Cleaner continuous photo-oxidation using singlet oxygen in supercritical carbon dioxide

Full text of DOI:10.1002/anie.200901731

Communications
DOI: 10.1002/anie.200901731
Clean Oxidation
Cleaner Continuous Photo-Oxidation Using Singlet Oxygen in
Supercritical Carbon Dioxide**
Richard A. Bourne, Xue Han, Martyn Poliakoff,* and Michael W. George*
1
Singlet oxygen, O , is widely used in contemporary organic
14.7 MPa and 314 K) in supercritical carbon dioxide
2
[
1] 1
[2]
[13–16]
[17]
1
synthesis. O , is also an efficient electron acceptor leading
to the formation of superoxide radical ion O C . Both O and
O C have been implicated in the biosynthesis of some natural
(scCO ).
Recently we demonstrated how O can be
2
2
2
À
1
used in homogeneous reactions by using a perfluorinated
2
2
À
photosensitizer which is soluble in scCO . In addition, both
2
2
products and in biochemical processes and harnessing singlet
oxygen to aid in the construction of the complex molecular
motifs found in natural products is a subject of current
the non-flammability of CO and high solubility of O in the
2
2
[
18,19]
compressed-gas solvent
give potential advantages. The
complete miscibility of O with CO allows the system to be
2
2
[3]
research. Two of the most convenient and “clean” methods
safely kept below the explosion limits, significantly reducing
for introducing oxygen into hydrocarbon substrates involve
the risks of the reaction. scCO also has much lower viscosity
2
1
reactions of O either with alkenes, leading to allylic alcohols,
and higher diffusivity than traditional solvents thus over-
coming mass-transfer limitations that are found in traditional
2
or with 1,3-dienes leading to endoperoxides. Both of these
synthetic transformations have been used as pivotal steps in a
[
20]
multiphasic systems.
[4]
number of natural-product syntheses and also in the syn-
Using the batch synthesis of ascaridole (2) from a-
terpinene (1) as an example, we monitored the kinetics
[
5]
thesis of perfume ingredients. Furthermore, the addition of
O2 is 100% atom efficient, unlike more conventional
oxidations where half of the oxygen atoms are usually
wasted, often in the form of H O or CO , although, some of
[17]
using in situ FTIR which demonstrated that scCO can give
2
improved performance over CCl , with the reactions being at
4
least two-times faster in scCO (Scheme 1). However, in the
2
2
2
[17]
this atom efficiency may be lost in subsequent steps.
interests of safety, our high-pressure experiment
was
1
The most common method used to generate O in the
carried out on a very small scale with no more than 30 mL
2
laboratory, is by photoexcitation of molecular oxygen, gen-
of product being isolated in any experiment.
[
6] 1
erally in the presence of a photosensitizer. O can also be
2
generated thermally with sodium hypochlorite and hydrogen
[
7]
[8]
[9]
peroxide or from Ph PO3 or oxone. The widespread
3
application of photochemistry on a large scale in industry has
[
10]
not yet proved feasible. There are severe scale-up limita-
tions with photochemical reactors and new reactor designs
are required to overcome the decrease in photochemical
reaction efficiency as the concentration of reagents is
increased to the high-concentration conditions compatible
Scheme 1. Photo-oxidation of a-terpinene to ascaridole;
TPFPP=5,10,15,20-tetrakis(pentafluorophenyl)porphyrin.
with industrial-scale processes. The higher reactivity and
1
relatively short lifetime of O gives rise to a serious technical
Clearly any useful process in scCO will require scaling up
2
2
problem; namely a solvent is required which is non-flamma-
and scaling up-high pressure systems presents several techni-
cal challenges. The first is to increase the volume of the
reaction vessel without increasing the optical pathlength
unduly and thereby avoiding “inner filter” effects which
would prevent the bulk of the fluid from reacting photo-
chemically. More generally, the cost of high-pressure vessels
themselves becomes prohibitive as they are made larger. Such
considerations have dictated that scale-up of supercritical
processes usually involves a switch from batch to continuous
1
ble and which is relatively slow in relaxing the O back to its
2
triplet ground state. Traditionally CCl has been the solvent of
4
choice but it is no longer acceptable from an environmental
standpoint. Other solvents, such as iso-propanol, have been
[11,12]
used
but there are problems of potential flammability.
Experiments have demonstrated that it is possible to
1
generate O2 with a reasonably long lifetime (5.1 ms at
[
21]
[*] Dr. R. A. Bourne, X. Han, Prof. M. Poliakoff, Prof. M. W. George
processing.
School of Chemistry, University Park
There have been a number of reports of using continuous
1
University of Nottingham, NG7 2RD (UK)
E-mail: martyn.poliakoff@nottingham.ac.uk
mike.george@nottingham.ac.uk
photochemical reactors for O both in conventional glass-
2
[22–24]
ware and microreactors.
Although relatively efficient in
terms of conversion, these reactors have been rather low in
their overall productivity (which, for example, is only in the
[
**] We thank the EPSRC (EP/FO15275) for financial support and Prof.
G. Pattenden, Dr. A. Wells, and Dr. S. K. Ross for helpful
discussions. We also thank M. Dellar, M. Guyler, D. Lichfield, R.
Wilson, and P.Fields for technical support. M.W.G. gratefully
acknowledges receipt of a Wolfson Merit Award from The Royal
Society.
À1
À1
mmoll min range) and require relatively high electrical
power consumption for the lamps.
Our group has pioneered the use of small-scale continu-
ous photochemical reactors for the synthesis of organome-
5
322
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 5322 –5325

Angewandte
Chemie
tallic compounds in scCO and other supercritical sol-
and not obscured by the mounts. Our design differs from
2
[18,25,26]
[32]
vents.
reactors
We have also developed larger scale continuous
more conventional sapphire tubular view cells
which
[
27–29]
for heterogeneous catalysis in scCO , culminat-
normally have much greater diameter with thicker walls
which necessitates long optical paths and hence the possibility
of inner filter effects. The internal diameter of our tube is
7.8 mm, the wall thickness is 1.2 mm, and the total volume is
5.7 mL of which the irradiated volume is 4.1 mL. In its present
configuration the cell is unheated because a sufficient
temperature rise is caused by the LEDs. We use eight
composite 1000 lumen white LEDs (OSTAR; Part Code:
LEUWE3B-PZQZ-4C8F), each of which is composed of a
cluster of six individual diodes; the LEDs are mounted as two
arrays of four LEDs on commercial aluminum heat sinks. On
the scale of our apparatus, the LEDs are effectively point light
sources with a divergence of 1308. This arrangement means
that when the LEDs are mounted approximately 0.5 cm from
the sapphire cell that the majority of the light passes through
the cell. With all eight LEDs running at full power, the
internal temperature of the sapphire cell can rise to as high
708C; therefore modest cooling is achieved by blowing air
over the cell using a commercial ventilation fan (Bion-
2
ing in the construction of a full-scale 1000 tons per annum
[21]
chemical plant. Herein we describe a continuous reactor for
1
the photochemical reactions of O2 which considerably
improves efficiency compared to reactors for conventional
solvents. Our reactor has two innovative features: 1) a very
simple tubular sapphire reactor which provides a relatively
long irradiation zone and 2) illumination by high-power
visible-light-emitting diodes (LED), such as used for auto-
mobile headlights, the small size of which permit efficient
illumination of the cell without any additional optics. Herein,
we describe our continuous-flow reactor, illustrate its oper-
ation with a successful 3000-times scale-up of our synthesis of
2, apply the reactor to the photo-oxidation of citronellol (3),
the first stage in the synthesis of Rose Oxide (8), and show
that the combination of our reactor and the LEDs can provide
a substantial improvement in efficiency compared to conven-
tional solvents.
Safety warning: the reactions described involve high
pressures and should only be carried out in an apparatus
with the appropriate pressure rating and with due regard to the
aire B299 fanheater).
1
It is important to stress that the generation of O only
2
potentially explosive reaction between O and organic com-
pounds.
requires visible light. The efficiency of current LEDs is much
higher in the visible region than in the UV. Therefore it is
possible to achieve high “wall plug” efficiency with visible
LEDs; indeed the efficiency could be increased further by
using narrowband LEDs with a wavelength tuned to the
specific photosensitizer being used. However, we have chosen
to use broadband visible LEDs so that we can change
photosensitizer without modifying the photolysis source.
The sapphire cell is mounted in the continuous-flow
2
The essential component in our approach to scale up is the
photolysis cell, shown in Figure 1, which is based on a
[
30,31]
miniature sapphire view cell which was successfully used
for the supercritical synthesis for the series of new gas solvates
of C60 including C (CO ) , C (C H ) , and C (C H ) .
60
2
x
60
2
4
x
60
2
6 x
The cell consists of a sapphire tube sealed into a stainless
steel holder by O-ring seals which can slide to accommodate
differential thermal expansion. The advantage of this arrange-
ment is that a relatively narrow sapphire tube can be used
with a large proportion of its length available for irradiation
system (Figure 2). In a typical experiment, liquid CO was
2
À1
pumped at 2.0 mLmin (pump head at À108C, 48 bar), the
organic reactant containing the photosensitizer was pumped
À1
at 0.2 mLmin , and O was dosed in (see Figure 1) at a rate of
2
two molar equivalents of O to organic reactant.
2
We have proved the principle and operation of our reactor
by performing the oxidation of 1 to 2 (Scheme 1), the reaction
that we have previously carried out under batch conditions in
3
a cell of total volume of 2 cm . The photosensitizer,
5
,10,15,20-tetrakis(pentafluorophenyl)porphyrin (TPFPP), is
soluble in both 1 and 2 and all three components are soluble in
[
17]
the scCO /O mixture. Quantitative conversion could be
2
2
achieved in a single pass through our reactor with flow rates of
À1
substrate up to 0.2 mLmin . The reactor operated with
unchanged efficiency over 8 h without any noticeable fouling
of the sapphire tube. This feature is particularly important as
fouling of windows is often a significant problem in many
[33]
photochemical reactions. This 8 h run yielded 96 mL of 2
1
with less than 0.5% 1 as shown by H NMR spectroscopy. This
result represents a ꢀ 3000 scale-up of our original batch
reaction.
As a more stringent test of our flow reactor, we have
studied the photo-oxidation of citronellol (3), a key step in the
synthesis of Rose Oxide (8), a fragrance of commercial value
Figure 1. a) Schematic diagram of the sapphire tube reactor (R),
shown in its reactor housing. The sapphire tube is held in place by a
pair of EPDM rubber O rings, and a 1/8 inch Autoclave Engineer fitting
allows connection to the apparatus. b) Diagrammatic view of the LED
irradiation of the sapphire tube reactor with two LED arrays (L), each
containing four 1000 lumen LEDs positioned approximately 5 mm
from the sapphire tube. Each array is mounted on an aluminum heat
sink (HS), cooled by two 5 cm diameter fans (not shown).
(Scheme 2).
This reaction presents a number of challenges to our
approach namely 1) the sensitizer TPFPP is insufficiently
Angew. Chem. Int. Ed. 2009, 48, 5322 –5325
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5323

Communications
[
35]
and therefore using O represents a safety hazard; N has
2
2
very similar properties to O in terms of phase behavior at
2
room temperature and is an acceptable substitute for O . Our
2
phase measurements on 3 + DMC + TPFPP + scCO + N ,
2
2
demonstrate that a single-phase mixture can be obtained
under the reaction conditions (Figure 3). Similar experiments
with the hydroperoxide products confirm that they are indeed
insoluble possibly because of increased hydrogen bonding but
our continuous reactor can cope with biphasic reaction
mixtures provided it is operated in a downflow configuration.
The next stage was to carry out a batch reaction to
establish the reaction kinetics using a single white LED as the
photolysis source. In the absence of suitable IR bands we
monitored the reaction by the pressure drop as the O was
2
consumed. Figure 4 shows that the pressure falls linearly with
time indicating a zero-order reaction, as might be expected if
the rate determining step is the absorption of light by TPFPP.
1
The reaction of O with 3, is relatively unselective and gives
2
an almost equimolar mixture of 4 + 5 which can be converted
into 8 in the subsequent work-up.
Figure 2. The photochemical flow system. CO is delivered from a
2
chilled Jasco PU-1580-CO2 pump. O is added at a measured rate by a
2
Rheodyne dosage unit. The organic substrate and photosensitizer
We repeated the reaction using our continuous-flow
À1
solution are pumped using a Jasco PU-980 HPLC pump. The CO and
2
reactor. The single-pass conversion of 3 at 0.1 mLmin
O are passed through a mixer (M1; 5 cm 1/4 inch SS316 tubes at
À1
2
(1:1 v:v DMC:3), 1.0 mLmin of CO , 2 equivalents of O
2
2
5
08C) and then mixed with the substrate stream (subs), which passes
at 180 bar was again 100% with 52% selectivity to 4 and 48%
selectivity to 5. The reaction was performed for approxi-
mately 4 h converting 12 mL of 3. Since 4 and 5 are
considerably less stable than ascaridole, 2, which is itself
potentially explosive, we carried out the majority of our
experiments by collecting the products directly into an
aqueous solution of Na SO . This approach ensured that 4
through a second mixer (M2). The mixed stream then passes into the
sapphire tube reactor (R), and then flows out continuously through a
Jasco BP-1580-81 back-pressure regulator (BPR). The material is then
collected in a glassware flask (F), with attached condenser (C).
soluble in pure 3 to be used in the same way as with 1 and
2
3
2
) the hydroperoxide products, 4 and 5 are virtually immis-
and 5 were converted into 6 and 7 as soon as they were
delivered from the reaction system. The co-solvent DMC is
completely soluble in water and so it did not present a
problem. In one experiment, we tried running the reactor in
cible with scCO . The solubility problem with TPFPP was
overcome by addition of a 1:1 v:v of dimethyl carbonate
2
(
DMC), to 3. DMC is a relatively highly oxidized compound
[
34]
which we have used for a range of reactions in scCO , in
addition DMC does not react with O .
The first stage of our investigation was phase-behavior
studies both of the continuous reaction mixture and the
products. The volume of our phase analyzer is relatively large
upflow but the immiscibility of 4 and 5 with scCO led to the
2
2
1
accumulation of liquid in the sapphire tube; therefore all
subsequent experiments were performed in downflow. It is
important to stress that any cloudiness caused by incipient
2
Figure 3. Photographs showing the phase behavior at 558C of a
mixture containing 3 (5.23 mmol) and DMC (0.0113 mmol) with a
mixture of scCO (0.432 mol) and N (10.5 mmol) using a variable
2
2
[
35]
volume view cell a) 120 bar: multiphase mixture, b) 140 bar: single
phase, c) 180 bar: also single phase, d) separate experiment with the
product mixture (replacing 3 with the 100% converted mixture of
peroxide products in the same molar ratios). Note the droplets
Scheme 2. Reaction pathway for the conversion of 3 into 8. Photo-
oxidation of 3, subsequent reduction to diol products 6 and 7, and
then acidification to yield Rose Oxide (8). DMC=dimethyl carbonate.
indicating a biphasic mixture with scCO under all conditions studied
2
(40–608C, 100–200 bar).
5
324
www.angewandte.org
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2009, 48, 5322 –5325

Angewandte
Chemie
[
[
1] E. L. Clennan, A. Pace, Tetrahedron 2005, 61, 6665 – 6691.
2] I. Saito, T. Matsuura, K. Inoue, J. Am. Chem. Soc. 1983, 105,
200 – 3206.
3] T. Montagnon, M. Tofi, G. Vassilikogiannakis, Acc. Chem. Res.
008, 41, 1001 – 1011.
3
[
2
[
[
[
4] N. Hoffmann, Chem. Rev. 2008, 108, 1052 – 1103.
5] N. Monnerie, J. Ortner, J. Sol. Energy Eng. 2001, 123, 171 – 174.
6] M. C. DeRosa, R. J. Crutchley, Coord. Chem. Rev. 2002, 233,
3
51 – 371.
7] C. S. Foote, S. Wexler, W. Ando, R. Higgins, J. Am. Chem. Soc.
968, 90, 975 – 981.
8] P. D. Bartlett, G. D. Mendenhall, D. L. Durham, J. Org. Chem.
980, 45, 4269 – 4271.
9] M. C. Carreꢁo, M. Gonzalez-Lopez, A. Urbano, Angew. Chem.
[
[
[
1
Figure 4. Monitoring the batch photo-oxidation of 3 (0.328 mmol) and
DMC (0.709 mmol) with O (0.595 mmol) in scCO (2067 psi) at 408C.
The batch cell was irradiated using a single white 1000 lumen LED.
1
2
2
2006, 118, 2803 – 2807; Angew. Chem. Int. Ed. 2006, 45, 2737 –
The pressure drop corresponds to one molar equivalent of O being
2
2741.
consumed for each mole of 3 present.
[
[
10] K. H. Pfoertner, J. Photochem. Photobiol. A 1990, 51, 81 – 86.
11] M. Oelgemꢂller, N. Healy, L. de Oliveira, C. Jung, J. Mattay,
Green Chem. 2006, 8, 831 – 834.
12] M. Oelgemꢂller, C. Jung, J. Ortner, J. Mattay, E. Zimmermann,
Green Chem. 2005, 7, 35 – 38.
phase separation in the reactor did not appear to cause any
reduction in the photochemical efficiency. Subsequent acidifi-
cation of the aqueous phase with HCl was performed
overnight and the organic products were extracted with n-
hexane. GC-MS analysis of the organic phase showed
selectivity of 97.6% 8 (mixture of cis and trans), 2.2% 6 + 7,
and 0.2% 3.
[
[13] A. A. Abdel-Shafi, F. Wilkinson, D. R. Worrall, Chem. Phys.
Lett. 2001, 343, 273 – 280.
[
14] A. A. Abdel-Shafi, D. R. Worrall, J. Photochem. Photobiol. A
007, 186, 263 – 269.
2
[
[
15] M. Okamoto, T. Takagi, F. Tanaka, Chem. Lett. 2000, 1396 – 1397.
16] D. R. Worrall, A. A. Abdel-Shafi, F. Wilkinson, J. Phys. Chem. A
One of the reasons for choosing 3 is that fairly detailed
data are available for the efficiency of the photochemistry in a
mulitpass continuous microreactor compared to a conven-
2001, 105, 1270 – 1276.
[17] R. A. Bourne, X. Han, A. O. Chapman, N. J. Arrowsmith, H.
[
24]
tional Schenk reactor under LED irradiation. According to
this analysis, the conventional Schenk reactor with an
Kawanami, M. Poliakoff, M. W. George, Chem. Commun. 2008,
4457 – 4459.
[
[
18] J. A. Darr, M. Poliakoff, Chem. Rev. 1999, 99, 495 – 541.
19] M. C. McHugh, V. J. Krukonis, Supercritical Fluid Extraction:
Principles & Practice, Butterworth-Heinemann, Boston, 1994.
20] E. J. Beckman, J. Supercrit. Fluids 2004, 28, 121 – 191.
[21] P. Licence, J. Ke, M. Sokolova, S. K. Ross, M. Poliakoff, Green
Chem. 2003, 5, 99 – 104.
22] A. A. Lapkin, V. M. Boddu, G. N. Aliev, B. Goller, S. Polisski, D.
Kovalev, Chem. Eng. J. 2008, 136, 331 – 336.
23] E. E. Coyle, M. Oelgemoller, Photochem. Photobiol. Sci. 2008, 7,
immersed LED array gave
0
a
space–time yield of
.1 mmoll min . By contrast a LED illuminated continu-
ous microreactor with a recirculating reaction mixture of
À1
À1
[
1
0
0 mL (0.1m of 3 in ethanol) gave a space–time yield of
.9 mmoll min ; that is, nine-times higher than the Schenk
À1
À1
[
[
[
reactor. In our system, we have a single pass reactor which is
converting 0.27 mmol of 3 per minute with 140 W of electrical
input giving
a
space–time yield of approximately
1313 – 1322.
À1
À1
7
0 mmoll min which is nearly two orders of magnitude
24] S. Meyer, D. Tietze, S. Rau, B. Schaefer, G. Kreisel, J. Photo-
chem. Photobiol. A 2007, 186, 248 – 253.
[25] J. A. Banister, P. D. Lee, M. Poliakoff, Organometallics 1995, 14,
higher than the previous microreactor. We can also compare
our a-terpinene results with a microfabricated nanoreactor
with a 20 W tungsten lamp which gave 80% conversion in a
3876 – 3885.
À1
À1 [24]
[
[
[
26] J. A. Banister, S. M. Howdle, M. Poliakoff, J. Chem. Soc. Chem.
Commun. 1993, 1814 – 1815.
27] P. Clark, M. Poliakoff, A. Wells, Adv. Synth. Catal. 2007, 349,
single pass at a flow rate of 0.0002 mmol min .
By
À1
À1
contrast we achieved 100% conversion at 1.24 mmol min
an improvement of approximately 6000 times.
2655 – 2659.
Herein, we have described an innovative continuous
28] R. A. Bourne, J. G. Stevens, J. Ke, M. Poliakoff, Chem. Commun.
2007, 4632 – 4634.
1
photocatalytic reactor for performing reactions of O2 in
scCO . The reactor has demonstrated the potential of using
[29] M. G. Hitzler, M. Poliakoff, Chem. Commun. 1997, 1667 – 1668.
[30] A. OꢃNeil, C. Wilson, J. M. Webster, F. J. Allison, J. A. K.
Howard, M. Poliakoff, Angew. Chem. 2002, 114, 3950 – 3953;
Angew. Chem. Int. Ed. 2002, 41, 3796 – 3799.
2
LEDs for performing synthetic photochemistry in a contin-
uous milliliter-scale reactor with demonstrably higher space–
time yields than have been previously observed. This has been
accomplished by the combination of high-power LED
[
31] C. N. Field, P. A. Hamley, J. M. Webster, D. H. Gregory, J. J.
Titman, M. Poliakoff, J. Am. Chem. Soc. 2000, 122, 2480 – 2488.
[32] C. M. M. Duarte, M. Crew, T. Casimiro, A. Aguiar-Ricardo,
M. N. da Ponte, J. Supercrit. Fluids 2002, 22, 87 – 92.
technology with a high pressure scCO reactor system capable
2
of supporting high concentrations of O with negligible mass-
2
transfer limitations.
[33] A. M. Bruun, M. T. Muurette, E. Oliveros, Photochemical
Technology, Wiley, Chichester, UK, 1991.
[
[
34] P. N. Gooden, PhD thesis, University of Nottingham, UK, 2006.
35] P. Licence, M. P. Dellar, R. G. M. Wilson, P. A. Fields, D.
Litchfield, H. M. Woods, M. Poliakoff, S. M. Howdle, Rev. Sci.
Instrum. 2004, 75, 3233 – 3236.
Received: March 31, 2009
Published online: June 12, 2009
Keywords: oxidation · photochemistry · singlet oxygen ·
.
supercritical fluids · sustainable chemistry
Angew. Chem. Int. Ed. 2009, 48, 5322 –5325
ꢀ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5325