)-propan-
2-ol 34 (0.38 g) (n.c) (Found: C, 46.1: H, 3.2% C8H7F4NO
requiresC,45.9;H 3.4%), ꢀH (acetoned6)1.7(t, JH, (4-F and 2-F)

P.L. Coe, A.J. Rees / Journal of Fluorine Chemistry 101 (2000) 45±60
53
3,5-Dichlorotri¯uoropyridine 9 is now commercially
available in bulk quantities and we used this as our starting
material. We found that the best conditions to obtain the
desired 3-chlorotetra¯uoropyridine 8 were to react 9 with a
slight excess of spray dried potassium ¯uoride using a
catalytic quantity of the salt 11 in sulfolane at 180±
1908C and to remove product as it was formed. Using these
conditions we obtained a mixture of penta¯uoropyridine 10
(21%), the required 3-chlorotetra¯uoropyridine 8 (61%) and
unreacted starting material (13%). We have been able to
repeat this work on the kilogram scale.
or better to triethylamine/ether. We found that using one
equivalent of triethylamine gave complete conversion of the
starting material after 13 days and although the GC yield of
product was 74%, it was extremely dif®cult to isolate the
pyridine from the reaction medium without resort to pre-
parative GC.
We decided therefore to abandon this route and to con-
centrate our attention on the Chambers gas phase route [5].
In this method the starting chloro¯uoropyridines were
passed in a stream of hydrogen over 10% palladium on
carbon at 2508C and the product collected in liquid air. In
this way it was found possible to obtain 2,3,4,6-tetra¯uor-
opyridine 7 from 3-chlorotetra¯uoropyridine 8, 2,4,6-tri-
¯uoropyridine 12 from 3,5-dichlorotri¯uoropyridine 9 and
interestingly at higher temperatures 2,3,5,6-tetra¯uoropyr-
idine 13 from penta¯uoropyridine 10, in the latter case in
relatively low yield with considerable decomposition. This
latter result is in accord with the work of Pummer and Wall
[10] who showed it was possible to reduce hexa¯uoroben-
zene to penta- and tetra¯uorobenzenes under similar con-
ditions but using platinum based catalysts. We attempted to
repeat the literature process using fresh palladium on carbon
with hydrogen and 3-chlorotetra¯uoropyridine 8, without
success. In an attempt to explain the difference between our
results and the literature procedure we repeated reaction in
exactly the same manner but then continued to pass the
carrier gas for 36 h, when we found that we obtained
approximately the same yield of product as had been
reported. We believe that there was a difference in the
activation and probably the nature of the carbon supports
used in the different sets of experiments. It is known that
¯uorocompounds are strongly absorbed by active carbon, a
process that can be readily reversed by heating the carbon
strongly [15]. We next tried a reduction of 8 using commer-
cially available 0.5% palladium on alumina as catalyst. We
found that at temperatures between 2508C and 2858C we
were able to obtain, under the best conditions, a smooth
reduction of 8 to a mixture of the desired 2,3,4,6-tetra¯uor-
opyridine 7 and a little unchanged starting material. Table 1.
We next investigated the reduction of 3-chlorotetra¯uor-
opyridine 8 under a variety of conditions to attempt to ®nd a
clean large scale route to 2,3,4,6-tetra¯uoropyridine 7.
There are reports [5,10] of the reduction of chlorine to
hydrogen in polyhalobenzenes and polyhalopyridines by
hydrogen and metal catalysts and of hydrode¯uorination by
complex metal hydrides, in per¯uoroarenes and heteroar-
enes [11,12]. In the latter case the reaction is only of
synthetic value if the ¯uorine atom to be reduced is acti-
vated. In our study we were concerned with the hydrode-
chlorination of 3-chlorotetra¯uoropyridine 8 in particular.
Literature evidence [13] clearly indicates that in reactions of
poly¯uoro- and chloropoly¯uoro-pyridines with complex
metal hydrides, preferential hydrode¯uorination at the 4-
position was the most likely outcome. Thus, we turned our
attention to the use of hydrogen and metal catalysts as a
means of achieving our objective. McNamara and Cook [14]
have shown that 4-amino-3,5-dichlorodi¯uoropyridine can
be reduced to 4-amino-2,6-di¯uoropyridine by reaction with
hydrogen at 3 bar pressure with palladium on carbon as
catalyst at room temperature using methanol/acetic acid as
the solvent, albeit rather slowly, over 3 days. It therefore
seemed sensible to study the reaction of 8 under the same
conditions. This reaction resulted in complete recovery of
the starting material, and so we concluded that the amino
function was in some way contributing to the successful
reduction of 4-amino-3,5-dichlorodi¯uoropyridine. We
changed the solvent system to either neat triethylamine
Table 1
Reduction results of 8 using 0.5% palladium on fused alumina catalysta
Run
Input (g)
T (8C)
H2 flow (ml/min) Addition rate (g/min) Mass output (g)
3-H (%)
3-Cl (%)
3-H yield (%)
1
2
5
250
250
275
285
285
285
280
275
275
285
290
50
55
50
50
65
70
75
60
50
60
60
0.07
0.12
0.22
0.83
0.49
0.46
0.53
1.0
0.3
0.8
16.7
46.2
41
91
88
76
26
71
82
33
43
67
71
86
9
12
24
74
29
18
67
57
33
29
10
5
3
19.7
50
79
30
72
81
15
44
65
68
77
4
5
46.2
41
6
40
7
50
44
44
42
8
9
42
0.84
0.88
0.44
39.5
36
10
11
39.5
53
49.7
a All experiments were run on the same column (packed and heated length 2.5 Â 15 cm) containing 0.5% Pd on fused alumina. Temperature is the
average over the time taken to add substrate. Addition rate calculated as the ratio of mass added to time of addition. 3-H is 2,3,4,6 tetrafluoropyridine, 3-Cl is
3-chlorotetrafluoropyridine. % Composition calculated from GC traces using a calibration curve obtained from known mixtures.

54
P.L. Coe, A.J. Rees / Journal of Fluorine Chemistry 101 (2000) 45±60
Scheme 2.
It appears that there is only a small temperature effect
within the temperature range used. Similarly the hydrogen
¯ow rate and mass input also have only small effects. The
main effect on conversion seems to be the rate and mode
of addition of the pyridine since the yields improved if
the pyridine was added dropwise to the top of a vertical
pyrolysis tube rather than by volatilisation into the hydrogen
stream. We also found that it was possible to re-pyrolyse
the products from runs which contained higher amounts
of starting material without any over reduction. The
desired 2,3,4,6-tetra¯uoropyridine 7 was obtained in
high purity by fractional distillation of the combined
products from several pyrolyses. This phase of our study
is summarised in Scheme 2.
We next extended the study of the reactions of 2,3,4,6-
tetra¯uoropyridine 7 that we believed would lead to the 3,4-
disubstituted tetra¯uoropyridines we required for our dea-
zapurine synthesis. Two areas appeared to be promising,
®rstly, to extend the organometallic work reported pre-
viously [5], taking into account some other reported reac-
tions of tetra¯uoro-3-lithiopyridine 14 to make the 3-formyl
tetra¯uoropyridine 15 and the corresponding 3-formyl-4-
hydroxy derivative 16 [16,17]. Secondly, to probe the
nucleophilic displacement reactions of 2,3,4,6-tetra¯uoro-
pyridine 7 and compounds derived from it (see Scheme 3).
We tried ®rst to optimise the yield of tetra¯uoronicotinic
acid 6. There have been some examples reported of
quite dramatic solvent effects in the carbonation of
organometallic reagents in the ¯uoroarene series. Early
work by ourselves showed that penta¯uorophenyl Grignard
and lithium reagents derived from the bromo or iodo
compounds reacted normally with most electrophiles in
ether or ether/hexane but under these conditions would
not react with carbon dioxide [18,19]. These observations
were con®rmed by Tamborski [20] who found that it
was necessary to use ether/THF to obtain good yields
of the carboxylic acid. He also generated the lithium reagent
from penta¯uorobenzene by deprotonation with butyl
lithium and his system was then essentially salt free. In
the original work [5] to prepare tetra¯uoronicotinic acid,
hexane was used as the solvent and the acid was obtained
in 62% yield. In this work the CO2 was added to the reagent
as a gas. We repeated the literature reaction exactly as
reported and also obtained a 65% yield of the desired
acid. The reaction was then repeated on the literature scale
(3.3 mmol) but solid carbon dioxide was added, we obtained
a 75% yield but when we scaled up the reaction to 50 mmol
we obtained only a 10% yield with much polymerisation.
We have no ®rm evidence to explain this result which
was consistent over several experiments. However, we
Scheme 3.

P.L. Coe, A.J. Rees / Journal of Fluorine Chemistry 101 (2000) 45±60
55
believe it may be a temperature effect, in that addition of the
solid, CO2 maintains a low temperature in the reaction
mixture for a long period of time and the CO2 probably
does not react at 788C. When the gaseous addition is used
the reaction mixture is slowly allowed to warm to ambient
temperature with the CO2 still ¯owing. We then changed the
solvent system to either ether/hexane or THF/ether/hexane
with gaseous CO2 addition. Over a number of large-scale
(50 mmol or greater) experiments, we obtained consistently
yields in the region of 75%. This afforded suf®cient of the
acid for further study. During the course of this part of the
work we also used lithium diisopropylamide (LDA) as the
base to form the lithio derivative. Although the reaction
proceeded smoothly, the yields were generally lower and
products were isolated, which on the basis of their NMR
spectra, arose from reaction of the diisopropylamine gen-
erated in the reaction with the ®rst formed product, pre-
sumably at the now very highly active 4-position (see
below). We wished to prepare the 3-formyl-tetra¯uoropyr-
idine (tetra¯uoronicotinaldehyde) 15, which has been sug-
gested to be too reactive towards nucleophiles, particularly
in the 4-position, to be used to prepare other derivatives
[16]. Since we wished to make 3,4-disubstituted compound
it seemed logical that we should investigate this report
further. Thus, reaction of the 3-lithio-tetra¯uoropyridine
14 with N-methylformanilide in ether/hexane afforded, after
work-up under strongly acidic conditions, the known tetra-
¯uoronicotinaldehyde 15 in 57% yield as a crystalline solid.
However, work-up with only the addition of water or
aqueous ammonium chloride gave a mixture of the aldehyde
15 and 2,5,6-tri¯uoro-4-hydroxynicotinaldehyde 16. If
water was added and the mixture was heated for 30 min
prior to isolation the product was solely 16 in 67% yield.
Both structures were con®rmed by physical means. We have
thus been able to con®rm that the earlier suggestion is
correct and further to show that it is possible to use the
very high reactivity of the 4-position to advantage in order to
obtain potentially useful 3,4-disubstituted compounds. We
did not in the event carry out a similar reaction with
ammonia to get the 4-amino compound since we found
(see below) a much simpler and more reliable route. These
reactions are summarised in Scheme 3.
bromo desilylation reactions, we felt it might be possible
to similarly nitrodesilylate or nitrodestannylate appropriate
derivatives of 2,3,4,6-tetra¯uoropyridine 7. We thus carried
out the reaction of 3-lithio-tetra¯uoropyridine 14 with
chlorotrimethylsilane and in this case somewhat surpris-
ingly obtained two products which could be separated by
column chromatography to give the desired 5-trimethylsilyl
derivative 17 as the major product (70%) and 2-butyl-3,4,6-
tri¯uoro-5-trimethylsilylpyridine 18 30% as the minor pro-
duct. The structure of both products was con®rmed by
analysis of their NMR spectra, elemental analysis and mass
spectrometry (see Section 3).
This result is unusual, as we would have expected the
second substitution of the butyl group to occur in the 4-
position. We could explain the result on steric grounds in
that the vicinal arrangement of the butyl and trimethylsilyl
groups is unfavourable. We have NMR evidence that a
second product obtained when we carried out a similar
reaction but using LDA as base was in fact the analogous
2-N,N-diisopropylamino derivative but we were unable to
isolate suf®cient of it to obtain full characterisation, thus
again indicating the selective substitution para to the tri-
methylsilyl group rather than the expected 4-substitution.
An argument could also be advanced suggesting that the
empty d-orbitals on silicon can readily stabilise the negative
charge on C5 of the canonical structure. There is evidence
for such species being important in nucleophilic attack on
halosilane derivatives [23].
We next carried out a similar reaction of 3-lithio-tetra-
¯uoropyridine 14 with tributylstannyl chloride and obtained
a clear liquid as the sole product in 73% yield which was
readily identi®ed by physical means as the expected 5-
tributylstannyl-2,3,4,6-tetra¯uoropyridine 19. The absence
of the butylated moiety analogous to 18 can be readily
explained by the much greater reactivity towards nucleo-
philes of the tributylstannylchloride relative to its silicon
counterpart. Reaction of the lithio-reagent 14 with iodine
readily afforded the expected 2,3,4,6-tetra¯uoro-5-iodopyr-
idine 20 in 52% yield again identi®ed by physical means.
This compound could be useful as a reactive starting
material in copper(1) mediated reactions which could be
used to prepare our target deazapurine. These reactions are
summarised in Scheme 4.
In the light of the observed high level of reactivity
towards electrophiles of the lithio compound 14 we decided
to investigate the possibility of forming a C±N bond directly.
It has been shown that reaction of organometallic reagents
with nitrogen dioxide afforded oximes [26,27], whilst reac-
tion with nitrosyl chloride yielded nitroso compounds e.g
from phenylmagnesium bromide and nitrosyl chloride,
nitrosobenzene was obtained in 56% yield [28]. As a result
of the high reactivity of tetra¯uoronicotinaldehyde 15
towards nucleophiles which we have described above, we
were concerned about a suitable work-up procedure and
equally in view of the known properties of penta¯uoroni-
trosobenzene, we were concerned about the volatility and
An attractive possibility would be to obtain tetra¯uoro-5-
nitro-pyridine since this, almost certainly, would aminate
under very mild conditions at the 4-position to give us our
desired pyridine derivative after a simple reduction step. It
seemed very unlikely that direct nitration of 2,3,4,6-tetra-
¯uoropyridine 7 would occur and indeed we readily con-
®rmed this view by reacting 7 with fuming nitric acid and
boron tri¯uoride/hydrate in sulfolane under conditions we
used to nitrate penta¯uorobenzene [21]. Unreacted starting
material was recovered almost quantitatively. To our knowl-
edge no other direct electrophilic substitution reactions have
been reported for any highly ¯uorinated pyridines. In the
light of our recent work with trimethylsilylated poly¯uor-
obenzenes [22] where we were able to effect nitro and

56
P.L. Coe, A.J. Rees / Journal of Fluorine Chemistry 101 (2000) 45±60
Scheme 4.
possible toxicity of the nitrosopyridine. We decided in the
®rst instance not to try to isolate the nitroso compound but to
either oxidise it to the nitro- compound, a known species, or
to aminate it to the nitrosoamine, or to reduce it to the 3-
amino compound. We proposed to do all of these reactions
in situ. The lithium reagent 14 was reacted at 788C with
gaseous nitrosyl chloride fed into the reaction from an
external reservoir of the liquid as a steady stream of the
gas diluted with dry nitrogen. During the passage of the gas
the solution of 14 was raised slowly to room temperature
(over about 2 h). The reaction mixture was then stirred at
room temperature for a further 2 h when the ¯ow of nitrosyl
chloride was stopped and air was drawn through the solution
for 6 h. After this time an aliquot of the solution was
examined by 19F NMR spectroscopy. The spectrum showed
three sets of signals attributable to three compounds con-
taining poly¯uoropyridine rings. Although the three com-
ponents could be observed by TLC, none could be separated
into a pure component by column chromatography. In the
light of later results which allowed us to compare NMR data
from pure compounds with the spectra of the mixture we
were able to identify the three components as bis-(2,4,5,6-
tetra¯uoro-3- pyridyl) amine 21, bis-(2,4,5,6-tetra¯uoro-3-
pyridyl)hydroxylamine 22 and bis-(2,4,5,6-tetra¯uoro-3-
pyridyl)nitroxide 23. We then repeated the experiment
except that 0.880M ammonia was added after the addition
of nitrosyl chloride was stopped. Work-up of the reaction
afforded a very dark crystalline solid in ca. 30% yield. TLC
showed two spots of equal intensity but no clean separation
was possible using column chromatography. The 19F NMR
spectrum of the mixture showed two sets of signals attri-
butable to bis-(2,4,5,6-tetra¯uoro-3-pyridyl) amine 21 and
N-(4-amino-2,5,6-tri¯uoro-3-pyridyl)-3-aminotetra¯uoro-
pyridine 24 by comparison with authentic samples (see
below). The reaction was repeated but at the end of the
addition of the nitrosyl chloride, water and zinc powder
were added to the reaction mixture, work-up afforded a pale
yellow solid. This product was con®rmed as bis-(2,4,5,6-
tetra¯uoro-3-pyridyl)amine 21 by the usual combination of
physical methods. This and the previous two experiments
again indicated the high reactivity of the lithium reagent 14.
The activating effect of the tetra¯uoropyridyl ring on
neighbouring electrophilic centres is again apparent since
again it appears that the ®rst formed product is much more
reactive than the starting material. This results in the for-
mation, as we found in the reaction with acetyl chloride and
tri¯uoroacetic anhydride, of the disubstituted products (see
below). The results of this set of reactions are illustrated in
Scheme 5.
Having obtained a pure sample of 21 we were able to
react it with 0.880 ammonia to obtain an amine. Analysis of
the structure by physical means showed it to be identical to
compound 24 above thus con®rming the structure of 24. We
also carried out a number of unsuccessful experiments with
‡
14 and nitrosonium tetra¯uoroborate (NO BF4) and with
‡
nitronium tetra¯uoroborate (NO2 BF4) to make the nitroso
and nitrocompounds, respectively.
We next investigated the reaction of some of the com-
pounds we had prepared. Firstly we studied some chemistry
of tetra¯uoronicotinic acid 6. Much to our surprise, we
found it impossible to directly esterify 6, using a variety of
procedures. Further we failed to obtain the ester directly by
reaction of the lithiocompound 14 with puri®ed ethyl chlor-
oformate, a reaction which readily proceeded with 2,3,5,6-
tetra¯uoro-4-lithiopyridine to yield ethyl tetra¯uoroisoni-
cotinate 25. At present we have no satisfactory explanation
for these observations. From previous observations and
from the result we obtained in the attempted preparation
of the aldehyde 15 above, when we isolated the 4-hydroxy
compound as the major product, we expected that the 4-
¯uorine atom in tetra¯uoronicotinic acid 6 would be very
reactive towards nucleophiles. This proved to be the case
since treatment of the acid 6 with excess 0.880 ammonia at
room temperature gave 4-amino-2,5,6-tri¯uoronicotinic
acid 26 in 72% yield, which was fully characterised by
physical methods. We found no evidence for the formation
of the analogous 2-amino compound. We subsequently
discovered that we could carry out the carbonation and
amination successively in a `one-pot reaction' in 83% yield.
Tetra¯uoroisonicotinic acid and its ethyl ester 25 are readily
available and it is reported that penta¯uorobenzoic acid will,
under carefully controlled conditions, give a high proportion
of the ortho substituted product in nucleophilic substitution
reactions [4]. We felt therefore that we should investigate

P.L. Coe, A.J. Rees / Journal of Fluorine Chemistry 101 (2000) 45±60
57
Scheme 5.
the reaction of 25 with ammonia. Thus, we treated ethyl
tetra¯uoroisonicotinate 25 with 0.880 ammonia. In addition
to unreacted starting material we obtained 2-amino-3,5,6-
tri¯uoroisonicotinamide 27 in 33% yield, ethyl-2-amino-
3,5,6-tri¯uoroisonicotinate 28 12% and a small amount of
2,3,5,6-tetra¯uoroisonicotinamide 29, but none of the
desired 3-amino products were observed. This result sug-
gests that the activating effect of the ring nitrogen atom is
still suf®ciently potent to overcome any directing effect of
the ester function and that derivatives of tetra¯uoroisoni-
cotinic acid are likely to be of little value in the quest for
suitable substrates for deazapurine formation. The results of
these reactions are shown in Scheme 6. Not surprisingly, in
the light of the failed direct nitration reported above,
Scheme 6.

58
P.L. Coe, A.J. Rees / Journal of Fluorine Chemistry 101 (2000) 45±60
Scheme 7.
nitrodesilyl- and nitrodestannylation reactions of 17 and 18
under conditions we had found gave reasonable results in
the poly¯uorobenzene series [22] gave only unreacted
starting materials.
in the spectrum of the starting 26 at ꢀ 61, 91 and 170
had been replaced by signals at ꢀ 92, 99 and 167; these
values compare very favourably with those reported by
Chambers [2] for 3,4-diaminotri¯uoropyridine at ꢀ 94,
104 and 169. This led us to believe we had produced a
tri¯uoropyridine with amino functions at the 3- and 4-
positions. Clearly it was not the free diamine or a salt of
it since it had a sharp melting point at 239±2418C, whereas
the diamine as reported by Chambers had a melting point of
117±1188C. The mass spectrum showed a peak of m/z 189 as
the highest mass ®tted with an empirical formula
C6H2F4N3O and this was con®rmed by the elemental ana-
lysis as being the molecular formula. The 13C NMR spec-
trum showed a signal at ꢀ 155.2 indicative of a carbonyl
group; this was con®rmed by the J-modulated spin echo
spectrum. This data lead us to assign the structure of the
product as 3-deaza-2,3,6-tri¯uoro-8-hydroxypurine (3-
deaza-2,3,6-tri¯urouric acid) 30. The proposed reaction
sequence leading to this unexpected product is shown in
Scheme 8.
The standard route for the conversion of uric acid 31 and
its derivatives to purines of the kind we required is by re¯ux
of 31 with acid amides. To obtain 8-unsubstituted com-
pounds, formamide itself is used, as exempli®ed by the
preparation of xanthine 32 from uric acid (Scheme 9). We
investigated the reaction of 30 under conditions which had
given us a good yield of 32 in a trial experiment, with no
success. A wide range of different reduction methods we
then tried all failed giving either unchanged 30 or in tar
formation. Thus, our ®nal goal eluded us.
The conversion of the carboxyl group in 4-aminotri¯uor-
onicotinic acid 26 to an amine should be possible by one of
several methods, e.g. the Hofmann degradation known to
proceed with tetra¯uorophthalimide [29] to yield tetra¯uor-
oanthranilic acid, or the Curtius reaction and its variations.
In the light of possible substitution at the 2 or 6 positions in
the amino-acid 26 by hydroxide in the strongly basic
conditions required for the Hofmann degradation, we
decided to investigate the Curtius reaction and the Schmidt
variation of it. We tried the Schmidt variation without
success, the only recoverable material was unreacted 26.
A relatively recent modi®cation of the Curtius reaction has
been reported to work well with relatively unactivated
carboxylic acids [30,31]. This modi®cation uses diphenyl
phosphoraziate as the azide source. The rationale behind the
reaction is that the carbonyl group in the acid is activated by
reaction at the phosphorus ultimately leading to the required
acylazide. This reaction is in some ways akin to the well-
known Mitsonobu reaction. The acyl azide so formed leads,
in the expected manner, by the Wolf-type rearrangment to
the isocyanate which then reacts with t-butanol to form the
carbamate ester which on acidic hydrolysis affords the
desired amine as its salt. This method was very attractive
to us since one of the cited examples of its ef®ciency was the
formation of 2-aminopyridine from pyridine-2-carboxylic
acid in 73% yield [31]. This reaction sequence is shown in
Scheme 7. The initial reaction we carried as a trial using
tetra¯uoronicotinic acid was not very encouraging, it gave a
mixture of a number of products which could not be readily
separated. This is perhaps not too surprising since in this
acid we still have the highly reactive 4-¯uorine atom present
which, especially in the presence of triethylamine, could
easily react with t-butanol, the preferred reaction solvent,
and any other nucleophilic species present or generated in
the reaction. We then reacted the amino acid 26 under the
literature conditions and obtained a single product albeit in a
rather low yield (19%). The 19F spectrum proved to be very
useful in elucidating the structure of the product, the signals
We next decided to study the reactions of 14 with carbon
and nitrogen electrophiles and these reactions proved to be
much more interesting than we expected. Reaction of the
lithiocompound 14 and benzaldehyde, gave phenyl-
(2,4,5,6- tetra¯uoro-3-pyridyl)-methanol 33 in 71% yield,
identi®ed by physical means. The reaction with propanone
proceeded smoothly to give a clear liquid in >90% yield, the
1H NMR spectrum of which indicated it to be almost pure 2-
(2,4,5,6-tetra¯uoro-3-pyridyl) propan-2-ol 34. The spec-
trum was notable for an unusual coupling pattern, a quintet
with a coupling constant of 2 Hz for the methyl groups.
However, all our attempts to completely purify 34 either by

P.L. Coe, A.J. Rees / Journal of Fluorine Chemistry 101 (2000) 45±60
59
Scheme 8.
386 and the IR spectrum showed the presence of a carbonyl
function consistent with the presence of an ester. These data
together with the 19F and 13C NMR spectra, accurate mass
measurement and the elemental analysis con®rm the empiri-
cal formula as C14H6F8O2, consistent with the structure
being 1,1-bis(2,4,5,6-tetra¯uoro-3-pyridyl)ethyl acetate 36.
The origin of the unexpected multiplicity of the methyl
signal we believe is best explained by a coupling to the ortho
¯uorine atoms of each of the pyridine rings, a study of
models of the compound reveals that the methyl groups and
these atoms are very close together. This kind of coupling
has been observed previously in other ¯uoroarenes [25] but
whether it is a ``through space'' or a long-range through
bond coupling is not clear. This result clearly indicated that
the pyridyl lithium 14 is very reactive and interestingly that
the presumed ®rst formed ketone, the expected product of
the reaction, appears to be more reactive than acetyl chloride
at least in the system we were using. A similar reaction using
Scheme 9.
high vacuum distillation at low temperature or by column
chromatography failed. Distillation at various temperatures
and pressures always afforded 2-(2,4,5,6-tetra¯uoro-3-pyr-
idyl) prop-2-ene 35, the dehydration product of 34, whilst
chromatography gave a mixture of 34 and 35. We were able
to obtain 35 more simply by distillation of 34 with a trace of
added p-toluenesulphonic acid. However, we were unable to
obtain a sample of 34 pure enough for either elemental
analysis or accurate mass measurement, although we were
able to obtain NMR parameters and a mass spectrum (GC/
MS) as partial characterisation. This ready dehydration of
poly¯uoroaryl methanols has been noted previously in the
penta¯uorobenzene series [18]. The reaction of 14 with an
excess of acetyl chloride, which we had thought would give
the methyl ketone analogous to the reaction of penta¯uor-
ophenylorganometallics [24] gave a somewhat unexpected
result. The 1H NMR spectrum of the crude product, obtained
in 80% yield, showed the presence of two methyl groups,
one of which was at ꢀ 2.14 as a singlet and the second one
was at ꢀ 2.36, again, as noted above for 34, a quintet with a
coupling of 2 Hz. The mass spectrum indicated a RMM of
1
tri¯uoroacetic anhydride afforded a single product, the H
NMR spectrum showed a broad signal characteristic of an
OH group at ꢀ 5.2 which disappeared on shaking with D2O.
The IR spectrum showed no carbonyl group to be present.
The 19F spectrum in addition to signals for the ¯uorine
atoms of the pyridine ring system showed a quintet at ꢀ
77.8, characteristic of a CF3 group and interestingly
showing as a quintet again clearly coupled to the ring
¯uorine atoms. The mass spectrum and elemental analysis
further con®rmed the structure as 1,1-bis(2,4,5,6-tetra-
¯uoro-3- pyridyl)-2,2,2-tri¯uoroethanol 37, again arising
from rapid attack of 14 onto the ®rst formed tri¯uoromethyl
ketone. Thus, we have further demonstrated in these experi-
ments the reactivity of 2,4,5.6-tetra¯uoro-3-pyridyllithium.
These results are shown in Scheme 10.

60
P.L. Coe, A.J. Rees / Journal of Fluorine Chemistry 101 (2000) 45±60
Scheme 10.
[11] R.E. Banks, J.E. Burgess, W.M. Cheng, R.N. Haszeldine, J. Chem.
4. Conclusion
Soc. (1965) 575
[12] L.A. Wall, W.J. Pummer, J.F. Fern, J.M. Antonucci, J. Res. Nat. Bur.
Stand. Sect A 67 (1963) 481.
We believe that we have considerably widened the scope
of poly¯uropyridine chemistry, particularly the develop-
ment of the preparation and reactions of 2,3,4,6-tetra¯uor-
opyridine.
[13] G.M. Brooke, R.D. Chambers, J. Heyes, W.K.R. Musgrave, Proc.
Chem. Soc. (1963) 213
[14] D.J. McNamara, P.D. Cook, J. Med. Chem. 30 (1987) 340.
[15] R.L. Powell, Personal communication.
[16] R.D. Chambers, G.A. Heaton, W.K.R. Musgrave, L. Chadwick, J.
Chem. Soc. (1969) 1700.
References
[17] G.M. Brooke, R.S. Matthews, N. Robson, J. Chem. Soc., Perkin
Trans. 1 (1980) 2520.
[18] E. Nield, R. Stephens, J.C. Tatlow, J. Chem. Soc. (1959) 166.
[19] P.L. Coe, R. Stephens, J.C. Tatlow, J. Chem. Soc. (1962) 3227.
[20] R.J. Harper, E.J. Soloski, C. Tamborski, J. Org. Chem. 29 (1964)
2385.
[1] R. Storer, C.J. Ashton, A.D. Baxter, M.M. Hamm, C.L.P. Marr, A.M.
Mason, C.L. Mu, P.L. Myers, S.A. Noble, C.R. Renn, N.A. Weir,
J.M. Wood, P.L. Coe, Nucleosides and Nucleotides 18 (1999) 203.
[2] R.D. Chambers, J. Hutchinson, W.K.R. Musgrave, J. Chem. Soc. C
(1966) 220.
[21] P.L.Coe, A.E. Jukes, J.C. Tatlow, J. Chem. Soc. C (1966) 2323.
[22] P.L. Coe, A.M. Stuart, D.J. Moody, J. Fluorine Chem. 92 (1998)
27.
[3] G.M. Brooke, J. Burdon, J.C. Tatlow, J. Chem. Soc. (1961) 802.
[4] J. Burdon, W.B. Hollyhead, J.C. Tatlow, J. Chem. Soc. (1965) 6336.
[5] R.D. Chambers, F.G. Drakesmith, W.K.R. Musgrave, J. Chem. Soc.
(1965) 5045.
[23] Organic Synthesis: The Roles of Boron and Silicon,Thomas Oxford
Science Publications, 1991, pp. 48 and 49.
[24] A.K. Barbour, M.W. Buxton, P.L. Coe, R. Stephens, J.C.Tatlow,
J.Chem. Soc. (1961) 808.
[25] J. Burdon, Tetrahedron 21 (1965) 1101.
[6] R.D. Chambers, J. Hutchinson, W.K.R. Musgrave, J. Chem. Soc.
(1964) 3573.
[7] R.E. Banks, R.N. Haszeldine, J.V. Latham, I.M. Young, J. Chem.
Soc. (1965) 594.
[26] H. Wieland, Ber. 36 (1903) 2315.
[27] V.H. Derner, O.C. Derner, J. Am.Chem. Soc. 64 (1942) 3056.
[28] B. Oddo, Gazz. Chim. Ital. 39 (1909) 659, C.A. (1911) 686.
[29] B.Gething, C.R. Patrick, J.C. Tatlow, J. Chem. Soc. (1961) 1574.
[30] T. Shioiri, K. Ninomiya, S. Yamada, J. Am. Chem. Soc. 94 (1972)
6203.
[8] B. Langlois, L. Gilbert, G. Forat, in: J.-R. Desmurs, S. Ratton (Eds.),
Industrial Chemistry Library, vol. 8, Elsevier, Amsterdam, 1996,
pp. 244±293.
[9] I. Horner, G. Mummenthey, H. Moser, P. Beck, Ber. 99 (1966) 2782.
[10] R.E. Florin, W.J. Pummer, L.A. Wall, J. Res. Nat. Bur. Stand. 62
(1959) 119.
[31] T. Shioiri, K. Ninimiya, S. Yamada, Tetrahedron 30 (1974) 2151.