Gas-liquid segmented flow microfluidics for screening Pd-catalyzed carbonylation reactions

Article abstract of DOI:10.1002/chem.201104059

Go with the (segmented) flow: A gas-liquid microfluidic reactor system has been developed to study Pd-catalyzed carbonylation reactions over a range of flow regimes and reaction conditions (see picture). The segmented gas-liquid flow regime, in comparison to annular flow, enables reactions to be studied over longer reaction times and without the buildup of unwanted Pd particles. Copyright

Full text of DOI:10.1002/chem.201104059

DOI: 10.1002/chem.201104059
Gas–Liquid Segmented Flow Microfluidics for Screening Pd-Catalyzed
Carbonylation Reactions
Xiuqing Gong,[a] Philip W. Miller,[a] Antony D. Gee,[b] Nicholas J. Long,[a]
Andrew J. de Mello,*[c] and Ramon Vilar*[a]
Chemical reactions that occur within nanoliter to microli-
ter volumes demonstrate many advantages over traditional
batch synthetic methods.[1–4] On the microscale, the inherent
high surface area-to-volume ratios facilitate highly efficient
heat transfer processes and rapid and controllable mixing re-
gimes enhancing mass transport. Both factors act to signifi-
cantly enhance product quality, reaction yields and analyti-
cal throughput. Reactions performed under continuous flow
within microfluidic channels also offer the advantages of im-
proved safety under extreme conditions, enhanced processa-
bility, facile in-line detection, and reagent economy.[5–12]
Gas–liquid phase reactions are particularly amenable to
processing within small volumes under continuous flow con-
ditions. The increased contact areas generated within such
devices are ideal for enhancing mass transport of gaseous re-
agents into the liquid phase, and the safety benefits associat-
ed with pressuring small gas volumes is clear. A diversity of
gas–liquid phase reactions are key components of many in-
dustrial processes. For example, hydrogenations, hydrofor-
mylations, carbonylations, chlorinations, and oxidations all
require a gaseous reagent to enter a liquid phase before a
reaction can occur. Typically, these gas–liquid reactions are
performed in pressurized containers under batch conditions,
however in many cases the reactions are amenable to flow
processing.[13–17]
regular and alternating formation of segments of gas and
liquid. The flow regime obtained is dependent on volumetric
flow rates, channel geometries, and the physical properties
of the liquid phase. Flow regime maps can be generated by
plotting the superficial gas velocity against liquid velocity.[18]
Although an annular flow regime will generate high surface
area contact between the gaseous and liquid phases within a
microchannel the high gas flow rates typically result in un-
desirably short residence times (few minutes) for the liquid
reagents.
Our group and others have previously used various gas–
liquid flow regimes to study the palladium-catalyzed carbon-
ylations of aryl halides.[19–22] This is a versatile and widely
used approach for synthesizing organic molecules containing
a carbonyl functional group.[23–25] In previous studies, using
annular flow conditions, we found that reaction yields were
improved compared to macroscale reactions, however, the
short reaction times imposed by the flow regime precluded
studies over longer timescales. Additionally, undesirable
traces of metallic palladium were found to accumulate on
channel surfaces, which occasionally resulted in channel
blockage. To study carbonylation reactions over longer time
periods and to prevent Pd aggregation we have developed a
microfluidic system for rapidly generating and incubating
segmented gas–liquid flows and used it to screen Pd-cata-
lyzed carbonylation reactions. Importantly, this has allowed
Gas–liquid flow within channels can be broadly catego-
rized as being either annular or segmented. Annular flow is
characterized by a rapid gas flow through the center of a
channel resulting in a thin film of liquid coating the internal
surface of the channel, whilst segmented flow describes the
us to study the scope of the palladium(I) dimer [Pd
(PtBu3)2] as pre-catalyst for carbonylative coupling reactions
using a range of different substrates.
2ACHTUNGTRENNUNG(m-I)2-
AHCTUNGTRENNUNG
The microfluidic reaction system contains two parts, a
glass chip for segment generation and a fluorinated ethylene
propylene (FEP) tube for incubation (Figure 1a and Fig-
ure S1 in the Supporting Information). To study flow gener-
ation within our device, the CO gas flow rate was fixed at
5 sccm and the liquid flow (toluene) rate was increased
steadily. By gradually raising the liquid flow rate the transi-
tion between annular and segmented flow could be visual-
ized by injecting markers into the flow. Flow regimes were
plotted as gas flow rate versus liquid flow rate (Figure S2 in
the Supporting Information).
The transition between annular and segmented flow is
known to be dependent on both the channel size and the
fluid properties, with smaller channel sizes favoring the for-
mation of segmented flows.[26–28] Accordingly, the glass seg-
ment generation chip has a small internal diameter (ID=
[a] Dr. X. Gong, Dr. P. W. Miller, Prof. N. J. Long, Prof. R. Vilar
Department of Chemistry, Imperial College London
Exhibition Road, South Kensington, London SW7 2AZ (UK)
E-mail: r.vilar@imperial.ac.uk
[b] Prof. A. D. Gee
Division of Imaging Sciences and Biomedical Engineering
Kingꢀs College, St Thomasꢀ Hospital, London, SE1 7EH (UK)
[c] Prof. A. J. d. Mello
Institute for Chemical and Bioengineering
Department of Chemistry and Applied Biosciences
ETH Zurich (Switzerland)
E-mail: andrew.demello@chem.ethz.ch
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201104059.
2768
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 2768 – 2772

COMMUNICATION
Figure 1. a) Experimental setup for gas–liquid flow reaction system. The upper right diagram indicates the connection between glass chip (d1) and FEP
tubing (d2) through an O-ring. The model Pd-catalyzed aminocarbonylation of iodobenzene is shown in the lower portion. b) Annular flow regime. c)
Segmented flow regime. Three segments flow in the sequence. From left to right: first the substrate solution (yellow), second the solvent segment
(white), and third the dye segment (blue). In the latter, the distance between blue-dye segments is 800 mm. d) Annular flow combined with segmented
flow. e) The liquid segment is extended and then recovered in the larger bore FEP tubing.
100ꢂ220 mm) to allow the creation of gas–liquid segments at
high frequency. After the gas–liquid segments are formed,
they flow into the FEP tubing which has a larger internal di-
ameter (ID=800 mm). The role of this tubing was twofold.
First it acts as a residence chamber to incubate the reaction
for a given period of time and second it reduces the liquid
segment thickness via vertical expansion into a larger
volume (Figure 1a). By thinning the liquid segments the
contact area between the CO gas and substrate is improved.
The length of the gas–liquid segment can be adjusted by
carefully controlling relative flow rates. A two-way valve
mounted at the outlet of the FEP tubing was used to shut-
off the flow and store the gas liquid segments for incuba-
tion.
The palladium-catalyzed aminocarbonylation of iodoben-
zene and benzylamine (Figure 1a) was selected as a model
reaction to test annular and segmented flow reactions. Previ-
ously we have performed a basic investigation of this reac-
tion under annular flow.[20,21] In a typical reaction the aryl
halide/amine solution was pre-mixed with the catalyst
reactions performed at higher temperatures and pressures
gave enhanced yields (see Figure S3a and b in the Support-
ing Information). Temperature variations had a more pro-
nounced effect on the reaction, with yields increasing from
5% at 808C to 85% at 1508C (at 2 bar of CO). An increase
in the CO pressure had a lower impact on yields. For exam-
ple, reaction yields at 1008C increased from 7% at 1 bar to
18% at 5 bar (Figure S3a in the Supporting Information).
To extend residence times under annular flow conditions the
glass chip was connected to 7 m of FEP tubing (with an ID
of 250 mm), to provide for a total channel length of 12 m.
This resulted in residence time of 4 min. By using these new
conditions, conversions were improved at 1008C and 2 bar
CO pressure (Figure S3b in the Supporting Information). To
further extend residence times under this flow regime would
require longer tubing. This is impractical due to the limits
set by increasing resistance in the system.[29] For example, to
generate a 10 min residence time, approximately 50 m of
tubing would be required (at 4 bar gas pressure) to sustain
annular flow. Significantly, utilization of segmented flow re-
gimes enables superior control over reaction times. Fig-
ure 2a–c show data describing the effect of pressure, reac-
tion time, and segment size on the yield of N-benzylbenza-
mide. Experiments profiling the effect of pressure were per-
formed at 1008C over a 10 min period and showed a notable
improvement in product yield on increasing the pressure
from 1 bar (39%) to 5 bar (75%). For reactions incubated
over longer time periods (at 2 bar and 1008C) completion of
the reaction was achieved within 40 min. A key advantage
of using a microfluidic system for pressurized reactions of
this kind is the benefit of only having to pressurize very
small quantities of gas. To perform equivalent reactions on a
(namely [PdACHTUNGTRENNUNG(PPh3)4]) in toluene and infused, via the injector
port, into a CO gas stream on the glass chip where segment-
ed or annular flow could be established (for full experimen-
tal details see the Supporting Information). To assess the
effect of flow regime on the reaction yield, annular flow was
initially investigated over an extremely wide range of varia-
bles. Annular flow within the system was achieved at a CO
pressure of 2 bar, gas flow rates of 9 sccm, and liquid flow
rates of 15 mLminÀ1 (Figure 1b). This resulted in a total resi-
dence time of 2 min. The effects of temperature over the
range 808C–1508C and pressure 1–5 bar on the carbonyla-
tion reaction were also investigated. As might be expected,
Chem. Eur. J. 2012, 18, 2768 – 2772
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2769

R. Vilar, A. J. deMello et al.
placed under a CO atmosphere and heated, the reaction
yields are disappointingly low. However, when the aryl
halide and amine substrate are added to [Pd2ACHTNUTRGNEN(UG m-I)2ACHTUNGTRENNUNG(PtBu3)2]
under a CO atmosphere then reaction yields are extremely
high. It is suspected that this pre-catalyst is deactivated in
the presence of excess amine. To achieve the correct se-
quence of reagent addition the microfluidic system was fur-
ther developed to include an extra injector valve and an ad-
ditional T-junction chip. A schematic of this is shown in Fig-
ure 1d and e. The reaction sequence involves the generation
of gas–liquid segmented flow between CO and the aryl
halide/amine solution at a T-junction prior to the first inlet
of the glass chip. This is then merged with a solution of [Pd2-
ACHUNTGRENUN(G m-I)2ACHTUGNTRENN(UNG PtBu3)2] in toluene that enters via the second chip
inlet. Mixing of the reagents is fast and occurs along the
glass chip. Once the solutions reach the FEP tubing at the
device outlet, the reagents are completely mixed and consis-
tent sized segments are generated (Figure 1e). Initially the
carbonylative coupling reaction was studied under an annu-
lar flow regime at temperatures between 808C and 1208C at
2 bar (Figure 2d) over a 4 min period.
Although yields of N-benzylbenzamide were >60% at
1208C considerable formation of metallic Pd occurred along
the length of the channel which, over time, resulted in flow
reduction and occasional blockage of the channel (Fig-
ure 3a). Recently sonication of microfluidic Pd-catalyzed re-
Figure 2. a) Percentage yield of N-benzylbenzamide as a function of pres-
sure within a segmented flow (12 m flow length, 10 min reaction time,
and a segment length of 600 mm). b) Percentage yield of N-benzylbenza-
mide as a function of time (12 m flow length, 2 bar reaction pressure, and
a segment length of 600 mm). c) Percentage yield of N-benzylbenzamide
as a function of segment length (12 m flow length, 10 min reaction time,
and a reaction pressure of 2 bar). Scale bar 1.5 mm. d) Percentage yield
of N-benzylbenzamide as a function of temperature in annular flow using
a Pd-dimer catalyst.
macroscale at 5 bar would require specialized high-pressure
equipment and safety procedures.
Besides the effects of CO pressure and temperature, the
length of the liquid segment within the channel was also
found to have a significant impact on product yields. Three
segment lengths were investigated (0.75, 0.5, and 0.25 mm)
using a 10 min reaction time at 2 bar and at 1008C (Fig-
ure 2d). The largest segment (0.75 mm) was found to give a
modest 36% yield over this time period, whilst the smallest
segment (0.25 mm) gave a significantly improved yield of
89%. This effect is directly attributable to the larger surface
area of the smaller segments. The smaller 0.25 mm segments
have surface-to-volume areas three times larger than the
0.75 mm segments, which results in improved mass transport
of CO gas in the liquid. Interestingly, the much smaller
liquid segments may also be thought of as acting as mem-
branes. According to Fickꢀs law, the net diffusion rate of a
gas across a fluid membrane is inversely proportional to the
thickness of the membrane.[30] Accordingly, a thin mem-
brane is preferable for the CO transport in solvent and will
accelerate the aminocarbonylation reaction.
The nature of the Pd catalyst is known to have a marked
effect on the rates and yields of carbonylation reactions. Re-
cently we have shown for the first time that Pd dimer pre-
catalysts of the type [Pd2ACHTNUTRGENN(UG m-I)2ACHUTNTGREN(NUNG PtBu3)2] are highly active for
performing carbonylation reactions in short reaction
times.[31] We have now used this pre-catalyst in combination
with our segmented flow microfluidic reactor to screen a
small number of substrates for carbonylation reactions. One
of the key challenges when performing carbonylation reac-
tions with this type of catalyst is ensuring the addition of re-
agents in a specified order. If the aryl halide and amine sub-
Figure 3. a) When performing the reaction under annular flow conditions,
metallic Pd adheres to the interior wall and impedes flow. The magnified
image corresponds to the square area on the lower glass chip. b) Pd parti-
cles are confined in the segments. A one pound sterling coin was inserted
to visualize the tube size. c) A gas–liquid segment before reaction. d) A
gas–liquid segment after reaction. e) Segments merge if the CO is ex-
hausted.
strates are simply mixed together with [Pd2ACTHNUTRGNEN(UG m-I)2CAHTUNGTREN(NUNG PtBu3)2],
2770
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 2768 – 2772

Microfluidic Systems for Screening Pd-Catalyzed Reactions
actions has been reported as a method of avoiding
COMMUNICATION
Table 1. Substrates screened by using the developed microfluidic method.[c,d]
Pd buildup within microchannels.[32] However, when
the flow regime was changed to segmented flow we
discovered that Pd buildup on the channel surface
was no longer a problem; the Pd particles formed
in the liquid segments flowed unhindered through
the channels (Figure 3b). An additional and excit-
ing feature of utilizing segmented flows is that reac-
tions could be monitored by observing the decrease
in CO bubble size as the reaction progresses (Fig-
ure 3c–e, and the Supporting Information). Notably,
segmented flow and incubation of the gas–liquid
segments enable control over residence times with
reactions requiring small amounts of reagents and
catalysts (5 mg and <0.5 mg respectively). Under
annular flow residence times are restricted to only
2–4 min. A range of substrates were screened by
using the described microfluidic method. Table 1
summarizes the results of various carbonylation re-
Aryl
halide
Amine
Product
Pd dimer[a]
[PdACHTUNGTRENNUNG
(PPh3)4][b]
96% (10 min)
99% (1 h)
66% (10 min)
99% (1 h)
60% (1 h)
43% (1 h)
99% (12 h)
99% (12 h)
60% (1 h)
19% (1 h)
99% (12 h)
30% (12 h)
12% (1 h)
no yield (1 h)
78% (12 h)
<5% (12 h)
52% (1 h)
82% (12 h)
no yield (1 h)
26% (12 h)
[a] 2 mol% of [Pd
2ACHTUNGTREN(NUG m-I)2AHCTUNGTRENN(UNG PtBu3)2] was used for these reactions. [b] 2 mol% of [Pd-
AHCTNUTGREGN(NNU PPh3)4] was used for these reactions. [c] Although the palladium dimer contains twice
as much palladium than [PdACHTUNTRGNEUNG(PPh3)4], we decided to use the same mol% of the com-
pound since, as we have shown recently,[31] the second Pd in the dimer is likely to be a
“sacrificial” palladium and is not involved in the catalysis. [d] Yields were determined
by GC.
actions over a range of reaction times using [Pd
2ACHTUNGERTN(NUNG m-
I)2A(PtBu3)2] as catalyst (results with [Pd(PPh3)4] are
G
ACHTUNGTRENNUNG
also shown for comparison). Mild reaction conditions of
2 bar CO and 1008C were used in all cases. The model cou-
pling reaction was found to give almost quantitative yields
reactions,[33–35] their role in carbonylative coupling reactions
has not been explored thus far. The results we present here
show this type of palladium compound to be very promising
as a highly active catalyst for this type of reaction.
when using [Pd2ACHTUNGTRENNUNG(m-I)2ACHTUNGTRNE(NUGN PtBu3)2] within only 10 min. More
challenging reactions however required longer reaction
times to generate sufficiently high yields. For example, when
aniline, a less potent nucleophile, was used, reasonable
yields of the benzanilide product were achieved only after
1 h, and quantitative conversions after 12 h. The activity of
Acknowledgements
EPSRC is thanked for funding this project (EP/G027749/1) and for a
Leadership Fellowship (EP/H005285/1) for RV. Johnson Matthey is
thanked for a loan of palladium.
the dimer was again better than that of [PdACHTNUTRGNE(UNG PPh3)4]. As
might be expected, reactions with aryl bromo substrates also
proved more challenging compared to phenyl iodide. An
ortho-CF3 activating group resulted in good yields after 1 h
(60%), whereas the equivalent reaction with a deactivating
ortho methoxy group only achieved a 12% yield within 1 h.
The intramolecular carbonylation reaction of 2-bromobenzyl
alcohol forming phthalide also proved to be quite challeng-
ing under these conditions and required 12 h reaction to
achieve an 82% yield. For these reactions, the higher activi-
Keywords: carbonylation
palladium · segmented flow
·
catalysis
·
microfluidics
·
[1] T. Wirth, Microreactors in Organic Chemistry and Catalysis, Wiley-
VCH, Weinheim, 2008.
[2] W. Ehrfeld, V. Hessell, H. Lowe, Microreactors: new technology for
modern chemistry, Wiley-VCH, Weinheim, 2000.
[3] R. L. Hartman, K. F. Jensen, Lab Chip 2009, 9, 2495.
[4] A. J. de Mello, Nature 2006, 442, 394.
[5] J. I. Yoshida, A. Nagaki, T. Yamada, Chem. Eur. J. 2008, 14, 7450.
[6] L. Ducry, D. M. Roberge, Angew. Chem. 2005, 117, 8186; Angew.
Chem. Int. Ed. 2005, 44, 7972.
[7] T. Kawaguchi, H. Miyata, K. Ataka, K. Mae, J. Yoshida, Angew.
Chem. 2005, 117, 2465; Angew. Chem. Int. Ed. 2005, 44, 2413.
[8] A. Iles, R. Fortt, A. J. de Mello, Lab Chip 2005, 5, 540.
[9] K. Jꢃhnisch, V. Hessel, H. Lçwe, M. Baerns, Angew. Chem. 2004,
116, 410; Angew. Chem. Int. Ed. 2004, 43, 406.
[10] M. Brivio, W. Verboom, D. N. Reinhoudt, Lab Chip 2006, 6, 329.
[11] P. Watts, S. Haswell, J. Chem. Eng. Technol. 2005, 28, 290.
[12] K. F. Jensen, Chem. Eng. Sci. 2001, 56, 293.
[13] N. de Mas, A. Gunther, T. Kraus, M. A. Schmidt, K. F. Jensen, Ind.
Eng. Chem. Res. 2005, 44, 8997.
[14] R. D. Chambers, M. A. Fox, D. Holling, T. Nakano, T. Okazoe, G.
Sandford, Lab Chip 2005, 5, 191.
ty of the palladium(I) dimer over [PdACHTNURTGNE(UNG PPh3)4] was even
more clear than in the previous cases (see last three entries
in Table 1).
In conclusion, we have demonstrated that although annu-
lar flow regimes are useful for performing carbonylation re-
actions they are inherently limited by short reaction times
and detrimental Pd deposition on channel surfaces. A seg-
mented flow regime on the other hand provides superior
control over both reaction conditions and processing times,
and importantly minimizes the possibility of channel block-
age through Pd particle formation. Moreover, the system is
extremely effective as a method for screening carbonylation
reactions in a safe and reliable way. Indeed we have used
this system to show that the dimer [Pd2ACHTUNGRTENNG(U m-I)2ACHTUNTGREN(NUNG PtBu3)2] is an
excellent pre-catalyst for the aminocarbonylation of a wide
range of substrates. While palladium dimers have been pre-
[15] K. Kawai, T. Ebata, T. J. Kitazume, Fluorine Chem. 2005, 126, 956.
[16] J. Kobayashi, Y. Mori, K. Okamoto, R. Akiyama, M. Ueno, T. Kita-
mori, S. Kobayashi, Science 2004, 304, 1305.
À
viously shown to be excellent pre-catalysts for C N coupling
Chem. Eur. J. 2012, 18, 2768 – 2772
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
2771

R. Vilar, A. J. deMello et al.
[17] R. S. Besser, X. Ouyang, H. Surangalikar, Chem. Eng. Sci. 2003, 58,
19.
[26] K. A. Triplett, S. M. Ghiaasiaan, S. I. Abdel-Khalik, D. L. Sadowski,
Int. J. Multiphase Flow 1999, 25, 377.
[27] T. Cubaud, C. M. Ho, Phys. Fluids 2004, 16, 4575.
[28] T. S. Zhao, Q. C. Bi, Int. J. Multiphase Flow 2001, 27, 765.
[29] J. Judy, D. Maynes, B. W. Webb, Int. J. Heat Mass Transfer 2002, 45,
3477.
[18] A. Gꢄnther, K. F. Jensen, Lab Chip 2006, 6, 1487.
[19] E. R. Murphy, J. R. Martinelli, N. Zaborenko, S. L. Buchwald, K. F.
Jensen, Angew. Chem. 2007, 119, 1764; Angew. Chem. Int. Ed. 2007,
46, 1734.
[30] J. D. Way, R. D. Noble, T. M. Flynn, E. D. Sloan, J. Membr. Sci.
1982, 12, 239.
[31] G. Buscemi, P. W. Miller, S. Kealey, A. D. Gee, N. J. Long, J. Passchi-
er, R. Vilar, Org. Biomol. Chem. 2011, 9, 3499.
[32] R. L. Hartman, J. R. Naber, N. Zaborenko, S. L. Buchwald, K. F.
Jensen, Org. Process Res. Dev. 2010, 14, 1347.
[33] J. P. Stambuli, R. Kuwano, J. F. Hartwig, Angew. Chem. 2002, 114,
4940; Angew. Chem. Int. Ed. 2002, 41, 4746.
[34] U. Christmann, R. Vilar, A. J. P. White, D. J. Williams, Chem.
Commun. 2004, 1294.
[35] U. Christmann, D. A. Pantazis, J. Bene tBuchholz, J. E. McGrady,
M. Maseras, R. Vilar, J. Am. Chem. Soc. 2006, 128, 6376.
[20] P. W. Miller, N. J. Long, A. J. de Mello, R. Vilar, J. Passchier, A.
Gee, Chem. Commun. 2006, 546.
[21] P. W. Miller, N. J. Long, A. J. de Mello, R. Vilar, H. Audrain, D.
Bender, J. Passchier, A. J. Gee, Angew. Chem. 2007, 119, 2933;
Angew. Chem. Int. Ed. 2007, 46, 2875.
[22] M. T. Rahman, T. Fukuyama, N. Kamata, M. Sato, I. Ryu, Chem.
Commun. 2006, 2236.
[23] C. Csajꢅgi, B. Borcsek, K. Niesz, I. Kovacs, Z. Szekelyhidi, Z. Bajko,
L. Urge, F. Darvas, Org. Lett. 2008, 10, 1589.
[24] H. M. Colquhoun, D. J. Thompson, M. V. Twigg, Carbonylation
Direct Synthesis of Carbonyl Compounds, Plenum Press, New York,
1991, p. 37.
[25] A. Brennfꢄhrer, H. Neumann, M. Beller, Angew. Chem. 2009, 121,
4176; Angew. Chem. Int. Ed. 2009, 48, 4114.
Received: November 11, 2011
Published online: February 13, 2012
2772
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 2768 – 2772