9750 Ind. Eng. Chem. Res.

2005, 44, 9750-9769

Gas-Liquid and Gas-Liquid-Solid Microstructured Reactors:

Contacting Principles and Applications
Volker Hessel,*,†,‡ Panagiota Angeli,§ Asterios Gavriilidis,*,§ and Holger Lo1 we†,|
Institut für Mikrotechnik Mainz GmbH, Carl-Zeiss-Str. 18-20, 55129 Mainz, Germany; Department of
Chemistry and Chemical Engineering, Eindhoven University of Technology, 5600 MB Eindhoven, P.O. Box
513, The Netherlands; Department of Chemical Engineering, University College London, Torrington Place,
London WC1E 7JE, UK; and Department of Chemistry, Pharmaceutics and Earth Sciences, Institute of
Organic Chemistry, Johannes-Gutenberg-Universität Mainz, Duesbergweg 10-14, 55128 Mainz

A variety of gas-liquid microchannel reactors have been developed so far, using different
contacting principles. Some devices utilize continuous-phase contacting (i.e., nondispersed
separate phases with large specific interfaces). Among these are microstructured falling film,
overlapping channel, and mesh reactors. Dispersed-phase contacting is obtained when one of
the phases is interdispersed into the other phase. Regular flow patterns are provided by the
Downloaded via CSIR-INDIAN INST CHEMICAL TECHLGY on May 16, 2019 at 07:09:02 (UTC).
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

segmented (Taylor) flow in a single microchannel or numbered-up versions such as the

microbubble column; other flow patterns such as annular flow may be achieved as well. Foam
microreactors utilize a moving rigid 3-D bubble network at high gas content. Miniaturized packed-
bed microreactors follow the paths of classical engineering by enabling trickle-bed operation.
Because of the often highly regular flow pattern, not obtained in conventional gas-liquid
contactors, an understanding of the underlying hydrodynamics and heat and mass transfer is
crucial for optimal performance of all types of gas-liquid microstructural reactors. Several
examples are given, including film-thickness measurements, flow-pattern maps, determination
of mass-transfer coefficients, residence-time distributions, scale-out issues, etc. Numerous
applications demonstrate the improved performance of gas-liquid microreactors. Among these
are fluorinations, chlorinations, hydrogenations, sulfonations, photo-oxidations, etc. Recently,
the scope of reactions has been widened, since there is now the possibility to carry out gas-
liquid-solid processes in the same microreactors as used for noncatalytic reactions because of
the development of catalyst washcoats and other materials deposited onto microchannels. Some
relevant examples are given for illustration.

1. Introduction especially in terms of enhanced mass and heat transfer,

making them suitable for fast exothermic reactions. This
In traditional chemical engineering, there is a large is simply the result of the large surface-to-volume ratio
variety of reactors that can be utilized for gas-liquid that they offer. This feature, though, affects their
(GL) and gas-liquid-solid (GLS) reactions. These hydrodynamics. As such, some of the microstructured
include the well-known mechanically agitated tanks, GL and GLS reactors have macroscale analogues offer-
slurry reactors, bubble, packed, and spray columns, ing similar but improved contacting patterns, while
falling-film, loop, and trickle-bed reactors, and the less others are completely novel and benefit from the unique
widespread static mixer, venturi, and spinning disk advantages of microscale operation.
reactors.1,2 The choice of reactor depends on various
In the following, contacting principles employed to
criteria, such as production output, reaction rate, resi-
current microstructured reactors are presented, along
dence time, volumetric ratio of phases, exothermicity,
with information about the associated hydrodynamics
heating-cooling requirements, ease of scale-up, etc.
and transport properties. Applications of the microengi-
With the advent of microreaction technology a new
neered reactors in demanding GL and GLS reactions
family of reactors using microengineered structures has
are further discussed and demonstrate that for mul-
become available. Most of these reactors are at the
tiphase reactions, which are often limited by mass
research stage but a few are already commercially
transfer, improved overall performance can be obtained.
available. Because of their (in most cases) relatively
small volume, they can be used only for small scale
production, which makes them particularly suited for 2. Contacting Principles for Microscale
local and safe chemical production. In addition, they Gas-Liquid Operation
extend the range of capabilities of conventional reactors There are basically two approaches for bringing two
phases into contact: the first is to keep both phases
* To whom correspondence should be addressed. E-mail: continuous and use the reactor to create an interface
[email protected]. Phone: ++49-6131-990 450. Fax: between them (continuous-phase microreactors), while
++49-6131-990 305. E-mail: [email protected].
the second is to disperse one phase into the other using
Phone: ++44-20-7679-3811. Fax: ++44-20-7383-2348.
†
Institut für Mikrotechnik Mainz GmbH. an appropriate reactor inlet or a micromixer upstream
‡
Eindhoven University of Technology. of the reactor section (dispersed-phase microreactors).
§
University College London. In continuous-phase microreactors the gas and liquid
|
Johannes-Gutenberg-Universität Mainz. phases form two streams which are fed separately in
10.1021/ie0503139 CCC: $30.25 © 2005 American Chemical Society
Published on Web 10/26/2005
 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005 9751

the liquid and gas region of the reactor and are (ideally)
also withdrawn separately at the reactor outlet. The
crucial design issue of these reactors is the way the
interface, liquid film, or both are stabilized. In the
falling-film microreactor, gravity, in conjunction with
microstructured plates, create and stabilize the liquid
film. In overlapping-microchannel and mesh microre-
actors, the gas-liquid interface is stabilized by the
menisci that form in small (<5 µm) openings. The
advantage of such reactors is that the phases are not
intermixed and the gas-liquid interfaces are well-
defined. However, the pressures of the two phases have Figure 1. Picture of the falling-film microreactor and schematic
to be carefully controlled to avoid phase intermixing. showing the gas-liquid contacting. Reprinted with permission
In dispersed-phase reactors, the reactor is simply a from ref 9. Copyright 2003 Elsevier Ltd.
microchannel or even a larger size tube. A gas-liquid
dispersion is created by an inlet which induces merging continuous sheets, which may demonstrate waviness.
of the gas and liquid streams. It contains a dual feed At either the front end of those sheets or at low flow
arrangement or a multiple feed structure that splits the rates, the flow breaks up into rivulets, fingers, or a
phases in thin lamellae, which in the mixing section series of droplets.5 A falling-film reactor microreactor
form a dispersion. After they pass through the inlet developed by the Institut für Mikrotechnik Mainz
section, both phases are delivered to a single reactor GmbH (IMM)6,7 can generate stable films less than 100
region. The hydrodynamics obtained depend on various µm thick. The most critical part of the reactor is the
parameters, but one of the most critical is the ratio of stainless steel plate where the falling film is generated.
the gas-to-liquid flow rates. The flow obtained ranges Microchannels (typically 300 µm wide, 100 µm deep, 78
from bubble flow, the bubbles have a smaller diameter mm long separated by 100 µm wide walls) are fabricated
than the reactor channel (microbubble column reactor), using electrodischarge machining or wet-chemical etch-
to slug flow (or segmented or Taylor), the bubbles have ing. Supply and withdrawal of liquid are through
larger equivalent diameter than the reactor channel boreholes which are connected via one large slit to
(segmented-Taylor flow microreactor), as the ratio in- numerous small orifices at the top of the micro channels.
creases. Unlike conventional bubble columns, a uniform The slit acts as flow restrictor and aids the equipartition
bubble-size distribution can be obtained in these reac- of the liquid phase to parallel streams. The entire plate
tors. At very high values of the gas-to-liquid flow rate measures 89 × 46 mm and is housed in a stainless steel
ratio, annular flow is obtained. A thin liquid annulus enclosure shown in Figure 1. A structured heat-
film surrounds a gas core (annular flow microreactor), exchanger copper plate is inserted into a cavity beneath
and hence, the reactor becomes a continuous-phase the falling-film plate for temperature control. The top
reactor. Under these conditions flow disturbances such part of the housing has a view port covered by a thick
as waves can occur and entrainment of the liquid film piece of glass that allows inspection of the entire channel
into the gas core may take place. When a micromixer is section of the falling-film plate. In this way, the reactor
used upstream of the reactor channel and when the can also be used for photochemical reactions provided
latter has much larger diameter than the fluid lamella the window material is transparent to the wavelength
created in the micromixer, a foam is created (foam of interest.8 When both the top and bottom parts of the
microreactor). These reactors can provide a large gas- housing are placed together, a cavity is created above
phase void fraction. However, coalescence may occur, the plate through which the gas flows.
especially at large residence times. It must be noted 3.1.2. Overlapping-Channel and Mesh Micro-
that, although in dispersed-phase reactors transport reactors. In these types of reactors, the interface is
limitations between the two phases are significantly stabilized by well-defined openings between the gas and
reduced, there is generally a degree of uncertainty the liquid. The fluid layers have to be thin enough to
concerning the actual fluid geometries and available enhance mass transfer. One such reactor configuration
interfacial areas. Furthermore, the separation of the two suggested for liquid-liquid applications, where the
phases at the end of the process could be problematic partial overlapping of the (open on one side) channels
as gravity, often used in large scale-operations, is where the two fluids flow defined the available inter-
negligible. Phase-separation structures or devices may facial area, was found to have limited stability and a
therefore need to be incorporated downstream of the small interfacial area (2-15% of the total).10
reaction region, if the phases need to be collected A novel configuration was suggested by Central
separately. Both of the above types of gas-liquid Research Laboratories (CRL) that implements a mesh
microreactors (continuous-phase and dispersed-phase) structure to separate the planar chambers which con-
can be rendered catalytic by appropriate modifications tain the two fluids.11 The mesh has large open interfa-
as discussed in the next section. cial area (which can be as high as 40%) while the mesh
to wall distances can be set, generally, to 80-140 µm
3. Multiphase Microstructured Devices providing chamber volumes of ∼100 µL. The pore widths
3.1. Continuous-Phase Microreactors. 3.1.1. Fall- are in the 1-10 µm range, generally ∼5 µm, and provide
ing-Film Microreactor. Falling-film reactors employ adequate stability. The pore length (mesh thickness)-
thin liquid films that are created by a liquid feed falling to-width ratio is ∼1/1 and ensures low diffusive-
under gravitational pull. The liquid film is in contact transport resistance through the mesh. Minimization
with a solid support, which is usually either a thin wall of mixing between consecutive samples can be achieved
or stack of pipes. Conventional falling-film systems by radial flow from an entrance port that would flush
generate films with thickness on the order of 0.5-3 the reactor region. A quadrant-reactor configuration
mm.3,4 At large flow rates, the film takes the form of with a deep outlet channel at the circumference provides
9752 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005

Figure 2. Fully assembled micromesh reactor and microstructured mesh.

radial flow and minimizes flow resistances ensuring diameters (7 µm and 20 µm, respectively). The reaction
even flow of the fluidic streams (and consequently channels were 50 µm × 50 µm or 300 µm × 100 µm in
residence time) in the reactor (see Figure 2). cross section. Efficient temperature control was provided
The mesh is fabricated in nickel using a two-stage by a heat-transfer medium flowing in counter-current
electroplating method. It is placed between two glass mode in mini-channel arrays adjacent to the reaction-
layers that form the chambers for the two fluids. Struts channel array on both sides.
fabricated on the mesh align with pillars on the glass A two-channel dispersed phase microreactor was
to provide the necessary channel width of the reactant designed by de Mas et al. for selective direct fluorination
channels and provide additional support to the mesh. of aromatics.17,18 A single inlet was used for the liquid,
The reactor can be used in a stopped-flow mode for slow and there were two gas ports. At the outlet, the two
reactions or small sample volumes, where dispersion of reaction channels merge into a single channel and the
reactants and products can be minimized by withdraw- fluids exited via a single port. Each reaction channel
ing the product through the inlet port. With larger had a triangular cross section. For heat loads up to 0.3
sample volumes or faster reactions, flow-through opera- W, no external cooling was required.
tion is also possible.
Another type of segmented-flow microreactor has been
3.2. Dispersed-Phase Microreactors. 3.2.1. Seg-
proposed by Hardt et al.19 It consists of parallel plates
mented (Taylor) Flow. In the simplest form of these
which contain stripes of different wettability regions.
reactors, the two fluids are brought into contact in a
Liquid streams are injected though orifices in one of the
single channel. A variety of flow patterns can form but
plates and form lamellae spanning the space between
segmented or Taylor-bubble flow is the preferred op-
the plates and are confined within the stripes of highest
tion.12 These reactor configurations have many similari-
ties with the catalytic monolith reactors and a lot of wettability. By incorporating flow restrictions in the
work in Taylor-microchannel flow is aimed at under- liquid inlet region droplets are generated and are
standing and improving the conditions within the transported downstream by the gas flow. The wetting
monolithic reactor channels.13 The separation of phases properties of the regions can affect droplet shape and
at the reactor outlet can be a problem and a number of integrity.
approaches have been suggested that aid separation and 3.2.3. Foam Microreactor. One of the simplest
reduce pressure fluctuations as the bubbles exit. Hsieh systems for generating dispersed gas-liquid flows is to
and Yao used a separator consisting of two parallel use micromixers, followed by a tube of variable length
silicon wafers both containing an array of etched holes.14 and typical diameter of a few mm. The whole system
One wafer was coated with a hydrophilic film (liquid can be immersed in a thermostatic bath for temperature
separator that facilitates the removal of the liquid) and control. The mixers induce the gas and liquid streams
the other with a hydrophobic film (gas separator that to merge with or without prior splitting of the gas and
prevents the liquid from leaking through it). Günther liquid streams into finer substreams. The first category
et al.15 separated liquid from gas by a capillary separa- includes the interdigital (rectangular, triangular, and
tor at the end of the segmented flow channel. The slit-type mixers; Figure 3)20 and cyclone21 mixers, which
separator comprised 16 capillaries, each one ∼20 µm incorporate feed-channel arrays. The second category
wide. Separation was achieved by applying a pressure includes the caterpillar mixer (with22 or without23,24 a
difference smaller than the capillary pressure across the splitting plane). The incoming streams are fed into a
capillary microchannel structure. In this way, only gas dual-feed arrangement and the mixer subsequently
remained in the main channel. performs multiple splittings and recombinations of the
3.2.2. Microbubble Column. A scaled-out version flow. In contrast to reactors forming slug and annular
of a single-channel Taylor-flow reactor was presented flow patterns, such as the microbubble column, the
by Hessel et al.6,16 In this device, a microbubble column, reaction channel downstream of the mixing section is
the gas and liquid feeds were split into a number of of sufficiently large diameter so that the small bubbles
substreams and were subsequently brought into contact generated in the mixing section, pack together in the
in the reaction channels so that one gas and one liquid reaction channel resulting in foam flows. Coalescence
substream were introduced into one reaction channel. may take place, which can be addressed by controlling
To achieve flow equipartition, the gas- and liquid-inlet viscosity and surface tension, provided the reaction
channels were designed with very different hydraulic chemistry is not affected.
 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005 9753

Figure 3. Micromixers used for gas-liquid dispersions. Reprinted with permission from ref 20. Copyright 2004 Elsevier Ltd.

3.3. Catalytic Gas-Liquid-Solid Microreactors. channel is used to deliver catalyst as a slurry to the
3.3.1. Catalytic Falling-Film Microreactor. Cata- reaction channel and contains two ports to enable cross-
lysts can be incorporated in multiphase microreactors flow of the slurry. This provides even loading of particles
either as coatings or as packed beds. In the falling-film in multichannel reactors. To avoid the randomness in
microreactor, the catalyst was incorporated as thin packing that is inherent in small catalyst particle beds,
nonporous films or as particles in alumina-coated plates. which can result in flow maldistribution and a high-
For the former, sputtering of the microstructured plate pressure drop in multichannel reactors, Losey et al.
with 100 nm thick palladium film or UV-decomposition designed a fully micromachined packed-bed reactor
of a palladium acetate solution was employed. These made out of silicon.26 An ordered array of columns, 50
films showed evidence of cracking and flaking during µm in diameter, was etched into the reactor channels
nitrobenzene hydrogenation. Catalyst incorporation in mimicking the arrangement of a regular packed bed.
a high surface area 10 µm thick alumina film, deposited The columns were porosified and loaded with catalyst
using the slurry/washcoating method, followed by im- using conventional impregnation. Heating was ac-
pregnation or incipient wetness with palladium nitrate, complished by heating cartridges in the cover plates of
turned out to be superior for long-term activity, while the reactor used for sealing or by incorporating micro-
deactivation by organic compounds could be reversed fabricated heaters directly on the reactor. Because of
by suitable regeneration procedures.9 the large exposed specific surface area of the micro-
3.3.2. Packed-Bed Microreactor. Losey et al. in- reactors, the heat losses were very large and, even for
corporated standard porous catalysts in silicon-glass exothermic reactions, heating (not cooling) was required
microfabricated reactors consisting of a microfluidic- to keep the temperature constant.
distribution manifold, a single microchannel reactor or 3.3.3. Wall-Coated Microchannel Reactor. Fö-
a 10 microchannel array, and a 25 µm microfilter for disch et al. used a reactor module that can incorporate
immobilizing the solid particles in the reactor (see either a conventional packed bed or a stack of micro-
Figure 4).25,26 The fluid streams are brought into contact structured aluminum wafers.27 In the latter case, the
by a series of interleaved high aspect ratio inlet chan- reactor resembles monolith reactors,12 with the major
nels. Perpendicular to these channels, a 400 µm wide difference being the smaller size of the channels (0.3 ×
9754 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005

Figure 4. Single-channel microfabricated packed-bed reactor. Reprinted from ref 25. Copyright 2001 American Chemical Society.

Figure 5. Surface profile of an ethanol film at various flow rates in a falling film microreactor with 300 µm × 100 µm microchannels.

0.7 mm). The wafers were anodized and a Pd catalyst cause the liquid to bulge out of the channel until the
was deposited by electrochemical deposition or conven- hydrostatic pressure exceeds the surface-tension forces
tional impregnation. Both methods of catalyst incorpo- and the liquid bursts out of its channel-wall confines.
ration resulted in transverse Pd concentrations gradi- The film thickness observed for various microchannel
ents even in a 40 µm porous layer, except when long sizes was mostly smaller than existing correlations and
impregnation times were employed.28 No special inlet equations, which are for two-dimensional films and do
manifold was used and the results indicated that phase not account for gas-liquid-solid contact and surface
separation may have occurred (i.e., some microchannels tension. The highest specific interfacial area that cor-
could have been filled only with liquid while others by responds to these measurements is on the order of
gas). 20 000 m2/m3. This is much larger than that of conven-
tional gas-liquid reactors, such as bubble columns and
4. Hydrodynamics and Mass and Heat Transfer agitated tanks (up to 200 m2/m3).2,30
4.1. Continuous-Phase Microreactors. 4.1.1. Fall- Flow Equipartition. The quality of equipartition of
ing-Film Microreactor. Liquid-Film Profiles. In the the falling-film microreactor was evaluated by imaging
falling-film microreactor, the microchannels prevent the the wetting behavior of the uncovered reaction plate
breakup of the liquid film at low-flow rates. Because of with a 20 mL/h flow of acetonitrile.16 It was established
the combination of capillary forces and small channel that 90% of all substreams reached the bottom of the
widths, liquid being pulled up along the sides of the reaction zone within 17-18 s.
channels takes up a significant portion of the channel Axial Dispersion. Despite the small film size of the
width and the surface of the liquid film takes the form falling-film microreactors, simulations as well as experi-
of a flowing meniscus. The profile of the film surface ments have shown that significant axial dispersion can
has been characterized using laser-scanning confocal be present for typical liquid-phase diffusivity values,
microscopy29 (Figure 5). and this is attributed to convective rather than diffusive
At low-flow rates, the surface of the film curves effects.31 Depending on the channel cross-section (which
upward while moving from the center of the channel to in turn depends on the microstructuring method uti-
the walls. As the flow rate increases, the film thickness lized), the cross-sectional velocity profile displays a
also increases and the surface profile becomes flatter. single (for rectangular microchannels made by electrical
When the channel is completely filled, the profile discharge machining) or a double maximum (for rounded
becomes flat, while further increases in the flow rate corner microchannels made by wet-chemical etching).
 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005 9755

transfer in the liquid phase. The transverse concentra-

tion profiles of CO2 in the gas phase are not as steep as
the liquid phase as shown in parts c and d of Figure 6.
Decreasing the gas chamber depth from 5.5 to 2.5 mm
makes the concentration gradients slightly flatter (Fig-
ure 6e) thus improving CO2 transfer to the interface.
Higher NaOH concentrations, for similar gas-residence
times provide high reaction rates and lead to higher
conversion, not only resulting from the direct effect of
the concentration on the reaction rate but also from the
indirect effect of the ionic strength on the rate constant.
Heat Transfer. The good heat-transfer capability of
the falling-film microreactor was demonstrated by
monitoring the heat released during pure CO2 adsorp-
tion in a NaOH solution using an IR camera.34 In that
case, a 625 µm thick Si window was employed. A
maximum temperature rise of 0.6 K was observed for a
2 M NaOH solution. Maximum axial temperature
variations of 1.5 K were measured but this was caused
by the batchwise addition of CO2, which was admitted
in the gas chamber (in counter-current mode) at t ) 0
and its supply was stopped after ca. 20 s.
Mass-Transfer Coefficient. For nitrobenzene hy-
drogenation, the overall mass-transfer coefficient, kLa,
was conservatively estimated (on the basis of the film
thickness in the middle of the channels) to be in the
range of 3-8 s-1.35 By reducing the flow rate, and
therefore the liquid film thickness, one would obviously
increase the kLa values. However, the need to avoid
liquid film dry-out imposes a minimum practical flow
Figure 6. Transverse dimensionless CO2 concentration profiles rate. It is worth noting that decreasing the flow rate
across the liquid and gas films for an FF I falling film microreactor also results in an increase of the residence time and
(64 channels of dimensions 300 µm × 100 µm × 66.4 mm): (a) specific interfacial area. Despite the large mass-transfer
and (c) gas chamber depth 5.5 mm, 0.8% CO2, and 0.1 M inlet
coefficient, an analysis of the results indicated that the
NaOH concentration; (b) and (d) gas chamber depth 5.5 mm, 8%
CO2, and 1 M inlet NaOH concentrations and (e) gas chamber system was operating between the mass-transfer and
depth 2.5 mm, 8% CO2, and 1 M inlet NaOH concentration. Profiles kinetic-control regimes. As a comparison, for intensified
are shown for the axial dimensionless positions ζ ) 0.1 and 0.7. gas liquid contactors, kLa can reach 3 s-1, but for bubble
Zero denotes the gas-liquid interface. Reprinted from ref 33. columns and agitated tanks, it does not exceed 0.2 s-1.
Copyright 2005 American Chemical Society. For another hydrogenation reaction, cyclododecatriene
The dispersion can be reduced by a design composed of hydrogenation, significant reactant concentration gra-
alternating platelets wetted by the liquid film.31 The dients were observed during simulations in a 40 µm
platelets are arranged in such a way that the region of thick porous catalytic layer located on the external
the liquid film in contact with the wall of one platelet surface of microchannels.28
forms the free surface of the film of the next platelet. 4.1.2. Mesh Microreactor. Meniscus Stability. In
Such a configuration can enhance heat transfer by a the mesh microreactor, to avoid mixing of the two
factor of 3, compared to that of continuous plates, phases (breakthrough of the interface into one or the
because of the permanent entrance effect induced by the other phase), the meniscus should be stable and remain
alternating plates.32 within the pore of the mesh for the range of pressure
Mass Transfer. Mass transfer in the falling-film differences between the two phases established during
microreactor has been assessed for homogeneous and operation. Theoretical studies in single capillaries,
catalytic reactions. Specifically, carbon dioxide absorp- confirmed by experimental observations, allowed the
tion in sodium hydroxide solution was studied both range of pressure differences between the two phases
experimentally and theoretically.6,33 The model showed that maintain a stable meniscus to be evaluated.36
reasonable agreement with the experimental results Phenomena such as contact-angle hysteresis and me-
without any adjustable parameters, even though a two- niscus contortion at the pore ends increase the available
dimensional film geometry was utilized. This was partly range of pressure differences compared to those found
caused by the relative insensitivity of the model to the by simply using the Laplace equation. An example is
film thickness, which was in turn attributed to the fact given in Figure 7 for a system with pore radius of 10
that the reaction was under mass-transfer control. Parts µm and for a fluid pair with interfacial tension of 0.072
a and b of Figure 6 show the transverse CO2 concen- N/m and advancing and receding contact angles of 45°
trationprofiles in the liquid film for two cases that have and 15°, respectively. The position and shape of the
the same residence time, 6.4 s. The concentration of meniscus within the pore, as well as the available
OH- did not decrease to zero near the interface for any interfacial area, are given for the range of allowed
of the cases investigated, while the CO2 dissolved in the pressure differences between the nonwetting and wet-
liquid phase is consumed at the most within 25% of the ting phases (PNW - PW).11
liquid thickness near the interface. This shows that the Interfacial Area and Mass Transfer. The use of
limiting step for the gas-liquid reaction is CO2 mass the reactor has been demonstrated in gas-liquid and
9756 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005

Figure 7. Meniscus area and position as a function of the

pressure difference between phases (PNW - PW).

gas-liquid-solid catalytic chemistries.37 Solid catalyst

inserts were mounted on the glass wall of the liquid
phase. The mesh open area was about 20-25% which
led to gas-liquid interfacial area of 2000 m2/m3, well Figure 8. Taylor flow and streamlines within the liquid slug.
above the values (100-300 m2/m3) obtained in tradi-
tional stirred-tank reactors. Some mass transfer limita- the liquid slugs. The film surrounding the bubbles is
tions were found when a catalyst with high activity was the only means of communication between two succes-
used. The reactor was also used successfully in screen- sive slugs and in the majority of cases its thickness is
ing applications with a chiral ligand inventory down to only a fraction of a percentage of the tube diameter.
10 nmole. Also, within the slugs, the recirculation of the liquid
4.2. Dispersed-Phase Microreactors. 4.2.1. Seg- improves radial-mass transfer from liquid to wall and
mented Flow Microreactors. Hydrodynamics. When from gas to liquid.44,45 The combination of good radial-
the two fluids are brought into direct contact in a mass transfer and low axial-mass transfer in the liquid
channel, flow patterns similar to those seen in large- makes Taylor flow suitable for two-phase applications
diameter channels are obtained.38-40 These are bubbly, that involve mass transfer (fluid-fluid or wall-fluid)
churn, slug (segmented), and annular or ring annular or single-phase applications which suffer from a large
flow. Varying the wetting conditions of the channel wall amount of back-mixing.
was found to affect the flow patterns formed or even Recently, meandering channels were used to further
resulted in new ones such as the “skewed flow” pattern improve radial mixing in Taylor flow.46 Using micro-
seen in 100 µm quartz capillaries where the gas bubbles particle image velocimetry, Günther et al.46 observed
in slug flow are connected via gas stems (Yakitori that mixing is accelerated by the periodic switching of
flow39). Chen et al. also observed “bubble-train” flow (a the recirculation patterns within the liquid slugs. In
name usually reserved for Taylor-bubble flow) where addition, surface roughness and compressibility of the
several Taylor bubbles are connected together and only gas phase further improve radial mixing.
separated by a thin membrane between neighboring Bubble Shape and Dimensions. For reactive sys-
bubbles.40 Currently there is no agreement as to whether tems, the hydrodynamics in the liquid phase, the bubble
models for flow-pattern transition boundaries, pressure shape, length, and velocity, and the film thickness need
drop, and gas hold up derived from large-scale systems to be known. These depend on a number of dimension-
can be used satisfactorily to predict the data in micro- less groups derived from the Navier-Stokes equations.
channels39,41,42. In general, an increase in the capillary number, Ca, will
In microchannels, slug or segmented flow takes the result in an increase in the film thickness. At low
form of Taylor-bubble flow (Figure 8) which is one of velocities, inertia effects can be ignored. For Ca < 10-3,
the preferred patterns for two-phase operations. In this lubrication analysis shows the film thickness δ to be
pattern, long gas bubbles of an equivalent diameter ∼Ca1/3.47 Surprisingly, experiments confirmed this scal-
larger than the channel diameter are separated by ing law for higher capillary numbers, but for lower ones,
liquid slugs. The bubbles adopt a characteristic capsular the film thickness was substantially larger than that
shape and completely fill the channel cross section with predicted by theory. Ratulowski and Chang showed that
only a thin liquid film separating them from the tube Marangoni effects resulting from the presence of trace
wall. The presence of bubbles in front and at the back impurities in the flow can explain the discrepancies
of the liquid slugs modifies the flow field in the liquid between the theory and experiments at low Ca.48 In fact,
compared to single-phase flow. As first suggested by numerical simulations, where impurities are avoided,
Taylor, at high capillary numbers (Ca ) µUb/γ, where have confirmed the theory (e.g., in refs 49 and 50). At
µ is the liquid viscosity, Ub is the bubble velocity, and γ low Ca the shape of the bubble caps can be assumed to
is the interfacial tension), the liquid will completely be spherical and the film quickly approaches its final
bypass the bubble, but more commonly, at lower Ca, thickness away from the bubble caps, resulting in a
toroidal vortices form into the slug43 (Figure 8). cylindrical bubble body. With an increase in Ca, ripples
Axial Mixing. The separation of the bulk liquid by start to form at the rear end of the bubble. At Ca > 10-3,
the gas bubbles results in reduced axial mixing between the Reynolds number (Re ) FUbd/µ, where d is the
 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005 9757

channel diameter and F is the liquid density) also has

an effect. As the Re increases, the rear ends of the
bubbles are flattened and the amplitude of the ripples
increases,49 while the film thickness also increases.51
The flow pattern inside long slugs acquires a parabolic
profile at some distance away from the bubble caps. This
distance increases with increasing Re. For vertical
arrangements, the Froude number (Fr ) Ub2/gd) is also
important and its increase causes an increase in the
rising-bubble velocity. The effects of Re and Fr can be
neglected for low Ca.
Bubble and slug lengths in microchannels mainly
depend on the inlet conditions, as the significance of
surface tension forces means that once the Taylor
bubbles form little change to their size is expected
within the channel as a result of breakup or coalescence.
Although there is abundant literature on the formation
of small bubbles, usually in an unconfined liquid, very
little information is available on Taylor-bubble forma-
tion in microchannels. Amador et al. carried out experi-
ments in different microchannel inlet configurations
(concentric and T- and Y-junctions) using water and
octane that have different interfacial tensions and
identified three main mechanisms of bubble formation.52
These, in some cases, were accompanied by bubble Figure 9. Effect of slug length, Ls, on (a) the tracer concentration
pairing at the gas inlet or, when bubbles smaller than of exiting slugs and (b) on the reactor residence-time distribution
the tube diameter were formed, by coalescence. Bubble curves. Input concentration 1 in cell 1, dimensionless film thick-
lengths depended on the ratio of gas-to-liquid, the ness ) 10-2, dimensionless reactor length ) 1000. Reprinted with
superficial velocities, and the size of the gas inlet. The permission from ref 57. Copyright 2005 Wiley-VCH.
fluid with the lower-interfacial tension (octane) pro-
duced smaller bubbles. nounced, which was taken into account with a variation
of the CSTR-PFR model. The above models allow
Pressure Drop. The presence of the bubbles in a
analytical expressions to be derived for the reactor-exit
single-phase liquid flow results in an additional pres-
concentration.
sure drop-term because the different curvatures at the In more sophisticated models, a unit cell, composed
front and back of the bubble result in capillary pressure of a bubble and a slug or more commonly of a slug and
(Jamin effect). Bretherton’s lubrication analysis pro- two-half bubbles, is considered. The region of the slug
vides an expression for pressure drop.47 At higher Ca, can be divided into two separate regions, a closed-vortex
an empirical correlation was suggested by Kreutzer that one and an enclosing-open thin-liquid annulus film. This
related the friction factor for Taylor flow to the slug forms the basis of the two-region models. Convolution
length and the Re and Ca numbers.12 The underpre- methods are then used to obtain, from the results in a
diction of the experimental data by the numerical single cell, the residence-time distributions in a whole
simulations was attributed to the Marangoni phenom- Taylor-flow channel. In their model, Pedersen and
ena. The same trends were found for channels with a Hovarth assumed that the two regions are well-mixed
rectangular cross section.15 In general, bubbles cause and a mass-transfer coefficient is used to account for
an increase in the pressure drop compared to that in a the mass transfer between them.55 Thulasidas et al.
single-phase liquid flow, and more bubbles (per unit assumed that the two regions are well mixed in the axial
channel length) result in a higher pressure drop. direction and exchange mass only by diffusion in the
Residence Time Distribution. A possible applica- radial direction.56 A one-dimensional diffusion equation
tion of Taylor flow is in the field of combinatorial is then used to account for mass transfer. The equation
chemistry where segmentation can be utilized for in- is solved for a time equivalent to that of a single
troducing sequentially different reactants/samples within circulation time of the vortex, and then the concentra-
a microchannel reactor. A narrow residence-time dis- tions of the two regions are averaged. This model
tribution during Taylor flow, compared to that of a showed good agreement with their experimental data.
single-phase flow, has been demonstrated experimen- Kreutzer developed a continuous model where the film
tally.46 Residence-time data can be used to show how is modeled as stagnant zone and the slugs as plug-flow
samples spread in the reactor and consequently provide zones with no axial dispersion.12 Mass transfer between
guidelines regarding the frequency of sample injection. the film and the slugs was accomplished via a mass-
In an early attempt to predict axial mixing in Taylor transfer coefficient. The model12 also predicted the
flow, Thiers et al. assumed perfect mixing between the experimental residence-time distributions by Thulasidas
slugs and the film and derived a tank-in-series model et al.56 satisfactorily. Recently, Salman et al., using CFD
connected by the film surrounding the bubbles.53 Sal- analysis to obtain the residence-time distribution in a
man et al. restricted the assumption of perfect mixing single unit cell, demonstrated that an increase in the
to low Bodenstein numbers, Bo, (Bo < 500, where Bodenstein number, the capillary number, or the slug
Bo) Ubd/D, D is the diffusivity in the liquid phase) and length results in larger tracer spreading in terms of
developed a CSTR-PFR in series model.54 It was also time.57 The effect of slug length on the concentration of
shown that at very low Bo, because of the increased the exciting slugs and on the residence-time distribution
influence of diffusion, axial mixing becomes more pro- can be seen in Figure 9 parts a and b, respectively. It is
9758 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005

worth noting that when short slugs are used the tracer
spreads to a larger number of slugs, but its spread over
time is lower.
Mass-Transfer Coefficient. In the above models,
quite often mass transfer coefficients are required from
the gas to the liquid and from the liquid to the wall.
Berčič and Pintar, using methane as the transferred
species, experimentally obtained a correlation for an
overall gas-to-liquid mass-transfer coefficient,45 which
was later generalized by Berčič for other gases.58 kLa
values up to about 0.25 s-1 were found. In the majority,
however, investigators consider two different contribu-
tions to the overall mass-transfer coefficient, from the
cylindrical part of the bubble and from the bubble ends.
Irandoust et al. experimentally studied the mass trans-
fer of oxygen to different liquids and found kLa values Figure 10. Flow-pattern map derived for single-channel (300 µm
as high as 0.3 s-1.59 In their model for vertical config- × 100 µm) operation in a microbubble column for water and
uration, the film was treated as having the parabolic isopropanol using a dimensionless representation based on the
profile of a falling film with convection dominating in Weber number and the ratio of the gas and liquid velocities.
Reprinted with permission from ref 62. Copyright 2002 VDI-
the axial direction and diffusion in the radial, while for Verlag.
the bubble ends a correlation of the mass-transfer
coefficient derived from solid spheres was used. The
final expression was found to predict reasonably well
the experimental data when a correction factor was
applied. Numerical simulations showed that both the
cylindrical part of the bubble and the spherical caps
made similar contributions to mass transfer. Their
expression however, was found to underpredict the
results by Berčič and Pintar.45 In contrast to their
findings, Berčič and Pintar also found that the mass-
transfer coefficient was independent of the bubble
length and only dependent on slug length.45 Van Baten
and Krishna also suggested two contributions for the
mass transfer.60 For the bubble ends, penetration theory
was used, while for the liquid film, a model based on
transient mass transfer to a falling film by Pigford was Figure 11. Specific interfacial areas derived for single-channel
considered. Computational fluid dynamics simulations (300 µm × 100 µm) operation in a microbubble column for water
confirmed the validity of their simple model. The and isopropanol versus the dimensionless Weber number. The
simulations gave kLa values between 0.04 and 0.25 s-1. straight and dotted lines refer to interpolations. Reprinted with
Liquid-to-wall mass transfer coefficients have also permission from ref 62. Copyright 2002 VDI-Verlag.
been measured (e.g., ref 61). Kreutzer suggested a
simple model for this mass transfer by ignoring the film changes between the different flow patterns are nearly
surrounding the bubbles and simplifying the gas-liquid the same for both systems at similar Weber numbers.
interface to flat ends.12 It was shown that mass transfer Thus, a prediction for similar changes is possible for
from the liquid to the wall could be an order of different gas-liquid systems taking into account that
magnitude faster than that in single-phase flow through the same channel geometries and flow velocities are
capillary. used.62
4.2.2. Microbubble Column. Flow-Pattern Maps. A variety of flow patterns was also seen in the dual
For the microbubble column, flow-pattern maps were channel gas-liquid reactor developed at MIT.17 Bubble,
given by Haverkamp.62 A dimensionless flow-pattern slug-annular, and churn flows were observed, while
map is obtained when plotting the Weber number during annular flow, dry patches can appear on the wall.
versus the velocity ratio between the gas and liquid Interfacial Area. Gas-liquid systems can also be
phases. The latter ratio is related to the shear forces characterized by the interfacial area depending on the
and the Weber number is defined as Weber number. When the liquid velocity is kept con-
stant and the gas velocity is changed, the Weber number
wG2deqFL
We ) can be used to predict interfacial areas for different
δ liquid-flow velocities. Figure 11 shows the phase-bound-
ary areas for the systems isopropanol/nitrogen and
with
water/nitrogen in the microbubble column.
wG ) gas velocity Measured data (circles, squares) are compared to
continuous lines which denote interpolated functions of
δ ) liquid film thickness
the Weber number. The maximum interfacial areas of
FL ) liquid density 19 000 m2/m3 for isopropanol/nitrogen and of 18 000 m2/
deq ) equivalent channel diameter m3 for water/nitrogen are observed at similar Weber
numbers above 1.5.62
In Figure 10, a flow-pattern map for the systems Figure 12 shows the calculated degree of conversion
isopropanol/nitrogen and water/nitrogen is shown. The for the catalytic oxidation of butyraldehyde as a function
 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005 9759

bubble and the wall, while the other two were obtained
from experimental correlations.63 The model was
used for parametric studies in monolithic reactors. In
Kreutzer’s model, mass transfer between the bubble and
the wall and between the liquid slug and the wall were
both calculated from film theory, while for the mass
transfer from the bubble caps to the liquid slug,
penetration theory was used. Discrepancies were found
at high Bo when compared to a numerical solution and
were attributed to high-concentration gradients near the
bubble caps.
4.2.4. Foam Microreactors. Flow Patterns. The
foam flows generated by interdigital micromixers under
certain conditions consist of closed packing of bubbles
of very uniform size (Figure 13). This type of flow has
been termed “hexagon flow” and can contain bubbles as
small as 120 µm.64 Discontinuous “stop-and-go” flow has
Figure 12. Calculated degree of conversion for the oxidation of also been observed for a large liquid/gas flow rate ratio.
butyraldehyde as a function of the specific interfacial areas for
For mixers with multiple inlet channels, the mixing
operation in a microbubble column. Reprinted with permission
from ref 62. Copyright 2002 VDI-Verlag. chamber geometry was shown to affect bubble size-
distribution and average bubble size.20 A slit-shaped
mixing section (with the smallest slit width), which
contains restrictions for bubble formation and bubble
pinch-off, led to the smallest bubbles and a narrow
bubble size distribution. Not all of the gas-feed channels
were found to be active during bubble formation and
most of them were flooded by the liquid. The influence
of capillary forces decreases with increasing flow rate
and an increasing number of active gas-feeding channels
were observed accordingly. However, this was associated
with a wider bubble-size distribution. A mixer with a
single gas feed encompassed by two liquid-feed channels
(Figure 3) circumvented the problem of flow nonequi-
partition and improved the bubble-size distributions
even further.
Figure 13. Foam formation in a rectangular interdigital mixer.
Residence-Time Distribution. The residence-time
Top image shows the flow in the mixing chamber and the bottom
image shows it in a glass tube connected to the mixer outlet. distribution of foam microreactors was characterized by
Reprinted with permission from ref 20. Copyright 2004 Elsevier Pennemann et al. for a standard stainless-steel multi-
Ltd. channel interdigital mixer.65 Most of signal dispersion
(in dimensionless terms) was found to occur in the
of the specific interfacial areas for operation in a mixer, while in the reactor tube, it was found to be
microbubble column.62 negligible. Dispersion increased with the total flow rate
A simple reactor model given in ref 62 is used for and gas-liquid ratio and was affected by reactor tube
rough estimation of the conversion in the microbubble orientation; gravity led to up to five times higher
column. It turns out that an increase in gas velocity dispersion for the vertical tube than the horizontal one.
leads to an increase in the interfacial area. However, If instead of the multichannel feed the single gas
this is done at the expense of the residence time channel feed interdigital mixer is utilized, tracer disper-
(reduction). These competing effects result in a maxi- sion can be reduced.20
mum for conversion as shown in Figure 12. At low 4.2.5. Packed-Bed Microreactors. Interfacial Area.
interfacial areas, which are established at low-gas For reactors where a regular array of microcolumns is
volumetric flow rates, residence time, although high, is incorporated into the microchannel, in addition to slug
not sufficient and conversion decreases. If the flow rate and annular, other flow patterns have been observed.26
becomes higher, a maximum of conversion is achieved These are “dispersed flow” which is encountered when
because of an optimal combination of interfacial area both the gas- and liquid-flow rates are increased and is
and residence time. Then, conversion decreases again characterized by an intimate dispersion similar to the
as a result of a shorter residence time. As a consequence, bubbly and foam flow obtained by micromixers. In the
for every gas-liquid reaction, both parameters, inter- microfabricated packed-bed reactor, reported by Losey
facial area and residence time, have to be optimized.62 et al., pulsating flow has been observed, similar to
4.2.3. Segmented-Flow Catalytic Systems. Mass macroscale analogues but at slightly different flow
Transfer. Taylor-flow reactor models in which the conditions.25 In the dispersed-flow regime, the inter-
catalyst is deposited on channel wall have been sug- facial area was estimated to be 16 000 m2/m3, while for
gested by Cybulski et al.63 and Kreutzer.12 In both slug flow obtained at lower liquid-flow rates, it was
models, three different resistances to mass transfer are estimated to be 1500 m2/m3. In conventional trickle
used, namely from the cylindrical part of the bubbles beds, specific interfacial area can reach (for pulse flow)
to the wall, from the gas caps to liquid, and from the up to 500 m2/m3.66
liquid slug to wall. Cybulski used film theory to calcu- Mass-Transfer Coefficient. For cyclohexene hydro-
late the mass-transfer coefficients between the gas genation in micro-packed-bed reactors, the mass-
9760 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005

transfer coefficient was found to be in the range of 5-15 ucts.73,74 Because of the large heats released and the
s-1 (for powdered catalyst25) and 3-7 s-1 (for porous corresponding autoaccelerating effect on the rate of
silicon microcolumns26). Typical values for laboratory fluorination, explosions happened quite frequently with
trickle-bed reactor systems are 0.01-0.08 s-1.67 The such direct fluorinations.74 Furthermore, the insolubility
good mass-transfer characteristics of the microreactors of fluorine in most solvents hinders good control of
are the result of the high gas-liquid interfacial area reactant dosing because the gas-liquid interface, which
generated by the microreactor, which is unattainable is typically not actively tuned, is now the means to
with its macro-scale counterpart because the two phases determine and control the mass transfer, reaction rate,
are forced to mix over the catalyst in a constrained and selectivity.
volume relative to the large pellet size in trickle-bed Conventional ways to overcome the problems encoun-
reactors. However, if one considers the penalty in terms tered with the direct fluorination were to reduce side
of increased resistance to flow and hence the associated and consecutive reactions by working at very low
energy dissipation, it was concluded that the micro- temperatures down to -80 °C and the dilution of the
reactor was operating as efficiently as a standard substrates to very low concentrations.73,75-78 In this way,
reactor. More specifically, the energy dissipation of the fundamental studies could be made, and indeed, it was
microreactor process was higher (V ) 2-5 kW/m3) as confirmed that an electrophilic pathway is possible with
compared to laboratory trickle-bed reactors (V ) 0.01- direct fluorination as it is for the other halogenations,
0.2 kW/m3).25 such as chlorinations and brominations.
4.2.6 Scale-Out Issues. Scaling-up dispersed-phase Multiphase microstructured reactors approach in a
microreactors generally seems to be more complicated more direct manner the issues relevant for direct
than it is for continuous-phase ones because interfacial fluorinations.17,79-83 The specific interfaces are large
forces cause the phases to rapidly coalesce. As charac- (e.g., 20 000 m2/m3 and more). The magnitudes of the
teristic size decreases, the capillary forces become more gas-liquid interfaces are precisely known and, for the
important. According to the Young-Laplace equation68 continuous-phase microreactors, are quasi-constant for
a pressure difference of 0.03 atm would be needed to the whole operation time (i.e., fluctuations of the
create bubbles 100 µm in diameter in water. Since interfacial area have much less impact). Thus, high-local
bubbles in different channels do not form simulta- fluorine concentrations at the interfaces can be achieved
neously, dynamic pressure fluctuations during bubble and conversion is greatly enhanced. The large heat
formation occur and these can lead to flow instabilities, released is efficiently transferred by integrated micro-
channeling, and unequal flow distribution between heat exchangers with large specific transfer areas.
channels.7 Thus, to avoid maldistribution, the gas and Finally, the short channel length allows a defined short
liquid should be introduced through individual inlet reaction time to be set, which limits the exposure under
“pressure-drop” channels that present a pressure drop the aggressive conditions to the kinetically needed time.
significantly higher than that across the reaction chan- All microreactors used for fluorinations bear some or
nels.17 De Mas et al.18 incorporated pressure-drop chan- all these traits.
nels in a 20 channel reactor and showed that the Initial work by Chambers et al. started with ethyl
pressure-drop channels prevented cross talk between acetoacetate, a β-keto ester.81,82 High yields of the mono-
reaction channels and allowed them all to operate under fluorinated product at short reaction times were ob-
the same flow regime avoiding channeling. However, at tained, often better than standard batch laboratory
high gas-flow rates, the reactor operates under annular- protocols. As a particular microreactor advantage, in
flow conditions, and under these circumstances, pres- addition to those listed above, it was found that the
sure-drop channels are not necessary. Instead, manifold metallic construction material changes the keto/enol
channels with a larger equivalent diameter than the equilibrium. The enol species, much more labile to
reaction channels can ensure flow equipartition.69 This fluorination, is formed in a higher concentration so that
approach is similar to that employed for single-phase the chemistry in the multiphase microreactor seems to
flow,70 according to which a ratio of equivalent diam- be surface-driven.
eters of distribution to reaction channel of 20 would
ensure flow equipartition (within 1%) for up to 250
reaction channels.

5. Applications
More details on gas-liquid applications in micro-
structured reactors than given here can be found in refs
23b, 79-81, and 93.
5.1. Gas-Liquid Reactions. 5.1.1. Direct Fluori- Later, the novel continuous-chemical protocol was
nation of Aliphatics and Non-C Moieties. Fluorina- verified for a number of other β-keto esters. such as
tions are typically performed by circumventing routes, ethyl 2-chloro-3-oxobutanoate82,83 or ethyl 2-methyl-3-
such as the Balz-Schiemann and Halex processes, oxobutanoate.83 Five- and six-ring β-keto ester struc-
because the direct route using elemental fluorine and tures, such as 3-acetyl-3,4,5-trihydrofuran-2-one,83 2-ace-
an organic substrate is known to be unselective and tylcyclohexan-one,83 and ethyl 2-oxocyclohexane
harmful.71,72 The latter reaction is so fast and exother- carboxylate,83 were directly fluorinated as well.
mic, when initiated, that it is accompanied by large heat 5.1.2. Direct Fluorination of Benzenoid Aromat-
release. The local rise in temperature increases consid- ics. Fluorinated aromatics are particularly important
erably the rate of radical scission of the elemental as intermediates for drug synthesis in the pharmaceuti-
fluorine. The fluorine radicals thus formed react via cal industry. The use of elemental fluorine presents the
nonselective radical and oligomerization pathways to same problems as mentioned above for the fluorination
give a broad spectrum of side and consecutive prod- of aliphatics. Because of the sp2 hybridization of the
 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005 9761

Figure 14. Chromatogram of the reaction products of the direct

fluorination of toluene showing the substitution pattern
(3.3:1.0:1.9) of the ortho-, meta-, and para-fluorotoluenes. Re-
printed with permission from ref 17. Copyright 2003 American
Chemical Society.

aromatic core, there is, in addition, a chance for addition

besides the desired electrophilic substitution pathway,
especially if fluorine radicals attack. Such reactions are Figure 15. Pilot-scale microstructured reactor for direct fluorina-
tion with elemental fluorine composed of 30 parallel-operated
not selective; they often undergo C-C bond cleavage to channels. Reprinted with permission from ref 69. Copyright 2005
undefined ring-ruptured products. Since the latter Royal Society of Chemistry.
initially are radicals, oligomerizations to higher-molec-
ular species of an undefined chemical nature may follow. solvent.17 While fluorinations in laboratory flasks take
Another peculiarity of direct aromatic fluorination is hours, a falling-film microreactor does the same job
that substitution isomerism and follow-up reactions to within seconds, and a microbubble column can do it even
multiple-fluorinated species are likely to happen in in microseconds.7 Accordingly, the respective space-
competition to the usually favored monosubstitution to time yields of the microstructured reactors are higher
one defined isomer. by orders of magnitude.7
Electrophilic substitution was achieved using micro- de Mas et al. obtained conversions of 44-77% and
structured reactors, such as the falling-film microreactor yields of 60-78% for different contents of the solvents
and the microbubble column by Jähnisch et al.7,79,80 or and different flow rates, using formic acid/acetonitrile
the dual-channel microreactor by de Mas et al.,17,18 even mixtures.17
at high-substrate concentration (typically about 0.1 M,
but also up to 1.0 M84) and high fluorine-gas content
(up to 50%) and the temperatures were not required to
be so low (-10 °C up to room-temperature being
sufficient). In addition to the fact that substituted mono-
fluoro products were formed with reasonable yield, the
distribution of isomers found is a clear indicator that
an electrophilic pathway with fluorine cation intermedi- A scale-out concept for performing fluorinations was
ates indeed occurred7,17 (Figure 14). recently presented by Chambers et al.69 They used the
direct fluorination of ethyl acetoacetate in formic acid
by fluorine (10-50%) as model reaction and constructed
reactors with 9, 18, and 30 microchannels (Figure 15).
When an average value of about 0.2 g product per
hour per microchannel is assumed, the productivity of
a two-sided 30-channel reactor device for fluorinations
Addition and oligomerization products, the latter are amounts to about 300 g per day. By scale-out to 10
often solids which may clog a microreactor, were not microstructured reactors operated in parallel, the ca-
found under regular conditions.7 The role of the solvent pacity is increased to about 3 kg of product per day,
is crucial. Inert solvents are required. Fully fluorinated which is in the range of pilot-plant operation.
apolar solvents, such as octafluorotoluene, give worse 5.1.3. Direct Fluorination of Heterocyclic Aro-
results than those of polar environments, which favor matics. Selective fluorination of quinoline aromatics
the scission of fluorine into (partly) charged species. leads to various commercially important life-science
Thus, methanol and, still better, acetonitrile turned out products such as 5-fluoroacil, 5-fluoroprimaquine, and
to be the solvents of choice for microreactor fluorina- ciprofloxacin where the fluorine moiety replacing the
tions.7 Formic acid is even more effective, probably via hydrogen has a significant impact of the chemical and
the protons released and the fact that it hardly offers biological properties of the whole molecule.
any reactive site for the fluorine.69 The fluorination of quinoline in acetonitrile using a
Yields of up to 28% were obtained by Jähnisch et al. microstructured reactor by Chambers et al. gave only
when using the solvent acetonitrile at conversions tar-like products.86,87 The use of formic acid resulted in
ranging from 7 to 76% and at selectivities ranging from a complex mixture of “regular” mono- and poly-fluori-
31 to 43%.7 In later work, under more optimized nated low-molecular weight products in addition to tar.
conditions, conversions from 17% to 95% and selectivi- When concentrated sulfuric acid was used, no tar
ties from 37% to 10% were achieved using methanol as formation was observed at 0-5 °C. At moderate conver-
9762 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005

sion, high selectivities were achieved, totaling about The Bayer-Villiger oxidation of cyclohexanol to cyclo-
91%. Substitution was achieved in the electron-rich hexanone was made with fluorine and aqueous formic
benzenoid core only, and not in the electron-poor pyri- acid (5% water) with 60% conversion and 88% selectiv-
dine-type core. The positional order of the reactivity ity.88 In all cases, the use of a microreactor circumvents
within the quinoline molecule followed the order of C the application of toxic heavy-metal oxidants.88 The only
atoms 5 > 8 > 6 which is in line with the electrophilic microflow-processing byproduct is HF which could be
reactivity determined by proton/deuterium exchange recycled by electrolysis.
studies in strong acid media. This means that positions 5.1.5. Photochlorination of Aromatic Isocyan-
R to the heteroring are more readily attacked than the ates. Side-chain photochlorination of toluene isocyan-
β positions. ates yields important industrial intermediates for poly-
urethane synthesis, one of the most important classes
of polymers.89 Illuminated thin-liquid layers in micro-
channels should have a much higher photon efficiency
(quantum yield) than that given for conventional pro-
cessing.
Ehrich et al. investigated the toluene-2,4-diisocyanate
reaction with chlorine to 1-chloromethyl-2,4-diisocyana-
tobenzene, a gas-liquid reaction of interest to BASF.89
There is a radical route to the desired chain-substituted
product, and an electrophilic route to the ring-substi-
tuted toluene-5 chloro-2,4-diisocyanate, dependent on
the processing conditions.

For substituted quinolines, the manner in which

electron-releasing and -withdrawing groups may affect
the regioselectivity was investigated. Generally, no
directing effect of the first substituent was observed (i.e.,
the product profile was the same as that found for
quinoline itself). Fluorination of electron-releasing groups
such, as -CH3 and -OCH3, yielded an ortho and para
pattern, consistent with an electrophilic-substitution
path, and electron-withdrawing groups, such as -NO2,
yielded a meta-substitution pattern.

Yields of the product, 2,4-diisocyanatobenzene, were

from 24% to 54% when a falling-film microreactor (130
°C) was used.89 The conversions ranged from 30% to
81% at selectivities from 79% to 67%, respectively. A
selectivity-conversion plot (Figure 18) shows that the
5.1.4. Oxidations of Alcohols, Diols, and Ketones content of the ring-substituted product decreases with
with Fluorine. Fluorine can also be used to mediate increasing conversion (12-5%), while simultaneously,
oxidation reactions.88 This is possible by direct fluorine the other byproducts were formed in a much larger
reaction, or it can be done in an indirect manner. The amount (8-29%). Since conversion goes parallel to
latter relies on the intermediate generation of oxygen residence time, this clearly points to the fact that the
transfer reagents such as HOF • MeCN by the reaction byproducts are mainly formed by consecutive reaction
of aqueous acetonitrile with elemental fluorine. steps. Control experiments in a 30 mL batch-reactor
The oxidation of cyclohexanol to cyclohexanone with reaction volume took 30 min of reaction time to result
fluorine and aqueous acetonitrile at room temperature in a conversion of 65% at a selectivity of 45%, whereas
by Chambers et al. gave a conversion of 84% and a the falling-film reactor needed only about 14 s for the
selectivity of 74%.88 In a similar way, diols such as 1,2- same performance. In terms of space-time yields, the
cyclohexan-diol were partly or fully oxidized. At a falling film-microreactor achieved up to 401 mol/h, while
conversion of 87%, the monooxidation product was the batch reactor yielded only 1.3 mol/h.89 The impact
obtained with 53% selectivity and the dioxidation of Lewis-acid formation on the reaction course was
product with a 30% yield. tested by replacing the nickel-plated microreactor con-
struction material with iron.89 The plate surface is
converted to iron chloride under the reaction conditions.
The selectivity to the target product then drops from
67% to 50%.
5.1.6. Monochlorination of Acetic Acid. Clariant
GmbH headed for increasing selectivity for the monochlo-
rination of acetic acid to give chloro-acetic acid.90 This
product is amenable to further chlorination to give
dichloro-acetic acid under the reaction conditions. The
removal of this impurity demands a laborious and costly
 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005 9763

Figure 16. Microplant for toluene sulfonation with a falling-film microreactor as the central apparatus and unitized backbone as the
fluidic bus system. Reprinted with permission from ref 91. Copyright 2004 Elsevier Ltd.

separation, either by crystallization or hydrogen reduc- The latter were attached to a newly developed fluidic
tion with a Pd catalyst. backbone providing unitized ports and plant building-
block sizes, yielding a fluidic bus. A delay-loop reactor
was added to complete the toluene pyrosulfonic acid
conversion. Anhydride hydrolysis was performed in a
tempered liquid-liquid microstructured reactor by mix-
ing with water. Several sampling stations were inserted
in the microreactor plant to monitor the impact of the
reaction time and the type of operations on the distribu-
tion of the products.91,92
For molar ratios of sulfur trioxide/toluene from 5/100
to 15/100, the selectivity to toluenesulfonic acid is
Wehle et al. obtained a yield of 90% with a falling- constant at about 73%, while the selectivity for anhy-
film microreactor which is above that of large-scale dride formation decreases from 8 to 2%. The selectivity
bubble-column processing.90 Selectivity was much bet- for the pathway to ditolyl sulfone is low at about 3%. If
ter, as less than 0.05% dichloro-acetic acid was formed, sampling is done downstream of the delay loop, the
while conventional processing typically gives 3.5%. selectivity of toluenesulfonic acid increases to 82%
5.1.7. Sulfonation of Toluene. The sulfonation of because of the conversion of toluene pyrosulfonic acid
toluene has a more complex scenario of elemental with toluene. The anhydride selectivity is further de-
reactions, compared to those of the other gas-liquid creased to 2%, while that of ditolyl sulfone is not altered.
reactions presented above. Numerous side and consecu- After the reaction mixture is hydrolyzed with water, a
tive reactions are possible as follows:91,92 toluene pyro-
nearly quantitative conversion of the anhydride to the
sulfonic acid, toluenesulfonic acid, ditolyl sulfone,
acid is achieved, if mixing conditions and excess of water
toluenesulfonic anhydride, and a mixed anhydride are
are optimized. Totally, a selectivity of 82% for the target
formed.
product, toluenesulfonic acid, is achieved at nearly
Thus, the sulfonation of toluene demands precise
control of the concentration and residence time.91,92 In complete conversion.
addition, temperature control is an issue because the 5.1.8. Photo-oxidation of R-Terpinene and Cy-
reaction is highly exothermic. To have this kind of clopentadiene. The photo-oxygenation of cyclopenta-
process control, a whole continuously operated plant was diene with Rose Bengal as photosensitizer was carried
equipped with many microflow components (Figure 16). out by Jähnisch et al. in methanol in the presence of
oxygen8 (see refs 93 and 94 also). The endoperoxide
forms as intermediate and is reduced to give 2-cyclo-
penten-1,4-diol, which is used for pharmaceutical-drug
synthesis. Batch processing is hindered, since the
explosive intermediate is formed in greater amounts.
9764 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005

The microreactor photo-oxidation was carried out coated.25,26 The resulting surface of 100 m2 is not far
using a xenon lamp as a light source, and the irradiation from the porosity provided by the catalyst particles
was made via a quartz-glass window.8 Thiourea in employed otherwise as fixed bed. In one study, a reactor
methanol reduced the labile endoperoxide online to the with such catalytic-wall microcolumns was compared to
stable diol product. The yield of 2-cyclopenten-1,4-diol another reactor containing the catalyst particles as fixed
amounts to 20% at 10-15 °C. bed. When normalized for the metal content of the
A [4+2]-cycloaddition of singlet oxygen to R-terpinene device, the reaction rates of the columnar and the
in methanol yielding ascaridole was conducted by Woot- particle-containing reactor become similar, 6.5 × 10-5
ton et al. in the presence of catalytic amounts of Rose mol/min m2 and 4.5 × 10-5 mol/min m2, respectively.
Bengal as photosensitizer in a one-channel microreactor The increase in the interfacial area in the micro-
made of glass.95 Enrichment of the explosive epiperoxide reactor is paid by more flow resistance, which means
ascaridole is avoided because of the minute active energy dissipation. The energy dissipation factor, the
microreactor internal volumes. The quantum yield is power unit per reactor volume, of the microreactor
also enhanced by the large penetration of light through process was thus higher (V ) 2-5 kW/m3) as compared
the thin liquid layer. This is of particular importance to the laboratory trickle-bed reactors (V ) 0.01-0.2 kW/
for this reaction because the dissolved Rose Bengal m3).25 When the still larger gain in mass transfer is
possesses a large extinction coefficient and thus absorbs considered, an overall gain in performance for the
the radiation after a short path length. The operation microreactor can be stated.
was thus feasible at a high-sensitizer concentration of 5.2.2. Hydrogenation of p-Nitrotoluene and Ni-
5 × 10-3 M, while still profiting from a large optical trobenzene over Pd/C and Pd/Al2O3. Nitroaromatics
transmittance of 95%. Because of the high number of owe their great importance in organic synthesis to being
active photons in the microreactor channel, a high intermediates in the generation of their respective
conversion of about 80% is obtained at an irradiation anilines by hydrogenation.27,97 For instance, pharma-
time as short as 5 s.96 ceuticals are produced via that route.97 The hydrogena-
tions of nitroaromatics have high intrinsic reaction
rates, which however cannot be exploited by conven-
tional reactors as they are unable to cope with the large
heat releases resulting from the large reaction enthal-
pies (500-550 kJ/mol)9,27,97. For this reason, the hydro-
gen supply is restricted, thereby controlling the reaction
The microreactor photoprocessing is thought to fa-
rate. Otherwise, decomposition of the nitroaromatics or
cilitate scale-up and unwanted sample heating, avoiding
of the partially hydrogenated intermediates can occur.9
the expeditious use of collimators and refrigeration to
As model reactions, the hydrogenations of nitroben-
control the tungsten lamp power.96 In addition, safety
zene and p-nitrotoluene over supported noble metal
issues which arise for aerated-oxygenated organic sol-
catalysts were investigated. These hydrogenations in-
vents can be addressed differently. No presaturation of
volve various elemental reactions with different inter-
the R-terpinene with oxygen is required, and directly
mediates which can react with each other, as depicted
after the reaction, the sampling volume can be instantly
in the scheme below for p-nitrotoluene.9
degassed wih nitrogen.
5.2. Gas-Liquid-Solid Reactions and Homoge-
neously Catalyzed Gas-Liquid Reactions. 5.2.1.
Cyclohexene Hydrogenation over Pt/Al2O3. Cyclo-
hexene hydrogenation is a well-studied reaction espe-
cially in conventional trickle-bed reactors (see original-
literature citations in refs 25 and 26) and thus serves
well as model reaction. It is a fast process so mass-
transfer limitations are likely.26 Because of the exother-
mic nature of hydrogenations, the avoidance of hot spots
resulting in a reduction of selectivity and of catalytic
activity is a second driver.25 The hydrogenation of p-nitrotoluene in the presence
of a supported Pd catalyst by Födisch et al. gave 100%
selectivity.27,97 Three different ways to prepare the Pd
catalyst were used. For an impregnated catalyst on an
aluminum oxide wash coat, the conversions were 58-
Conversions ranging from 2.8% to 16.0% were deter- 98% depending on the process conditions (Table 1). The
mined for a packed-bed microreactor by Losey et al.25 conversion for an electrodeposited catalyst was 58%,
(see also ref 96). Average reaction rates ranged from 8.6 whereas an impregnated catalyst on an electro-oxidized
× 10-4 mol/min gcat to 1.4 × 10-3 mol/min gcat25 (see also nanoporous substrate had an 89% conversion. The best
ref 96), which is close to the intrinsic reaction rate of microreactor conversion was similar to that of a con-
3.4 × 10-3 mol/min gcat. As an excess of cyclohexene was ventional fixed-bed reactor (85%), while the maximum
used, the kinetics were zero order for this species’ microreactor yield of 30% was superior.
concentration and first order with regard to hydrogen.25 For the hydrogenation of nitrobenzene by Yeong et
For this pseudo-first-order reaction, a volumetric rate al., an even wider variation of preparation procedures
constant of 16 s-1 was determined with a catalyst for the palladium catalyst was tested.9
surface area of 0.57 m2/g and a catalyst loading density
of 1 g/cm3.
An array of columns was formed using a plasma-etch
process; they were subsequently porosified and catalyst
 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005 9765

Table 1. Results for the Hydrogenation of p-Nitrotoluene

Using Different Microchannel Reactorsa
MPd Tcalc Mp-nitrotoluene/Ageom τ conversion
catalyst (mg) (K) (g/h cm2) (s) rr (%)
MEIb 23 773 0.013 280 43 58
MChc 24 773 0.013 280 43 96
MChc 21 903 0.013 280 43 98
MChc 21 903 0.045 85 21 58
WChd 20 903 0.045 260 21 89
FBe 10 773 1.7 × 10-6 90 21 85
a Comparison of those results with ones obtained by applying a

conventional fixed-bed catalyst (from ref 27). b Electrochemically

deposited Pd. c Chemically deposited Pd using formalin. d Chemi-
cally deposited Pd on anodized aluminum wires using formalin.
e Conventional fixed-bed catalyst.

Figure 18. Arrhenius plot for the Pd (O) and Pt ([) alumina-
supported catalysts. Continuous microreactor operation, 2-3 bar,
1 M reactant solution in methylcyclohexane, 0.1 cm3/min liquid
flow rate. Reprinted with permission from ref 37. Copyright 2004
Royal Society of Chemistry.

different Reynolds numbers. When the flow rate was

increased, the reaction time was reduced and the film
thickness was increased. Conversion decreased, showing
a linear dependence on log Re. There was only a weak
dependence of hydrogenation conversion on hydrogen
pressure.
Figure 17. Nitrobenzene conversion and aniline selectivity as a 5.2.3. Heterogeneous Hydrogenation of R-Methyl
function of reaction time for a plate coated by incipient wetness Styrene. The hydrogenation of R-methylstyrene is a
Pd catalyst. Reprinted with permission from ref 9. Copyright 2003 standard process for elucidating mass-transfer effects
Elsevier Ltd. in catalyst pellets and in fixed-bed reactors (see ref 98
and original literature cited therein). The physical
A sputtered palladium catalyst exhibited low conver- properties of R-methylstyrene are well described in the
sion and large deactivation of the catalyst (60 °C, 4 bar). literature, and the intrinsic kinetics are known.
The corresponding selectivity was also low. A slightly
better performance was obtained after an oxidation/
reduction cycle. Following a steep initial deactivation,
the catalyst activity stabilized at 2-4% conversion and
at about 60% selectivity. After reactivation, the selectiv-
ity initially approached 100%. As side products, all
intermediates except phenylhydroxylamine were identi- The reaction was carried out by Losey et al. with a
fied. For a UV-decomposed palladium catalyst, conver- palladium catalyst supported on activated carbon using
sion was found to be slightly higher than that for the a microstructured packed-bed reactor.98 It is moderately
sputtered one. A spectrum of side products similar to fast at room temperature and 1 atm of hydrogen. The
that for the sputtered catalyst was obtained. For an microreactor, however, was operated at 50 °C. The
impregnated palladium catalyst, complete conversion reaction is first order with regard to hydrogen and zero
was achieved and maintained for 6 hours. Selectivity order with regard to R-methylstyrene. Conversions
decreased with time, but still remained at a high level. ranging from 20% to 100% were determined for a
The best performance of all catalysts investigated was packed-bed microreactor.98 Initial reaction rates were
found for an incipient-wetness palladium catalyst, which close to 0.01 mol/min per reaction channel without prior
initially had more than 90% conversion. A 75%-conver- activation of the catalyst.98 This is in agreement with
sion at selectivity of 80% was reached for long times on literature data on intrinsic kinetics.
stream (see Figure 17). Similar findings were made with Pd/Al2O3 and Pt/
The catalyst lifetimes for the four types of catalysts, Al2O3 catalysts supported on a microstructured mesh
prepared by different preparation routes, depended on reactor by Abdallah et al.11,37 The global rate constants
the catalyst loading which was related to the prepara- were 56 s-1 and 1.4 s-1 for the Pd and Pt catalysts,
tion route.9 The larger the loading, the longer the respectively.37 An activation energy of 46 ( 5 kJ mol-1
catalysts could be used before reactivation. The four was found for the Pt catalyst in the mesh reactor, being
catalysts had the following sequence of lifetime and comparable to the value of 39 kJ mol-1 reported for a
activity: wet impregnation > incipient wetness > UV commercial Pt/Al2O3 powder catalyst in a well-behaved
decomposition of precursors > sputtering. Several routes batch reactor (Figure 18).
for the reactivation of the used catalyst were tested, The results thus demonstrate operation in the chemi-
such as dissolution of organic residues by dichloro- cal regime and not in a mass-transfer controlled one.
methane or burning them by heating in air. In this way, For the Pd catalyst, however, an activation energy of
the initial activity was recovered, thus regaining com- close to zero is found in the mesh reactor, different from
plete conversion. the value of 41 kJ mol-1 reported for a commercial
The impact of the hydrodynamics of the falling film powder catalyst in a well-behaved batch reactor. The
on the hydrogenation of nitrobenzene was also investi- activity of the catalyst is too high and mass transfer is
gated.35 The flow rate was varied to give processing at limiting.
9766 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005

Figure 19. SEM micrograph of monodisperse silica nanoparticles generated by a segmented-flow microreactor (left). Percentage of mean
diameter of the silica nanoparticles as a function of the residence time for a batch reactor and a segmented-flow microreactor (SFR).
Reprinted from ref 103. Copyright 2004 American Chemical Society.

5.2.4. Asymmetric Hydrogenation of Cinnamic Table 2. Benchmarking of Figures of Merit for the
Acid Derivatives. The suitability of microreactor Gas-Liquid Screening Using a Well-Behaved Mini Batch
and a Continuous Foaming-Flow Microreactor102
operation for catalyst and ligand screening for a homo-
geneously catalyzed gas-liquid reaction in a micro- benchmarked property of
reactor was demonstrated for the asymmetric hydroge- the g/l screening mini batch microflow
nation of Z-methylacetamidocinnamate (mac) with rhod- reaction vol (ml) 10 0.1
ium chiral diphosphine complexes. A microstructured av amount of Rh/expt (µg) 500-1000 5-20
mesh reactor was used providing a stable gas-liquid typical amount of ligand/expt (µmol) 10 0.1
temp range (°C) 20-100 20-80
interface.37 Up to 20 chiral diphosphines were evaluated pressure range (bar) 1-100 1-11
by Abdallah et al. For the very active catalysts, continu- residence time >10 min 1-30
ous-flow operation with a residence time of less than 1 av TTF achieved during study (d-1) 2 15
min was sufficient. For less active catalysts, the flow max actual achievable TTF (d-1) 3 40
was stopped and a batch-wise operation (e.g., a 30 min range of solvents large restricted
reaction time) was conducted. A fairly good agreement automation of reagent/catalyst injection no yes
automation of sample collection no yes
between the published and measured enantiomeric
excess (ee) data was found, which underlines the suit-
ability of the microreactor for catalyst screening con-
(60/40 wt %) and hydrogen.99-101 To stabilize the gas-
cerning gas-liquid homogeneous processes. In addition,
liquid interface, sodium dodecyl sulfate was added as
the mesh microreactor allowed facile investigation of the
surfactant. In this way, small pulses of 200 µl volume
effect of hydrogen pressure on ee.
with well-characterized and diminished pulse broaden-
ing (see ref 65) were achieved. Efficient mass transfer
in a kinetically controlled manner was achieved by the
large specific gas-liquid interfaces of the foams of up
to 50 000 m2/m3.99,100
A benchmarking of the efficiency of the new screening
method was done by comparison to a well-behaved mini
Another kind of screening technique for the hydro- batch reactor (see Table 2). A microreactor test unit runs
genation of a cinnamic methyl ester was developed by 15 and 40 tests per day in average and at maximum,
the transient serial screening of rigid foam flows by de respectively, whereas only 2 and 3 such tests can be
Bellefon et al.99-102 The microreactor setup included processed in a mini batch.102 Two hundred fourteen tests
valves for pulse injection, a micromixer for foaming, and were conducted in one investigation.101 The consumption
a delay tube for reaction of the flowing foam. The of noble metal and chiral ligands per test was orders of
reaction was carried out using ethylene glycol/water magnitude lower for a microreactor test unit compared
 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005 9767

to that of a mini batch (5-20 µg of noble metal and 10 reactions. Fluorinations, oxidations, hydrogenations,
µmol chiral ligand for the microreactor and 500-1,000 and sulfonations, as well as photochemical reactions,
µg of noble metal and 0.1 µmol chiral ligand for the mini have been carried out with such reactors and significant
batch).102 performance advantages have been demonstrated in
The enantiomeric excess (ee) distribution of the terms of selectivity, space-time yield, and safety. Mul-
hydrogenation of a cinnamic methyl ester was quite tiphase microstructured reactors thus offer novel tools
narrow: 90% of all data were within 40% to 48% ee.100 for demanding chemical gas-liquid processes.
Experimental data were fitted to several empirical
models, the best agreement was found with a statistical
model with first-order kinetics for hydrogen.100 With this Literature Cited
model, a parity diagram was given, showing that 29
(1) Mills, P. L.; Ramachandran, P. A.; Chaudhari, R. V. Mul-
(17%) of the experiments from 170 had to be rejected, tiphase reaction engineering for fine chemicals and pharmaceu-
whereas all the others were adequately described. Both ticals. Rev. Chem. Eng. 1992, 8, 5-176.
the kinetic constant and the activation energy were (2) Lee, S.-Y.; Tsui, Y. P. Succeed at gas-liquid contacting.
lower for a microreactor test unit than for a batch Chem. Eng. Prog. 1999, 95, 23-49.
reactor (micro/batch: k323 ) 9.3 vs 19.1 m3/kmol MPa (3) Dabir, B.; Riazi, M. R.; Davoudirad, H. R. Modelling of
min, Ea ) 31 vs 40 kJ/mol).100,102 This was attributed falling film reactors. Chem. Eng. Sci. 1996, 51, 2553-2558.
to the very low inventory of material.100 (4) Karimi, G.; Kawaji, M.; An experimental study of freely
falling films in a vertical tube. Chem. Eng. Sci. 1998, 53, 3501-
5.3. Specialty Applications. 5.3.1. Use of Seg- 3512.
mented Gas-Liquid Flow for Particle Synthesis. (5) Oron, A.; Davis, S. H.; Bankoff, S. G. Long-scale evolution
Segmented Taylor gas-liquid flows (i.e., alternating of thin liquid films. Rev. Mod. Phys. 1997, 69, 931-980.
regular-sized gas and liquid portions) can be used to (6) Hessel, V.; Ehrfeld, W.; Herweck, T.; Haverkamp, V.; Löwe,
separate small liquid reacting entities from each other H.; Schiewe, J.; Wille, C.; Kern; T., Lutz, N. Gas-liquid microre-
by gas bubbles for the benefit of well-defined residence actors: hydrodynamics and mass transfer. In IMRET 4, Proceed-
time with reduced axial dispersion (see Section 4.2). ings of the 4th International Conference on Microreaction Tech-
nology, Atlanta, GA, March 5-9, 2000; AIChE: New York, 2000;
Gas-liquid segmented flow was applied to particle pp 174-186.
synthesis in a micromixer with two liquid inlets for (7) Jähnisch, K.; Baerns, M.; Hessel, V.; Ehrfeld, W.; Haverka-
mixing the precursors of the particle synthesis and a mp, V.; Löwe, H.; Wille, C.; Guber, A. Direct fluorination of toluene
gas inlet for separating the continuous liquid streams using elemental fluorine in gas-liquid microreactors. J. Fluorine
into fragments separated by air gas segments.46,103 Chem. 2000, 105, 117-128.
Using the Stöber synthesis, tetraethyl orthosilicate (8) Jähnisch, K. Photochemical generation and [4+2]-cyclo-
(TEOS) in ethanol was hydrolyzed by a solution of addition of singlet oxygen in a falling film microreactor. Chem.
Ing. Tech. 2004, 76, 630-632.
ammonia, water, and ethanol by Khan et al.103 In this
(9) Yeong, K. K.; Gavriilidis, A.; Zapf, R.; Hessel, V. Catalyst
way, silicic acid monomer Si(OH)4 is produced, which preparation and deactivation issues for nitrobenzene hydrogena-
then undergoes polycondensation to form colloidal silica tion in a microstructured falling film reactor. Catal. Today 2003,
SiO2. In this way, monodisperse nanoparticles of a 81, 641-651.
diameter of 200-500 nm were generated (Figure 19). (10) Bibby, I. P.; Harper; M. J.; Shaw, J. Design and optimi-
Such particles have industrial applications (e.g., in sation of micro-fluidic reactors through CFD and analytical
pigments, catalysts, sensors, health-care, antireflective modelling. In IMRET 2, Proceedings of the 2nd International
Conference on Microreaction Technology, New Orleans, 1998;
coatings, and chromatography). AIChE: New York, 1998; pp 335-339.
5.3.2. Falling-Film H2O2 Evaporation. The driver (11) Wenn, D. A.; Shaw, J. E. A.; Mackenzie, B. A mesh
for using a microstructured falling-film evaporator by microreactor for 2-phase reactions. Lab Chip 2003, 3, 180-186.
Klemm et al. was to reach maximum hydrogen peroxide (12) Kreutzer, M. T. Hydrodynamics of Taylor flow in capillaries
evaporation per unit area at minimum decomposition and monolith reactors, Ph.D. Thesis, Technical University of Delft,
and minimum demand of carrier gas.104 A carrier gas Delft, The Netherlands, 2003.
(nitrogen) transports the hydrogen peroxide which is (13) Irandoust, S.; Andersson, B. Mass transfer and liquid-
phase reactions in a segmented two-phase flow monolithic catalyst
evaporated from its aqueous solution. Throughputs,
reactor. Chem. Eng. Sci. 1988, 43, 1983-1988.
increasing with temperature, were from 1 to 3 kg of (14) Hsieh, C.-C.; Yao, S.-C. Development of a microscale
hydrogen peroxide/h m2. The relative decomposition was passive gas-liquid separation system. Presented at the 5th
typically lower than 10% and could be reduced to 1-3% International Conference on Multiphase Flow, Yokohama, Japan
for optimum operation conditions (e.g., by increasing 2004; Paper 566.
temperature). It is assumed that decomposition does not (15) Günther, A.; Jhunjhunwala, M.; Thalmann, M.; Schmidt,
take place in the liquid phase, but rather in the vapor M. A.; Jensen, K. F. Micromixing of miscible liquids in segmented
phase by contact with non-wetted parts of the evapora- gas-liquid flow. Langmuir 2005, 21, 1547-1555.
(16) Hessel, V.; Ehrfeld, W.; Golbig, K.; Haverkamp, V.; Löwe,
tor. H.; Storz, M.; Wille, C.; Guber, A.; Jähnisch, K.; Baerns, M. Gas-
liquid micro reactors for direct fluorination of aromatic compounds
using elemental fluorine. In IMRET 3, Proceedings of the 3rd
6. Conclusions International Conference on Microreaction Technology, Frankfurt,
Germany, 1999; AIChE: New York, 1999; pp 526-540.
Many gas-liquid and gas-liquid solid reactors manu- (17) de Mas, N.; Günther, A.; Schmidt, M. A.; Jensen, K. F.;
factured by microstructuring and precision engineering Microfabricated multiphase reactors for the selective direct fluo-
have been evaluated in terms of hydrodynamics and rination of aromatics. Ind. Eng. Chem. Res. 2003, 42, 698-710.
mass and heat transfer. Interfacial areas as high as (18) de Mas, N.; Gunther, A.; Schmidt, M. A.; Jensen, K. F.
50 000 m2/m3 and mass-transfer coefficients on the order Scalable microfabricated multiphase reactors for direct fluorina-
tion reactions. In Digest of Technical Papers, Boston Transduc-
of 10 s-1 have been obtained. Both constitute significant ers’03, Boston, MA, June 8-12, 2003; IEEE: Washington, D.C.,
improvements compared with those of conventional 2003; Vol. 1-2, pp 655-658.
reactors. Such excellent characteristics provide unique (19) Hardt, S.; Pennemann, H.; Jiang, F. CFD-based optimiza-
reactive environments particularly for fast exothermic tion of gas-liquid micro reactors with droplet and bubble transport.
9768 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005

In Technical Proceedings of the 2003 Nanotechnology Conference (38) Triplett, K. A.; Ghiaasiaan, S. M.; Abdel-Khalik, S. I.;
and Trade Show; Computational Publications: Boston, 2003; pp Sadowski, D. L. Gas/liquid two-phase flow in micro channels. Part
254-257. I: two-phase flow patterns. Int. J. Multiphase Flow 1999, 25, 377-
(20) Löb, P.; Pennemann, H.; Hessel, V. g/l-Dispersion in 394.
interdigital micro mixers with different mixing chamber geom- (39) Serizawa, A.; Feng, Z.; Kawara, Z. Two-phase flow in micro
etries. Chem. Eng. J. 2004, 101, 75-85. channels. Exp. Therm. Fluid Sci. 2002, 26, 703-714.
(21) Hardt, S.; Dietrich, T.; Freitag, A.; Hessel, V.; Löwe, H.; (40) Chen, W. L.; Twu, M. C.; Pan, C. Gas/liquid two-phase flow
Hofmann, C.; Oroskar, A.; Schönfeld, F.; Vanden Bussche, K. in microchannels. Int. J. Multiphase Flow 2002, 28, 1235-1247.
Radial and tangential injection of liquid/liquid and gas-liquid (41) Triplett, K. A.; Ghiaasiaan, S. M.; Abdel-Khalik, S. I.;
streams and focusing thereof in a special cyclone mixer. IMRET LeMouel, A.; McCord, B. N. Gas/liquid two-phase flow in micro
6, Proceedings of the 6th International Conference on Microreac- channels. Part II: void fraction and pressure drop. Int. J. Multi-
tion Technology, New Orleans, March 11-14, 2002; AIChE: New phase Flow 1999, 25, 395-410.
York, 2002; Publication 164, pp 329-344. (42) Akbar, M. K.; Plummer, D. A.; Ghiaasiaan, S. M. On gas-
(22) Schönfeld, F.; Hessel, V.; Hofmann, C. An optimised split- liquid two-phase flow regimes in micro channels. Int. J. Multiphase
and-recombine micro mixer with uniform ‘chaotic’ mixing. Lab Flow 2003, 29, 855-865.
Chip 2004, 4, 65-69. (43) Thulasidas, T. C.; Abraham, M. A.; Cerro, R. L. Flow
patterns in liquid slugs during bubble-train flow inside capillaries.
(23) Hessel, V.; Löwe, H.; Müller, A.; Kolb, G. Chemical Micro
Chem. Eng. Sci. 1997, 52, 2947-2962.
Process Engineering-Processing and Plants; Wiley-VCH: Wein-
(44) Irandoust, S.; Andersson, B. Simulation of flow and mass
heim, Germany, 2005.
transfer in Taylor flow through a capillary. Computers Chem. Eng.
(24) Hessel, V.; Hardt, S.; Löwe, H. Chemical Micro Process 1989, 13, 519-526.
Engineering-Fundamentals, Modelling and Reactions; Wiley- (45) Berčič, G.; Pintar, A. The role of gas bubbles and liquid
VCH: Weinheim, Germany, 2004. slug lengths on mass transport in the Taylor flow through
(25) Losey, M. W.; Schmidt, M. A.; Jensen, K. F. Microfabri- capillaries. Chem. Eng. Sci. 1997, 52, 3709-3719.
cated multiphase packed-bed reactors: characterization of mass (46) Günther, A.; Khan, S. A.; Thalmann, M.; Trachsel, F.;
transfer and reactions. Ind. Eng. Chem. Res. 2001, 40, 2555-2562. Jensen, K. F. Transport and reaction in microscale segmented
(26) Losey, M. W.; Jackman, R. J.; Firebaugh, S. L.; Schmidt, gas-liquid flow. Lab Chip 2004, 4, 278-286.
M. A.; Jensen, K. F. Design and fabrication of microfluidic devices (47) Bretherton, F. P. The motion of long bubbles in tubes. J.
for multiphase mixing reaction. J. Microelectromech. Systems Fluid Mechanics 1961, 10, 166-188.
2002, 11, 709-717. (48) Ratulowski, J.; Chang, H. C. Marangoni effects of trace
(27) Födisch, R.; Hönicke, D.; Xu, Y.; Platzer, B. Liquid phase impurities on the motion of long gas bubbles in capillaries. J. Fluid
hydrogenation of p-nitrotoluene in micro channel reactors. In Mechanics 1990, 210, 303-328.
Microreaction Technology-IMRET 5, Proceedings of the 5th (49) Edvinsson, R. K.; Irandoust, S. Finite-element analysis of
International Conference on Microreaction Technology; Matlosz, Taylor flow. AIChE J. 1996, 42, 2, 1815-1823.
M., Ehrfeld, W., Baselt, J. P., Eds.; Springer-Verlag: Berlin, 2001; (50) Taha, T.; Cui Z. F. Hydrodynamics of slug flow inside
pp 470-478. capillaries. Chem. Eng. Sci. 2004, 59, 1181-1190.
(28) Platzer, B.; Xu, Y.; Rossmann, T.; Födisch, R.; Hönicke, (51) Aussilous, P.; Quére, D. Quick deposition of a fluid on the
D. Consideration of the effect of irregular catalytic active compo- wall of a tube. Phys. Fluids 2000, 12, 10, 2367-2371.
nent distributions in mesopores-extension of a model for wall- (52) Amador, C.; Salman, W.; Sanguanpiyapan, S.; Gavriilidis,
catalyzed reactions in micro channel reactors. Chem. Eng. Technol. A.; Angeli, P. Effect of gas inlet conditions on the mechanism of
2003, 26, 765-773. Taylor flow formation. Presented at the 5th International Confer-
(29) Yeong, K. K.; Gavriilidis, A.; Zapf, R.; Kost, H. J.; Hessel, ence on Multiphase Flow, Yokohama, Japan, 2004; Paper 515.
V.; Boyde, A. Characterisation of liquid film in a microstructured (53) Thiers, R. E.; Reed, A. H.; Delander, K. Origin of the lag
falling film reactor using laser scanning confocal microscopy. Exp. phase of continuous flow analysis curves. Clin. Chem. 1971, 17,
Therm. Fluid Sci. 2005, accepted for publication. 1, 42-48.
(30) Levenspiel, O. Chemical Reaction Engineering, 3rd edition; (54) Salman, W.; Gavriilidis, A.; Angeli, P. A model for predict-
Wiley: New York, 1999. ing axial mixing during gas-liquid Taylor flow in micro channels
(31) Hardt, S.; Doffing, F.; Pennemann, H. Simulation of at low Bodenstein numbers. Chem. Eng. J. 2004, 101, 391-396.
hydrodynamic dispersion in gas-liquid micro reactors. In Proceed- (55) Pedersen, H.; Hovarth, C. Axial dispersion in a segmented
ings of the 5th International Conference on Modeling and Simula- gas-liquid flow. Ind. Eng. Chem. Fundam. 1981, 20, 181-186.
tion, San Juan, Puerto Rico, 22-25 April 2002; Computional (56) Thulasidas, T. C.; Abraham, M. A.; Cerro, R. L. Dispersion
Publications: Boston, 2002; pp 54-57. during bubble-train flow in capillaries. Chem. Eng. Sci. 1999, 54,
61-76.
(32) Hardt, S.; Schönfeld, F. Simulation of hydrodynamics in
(57) Salman, W.; Angeli, P.; Gavriilidis, A.; Sample pulse
multiphase microreactors. In Proceedings of the 5th World Con-
broadening in Taylor flow micro channels for screening applica-
gress on Computional Mechanics, Vienna, Austria, 7-12 July,
tions. Chem. Eng. Technol. 2005, 28, 509-514.
2002; Mang, H., Rammersdorfer, F., Ebersteiner, J., Eds;
WCCM: Vienna, Austria, 2002; http://wccm.tuwien.ac.at. (58) Berčič, G. Influence of operating conditions on the observed
reaction rate in the single channel monolith reactor. Catal. Today,
(33) Zanfir, M.; Gavriilidis, A.; Wille, Ch.; Hessel, V. Carbon 2001, 69, 147-152.
Dioxide Absorption in a Falling Film Microstructured Reactor:
(59) Irandoust, S.; Ertlé, S.; Andersson, B. Gas/liquid mass
Experiments and Modelling. Ind. Eng. Chem. Res. 2005, 44, 1742-
transfer in Taylor flow through a capillary. Can. J. Chem. Eng.
1751.
1992, 70, 115-119.
(34) Wille, C.; Ehrfeld, W.; Herweck, T.; Haverkamp, V.; Hessel, (60) van Baten, J. M.; Krishna, R. CFD simulations of mass
V.; Löwe, H.; Lutz, N.; Möllmann; K.-P.; Pinno, F. Dynamic transfer from Taylor bubbles rising in circular capillaries. Chem.
monitoring of fluid equipartition and heat release in a falling film Eng. Sci. 2004, 59, 2535-2545.
micro reactor using real-time thermography. In Proceedings of the (61) Gruber, R.; Melin, T. Radial mass transfer enhancement
VDE World Microtechnologies Congress, MICRO.tec 2000, EXPO in bubble-train flow. Int. J. Heat Mass Transfer 2003, 46, 2799-
Hannover, Germany, 25-27 September 2000; VDE Verlag: Berlin, 2808.
2000; pp 349-354. (62) Haverkamp, V. Charakterisierung einer Mikroblasensäule
(35) Yeong, K. K.; Gavriilidis, A.; Zapf, R.; Hessel, V. Experi- zur Durchführung stofftransportlimitierter und/oder hochexo-
mental studies of nitrobenzene hydrogenation in a microstructured thermer Gas/Flüssig-Reaktionen. Dissertation, Universität
falling film reactor. Chem. Eng. Sci. 2004, 59, 3491-3494. Erlangen-Nürnberg, Erlangen, Germany, 2002 (in Fortschritt-
(36) Amador, C.; Angeli, P.; Gavriilidis, A.; Shaw, J.; Wenn, D. Bericht VDI, Reihe 3, Nr. 771).
Meniscus shape, position and stability in straight pores. In (63) Cybulski, A.; Edvinsson, R.; Irandoust, S.; Andersson, B.
Proceedings of the AICHE Meeting, San Francisco, 16-21 Novem- Liquid-phase methanol synthesis: modelling of a monolithic
ber, 2003; AIChE: New York, 2003; Paper 277a. reactor. Chem. Eng. Sci. 1993, 48, 3463-3478.
(37) Abdallah, R.; Meille, V.; Shaw, J.; Wenn, D.; de Bellefon, (64) Hessel, V.; Ehrfeld, W.; Golbig, K.; Haverkamp, V.; Löwe,
C. Gas-liquid and gas-liquid-solid catalysis in a mesh micro- H.; Richter, T. Gas-liquid dispersion processes in micro mixers:
reactor. Chem. Commun. 2004, 372-373. the hexagon flow. In Process Miniaturization. Proceedings of the
 Ind. Eng. Chem. Res., Vol. 44, No. 25, 2005 9769

2nd International Conference on Microreaction Technology, IM- the photochemical chlorination of alkylaromatics. Chimia 2002,
RET 2, Topical Conf. Preprints, New Orleans, 1998; Ehrfeld, W., 56, 647-653.
Rinard, I. H., Wegeng, R. S., Eds.; AIChE, New York, 1998; pp (90) Wehle, D.; Dejmek, M.; Rosenthal, J.; Ernst, H.; Kamp-
259-266. mann, D.; Trautschold, S.; Pechatschek, R. Verfahren zur Her-
(65) Pennemann, H.; Hessel, V.; Kost, H.-J.; Löwe, H.; de stellung von Monochloressigsäure in Mikroreaktoren. 27.07.2000,
Bellefon, C. Investigations on pulse broadening for transient 2000.
catalyst screening in gas/liquid systems. AIChE. J. 2004, 50, (91) Müller, A.; Cominos, V.; Horn, B.; Ziogas, A.; Jähnisch,
1814-1823. K.; Grosser, V.; Hillmann, V.; Jam, K. A.; Bazzanella, A.; Rinke,
(66) Ramachandran, P. A.; Chaudhari, R. V. Three-Phase G.; Kraut, M. Fluidic bus system for chemical micro process
Catalytic Reactors; Gordon and Breach: New York, 1983. engineering in the laboratory and for small-scale production.
(67) Al-Dahhan, M. H.; Larachi, F.; Dudukovic, M. P.; Laurent, Chem. Eng. J. 2005, 107 (1-3), 205-214.
A. High-pressure trickle-bed reactors: a review. Ind. Eng. Chem. (92) Müller, A.; Cominos, V.; Hessel, V.; Horn, B.; Schürer, J.;
Res. 1997, 36, 3292-3314. Ziogas, A.; Jähnisch, K.; Hillmann, V.; Grosser, V.; Jam, K. A.;
(68) Adamson A.; Gast, A. Physical Chemistry of Surfaces, 6th Bazanella, A.; Rinke, G.; Kraut, M. Fluidisches Bussystem für die
edition; Wiley: New York, 1997 . chemische Verfahrenstechnik und für die Produktion von
(69) Chambers, R. D.; Fox, M. A.; Holling, D.; Nakano, T.; Feinchemikalien. Chem. Ing. Tech. 2004, 76, 641-651.
Okazoe T.; Sandford, G. Elemental fluorine Part 16. Versatile thin- (93) Jähnisch, K.; Hessel, V.; Löwe, H.; Baerns, M. Chemistry
film gas-liquid multichannel microreactors for effective scale-out. in Microstructured Reactors. Angew. Chem., Int. Ed.2004, 43,
Lab Chip 2005, 5, 191-198. 406-446.
(70) Amador, C.; Gavriilidis, A.; Angeli, P. Flow distribution
(94) Jähnisch, K.; Baerns, M. Verfahren zur Photooxygenierung
in different microreactor scale-out geometries and the effect of
von Olefinen. 04.12.2002, 2002.
manufacturing tolerances and channel blockage. Chem. Eng. J.
2004, 101, 379-390. (95) Wootton, R. C. R.; Fortt, R.; de Mello, A. J. A Microfabri-
(71) Schiemann, G.; Cornils, B. Chemie und Technologie cated nanoreactor for safe, continuous generation and use of singlet
Cyclischer Fluorverbindungen; Ferdinand Enke Verlag: Stuttgart, oxygen. Org. Proc. Res. Dev. 2002, 60, 187-189.
Germany 1969; pp 188-193. (96) Losey, M. W.; Isogai, S.; Schmidt, M. A.; Jensen, K. F. In
(72) Balz, G.; Schiemann, G. Ber. Dtsch. Chem. Ges. 1927, 60, Microfabricated devices for multiphase catalytic process, Proceed-
1186. ings of the 4th International Conference on Microreaction Tech-
(73) Grakauskas, V. Direct liquid-phase fluorination of halo- nology, IMRET 4, Atlanta, USA, 5-9 March, 2000; AIChE: New
genated aromatic compounds. J. Org. Chem. 1969, 34, 2835-2839. York, 2000; pp 416-424.
(74) Purrington, S. T.; Kagen, B. S. The application of elemental (97) Födisch, R.; Reschetilowski, W.; Hönicke, D. Heteroge-
fluorine in organic chemistry. Chem. Rev. 1986, 86, 997-1018. neously catalyzed liquid-phase hydrogenation of nitro-aromatics
(75) Grakauskas, V. Direct liquid-phase fluorination of aromatic using micro channel reactors. In The Future Role of Aromatics in
compounds. J. Org. Chem. 1970, 35, 723-728. Refining and Petrochemistry, Proceedings of the DGMK-Confer-
(76) Cacace, F.; Giacomello, P.; Wolf, A. P. Substrate selectivity ence, Erlangen, Germany, 1999; German Society for Petroleum
and orientation in aromatic substitution by molecular fluorine. J. and Coal Science and Technology: Hamburg, Germany, 1999; pp
Am. Chem. Soc. 1980, 102, 3511-3515. 231-238.
(77) Cacace, F.; Wolf, A. P. Substrate selectivity and orientation (98) Losey, M. W.; Schmidt, M. A.; Jensen, K. F. A micro
in aromatic substitution by molecular fluorine. J. Am. Chem. Soc. packed-bed reactor for chemical synthesis. Springer-Verlag: Berlin,
1978, 100, 3639-3641. 2000; pp 277-286.
(78) Conte, L.; Gambaretto, G. P.; Napoli, M.; Fraccaro, C.; (99) de Bellefon; Tanchoux, N.; Caravieilhes, S.; Grenouillet,
Legnaro, E. Liquid-phase fluorination of aromatic compounds by P.; Hessel, V. Microreactors for dynamic high throughput screening
elemental fluorine. J. Fluorine Chem. 1995, 70, 175-179. of fluid-liquid molecular catalysis. Angew. Chem. 2000, 112,
(79) Löb, P.; Löwe, H.; Hessel, V. Fluorinations, chlorinations 3584-3587.
and brominations of organic compounds in microstructured reac- (100) de Bellefon, C.; Pestre, N.; Lamouille, T.; Grenouillet, P.
tors. J. Fluorine Chem. 2004, 125, 1677-1694. High-throughput kinetic investigations of asymmetric hydrogena-
(80) Hessel, V.; Löb, P.; Löwe, H. Direct fluorination of aromat- tions with microdevices. Adv. Synth. Catal. 2003, 345, 190-193.
ics with elemental fluorine in microstructured reactors. Chem. (101) de Bellefon, C.; Caravieilhes, S.; Grenouillet, P. Applica-
Today 2004, No. 5, 10-15. tion of a micro mixer for the high throughput screening of fluid-
(81) Chambers, R. D.; Sandford, G. Durham microreactors for liquid molecular catalysts. Springer-Verlag: Berlin, 2001; pp 408-
direct fluorination. Chem. Today 2004, No. 6, 6-8. 413.
(82) Chambers, R. D.; Spink, R. C. H. Microreactors for (102) de Bellefon, C.; Abdallah, R.; Lamouille, T.; Pestre, N.;
elemental fluorine. Chem. Comm. 1999, No. 10, 883-884. Caravieilhes, S.; Grenouillet, P. High-throughput screening of
(83) Chambers, R. D.; Holling, D.; Spink, R. C. H.; Sandford, molecular catalysts using automated liquid handling, injection and
G. Elemental fluorine Part 13. Gas/liquid thin film reactors for microdevices. Chimia 2002, 56, 621-626.
selective direct fluorination. Lab Chip 2001, 1, 132-137.
(103) Khan, S. A.; Günther, A.; Schmidt, M. A.; Jensen, K. F.
(84) IMM Institut für Mikrotechnik Mainz GmbH. unpublished.
Microfluidic synthesis of colloidal silica. Langmuir 2004, 20, 8604-
(85) Autorenkollektiv. Organikum, 16th ed.; VEB Deutscher
8611.
Verlag für Grundstoffindustrie: Berlin, 1986; pp 530-563.
(86) Chambers, R. D.; Holling, D.; Sandford, G.; Batsanov, A. (104) Klemm, E.; Albrecht, J.; Lange de Oliveira, A.; Markowz,
S.; Howard, J. A. K. Elemental fluorine Part 15. Selective direct G.; Gross, S.; Schütte, R.; Ehrlich, J.; Schirrmeister, S.; von
fluorination of quinoline derivatives. J. Fluorine Chem. 2004, 125, Morstein, O. Method for obtaining a gaseous phase from a liquid
661-671. medium and device for carrying out the same, WO2004/036137,
(87) Chambers, R. D.; Holling, D.; Sandford, G.; Puschmann, 29.04.2004.
H.; Howard, J. A. K. Selective direct fluorination of quinoline
derivatives. J.Fluorine Chem. 2002, 117, 99-101. Received for review March 4, 2005
(88) Chambers, R. D.; Holling, D.; Rees, A. J.; Sandford, G. Revised manuscript received August 10, 2005
Microreactors for oxidations using fluorine. J. Fluorine Chem. Accepted August 16, 2005
2003, 119, 81-82.
(89) Ehrich, H.; Linke, D.; Morgenschweis, K.; Baerns, M.;
Jähnisch, K. Application of microstructured reactor technology for IE0503139