A one-pot diazotation-fluorodediazoniation reaction and fluorine gas for the production of fluoronaphthyridines

Article abstract of DOI:10.1021/op500100b

Several synthetic routes to 7-fluoro-2-methoxy-8-methyl-1,5-naphthyridine (1) are presented, and their suitability for scale-up is discussed. The way of introducing the fluorine atom is crucial. Early routes start from commercially available fluorinated building blocks or employ F⁺ reagents like SelectFluor and delivered up to 70 kg of 7-fluoro-2-methoxy-1,5-naphthyridine (18). To prepare for larger scales, the focus turned to the use of HF or elemental fluorine, both one of the cheapest sources of fluorine. The first method, a one-pot diazotation-fluorodediazoniation with 6-methoxy-1,5- naphthyridin-3-amine (9) in HF gave the fluorinated naphthyridine 18 in high yield and purity without isolation of the unstable diazonium salt, the latter being a severe drawback of the related Balz-Schiemann protocol. The second method relies on the use of fluorine gas for a surprisingly selective ortho-fluorination of 6-methoxy-1,5-naphthyridin-4-ol (10).

Full text of DOI:10.1021/op500100b

Article
pubs.acs.org/OPRD
A One-Pot Diazotation−Fluorodediazoniation Reaction and Fluorine
Gas for the Production of Fluoronaphthyridines
Stefan Abele,*,§ Gunther Schmidt,§ Matthew J. Fleming,⊥ and Heinz Steiner⊥
§Process Research Chemistry, Actelion Pharmaceuticals Ltd., Gewerbestrasse 16, CH-4123 Allschwil, Switzerland
⊥Solvias AG, Romerpark 2, CH-4303 Kaiseraugst, Switzerland
̈
S
* Supporting Information
ABSTRACT: Several synthetic routes to 7-fluoro-2-methoxy-8-methyl-1,5-naphthyridine (1) are presented, and their suitability
for scale-up is discussed. The way of introducing the fluorine atom is crucial. Early routes start from commercially available
fluorinated building blocks or employ F+ reagents like SelectFluor and delivered up to 70 kg of 7-fluoro-2-methoxy-1,5-
naphthyridine (18). To prepare for larger scales, the focus turned to the use of HF or elemental fluorine, both one of the
cheapest sources of fluorine. The first method, a one-pot diazotation−fluorodediazoniation with 6-methoxy-1,5-naphthyridin-3-
amine (9) in HF gave the fluorinated naphthyridine 18 in high yield and purity without isolation of the unstable diazonium salt,
the latter being a severe drawback of the related Balz−Schiemann protocol. The second method relies on the use of fluorine gas
for a surprisingly selective ortho-fluorination of 6-methoxy-1,5-naphthyridin-4-ol (10).
The hydroxy-naphthyridine 10 is the common early inter-
mediate of routes 4 and 5, ultimately derived from 7 via another
high-temperature reaction, the Conrad−Limpach reaction. All
approaches requiring the handling of fluoroacetates9 were
discarded due to their insidious toxicity. Some syntheses could
conceivably start from dichlorofluoromethane;10 however,
Freon 21 is banned by the Montreal protocol. Starting from
various 3-fluoro-5-amino-6-halopyridines seemed attractive, but
their high costs ($2,400−10,000/kg) made them less attractive
as raw materials.11
The first “fit-for-purpose” route (Route 1, Scheme 1) started
from chloronitropyridine 2 that underwent a Stille coupling
with tributyl(1-ethoxyvinyl)tin.12 The fluorine was introduced
with SelectFluor and the ensuing α-fluoroketone 4 was
condensed with DMF dimethylacetal (DMF−DMA) to get
the substrate 12 for the nitro reduction−deoxobromination
sequence to bromo-fluoro-naphthyridine 13. Bromine−lithium
exchange with hexyllithium followed by MeI quench gave the
targeted fluoronaphthyridine 1.13 The benefits of this route are
its conciseness and the good yields−the reasons why it served
well as “fit-for-purpose” route to deliver the first 10 kg of 1 for
the downstream steps towards the API. Still, it suffered from
the toxic tin organyls, two cryogenic steps, and the high cost of
SelectFluor precluding a further scale-up.14
INTRODUCTION
■
More and more methods are emerging for the late-stage
introduction of fluorine into advanced intermediates.1−7
However, once a route has to meet the stringent requirements
for scale-up, the range of technologies is still restricted to
fluorinations that have proven viable on very large scale, for
example for the manufacturing of fluorinated heterocycles as
raw materials in the fine chemicals or agrochemical industry.
During the route finding and process development for the
scale-up of 7-fluoro-1,5-naphthyridines 1 and 18 as inter-
mediates for active pharmaceutical ingredients (API) at
Actelion Pharmaceuticals Ltd., different scalable fluorination
processes were tested. Amongst these, a diazotation−
fluorodediazoniation protocol and an unprecedented selective
fluorination with fluorine gas have been most attractive in
comparison with other strategies like “buying in” the fluorine
via a cheap raw material or using F+ reagents.8 In the following,
five routes towards 1 are presented with a focus on the different
strategies used to introduce the fluorine atom on low-kg to 70
kg-scale. Availability, cost, ease of processing, and the number
of chemical steps are finally balanced in order to select the most
economical route for scales larger than 100 kg.
The routes 1−5 are summarized in Figure 1. Route 1 uses
the F+ reagent SelectFluor for the fluorination of the enol ether
3. Route 2 follows the strategy “buying-in F” and starts from
readily available dichlorofluoronicotinic acid (5). 3-Amino-6-
methoxypyridine (7) is the common starting material for routes
3, 4, and 5. In route 3, it serves as substrate for the Gould−
Jacobs reaction at high temperature for the construction of the
1,5-naphthyridine core that is further elaborated to the 3-
aminonaphthyridine 9. The latter is a common intermediate of
routes 3 and 4. Two diazotation−fluorodediazoniation methods
then shall deliver the targeted fluorinated product 18, either
using tert-butylnitrite/BF3·OEt2 or NaNO2 in HF for the
diazotation. The target product 1 is then conceivably produced
from 18 following an ortho-lithiation−methylation protocol.
Dichlorofluoronicotinic acid (5) is relatively affordable (<
$165/kg) and was considered an economically viable
fluorinated building block (Scheme 2). Route 2 started with
the regioselective dechlorination to 14 under transfer hydro-
genation conditions. The carboxyl group was transformed into
the amino group via a Curtius reaction with diphenylphos-
phoryl azide (DPPA) in tert-butanol followed by Boc-
Special Issue: Fluorine Chemistry 14
Received: March 24, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op500100b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 1. Synopsis of routes 1−5 to 1 discussed in the text. The source of the fluorine atom is highlighted in yellow.
Scheme 1. Route 1 (Stille) to 1 using F+ reagent SelectFluor (10-kg scale)
Scheme 2. Route 2 (Heck) from fluorinated pyridine building block 5 (70-kg scale)
deprotection. The chlorofluoropyridine 6·HCl was coupled
with n-butyl acrylate in a Heck reaction to 15 that was cyclized
to the 1,5-naphthyridinone 16 in HOAc in the presence of
tributyl phosphine. The methoxy group was introduced via the
chloride 17 by a SNAr reaction with NaOMe to afford the
fluoro-naphthyridine 18 that is the precursor of 1 (Scheme 5
below). This Route 2 proved to be robust and reproducibly
delivered 70 kg of 18, but the cost of goods and the number of
steps should be further decreased (see Table 2). In addition, Pd
was used in two steps, the phosphine should be avoided, and
the Curtius reaction being associated with azides was not
preferred for large-scale production.15
Diazotation−fluorodediazoniation is a preferred method for
the regioselective introduction of fluorine into aromatic
rings.16,17 In the Balz−Schiemann reaction, the diazonium
tetrafluoroborates are isolated and thermally decomposed to
the fluoroaromatic products either neat or in solution. Two
major risks have to be assessed when scaling up diazotation−
fluorodediazoniations. First, the thermal instability of the
diazonium salt might lead to premature loss of nitrogen
B
dx.doi.org/10.1021/op500100b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 3. Route 3 (Gould−Jacobs, Balz−Schiemann) to 18 (100-g scale)
(sudden pressure buildup) so prolonged storage in tanks or
transfer through lines is dangerous. In addition, some
diazonium salts are friction sensitive and explode after
subjection to mechanical stress. Second, the fluorodediazonia-
tion, the release of nitrogen, is performed at elevated
temperature and the exothermic reaction, gas evolution,
foaming, and accumulation have to be controlled. Three
families of NO+ reagents are mainly used for the production of
the diazonium salts: (i) NaNO2,18,19 nitrosonium tetrafluor-
oborate (NOBF4),20 or alkyl nitrites.21,22
The required aniline 9 for such an approach was produced as
shown in Scheme 3. Gould−Jacobs reaction of the con-
densation product 19 of the aminopyridine 7 with ethoxy-
methylene malonate at 280−290 °C in Dowtherm A23 gave the
hydroxy-naphthyridine 20. Bromination to 21 and hydro-
genation delivered the ester 8 that was transformed into the
primary amide 22. The Hofmann rearrangement was run in
NaOH and pyridine with bromine and yielded the desired
aniline 9 for the Balz−Schiemann reaction. First, the diazonium
tetrafluoroborate 23 was prepared with tert-butylnitrite and
BF3·OEt2 in THF. The diazonium salt was then dosed in
portions to heptane at 85 °C to trigger the fluorodediazoniation
to 18. However, the salt 23 decomposed already at rt, as
corroborated by a TLeft Limit of 39 °C in the differential scanning
calorimetry (DSC, see Figure 2). Stabilizing with heptane was
only partially successful, raising the left limit of the exothermic
decomposition by 26 to 65 °C, still too low a temperature to
allow for handling of 23 on large scale.24 Running the
dediazoniation of 23 at higher temperature in xylenes (70−
100 °C) led to 20% of a byproduct (by LC−MS), most
probably being derived from a Friedel−Crafts reaction of the
diazonium salt with the solvent.25 Clearly, isolation of the
diazonium salt was not acceptable for scale-up. As the
fluoronaphthyridine 18 was very pure (99% a/a by LC−MS),
we studied this type of transformation further.
For a safe scale-up of such deaminofluorinations, continuous
flow applications have been developed encompassing the
fluorodediazoniation step, however, these methods still require
the bulk isolation of dangerous diazonium salts.26 Methods are
known where the labile diazonium salt is not isolated.27−31 We
felt that these fluorodediazoniation reactions are best suited for
the scale-up as the diazonium salt 23 posed severe risks of
explosion. The following reagents have mostly been used:
Figure 2. DSC of the diazonium salt 23. (a) Dry, TLeft Limit = 39 °C
(−552 kJ/kg). (b) Heptane-wet, TLeft Limit = 65 °C (−470 kJ/kg).
NaNO2 or NOBF4 in HF or in pyridine/HF solutions (Olah’s
reagent).
We found that the substrate for the diazotation−
fluorodediazoniation 9 can more conveniently be produced
from the readily accessible hydroxy-naphthyridine 10. The
latter is derived from 5-amino-2-methoxypyridine (7), Mel-
drum’s acid (24), and triethyl orthoformate (Scheme 4),32 or
from methyl propiolate.33 For larger scales, reusing the
Dowtherm A for several batches could be an opportunity for
savings.
The ortho-nitration of hydroxy-naphthyridine 10 proceeded
with high selectivity (Scheme 5). Bromination of hydroxy-nitro-
C
dx.doi.org/10.1021/op500100b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 4. Conrad−Limpach reaction used for the synthesis of 10
Scheme 5. Route 4 (HF) and Route 5 (F2) to 1
Figure 3. Unsuccessful ortho-fluorinations of 10 with the structures of the used F+ reagents. Screening parameters: solvent (DMF, MeOH, TFA,
AcOH, ACN, water), additive (H2SO4), temperature (20 °C, 60 °C).
naphthyridine 26 with PBr3 in DMF to 11 followed by an
excessive reduction of both the bromide and the nitro group
afforded the naphthyridylamine 9 in good yields. For the
ensuing pivotal transformation, two different conditions turned
out to yield 18 in high purity and 90% yield.34 If HF (bp 19.5
°C) was used as a single solvent, the reaction was carried out in
an autoclave. The addition of pyridine reduces the vapor
pressure of HF, thus allowing to perform the reaction at normal
pressure at reduced operation costs.35 According to the first
method, NaNO2 was added to a solution of 9 in HF at −5 °C
in a perfluoralkoxyalkane flask. Only traces of fluoro-
naphthyridine 18 were detected after warming up to 10 °C
within 1 h. The mixture was transferred into a Monel autoclave,
and the fluoro-dediazoniation was triggered by heating the
mixture at 65 °C for 30 min, recording a pressure of 6.6 bar. HF
was removed by distillation, and 18 was obtained with >99%
a/a purity after aqueous work-up. The second method differed
from the first one by the addition of pyridine and by a different
order of addition: in a perfluoralkoxyalkane flask, NaNO2 was
added to 30% w/w pyridine/HF at −5 °C, followed by
charging of aniline 9 at −40 °C. The mixture was then heated
to 60 °C until the gas evolution ceased. After quenching the
mixture on ice and neutralization with aqueous ammonia, 18
was obtained in the same yield of 90% and >94% a/a purity.
The methyl group was installed onto 18 by ortho-lithiation
with lithium diisopropylamide (LDA), followed by methylation
with MeI to yield 1 (80% yield). Route 4 was uneventfully
demonstrated on pilot plant scale (250−1000 L) by a CRO up
to the naphthyridylamine 9. The diazotation−fluorodediazo-
niation was demonstrated on 50-g scale, the larger-scale
production depending on clinical milestones. It would be a
conceivable approach for our intended larger scales.
The rapid access to hydroxy-naphthyridine 10 (Scheme 5)
rendered an ortho-fluorination with a F+ reagent an attractive
option. To estimate the reactivity of 10, this hydroxy-
naphthyridine was successfully ortho-iodinated, -brominated,
and -chlorinated in HOAc with NIS, NBS, and NCS,
respectively. The yields were higher than 81% on 50-g scale.
This prompted us to examine the ortho-fluorination with
commercially available F+ reagents (Figure 3). However, under
the depicted reaction conditions, the best result was 16%
conversion to 27 with Accufluor NFTh in MeOH at 60 °C, the
remainder being mostly unreacted starting material.
Elemental fluorine would be the ideal F+ reagent from a cost
perspective, most electrophilic F+-reagents being produced with
F2. However, fluorine gas suffers from extreme reactivity
D
dx.doi.org/10.1021/op500100b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 6. Published fluorinations with F2 of related substrates
(nonselectivity), associated with exothermic reactions, oxida-
tions, and polymerization side reactions.36−40 F2 was applied for
the fluorination of phenols41−44 and pyridines,45,46 albeit in
moderate yields (Scheme 6). A notable example is 5-
fluorouracil that is produced from uracil with fluorine in
water.47,48 High functionalization is typically not tolerated, and
moderate yields are prevailing.8 An example most similar to 10
is 6-methoxyquinoline that has been fluorinated in 5-position in
low yield, accompanied by an overfluorinated byproduct.49
Hence, the chances for a highly regioselective fluorination of 10
were low.
Surprisingly, an excellent selectivity was observed when 10%
F2 in N2 was bubbled through a solution of 10 in conc. H2SO4
(100 vol) at 80 °C for 4 h, the IPC (HPLC) indicating a 89:11
mixture of starting material and product without any by-
products or side products (Table 1, entry 1). Heating to higher
temperatures to accelerate the reaction led to significant
byproduct formation (entry 2, byproducts were not identified).
The use of a sintered glass frit for a better dispersion of F2
doubled the reaction velocity (entry 3). When the concen-
tration was increased by a factor of 10 to 10 vol, excessive
foaming resulted in a slower reaction rate (entry 4). A higher
concentration is desirable as it will reduce the amount of base
required for the quench. The slower reaction rate could
successfully be compensated by the use of a glass frit with
double-sized pores (20 instead of 10 μm, entry 5). On 50-g
scale, 10% F2 in N2 was bubbled into the solution of 10 in conc.
H2SO4 (10 vol) via a 20-μm frit at 80 °C for 27 h, leading to
93% conversion (no byproduct by HPLC). SiO2 was added as
antifoaming agent and the flow rate was varied between 30 and
60 L/h to control remaining foaming (entry 6).50 27 was
obtained in a moderate yield of 41% yield after a quench with
aqueous ammonia and filtration (nonoptimized work-up).51 All
six alternative solvents tested (entries 7−15) either led to no
conversion or to multiple byproducts. HF would be a preferred
solvent due to the ease of its removal by distillation. The
reaction in HF was carried out in an autoclave and did give the
expected product, albeit accompanied by side products (entry
16). Remaining challenges for the scale-up of the ortho-
fluorination with elemental fluorine are the high corrosivity of
the reaction media, the foaming, and the isolation of the
product from conc. H2SO4 leading to large volumes during the
aqueous work-up.52
Whereas Routes 1 and 2 delivered enough material for
preclinical and phase 1 clinical development, Routes 3, 4, and 5
were proposed for larger scales to reduce the cost of goods. A
rough cost estimate was performed based on a standard dilution
(10 vol), 1 day cycle time for simple reactions (e.g., no work-up
required), 2 days cycle time for standard reactions, 3 days cycle
time for special reactions (nitration, azide, fluorination), and
commercial scale raw material prices with a 1,000 kg API target
(Table 2).53 Route 1 was not considered for this analysis due to
the toxicity of tin organyls and high costs of SelectFluor. For
reasons of confidentiality, the estimated costs are presented
relative to the normalized costs of Route 2. Interestingly, the
major cost driver of all routes is the step that introduces the
fluorine atom, except for Route 2 where the fluorine source is a
purchased raw material. The overall yield is similar for all five
routes although the number of steps varies from 5−8. The one-
pot diazotation−fluorodediazoniation was calculated for routes
3 and 4. Routes 3 and 4 are roughly 20% and 40% less
expensive than Route 2, respectively. As Route 5 (F2 gas) is not
yet deemed fit for pilot-plant scale due to its cumbersome
work-up, the costs have not been assessed. Still, the low
E
dx.doi.org/10.1021/op500100b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(SelectFluor) or started from a commercially available
fluorinated building block (5) and successfully delivered the
fluoronaphthyridines 1 and 18 with high purity for the GMP
production. Later introduction of the fluorine was accom-
plished by the following routes, employing lower-cost
fluorination reagents HF or F2 gas, both being routinely used
on large scale by specialists in the fine chemicals industry. The
one-pot diazotation−fluorodediazoniation protocol of aniline 9
with NaNO2 in HF or pyridine/HF is an attractive option for
scale-up as it afforded 18 in high yield. An additional alternative
is offered by the surprisingly selective ortho-fluorination of
hydroxy-naphthyridine 10 with F2 gas, without the ubiquitous
tarry byproducts reported for such transformations. A critical
parameter was the mass transfer, hence the dispersion of F2 gas
in the mixture, and the work-up. The preferred solvent was
conc. H2SO4. It might be worthwhile to further investigate
other solvents (ideally acidic solvents that can be removed by
distillation) with additives (e.g., HBF4) that will facilitate the
work-up of fluorohydroxynaphthyridine 27.
Table 1. Optimization of the ortho-fluorination of 10. 10/
27/byproducts denoting the ratio of starting material/
a
desired product/byproducts (HPLC)
frit pore size
T
10/27/
byproducts
entry
solvent
(μm)
(°C) t (h)
1
2
3
conc. H2SO4
conc. H2SO4
conc. H2SO4
conc. H2SO4
conc. H2SO4
conc. H2SO4
−
−
10
10
20
20
10
80
125
80
4
2
89/11/0
40/18/42
10/90/0
24/76/0
0/100/0
7/93/0
2
b
4
80
10
10
27
1
c
5
d
80
6
80
7
conc. H2SO4/
dioxane
80
94/0/6
8
HCO2H
HCO2H
AcOH
CHCl3
DMF
−
−
−
−
−
−
−
−
20
60
20
60
20
0
1.5
44/3/53
20/3/77
0/0/100
93/2/5
9
3
1
7
2
1
2
1
EXPERIMENTAL SECTION
10
11
12
13
14
15
■
Caution: strict precautions have to be taken when working with
HF and F2!54−56
45/0/55
85/0/15
0/0/100
43/6/51
MeCN
MeCN
5-(((6-Methoxypyridin-3-yl)amino)methylene)-2,2-di-
methyl-1,3-dioxane-4,6-dione (25). A double-jacketed flask
was charged with 5-amino-2-methoxypyridine (7, 0.5 kg, 1
equiv), 2,2 dimethyl-1,3-dioxane-4,6-dione (0.7 kg, 1.2 equiv),
triethyl orthoformate (0.74 L, 1.1 equiv), and EtOH (4 L). The
mixture was heated to reflux for 1 h. The dark suspension was
cooled to 5 °C, and the mixture was filtered. The product was
washed with EtOH (1 L) and dried on a rotary evaporator to
obtain 25 as a purple solid. Yield: 1.058 kg (95%). Purity (LC−
20
20
10% aqueous
H2SO4
e
16
anhydrous HF
−
60
2
68/24/8
a
b
1 g 10, 100 vol H2SO4, 10% F2 in N2 at 20 L/h. 10 g 10, 10 vol
H2SO4, heavy foaming observed. 10 g 10, 10 vol H2SO4, 30 L/h,
foaming. 50 g 10, 10 vol H2SO4, 30−60 L/h, SiO2 (0.2 wt), foaming.
In a Monel autoclave.
c
d
e
1
MS method 1): 100% a/a, tR = 1.22 min, [M + 1]+ = 279; H
number of steps and the 100% selectivity bear a large potential
for a significant increase of the yield and further cost reduction.
NMR (400 MHz, CDCl3): δ 11.20 (d, J = 13.9 Hz, 1H), 8.51
(d, J = 14.2 Hz, 1H), 8.14 (d, J = 2.9 Hz, 1H), 7.54 (dd, J = 8.9,
3.0 Hz, 1H), 6.85 (d, J = 8.9 Hz, 1H), 3.98 (s, 3H), 1.77 (s,
6H).
Table 2. Comparison of the Routes 1−5 to fluoro-
naphthyridine 1
6-Methoxy-1,5-naphthyridin-4-ol (10). Caution: high
temperature reaction with release of gas! A 4-L glass flask
was equipped with a still head and a large reflux condenser (no
water cooling), connected to a Liebig condenser (no water
cooling) with a receiving flask. The receiving flask was equipped
with a reflux condenser cooled by water. Dowtherm A (1.3 L)
was heated to 255 °C under N2 with an electrical heating
mantle. 25 (161 g, 1 equiv) was dissolved in 1,3-dimethyl-
imidazolidinone (0.5 L) at 80 °C. The hot solution was added
to the boiling Dowtherm A at 255−235 °C over a period of
about 35 min. Caution: gas release; foaming is controlled by
dosage speed. Liquid (acetone) distilled into the receiver. The
reaction mixture was cooled to 20 °C. At approximately 140
°C, a precipitation occurred. The suspension was filtered and
slurried in EtOH (0.8 L) at 80 °C. The mixture was cooled to
20 °C, filtered, and washed with EtOH (0.15 L). The product
was dried on a rotary evaporator at 50 °C at 5 mbar to yield 10
as a brown solid. Yield: 69.6 g (68%). Purity (LC−MS method
overall
yield
(%)
relative
costs
(%)
major cost
drivers
route (F-source)
steps
No. 1 Stille, (Selectfluor)
No. 2 Heck, (5)
6
8
35
25
SelectFluor
a
Pd(OAc)2 DPPA
azide reaction
100
No. 3 Gould−Jacobs,
fluorodediazoniation,
(HF)
8
7
5
21
32
amino-pyridine 7
HF reaction
84
No. 4 Conrad−Limpach,
fluorodediazoniation,
(HF)
amino-pyridine 7
HF reaction
64
b
No. 5 ortho-fluorination,
(F2)
30
F2 reaction
n.a.
a
Not considered due to toxicity of tin organyls and high costs of
SelectFluor.14 Calculated with 41% isolated yield of the fluorination
b
step (>93% purity of the reaction mixture).
1
1): 99% a/a, tR = 0.49 min, [M + 1]+ = 177; H NMR (400
CONCLUSION
MHz, D6-DMSO): δ 11.78 (m, 1H), 7.97 (m, 2H), 7.17 (d, J =
9.0 Hz, 1H), 6.28 (m, 1H), 3.94 (s, 3H).
■
Several routes to 7-fluoro-2-methoxy-8-methyl-1,5-naphthyr-
idine (1) have been developed as kg-amounts of this
intermediate were urgently required for the production of an
API at Actelion Pharmaceuticals Ltd. The synthetic route
design was guided by an efficient introduction of the fluorine
atom. The first two routes either used a F+ reagent
6-Methoxy-3-nitro-1,5-naphthyridin-4-ol (26). Hy-
droxy-naphthyridine 10 (80 g, 1 equiv) was added in portions
to fuming HNO3 (0.5 L) at 10−15 °C within 20 min. The
reaction mixture was heated to 67 °C for 4 h. The mixture was
cooled to 20 °C and added on ice (2 kg). The yellow
F
dx.doi.org/10.1021/op500100b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
suspension was filtered and the product was washed with water
(1.5 L). After drying, 26 was obtained as a yellow solid. Yield:
70 g (70%). Purity (LC−MS method 2): 100% a/a, tR = 0.540
min, [M + 1]+ = 222; 1H NMR (400 MHz, D6-DMSO) δ 13.00
(m, 1H), 9.15 (s, 1H), 8.08 (d, J = 9.0 Hz, 1H), 7.29 (d, J = 9.0
Hz, 1H), 3.99 (s, 3H).
8-Bromo-2-methoxy-7-nitro-1,5-naphthyridine (11).
PBr3 (68 mL, 1.2 equiv) was added to a suspension of
hydroxy-nitro-naphthyridine 26 (132.3 g, 1 equiv) in DMF (1.2
L) at 20 °C and within 15 min. The mixture was stirred at 65
°C for 1 h, cooled to 20 °C, poured on ice (0.8 kg), and
filtered. The product was slurried in EtOH (0.5 L), filtered and
dried on a rotary evaporator at 65 °C, affording 11 as a yellow
solid. Yield: 155.8 g (92%). Purity (LC−MS method 2): 100%
a/a, tR = 1.624 min, [M + 1]+ = 283, 285; 1H NMR (400 MHz,
D6-DMSO): δ 9.21 (s, 1H), 8.44 (d, J = 9.1 Hz, 1H), 7.52 (d, J
= 9.1 Hz, 1H), 4.13 (s, 3H).
6-Methoxy-1,5-naphthyridin-3-amine (9). Bromo-nitro-
naphthyridine 11 (60 g, 1 equiv) and Raney nickel
(approximately 20 g) were suspended in MeOH (600 mL) in
a 1-L autoclave equipped with gas stirrer and thermometer. The
autoclave was inertized before being pressurized with hydrogen
(5 bar), and the mixture was stirred for 2 h. Et3N (59 mL, 2
equiv) and Raney nickel (approximately 10 g) were added after
inertization. The mixture was hydrogenated at 50 °C and 10 bar
for 3 h. IPC (LC−MS) was indicating full conversion. Activated
charcoal (7 g) was added, and the reaction mixture was heated
to 60 °C and filtered hot over Celite (70 g). The filter cake was
rinsed with MeOH (280 mL) at 20 °C. Water (120 mL) was
added to the combined filtrates, the yellow solution was
concentrated on a rotary evaporator at 60 °C under reduced
pressure to remove solvent (800 mL). The suspension was
cooled to 5 °C and filtered. The product was washed with water
(50 mL) and dried on a rotary evaporator at 75 °C below 20
mbar to afford 9 as an off-white solid. Yield: 28.4 g (77%).
Purity (LC−MS method 2): 100% a/a, tR = 0.71 min, [M + 1]+
= 176; 1H NMR (400 MHz, D6-DMSO): δ 8.28 (d, J = 2.5 Hz,
1H), 7.97 (d, J = 8.9 Hz, 1H), 7.04 (d, J = 2.5 Hz, 1H), 6.79 (d,
J = 8.9 Hz, 1H), 5.92 (s, 2H), 3.94 (s, 3H); 13C NMR (125
MHz, D6-DMSO): δ 162.4 (s), 146.5 (s), 143.8 (s), 140.6 (s),
139.8 (s), 133.5 (s), 111.6 (s), 110.5 (s), 53.6 (s).
J = 9 Hz), 140.4 (s), 139.1 (s), 138.9 (d, J = 26 Hz), 118.8 (d, J
= 17 Hz), 116.4 (d, J = 3 Hz), 54.3 (s).
One-Pot Diazotation−Fluorodediazoniation; in 30 wt %
Pyridine/HF. A 200 mL perfluoralkoxyalkane flask was charged
with liquid HF (60 g, 105 equiv) at −40 °C, and pyridine (26.1
g, 11.56 equiv) was added at −50 to −40 °C within 30 min.
Caution: very exothermic! NaNO2 (2.2 g, 1.1 equiv) was added
at −50 to −60 °C within 5 min. The aniline 9 (5 g, 1.0 equiv)
was added to the reaction mixture at −50 to −40 °C within 10
min. The yellow solution was warmed to 20 °C within 30 min,
and then to 65 °C. Caution: an exothermic reaction with gas
(N2) evolution occurred! The mixture was stirred at 60 °C for 2
h until the gas evolution ceased and IPC (TLC) indicated
complete conversion. The mixture was cooled to 0 °C and
poured onto ice (100 g) and EtOAc (80 mL). Aqueous 25%
NH3 (90 mL) was carefully added (pH 4−5). The aqueous
layer was extracted with EtOAc (2 × 20 mL). The organic
layers were washed with water (20 mL). The combined organic
layers were dried over MgSO4 and evaporated to dryness at 50
°C under reduced pressure, yielding 18 as a yellow crystalline
solid. Yield: 4.80 g (90%). Purity (LC−MS method 2): 94.4%
a/a.
In HF. A 200 mL perfluoralkoxyalkane flask was charged with
liquid HF (60 g, 105 equiv) at −40 °C. The aniline 9 (5 g, 1.0
equiv) was added at −50 to −40 °C within 10 min. The yellow
solution was warmed to −5 °C. NaNO2 (2.2 g, 1.1 equiv) was
added in portions at −9 to −5 °C within 5 min. Caution:
exothermic! The reaction mixture was warmed to 10 °C and
stirred for 1 h before being cooled to −30 °C and transferred to
a Monel stirring autoclave. It was heated to 65 °C within 30
min. During the heating-up a pressure increase to 6.6 bar
occurred, indicative of the liberation of N2. Upon cooling to 0
°C, IPC (TLC) showed complete conversion. HF was
evaporated using a water jet pump, and the residue was poured
on ice/EtOAc. Aqueous 25% NH3 (40 mL) was carefully added
(pH 8). The aqueous phase was extracted with EtOAc (2 × 20
mL), and the organic phases were washed with water (20 mL).
The organic phases were dried over MgSO4 and evaporated to
dryness at 50 °C under reduced pressure to yield 18 as a yellow
crystalline solid. Yield: 4.84 g (90%). Purity (LC−MS method
2): 99.3% a/a.
7-Fluoro-2-methoxy-1,5-naphthyridine (18). Balz−
Schiemann. BF3·OEt2 (49 mL, 2.5 equiv) was added dropwise
to a suspension of naphthyridylamine 9 (27 g, 1.0 equiv) in
THF (200 mL) at −20 °C. tert-Butyl nitrite (22.3 mL, 1.1
equiv) was added at −20 °C. The yellow suspension was
allowed to warm to 25 °C and stirred at 25 °C for 15 min prior
to filtration and washing of the filter cake with heptane (3 × 25
mL). After drying on a rotary evaporator at 20 °C and 5 mbar
pressure for a few min, the diazonium salt 23 (41 g) was added
to heptane (200 mL) in 10 portions at 85 °C within 30 min.
The mixture was stirred at 85 °C for 10 min before it was
cooled to 20 °C. EtOAc (250 mL) and water (250 mL) were
added. After layer separation, the organic layer was washed with
water (200 mL), filtered over Na2SO4 (20 g), and concentrated
to dryness on a rotary evaporator at 55 °C under reduced
pressure to afford 18 as a brown crystalline solid. Yield: 22.3 g
(81%). Purity (LC−MS method 2): 98.6% a/a, tR = 1.302 min,
[M + 1]+ = 176; 1H NMR (400 MHz, D6-DMSO): δ 8.86 (d, J
= 2.7 Hz, 1H), 8.32 (d, J = 9.0 Hz, 1H), 8.07 (dd, J = 10.0, 2.4
Hz, 1H), 7.26 (d, J = 9.1 Hz, 1H), 4.03 (s, 3H); 13C NMR (100
MHz, D6-DMSO): δ 163.3 (s), 158.8 (d, J = 258 Hz), 142.5 (d,
Ortho-Fluorination with F2. 3-Fluoro-6-methoxy-1,5-
naphthyridin-4-ol (27). Conc. H2SO4 (500 mL) was charged
in a 1-L fluorination apparatus. Hydroxy-naphthyridine 10 (50
g, 1 equiv) was added, followed by SiO2 (10 g). The suspension
was magnetically stirred. Caution: very exothermic! After 5 min,
a clear brown solution was obtained. The solution was heated
to 80 °C, and 10% F2 in N2 was bubbled through the reaction
mixture at 80 °C with a rate of approximately 30 L/h using a
Teflon tube with a glass frit (20-μm pore size) at the lowest
point of the apparatus (note: the gas flow rate was adjusted to
30−60 L/h throughout the reaction to control foaming). IPC
(LC−MS) showed 93% conversion after 27 h. The reaction
mixture was slowly added on crushed ice (100 g) at 0−20 °C.
Caution: very exothermic! This mixture was added dropwise to
a precooled (about 5 °C) aqueous 28% NH3 solution (1.65 L)
at 0−20 °C. Caution: very exothermic, external cooling with
−70 °C. The final pH of the suspension was 7.4. The mixture
was stirred at rt for 2 h. The product was filtered and dried at
70 °C in a vacuum oven for 16 h to yield 27 as a grey solid.
Yield: 22.5 g (41%). Purity (LC−MS method 3): 94% a/a (10:
1
6% a/a), tR = 3.2 min, [M + 1]+ = 195; H NMR (400 MHz,
G
dx.doi.org/10.1021/op500100b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
D6-DMSO): δ 12.05 (br. s, 1H), 8.39 (br. s, 1H), 8.01 (br. s,
1H), 7.20 (d, J = 9.0 Hz, 1H), 3.97 (s, 3H).
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: stefan.abele@actelion.com.
8-Bromo-7-fluoro-2-methoxy-1,5-naphthyridine (13).
PBr3 (0.82 mL, 1.1 equiv) was added to a mixture of fluoro-
hydroxy-naphthyridine 27 (1.50 g, 1 equiv) in DMF (10 mL) at
50 °C. The mixture was stirred at 70 °C for 1 h. After cooling
to rt, the reaction mixture was diluted with water (500 mL),
and 6 N NaOH (1.3 mL) was added. The solid was filtered off
and thoroughly washed with water. The resulting solid was
taken up in EtOAc, and the solution was filtered through a pad
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
S.A. and G.S. thank Dr. Martin Kesselgruber for assistance in
the cost-of-goods analysis and are grateful to Dr. Thomas
Weller for his support. Dr. Cornelia Zumbrunn and Dr. Jean-
Philippe Surivet are acknowledged for their first synthesis of 18
via Route 2, and we are indebted to Dr. Reber for help on
Route 3.
1
of SiO2 to afford 13 as a beige solid. Yield: 2.0 g (100%). H
NMR (400 MHz, D6-DMSO): δ 8.86 (s, 1H), 8.34 (d, J = 9.0
Hz, 1H), 7.31 (d, J = 9.1 Hz, 1H), 4.08 (m, 3H).
7-Fluoro-2-methoxy-8-methyl-1,5-naphthyridine (1).
From Bromo-fluoro-naphthyridine 13. Bromo-fluoro-naph-
thyridine 13 (50 g, 1 equiv) was dissolved in THF (500 mL)
and cooled to −78 °C. To the suspension was added 2.5 M
hexyllithium in hexane (83 mL, 1.1 equiv) at −78 to −70 °C.
The mixture was stirred at −78 °C for 1 h. To the mixture was
added MeI (13.4 mL, 1.1 equiv) at −78 to −70 °C. The
mixture was allowed to warm to 20 °C. To the mixture was
added water (2 mL) and TBME (250 mL). The mixture was
washed with water (2 × 250 mL). The organic layer was dried
with MgSO4 and concentrated to dryness to give 1 as yellow
solid. The crude product (61.7 g) was purified by
chromatography using silica gel (500 g) and, as eluent, a
mixture of heptane/EtOAc (15:1 v/v). Yield: 32.3 g (86%).
Purity (LC−MS method 2): 99% a/a, tR = 1.465 min, [M + 1]+
REFERENCES
■
(1) Watson, D. A.; Su, M.; Teverovskiy, G.; Zhang, Y.; Garcia-
Fortanet, J.; Kinzel, T.; Buchwald, S. L. Science 2009, 325, 1661−1664.
(2) Fier, P. S.; Hartwig, J. F. J. Am. Chem. Soc. 2012, 134, 10795−
10798.
(3) Teare, H.; Robins, E. G.; Kirjavainen, A.; Forsback, S.; Sandford,
G.; Solin, O.; Luthra, S. K.; Gouverneur, V. Angew. Chem., Int. Ed.
2010, 49, 6821−6824.
(4) Katcher, M. H.; Sha, A.; Doyle, A. G. J. Am. Chem. Soc. 2011, 135,
12990−12993.
(5) Liu, W.; Huang, X.; Cheng, M. J.; Nielsen, R. J.; Goddard, W. A.;
Groves, J. T. Science 2012, 337, 1322−1325.
(6) Sladojevich, F.; Arlow, S. I.; Tang, P.; Ritter, T. J. Am. Chem. Soc.
2013, 135, 2470−2473.
1
(7) Campbell, M. G.; Ritter, T. Org. Process Res. Dev. 2014, 18, 474−
480.
= 193; H NMR (400 MHz, D6-DMSO): δ 8.75 (s, 1H), 8.26
(d, J = 9.0 Hz, 1H), 7.22 (d, J = 9.0 Hz, 1H), 4.05 (s, 3H), 2.57
(d, J = 2.1 Hz, 3H); 13C NMR (125 MHz, D6-DMSO): δ 162.4
(s), 157.2 (d, J = 254 Hz), 141.6 (d, J = 7 Hz), 140.7 (s), 138.3
(d, J = 2 Hz), 138.0 (d, J = 27 Hz), 128.0 (d, J = 13 Hz), 115.6
(d, J = 3 Hz), 54.0 (s), 8.8 (d, J = 3 Hz).
(8) Kirsch, P. Modern Fluoroorganic Chemistry: Synthesis, Reactivity,
Applications, 2nd ed.; Wiley VCH: Weinheim, Germany, 2013.
(9) Jennings, R. A. U.S. Patent 5,204,478, 1993.
(10) 2-Bromo-3-fluoroquinolines: Ondi, L.; Volle, J.-N.; Schlosser,
M. Tetrahedron 2005, 61, 717−725.
(11) The bromo-fluoronaphthyridine 13 was offered for $44,000−
200,000/kg by catalogue companies.
From Fluoro-naphthyridine 18. Hexyllithium in hexane (2.5
M, 400 mL, 1.15 equiv) was added to a solution of
diisopropylamine (144 mL, 1.15 equiv) in THF (1.4 L) at
−26 to −20 °C. The solution was cooled to −75 °C. A solution
of fluoro-naphthyridine 18 (158 g, 1 equiv) in THF (500 mL)
was added to the LDA solution at −75 to −70 °C. The
suspension was stirred at −74 °C for 0.5 h. MeI (68 mL, 1.2
equiv) was added at −74 to −72 °C over a period of 1 h. The
mixture was allowed to warm to 20 °C prior addition of water
(10 mL) and TBME (530 mL). The mixture was washed with
water (2 × 530 mL). The organic layer was concentrated to
dryness to give crude 18 as a beige solid. Yield: 182 g (107%).
Purity (LC−MS method 2): 100% a/a. Four batches were
produced and combined for the following crystallization. Crude
18 (703 g in total) was dissolved in acetonitrile (2 L) at 65 °C,
the solution was cooled to rt and stirred at rt for 2 h. The
suspension was cooled to 5 °C, stirred at 5 °C for 2 h, and
filtered. The cake was washed with cold acetonitrile (0.3 L) and
dried at 60 °C under reduced pressure to yield 18 as an off-
white solid. Yield: 534 g (76%). Purity (LC−MS method 2):
(12) Brown, P.; Dabbs, S. WO/2008/128953, 2008.
(13) This step was not optimized as, originally, 13 was transformed
into another intermediate.
(14) Costs for SelectFluor on 100-kg scale: >$1,100/kg.
(15) Later, GSK published a similar route: Voight, E. A.; Yin, H.;
Downing, S. V.; Calad, S. A.; Matsuhashi, H.; Giordano, I.; Hennessy,
A. J.; Goodman, R. M.; Wood, J. L. Org. Lett. 2010, 12, 3422−3425.
(16) Langlois, B. In Methods of Organic Chemistry (Houben-Weyl), 4th
ed.; Georg Thieme Verlag: Stuttgart, Germany, 2000; Vol. E10a;
Chapter 26, pp 686−740.
(17) Osio Barcina, J.; de la Moya Cerero, S. In Methods of Organic
Chemistry (Houben-Weyl), 4th ed.; Georg Thieme Verlag: Stuttgart,
Germany, 2000; Vol. E10b Part 1; Chapter 1.1.8.5., pp 241−268.
(18) A study of solvents for the fluorodediazoniation: Matsumoto, J.-
i.; Miyamoto, T.; Minamida, A.; Nishimura, Y.; Egawa, H.; Nishimura,
H. J. Heterocycl. Chem. 1984, 21, 673−679.
(19) Dosage of a slurry of a diazonium salt: Hogan, P. J.; Cox, B. G.
Org. Process Res. Dev. 2009, 13, 875−879.
(20) Applied for the synthesis of 4-bromo-3-fluoro-6-methoxy-[1,5]
naphthyridine: Miller, W. H.; Rouse, M.; Seefeld, M. A. WO/2006/
081181 A2, 2006.
(21) Seminal paper: Doyle, M. P.; Bryker, W. J. J. Org. Chem. 1979,
44, 1572−1574.
(22) Applied for the synthesis of 3-fluoroquinolines: El-Ahmad Y.;
Leconte, J.-P.; Malpart, J.; Mignani, S.; Mutti, S.; Tabart, M. WO/
2005/049575 A1, 2005.
1
100% a/a. H NMR assay: 98.5% w/w.
ASSOCIATED CONTENT
■
S
* Supporting Information
Conditions of the LC−MS methods and H and 13C NMR
1
(23) Dowtherm A heat transfer fluid is an eutectic mixture of two
very stable compounds, biphenyl and diphenyl oxide with a bp of 257
°C.
spectra. This material is available free of charge via the Internet
at http://pubs.acs.org.
H
dx.doi.org/10.1021/op500100b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(24) On a manufacturing scale, a safe processing temperature is
considered to be 100 K below the onset temperature of the critical
exothermic events.
(25) Byproduct detected when running the decomposition of 23 in
xylenes:
(26) Yu, Z.; Lv, Y.; Yu, C. Org. Process Res. Dev. 2012, 16, 1669−
1672.
(27) Boudakian, M. M. U.S. Patent 4,096,196, 1978.
(28) Olah, A. G.; Welch, J. T.; Vankar, Y. D.; Nojima, M.; Kerekes, I.;
Olah, J. A. J. Org. Chem. 1979, 44, 3872−3881.
(29) Yoneda, N.; Fukuhara, T. Tetrahedron 1996, 52, 23−36.
(30) Schurter, R.; Siegrist, U.; Rempfler, H.; Baumeister, P. U.S.
Patent 4,831,148A, 1989.
(31) With NOBF4: Milner, D. J. EP0430434, 1991.
(32) Davies, D. T.; Jones, G. E.; Markwell, R. E.; Pearson, N. D.
WO/2003/010138 A2, 2006.
(33) Brooks, G.; Davies, D. T.; Jones, G. E.; Markwell, R. E.; Pearson,
N. D. WO/2003/87098 A1, 2003.
(34) Abele, S., Schmidt, G. WO/2013/118086 A1, 2013.
(35) Bp of 30% and 40% pyridine in HF: 55 and 90 °C, respectively.
(36) Rock, M. H. In Methods of Organic Chemistry (Houben-Weyl),
4th ed.; Georg Thieme Verlag: Stuttgart, Germany, 2000; Vol. E10b
Part 1; Chapters 1.1.2.7. and 1.1.2.8., pp 32−49.
(37) Rozen. S. In Methods of Organic Chemistry (Houben-Weyl), 4th
ed.; Georg Thieme Verlag: Stuttgart, Germany, 2000; Vol. E10a;
Chapter 2.3.1.4., pp 176−178.
(38) Chambers, R. R.; Skinner, C. J.; Hutchinson, J.; Thomson, J. J.
Chem. Soc., Perkin Trans. 1 1996, 605−609.
(39) Sandford, G. J. Fluorine Chem. 2007, 128, 90−104.
(40) A continuous flow process using F2 gas copes well with these
challenges: McPake, C. B.; Sandford, G. Org. Process Res. Dev. 2012,
16, 844−851.
(41) Misaki, S. J. Fluorine Chem. 1981, 17, 159−171.
(42) Misaki, S. J. Fluorine Chem. 1982, 21, 191−199.
(43) Chirakal, R.; Brown, K. L.; Firnau, G.; Garnett, E. S. J. Fluorine
Chem. 1987, 37, 267−278.
(44) Chambers, R. D.; Hutchinson, J.; Sparrowhawk, M. E.;
Sandford, G.; Moilliet, J. S.; Thomson, J. J. Fluorine Chem. 2000,
102, 169−173.
(45) Van der Puy, M. Tetrahedron Lett. 1987, 28, 255−259.
(46) With F2/I2: Chambers, R. D.; Parsons, M.; Sandford, G.;
Skinner, C. J.; Atherton, M. J.; Moilliet, J. S. J. Chem. Soc., Perkin Trans.
1 1999, 803−810.
(47) Schuman, P. D.; Tarrant, P.; Warner, D. A.; Westmoreland, G.
U.S. Patent 3,954,758, 1971.
(48) Barton, D. H. R.; Hesse, R. H.; Toh, H. T.; Pechet, M. M. J. Org.
Chem. 1972, 37, 329−330.
(49) Chambers, R. D.; Holling, D.; Sandford, G.; Puschmann, H.;
Howard, J. A. K. J. Fluorine Chem. 2002, 117, 99−101.
(50) The glass frit was not stable and had to be replaced every 4 h. A
Teflon frit would be required for prolonged reaction.
(51) Abele, S., Meier, H.; Schmidt, G. Steiner, H. WO/2013/16087
A1, 2013.
(52) In aqueous HCl, the product 27 was soluble and not stable,
leading to partial demethylation.
(53) The software ChemProject was used.
(54) Mietchen, R.; Peters, D.; Rock, M. H. In Methods of Organic
Chemistry (Houben-Weyl), 4th ed.; Georg Thieme Verlag: Stuttgart,
Germany, 2000; Vol. E10a; Chapter 1.1.4., pp 98−100.
(55) http://web.ornl.gov/sci/physical_sciences_directorate/mst/
RSG/pdf/HF_Safety_1.pdf.
(56) Search for fluorine or hydrogen fluoride on: http://gestis-en.
itrust.de.
I
dx.doi.org/10.1021/op500100b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX