Flow Photochemistry PDF

Flow photochemistry: Old light through new windows
Jonathan P. Knowles, Luke D. Elliott and Kevin I. Booker-Milburn*
Review Open Access
Address: Beilstein J. Org. Chem. 2012, 8, 2025–2052.

School of Chemistry, University of Bristol, Cantock’s Close, Bristol, doi:10.3762/bjoc.8.229
BS8 1TS, UK.
Received: 31 August 2012
Email: Accepted: 29 October 2012
Kevin I. Booker-Milburn* - [email protected] Published: 21 November 2012
* Corresponding author Associate Editor: C. Stephenson
Keywords: © 2012 Knowles et al; licensee Beilstein-Institut.

cycloaddition; flow chemistry; photocatalysis; photochemistry; License and terms: see end of document.
photooxygenation
Abstract
Synthetic photochemistry carried out in classic batch reactors has, for over half a century, proved to be a powerful but under-
utilised technique in general organic synthesis. Recent developments in flow photochemistry have the potential to allow this tech-
nique to be applied in a more mainstream setting. This review highlights the use of flow reactors in organic photochemistry,
allowing a comparison of the various reactor types to be made.
Introduction
The use of ultraviolet light to carry out bond-forming reactions the application of photochemistry in synthesis were reported.
in synthetic organic chemistry has a long history dating back to Eaton's cubane synthesis [4], Corey's synthesis of carophyllene
the mid-19th century. The observation by Trommsdorff [1] that alcohol [5,6] and Wender's synthesis of cedrene [7] are just
crystals of the sesquiterpene santonin would literally burst open three outstanding examples to highlight. Photochemistry has
upon exposure to sunlight can perhaps be considered as the also made the transition to industrial-scale synthesis. For
beginning of organic photochemistry. In 1883 Cannizzaro and example the Toray process [8-10] for the synthesis of capro-
Sestini [2] investigated this further and reported the formation lactam, used to manufacture Nylon 6, proceeds by irradiation of
of photosantonic acid upon irradiation of santonin. It is general- cyclohexane with NOCl and HCl, and is carried out in dedic-
ly regarded that the systematic and ground-breaking investi- ated plants producing >100,000 tons per annum.
gations of Ciamician and Silber [3] paved the way for modern
synthetic photochemistry. At the turn of the 20th century they Conventional techniques & equipment
described the first examples of now common reactions such as For well over half a century the most dependable apparatus for
intramolecular [2 + 2] cycloaddition; basic ketone photochem- laboratory scale organic photochemistry has been the immer-
istry such as α- and β-cleavage; as well as fundamental concepts sion-well photoreactor in conjunction with mercury-vapour-
such as the singlet and triplet states and n,π* and π,π* excited discharge lamps (Figure 1). This compact batch reactor is an
states. From the late 1950s onwards thousands of examples of excellent device to carry out preparative photochemistry on
2025
Beilstein J. Org. Chem. 2012, 8, 2025–2052.
scales of milligrams up to a few grams. The lamp is contained range 6–16 W are low-priced devices and are generally very
in a double-jacketed water-cooled immersion well. This is then efficient in their conversion of input power into UV (30%).
placed into a reaction flask containing the chromophoric sub- Uncoated lamps emit the bulk (90%) of their spectral output at
strate. This flask is usually standard Pyrex glassware. The solu- 254 nm (UVC) and are particularly suited to carbonyl and arene
tion is normally degassed to remove oxygen in order to photochemistry as well as halogenation chemistry. These lamps
diminish the possibility of quenching and other reactions, such are also available with a range of phosphor coatings to emit
as peroxide formation (conveniently achieved with a long both UVB and UVA radiation. They have found commercial
needle and nitrogen stream). For safety, the cooling water use in medical, tanning and insect-attraction applications. These
should be connected to a flow sensor in order to shut down the lamps have very long lifetimes, often in excess of several thou-
lamp should the water pressure drop. The whole apparatus can sand hours.
be shielded in a cabinet to avoid exposure to powerful UV radi-
ation. Alternatively aluminium foil can be wrapped around the • Medium pressure: These are much higher power lamps of
glassware to achieve a very effective level of shielding. UV input powers ranging from 125 W up to large 60 kW lamps for
goggles/visor should be worn if unshielded apparatus is oper- industrial purposes (e.g., the Toray process). In standard labora-
ated, and especially during sampling of the reaction mixture. tory use, lamps of 125 and 400 W are the most common. These
Once all this is in place the lamp is then switched on and the lamps are broadband emitters with the most powerful UV
progress of the photochemical reaction can be monitored by output in the 300–370 nm region. Strong emissions in the IR
conventional means (TLC, GC, LCMS). As often no reagents region account partly for their high operating temperatures
are used, workup is by simple evaporation of solvent, and the meaning they must be used in an appropriate water-cooled
product is purified by conventional means. immersion-well apparatus. They usually have reliable lifetimes
of a few hundred hours, which can be extended considerably if
left on. They can be used for general-purpose photochemistry
and are particularly suited for chromophores absorbing strongly
in the 290–400 nm region.
The glassware used for the immersion-well is particularly

important as it functions as a useful filter with medium pres-
sure lamps: quartz is essentially transparent from 200 nm to
visible; Vycor >240 nm; Pyrex >300 nm; uranium glass
>350 nm. It should be noted that Vycor and uranium glass are
now difficult to source due to manufacturing issues. Often it is
more convenient to purchase a quartz immersion well and use a
glass filter, e.g., a tube of Pyrex can be placed between the lamp
and the inner wall of the immersion well, thus filtering out radi-
ation below 300 nm.
Finally, solvent choice is also a key factor. The solvent must be

Figure 1: An immersion-well batch reactor with 125 W medium pres- able to dissolve a range of different substrates but must not be a
sure Hg lamp.
strong UV absorber itself. Other factors to consider are that the
solvent should not undergo quenching or hydrogen-atom
The most common UV sources are the commercially available abstraction or other reactions with the excited state (although
mercury-discharge lamps. These are evacuated glass tubes some solvents can be useful sensitisers, e.g., acetone). Acetoni-
containing mercury vapour through which an electrical trile has proved to be a particularly versatile solvent as it is
discharge passes. This results in the excitation of the Hg atoms economical, good at dissolving polar substrates, does not absorb
and a subsequent emission of UV radiation. The two most above 200 nm, and is easy to remove on a rotary evaporator.
common Hg lamps are
Other reactor systems have been developed over the years and
• Low pressure: These are similar to everyday fluorescent include
lamps, and input powers range from 6 up to 300 W and above.
However, the latter are generally very large (1–2 metres in • Multiple lamp or “Rayonet” reactors. This is a cabinet where
length) and not suited to general laboratory use. Lamps in the multiple lamps direct their radiation towards a sample at the
2026
centre. There is usually a fan located at the bottom of the One of the key issues of scaling-up organic photochemistry in
cabinet to ensure sufficient cooling. These are useful for scaling an immersion-well (batch) reactor is that light penetration to the
up batch reactions that use low-pressure lamps. surrounding solution is limited by the high absorption of the
substrate and falls off rapidly with distance from the lamp. This
• Falling-film reactors. A falling-film reactor is particularly effect is best explained by considering a few basic equations.
useful for scaling up the photolysis of a strongly absorbing The absorption of light by a solution (A) shows a linear relation-
chromophore. Here, a thin film of the substrate solution flows ship with the extinction coefficient (ε), the molar concentration
down a glass plate or tube in close proximity to the light source. (c) and the path length (I) as described by the Beer–Lambert
The short path length leads to very efficient irradiation. The law (Equation 1). The absorption, however, is expressed as a
downside is that residence time is short and often the solution logarithmic function of the ratio of transmitted light (I) to
has to be recirculated, leading to the possibility of side reac- incident light (I 0 ) (Equation 2).
tions. Nonetheless, this has proved to be a valuable device in
the right circumstances. For example, Griesbeck [11] reported (1)
the design of a particularly useful version in conjunction with a
high power 308 nm XeCl excimer source. (2)
Why then, with this wealth of useful reactions and established

techniques at hand, do mainstream organic chemists tend to An absorption of 1 therefore represents a situation where 90%
avoid photochemistry as a routine synthetic tool? of light is absorbed (I = 0.1 I0). For example, a weak π, π*
absorption such as the forbidden band of benzene at 254 nm has
There are a number of likely contributing factors: an extinction coefficient of about 200 M −1 cm −1 . The path
length required for a solution of a modest concentration of
• Equipment. The first time user is often confronted by a lack of 0.05 M to absorb 90% of the incident light will be just 0.1 cm,
suitable equipment or know-how, e.g., which lamp, glassware or 1 mm. The transmission profile for such a solution is shown
and solvent to use. in Figure 2.
• Safety. Medium pressure mercury lamps operate optimally at

~600 °C and emit intense and potentially damaging UV radi-
ation.
• Difficulty in scaling up. The Toray caprolactam process

proves this is not a problem at the industrial scale. However, in
the lab it is often very difficult to scale up above a few grams in
a classic immersion-well reactor (see below).
It is perhaps then not surprising that synthetic chemists, as

potential first-time users, avoid this medium. This in turn leads
to the more fundamental problem: synthetic chemists do not
generally think photo-retrosynthetically. As a result potentially
Figure 2: Transmission profile of a 0.05 M solution, ε = 200 M−1 cm−1.
shorter and more efficient synthetic routes to complex organic
molecules, as well as access to new molecular space have long
been avoided by mainstream synthetic chemists. A more typical π, π* transition may have an extinction coeffi-
cient of about 20,000 M−1cm−1 and in this case a 0.05 M solu-
Flow to the rescue? tion will absorb 90% if incident light at a distance of 0.01 mm.
Over the last 15 years flow chemistry has begun to make a
major impact in the way many organic chemists perform syn- Essentially the reaction solution (photolysate) nearest the lamp
thesis. The pioneering work of Ley [12-14] and others [15] has “screens” the bulk of the reaction solution from UV. This effect
demonstrated that complex organic molecules can be is also amplified if the reaction solution is concentrated. When
constructed continuously in well-designed multireactor systems the scale of the reaction is increased with the same lamp it
linked in sequence and under precise software control. Nearly becomes increasingly more difficult to drive the reaction to
all common batch reactions [16] can now be carried out in flow. completion. Attempts to do so often result in the “curse” of syn-
2027
thetic photochemistry, i.e., over-irradiation of the product and highly engineered glass and metal systems. They are generally
formation of side products and photopolymers. defined as having channels less than 1 mm in thickness and
typical throughput flow rates range from a few microliters up to
This is where flow photochemistry becomes a very attractive 1 mL per minute. A syringe pump is ideally suited to delivering
proposition, as in principle it can overcome all the key prob- solutions to these reactors (Figure 3).
lems of batch photochemistry in the laboratory.
• At any one time under flow conditions only a very small

amount of the total reaction solution “sees” intense UV irradi-
ation from the UV source. This leads to very efficient, uniform
irradiation of the whole reaction solution over time.
• The UV exposure time can be precisely controlled by the

flow-rate and reactor volume. This can address both the under-
and over-irradiation problems encountered with batch reactors.
• As a continuous flow device is scale independent, a single

reactor can in principle be used to process a few milligrams of
substrate up to nearly a kilogram per day (see macroreactors).
• Due to much shorter path lengths high concentration solutions

can be irradiated effectively.
• Large volumes of very low concentration solutions can be irra-

diated. This is particularly useful for reactions with competing
intermolecular side reactions, e.g., dimerisation and polymerisa-
tion. Figure 3: Schematic of a typical microflow photochemical reactor
(above) and detail of a triple-channel microflow reactor (below) used
for the photooxygenation of citronellol [20].
• The photolysate can be concentrated by continuous evapor-
ation and the solvent recycled with the starting material. This
can dramatically cut down the solvent footprint, particularly in • Macroflow. These devices generally involve UV-transparent
dilute reactions where large volumes of solvent would be tubing (>0.5 mm, i.d.) wrapped around a high-power UV source
required to process quantities of substrate. and have flow rates usually greater than 1 mL/min (Figure 4).
• Safety. By allowing the bulk solution to be kept remote from The primary purpose of any photochemical reactor is to allow a
the lamp, only a minimal amount of flammable solvent is near a solution to be irradiated by the emissions from a light source in
potential ignition source at any one time. a controlled manner. It is useful to consider a lamp as emitting a
flux of photons. An equation to calculate the number of moles
Historically, there have been a few reports of the applications of of photons (einsteins) per hour at a given wavelength (λ) if the
rudimentary flow techniques in photochemistry, such as the use total power of emissions at that wavelength is known, is shown
of a spiral glass reactor in vitamin D synthesis (1959) [17] and in Equation 3.
the use of coiled teflon tubing as a gas-phase reactor for the
synthesis of methyl chloride (1971) [18]. However, it was not (3)
until the turn of the 21st century that the application of flow
devices to synthetic photochemistry really started to grow. For
the purposes of this review, there are broadly speaking two This equation is particularly useful when considering selective
types of flow-reactor system that have been used for synthetic narrowband emitters such as a low-pressure mercury lamp or
photochemistry. LEDs. For example, a 15 W low-pressure lamp, if operating at
30% efficiency will have a total UV power of 4.5 W at 254 nm.
• Microflow [19]. These devices consist of fabricated micro- According to Equation 3 this corresponds to a photon flux of
channels and range from bespoke “lab-on-a-chip” designs to 34 mmol photons per hour. When multiplied by the quantum
2028
Figure 4: Schematic of a typical macroflow photochemical reactor (above) and images of the FEP photochemical flow reactor developed by Booker-
Milburn and Berry [21].
yield of a reaction driven by 254 nm UV, a figure is obtained strongly absorbing solutions. Temperature control is also more
that represents the maximum theoretical productivity of the effective than in larger systems offering the possibility of
reaction if all the emitted photons are absorbed. studying photochemistry outside the normal range. When
coupled with online analysis, these reactors can potentially
The small aperture and limited channel coverage of a micro- enable the rapid screening of reactions and conditions for
flow reactor puts it at a distinct disadvantage in this aspect. The optimisation and discovery.
microchannels are milled or etched into a planar surface and
cannot efficiently capture the radial emission from most The most efficient method of capturing the maximum number
common light sources. Even LEDs, which can be considered as of photons, and hence to maximise productivity, is to construct
planar light sources, often have a beam width wider than the the reactor around the lamp. This approach has been met with
actual channels; however, this issue could easily be addressed great success in the field of macroflow photochemistry. These
by a more directed reactor design. As a result microflow reactors can easily be constructed, even by a novice, using
reactors are often inefficient at capturing light and suffer poor cheap, readily available materials. Essentially, all that is
productivity. required is some UV-transparent tubing, a pump and a lamp. A
water cooled jacket is not even required if a low pressure lamp
The main advantage at present of the microflow photochemical is used, due to the mild operating temperatures (40 °C).
reactors is the exquisite control over reaction conditions they
can offer. The precisely engineered channels can be made Such a reactor can make efficient use of the 15 W lamp
shallow enough to ensure uniform irradiation of concentrated or described earlier. For example, when 80% of the available
2029
photons are delivered to the solution to drive a reaction with a

quantum yield of just 10%, the productivity can be estimated as
2.7 mmol/h. The photons are acting as a “reagent” but not one
that can be added all at once; rather, they are introduced as a
stream (flux). As the reactant solution is also introduced as a
stream, the flow rate can be precisely tailored to match the
photon flux of the lamp and the quantum yield of the substrate.
The main aim of this review is to illustrate the advances made

in flow photochemistry over the last 10 years. This also
presented an opportunity to compare different reactor types in
Scheme 1: [2 + 2] photocycloadditions of enones with enol derivatives.
areas such as selectivity, yield and productivity [22].
Review flow system was lower than that in the batch reactor [25]. This
Microflow photochemistry was later shown to be due to the reversibility of the cycloaddi-
Photocycloadditions tion to yield 5, whilst the reaction to yield 6 is irreversible.
Although many photocycloadditions have been performed by Thus, attempts to achieve high conversions always increased
using other types of reactor, only [2 + 2] cycloadditions have the proportion of product 6 [26]. Use of the microreactor proved
been performed in microflow systems. The first of these was an to be an advantage in this situation as compound 5 was removed
intermolecular reaction between enones and vinyl esters or as it was formed and was not exposed to further irradiation. In
ethers (Scheme 1). By using a 300 W Hg lamp and a this way high selectivity (96:4) could be achieved in flow whilst
FOTURAN glass reactor the reaction gave moderate to high little selectivity (55:45) was achieved in batch for the same
yields, albeit with poor diastereoselectivity [23]. Subsequent conversion. Increasing the width of the channel in the flow
work demonstrated that the photon efficiency of the reaction reactor allowed flow rates to be increased whilst retaining this
could be improved through the use of a 15 W black light and a level of selectivity; however, productivity even with this larger
quartz-plate-covered microreactor [24]. Although faster reac- reactor was still extremely low (0.014 mmol/h) [26].
tion times are claimed for the microflow system as compared to
batch, it is probably unrealistic to compare reaction completion Asymmetric induction can be achieved in [2 + 2] cycloaddi-
times in two reactors whose volumes are so different. Material tions through the use of a chiral auxilliary. The [2 + 2] reaction
output per unit time provides a better comparison, and in this shown in Scheme 3 employed a chiral ester function to direct
case shows the batch reactor to be uncompetitive when the facial selectivity of the addition of the enone to cyclo-
compared to two microflow reactors connected in series pentene. Diastereoselectivity was found to be largely inde-
(0.35 mmol/h in flow versus 0.02 mmol/h in batch) [23]. pendent of solvent (DCM versus toluene) but was dependent on
temperature. Thus, cooling the reaction from 0 °C to −40 °C
Intramolecular [2 + 2] photocycloadditions have also been gave an increase in d.e. of compound 9 from 71% to 82%; this
performed by using microflow apparatus. Mizuno et al. reported also affected the selectivity of 9 versus 10, which changed from
the reaction shown in Scheme 2 using a Xe lamp (λ > 290 nm) 39:61 to 1:1 under the same conditions. Interestingly, the micro-
in a poly(dimethylsiloxane) reactor [25]. The microreactor was flow reactor gave better diastereoselectivity for compound 9
compared with a batch reactor and was found to give slightly than a batch reactor under the same conditions (82% d.e. versus
better selectivity (59:9 versus 56:17) for 4,5-fused system 5 72% d.e.). This effect was ascribed to the more efficient
over a 4,6-fused system 6. However, conversion in the micro- temperature control due to the smaller reactor volume. How-
Scheme 2: Competing reactions in an intramolecular [2 + 2] photocycloaddition.
2030
Scheme 3: Diastereocontrolled cycloaddition of a cyclic enone with cyclopentene.
ever, at 0.02 mmol/h the productivity of the microreactor was In a similar [2 + 2] photocycloaddition (Scheme 5), the authors
lower than would be expected of a batch reactor [27]. A similar, demonstrated that by extending the chain length by one carbon,
non-diastereoselective version of this reaction has been the 4,6-fused ring system 14 could be formed, albeit with some
performed by using a medium-pressure Hg lamp [28]. formation of regioisomer 15. In this instance the batch reaction
gave slightly better selectivity than that seen in the microflow
Another [2 + 2] reaction has been performed in the LOPHTOR system [29].
stainless steel channel reactor. The cycloaddition shown in
Scheme 4 was performed in this reactor and compared to the The Paterno–Büchi reaction has also been explored in a micro-
same reaction performed in a conventional batch reactor. flow setting (Scheme 6). A 15 W black light was again found to
Despite the increased yield and decreased irradiation time, be a more photon-efficient light source than a 300 W Hg lamp;
achieving a real comparison of the two reactor outputs is diffi- however, longer reaction times were required to achieve
cult given the lack of details regarding the batch process. The comparable yields with the lower-power light source. These
highest productivity from the microflow system was longer residence times further lowered the reactor productivity
0.22 mmol/h. It is hard to give a meaningful comparison with to 0.15 mmol/h [24].
the batch reaction as no scale is reported; however, it is possible
to state that the batch reaction would have had to be performed Photooxygenations
on a 17.6 mmol scale in the given time period to be competitive Direct oxygenation of organic molecules through the photo-
[29]. sensitised addition of singlet oxygen represents an atom-
Scheme 4: Comparison of yields and reaction times for a batch reactor with a microflow system.
Scheme 5: Intramolecular [2 + 2] photocycloaddition.
2031
Scheme 6: Paterno–Büchi reaction of benzophenone with an allylic alcohol.
Scheme 7: Photooxygenation of cyclopentadiene.
economic method of functionalisation and is employed in the Rose Bengal has also been employed as a sensitiser in a micro-
synthesis of a range of compounds of commercial interest chip reactor equipped with a 20 W tungsten lamp for the addi-
including fragrances and pharmaceuticals. One major disad- tion of singlet oxygen to α-terpinene to yield the anthelmintic
vantage of using this reaction on a large scale is the potential for asaridole (23, Scheme 8) [32]. Comparison of this microflow
fires or detonation of accumulating peroxide products. Other reaction to a batch reaction using a 500 W tungsten lamp
disadvantages include inefficient irradiation of bulk solutions, showed that although the microflow reaction gave a higher
which combined with the extremely short lifetime of singlet yield (85% versus 67%), the productivity of the flow reactor
oxygen means that lengthy irradiations are often required. was markedly lower (1.5 mg/h versus 175 mg/h). This high-
These issues can be overcome through the use of continuous- lights one common issue with moving to microflow photochem-
flow chemistry: reactions performed in this manner have only a istry: although yields may increase, productivity can be signifi-
small amount of oxygenated solvent and peroxide product cantly lower.
present at any one time, and this can be reduced immediately
upon leaving the reactor. The smaller reactor volumes involved
in flow chemistry also mean that irradiation is efficient and
hence that the singlet oxygen generated can react within its
short lifetime.
Falling-film microreactors have been employed in the Rose

Scheme 8: Preparation of the anthelmintic ascaridole 23.
Bengal-sensitised oxygenation of cyclopentadiene 19
(Scheme 7) [30,31]. Although this method allowed good
temperature control and the immediate quenching of the poten- A glass-loop microreactor was employed in the sensitised
tially explosive peroxide intermediate 20 as it formed, the oxygenation of (−)-β-citronellol (24) shown in Scheme 9, an
process was low yielding (20%), and whilst 0.95 g of product important reaction for the synthesis of the fragrance rose oxide
21 was produced the lack of reported details make it difficult to 27. It was shown that Rose Bengal was approximately twice as
say how scaleable this is likely to be. effective a sensitiser as Ru(t-bpy)3Cl2 when a 450 W Xe lamp
Scheme 9: Production of rose oxide 27 from (−)-β-citronellol (24).
2032
was employed. Through the use of an LED light source in the by using methylene blue as the sensitiser and a 16 W LED light
microreactor, the authors reported that the microflow system source, despite the reactor volume being only 285 µL [39]. This
was slightly superior to the batch-type Schlenk reactor in terms output was 2.6 times that of a 50 mL batch reactor. The same
of space–time turnovers and photon efficiency; however, the reactor was also successfully applied to the oxidation of allylic
required 400 min irradiation to achieve reasonable conversion alcohols for the synthesis of the antimalarial artemisinin, and
on a 1 mmol scale would likely cause difficulty in finding syn- the conversion of α-terpinene to ascaridole. Addition of a
thetic utility for the reactor [33,34]. It is also interesting to second oxygen-containing channel gave a triple channel reactor
compare the efficiency of the reactor with that of alternatives; that showed even higher space–time yields; however, flow rates
despite the reported high photon efficiency, an LCE compari- are not reported, making it impossible to calculate the
son [35] of six methods for the conversion of β-citronellol into productivity of the reactor [20].
rose oxide showed the above method to be less competitive,
mainly due to the long irradiation times causing high consump- Whilst yet to be employed in a synthetic context, it has been
tion of electrical power. It should be noted that an alternative shown that porous silica nanoparticles can effectively produce
photochemical oxidation using solar radiation was found to be singlet oxygen when irradiated with LEDs in the oxidative de-
highly efficient, second only to the industrial Dragoco protocol gradation of 1,3-diphenylisobenzofuran [40].
(mercury-arc lamp) [35].
Photocatalytic reactions
Microflow photochemistry can add a further advantage in Photocatalytic reactions are an area in which continuous flow
photooxygenation reactions through the use of dual-channel can be particularly advantageous due to the large surface-to-
reactors. With this type of reactor, oxygen is passed down a volume ratio ensuring efficient irradiation of the whole reaction
second channel, which runs parallel to the channel containing media; this is especially useful if the photocatalyst is solid
the reaction mixture, and a porous wall allows the diffusion of supported. The immobilised catalyst of choice is titanium
the oxygen into the reaction mixture. This method avoids the dioxide, both with and without Pt doping, due to its photochem-
need to oxygenate the solvent before injection, as well as issues ical stability. The photocatalyst can be effectively deposited
such as oxygen depletion in the reaction mixture during the onto a microreactor surface, thus eliminating the need for
reaction, and bubble formation. Such reactors have been shown subsequent removal of the catalyst in dispersed powder form as
to be effective in the oxidative degradations of para-chloro- would be required in a batch reactor. This methodology has
phenol, toluene [36], phenol and methylene blue [37] with a been shown to be effective in the oxidative degradation of a
deposited TiO2 photocatalyst. range of organic compounds, including methylene blue, [41-43]
o-cresol [44], perchloroethylene [45] and 4-chlorophenol [46],
Aside from oxidative degradations, dual-channel reactors have mainly with a view to air and water purification [47]. The
been used for synthetically useful transformations. Returning to following section will focus only on synthetic applications of
the oxidation of α-terpinene, moderate yields of ascaridole were such reactors.
obtained by using a silica-supported fullerene promoter [38].
Similarly, L-methionine was efficiently oxidised to the corres- Photoexcitation of titanium dioxide semiconductors leads to the
ponding sulphoxide in the same reactor [38]. Again, an issue promotion of an electron to the conduction band, leaving behind
with the reactor was the low productivity: in the case of the oxi- a positive hole in the valence band. Thus titanium dioxide can
dation of α-terpinene, productivities were in the order of function either as an oxidant by donation of an electron of a
10 mg/h, whilst the oxidation of methionine proved less reacting molecule into an electron hole, or as a reductant by the
productive at 4.5 mg/h. Neither of these seems likely to be of donation of an electron in the conduction band of titanium
synthetic use. However, not all dual channel microreactors need dioxide to another molecule. One synthetic use of titanium
suffer such low productivity; the oxidation of β-citronellol (24) dioxide as a photocatalyst is in the alkylation of amines. As
has been performed in the dual channel reactor designed by shown in Scheme 10, photolysis of a mixture of benzylamine
Kim et al. and gave a daily output of 45.5 mmol (1.9 mmol/h) (28) and ethanol in the presence of a titanium dioxide
Scheme 10: Photocatalytic alkylation of benzylamine.
2033
photocatalyst gave the ethylamine 29 in high conversion, The conversion of L-lysine (32) to L-pipecolinic acid (33,
employing both the oxidative (ethanol to acetaldehyde) and Scheme 12) has also been investigated by using a titanium
reductive (imine to amine) activity of the photocatalyst [48]. dioxide photocatalyst. The reaction was found to give poor
results under both microflow and batch conditions, both giving
The reaction was also applied to the alkylation of aniline and a yield of 14%. Although the reaction is stated to have had a
piperidine. Use of a microsystem conferred a number of advant- significantly faster conversion rate in the microreactor, when
ages over the same reaction under batch conditions: dialkyla- the volumes involved are taken into account, output of the batch
tion could be suppressed as monoalkylated product was reaction is reported to be 67 times greater than that of the
removed from the reactor as it formed, avoiding over-reaction; microflow setup. A further problem with the reaction under
UV-LED light sources could be employed, requiring less power both sets of conditions was the erosion of enantiopurity, the ee
than the lamps typically used for the same reaction under batch of the final product being 50% and 47% in microflow and batch
conditions; and the reaction could be performed by using reactors, respectively [53].
Pt-free titanium dioxide, something which had been shown to
be unsuccessful under batch conditions. Nevertheless, the low Photocatalysis can also be performed by using visible light.
flow rates and low concentrations involved limited output to This has been applied to the hydrodehalogenation of
2.4 µmol/h, leaving significant questions over its synthetic α-haloketones by using the dye Eosin Y (36) as a photocatalyst,
utility [48-50]. and both DIPEA and Hantzsch ester 35 as electron donors
(Scheme 13). The reaction was shown to be high yielding for a
Other redox chemistry to be performed by using titanium number of chlorides and bromides. Again, lower reaction times
dioxide photocatalysts includes the selective reduction of nitro were reported for the microflow than for the batch reaction, and
groups to amines in the presence of ketones (Scheme 11) in this case the productivity of the microflow reactor was shown
[50,51], and the oxidation of benzaldehyde to benzoic acid, to be significantly higher than that of the batch reactor
albeit with incredibly poor conversion [52]. (2.5 mmol/h versus 0.4 mmol/h) [54,55].
Scheme 11: Photocatalytic reduction of 4-nitroacetophenone.
Scheme 12: Conversion of L-lysine to L-pipecolinic acid.
Scheme 13: Photocatalytic hydrodehalogenation.
2034
Aza-Henry reactions can also be performed with visible-light online QMS permitted mechanistic studies to be performed.
photocatalysis, in this case with either ruthenium or iridium Although the results demonstrated the quantum yield for the
catalysts (Scheme 14). Again, residence times in the 100 µL gas-phase reaction in simulated solar light to be substantially
microreactor were significantly shorter than those for the batch lower than that of the solution-state reaction (0.16% versus
reactor, and in this case the microflow system also allowed 5.5%), it was shown that conversions of up to 43% could be
reactions that were entirely unsuccessful under batch condi- achieved [57].
tions to be conducted [54].
Photodecarboxylation reactions
Eosin Y (36) catalysis was also applied to an organocatalytic The acetone-sensitised photodecarboxylation chemistry initially
(42) photoredox α-alkylation of octanal (40, Scheme 15) to developed by Griesbeck [58] under batch conditions was
aldehyde 43. The reaction proved to be high yielding under both suggested as being ideally suited to microflow conditions
batch and microflow conditions, and a reduced temperature [59,60]. The chemistry involves the decarboxylative addition of
gave high enantioselectivity. The low productivity of the potassium carboxylates to phthalimides, thus offering an alter-
100 µL microreactor led the reaction to be transferred to an FEP native to Grignard reagents that can be employed under aqueous
macroflow reactor for further scale-up, which led to hugely conditions [58]. Initially, the α-photodecarboxylation of
increased productivity. This does, however, demonstrate the phthaloyl glycine 44 (Scheme 16) was investigated in a micro-
utility of microreactors for initial screening of reaction condi- flow Dwell device and compared with the reaction under batch
tions prior to scale-up [54]. Later work demonstrated that the conditions. The microflow reactor required a shorter residence
reaction could be extended to a number of aldehydes with good time than the Rayonet reactor, which also required 220% more
yields and high to excellent ee’s [55]. irradiation, but the reactor volumes are too different to make
really meaningful comparisons. If productivity is compared,
Photocatalysis in microreactors can also be applied to gas-phase the batch reactor is higher at 0.89 mmol/h compared to
reactions. For instance, the titanium dioxide photocatalysed oxi- 0.025 mmol/h for the microflow setup [61].
dation of both carbon monoxide and methanol has been fol-
lowed in a microreactor through the use of quadrupole mass Of more synthetic interest is the addition of carboxylates 46 to
spectrometry (QMS). The levels of time resolution and versat- phthalimides 45 shown in Scheme 16. This reaction has also
ility of detection offered by this method were reported to be far been compared for microflow and batch reactors for a number
better than that available from conventional GC analysis and, of substrates [61,62], and again, although residence times are
thus, gave data ideally suited to mechanistic studies [56]. lower under microflow conditions, productivity is higher for the
batch reactor. For instance, with 46 (R1 = Ph), productivity is
The potential of microreactors in the photocatalytic splitting of 0.07 mmol/h versus 4.0 mmol/h for the microflow and Rayonet
water has also come under investigation. In this case a rhodium- reactors, respectively. When the same light source is used the
containing inorganic photocatalyst was employed, and again Rayonet remains the most productive at 0.44 mmol/h [61].
Scheme 14: Photocatalytic aza-Henry reactions.
Scheme 15: Photocatalytic α-alkylation of aliphatic ketones.
2035
Scheme 16: Decarboxylative photochemical additions.
Tethering the potassium carboxylate function to nitrogen (e.g., the flow reaction proved successful, the final ratio of the desired
48) allows the decarboxylative additions to be performed as product 51 to the unwanted reductive dimer 52 was identical to
intramolecular cyclisations to the heterocyclic systems 49 that achieved in a batch reactor [63].
(Scheme 16). The reactions of both substrates proved successful
in both batch and microflow reactors with the batch reactor It has been shown that 4,4’dimethoxybenzophenone (DMBP)
proving the most productive (0.54 versus 0.02 mmol/h). When can be used instead of acetone as a sensitiser in these reactions,
the flow-reactor light source was employed in the batch reactor, thus allowing the use of UVA rather than UVB irradiation.
a substantial decrease in conversion was observed in the same Although this is desirable from a technical point of view, partic-
time period (0.54 versus 0.16 mmol/h). Whilst this remains 7.5 ularly with regard to the use of LED light sources, it does intro-
times greater than the flow reaction, the sense of sacrificing duce further purification issues. The reactions shown in
conversion for productivity would depend on a number of Scheme 16 were performed in both microflow and batch
factors, particularly the value of time and products versus the reactors, and although the microflow reactor was seen, in some
value of the starting material. However, as the batch reaction circumstances, to be more selective for the desired reaction
using the lower powered light source was not run for an versus reduction, the productivity of the batch reactor was
extended time it is impossible to say what the productivity consistently superior [64].
would have been had the reaction been allowed to near comple-
tion [61]. Miscellaneous photochemical reactions
DMBP has been used as a sensitiser for another reaction
In an extension of the above reactions, it was shown that performed in a microflow reactor. Oelgemöller et al. showed
α-thioalkyl-substituted carboxylates 50 could be added to that the addition of isopropanol (54) to furanones 53
phthalimides 45 in a microflow reactor (Scheme 16). Although (Scheme 17) could be performed in a microchip reactor using
2036
Scheme 17: Photochemical addition of isopropanol to furanones.
an LED light source [65]. The results showed the reaction time
to be shorter than for a Rayonet reactor with flow rates of
2.6 µL/min. Further work focussed on the use of a dual micro-
capillary tower reactor [66] and a Dwell device, both of which
gave similarly high conversions but with much improved flow
rates (230 µL/min and 340 µL/min, respectively). Both devices
approach the productivity regime of the batch reactor with
which they were compared (1.4 mmol/h batch versus
0.46 mmol/h tower and 0.67 mmol/h dwell) [67]. This progress
Scheme 18: Photochemical addition of methanol to limonene.
was built upon by the manufacture of a ten-capillary device
fabricated from FEP tubing (1.6 mm o.d., 0.8 mm i.d.), each
tube having an internal volume of 5 mL. The tubing was Microflow photochemistry has been applied to the challenging
wrapped around two 18 W UVA (365 nm) lamps, five tubes per 1,4-reduction of flavones 60 with NaBH4 (Scheme 19). It was
lamp, giving a device that clearly fits into the macroflow regime shown that under photochemical reduction conditions a number
as defined in this review. By making more efficient use of the of products, including ethyl salicylate and various dimers, were
light, productivities were increased to 1.8 mmol/h per tube. formed. Use of a photochemical flow system rather than a batch
Additionally, it was demonstrated that the ten capillaries could reactor was found to give a greater conversion at the expense of
be used either for parallel synthesis or, by performing the same increased dimer formation, both effects being ascribed to the
reaction in each capillary, a single product could be formed at greater average photon density in the flow system [70].
up to 18 mmol/h, which is a significant improvement on the
microreactor starting point [68]. Mechanistic studies of the photoreduction of benzophenone
(63) with benzhydrol (64) (Scheme 20) have been performed in
Toluene has been employed as a photosensitiser in the addition a microflow reactor, allowing the quantum yield of the reaction
of methanol to limonene (56, Scheme 18). The reaction was to be determined by using far less solvent than in standard
performed in a quartz microreactor; however, the reaction methods [71].
suffered from poor product selectivity and d.e. Comparison with
a batch reactor showed that the batch reactor gave much higher A microflow reactor was used in the scaling up of the produc-
conversion [50,69]. Subsequent work has shown that the tion of a key intermediate for the endothelin receptor antagonist
conversion of this reaction in a microflow reactor can be impro- myriceric acid A (Scheme 21). The reaction was optimised in a
ved through the use of high-power, high-pressure Hg lamps single-channel microreactor, which demonstrated that doubling
[28]. the residence time allowed the switch from a 300 W Hg lamp to
Scheme 19: Light-promoted reduction of flavone.
2037
Scheme 20: Photoreduction of benzophenone with benzhydrol.
Scheme 21: Barton reaction in a microflow system.
a 15 W black light, i.e., a significant increase in photon effi- the productivity of the reactor was very low due the optimal
ciency, and resulted in an increased percentage yield (71% flow rate being 5 µL/min [74].
versus 56%). The same yields could be achieved by using a
1.7 W UV-LED under identical conditions, making the reaction Photocyanation and photochlorination have both been investi-
more photon-efficient still. These conditions were then applied gated as reactions for microflow systems. Use of a glass dwell
to a multichannel microreactor, which allowed the synthesis of device allowed the formation of chlorocyclohexane (73) by
multigram quantities of the desired intermediate at rates of up to chlorination of cyclohexane (72) with sulfuryl chloride under
0.155 g/h [72,73]. This is an example where the high value of 15 W black-light irradiation (Scheme 23); however, the reac-
the starting material 66 makes achieving a high percentage yield tion was relatively inefficient and productivity was low
very important, and although it remains possible that switching (0.18 mmol/h) [75]. A biphasic photocyanation of pyrene (74,
from a microflow system to batch reactor would increase Scheme 24) was shown to be efficient in a polymer micro-
productivity at the expense of percentage yield, it is unlikely channel reactor, but the low flow rate led to very low
that this would prove more cost effective. productivity (0.24 µmol/h) [76].
Microflow photochemistry has also been applied to the syn-

thesis of another steroidal compound, vitamin D 3 (71,
Scheme 22). By performing the synthesis in a microreactor, the
two stages could be performed consecutively under two
different conditions, giving good selectivity for the desired
Scheme 23: photochemical chlorination of cyclohexane.
product from a number of common byproducts. Unfortunately,
Scheme 22: Microflow synthesis of vitamin D3.
2038
Scheme 24: photochemical cyanation of pyrene.
Finally, two studies have demonstrated the utility of coupling construct a simple but highly effective single-pass continuous-
online analysis to photomicroreactors, for following reactions flow reactor (Figure 3). This reactor demonstrated for the first
and detecting short-lived intermediates. A microflow reactor time how synthetic organic photochemical reactions can be
using photonic crystal fibres was coupled to a mass spectro- scaled up in a traditional laboratory fume hood to produce
meter to allow the conversion of cyanobalamin to aquabalamin multigram quantities of materials without the need for particu-
to be followed. The technique was found to allow much more larly specialist equipment.
rapid analysis when compared to a cuvette based approach;
however, one issue highlighted was that in general it did not By using an inexpensive peristaltic pump, the [2 + 2] cycloaddi-
permit quantitative analysis [77]. Another approach to tion of maleimide (76) and n-hexyne was run continuously for
following microphotochemical reactions by online analysis 24 hours under optimised conditions for a custom-built Pyrex
involved UV detection. The formation of benzopinacol from reactor (Scheme 25). This reaction produced 85 g of isolated
benzophenone was used as a model reaction and quartz reactor cyclobutene product 77. A Vycor reactor, driven by a 600 W
construction allowed both the use and detection of shorter lamp gave an 83% conversion when a 0.4 M solution was
wavelengths than would be allowed by Pyrex. The data gener- passed through at 8 mL/min. This corresponds to a productivity
ated was compared to that generated by HPLC analysis; how- of 159 mmol/h, which if run over the same 24 hour period
ever, although both sets of data showed the same trend, there would yield 685 g of product. The higher productivity of the
were differences in concentration values. This was assumed to Vycor reactor illustrates the importance of glassware choice for
be due to continued photoreaction in the HPLC samples (thus UV transmission. The N-pentenyl substituted dimethyl
leading to inaccurate HPLC data), and indeed use of the online maleimide 78 underwent a [5 + 2] photocycloaddition to the
analysis did allow a profile of conversion versus flow rate to be corresponding azepine 79 with a productivity of 39 mmol/h.
determined [78]. Although at a relatively early stage, both
devices show how microflow photochemistry is ideally suited to The same reactor was also used to optimise the [5 + 2] cycload-
rapid process optimisation, and it is to be hoped that this poten- dition of N-pentenyl-3,4-dichloromaleimide 80 (Scheme 26), a
tial receives further investigation in the future. substrate sensitive to over-irradiation due to secondary reac-
tions of the bicyclic azepine product 81. The scale-up of this
Macroflow photochemistry reaction was hugely impractical as a batch process. For the
Photocycloadditions maximum tolerated concentration of 0.02 M, the product can
The reactor that has served as the prototype for many of the only be produced in ~0.5 g amounts, at 50–65% yield, using a
studies summarised in this review of macroflow photochem- conventional batch apparatus. The convenience and superiority
istry was first reported by Booker-Milburn and Berry in 2005 of flow photochemistry over batch can be graphically illus-
[21]. In this detailed study, UV-transparent fluorinated ethylene trated by the fact that 60 g of 80 was converted to 38.5 grams
propylene (FEP) tubing (3.1 mm o.d., 2.7 mm i.d.) was used to (64% isolated yield) of 81 in a single 11 h run (0.1 M,
Scheme 25: Intermolecular [2 + 2] cycloaddition of maleimide (76) and intramolecular [2 + 2] cycloaddition of dimethylmaleimide derivative 78 under
flow conditions.
2039
Scheme 26: Intramolecular [5 + 2] cycloaddition of maleimide under flow conditions.
4 mL/min), with recovery of 15.6 g of starting material [79]. To The in situ generated species underwent spontaneous
process the same amount in batch at 65% conversion would Diels–Alder cycloaddition in the presence of cyclopentadiene.
require 120 individual 0.5 g scale reactions with no recovery of The reaction was performed on 18.2 g (60 mmol) of phenacyl
starting material. sulfide 84 under batch conditions in neat cyclopentadiene to
give a 65% yield after 9 hours. Under optimised flow condi-
This FEP reactor was essential for the protecting-group-free tions 38 g (126 mmol) of the sulfide was irradiated in DCM
synthesis of (±)-neostenine [80]. The key step utilised the (0.2 M) in the presence of 40 equiv cyclopentadiene at
[5 + 2] cycloaddition (Scheme 27) for the construction of the 2 mL/min (5.25 hours in total) to give a 75% yield of 86.
pyrrolo[1,2-a]azepine core 83. As a batch process, this particu-
larly sensitive reaction could only be performed on a 50 mg One issue with photochemical flow reactions, which can some-
scale giving yields from 40–60% with full consumption of times be overlooked, is the reduction in performance of the
starting material. It would therefore require 42 individual batch reactor as material is deposited on the tubing walls. This is
reactions to process the 2.1 g of available precursor. When rarely an issue when the reaction is performed on a small scale/
performed under flow conditions, this material was processed in short run time but a thorough evaluation of the performance of a
a single 9 hour run with the recovery of 23% starting material reaction must also take the maximum operational time into
82. account.
Soon after it was first reported, the Booker-Milburn/Berry The photodimerisation of maleic anhydride (87, Scheme 29) is
reactor was utilised by Aggarwal et al. to scale up the photo- one reaction that poses such a problem since the product 88 is
chemical generation of a thioaldehyde 85 (Scheme 28) [81]. insoluble in common organic solvents. As a batch process the
Scheme 27: Intramolecular [5 + 2] cycloaddition as a key step in the synthesis of (±)-neostenine.
Scheme 28: In situ generation of a thioaldehyde by photolysis of a phenacyl sulfide.
2040
reaction is particularly inefficient since the product suspension reaction temperature and a distinct advantage of the flow reactor
scatters light and coats the immersion well. The precipitated is that the temperature can be more precisely controlled than in
product can also pose a problem in a flow reactor as it can a batch process. This is achieved by immersing the FEP-
adhere to the walls and even block the flow channels wrapped well in a methanol-containing cooling bath. At a given
completely. temperature the d.e. of the products 90/91 under flow condi-
tions was superior to that when carried out as a batch process.
Flow conditions were employed by Seeberger et al. in the

[2 + 2] cycloaddition of maleimide-functionalised poly-L-lysine
with alkyne tethered glycol-dendrons to form cyclobutenes [84].
The reactions were performed on a sub-millimolar scale, but the
precise control over reaction conditions with the flow apparatus
allowed for high yields of the complex dendronic products.
Scheme 29: Photodimerisation of maleic anhydride.
Nettekoven et al. [85] trialled a continuous flow reactor

This issue has been addressed in a study by Horie et al. [82]. consisting of an Ehrfeld Photoreactor XL driven by a custom
The FEP tubing of a flow reactor clogged up within an hour built bank of four 8 W low-pressure lamps. The intramolecular
under conventional flow conditions with a single solvent. cycloaddition of cyclopentenone 92 (Scheme 31) was opti-
Nitrogen gas was introduced creating a regular stream of mised by varying the reactor channel thickness (20–90 μm),
bubbles, which broke the solution up into a series of discrete concentration and flow rate. Under optimised conditions 100 g
liquid portions in the flow channel. This gas/liquid "slug flow" of starting material in acetone (0.026 M) was irradiated at a
enabled the precipitated product to be transported more effi- flow rate of 3.0 mL/min to give 93 in an isolated yield of 48%.
ciently through the reactor. The individual segments are The reactor was flushed through with methanol every 24 hours
described as acting like a “micro-batch reactor” containing the to remove deposits of polymeric side-products, which can
mobilised solid. It was found that tubing of i.d. around 1 mm or reduce the yield of the reaction.
less was required to produce a regular gas/liquid slug flow. The
reactor was also immersed in an ultrasonic bath to further
reduce the risk of product adhesion. In this way the reactor
could be run continuously for over 16 hours. Compared to the
batch process the maleic anhydride dimer 88 was formed in a
higher purity since over-irradiation is avoided. A higher overall
conversion can be achieved by continuous filtering of the prod-
uct and recycling of the solution.
Scheme 31: Intramolecular [2 + 2] cycloaddition of a cyclopentenone.

A gas-liquid slug flow was also recently used for the [2 + 2]
cycloaddition of a chiral cyclohexenone 89 with ethylene
(Scheme 30) [83]. The reactor comprised of FEP tubing Isomerisations and rearrangements
(1.0 mm i.d.) wrapped around a quartz immersion well with The irradiation of diazo compounds often leads to the loss of
nine windings. The reaction was driven by a 500 W medium molecular nitrogen along with the formation of a highly reactive
pressure mercury lamp and the solution delivered by a syringe carbene species. The carbene generated from the photolysis of
pump. The d.e. of the product was found to be dependent on the α-diazo-β-ketoamide 94 (Scheme 32) underwent a Wolff-
Scheme 30: [2 + 2] cycloaddition of a chiral enone with ethylene.
2041
Scheme 32: Photochemical Wolff rearrangement and cyclisation to β-lactams.
rearrangement to ketene 95, which cyclized to β-lactam dia- had previously been reported to give a 45% yield in 20 hours,
stereomers 96 and 97 [86]. As a batch process the irradiation representing a productivity of 0.25 mmol/h [87]. Seeberger and
was performed by irradiating a cooled toluene solution workers have since revisited the reaction and, through the use of
(1.1 mmol, 0.01 M) of the diazo compound in a Pyrex flask a continuous flow reactor, obtained 99 with a productivity of
with an external medium-pressure mercury lamp. After 7 hours 0.84 mmol/h when irradiating the same substrate [88]. Flow
an overall yield of 90% was obtained (0.14 mmol/h). conditions allowed for the precise control over the reaction time
necessary to optimise this sensitive reaction.
In order to increase the productivity and scalability of the reac-
tion the cooled solution (1.82 mmol, 0.01 M) was continuously
circulated around the lamp through a coil of FEP tubing
(3.2 mm o.d., 1.6 mm i.d.) with a total volume of 15 mL. The
flow rate was set to 12.5 mL/min although this was of course
not a single-pass operation. This setup gave an overall yield of
81% after just 3.5 hours (0.42 mmol/h), a productivity which
could probably be improved by wrapping the FEP tubing
Scheme 33: Photochemical rearrangement of aryl azides.
around the full length of the lamp.
As a safer alternative to the medium-pressure lamp a 100 W Photochemical rearrangements of the N-oxide moiety represent
CFL was used to drive the reaction. When used in an analogous another important class of reactions that are able to transform an
flow configuration, the solution (1.82 mmol, 0.01 M) was circu- aromatic ring [89]. When synthesising a range of 4-substituted
lated at 5 mL/min for 48 hours to give an overall yield of 91% quinolone 101 derivatives from quinoline N-oxides 100
(0.035 mmol/h). The corresponding batch reaction required (Scheme 34), Bach and co-workers found that the yield was
18 hours to process just 0.18 mmol of the diazo compound in a reduced as a result of [2 + 2] dimerisation of the photochemic-
95% yield (0.01 mmol/h). ally active product [90]. This was prevented by performing the
reaction at reduced concentration (6–7 mM) using fluorescent
The photolysis of aryl azides in the presence of water provides UVA lamps (419 and 366 nm) in the presence of oxygen as a
an easy access to the 3H-azepinone ring system. Unfortunately, triplet quencher. In order to efficiently irradiate the large
the reaction suffers from the need for extensive irradiation times volumes of dilute reactant solution required, the reaction was
and is prone to the formation of byproducts and product decom- carried out as a flow process. A flow reactor was assembled
position. As a batch process, the photolysis of 2.0 g of methyl consisting of a 7 mm Duran tube, double coiled with an outer
4-azidobenzoate (98, Scheme 33) in THF and water (0.05 M) diameter of 75 mm and a height of 200 mm. This was posi-
Scheme 34: Rearrangement of quinoline N-oxides to quinolones.
2042
tioned in the centre of a Rayonet reactor with 16 lamps of the was investigated (Scheme 36) [94]. A continuous-flow photo-
required wavelength. chemical reactor was constructed by using the Booker-Milburn/
Berry configuration in the hope that flow conditions would
The reactor was also used to produce quinolones bearing overcome the inherent difficulties in scaling up a batch process.
tethered alkenes at the 4-position for subsequent intramolecular A single layer of tightly coiled FEP tubing (3.18 mm o.d.,
[2 + 2] cycloaddition [91]. This reaction was completely 1.59 mm i.d.) was wrapped around a cooled immersion well
suppressed during the singlet-mediated N-oxide rearrangement with a Pyrex filter and the reaction was driven by a 450 W MP
by the presence of triplet-quenching oxygen. lamp. In order to optimise the reaction a full factorial design-of-
experiments (DoE) study was performed with temperature,
Another reaction that has recently been reinvestigated by using concentration, flow rate, and 9-acetylanthracene loading as
a continuous-flow process is the photochemical rearrangement factors. Optimal conditions correspond to low concentration
of arylcyclobutenone 102 to 5H-furanone 103 (Scheme 35). The (5.0 mg/mL) and high flow rate. The reaction temperature and
original report described a batch reaction utilising a quartz well sensitiser loadings were not found to be significant factors over
and a 400 W medium-pressure mercury lamp [92]. A 27% yield the range investigated in the design space.
of phenyl substituted furanone 103 was obtained after 4 hours,
and it was noted that the products were unstable under the Photooxidations
photolysis conditions. Harrowven and co-workers constructed a As highlighted in the microflow section the in situ generation of
photochemical flow reactor by wrapping PFA tubing (1.6 mm singlet oxygen (1O2) by sensitisers is in principle a simple and
o.d., 1.0 mm i.d.) around a quartz tube [93]. Driving the reac- environmentally benign method to produce such a reactive
tion from within this tube was a compact 9 W PL-S lamp, avail- reagent. Whilst the synthetic potential for 1O2 has been demon-
able with a UVA- or UVB-emitting phosphor in addition to the strated extensively, several issues with scale-up have prevented
uncoated 254 nm lamp. The use of the UVB lamp allowed for its widespread industrial use. As with all photochemical reac-
selective irradiation of the cyclobutenone over the product, thus tions, increasing the scale of a batch reaction by increasing
minimising secondary reactions. The resourceful choice of light reactor volume alone has an adverse effect on the efficiency.
source combined with fine control over irradiation time allowed This is a result of the majority of photons being absorbed a
for near quantitative yields to be obtained for a range of deriva- short distance from the lamp and is exacerbated by the intense
tives in this useful but previously capricious reaction. absorption of many photosensitisers.
A widely used solution to this problem is to recirculate solution

from a reservoir through an annular reactor. In a particularly
novel example, researchers generated 1 O 2 on nanoporous
silicon excited by the emissions from green LEDs [40]. The
reactor was used to decompose diphenylisobenzofuran with an
estimated quantum yield of 34% although it was not put to any
Scheme 35: Photochemical rearrangement of cyclobutenones.
synthetic use. The scale-up of 1O2 reactions is also complicated
by the need for efficient oxygen delivery to the system. A reser-
As part of a new route to the therapeutic vitamin D derivative voir of oxygenated solvent poses a considerable safety risk on
doxercalciferol, the sensitized photoisomerisation of 104 to 105 any scale and a far safer option is to introduce the gas as and
Scheme 36: Photoisomerisation en route to a vitamin-D derivative.
2043
when required. This technique, however, can be limited by the duction of oxygen to the solution, and efficient irradiation from
low mass transfer of oxygen into many solvents. The use of the light source.
supercritical carbon dioxide as a solvent in a single-pass reactor
enabled the oxidation of citronellol at a rate of 0.27 mmol/min The synthetic use of the above reactor was demonstrated
[95]. This method exploited the high solubility of O2 in scCO2, dramatically in the synthesis of artemisinin 108 [97]. This first-
along with the low viscosity of the fluid, to overcome the mass- line antimalaria drug was produced as a continuous-flow
transfer issues. The negligible environmental impact of scCO2 process from dihydroartemisinic acid in three consecutive steps.
along with its nonflammable nature could see its use in a par- TPP sensitised photooxidation of 106 produced the
ticular industrial setting, but the equipment is likely far too allylic hydroperoxide at a rate of 1.5 mmol/min in 75% yield
specialised to be taken up by many. (Scheme 38). This was followed by acid-catalysed Hock
cleavage and triplet-oxygen (3O2) oxidation. The resulting com-
Fortunately, Seeberger and co-workers have recently shown that pound underwent a series of spontaneous condensations to give
sufficient mass transfer of O 2 can be achieved with more artemisinin in 45% yield. This is a highlight for flow chemistry
conventional equipment if slug flow conditions are employed in general and demonstrates what can be achieved by the
[96]. A solution of citronellol was mixed with oxygen gas in a marriage of chemistry with technology.
PTFE T-mixer before entry into a Booker-Milburn/Berry type
FEP flow reactor (1.59 mm o.d., 0.76 mm i.d.) (Scheme 37). Photocatalytic reactions
The gaseous segments of the biphasic mixture enabled a huge As with photooxidations, the widespread use of reactions
surface area of the solution to be exposed to oxygen. The involving photoactivated catalysts has been marred by the inab-
concentration of oxygen in solution was further increased by ility to scale up batch processes. The highly reactive cyclo-
increasing the pressure with a 6.9 bar back-pressure regulator. pentadienylruthenium complex 110 can be prepared photo-
Under optimised conditions near quantitative yields were chemically from the corresponding benzene sandwich complex
obtained with a productivity of 2.5 mmol/min. This simple but 109 in near quantitative yield (Scheme 39). When performed as
hugely effective reactor configuration addresses all major issues a batch reaction the concentration was limited to 0.02 M and ir-
of sensitised photooxygenation reactions: safe, controlled intro- radiation times in excess of 12 hours were required. A
Scheme 37: Schematic of the Seeberger photooxygenation apparatus and sensitised photooxygenation of citronellol.
2044
Scheme 38: Sensitised photooxygenation of dihydroartemisinic acid.
continuous-flow reactor was constructed by wrapping high- It was later shown that a catalytically active [CpRu]+ species
purity perfluoro alkoxy alkane (HPFA) tubing (1.58 mm o.d., could be generated in situ by direct photolysis of the aryl com-
0.79 mm i.d.) around a quartz immersion well [98]. This reactor plex and intercepted by reactants [99]. This avoided the need to
enabled catalyst 110 to be produced at a rate of 1.56 g/h when a prepare and isolate the tris(acetonitrile) complex. The reactivity
0.06 M solution was pumped at 1 mL/min. The high purity PFA of the photoactivated species was first demonstrated with an
was required since standard PFA tubing leeched plasticiser into intramolecular ene–yne cycloisomerisation (Scheme 40). No
solution and reduced conversion. product was observed under batch conditions but the optimised
flow conditions gave complete conversion within very short
residence times. The reactor consisted of a single coil of quartz
tubing (3.18 mm o.d., 0.73 mm i.d.) positioned around a water
cooled 450 W MP lamp, with or without a Pyrex filter.
Although the reported isolated yield of 90% was obtained with
flow rate of 12.5 μL/min, 98% conversion was observed at flow
rates up to 125 μL/min through a Pyrex filter and 333 μL/min in
the absence of a filter.
The intermolecular alkene–alkyne coupling was also successful

Scheme 39: Photochemical preparation of CpRu(MeCN)3PF6.
for a range of substrates (Scheme 41). Whilst the productivity
Scheme 40: In situ photochemical generation and reaction of a [CpRu]+ catalyst.
Scheme 41: Intermolecular alkene–alkyne coupling with photogenerated catalyst.
2045
of the reaction for many of the isolated products appears rather tion tested (1 g/L, 1 mL/min = 0.34 mmol/h). The reaction was
modest (<1 mmol/h) given the power of the lamp, the reactor not optimised, however, and was not put to any synthetic use.
consisted of a single loop of quartz tubing with internal volume
250 μL. A modified setup that captures the full emissions of the The above examples have relied heavily on the UV light
450 W mercury lamp would likely improve the productivity emitted by medium-pressure mercury lamps. Exposure of the
rate. products to the high-energy photons can cause unwanted side
reactions. In recent years a vast amount of progress has been
A more efficient development of the quartz tubing reactor has made in the field of visible-light-activated photocatalysts.
since been reported by the same group [100]. The reactor Although central to the discovery of the new reactions, the
features a coil of quartz tubing (3.18 mm o.d., 1.0 mm i.d.) with batch apparatus used has often been ineffective in allowing
multiple turns so as to span almost half the length of the 450 W scale-up to synthetically useful quantities. This is once again
MP lamp. The loss of UV by transmission through the wall was due to the strong absorption of the photocatalysts preventing the
minimised through the use of a highly reflective aluminium penetration of light into the bulk of the solution. The first
mirrored cylinder. Although quartz is superior in terms of trans- examples of continuous-flow processes being used to carry out
parency and is chemically inert, the fabrication constraints may such visible-light-mediated photocatalytic reactions in the field
cause limitations. The reactor was used for a catalysed (118) of organic synthesis have only just emerged.
photoinduced-electron-transfer (PET) deoxygenation reaction
to produce 2‘-deoxy and 2‘,3‘-dideoxynucleosides. The Seeberger and co-workers constructed a reactor with a 4.7 mL
2‘-deoxynucleosides 119 were typically produced in yields of volume by wrapping FEP tubing (1.59 mm o.d., 0.76 mm i.d.)
around 80% with productivities around 0.1 mmol/h around two parallel metal rods held apart so as to form a planar
(Scheme 42). surface [102]. The tubing was positioned between two 17 W
white LED lamps, a configuration that efficiently captures the
An earlier example of a quartz-tubing photochemical reactor light from the planar light sources. A range of Ru(bpy)3Cl2
utilised a squared coil of tubing (5.0 mm o.d., 1.5 mm i.d.) 7 cm catalysed reactions were trialled and compared to their batch
wide and 23.5 cm high [101]. The thickness of the tubing walls counterparts. The productivity of the flow processes were
was compensated for by the impressive array of 15 W LP consistently higher than the previously reported batch results,
lamps: six inside the coil and another six external. The reactor tolerated lower catalyst loadings and proceeded well in the
was used for the photochemical defluorination of 3,5-diamino- absence of Hanzsch ester 35. For example, the reduction of
trifluoromethylbenzene (DABFT) 120 in water to give 3,5- methyl 4-azidobenzoate (98) gave 122 in 89% yield at a flow
diaminobenzoic acid (121, Scheme 43). Complete conversion of rate of 2.36 mL/min and concentration of 0.1 M in the presence
DABFT was observed at the highest flow rate and concentra- of 1.2 equiv of a Hantzsch ester (Scheme 44). This corresponds
Scheme 42: PET deoxygenation of nucleosides.
Scheme 43: Photochemical defluorination of DABFT.
2046
to a productivity of over 12 mmol/h. The reaction also capture the light from an assembly of seven blue (447 nm)
proceeded in the absence of the Hantzsch ester, but the flow LEDs operating at 5.88 W. The internal volume of the reactor
rate was reduced to 236 μL/min giving a productivity of was 497 μL and a silver mirrored flask was used as a reflector
1.2 mmol/h. behind the tubing. Impressive productivities were observed
when trapping oxidatively generated iminium ions with a range
of nucleophiles. For example, a solution of N-phenyltetrahy-
droisoquinoline 129, BrCCl3 and Ru(bpy)3Cl2 in DMF was
passed through the reactor at a rate of 5.75 mmol/h
(Scheme 46). This was sufficient for full conversion to the
iminium salt 130, which was reacted immediately with a
nucleophile present in the dark receiving flask. In addition to
the nitromethane adduct 131, sodium cyanide, an allylic silane,
and an acetylene acted as nucleophiles giving the trapped
Scheme 44: Aromatic azide reduction by visible-light-mediated products in yields of over 80%. When performed on batch,
photocatalysis.
the nitromethane adduct was produced at a rate of just
0.081 mmol/h.
Similar productivities [102] and yields were also observed for
the reductive dechlorination, reductive epoxide opening and Examples are also given for additional reactions that showed a
alcohol bromination all in the absence of the Hantzsch ester great improvement in productivity when performed with the
(Scheme 45). In the case of the bromination, the solution was flow reactor compared to the batch conditions used during
passed through the photochemical reactor at 25 °C before initial studies. These included intramolecular radical cyclisa-
flowing directly through a PTFE reactor at 100 °C to drive the tions, intermolecular radical indole functionalisations, and inter-
bromination to completion. molecular atom-transfer radical additions (ATRA) using an
iridium catalyst (Scheme 47).
A similar reactor design was reported soon after by Stephenson
et al. [103]. This was constructed from PFA tubing (0.76 mm The same reactor was also used to demonstrate a new method
i.d.) wrapped in figures of eight around two glass tubes so as to for the synthesis of symmetric anhydrides through the light-
Scheme 45: Examples of visible-light-mediated reactions.
Scheme 46: Visible-light-mediated formation of iminium ions.
2047
Scheme 47: Examples of visible-light-mediated photocatalytic reactions.
mediated generation of the Vilsmeier reagent by using photochemistry can be rendered useful in flow by using a well
Ru(bpy) 3 2+ and CBr 4 in DMF (Scheme 48) [104]. thought out, but simple, reactor design that utilises common
laboratory equipment.
Another continuous-flow visible-light reactor was also reported
at the same time by Gagné and co-workers [105]. The light- A recently reported example of a visible-light-mediated
mediated conjugate addition of glycosyl radicals to acrolein 142 photocatalytic process utilising flow conditions involved
(Scheme 49) was high yielding on a small scale in batch the cyclisation of stilbene derivative 144 to [5]helicene
(0.06 mmol in 1 h, 70%), but it could not be scaled up effect- (Scheme 50) [106]. A reactor was constructed by wrapping FEP
ively without extensive irradiation times (2.43 mmol in 24 h, tubing (2.0 mm o.d., 1.0 mm i.d.) around two 30 W CFLs.
85%). This issue was solved by conducting the reaction under Under optimised reaction conditions in a batch apparatus, a
continuous-flow conditions. The reactor used consisted of FEP 57% yield of [5]helicene was obtained after 120 hours irradi-
tubing (1.59 mm i.d.) conveniently wrapped around a standard ation with one 30 W CFL. A solution of identical concentration
Liebig condenser containing three strips of blue LEDs. Two of and scale was irradiated by using the flow reactor to give the
these were connected in series to give the alkylated glycoside product in 40% yield after just 10 hours. The reaction mixture
143 with a productivity of 0.55 mmol/h. The reactor was run still had to be passed through the reactor 20 times to obtain this
continuously for 24 h to yield 5.46 g of this key intermediate for conversion but the flow apparatus clearly allows more efficient
further studies. This example serves to illustrate how difficult irradiation.
Scheme 48: Anhydride formation from a visible-light-mediated process.
Scheme 49: Light-mediated conjugate addition of glycosyl bromide 141 to acrolein.
2048
Scheme 50: Visible-light-mediated photocyclisation to [5]helicene.
Conclusions emerging area for synthetic chemistry and promises to have

Flow photochemistry has developed rapidly since the early wide application. The value of FEP and related tube designs lies
reports just over 10 years ago. Initial studies focussed on the in the simplicity of their construction: all the tubing, glassware,
microflow regime, which itself was born out of the “lab-on-a- lamps and pumps are commercially available at a very econom-
chip” arena. Since then there have been many reports of various ical price and a functioning reactor can be set up in a matter of
well engineered microflow photochemical reactors. Most of hours in a standard fume hood.
these have shown that many photochemical reactions can be
carried out with higher yields (space/time) and selectivities and With now easy access to flow photochemistry we hope that the
with fewer side reactions than comparable batch reactors. On synthetic community at large will make more use of photo-
the whole, however, microreactors are uncompetitive with chemical bond-forming reactions and apply them to their
classic immersion-well batch reactors when it comes to the key general synthetic problems. As way of stimulus, the following
issue of productivity. This is unsurprising given the very low provocative question can be asked: can your ground-state chem-
reaction volumes and flow rates involved, and as such compari- istry give you easy, clean and reagentless access to 100 g quan-
son of two such different reactor topologies is not useful. tities of molecules with high structural complexity? Flow photo-
Microflow reactors are particularly well placed to make best use chemistry can.
of the current developments in LED technology. As microflow
reactors cannot make use of a large photon flux, much of the References
radiation from powerful UV lamps is wasted. Use of arrays of 1. Trommsdorf, H. Ann. Chem. Pharm. 1834, 11.
compact LEDs is much more suitable and efficient. At the 2. Cannizzaro, S.; Sestini, F. Gazz. Chim. Ital. 1873, 3, 241–251.
3. Ciamician, G.; Silber, P. Ber. Dtsch. Chem. Ges. 1901, 34,
moment LEDs of λ < 365 nm are expensive, prohibitively so at
2040–2046. doi:10.1002/cber.190103402118
wavelengths of 300 nm and below where a single LED can cost 4. Eaton, P. E.; Cole, T. W., Jr. J. Am. Chem. Soc. 1964, 86,
as much €300. This price will come down in future, but it is 3157–3158. doi:10.1021/ja01069a041
likely that only a microflow reactor could benefit from this. 5. Corey, E. J.; Nozoe, S. J. Am. Chem. Soc. 1964, 86, 1652–1653.
With further developments photochemical microflow reactors doi:10.1021/ja01062a052
6. Corey, E. J.; Nozoe, S. J. Am. Chem. Soc. 1965, 87, 5733–5735.
are likely to find many applications, particularly if they can be
doi:10.1021/ja00952a036
coupled with automation: screening for new photoreactions,
7. Wender, P. A.; Howbert, J. J. J. Am. Chem. Soc. 1981, 103, 688–690.
reaction and wavelength optimisation, drug discovery, micro- doi:10.1021/ja00393a041
actinometers for quantum yield measurements, etc. 8. Metzger, R.; Fries, D.; Heuschkel, U.; Witte, K.; Waidelich, E.;
Schmid, G. Angew. Chem. 1959, 71, 229–236.
Since its introduction in 2005, the FEP macroflow reactor of doi:10.1002/ange.19590710702
9. Fischer, M. Angew. Chem., Int. Ed. Engl. 1978, 17, 16–26.
Booker-Milburn and Berry has demonstrated that batch-locked
doi:10.1002/anie.197800161
reactions can be scaled up from sub-gram quantities to over 10. Pfoertner, K. H. J. Photochem. Photobiol., A 1990, 51, 81–86.
500 g per day in a single pass. A flagship example of this was doi:10.1016/1010-6030(90)87045-D
recently reported by Seeberger and Lévesque in their 11. Griesbeck, A. G.; Maptue, N.; Bondock, S.; Oelgemöller, M.
continuous (>200 g per day) synthesis of artemisinin, the Photochem. Photobiol. Sci. 2003, 2, 450–451. doi:10.1039/b212357f
12. Baxendale, I. R.; Deeley, J.; Griffiths-Jones, C. M.; Ley, S. V.;
current front-line treatment for malaria. Related designs have
Saaby, S.; Tranmer, G. K. Chem. Commun. 2006, 2566–2568.
very recently demonstrated that photocatalysis can be carried
doi:10.1039/b600382f
out in macroflow devices with high productivities. This is a
very significant development as photocatalysis is a powerful
2049
13. Baxendale, I. R.; Ley, S. V.; Mansfield, A. C.; Smith, C. D. 38. Carofiglio, T.; Donnola, P.; Maggini, M.; Rossetto, M.; Rossi, E.
Angew. Chem., Int. Ed. 2009, 48, 4017–4021. Adv. Synth. Catal. 2008, 350, 2815–2822.
doi:10.1002/anie.200900970 doi:10.1002/adsc.200800459
14. Baxendale, I. R.; Schou, S. C.; Sedelmeier, J.; Ley, S. V. 39. Park, C. P.; Maurya, R. A.; Lee, J. H.; Kim, D.-P. Lab Chip 2011, 11,
Chem.–Eur. J. 2010, 16, 89–94. doi:10.1002/chem.200902906 1941–1945. doi:10.1039/c1lc20071b
15. Jas, G.; Kirschning, A. Chem.–Eur. J. 2003, 9, 5708–5723. 40. Lapkin, A. A.; Boddu, V. M.; Aliev, G. N.; Goller, B.; Polisski, S.;
doi:10.1002/chem.200305212 Kovalev, D. Chem. Eng. J. 2008, 136, 331–336.
16. Mason, B. P.; Price, K. E.; Steinbacher, J. L.; Bogdan, A. R.; doi:10.1016/j.cej.2007.04.013
McQuade, D. T. Chem. Rev. 2007, 107, 2300–2318. 41. Teekateerawej, S.; Nishino, J.; Nosaka, Y. J. Appl. Electrochem.
doi:10.1021/cr050944c 2005, 35, 693–697. doi:10.1007/s10800-005-1623-x
17. Doede, C. M.; Walker, C. A. Chem. Eng. 1959, 62, 159. 42. Teekateerawej, S.; Nishino, J.; Nosaka, Y.
18. Feehs, R. H.; Woodstown, N. J. Photochemical Apparatus and J. Photochem. Photobiol., A 2006, 179, 263–268.
Process. U.S. Patent 3,554,887, Jan 12, 1971. doi:10.1016/j.jphotochem.2005.08.024
19. Coyle, E. E.; Oelgemöller, M. Photochem. Photobiol. Sci. 2008, 7, 43. Li, X.; Wang, H.; Inoue, K.; Uehara, M.; Nakamura, H.; Miyazaki, M.;
1313–1322. doi:10.1039/b808778d Abe, E.; Maeda, H. Chem. Commun. 2003, 964–965.
20. Maurya, R. A.; Park, C. P.; Kim, D.-P. Beilstein J. Org. Chem. 2011, 7, doi:10.1039/b300765k
1158–1163. doi:10.3762/bjoc.7.134 44. Chen, H.-W.; Ku, Y.; Irawan, A. Chemosphere 2007, 69, 184–190.
21. Hook, B. D. A.; Dohle, W.; Hirst, P. R.; Pickworth, M.; Berry, M. B.; doi:10.1016/j.chemosphere.2007.04.051
Booker-Milburn, K. I. J. Org. Chem. 2005, 70, 7558–7564. 45. Chen, D. H.; Ye, X.; Li, K. Chem. Eng. Technol. 2005, 28, 95–97.
doi:10.1021/jo050705p doi:10.1002/ceat.200407012
22. Productivities are generally expressed as mmol/h and have been 46. Gorges, R.; Meyer, S.; Kreisel, G. J. Photochem. Photobiol., A 2004,
calculated for the purposes of this review using data available in the 167, 95–99. doi:10.1016/j.jphotochem.2004.04.004
original publications. We have made every effort to ensure that these 47. Yang, T. C.-K.; Wang, S.-F.; Tsai, S. H.-Y.; Lin, S.-Y. Appl. Catal., B
are accurate. Any errors in our calculations are purely unintentional. 2001, 30, 293–301. doi:10.1016/S0926-3373(00)00241-1
23. Fukuyama, T.; Hino, Y.; Kamata, N.; Ryu, I. Chem. Lett. 2004, 33, 48. Matsushita, Y.; Ohba, N.; Suzuki, T.; Ichimura, T. Catal. Today 2008,
1430–1431. doi:10.1246/cl.2004.1430 132, 153–158. doi:10.1016/j.cattod.2007.12.078
24. Fukuyama, T.; Kajihara, Y.; Hino, Y.; Ryu, I. J. Flow Chem. 2011, 1, 49. Matsushita, Y.; Ohba, N.; Kumada, S.; Suzuki, T.; Ichimura, T.
40–45. doi:10.1556/jfchem.2011.00007 Catal. Commun. 2007, 8, 2194–2197.
25. Maeda, H.; Mukae, H.; Mizuno, K. Chem. Lett. 2005, 34, 66–67. doi:10.1016/j.catcom.2007.04.033
doi:10.1246/cl.2005.66 50. Matsushita, Y.; Ichimura, T.; Ohba, N.; Kumada, S.; Sakeda, K.;
26. Mukae, H.; Maeda, H.; Nashihara, S.; Mizuno, K. Suzuki, T.; Tanibata, H.; Murata, T. Pure Appl. Chem. 2007, 79,
Bull. Chem. Soc. Jpn. 2007, 80, 1157–1161. doi:10.1246/bcsj.80.1157 1959–1968. doi:10.1351/pac200779111959
27. Tsutsumi, K.; Terao, K.; Yamaguchi, H.; Yoshimura, S.; Morimoto, T.; 51. Matsushita, Y.; Ohba, N.; Kumada, S.; Sakeda, K.; Suzuki, T.;
Kakiuchi, K.; Fukuyama, T.; Ryu, I. Chem. Lett. 2010, 39, 828–829. Ichimura, T. Chem. Eng. J. 2008, 135 (Suppl. 1), S303–S308.
doi:10.1246/cl.2010.828 doi:10.1016/j.cej.2007.07.045
28. Pimparkar, K.; Yen, B.; Goodell, J. R.; Martin, V. I.; Lee, W.-H.; 52. Matsushita, Y.; Kumada, S.; Wakabayashi, K.; Sakeda, K.;
Porco, J. A., Jr.; Beeler, A. B.; Jensen, K. F. J. Flow Chem. 2011, 2, Ichimura, T. Chem. Lett. 2006, 35, 410–411. doi:10.1246/cl.2006.410
53–55. doi:10.1556/jfchem.2011.00006 53. Takei, G.; Kitamori, T.; Kim, H.-B. Catal. Commun. 2005, 6, 357–360.
29. Vasudevan, A.; Villamil, C.; Trumbull, J.; Olson, J.; Sutherland, D.; doi:10.1016/j.catcom.2005.02.010
Pan, J.; Djuric, S. Tetrahedron Lett. 2010, 51, 4007–4009. 54. Neumann, M.; Zeitler, K. Org. Lett. 2012, 14, 2658–2661.
doi:10.1016/j.tetlet.2010.05.119 doi:10.1021/ol3005529
30. Jähnisch, K.; Dingerdissen, U. Chem. Eng. Technol. 2005, 28, 55. Neumann, M.; Füldner, S.; König, B.; Zeitler, K.
426–427. doi:10.1002/ceat.200407139 Angew. Chem., Int. Ed. 2011, 50, 951–954.
31. Jähnisch, K. Chem. Ing. Tech. 2004, 76, 630–632. doi:10.1002/anie.201002992
doi:10.1002/cite.200403392 56. Vesborg, P. C. K.; Olsen, J. L.; Henriksen, T. R.; Chorkendorff, I.;
32. Wootton, R. C. R.; Fortt, R.; de Mello, A. J. Org. Process Res. Dev. Hansen, O. Chem. Eng. J. 2010, 160, 738–741.
2002, 6, 187–189. doi:10.1021/op0155155 doi:10.1016/j.cej.2010.03.083
33. Meyer, S.; Tietze, D.; Rau, S.; Schäfer, B.; Kreisel, G. 57. Dionigi, F.; Vesborg, P. C. K.; Pedersen, T.; Hansen, O.; Dahl, S.;
J. Photochem. Photobiol., A 2007, 186, 248–253. Xiong, A.; Maeda, K.; Domen, K.; Chorkendorff, I.
doi:10.1016/j.jphotochem.2006.08.014 Energy Environ. Sci. 2011, 4, 2937–2942. doi:10.1039/c1ee01242h
34. Kreisel, G.; Meyer, S.; Tietze, D.; Fidler, T.; Gorges, R.; Kirsch, A.; 58. Oelgemöller, M.; Cygon, P.; Lex, J.; Griesbeck, A. G. Heterocycles
Schäfer, B.; Rau, S. Chem. Ing. Tech. 2007, 79, 153–159. 2003, 59, 669–684. doi:10.3987/COM-02-S77
doi:10.1002/cite.200600092 59. Gallagher, S.; Hatoum, F.; Zientek, N.; Oelgemöller, M.
35. Ravelli, D.; Protti, S.; Neri, P.; Fagnoni, M.; Albini, A. Green Chem. Tetrahedron Lett. 2010, 51, 3639–3641.
2011, 13, 1876–1884. doi:10.1039/c0gc00507j doi:10.1016/j.tetlet.2010.05.020
36. Matsushita, Y.; Iwasawa, M.; Suzuki, T.; Ichimura, T. Chem. Lett. 60. Hatoum, F.; Gallagher, S.; Oelgemöller, M. Tetrahedron Lett. 2009,
2009, 38, 846–847. doi:10.1246/cl.2009.846 50, 6593–6596. doi:10.1016/j.tetlet.2009.09.067
37. Aran, H. C.; Salamon, D.; Rijnaarts, T.; Mul, G.; Wessling, M.; 61. Shvydkiv, O.; Gallagher, S.; Nolan, K.; Oelgemöller, M. Org. Lett.
Lammertink, R. G. H. J. Photochem. Photobiol., A 2011, 225, 36–41. 2010, 12, 5170–5173. doi:10.1021/ol102184u
doi:10.1016/j.jphotochem.2011.09.022
2050
62. Belluau, V.; Noeureuil, P.; Ratzke, E.; Skvortsov, A.; Gallagher, S.; 85. Nettekoven, M.; Püllmann, B.; Martin, R. E.; Wechsler, D.
Motti, C. A.; Oelgemöller, M. Tetrahedron Lett. 2010, 51, 4738–4741. Tetrahedron Lett. 2012, 53, 1363–1366.
doi:10.1016/j.tetlet.2010.07.017 doi:10.1016/j.tetlet.2012.01.010
63. Tan, S. B.; Shvydkiv, O.; Fiedler, J.; Hatoum, F.; Nolan, K.; 86. Mimieux Vaske, Y. S.; Mahoney, M. E.; Konopelski, J. P.;
Oelgemöller, M. Synlett 2010, 2240–2243. Rogow, D. L.; McDonald, W. J. J. Am. Chem. Soc. 2010, 132,
doi:10.1055/s-0030-1258032 11379–11385. doi:10.1021/ja1050023
64. Shvydkiv, O.; Nolan, K.; Oelgemöller, M. Beilstein J. Org. Chem. 87. Lamara, K.; Smalley, R. K. Tetrahedron 1991, 47, 2277–2290.
2011, 7, 1055–1063. doi:10.3762/bjoc.7.121 doi:10.1016/S0040-4020(01)96138-1
65. Shvydkiv, O.; Yavorskyy, A.; Nolan, K.; Youssef, A.; Riguet, E.; 88. Bou-Hamdan, F. R.; Lévesque, F.; O’Brien, A. G.; Seeberger, P. H.
Hoffmann, N.; Oelgemöller, M. Photochem. Photobiol. Sci. 2010, 9, Beilstein J. Org. Chem. 2011, 7, 1124–1129. doi:10.3762/bjoc.7.129
1601–1603. doi:10.1039/c0pp00223b 89. Albini, A.; Alpegiani, M. Chem. Rev. 1984, 84, 43–71.
66. Yavorskyy, A.; Shvydkiv, O.; Nolan, K.; Hoffmann, N.; Oelgemöller, M. doi:10.1021/cr00059a004
Tetrahedron Lett. 2011, 52, 278–280. doi:10.1016/j.tetlet.2010.11.018 90. Bakowski, A.; Dressel, M.; Bauer, A.; Bach, T. Org. Biomol. Chem.
67. Shvydkiv, O.; Yavorskyy, A.; Tan, S. B.; Nolan, K.; Hoffmann, N.; 2011, 9, 3516–3529. doi:10.1039/c0ob01272f
Youssef, A.; Oelgemöller, M. Photochem. Photobiol. Sci. 2011, 10, 91. Müller, C.; Bauer, A.; Maturi, M. M.; Cuquerella, M. C.; Miranda, M. A.;
1399–1404. doi:10.1039/c1pp05024a Bach, T. J. Am. Chem. Soc. 2011, 133, 16689–16697.
68. Yavorskyy, A.; Shvydkiv, O.; Hoffmann, N.; Nolan, K.; Oelgemöller, M. doi:10.1021/ja207480q
Org. Lett. 2012, 14, 4342–4345. doi:10.1021/ol301773r 92. Perri, S. T.; Foland, L. D.; Moore, H. W. Tetrahedron Lett. 1988, 29,
69. Sakeda, K.; Wakabayashi, K.; Matsushita, Y.; Ichimura, T.; Suzuki, T.; 3529–3532. doi:10.1016/0040-4039(88)85284-5
Wada, T.; Inoue, Y. J. Photochem. Photobiol., A 2007, 192, 166–171. 93. Harrowven, D. C.; Mohamed, M.; Gonçalves, T. P.; Whitby, R. J.;
doi:10.1016/j.jphotochem.2007.05.019 Bolien, D.; Sneddon, H. F. Angew. Chem., Int. Ed. 2012, 51,
70. Ouchi, A.; Sakai, H.; Oishi, T.; Kaneda, M.; Suzuki, T.; Saruwatari, A.; 4405–4408. doi:10.1002/anie.201200281
Obata, T. J. Photochem. Photobiol., A 2008, 199, 261–266. 94. Anderson, B. G.; Bauta, W. E.; Cantrell, W. R., Jr.
doi:10.1016/j.jphotochem.2008.05.024 Org. Process Res. Dev. 2012, 16, 967–975. doi:10.1021/op200346g
71. Borderie, B.; Lavabre, D.; Levy, G.; Laplante, J. P.; Micheau, J. C. 95. Bourne, R. A.; Han, X.; Poliakoff, M.; George, M. W.
J. Photochem. Photobiol., A 1991, 56, 13–23. Angew. Chem., Int. Ed. 2009, 48, 5322–5325.
doi:10.1016/1010-6030(91)80002-Y doi:10.1002/anie.200901731
72. Sugimoto, A.; Sumino, Y.; Takagi, M.; Fukuyama, T.; Ryu, I. 96. Lévesque, F.; Seeberger, P. H. Org. Lett. 2011, 13, 5008–5011.
Tetrahedron Lett. 2006, 47, 6197–6200. doi:10.1021/ol2017643
doi:10.1016/j.tetlet.2006.06.153 97. Lévesque, F.; Seeberger, P. H. Angew. Chem., Int. Ed. 2012, 51,
73. Sugimoto, A.; Fukuyama, T.; Sumino, Y.; Takagi, M.; Ryu, I. 1706–1709. doi:10.1002/anie.201107446
Tetrahedron 2009, 65, 1593–1598. doi:10.1016/j.tet.2008.12.063 98. Gutierrez, A. C.; Jamison, T. F. J. Flow Chem. 2011, 1, 24–27.
74. Fuse, S.; Tanabe, N.; Yoshida, M.; Yoshida, H.; Doi, T.; Takahashi, T. doi:10.1556/jfchem.2011.00004
Chem. Commun. 2010, 46, 8722–8724. doi:10.1039/c0cc02239j 99. Gutierrez, A. C.; Jamison, T. F. Org. Lett. 2011, 13, 6414–6417.
75. Matsubara, H.; Hino, Y.; Tokizane, M.; Ryu, I. Chem. Eng. J. 2011, doi:10.1021/ol2027015
167, 567–571. doi:10.1016/j.cej.2010.08.086 100.Shen, B.; Bedore, M. W.; Sniady, A.; Jamison, T. F. Chem. Commun.
76. Ueno, K.; Kitagawa, F.; Kitamura, N. Lab Chip 2002, 2, 231–234. 2012, 48, 7444–7446. doi:10.1039/c2cc33356b
doi:10.1039/b207991g 101.Chaignon, P.; Cortial, S.; Guerineau, V.; Adeline, M.-T.; Giannotti, C.;
77. Unterkofler, S.; McQuitty, R. J.; Euser, T. G.; Farrer, N. J.; Fan, G.; Ouazzani, J. Photochem. Photobiol. 2005, 81, 1539–1543.
Sadler, P. J.; Russell, P. St. J. Opt. Lett. 2012, 37, 1952–1954. doi:10.1562/2005-08-03-RA-637
doi:10.1364/OL.37.001952 102.Bou-Hamdan, F. R.; Seeberger, P. H. Chem. Sci. 2012, 3,
78. Lu, H.; Schmidt, M. A.; Jensen, K. F. Lab Chip 2001, 1, 22–28. 1612–1616. doi:10.1039/c2sc01016j
doi:10.1039/b104037p 103.Tucker, J. W.; Zhang, Y.; Jamison, T. F.; Stephenson, C. R. J.
79. Hook, B. D. A. The Development of a Continuous-Flow Photochemical Angew. Chem., Int. Ed. 2012, 51, 4144–4147.
Reactor and its Application to the Synthesis of Stemoamide. Ph.D. doi:10.1002/anie.201200961
Thesis, University of Bristol, U.K., 2007. 104.Konieczynska, M. D.; Dai, C.; Stephenson, C. R. J.
80. Lainchbury, M. D.; Medley, M. I.; Taylor, P. M.; Hirst, P.; Dohle, W.; Org. Biomol. Chem. 2012, 10, 4509–4511. doi:10.1039/c2ob25463h
Booker-Milburn, K. I. J. Org. Chem. 2008, 73, 6497–6505. 105.Andrews, R. S.; Becker, J. J.; Gagné, M. R. Angew. Chem., Int. Ed.
doi:10.1021/jo801108h 2012, 51, 4140–4143. doi:10.1002/anie.201200593
81. Aggarwal, V. K.; Fang, G.; Kokotos, C. G.; Richardson, J.; 106.Hernandez-Perez, A. C.; Vlassova, A.; Collins, S. K. Org. Lett. 2012,
Unthank, M. G. Tetrahedron 2006, 62, 11297–11303. 14, 2988–2991. doi:10.1021/ol300983b
doi:10.1016/j.tet.2006.06.044
82. Horie, T.; Sumino, M.; Tanaka, T.; Matsushita, Y.; Ichimura, T.;
Yoshida, J. Org. Process Res. Dev. 2010, 14, 405–410.
doi:10.1021/op900306z
83. Terao, K.; Nishiyama, Y.; Tanimoto, H.; Morimoto, T.;
Oelgemöller, M.; Kakiuchi, K. J. Flow Chem. 2012, 2, 73–76.
doi:10.1556/JFC-D-12-00005
84. Laurino, P.; Kikkeri, R.; Azzouz, N.; Seeberger, P. H. Nano Lett. 2011,
11, 73–78. doi:10.1021/nl102821f
2051
License and Terms

This is an Open Access article under the terms of the
Creative Commons Attribution License
(http://creativecommons.org/licenses/by/2.0), which
permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.
The license is subject to the Beilstein Journal of Organic

Chemistry terms and conditions:
(http://www.beilstein-journals.org/bjoc)
The definitive version of this article is the electronic one

which can be found at:
doi:10.3762/bjoc.8.229
2052

Flow Photochemistry PDF

Uploaded by

Copyright:

Available Formats

Flow Photochemistry PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Flow Photochemistry PDF

Uploaded by

Copyright:

Available Formats

Flow photochemistry: Old light through new windows

Jonathan P. Knowles, Luke D. Elliott and Kevin I. Booker-Milburn*

Review Open Access

Address: Beilstein J. Org. Chem. 2012, 8, 2025–2052.

* Corresponding author Associate Editor: C. Stephenson

Keywords: © 2012 Knowles et al; licensee Beilstein-Institut.

The glassware used for the immersion-well is particularly

Finally, solvent choice is also a key factor. The solvent must be

Why then, with this wealth of useful reactions and established

• Safety. Medium pressure mercury lamps operate optimally at

• Difficulty in scaling up. The Toray caprolactam process

It is perhaps then not surprising that synthetic chemists, as

• At any one time under flow conditions only a very small

• The UV exposure time can be precisely controlled by the

• As a continuous flow device is scale independent, a single

• Due to much shorter path lengths high concentration solutions

• Large volumes of very low concentration solutions can be irra-

photons are delivered to the solution to drive a reaction with a

The main aim of this review is to illustrate the advances made

Scheme 2: Competing reactions in an intramolecular [2 + 2] photocycloaddition.

Scheme 3: Diastereocontrolled cycloaddition of a cyclic enone with cyclopentene.

Scheme 5: Intramolecular [2 + 2] photocycloaddition.

Scheme 6: Paterno–Büchi reaction of benzophenone with an allylic alcohol.

Scheme 7: Photooxygenation of cyclopentadiene.

Falling-film microreactors have been employed in the Rose

Scheme 9: Production of rose oxide 27 from (−)-β-citronellol (24).

Scheme 10: Photocatalytic alkylation of benzylamine.

Scheme 11: Photocatalytic reduction of 4-nitroacetophenone.

Scheme 12: Conversion of L-lysine to L-pipecolinic acid.

Scheme 13: Photocatalytic hydrodehalogenation.

Scheme 14: Photocatalytic aza-Henry reactions.

Scheme 15: Photocatalytic α-alkylation of aliphatic ketones.

Scheme 16: Decarboxylative photochemical additions.

Scheme 17: Photochemical addition of isopropanol to furanones.

Scheme 19: Light-promoted reduction of flavone.

Scheme 20: Photoreduction of benzophenone with benzhydrol.

Scheme 21: Barton reaction in a microflow system.

Microflow photochemistry has also been applied to the syn-

Scheme 22: Microflow synthesis of vitamin D3.

Scheme 24: photochemical cyanation of pyrene.

Scheme 26: Intramolecular [5 + 2] cycloaddition of maleimide under flow conditions.

Scheme 27: Intramolecular [5 + 2] cycloaddition as a key step in the synthesis of (±)-neostenine.

Scheme 28: In situ generation of a thioaldehyde by photolysis of a phenacyl sulfide.

Flow conditions were employed by Seeberger et al. in the

Nettekoven et al. [85] trialled a continuous flow reactor

Scheme 31: Intramolecular [2 + 2] cycloaddition of a cyclopentenone.

Scheme 30: [2 + 2] cycloaddition of a chiral enone with ethylene.

Scheme 32: Photochemical Wolff rearrangement and cyclisation to β-lactams.

Scheme 34: Rearrangement of quinoline N-oxides to quinolones.

A widely used solution to this problem is to recirculate solution

Scheme 36: Photoisomerisation en route to a vitamin-D derivative.

Scheme 38: Sensitised photooxygenation of dihydroartemisinic acid.

The intermolecular alkene–alkyne coupling was also successful

Scheme 40: In situ photochemical generation and reaction of a [CpRu]+ catalyst.

Scheme 41: Intermolecular alkene–alkyne coupling with photogenerated catalyst.

Scheme 42: PET deoxygenation of nucleosides.

Scheme 43: Photochemical defluorination of DABFT.

Scheme 45: Examples of visible-light-mediated reactions.

Scheme 46: Visible-light-mediated formation of iminium ions.