One-Step Synthesis of 2-Fluoroadenine Using Hydrogen Fluoride Pyridine in a Continuous Flow Operation

Article abstract of DOI:10.1021/acs.oprd.9b00178

We report the development of a one-pot synthesis of 2-fluoroadenine from an inexpensive 2,6-diaminopurine starting material using diazonium chemistry in a continuous fashion. Given the sensitivity of this transformation to temperature, we conducted critical experiments to study the exothermicity of the reaction and the heat removal, which were critical for the development of the process. Our goal was to improve the yield and purity of this pharmaceutical intermediate (2-fluoroadenine) and develop a more robust process.

Full text of DOI:10.1021/acs.oprd.9b00178

Article
pubs.acs.org/OPRD
Cite This: Org. Process Res. Dev. XXXX, XXX, XXX−XXX
One-Step Synthesis of 2‑Fluoroadenine Using Hydrogen Fluoride
Pyridine in a Continuous Flow Operation
†
Nastaran Salehi Marzijarani,*,†,§ David R. Snead,†,§ Jonathan P. McMullen,† Franco
̧ is Levesque,
́
Mark Weisel,† Richard J. Varsolona,† Yu-hong Lam,‡ Zhijian Liu,† and John R. Naber*,†
†Process Research & Development, MRL, Merck & Co. Inc., P.O. Box 2000, Rahway, New Jersey 07065, United States
‡Modeling and Informatics, Merck & Co. Inc., P.O. Box 2000, Rahway, New Jersey 07065, United States
S
* Supporting Information
ABSTRACT: We report the development of a one-pot synthesis of 2-fluoroadenine from an inexpensive 2,6-diaminopurine
starting material using diazonium chemistry in a continuous fashion. Given the sensitivity of this transformation to temperature,
we conducted critical experiments to study the exothermicity of the reaction and the heat removal, which were critical for the
development of the process. Our goal was to improve the yield and purity of this pharmaceutical intermediate (2-fluoroadenine)
and develop a more robust process.
KEYWORDS: flow chemistry, continuous, exothermic, fluorination, diazonium, nitrite, temperature-sensitive
encouraged us to explore the use of continuous flow processing
for this transformation.8c
INTRODUCTION
■
The selective introduction of a fluorine atom in biologically
active molecules, especially in nucleosides for interrupting
cancer cell or viral replication, has received significant attention
in recent years.1 The main driver for the incorporation of
fluorine is the similarity in size of fluorine and hydrogen atoms
and the higher biological and chemical stability of organo-
fluorine compounds.1a,2 The interest in 2-fluoroadenine
nucleosides is increasing, with cytotoxic 2-fluoroadenosine
and modified sugar analogues demonstrating resistance toward
deamination by the catabolic enzyme adenosine deaminase and
thus longer lifetimes in vivo.3 Many efficient synthetic
methodologies have been reported for the formation of sp2
C−F bonds, though many require a directing group, such as a
weak amide, or have a limited substrate scope.4 Many of the
more efficient approaches have not been applied to the
synthesis of 2-fluoroadenine because of the high polarity and
lack of solubility of this compound in most organic solvents.
The existing methods are lengthy, require protecting group
manipulations, or employ expensive starting materials. They
also suffer from low overall yield or require the use of
chromatographic purification because of low selectivity.5 The
most commonly reported approach for the synthesis of 2-
fluoroadenine is the nucleophilic fluorination of aryl diazonium
salts in tetrafluoroboric acid (the Balz−Schiemann reaction).6
However, the reactivity of this polar compound necessitates
the use of aqueous conditions because of its low solubility in
organic solvents, and the fluorination step has been shown to
be very challenging under these conditions with low yields,
mainly because of competition between fluorine addition and
formation of the appropriate HO-purine as a byproduct.7 To
avoid the use of water, HF/Pyridine has been shown to be an
effective fluorinating reagent.8 Moderate yields reported for the
synthesis of 2-fluoroadenine and scale-up challenges we
encountered with the use of HF/Pyridine in batch mode
Continuous flow systems are gaining popularity for the
synthesis of pharmaceutical ingredients and intermediates.9
Importantly, improved control of reaction parameters such as
heat and mass transfer, mixing, and residence time are some of
the benefits arising from continuous flow compared with
related batch processes, and these advantages can often lead to
improved process safety and robustness.10 Herein we present
the synthesis of 2-fluoroadenine in continuous flow, where a
thorough understanding of heat transfer and several rounds of
optimization led to an efficient synthesis in high yield with high
selectivity.
RESULTS AND DISCUSSION
■
Part 1: Reaction Optimization and Understanding of
Batch Operation. The initial exploration of the reaction in
batch mode focused on screening of temperature, nitrite
sources/equivalents, solvents, acid sources, and fluorine
sources. The starting material, 2,6-diaminopurine (1a), is
sparingly soluble (less than 2 mg/mL) in most organic
solvents, with a maximum room-temperature solubility of only
17 mg/mL in N,N-dimethylacetamide (DMAc). Most of the
conditions screened led to low reactivity. The highest
conversions were found in aqueous environments. However,
these conditions resulted in low yields due to trapping of the
diazonium salt intermediate with water (see Table S1 in the
Supporting Information (SI)). After several rounds of
optimization, we found that the highest yields and conversions
could be achieved using tert-butyl nitrite (tBuONO) in neat
hydrogen fluoride pyridine solution (HF/Pyridine). Further
Special Issue: Honoring 25 Years of the Buchwald-Hartwig
Amination
Received: April 26, 2019
© XXXX American Chemical Society
A
DOI: 10.1021/acs.oprd.9b00178
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Table 1. Optimization of the Synthesis of 2-Fluoroadenine (2a) in Batch Operation
a
entry
condition
scale (g)
Tj or Tr
presssure (psi)
addition time (min)
conv. (%)
2a (%)
2b (%)
yield (%)
1
2
A
A
A
A
A
B
0.8
4.0
4.0
4.0
4.0
Tj = 0 °C
Tj = 0 °C
Tr = 0 °C
Tr = 0 °C
Tr = 0 °C
Tr = −5 °C
−
−
8
8
45
45
45
25
99.6
78.1
97.7
96.7
97.1
99.2
88.5
62.1
90.0
85.5
88.2
82.4
8.8
0.8
3.5
3.7
4.0
−
−
74
68
69
72
b
3
40−350
40−45
0.5
b
4
5
b
bc
,
6
10.0
0.5
13.8
a
Tj is the jacket temperature, and Tr is the temperature inside the reaction vessel. A discussion of safety issues associated with handling HF/
b
c
Pyridine and how to minimize them is provided in the Supporting Information. A Multimax/Easymax system was used for Tr control. 1.1 equiv of
nitrite was used after rounds of optimization.
Figure 1. Lab-scale flow reactor setup.
optimization led to the reaction conditions shown in Table 1
(see Table S2 for more details). Selectivity for the reaction of
the amine at the 2-position in preference to the 6-position was
observed; however, overfluorination was observed in some
cases. In our initial optimization, we observed variability in
conversion when the reaction scale was increased in batch
(Table 1, entries 1 and 2). The conversion dropped
significantly, with 5−12% of 1a remaining unreacted with the
same nitrite addition time (8 min) when the scale was changed
from 800 mg to 4 g. This was explained by the temperature
that was measured inside the reaction vessel, which reached 35
°C. We discovered that both the conversion and the
percentage of overfluorinated impurity 2b can be impacted
significantly by the temperature and the rate of nitrite addition.
Although the overfluorinated impurity 2b could be easily
removed during isolation (precipitation of the reaction stream
with EtOAc), the reaction yield was directly impacted by
higher amounts of this impurity. Achieving a purity of more
than 98% was essential for our process, and our goal was to
develop a robust crystallization and to avoid chromatographic
purification. We were successful in this goal and discovered a
crystallization method that was optimized to improve the
purity of 2-fluoroadenine to more than 98% (see the SI). It was
found that complete rejection of starting material 1a during
this crystallization was difficult at conversions lower than 95%,
and residual 2,6-diaminopurine was the main impurity after
our crystallization process. As a result, the reaction was
optimized to target complete conversion while simultaneously
minimizing impurity 2b, as shown in Table 1, entries 3−6. We
found that the headspace pressure did not affect the
conversion, but we observed an increase in overfluorinated
byproduct when the pressure was reduced from 350 to 50 psi
(entries 3−5).
The high sensitivity of the reaction to temperature was
explained by the short half-life of tBuONO in HF/Pyridine at
higher temperatures. Subsequent investigation led to the
realization that the reaction is rapid, highly exothermic, and
sensitive to temperature rise. The active species to form
diazonium salts, NOF, which was observed in situ by Raman
spectroscopy at ∼2340 cm−1, was very unstable above ambient
temperatures and found to decompose within 1 h of incubation
(see the SI).11 Therefore, the longer incubation of tBuONO in
HF/Pyridine (liquid or vapor in the headspace) resulted in
higher levels of this decomposition and lower conversion. This
hypothesis was confirmed by a premixing experiment in both
temperature-controlled and temperature-uncontrolled experi-
ments (Figure S4 in the SI), in which the degradation of
tBuONO was faster at higher temperature. The lack of
robustness toward minor changes in the process led us to
explore the use of a continuous flow system to ensure better
heat transfer and ultimately achieve a more robust process.
Part 2: Reaction Optimization in Flow Operation.12
The study of the reaction in continuous flow was performed in
the reactor setup shown in Figure 1. A solution of 1a in HF/
Pyridine was prepared and loaded into a stainless steel (SS)
syringe, and neat tBuONO was loaded into a second SS
syringe. Solutions were pumped with a syringe pump at a total
flow rate of 0.2 mL/min. Perfluoroalkoxyalkane (PFA)
precooling loops (0.063″ o.d., 0.03″ i.d.), constructed to give
1 min of precooling residence time (tR), were made for each
stream and joined by a PEEK tee mixer (0.02″ i.d.). The length
of the PFA flow reactor loop (0.063″ o.d., 0.03″ i.d.) was
varied to adjust the tR for kinetic analysis. Pressure was
B
DOI: 10.1021/acs.oprd.9b00178
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
controlled with a 100−250 psi back-pressure regulator (BPR).
The precooling loops, reactor, and BPR were cooled to 0 °C in
an ice bath.
The initial results with a flow rate of 0.2 mL/min in a plug
flow reactor were encouraging, giving slight improvements to
the product distribution (Table 2). More in-depth examination
respectively, the tubing size was changed from 0.03″ i.d. to
0.063″ i.d.. This led to higher levels of overfluorinated impurity
2b and a drop in 2-fluoroadenine (2a) from 93.9% to 86.9%
(Table 3, entries 2 and 3). Examination of the system revealed
that the result was likely caused by an internal temperature rise
of the reaction solution due to poor heat transfer in the wider-
bore tubing. This hypothesis was tested at low flow rates by
increasing the reactor temperature from 0 °C to ambient
temperature while maintaining the precooling loop temper-
atures at 0 °C (entries 1 and 2), which showed the same trend
with lower conversion (93.5%) and higher formation of
overfluorinated impurity 2b (8.9%). Next, we reduced the
temperature of reactor L3 from 0 to −8 °C at a flow rate of 2
mL/min. Decreasing the L3 temperature helped deliver higher
reactivity and selectivity (98.7 conversion, 92.0% 2a), as shown
in Table 3, entries 3−5. Lowering the temperature below −8
°C led to freezing of the tBuONO stream and clogging. We
realized that the temperature and tube diameter in reactor L3
were more important than the precooling temperatures in L1
and L2. Using narrower tubing (0.032″ i.d.) for the precooling
loops L1 and L2 at a flow rate of 2 mL/min did not result in a
significant improvement in the selectivity (entry 6). Similarly,
when we used −5 °C for the precooling tubes and 0 °C for
reactor L3, we did not observe significant improvement
(entries 3 and 7). The next step in finding a way to reproduce
our best results at the 0.2 mL/min scale (Table 3, entry 2) at
higher flow rates was to better understand the root cause of
this temperature dependence by measuring the temperature
rise inside the tube at different scales.
Part 3: Understanding the Reaction Exotherm and Its
Impact on Selectivity in Flow Operation. We first
investigated the reaction exotherm and kinetics (see the SI).
Because of the hazardous, corrosive nature of the reaction and
the lack of equipment compatibility with HF/Pyridine, heats of
reaction for mono- and difluorination were estimated using
computational chemistry models. Those models indicated that
both transformations are exothermic, with the first fluorination
(−58.9 kcal/mol) being more exothermic than the second
fluorination (−50.1 kcal/mol) by 8.8 kcal/mol. The rate of the
first fluorination at position 2 was compared with the rate of
the second fluorination at position 6 by stepwise addition of
tBuONO (see the SI). We observed fluorination only at
position 6 when starting from 2-fluoroadenine, which also
proceeded at a much lower rate.
Table 2. Initial Optimization of the Flow Process (0.2 mL/
a
min)
equiv of
tBuONO
L3 tR
T
pressure
(psi)
conv.
(%)
2a
(%)
2b
(%)
entry
(min) (°C)
1
2
3
4
5
6
1.47
1.16
1.31
1.31
1.31
1.35
5.0
20.0
7.5
7.5
20.0
20.0
0
0
0
0
0
0
100
100
100
250
100
100
97.9
94.0
96.6
96.9
97.1
97.8
77.9
90.0
91.9
86.5
93.9
91.4
15.7
2.7
0.5
6.8
0.7
4.2
a
Flow optimizations on a small scale (0.2 mL/min). A residence time
(tR) of 1 min was used for L1 and L2 for all entries at 0 °C (all used a
0.02″ i.d. tee mixer).
of the amount of tBuONO, the residence time, and the
temperature demonstrated an ability to tune the conversion by
varying the amount of oxidant. A higher amount of oxidant
(1.47 equiv, entry 1) resulted in overfluorination (15.7%) and
a loss in yield. Reducing the amount of tBuONO to 1.16 equiv
led to low conversion (94%, entry 2), and extending the
residence time for longer aging did not improve the
conversion. Finally, we found the optimal amount of tBuONO
to be 1.31 equiv (entries 3−5). The appropriate tR was
determined in a semicontinuous fashion where the flow reactor
effluent was collected and analyzed for reaction conversion at
various time points with agitation at 0 °C. We discovered that
the reaction approaches 80% conversion very quickly, within 1
min upon mixing, but that the reaction requires 20 min to
achieve the targeted conversion. Repeating the flow experiment
with 1.31 equiv of tBuONO at 0 °C with a residence time of
20 min provided 97.1% conversion with a high selectivity of
93.9% (Table 2, entry 5). More details on high-throughput
screening of different variables are discussed in the SI.
Substituting PFA tubing for stainless steel led to improved
temperature control and slight improvements in the product
profile (see the SI). However, for simplicity of operation,
transparency, and monitoring inside the flow reactors, PFA
tubing was utilized for all the optimizations and scale-ups in
the next sections.
The adiabatic temperature rise was estimated to be between
45 and 60 °C, with the reaction reaching ∼80% completion
within 1 min. With so much heat released in a short period of
time, it was not surprising that overfluorination presented an
issue. Additionally, the temperature rise could not only
accelerate the decomposition of the active species (nitrosyl
The reaction was then scaled up from 0.2 to 2 mL/min with
the best conditions from the initial optimization. While the
temperature, pressure, amount of tBuONO, and residence time
were held constant at 0 °C, 100 psi, 1.31 equiv, and 20 min,
Table 3. Flow Scale-Up Challenges and Importance of Temperature
T (°C)
PFA tubing i.d.
entry
L1/L2
L3
flow rate (mL/min)
L1/L2
L3
conv. (%)
2a (%)
2b (%)
1
2
3
4
5
6
7
0
0
0
0
0
0
−5
25
0
0
−5
−8
0
0.2
0.2
2.0
2.0
2.0
2.0
2.0
0.02″
0.02″
0.02″
0.02″
93.5
97.1
97.7
98.1
98.7
97.9
97.8
79.9
93.9
86.9
89.8
92.0
89.6
88.5
8.9
0.7
7.8
6.5
5.6
6.5
6.9
0.063″
0.063″
0.063″
0.032″
0.063″
0.063″
0.063″
0.063″
0.063″
0.063″
0
C
DOI: 10.1021/acs.oprd.9b00178
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 2. Reaction temperature as a function of tube diameter and tR at a flow rate of 2 mL/min. Conditions: 1.31 equiv of tBuONO, 1a (0.67 M in
HF/Pyridine), Tj = 0 °C.
fluoride), affecting the conversion and selectivity, but also
generate more safety issues with handling of the diazonium
intermediate. Diazonium moieties are known to be highly
energetic, and detonations can be readily initiated by heat and
shock, especially with solid-state diazoniums. Therefore,
diazonium species often are not isolated but instead are
directly used in further reactions. Buildup of diazonium salt
was not a concern in our system, in which desired product was
formed with no isolation of the diazonium intermediate,
because of the high electrophilicity and rapid fluorination.
Additionally, several reports have shown that the continuous
flow can be used as a tool to mitigate safety concerns related to
diazonium salts.6b,13 We utilized flow chemistry to continu-
ously prepare our transient diazonium species and directly
couple it into a subsequent consuming step with an enhanced
safety profile. Next, we performed various experiments to find
the best conditions for obtaining the best heat control.
maximum temperature observed was 27 °C at T1 and T2 (tR =
3−6 s). Not only was the temperature higher, it also stayed
above 0 °C for a longer period of time (15 °C at T4), which
led to decomposition of the nitrite source and eventually to
lower conversions and lower selectivity.
Table 4 shows more details on the relationship between the
maximum temperature rise, tube diameter, and tube wall
Table 4. Maximum Temperature Rise as a Function of Tube
Diameter
tube
diameter
wall
thickness
surface area to volume
ratio
max T
(°C)
entry
1
2
3
0.063″
0.032″
0.010″
0.031″
0.016″
0.026″
9.9
21.0
62.0
27.0
9.5
1.8
While the large exotherm suggests that slow addition of the
nitrite may be advantageous, the short lifetime of the active
species complicates this process, and we observed decreased
conversion in our semibatch process. Subsurface feedings were
not explored, as it was expected that backflow into the feed
tube may pose a significant challenge. During our scale-up with
batch operations, we did observe low conversion and lack of
robustness when the tBuONO feeding tube was located above
the surface. This was mainly due to decomposition of the
oxidant at the tip of the dosing tube at longer dosing times
(along with solidification and freezing of the oxidant).
To obtain a reaction temperature profile for the 2 mL/min
scale reaction (Table 3), thermocouples were inserted into the
reactor tubing to measure the temperatures at various positions
(T1 to T4 in Figure 2) in the reactor loop (L3). Our goal was
to understand the temperature rise shortly after mixing (T1, 3 s
after mixing) and as the reaction stream continued through L3
(T2−T4, tR = 6, 12, and 24 s, respectively). The temperature
readings at these four locations were measured for two
experiments, one using 0.03″ i.d. tubing and another using
wider tubing (0.063″ i.d.). The difference in temperature
between the two configurations was striking. In the first system
with thinner tubing (0.03″ i.d.), the temperatures were never
observed to rise above 12 °C (3−24 s residence time). Within
3−5 s of mixing (T1), the reaction temperature rose to 10−12
°C and then quickly decreased to 4−5 °C prior to reaching T2
(tR = 6 s), and then at T4 (tR = 24 s) the reaction temperature
was close to zero. Next, we repeated the same experiment for
the second system with a larger tubing diameter (0.063″ i.d.)
and measured the temperatures at all four locations (T1−T4).
When the radial dimension was increased to 0.063″, the
thickness. Interestingly, for our system the use of tubing with
an even narrower channel width (0.01″ i.d., 0.026″ wall
thickness) resulted in the lowest rise in temperature (1.8 °C),
but the product distribution did not improve relative to the
baseline conditions (0.032″ i.d. tubing). This suggests that the
fluorination is tolerant toward some degree of warming. Our
results show a strong dependence of the maximum temper-
ature on the tube diameter. It is also well-known that for
exothermic reactions, the rate of heat transfer is highly
dependent on three factors: (1) the rate of heat generation by
the reaction, (2) the rate of heat removal to the reactor wall
through convection or conduction, and (3) the rate of heat
removal from the reactor wall to the surroundings. Our results
suggested that heat transfer through the solution was the
limiting factor, not heat transfer across the tube wall. By going
to larger-diameter tubing, the surface area to volume ratio
decreases, so a longer time for diffusion is required and the
heat transfer is less effective. Smaller-diameter tubing increases
the surface area to volume ratio and thus increases heat
transfer. The larger the surface area to volume ratio, the smaller
is the temperature rise at a flow rate of 2 mL/min.
Several strategies were explored to improve the reaction
upon scale-up, including lowering the temperature of the
cooling bath, applying a mix-then-reside approach, and/or
multiport addition of nitrite (Table 5). It was found that when
the temperature of the cooling medium was reduced to −8 °C,
the results were similar to those observed in thinner tubing.
This further confirmed that the internal temperature was the
critical variable impacting the ability to scale, as already
demonstrated in Table 3. Running at lower temperature
provided a solution at this scale; however, the strategy might
D
DOI: 10.1021/acs.oprd.9b00178
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
was successful at both the 2 and 10 mL/min scales, there were
concerns about the ability to scale-up this system to much
higher flow rates.
A more general approach might result from splitting the
addition of tBuONO into multiple portions to decrease the
energy released into the reaction stream at one time, and this
was tested next (Table 5). This strategy proved to be
successful for reactions run at both 2 and 10 mL/min
(Table 5, entries 4 and 6). A multiport system was constructed
with four nitrite streams using 0.063″ tubing for the reactor,
and experiments at 10 mL/min resulted in conversion and
selectivity identical to those obtained in thin tubing at 0.2 mL/
min (99.0% conversion, 90.8% selectivity). This approach is
viewed as scalable, where the number of sequential tBuONO
addition points is dependent upon the heat removal rates at
the desired reactor scale.
Figure 3. Successful scale-up to a flow rate of 20 mL/min on an
automated platform with multiport addition of tBuONO.
Further demonstrating the applicability of this concept, the
flow rate of the reaction was doubled to 20 mL/min, and
results identical to those obtained at 10 mL/min were
obtained. For improved safety, the reaction was performed
on an automated platform14 to enable hands-free operation in
a fully enclosed fume hood at higher reaction volumes (Figure
3). To this end, the syringe pump, peristaltic pump, and
sampling valve could all be controlled by computer. In this
way, the current reactor setup was deemed to have sufficient
throughput capability to meet a significant portion of potential
commercial demand as part of a small-volume program. The
overall process scheme is shown in Scheme 1. With the current
reaction setup and a flow rate of 20 mL/min, 120 g of crude 2-
fluoroadenine (2a) can be synthesized in 1 h. Next, the
collected stream mainly containing 2a and HF/Pyridine was
poured into ethyl acetate (25 volumes). The desired product
(crude) was precipitated with very small mother liquor loss,
and great rejection of overfluorinated impurity 2b and HF/
Pyridine reagent was observed. The crude solid was dissolved
in water and acetonitrile for a pH-swing crystallization using
sodium hydroxide and acetic acid in order to upgrade the
purity of 2a to more than 98% (for details, see the SI).
Table 5. Higher Flow Rates with Various Tubing Diameters
and/or Numbers of Oxidant Inlets
no. of
nitrite
inlets
flow rate
entry (mL/min)
conv.
(%)
2a
(%)
2b
(%)
PFA tubing i.d.
1
2
3
4
5
6
2.0
2.0
2.0
2.0
10.0
10.0
1
1
1
2
1
4
0.063″
0.032″
0.032″/0.063″
0.032″
0.032″/0.09″
0.063″
97.7
98.8
98.2
98.5
97.3
99.0
86.9
91.1
90.6
91.7
87.9
90.8
7.8
6.3
2.9
5.4
5.7
5.8
not be applicable to further increases in the radial dimension,
especially when scaling to significantly larger diameter tubing
with thicker walls. Another challenge with this approach is that
lower temperature (below 0 °C) increases the risk of tBuONO
freezing and clogging.
Next, we looked into a mix-then-reside approach to improve
the cooling process in the tubular reactor. We examined this
idea at a 2 mL/min scale by having a first reactor coil with a
narrow diameter and stainless steel construction for improved
heat transfer and a second coil with a wider bore for increased
residence time to allow for full reaction completion (Table 5,
entries 3 and 5). Mix-then-reside was successfully employed by
constructing a first reactor segment of 1/16″ stainless steel
tubing (4 mL) that then flowed into a 1/8″ PFA reactor (16
mL). This provided efficient heat transfer in the critical early
stages of the reaction (highest temperature rise) and a greater
volume for the reaction to reach completion in a larger
diameter tubing to reduce pressure drop. While this approach
CONCLUSION
■
We have reported the synthesis of 2-fluoroadenine from
commercially available and inexpensive 2,6-diaminopurine
through diazonium chemistry without prefunctionalization in
one step. We found that this reaction is rapid and exothermic
and that it is the root cause of the low conversion and
selectivity when the temperature is not well-controlled. By
comparison of results obtained in batch to those obtained in
flow at different scales, it was found that it was easier to control
Scheme 1. Overall Process Scheme
E
DOI: 10.1021/acs.oprd.9b00178
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
́
(3) (a) Robins, M. J.; Uznanski, B. Nucleic acid related compounds.
34. Non-aqueous diazotization with tert-butyl nitrite. Introduction of
fluorine, chlorine, and bromine at C-2 of purine nucleosides. Can. J.
Chem. 1981, 59 (17), 2608−2611. (b) Baer, H.-P.; Drummond, G. I.;
Duncan, E. L. Formation and Deamination of Adenosine by Cardiac
Muscle Enzymes. Mol. Pharmacol. 1966, 2 (1), 67−76.
the heat transfer with the continuous operation. Process
improvement was established by understanding the impact of
temperature, heat transfer, and residence time for this highly
exothermic and temperature-sensitive reaction using a tubular
reactor. Consistent and scalable results were delivered while
using less tBuONO (1.31 equiv in comparison with 1.40 equiv
in batch), and improvements observed in the 2a selectivity
translated to yields that were approximately 10% higher than
the batch analogues (82.4% isolated yield). The high selectivity
for 2-fluoroadenine achieved from this process helped us
deliver material that was >98% pure with a single crystallization
and avoided lengthy chromatographic purification. We
developed a safer process to synthesize an important
intermediate for a low-dose pharmaceutical drug by both
controlling the heat formation in the reaction and using an
automated platform to avoid operator contact with the reaction
stream. This platform not only allows for automated and
continuous operation but also enables the chemists to execute
more potentially hazardous transformations in a safe and
reliable fashion. We hope that this publication will serve as a
guide to some considerations that are important when
designing a continuous process for highly exothermic reactions
toward manufacturing route development.
(4) (a) Yamamoto, K.; Li, J.; Garber, J. A. O.; Rolfes, J. D.;
Boursalian, G. B.; Borghs, J. C.; Genicot, C.; Jacq, J.; van Gastel, M.;
Neese, F.; Ritter, T. Palladium-catalysed electrophilic aromatic C−H
fluorination. Nature 2018, 554, 511. (b) Campbell, M. G.; Ritter, T.
Modern Carbon−Fluorine Bond Forming Reactions for Aryl Fluoride
Synthesis. Chem. Rev. 2015, 115 (2), 612−633. (c) Zhang, Y.; Wen,
C.; Li, J. C5-Regioselective C−H fluorination of 8-aminoquinoline
amides and sulfonamides with Selectfluor under metal-free conditions.
Org. Biomol. Chem. 2018, 16 (11), 1912−1920. (d) Lin, A.; Huehls,
C. B.; Yang, J. Recent advances in C−H fluorination. Org. Chem.
Front. 2014, 1 (4), 434−438. (e) Grushin, V. V. The Organometallic
Fluorine Chemistry of Palladium and Rhodium: Studies toward
Aromatic Fluorination. Acc. Chem. Res. 2010, 43 (1), 160−171.
(f) Ning, X.-Q.; Lou, S.-J.; Mao, Y.-J.; Xu, Z.-Y.; Xu, D.-Q. Nitrate-
promoted Selective C−H Fluorination of Benzamides and Benzenea-
cetamides. Org. Lett. 2018, 20 (8), 2445−2448.
(5) (a) Wang, J.-j.; Sun, X.-y.; Chen, J.-b. New Method for the
Synthesis of 2-Fluoroadenine. Chin. J. Synth. Chem. 2007, 15 (4),
506−507. (b) Giner-Sorolla, A.; Burchenal, J. H. Substituted
hydroxylaminopurines and related derivatives. J. Med. Chem. 1971,
14 (9), 816−819.
(6) (a) Mohy El Dine, T.; Sadek, O.; Gras, E.; Perrin, D. M.
Expanding the Balz−Schiemann Reaction: Organotrifluoroborates
Serve as Competent Sources of Fluoride Ion for Fluoro-
Dediazoniation. Chem. - Eur. J. 2018, 24 (56), 14933−14937.
(b) Park, N. H.; Senter, T. J.; Buchwald, S. L. Rapid Synthesis of Aryl
Fluorides in Continuous Flow through the Balz−Schiemann Reaction.
Angew. Chem., Int. Ed. 2016, 55 (39), 11907−11911.
(7) Montgomery, J. A.; Hewson, K. Synthesis of Potential
Anticancer Agents. XX. 2-Fluoropurines2. J. Am. Chem. Soc. 1960,
82 (2), 463−468.
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.oprd.9b00178.
■
S
Remaining experimental procedures and a discussion of
the equipment used (PDF)
AUTHOR INFORMATION
Corresponding Authors
*E-mail: nastaran.salehi.marzijarani@merck.com.
*E-mail: john.naber@merck.com.
■
(8) (a) Eaton, C. N.; Denny, G. H. Synthesis of 2-fluoroadenine. J.
Org. Chem. 1969, 34 (3), 747−748. (b) Olah, G. A.; Welch, J. T.;
Vankar, Y. D.; Nojima, M.; Kerekes, I.; Olah, J. A. Synthetic methods
and reactions. 63. Pyridinium poly(hydrogen fluoride) (30% pyridine-
70% hydrogen fluoride): a convenient reagent for organic fluorination
reactions. J. Org. Chem. 1979, 44 (22), 3872−3881. (c) Saischek, G.
Process for the preparation of 2-fluoroadenine. US 2009/0163713 A1,
2009.
ORCID
Nastaran Salehi Marzijarani: 0000-0001-8940-4674
Jonathan P. McMullen: 0000-0001-5969-2396
́
Franco̧ is Levesque: 0000-0001-9529-4993
Zhijian Liu: 0000-0002-5750-9890
(9) (a) McMullen, J. P.; Marton, C. H.; Sherry, B. D.; Spencer, G.;
Kukura, J.; Eyke, N. S. Development and Scale-Up of a Continuous
Reaction for Production of an Active Pharmaceutical Ingredient
Intermediate. Org. Process Res. Dev. 2018, 22 (9), 1208−1213.
(b) Gutmann, B.; Cantillo, D.; Kappe, C. O. Continuous-Flow
TechnologyA Tool for the Safe Manufacturing of Active
Pharmaceutical Ingredients. Angew. Chem., Int. Ed. 2015, 54 (23),
6688−6728. (c) Baumann, M.; Baxendale, I. R. The synthesis of
active pharmaceutical ingredients (APIs) using continuous flow
chemistry. Beilstein J. Org. Chem. 2015, 11, 1194−1219. (d) Cole,
K. P.; Groh, J. M.; Johnson, M. D.; Burcham, C. L.; Campbell, B. M.;
Diseroad, W. D.; Heller, M. R.; Howell, J. R.; Kallman, N. J.; Koenig,
T. M.; May, S. A.; Miller, R. D.; Mitchell, D.; Myers, D. P.; Myers, S.
S.; Phillips, J. L.; Polster, C. S.; White, T. D.; Cashman, J.; Hurley, D.;
Moylan, R.; Sheehan, P.; Spencer, R. D.; Desmond, K.; Desmond, P.;
Gowran, O. Kilogram-scale prexasertib monolactate monohydrate
synthesis under continuous-flow CGMP conditions. Science 2017, 356
(6343), 1144.
John R. Naber: 0000-0002-4390-3467
Author Contributions
§N.S.M. and D.R.S. contributed equally.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We acknowledge Simon Hamilton and David Waterhouse for
their analytical support, George Zhou and Zachary Dance for
their help with Raman spectroscopy, Ryan Cohen for NMR
analysis, Thomas Vickery for DSC measurements, Shane
Grosser for help with automation, and Benjamin Sherry for all
the support throughout this project.
REFERENCES
■
(10) (a) Plutschack, M. B.; Pieber, B.; Gilmore, K.; Seeberger, P. H.
The Hitchhiker’s Guide to Flow Chemistry. Chem. Rev. 2017, 117
(18), 11796−11893. (b) Hartman, R. L.; McMullen, J. P.; Jensen, K.
F. Deciding Whether To Go with the Flow: Evaluating the Merits of
Flow Reactors for Synthesis. Angew. Chem., Int. Ed. 2011, 50 (33),
7502−7519. (c) Yoshida, J.-i.; Takahashi, Y.; Nagaki, A. Flash
(1) (a) Liu, P.; Sharon, A.; Chu, C. K. Fluorinated nucleosides:
Synthesis and biological implication. J. Fluorine Chem. 2008, 129 (9),
́
743−766. (b) Wojtowicz-Rajchel, H. Synthesis and applications of
fluorinated nucleoside analogues. J. Fluorine Chem. 2012, 143, 11−48.
(2) Wilkinson, J. A. Recent advances in the selective formation of the
carbon-fluorine bond. Chem. Rev. 1992, 92 (4), 505−519.
F
DOI: 10.1021/acs.oprd.9b00178
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
chemistry: flow chemistry that cannot be done in batch. Chem.
́
Commun. 2013, 49 (85), 9896−9904. (d) Kockmann, N.; Thenee, P.;
̈
Fleischer-Trebes, C.; Laudadio, G.; Noel, T. Safety assessment in
development and operation of modular continuous-flow processes.
Reaction Chemistry & Engineering 2017, 2 (3), 258−280.
̈
̈
(11) (a) Ruff, O.; Stauber, K. Uber das Nitrosylfluorid (NOF). Z.
Anorg. Chem. 1905, 47 (1), 190−202. (b) Flores, A. L.; Darwent, B. d.
Photochemical decomposition of nitrosyl fluoride. J. Phys. Chem.
1969, 73 (7), 2203−2208. (c) Addison, C. C.; Lewis, J. The
chemistry of the nitrosyl group (NO). Q. Rev., Chem. Soc. 1955, 9 (2),
115−149.
(12) Britton, J.; Jamison, T. F. The assembly and use of continuous
flow systems for chemical synthesis. Nat. Protoc. 2017, 12, 2423.
(13) (a) Deadman, B. J.; Collins, S. G.; Maguire, A. R. Taming
Hazardous Chemistry in Flow: The Continuous Processing of Diazo
and Diazonium Compounds. Chem. - Eur. J. 2015, 21 (6), 2298−
2308. (b) Hu, T.; Baxendale, I. R.; Baumann, M. Exploring Flow
Procedures for Diazonium Formation. Molecules 2016, 21 (7), 918.
(14) Ley, S. V.; Fitzpatrick, D. E.; Ingham, R. J.; Myers, R. M.
Organic Synthesis: March of the Machines. Angew. Chem., Int. Ed.
2015, 54 (11), 3449−3464.
G
DOI: 10.1021/acs.oprd.9b00178
Org. Process Res. Dev. XXXX, XXX, XXX−XXX