Teflon AF-2400 mediated gas-liquid contact in continuous flow methoxycarbonylations and in-line FTIR measurement of CO concentration

Article abstract of DOI:10.1039/c1ob06017a

We report on the development of a continuous flow process for the palladium catalysed methoxycarbonylation of aryl, heteroaromatic and vinyl iodides and an aryl bromide using a Teflon AF-2400 based Tube-in-Tube reactor to mediate the selective permeation of carbon monoxide into solution at elevated pressures. The low volume of pressurised gas within the reactor (5.6 mL) offers the potential for an enhanced safety profile compared to batch processes. We also present preliminary results for the use of in situ FTIR to measure solution concentrations of carbon monoxide and demonstrate the use of a second reactor to effect the removal of carbon monoxide from the flow stream.

Full text of DOI:10.1039/c1ob06017a

View Article Online / Journal Homepage / Table of Contents for this issue
Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Organic &
Dynamic Article Links
Biomolecular
Chemistry
Cite this: Org. Biomol. Chem., 2011, 9, 6903
www.rsc.org/obc
COMMUNICATION
Teflon AF-2400 mediated gas–liquid contact in continuous flow
methoxycarbonylations and in-line FTIR measurement of CO concentration†
Peter Koos, Ulrike Gross, Anastasios Polyzos, Matthew O’Brien, Ian Baxendale and Steven V. Ley*
Received 25th June 2011, Accepted 27th July 2011
DOI: 10.1039/c1ob06017a
We report on the development of a continuous flow process
for the palladium catalysed methoxycarbonylation of aryl,
heteroaromatic and vinyl iodides and an aryl bromide using
a Teflon AF-2400 based Tube-in-Tube reactor to mediate
the selective permeation of carbon monoxide into solution
at elevated pressures. The low volume of pressurised gas
within the reactor (5.6 mL) offers the potential for an
enhanced safety profile compared to batch processes. We
also present preliminary results for the use of in situ FTIR
to measure solution concentrations of carbon monoxide and
demonstrate the use of a second reactor to effect the removal
of carbon monoxide from the flow stream.
synthesis which offers several advantages to batch chemistry. This
is particularly true for reactions involving hazardous/unstable
intermediates and reagents and/or extreme reaction conditions
(e.g. high temperature or pressure).3 This being the case, flow
chemical methods should overcome some of the obstacles to the
broader application of CO-mediated carbonylation as a useful
synthetic transformation. The majority of recent developments in
gas–liquid phase contact under flow conditions involve ‘segmented
flow’,4 in which the gas and the liquid are in two separate phases.
The reaction rate, which is partly determined by the concentration
of dissolved gas, is therefore dependent on the morphology of
the biphasic system (segment size, interfacial area etc.). Control
and prediction of this morphology (which depends on factors
including solvent/gas pressure and flow rate, geometry of the
mixing junction etc.) is by no means trivial and many of the
existing methods aimed at increasing the mixing of gas and
liquid phases, although involving some interesting engineering
developments,5 are often technically quite complex and require
fabrication techniques which are beyond the means of most
synthesis laboratories. Seeking a simpler, more accessible and more
controllable means of achieving gas–liquid contact, we conceived
the use of gas-permeable polymer membranes in flow synthesis.
We have previously reported on the use of the amorphous
fluoropolymer Teflon AF-24006 as a particularly effective method
of delivering ozone (in oxygen),7 carbon dioxide8 or hydrogen9
The carbonyl group is integral to many of the key bond forming
reactions of organic synthesis.1 Its success is dependent upon the
availability of methods to access carbonyl substrates, of which the
transition metal-mediated insertion of carbon monoxide (CO) is
one of the most efficient (Scheme 1).2
Scheme 1 General transition metal-catalysed carbonyl insertion reaction.
This atom-economical process combines the ready availability
of CO, the ability to drive reactions to completion through
pressurisation, and ease of workup (as the excess reagent is a gas),
making this transformation particularly attractive to synthesis
chemists. However, there still exists a general reluctance to use
CO gas in research laboratories, and there are good reasons for
this. The concerns relate to the pressurised containment, toxicity
and flammable nature of this reactive gas input. As such the
safe and scalable use of CO and other gases within laboratory
synthesis programmes represents an unmet need. Flow chem-
istry (in which the substrates/reagents are continuously pumped
through a relatively low volume reaction zone so that only a very
small fraction of the total is exposed to the reaction conditions
at any one time) has emerged as a new paradigm of chemical
Whiffen Laboratory, University Chemical Laboratory, University of Cam-
bridge, Lensfield Road, Cambridge, UK CB2 1EW. E-mail: svl1000@
cam.ac.uk
† Electronic supplementary information (ESI) available: Further experi-
mental detail and spectral data. See DOI: 10.1039/c1ob06017a
Fig. 1 The concentric tubular (Tube-in-Tube) gas-liquid reactor.
Org. Biomol. Chem., 2011, 9, 6903–6908 | 6903
This journal is
The Royal Society of Chemistry 2011
©

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
View Article Online
to a continuous (liquid) flow stream in a controlled and efficient
manner.
Here, we report our initial findings on the use of a gas–liquid
flow reactor employing Teflon AF-2400 in a concentric tubular
design for palladium catalysed methoxycarbonylation reactions
and we also discuss preliminary results from the use of an in-
line FTIR flow-cell to measure concentrations of CO in the
solutions so generated. In this reactor (Fig. 1),10 the liquid stream
is directed through the inner Teflon AF-2400 membrane tubing
and CO gas fills the outer tubing (standard PTFE). CAUTION!!:
This reactor is a prototype and not fully tested for safety! All
reactions involving pressurised gas should be carried out in an
efficient fumehood with adequate safety systems in place capable of
handling an unexpected rupture and/or ignition. Carbon monoxide
is highly toxic. This is made worse by the odourless and tasteless
nature of the gas. Excess carbon monoxide released was directed
into the exhaust of the fumehood. Carbon monoxide monitoring
equipment was used inside and outside the fumehood at all times!
The Teflon AF-2400 (3.8 m, 1.91 mL) and the outer PTFE tubing
are separated by stainless steel T-pieces, allowing the liquid and
gas stream to be independently introduced into the reactor with
accurate control of flow rate and pressure for each input. This
maximises membrane surface area, ensuring efficient permeation
of CO and, as the volume of gas occupying the reactor at any
given instance does not exceed 4.56 mL (1.2 mL of gas per meter
of tubing, which can be tailored by simply modifying the length of
tubing), the hazards, risks and precautionary measures associated
with the use of traditional high-pressure batch equipment are
largely circumvented. The study was initiated by evaluating
the methoxycarbonylation of m-iodotoluene (1a) using the flow
arrangement shown in Fig. 2. Vapourtec R2/R4 units were used in
conjunction with a Knauer 100 pump but any other commercially
available flow synthesis platform could also be used. Substrate (1a)
and triethylamine were premixed and loaded into a PEEK sample
loop, a second loop was filled with the palladium acetate and the
ligand xantphos.
Fig. 2 A schematic of the optimised gas-flow reactor configuration
involving a third solvent stream.
from the product stream. Shown in Table 1 are the conversions
obtained in several experiments where conditions were varied
according to flow rates, CO pressure, and heating coil length and
temperature. The initial conditions (entry 1) gave only a low 18%
conversion. Increasing the ratio of CO enriched solvent stream led
to a dramatic increase to 73% conversion (entry 2). Increasing the
total volume (and therefore residence time) of the heating coils
increased the conversion to 86%. It was found that decreasing the
amount of the relatively expensive xantphos ligand (to just over
stoichiometric with respect to the Pd) gave the same conversion
(entry 4). Increasing the CO pressure to 10 bar led to a slightly
increased conversion of 90% (entry 6). Lowering the temperature
gave a corresponding reduction in conversion (entry 5).
Using the conditions from Table 1, entry 4, we then evaluated
the scope of the process. As the fact that the swaged joints between
the PTFE outer tubing and stainless steel connectors in the Tube-
in-Tube reactor/injector are non-standard (normally joints are
made between like materials) we decided, due to the high toxicity
and odourless nature of CO, to carry out preparative reactions
initially at 7 bar. For some later reactions, when we were now
confident that the interconnects were robust, we used pressures
of up to 15 bar where required but with the added precaution
that the main gas cylinder was used only periodically to charge
an intermediate ‘reservoir’ section of high-pressure hose tubing,
the volume of which was sufficient to maintain constant pressure
for several hours. These manifolds have now been used with other
gases at pressures of up to 35 bar without incident. The results of
the methoxycarbonylation reactions of a series of aryl/alkenyl
iodides and an aryl bromide are shown in Table 2. Electron-
deficient aryl iodides 3a, 4a, 7a and 9a afforded the corresponding
The reagent streams were pumped at a rate of 0.1 mL min-1
and united at a T-piece, then combined with a third CO-enriched
stream from the gas–liquid reactor. This is an alternative configu-
ration to those described in our earlier work with these reactors,
as the substrate/catalyst does not pass through the Tube-in-Tube
device which now acts purely as a means to inject gas into the flow
stream. A 100-psi (6.9 bar) back-pressure regulator was installed
after the gas–liquid reactor to prevent out-gassing of the dissolved
CO. The presence of this back-pressure regulator is vital to ensure
that the gas–liquid stream remains a homogeneous solution, thus
avoiding segmented flow. The combined flow streams were then
directed into the heating coils and a QP-TU (polymer supported
thiourea) cartridge was incorporated to scavenge the palladium
Table 1 Results for investigation of reaction conditions for the conversion of 1a to 1b
Entry
Pd(II)/L mol (%)
CO (bar)
Pump a = b (mL min-1)
Pump c (mL min-1)
Coils d total (mL)
Coils d (◦C)
Conv. %
1
2
3
4
5
6
2.5/5
2.5/5
2.5/5
2.5/3.0
2.5/3.0
2.5/3.0
7
7
7
7
7
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.6
0.6
0.6
0.6
0.6
20
20
30
30
30
30
100
100
100
100
70
18
73
86
86
80
90
10
100
6904 | Org. Biomol. Chem., 2011, 9, 6903–6908
This journal is
The Royal Society of Chemistry 2011
©

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
View Article Online
Table 2 Results of continuous flow methoxycarbonylations
Entry
1
Substrate (a)
Product (b)
Solventa
A
Yldb/(Conv.c)
70/(85)
2
3
A
A
63/(78)
81/(95)d
4
5
A
A
88/(95)
(11)e
6
7
A
B
38/(95)f
83/(100)g,h
A
B
93/(95)
8
9
69/(77)h,i,j
A
A
72/(90)
71/(85)
10
11
A
B
43/(60)e
65/(73)h,j
12
13
B
62/(88)h,j
62/(100)j
B
14
15
B
B
81/(100)i,k
72/(94)h,i,j,l
77/(90)h,i,j
a A = PhMe : MeOH : DMF 45 : 45 : 10, B = dioxane : MeOH 1 : 1. b isolated yield after column chromatography on silica gel. c conversion based on signals
1
from product and starting material in H NMR spectrum of crude product material (no reference compound was added). d Substrate and base were
introduced in DMF. e Rapid Pd(0) precipitation observed. f Reaction performed at 50 ◦C, side products observed. g Reaction performed at 25 ◦C, 15 bar
CO, no xantphos, 5 mol% Pd(OAc)2. h No thiourea cartridge used. i 5 mol% Pd(OAc)2/6 mol% xantphos used. j 15 bar CO used and 30 mol% hydrazine
(from 1.0 M THF solution) added. k reaction performed at 25 ◦C, 15 bar CO. l reaction performed at 120 ◦C.
This journal is
The Royal Society of Chemistry 2011
Org. Biomol. Chem., 2011, 9, 6903–6908 | 6905
©

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
View Article Online
methyl esters with high conversions and high isolated yields, while
yields and conversions for electron-rich aromatic iodides 1a and
2a were more moderate. The ortho-keto substrate 5a resulted in
low conversion, perhaps due to catalyst deactivation, evidenced
by significant visible Pd black formation. With the exception of
the reactive vinyl iodide 6a, all of the methoxycarbonylations
proceeded cleanly without the formation of unwanted by-products
and within short reaction times (37 min in heating coils).
While the thiophene 9a reacted cleanly under these conditions,
affording the product in good conversion and yield, the pyrazole
11a gave only modest conversion and low yield.
Fig. 3 Flow configuration for in-line real-time FTIR CO monitoring.
After a brief screen of solvents and additives, we found that using
a 1 : 1 mixture of dioxane : methanol with the addition of 30 mol%
hydrazine (1.0 M THF solution)11 at an increased CO pressure
of 15 bar led to an improvement in the yield of 11b (65%). We
used these conditions for the reaction of two further nitrogenous
heterocycles, 12a and 13a, which both afforded products in 62%
yield. We found that with 12a the use of the polymer supported
thiourea cartridge led to poor recovery, perhaps due to the resin
binding to the product, and for this reaction this was removed.
Interestingly, we found that the sensitive vinyl silane product
6a was unstable to both the thiourea resin and the presence of
the xantphos ligand during the reaction and, when these were
removed and the reaction carried out at 15 bar CO pressure at
room temperature, a much improved 83% yield of the product
was isolated. The E-vinyl iodide 14a also reacted more smoothly
at room temperature with quantitative conversion to the product
14b which was recovered in high yield.
We have also begun to investigate the alkoxycarbonylation of the
less active but potentially more useful aryl bromides. Focussing on
4-nitro-bromobenzene, we found that using the dioxane–MeOH
system with 30 mol% hydrazine and twice the catalytic loading
gave a 77% isolated yield of product (90% conversion). A higher
conversion could be obtained with a higher reaction temperature
(120 ◦C) but only at the expense of yield, presumably due to side
reactions (entry 15). The 2-bromo-iodobenzene 8a reacted cleanly
and selectively at the iodo position only. Further investigation with
aryl and heteroaryl bromides in our laboratory is ongoing.
It is noteworthy that the corresponding batch reactions of
these and similar substrates can often require significantly longer
reaction times (12–24 h vs. 37 min) in addition to CO pressure.12
This is perhaps attributable to the very high effective surface-
area : volume ratio provided by the Teflon AF-2400 membrane
which leads to rapid dissolution of CO into the solvent.13 In order
to gain quantitative insight into the uptake of CO into the solvent
stream as a function of pressure and flow rate, we sought a means
of measuring the solution concentration of the gas in flow and
conjectured that the distinctive IR absorption of CO might be
used for this purpose. To assay this possibility an in-line real-time
FTIR experiment was performed,14 monitoring the intensity of
the dissolved CO IR stretching frequency15 at 2133 cm-1 using
a ReactIRTM15 unit fitted with a silicon (SiComp) flow probe
(Fig. 3).16 In these experiments, CO was dissolved in toluene and
this flow stream was then directed to the input of the flow probe.
The inclusion of a back-pressure regulator before the probe was
necessary to ensure that the liquid stream pressure did not exceed
8 bar (the current maximum pressure tolerance of this cell is 10
bar). Initially, the influence of CO pressure on the intensity of
the peak at 2133 cm-1 was examined using a constant flow rate
Fig. 4 (a) Intensity of the CO stretching frequency at 2133 cm-1 against
pressure at constant flow rate (0.6 mL min-1; 2, 5, 7, 9 bar of CO; 0 bar is
without CO); (b) Real time, in-line monitoring of dissolved CO in toluene
at various flow rates with constant pressure of CO (7 bar of CO; 1.0, 0.6,
0.3 mL min-1); (c) average peak height from A against CO pressure; (d) A
plot of volume of CO (mL) degassed per volume of toluene (mL) using the
gas burette measurements.
(0.6 mL min-1) (Fig. 4a). The peak intensity of this spectral region
clearly shows a dependence on applied CO pressure, which results
in increased CO dissolution.
As shown in Fig. 4c, a plot of the relative peak intensity
at 2133 cm-1 against CO pressure increases linearly prior to
approaching approximately 9 bar of CO. Separately, a gas burette
was used to measure the volume of CO, which evolved from the
solvent on depressurisation of the gas–liquid stream to ambient
pressure.
The volume of CO collected per unit volume of toluene at
0.6 mL min-1 as a function of applied CO pressure is shown
in Fig. 4d. A similar dependence on pressure was observed
for the gas burette and IR measurements (Fig. 4c and 4d),
demonstrating that the burette measurements can be used to
quantify the intensities in the IR experiments and that the IR
intensity is a true reflection of the CO concentration. With a view
to determining the concentration of dissolved CO over a range
of flow rates, additional in-line IR monitoring experiments were
undertaken. The preliminary results are shown in Fig. 4b, in which
the dissolved CO IR stretching frequency at 2133 cm-1 (in toluene)
6906 | Org. Biomol. Chem., 2011, 9, 6903–6908
This journal is
The Royal Society of Chemistry 2011
©