Flow Chemistry on Multigram Scale: Continuous Synthesis of Boronic Acids within 1 s

Article abstract of DOI:10.1021/acs.orglett.6b01681

The benefits and limitations of a simple continuous flow setup for handling and performing of organolithium chemistry on the multigram scale is described. The developed metalation platform embodies a valuable complement to existing methodologies, as it combines the benefits of Flash Chemistry (chemical synthesis on a time scale of <1 s) with remarkable throughput (g/min) while mitigating the risk of blockages.

Full text of DOI:10.1021/acs.orglett.6b01681

Letter
pubs.acs.org/OrgLett
Flow Chemistry on Multigram Scale: Continuous Synthesis of Boronic
Acids within 1 s
Andreas Hafner, Mark Meisenbach, and Joerg Sedelmeier*
Novartis Campus, Novartis Pharma AG, 4002 Basel, Switzerland
S
* Supporting Information
ABSTRACT: The benefits and limitations of a simple continuous flow setup for handling and performing of organolithium
chemistry on the multigram scale is described. The developed metalation platform embodies a valuable complement to existing
methodologies, as it combines the benefits of Flash Chemistry (chemical synthesis on a time scale of <1 s) with remarkable
throughput (g/min) while mitigating the risk of blockages.
thereby increasing the half-life of an aryl lithium intermediate
ontinuous manufacturing (CM) as a technical innovation
Caims to complement traditional batch operations by allowing for extended residence times (up to minutes) prior to
an electrophilic quench (Figure 1a).2 In this case, mixing
opening access to chemistries not easily accomplished in
efficiency plays a subordinate role and residence time (reaction
time) is optimized to maximize the conversion of an aryl halide
to the corresponding aryl lithium intermediate. A typical
temperature profile (Figure 1a) for this approach starts at low
temperature, e.g. −78 °C, followed by a first exotherm for the
Hal/Li exchange, relaxation to −78 °C prior to a second
exotherm for the electrophilic quench reaction, and relaxation
to the initial starting temperature. This flow regime has several
advantages such as an excellent temperature control and a
reduced risk for clogging due to the wider reactor dimensions
(compared to microstructured devices) and can be successfully
applied in case the newly generated aryl lithium species is
sufficiently stable at a given temperature and residence time
prior to a desired quench.
standard batch vessels, such as the handling of highly reactive
organolithium reagents and the generation and consumption of
unstable intermediates.1 Flow technology permits very good
temperature control of low temperature transformations and
rapid mixing of organolithium reagents and substrates, all
examples of potential benefits over traditional batch mode.
When performing organometallic chemistry in continuous flow
mode, two alternative strategies apply to control the stability of
an aryl lithium intermediate as a function of inner temperature
(IT) and residence time (τ) (Figure 1).
As for the first strategy, the inner temperature is accurately
controlled to low (cryogenic) temperatures (−50 to −100 °C),
The second strategy, referred to as Flash Chemistry (chemical
synthesis on a time scale of <1 s),3 on the contrary focuses on
instantaneous, highly efficient mixing at noncryostatic temper-
ature in combination with very short residence times (<1 s)
(Figure 1b). In this case the reaction time is defined by the
mixing time and transformations occur under quasi-adiabatic
conditions. This type of operation is typically performed in
narrow dimensioned micromixers (ID = ∼ 0.25 mm) with
excellent mixing properties which avoids decomposition
through rapid trapping of the unstable aryl lithium species.
Figure 1. Temperature profile of a low-flow (a) and an quasi-adiabatic
flow regime (b).
Received: June 10, 2016
© XXXX American Chemical Society
A
DOI: 10.1021/acs.orglett.6b01681
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
There are various examples reported in literature demonstrating
the synthetic power of chemical synthesis on a time scale of <1
s such as the generation and quenching of fragile aryl lithium
intermediates in the presence of pendant electrophilic
functionalities giving access to novel, impressive chemical
transformations.4
Despite significant developments and advances in the area of
organolithium chemistry applying CM technology,5 an unmet
need becomes apparent, when multigram quantities are
required in a short period of time. Microscale flow devices
used to prepare the initial small quantities of material for
medicinal chemistry applications are no longer appropriate for
larger scale, since their throughput is in general low (g/h) and
maintaining these setups effective over an extended period of
time is not easy due to its tiny channels which tend to clog.2d
Thus, microscale flow processes require time-consuming
redevelopment when more material is needed or high
throughputs (g/min) are important; e.g during process
development in pharmaceutical industry (Figure 2).5
operator predefined parameters thereby mitigating the risk of
safety relevant incidents.
Prior to initiating the practical work, we compromised and
defined essential prerequisites to accelerate the development
process: (a) usage of commercially available nBuLi (1.6 M in
hexanes), as no predilution is required and the reagent can be
pumped directly from the cylinder; (b) aryl bromide
concentration = 0.3 M, as our experience showed that this
concentration can be applied to most substrates without
problems of clogging; (c) use of PFA tubing as reactors (ID =
0.8 mm); (d) use of PTFE T-pieces7 (ID = 0.5 mm) as mixing
elements; (e) jacket temperature (JT) set to −30 °C, as this is
the lowest temperature which is readily available using standard
cooling devices (Scheme 1).
Scheme 1. General Setup for the Investigation of Mixing
Performance for the Metalation Step: Flow Rate of Aryl
Bromide (1) and nBuLi Was Changed over Time
Figure 2. Development phases of a continuous manufacturing process.
Based on the assumption that Hal/Li exchange reactions are
mixing-controlled, we started to investigate the mixing time (=
reaction time) using the model transformation of 1-bromo-4-
fluoro-2-(trifluoromethyl)benzene (1a) to 1-fluoro-3-(trifluoro-
methyl)benzene (3a) as depicted in Scheme 1. We assumed
that the yield of the proton quenched product 3a directly
corresponds to the mixing performance of our setup and that
the mixing performance of a simple T-piece is determined by
the total flow rate applied. Therefore, in contrast to the general
approach of optimizing the residence time of a reaction to
achieve high conversion of starting material (strategy 1), we
started optimizing the yield of 3a as a function of the total flow
rate (Figure 3).
Inspired by the various advantages of Flash Chemistry, we
envisioned to address this issue by combining the benefits of
very fast chemical transformations with high throughputs and a
straightforward setup broadly available to the scientific
community. A setup fulfilling this criteria not only wouldallow
mimicing the concept of flash synthesis in a macroscale device
but also would accelerate a potential scale-up process, as it
already produces enough material for early development studies
(kg/d).
With respect to hardware selection, we were aiming for a
simple, easy-to-use, broadly available setup ideally consisting of
only ordinary, commercially available T-pieces (ID = 0.5 mm)
and PFA tubing (OD = 1.6 mm or OD = 3.2 mm) allowing for
flexibility and modularity. However, due to the fact that this
setup does not deliver very short (<1 s) mixing times by
default, we had to find a way to increase the mixing
For this set of experiments, parameters including concen-
trations, flow rate ratio, temperature, tubing, and T-piece
performance within the T-piece mixer. Roder et al. already
̈
demonstrated that the mixing efficiency of a T-piece strongly
depends on the applied total flow rate.6 Therefore, we started
investigating the minimum total flow rate required (based on a
given macro-lab-scale T-piece mixer with ID = 0.5 mm) to
achieve efficient mixing in the range of <1 s under quasi-
adiabatic reaction conditions.
For our investigations we used a flow platform for laboratory
process development consisting of precooling loops for all
reagent streams, PFA tubular reactors, continuous syringe
pumps (Syrdos2), and pressure sensors for each reagent stream.
The platform is software-controlled (HiTecZang) enabling for
online monitoring of temperatures, flow rates, and system
pressures as well as safety shut-down actions based upon
Figure 3. Yield (%) of electron-poor arene 3a (blue) and electron-rich
arene 3b (red) vs total flow rate (mL/min) using a 100 ms reactor for
Hal/Li exchange. Yield was determined via HPLC using biphenyl as
internal standard.
B
DOI: 10.1021/acs.orglett.6b01681
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
remained unchanged (Scheme 1). The only parameter being
adjusted over the course of the experiments was the overall flow
rate for the Hal/Li exchange maintaining a flow rate ratio of
4.85:1. The conversion of starting material 1a to the
corresponding MeOH-quenched product 3a at different flow
rates is depicted in Figure 3 (blue).
intermediate 1 using trimethyl borate is also mixing-controlled,
we imagined that the borylation step can also be performed
within 500 ms due to the fact that the second step will benefit
from an even higher total flow rate and mixing efficiency within
the T-piece mixer. The reaction stream containing the boronate
is finally quenched by aqueous citric acid in batch mode.
As summarized in Scheme 2, various boronic acids were
rapidly synthesized using the described setup without any
It can be clearly noticed that at the lower end of flow rates
only moderate conversion can be achieved. However, the
conversion increases significantly when flow rates >14 mL/min
are applied, directly corresponding to the increased mixing
efficiency within the T-piece mixer. Full conversion of 1a → 3a
(yield >99%) was achieved using the described setup at a total
flow rate (1a + nBuLi) of 24.1 mL/min at JT = −30 °C, which
is in agreement with mixing times previously investigated by
Scheme 2. Multigram Synthesis of Different Boronic Acids
a
within 1 s
Roder et al.6
̈
We were curious if changing the electronic nature of the aryl
bromide from electron poor (1a) to an electron rich substrate
(1b) would impact our initial finding, as these substrates tend
to react slower in the Hal/Li exchange reaction.
Indeed, under analogous reaction conditions compared to
conversion of 1a → 3a, 4-bromoanisole (1b) was not fully
converted with nBuLi into the desired proton quenched
product and 3b could be only obtained in 77% yield (Figure 3,
red). However, a similar significant increase in yield was
observed when a total flow rate (1b + nBuLi) > 14 mL/min
was applied. We decided to extend the residence time of our
setup to make our setup more general and suitable for electron-
rich as well as electron-poor aryl bromides. Eventually, when
the residence time was adjusted to 500 ms, 4-bromoanisole
(1b) was reacted to anisole (3b) in 90% yield at 14.5 mL/min
and in 95% yield at a flow rate of >18 mL/min (Figure 4).
a
Reactions were performed on 61.2 mmol scale (12 min run). Isolated
b
yield is given. Starting from the corresponding diethyl acetate.
c
System blocked after 8 min, HPLC yield is given (λ = 210 nm).
d
HPLC yield is given (λ = 210 nm) as boronic acid decomposed
during workup.
further optimization or adjustments required. Electron-rich (4b,
4e) as well as electron-poor substrates (e.g., 4a, 4g, 4i) were
readily converted into the corresponding boronic acids in good
to excellent yields with a throughput of 5.1 mmol/min (∼1 g/
min). Most interestingly, aryl bromides bearing fluorine as well
as cyano substituents, which are prone to undergo undesired
side reactions or to decompose in the presence of nBuLi under
conventional batch conditions, could be lithiated and borylated
in high yield. Boronic acids bearing an aldehyde functionality
(4f) are accessible from the corresponding 1-bromo-4-
(diethoxymethyl)benzene. It should be noted that all boronic
acids were synthesized in high purity (>95% by HPLC at 210
nm) following a simple extractive workup, allowing the
straightforward synthesis of highly pure boronic acids in gram
scale without the need for a time-consuming purification
procedure.
Figure 4. Yield of 3b (%) vs total flow rate (mL/min) using a 500 ms
reactor for Hal/Li exchange. Yield was determined via HPLC using
biphenyl as internal standard.
Based on our gained insight and experience we defined the
optimal reaction and setup conditions suitable for the
transformation of electron-rich and electron-poor aryl
bromides. For future projects a total flow rate of 20.5 mL/
min was used, as this is seen as a good compromise between
throughput, consumption of starting material, and total flow
rates achievable for most common laboratory pump modules,
which should make our approach attractive also for academic
laboratories.8
Boronic acids not only are important intermediates for
Suzuki cross-coupling reactions, a reaction class of outstanding
importance in pharmaceutical industry,9 but also are coupling
partners in many more modern chemical transformations.10
To demonstrate the usefulness of the described flow setup
we applied our process to the synthesis of a variety of boronic
acids. As we assume that the borylation of the lithiated
In an effort to further expand the product diversity as well as
to demonstrate the broad applicability of our flow platform, the
third input stream was changed from trimethyl borate to other
electrophiles to prepare compounds such as aldehydes and
alcohols in one single operation (Scheme 3).
C
DOI: 10.1021/acs.orglett.6b01681
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
Scheme 3. Multigram Synthesis of Alcohols and Aldehydes
within 1 s
AUTHOR INFORMATION
Corresponding Author
■
a
*E-mail: joerg.sedelmeier@novartis.com.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors are very thankful for the support of the Novartis
Continuous Manufacturing Unit. In particular, we would like to
thank Francesco Venturoni, Julien Haber, Jutta Polenk,
Berthold Schenkel, and Christian Fleury. In addition, we
would like to thank Christine Waldraff and Stephane Schmitt
for analytical support.
REFERENCES
■
(1) (a) Gutmann, B.; Cantillo, D.; Kappe, C. O. Angew. Chem., Int.
Ed. 2015, 54, 6688. (b) Teoh, S. K.; Rathi, C.; Sharratt, P. Org. Process
Res. Dev. 2016, 20, 414. (c) Desai, A. A. Angew. Chem., Int. Ed. 2012,
51, 9223. (d) Hartman, R. L.; McMullen, J. P.; Jensen, K. F. Angew.
Chem., Int. Ed. 2011, 50, 7502.
(2) (a) Browne, D. L.; Baumann, M.; Harji, B. H.; Baxendale, I. R.;
Ley, S. V. Org. Lett. 2011, 13, 3312. (b) Newby, J. A.; Blaylock, D. W.;
Witt, P. M.; Pastre, J. C.; Zacharova, M. K.; Ley, S. V.; Browne, D. L.
Org. Process Res. Dev. 2014, 18, 1211. (c) Newby, J. A.; Blaylock, D.
W.; Witt, P. M.; Turner, R. M.; Heider, P. L.; Harji, B. H.; Browne, D.
L.; Ley, S. V. Org. Process Res. Dev. 2014, 18, 1221. (d) Laue, S.;
Haverkamp, V.; Mleczko, L. Org. Process Res. Dev. 2016, 20, 480.
(e) Sleveland, D.; Bjørsvik, H.-R. Org. Process Res. Dev. 2012, 16, 1121.
(f) Kupracz, L.; Kirschning, A. Adv. Synth. Catal. 2013, 355, 3375.
(3) (a) Yoshida, J. Flash Chemistry: Fast Organic Synthesis in
Microsystems; Wiley-Blackwell: Hoboken, NJ, 2008. (b) Yoshida, J.
Chem. Commun. 2005, 4509. (c) Yoshida, J.; Nagaki, A.; Yamada, T.
Chem. - Eur. J. 2008, 14, 7450.
a
Reactions were done on 25.5 mmol scale (5 min run). Isolated yield
b
is given. HPLC yield is given (λ = 210 nm) as product is volatile.
In all cases the desired product was obtained in remarkable
purity and high yield using the same reaction conditions as
previously outlined. This shows the general applicability of this
setup for different metalation reactions and allows the rapid and
simple synthesis of different compounds in a straightforward
manner.
In conclusion, we developed a simple flow setup consisting of
broadly available, easy-to-handle macroscale equipment for the
synthesis of various compounds following a reaction sequence
of Hal/Li exchange and electrophilic quench. By optimizing the
yield as a function of the total flow rate rather than residence
time, it was possible to synthesize building blocks within a 1 s
total reaction time and with a remarkable throughput of ∼60 g/
h.
This one setup allows for the preparation of small quantities
for medicinal chemistry applications as well as for the delivery
of larger quantities for early development phases with no need
for a redevelopment of the chemical transformation. The
straightforward concept greatly enhances the synthetic utility of
organolithium moieties and appeals attractively to the scientific
community. A concept for the seamless scale-up to larger
quantities for late phase development will be reported in due
time.
(4) (a) Nagaki, A.; Kim, H.; Yoshida, J. Angew. Chem., Int. Ed. 2008,
47, 7833. (b) Nagaki, A.; Ichinari, D.; Yoshida, J. J. Am. Chem. Soc.
2014, 136, 12245. (c) Kim, H.; Min, K.; Inoue, K.; Im, D. J.; Kim, D.;
Yoshida, J. Science 2016, 352, 691. (d) Yoshida, J.; Takahashi, Y.;
Nagaki, A. Chem. Commun. 2013, 49, 9896.
(5) Wheeler, R. C.; Benali, O.; Deal, M.; Farrant, E.; MacDonald, S. J.
F.; Warrington, B. H. Org. Process Res. Dev. 2007, 11, 704.
(6) Schwolow, S.; Hollmann, J.; Schenkel, B.; Roder, T. Org. Process
̈
Res. Dev. 2012, 16, 1513.
(7) Bohlender GmbH. http://www.BOLA.de (accessed Jun 27,
2016).
(8) For example: (a) Syrris Ltd. http://www.syrris.com (accessed Jun
27, 2016). (b) Uniqsis Ltd. http://www.uniqsis.com (accessed Jun 27,
2016). (c) Vapourtec Ltd. http://www.vapourtec.com (accessed Jun
27, 2016).
(9) (a) Miyaura, N.; Suzuki, A. Chem. Rev. 1995, 95, 2457.
(b) Brown, D. G.; Bostrom, J. J. Med. Chem. 2016, 59, 4443.
̈
(c) Roughley, S. D.; Jordan, A. M. J. Med. Chem. 2011, 54, 3451.
(d) Schneider, N.; Lowe, D. M.; Sayle, R. A.; Tarselli, M. A.; Landrum,
G. A. J. Med. Chem. 2016, 59, 4385.
(10) For example: (a) Barluenga, J.; Tomas-Gamasa, M.; Aznar, F.;
Valdes, C. Nat. Chem. 2009, 1, 494. (b) Battilocchio, C.; Feist, F.;
Hafner, A.; Simon, M.; Tran, D. N.; Allwood, D. M.; Blakemore, D. C.;
Ley, S. V. Nat. Chem. 2016, 8, 360. (c) Alam, R.; Vollgraff, T.;
Eriksson, L.; Szabo, K. J. J. Am. Chem. Soc. 2015, 137, 11262.
(d) Seiple, I. B.; Su, S.; Rodriguez, R. A.; Gianatassio, R.; Fujiwara, Y.;
Sobel, A. L.; Baran, P. S. J. Am. Chem. Soc. 2010, 132, 13194.
ASSOCIATED CONTENT
■
S
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.or-
glett.6b01681.
Detailed experimental procedures and spectral data for
all new compounds (PDF)
D
DOI: 10.1021/acs.orglett.6b01681
Org. Lett. XXXX, XXX, XXX−XXX