A fully automated, multistep flow synthesis of 5-amino-4-cyano-1,2,3- triazoles

Article abstract of DOI:10.1039/c0ob00815j

Having demonstrated in the preceding publication the flow synthesis of aryl azides, we describe here a general protocol for the in-line purification of these versatile intermediates. As part of this investigation, we evaluated the use of ReactIR 45m as a

Full text of DOI:10.1039/c0ob00815j

View Article Online / Journal Homepage / Table of Contents for this issue
Organic &
Biomolecular
Chemistry
Dynamic Article Links
Cite this: Org. Biomol. Chem., 2011, 9, 1938
www.rsc.org/obc
PAPER
A fully automated, multistep flow synthesis of
5-amino-4-cyano-1,2,3-triazoles†
Catherine J. Smith, Nikzad Nikbin, Steven V. Ley, Heiko Lange and Ian R. Baxendale
Received 1st October 2010, Accepted 22nd November 2010
DOI: 10.1039/c0ob00815j
Having demonstrated in the preceding publication the flow synthesis of aryl azides, we describe here a
general protocol for the in-line purification of these versatile intermediates. As part of this investigation,
we evaluated the use of ReactIR 45m as a tool for real-time detection of hazardous azide contaminants.
This azide synthesis and purification process was then incorporated into a multistep flow sequence to
generate a small collection of 5-amino-4-cyano-1,2,3-triazoles directly from aniline starting materials in
a fully automated fashion.
flow processes,6 with the goal of enabling 24/7 operation by
Introduction
carrying out multiple sequential reactions in a fully automated
fashion. Here we report on the use of Gilson Trilution LC and/or
Unipoint software7 to successfully automate the flow preparation
of a small collection of 5-amino-4-cyano-1,2,3-triazoles from
simple aniline starting materials.
Following the successful application of aryl azides in the
Staudinger aza-Wittig reaction,1 we were interested in exploring
other uses of these valuable building blocks in multistep flow pro-
cesses. Although we were able to make use of a convenient catch-
react-and-release purification in the flow Staudinger procedure, we
were aware that other potential sequences employing aryl azides
might not lend themselves to this method of purification. Given
that the preparation of aryl azides involves the use of toxic anilines
and trimethylsilyl azide, it was of critical importance not only for
product purity but also for process safety to remove any unreacted
starting materials from the flow stream before isolation or further
use. We describe here the development of a general purification
protocol that enables the generation and purification of aryl azides
within a contained flow system, providing safe and efficient access
to clean aryl azides as intermediates or final products.
With pure aryl azides in hand, one can envision a wide range
of possible derivatisation reactions which can be applied to these
versatile intermediates.2 Building on previous work with similar
scaffolds,3 we were particularly interested in designing a multistep
flow route for the synthesis of the biologically interesting 5-amino-
4-cyano-1,2,3-triazole templates generated by the reaction of aryl
azides with malononitrile.4
Results and discussion
Purification of aryl azides in flow
While the flow method for aryl azide generation described in
the preceding publication generally provides aryl azides in high
conversion from the corresponding anilines, the desired azide
products are potentially contaminated with unreacted trimethylsi-
lyl azide (which is readily hydrolysed to toxic, volatile and explosive
hydrazoic acid) and aniline starting materials. When these aryl
azides were carried on directly to the Staudinger reaction in
flow,1 these contaminants could be washed to waste and collected
for immediate destruction or disposal. If aryl azides are to be
carried on to further reactions that do not provide this convenient
purification step, however, the benefits of generating aryl azides
in flow are greatly diminished by having to carry toxic materials
through with the desired product. In this case, contamination with
unreacted starting material is not only a concern with respect to
product purity, but presents an unacceptable situation in terms of
process safety.
Therefore, we have developed a scavenging protocol by which
the aryl azide stream can be purified in-line using readily available
and inexpensive scavenger resins, greatly improving both product
purity and process safety. As shown in Scheme 1, following azide
synthesis (1 mmol scale), the reaction stream passes through a
10 mm i.d. ¥ 8 cm height glass column containing 1.3 g of
polymer-supported sulfonic acid (QP-SA), followed by 1.3 g of
polymer-supported dimethylamine (QP-DMA). The QP-SA traps
As discussed in the preceding publication, safety, product purity
and efficiency – in terms of chemist and laboratory time as well
as materials – are high priorities for the design of flow synthesis
routes.5 In order to address the efficiency aspect, we set out to
explore further options for the automation of complex multistep
Innovative Technology Centre, Department of Chemistry, University of
Cambridge, Lensfield Road, Cambridge, UK CB2 1EW. E-mail: theitc@
ch.cam.ac.uk
† Electronic supplementary information (ESI) available. CCDC reference
numbers 791579, 791580, 791582, 791583. For ESI and crystallographic
data in CIF or other electronic format see DOI: 10.1039/c0ob00815j
1938 | Org. Biomol. Chem., 2011, 9, 1938–1947
This journal is
The Royal Society of Chemistry 2011
©

View Article Online
Scheme 1 Synthesis and purification of aryl azides in flow. Inset: Used scavenging column containing QP-SA (left) and QP-DMA (right) with
discolouration indicative of scavenged impurities.
any unreacted aniline whilst at the same time converting any
remaining trimethylsilyl azide to hydrazoic acid, which is in turn
trapped onto the QP-DMA.
Since this scavenging step is necessary for both purity and
safety, it was important to ensure that enough scavenger was
used to completely remove the undesired contaminants. Although
all of the azide syntheses investigated proceeded with at least
80% conversion, this scavenging protocol was tested for its
effectiveness in both a successful reaction and a hypothetical
worst-case scenario in which only a 10% reaction had occurred.
This ‘failed’ reaction was simulated by carrying out the azide
synthesis with only 0.1 equivalent of tert-butyl nitrite, resulting in a
maximum 10% conversion and requiring the scavengers to remove
the remaining 90% of unreacted starting aniline and trimethylsilyl
azide.
The scavenging protocol was evaluated by monitoring the
reaction output for the presence of azide or aniline contaminants in
standard and failed reactions with and without scavenging. Resid-
ual aniline starting material was detected by 1H NMR analysis of
the crude output stream (no evaporation). Residual hydrazoic acid
or trimethylsilyl azide was detected by colourimetric evaluation
of the output stream at periodic intervals using a ferric chloride
indicator.8
Fig. 1 Ferric chloride controls. Top: Concentration screen of trimethylsi-
lyl azide in MeCN; a positive result (red colour) is clearly visible down to
0.005 M. Bottom: Starting materials and product for test reactions; negative
control vials (-) contain only the reaction component (0.5 M in MeCN),
The substrate 3,5-bis(trifluoromethyl)aniline was chosen for
these control reactions, as the colours of both the aniline starting
material and aryl azide product would not interfere with the
colourimetric analysis. As shown in Fig. 1, it was possible to
clearly detect the presence of trimethylsilyl azide in MeCN by
visual inspection down to a concentration of 0.005 M, and it was
confirmed that none of the reaction components either gave a false
positive or interfered with the detection of trimethylsilyl azide
in a positive control sample. Consequently, flow reactions were
carried out as described above (Scheme 1), with a sample of the
output stream being tested with the ferric chloride assay every two
minutes. The total combined reaction output was also collected
and examined by 1H NMR.
positive control vials (+) contain the reaction component and trimethylsilyl
azide (1 : 1, 0.5 M in MeCN).
Evaluation of in-line monitoring of trimethylsilyl azide
While colourimetric testing provides a reliable off-line method
for detection of trimethylsilyl azide or hydrazoic acid, an in-line
monitoring technique would allow real-time detection9 of these
undesired materials within a multistep flow synthesis, without ex-
posure to potentially contaminated reaction streams. We therefore
evaluated the potential of monitoring for residual trimethylsilyl
azide or hydrazoic acid using the ReactIR flow cell recently
developed by Mettler-Toledo.10,11 The original diamond (DiComp)
Table 1 Evaluation of scavenging procedure by 1H NMR and FeCl3
As shown in Table 1, both the successful and failed reactions
without scavenging contained residual aniline and azide starting
Test reaction
(RNH2 : TMSN3 : t-BuONO)
1H NMR
(Aniline : Azide)
1
materials. However, colourimetric testing and H NMR analysis
Scavenge
FeCl3
confirmed that, within the limits of detection, azide synthesis
followed by in-line scavenging (Scheme 1) yields an output stream
free from aniline, trimethylsilyl azide or hydrazoic acid, even in
the case of a failed reaction with less than 10% conversion to the
desired product.
Standard (1 : 1 : 1)
Standard (1 : 1 : 1)
Failed (1 : 1 : 0.1)
Failed (1 : 1 : 0.1)
No
Yes
No
Yes
12 : 88
Azide only
92 : 8
+
-
+
-
Azide only
This journal is
The Royal Society of Chemistry 2011
Org. Biomol. Chem., 2011, 9, 1938–1947 | 1939
©

View Article Online
sensor for this equipment was not ideal for the monitoring of azide
species due to the inherent ‘blind spot’ of the diamond sensor
(2250–1950 cm-1). This allows only weak absorbance in the region
N
N
N asymmetric stretching region is much improved for the
silicon (Fig. 2, bottom) over the original diamond sensor (Fig. 2,
top).12
of the normally strong azide N
N
N asymmetric stretching
We carried out a calibration of the detector using varying
concentrations of trimethylsilyl azide in acetonitrile. As shown
bands (n = 2160–2120 cm-1). We report here upon the evaluation
of a prototype silicon (SiComp) sensor, which does not suffer from
this blind spot, for the in-line detection of trimethylsilyl azide or
hydrazoic acid. As shown in Fig. 2, the sensitivity in the azide
in Fig. 3, the strong N
N
N asymmetric stretching band of
trimethylsilyl azide (n = 2141 cm-1) could be clearly detected at
concentrations of 0.05 M or greater. Therefore, the detection limits
of the device would require improvement in order to comfortably
rely on ReactIR as a replacement for the established off-line
colourimetric test to confirm complete azide removal. We were still
interested, however, in evaluating this device as a tool to provide
real-time information about reaction and scavenging processes
within the range of its current sensitivity.
We began by monitoring the simple test case of scavenging
trimethylsilyl azide in isolation. A 0.5 M solution of trimethylsilyl
azide in acetonitrile was pumped at 0.2 mL min-1 through a column
packed with either QP-SA or QP-DMA independently, or through
both reagents in series, and the output monitored by both ReactIR
and the ferric chloride assay. Fig. 4 shows an overlay of the IR
spectrum at the most concentrated point in the reaction plug for
each run, along with a reference spectrum of trimethylsilyl azide at
0.5 M. In each case there was agreement between the ferric chloride
test and the ability to detect a N
N N band in the IR spectrum.
Passage of trimethylsilyl azide through either QP-DMA or QP-
SA alone resulted in the detection of an azide species by both
IR and ferric chloride. However, no residual azide was detected
by either method after passage through a combined QP-SA/QP-
DMA scavenging column.
In order to provide useful information about real reactions
and scavenging processes, it is necessary for the ReactIR to
be able to detect trimethylsilyl azide or hydrazoic acid in the
presence of other reaction components, including the azide
Fig. 2 IR spectrum (4000–650 cm-1) of trimethylsilyl azide (0.5 M in
MeCN) using the ReactIR. Top: Diamond sensor, with grey box indicating
the diamond ‘blind spot’ (2250–1950 cm-1). Bottom: Silicon sensor.
Fig. 3 IR spectra (4000–650 cm-1) using the ReactIR silicon sensor for a concentration screen of trimethylsilyl azide in MeCN. Red line = 0.5 M, orange
line = 0.05 M, green line = 0.005 M, blue line = 0.0005 M. Inset: Zoom of the azide N
N asymmetric stretching region (2240–2000 cm-1). The strong
N asymmetric stretching band of trimethylsilyl azide (n = 2141 cm-1) is clearly visible at concentrations of 0.05 M or greater.
N
N
N
1940 | Org. Biomol. Chem., 2011, 9, 1938–1947
This journal is
The Royal Society of Chemistry 2011
©

View Article Online
synthesis of 1-azido-3,5-bis(trifluoromethyl)benzene. ReactIR re-
sults are again presented as an overlay of the IR spectrum at the
most concentrated point in the reaction plug for each run, along
with reference spectra of trimethylsilyl azide (Fig. 6, red dotted
lines) and 1-azido-3,5-bis(trifluoromethyl)benzene (Fig. 6, blue
dotted lines). In agreement with our previous control reactions,
no undesired azide species were detected by either ferric chloride
testing or ReactIR in the reactions with in-line scavenging (Fig. 6,
solid green lines). The solid purple lines in Fig. 6 represent the test
reactions without scavenging. In the standard reaction without
scavenging (Fig. 6, left, purple line), the small amount of residual
trimethylsilyl azide was detected by the ferric chloride test, but
was not visible in the IR spectrum. In the corresponding failed
reaction, however, the unreacted trimethylsilyl azide can be clearly
detected in the IR spectrum (Fig. 6, right, purple line).
Fig. 4 IR spectra for trimethylsilyl azide scavenging tests, zoomed into
the azide N
N asymmetric stretching region (2240–2000 cm-1). Red
dotted line = trimethylsilyl azide reference, dark purple solid line =
QP-DMA only, light purple solid line = QP-SA only, green solid line =
scavenging with both QP-SA and QP-DMA.
N
As demonstrated by these control reactions, the ReactIR
cannot yet compete with off-line colourimetric testing for reliable
detection of small amounts of trimethylsilyl azide in reaction
streams. However, the ReactIR can provide important information
in terms of safety in the flow synthesis of aryl azides by enabling
early in-line detection of critical process failure.
product. We therefore returned to the synthesis of 1-azido-3,5-
bis(trifluoromethyl)benzene and repeated the control reactions
described in Table 1, with the ReactIR cell incorporated just
prior to the exit of the flow system. The ReactIR results were
validated by comparison with ferric chloride testing and 1H NMR
analysis of the output as previously outlined. Fig. 5 shows the
IR reference spectra of the starting materials and 1-azido-3,5-
bis(trifluoromethyl)benzene product. As shown in Fig. 5 (inset),
the detection of undesired trimethylsilyl azide or hydrazoic acid
contaminants by ReactIR may be complicated by the presence and
Synthesis of 5-amino-4-cyano-1,2,3-triazoles
With access to a flow stream of pure aryl azide, we investigated the
further reaction of these azide intermediates to generate a small
collection of 5-amino-4-cyano-1,2,3-triazoles (Scheme 2) in flow.4
A seemingly straightforward method for the combination of
this azide stream with the base and malononitrile would be to
introduce these components in a third flow stream. However,
working in segmented flow presents a technical challenge in terms
potential overlap of the N
N N asymmetric stretching band for
the aryl azide product (in this case, n = 2122 cm-1).
1
Fig. 6 shows the ReactIR results, as well as the H NMR and
ferric chloride test results for the standard (left) and failed (right)
Fig. 5 Reference IR spectra (4000–650 cm-1) for the starting materials and product for scavenging test reactions. Dark blue line = 1-azido-
3,5-bis(trifluoromethyl)benzene product, red line = trimethylsilyl azide starting material, light blue line = 3,5-bis(trifluoromethyl)aniline starting
material, green line = tert-butyl nitrite starting material; each 0.5 M in MeCN. Inset: Zoom of the azide N
N
N asymmetric stretching region
(2240–2000 cm-1).
This journal is
The Royal Society of Chemistry 2011
Org. Biomol. Chem., 2011, 9, 1938–1947 | 1941
©

View Article Online
Fig. 6 Evaluation of ReactIR monitoring for trimethylsilyl azide scavenging in the flow synthesis of 1-azido-3,5-bis(trifluoromethyl)benzene. Presented
here are ReactIR, 1H NMR and ferric chloride test results for these control reactions. IR spectra (2240–2000 cm-1) are representative spectra for the flow
reactions with (green solid line) and without (purple solid line) scavenging, overlaid with reference spectra for trimethylsilyl azide (red dotted line) and
1-azido-3,5-bis(trifluoromethyl)benzene (blue dotted line). Left: Standard reaction with (green) and without (purple) scavenging. Right: Failed reaction
with (green) and without (purple) scavenging.
in the triazole synthesis. It was found that the loaded malononitrile
column could be stored in a sealed column in THF on the benchtop
for up to 6 weeks prior to use, indicating the potential for this toxic
material to be supplied as pre-loaded cartridges of ready to use
immobilised reagent.
The loading of the final malononitrile resin was difficult to
quantify because the A900 hydroxide resin is supplied water wet,
however, this procedure consistently resulted in a loading of 4–
5 mmol of captured malononitrile per column, based on the
amount of recovered malononitrile. As described below, these
columns were sufficient to effect full conversion of various aryl
azides to the desired triazole products on 1 mmol scale.
Scheme 2 General preparation of 5-amino-4-cyano-1,2,3-triazoles.
of the introduction of a third stream of reagents with consistent
stoichiometry relative to the in situ generated intermediate.1 In
this case, the azide intermediate exits the scavenging column as a
disperse plug with an unknown and varying concentration profile.
Without the ability to precisely monitor this output and effect
the corresponding delivery of the malononitrile/base stream, it is
a more attractive strategy to introduce them in an immobilised
form.
Synthesis of 5-amino-4-cyano-1,2,3-triazoles in flow
Azide synthesis and purification was carried out using the general
procedure shown in Scheme 1. The pure azide stream was
then passed directly through a column of supported anionic
malononitrile before exiting the flow system as shown in Scheme 3
(Step 1 only). Early optimisation studies indicated that for some
substrates, the reactions did not proceed to completion at room
temperature. Carrying out the triazole synthesis step at 60 ◦C by
heating the malononitrile reaction column in the R4 convection
heater resulted in full conversion for a wide range of substrates.
While some of the desired product was obtained by simply
washing the malononitrile reaction column with acetonitrile
following the reaction, low mass returns indicated that the triazole
product was being retained as the corresponding anion in the
malononitrile reaction column.13 A variety of proton sources were
evaluated for the release of the desired product. The most effective
reagent for clean release of the desired triazole products was found
to be malononitrile itself (Scheme 3, Step 2). This did result in
mixed stream of the desired products with malononitrile, however
with the base component immobilised, this excess malononitrile
could be safely and easily removed from the desired products
by evaporation within a contained system using a GeneVac
evaporator (HPLC setting, 60 ◦C). This release protocol had
Immobilisation of malononitrile
We planned to effect conversion to the desired triazole products
by directing the stream of azide intermediate through a column
of malononitrile, immobilised as its corresponding anion on
polymeric support. This would also enable the use of an excess
of malononitrile and base without contamination of the resulting
product stream.
The malononitrile reaction column was thus generated using
Ambersep 900 (A900) hydroxide form resin. Resin loading was
carried out in THF, as co-workers in our laboratory had previously
found A900 hydroxide resin to cause the hydrolysis of acetonitrile
to acetamide. A 15 mm i.d. glass column was packed to 3 cm height
with a slurry of A900 hydroxide resin in ethanol. This column was
then washed with dry THF at 1 mL min-1 for 20 min, causing
the resin to shrink. The column plunger ends were adjusted to
approximately 1.5 cm height and the column was then loaded by
washing with excess malononitrile (10 mmol, 1 M in dry THF) at
a flow rate of 1 mL min-1. The resulting malononitrile column
was washed at 1 mL min-1 with dry THF to remove excess
malononitrile and residual water before switching to MeCN for use
1942 | Org. Biomol. Chem., 2011, 9, 1938–1947
This journal is
The Royal Society of Chemistry 2011
©

View Article Online
Scheme 3 Flow synthesis of 5-amino-4-cyano-1,2,3-triazoles.
the additional benefit of regenerating the reaction column, which
could be successfully reused for at least six consecutive reactions.14
A collection of 5-amino-4-cyano-1,2,3-triazoles were thus gen-
erated in flow from aniline starting materials as shown in Scheme 4,
with the triazole products being purified simply by evaporation of
solvent and malononitrile. Final products were obtained in ≥90%
purity as determined by 1H NMR.
scavenging column. Sample Loop B was switched into line, and
Pumps A and B (1 : 9) were used to deliver a dilute wash solution of
malononitrile into the system. This released any remaining triazole
product from the reaction column, which exited through the UV
detector and was collected into the same flask.
For each iterative reaction, Valves B and C were used to
switch in a fresh scavenging column and Valve D to select a new
collection flask. In this way, as many as six sequential reactions,
corresponding to 55 continuous hours of reaction time, were
easily performed by the automated reactor with no manual input
required. The reaction sequence included two reaction steps, one
purification step and one release step all carried out within a
contained flow system, and provided the triazole products shown
in Table 2 in good overall yields and purities following only
evaporation of solvent and malononitrile.
While the general automated procedure shown in Scheme 4
provided access to useful quantities of triazole products in good
overall yields and purities for a range of substrates, several
reactions either proved unsuitable for flow processing due to low
solubility of the products (Table 2, Entries 5, 8 and 10) or required
slight modification of the general reaction procedure in order to
obtain the desired triazoles in good yields and purities.
Fully automated library synthesis of
5-amino-4-cyano-1,2,3-triazoles in flow
Ideally, automation of flow processes should increase process
efficiency both in terms of chemists’ time and laboratory utility.
Thus, our goal here was to carry out multiple sequential flow
reactions in a fully automated fashion, allowing optimal use of
laboratory facilities through 24/7 working.
Gilson Trilution LC and Unipoint software, while not specif-
ically designed for conducting flow synthesis, allow the control
and integration of various modular components. This enables the
flexible construction of versatile flow platforms for automating
complex multistep sequences. Using these software packages
to control the readily assembled reactor shown in Fig. 7 and
Scheme 4, the products in Table 2 were produced in a fully
automated fashion. With the ability to program the filling of
sample loops from a selection of reagents in an autosampler and
to control switching multiple valves, it was possible to automate
not only Steps 1 and 2 of a single reaction, but also to conduct
multiple sequential reactions without manual intervention.7
In Step 1, Sample Loops A and B were loaded with starting
materials from the autosampler, then switched into line and driven
with a constant stream of MeCN by Pumps A and B. The
starting materials mixed in a T-piece, and the combined flow
stream was directed by Valve A through a CFC, followed by a
scavenging column (selected by Valves B and C). The resulting
azide intermediate then passed into the malononitrile reaction
column, followed by a UV detector. The reactor output was
directed to a collection flask by Valve D.
Under the general reaction conditions, 5-amino-1-(4-
methoxyphenyl)-1H-1,2,3-triazole-4-carbonitrile was obtained in
an overall isolated yield of only 10% (although still in ≥90%
1
purity as determined by H NMR, Table 2, Entry 3a). Further
investigation revealed that this low yield was due to decomposition
of the 1-azido-4-methoxybenzene intermediate upon passage
through the QP-SA portion of the scavenging column. This
reaction was repeated with the scavenging column moved to after
the malononitrile reaction column, which resulted in isolation
of the triazole product in superior 73% yield and ≥90% purity
(Table 2, Entry 3b).
While early optimisation indicated that many azide interme-
diates required heating of the malononitrile column in order to
achieve full conversion to the corresponding triazole products,
the reaction of 1-azido-4-nitrobenzene was found to proceed with
full conversion at room temperature. Reaction of this substrate
at 60 ◦C, however, led to isolation of the desired triazole (26%)
In Step 2, Sample Loop B was filled with a concentrated solution
of malononitrile, and Valve A was switched to bypass the CFC and
This journal is
The Royal Society of Chemistry 2011
Org. Biomol. Chem., 2011, 9, 1938–1947 | 1943
©

View Article Online
Scheme 4 Automated flow synthesis of 5-amino-4-cyano-1,2,3-triazoles. Step 1: Sample loops A and B are filled by the autosampler, then these starting
materials injected and pumped through the system by Pumps A and B to effect azide generation, scavenging and triazole synthesis. Step 2: Sample loop B
is filled by the autosampler with a concentrated solution of malononitrile (0.5 M), which is injected, diluted by Pump A and pumped through the system
at 0.05 M. Valve A is now in position 2, allowing the wash solution to bypass the CFC and scavenging column to release the desired product from the
malononitrile reaction column. Triazole products are collected, then purified by evaporation of malononitrile and MeCN in the GeneVac.
1944 | Org. Biomol. Chem., 2011, 9, 1938–1947
This journal is
The Royal Society of Chemistry 2011
©

View Article Online
Fig. 7 Automated reactor for the flow synthesis of 5-amino-4-cyano-1,2,3-triazoles as depicted in Scheme 4.
contaminated with 4-nitroaniline (5%) (Table 2, Entry 4a), likely
due to decomposition of 1-azido-4-nitrobenzene in the heated
malononitrile column. Thus, for this substrate, the general reaction
procedure was carried out with the malononitrile reaction column
at ambient temperature, providing 5-amino-1-(4-nitrophenyl)-
1H-1,2-3-triazole-4-carbonitrile in 87% yield and ≥95% purity
(Table 2, Entry 4b).
Given that the azide generation step proceeded with high
conversion for all substrates examined (86–95%, see preceding
publication, Table 1), the lower overall yields generally observed
for triazole synthesis may indicate similar decomposition of other
azide intermediates in the purification or reaction columns.
Finally, 5-amino-1-(2-iodophenyl)-1H-1,2-3-triazole-4-carbo-
nitrile (Table 2, Entry 6) was found to be a low-melting solid and
proved unsuitable for purification under the general evaporation
conditions. Thus a scavenging process was developed in order to
remove the excess malononitrile within the flow system.
fore the exit of the flow system. This scavenging sequence was
incorporated into the automated reaction process by extending
the overall reaction time and incorporating a second set of column
switching valves for the PS-TBD scavenging resin, making it
possible to fully automate multiple sequential reactions. This
PS-TBD scavenge step provided the desired triazole product
completely free from malononitrile, but unfortunately in only
1
approximately 80% purity as indicated by H NMR. The signifi-
cant impurities were acetamide and cyanoacetamide, along with
two additional unidentified by-products, which were presumably
formed by malononitrile and acetonitrile during the long residence
time in the basic scavenging column. With the toxic malononitrile
sequestered by this in-line scavenging process, these other impu-
rities could be removed by column chromatography to provide
the desired product in high purity and 55% yield (Table 2, Entry
6). Although this scavenging process allows successful removal of
toxic malononitrile in this case, the expense of the PS-TBD reagent
and the additional impurities generated during the scavenging
step render the original evaporation protocol more attractive for
general application.
Malononitrile could be successfully removed from the product
stream by including a 15 mm i.d. glass column packed with 1,5,7-
triazabicyclo[4.4.0]dec-5-ene polystyrene (PS-TBD, 11.25 g)15 be-
This journal is
The Royal Society of Chemistry 2011
Org. Biomol. Chem., 2011, 9, 1938–1947 | 1945
©

View Article Online
Table 2 5-Amino-4-cyano-1,2,3-triazoles generated by the automated flow synthesis shown in Scheme 4
Entry Aniline
1
Isolated product
Isolated yielda
83%
Entry Aniline
8
Isolated product
Isolated yielda
N.Y.d
2
3
4
5
73%
9
53%
N.Y.d
60%
76%
Standard: 10%
Modified: 73%b
10
11
12
Standard: 26% product
+ 5% aniline
Modified: 87%c
N.Y.d
6
7
55%e
13
61%
85%
a All reactions on 1 mmol scale. b Modified procedure: The standard procedure was carried out with the scavenging column (2.6 g QP-SA + 2.6 g
QP-DMA) located after the malononitrile reaction column. c Modified procedure: The standard procedure was carried out with the malononitrile column
at room temperature. d N.Y. = no yield. The automated reactor shut down automatically mid-reaction, due to high system pressure caused by precipitate
formation. The products shown were isolated: by flushing the system with acetonitrile (Entry 5), as the material collected in the output flask before reactor
shutdown (Entry 8), or as crystalline solid recovered from the malononitrile reaction column (Entry 10). e This low-melting solid could not be purified
by evaporation under the general conditions in the GeneVac. Instead, malononitrile was removed in-line with PS-TBD scavenger. The resulting crude
product was contaminated with by-products from the scavenging process, which were removed by column chromatography.
1946 | Org. Biomol. Chem., 2011, 9, 1938–1947
This journal is
The Royal Society of Chemistry 2011
©

View Article Online
D. Segnini, A. Da Settimo and O. Livi, Ital. J. Biochem., 1979, 28, 194–
206; D. R. Sutherland and G. Tennant, J. Chem. Soc. C, 1971, 706–713;
A. Dornow and J. Helberg, Chem. Ber., 1960, 93, 2001–2010.
5 M. Baumann, I. R. Baxendale and S. V. Ley, Mol. Diversity, 2011,
DOI: 10.3987/COM-10-S(E)77; J. Yoshida, The Chemical Record,
2010, 10, 332–341; V. Hessel, Chem. Eng. Technol., 2009, 32, 1655–
1681; Microreactors in Organic Synthesis and Catalysis, T. Wirth,
ed., Wiley-VCH, Weinheim, 2008; N. Kockmann, M. Gottsponer, B.
Zimmermann and D. Roberge, Chem.–Eur. J., 2008, 14, 7470–7477;
B. P. Mason, K. E. Price, J. L. Steinbacher, A. R. Bogdan and D. T.
McQuade, Chem. Rev., 2007, 107, 2300–2318; A. J. deMello, Nature,
2006, 442, 394–402.
6 F. Venturoni, N. Nikbin, S. V. Ley and I. R. Baxendale, Org. Biomol.
Chem., 2010, 8, 1798–1806; M. D. Hopkin, I. R. Baxendale and S. V.
Ley, Chem. Commun., 2010, 46, 2450–2452; Z. Qian, I. R. Baxendale
and S. V. Ley, Synlett, 2010, 505–508; I. R. Baxendale, J. Deeley, C. M.
Griffiths-Jones, S. V. Ley, S. Saaby and G. K. Tranmer, Chem. Commun.,
2006, 2566–2568.
7 For examples of other automated sequences carried out using this
software see: I. R. Baxendale, S. V. Ley, C. D. Smith, L. Tamborini and
A.-F. Voica, J. Comb. Chem., 2008, 10, 851–857; C. M. Griffiths-Jones,
M. D. Hopkin, D. Jo¨nsson, S. V. Ley, D. J. Tapolczay, E. Vickerstaffe
and M. Ladlow, J. Comb. Chem., 2007, 9, 422–430; M. Baumann, I. R.
Baxendale, S. V. Ley, C. D. Smith and G. K. Tranmer, Org. Lett., 2006,
8, 5231–5234; I. R. Baxendale, C. M. Griffiths-Jones, S. V. Ley and
G. K. Tranmer, Synlett, 2006, 427–430.
8 M. E. Kopach, M. M. Murray, T. M. Braden, M. E. Kobierski and
O. L. Williams, Org. Process Res. Dev., 2009, 13, 152–160; F. Feigl and
V. Anger, Spot Tests in Inorganic Analysis, trans. R. E. Oesper, Elsevier,
Amsterdam, 6th English edn, 1972, p. 357.
Conclusions
The method described here for the flow synthesis and purification
of aryl azides provides safe, scalable access to these versatile and
important compounds. This general method can be used to provide
a wide range of aryl azides in high conversions and purities, ready
for isolation as final products or application in a variety of further
transformations, such as the Staudinger reaction1 or heterocycle
formation.
The flow synthesis of 5-amino-4-cyano-1,2,3-triazoles involved
four to five unit operations per reaction and accomplished the safe
handling of toxic materials, unstable intermediates and hazardous
reactions within a contained flow system, requiring only evapora-
tion of solvent and malononitrile to furnish the desired products
in high purity. As a result, a small library synthesis involving this
complex multistep flow process was successfully carried out in a
fully automated fashion, conducting up to six sequential reactions
over 55 h with no manual intervention.
The development of improved technologies and techniques for
flow processes and their automation is a significant challenge, but
one which gives us increasing confidence that even longer, more
challenging reaction sequences are possible. These techniques
provide chemists with valuable tools with which to enhance
productivity through an increased working schedule capable of
a 24/7 operating regime.
9 J. Wiss, C. Fleury and U. Onken, Org. Process Res. Dev., 2006, 10,
349–353.
10 The ReactIRTM 45 m and iC IRTM software are available from Mettler-
Toledo AutoChem, 7075 Samuel Morse Drive, Columbia, Maryland,
USA. Website: www.mt.com.
Acknowledgements
11 H. Lange, C. F. Carter, M. D. Hopkin, A. Burke, J. G. Goode, I. R. Bax-
endale and S. V. Ley, Chem. Sci., 2011, DOI: 10.1039/C0SC00603C;
Z. Qian, I. R. Baxendale and S. V. Ley, Chem.–Eur. J., 2010, 16, 12342–
12348; C. F. Carter, H. Lange, S. V. Ley, I. R. Baxendale, B. Wittkamp,
J. G. Goode and N. L. Gaunt, Org. Process Res. Dev., 2010, 14, 393–
404; C. F. Carter, I. R. Baxendale, M. O’Brien, J. B. J. Pavey and S. V.
Ley, Org. Biomol. Chem., 2009, 7, 4594–4597.
12 All spectra shown here have had MeCN solvent background subtracted.
See supplemental information for specific details about ReactIR
monitoring.
13 Triazole products were only partially trapped on the reaction column,
ruling out the possibility for a catch-and-release procedure. The extent
of product trapping seemed to vary from substrate to substrate,
presumably according to the pKa of the free amine protons.
14 Reuse of the same malononitrile column for six sequential reactions
resulted in minimal cross-contamination; each triazole product was
found to contain £1% of each of the previous products by 1H
NMR. It was found that further washing with malononitrile did little
to improve this cross-contamination. For the purposes of this study,
it was determined to be acceptable to obtain products with traces
of cross-contamination in favour of the process efficiency benefits of
reusing malononitrile columns. If this degree of cross-contamination
is unacceptable for a given library synthesis, it was found that cross-
contamination could be completely eliminated by swapping in a fresh
malononitrile column between sequential reactions.
We gratefully acknowledge Anachem Instruments Limited for
assistance with Trilution LC software, Mark D. Hopkin for advice
on automation, Mettler-Toledo AutoChem for assistance with the
ReactIR, the Paul Mellon Fellowship (C. J. S), the EPSRC (N. N.
and H. L.), the Royal Society (I. R. B.) and the BP 1702 Fellowship
(S. V. L.) for funding. We also wish to thank J. E. Davies for
crystal structure determination and the EPSRC for a financial
contribution toward the purchase of the X-ray diffractometer. We
are grateful to the Department of Chemistry Photography Services
for assistance with graphics.
Notes and references
1 C. J. Smith, C. D. Smith, N. Nikbin, S. V. Ley and I. R. Baxendale, Org.
Biomol. Chem., 2011, DOI: 10.1039/c0ob00813c.
2 S. Bra¨se, C. Gil, K. Knepper and V. Zimmermann, Angew. Chem., Int.
Ed., 2005, 44, 5188–5240.
3 C. J. Smith, F. J. Iglesias-Siguenza, I. R. Baxendale and S. V. Ley, Org.
Biomol. Chem., 2007, 5, 2758–2761.
4 V. Y. Rozhkov, L. V. Batog, E. K. Shevtsova and M. I. Struchkova,
Mendeleev Commun., 2004, 14, 76–77; G. Biagi, I. Giorgi, O. Livi, A.
Nardi and V. Scartoni, J. Heterocycl. Chem., 2002, 39, 1293–1298; G.
L’Abbe and L. Beenaerts, Bull. Soc. Chim. Belg., 1989, 98, 421–422; A.
Da Settimo, O. Livi, P. L. Ferrarini and G. Biagi, Farmaco, Ed. Sci.,
1980, 35, 308–323; A. Da Settimo, O. Livi, P. L. Ferrarini, I. Tonetti
and E. Ciabattini, Farmaco, Ed. Sci., 1979, 34, 371–382; A. Lucacchini,
15 PS-TBD resin from several sources was evaluated for use in this
scavenging protocol. Performance was found to vary both in terms
of purity of the isolated products and also in terms of backpressure
characteristics for use in flow. PS-TBD from Biotage (part number
800321) was used for the work described here.
This journal is
The Royal Society of Chemistry 2011
Org. Biomol. Chem., 2011, 9, 1938–1947 | 1947
©