Selective DIBAL-H Monoreduction of a Diester Using Continuous Flow Chemistry: From Benchtop to Kilo Lab

Article abstract of DOI:10.1021/acs.oprd.0c00158

Herein we report a selective DIBAL-H-mediated reduction of a heterocyclic diester to the corresponding monoaldehyde using continuous flow chemistry. The use of continuous flow enabled operation at lower temperatures and better control of the reaction time, thereby allowing for a significant increase in reaction selectivity and yield compared with batch conditions. The reaction's development as a continuous flow process and its scale-up from laboratory gram scale to multikilogram scale are discussed, including design of experiments studies to probe the optimal reaction window.

Full text of DOI:10.1021/acs.oprd.0c00158

Subscriber access provided by University of Birmingham
Communication
Selective DIBAL-H Monoreduction of a Diester Using
Continuous Flow Chemistry: From Benchtop to Kilo Lab
Nick Uhlig, Andrew Martins, and Detian Gao
Org. Process Res. Dev., Just Accepted Manuscript • DOI: 10.1021/acs.oprd.0c00158 • Publication Date (Web): 08 May 2020
Downloaded from pubs.acs.org on May 10, 2020
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 44
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Selective DIBAL-H Monoreduction of a Diester
Using Continuous Flow Chemistry: From Benchtop
to Kilo Lab
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
AUTHORS: Nick Uhlig*, Andrew Martins, and Detian Gao
AUTHOR ADDRESS: Gilead Alberta ULC, 1021 Hayter Rd. NW, Edmonton, Alberta,
T6S 1A1 Canada.
ACS Paragon Plus Environment
1

Organic Process Research & Development
Page 2 of 44
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
2

Page 3 of 44
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
ABSTRACT. Herein we report a selective DIBAL-H-mediated reduction of a heterocyclic
diester to the corresponding mono-aldehyde using continuous flow chemistry. The use of
continuous flow enabled operation at lower temperatures and better control of residence
time, thereby allowing for a significant increase in reaction selectivity and yield compared
to batch conditions. The reaction’s development as a continuous flow process, and its
scale-up from laboratory gram scale to multi-kilogram scale are discussed, including
design-of-experiments studies to probe the optimal reaction window.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
KEYWORDS: Flow chemistry, reduction, metal hydrides, DIBAL-H, ester, aldehyde
INTRODUCTION
The selective reduction of carbonyl compounds represents an important and frequently
used functional group transformation in both academic studies and industrial applications.
Between hydrogenations and hydride reductions, a wealth of methods and reagents is
available to effect such reductions.1 While many of these chemical approaches have been
reported at large scale (>100 mmol) in the pharmaceutical industry,2 one area which
ACS Paragon Plus Environment
3

Organic Process Research & Development
Page 4 of 44
1
2
3
4
5
6
7
8
9
shows a conspicuous lack of large-scale examples is the direct reduction of esters to
aldehydes.3 A commonly invoked reagent for such a transformation on lab scale is
diisobutylaluminum hydride (DIBAL-H),4 yet this reagent is not nearly as commonly
encountered on large scale, likely due to multiple process challenges, including the
strongly exothermic nature of the reaction, and the formation of high levels of impurities
as a result of overreduction, especially at non-cryogenic temperatures.5
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
RESULTS AND DISCUSSION
Ethyl 2-formylpyridine-4-carboxylate (2) is an intermediate en route to an investigational
compound for the treatment of hepatitis B and inflammatory diseases. The synthesis of 2
(Scheme 1) consists of two steps starting from commercially available pyridine-2,4-
dicarboxylic acid. Esterification to form diethyl pyridine-2,4-dicarboxylate (1), followed by
selective monoreduction of the 2-carboxylate ester affords compound 2.
ACS Paragon Plus Environment
4

Page 5 of 44
Organic Process Research & Development
1
2
3
4
5
Scheme 1. Two-step route to ethyl 2-formylpyridine-4-carboxylate, 2.
6
7
HO
O
EtO
O
EtO
O
8
9
1.5 equiv. DIBAL-H,
then EtOAc
SOCl2
O
O
O
THF, -20 °C
EtOH, reflux
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
N
N
N
OH
OEt
H
1
2
During an early manufacturing campaign, the reduction of 1 was performed in batch
mode at 18 kg scale. While the metered addition of 1.5 equivalents of DIBAL-H to a
solution of 1 in THF at -20 °C was successful in mitigating a substantial exotherm,
significant amounts of overreduction were observed at both the 2- and 4-carboxylate
positions, yielding side products 3, 4, and 5 (Scheme 2). It was found that compound 3
purged via downstream processing, however compounds 4 and 5 proved to be
problematic, due to the formation of persistent downstream impurities in subsequent
transformations, making their control critical to final product quality.
ACS Paragon Plus Environment
5

Organic Process Research & Development
Page 6 of 44
1
2
3
4
5
Scheme 2. Proposed reaction pathway for the formation of impurities in the reduction of
6
7
8
1.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
poor purge of
downstream impurities
H
O
OH
O
O
N
N
H
H
4
5
EtO
O
EtO
O
EtO
- i-Bu2AlOEt
O
EtO
O
4
3
DIBAL-H
H
O
O
OH
2
N
N
N
N
OEt
Al
OEt
H
O
i-Bu
i-Bu
int-1
1
2
3
While monoreduction at the 2-position was the dominant pathway, the overall
chemoselectivity was insufficient (Scheme 2), with formation of compounds 3, 4, and 5
occurring concomitantly with the formation of 2. To obtain sufficiently pure 2 for
downstream processing, additional purification steps (crystallization and column
chromatography) were necessary to remove the impurities arising from this
overreduction, resulting in significant yield loss (22% isolated yield). To better understand
the chemo- and regioselectivity of the conversion of 1 to 2, experiments were performed
ACS Paragon Plus Environment
6

Page 7 of 44
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
to further assess reaction enthalpy and temperature control, in addition to reaction
stoichiometry, which revealed a combination of factors affecting reaction performance.
The formation of tetrahedral intermediate int-1 (Scheme 2) was observed to be relatively
fast, on the order of 1-2 minutes to reach full conversion at -20 °C, based on lab
experiments. Collapse of this intermediate yields compound 2 which, in the presence of
additional DIBAL-H, will lead to the formation of impurity 3.5 While the stability of int-1
should be improved via presumptive chelation to the pyridine nitrogen, similar to the
chelation effect observed in Weinreb amides, the reaction temperature is still expected to
strongly affect the rate of this collapse. For this reason, similar reductions often employ
cryogenic temperatures (commonly -78 °C) to extend the intermediate’s lifetime, thereby
delaying its collapse until aqueous workup.7 Similarly, the formation of impurities 4 and 5
can occur by overreduction of the pyridine-4-carboxylate of int-1, or of compound 2 after
the collapse of int-1. Without the benefit of chelation, the 4-carboxylate is more prone to
complete reduction to form 5. Selectivity for 2- versus 4-carboxylate reduction was also
observed to be temperature-influenced, with higher temperatures leading to poorer 2-
carboxylate selectivity. On large scale, reactor cooling was limited to approximately -
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
7

Organic Process Research & Development
Page 8 of 44
1
2
3
4
5
6
7
8
9
20 °C, far above the typically employed cryogenic temperatures. This was suspected to
result in comparatively faster breakdown of int-1, as well as diminished 2-carboxylate
regioselectivity. Additionally, calorimetry experiments conducted on the DIBAL-H
reduction of 1 at -20 °C revealed an enthalpy of reaction of -319 kJ/mol, corresponding to
an adiabatic temperature rise of 119 K (further details on calorimetry measurements can
be found in the Supplementary Information). Control of this vigorous exotherm was only
possible by slow, metered addition of DIBAL-H, which required long addition times (>60
minutes at 18 kg scale). Coupled with the reaction’s sensitivity to temperature, these
extended addition times at -20 °C offered ample opportunity for both the breakdown of
int-1 and reduction at the 4-carboxylate position, exacerbating impurity formation during
scale-up.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Portion-wise dosing experiments indicated that the yield of compound 2 peaks at
approximately 1.2 equivalents of DIBAL-H, with conversion reaching completion at 1.3
equivalents (Figure 1). Even at these lower loadings of DIBAL-H, the formation of
impurities 3, 4, and 5 was significant in batch mode, occurring concomitantly with
formation of compound 2. These experiments suggested that batch or semi-batch
ACS Paragon Plus Environment
8

Page 9 of 44
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
reactions to form 2 would not be feasible, so alternatives to batch processing would need
to be explored to generate sufficient amounts of 2 the required purity for downstream
synthesis efforts.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. LC analysis of the reaction mixture during portion-wise addition of DIBAL-H.
Impurity 3 is not shown due to its poor absorbance at 257 nm and co-elution with toluene
from the DIBAL-H solution (refer to the Supplementary Information for more details).
While compound 1 represented a challenging substrate for selective monoreduction via
batch or semi-batch processing, we were encouraged by several reports in the literature
of successful lab-scale DIBAL-H reductions of mono-esters to aldehydes using
continuous flow chemistry,6a,8 including one example of such on >100 mmol scale.8d By
ACS Paragon Plus Environment
9

Organic Process Research & Development
Page 10 of 44
1
2
3
4
5
6
7
8
changing the reaction paradigm from batch to continuous flow, several advantages could
be realized:9
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
 greater control over reaction/residence time;
 improved heat transfer and temperature control due to increased surface area-
to-volume ratio (SAV) in flow; and
 access to operating temperatures outside the range of typical batch reactors.
To investigate the feasibility of developing a continuous manufacturing process, an
evaluation of different conditions was executed with a simple lab-scale flow reactor and
variable lengths of PFA tubing (Table 1).10 A full description of equipment, set-up, and
conditions can be found in the Experimental Details and Supplementary Information.
Initial runs at low concentration showed poor conversion, even at elevated temperatures
and extended residence times (Table 1, entries 1-6). Increasing the concentration of
compound 1 to 14.3 wt% in THF dramatically accelerated the reaction rate, enabling
complete conversion even at -40 °C (entry 7-9). Lower temperatures afforded incomplete
conversion even at extended residence times (entry 10), while higher temperature led to
increased impurity levels (entries 7, 8) even with reduced equivalents of DIBAL-H (entry
ACS Paragon Plus Environment
10

Page 11 of 44
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
13). These results highlight the sensitivity of this reaction to temperature. Ultimately,
shortening the residence time to 75 seconds and using a bath temperature of -40 °C with
1.5 equivalents of DIBAL-H yielded the most promising result (entry 12). As compared to
both lab-scale and pilot plant batch results (entries 14 and 15 respectively), the
continuous flow synthesis of 2 using these conditions resulted in dramatically lowered
impurity levels, most notably in the combined levels of impurities 4 and 5.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
11

Organic Process Research & Development
Page 12 of 44
1
2
3
4
5
Table 1. Screening of temperature, residence time (τ), DIBAL-H equivalents, and concentration for the reduction of 1 in
6
7
8
flow.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
cooling bath
pre-cooling
H
Al
reactor loop
DIBAL-H
25 wt% in PhMe
Impurities
quench loop
V = 13 mL
EtO
O
pre-cooling
EtO
O
EtO
O
H
O
OH
O
N
O
OH
O
O
N
N
N
N
OEt
H
H
H
1
pre-cooling
2
3
4
5
in THF
O
O
EtOAc
13 equiv.
Post-workupd LC %AN,
Compound #
[1] in
THF
Temp.a
Flow rate
(mL/min)b
τc
DIBAL-H
(equiv.)
Entry
Variable
(°C)
(s)
(wt%)
1
2
3
4
5
1
2
3
Temperatur
e
22
23.5
23.1
23.0
114
114
114
1.5
1.5
1.5
3.8
3.8
3.8
0.5
4.0
59.7 11.4
74.1 10.0
1.1
1.5
ND
24.1
9.1
5.6
0
(low conc.)
-20
20.2 59.7 12.0
ACS Paragon Plus Environment
12

Page 13 of 44
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Post-workupd LC %AN,
Compound #
[1] in
Temp.a
(°C)
Flow rate
(mL/min)b
τc
DIBAL-H
(equiv.)
THF
Entry
Variable
(s)
(wt%)
1
2
3
4
5
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
4
5
-40
-78
23.0
23.1
114
114
1.5
1.5
3.8
3.8
20.3 63.7
8.7
ND
5.7
Trac
e
89.9
9.1
0.1
ND
6
7
8
9
-78
0
23.1
23.0
23.0
23.2
228
114
114
114
1.5
1.5
1.5
1.5
3.8
79.3 19.3
0.6
7.5
5.6
0.5
ND
23.9
6.7
Temperatur
e
14.3
14.3
14.3
0.1
ND
47.9
60.0
17.3
27.1
-20
-40
(high conc.)
Trac
e
75.6
3.6
3.0
3.9
4.1
18.5
2.6
1.7
0.2
3.8
4.7
10
11
-60
-40
27.0
26.2
98
1.5
1.5
14.3
14.3
17.5 73.4
Time
101
Trac
75.0
e
16.8
2.9
12
-40
27.0
75
1.5
14.3
1.5
84.2
ACS Paragon Plus Environment
13

Organic Process Research & Development
Page 14 of 44
1
2
3
4
5
6
7
8
9
Post-workupd LC %AN,
Compound #
[1] in
Temp.a
(°C)
Flow rate
(mL/min)b
τc
DIBAL-H
(equiv.)
THF
Entry
Variable
(s)
(wt%)
1
2
3
4
5
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
13
DIBAL-H
equiv.
-20
-20
-23
23.2
N/A
N/A
115
N/A
N/A
1.3
1.5
1.5
14.3
1.6
72.4
6.0
15.9
3.5
14
15
Batch mode
(lab scale)
ca. 14.3
ca. 14.3
ND
0.5
47.0
53.0
4.9
5.4
25.0
4.0
21.6
34.0
Batch mode
(18 kg
scale)
a For flow experiments, bath temperature is given. For batch experiments, pot contents temperature is given.
b Represents the combined flow rates of DIBAL-H and 1 feeds in the reactor loop.
c τ = residence time.
d
LC results are reported after quenching remaining DIBAL-H with EtOAc, and partial aqueous work-up by stirring the
resulting organic reaction mixture over 33 wt% aqueous sodium potassium tartrate until layer separation occurred. %AN of
the organic layer was measured at a wavelength of 257 nm. <0.05 %AN = ND, <0.1 %AN = trace.
ACS Paragon Plus Environment
14

Page 15 of 44
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Using the chosen conditions (Table 1, entry 12), larger scale test runs were performed
using the above lab-scale setup by simply extending the pumping time, allowing for the
conversion of 500 g of 1 (2.24 mol) to 2 in a single run over approximately 4 h (Table 2).
Workup was accomplished using an in-line quench of remaining DIBAL-H with ethyl
acetate followed by delivery of the quenched reaction mixture into a reactor containing an
aqueous sodium potassium tartrate (Rochelle salt) solution to facilitate solubilization and
removal of the aluminum salts via a batch work-up process.11 Water and brine washes of
the organic layer, followed by filtration through silica for the purposes of color removal
and evaporation to dryness, yielded 2 which was carried forward into downstream
chemistry without further purification. This workup and silica treatment also demonstrated
the ability to purge the lower levels of compounds 4 and 5 generated using this flow
process (Table 2). This process was performed in triplicate at 500 g scale (entries 1-3),
affording improved yields ranging from 64-67% versus the 22% yield obtained during
previous batch manufacturing. This improved yield reflects the dramatically lowered
impurity burden, and lower product losses due to the lessened need for purification of the
resulting material.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
15

Organic Process Research & Development
Page 16 of 44
1
2
3
4
5
Table 2. 500
g
test run preparation of
2
under continuous flow.
6
7
8
cooling bath
-40 °C
pre-cooling
H
9
Al
reactor loop
 = 75 s
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
DIBAL-H
1.5 equiv.
25 wt% in PhMe
EtO
O
quench loop
 = 22 s
-40 °C
pre-cooling
EtO
O
O
N
O
-40 °C
OEt
N
1
H
500 g
2
pre-cooling
ca. 14.3 wt% in THF
64-67%
O
O
EtOAc
13 equiv.
Pre-workupa LC%AN,
Compound #
Final solidsb LC %AN,
Compound #
Flow
Run #
Yield
(%)
1
2
3
4
5
1
2
3
4
5
1
2
3
1.2
1.3
0.6
57.5 N/Ac 6.0
65.5 N/Ac 8.8
61.2 N/Ac 9.9
9.5
6.7
7.0
2.1
0.9
0.8
97.5 ND
98.7 ND
0.2
0.2
0.1
ND
ND
ND
67
67
64
98.7
0.2
a Purity of reaction mixture sampled directly from reactor outlet (after mixing with EtOAc,
but prior to aqueous workup). %AN was measured at a wavelength of 257 nm.
b Purity of final solids after aqueous workup and filtration through silica gel. %AN was
measured at a wavelength of 257 nm, <0.1 %AN = trace, <0.05 %AN = ND. Please refer
to the Experimental Section for further details.
c Compound 3 was not quantitated in pre-workup samples due to co-elution with toluene.
ACS Paragon Plus Environment
16

Page 17 of 44
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Following the success of the 500 g test runs, additional lab-scale experiments were
performed to better understand the mixing, flow rate, stoichiometry, and temperature
requirements of the reaction.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
While highly effective, the mixing device used to combine the feeds of compound 1 and
DIBAL-H in the 500 g test runs (a “caterpillar” split-and-recombine micromixer from Institut
für Mikrotechnik Mainz GmbH, IMM) was limited in its throughput due to significant back
pressure at higher flow rates, a known characteristic of this micromixer design.12 Several
sizes of simple T-mixer were subsequently evaluated to better understand the criticality
of mixing to product quality. After partial workup with aqueous Rochelle salt, product
distributions were noticeably different between mixing apparatuses (Table 3). At large
inner diameters of up to 3.18 mm (1/8”), both lower conversion and increased
overreduction were observed using a standard T-mixer (entry 1). At small inner diameters,
T-mixers proved similar in reaction outcomes to the caterpillar micromixer (entry 4). Thus,
while mixing proved influential in terms of reaction purity, adequate mixing did not require
specialized equipment, and mixing efficiency could be delivered with a much lower
pressure drop by using a simple T-mixer as opposed to the caterpillar mixer.12,13
ACS Paragon Plus Environment
17

Organic Process Research & Development
Page 18 of 44
1
2
3
4
5
Table 3. Effect of different mixing apparatuses on the preparation of 2 under flow
6
7
8
conditions.a
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Post-workupc LC %AN,
Compound #
Entry
Mixera
Sizeb
1
2
3
4
5
1
Caterpillar
micromixer
300 µL
0.7
85.0
6.3
2.7
0.8
2
3
T-mixer
T-mixer
T-mixer
3.18 mm
1.59 mm
0.254 mm
2.4
1.0
1.4
75.3
80.9
83.2
3.7
5.9
6.9
10.0
5.8
1.7
2.0
1.0
4
4.0
a
All entries used 1.5 equiv. DIBAL-H, a 75 second residence time, and a bath
temperature of -40 °C. For a detailed description of the standard flow conditions, please
refer to the Experimental Section.
b
Caterpillar mixer size represents internal volume. T-mixer sizes represent the inner
diameter.
c LC results are reported after quenching with EtOAc, and partial work-up by stirring the
resulting organic reaction mixture over 33 wt% aqueous sodium potassium tartrate for a
minimum of 4 h. %AN was measured at a wavelength of 257 nm.
A design-of-experiments (DoE) study was next initiated to examine and quantify the
effects of several factors on reaction purity. Factors selected for study included bath
temperature, residence time and DIBAL-H equivalents. Flow rate and mixing apparatus
ACS Paragon Plus Environment
18

Page 19 of 44
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
were held constant between experiments, as mixing was not evaluated as part of the
study. The study was performed using a 3-level face-centered central composite design
(CCF), which included 15 individual experiments with statistical analysis performed using
Fusion QbD software.14
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In examining the generated dataset, the two responses selected for optimization were
maximization of the pre-workup %AN of 2 and minimization of the combined pre-workup
%AN for 4 and 5. The model fit for both responses was high (R2 = 0.9881; R2
=
[2]
[4+5]
0.9938). For both responses, within the ranges studied, temperature and DIBAL-H
equivalents proved to be the factors with the strongest effects on reaction outcomes,
similar to observations in earlier condition screening.
Using the obtained data, limits were then set within the generated model to keep levels
of 2 >88 %AN and 4 + 5 <6 %AN, and the model was used to predict a design space
satisfying both of these criteria, that maximized yield of 2. These studies demonstrated
that the previously identified optimal conditions reside within the chosen design space
(Figure 2c), and were near the model’s predicted maximum in terms of yield of compound
2 (Figure 2a). Model verification and refinement experiments demonstrated that modest
ACS Paragon Plus Environment
19

Organic Process Research & Development
Page 20 of 44
1
2
3
4
5
6
7
8
9
improvements in reaction outcomes could be obtained using alternate conditions, but that
such improvements would necessitate longer residence times of up to 120 seconds. For
the full dataset generated by this study, including model refinement and verification runs,
model term ranking, and regression and ANOVA statistics, please refer to the
Supplementary Information.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2. a) %AN 2 response surface at different DIBAL-H loadings and temperatures.,
3-D (left) and 2-D (right). b) 2-D %AN 4+5 response surface at different DIBAL-H loadings
ACS Paragon Plus Environment
20

Page 21 of 44
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
and temperatures. c) superimposed limits for 1 (<4 %AN, blue), 2 (>88 %AN, green), and
4 + 5 impurity contents (<6 %AN, red) yield a predicted design space shown in white, with
crosshairs indicating the chosen setpoints for DIBAL-H equivalents (1.50) and
temperature (-40 °C). All contours use a constant residence time of 80 seconds, the
center point in the range of residence times studied. All images were generated using
Fusion QbD software version 9.9.0. %AN values measured and predicted at 242 nm.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
With a thorough understanding of the reaction’s performance, a reactor system was
assembled to perform a 20 kg scale reduction of compound 1 in continuous flow. A
simplified representation of the flow reactor design and the chosen reaction parameters
can be viewed in Scheme 3.
DIBAL-H, a 13.9 wt% stock solution of compound 1 in THF, and EtOAc were pumped
through the system using gear pumps to ensure smooth and pulseless flow. Flow was
monitored using floor scales, allowing direct measurement of mass flow for both the
DIBAL-H and compound 1 stock solution feeds. Flow rates were then manually controlled
by adjusting the speed of each pump as needed to maintain target flow rates. Pump
ACS Paragon Plus Environment
21

Organic Process Research & Development
Page 22 of 44
1
2
3
4
5
6
7
8
discharge pressure was maintained at 40 psi using manually-adjusted back-pressure
regulators to help maintain consistent process flow rates.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Stainless steel shell and tube heat exchangers were used both for pre-cooling all
reagent feeds and as reaction and quench spaces. Each heat exchanger consisted of 37
tubes of 1.88 mm inner diameter inside an outer jacket. The reactor space had a total
volume of 77.5 mL and the quench space a volume of 40.6 mL. This design allowed
narrow tubing widths to improve temperature control and heat transfer,9b,15 while also
reducing overall path length to mitigate system pressure drop. Jacket temperature of the
heat exchangers was controlled using a recirculating chiller, and both the process and
coolant temperatures were monitored using in-line thermocouple temperature indicators.
On the process side, these indicators were located at the outlets of all pre-cooling heat
exchangers and at the reaction space outlet.
Workup was accomplished similarly to lab-scale conditions, by collecting the reaction
eluent directly into a stirred solution of Rochelle salt in water at room temperature. The
organic layer was subsequently washed with water and brine to produce a stock solution
ACS Paragon Plus Environment
22

Page 23 of 44
Organic Process Research & Development
1
2
3
4
5
of compound 2. Detailed equipment, process, and operation descriptions can be found in
6
7
8
the Experimental Details and the Supplementary Information.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 3. Illustration of the flow reactor design used for the multikilogram-scale
synthesis of 2.
pre-cooling
TI
-40 °C
DIBAL-H
1.5 equiv.
25 wt% in PhMe
Target 19.1 g/min
0.0336 mol/min
reaction space
V = 77.5 mL
TI
= 75 s

1.0 mm ID t-mixer
cooling bath, -40 °C
-40 °C
EtO
O
EtO
O
quench space
V = 40.6 mL
 = 26 s
pre-cooling
-40 °C
TI
O
O
-40 °C
N
N
H
OEt
2
pre-cooling
-40 °C
1
TI
Theor. 0.24 kg/h
1.34 mol/h
0.0224 mol/min
13.9 wt% in THF
Target 36.0 g/min
( = 5.0 g/min pure)
0.0224 mol/min
EtOAc
~13 equiv.
Target 26.5 g/min
0.3 mol/min
Legend
TI
temperature
indicator
pump
back-pressure
regulator
shell and tube
heat exchanger
During scale-up, the synthesis of 2 was split into seven individually executed modules,
each at a scale of 1.1-3.4 kg input of 1, to allow the process to fit into the available batch
equipment for workup. Execution of the seven modules took a total of 14 days with each
module consisting of approximately 11-12 hours of pumping time, and approximately one
ACS Paragon Plus Environment
23

Organic Process Research & Development
Page 24 of 44
1
2
3
4
5
6
7
8
9
day for workup. From these combined modules, 10.85 kg of product was obtained (66%
overall yield, as a 4.5 wt% stock solution in EtOAc and toluene by quantitative 1H NMR),
which was similar to our initial 500 g test run yields.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The stock solution of 1 used in manufacturing was produced using a newly-developed
process, wherein the solid hydrochloride salt of 1, 1•HCl, was freebased using
triethylamine in a slurry-to-slurry transformation in THF (Scheme 4). When the resulting
stock solution was used in the continuous flow manufacturing of 2, the residual presence
of the byproduct triethylamine hydrochloride (TEA•HCl) led to significant reactor fouling
in the pre-cooling heat exchanger for compound 1, leading to partial clogging of the
reactor. This fouling also affected the narrow tubing (1 mm ID) of the mixing apparatus
and the lines connecting it to the heat exchanger, and resulted in higher system pressure
drop during operation, increasing from approximately 12 psi at the beginning of a module
to as high as 30-35 psi near the end. Because these pressures remained below the 40
psi setpoint of the manual back pressure regulators with which the pumps were each
equipped, pump performance was easily maintained throughout each module. Between
ACS Paragon Plus Environment
24

Page 25 of 44
Organic Process Research & Development
1
2
3
4
5
modules, flushing with room-temperature water, followed by THF, was sufficient to clear
6
7
8
the lines of the fouling.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 4. Freebasing of 1•HCl by triethylamine to produce 1 freebase as a stock solution
in THF.
CO2Et
CO2Et
NEt3
+
• HCl
CO2Et
1•HCl
NEt3 • HCl
+
THF
22 °C
N
N
CO2Et
1
TEA•HCl
15 ± 1.5 wt% in THF
While the stock solution of 1 had been previously polish filtered as part of the freebasing
process, and both feed lines for compound 1 and DIBAL-H were equipped with in-line
filters to prevent solids ingress, solubilized TEA•HCl precipitated from solution upon
cooling to -40 °C in the heat exchanger, leading to significant buildup in the lines. This
issue was confirmed by cooling a sample of the 1 stock solution to -20 °C and crystallizing,
1
isolating and characterizing by H NMR the resulting TEA•HCl salt. Based on these
studies, it was determined that the stock solution contained approximately 5 mol% of
TEA•HCl relative to 1.
ACS Paragon Plus Environment
25

Organic Process Research & Development
Page 26 of 44
1
2
3
4
5
6
7
8
9
The presence of such quantities of TEA•HCl in the reaction stream also led to lowered
reaction conversion due to partial quenching of DIBAL-H. This lowered conversion was
most pronounced in modules 1 and 2, (Table 4, entry 2), and could be reproduced by
using the stock solution of 1 from manufacturing in smaller-scale experiments on the lab-
scale flow system (Table 4, entry 3), confirming that it was not equipment-specific.
During manufacturing, this observed lowered conversion was counteracted by slightly
increasing the flow of DIBAL-H in all modules to 20 g/min (an increase of approximately
5%, translating to an additional 0.075 equiv. of DIBAL-H relative to 1) to drive the reaction
to completion. While the reaction conditions were robust enough to tolerate up to 10%
(0.15 equiv.) excess DIBAL-H with minimal impact to final purity after workup (details on
ranging studies can be found in the Supplementary Information), this increased quantity
of DIBAL-H led to complications in the resulting aqueous workup. With an increased flow
rate of DIBAL-H—and thus a larger excess entering the workup vessel—the aqueous
workups with sodium potassium tartrate were also observed to become exceedingly dark
and opaque, large amounts of insoluble particulates also forming.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
26

Page 27 of 44
Organic Process Research & Development
1
2
3
4
5
Table 4. Summary of pre- and post-workup sample analyses for flow modules 1-7, and
6
7
8
comparison to lab-scale experiments.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Pre-workupa LC %AN
Compound #
Post-workupb LC %AN
Compound #
Entry
Experiment
Scale
1
2
3
4+5c
1
2
3
4+5c
1
Lab-scale
30.0 g ND 88.0 1.3 8.3 ND 94.4 1.4 0.4
standard conditions
2
Mfg. module 1
Mfg. module 2
Mfg. module 3
Mfg. module 4
Mfg. module 5
Mfg. module 6
Mfg. module 7
Lab-scaled
1.1 kg 2.0 85.7 0.7 8.6 2.2 88.7 1.0 0.4
3.4 kg 2.2 84.0 0.9 8.4 2.6 88.8 1.3 0.5
3.4 kg 1.1 83.5 0.8 10.8 1.4 89.0 1.5 0.4
3.4 kg 1.3 84.5 0.7 9.2 1.6 88.1 1.5 0.6
3.4 kg 1.2 84.6 1.0 9.7 1.3 89.3 1.4 0.7
3.4 kg 1.2 84.9 0.8 10.1 1.3 88.9 1.1 0.5
2.0 kg 1.0 85.1 0.8 11.2 1.2 87.9 1.1 0.5
30.0 g 2.6 90.1 1.1 6.0 2.7 89.1 1.5 0.6
3
4
Post-color removal, 30.0 g
lab-scalee
—
—
—
—
1.7 94.1 0.5 0.2
a
Samples taken directly from reactor outlet (after mixing with EtOAc, but prior to
aqueous workup). %AN measured at 242 nm. <0.05 = ND, ≤0.1 = trace. Values presented
are averages across all pre-workup samples taken throughout each module. For sampling
procedure, please refer to the Supplementary Information.
ACS Paragon Plus Environment
27

Organic Process Research & Development
Page 28 of 44
1
2
3
4
5
b Purity of organic stock solution after full aqueous workup. %AN measured at 242 nm.
6
7
c Compounds 4 and 5 co-eluted in this analytical method and were integrated together.
8
9
d
Results obtained using a sample of the stock solution of compound 1 from
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
manufacturing modules 1-7 and lab-scale flow system.
e Details can be found in the Experimental Section and Supplementary Information.
While the solid particulates could not be conclusively identified by analysis, this dark
color had been observed on lab scale when using larger excesses of DIBAL-H in
continuous flow, and was hypothesized to be related to the quantity of aluminum species
present during the aqueous workup. The color change and particulate formation also
seemed to have negative effects on the impurity purge during the workup, resulting in
lower, albeit still acceptable post-workup purity for the manufacturing modules (Table 4,
entry 2), which aligned with our observations when this chemistry was replicated on lab
scale (Table 4, entry 3). Although the purity profile of the resulting darkly colored stock
solution was still acceptable for the material’s use in downstream chemistry on the basis
of use tests, we opted to upgrade the color and remove the insoluble particulates using a
simple filtration through a pad of silica, similar to the color removal procedure performed
in the 500 g test runs. This also offered a modest purity upgrade (Table 4, entry 4),
ACS Paragon Plus Environment
28

Page 29 of 44
Organic Process Research & Development
1
2
3
4
5
6
7
8
allowing successful use of the material in downstream chemistry to yield on-spec target
compound.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
CONCLUSION
The scale-up of the selective DIBAL-H mediated reduction of a highly challenging
heterocyclic diester to the corresponding mono-aldehyde was accomplished through use
of continuous flow chemistry. The rapid, highly exothermic, and temperature-sensitive
reaction was successfully scaled from multi-gram lab scale to a multikilogram scale of 20
kg input of compound 1 using a flow reactor system. Use of continuous flow chemistry
offered a significant advantage over batch chemistry for this transformation in terms of
quality and yield due to inherently better control of temperature and residence time,
offering a dramatic improvement in yield.
EXPERIMENTAL SECTION
Batch Reduction of 1 to 2
1 (18.0 kg, 80.6 mol) was charged to the reactor as a stock solution in ethyl acetate. A
solvent swap was performed by repeated put-and-take vacuum distillations with 90 kg
THF charged on each iteration (up to 4 iterations, using a jacket temperature of ≤65 °C).
ACS Paragon Plus Environment
29