A monolith immobilised iridium Cp catalyst for hydrogen transfer reactions under flow conditions

Article abstract of DOI:10.1039/c4ob02376e

An immobilised iridium hydrogen transfer catalyst has been developed for use in flow based processing by incorporation of a ligand into a porous polymeric monolithic flow reactor. The monolithic construct has been used for several redox reductions demonstrating excellent recyclability, good turnover numbers and high chemical stability giving negligible metal leaching over extended periods of use.

Full text of DOI:10.1039/c4ob02376e

Organic &
Biomolecular Chemistry
View Article Online
View Journal | View Issue
PAPER
A monolith immobilised iridium Cp* catalyst for
hydrogen transfer reactions under flow conditions
Cite this: Org. Biomol. Chem., 2015,
3, 1768
1
a
a
a,b
Maria Victoria Rojo,* Lucie Guetzoyan and Ian. R. Baxendale
Received 10th November 2014,
Accepted 2nd December 2014
An immobilised iridium hydrogen transfer catalyst has been developed for use in flow based processing
by incorporation of a ligand into a porous polymeric monolithic flow reactor. The monolithic construct
has been used for several redox reductions demonstrating excellent recyclability, good turnover numbers
and high chemical stability giving negligible metal leaching over extended periods of use.
DOI: 10.1039/c4ob02376e
www.rsc.org/obc
Introduction
Transition metal mediated hydrogen transfer reactions are nor-
mally characterised by relatively mild conditions enabling the
reduction of aldehydes, ketones and imines or the corres-
1
ponding oxidation of alcohols and amines in high yields. In
these processes the catalyst typically acts as a hydride relay
mediating transfer between two starting materials in different
oxidation states (i.e. alcohol/aldehyde) operated under pseudo-
Fig. 1 Parent iridium Cp* type catalyst 1, the parent ligand 2 and the
polymerisable ligand derivative 3.
2
equilibrium type conditions. Consequently, the processes are
often described as “hydrogen transfer” or “borrowing hydro-
gen” reflecting the overall redox-neutral transformation.
Although the use of homogeneous metal catalysts have
shown significant scope allowing hydrogen transfer reactions
(
2 → 3) for rapid incorporation into a polymeric matrix (Fig. 1).
Interestingly, complex 1 had shown reasonable activity under
photoirradiation for the catalytic oxidation of water indicating
a robust complex, stable ligand association and proven poten-
3
to be scaled to kilogram levels, the issue of catalyst separation
6
tial for redox cycling.
from the product, and thereby achieving acceptable ppm con-
tamination levels is still a major challenge. In addition many
of the most effective catalysts are based upon precious tran-
sition metals such as Pt, Ru, Ir and Rh that, without an
Results and discussion
effective recycling strategy, preclude their use in industrial Our target was to prepare a monolith based reactor to act as
7,8
important processes. An obvious solution is to therefore carry both the catalyst support and a direct plug in flow cartridge.
out immobilisation of the catalyst onto a solid support facili- Monoliths are a single continuous piece of permeable media
tating post reaction purification by filtration or engineering constructed from either organic or inorganic materials. In the
fixed bed catalyst reactor systems. Consequently several immo- context of this work, the monolith was generated via suspen-
bilisation strategies for this class of catalyst have previously sion polymerisation to create a permanent, well-defined
4
been investigated. However, in these cases high metal leach- porous architecture that retained its structural integrity inde-
ing or significant catalyst deactivation compared to the non- pendently of the solvent or reagents used due to a high degree
immobilised species has been observed. To overcome some of of crosslinking within the polymer matrix. This type of con-
these issues we herein describe an alternative immobilisation struct is also known to offer significantly higher mass transfer
strategy designed specifically for use in flow based processing compared to traditional bead immobilisation formats as it
5
7
scenarios. We made use of the known iridium complex 1
relies on convective flow instead of diffusion. Furthermore,
which seemed highly amenable to simple ligand modification large pressure drops are not usually observed using such
9
monoliths in flow processes, this is desirable to ensure the
consistency of the flow process over time.
a
Our plan was therefore to generate the monomer ligand 3,
polymerise it into a monolith and then finally load the iridium
metal. The functional monomer ligand 3 was easily prepared
Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge,
CB2 1EW, UK. E-mail: mavirm@hotmail.com
b
Department of Chemistry, University of Durham, South Road, Durham, DH1 3LE, UK
1768 | Org. Biomol. Chem., 2015, 13, 1768–1777
This journal is © The Royal Society of Chemistry 2015

View Article Online
Organic & Biomolecular Chemistry
Paper
9
: 1 hexane–EtOAc (R
f
0.32) solvent system, affording 94–96%
isolated yield of 7.
An initial comparative test of the catalytic activity of the
parent complex 1 alongside the new vinyl analogue 7 was
undertaken in batch. In parallel experiments benzaldehyde
and acetophenone (1 mmol) were heated at reflux in iPrOH
with 3 mol% of complex (1 or 7) and 5 mol% of tBuOK. After
3
h the reactions were stopped and analysed, pleasingly only
the presence of the corresponding alcohols were detected in
each of the four vessels. This result encouraged us to further
progress the study.
Many immobilised catalysts are limited in their substrate
scope as they suffer from metal leaching induced by competi-
tive ligand exchange. This is especially true for heterocyclic
substrates commonly found in many pharmaceutical and agro-
chemicals, where exchange and sequestration from the solid
phase is a notorious problem. Therefore to quickly assess the
stability of complex 7 and evaluate the potential for deleterious
ligand exchange, we conducted several experiments using
various chelating ligands including under simulated reaction
conditions. As an illustration, solutions of complex 7 and
2-phenylpyridine (equimolar) were heated at 60–100 °C (5 °C
increments) for 1, 4, 12 and 24, 120 h in iPrOH and acetone
with additionally added tBuOK (20 mol%) or tetraethyl-
ammonium formate (2 equiv.). In all cases no evidence was
found by NMR or MS for ligand exchange. In the cases of reac-
tions involving iPrOH, small amounts of acetone were detected
in the crude reactions, although this was not quantified.
Having established the general stability of the complex we next
Scheme 1 Synthesis of the functional monomer 3 (A) and preparation
of complex 7 (B).
from commercially available starting materials; 2-bromopyri- turned our attention to the formation of the monolith.
dine (4) and 4-vinylphenylboronic acid (5) were reacted via a
To make the most effective use of the valuable precursor
palladium catalysed cross coupling reaction reproducibly in complex 6 we elected to pursue a strategy involving initial poly-
1
0
>
82% yield on a 5–10 mmol scale (Scheme 1). With easy merisation of ligand 3 within a monolith followed by a flow
access to gram quantities of the monomer we first evaluated based loading of the iridium complex. Using a Vapourtec R4
its coordination and stability in association with the iridium unit we were able to rapidly screen conditions for the monolith
metal generating complex 7.
formation and successive iridium loading. The formation of
The original complex 1 is synthesised by ligand exchange of the monolith takes place inside a sealed column (Omnifit
the dimeric complex pentamethylcyclopentadienyliridium(III) column with fixed end pieces) using an initially homogeneous
chloride (6, CAS number 12354-84-6) with 2-phenylpyridine. mixture of the functionalised monovinyl monomer 3, styrene,
Although precursor 6 is commercially available, it is extremely divinylbenzene, a porogenic solvent (dodecanol) and a radical
expensive. We alternatively found it convenient to prepare the initiator, namely 4,4′-azobis(4-cyanovaleric acid). Radical poly-
iridium dimer 6 by simply heating iridium(III) chloride hydrate merisation was induced at elevated temperatures (>90 °C using
and pentamethylcyclopentadiene in MeOH at 110 °C for the Vapourtec air heating system) generating a rigid mono-
1
1
1
0 minutes under microwave conditions (Scheme 1A). With lithic matrix possessing high levels of crosslinking (Fig. 2).
dimer 6 in hand, we hoped that addition of the monomeric The porogen and any residual non-polymeric material can be
ligand 3 to the crude reaction mixture would directly yield the easily flushed out leaving a porous functionalised monolith by
desired complex 7. However, under a range of conditions, even attaching the column in-line to the Vapourtec solvent delivery
in the presence of various organic and inorganic bases, this system.
failed to generate any adduct. It was found necessary to first
Two monolith column sizes were prepared; 6.6 × 50 mm
evaporate the MeOH and then redissolve the complex in DCM. column equating to a dry polymer weight of 0.73 g (theoretical
Subsequent treatment of this solution with either NaOAc·3H O loading of 0.44 mmol ligand) and a larger 10 × 100 mm
2
or Bu NOAc (TBAA) in the presence of monomer 3 cleanly column with a dry polymer mass of 4.82 g (theoretical loading
4
furnished the desired complex 7 in quantitative conversion. of 3.2 mmol ligand); four monoliths were made simul-
Experiments carried out in the absence of base failed to show taneously using a single Vapourtec R4 system. Following
any reaction. If required, the ligated complex 7 could also be washing, the monoliths were loaded with iridium by eluting a
easily purified by column chromatography on silica using a solution of complex 6 and equimolar TBAA in DCM (0.05 M;
This journal is © The Royal Society of Chemistry 2015
Org. Biomol. Chem., 2015, 13, 1768–1777 | 1769

View Article Online
Paper
Organic & Biomolecular Chemistry
of 0.12 mL min− , 42% conversion to 1-phenylethanol was
1
−
1
achieved. Reducing the flow rate to 0.08 mL min gave a
corresponding increase to 62% conversion. A total of 93%
−1
conversion was obtained at a flow rate of 0.05 mL min , the
practical limit of the Uniqsis system for pumping. By demon-
stration, diluting the input of the acetophenone (to 1 M)
allowed complete conversion to the alcohol at a flow rate of
.05 mL min− . Importantly, the addition of base to the solu-
1
0
tion of the starting material was found to be vital for these
transformations; experiments performed without tBuOK or
mixing a solution of base injected from an additional sample
loop led to unreacted starting materials being isolated.
Having obtained these preliminary results relating to redox
reduction, we next investigated the inverse oxidation process.
A 1 M solution of 1-phenylethanol in acetone containing 3 mol%
−
1
2 3
Cs CO was passed through the monolith (0.1 mL min ),
Fig. 2 Top image: White column containing polymerised monolith. which was heated at 60 °C. However, this failed to yield any of
Yellow column containing Iridium loaded monolith. Bottom image: Mono- the desired ketone. Similarly no conversion was obtained
lith composition (wt%).
when tBuOK, DBU or pyridine were used as bases or when the
temperature of the reactor was raised to 90 °C. In each case,
flow rate 0.15 mL min−1) through the cartridge using a re-
only crude aldol adducts resulting from acetone enolate attack
1
could be identified by H-NMR and GC-MS. During the course
of this investigation, we subsequently uncovered a publication
by Fujita et al. where the parent dimer complex 1 had pre-
cycling strategy (total time 20 h). The solution was then changed
to neat DCM and the column washed for a further 4 h.
Based upon inductively coupled plasma-mass spectroscopy
viously been tested in the oxidation of benzyl alcohol but
(ICP-MS) iridium analysis, an accumulated iridium content of
1
2
showed little activity (8% after 20 h). In additional testing we
also confirmed that both complex 1 and its derivative 7
showed negligible activity for the oxidation of either benzyl
alcohol or 1-phenylethanol using acetone as the reciprocal
partner (<4%). Of some note was that on injecting further runs
of benzaldehyde and acetophenone through the same small
monolith reactor in iPrOH we were again able to obtain com-
plete reduction, confirming that it was still active.
0
.33 (±0.04–4 columns) and 2.54 (± 0.19–3 columns) mmol for
the two column sizes was achieved. Using alternative solvents
or heating the column during loading failed to improve upon
these loadings which are 75% (small; 6.6 × 50 mm) and 79%
(large; 10 × 100 mm) of the theoretical maximum.
To test the monoliths, we placed one of the small units
0.73 g, 0.44 mmol theoretical) into the flow path of a Uniqsis
(
FlowSyn (Scheme 2). A 1 mL aliquot of benzaldehyde (1.5 M)
containing 3 mol% tBuOK in iPrOH was injected into the flow
path and directed through the monolith, which was heated to
To evaluate the full utility of the flow reactor we set up a
flow system with an autosampler input and a fraction collector
output (Uniqsis ALF system). This was used in two modes
0 °C, at a flow rate of 0.12 mL min− , giving a residence time
1
9
(
Fig. 3). First, we placed a new 10 × 100 mm iridium cartridge
of approximately 6 minutes. This led to complete conversion
to the corresponding alcohol with an effective catalyst loading
of ∼35 mol% (theoretical or 26% corrected based upon
ICP-MS). Injecting a further 5 repeat plugs of the benzaldehyde
solution each gave complete conversion to benzyl alcohol
demonstrating the immobilised iridium species functioned
catalytically. Also, no iridium species were found on analysis
of the final solutions by MS. Next, we examined the reduction
of acetophenone, a substrate which is known to require longer
reaction times, using the same column reactor. At a flow rate
into the system and started a continuous flow of 4-chloroaceto-
phenone (1 M in iPrOH containing 3 mol% tBuOK) at a rate of
1
0
.10 ml min− (residence time of 52 min) through the reactor.
The fraction collector was used to aliquot 5 mL samples,
which were individually analysed. The first sample through the
system showed only 18% conversion, the second 83% but by
Scheme 2 Monolith reactor testing configuration.
Fig. 3 Flow reactor with autosampler and fraction collection.
1770 | Org. Biomol. Chem., 2015, 13, 1768–1777
This journal is © The Royal Society of Chemistry 2015

View Article Online
Organic & Biomolecular Chemistry
Paper
the third sample quantitative conversion was reached. We col- from hydride conjugate addition were detected in the reaction
lected a further total of 112 samples, each of which showed com- mixtures although this was not significant (entries 9 and 34).
plete conversion, before stopping the reactor after 100 h whilst The vinylogous amide (entry 35) being a more electron rich
still at full conversion. This equates to 0.56 M of substrate pro- system gave no detectable reaction. This is consistent with a
cessed using 2.54 mmol of immobilised catalyst, giving a catalyst system requirement for a more electron deficient carbonyl for
turnover of 220 and a catalyst usage of 0.45 mol%. To validate effective reduction.
this result, a new cartridge was placed in-line and the reaction
During our testing we also made two additional discoveries
repeated using acetophenone. Apart from a slightly longer induc- regarding the storage and stability of the immobilised catalyst.
tion phase until it reached full conversion the system behaved It was found that if the catalyst was stored in iPrOH or with
identically, processing over 600 mmol of substrate before being residual alcohol still present from its use in the reduction reac-
stopped still functioning at full catalytic activity.
tions, this led to its deactivation after only a few hours. This
We also wished to test the extent of the substrate scope by was immediately identifiable as the cartridge change from the
setting the reactor to automatically progress through a small standard characteristic canary yellow to a more golden orange
library of starting materials. To this end, a collection of 40 in colour.
aldehydes and ketones were prepared as 1.5 M stock solutions.
This visual indicator was consistent with significantly
These were loaded into the autosampler. Each sample was reduced activity. A previously used cartridge stored in iPrOH
then processed consecutively using a 5 mL sample injection overnight gave only 6% conversion of benzaldehyde to benzyl
−
1
(
7.5 mmol) followed by a 15 mL iPrOH wash, which was col- alcohol (10 × 100 mm, 3 mol% tBuOK, 90 °C; 0.25 mL min ).
1
lected as a single combined fraction for analysis by H-NMR. After equivalent storage for 3 days in iPrOH the cartridge gave
For consistency purposes each sample was run in duplicate in no activity. Interestingly, a fresh column could be washed and
1
3
a randomised sequence (Table 1).
stored in iPrOH without any dertious results (10 × 100 mm,
As can be seen from the tabulated data, the reactor showed 3 mol% tBuOK, 90 °C; 0.25 mL min− – quantitative conver-
1
1
3
excellent activity with most substrates, giving full conversion.
sion). However, adding any form of base to the iPrOH storage
Both benzylic and aliphatic aldehydes and ketones were suc- solution (tBuOK, NaOMe) resulted in slow deactivation. This
cessfully reduced. However, a number of substrates (entries 11, was also found to be the case if EtOH or MeOH was used
1
5, 22, 23, 26, 29, 32 and 37) required extended residence although the deactivation took much longer, 5 and 8 days
times in order to obtain complete conversion. This was respectively which is in accordance with their relative oxidation
achieved by simply reducing the flow rate to 0.12 mL min−
1
,
potentials. We account for this behaviour by assuming that a
giving a doubled residence time which allowed these less reac- iridium hydride like species formed from abstraction of a
tive compounds to be fully transformed. An exception was hydrogen from an alcohol under basic conditions eventually
entry 28, which failed to show any improvement at reduced undergoes a competitive reductive elimination in the absence
flow rates. Indeed, the same conversion of ∼95% was obtained of a carbonyl acceptor, leading to a reduced and inactive
−1
even at a flow rate of 0.5 mL min and residence time of only iridium species. As a result of these observations we deter-
1
4
1
5 min. In addition, a notable few materials (entries 10 and mined that post-washing and storage of the reaction cartridge
1) failed to show any reaction allowing partial recovery of the in acetone stabilised the catalyst. After 3 days, a cartridge
2
starting materials. In each instance the compounds possess an stored in acetone could be used again with no discernible
acidic hydrogen. This is also consistent with the results difference in reactivity. However, indefinite storage of used
obtained from the phenolic compound entry 8, which showed monoliths was found not to be possible, with cartridges stored
low conversions even using extended reaction times. It was also for >12 days showing gradual and accelerated deactivation
identified that these compounds (entries 8, 10 and 21) were being completely inactive after 3 weeks.
retained within the reactor, causing cross contamination of
Our final assessment was to determine the extent of metal
latter samples by slow leaching of the compound into the flow leaching during a extended run. For this experiment, a freshly
stream. The compounds also retarded the activity of the reactor prepared and washed cartridge (10 × 100 mm) was inserted in-
until they had been fully purged from the system. In a separate line and a 1.35 M solution of 3-chloroacetophenone (3 mol%
reaction, we showed that 4-bromobenzaldehyde in the presence tBuOK) was eluted through the monolith (heated at 90 °C) as a
of 25 mol% of 4-chlorophenol using 3 mol% tBuOK gave only 3 continuous flow (0.25 mL min− ). The output was collected as
and 6% conversion respectively under the previously successful 200 mL fractions, which were concentrated under reduced
conditions. It is difficult to access if this result is simply a pressure and analysed for conversion. The catalyst performed
1
a
problem of depletion of the base (pK of the phenol) inhibiting well, allowing a total processed volume of 1.4 L in >98% con-
1
the reaction. However, extensive retention of the phenol com- version as determined by H-NMR (93.3 h, 1.89 M processed,
ponents by the reactor cartridge indicates indirectly a stronger TON 744, 2.45 mmol of catalyst −0.13 mol% of catalyst usage).
interaction with the immobilised catalyst. In support of this A gradual deactivation of the system was observed linearly over
hypothesis, a non-iridium loaded monolithic cartridge showed the next 6 samples (1.62 M), decreasing the conversion to 92%.
no chromatographic retention of 4-chlorophenol.
A further more rapid reduction in the conversion was observed
Finally, α,β-unsaturated ketones failed to give any carbonyl over the next 7 samples (1.89 M), reducing the conversion to
reduction; instead small quantities of the product derived 77% when the reactor was manually stopped. This again
This journal is © The Royal Society of Chemistry 2015
Org. Biomol. Chem., 2015, 13, 1768–1777 | 1771

View Article Online
Paper
Organic & Biomolecular Chemistry
Table 1 Library production using an iridium catalyst monolith reactor
a
a
Entry
1
Substrate
Conversion (run no.)
Entry
21
Substrate
Conversion (run no.)
d
100% (1, 58)
0%, (12) only 42% SM recovery
e
c
2
3
4
100% (2, 79)
100% (3, 42)
100% (4, 60)
22
23
24
55% (13) , 94% (61), 100%
72% (14), 74% (48)
100% (15, 66)
5
6
100% (5, 63)
100% (6, 68)
100% (7, 73)
25
26
100% (16, 75)
c
90% (17), 85% (57), 100%
7
8
27
28
100% (18, 50)
d
c
c
18% (8) only 57% SM recovery 24%
95% (19), 96% (44), 94%
b,e
b
c
9
33% (9) , 28 (54)
29
30
47% (20), 49% (77), 100%
d
10
0% (10) only 71% SM recovery
100% (21, 80)
100% (22, 76)
e
c
1
1
1
2
42% (11) , 82% (41), 100%
31
32
c
98% (23), 100% (67)
76% (32), 79% (51), 100%
1
3
4
100% (24, 45)
33
34
100% (33, 53)
b
b
1
97% (25), 100% (52)
5% (34) , 7 (46)
1772 | Org. Biomol. Chem., 2015, 13, 1768–1777
This journal is © The Royal Society of Chemistry 2015

View Article Online
Organic & Biomolecular Chemistry
Paper
Table 1 (Contd.)
a
a
Entry
Substrate
Conversion (run no.)
Entry
35
Substrate
Conversion (run no.)
c
c
15
43% (26), 38% (43), 100%
0% (35, 78), 0%
1
6
7
100% (27, 47)
100% (28, 70)
36
37
100% (36, 69)
c
1
44% (37), 48% (71), 100%
1
8
9
100% (29, 59)
100% (30, 76)
38
39
100% (38, 64)
100% (39, 72)
1
20
100% (31), 97% (62)
40
100% (40, 55)
a
.5 M in iPrOH with 3 mol% tBuOK; flow rate 0.25 mL min−1, residence time ∼32 min; reactor heated at 90 °C. GC-MS analysis of the reaction
b
1
c
d
mixture indicated conjugate addition, the reaction was not analysed by NMR. Flow rate 0.12 mL, residence time ∼65 min. Duplicate reactions
e
were not run due to cross contamination issues. Sample contaminated with starting material from previous run. The reactor was stopped and
purged by flowing iPrOH at a flow rate of 0.5 mL min− for 1 h.
1
coincided with an observable colour change of the monolith
reactor (canary yellow → pale orange). As a simple proof of
concept, one incomplete 200 mL reaction sample of 78%
Experimental
General information
conversion was passed through a new monolith reactor under Unless otherwise specified, reagents were obtained from com-
identical processing conditions, giving full conversion upon mercial sources and used without further purification. Sol-
1
H-NMR analysis. In addition six samples (1, 8, 10, 14, 17 vents were obtained from Fisher Scientific and distilled before
1
and 20) were selected for ICP-MS analysis. It was found use. H-NMR spectra were recorded on a Bruker Avance
that none of the concentrated samples contained any iridium DPX-400, with the residual solvent peak as the internal refer-
1
3
(
at the limits of detection used <0.1 ppm), which confirms the ence (CDCl
3
= 7.26 ppm). C-NMR spectra were recorded on
strong retention of the iridium within the polymer ligating the same spectrometer with the central resonance of the
matrix.
3
solvent peak as the internal reference (CDCl = 77.16 ppm). IR
spectra were recorded neat on a PerkinElmer Spectrum One
FTIR spectrometer with Universal ATR sampling accessories.
Letters in parentheses refer to the relative absorbance of the
peak: w = weak (<40% of the most intense peak), m = medium
Conclusions
In conclusion, we have successfully prepared a monolithic (40%–70% of the most intense peak), s = strong (>70% of the
iridium hydride transfer catalyst which has shown high activity most intense peak). High resolution mass spectra (HRMS) were
and excellent retention of the metal under flow processing recorded on a Waters Micromass LCT Premier Q-TOF spectro-
conditions. We have been able to demonstrate its use in meter by electrospray ionisation (ESI) or an ABI/MDS Sciex
a series of transfer reductions of aldehydes and ketones Q-STAR Pulsar. The mass reported is containing the most abun-
comprising both aliphatic and benzylic chemical structures. dant isotopes. Limit: ±5 ppm. LC-MS analysis was performed on
A limitation regarding the processing of substrates containing an Agilent HP 1100 series chromatograph (Mercury Luna 3μ
phenolic and N–H indole functional groups was identified. C18 (2) column) attached to a Waters ZQ2000 mass spectro-
We have furthermore conducted several scale-up experiments meter with ESCi ionisation source in ESI mode. Microwave
exploring the processing potential of these small scale assisted reactions were performed in a Biotage Initiator micro-
monoliths.
wave device (see http://www.biotage.com/). Elemental analysis
This journal is © The Royal Society of Chemistry 2015
Org. Biomol. Chem., 2015, 13, 1768–1777 | 1773

View Article Online
Paper
Organic & Biomolecular Chemistry
and ICP-MS was performed at Butterworth Laboratories Ltd.
Samples were run in triplicate and are taken as averages of the 3
runs. Digestion was conducted by microwave (BLM486G).
VapourTec® R2+ R4 unit: (see http://www.vapourtec.co.uk)
The R2+ unit consists of two HPLC pumps that can be easily
monitored and adjusted. The pumps can operate between reac-
tion pressures of 1–50 bar and flow rates of 0.05–10.0 mL
−
1
min per channel with easy switching between reagent and
solvent inputs. Reagent addition can be achieved via a continu-
ous mode by passage of a stock solution through the pump
heads or employing single alloquot injection via the two in-
line sample loops. The R4 is a heater unit, which can accom-
modate up to four independently heated reactors. Precise
temperature control and measurement is possible over the temp-
erature range of ambient to 150 °C (Fig. 4). Very high heating
Fig. 5 Uniqsis FlowSyn unit with ALF automation.
vials. The module also includes a laptop with software for
both programming and real-time monitoring of the exper-
iments (pressure and temperature plots), generating a sounding
alarm when the pressure of the system drops due to air bubbles
and giving 1 min to purge the pumps before the system stops.
−1
and cooling rates (up to 80 °C min ) allow for quick temperature
changes. Glass Omnifit® columns and supplied housing were General procedures
8
used to prepare the monoliths as previously described.
Preparation of 2-(4-vinylphenyl)pyridine monomer (3). Pro-
cedure A: A mixture of 4-vinylbenzyl boronic acid (1.03 g,
mmol), Cs CO (2.60 g, 8 mmol) and 2-bromopyridine
0.79 g, 5 mmol) were dissolved in solvent mixture of toluene–
EtOH (10 : 1 mixture, 40 mL) and tetrakis-triphenylphosphine
palladium(0) was added (0.29 g, 0.25 mmol). The mixture
Uniqsis unit: (see http://www.uniqsis.com)
7
(
2
3
The FlowSyn PEEK unit consists of two independently con-
trolled HPLC pumps with PEEK channeled flow paths which
can operate at reaction pressures of up to 70 bar and flow rates
−
1
of 0.05–10.0 mL min per pump head. Each channel can be
easily switched between a reagent or solvent supply. Reagent
addition can be conducted in a continuous manner by elution
of a stock solution through the pump heads or as a plug flow
by employing two in-line injection sample loops. The FlowSyn
includes two heatable reactor modules: a convective heating
coil and a column holder, which can be heated up to 260 °C.
Glass Omnifit® columns and backpressure regulators were
used to perform the experiments.
The FlowSyn Automated Loop Filling (FlowSyn Auto-LF) (Fig. 5)
is a module which includes an auto-sampler and a fraction collec-
tor, allowing the preparation of libraries. It loads and collects reac-
tions simultaneously, integrating wash steps to prevent cross-
contamination using a Gilson 10 ml Syringe pump. The solu-
tions of the starting materials are prepared in vials including
septum-protected caps and loaded into an intermediate
holding coil, which incorporates air bubbles to prevent
sample dispersion and dilution, and directly transferred to the
sample loops. The final products are collected using a rack of
(
which changed from bright yellow to orange) was degassed
and stirred under nitrogen while heated at 100 °C for 6 hours.
The reaction was then quenched by addition of water and
extracted with EtOAc (filtration over celite or a plug of silica
was used to remove insoluble palladium residues). The organ-
4
ics were dried over anhydrous MgSO and the solvent concen-
trated under vacuum. The crude product was purified by
column chromatography, eluting with hexane–EtOAc (95 : 5) to
give the desired monomer as a colourless oil (750 mg, 83%
yield).
Procedure B via microwave-assisted reaction: Into a 20 mL
microwave vial was added 4-vinylbenzyl boronic acid (740 mg,
5
(
mmol), Cs CO (1.79 g, 5.5 mmol) and 2-bromopyridine
2 3
569 mg, 3.6 mmol) and dissolved in toluene–EtOH
(10 : 1 mixture, 20 mL). Tetrakis-triphenylphosphine palladium(0)
was added (0.29 g, 0.25 mmol) and the mixture (which
changed from bright yellow to orange) was heated at 120 °C for
3
tion was quenched by addition of water and the crude treated
as described in the procedure above to yield the desired
0 min using a Biotage Initiator microwave device. The reac-
monomer as colourless oil (560 mg, 86% yield).
1
2
-(4-Vinylphenyl)pyridine. H NMR (400 MHz, CDCl
3
, δ/ppm,
J/Hz): 8.70 (1H, d, J = 4.6, HetAr), 8.01 (2H, d, J = 8.1, Ar),
.68–7.62 (2H, m, HetAr), 7.52 (2H, d, J = 8.4, Ar), 7.17–7.13
1H, m, HetAr), 6.77 (1H, dd, J = 17.5, 10.9, –CHvCH ), 5.83
1H, dd, J = 7.5, 0.7, –CHvCH ), 5.30 (1H, dd, J = 11.0, 0.9,
7
(
2
(
–
2
1
3
2 3
CHvCH ). C NMR (100 MHz, CDCl , δ/ppm, J/Hz) 156.82
(
(
(
(
C), 149.67 (CH), 138.69 (C), 138.17 (C), 136.73 (CH), 136.43
CH), 127.05 (2 × CH), 126.67 (2 × CH), 122.12 (CH), 114.53
−1
CH
2
). IR (νmax/cm ): 3049.2 (w), 3011.5 (w), 1628.6 (w), 1588.2
m), 1572.9 (m), 1466.7 (m), 1435.2 (m), 1297.6 (w), 1265.0 (m),
Fig. 4 VapourTec® R2+ R4 unit.
1153 (w), 1013.2 (w), 989.2 (m), 907.5 (s), 851.4 8s), 786.0 (s),
1774 | Org. Biomol. Chem., 2015, 13, 1768–1777
This journal is © The Royal Society of Chemistry 2015

View Article Online
Organic & Biomolecular Chemistry
31.9 (s), 703.7 (m). HRMS (m/z) calculated for C12
Paper
7
H
13
N
2
convective heating unit was used to heat the column at 90 °C
(
M + H), 182.0973; found 182.0970. Anal. Calculated for during 20 h. The resulted white polymeric monolith was
13
C H
11N: C, 86.15; H, 6.12; N, 7.73. Found: C, 86.16; H, 6.23; washed with dry THF at 0.1–1 mL min− (40 min ramp) for 2 h
1
N, 7.69. No P found.
at 60 °C using the R2+ and R4 combination unit. The back-
Procedure for the preparation of pyridinyl Ir complex (7)
pressure for the washing step was 1–2 bars and no monomer
Synthesis of Chloro(pentamethylcyclopentadienyl)iridium(III) was observed when the output solution was analysed by
1
dimer (6) [Ir(Cp*Cl
1
2
2
)
2
]. A mixture of IrCl
.35 mmol) and pentamethylcyclopentadiene (0.375 mL, 6.91; N, 2.17 (3.12 mmol calculated loading based upon
.4 mmol) was dissolved in methanol (6 ml, generating a dark N content).
General procedure for the loading of the monolith. A stock
3 2
·3H O (500 mg, LC-MS or H-NMR. Elemental analysis: Found C, 90.80; H,
green-black solution) and heated at 110 °C with very high
absorption for 10 min using a Biotage Initiator microwave solution (0.05 M) of iridium dimer 6 (570 mg, 3.2 mmol) and
device. The bright orange precipitate formed was filtered and equimolar tetrabutyl ammonium acetate in dry dichloro-
washed with fresh methanol (2 × 5 mL) to furnish pure methane were maintained under an N2 atmosphere and
Ir(Cp*Cl
2 2
) . Spectroscopic data was consistent with literature pumped continuously through the monolith column prepared
−
1
values.
in the previous step over 20 h at a flow rate of 0.15 mL min
using the Uniqsis FlowSyn device. The monolith turned from
100 mg, 0.125 mmol), 2-(4-vinylphenyl)pyridine (45 mg, white to yellow (see Fig. 2 main body text) and the Ir loading
Formation of ligated complex (7). A mixture of [Cp*IrCl2]
2
(
0
.25 mmol) and NaOAc·3H O (85 mg, 0.625 mmol) [alterna-
2
was calculated by inductively coupled plasma-mass spec-
troscopy (ICP-MS) iridium analysis: calculated (small column
tively (Bu) NOAc (188 mg, 0.625 mmol) can be used] was dis-
4
solved in dichloromethane (5 mL) and stirred at room 6.6 × 50 mm) iridium content 0.33 mmol (±0.04–4 columns),
temperature under argon for 24 h. The crude mixture was fil-
tered through celite to remove the base and the resultant red
solution was concentrated under vacuum. The product was
then washed with hexane to give the desired complex as a
bright orange solid in quantitative yield.
(
(
large column 10 × 100 mm) iridium content 2.54 mmol
±0.19–3 columns).
Transfer hydrogenation in flow
A set of carbonyl derivatives were reduced in flow according to
the following procedures.
Chloro(pentamethylcyclopentadienyl)[(2-pyridinyl)4-vinyl-
1
phenyl]iridum(III) (7). H NMR (400 MHz, CDCl , δ/ppm,
3
General procedure for individual experiments. A solution of
the carbonyl containing 3 mol% tBuOK in iPrOH was injected
into the sample loop of the Uniqsis FlowSyn device. The valve
was set to load and the material pumped through the mono-
lith, which was heated at 90 °C, using iPrOH as the system
solvent. The output of the reactor was directed through a
column packed with a solid-supported sulfonic acid (Amberlyst
A15 6.2 g) to sequester the base and a back-pressure regulator
J/Hz): 8.66 (1H, d, J = 5.5, Ar), 7.84 (1H, s, Ar), 7.76 (1H, d, J =
8
7
5
.0, Ar), 7.63–7.59 (2H, m, Ar), 7.11 (1H, d, J = 8.1, Ar),
.12–7.03 (1H, m, Ar), 6.78 (1H, dd, J = 17.6, 11.0, –CHvCH2),
.81 (1H, d, J = 17.5, –CHvCH ), 5.26 (1H, d, J = 17.7,
2
3
1
–
CHvCH
δ/ppm): 166.9 (C), 163.3 (C), 151.4 (CH), 144.1 (C), 139.4 (CH),
37.3 (CH), 137.0 (C), 133.9 (CH), 123.9 (CH), 122.2 (CH), 120.3
CH), 119.0 (CH), 113.7 (CH ), 88.5 (Cp*Me ), 8.97 (Cp*Me ).
HRMS (m/z) calculated for C H NClIr (M + H), 544.1378;
2 3
), 1.72 (15H, s, Cp*); C NMR (100 MHz, CDCl ,
1
(
2
5
5
(
100 psi) was added in-line. The output of the reactor was col-
2
3
26
lected as a single fraction and the solvent evaporated to yield
the corresponding alcohol which was immediately analysed.
found 544.1370.
Monolith preparation
Benzyl alcohol
General procedure for the polymerisation. The monoliths
were prepared according with the following general procedure.
The reaction scale specified below (4.8 g of mixture) provides
sufficient monolith material to fill a 10 mm i.d. × 100 mm
Omnifit® glass column. Technical grade solution (80%) of
divinylbenzene (DVB) containing a mixture of isomers was
used and percentages are given on a w/w basis (reagent/total
mixture).
Prepared from a solution of benzaldehyde in iPrOH (1.5 M,
2
mL), which was injected and passed through the monolith
(
0.7 g, 0.44 mmol theoretical loading) at a flow rate of 0.12 ml
−
1
min
(residence time of approximately 6 minutes). The
output from the reactor was collected for 80 min, giving
enough time to wash the column once the reaction was
finished. The alcohol was obtained in quantitative yield and
NMR data was consistent with literature values.
To a mixture of the functionalised monovinyl monomer 3
(
5
576 mg, 3.2 mmol, 12%), styrene (0.636 mL, 0.578 g,
.54 mmol, 12%), DVB (0.945 mL, 0.864 g, 6.64 mmol, 18%),
1-Phenylethanol
dodecanol (3.34 mL, 57%) was added 1,1′-azobis(cyclohexane- Prepared from a solution of acetophenone in iPrOH (1.5 M,
carbonitrile) (0.24 g, 1.0% relative to monomer + styrene + 2 mL), which was injected and passed through the monolith
DVB). The mixture was vigorously stirred trying to dissolve the (0.7 g, 0.44 mmol theoretical loading) at a flow rate of 0.05 ml
initiator and a 10 mm i.d × 100 mm Omnifit glass column was min− (residence time of approximately 14 minutes). The
filled from this colourless, clear solution. Both ends of the output from the reactor was collected for 80 min, giving
columns were sealed using PTFE plugs and the VapourTec R4 enough time to wash the column once the reaction was
1
This journal is © The Royal Society of Chemistry 2015
Org. Biomol. Chem., 2015, 13, 1768–1777 | 1775

View Article Online
Paper
Organic & Biomolecular Chemistry
finished. The alcohol was obtained in quantitative yield and
NMR data was consistent with literature values.
P. G. Andersson and P. Brandt, Chem. Soc. Rev., 2006, 35,
237–248; (b) T. Ikariya and A. J. Blacker, Acc. Chem. Res.,
2007, 40, 1300–1308; (c) M. H. S. A. Hamid, P. A. Slatford
and J. M. J. Williams, Adv. Synth. Catal., 2007, 349, 1555–
1575; (d) T. Suzuki, Chem. Rev., 2011, 111, 1825–1845;
(e) C. Mazet, Chimia, 2011, 65, 802–805.
General procedure for experiments using a single input and
a fraction collector. The fraction collector and the laptop were
connected to the FlowSyn reactor and the ALF software used to
program the experiment to run with a continuous input of
starting material and the product to be collected in 5 mL frac-
tions. A stock solution of the 4-chloroacetophenone (1 M in
iPrOH) containing 3 mol% tBuOK was pumped continuously
2 (a) G. Z. Wang, U. Andreasson and J. E. Backvall, J. Chem.
Soc., Chem. Commun., 1994, 1037–1038; (b) U. Karlsson,
G. Z. Wang and J. E. Backvall, J. Org. Chem., 1994, 59, 1196–
1198; (c) A. M. Hayes, D. J. Morris, G. J. Clarkson and
M. Wills, J. Am. Chem. Soc., 2005, 127, 7318–7319; (d) X. Wu
and J. Xiao, Chem. Commun., 2007, 2449–2466; K. Fujita,
N. Tanino and R. Yamaguchi, Org. Lett., 2007, 9, 109–111;
(e) K. Fujita, Y. Enoki and R. Yamaguchi, Tetrahedron, 2008,
64, 1943–1954; (f) O. Saidi, A. J. Blacker, M. M. Farah,
S. P. Marsden and J. M. J. Williams, Angew. Chem., Int. Ed.,
2009, 48, 7375–7378; (g) M. Ito, Y. Endo, N. Tejima
and T. Ikariya, Organometallics, 2010, 29, 2397–2399;
(h) O. Saidi, A. J. Blacker, M. M. Farah, S. P. Marsden and
J. M. J. Williams, Chem. Commun., 2010, 46, 1541–1543.
3 (a) M. A. Berliner, S. P. A. Dubant, T. Makowski, K. Ng,
B. Sitter, C. Wager and Y. Zhang, Org. Process Res. Dev.,
2011, 15, 1052–1062.
4 (a) D. J. Bayston, C. B. Travers and M. E. C. Polywka, Tetra-
hedron: Asymmetry, 1998, 9, 2015–2018; (b) Y. C. Chen,
T. F. Wu, J. G. Deng, H. Liu, X. Cui, J. Zhu, Y. Z. Jiang,
M. C. Choi and A. S. Chan, J. Org. Chem., 2002, 67, 5301–
5306; (c) X. Li, X. Wu, W. Chen, F. E. Hancock, F. King and
J. Xiao, Org. Lett., 2004, 6, 3321–3324; (d) P. N. Liu,
J. G. Deng, Y. Q. Tu and S. H. Wang, Chem. Commun., 2004,
2070–2071; (e) P. N. Liu, P. M. Gu, J. G. Deng, Y. Q. Tu
and Y. P. Ma, Eur. J. Org. Chem., 2005, 3221–3227;
(f) Y. Arakawa, N. Haraguchi and S. Itsuno, Tetrahedron
Lett., 2006, 47, 3239–3243; (g) L. Jiang, T. F. Wu, Y. C. Chen,
J. Zhu and J. G. Deng, Org. Biomol. Chem., 2006, 4, 3319–
3324; (h) G. H. Liu, J. Y. Wang, T. Z. Huang, X. H. Liang,
Y. L. Zhang and H. X. Li, J. Mater. Chem., 2010, 20, 1970–
1975; (i) S. J. Lucas, B. D. Crossley, A. J. Pettman,
A. D. Vassileiou, T. E. O. Screen, A. J. Blacker and
P. C. McGowan, Chem. Commun., 2013, 49, 5562–5564.
5 (a) C. Scheeren, F. Maasarani, A. Hijazi, J. .-P. Djukic,
M. Pfeffer, S. D. Zarić, X.-F. Le Goff and L. Ricard, Organo-
metallics, 2007, 26, 3336–3345; (b) L. Li, W. W. Brennessel
and W. D. Jones, J. Am. Chem. Soc., 2008, 130, 12414–
−
1
(
flow rate 0.10 ml min , residence time 52 min) through a
new monolith (10 × 10 mm i.d., 4.82 g, theoretical loading of
.54 mmol), which had previously been washed with fresh
2
iPrOH. The column was heated to 90 °C and the output col-
lected into a rack of vials which were individually analysed by
NMR. 4-Chloro-1-phenyl ethanol: Obtained as a pale brown
1
oil. H NMR (400 MHz, CDCl
3
, δ/ppm, J/Hz): 7.34–7.30 (4H, m,
Ar), 4.90–4.85 (1H, m, –CH), 2.14 (1H, OH), 1.48 (3H, d, J = 6.6,
1
3
–CH
3
); C NMR (100 MHz, CDCl
3
, δ/ppm): 144.3 (C), 133.1 (C–
).
Cl), 128.6 (2 × CH), 126.8 (2 × CH), 69.7 (CH–OH), 25.3 (CH
3
General procedure for the preparation of a library of com-
pounds using automated injection and a fraction collector.
The FlowSyn Automated Loop Filling (FlowSyn Auto-LF)
module was connected to the flow reactor and programmed
using the laptop, so the conditions and reagent rack layout for
each experiment were established. A new monolith (10 ×
1
0 mm i.d., 4.82 g, theoretical loading of 3.2 mmol) was
attached to the system and washed with fresh iPrOH. Stock
solutions of the starting materials (1.5 M in iPrOH, 12 mL)
containing 3 mol% tBuOK were placed in the sample rack, and
loaded via the 5 mL holding coil, with air bubbles incorpor-
ated in between samples to avoid dispersion. Reagent solu-
tions were transferred directly to the sample loops of the
Uniqsis reactor and pumped through the monolith, which was
heated at 90 °C, at a flow rate of 0.25 mL min− using iPrOH
as the system solvent. The experiment was set up so that the
output was collected as 20 mL fractions (5 mL experiment +
1
1
5 mL wash), which were individually analysed by NMR. The
software provided real-time monitoring, with pressure and
temperature plots ensuring that the system was stable all over
the run. Each experiment was run in duplicate in a random
sequence. Compounds were assessed for conversion by com-
parison to authentic samples of the starting material and
product. Yield was determined by gravimetric assessment of the
evaporated samples and comparison against authentic sample.
1
2419; (c) Y. Boutadla, O. Al-Duaji, D. L. Davies,
G. A. Griffith and K. Singh, Organometallics, 2009, 28,
33–440.
Acknowledgements
4
6
(a) J. F. Hull, D. Balcells, J. D. Blakemore, C. D. Incarvito,
O. Eisenstein, G. W. Brudvig and R. H. Crabtree, J. Am.
Chem. Soc., 2009, 131, 8730–8731; (b) J. D. Blakemore,
N. D. Schley, D. Balcells, J. F. Hull, G. W. Olack,
C. D. Incarvito, O. Eisenstein, G. W. Brudvig and
R. H. Crabtree, J. Am. Chem. Soc., 2010, 132, 16017–
We would like to thank AstraZeneca, Pfizer, Reaxa and the
Technology Strategy Board for funding (TP14/HVM/6/
I-BD368B) (MVR & LG) and the Royal Society for a URF (IRB).
Notes and references
16029.
1
For reviews on transition metal hydrogen transfer catalyzed
reactions see: (a) J. S. M. Samec, J.-E. Bäckvall,
7 (a) F. Svec and J. M. J. Frechet, Chem. Mater., 1995, 7,
707–715; (b) F. Svec and J. M. J. Frechet, Science, 1996, 273,
1776 | Org. Biomol. Chem., 2015, 13, 1768–1777
This journal is © The Royal Society of Chemistry 2015

View Article Online
Organic & Biomolecular Chemistry
Paper
2
05–211; (c) E. C. Peters, F. Svec and J. M. J. Frechet, Adv. 10 A. Núñez, A. M. Cuadro, J. Alvarez-Builla and J. J. Vaquero,
Mater., 1999, 14, 1169–1181. Org. Lett., 2007, 9, 2977–2980.
(a) M. Baumann, I. R. Baxendale, S. V. Ley, N. Nikbin and 11 A. J. Blacker, St. Brown, B. Clique, B. Gourlay, C. E. Headley,
8
C. D. Smith, Org. Biomol. Chem., 2008, 6, 1587–1593;
b) K. A. Roper, H. Lange, A. Polyzos, M. B. Berry,
S. Ingham, D. Ritson, T. Screen, M. J. Stirling and
R. G. Thompson, Org. Process Res. Dev., 2009, 13, 1370–1378.
(
I. R. Baxendale and S. V. Ley, Beilstein J. Org. Chem., 2011, 12 K.-I. Fujita, T. Yoshida, Y. Imori and R. Yamaguchi, Org.
, 1648–1655; (c) C. J. Smith, N. Nikbin, C. D. Smith, Lett., 2011, 13, 2278–2281.
S. V. Ley and I. R. Baxendale, Org. Biomol. Chem., 2011, 9, 13 The random sequence was generated using http://www.
927–1937; (d) H. Lange, M. J. Capener, A. X. Jones, random.org/sequences/.
C. J. Smith, N. Nikbin, I. R. Baxendale and S. V. Ley, Synlett, 14 We were unable to achieve full conversion with this sub-
7
1
2
011 (6), 869–873; (e) R. J. Ingham, E. Riva, N. Nikbin,
I. R. Baxendale and S. V. Ley, Org. Lett., 2012, 14, 3920–
923.
L. Kiwi-Minsker, Chimia, 2002, 56, 143–147.
strate even under batch conditions. Spectral analysis of the
starting material along with m.p. and elemental analysis
indicated it was of >99% purity. We can not explain this
result at this time.
3
9
This journal is © The Royal Society of Chemistry 2015
Org. Biomol. Chem., 2015, 13, 1768–1777 | 1777