A falling-film microreactor for enzymatic oxidation of glucose

Article abstract of DOI:10.1002/cctc.201400028

Many oxidation processes require the presence of molecular oxygen in the reaction media. Reactors are needed that provide favorable conditions for the mass transfer between the gas and the liquid phase. In this study, two recent key technologies, microreactor technology and biotechnology, were combined to present an interesting alternative to conventional methods and open up excellent possibilities to intensify chemical processes in the field of fine chemicals. An enzyme-catalyzed gas/liquid phase reaction in a falling-film microreactor (FFMR) was examined for the first time. The test reaction was the oxidation of β-D-glucose to gluconic acid catalyzed by glucose oxidase (GOx). Various factors influencing the biotransformation, such as oxygen supply, temperature, enzyme concentration, and reaction time were investigated and compared to those in conventional batch systems. The most critical factor, the volumetric mass-transfer coefficient for the efficient use of oxygen-dependent enzymes, was determined by using the integrated online detection of dissolved oxygen in all systems. The extremely large surface-to-volume ratio of the FFMR facilitated the contact between the enzyme solution and the gaseous substrate. Hence, in a continuous bubble-free FFMR system with a residence time of 25 seconds, a final conversion of up to 50 % in enzymatic oxidation was reached, whereas conversion in a conventional bubble column resulted in only 27 %. Finally, an option for scale-up was shown through an enlarged version of the FFMR. Oxygen intake: An enzyme-catalyzed gas/liquid phase reaction in a falling film microreactor (FFMR) was examined for the first time. The test reaction was the oxidation of β-D-glucose to gluconic acid catalyzed by glucose oxidase. As a result of to the large surface-to-volume ratio, extremely high oxygen transfer rates are achieved.

Full text of DOI:10.1002/cctc.201400028

CHEMCATCHEM
FULL PAPERS
DOI: 10.1002/cctc.201400028
A Falling-Film Microreactor for Enzymatic Oxidation of
Glucose
[a]
[b]
[b]
[a]
Sabine Illner, Christian Hofmann, Patrick Lçb, and Udo Kragl*
Many oxidation processes require the presence of molecular
oxygen in the reaction media. Reactors are needed that pro-
vide favorable conditions for the mass transfer between the
gas and the liquid phase. In this study, two recent key technol-
ogies, microreactor technology and biotechnology, were com-
bined to present an interesting alternative to conventional
methods and open up excellent possibilities to intensify chemi-
cal processes in the field of fine chemicals. An enzyme-cata-
lyzed gas/liquid phase reaction in a falling-film microreactor
tration, and reaction time were investigated and compared to
those in conventional batch systems. The most critical factor,
the volumetric mass-transfer coefficient for the efficient use of
oxygen-dependent enzymes, was determined by using the in-
tegrated online detection of dissolved oxygen in all systems.
The extremely large surface-to-volume ratio of the FFMR facili-
tated the contact between the enzyme solution and the gas-
eous substrate. Hence, in a continuous bubble-free FFMR
system with a residence time of 25 seconds, a final conversion
of up to 50% in enzymatic oxidation was reached, whereas
conversion in a conventional bubble column resulted in only
27%. Finally, an option for scale-up was shown through an en-
larged version of the FFMR.
(FFMR) was examined for the first time. The test reaction was
the oxidation of b-d-glucose to gluconic acid catalyzed by glu-
cose oxidase (GOx). Various factors influencing the biotransfor-
mation, such as oxygen supply, temperature, enzyme concen-
Introduction
[
4–6]
High yields, safe operation conditions, small environmental im-
pacts and costs, and fast and flexible process development
characterize efficient production processes for chemical sub-
stances. Continuous processes using microstructured devices
ter treatment.
Many processes have been developed and
improved significantly regarding oxygen entry through mem-
[7]
[8,9]
[10]
brane technology, thin silicon tubes,
or cyclone reactors
for aeration. Dispersive methods potentially inactivate the en-
zymes by the complex effects of hydrodynamic shear stress
[
1]
can fulfill a certain number of these demands. Particularly in
comparison with conventional batch systems, microreactors
are characterized by shorter mixing times, an improved heat
and mass transfer, precise residence times, and a small liquid
hold-up. Microreactors were built to realize difficult or unfeasi-
ble-to-control procedures of highly exothermic or fast reac-
tions with toxic compounds for which the reaction volume has
[11]
and the gas/liquid interface. Bubble-free oxygenation in an
FFMR, developed by IMM (now Fraunhofer ICT–IMM, Germany),
provides a solution to prevent the inactivation of enzymes.
The use of an FFMR with a large specific phase–boundary sur-
face for bubble-free oxygenation has not yet been tested for
oxygen-dependent biocatalysts and especially for oxidoreduc-
tases so far. The FFMR has, however, found several successful
applications in the field of fast exothermic gas/liquid reac-
[
2,3]
to be as low as possible.
We found that the advantages of
microscale reaction technologies for biotransformation are sim-
ilar to those in chemocatalysis. This is mainly the very large
surface to volume ratio in, e.g., a falling-film microreactor
[12–17]
tions.
Generally, for enzyme catalysis in microreactors, sev-
eral factors, such as the complete solubility of all reaction com-
ponents and high enzyme activity and stability, must be taken
into account. The major challenge is the immobilization of
active cells or enzymes in or on microchannels. There are
a number of examples in the field of analytical biochemistry,
e.g., microsensors with immobilized enzymes or lab-on-a-chip
(FFMR). Uniform bubble-free gas entry with high oxygen uti-
lization coefficients can resolve the well-known problem of
gas-restricted biocatalysis at higher concentrations. In particu-
lar, oxygen is generally known as a limiting factor for numer-
ous oxidation processes, aerobic fermentations, and wastewa-
[18–20]
biosystems for the detection of glucose in blood or wine.
However, the adaptation of microreactors for biocatalytic reac-
tions in industrial processes is not yet visible. A key reason for
the lack of implementation is the fact that both key technolo-
gies are still very young and must overcome their own difficul-
ties. Some recent articles, such as that of Matsuura et al., have
reported enzyme-encapsulated microreactors for efficient syn-
[
a] S. Illner, Prof. Dr. U. Kragl
Department of Chemistry
University of Rostock
Albert Einstein Str. 3a, 18059 Rostock (Germany)
Fax: (+49)381-498-6452
E-mail: udo.kragl@uni-rostock.de
[b] C. Hofmann, Dr. P. Lçb
[21]
thesis. Zhang et al. used an open tubular microreactor with
Continuous Chemical Engineering Department
Fraunhofer ICT-IMM
Carl-Zeiss-Strasse 18–20, 55129 Mainz (Germany)
an enzyme-functionalized microfluidic channel for the ampero-
[22]
metric detection of glucose. The linear range for the detec-
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1748 – 1754 1748

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
ꢀ
1
tion of glucose was 0.05–7.5 mmolL with a detection limit of
In Figure 1, the setup used to operate the FFMR is shown;
ꢀ
1
[12,25]
2
3 mmolL . In most cases, researchers worked with very low
details can be found in the literature.
The FFMR has a low
concentrations because high activities in closed channels are
still not attainable by covalent immobilization. Hence, micro-
channels were packed with encapsulated enzymes and called
liquid hold-up at open channels in the front and an efficient
heat exchanger in the back part and is combined with online
measurement technology for oxygen control and mass trans-
fer. In further experiments, this standard FFMR-S was replaced
by a large FFMR-L to test the scale-up conditions. To evaluate
this novel microreactor system, we compared this system with
conventional batch systems. A theoretical evaluation of contin-
uous-flow microreactors versus conventional processing op-
tions for glucose oxidation was described by Dencic et al. in
[
21,23,24]
“
packed-bed micro-bioreactors.”
In this paper, we dem-
onstrate that a continuously operated microreactor system can
be used for oxidative enzyme catalysis even without immobili-
zation. For the appropriate test reaction, the biocatalyst should
have high activity and a moderate price. All reaction compo-
nents should be soluble in high concentrations (up to
ꢀ
1
[26]
2
00 mmolL ) making the oxygen transfer the limiting factor.
All these points are united in the glucose oxidase (GOx, EC
.1.3.4) catalyzed conversion of b-d-glucose under oxygen con-
sumption (Scheme 1).
2012. With respect to these calculations, it must nevertheless
be noted that the assumption of an enzyme immobilization in
an FFMR is aggravated by the fact that the residence time is in
the range of seconds, and for high yields, the enzyme activity
in every open channel has to be
1
more than 100 units for a prepa-
rative scale. There is ongoing
work to develop such systems
by surface modification of the
reactor parts directing the liquid
flow. For demonstrating the fea-
sibility of the system, the soluble
enzyme was used. Another ap-
proach would be to use an ultra-
Scheme 1. High-specific GOx-catalyzed reaction of glucose to gluconic acid under oxygen consumption for cofac-
tor recycling.
filtration membrane to retain the
[5]
enzyme in the reactor.
GOx, a flavoprotein from Aspergillus niger, can oxidize b-d- Results and Discussion
glucose to d-glucono-d-lacton, which spontaneously hydrolyz-
Oxygen mass transfer characteristics of the compared
systems
es to gluconic acid (pK =3.76). The prostetic group flavin ade-
a
nine dinucleotide (FAD) reacts thereby to FADH . Molecular
2
oxygen, as a hydrogen acceptor, ensures FAD recycling under
the formation of hydrogen peroxide (Scheme 1). The consump-
tion of dissolved oxygen or the conversion of glucose is cou-
pled stoichiometrically with the production of gluconic acid.
Clearly the limiting factor is the low oxygen solubility of
Oxygen is passed from air with the entry essentially deter-
mined by the phase boundary, temperature, pressure, turbu-
lence, and air movement. First, the determination of the volu-
metric mass-transfer coefficient (k a) in the compared reactors
L
is crucial for evaluating aeration efficiency and quantifying the
effects of the operating variables. The increase of the oxygen
ꢀ
1
ꢀ1
8.11 mgL (258C, 1013 hPa) or 0.25 mmolL in water and,
therefore, requiring efficient methods for oxygen supply in the
reaction mixture.
concentration in the liquid c over time can be described as:
L
dcO2
¼
kL aðcꢁ ꢀ c Þ
ð1Þ
O
2
O2
dt
in which (c*ꢀc) defines the driving force and k a the volumet-
L
ric mass transfer coefficient.
As shown in Figure 2, the logarithmic plot of oxygen con-
centration versus time forms a straight line determining the
kLa in conventional batch systems. The dissolved oxygen con-
centration profile was measured three times for each set of op-
erating conditions after the liquid in the reactor was deoxy-
genated by argon sparging and the stirrer or air inflow was
2
ꢀ3
started. With a surface-to-volume ratio of 28 m m and a volu-
ꢀ
1
metric mass transfer coefficient of 1.13ꢂ0.004 h for oxygen,
the fast oxidation reaction in the stirred beaker (200 rpm) will
be highly limited. For the bubble column (BC), a large range
Figure 1. Flow scheme of continuously operated FFMR.
for the k a value can be found in the literature because of the
L
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1748 – 1754 1749

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
sens, Germany) for different gas volume flows. Weuster-Botz
ꢀ
1
et al. reported k a values of up to 576 h for a superficial air
L
ꢀ
1
[28]
velocity of ꢃ0.6 cms in small-scale BCs. BCs have, in gen-
eral, higher volumetric oxygen-transfer coefficients (250 to
ꢀ
1
ꢀ1 [29]
>
500 h ) than shaken vessels (ꢄ140 h ). With an air veloc-
ꢀ
1
ꢀ1
ity of 0.24 cms , we found a k a value of 234 h in the BC.
L
However, the k a values of conventional batch systems can in
L
no way be compared to those achievable in the FFMR system.
2
ꢀ3[30]
The surface-to-volume ratio of up to 20000 m m
is several
orders of magnitude larger than in the beaker and much
higher than in the BC. In the FFMR an oxygen saturation of
ꢀ
1
8
.9 mgL is reachable in less than six seconds, which was
chosen as the residence time for the liquid. Under these condi-
ꢀ
1
ꢀ1
tions the k a value was estimated to be 20590 h (6 s ). This
L
is in good agreement with previously published data. Yeong
ꢀ
1
et al. reported k a values from 3 to 8 s corresponding to
L
ꢀ
1
[31]
1
0800 to 28000 h depending on liquid flow. Owing to the
setup (pumps, diameter of tubing), shorter residence times
were not possible in our case.
The huge volumetric oxygen mass transfer coefficient results
from the large surface-to-volume ratio and the almost exclu-
sively laminar boundary layer at the grooves, guaranteeing un-
surpassable bubble-free oxygen entry (Table 1). The initial in-
crease in the measured values was only attributable to the
start-up response of the FFMR and the oxygen flow-through
cells.
Bioconversion experiments
The glucose oxidation catalyzed by GOx was investigated in an
FFMR and the two batch systems at reaction times of approxi-
mately 5–37 s or 7200 s (2 h), respectively, at 258C. In the
FFMR, different residence times were obtained by changing
the flow rate, with lower flow rates giving longer residence
times and vice versa (see Figure 6A). In batch mode, samples
were taken at predetermined intervals. The values of glucose
conversion versus time were measured by using HPLC. First, re-
garding the batch systems, the conversion of glucose to glu-
conic acid is compared in Figure 3 after a reaction time of
Figure 2. Determination and comparison of the volumetric mass-transfer co-
120 min using several enzyme concentrations.
efficient for oxygen without enzyme using A) a beaker, stirred at 200 rpm;
ꢀ
1
The conventional systems reveal different trends of product
B) BC, air flow rate: 100 mLmin ; C) FFMR-S, 64-channel reaction plate (tem-
ꢀ1
concentration versus time depending on the enzyme concen-
tration used. Despite the same high enzyme amounts, the con-
version in the beaker is strongly limited. Immediately after the
addition of GOx, the oxygen content decreases within less
pered at 258C), liquid flow: 1 mLmin , compressed air counterflow:
ꢀ
1
ꢀ1
1
2
00 mLmin . Conditions: c*=8.11 mgL (258C, 1013 mbar), volume of
1
ꢀ
5 mL previously degassed with argon, 0.1 molL glucose in phosphate
buffer (pH 7) at 258C.
differences in gas-distributor designs, liquid proper-
ties, and column dimensions. Garcia-Ochoa and
Gomez provided a comprehensive overview of the
Table 1. A comparison of oxygen supply in the beaker, the bubble column, and the
FFMR.
[
27]
[
a]
oxygen transfer rate in various bioreactors. Accu-
rate bubble-size measurements for theoretical calcu-
lations are difficult to obtain. The intensity of the tur-
bulence and shearing fields determine the size of the
bubbles and thereby the size of the boundary surface
of the phase. Thus, for the experimental determina-
Reactor
type
Gas flow rate
Superficial air velocity
kLa
[h
OTRmax
ꢀ
1
ꢀ1
ꢀ1
ꢀ1 ꢀ1
[mLmin
]
[cms
]
]
[gO L h ]
2
Beaker
0
15
0
00
0
1.13ꢂ0.004
38.8ꢂ0.4
ꢄ0.01
0.32
1.13
1.93
170
0.035
0.118
0.236
0.012
Bubble
column
5
1
137.3ꢂ2.2
233.7ꢂ6.2
20590ꢂ500
[b]
FFMR
100
tion of k a, we measured the dissolved oxygen con-
L
ꢀ
1
[
2
a] OTR: Oxygen transfer rate; c(O )*=8.11 mgL . [b] 64 channels.
tent over time with optical oxygen sensor spots (Pre-
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1748 – 1754 1750

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Figure 3. Comparison of conversions in the batch mode after 120 min.
ꢀ1
than a minute to 0 gL . Complete conversion was reached
after approximately 20 h using an enzyme activity of
ꢀ
1
5
0 UmL . In the BC, glucose was completely converted to glu-
conic acid within approximately 2 h by using high enzyme ac-
ꢀ
1
tivities (200 Uml ) and without the formation of byproducts
as shown by HPLC analysis. However, in bubble column experi-
ments the formation of foam was observed caused by protein
denaturation at the gas/liquid interface. This foam formation
can be reduced by reducing the gas flow, but, at the same
time, reducing the reaction velocity as a result of the limiting
oxygen supply (Figure 4A). Despite the reduced gas flow, the
foam forming is not completely prevented. With the native
enzyme, this is a great disadvantage of an otherwise successful
batch system regarding the high volumetric oxygen mass
transfer coefficients and low system costs. In Figure 4B, the
conversion is shown during the first minutes of the reaction
under variation of the enzyme concentration. The results clear-
ly show that high concentrations of GOx are necessary to com-
pare the beaker and BC with the falling film microreactor.
Owing to the design developed for fast and exothermic chemi-
cal reactions, only short residence times of less than one
minute are possible in the FFMR. Under initial rate conditions
without any limitations, the full conversion of 2.5 mmol glu-
cose should be achieved in approximately 30 s with an
Figure 4. Conversion of glucose to gluconic acid as a function of reaction
ꢀ1
time at 258C using A) different gas flows, conditions: 0.1 molL glucose in
ꢀ
1
phosphate buffer (pH 7), 25 UmL GOx (2.96 mg), 25 mL, beaker: 200 rpm;
ꢀ1
B) various GOx concentrations (UmL ) under otherwise the same conditions
ꢀ
1
and 100 mLmin gas flow in BC.
ample, the reaction times to provide the same conversion
differ between hours in the beaker and seconds in the FFMR.
Hence, improvements in the continued supply of oxygen are
better reflected in the volumetric productivity or space–time
yield (STY) calculated by using Equation (2):
m C S c0 M C S
0
ꢀ
1
STY ¼
¼
ð2Þ
enzyme concentration of 200 UmL . After 24 h, complete con-
version in all setups with high enzyme concentrations was
achieved.
V t
t
R
in which m stands for the amount of glucose, C for conver-
0
The low effective initial activities determined from the slope
of the linear part of the gluconic acid production curve can be
sion, S for selectivity, t for residence time, M for the molecular
mass of glucose, and V for the reactor volume or volume of
R
correlated with the determined k a values for oxygen. These
L
the liquid film.
investigations in conventional batch systems still offer consid-
erable scope for optimization and impressively confirmed that
the oxygenation limits the reaction rate at high substrate
concentrations.
The necessary periods for emptying, cleaning, filling, and
maintenance of the batch systems were not considered, but
the advantage of the continuous system is obvious. However,
it also has to be considered that a comparison of batch and
continuous systems is difficult owing to huge differences in re-
actor volume and residence time. The liquid film thickness
amounts to 25–100 mm, and the film volume was calculated
To overcome the oxygen limitation, we used a continuously
operating FFMR system. This reactor guarantees permanent
oxygen saturation in the reaction solution even at high reac-
tion rates resulting from high enzyme concentrations. The very
short residence times of only a few seconds limits the achieva-
ble conversion. The results obtained in the continuously oper-
ated FFMR cannot be compared directly through the plots in
Figure 4 with the reaction times in the batch systems. For ex-
over the liquid flow rate F multiplied by the residence time t
L
using Equation (3):
V ¼ F t
ð3Þ
F
L
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1748 – 1754 1751

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Figure 5. Comparison of STY in different reaction systems at 30% conver-
ꢀ1
sion. Conditions: 0.1 molL glucose, phosphate buffer (pH 7), 25 mL, FFMR/
ꢀ
1
BC: compressed air cross-flow: 100 mLmin , FFMR 32 channels, beaker:
200 rpm.
The significantly higher STYs in the microreactor (780 and
ꢀ
1
ꢀ1
ꢀ1 ꢀ1
1
300 gh L ) than in the beaker (2.3 gh L ) and the bubble
ꢀ1
ꢀ1
column (117 gh L ) are highlighted in Figure 5, reflecting
the improvements in oxygen supply and shortened reaction
time. This result corresponds to an increase in STY by a factor
of 300 in the microreactor relative to the beaker system allow-
ing more efficient use of the enzyme activity.
ꢀ
1
The reaction time for a conversion of 30% (5.4 gL glucose)
differs from 142 min in the beaker or 3 min in the BC to some
seconds in the FFMR. Thus, we have successfully demonstrated
that a continuous-flow microreactor is a promising tool for the
development of efficient enzyme reaction systems requiring an
effective supply of a gaseous substrate. A defined and con-
stant yield between 20 and 40% can be reached within sec-
onds depending on the chosen enzyme concentration and the
residence time on the microstructured reaction plates (Fig-
ure 6B). By reducing the flow rate or the reactor angle of incli-
nation from 908 to 458 and varying the channel geometry and
number on the reaction plate from 32 channels with 600 mmꢁ
Figure 6. Comparison of FFMR-S and FFMR-L: A) range of varying residence
times in FFMR-S (32 channels) and FFMR-L (50 channels) with an inset graph
for FFMR-S (64 channels); conditions: averaged residence time of 10-fold
measurement with isopropyl alcohol as the liquid phase; B) yield with re-
spect to gluconic acid concentration at high enzyme concentrations at dif-
ferent flow rates in standard FFMR-S with 32 (squares, blue-shaded area)
and 64 (circles, gray-shaded area) channels and large FFMR-L with 50 chan-
ꢀ1
nels (triangles, yellow-shaded area); conditions: 0.1 moll glucose, phos-
ꢀ1
phate buffer (pH 7); compressed air cross-flow: 100 mLmin ; note: a product
ꢀ1
concentration of 50 mmolL complies to a yield of 50%.
Scale-up from FFMR-S to FFMR-L
200 mm cross-section to 64 channels with 300 mmꢁ100 mm
cross-section, the residence time can be further enhanced, and
a conversion rate of 50% was reached (data not shown). How-
ever, higher or full conversion remains problematic because of
the low volumetric flow necessary. In addition, diffusion pro-
cesses within the liquid phase might become limiting as well.
Also, the process of mutarotation of the glucose can be a fur-
ther restriction. The timescale for reaching the equilibrium be-
tween a- and b-d-glucopyranose (34% and 66%, respectively,
in the equilibrium at 258C) is also in the range of minutes. The
rate constant for a noncatalyzed reaction of the [a]![b] form
To investigate longer residence times and a first step for an up-
scale of the volume, a repetition of the experiments in the
FFMR-L was performed. This reactor bears a larger reaction
plate with a tenfold larger surface than the FFMR-S working
with a cross-section of the 50 channels of 1200 mmꢁ400 mm
(Figure 7). Measurement of the residence-time distribution
yielded the highest residence time of approximately 40 s with
ꢀ
1
a flow rate of 1 mLmin in the FFMR-L. Regarding the strong
broadening in residence time distribution, a lower flow rate is
not recommended. For many biotransformations, the range of
residence times between 1 and 25 seconds in FFMR-S or 7 and
40 s in FFMR-L is far too small. Only with highly active enzymes
and high catalyst concentration, high conversion is possible. In
Figure 6B, the yield is shown if varying the reactor, the channel
geometry, and the residence time. All the FFMR experiments
exhibit unusually small standard deviations through well de-
fined contact times between the liquid and the gas phase. The
yield of gluconic acid depends linearly on the residence time
ꢀ
1
is calculated at 0.0135 min from kinetic measurements fol-
[
32]
lowing a first-order kinetics. GOx converts only the b-d-glu-
copyranose. The prevention of enzyme inactivation at the
liquid/gas interface is another big advantage of bubble-free
oxygenation in an FFMR.
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1748 – 1754 1752

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
12 times faster than that in the bubble column and 588 times
higher than in the beaker using equal amounts of enzyme. Full
conversion as observed in the bubble column is not possible
because of the not yet optimized reactor geometry for longer
residence times. The initial concentration of a-d-glucose lies in
ꢀ
1
the range of 50 to 200 mmolL , proving that the system is
indeed suitable for preparative-scale synthesis. This result
opens the door for the application of continuously operating
microbioreactors for the flexible biochemical processing of fine
chemicals.
Experimental Section
Chemicals and materials
All chemicals, such as isopropyl alcohol, phosphate buffer salts, b-
d-glucose, and d-glucono-d-lacton, were obtained from Merck or
Sigma–Aldrich. Glucose oxidase (EC 1.1.3.4 from A. niger, type X-S,
ꢀ
1
lyophilized powder, 211 Umg ) was purchased from Sigma–Al-
drich. All chemical reagents used in the experiment were of analyt-
ical grade.
Experimental setup and procedure
The biotransformation was performed in a stirred beaker (B),
a bubble column (BC) constructed of glass with sintered glass
bottom as a gas diffuser, and a standard falling-film microreactor
FFMR (IMM, Mainz, Germany). Beaker and BC as conventional sys-
tems were each characterized by a diameter of 3 cm and a reaction
Figure 7. Falling-film microreactor developed by IMM; left: FFMR-S, right:
FFMR-L.
ꢀ
1
ꢀ1
volume of 25 mL. A preparation of 0.1 molL glucose (18 gL ) in
a phosphate buffer solution (pH 7) was converted under varying
up to 50%. Despite longer reaction times in the FFMR-L, the
conversion is also limited to 40% at an initial glucose concen-
tration of 100 mm. The highest yields at an enzyme concentra-
ꢀ1
enzyme concentration (50–400 UmL ). The reaction in the contin-
uous FFMR system occurred along three vertical microstructured
reaction plates (16 parallel open microchannels, widthꢁthickness:
1200ꢁ400 mm; 32 channels, 600ꢁ200 mm; 64 channels, 300ꢁ
100 mm) with a size of 76ꢁ25.6 mm. The liquid film of glucose–
enzyme solution was transported within the grooves by the action
of gravitational forces, and the open form ensured contact with
the gas phase. The two pulsation-free pumps were necessary to
provide a constant liquid-film thickness, typically in the range of
several dozens of micrometers. The liquid hold-up in the FFMR was
varied and adjusted depending on the reaction plate. The volume
before and after the FFMR was kept to a minimum, resulting in
a total volume of approximately 5 mL. In front of the grooves was
a gas-phase chamber. In this chamber, a counter-flow of com-
pressed air or argon was controlled by a pre-pressure regulator
ꢀ
1
ꢀ1
tion of 200 Uml are obtained with a flow rate of 1 mLmin
in the FFMR-L. This confirms the previously described limita-
tion through slow diffusive processes during decreased glu-
cose concentrations or high conversion. The simple transfer of
experiments of FFMR-S to FFMR-L (Figure 7) confirmed an es-
pecially risk-free scale-up, as often described in the litera-
[
12,16,33]
ture.
The conversion in different FFMRs is nearly con-
stant, as the production rate in the reactor increases through
higher capacity and flow rates.
Conclusions
(
(
Parker, Porter 4000, 0–60 psi) and a digital mass-flow controller
Bronkhorst, El-Flow Select F-201CV). For online oxygen detection,
For the first time, an enzyme-catalyzed oxidation in a falling-
film microreactor was investigated. Microstructured devices are
new and flexible tools that show high potential to enabling
chemical reactions without mass- or heat-transport limitations
with additional low risks in process up-scaling. Thus, it is not
necessary to substitute all process steps through microreactor
technology, but only if this represents an advantage for the
whole process; the same applies to biotechnology. In the
FFMR system, significantly higher specific interfacial areas im-
prove the oxygen transfer rate. The space–time yields exceed
those obtained with conventional reactors by orders of magni-
tude. Herein, we describe a system that allows continuous con-
version of the substrate up to 50% in less than 10 seconds of
residence time. The initial reaction rate in the FFMR was
we used optical oxygen sensor spots (B, BC) and flow-through cells
FFMR) from Presens GmbH. Directly before and after the FFMR
(
were two oxygen flow-through cells and two microannular gear
pumps (HNP, mrz-2942) with a particle filter (Swagelok, 7 mm) com-
bined through Swagelok connections in a 1/8“ periphery. Negative
effects of shearing forces on enzyme stability by pumping through
the FFMR system were not observed. Stability experiments (closed-
loop circulation) displayed no loss of enzyme activity over 22 h.
The FFMR was temperature-controlled by a cryostat (Huber, CC 3).
In the BC, the gas flow rate was equally controlled. In the beaker,
a stirring rate of 200 rpm was adjusted without additional gassing.
Samples (0.5 mL) were taken at the outlet of the FFMR system,
with the number of samples based on the liquid flow rate in the
FFMR. All samples were immediately heated to 998C resulting in
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1748 – 1754 1753

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
a complete loss of enzyme activity (Eppendorf, Thermomixer com-
fort, holding time: 10 min, stirrer speed: 300 rpm) and measured
with HPLC.
[1] V. Hessel, I. V. Gursel, Q. Wang, T. Noel, J. Lang, Chem. Eng. Technol.
2
012, 35, 1184.
[
[
2] P. Lçb, H. Lçwe, V. Hessel, J. Fluorine Chem. 2004, 125, 1677.
3] G. Kolb, V. Hessel, V. Cominos, C. Hofmann, H. Lçwe, G. Nikolaidis, R.
Zapf, A. Ziogas, E. R. Delsman, M. H. J. M. de Croon, J. C. Schouten, O. de
La Iglesia, R. Mallada, J. Santamaria, Catal. Today 2007, 120, 2.
4] J. M. Bolivar, T. Consolati, T. Mayr, B. Nidetzky, Biotechnol. Bioeng. 2013,
110, 2086.
Analytical procedure
[
[
The concentration of gluconic acid was determined by HPLC using
an Aminex HPX-87H column (Biorad, 300ꢁ7.8 mm) and a Cation-H
Refill Cartridge (Supleco, 30ꢁ4.6 mm) as pre-column at 658C. The
5] E. U. C¸ okgçr, S. Sozen, G. Insel, D. Orhon, Environ. Technol. 2009, 30,
1169.
ꢀ1
[6] J. L. Casas Lꢂpez, J. A. S. Perez, J. M. F. Sevilla, F. G. A. Fernandez, E. M.
eluent was a 6 mm sulfuric acid with a flow rate of 0.8 mLmin
,
Grima, Y. Chisti, J. Chem. Technol. Biotechnol. 2004, 79, 1119.
coupled with an RI and UV detector in a Knauer HPLC system. The
injection volume was 20 mL, and the retention time of the gluconic
acid was 6.8 min. Gluconic acid was calibrated through its peak
area (l=210 nm), and the initial concentration of glucose was
measured by RI detector.
[
[
7] S. Lꢃtz, N. N. Rao, C. Wandrey, Chem. Eng. Technol. 2006, 29, 1404.
8] S. Rissom, U. Schwarz-Linek, M. Vogel, V. I. Tishkov, U. Kragl, Tetrahedron:
Asymmetry 1997, 8, 2523.
[9] U. Schwarz-Linek, A. Krçdel, F. A. Ludwig, A. Schulze, S. Rissom, U. Kragl,
V. I. Tishkov, M. Vogel, Synthesis 2001, 0947.
[
10] D. Weuster-Botz, E. Hunnekes, A. Hartbrich, Bioprocess Eng. 1998, 18,
433.
Large-FFMR experiments
[11] W. Van Hecke, R. Ludwig, J. Dewulf, M. Auly, T. Messiaen, D. Haltrich, H.
Van Langenhove, Biotechnol. Bioeng. 2009, 102, 122.
Apart from the standard FFMR, a large FFMR was used with a 10-
fold increase in the structured surface area on the reaction plate.
Therefore, the length and the number of channels were increased
by a factor of 10 . In the general design, the FFMR-L was relatively
similar to that of FFMR-S. Prior to the biotransformation, the resi-
dence time distribution in the FFMR-L was measured. All other
steps are identical to the previous FFMR-S procedure.
[12] K. Jꢄhnisch, M. Baerns, V. Hessel, W. Ehrfeld, V. Haverkamp, H. Lçwe, C.
Wille, A. Guber, J. Fluorine Chem. 2000, 105, 117.
[13] K. Jꢄhnisch, U. Dingerdissen, Chem. Eng. Technol. 2005, 28, 426.
0
.5
[
14] N. Steinfeldt, R. Abdallah, U. Dingerdissen, K. Jꢄhnisch, Org. Process Res.
Dev. 2007, 11, 1025.
[
[
15] T. Xie, C. Zeng, C. Wang, L. Zhang, Ind. Eng. Chem. Res. 2013, 52, 3714.
16] B. K. Vankayala, P. Lçb, V. Hessel, G. Menges, C. Hofmann, D. Metzke, U.
Krtschil, H.-J. Kost, Int. J. Chem. React. Eng. 2007, 5, 1.
[
17] O. Shvydkiv, C. Limburg, K. Nolan, M. Oelgemçller, J. Flow Chem. 2012,
2
, 52.
18] T. B. Goriushkina, A. P. Soldatkin, S. V. Dzyadevych, J. Agric. Food Chem.
009, 57, 6528.
19] J. Wang, M. P. Chatrathi, G. E. Collins, Anal. Chim. Acta 2007, 585, 11.
20] S. Wang, P. Su, Y. Yang, Anal. Biochem. 2012, 427, 139.
Measurement of residence time in microreactor systems
[
2
Experimental measurement of the mean residence time (t) was
achieved through a sight window in the upper part of the housing.
Detection through video images was not possible over the whole
microstructured plate through a limited frontal view, therefore,
manual measurements with a stopwatch and additional lighting
covering the complete plate was favored. The flow distribution of
isopropyl alcohol while wetting was approximately representative
of a stationary flow condition. For the statistical evaluation of flow-
front measurements, 10-fold repetition was performed. Theoretical
simulations or mathematical calculations of flow distribution have
[
[
[
21] S. Matsuura, T. Yokoyama, R. Ishii, T. Itoh, E. Tomon, S. Hamakawa, T. Tsu-
noda, F. Mizukami, H. Nanbu, T. Hanaoka, Chem. Commun. 2012, 48,
7058.
[22] L. Zhang, P. Qu, J. Sheng, J. Lei, H. Ju, Chin. J. Chem. 2012, 30, 2145.
[23] A. S. Bhangale, K. L. Beers, R. A. Gross, Macromolecules 2012, 45, 7000.
´
-Fura cˇ , P. Zˇ nidar sˇ i cˇ -Plazl, Process Biochem. 2012,
[24] M. Cvjetko, J. Vorkapic
7, 1344.
4
[
25] M. Zanfir, A. Gavriilidis, C. Wille, V. Hessel, Ind. Eng. Chem. Res. 2005, 44,
1742.
26] I. Dencic, J. Meuldijk, M. de Croon, V. Hessel, J. Flow Chem. 2012, 1, 13.
27] F. Garcia-Ochoa, E. Gomez, Biotechnol. Adv. 2009, 27, 153.
28] D. Weuster-Botz, J. Altenbach-Rehm, A. Hawrylenko, Bioprocess Biosyst.
Eng. 2001, 24, 3.
[15,34]
been described previously in the literature.
[
[
[
Acknowledgements
[
[
[
29] S. D. Doig, K. Ortiz-Ochoa, J. M. Ward, F. Baganz, Biotechnol. Prog. 2005,
Financial support from Fonds der chemischen Industrie (FCI,
Frankfurt, Germany) and Deutsche Bundesstiftung Umwelt (DBU,
Osnabrꢀck, Germany, grant no. AZ 13191) are gratefully acknowl-
edged. The authors want to thank Carsten Damerau, HNP Mikro-
systeme, for competent advice on micro-annular gear pumps
and Peter Kumm for his help during the construction of the con-
tinuous operating microsystem. The Institut fꢀr Mikrotechnik
Mainz GmbH (now: Fraunhofer ICT-IMM) is acknowledged for
providing the large FFMR.
21, 1175.
30] E. V. Rebrov, T. Duisters, P. Lçb, J. Meuldijk, V. Hessel, Ind. Eng. Chem.
Res. 2012, 51, 8719.
31] K. Kin Yeong, A. Gavriilidis, R. Zapf, V. Hessel, Chem. Eng. Sci. 2004, 59,
3
491.
[32] J. Pazourek, J. Sep. Sci. 2010, 33, 974.
[
[
33] R. D. Chambers, R. C. H. Spink, Chem. Commun. 1999, 883.
34] J. M. Commenge, T. Obein, X. Framboisier, S. Rode, P. Pitiot, M. Matlosz,
Chem. Eng. Sci. 2011, 66, 1212.
Keywords: enzyme catalysis · continuous processes · gluconic
acid · microreactor · oxygen transfer
Received: January 12, 2014
Published online on May 21, 2014
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1748 – 1754 1754

Copyright of ChemCatChem is the property of Wiley-Blackwell and its content may not be
copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for
individual use.