Continuous Liquid Vapor Reactions Part 1: Design and Characterization of a Reactor for Asymmetric Hydroformylation

Article abstract of DOI:10.1021/acs.oprd.5b00407

A research scale continuous reactor system was designed and developed for high pressure asymmetric hydroformylation (AHF) reactions with an 8 h reaction time. The continuous reactor achieved high k_La, low axial dispersion, and an 8 mL liquid holdup volume. The reactor consisted of 20 vertical bubble flow pipes-in-series, connected by small diameter tubing jumpers. This type of continuous reactor is proven to be scalable up to 360 L in our GMP pharmaceutical manufacturing plant for high pressure hydrogenation. The continuous reactor was used for the AHF of styrene and 2-vinyl-6-methoxynaphthalene catalyzed by rhodium(bisdiazaphospholane) (BDP) complexes. The CSTRs-in-series numerical model fit the experimental data better than the plug flow with dispersion model. Samples were taken along the length of the continuous reactor and used for kinetic data modeling. Vapor liquid mass transfer rate constants were about 3 orders of magnitude higher than reaction rate constants.

Full text of DOI:10.1021/acs.oprd.5b00407

Subscriber access provided by George Washington University Libraries
Full Paper
Continuous Liquid Vapor Reactions Part 1: Design and
Characterization of a Reactor for Asymmetric Hydroformylation
Martin David Johnson, Scott A May, Joel Calvin, Gordon Lambertus, Prashant B
Kokitkar, Clark R. Landis, Bradley R. Jones, Martha Leigh Abrams, and James Stout
Org. Process Res. Dev., Just Accepted Manuscript • DOI: 10.1021/acs.oprd.5b00407 • Publication Date (Web): 31 Mar 2016
Downloaded from http://pubs.acs.org on April 5, 2016
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 free 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 accessible to all
readers and 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.
Organic Process Research & Development 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 40
Organic Process Research & Development
1
2
3
4
5
6
Continuous Liquid Vapor Reactions Part 1: Design and Characterization of a
Reactor for Asymmetric Hydroformylation
7
Martin D. Johnson,†,* Scott A. May,† Joel R. Calvin,† Gordon R. Lambertus,† Prashant B.
Kokitkar,† Clark R. Landis,‡ Bradley R. Jones,‡‡, M. Leigh Abrams,‡ ‡, James R. Stout#
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
†Small Molecule Design and Development, Eli Lilly and Company, Indianapolis, Indiana 46285, Unites States
‡
Department of Chemistry, University of Wisconsin—Madison, 1101 University Avenue, Madison Wisconsin
53706, United States
#D&M Continuous Solutions, LLC, Greenwood, IN 46143, United States
ACS Paragon Plus Environment

Organic Process Research & Development
Page 2 of 40
1
2
3
4
TOC Graphic
5
6
7
360 L liquid in reactor
8 mL liquid in reactor
8
To 43 more pipes
53 mm
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
8 mm
To 18 more pipes
Scale
Down
Gas
Bubbles
45,000X
Gas and
Liquid In
Liquid slug:
~0.001 mL,
1-3 cm/s
3.7 m of tubing
(0.56 mm id)
between pipes
Gas
Bubbles
Gas and
Liquid In
4.27 m of tubing
(4.57 mm id)
between pipes
Liquid slug:
0.1 – 10 mL,
25-50 cm/s
ACS Paragon Plus Environment

Page 3 of 40
Organic Process Research & Development
1
2
3
4
Abstract
5
6
7
8
A research scale continuous reactor system was designed and developed for high pressure
asymmetric hydroformylation (AHF) reactions with 8 hour reaction time. The continuous
reactor achieved high kla, low axial dispersion, and 8 mL liquid holdup volume. The reactor
consisted of 20 vertical bubble flow pipesꢀinꢀseries, connected by small diameter tubing jumpers.
This type of continuous reactor is proven to be scalable up to 360 L in our GMP pharmaceutical
manufacturing plant for high pressure hydrogenation. The continuous reactor was used for the
AHF of styrene and 2ꢀvinylꢀ6ꢀmethoxynaphthalene catalyzed by rhodium(bisdiazaphospholane)
(BDP) complexes. The CSTRsꢀinꢀseries numerical model fit the experimental data better than
the plug flow with dispersion model. Samples were taken along the length of the continuous
reactor and used for kinetic data modeling. Vapor liquid mass transfer rate constants were about
3 orders of magnitude higher than reaction rate constants.
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
Key words
continuous reaction, plug flow reactor, pipes in series, hydroformylation
ACS Paragon Plus Environment

Organic Process Research & Development
Page 4 of 40
1
2
3
4
Introduction
5
6
7
8
Part 2 of this work (companion paper) documents the results of an asymmetric
hydroformylation (AHF) in a plug flow reactor1. That paper describes the efficient
enantioselective synthesis of (S)ꢀNaproxen via asymmetric hydroformylation of 2ꢀvinylꢀ6ꢀ
methoxynaphthalene with RhꢀBDP. Continuous reactors are advantageous for high pressure
hydroformylation reactions in the pharmaceutical industry because of safety, cost,2 and
consistent quality. There are many examples of reduced cost and improved safety and quality of
continuous reactions compared to batch.3 Hydrogen is an explosion hazard4 and carbon
monoxide has high toxicity.5 They both have relatively low solubility in many organic solvents
at atmospheric pressure and therefore often require high reactor pressures, which increases the
risk of gas releases and the amount of gas that could be released. The manufacturing scale
vertical bubble flow pipes in series reactors are designed to operate more than 98% liquid filled,
which minimizes the mass of hazardous reagent gas that could be released in the event of a leak
or rupture. Furthermore, continuous reactors made of pipes and tubing are relatively low cost for
achieving high pressure rating and sufficient heat and mass transfer rates, with low internal gas
volume.
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
One of the main challenges for this work was to design and develop a research scale
continuous hydrofomylation reactor capable of high pressures (60 bar), long reaction times (2 to
20 hours residence time), small reactor liquid volume (less than 10 ml), intermediate sample
points along the length or the reactor, optional catalyst preꢀactivation in flow prior to mixing in
substrate, very low liquid flow rates (0.02 ml/min or less), and high vapor liquid mass transfer
rates (kla about 0.1 sꢀ1). Specifically, the desired research scale continuous reactor would meet
the design criterial listed in Table 1. The reason for the long reaction times was to achieve high
ACS Paragon Plus Environment

Page 5 of 40
Organic Process Research & Development
1
2
substrate to catalyst ratio and minimize the quantity6 and removal burden of precious metals.7
While vertical bubble flow pipes in series is an effective reactor design of choice for continuous
high pressure hydrogenation reactions using homogeneous catalysts, it is difficult to scale this
type of reactor down to research scale and maintain high vapor liquid mass transfer rates, and
also meet all 13 of the design requirements in Table 1.
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
Table 1. Design criteria for research scale continuous hydroformylation reaction system
1. Mass flow metering of reagent gas with redundant
measurements of gas flow for different ratios of H2 and CO
2. mass flow metering of substrate solution and catalyst/ligand
solution
0.05 to 1.0 mmol/min,
+/- 5% of setpoint
0.001 to 0.1
Flows
ml/min, +/- 2%
2 to 20 hours
3. Mean hydraulic residence time (τ) adjustable
τ
4. Small reactor volume to minimize material usage in
development.
Volume
< 10 mL liquid holdup
1 to 70 bar,
+/- 0.1% of setpoint
kla > 0.01/s
Pressure
5. Reactor pressure control.
6. High vapor liquid mass transfer rates.
Within 2 °C of desired
reaction temperature
for >99% of reactor
length.
7. High heat transfer rates so that reaction is isothermal along
entire length of continuous reactor.
Heat and
mass transfer
8. Low axial dispersion.
D/uL < 0.025
9. Intermediate sample points along the length of the PFR, to
generate conversion versus time data.
10. Ability to significantly change syngas equivalents or reactor
pressure without significantly changing τ.
Extra
capabilities
11. Optional Pre-activation in flow for catalyst/ligand mixing with
syngas at reaction pressure and elevated temperature.
12. O2 free system. Inert transfers of catalyst/ligand solution
into feed containers/pumps in an inert fashion.
13. Scalable to commercial volume of >1000L
a Constant pressure +/- 0.1%. facilitates constant, accurate, precise reagent gas flow through the entire reactor.
This is necessary to establish the segmented flow needed for high vapor/liquid mass transfer.
The continuous membrane reactors developed by Ley and coꢀworkers are a good option
for research scale experimental continuous gas/liquid reactions8. Reagent gas diffuses into a
liquid filled tube from the pressurized gasꢀfilled shell side through permeable tubing. The tubeꢀ
inꢀtube semiꢀpermeable membraneꢀbased gas reactor has been used for continuous
homogeneously catalyzed hydrogenations8,9 and continuous reactions with CO.10 However, the
ACS Paragon Plus Environment

Organic Process Research & Development
Page 6 of 40
1
2
3
4
5
6
7
8
9
desire was to develop a reactor that could be scaled to >1000 L at manufacturing scale. The
larger volumes are needed for the longer residence times to reduce catalyst loading. Continuous
asymmetric hydrogenation reactions have also successfully been run in a helicoidal single
channel falling film micro reactor11, and a micro channel mixer followed by tube reactor12. A
recent review of continuous flow reactions at research scale using reagent gas was provided by
Baxendale, in which examples using CO, H2, and syngas were described.13 In addition, see
Cossar et al for a review of continuous hydrogenation reactions highlighting safety and
performance advantages of flow.14 Continuous hydrogenation with heterogeneous catalysts in a
packed bed, for example the Thales Nano HꢀCube®, often require short reaction times for
example less than 5 minutes.15 Furthermore, many continuous reactions with CO gas have been
done in microfluidic devices with fast reaction times on the order of 2 to 15 minutes.16 However,
one of the design criteria in this work was to enable flow chemistry with 2ꢀ20 hour reaction
times. We have run continuous gas/liquid reactions in coiled tube reactors with 12 hours
reaction time, and scaled these reactors up to 73 L. 17 However, the coiled tube reactors are not
practically scalable to volumes larger than about 200 L due to fabrication and heating/cooling
challenges. Instead, a vertical pipes in series reactor design was envisaged, where the liquid and
gas flow coꢀcurrently up through larger diameter bubble flow pipes and down through smaller
diameter jumper tubing, with 15 to 45 pipes in series.18 Compared to the coiled tube reactors,
this reactor type is scalable to larger volumes, it operates at higher percent liquid filled, it has
higher vapor liquid mass transfer rate in the small diameter jumpers, and the internal surfaces of
the straight pipes and jumpers are more easily inspected after cleaning.
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
Continuous processing is gaining importance in pharmaceutical development,
manufacturing, and academic research.19 Quality control advantages of continuous processing
ACS Paragon Plus Environment

Page 7 of 40
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
can be achieved because of steady state operation and realꢀtime product quality information by
onꢀline process analytical technology. Real time adjustments can be made to keep product
quality high at all times. This aligns with quality by design principles, and implementation of
continuous processes is supported by the FDA.20 ICH Q8 is a guide to design a quality product
and its manufacturing process to consistently deliver the intended performance of the product.
The guideline explains that a process is designed to consistently meet product critical quality
attributes, and that the process is continually monitored and updated to assure consistent quality
over time. A steady state process (continuous) with onꢀline analytical is more intrinsically
suited for consistency than a transient process (batch). In this manuscript and the Supporting
Information, a custom onꢀline HPLC system is described, which was developed and used to
convert a standard offꢀline HPLC into a fully automated onꢀline system.
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
First generation coiled tube reactor system suffered from poor mass transfer
The first generation coiled tube reactor system gave slower reaction rates than batch.
Continuous AHF of styrene was run using a coiled tube reactor similar to the lab scale reactor
that we reported in a previous study for asymmetric hydrogenation of an enone.17 The coiled
tube reactor was 68 ml, made from a coil of 1.75 mm i.d. stainless steel tubing. The reactor used
standard 1/8ꢀinch o.d. commercially available 316L stainless steel tubing, high pressure pumps,
and a commercially available back pressure regulator.21 A picture of the coiled tube reactor and
the constant temperature oil bath used to submerge the reactor is shown in Figure 1.
ACS Paragon Plus Environment

Organic Process Research & Development
Page 8 of 40
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
a
b
Figure 1. Picture of (a) 68 mL coiled tube continuous reactor and (b) constant temperature oil
bath.
The two continuous feed solutions were 2.01 M styrene in toluene and 7.766 mM catalyst
in toluene. Reagent gas was 50:50 CO:H2. The reaction schematic is shown in Figure 2.
Rh(CO)2(acac), (S, S)-Ph-BPE
S/C 2000
O
O
O
CO/H2 (200 psi)
Toluene, 80 °C
Ph
Ph
(S)-2-phenyl-
propionaldehyde
(R)-2-phenyl-
propionaldehyde
3-phenyl-
propionaldehyde
P
P
(S)-Branched
Ph
(R)-Branched
Linear
Ph
(S, S)-Ph-BPE
Figure 2. Asymmetric Hydroformylation of Styrene with Rhꢀ(S,S)ꢀPhꢀBPE
Two sets of flow conditions were used in the continuous AHF experiments, as listed in
Table 2. Conversion versus time is shown in Figure 3, and B/L ratio and ee versus time are
shown in Figure 4. With 2 hour τ, conversion was about 47%, and with 4 hour τ, conversion was
about 65%. The data in the figures was generated using an onꢀline GC platform, as described in
the experimental section.
ACS Paragon Plus Environment

Page 9 of 40
Organic Process Research & Development
1
2
Table 2. AHF of styrene in first generation coiled tube continuous reactor at 80 oC and 200 psig
Syngas.
3
4
5
6
7
8
9
styrene
mean
solution
residence
time,
catalyst flow conversion B/L
ee
flow
rate
τ
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
0.0987
mL/min
0.0493
mL/min
0.0128
mL/min
0.00638
mL/min
35-
45
2 hour
47%
65%
45-50%
35%
4 hours
15
Gas/liquid volumetric flow rate was about 5/1 but not consistently controlled.
80
4 hour residence
time
70
2 hour residence
60
50
40
30
20
10
0
0.00
5.00
10.00
15.00
20.00
25.00
30.00
Reaction Time (hours)
Figure 3. Conversion of styrene in first generation continuous AHF reactor.
ACS Paragon Plus Environment

Organic Process Research & Development
Page 10 of 40
1
2
3
4
5
6
7
8
9
50
45
40
35
30
25
20
15
10
5
70
60
50
40
30
20
10
0
2 hour residence
time
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
4 hour residence
time
B/Lratio
ee
0
0.00
5.00
10.00
15.00
20.00
25.00
30.00
Reaction Time (hours)
Figure 4. B/L ratio and ee for styrene asymmetric AHF in first generation flow reactor.
For comparison, the reaction was run batch in an Endeavor® reactor at 80 °C and 100,
200, 300 psig, using the same feeds solutions as the continuous reactions.22 The batch reactions
achieved 100% conversion in 4 hours. The continuous flow conversion versus time did not
match batch. Furthermore, low selectivity in flow pointed to poor vapor liquid mixing. One
problem was poor control of gas mass flow rate and reactor pressure. The commercially
available back pressure regulator was allowing the pressure to vary between about 198ꢀ202 psig.
This resulted in liquid and gas segments that were very long, causing low vapor/liquid interfacial
surface area. Some liquid segments were 0.24 mL volume. The smallest liquid segments were
about 0.01 mL. Vapor liquid mass transfer would be much better if all liquid segments were
about 0.01 mL because interfacial surface area in the reactor would be much higher. This was
verified visually by changing the first 3 feet of the reactor tube from stainless steel to transparent
PFA, so that the bubble lengths and liquid slug lengths could be directly observed. The large
internal diameter of the tubing (1.75 mm) was another reason for larger liquid segment volume in
the tubes and thus lower gas/liquid interfacial surface area. Linear velocity in the tube reactor
ACS Paragon Plus Environment

Page 11 of 40
Organic Process Research & Development
1
2
3
4
was only about 24 cm/min at 2 h τ and 12 cm/min at 4 h τ. Data in the Supporting Information
5
shows that gas/liquid mass transfer rate is higher for higher linear velocities. Therefore, we
sought a reactor with higher gas/liquid interfacial surface area and higher linear velocities, which
should lead to higher gas/liquid mass transfer rates.
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
Second generation coiled tube reactor system had better gas+liquid mixing
The second generation coiled tube reactor system was designed for better pressure
control, better gas mass flow rate control, and smaller diameter tubing. Takebayashi et al had
shown that changing to smaller i.d. tubing, 0.5 mm versus 1.0 mm, made a dramatic
improvement in reaction efficiency for segmented flow with CO reagent gas,23 reasoning that it
was because of the increased surface area of CO interface per unit reactor volume. Likewise, our
second generation coiled tube reactor achieved about 10ꢀ300X smaller liquid segment size and
thus much higher interfacial surface area for gas/liquid mass transfer. Changes to the reactor
inlet and outlet side flow control and pressure control resulted in less oscillations in pressures
and gas flow rate, and much smaller and more consistent liquid slug length and gas bubble
length. The reꢀdesigned method of pressure control allowed the reactor outlet pressure to be held
very close to setpoint, with less than 0.1% variation, for example 200.0 psig ± 0.2 psig. A
pressure versus time trend is shown in the Supporting Information. No pressure regulators were
used for pressure control, only sequenced automated block valves, finite volume chambers, and
pressure transmitters. This is described in more detail in the experimental section and
Supporting Information.
Using the revised reactor system design, a hydroformylation reaction with styrene was
run in a 14.3 mL reactor made from standard 1/16ꢀinch o.d. 316L stainless steel tubing. The
ACS Paragon Plus Environment

Organic Process Research & Development
Page 12 of 40
1
2
3
4
5
6
reactor tube was 0.56 mm i.d. and 58 m long, coiled and submerged inside a constant
temperature bath.
7
8
9
Conversion versus time matched closely between the second generation continuous
coiled tube reactor and the batch reactor. The reaction scheme was shown in Figure 2.
Flow conditions for the styrene AHF reaction were 80 °C, 200 psig pressure, 2.03 M
styrene in toluene, S/C = 2000, total liquid flow rate of 0.02 ml/min, gas feed rate 0.13
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
mmol/min at 50:50 CO:H2, and about 60 minutes τ. Conversion was 60%, ee was 60ꢀ70%
(rising with time from about 60% after one reactor turnover to about 70% after 10 turnovers),
and B/L was ∼40. With the same styrene and catalyst solutions used in the continuous reactor, a
batch reaction which was run for 16 hours and achieving 100% conversion resulted in ee of 61%
and the B/L was 42. The ee and B/L results were almost identical to the continuous results.
Five batch reactions were done using an Endeavor, using 2 mL of the styrene solution
and 0.25 mL of the catalyst solution in each reactor (S/C 2000). Various pressure and
temperatures were screened. Results for the 16 hour reactions are shown in Table 3.
Table 3. Batch AHF of styrene with Rhꢀ(S,S)ꢀBPE.
temp pressure conversion
Entry
B/L
ee (%)
(oC)
60
60
80
80
80
(psig)
100
400
100
200
400
(%)
93
1
2
3
4
5
55.5
50.7
33.4
42.2
40.6
89.5
90.9
59.2
60.9
63.8
52
100
100
98
In the batch reactions, ee was higher (∼90%) and the B/L ratio also increased to > 50 at
60 °C and 400 psig. However, the conversion was lower at 60 °C with the 100 psi reaction
reaching only 93% conversion after 16 hours and the 400 psi only 52%. At 80 °C the ee was ~
60% and the B/L ratio was between 33ꢀ42. Reaction rate was faster at lower pressure.
ACS Paragon Plus Environment

Page 13 of 40
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Conversion after 60 minutes in the continuous reactor was as high as conversion after 120
minutes batch, and B/L ratio and ee were similar. As described in the companion paper1, it has
been shown that regioꢀ and enantioselectivity of styrene hydroformylation as catalyzed by
rhodium(bisdiazaphospholane) complexes are sensitive to the CO partial pressure, with low
partial pressures giving low selectivity.24 The Endeavor provides exceptional mixing (high kLA)
and therefore maximum selectivity would be expected25. The selectivity from the continuous run
matched that observed in the batch run, thus evidence that vapor liquid mixing rate was sufficient
to maintain high CO concentration in solution.
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 improvement to gas flow control and smaller diameter tubing in the 2nd generation
coiled tube allowed for much better results in flow, both the 1st and 2nd generation reactors fail
to meet design criteria 3, 9, 10, 11, and 13 listed in Table 1. Notably, it was difficult to achieve
desired residence time accurately with the coiled tube design. In order to achieve a desired
% liquid filled must be known and maintained. depends on conversion, and conversion
depends on , because % liquid filled depends on how much of the gas reacts. Thus, we sought a
τ, the
τ
τ
reactor design for which liquid
conversion.
τ was not so significantly impacted by gas flow rate or reaction
Measurement of vapor/liquid mass transfer rates in small diameter tubing
One of the challenges in flow reactors is quantifying vapor/liquid mass transfer rates,
especially at research scale. It is difficult to make direct measurements of dissolved gas
concentration as a function of time in micro tube reactors at low flow rates. In batch, the gas
uptake is measured by change in pressure versus time. As gas dissolves, pressure decreases.
This is not possible in the flow tube reactors. In the continuous reactors, pressure does not
change. There have been several examples of research scale gas/liquid reactions with high mass
ACS Paragon Plus Environment

Organic Process Research & Development
Page 14 of 40
1
2
transfer rates, on the order of 0.1 to 15 sꢀ1. Ley and coꢀworkers measured gas/liquid mass
transfer rates in membrane reactors using total liquid flow rates of about 1ꢀ2 ml/min. 8 Mass
transfer rate was measured indirectly, by measuring how much gas was evolved from the liquid
during degassing. A research group at University of Huddersfield measured vapor liquid mass
transfer rates in small scale bubble columns.26 In 3 mL bubble flow reactors, kla was on the order
of 0.1 sꢀ1. Hessel and coꢀworkers have used falling film micro reactors at research scale and
achieved kla 3 sꢀ1 to 8 sꢀ1 .27 Flow rate range was 0.5 to 3 mL/min and reactor pressure was 1 to 6
bar. A microchannel device with sequestered heterogeneous catalyst was used by Jensen and
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
coꢀworkers for the hydrogenation of cyclohexene, where kla was determined to be 5 sꢀ1 to 15 sꢀ
1 28
.
A reason for their high kla was the high vapor/liquid interfacial area in the micro structured
reactor. We also sought to achieve 0.1 sꢀ1 or greater in this present work, but to sustain the high
kla for longer (8 h) in the flow reactor, using <0.02 ml/min total liquid flow rate.
τ
In this present work, mass transfer rate of syngas (1:1 H2:CO molar ratio) from the gas
phase to the liquid phase was measured experimentally. The approach was to measure
dissolving rate of individual gas slugs moving through the reactor tubing, measure gas and liquid
slug size in the tubing during reaction conditions, and use these quantities to calculate kla.
Downstream from the stainless steel feed lines and the stainless steel mixing Tees, a section of
the 1.59 mm o.d. steel tubing was replaced with a transparent PFA tube (1.59 mm o.d. and 0.79
mm i.d.) so that a vapor slug length could be observed and measured as it traveled along the
length of the reactor tube. The higher the linear velocity in the tube, the higher the rate of gas
bubble dissolving. The maximum syngas bubble shrinking rate is about 0.3 cm/min which
occurs at the highest linear velocity tested, about 40 cm/min. The results are shown in the
Supporting Information. Syngas bubble shrinking rate of 0.3 cm/minute corresponds to
ACS Paragon Plus Environment

Page 15 of 40
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
dissolving rate of 0.00048 mmol/min for a single gas bubble flowing through the small diameter
tubing. This was used as a conservative approximation for gas slug dissolving rate in the AHF
reaction experiments. It was conservative because it was measured for lower linear velocity and
lower operating pressure and temperature than the actual AHF reaction experiments. The
measured gas dissolving rate and experimentally observed liquid slug size were used to calculate
kla = 0.1 sꢀ1, which corresponds to 46 seconds required to reach 99% of saturation. See
Supporting Information for details of the measurements and calculations.
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
Third generation reactor system achieved additional design criteria
There were still some important limitations of the second generation continuous reactor
system. Of the 13 design criteria listed in Table 1, the 2nd generation reactor only met 8 of the
criteria. This coiled tube design could not satisfy criteria for the following:
• Criteria 3: Adjustable mean hydraulic residence time (τ) between 2 ꢀ 20 h
• Criteria 9: Intermediate sample points along the length of the PFR, to generate
conversion versus time data.
• Criteria 10: Ability to significantly change syngas equivalents or reactor pressure
without significantly changing τ.
• Criteria 11: Optional Preꢀactivation in flow for catalyst/ligand mixing with syngas at
reaction pressure and elevated temperature.
• Criterial 13: Scalable to commercial volume of >1000L
Therefore, design changes were made to achieve these goals while retaining the previous criteria.
We had previously developed a pipesꢀinꢀseries reactor platform29 that was capable of achieving
the design criteria proven at scales up to 360 L, but a small research scale version (volume <10
mL) of this design did not exist. The advances in gas & pressure control developed in the coiled
ACS Paragon Plus Environment

Organic Process Research & Development
Page 16 of 40
1
2
3
4
5
6
7
8
9
tubes led us to explore the pipes in series design at a much smaller scale. Thus, 20 vertical pipes
in series were installed at equal distances along the 73 m long, 0.56 mm i.d. tube reactor. The
reason that 20 pipes were chosen was to maintain low overall axial dispersion. If most of the
liquid is in the 20 pipes, and if each of them has infinite axial dispersion, then the RTD is not
worse than that of 20 CSTRs in series, which is approximately the same as a PFDR with axial
dispersion number 0.024. This means 99% pushout is accomplished after 1.5 volume turnovers.
Flow of gas and liquid were coꢀcurrent in the upward direction through each of the vertical pipes.
Internal dimensions for each 0.3 mL pipe were 8 mm wide and 7.3 mm tall, with a 1 mm tall
conical top and a 1 mm tall conical bottom.30
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 5. Research scale pipes in series design concepts
Figure 5 graphically describes the design of the research scale pipes in series reactor. Pictures of
the reactor are shown in the companion paper.1 The vertical pipes remained almost completely
ACS Paragon Plus Environment

Page 17 of 40
Organic Process Research & Development
1
2
3
4
5
6
liquid filled at all times because the H2/CO reagent gas bubbled up through the liquid in each
pipe. This allowed the system to run at higher gas flow rates without significantly changing the
7
8
9
volume of liquid in the vertical pipes. The revised reactor easily operated with 8 h τ. Total
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
reactor volume was 23.8 mL. Volume in the 20 vertical upflow bubble pipes was 6 mL and
volume in the 0.56 mm i.d. connecting tubes was 17.8 mL. The 0.56 mm i.d. connecting tubes
operated 90ꢀ93% vapor filled at baseline conditions. The total liquid volume of the reactor was
~8 mL which met our design criteria. The reactor was designed to provide long residence time
in the bubble flow pipes, and to provide high vaporꢀliquid mass transfer rates in the 0.56 mm i.d.
connecting tubes.
High vaporꢀliquid mass transfer was achieved by establishing very small liquid segments
flowing through the small diameter tubing. Typical liquid slug size in the 0.56 mm i.d. tubing
was on the order of about 0.001 mL, as verified by replacing a length of stainless steel tubing
with transparent PFA tubing at the lower pressure conditions. A picture of ∼0.001 mL liquid
slugs and ∼0.01 mL vapor slugs flowing in the tubing is shown in Figure 6. Flow conditions
were 150 psig, 0.0167 mL/min total liquid feed rate, and 0.161 mmol/min gas feed rate.
ACS Paragon Plus Environment

Organic Process Research & Development
Page 18 of 40
1
2
3
4
5
Figure 6. Picture of liquid segments (smaller) and vapor segments (larger) flowing through 0.79
mm i.d. transparent tubing.
6
7
8
9
Jumper length was chosen so that liquid would saturate with syngas before the first pipe,
and it would subsequently reꢀsaturate with syngas between each pipe. Table 4 shows calculated
time in each 3.66 m jumper for each of the experiments, which was about 140 to 270 seconds
depending on the experimental conditions. If kla is 0.1 secꢀ1, then time to reach 99% of saturation
is 46 s. Thus, jumper length was conservatively chosen so that time in jumper between pipes
would be at least 3X longer than needed to reꢀsaturate.
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
Table 4. Calculated time in 0.56 mm i.d. tubing jumper between each 0.3 mL vertical pipe during
flow reaction of 2ꢀvinylꢀ6ꢀmethoxynaphthalene.
linear
gas feed
rate
(mmol/min) i.d. tubing
(cm/min)
velocity in
0.56 mm
time in 0.56 mm i.d. tubing jumper
between vertical pipes (seconds)
Reaction*
P (psig)
reaction1
reaction2
Reaction3
400
400
800
0.2
0.22
0.651
82
89
268
247
158
139
Reaction4
Reaction5
150
400
0.161
0.244
159
97
138
227
*Reaction temperature = 70 oC,
Total liquid feed rate = 0.0167 mL/min,
A 14.3 mL 0.56 mm i.d. 316L stainless steel tube was installed for optional preꢀactivation
in flow. It was the same coiled tube that had been previously used for second generation
continuous reactor. It was used for catalyst/ligand mixing with syngas at reaction pressure and
elevated temperature for controlled amount of time. 3ꢀway valves were installed so that the
operator could switch to either flow through preꢀactivation tubing section or bypass it. The small
diameter tubing was selected for segmented flow, so that catalyst molar flow rate was
independent of the amount of toluene that strips off into the syngas in the preꢀactivation section,
and thus independent of the excess molar equivalents gas used in the reaction. This preꢀ
activation loop was submerged in a separate constant temperature oil bath, so that preꢀactivation
temperature could be set independently from reaction temperature.
ACS Paragon Plus Environment

Page 19 of 40
Organic Process Research & Development
1
2
3
4
Quantifying axial dispersion and mean residence time in continuous reactor
5
6
7
8
Low axial dispersion was desired, so that the impact of dispersion on conversion versus
would be minimal, and so that residence time distribution (RTD) would be narrow. Axial
τ
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
dispersion number was measured to be D/uL = 0.014ꢀ0.018, and the reactor had RTD equivalent
of about 32 to 38 equal sized continuous stirred tank reactor (CSTR)s in series, depending on the
flow conditions. There is a detailed section on quantifying D/uL, equivalent number CSTRs in
series, and τ in the Supporting Information. Fꢀcurves were measured and the data was modeled
by the plug flow with dispersion reactor (PFDR) model and the CSTRs in series model. The
CSTRs in series model fit the RTD profile better than the PFDR model for this reactor (Table 5).
Interestingly, the model fit for τ was about 0.1 to 0.2 hours longer using the CSTRsꢀinꢀseries
model compared to the PFDR model. In these 4 RTD measurements, the average axial
dispersion number was 0.015 when the PFDR model was used, and the average number of
CSTRs in series was 36 when that model was used. Each model was fit to the data by
minimizing the sum of squared residuals (SSR). By comparing minimum SSR for the model fits
in the table, it is clear that the CSTRs in series model fits the experimental data better (Table 5).
Therefore the
τ
determined by fitting the CSTRs in series model was considered more
. All of the Fꢀcurve data and model fits to the experimental data are
representative of actual
τ
shown in the Supporting Information.
Table 5. Summary of τ and RTD tests with nonꢀreacting solvents, fit to both the PFDR and the
CSTRs in series reactor models.
CSTRs
in
series
model,
number
of
Sum of
squared
residuals
model
fit
Sum of
squared
residuals
model
fit
CSTRs
in
series
model
PFDR
model
actual gas
flow
PFDR
model
D/uL
solvent
test
P
(psig)
(mmol/min)
τ
CSTRs
τ
PFDR
CSTRs in series
1
400
0.2
8.13
0.0137
339
8.25
38 122
ACS Paragon Plus Environment

Organic Process Research & Development
Page 20 of 40
1
2
3
4
5
6
2
3
4
400
150
800
0.116
0.14
8.78
7.25
8.28
0.0137
0.0162
0.015
272
377
163
9
38
32
35
39.8
97
7.4
8.45
7
0.513
18.9
8
9
Temperature = 70 oC.
Total liquid flow = 0.0167 mL/min
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 pipes dominate the RTD. We know from similar axial dispersion measurements that
D/uL is about 0.0002 to 0.0004 for solvent flow in 73 m long tubing with 0.56 mm i.d..31
Furthermore we know that axial dispersion number is smaller for 2ꢀphase gas + liquid segmented
flow in tube reactors,17 therefore D/uL would be less than 0.0002 for segmented gas + liquid
flow. Because D/uL is actually 0.015 on average for the measurements shown in Table 5, we
conclude that the 20 vertical pipes dominate RTD, making the reactor significantly less plug
flow but still achieving design goals.
Calculating mean residence time
One of the biggest experimental challenges was knowing
condition. It is much more difficult for gas + liquid reactions than for liquid only reactions. The
percent conversion depends on liquid , and the liquid depends on % conversion. 17.8 mL of
τ for each steady state reaction
τ
τ
the 23.8 mL reactor is the 0.56 mm i.d. tubing. Gas and liquid flow are segmented in the tubing.
A fraction of the gas dissolves in the liquid, a fraction of the gas is used up by reaction for which
rate decreases along the length of the reactor, a fraction of the gas passes all the way through the
reactor in excess, and a fraction of the solvent partitions into the gas phase. The gas and liquid
phases both expand in the reactor because of the higher temperature, although the gas phase has
greater thermal expansion than the liquid. We used RedlichꢀKwongꢀSoave (RKS) equation of
state (EOS) model for CO and H2 solubility in the toluene and did the numerical calculations
using ASPEN PLUS®. We also compared to other physical property methods, such as NRTLꢀIG
ACS Paragon Plus Environment

Page 21 of 40
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
with Henry’s Law, NRTLꢀRK with Henry’s Law, and PengꢀRobinson EOS. The choice of RKS
EOS was based on a comparison of the model predictions to literature data32 for hydrogen
solubility in toluene, which is shown in Figure 7. The other EOS methods and the NRTL models
predict significantly higher dissolved hydrogen concentration than RKS EOS.
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
Literature Data
PENG-ROB
NRTL-RK w/HL
NRTL w/HL
RK-SOAVE
0.04
0.035
0.03
0.025
0.02
0.015
0.01
0.005
0
0
200
400
600
800
1000
1200
Pressure / psig
Figure 7. Comparison of hydrogen solubility predictions from different thermodynamic models
with literature data.
For the RKS EOS method, parameters are available for H2ꢀToluene and H2ꢀCO systems but the
parameters for COꢀtoluene binary system are missing. No data was available for COꢀtoluene or
for COꢀH2ꢀtoluene systems to enable regression of those parameters. The prediction of toluene
in vapor phase from EOS and activity coefficient based models match each other well. The
predictions for gas concentrations in the liquid phase are likely to be accurate based on results in
Figure 7. A graphical depiction of the predictions is shown in Figure 8 and Figure 9 using RKS
EOS for CO and H2 solubility in the toluene.
ACS Paragon Plus Environment

Organic Process Research & Development
Page 22 of 40
1
2
3
4
5
6
Toluene in Vapor Phase
0.30
0.25
0.20
0.15
0.10
0.05
0.00
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
0.
200.
400.
600.
800.
1000.
Pressure / psig
Figure 8. Numerical prediction of toluene mole fraction in the vapor phase in the tubular reactor
at 70°C as a function of pressure.
CO and H2 Concentration in Toluene
CO
H2
0.25
0.20
0.15
0.10
0.05
0.00
0.
200.
400.
600.
800.
1000.
Pressure / psig
Figure 9. Numerical prediction of dissolved concentration of H2 and CO in the liquid phase in
the tubular reactor at 70°C as a function of pressure.
For the continuous reactions, gas flow was higher near the beginning of the reactor than the end,
because gas is reacting along the length. Gas uptake is highest near the beginning of the reactor
and slows down near the end because it is approximately a first order reaction. The liquid holdꢀ
up in the 0.56 mm tubing was calculated along the length of the reactor with the Beggs and
ACS Paragon Plus Environment

Page 23 of 40
Organic Process Research & Development
1
2
Brill33 correlation for 2ꢀphase flow using a pipeline block in ASPEN PLUS®. The coefficients
for the correlation were adjusted to provide the best fit for all 5 solventꢀonly cases. The values of
the 3 BeggsꢀBrill coefficients, A, α, and β for the best fit were determined to be 0.98, 0.96, and
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
0.10875, respectively.
The results of these calculations are shown in Table 6.
transition data (“measured ”) was included in the table for comparison. The predicted values
for are within 10% of the measured values, as seen in Table 6.
τ
was calculated for each of the 5 solvent axial dispersion experiments.
τ
quantified by modeling the Fꢀcurve
τ
τ
Table 6. Predicted and measured values for
τ in the solvent tests and the actual 2ꢀvinylꢀ6ꢀ
methoxynaphthalene reactions in the pipes in series continuous reactor.
predicted
from
numerical
calculations
(h)
τ
first order
reaction
rate
Reaction
measured τ
(h)
τ (h)
actual gas
feed
solvent
test
constant
P (psig)
400
(mmol/min) (1/h)
8.25a
9 a
7.4 a
8.45 a
n/a
n/a
n/a
n/a
solvent1
solvent2
solvent3
solvent4
0.2
n/a
8.15
9.85
7.11
7.60
400
0.116
0.14
n/a
n/a
n/a
150
800
0.513
n/d
n/d
8.3b
8.2 b
reaction1
reaction2
400
400
0.2
0.8
n/d
n/d
0.22
0.97
n/d
8.2c
Reaction3
Reaction4
Reaction5
800
150
400
0.651
0.161
0.244
0.55
0.72
1.29
n/d
n/d
n/d
7.5d
8.5 d
7.5
8.5
7e
n/a
solvent
pushout
400
0.258
n/a
7.62
Temperature = 70 oC
Olefin molarity in the reactions was 0.532 M, and olefin solution feed rate was 0.01546 mL/min.
Total liquid feed including catalyst solution was 0.0167 mL/min. Total toluene feed for the
solvent tests was 0.0167 mL/min.
a from fitted CSTRs in series model to Fꢀcurve transition data
b Not measured during actual flow reaction experiment. Use the same
run 1, corrected for difference in gas feed rate and gas reacted.
τ
as measured for solvent
c Not measured during actual flow reaction experiment. Use the
τ measured for the 800 psig
solvent run, but corrected for the higher gas feed rate compared to the solvent run and corrected
for the amount of gas that reacts.
ACS Paragon Plus Environment

Organic Process Research & Development
Page 24 of 40
1
2
3
4
5
d from midpoint of transition from previous steady state.
e from shutdown transition curve.
6
7
8
9
Experimental Results for AHF reactions in flow
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 third generation reactor was used for AHF of 2ꢀvinylꢀ6ꢀmethoxynapthalene with Rhꢀ
bisdiazaphospholane catalysts, shown in Figure 10.
Figure 10. AHF of 6ꢀmethoxyꢀ2ꢀvinylnaphthalene
A summary of the continuous reaction results are given in Table 7. These results are discussed
in the companion paper.1
Table 7. Experimental Results for continuous AHF of 2ꢀvinylꢀ6ꢀmethoxynapthalene with Rhꢀ
bisdiazaphospholane catalyst.*.
%
%
%
conv.
at
Reaction τ
pre-
actual gas
feed
conv. conv.
at at
pipe3 pipe7 pipe13
(h)
Reaction activation
Conversion
%
#
1
loop?
P (psig)
400
(mmol/min)
B/L
Ee
8.3
8.2
no
0.2
98
15-17
92
62.6
89.9
97.3
yes, 1.28h
at 60 C
2
3
4
5
400
800
150
400
0.22
0.651
0.161
0.244
98
97
98
97
16-17
92
63.7
92.2
81.2
84
97.1
93.8
95.3
96.8
8.2
7.5
8.5
no
no
no
11-12 79-80 43.2
9
88-89 51.9
12
90 78.4
90.6
*Temperature was 70 oC, Olefin molarity in the reactions was 0.532 M,Olefin solution feed rate was 0.01546
mL/min, Total liquid feed including catalyst solution was 0.0167 mL/min, S/C was 1321, and reagent gas was
50:50 CO:H2.
ACS Paragon Plus Environment

Page 25 of 40
Organic Process Research & Development
1
2
3
4
Reactor modeling and quantifying rate constants.
5
6
7
8
9
Fractional conversion of reactant A, as it passes through a tubular reactor, is governed by
axial dispersion, advection, reaction rate, and reaction order, as shown in the following
dimensionless form of the steady state material balance equation for component A34.
ꢀ ꢃꢄꢅꢆ ꢃꢅꢆ
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
ꢌꢍꢎꢏ
ꢐꢌ
−
+ ꢈꢉꢊꢆꢋ 1 − ꢅꢆ = 0
ꢁꢂ ꢃꢇꢄ
ꢃꢇ
XA = Fractional conversion of reagent A
CA0= Initial concentration
z = Fraction of total reactor length (l/L)
k = Reaction rate constant
n = Reaction order
The first term represents flux in and out of the differential volume by dispersion, the second term
represents flux in and out by bulk flow, and the third term represents reaction. Steady state
means that the differential change in concentration with time equals zero, which is why the term
on the right side of the equation is 0.
The companion paper explains that the reaction is observed to be approximately first order in 2ꢀ
vinylꢀ6ꢀmethoxynapthalene.1 Therefore it was modeled as a first order reaction. The above
equation has been solved analytically for a first order reaction34.
ꢎꢔꢕ
ꢗ
ꢊꢆ
4ꢑꢒꢓ
ꢄ ꢖ
= 1 − ꢅꢆ =
ꢘꢔꢕ
ꢘꢔꢕ
ꢄ ꢖ
ꢊꢆꢋ
ꢓ
ꢗ
ꢓꢍ
ꢗ
ꢐꢄ
1 + ꢑ ꢒ
ꢏ
ꢐꢄ
− 1 − ꢑ ꢒ
ꢄ ꢖ
ꢏ
ꢀ
ꢙ
ꢑ = 1 + 4ꢈꢉꢚ ꢛ
ꢜ
ꢁꢂ
ꢈꢉ = reaction Damkohler Number = reaction rate times mean hydraulic residence time. These
equations for conversion versus distance along the PFDR as a function of Dispersion Number
and Damkohler number are used to solve for conversion versus time in the reactor for a given
axial dispersion number.
ACS Paragon Plus Environment

Organic Process Research & Development
Page 26 of 40
1
2
3
4
5
6
In addition, the reactor was modeled as a large number of equal volume CSTRs in series. The
reactor design equation for first order irreversible reaction in equalꢀsized CSTRs in series is35
7
8
9
ꢊꢆꢌ
ꢊꢆꢋ
1
=
ꢏ
ꢐꢌ
1 + ꢈꢉ
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
CAn = concentration of reagent A in the nth CSTR in series.
CA0 = initial concentration of reagent A
n = number of CSTRs in series.
Both of these reactor models were fit to the experimental data shown in Table 7. Samples were
taken from intermediate points along the length of the 23.8 mL reactor at steady state for each
experimental condition. Conversion versus distance through reactor was translated to conversion
versus τ. The results, along with the fitted first order rate constant, are summarized in Table 8.
Table 8. Conversion versus τ for 5 steady state conditions for continuous 2ꢀvinylꢀ6ꢀ
methoxynaphthalene reaction in the 23.8 mL reactor, and fitted first order rate constant.
first order
Reaction
pre-
rate
constant k
(1/h)a
experiment activation
P
time
#
1
loop?
no
(psig)
τ (h)
pipe (h) conversion
3
7
1.2
2.9
5.4
62.6
89.9
97.3
400
8.3
13
0.86
3
7
1.2
2.9
63.7
92.2
yes,
1.28h at
60 C
2
3
4
5
400
800
150
400
8.2
8.2
7.5
8.5
13
3
5.3
1.2
2.9
5.3
1.1
2.6
4.9
1.3
3.0
5.5
97.1
43.2
81.2
93.8
51.9
84
0.97
0.55
0.72
1.29
7
no
no
no
13
3
7
13
3
95.3
78.4
90.6
96.8
7
13
a from fit of CSTRs in series model, with 36 equal sized CSTRs.
Temperature was 70 oC for all reactions.
ACS Paragon Plus Environment

Page 27 of 40
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Figure 11 shows conversion versus time and fitted reaction models for the first steady
state reaction condition. The 36 CSTRs in series model was fit to the experimental data to
minimize the sum of squared residuals between predicted and measured conversion at each
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
intermediate τ. First order rate constants, k, quantified by this method are listed in the last
column in Table 8. Although the CSTRs in series model was used to calculate k, the other two
model fits are shown in Figure 11 using the same k values to show the comparison (ideal PFR
and PFDR with D/uL = 0.015). As seen in Figure 11, calculated conversion versus time for each
of the 3 models is almost identical. This means that axial dispersion is low enough so that it has
negligible impact on conversion versus time. The ideal PFR model could have been used
without significant error for this reactor. Similar reactor model fits to the experimental data for
the four other steady state reaction conditions are shown in the Supporting Information.
100
90
80
70
batch or ideal PFR
60
36 CSTRs in series
50
40
30
20
10
0
PFDR with D/uL=0.015
experimental data
0
1
2
3
4
5
6
time (hours)
Figure 11. Conversion versus τ data and reactor model equation fits for continuous AHF reaction
of 2ꢀvinylꢀ6ꢀmethoxynaphthalene, at 70 oC, 400 psig, and 8.3 h total τ. Fitted first order k =
0.86/h.
ACS Paragon Plus Environment

Organic Process Research & Development
Page 28 of 40
1
2
First order rate constants range from 0.55 to 1.29 hꢀ1 (0.00015 to 0.00036 sꢀ1). This is about 3
orders of magnitude less than kla in the reactor, which is one reason why we claim that kla = 0.1 sꢀ
1 is sufficiently high for this type of continuous reaction.
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
Design and implementation of custom onꢀLine HPLC
An example of onꢀline HPLC data for one of the experimental continuous reactions is
shown in Figure 12. This data is valuable to show time to line out to steady state for each of the
reaction experiments. It also shows steady state conversion, ee, and B/L ratio for each reaction.
13
99
12.5
97
95
93
91
89
87
85
12
11.5
11
Conv vs Time
EE vs Time
10.5
10
B/L
9.5
9
B/L vs Time
8.5
8
0
5
10
15
20
Time (h)
Figure 12. Onꢀline HPLC data for conversion, ee, and branched/linear ratio versus time for
continuous reaction of 2ꢀvinylꢀ6ꢀmethoxynaphthalene. This shows the transition from 150 psig
to 400 psig (condition 4 to condition 5).
Onꢀline HPLC was achieved by converting an offꢀline Agilent 1100 HPLC into an onꢀline HPLC
by implementing a custom built, automated sampling/diluting/parking system. This is described
in more detail in the experimental section and Supporting Information.
ACS Paragon Plus Environment

Page 29 of 40
Organic Process Research & Development
1
2
3
4
Heat transfer rates
5
6
7
8
The heat transfer properties of the process fluids were modeled using Aspen Properties
software. These parameters were used in a custom MATLAB code to calculate the steady state
temperature profile along the length of the 0.56 mm i.d. tube. In order to account for the
segmented flow within the reactor, it was assumed that the larger volumetric gas phase
dominated the heat transfer within the 0.56 mm i.d. tube. This allowed for calculation of the
steady state temperature profile along the tube of a gas only system. From this steady state
temperature profile, the internal wall temperature was calculated. The internal wall temperature
was used with the heat transfer properties of toluene to determine the temperature profile of the
toluene along the reactor. Under these conditions, the toluene was within 1ºC of the bath
temperature (70 ºC) before 1 centimeter into the reactor. At process conditions, the gas phase
features a higher heat transfer coefficient than the liquid phase. Therefore, a conservative
heating time for this system would be bounded by a toluene only system. Therefore, a similar,
toluene only system was modeled. This model calculation showed that the process fluids
reached within 1ºC of the bath temperature (70 ºC) around 10 cm into the reactor, which is only
0.1% of total reactor length.
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
Scale up reactor design
At manufacturing scale, the vertical pipesꢀinꢀseries reactor operates about 98% liquid
filled when liquid τ
is on the order of 4 to 12 hours.36 This is because more than 99% of the
volume is in the vertical upꢀflow pipes (Table 9). The 360L reactor in the table is installed at an
Eli Lilly GMP manufacturing site. The pipes are large enough diameter (more than 2 inches i.d.)
so that reagent gas bubbles up through the liquid, and thus
τ for the reagent gas is much less than
τ
for liquid. Downꢀflow through the smaller diameter vertical downꢀjumper tubing between each
ACS Paragon Plus Environment