High-Throughput reaction optimisation and activity screening of ferrocene-based lewis acid-catalyst complexes by using continuous-flow reaction detection mass spectrometry

Article abstract of DOI:10.1002/chem.200900317

Optimising synthetic conversions and assessing catalyst performance is a tedious and laborious endeavour. Herein, we present an automated alternative to the commonly applied sequential approaches that are used to increase catalyst discovery process efficiencies by increasing the number of entities that can be tested. This new approach combines conversion of the reactants and determination of product formation into a single comprehensive reaction detection system that can be operated with minimal catalyst and reactant consumption. With this approach, rudimentary reaction conditions can be quickly optimised and the same system can then be used to screen for the optimal homogenous catalyst in a selected solution-phase synthetic conversion. The system, which is composed of standard HPLC components, can be used to screen catalyst libraries at a repetition rate of five minutes and can be run unsupervised. The sensitive mass spectrometric detection that is implemented in the reaction detection methodology can be used for the simultaneous monitoring of reactants, catalysts and product ions. In the experiments, the three-component reaction that gives a substituted 2imidazoline was optimised. Afterwards, the same method was used to assess a library of ferrocene-based Lewis acid catalysts for performance in the aforementioned conversion in six different solvents. We demonstrate the feasibility of using this methodology to directly compare the performance results obtained in different solvents by calibrating the solvent-specific MS responses.

Full text of DOI:10.1002/chem.200900317

DOI: 10.1002/chem.200900317
High-Throughput Reaction Optimisation and Activity Screening of
Ferrocene-Based Lewis Acid–Catalyst Complexes by Using Continuous-
Reaction Detection Mass Spectrometry
ACHTUNGTRENNUNGF low
[
a]
[a]
[a]
[b]
Cornelius T. Martha,* Anton Heemskerk, Jan-Carel Hoogendoorn, Niels Elders,
[
a]
[b]
[a]
Wilfried M. A. Niessen, Romano V. A. Orru, and Hubertus Irth*
Abstract: Optimising synthetic conver-
sions and assessing catalyst perfor-
mance is a tedious and laborious en-
deavour. Herein, we present an auto-
mated alternative to the commonly ap-
plied sequential approaches that are
used to increase catalyst discovery pro-
cess efficiencies by increasing the
number of entities that can be tested.
This new approach combines conver-
sion of the reactants and determination
of product formation into a single com-
prehensive reaction detection system
that can be operated with minimal cat-
alyst and reactant consumption. With
this approach, rudimentary reaction
conditions can be quickly optimised
and the same system can then be used
to screen for the optimal homogenous
catalyst in a selected solution-phase
synthetic conversion. The system,
which is composed of standard HPLC
components, can be used to screen cat-
alyst libraries at a repetition rate of
five minutes and can be run unsuper-
vised. The sensitive mass spectrometric
detection that is implemented in the re-
action detection methodology can be
used for the simultaneous monitoring
of reactants, catalysts and product ions.
In the experiments, the three-compo-
nent reaction that gives a substituted 2-
imidazoline was optimised. Afterwards,
the same method was used to assess a
library of ferrocene-based Lewis acid
catalysts for performance in the afore-
mentioned conversion in six different
solvents. We demonstrate the feasibility
of using this methodology to directly
compare the performance results ob-
tained in different solvents by calibrat-
Keywords: high-throughput screen-
ing · homogeneous catalysis · mass
spectrometry
·
multicomponent
reactions · reaction optimization
ing the solvent-specific MS ACHUTNGRNEGUNr esponses.
Introduction
costs. Nowadays, industrial progress includes the need to de-
velop sustainable processes in which environmental impact
is not amplified when economic growth is achieved. Of all
the approaches that are used to decrease the environmental
footprint, catalysis is one of the most important tools due to
its principal role in industrial chemical synthesis.
Innovation in chemical industrial processes is a perpetual
process driven by an incessant need to reduce the consump-
tion of resources and energy and the production of waste in
an everlasting attempt to be more competitive by reducing
With industry demands to lower development costs, cata-
lyst discovery process efficiencies have to be increased. A
possible strategy is to replace conventional sequential syn-
thesis and activity assessment methods by automated high-
throughput experimental methodologies. This combinatorial
approach requires the development of specific instruments
and robotics to cope with the demanded throughput that
[
a] C. T. Martha, A. Heemskerk, J.-C. Hoogendoorn,
Prof. Dr. W. M. A. Niessen, Prof. Dr. H. Irth
Department of Analytical Chemistry and Applied Spectroscopy
Faculty of Sciences, VU University Amsterdam
De Boelelaan 1083, 1081 HV Amsterdam (The Netherlands)
Fax : (+31)205987542
[1]
E-mail: ctmartha@few.vu.nl
originates from the large number of samples. Spectrosco-
py-based techniques for the detection of catalytic activity in
high-throughput screening assays are still the most ubiqui-
tous methods in this field despite the restricted applicability
caused by the prerequisite of chromophores and fluoro-
irth@few.vu.nl
[
b] N. Elders, Prof. Dr. R. V. A. Orru
Department of Organic and Inorganic Chemistry
Faculty of Sciences, VU University Amsterdam
De Boelelaan 1083, 1081 HV Amsterdam (The Netherlands)
[2]
phores in either substrate or product for detection. In this
perspective, mass spectrometry (MS) offers an outstandingly
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200900317.
7368
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 7368 – 7375

FULL PAPER
wide range of applicability for the sensitive detection of
4) the determination of product formation at associated pa-
rameter settings to optimise product formation. Once the
optimal settings have been determined, a catalyst perfor-
mance screening can be used to identify the best-performing
catalyst for the selected synthetic conversion.
[3]
(
multiple) selected target analytes in complex samples. Al-
though quantification can be affected by ion suppression
and there are compounds that are not ionisable, the multi-
tude of gas- and solution-phase mechanisms that convert
neutral molecules into ionic species (i.e., protonation, metal-
ation, chemical reactions, quaternisation and oxidation/re-
duction) results in the eminent potential of MS for the iden-
In commonly applied catalyst performance screening
methods, catalyst performance is rated based on kinetics
that are constructed from the increased slope of the initial
rate of the reaction. To obtain this data in traditionally ap-
plied methodologies, the conversion of reactants into prod-
ucts has to be determined at different reaction times. In the
continuous-flow reaction detection methodology, an in-
creased throughput is accomplished by a reduction in the
number of data points that are obtained. In the reaction de-
tection approach, the conversion into product takes place in
a thermostated reactor with a fixed internal volume. There-
fore, the catalyst performance is assessed by determining
the peak area from the selected ion monitoring (SIM) prod-
uct trace in atmospheric-pressure chemical ionisation
(APCI) MS detection. The optimal continuous-flow reactor
volume in this system associates with a reaction time that
only slightly exceeds the time period required for the reac-
tion to reach its initial rate. For this methodology to be ap-
plicable, it should allow for the sensitive quantification of
the product with high accuracy. In the current continuous-
flow reaction detection setup, this is accomplished with
system parameters, that is, sampling time, reaction time, re-
action pressure and reaction temperature, that are either
fixed or accurately controlled and result in high system re-
peatability and reproducibility.
[4]
tification of catalyst complexes.
Although several combinatorial tools have been devel-
oped for the catalysis toolbox, their development into high-
throughput reaction optimisation strategies is relatively un-
tapped despite the fact that attractive new analytical plat-
[5]
forms have been introduced. For reaction optimisation in
material research, research efforts are mainly focused on the
development of new in silico design, improved reactor plat-
forms and analysis tools.
Commonly, the optimisation of reaction parameters is per-
formed by sequential unit operations, such as transfer of liq-
uids, quenching of the reaction and storage of the samples
prior to analysis. In the most advanced unit-operation-based
methodologies, automated workstations and robotics are
[6]
used to complete the sequential steps.
Recently, a continuous-flow reaction detection screening
system was introduced that was used for the determination
[7]
of catalyst performance in synthetic conversions. The feasi-
bility of using this approach for the assessment of catalyst
activity towards a selected synthetic conversion was demon-
strated by the performance screening of a selected number
of homogeneous Lewis acid catalysts for a synthetic conver-
sion in tetrahydrofuran (THF). The results proved to be
comparable to the results obtained with a traditional bench-
scale experiment. Moreover, the mass spectrometric detec-
tion proved to be very sensitive, with an achieved detection
limit of as low as 1.6 nm for the formed product, while the
ruggedness of the method was illustrated with a 24-hour
screening that was completed with a peak-area variation of
within 7% relative standard deviation.
The optimisation of a certain reaction is usually per-
formed by relating product formation to different reaction
parameter settings. Herein, we present an integrated analyti-
cal reaction detection methodology based on standard
HPLC equipment (see Figure 1). An advantage of this con-
tinuous-flow total-analysis system is that all the intermediate
liquid transfer steps are made superfluous and the necessity
of applying robotics to achieve a high sample throughput is
circumvented. In a sequential unit operation system, product
formation is determined at different reaction settings. We
Herein, the applicability of the novel continuous-flow re-
action detection methodology is explored. Initially, the feasi-
bility of using this reaction detection methodology for the
high-throughput reaction optimisation of a Lewis acid-cata-
lysed multicomponent synthesis of substituted 2H-2-imida-
zolines was investigated. Secondly, by utilising the same
comprehensive system, the catalyst–ligand activity of differ-
ent ferrocene-based catalyst complexes in this atom-efficient
multicomponent reaction (MCR) in six different commonly
used solvents was assessed.
Results and Discussion
Figure 1. Schematic overview of the continuous-flow reaction detection
system. This system is comprised of two HPLC pumps labelled react-
In general, the workflow of a catalysed reaction optimisa-
tion strategy includes 1) the selection of the synthetic con-
version, 2) a preliminary selection of potential catalysts,
ACHTUNGERTNNUNGa nt A and reactant B, which pump the reactants (3 and 4, respectively,
see Scheme 1) into a mixer. An injector is used to introduce the catalysts
into the stream of reactants. The conversion of reactants into product
takes place in the thermostated coiled open tubular reactor. A third
HPLC pump (make-up) can be used to add water to quench the reaction
and increase APCI-MS detection compatibility.
3) the optimisation of the synthetic conversion of reactants
by selected variation of involved system parameters and
Chem. Eur. J. 2009, 15, 7368 – 7375
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
7369

H. Irth, C. T. Martha et al.
demonstrate that accurately controlled system parameters
can be used for the optimisation of reaction parameters.
After reaction optimisation, the flexibility of the system
was demonstrated by employing the same mass spectromet-
ric detection system for the screening of catalyst libraries for
activity towards a selected synthesis. Minute amounts of re-
actants and catalysts were used, without the necessity of al-
tering any system parameters.
Lewis acid catalyst selection: Lewis acids have been report-
ed to be excellent catalysts in the MCR of a substituted 2H-
[8]
2
-imidazoline of type 5 (Scheme 1). In this MCR, an
Figure 2. Integrated peak areas of the Lewis acid catalyst screening ob-
tained with the continuous-flow reaction detection system, showing the
activity of eight different metal-based triflates in the synthesis of 5.
Reaction optimisation: In the reaction optimisation experi-
ment described hereafter, the concentration of the reactants
was fixed at 4 mm. The reactant ratio, catalyst concentration
and reaction temperature were optimised by varying the as-
sociated system parameters in the analysis system. In
Figure 3, the optimisation of the reactant ratio is presented.
Scheme 1. Lewis acid-catalysed three-component reaction giving 2H-2-
imidazoline (R =Benzyl, R =R =CH , R =p-nitrobenzyl, R =H).
3
1
2
3
4
5
amine (1) condenses with a ketone or aldehyde (2) to form
an intermediate imine (3) that subsequently reacts with an
a-acidic isocyanide (4). This atom-efficient reaction, with
water as the only byproduct, is ideally suited for the genera-
tion of focused libraries of pharmaceutically interesting ana-
logues by using a combinatorial approach in which molecu-
lar diversity is rapidly introduced by varying the substituents
[9]
of the reactants. Additionally, sample pre-cleanup costs
are significantly reduced due to the atom efficiency. New 2-
[10]
imidazolines are promising pharmaceutical candidates.
For example, their affinity towards the imidazoline binding
site result in a broad range of pharmaceutical applications
with a wide variety of associated biological functions, such
as hypertension, blood pressure regulation, insulin secretion
Figure 3. Relative peak areas of formed product determined by using MS
detection at different substrate ratios. The normalised peak area illus-
trates the impact of reactant ratio on product formation.
[11]
control and numerous human brain disorders. Moreover,
2
-imidazolines are used as chiral templates to increase reac-
[12]
tion enantioselectivity in asymmetric synthesis.
The experiments herein describe the conversion of ben-
In this experiment, the percentage of 4 was increased step-
wise in increments of 10% (nine steps).
1
2
3
zylamine (1; R =benzyl) and acetone (2; R =R =CH )
3
into the corresponding intermediate imine (3) followed by a
Lewis acid-catalysed reaction with p-nitrobenzylisocyanide
The product formation was determined at different reac-
tant ratios, whereas the conversion was catalysed by inject-
ing 10 mL of the ligated AgOTf catalyst complex formed by
mixing 1 mm AgOTf with an equal amount of ferrocene 6b
(Figure 4). With a repetition time of five minutes, this rapid
optimisation strategy was performed by automatically
changing the reactant A/reactant B ratio by altering the flow
rate of the HPLC pumps, as depicted in Figure 1, while the
total flow remained unchanged. The associated product for-
mation was determined by integration of the peak in the re-
corded SIM trace of the protonated product molecule (m/z
310.2). As shown in Figure 3, the ratio of reactants in the
MCR has a significant effect on the overall product forma-
tion that can be achieved. If the peak areas at different
ratios are normalised, a ratio of 50:50 3/4 results in the high-
(
4) to form the corresponding 2H-2-imidazoline (5).
In the first step of the reaction optimisation experiments,
eight triflates (Zn
AgOTf, Sc(OTf) , Cu ACHTUNGTREUNNN(G OTf) and Bi ACHTNUGTRENNUNG( OTf) ) were screened
2 3
ACHTUNGERTNUNG( OTf) , Yb ACHNUTGERTNNGNU( OTf) , LiOTf, Hf ACHNUTGTREUNNGN( OTf) ,
2
3
4
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
3
for activity towards the synthesis of 5. In this screening ex-
periment, the triflates were injected at a repetition rate of
five minutes. During the 40-minute experiment, 10 nmol
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
(ꢀ4 mg) of catalyst and 12 mmol (ꢀ2 mg) of reactants were
consumed. As indicated in Figure 2, the best-performing tri-
flate for this conversion is AgOTf. This Lewis acid catalyst
was combined with a representative ferrocene (6b) to form
the catalyst complex that was used for the further reaction
optimisation experiments.
7370
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 7368 – 7375

High-Throughput Reaction Optimisation and Activity Screening
FULL PAPER
alyst concentrations the product
formation increased significant-
ly, with a maximum at the high-
est catalyst concentration. How-
ever, the product that formed
did not increase linearly with
catalyst concentration and an
asymptotic curve was deter-
mined due to saturation of the
reaction mixture. From the re-
sults of this experiment, the op-
timal catalyst–reactant ratio
could be determined. At the
lowest ferrocene-based catalyst
concentration that showed ac-
tivity, a catalyst/reactant ratio
of 0.0125% was applied. This
ratio increases to 12.5% (cata-
lyst/reactant) at the highest cat-
alyst concentration (500 mm).
From Figure 5, it can be con-
cluded that increasing the cata-
lyst
concentration
beyond
1
00 mm (2.5%) does not signifi-
cantly improve the amount of
product formed. Therefore, the
optimum catalyst/reactant ratio
is found to be 2.5%.
Because the rate constant of
a reaction is dependent on the
reaction temperature, increas-
ing the reactor temperature re-
sults in enhanced product for-
mation. The temperature de-
pendence of this MCR to give 5
has been investigated previous-
[7]
ly. By introducing the AgOTf
Figure 4. Stabilising ferrocene catalyst ligands. Ferrocenes 6a–h share a common ferrocenic backbone with dif- catalyst at different tempera-
ferent substituents. Mixing the ferrocenes with AgOTf resulted in catalyst complexes that were screened for tures, an optimum temperature
activity towards the synthesis of 5.
curve was obtained. Although
the highest temperature yielded
est product yield. The ratio proved to be very important be-
cause a 20% reduction in either of the reactants resulted in
a 25% decrease in product formation.
To optimise the catalyst concentration at the determined
reactant concentration (4 mm), equal amounts of ferrocene
6
b and AgOTf were mixed and injected into the reaction
detection system at different concentrations. The amount of
product formed was determined by using the same method-
ology that was used in the previous experiment.
Figure 5 presents the product formation at twelve differ-
ent catalyst concentrations between 500 nm and 500 mm. Al-
though the presence of the lowest catalyst concentration
could be confirmed in the total-ion current (TIC) trace re-
corded by using a single quadrupole mass spectrometer, it
did not result in enhanced product formation. At higher cat-
Figure 5. Peak areas of formed product determined at different catalyst
concentrations during optimisation.
Chem. Eur. J. 2009, 15, 7368 – 7375
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
7371

H. Irth, C. T. Martha et al.
the highest product formation, the curve approached an op-
timum at 808C. To determine the optimum temperature
when stabilising ferrocene catalyst ligands are utilised, a
mixture of 1 mm AgOTf and 1 mm ferrocene 6b was injected
at different reactor ACHTUNGTRENNUNGt emperatures.
The results of the temperature optimisation experiment
are presented in Figure 6. The data present an obvious in-
product standard into the online system. To replicate prod-
uct responses under the actual reaction detection conditions,
the standards were injected into a stream of reactants (3 and
4). Due to the absence of catalyst, the SIM (m/z 310.2)
product response directly relates to the concentration of in-
jected product. From the integrated SIM traces, solvent-spe-
cific calibration lines were constructed with good linearity
(
1
correlation coefficients between 0.9920 (acetonitrile) and
.0000 (methanol) with a maximum residual standard devia-
tion (RSD) of 4.9%). The normalised slopes of the product
MS response calibration lines are compared in Figure 7.
Figure 6. Reaction temperature optimisation.
crease in product formation (rate constant) at elevated tem-
peratures. However, compared with previously obtained re-
sults the effect of enhanced reactor temperature is signifi-
cantly larger. The differences are quite significant, especially
when higher temperatures are applied. When recording the
temperature curve, an optimum was not achieved because
only reactor temperatures of up to 908C were investigated.
As the differences between the results of the temperature
optimisation experiments with the two different catalyst
classes suggest, temperature optima proved to be catalyst-
class specific.
Figure 7. Relative APCI-MS response of product standard in THF,
MeOH, isopropylalcohol (IPA), acetonitrile (ACN), ethylacetate
(ETOAc) and dichloromethane (CH Cl ).
2 2
Activity assessment of a ferrocene library: An important
tool for optimising homogeneous catalysis is the application
of stabilising ligands to alter the catalytically active (transi-
tion) metal centre. By applying these complexes, it is possi-
ble to optimise substrate conversion kinetics and to steer
stereo-, regio- and/or enantioselectivities. This ability to
affect the synthesis is based upon a combination of the elec-
tronic and steric parameters of the AgOTf–ferrocene cata-
lyst complex involved.
Ever since the discovery of ferrocene, the special proper-
ties of this metallocene have led to an ever-increasing
number of applications involving this aromatic scaffold in
catalysis, in electrochemistry and in coordination chemistry.
In addition to the possibility of designing specific ferrocene
ligands for specific applications, a major advantage of the
scaffold is its very high stability; ferrocene-based catalysts
are resistant to high temperatures and humidities and can
cope with a multitude of different reagents. Combined with
the low prices for ferrocene, the aforementioned factors
result in great application potential of ferrocene-based cata-
lysts in industrial processes.
System development: One of the reasons for the popularity
of spectroscopic detection techniques in the determination
of catalyst performance is the versatile applicability of those
methods. Although MS detection offers superior selectivity
and sensitivity, and an ample number of applications for or-
ganometallic and inorganic compounds in solvents other
than those commonly used in HLPC analysis are reported,
the merits of this detection technique for synthetic conver-
sions have not been fully explored yet. One of the important
factors when considering the application potential in syn-
thetic organic chemistry is the possibility of applying MS de-
tection in virtually any desirable solvent. To demonstrate
this ability, the catalyst-mediated conversion of reactants is
monitored in six different solvents, all commonly used in
synthetic organic chemistry. However, to be able to compare
product formation in different solvents in an activity assess-
ment, the solvent-specific APCI-MS response of the product
has to be calibrated.
In the first stage of the reaction optimisation strategy, a
number of metal triflates were screened for activity in the
synthetic conversion that gives target molecule 5. To broad-
en the range of the methodology, we screened a small ferro-
cene library for activity. The ferrocenes were mixed with
AgOTf (1:1) prior to the activity assessment to form catalyst
complexes. This ferrocene library is composed of some in-
The relative MS responses of the product in the different
solvents were determined by the triplicate injection of dif-
ferent concentrations (5–250 nm) of in-house-synthesised
7372
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 7368 – 7375

High-Throughput Reaction Optimisation and Activity Screening
FULL PAPER
house-produced ferrocene catalyst ligands (6 a–h) and sever-
al commercially available Josiphos (7 a–d), Taniaphos
Conclusion
(
8 a,b), Walphos (9 a,b) and a Mandyphos (10) ferrocene
We have presented an alternative integrated continuous-
flow reaction detection system for common homogeneous
catalyst development approaches based on autonomous sys-
tems for the catalysed conversion of reactants and the deter-
mination of product formation. In the comprehensive inte-
grated reaction detection analysis system, these segregated
systems were merged into a single integrated continuous-
flow system. A great virtue of the screening methodology
currently described is that the versatility of traditional ap-
proaches is preserved. This is demonstrated by the employ-
ment of the same system for an automated optimisation of
rudimentary reaction parameters and for the performance
screening of ferrocene-based catalyst complexes. In the first
type of experiment, optimum
catalyst ligand(s).
The activity of the ferrocenes was determined in a high-
throughput experiment and involved performance screening
in six different solvents.
The results of this screening confirm the ability to screen
ferrocene-ligated catalyst complexes in the selected reaction
by using the MS-based continuous-flow reaction detection
system. The different product yields in this activity assess-
ment can be directly compared by correcting the results for
the different catalyst concentrations and APCI-MS respons-
es in different solvents. The results of this assay are present-
ed in Figure 8. The steric effects introduced by the different
substrate ratios, catalyst con-
centrations and reaction tem-
peratures were determined un-
supervised. In the second type
of experiment, the applicability
of the system was expanded by
proof of its ability to screen for
different activities related to
steric parameters introduced by
different ferrocene backbone
substituents. Because of the
HPLC equipment on which this
system is based, a high degree
of automation could be ach-
ieved while the reaction detec-
tion methodology circumvented
Figure 8. Activity assessment of the ferrocene library in different solvents.
the need to use sequential unit
operations for the transfer of
catalyst complexes are of the same order of magnitude and,
therefore, catalyst performance is found to be comparable.
The catalyst activity in the different solvents is influenced
by the solvent properties. The activity in ethylacetate was
found superior and was significantly better than the perfor-
mance in dichloromethane. In general, ferrocenes 10, 8b
and 9b were rated the poorest performing catalysts for this
conversion, whereas ferrocenes 7a, 6c and 6h were found to
be the best performing catalysts. From these results, differ-
ent activities introduced by different ferrocene types can be
distinguished. Moreover, additional steric hindrance intro-
duced by different substituents, such as in ferrocenes 9a and
liquids. Moreover, by integrating the workflow into a single
system setup, reaction optimisation and activity assessment
procedures are significantly simplified.
The pertinence of the reaction detection methodology
was further improved by the considerably increased
throughput accomplished by the reduction of the number of
data points required for an activity assessment. For this
methodology to be able to screen catalysts based on a single
time-point, it must allow for sensitive quantification of the
product with high accuracy at a reaction time that only
slightly exceeds the time period required for the reaction to
reach the initial rate. This was accomplished by the imple-
mentation of MS detection for sensitive simultaneous moni-
toring of the reactant- and product-related ions involved in
catalyst-induced product formation. The current trend in
screening assays is the downscaling of assays to reduce the
amount of involved entities (catalysts and solvents) in an at-
tempt to preserve both economical and environmental re-
sources. The main disadvantage of miniaturising dimensions
is the proportional reduction in formed product and the in-
creased detection sensitivities required. Because of the in-
trinsic nature of the APCI process, APCI-MS is ideally
suited to cope with these demands. We have proved this by
9
b, result in reduced activities. Because the HPLC pumps
used in the setup were unable to cope with CH Cl , a differ-
2
2
ent setup was used for assaying the conversion in this sol-
vent. The two reactant pumps were replaced by pumps that
could pump CH Cl and the water make-up flow was omit-
2
2
ted because of the immiscibility. Although this setup
changes the related flow rates and APCI performance, the
results could still be directly compared by accurately con-
trolling reactant flow rates and the calibration of the MS
ACHTUNGTRENNUNGr esponses.
Chem. Eur. J. 2009, 15, 7368 – 7375
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
7373

H. Irth, C. T. Martha et al.
demonstrating the MS detection of low concentrations of
converted product, thereby preserving valuable catalysts and
reactants.
to increase detector compatibility by adding a water make-up flow. If di-
chloromethane was used the substrate delivery was performed by using
two Gilson model 302 HPLC pumps (Gilson, Villiers-le-Bel, France) due
to equipment incompatibilities. Due to solvent immiscibility, the water
make-up flow had to be omitted in experiments conducted in dichloro-
methane. The injection of catalyst–ligand complexes into the stream of
reactants was performed by using a fully programmable Gilson 231 auto-
injector (Gilson, Villier-le-Bel, France). The conversion into product oc-
curred in a coiled open tubular reactor (produced in-house) immersed in
a temperature-controlled water bath (Grant, Cambridge, England). The
quantification of catalyst-mediated product formation was performed by
using SIM in a LCMS-2010EV single-stage quadrupole mass spectrome-
ter (Shimadzu, ꢁs-Hertogenbosch, the Netherlands) equipped with an
APCI source and operating in positive-ion mode. Measurements were
performed by alternating between the SIM of the product and intermedi-
ate imine (m/z 310.2 and 148.1, respectively) and a full-spectrum TIC.
The main prerequisite of a detector in catalyst perfor-
mance screening of synthetic conversions is its applicability
in different solvents. APCI-MS can be used to determine
product-related ions in a wide variety of different applica-
tions, due to the multitude of gas- and solution-phase mech-
anisms that convert neutral molecules into ionic species.
However, concerning the applicability in different solvents,
the usefulness is somewhat restricted to solvents associated
with LC and GC separations. We demonstrated the feasibili-
ty of using APCI-MS for reaction detection in six different
solvents that are commonly used in synthetic chemistry. To
be able to compare catalyst-mediated product formation in
different solvents, the system was calibrated for solvent-spe-
cific MS responses before the activity assessment was con-
ducted. Although dichloromethane required slight system
modifications due to the calibration of the MS responses, a
direct comparison between the activity assays conducted in
the six different solvents could still be achieved.
Acknowledgements
This work was financially supported by an ECHO grant from the Dutch
Scientific Society NOW. Shimadzu Benelux is acknowledged for the loan
of the LCMS-2010EV single-stage quadrupole mass spectrometer. Solvias
is acknowledged for providing Josiphos, Walphos, Mandyphos and Tania-
phos ferrocene catalyst ligands.
Experimental Section
[
1] For selected reviews on combinatorial methods for the discovery of
homogeneous catalysts, see: a) A. Hagemeyer, B. Jandeleit, Y. Liu,
D. M. Poorjary, H. W. Turner, A. F. Volpe, Jr., W. H. Weinberg,
Appl. Catal. A 2001, 221, 23–43; b) B. Jandeleit, D. J. Schaefer, T. S.
Powers, H. W. Turner, W. H. Weinberg, Angew. Chem. 1999, 111,
Materials: The activity assay of the ferrocene library was conducted in
six different solvents; GC-grade (>99%) THF and dichloromethane
were purchased from Fluka (Buchs, Switzerland), methanol (absolute
ULC/MS grade) and acetonitrile (HPLC-S grade) were purchased from
Biosolve (Valkenswaard, the Netherlands). 2-propanol (>99.8%) and
ethyl acetate (>99.5%) were purchased from Sigma Aldrich (Steinheim,
Germany).
2
648–2689; Angew. Chem. Int. Ed. 1999, 38, 2494–2532; c) W. F.
Maier, K. Stçwe, S. Sieg, Angew. Chem. 2007, 119, 6122–6179;
Angew. Chem. Int. Ed. 2007, 46, 6016–6067; d) M. T. Reetz, Angew.
Chem. 2008, 120, 2592–2626; Angew. Chem. Int. Ed. 2008, 47, 2556–
The applied MCR involved three different reactants; acetone (>99%)
was purchased from J. T. Baker (Deventer, the Netherlands), benzyl-
amine was purchased from Fluka (Buchs, Switzerland), and the product
standards of 4 and 5 were synthesised in-house. The silver in the catalyst
complexes was obtained by dissolving silver trifluoromethanesulfonate
2
588; e) J. M. Serra, A. Corma, S. Valero, E. Argente, V. Botti,
QSAR Comb. Sci. 2007, 26, 11–26.
[
2] a) R. A. Potyrailo, TrAC, TrAC Trends Anal. Chem.Trends Anal.
Chem. 2003, ##23##22, 374–384; b) M. B. Onaran, C. T. Seto, J. Org.
Chem. 2003, 68, 8136–8141; c) M. T. Reetz, A. Zonta, K. Schimos-
sek, K. Liebeton, K.-E. Jaeger, Angew. Chem. 1997, 109, 2961–2963;
Angew. Chem. Int. Ed. Engl. 1997, 36, 2830–2832.
(
silver triflate, AgOTf, >98%) from Fluka (Buchs, Switzerland) in the
appropriate solvent. The other triflates that were used in the preliminary
Lewis acid catalyst screening (Zn(OTf) , Yb(OTf) , LiOTf, Hf(OTf)
Sc(OTf) , Cu(OTf) and Bi(OTf) ) were obtained from either Sigma–Al-
A
H
U
G
R
N
N
2
A
H
U
G
R
N
U
G
3
A
T
G
R
N
U
G
4
,
[
[
3] D. Fabris, Mass Spectrom. Rev. 2005, 24, 30–54.
4] a) W. Henderson, B. K. Nickleson, L. J. McCaffrey, Polyhedron
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
3
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
2
A
T
G
R
N
U
G
3
drich or Fluka. Water was produced in-house by using a Millipore Aca-
demic water purifier (Molsheim, France) and degassed by using a Bran-
son ultrasonic bath (Danbury, USA). Formic acid (98%) was purchased
from J. T. Baker (Deventer, the Netherlands). The ferrocene catalyst
ligand library that was assayed was obtained from two different sources.
The first part of the library (6a–g; see Figure 2), were synthesised in six
steps starting from ferrocenecarboxaldehyde by using literature proce-
1
2
998, 17, 4291–4313; b) J. C. Traeger, Int. J. Mass Spectrom. 2000,
00, 387–401.
[
5] a) H. R. Sahoo, J. G. Kralj, K. F. Jensen, Angew. Chem. 2007, 119,
806–5810; Angew. Chem. Int. Ed. 2007, 46, 5704–5708; b) D.
Belder, M. Ludwig, L.-W. Wang, M. T. Reetz, Angew. Chem. 2006,
18, 2523–2526; Angew. Chem. Int. Ed. 2006, 45, 2463–2466; c) O.
5
1
Trapp, S. K. Weber, S. Bauch, T. Bꢂcker, W. Hofstadt, B. Spliethoff,
Chem. Eur. J. 2008, 14, 4657–4666; d) O. Trapp, S. K. Weber, S.
Bauch, W. Hofstadt, Angew. Chem. 2007, 119, 7447–7451; Angew.
Chem. Int. Ed. 2007, 46, 7307–7310; e) N. Nikbin, M. Ladlow, S. V.
Ley, Org. Process Res. Dev. 2007, 11, 458–462; f) J. Kobayashi, Y.
Mori, K. Okamoto, R. Akiyama, M. Ueno, T. Kitamori, S. Kobaya-
shi, Science 2004, 304, 1305–1308.
[
13]
dures.
The second part of the ferrocene library contained ligands of
the Josiphos, Walphos, Mandyphos, and Taniaphos ferrocene type, which
were obtained from Sigma–Aldrich (Steinheim, Germany). The synthesis
procedure and NMR spectroscopy and high-resolution MS conformation
of these ferrocenes is described in the Supporting Information.
Reaction detection system: The analytical system (see Figure 1) that was
used for the reaction optimisation and activity assessment of the ferro-
cene library involved mixing the reactants, injecting the catalysts, con-
verting the reactants and detecting the product in a single comprehensive
continuous-flow reaction detection system. The substrate delivery config-
uration was adapted based on the applied solvent. If THF, acetonitrile.
methanol, ethyl acetate or 2-propanol were used, the mixing of reactants
was performed by using two Shimadzu LC-20AD HPLC pumps (Shi-
madzu, ꢁs-Hertogenbosch, the Netherlands) into a mixer (Valco Interna-
tional, Schenkon, Switzerland). A third LC-20AD HPLC pump was used
[6] a) N. W. Hird, Drug Discovery Today 1999, 4, 265–274; b) M. S.
Congreve, C. Jamieson, Drug Discovery Today 2002, 7, 139–142;
c) M. D. Bowman, M. M. Jacobson, B. G. Pujanauski, H. E. Black-
well, Tetrahedron 2006, 62, 4715–4727.
[7] C. T. Martha, N. Elders, J. G. Krabbe, J. Kool, W. M. A. Niessen,
R. V. A. Orru, H. Irth, Anal. Chem. 2008, 80, 7121–7127.
[8] a) R. S. Bon, C. Hong, M. J. Bouma, R. F. Schmitz, F. J. J. de Kanter,
M. Lutz, A. L. Spek, R. V. A. Orru, Org. Lett. 2003, 5, 3759–3762;
b) R. S. Bon, B. van Vliet, N. E. Sprenkels, R. F. Schmitz, F. J. J.
7374
www.chemeurj.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2009, 15, 7368 – 7375

High-Throughput Reaction Optimisation and Activity Screening
FULL PAPER
de Kanter, C. V. Stevens, M. Swart, F. M. Bickelhaupt, M. B. Groen,
R. V. A. Orru, J. Org. Chem. 2005, 70, 3542–3553; c) N. Elders, R. F.
Schmitz, F. J. J. de Kanter, E. Ruijter, M. B. Groen, R. V. A. Orru, J.
Org. Chem. 2007, 72, 6135–6142.
9] For illustrative reviews concerning multicomponent reactions, see:
a) R. W. Armstrong, A. P. Combs, P. A. Tempest, S. D. Brown, T. A.
Keating, Acc. Chem. Res. 1996, 29, 123–131; b) L. F. Tietze, A.
Modi, Med. Res. Rev. 2000, 20, 304–322; c) A. Dçmling, I. Ugi,
Angew. Chem. 2000, 112, 3300–3344; Angew. Chem. Int. Ed. 2000,
venport, D. L. Davies, J. Fawcett, D. R. J. Russell, J. Chem. Soc.
Perkin Trans. 1 2001, 1, 1500–1503; d) F. Menges, M. Neuburger, A.
Pfaltz, Org. Lett. 2002, 4, 4713–4716; e) C. A. Busacca, D. Gross-
bach, R. C. So, E. M. OꢁBrien, E. M. Spinelli, Org. Lett. 2003, 5,
595–598; f) M. Casey, M. P. Smyth, Synlett 2003, 102–106; g) E.
Guiu, C. Claver, J. Benet-Buchholz, S. Castillon, Tetrahedron:
Asymmetry 2004, 15, 3365–3373; h) A. Bastero, C. Claver, A. Ruiz,
S. Castillon, E. Daura, C. Bo, E. Zangrando, Chem. Eur. J. 2004, 10,
3747–3760; i) T. Arai, T. Mizukami, N. Yokoyama, D. Nakazato, A.
Yanagisawa, Synlett 2005, 2670–2672; j) G. Anilkumar, S. Bhor,
M. K. Tse, M. Klawonn, B. Bitterlich, M. Beller, Tetrahedron: Asym-
metry 2005, 16, 3536–3561; k) M. E. Weiss, D. F. Fischer, Z.-Q. Xin,
S. Jautze, W. B. Schweizer, R. Peters, Angew. Chem. 2006, 118,
5823–5827; Angew. Chem. Int. Ed. 2006, 45, 5694–5698; l) K. Ma, J.
You, Chem. Eur. J. 2007, 13, 1863–1871; m) S. Jautze, P. Seiler, R.
Peters, Angew. Chem. 2007, 119, 264–268; Angew. Chem. Int. Ed.
2007, 46, 1260–1264.
[
3
9, 3168–3210; d) A. Domling, Comb. Chem. High Throughput
Screening 1998, 1, 1–22; e) R. V. A. Orru, M. de Greef, Synthesis
003, 1471–1499; f) J. Zhu, Eur. J. Org. Chem. 2003, 1133–1144;
g) H. Bienaymꢃ, C. Hulme, G. Oddon, P. Schmitt, Chem. Eur. J.
000, 6, 3321–3329.
10] a) P. Bousquet, J. Feldman, J. Schwartz, J. Pharmacol. Exp. Ther.
984, 230, 232–236; b) A. Parini, C. Gargalidis Moudanos, N. Pizzi-
2
2
[
1
nat, S. M. Lanier, Notes TiPS Zeitschrift wurde bereits 1975 stillge-
legt! 1996, 17, 13–16; c) C. Dardonville, I. Rozas, Med. Res. Rev.
[13] Synthesis of the ferrocene template 6h: a) C. R. Hauser, J. K. Lind-
say, J. Org. Chem., 1957, 22, 906–908; b) D. Marquarding, H. Klusa-
cek, G. Gokel, P. Hoffmann, I Ugi, J. Am. Chem. Soc., 1970, 92,
5389–5393; c) T. Hayashi, T. Mise, M. Fukushima, M. Kagotani, N.
Nagashima, Y. Hamada, A. Matsumoto, S. Kawakami, M. Konishi,
K. Yamamoto, M. Kumada, Bull. Chem. Soc. Jpn., 1980, 53, 1138–
1151. The substitution (with retention of configuration): d) T. Haya-
shi, A. Yamazaki, J. Organomet. Chem., 1991, 413, 295–302; e) G.
W. Gokel, D. Marquarding, I. Ugi, J. Org. Chem., 1972, 37, 3052–
3058).
2
004, 24, 639–661.
[
11] Examples of biological functions of the imidazoline binding site: hy-
pertension: a) P. Bousquet, J. Feldman, Drugs 1999, 58, 799–812;
anti-inflammation: b) M. Ueno, K. Imaizumi, T. Sugita, I. Takata,
M. Takeshita, Int. J. Immunopharmac. 1995, 1, 597–603; anti-hyper-
glycemia: c) M. E. Doyle, J. M. Egan, Pharmacol. Rev. 2003, 55,
1
05–131; hormonal effects: d) M. von Rauch, M. Schlenk, R. Gust,
J. Med. Chem. 2004, 47, 915–927; human brain disorders: e) A.
Holt, J. Psychiatry Neurosci. 2003, 28, 409–414.
[
12] a) C. Botteghi, A. Schionato, G. Chelucci, H. Brunner, A. Kurzinger,
U. Obermann, J. Organomet. Chem. 1989, 370, 17–31; b) T. Mori-
moto, K. Tachibana, K. Achiwa, Synlett 1997, 783–785; c) A. J. Da-
Received: February 5, 2009
Published online: June 23, 2009
Chem. Eur. J. 2009, 15, 7368 – 7375
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
7375