Process development and manufacture of potassium 2-fluoro-6-hydroxyphenyltrifluoroborate

Article abstract of DOI:10.1016/j.tet.2019.05.063

The development of a phase-appropriate manufacturing-scale synthesis of potassium 2-fluoro-6-hydroxyphenyltrifluoroborate was achieved. Investigations into improving the yield and robustness indicated that pH of the reaction medium is a critical process parameter. Additional development resulted in replacing tartaric acid with citric acid, resulting in improved process robustness and enabling scale-up to >10 kg.

Full text of DOI:10.1016/j.tet.2019.05.063

Accepted Manuscript
Process development and manufacture of potassium 2-fluoro-6-
hydroxyphenyltrifluoroborate
Lokesh Pawar, Ramesh Jayaramaiah, Baburaj Krishnan, Athimoolam
Arunachalampillai, Ying Chen, Andrew T. Parsons, Jo Anna Robinson, Jason S.
Tedrow
PII:
S0040-4020(19)30625-8
https://doi.org/10.1016/j.tet.2019.05.063
TET 30384
DOI:
Reference:
To appear in: Tetrahedron
Received Date: 25 April 2019
Revised Date: 24 May 2019
Accepted Date: 29 May 2019
Please cite this article as: Pawar L, Jayaramaiah R, Krishnan B, Arunachalampillai A, Chen Y, Parsons
AT, Robinson JA, Tedrow JS, Process development and manufacture of potassium 2-fluoro-6-
hydroxyphenyltrifluoroborate, Tetrahedron (2019), doi: https://doi.org/10.1016/j.tet.2019.05.063.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
Process Development and Manufacture of Potassium 2-
Fluoro-6-hydroxyphenyltrifluoroborate
Lokesh Pawar,a Ramesh Jayaramaiah,a Baburaj Krishnan,a Athimoolam Arunachalampillai, a*
Ying Chen,b Andrew T. Parsons,b Jo Anna Robinson,b Jason S. Tedrowb
aSyngene International Ltd. Biocon Park, Plot 2 & 3, Jigani Link Road, Bangalore 560099
bAmgen, Inc., 360 Binney Street, Cambridge, MA 02139, United States
Abstract:
The development of a phase-appropriate manufacturing-scale synthesis of potassium 2-
fluoro-6-hydroxyphenyltrifluoroborate was achieved. Investigations into improving the yield
and robustness indicated that pH of the reaction medium is a critical process parameter.
Additional development resulted in replacing tartaric acid with citric acid, resulting in
improved process robustness and enabling scale-up to >10 kg.
1

ACCEPTED MANUSCRIPT
Introduction
Since the discovery of the Suzuki-Miyaura coupling (SMC),1,2 significant research
efforts have focused on designing catalyst systems that improve the efficiency of coupling
unstable boronic acids3 (typically heterocyclic4 or polyfluorinated5 arenes). Further
improvements to the SMC were realized with the development of stable boron reagents,6
which are resistant to non-productive decomposition during the reaction7. Over the past two
decades, potassium organotrifluoroborate (BF3K) salts have emerged as an attractive
alternative to boronic acid/ester reagents due to their improved stability and high performance
in the SMC.8,9 BF3K salts are typically prepared by the reaction of the corresponding boronic
acid or esters with 4-6 equivalents of KHF2.1011 One disadvantage of this method is that the
reaction conditions are corrosive and can result in reactor etching. Though this method is
suitable for lab scale synthesis, to the best of our knowledge there have been no reports of
large-scale batch production.12 Recently, Lennox and Lloyd-Jones13 reported a protocol for
conversion of boronic acids to BF3K salts using potassium fluoride and acetic/tartaric acid,
resulting in improved processability and throughput of potassium organotrifluoroborate salt
preparation. Scheme 1 depicts the mechanism of trifluoroborate generation using this
protocol (Scheme 1).
Scheme 1: Mechanistic proposal for the preparation of aryl trifluoroborates from boronic
acids using KF/acid (adapted from reference)13
Herein we report the development and implementation of a modified Lennox/Lloyd-
Jones protocol for the manufacture of potassium 2-fluoro-5-hydroxyphenyltrifluoroborate 2.
The present studies involve a discussion of process development activities and investigations
aimed towards improving the yield and robustness to enable implementation on multi-
kilogram production scale.
2

ACCEPTED MANUSCRIPT
Scheme 2: Synthesis of potassium 2-fluoro-5-hydroxyphenyltrifluoroborate (2) using tartaric
acid
Results and Discussion
The discovery preparation of 2 utilized tartaric acid (2.5 equiv) and potassium
fluoride (4.0 equiv) in 39 L/kg of solvent (CH3CN:THF:H2O = 50:25:3). Upon reaction
completion, the mixture was filtered and concentrated to a minimum volume under vacuum at
< 30 °C followed by crystallization from isopropanol. After a 2nd crop crystallization, 2 was
isolated in 80% yield and 70% wt/wt assay by quantitative 1H NMR.
Due to the aggressive timeline for this program, the initial multi-kilo delivery
implemented a modified discovery process that was optimized to reduce solvent volumes.
The overall reaction volume was decreased by reducing the acetonitrile charge from 25 to 15
L/kg (29 overall L/kg 1). A final reslurry in isopropanol was also implemented to improve
product potency. Using this protocol, multiple < 1 kg batches were executed to deliver 4 kg
of 2, with yields ranging between 65-70% and wt/wt potencies of > 90%. Further reduction
of solvent quantities resulted in generation of an oily reaction mass instead of the desired
free-flowing slurry. Furthermore, attempts to use fewer equivalents of tartaric acid and
potassium fluoride resulted in incomplete conversion. With additional material required to
support the program, efforts were focused on further development and characterization to
improve process robustness for larger scale deliveries. The scale increase also prompted
safety and thermal hazard evaluations prior to manufacture on pilot plant scale.14 For
subsequent batches, a Hastelloy reactor was utilized due to etching that occurred during when
using glass reactors, suggesting the generation of hydrofluoric acid occurs during the process.
Two batches were conducted with 2.0 kg and 2.8 kg input of 2-fluoro-6-hydroxyboronic acid.
The first batch yielded 1.88 kg (65%) of 2 with an assay (potency) of 94 wt% by qNMR,
which is within the expected range. However, the yield of the second batch (2.8 kg scale) was
only 1.5 kg with a potency of 86 wt%. An additional purification (methanol reslurry) was
implemented to meet the specifications, resulting in additional liquor loss and a final yield of
3

ACCEPTED MANUSCRIPT
1.1 kg (31.5%). Analysis of the mother liquor indicated the presence of 3-fluorophenol,
suggesting decomposition of the starting material or product had occurred. Additional
optimization resulted in increased lab-scale yield to > 80%, however similar results on > 2.0
kg scale were observed, prompting a more detailed investigation.
Investigations on the low yield batches
To achieve a better understanding of the process and parameters that may be resulting in
3-fluorophenol generation, several different factors were considered:
1. pH and water content: The pH of the reaction mixture was monitored throughout the
process during lab and production batches. It was observed that the pH of production
batches were more acidic, ranging from 1.5–2.5 while development batches ranged
from 4–5.5. It was also observed that water content in the mother liquor from
production batches were also higher than the lab batches.
2. Equipment: The reactor used for these batches was 600 L Hastelloy vessel where the
occupancy was only 16% during the concentration of the filtrate. Occupancy during
development batches was always >70% during this operation.
3. Addition rates: Total time for the tartaric acid addition during production was 3.5
times longer than in lab batches
With the information from the investigation report, downscaling experiments were performed
to mimic the exact plant conditions (e.g. longer addition times, lower pH) to identify the root
cause for 3-fluorophenol generation observed during the 2.8 kg batch.
4

ACCEPTED MANUSCRIPT
Mitigation strategy for the pilot plant-batches
The reaction mixture containing 1 and potassium fluoride is basic, with the reaction only
proceeding upon acidification via tartaric acid addition. To better understand the impact of
pH on the stability of the reaction mixture, several experiments were carried out at three
different pH values (1.9 and 3.6 and 7) and 3-fluorophenol content was monitored (
Table 1). These experiments suggest that high acidity of the reaction mixture can lead
to decomposition of the product or starting material into 3-fluorophenol.
Table 1 Stability studies on the filtrate from reaction mixture at various pH
3-Fluorophenol content
Entry
Process changes
pH
4 h
7 h
pH adjusted by adding additional
tartaric acid in THF
1
2
1.9
3.6
6.5%
25.1%
No changes
3.1%
3.3%
3.2%
3.2%
pH adjusted by adding additional
KF in water
3
7.3
Subsequent studies aimed to determine the impact of tartaric acid solution addition
rate on the pH of the reaction mixture (Table 2). Two separate small-scale experiments were
run with slow (4 mL/min) and fast (12 mL/min) addition rates of tartaric acid solution. It was
observed that slow addition resulted in a greater acidity of the reaction mixture (pH = 2.7)
compared to fast addition, (pH = 5.9). This is consistent with the high acidity observed
during the production runs in which the tartaric acid solution was added at a rate
corresponding to 0.85 mL/min. These studies suggest that high acidity of the reaction
mixture during production was caused by long tartaric acid addition times, resulting in 3-
fluorophenol generation via protodeboronation. While it is unclear why addition rate affects
the reaction pH, we postulate that this may be due to different tartrate salt speciation (mono-
vs dipotassium tartrate) and their rate of formation.
Table 2. Effect of tartaric acid addition rate on pH
5

ACCEPTED MANUSCRIPT
Entry
Addition rate
4 mL/min
pH of reaction mixture
1
2
3a
2.7
5.9
1.5
12 mL/min
0.85 mL/min
aProduction-scale run
To provide additional control over the process and reduce mother liquor losses during
the final crystallization, focused solubility studies were performed on the trifluoroborate
product 2 (
Table 3). This data was used to determine an in-process control testing limit to
determine an appropriate endpoint for the solvent switch into isopropanol and water content
target. While water content had little impact on the solubility in pure isopropanol, the product
has high solubility in THF/acetonitrile mixtures (Table 3, entry 5), which nearly doubles in
the presence of water (Table 3, entry 6). Based on this data, reduction of THF and acetonitrile
content is critical to avoid high liquor losses.
Table 3 Solubility studies of 2-fluoro-5-hydroxyphenyltrifluoroborate (2)
Entry
Solvent system
THF
Concentration (mg/mL)
1
2
3
4
5
6
7
8
9
30.2
Not detected
15.8
IPA
MeOH
Acetonitrile
204.1
176.6
351.3
3.0
THF:CH3CN (12:15)
THF:CH3CN:H2O (12:15:2)
4% water in IPA
8% water in IPA
12% water in IPA
1.6
2.6
Based on the above studies, the following process controls were implemented to improve the
robustness and yield of 2:
1. pH of process stream: Maintain a pH of 4 – 6 throughout the process by implementing
i) an increased addition rate of tartaric acid; ii) continuous pH monitoring throughout
6

ACCEPTED MANUSCRIPT
the process; and iii) Optional KF charge after the tartaric acid addition if the pH is <
4.
1
2. Crystallization solvent composition: control acetonitrile and THF to < 5% by H
NMR prior to product isolation
Through implementation of the above improvements, multiple >5 kg batches were executed
to consistently provide yields of >85% with >97 wt% potency.
Citric Acid Process
Prior to production of subsequent batches of 2, additional development activities
focused on improving the trifluoroborate conversion process. Due to the pH dependence on
the final yield of 2, we sought to determine if an alternative acid would provide a more robust
process. After screening numerous acids, citric acid was identified as a potential replacement
for tartaric acid (Scheme 3). Addition rate experiments were performed using citric acid and
found no effect on pH was found (Entry 1 and 2, Table 4). Both compounds 1 and 2 were
found to be stable in the citric acid solution in acetonitrile during the prolonged stirring for 24
h. Addition of potassium fluoride solution in water to the pre-mixed solution of citric acid
and 2 was also found to be within the desired pH range.
Table 4. Effect of addition rate on pH in the citric acid process on lab scale
Entry
Addition details
8 mL/min of citric acid
pH of reaction mixture
1
2
5.5
5.4
30 mL/min of citric acid
6.7 mL/min of KF solution in water to a
mixture of 1 and citric acid
3a
5.4
a Condition followed in the production batches
The use of citric acid provided the following advantages: i) citric acid is soluble in
acetonitrile, allowing for the elimination of THF from the process and reduction of reaction
volume to 13 L/kg; and ii) the pH of the reaction mixture is not dependent on the addition
rate of citric acid solution. The improved citric acid process was executed on multiple 14 kg
batches, providing 90-95% yield of 2 with an assay of > 97 wt%.
7

ACCEPTED MANUSCRIPT
Scheme 3 Synthesis of potassium 2-fluoro-5-hydroxyphenyltrifluoroborate (2) using citric
acid
Conclusion
Herein we described the development of a scalable process to produce 2-fluoro-6-
hydroxyphenyltrifluoroborate from its corresponding boronic acid using a modified
Lennox/Lloyd-Jones protocol. We identified the dependence of pH on tartaric acid addition
rate, leading to the implementation of appropriate pH control, resulting in improved yields.
Subsequent development resulted in the implementation of citric acid as a replacement for
tartaric acid, which provided improved robustness by eliminating the pH-dependence on acid
addition rate while simplifying the process. The improved process has been demonstrated on
> 10 kg scale, providing high yield and potency of the desired trifluoroborate product 2.
Acknowledgements
This paper is dedicated to Professors Stephen L. Buchwald and John F. Hartwig to
commemorate the 25th anniversary of the Buchwald-Hartwig coupling. We thank Dr. Nilesh
Joshi, Krishna Sheena, Rafi Shaik (Syngene Amgen Research Center) and Ramesh Sela
(Syngene International Ltd.) for providing technical support during our development. We
also thank Dr. Narendra Ambhaikar (Syngene International Ltd) and Dr. Debasis Patra
(Amgen, Inc.) for their continued support during this project.
Experimental Section
General Considerations:
The starting material boronic acid 1 was procured from Wuxi AppTech. Tartaric acid was
procured from Kaival chemicals. Potassium fluoride was procured from Chempure. Citric
acid was procured from Sony Chem. Unless otherwise stated all the chemicals were used
1
without further purification. All reaction solvents employed were of commercial grade. H
19
11
and F NMR and B NMR spectra were recorded on Bruker AV3-400 (400/376.69/128.51
MHz) spectrometer using the DMSO-d6 as an internal standard. The values of chemical shifts
8

ACCEPTED MANUSCRIPT
are reported in parts per million (ppm) with the multiplicities of the spectra reported as
follows: singlet (s), doublet (d), triplet (t), quartet (q), multiplet (m) and broad (br), values for
coupling constants (J) are assigned in Hz. Residual solvent analysis was recorded using
Agilent GC 6890N/7890A series equipped with Head space injection auto sampler. The
potassium content was analysed using Agilent (Varian) L720ES instrument. Moisture
content was measured using METROHM 907 Titrando (Karl-Fischer titration). Assay purity
was measured by quantitative NMR (qNMR) assay using benzyl benzoate as internal
standard (procured from Sigma Aldrich).
Tartaric acid procedure:
Lab scale procedure:
To a 4-neck 3 L round-bottom flask (RBF) equipped with thermo pocket and mechanical
stirrer was charged (2-fluoro-6-hydroxyphenyl)boronic acid (350.0 g, 2.25 mol) and
acetonitrile in (8750 mL, 25 L/kg). To the resulting reaction mixture was added a solution of
potassium fluoride in water (522 g, 8.98 mol in 700.0 mL, 2.0 L/kg) at 22-25 °C followed by
dropwise addition of (2R,3R)-2,3-dihydroxysuccinic acid (842 g, 5.61 mol) in
tetrahydrofuran (4375 mL, 12.5 L/kg) and water (700 mL, 2 L/kg) over a period of 30 min.
The reaction mixture was stirred for 16 h at 22-25 °C. After completion of the reaction is
indicated by HPLC, the reaction mixture was filtered through celite, washing the bed with
THF (5 L/kg). The filtrate was concentrated in vacuo (50 mm Hg) by rotavap (bath temp <
30° C) to approx. 3-4 total volumes (1050-1400 mL). Isopropanol (3500 mL, 10 L/kg) is
added and concentrated to approx. 9-10 total volumes (3.0-3.5 L). The resulting slurry was
cooled to 0-5 °C and stirred for 2 h. The product was isolated by filtration and washed with
isopropanol (2 L/kg). The resulting solid was then dried under vacuum for 4-5 h to afford the
product as off-white crystalline solid. Yield: 378.0 g (77%)
Pilot Plant scale procedure consisted of three parts:
Part 1: Tartaric acid solution preparation: To a clean and dry 500 L Hastelloy reactor was
charged tetrahydrofuran (72 L, 12 L/kg) and tartaric acid (14.44 kg, 96.2 mol). The resulting
mixture was stirred for 10 min at 22-25°C. The temperature was increased to 45-47 °C and
aged for 1-2 h. The reaction mass was cooled to 22-28 °C and transferred to a high-density
polyethylene (HDPE) container for storage, rinsing the reactor with additional THF (3.0 L,
0.5 L/kg).
9

ACCEPTED MANUSCRIPT
Part-2: Preparation of Potassium fluoride solution: To a HDPE (80 L) container
potassium fluoride (8.94 kg, 153.93 mol) and water (12.0 L, 2 L/kg) was charged. The
resulting mixture was agitated until complete dissolution (10-20 min).
Part-3 Preparation 2-Fluoro-6-hydroxyphenyltrifluoroborate: To a clean and dry 500 L
Hastelloy reactor was charged acetonitrile (90 L, 15.0 L/kg) and (2-fluoro-6-hydroxyphenyl)
boronic acid (6.0 kg, 38.48 mol). The resulting slurry was stirred for stirred for 5-10 min at
22-28 °C. To this reaction mixture was charged the potassium fluoride solution over 2-5-min.
(The reaction mixture initially becomes a solution, but precipitation occurs upon complete
addition). The reaction mixture was stirred for 30 min at 25-28 °C and then the tartaric acid
solution was added to the reaction mass over a period of 10-12 mins and stirred for an
additional 30 min. The progress of the reaction was monitored by TLC. The pH of the
reaction mixture was monitored by diluting 2 mL with 2 mL of water. A pH of 2.3 (Limit ≤
5.5) indicated additional potassium fluoride was required. A potassium fluoride solution (0.24
g was dissolved in 0.32 L water) was charged to adjust the pH between 5.5-6.5. Once the pH
was within the desired range, the reaction mass was filtered through celite, washing with THF
(12.0 L, 2.0 vol). The combined filtrates were polish filtered through 5.0 µm stainless steel
and 0.2 µm polypropylene cartridge into a clean reactor. The solution was concentrated
under vacuum (30-50 mm Hg) to approx. 10 total volumes (60 L) at 20-28 °C (< 30 °C).
Isopropanol (60 L, 10 L/kg) was added to the reactor and re-distilled to 10 total volumes (60
L). This process was repeated three time until water content reached the desired limit of
<12%. The resulting slurry was cooled to 15±5°C and stirred for 1 h. The product was
isolated by filtration, washing with isopropanol (2 x 15 L). The product was vacuum dried at
< 30 °C to afford the product as off-white crystalline solid. Yield: 7.02 kg (83.6%).
Citric acid procedure:
Procedure: To an 80 L HDPE container was charged water (28 L, 2 L/kg) and potassium
fluoride (21 kg, 361.57 mol). The mixture was stirred until a homogenous solution is
obtained (approx. 15 min). To a 500 L Hastelloy reactor, acetonitrile (280 L, 10 L/kg), 2-
fluoro-6-hydroxyphenyl boronic acid (14 kg, 89.79 mol), and citric acid (29.4 kg, 134.68
mol) was charged. The charging funnel was rinsed with additional acetonitrile (14 L, 1 L/kg)
°
and the resulting reaction mixture was stirred for 15-20 min at 25-28 C. The premade
solution of potassium fluoride was then charged over a period of 60 min, maintaining the
batch temperature between 25-30 °C. The reaction mixture is monitored for conversion by
TLC. Upon completion (approx. 10 h), the mixture was filtered through celite, rinsing the
10

ACCEPTED MANUSCRIPT
reactor and filter with acetonitrile (48 L, 3.43 L/kg). The filtrate was charged back into the
500 L Hastelloy reactor. Isopropanol (56 L, 4 L/kg) was added and the mixture was
concentrated under vacuum (30-50 mm Hg) at 30-35 °C to 5-8 total volumes (70-112 L). This
process was repeated an additional 2-3 times until acetonitrile content was < 5% and water
content was < 6%. The resulting product was isolated on a Nutsche filter. The reactor and
filter cake were washed with isopropanol (28 L, 2 L/kg). The resulting product was vacuum
dried at < 30 °C to obtain the product as off-white crystalline solid. Yield: 18.9 kg (96.5%).
Analytical data
1H NMR (400 MHz, DMSO-d6): δ 8.08 (q, J = 14.36 Hz, 1H), 6.95 (q, J = 7.92 Hz, 1H),
6.36 (q, J = 8.60 Hz, 2H). (J= H-F coupling constants):19F NMR (376.69 MHz, DMSO--d6): δ
-105.33 (s), -130.0 (q), -148.3 (s) (KHF2 contaminant).11B NMR (128.51 MHz, DMSO-d6): δ
3.051(br s). Moisture Content (KF titration) = 0.17%, Heavy metal content (ICP-OES:
Dilution method) = Other metals (Pd. Hg, As, Pb and Sb) < 5 ppm, potassium content
=17.18%
References
1 N. Miyaura and A. Suzuki, Chem. Rev. 1995, 95, 2457–2483.
2 Martin, R.; Buchwald, S. L. Acc. Chem. Res. 2008, 41, 1461–1473.
3 Kinzel, T.; Zhang, Y.; Buchwald, S. L. J. Am. Chem. Soc. 2010, 132, 14073–14075.
4 Cox, P. A.; Leach, A. G.; Campbell, A. D.; Lloyd-Jones, G. C. J. Am. Chem. Soc. 2016,
138, 9145–9157.
5 Cox, P. A.; Reid, M.; Leach, A. G.; Campbell, A. D.; King, E. J.; Lloyd-Jones, G. C. J. Am.
Chem. Soc. 2017, 139, 13156–13165.
6 Lennox, A. J. J.; Lloyd-Jones, G. C. Chem. Soc. Rev. 2014, 43, 412-443.
7 Miyaura, N.; Yamada, K.; Suzuki, A. Tetrahedron Letters. 1979, 20, 3437–3440.
8 Darses, S.; Genet, J.-P. Chem. Rev. 2008, 108, 288–325.
9 Molander, G. A.; Sandrock, D. L. Curr. Opin. Drug Discov. Devel. 2009, 12, 811–823.
10 Molander, G. A.; Jean-Gerard, L. Org. React. 2013,79, 1-316.
11 Lennox, A. J. J. Organotrifluoroborate Preparation, Coupling and Hydrolysis, Springer
Theses, DOI: 10.1007/978-3-319-01134-9_2.
12 Broom, T.; Hughes, M.; Szczepankiewicz, B. G.; Ace, K.; Hagger, B.; Lacking, G.; Chima,
R.; Marchbank, G.; Alford, G.; Evans, P.; Cunningham, C.; Roberts, J. C.; Perni, R. B.;
Berry, A.; Rutter, A.; Watson, A. A. Org. Process Res. Dev. 2014, 18, 1354−1359.
13 Lennox, A. J. J.; Lloyd-Jones, G. C. Angew. Chem. Int. Ed. 2012, 51, 9385–9388
14 Process Hazard Identification and Risk Assessment: Basic recommendation for engineering
controls such as vacuum and temperature requirement for the distillation were put in place.
DSC studies: No exothermic decomposition was observed in the analysis of the boronic acid
starting material 1; a moderate exotherm at 191.5°C was observed for the product
11

ACCEPTED MANUSCRIPT
trifluoroborate salt 2; and low exotherm for the reaction mixture at 166.5°C. These
exothermic events were deemed low risk since the operating temperature throughout the
process is < 30 °C. Compatibility test for reactor material of construction (coupon test):
Moderate etching of stainless steel and glass and no etching of the Hastelloy. This suggests
hydrofluoric acid is generated during the process. Filter cake resistance studies: Moderate
specific cake resistance on the order of e10 ft/lb.
12