Use of a Droplet Platform to Optimize Pd-Catalyzed C-N Coupling Reactions Promoted by Organic Bases

Article abstract of DOI:10.1021/acs.oprd.9b00236

Recent advances in Pd-catalyzed carbon-nitrogen cross-coupling have enabled the use of soluble organic bases instead of insoluble or strong inorganic bases that are traditionally employed. The single-phase nature of these reaction conditions facilitates their implementation in continuous flow systems, high-throughput optimization platforms, and large-scale applications. In this work, we utilized an automated microfluidic optimization platform to determine optimal reaction conditions for the couplings of an aryl triflate with four types of commonly employed amine nucleophiles: anilines, amides, primary aliphatic amines, and secondary aliphatic amines. By analyzing trends in catalyst reactivity across different reaction temperatures, base strengths, and base concentrations, we have developed a set of general recommendations for Pd-catalyzed cross-coupling reactions involving organic bases. The optimization algorithm determined that the catalyst supported by the dialkyltriarylmonophosphine ligand AlPhos was the most active in the coupling of each amine nucleophile. Furthermore, our automated optimization revealed that the phosphazene base BTTP can be used to facilitate the coupling of secondary alkylamines and aryl triflates.

Full text of DOI:10.1021/acs.oprd.9b00236

Article
Cite This: Org. Process Res. Dev. XXXX, XXX, XXX−XXX
pubs.acs.org/OPRD
Use of a Droplet Platform To Optimize Pd-Catalyzed C−N Coupling
Reactions Promoted by Organic Bases
†
,§
‡,§
Nicholas A. White,‡ Stephen L. Buchwald,*,‡
Lorenz M. Baumgartner,
and Klavs F. Jensen*
Joseph M. Dennis,
,
†
†
Department of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States
Department of Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States
‡
*
S Supporting Information
ABSTRACT: Recent advances in Pd-catalyzed carbon−nitrogen cross-coupling have enabled the use of soluble organic bases
instead of insoluble or strong inorganic bases that are traditionally employed. The single-phase nature of these reaction
conditions facilitates their implementation in continuous flow systems, high-throughput optimization platforms, and large-scale
applications. In this work, we utilized an automated microfluidic optimization platform to determine optimal reaction
conditions for the couplings of an aryl triflate with four types of commonly employed amine nucleophiles: anilines, amides,
primary aliphatic amines, and secondary aliphatic amines. By analyzing trends in catalyst reactivity across different reaction
temperatures, base strengths, and base concentrations, we have developed a set of general recommendations for Pd-catalyzed
cross-coupling reactions involving organic bases. The optimization algorithm determined that the catalyst supported by the
dialkyltriarylmonophosphine ligand AlPhos was the most active in the coupling of each amine nucleophile. Furthermore, our
automated optimization revealed that the phosphazene base BTTP can be used to facilitate the coupling of secondary
alkylamines and aryl triflates.
KEYWORDS: microfluidic optimization platform, homogeneous catalysis, palladium, amination
INTRODUCTION
greater functional group tolerance than alkoxide bases, but
their use is complicated by their poor solubility in organic
■
Over the last 25 years, methods for palladium-catalyzed
carbon−nitrogen bond formation have enabled the discovery
and production of many fine chemicals, including active
11
solvents. Accordingly, reactions with these systems often
involve heterogeneous reaction mixtures with complex mass
transfer phenomena, which encumbers their application in
1
−3
pharmaceutical ingredients (APIs).
Generations of new
12−14
HTE technologies.
catalysts and protocols have been discovered that have led to
robust synthetic processes. For example, phosphine-bound
15,16
The use of organic bases (Figure 2)
offers a simple,
4
,5
practical solution to these problems. Such bases exhibit
excellent functional group tolerance and solubility in
commonly used reaction solvents. As a result, reactions
facilitated by them take place in a single liquid phase, which
precatalysts, including air-stable oxidative addition com-
6
plexes, were developed to conveniently and reliably deliver
active Pd(0) complexes. These catalysts, including those
bearing dialkylbiarylmonophosphine ligands (Figure 1),
facilitate the coupling of a diverse set of amine nucleophiles
enables straightforward drug discovery with HTE equip-
17−19
ment
and simplifies the development of large-scale
7
and aryl (pseudo)halides under mild reaction conditions.
20
production processes. Despite these key advantages, the
use of organic bases in Pd-catalyzed amination has been
relatively uncommon. Early methods that employed organic
bases suffered from limited substrate scope and a requirement
to employ harsh reaction conditions. For example, the coupling
of aryl nonaflates with anilines was facilitated by an XPhos
While the number and quality of amination methodologies
have steadily increased over time, the majority of protocols
continue to employ inorganic alkoxide, carbonate, or
phosphate bases. The use of these bases suffers from a number
of operational challenges that hinder the application of Pd-
catalyzed amination as a general synthetic tool, especially in
emerging industrial technologies such as flow chemistry and
high-throughput experimentation (HTE). For example,
nucleophilic metal alkoxide and lithium amide bases can
complicate the development of continuous flow processes and
large-scale drug production because of the precipitation of
(
L4)-supported catalyst and 1,8-diazabicyclo[5.4.0]undec-7-
ene (DBU) but required microwave heating to temperatures
up to 150 °C.
21
Within the past few years, the combination of sterically
bulky ligands like t-BuXPhos (L1) or t-BuBrettPhos (L2) with
strong, sterically hindered organic bases has expanded the
8
,9
insoluble halide salts during the course of the reaction.
Furthermore, these bases, especially in combination with
amines, are incompatible with a number of base-sensitive and
electrophilic functional groups, which limits the synthetic
Special Issue: Honoring 25 Years of the Buchwald-Hartwig
Amination
10
utility of methods requiring their use. Non-nucleophilic
Received: May 20, 2019
inorganic bases, including Cs CO , K CO , and K PO , offer
2
3
2
3
3
4
©
XXXX American Chemical Society
A
DOI: 10.1021/acs.oprd.9b00236
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
relevant consideration for downstream kilogram-scale produc-
tion processes.
Using an automated microfluidic optimization platform, we
carried out catalytic reactions and analyzed the effects of
multiple variables on the product yield. We determined
reaction conditions that maximized the yields for four model
amine nucleophiles. The explored variables included the
identity of the ligand, selected from a small pool of
commercially available phosphine ligands, and the choice of
soluble organic base. Additionally, we examined continuous
process variables such as the temperature, time, and amount of
organic base. The resulting data provide general guidelines for
selecting suitable reaction conditions for Pd-catalyzed C−N
cross-coupling using soluble bases.
Figure 1. Selection of amination ligands L1−L7 for Pd-catalyzed C−
N coupling and the corresponding oxidative addition precatalysts
P1−P7.
RESULTS AND DISCUSSION
■
The results of the preliminary experiments and detailed data
associated with the four optimization efforts are included in the
Supporting Information. These results showed that the highest
product yields were obtained with precatalysts P1−P3 bearing
ligands L1−L3 for the coupling of a range of amine
nucleophiles. Precatalysts bearing L4−L7 were much less
effective under identical conditions and therefore were not
utilized in our optimization studies.
Cross-Coupling of Primary Anilines. As shown in Figure
, the coupling of p-tolyl triflate and aniline (1) occurs in near
3
quantitative yield over a wide range of conditions, with the
exception of reactions facilitated by TEA. For this reason, the
algorithm was unable to find a global optimum during the
screening of precatalysts P1−P3. Despite the selective
generation of product and minimal side-product formation,
low conversion was observed for all reactions performed with
TEA and in some cases with TMG. This observation suggests
that these bases may not be strong enough to facilitate the
reaction under the tested conditions.
Figure 2. Selection of soluble organic bases with different degrees of
15,16
+
steric hindrance, base strengths (pKBH in ACN),
and prices.
Commercial base costs were calculated from the largest batch
available from common chemical vendors at the time of this study.
Comparison of the results associated with TMG and BTMG
shows that the base structure plays an important role in the
success of the reaction. Although TMG and BTMG exhibit
very similar pKBH+ values, the addition of a t-Bu group to the
basic nitrogen of TMG seems to be beneficial for the reaction
yield. Previous studies on the role of organic bases in arylamine
couplings have shown that the formation of base-bound Pd
scope of C−N cross-coupling reactions. Nitrogen-containing
bases, such as 7-methyl-1,5,7-triazabicyclo[4.4.0]dec-5-ene
(
(
MTBD) and 2-tert-butyl-1,1,3,3-tetramethylguanidine
BTMG), as well as phosphazene bases, such as BTTP and
22
P2Et, have recently been utilized in a number of homogeneous
amination methods. For example, researchers at Merck first
reported the successful arylation of benzamide at room
temperature using L1 or L2 in combination with BTMG,
26
species can inhibit catalysis. In this case, the t-Bu group may
prevent unfavorable binding and thus increase the overall
reactivity. In an additional comparison, the stronger but less
sterically hindered base DBU exhibits a lower reactivity. These
results are consistent with the notion that the most effective
1
6
DBU, MTBD, or P Et. Their work in this area also
2
encompassed the arylation of other nucleophiles, such as 1°
1
6
and 2° alkylamines, with P Et, demonstrating the diversity of
2
16
1
9,23
organic bases are strong and moderately sterically hindered.
this system.
Despite the mild reaction conditions of these
reactions, the methodology is still reliant on a very expensive
base ($40/mmol). A more affordable alternative would thus be
highly desirable.
For the coupling reactions conducted with DBU, we
observed the precatalyst activity trend P1 < P2 < P3, where
the highest yields were obtained with P3. As an illustration, the
use of P2 led to 51% yield in comparison with only 9% yield
achieved with P1 under the same conditions (Figure 3,
reaction 9 vs 45). Likewise, the reaction with precatalyst P3
reached a nearly quantitative yield of 97% at 30 °C, while that
with P2 reached only a 42% yield at 57 °C (Figure 3, reaction
69 vs 64). This activity trend is consistent with the results of
our previous study. On the basis of NMR analyses and DFT
calculations, we previously proposed that ligands with bulkier
substituents (Cy < t-Bu < Ad) form a more electron-deficient
Pd complex, which acidifies the bound amine and promotes
the deprotonation of the amine by weaker bases.
We recently reported the use of a sterically demanding
24
AlPhos (L3)-supported catalyst to facilitate the coupling of
anilines, amides, and 1° amines using DBU, a common and
2
5
inexpensive base. We also demonstrated that L2 could be
employed as a less expensive alternative, albeit with a slightly
lower level of efficiency and generality. These results raise the
question whether AlPhos or other dialkylbiaryl mono-
phosphine ligands in combination with stronger, less sterically
21
hindered bases can offer an alternative to P Et for the coupling
2
of 2° alkylamines. While the cost of a given base is often of
secondary importance for drug discovery, it becomes a more
B
DOI: 10.1021/acs.oprd.9b00236
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 3. Online optimization results for aniline (1). Representative
data points are shown for clarity. All of the data points are provided in
the Supporting Information. The diagram position indicates the
combination of base (row) and ligand (column). Continuous variable
settings are represented by the position in the xy plane of the three-
dimensional diagram (time, base equivalents) and the color bar
Figure 4. Online optimization results for benzamide (2).
Representative data points are shown for clarity. All of the data
points are provided in the Supporting Information. Axis labels and
plot orientations are analogous to those in Figure 3. Asterisks (*)
denote limited accuracy due to product peak tailing in LC traces: yield
≥ 90% and conversion ≥ 90%. See the Supporting Information for full
discussion.
(
temperature). The resulting yield is indicated by the vertical position
on the z axis. Reaction numbers next to data points indicate the order
of execution. Aniline was added to the reaction mixture by online
injection.
optimization data showed clear reactivity trends in regard to
precatalyst and base combinations. The data in Figure 4 show
that benzamide is more difficult to couple than aniline. Under
these reaction conditions, use of the bases TMG and DBU
leads to low yields, and the reactions require higher
temperature for all of the precatalysts to increase the reactivity.
In contrast, use of the more sterically hindered bases BTMG
and MTBD allowed for excellent product formation under
milder conditions. For example, the substitution of DBU with
MTBD increased the yield from 11% to 57% under similar
conditions (Figure 4, reaction 67 vs 83). The use of BTMG
further increased the reaction rate, leading to a conversion of
greater than 90% at room temperature in 25 min (reaction 80).
The use of precatalyst P1 bearing t-BuXPhos as the ligand
led to poor performance for the arylation of benzamide. For
instance, a maximum yield of 26% with BTMG was measured
in reaction 10 (Figure 4). The same precatalyst and base
combination provided yields of 84% to 100% for aniline. In
contrast, employing other precatalysts led to more successful
For the coupling of aniline, our online optimization data
suggest that increasing the base concentration does not
increase the reactivity. These data are also consistent with
our mechanistic studies, which show that the rate of this
coupling reaction displays an inverse dependence on the base
2
7
concentration. As a result, an efficient process for the
coupling of primary anilines requires only a slight excess of the
base relative to the electrophile.
Cross-Coupling of Primary Amides. The optimization
results for the coupling of benzamide (2) are summarized in
Figure 4. To overcome the low solubility of this amide in
ethereal solvents, we chose DMSO as the reaction solvent.
While the high viscosity of this solvent led to reaction mixtures
28
that periodically clogged the reactor tubing, this solvent was
ideal for catalysis based on batch reactions featuring polar
aprotic solvents. Despite this operational difficulty, the
C
DOI: 10.1021/acs.oprd.9b00236
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
product formation. For example, the use of P2 and P3 resulted
in yields of 51% and 67%, respectively, at 65 °C (Figure 4,
reactions 46 and 72). These data points suggest that
precatalyst P3 is slightly more reactive than P2 under these
reaction conditions. However, reactions 34 and 66 (Figure 4)
show that both precatalysts can be used to obtain quantitative
yields. Therefore, P2 and P3 are equally suitable catalysts for
the coupling of primary amides such as benzamide.
precatalyst. For example, when BTMG was used with P1, we
observed a maximum yield of only 24% (Figure 5, reaction 32).
However, performing the reaction with precatalyst P2
increased the maximum observed yield to 92% (reaction 68).
The use of P3 resulted in quantitative yields with both of these
bases. Very low yields were observed for reactions carried out
with the inexpensive base DBU in combination with either P1
and P2. In contrast, P3 coupled β-phenethylamine at high
temperatures (reaction 84). Although the yield is limited to
Cross-Coupling of Primary Aliphatic Amines. Arylation
of the primary amine β-phenethylamine (3) is very effective
with the phosphazene base BTTP (Figure 5). The
optimization algorithm showed that very high yields (82−
4
2% within ∼30 min, the reaction is expected to reach full
25
conversion within several hours. These conditions could be
exploited for the development of an economical process with
the inexpensive base DBU provided that the substrates and
product are stable at higher temperatures (Table 1).
1
6
00%) can be reached within minutes at temperatures above
5 °C using any of the precatalysts tested. For P3, a 93% yield
was realized at 30 °C (reaction 85), suggesting that the
combination of P3 and BTTP is effective under mild
conditions.
For reactions facilitated by the bases BTMG and MTBD, the
product formation depended more strongly on the choice of
Furthermore, the data show a positive effect of increasing
the concentration of BTMG, MTBD, and BTTP on the
reaction yield. As an example, increasing the stoichiometric
amount of MTBD from 1.1 to 2.0 equiv elevated the yield from
65% to 99% (Figure 5, reaction 72 vs 70). This result suggests
that the base is involved in the rate-limiting step and implies
Figure 5. Online optimization results for β-phenethylamine (3).
Representative data points are shown for clarity. All of the data points
are provided in the Supporting Information. Axis labels and plot
orientations are analogous to those in Figure 3.
Figure 6. Online optimization results for morpholine (4).
Representative data points are shown for clarity. All of the data
points are provided in the Supporting Information. Axis labels and
plot orientations are analogous to those in Figure 3.
D
DOI: 10.1021/acs.oprd.9b00236
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Table 1. Recommended Reaction Conditions To Obtain High Yields (70−100%) in Arylations of Four Classes of Amine
Nucleophiles
substrate
profile
catalyst
base
equiv of base
T (°C)
time
a
1
° anilines
° amides
° amines
° amines
mild
P1−P3
P1, P2
P2, P3
P2, P3
P3
P3
P3
P3
BTMG, DBU
TMG, DBU
BTMG, MTBD
DBU
BTMG, MTBD, BTTP
DBU
1.1
1.1
30
100
30
100
30
100
30
70
<30 min
<5 min
b
low cost
mild
low cost
mild
low cost
mild
most effective
c
c
1
1
2
1.1−2.0
1.1−2.0
1.1−2.0
1.1−2.0
1.1−2.0
1.1−2.0
<90 min
<15 min
<45 min
c
>2 h
>6 h
c
BTTP, P Et
2
c
BTMG, MTBD, BTTP
<60 min
a
b
Temperature-sensitive reagents require the use of more expensive catalysts and bases. Temperature-tolerant substrates permit the use of less
expensive catalysts and bases. Estimate based on extrapolation of optimization results.
c
that the base stoichiometry may be important to consider
during optimization of arylation reactions of primary amines.
Cross-Coupling of Secondary Aliphatic Amines. The
secondary amine morpholine (4) was the most difficult to
couple among the nucleophiles examined (Figure 6). Reactions
using precatalyst P2 (1 mol %) gave extremely poor yields in
general: the optimum conditions determined by the algorithm
resulted in only 11% yield with 26% selectivity toward the
desired coupling product (Figure 6, reaction 29). In this case,
the mass balance contained a number of side products,
including p-cresol (Ar−OH) and p-tolyl ether (Ar−O−Ar).
These are commonly formed when the Pd−N bond-forming
step of the catalysis is slow and water is present in the reaction
reagent is sufficient for the coupling of 1° anilines, which is in
line with other experimental studies. For the coupling of 1°
amines, product formation can be enhanced by increasing the
base concentration for the bases BTMG, MTBD, and BTTP.
Finally, in coupling of 2° amines, the temperature must be
well-controlled to limit the rate of competing side reactions,
including hydrolysis of the electrophile.
The recommended conditions can serve as a starting point
for developing economical production processes with soluble
organic bases. We note that these conditions are best applied
to cross-coupling reactions employing aryl triflate electrophiles.
While previous work has shown that certain ligand and base
combinations are suitable for coupling of aryl triflates,
8
27
mixture. In some cases, this pathway might be mitigated
bromides, and chlorides, we cannot generalize these specific
through the use of aryl nonaflates instead of aryl triflates as
conditions to other electrophiles without further studies.
Nonetheless, this study shows that a number of organic
bases can serve as operationally practical and synthetically
useful alternatives to the bases traditionally used in Pd-
catalyzed amination reactions.
2
1
previously shown. Carrying out the reaction with an
increased catalyst loading (2 mol %) with MTBD for 20 min
at 100 °C resulted in the formation of only a trace amount of
product (∼1% yield). The data for this preliminary
optimization run are included in the Supporting Information.
These results confirm that P2 is not an effective precatalyst for
EXPERIMENTAL METHODS
14,18
■
the coupling of 2° amines such as morpholine.
We carried out the process optimization with a slightly
modified version of the automated microfluidic reaction
platform used for our recent C−C coupling experiments.
Reactions using P3 (2 mol %) also exhibited limited
selectivity at high temperatures, and the same side products
were observed. For instance, we measured a 47% yield of the
desired product with 98% conversion of the aryl triflate at
29
This platform allows material-efficient online optimization with
continuous and discrete variables. The tube-based reactor
design facilitates the transfer of conditions to flow. We
narrowed down the parameter space with preliminary experi-
ments. More detailed information about experimental
procedures, the reaction system, and the algorithm is included
in the Supporting Information.
For each optimization run, a stock solution containing the
desired precatalyst, p-tolyl triflate, solvent (2-MeTHF or
DMSO), and 1-fluoronaphthalene (internal standard) was
prepared in a nitrogen-filled glovebox. Under these inert-
atmosphere conditions, the fine slurry was transferred to a
tapered vial equipped with a stir bar. The vial was sealed with a
screwtop cap fitted with a silicone septum. Vials containing
neat amine nucleophiles (with the exception of benzamide)
were prepared in an analogous manner. All stock solution vials
were removed from the glovebox and transferred to the vial
rack of the liquid handler, which was enclosed in a
polyethylene glovebag (AtmosBag, Sigma-Aldrich) filled with
a dry nitrogen atmosphere. A magnetic stir plate was used to
agitate the catalyst/electrophile stock solution such that the
precatalyst-containing slurry was sufficiently mixed. Glass
syringes were primed with solvent and were mounted on
automatic syringe pumps (Harvard Apparatus PHD Ultra).
1
00 °C (Figure 6, reaction 48). The algorithm determined that
optimal yields of about 70% could be obtained using a reaction
time of 30 min and a temperature of 70 °C with BTMG,
MTBD, and BTTP. Under these conditions, the selectivity
increased to 70−80%.
We then evaluated the transferability of these results from
the small scale of the microfluidic system (0.01 mmol) to
larger reaction scales. For this purpose, we performed a batch-
scale reaction (2.35 mmol) under the conditions of reaction 48
(
Figure 6). We obtained an isolated yield of 54% with a
conversion of 95%. This is in good agreement with the results
obtained with LC/MS, suggesting that our data are indicative
of the efficiency of these reactions independent of scale.
CONCLUSION
■
Using an automated microfluidic platform, we have developed
recommended conditions for Pd-catalyzed C−N cross-
couplings with soluble organic bases covering a range of
substrates (Table 1). The optimization algorithm showed that
a number of ligand and base combinations are suitable for the
coupling of primary anilines. The results also supported the
notion that a slight excess of base with respect to the limiting
E
DOI: 10.1021/acs.oprd.9b00236
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 7. Schematic of the optimization platform setup.
These syringes were used for the online injection of rinse and
quench solutions. For optimization of the reactions with
aniline, a syringe filled with this nucleophile was prepared for
components from the vials containing the reagent stock
solutions. The combination of these reagents and mixing in the
needle resulted in a 40 μL droplet containing all of the reaction
components. This droplet was loaded into a 15 μL sample loop
connected to the six-port, two-position valve of the liquid
handler. The injection of excess liquid into the injection
module ensured that the sample loop contained no gas. After
all of the rinse droplets had been cleared from the main path
into the waste container, the liquid handler switched the
position of the valve to transfer the droplet into the reaction
system.
30
injection at the reactor inlet.
The reaction system (Figure 7) was controlled with
MATLAB (version R2016a; The MathWorks, Inc.) and
LabVIEW (version 16.0f2; National Instruments). Nitrogen
gas was used to pressurize the main tubing lines of the system
with a positive pressure of 100 psig (6.9 bar). The nitrogen
also served as an inert carrier gas that allowed the carrier
syringe to move droplets inside the tubing. The main tubing
lines connected the carrier syringe, the reactor, and the
pressurized waste container. Syringes for online injection were
attached to the main path by T-unions.
The liquid handler (GX 241, Gilson) was connected to the
main lines between the carrier syringe and the reactor by a six-
port, two-position valve equipped with a sample loop. In its
load position, the sample loop was connected to the injection
module of the liquid handler. In its inject position, the sample
loop was connected to the main path of the system. This
configuration allowed the liquid handler to transfer a droplet
from the injection module to the pressurized main path by
switching the valve position. The LC/MS system was
connected with a second six-port, two-position valve located
before the pressurized waste container.
To begin an optimization run, the user started up the system
by pressurizing the main path and priming the syringe pumps.
After initialization, the system carried out experimental cycles
sequentially until the algorithm was terminated. Additionally,
each run was restricted to a maximum of 24 h to limit the
effect of any degradation of the catalyst stock solution. Each of
the experimental cycles consisted of droplet preparation, active
reaction, and product analysis.
For droplet preparation, a syringe injected four 15 μL
droplets of reaction solvent upstream of the liquid handler to
purge the lines of contaminants. Following this sequence, the
liquid handler equipped with a needle withdrew the reaction
To start the reaction, the carrier syringe moved the droplet
into the reactor. For the optimization of aniline cross-
couplings, up to 6 μL of the amine/solvent solution was
injected through the T-union at the reactor inlet prior to this
step. The droplet was then oscillated inside the U-shaped
reactor loop for the target residence time (1−30 min). A
photodetector was used to locate the droplet in the reactor.
This feature allowed the system to repeatedly reverse the flow
direction of the droplet when it reached the terminus of the U-
shaped reactor, which mixed the droplet through oscillatory
3
1
motion.
This oscillatory reactor was housed in a
thermocouple-controlled block that was heated by electrical
cartridge heaters or cooled by a fan between runs. Finally, after
the desired reaction time, solvent was injected into the tube
containing the reaction droplet via a T-union at the reactor
outlet. The injected solvent volume was set equal to that of the
reaction droplet, resulting in a 1:1 dilution. We assumed that
this dilution and cooling effectively quenched the reaction.
Next, the diluted droplet was loaded into the 1.7 μL sample
loop of the second six-port, two-position valve to start the
product analysis. Approximately 0.4 μL of the droplet was
transferred to the LC/MS system, and the product mixture was
analyzed with reversed-phase chromatography. A UV detector
(230 to 330 nm) was used to quantify the product
concentrations on the basis of an analytical method
precalibrated with product standards. Finally, a mass
F
DOI: 10.1021/acs.oprd.9b00236
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
spectrometer was used to verify peak identities. After the
chromatographic method was complete, the automatically
integrated peak areas were exported to an Excel file. MATLAB
was used to import these data and calculate the product yield.
The methods are described in the Supporting Information.
The optimization algorithm aimed to maximize the coupling
product yield. The continuous variables temperature, time, and
amount of base were manipulated within user defined intervals;
the discrete variable pool consisted of up to four mutually
exclusive organic bases. At the beginning of the optimization,
the algorithm used a set of D-optimal experiments for an initial
scan of the reaction space. The resulting yield data were used
to fit a quadratic response surface model for every base type.
After each following iteration, these models were further
refined to determine the discrete variable choice and the
continuous variable conditions that maximized the yield.
Discrete variable choices were dropped from the pool if their
performance was lower than the lower bound of the 99%
confidence interval of the predicted global yield maximum.
This procedure allowed us to focus the new experiments of the
next iteration step on the most promising candidates. The final
experiments were selected to minimize the uncertainty of the
predicted global maximum according to a G-optimal criterion.
The algorithm terminated if the best experimental yield failed
to improve by at least 1% after three consecutive iterations.
experimental design of this project and Jessica Xu (MIT) for
performing HRMS analyses. Finally, we thank Richard Liu, Dr.
Scott McCann, and Dr. Christine Nguyen (MIT) for assistance
in the preparation of this article.
REFERENCES
■
(
1) Beletskaya, I. P.; Cheprakov, A. V. The Complementary
Competitors: Palladium and Copper in C−N Cross-Coupling
Reactions. Organometallics 2012, 31, 7753−7808.
(
2) Ruiz-Castillo, P.; Buchwald, S. L. Applications of Palladium-
Catalyzed C−N Cross-Coupling Reactions. Chem. Rev. 2016, 116,
2564−12649.
(3) Brown, D. G.; Bostro
1
̈
m, J. Analysis of Past and Present Synthetic
Methodologies on Medicinal Chemistry: Where Have All the New
Reactions Gone? J. Med. Chem. 2016, 59, 4443−4458.
(4) Bruno, N. C.; Tudge, M. T.; Buchwald, S. L. Design and
Preparation of New Palladium Precatalysts for C−C and C−N Cross-
Coupling Reactions. Chem. Sci. 2013, 4, 916−920.
(5) Bruno, N. C.; Niljianskul, N.; Buchwald, S. L. N-Substituted 2-
Aminobiphenylpalladium Methanesulfonate Precatalysts and Their
Use in C−C and C−N Cross-Couplings. J. Org. Chem. 2014, 79,
4
(
161−4166.
6) Ingoglia, B. T.; Buchwald, S. L. Oxidative Addition Complexes as
Precatalysts for Cross-Coupling Reactions Requiring Extremely Bulky
Biarylphosphine Ligands. Org. Lett. 2017, 19, 2853−2856.
(7) Ingoglia, B. T.; Wagen, C. C.; Buchwald, S. L. Biaryl
Monophosphine Ligands in Palladium-Catalyzed C−N Coupling:
An Updated User’s Guide. Tetrahedron 2019, 75, 4199−4211.
ASSOCIATED CONTENT
Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.oprd.9b00236.
■
(8) Noel, T.; Buchwald, S. L. Cross-Coupling in Flow. Chem. Soc.
*
S
Rev. 2011, 40, 5010−5029.
(9) While the water content of the 2-MeTHF used in the
optimizations was low (29 ppm by Karl Fischer titration), water
was detected in MTBD by NMR analysis. This base is known to be
hygroscopic and should be stored in a desiccator or under an inert
atmosphere. For comments on the effect of water in cross-coupling
reactions, see: Falß, S.; Tomaiuolo, G.; Perazzo, A.; Hodgson, P.;
Yaseneva, P.; Zakrzewski, J.; Guido, S.; Lapkin, A.; Woodward, R.;
Meadows, R. E. A Continuous Process for Buchwald−Hartwig
Amination at Micro-, Lab-, and Mesoscale Using a Novel Reactor
Concept. Org. Process Res. Dev. 2016, 20, 558−567.
Detailed experimental procedures and characterization
data (PDF)
Raw optimization data, calibration curves, and HPLC
data (XLSX)
AUTHOR INFORMATION
Corresponding Authors
E-mail: kfjensen@mit.edu (K.F.J.).
E-mail: sbuchwal@mit.edu (S.L.B.).
■
(10) Sperry, J. B.; Price Wiglesworth, K. E.; Edmonds, I.; Fiore, P.;
*
Boyles, D. C.; Damon, D. B.; Dorow, R. L.; Piatnitski Chekler, E. L.;
Langille, J.; Coe, J. W. Kiloscale Buchwald−Hartwig Amination:
Optimized Coupling of Base-Sensitive 6-Bromoisoquinoline-1-car-
bonitrile with (S)-3-Amino-2-methylpropan-1-ol. Org. Process Res.
Dev. 2014, 18, 1752−1758.
*
ORCID
Lorenz M. Baumgartner: 0000-0002-6745-765X
Joseph M. Dennis: 0000-0003-3069-312X
Nicholas A. White: 0000-0001-5038-8865
Stephen L. Buchwald: 0000-0003-3875-4775
Klavs F. Jensen: 0000-0001-7192-580X
Author Contributions
(
11) Surry, D. S.; Buchwald, S. L. Dialkylbiaryl Phosphines in Pd-
Catalyzed Amination: A User’s Guide. Chem. Sci. 2011, 2, 27−50.
12) Meyers, C.; Maes, B. U.; Loones, K. T.; Bal, G.; Lemiere, G. L.;
(
̀
Dommisse, R. A. Study of a New Rate Increasing “Base Effect” in the
Palladium-Catalyzed Amination of Aryl Iodides. J. Org. Chem. 2004,
§
6
(
9, 6010−6017.
L.M.B. and J.M.D. contributed equally to this work.
13) Damon, D. B.; Dugger, R. W.; Hubbs, S. E.; Scott, J. M.; Scott,
Notes
R. W. Asymmetric Synthesis of the Cholesteryl Ester Transfer Protein
The authors declare the following competing financial
interest(s): MIT has filed patents on ligands and precatalysts
that are described in this article, from which S.L.B. and former
co-workers receive royalty payments.
Inhibitor Torcetrapib. Org. Process Res. Dev. 2006, 10, 472−480.
(14) Kuethe, J. T.; Childers, K. G.; Humphrey, G. R.; Journet, M.;
Peng, Z. A Rapid, Large-Scale Synthesis of a Potent Cholecystokinin
(CCK) 1R Receptor Agonist. Org. Process Res. Dev. 2008, 12, 1201−
1208.
ACKNOWLEDGMENTS
(15) Kutt, A.; Selberg, S.; Kaljurand, I.; Tshepelevitsh, S.; Heering,
̈
■
A.; Darnell, A.; Kaupmees, K.; Piirsalu, M.; Leito, I. pKa Values in
Organic ChemistryMaking Maximum Use of the Available Data.
Tetrahedron Lett. 2018, 59, 3738−3748.
We thank the Novartis Center for Continuous Manufacturing
for funding this project. We also recognize additional support
from the NSF Graduate Research Fellowship Program (Grant
(
16) Buitrago Santanilla, A.; Christensen, M.; Campeau, L.-C.;
1
122374, J.M.D.) and the NIH (Grant 1F32GM120847-01,
Davies, I. W.; Dreher, S. D. P Et Phosphazene: A Mild, Functional
2
N.A.W.). We acknowledge Novartis for a generous donation of
TMG and BTMG and Sigma-Aldrich for t-BuBrettPhos. We
recognize Connor Coley (MIT) for assisting in the
Group Tolerant Base for Soluble, Room Temperature Pd-Catalyzed
C−N, C−O, and C−C Cross-Coupling Reactions. Org. Lett. 2015, 17,
3370−3373.
G
DOI: 10.1021/acs.oprd.9b00236
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(
17) Buitrago Santanilla, A.; Regalado, E. L.; Pereira, T.; Shevlin, M.;
Bateman, K.; Campeau, L.-C.; Schneeweis, J.; Berritt, S.; Shi, Z.-C.;
Nantermet, P.; Liu, Y.; Helmy, R.; Welch, C. J.; Vachal, P.; Davies, I.
W.; Cernak, T.; Dreher, S. D. Nanomole-Scale High-Throughput
Chemistry for the Synthesis of Complex Molecules. Science 2015, 347,
4
(
9−53.
18) Ahneman, D. T.; Estrada, J. G.; Lin, S.; Dreher, S. D.; Doyle, A.
G. Predicting Reaction Performance in C−N Cross-Coupling Using
Machine Learning. Science 2018, 360, 186−190.
(
19) Gesmundo, N. J.; Sauvagnat, B.; Curran, P. J.; Richards, M. P.;
Andrews, C. L.; Dandliker, P. J.; Cernak, T. Nanoscale Synthesis and
Affinity Ranking. Nature 2018, 557, 228−232.
(
20) Hsieh, H.-W.; Coley, C. W.; Baumgartner, L. M.; Jensen, K. F.;
Robinson, R. I. Photoredox Iridium−Nickel Dual-Catalyzed Decar-
boxylative Arylation Cross-Coupling: From Batch to Continuous
Flow via Self-Optimizing Segmented Flow Reactor. Org. Process Res.
Dev. 2018, 22, 542−550.
(
21) Tundel, R. E.; Anderson, K. W.; Buchwald, S. L. Expedited
Palladium-Catalyzed Amination of Aryl Nonaflates through the Use of
Microwave-Irradiation and Soluble Organic Amine Bases. J. Org.
Chem. 2006, 71, 430−433.
(
22) Schwesinger, R.; Hasenfratz, C.; Schlemper, H.; Walz, L.;
Peters, E.; Peters, K.; von Schnering, H. G. How Strong and How
Hindered Can Uncharged Phosphazene Bases Be? Angew. Chem., Int.
Ed. Engl. 1993, 32, 1361−1363.
(
23) Kutchukian, P. S.; Dropinski, J. F.; Dykstra, K. D.; Li, B.;
DiRocco, D. A.; Streckfuss, E. C.; Campeau, L.-C.; Cernak, T.;
Vachal, P.; Davies, I. W.; Krska, S. W.; Dreher, S. D. Chemistry
Informer Libraries: A Chemoinformatics Enabled Approach To
Evaluate and Advance Synthetic Methods. Chem. Sci. 2016, 7,
2
(
604−2613.
24) Sather, A. C.; Lee, H. G.; De La Rosa, V. Y.; Yang, Y.; Muller,
P.; Buchwald, S. L. A Fluorinated Ligand Enables Room-Temperature
and Regioselective Pd-Catalyzed Fluorination of Aryl Triflates and
Bromides. J. Am. Chem. Soc. 2015, 137, 13433−13438.
(
25) Dennis, J. M.; White, N. A.; Liu, R. Y.; Buchwald, S. L. Breaking
the Base Barrier: An Electron-Deficient Palladium Catalyst Enables
the Use of a Common Soluble Base in C−N Coupling. J. Am. Chem.
Soc. 2018, 140, 4721−4725.
(
26) Sunesson, Y.; Lime, E.; Nilsson Lill, S. O.; Meadows, R. E.;
́
Norrby, P.-O. Role of the Base in Buchwald−Hartwig Amination. J.
Org. Chem. 2014, 79, 11961−11969.
(
27) Dennis, J. M.; White, N. A.; Liu, R. Y.; Buchwald, S. L. Pd-
Catalyzed C−N Coupling Reactions Facilitated by Organic Bases:
Mechanistic Investigation Leads to Enhanced Reactivity in the
Arylation of Weakly Binding Amines. ACS Catal. 2019, 9, 3822−
3
(
830.
28) The high viscosity of DMSO made the transfer of droplets
through the 1.7 μL sample loop of the LC/MS (inner diameter < 0.1
mm) more difficult than with 2-MeTHF. This effect led to a pressure
drop at the constriction point that periodically led to time-outs of the
system when the liquid failed to clear in a timely manner. The use of a
larger sample loop may mitigate this problem.
(
29) Baumgartner, L. M.; Coley, C. W.; Reizman, B. J.; Gao, K. W.;
Jensen, K. F. Optimum Catalyst Selection Over Continuous and
Discrete Process Variables with a Single Droplet Microfluidic
Reaction Platform. React. Chem. Eng. 2018, 3, 301−311.
(
30) In comparison with the other nucleophiles, the reactions with
aniline proceed much more rapidly at room temperature. Therefore,
mixing this substrate with the other reaction components in the liquid
handler would lead to significant conversion within the sample probe.
Preliminary experiments showed that a precipitate formed at the end
of some reactions, which clogged the probe before the reaction
mixture could be injected into the reaction system. As a result, the
aniline was added to the reaction mixture by online injection, which
alleviated this problem.
(
31) Abolhasani, M.; Bruno, N. C.; Jensen, K. F. Oscillatory Three-
Phase Flow Reactor for Studies of Bi-Phasic Catalytic Reactions.
Chem. Commun. 2015, 51, 8916−8919.
H
DOI: 10.1021/acs.oprd.9b00236
Org. Process Res. Dev. XXXX, XXX, XXX−XXX