Product recovery from ionic liquids by solvent-resistant nanofiltration: Application to ozonation of acetals and methyl oleate

Article abstract of DOI:10.1039/c003956j

In this work we tackle the problematic separation of reaction products from ionic liquid media. Solvent-resistant nanofiltration proves to be an attractive technique for the separation of non-volatile polar products from ionic liquids. In view of the high compatibility between ozone and ionic liquids, two ozone-mediated model reactions were chosen: firstly the oxidation of acetals to esters in the presence of ozone and secondly the ozonation of methyl oleate to monomethyl azelate and pelargonic acid. The objective was to retain the ionic liquid phase by means of a solvent-resistant nanofiltration membrane, while the organic reaction products permeate through the polymeric membrane. First, the ozonations were studied in order to know the characteristic product compositions. Next, a screening of membranes was performed on synthetic product mixtures. The second generation polyimide-based DuraMem membranes showed the highest rejection, up to 96%, for the evaluated ionic liquids. These DuraMem membranes also proved suitable for the separation of the products on real reaction mixtures, even in a single filtration step.

Full text of DOI:10.1039/c003956j

Cutting-edge research for a greener sustainable future
www.rsc.org/greenchem
Volume 12 | Number 10 | October 2010 | Pages 1677–1872
ISSN 1463-9262
PAPER
De Vos et al.
COMMUNICATION
Product recovery from ionic liquids
by solvent-resistant nanofiltration
Leitner et al.
Catalytic supercritical fluid extraction

PAPER
www.rsc.org/greenchem | Green Chemistry
Product recovery from ionic liquids by solvent-resistant nanofiltration:
application to ozonation of acetals and methyl oleate
a
a
a
a
a
Charlie Van Doorslaer, Daan Glas, Annelies Peeters, Angels Cano Odena, Ivo Vankelecom,
b
a
a
Koen Binnemans, Pascal Mertens and Dirk De Vos*
Received 8th March 2010, Accepted 25th June 2010
DOI: 10.1039/c003956j
In this work we tackle the problematic separation of reaction products from ionic liquid media.
Solvent-resistant nanofiltration proves to be an attractive technique for the separation of
non-volatile polar products from ionic liquids. In view of the high compatibility between ozone
and ionic liquids, two ozone-mediated model reactions were chosen: firstly the oxidation of acetals
to esters in the presence of ozone and secondly the ozonation of methyl oleate to monomethyl
azelate and pelargonic acid. The objective was to retain the ionic liquid phase by means of a
solvent-resistant nanofiltration membrane, while the organic reaction products permeate through
the polymeric membrane. First, the ozonations were studied in order to know the characteristic
product compositions. Next, a screening of membranes was performed on synthetic product
TM
mixtures. The second generation polyimide-based DuraMem membranes showed the highest
TM
rejection, up to 96%, for the evaluated ionic liquids. These DuraMem membranes also proved
suitable for the separation of the products on real reaction mixtures, even in a single filtration step.
Introduction
products from the ionic liquid phase remain in several cases a
challenge for industrial application. When the reaction products
are apolar, judicious choice of the ionic liquid (IL) enables in
specific cases their spontaneous expulsion from the ionic liquid,
Ionic liquids are a class of solvents consisting of large organic
cations and smaller (in)organic counter ions. The ideal ionic
liquids are liquid over a wide temperature range and exhibit a low
vapour pressure. Ionic liquids are considered ‘designer solvents’:
by a judicious choice of anion and cation the physicochemical
properties of the ionic liquid can be adapted to the needs of
a specific application. Ionic liquids have recently received wide
attention as an environmentally benign alternative for volatile
organic solvents.
In a recent communication we demonstrated the high compat-
ibility between ozone and ionic liquids. Ozone can be produced
via an electric discharge and is a worthwhile choice for complex,
6
as we recently reported. This offers an attractive alternative for
the classical extraction with organic solvents. However, if the
reaction products have a higher polarity, one has to revert to
extraction with more polar solvents, in which the ionic liquid
may partly or completely dissolve. This can potentially result in
7
significant losses of ionic liquid in the product stream.
1
Innovative unit operations like nanofiltration may show
7,8
promise to overcome such separation issues. The typical
2
objective is to retain the charged ionic liquid by means of
a membrane, while the organic reaction product permeates
through the membrane. For this purpose, rather hydropho-
bic membranes with the appropriate molecular weight cut-off
3
multi-electron oxidations as only dioxygen is obtained as the
by-product. However, the explosive nature of mixtures of ozone
with the vapours of conventional organic solvents is a significant
(MWCO) were selected. The recent introduction of chemically
4
disadvantage. Additionally, ozone is a detrimental agent in
stable nanofiltration membranes may allow separation of non-
volatile products from reaction media mainly consisting of ionic
atmospheric aerosol formation, especially in the presence of
organic compounds. The introduction of ionic liquids can solve
5
liquid. However, studies coupling solvent resistant nanofiltration
these issues as most ionic liquids have a close-to-zero vapour
pressure at room temperature, thereby minimizing the explosion
risk. Moreover, the formation of aerosols is suppressed, as the
escape of liquid droplets or vapours from the solvent surface
is impeded by the high viscosity, high surface tension and low
vapour pressure of ionic liquids.
7–10
(
SRNF) with ionic liquid reactions are very rare. For instance,
the group of Livingston has addressed the issue, using the Suzuki
7,10
cross-coupling as a model reaction.
Therefore two ozone-mediated model reactions were chosen
to study the product separation from ionic liquids applying
SRNF: first the oxidation of acetals to esters in the presence
of ozone, and second the ozonation of methyl oleate to
monomethyl azelate and pelargonic acid. Acetals derived from
aldehydes are known to react with ozone to yield ester type
Despite these advantages offered by ionic liquids, efficient re-
cycling of the ionic liquid and effective separation of the reaction
11
products. Scheme 1 depicts the two possible pathways in the
ozonation of 2-methoxytetrahydropyran (MTP): either lactone
a
Centre for Surface Chemistry and Catalysis, Department of Microbial
and Molecular Systems, K. U. Leuven, Kasteelpark Arenberg 23, box
(
d-valerolactone, LA) formation or transformation to the open
2
461, 3001, Leuven, Belgium. E-mail: dirk.devos@biw.kuleuven.be
b
hydroxyester form (methyl 5-hydroxypentanoate, HE) which can
be further oxidized to the carboxy ester 5-methoxy-5-oxovaleric
Division for Molecular Design and Synthesis, Department of Chemistry,
K. U. Leuven, Celestijnenlaan 200F, box 2404, 3001, Leuven, Belgium
1
726 | Green Chem., 2010, 12, 1726–1733
This journal is © The Royal Society of Chemistry 2010

that pyrrolidinium-based ionic liquids are stable to ozone, while
imidazolium-based ionic liquids proved sensitive to ozonolytic
2
degradation, as demonstrated by NMR studies. Pernak et al.
also observed this fast degradation of imidazolium ionic liquids
16
induced by ozone. This was ascribed to the sensitivity of the
C
C bonds to ozone. [BMPyr][N(CN)
2
] with a melting point
◦
below -50 C was shown to be a suitable solvent for the acetal
oxidation. The ozonation proceeds not only faster but also
2
Scheme 1 Ozonation of 2-methoxytetrahydropyran (MTP), with for-
mation of methyl 5-hydroxypentanoate (HE) or d-valerolactone (LA).
Methyl 5-hydroxypentanoate can be further oxidized to 5-methoxy-5-
oxovaleric acid (CE).
more selectively in the ionic liquid: in the ionic liquid only
the HE and the over-oxidized CE were formed, which contrasts
with the formation of the lactone product and condensation
products (e.g. methyl 5-(4-methoxy-4-oxobutoxy)pentanoate) in
the case of acetic anhydride as the solvent. Moreover, the
traditional procedure requires acetic anhydride/sodium acetate
for hydroxyl group trapping to inhibit over-oxidation to the
corresponding carboxy ester. From this reasoning, it is clear
that ionic liquids are a more practical media for such ozone me-
diated transformations than the traditional low-boiling organic
solvents.
However, considering the objective of product separation by
means of SRNF, [BMPyr][N(CN) ] is rather inappropriate in
view of its low molecular weight (MW = 208.3 g mol ). A
substantially higher molecular weight compared to the solutes’
molecular weight often facilitates the solute separation from the
acid (CE). We previously communicated that the pyrrolidinium-
based ionic liquid [BMPyr][N(CN) ] is an excellent reaction
2
11
medium compared to the conventionally used organic solvents
2
like acetic anhydride and ethyl acetate. The ozonation pro-
ceeded not only faster but also more selectively in the ionic
liquid.
The second reaction studied in this paper is the ozonolyis of
methyl oleate. The industrial attention towards renewables, like
methyl oleate, strongly increased as a result of the ever growing
use and future depletion of fossil resources for both energy con-
2
-
1
12
sumption and chemicals production. Fatty acid esters present
in seed oils are an alternative source of chemicals, since chem-
ically they are closely related to the hydrocarbons in crude oil.
Esters such as methyl oleate are normally derived from the trans-
esterification of these seed oils with methanol, together with glyc-
17
ionic liquid phase.
Screening of ionic liquids. Therefore various ionic liquids
of higher molecular weight were assessed on their sensitivity
towards ozone. Additional selection criteria were a low melting
point and low viscosity. In a screening of 10 ionic liquids,
only 1-butyl-1-pyrrolidinium bis(trifluoromethylsulfonyl)imide
13
erol as a by-product. Various transformations are possible with
methyl oleate, since both the carboxylic group and the double
bond can be modified. Possible C C functionalizations include
hydrogenation, hydroformylation, hydrosilylation, metathesis
reactions, epoxidation and oxidative cleavage towards aldehydes,
-
1
(
[BMPyr][Tf
2
N], MW = 422.4 g mol ) and alkyldimethylbenzy-
lammonium chloride (benzalkonium chloride, [BenzAlk][Cl]),
-
1
with alkyl = C
8
H17–C18H37; MW = 283.5–423.5 g mol ) fulfilled
14
ketones or acids. In particular, the oxidative cleavage of oleic
acid or its methyl ester, leading to azelaic acid and pelargonic
acid is interesting. Indeed, azelaic acid and its derivatives are
applied in a broad range of industrial applications such as
lubricants, plasticizers, adhesives, cosmetics and pharmaceuti-
cals. The production of polymers like polyamides or polyesters
represents another application area. Pelargonic acid is used as
these conditions.
These ionic liquids were tested as the reaction media in the
ozonation of 2-methoxytetrahydropyran (MTP). Fig. 1 reports
on the influence of the ionic liquid’s nature on the hydroxy
and carboxy ester yields. At 80 C, MTP is totally converted
towards the corresponding esters after only 1 h in the case of
◦
[
BMPyr][Tf N]. The respective yields of hydroxy and carboxy
2
14
starting material for plasticizers and herbicides. The industrial
ester were 55% and 45%. Note that the total ester yield after 1 h
is 100%. Applying [BenzAlk][Cl] as the reaction medium, the
reaction proceeded significantly more slowly, with a conversion
15
process uses ozone for the preparation of these acids. In that
process, the double bond, which is very reactive towards ozone,
is first transformed into an ozonide, which can subsequently be
split oxidatively to form the corresponding acids. Due to the high
compatibility between ozone and ionic liquids and based on our
experience with the acetal ozonations, it is also expected that
in this case ionic liquids can contribute to a safer and greener
transformation of methyl oleate, without aerosol formation.
In this work, we study these two model reactions in ionic liq-
uids, with the primary focus on post-reaction product separation
applying SRNF.
Results and discussion
Oxidation of acetals to esters in the presence of ozone
Fig. 1 Influence of choice of ionic liquid on the yield of hydroxy ester
-1
◦
A prerequisite for applying ionic liquids in ozonolysis reactions
is their chemical stability towards ozone. Before, we reported
and carboxy ester (Reaction conditions: [MTP] = 5 mol L ; 80 C; 1 h;
0.21 mmol/min O -flow).
3
This journal is © The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 1726–1733 | 1727

of only 42% after a 1 h reaction time. The presumable reason
for these observations is different ozone solubility. To the best
of our knowledge, ozone solubility data in ionic liquids are
at present not available. However, such data are available for
oxygen. Anthony et al. evaluated the solubility of various gases
in ionic liquids and stated that the oxygen solubility is generally
Table 1 Screening of membranes for rejection of the IL in test mixtures
ab
mimicking the acetal oxidation reaction mixtures
Permeability
(L m h bar)
d
-2
Entry Membrane
p (N
2
)/bar
R
IL (%)
c
1
2
3
4
5
6
7
8
9
1
Zeol-PDMS/PAI1 20
52 ± 3
87 ± 1
90 ± 2
86 ± 4
52 ± 4
95 ± 2
91 ± 1
54 ± 4
91 ± 1
88 ± 1
—
—
0.7 ± 0.3
—
c
Zeol-PDMS/PAI2 20
18
rather low. They also reported that pyrrolidinium ionic liquids
c
Zeol-PDMS/PI2
20
20
-
and especially those containing the Tf
2
N anion have slightly
c
Zeol-PDMS/PI3
TM
higher oxygen solubility values. It is plausible that the solubility
variations of ozone follow similar trends. However, determining
the ozone solubility in ionic liquids is not part of the scope of the
present work but will be addressed in a forthcoming study which
focuses on the detailed kinetics of ozone mediated reactions.
Next the effect of reaction temperature on the MTP ozonation
STARMEM 122 40
—
TM
DuraMem 200
40
40
20
40
40
0.15 ± 0.075
0.375 ± 0.05
0.175 ± 0.025
0.45 ± 0.125
0.5 ± 0.2
e
TM
DuraMem 200
TM
DuraMem 200
f
TM
DuraMem 200
TM
0
DuraMem 300
a
◦
b
Filtration conditions: 1000 RPM stirring, T = 20 C. Mixture: 0.5 g of
in [BMPyr][Tf
2
N] was investigated. Fig. 2 shows the evolution
2
[BMPyr][Tf N], 1 mmol of 2-methoxytetrahydropyran, 2 mmol of ethyl
c
3
-OH-butyrate, 7 mL of 2-butanone. Non-commercial asymmetric
in HE and CE yields as a function of the reaction time at three
◦
membrane coated with zeolite filled PDMS x times (x = 1, 2 or 3);
different reaction temperatures. At 25 C, the highest HE yield
d
PI = polyimide support layer; PAI = polyamidoimide support layer. The
was attained after 4 h, with HE and CE yields of respectively
indicated rejection (RIL = ((1 - C
P
/C
F
P
) ¥ 100%) with C the concentration
7
6% and 24%. Obviously, the reaction proceeded much faster
of the IL in the permeate and C
F
the concentration of the IL in the feed)
is an average of at least two filtrations. The standard deviation is also
at higher reaction temperatures. The maximum HE yield was
e
◦
f
◦
◦
shown. T = 50 C. Methanol was used as the solvent.
reached after 2 h and 30 min at 50 C (Y(HE) = 77%) and 80 C
Y(HE) = 75%) respectively. It is striking that the same maximum
(
HE yields are obtained at different reaction temperatures.
weight resemblance to the commercially unavailable methyl 5-
hydroxypentanoate, the hydroxy ester derived from MTP. 2-
Butanone was chosen as a solvent because it dissolves both the
ionic liquid and the ester products. Note that water admixture
was avoided to simplify the ensuing determination of the ionic
liquid concentration in the filtrate phase by conductometry; the
presence of water would require additional corrections. In a
first step, the performance of a given membrane was assessed
by its ionic liquid rejection (%). Because the MWCO of a
membrane depends on the filtration conditions (like solvent,
19
solute, transmembrane pressure and temperature) the selection
of a proper membrane for a given application often necessitates
9
the screening of multiple membranes. Recently, based on an
20
earlier concept, an adapted high-throughput filtration cell was
designed in-house, which allows rapid membrane screening of
very small feed volumes (see Experimental). The results of this
membrane screening are presented in Table 1. For the best
performing membranes (highest IL rejections) the permeabilities
were also determined. Note that because of the small volumes
collected in the membrane filtrations, the permeabilities are
indicative, which explains the rather large standard deviations.
Because of the ionic character of the IL, rather hydrophobic
membranes, with typically a low affinity for the IL compared to
hydrophilic membranes, were screened. First, a range of PDMS-
based membranes was prepared (Table 1, entries 1–4). Two
different PI and PAI based supports were coated with zeolite
ZSM-5 filled hydrophobic polydimethylsiloxane (PDMS) (see
Experimental), with x in PIx (or PAIx) representing the number
of PDMS coatings. Because PDMS is known to swell strongly
in apolar solvents like toluene and ethyl acetate, the PDMS
layer was filled with zeolites, which results in an increased cross-
linking density by interaction of the zeolite silanol groups with
Fig. 2 Yield of hydroxy ester (filled symbols) and carboxy ester
(
open symbols) for the ozonation of MTP in [BMPyr][Tf
2
N] (Reaction
conditions: [MTP] = 5 M; 0.21 mmol/min ozone flow).
In conclusion, [BMPyr][Tf
mild conditions, is an excellent alternative for [BMPyr][N(CN)
2
N], with a HE yield of 77% under
].
2
-
1
As the MW of the former ionic liquid is higher (422.4 g mol
-
1
vs. 208.3 g mol ), it is a more suitable candidate to be rejected
by a SRNF membrane.
Post-reaction product separation by solvent resistant nanofiltra-
tion. There are two strategies for the separation of a product
and the ionic liquid: either the product can be rejected while the
ionic liquid permeates through the membrane (strategy I), or the
9
IL can be rejected while the product permeates (strategy II). In
the acetal oxidation, the ionic liquids’ molecular weight is higher
than that of the products. Therefore strategy II will be used.
Before subjecting the reaction mixtures to SRNF, a screening of
polymeric membranes was performed. Therefore samples were
prepared for filtration closely mimicking the reaction mixtures
21
the PDMS polymer chains. We expect that 2-butanone, which
has a polarity comparable to that of ethyl acetate, would induce
a comparably strong swelling. The swelling of membranes causes
an increase in the free space between the polymer chains, leading
to lower solute rejections. By filling the PDMS layer with zeolites,
at 66% substrate conversion: 0.5 g of [BMPyr][Tf N], 1 mmol of
2
MTP and 2 mmol of ethyl 3-OH-butyrate were dissolved in 2-
butanone as solvent. Ethyl 3-OH-butyrate was chosen as an ester
model compound, because of its functional group and molecular
1
728 | Green Chem., 2010, 12, 1726–1733
This journal is © The Royal Society of Chemistry 2010

TM
if dispersed well, this swelling is strongly decreased, enhancing
the solute rejection in these solvents.
Table 2 Multi-step filtration with DuraMem 200 of acetal oxidation
ab
reaction mixtures
It emerges from Table 1 that for the PAI supports, the rejection
of the ionic liquid increased upon coating with additional PDMS
layers. In the case of PAI, 2 coatings of PDMS gave a rejection
c
Permeability (L m-2 h bar)
Filtration step
RIL (%)
1
2
3
71 ± 2
94 ± 2
96 ± 3
0.175 ± 0.07
0.275 ± 0.07
0.425 ± 0.09
of 87% at 20 bar of N (entry 2), whereas one coating only led to
2
a 52% rejection (entry 1). This could be expected since repeated
coating with PDMS makes the top section more dense. Therefore
the PI support was also coated with 2 coatings of PDMS. The
PDMS based membranes on top of PI supports prove even
more suitable for retaining the ionic liquid compared to those
based on PAI supports: PI2 exhibited a rejection of 90% at 20
a
◦
◦
Filtration conditions: p (N ) = 40 bar, 1000 RPM stirring, T = 20 C.
2
b
Reaction conditions: 2 g of [BMPyr][Tf
2
N], 10 mmol of MTP, 80 C,
.21 mmol min- of ozone. The indicated rejection R = ((1 - C
1
c
/C
O
) ¥
0
1
P
O
00%) with C the concentration of the IL in the permeate and C
P
the
concentration of the IL in the original mixture, is an average of at least
two filtrations. The standard deviation is also shown.
bar N
2
(entry 3) whereas PAI2 rejected 87% of the ionic liquid
(
entry 2). It appears that a coating with two PDMS layers is the
theoretically leads to a higher solvent permeation but the ionic
liquid permeation remains more or less constant, resulting in a
optimum, as a further increase in the number of PDMS layers
did not improve the performance of the membrane (entry 4). In
addition, with increasing PDMS coatings, the thickness of the
top layer increases, leading to lower membrane permeabilities.
Summarizing, irrespective of the support used, both PDMS
membranes retain the ionic liquid reasonably well from the test
mixture.
23
higher ionic liquid rejection.
TM
These results suggest that DuraMem 200 membranes can
be used to recycle ionic liquids for the acetal oxidation. This was
verified by applying these membranes in SRNF of real reaction
mixtures. Hereto 10 mmol of 2-methoxytetrahydrofuran was
-
1
ozonized (0.21 mmol min ozone flow) in 2 g of [BMPyr][Tf
2
N]
In a next step, commercially available stable SRNF mem-
◦
TM
at 80 C. After reaction the mixture was diluted in 2-butanone
branes with a low MWCO were investigated. As STARMEM
1
(7 mL) and subjected to the SRNF. As can be seen in Table 2
22 exhibited a very good performance in the rejection of
the one step filtration of the reaction mixture resulted in only
71% of ionic liquid rejection. Therefore, after dilution with 2-
butanone, the permeate was again filtered with the DuraMem
trihexyl(tetradecyl)phosphonium chloride (MW = 519.31 g
-
1
7
mol ) in a product mixture of a Suzuki coupling, a preliminary
filtration was performed with this membrane. However, this ‘first
generation’ STARMEM membrane with a reported MWCO of
TM
200 membrane: the ionic liquid rejection increased to 94%. A
third filtration step even led to an ionic liquid rejection of 96%.
Note that the permeability of the membrane was higher in the
third step compared to the first step. This could be due to the
decreased ionic liquid concentration in the reaction mixture, due
to the dilution with 2-butanone.
2
20 in toluene, led to a rejection of only 52% for [BMPyr][Tf
2
N]
-
1
(
MW = 422.41 g mol ) in 2-butanone (Table 1, entry 5). Recently
Membrane Extraction Technology Ltd developed ‘second gen-
eration’ SRNF membranes referred to as DuraMem . These
membranes were also tested on their applicability for rejecting
ionic liquids. With DuraMem 200, the highest rejections for
our application (95%, entry 7) were achieved. DuraMem 300
exhibited a lower rejection value of 88% (entry 10). Despite this
lower rejection, the permeability through DuraMem 300 is
higher (0.5 L m h bar) compared to DuraMem 200 (0.15 L
TM
TM
TM
Summarizing, DuraMem 200 was found to be an excellent
membrane for the separation of the non-volatile hydroxy and
TM
carboxy ester from [BMPyr][Tf N] in the ozonation of acetals,
2
TM
with an ionic liquid rejection of 96%.
-
2
TM
-
2
m
h bar).
Based on the good results with DuraMem 200, the effect
Ozonation of methyl oleate
TM
of the filtration conditions was investigated. Increasing the
filtration temperature to 50 C resulted in a slight decrease in
The ozonation of methyl oleate (MO) is generally accepted
to proceed via the Criegee mechanism: the electrophilic ozone
molecule reacts with the C C bond with formation of a primary
◦
rejection, but the permeability was strongly enhanced (entry 7).
With methanol as solvent, lower IL rejections (91%) but higher
24
ozonide (1,2,3-trioxolane) (step (1), Scheme 2). This unstable
primary ozonide splits into Criegee intermediates (peroxides)
-
2
permeabilities (0.45 L m h bar) were obtained (entry 9). This is
possibly due to the higher affinity of the ionic liquid for methanol
than for 2-butanone, leading to a better solvation of the ionic
liquid. The smaller methanol solvent molecules drag the ionic
liquid more effectively through the membrane, leading to lower
rejections and higher permeabilities. Lowering the pressure to 20
25
and aldehydes (step (2), Scheme 2). These intermediates can
recombine to form the secondary ozonide (step (3), Scheme 2).
In a second phase, the secondary ozonides can be split oxida-
15,26
tively or thermally.
In the oxidative splitting, the secondary
ozonide reacts with oxygen with formation of two carboxylic
acids, pelargonic acid and monomethyl azelate. The optimal
bar of N led to decreased IL rejections (54%, entry 8). A possible
2
◦
15
explanation is that upon applying higher pressure, the polymer
chains in the membrane become more compact, resulting in
higher rejections. Such lower rejections at lower pressure were
temperature for the latter reaction is just below 100 C. This
reaction is traditionally performed in pure methyl oleate.
However, because of drawbacks such as evaporation losses
and contamination of oxygen gas with organic vapors and
organic particles, additional gas purification steps are required,
9
observed before in literature and are also consistent with the
22
solution–diffusion theory as presented by Wijmans and Baker:
15
the driving force for the solvent is the pressure difference
across the membrane. In contrast, solute transport is rather
influenced by the concentration gradient. A higher pressure
at the expense of extra energy and effort. By performing the
ozonolysis reaction in ionic liquids, these extra steps can be
avoided.
This journal is © The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 1726–1733 | 1729

3 2
Table 3 Ozonation of MO in [MOct N][Tf N] towards nonanal (1),
a
pelargonic acid (2), methyl 9-oxononanoate (3) and methyl azelate (4)
Yield (%)
a
a
Entry
Ozone phase
Oxygen phase
1
2
3
4
◦
◦
1
2
3
4
5
2 h/20 C
5 h/20 C
15
14
16
4
35
36
34
46
47
14
6
15
2
36
44
35
48
48
◦
◦
2 h/20 C
5 h/100 C
b
◦
◦
2 h/20 C
5 h/20 C
b
◦
◦
◦
2 h/20 C
5 h/100 C
7 h/100 C
bc
◦
0.25 h/20 C
3
2
a
1
mmol methyl oleate, 1.5 g [MOct
.781 mmol O /min; stirring: 1000 RPM; phase 2 (oxygen phase): 0.25 L
/min; stirring: 500 RPM. The mixture contained 10 wt% water with
3 2
][Tf N], phase 1 (ozone phase):
0
O
3
b
2
c
-1
respect to the ionic liquid. 0.50 L min
2
O .
In Fig. 3 the evolution of the product yields as a function
of the reaction time is depicted. After the ozonation phase, the
yield of the four ozonation products is around 25%. During the
oxygen phase, the formed aldehydes are transformed towards
the corresponding acids, with a summed acid yield of 94%,
comprising both mono- and diacids, after 7 h.
Scheme 2 Ozonolysis of methyl oleate via the Criegee mechanism (R
and R are –(CH –COOMe and –(CH –CH ).
1
2
2
)
7
2
)
7
3
In a first step, a screening of ionic liquids was made based on
good dissolution of methyl oleate, stability towards ozone and
high MW for a good separation of the reaction products with
SRNF. Out of the 15 tested ionic liquids, only methyltrioctylam-
-
1
monium bistriflimide ([MOct
3
N][Tf
2
N], MW = 648.85 g mol ),
][CF COO],
MW = 481.73 g mol ) and trihexyltetradecylammonium di-
methyltrioctylammonium trifluoroacetate ([MOct
3
3
-
1
-
1
cyanamide ([Hex
3
TP][N(CN)
2
], MW = 549,9 g mol ) exhibited
N][Tf N] was chosen as a
the desired characteristics. [MOct
3
2
model ionic liquid for the ozonation reaction, because of its
high MW. The methyl oleate ozonation to pelargonic acid and
monomethyl azelate is generally performed in two consecutive
phases: an ozone phase, typically performed at lower temper-
Fig. 3 Evolution of the product yields as a function of the reaction
time (reaction conditions: 1 mmol methyl oleate, 1.5 g [MOct
3
][Tf
2
N],
◦
1
) ozone phase: 0.25 h, 20 C, 0.78 mmol O
3
/min, 1000 rpm; 2) oxygen
◦
phase: 7h, 100 C, 0.50 L O
2
/min, 500 rpm).
15
atures, followed by a phase under pure oxygen. With both
◦
the ozone and oxygen phase at 20 C, the methyl oleate was
Post-reaction product separation by solvent resistant nanofil-
tration. For the separation of the acid and ester products
formed in the ozonolysis of methyl oleate, an analogous strategy
was applied as for the acetals ozonation. Different membranes
were tested, which are capable of retaining the larger IL while
the smaller products permeate. For this preliminary screening
samples to be filtered were prepared that simulate the reaction
fully converted in the ionic liquid to aldehydes and acids. As
expected, there was a 50–50 ratio between the mono- (nonanal
and pelargonic acid) and bifunctional products (methyl 9-
oxononanoate and monomethyl azelate) (Table 3, entry 1). One
can see from Table 3 that in the absence of water, increasing the
◦
◦
reaction temperature of the oxygen phase from 20 C to 100 C
did not influence the acid/aldehyde ratio of the monofunctional
products much, with an acid/aldehyde ratio of 2.4 and 2.6 at
mixtures at 90% substrate conversion. 0.5 g [MOct
3
N][Tf N],
2
0.0333 mmol of methyl oleate, 0.3 mmol of azelaic acid and
0.3 mmol of pelargonic acid were dissolved in 2-butanone as
solvent. 2-Butanone was chosen as a solvent as it solubilizes
both the IL and the acid products. In this screening, a membrane
was evaluated by its IL rejection (%) (Table 4).
◦
◦
2
0 C and 100 C respectively (entries 1–2). In the case of the bi-
functional products an increase to 7 was observed. Upon adding
1
0 wt% water and increasing the temperature during the oxygen
◦
phase to 100 C, the yields of pelargonic acid and monomethyl
azelate increased from 34 and 35 to 46 and 48% respectively
As reported in Table 4, despite the higher MW of
-
1
(
1
entries 3–4). It is well known that a reaction temperature around
[MOct
3
N][Tf N] (MW = 648.85 g mol ) the first generation
TM
2
◦
00 C is ideal for the oxygen phase in the ozonation of methyl
STARMEM membranes were again not capable of retaining
the IL: STARMEM 120 rejected the IL for only 25% (entry
1), whereas STARMEM 122, which has a MWCO of 220
15
TM
oleate. It seems that the higher temperature only promotes the
aldehyde to acid oxidation if water is present. Water presumably
functions as a proton carrier and source. Shortening the ozone
phase and prolonging the oxygen phase (oxygen flow doubled)
did not further improve the results (entry 5).
TM
Da, showed a 40% rejection (entry 2). The rather non-selective
TM
STARMEM membranes still exhibited high fluxes: 413 and
-
2
TM
TM
275 L m h for STARMEM 120 and STARMEM 122
1
730 | Green Chem., 2010, 12, 1726–1733
This journal is © The Royal Society of Chemistry 2010

Table 4 Screening of membranes for rejection of the IL in test mixtures
oxidation and the ozonation of methyl oleate. The acetals
are ozonized not only faster but also more selectively in
ionic liquids compared to the traditional low-boiling organic
solvents, with the only products being the hydroxy ester and
the over-oxidized carboxy ester. Methyl oleate was oxidized
ab
mimicking the methyl oleate reaction mixtures
c
Permeability (L m-2 h bar)
Entry Membrane
RIL (%)
TM
1
2
3
4
5
6
7
STARMEM 120
25 ± 3
40 ± 5
81 ± 1
96 ± 2
90 ± 0
93 ± 1
86 ± 5
10.3 ± 3.0
6.9 ± 5.7
TM
STARMEM 122
in [MOct
3
N][Tf N] towards pelargonic acid and monomethyl
2
TM
DuraMem 200
0.1 ± 0.025
0.2 ± 0.1
TM
azelate, with a carboxylic acid yield of 94%. After selection
of two suitable ionic liquids, 1-butyl-1-methylpyrrolidinium
bistriflimide and methyltrioctylammonium bistriflimide, it was
shown that solvent-resistant nanofiltration is an effective tech-
nique for separation of the ozonation products from the ionic
liquid. Based on nanofiltration on synthetic product mixtures, it
emerged that the highest permeability combined with a good IL
rejection of 90% was achieved with the PI supported zeolite filled
DuraMem 300
TM
d
DuraMem 300e
0.375 ± 0.05
0.35 ± 0.1
0.5 ± 0.075
TM
DuraMem 300f
TM
DuraMem 300
a
b
◦
Filtration conditions: p (N
Mixture: 0.5 g of [MOct
2
) = 40 bar, 1000 RPM stirring, T = 20 C.
3 2
N][Tf N], 0.00333 mmol of methyl oleate,
0
.3 mmol of azelaic acid, 0.3 mmol of pelargonic acid, 7 ml of 2-
c
d
◦
butanone. Rejection RIL : see Tables 1 and 2. Filtration T = 50 C.
e
f
2
Filtration pressure = 20 bar N . Methanol as solvent.
TM
PDMS. However, the polyimide-based DuraMem membranes
exhibited the highest rejections of ionic liquid (up to 96%).
Moreover, these polymeric membranes proved highly efficient
in separation of the products from real reaction mixtures. To
our knowledge it is the first time that a SRNF membrane can
separate reaction products efficiently from an ionic liquid like
respectively. A rejection of 40% is, however, too low for applying
these SRNF membranes for the product recovery from the ionic
liquid medium.
TM
The DuraMem membranes exhibited a substantially better
-
1
filtration performance. At 40 bar and with 2-butanone as
[BMPyr][Tf N] with a MW of only 422 g mol . In only 3 steps
2
TM
solvent, DuraMem 200 and particularly DuraMem 300 led
the IL was rejected for 96%. For the higher MW ionic liquid
to higher IL rejections, with IL rejections of 81% and 96%
respectively (entries 3–4). However, low membrane permeabilites
were obtained. These low permeabilities could be improved
[MOct N][Tf N], such rejections were even obtained after one
single filtration step.
3
2
-
2
to 0.375 L m h bar by increasing the filtration temperature
◦
◦
from 20 C to 50 C, however at the expense of an ionic
liquid rejection decreasing from 96 to 90% (entry 5). Reducing
Experimental
Materials
the N driving pressure to 20 bar also resulted in a lower
2
IL filtration efficiency: whereas the ionic liquid rejection was
slightly decreased, the membrane permeability was increased
Ionic liquids. Abbreviation list of ionic liquids: [MOct N]-
3
[Tf N] (methyltrioctylammonium bistriflimide with bistri-
2
(
entry 6). Replacing 2-butanone by methanol (entry 7) led
flimide = bis(trifluoromethylsulfonyl)imide); [BMPyr][N(CN)2]
to lower ionic liquid rejections (86%) but higher membrane
(1-butyl-1-methylpyrrolidinium dicyanamide); [BMPyr][Tf N]
2
-
2
permeabilities (0.5 L m h bar), just as in the case of the acetal
ozonation. This is remarkable since one would expect that a
more polar solvent like methanol would cause less swelling of
the membrane, thus leading to lower permeabilities and higher
rejections. This indicates that the interactions between solvent,
solute and membrane impart an important effect during the
SRNF process.
Again, the separation performances of the optimal mem-
brane, DuraMem
(1-butyl-1-methylpyrrolidinium bistriflimide); [Benzalk][Cl]
(alkyldimethylbenzylammonium chloride); [MOct ][CF COO]
3
3
(methyltrioctylammonium trifluoroacetate); [Hex TP][N(CN) ]
3
2
(trihexyltetradecylphosphonium dicyanamide). All [Tf N] and
2
[N(CN)2] based ionic liquids were obtained from Io-
LiTec. [Benzalk][Cl] was purchased from Sigma Aldrich.
[MOct ][CF COO] was obtained from Merck. The water content
21
3
3
of the ionic liquid batches used in the ozonation reactions and
filtration experiments was less than 100 ppm for the hydrophobic
ionic liquids and less than 1000 ppm for the halide containing
ionic liquids. In non-halide ionic liquids, such as triflimides
or dicyanamides, the halide content was typically less than
100 ppm.
TM
300, were evaluated under the actual
reaction conditions. The post-reaction mixture from the ozona-
tion of methyl oleate (1 mmol of methyl oleate, 1.5 g of
-
1
◦
[
MOct
phase 2 : 7 h, 0.250 L min
-butanone, pressurized with 40 bar N
3
][Tf
2
N], phase 1 : 0.5 h, 0.781 mmol min
O
3
, 25 C,
-
1
◦
O
2
, 100 C) was diluted in
2
2
and filtered with
TM
◦
DuraMem 300 (stirring: 1000 RPM; T = 20 C). In contrast
with the reaction mixture from the acetal ozonation, a one step
filtration of the reaction mixture resulted in a 96% rejection of
the IL with a permeability of 0.38 L m h bar. The experiment
was repeated five times, leading to a standard deviation of 2%
and 0.05 L m- h bar for the rejection and the permeability
respectively.
Reagents. 2-Methoxytetrahydropyran, methyl oleate, 2-
butanone, ester and acid reference compounds were obtained
from commercial sources.
-
2
Membranes. Abbreviation list of membranes: Zeol-
PDMS/PAIx (zeolite CBV 3002 filled PDMS coated polyami-
doimide, with x indicating the number of PDMS coatings); Zeol-
PDMS/PIx (zeolite CBV 3002 filled PDMS coated polyimide,
with x indicating the number of PDMS coatings).
2
Conclusions
TM
TM
The STARMEM and DuraMem membranes, asymmet-
ric SRNF membranes with an active layer of polyimide, in
‘dry’ form, were supplied by Membrane Extraction Technology
In summary, we proved that ionic liquids are excellent media
for ozone mediated reactions, more specifically for the acetal
This journal is © The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 1726–1733 | 1731

a
Table 5 Molecular weight cut off (MWCO) of commercial membranes
Typical procedure for acetal ozonation
b
Membrane
MWCO (Da)
Solvent
The acetal (10 mmol) was dissolved in 2 g of ionic liquid and
◦
-1
ozonized at 0–80 C (rate of ozone supply: 0.21 mmol min )
for 8 h. Ozone was produced with a Fischer Ozone Generator
type OZ 500. After reaction, the excess ozone was removed
by flushing the system with nitrogen. Reaction mixtures in
organic solvents were injected directly in the GC. The solutions
in ionic liquids were first silylated with MSTFA (20 mol
TM
STARMEM 120
200
220
150
200
300
toluene
toluene
TM
STARMEM 122
TM
DuraMem 150
acetone
acetone
acetone
TM
DuraMem 200
TM
DuraMem 300
a
b
Obtained from Membrane Extraction Technology Ltd. Solvent in
which the MWCO is defined.
◦
MSTFA/hydroxyl group; 80 C; 2 h) and extracted with diethyl
ether. The ether layer was analyzed with a Shimadzu 2014 GC
equipped with a FID detector and an apolar CP-Sil 5 CB
column. Note that after optimization of the filtration procedure,
the extraction step was not needed anymore, because the IL
could be simply retained with SRNF. The identity of the
reaction products was verified by GC-MS. An Agilent 6890 gas
chromatograph, equipped with a HP-5MS column, coupled to
a 5973 MSD mass spectrometer was used.
Ltd (UK) (www.membrane-extraction-technology.com). Their
MWCO’s are presented in Table 5.
Synthesis procedure of PDMS membranes
The zeolite filled polydimethylsiloxane (PDMS) coated poly-
imide (PI) membranes were synthesized according to the method
27
described by Dobrak et al. This membrane consists of a cross
linked polyimide (Matrimid 5218) support layer, coated with
zeolite (CBV 3002) filled PDMS. The PDMS layer is made from
the pre-polymer RTV 615A and the cross-linker RTV 615B in a
Typical procedure for ozonolysis of methyl oleate
Typically 1 mmol of methyl oleate was dissolved in 1.5 g of
◦
[MOct N][Tf N]. First, the methyl oleate was ozonized at 20 C
3
2
-
1
1
0 : 1 ratio.
(rate of ozone supply: 0.78 mmol min ) for 0.25–2 h. In a
second phase, after flushing with nitrogen, oxygen was bubbled
The zeolite filled PDMS coated polyamidoimide (PAI) mem-
◦
branes were prepared as follows. The support layer was prepared
by casting a Torlon (Solvay) solution (18 wt% PAI, 61.5 wt%
NMP and 20.5 wt% THF) on a polypropylene non-woven (FO
through the solution at 20–100 C (rate of oxygen supply:
-
1
0.250 L min ) for 3.5–7 h. After reaction, the reaction mixture
◦
was transesterified overnight at 50 C with 2% sulfuric acid in
2
471) kindly provided by Freudenberg (Germany). Therefore
methanol (0.5 ml) and extracted with n-heptane. The heptane
layer was analyzed analogously to the ether layer in the acetal
oxidation. For the identification of the reaction products: see
acetal oxidation.
a custom-made casting knife, set at a gap of 250 mm, was
used. After evaporation of the solvent (30 s), the resulting
film was immersed in water (20 min) at room temperature.
The prepared supports were afterwards subjected to a solvent-
exchange procedure, involving immersion in isopropanol for
three hours and subsequently for three days in a 40 : 40 : 20 (vol-
ume ratio) mixture of toluene/4-methyl-2-pentanone/mineral
Typical filtration procedure
Filtration experiments were performed on circular membranes
2
◦
(3.24 cm ) using a stainless steel dead-end filtration cell in high
oil. Afterwards these films were wiped and dried at 60 C for at
throughput mode (10 ¥ 10 ml) (Fig. 4). The ten filtration cells
were filled with approximately 7 ml of the mixtures, stirred with
a magnetic stirrer on a stainless steel filter plate positioned 7 mm
above the membrane surface (1000 RPM) and pressurized with
least one hour. The first step in coating the support with PDMS is
the preparation of the PDMS/filler suspension. Therefore 0.5 g
of zeolite filler (CBV 3002 from PQ-corporation) was dispersed
in hexane in an ultrasonic bath for one hour, followed by addition
of 0.14 g of cross linker (RTV 615B). The mixture was stirred for
two hours at 40 C. Finally, 1.4 g of pre-polymer (RTV 615A) was
added and the mixture was stirred for another hour at 60 C (pre-
polymerization). The PDMS/filler solutions were then coated
on the PI support, which was taped onto a stainless steel plate,
under an angle of 60 . After evaporation of most of the solvent,
this coating procedure was repeated with the membrane turned
2
0–40 bar of N
2
. The filtrations were carried out at temperatures
◦
◦
between 20–50 C. The first ml of the filtrations was discarded
(to avoid measuring under initial non-equilibrium conditions).
For determination of rejections and fluxes the second and
third ml were used. The ionic liquid rejection was calculated as
◦
◦
28
◦
upside-down. The membrane was placed overnight at 110 C to
remove the rest of the solvent and to complete the cross linking.
Ozone stability test
Ozone resistance of ionic liquids can be evaluated by NMR,
as well as macroscopically by investigating colour and viscosity
changes. NMR spectroscopic measurements were carried out
1
with a Bruker AMX-300 at 300 MHz ( H) in deuterium oxide
1
3
(
D
2
O) and 75.5 MHz ( C) in deuterated chloroform (CDCl )
3
with tetramethylsilane (TMS) as standard. Additional NMR
spectra were recorded on a Bruker Avance 400 spectrometer.
Fig. 4 High-throughput filtration cell (10 ¥ 10 ml) designed in-house.
This journal is © The Royal Society of Chemistry 2010
1
732 | Green Chem., 2010, 12, 1726–1733

R
IL = ((1 - C
P
/C
F
) ¥ 100%) with C
P
the concentration of the
6 C. Van Doorslaer, J. Wahlen, P. G. N. Mertens, B. Thijs, P.
Nockemann, K. Binnemans and D. E. De Vos, ChemSusChem, 2008,
ionic liquid in the permeate and C
F
the concentration of the ionic
1
, 997.
liquid in the feed. The ionic liquid concentration in feed and
permeate was measured using conductometry using a Consort
K 620 laboratory conductometer with Pt electrode. Firstly, a
calibration curve which indicates the ionic liquid concentration
as a function of the measured current was made based on
mixtures with known ionic liquid concentrations. Based on
this curve, the ionic liquid concentration in the permeates was
determined.
7
H. Wong, C. J. Pink, F. C. Ferreira and A. G. Livingston, Green
Chem., 2006, 8, 373.
8 A. Cano-Odena, P. Vandezande, D. Fournier, W. Van Camp, F. E.
Du Prez and I. F. J. Vankelecom, Chem.–Eur. J., 2010, 16, 1061.
9
J. Kr o¨ ckel and U. Kragl, Chem. Eng. Technol., 2003, 26, 1166.
1
1
0 S. Han, H. T. Wong and A. G. Livingston, Chem. Eng. Res. Des.,
2005, 83, 309.
1 (a) P. Deslongchamps and C. Moreau, Can. J. Chem., 1971, 49,
2
465; (b) P. Deslongchamps, P. Atlani, D. Frehel, A. Malaval and
C. Moreau, Can. J. Chem., 1974, 52, 3651; (c) R. J. Taillefer, S. E.
Thomas, Y. Nadeau, S. Fliszar and H. Henry, Can. J. Chem., 1980,
58, 1138.
In the multi-step filtration experiments, the permeate was
collected from just after the pressurisation until 4 ml of the feed
1
1
1
2 A. Behr and J. P. Gomes, Eur. J. Lipid Sci. Technol., 2010, 112, 31.
3 P. A. Thomas and B. B. Marvey, Int. J. Mol. Sci., 2009, 10, 5020.
4 A. K o¨ ckritz, M. Blumenstein and A. Martin, Eur. J. Lipid Sci.
Technol., 2009, 111, 58.
(
9 mL) had permeated. The same high throughput filtration cell
and conditions were used as in the single step filtration. After
filtration, the filtration cell was depressurized and approximately
4
ml of 2-butanone was added to the permeate, to reach again the
15 C. Goebel, A. Brown, H. Oehlschlaeger, R. Rolfes, US Patent no.
2
813113, United States Patent Office, 1957.
initial feed volume (9 ml). This new feed was then subjected to a
next filtration step. In total, three filtration steps were performed.
1
1
6 J. Pernak and I. Goc, Pol. J. Chem., 2003, 77, 975.
7 P. Vandezande, L. E. M. Gevers and I. F. J. Vankelecom, Chem. Soc.
Rev., 2008, 37, 365.
1
1
8 J. L. Anthony, J. L. Anderson, E. J. Maginn and J. F. Brennecke,
J. Phys. Chem. B, 2005, 109, 6366.
9 (a) I. F. J. Vankelecom, K. De Smet, L. E. M. Gevers, A. Livingston,
D. Nair, S. Aerts, S. Kuypers and P. A. Jacobs, J. Membr. Sci., 2004,
231, 99; (b) L. E. M. Gevers, G. Meyen, K. De Smet, P. V. De Velde,
I. F. J. Vankelecom and P. A. Jacobs, J. Membr. Sci., 2006, 274, 173.
References
1
(a) P. Wasserscheid and T. Welton, Ionic Liquids in Synthesis, Wiley-
VCH, 2008; (b) T. Welton, Chem. Rev., 1999, 99, 2071; (c) N. V.
Plechkova and K. R. Seddon, Chem. Soc. Rev., 2008, 37, 123;
(
d) V. I. Parvulescu and C. Hardacre, Chem. Rev., 2007, 107, 2615;
e) U. Doma n´ ska and R. Bogel-Łukasik, J. Phys. Chem. B, 2005,
09, 12124; (f) Y. U. Paulechka, G. J. Kabo, A. V. Blokhin, O. A.
Vydrov, J. W. Magee and M. Frenkel, J. Chem. Eng. Data, 2003, 48,
57.
20 P. Vandezande, L. E. M. Gevers, J. S. Paul, I. F. J. Vankelecom and
P. A. Jacobs, J. Membr. Sci., 2005, 250, 305.
21 L. E. M. Gevers, I. F. J. Vankelecom and P. A. Jacobs, J. Membr. Sci.,
2006, 278, 199.
22 J. G. Wijmans and R. G. Baker, J. Membr. Sci., 1995, 107, 1.
23 L. G. Peeva, E. Gibbins, S. S. Luthra, L. S. White, R. P. Stateva and
A. G. Livingston, J. Membr. Sci., 2004, 236, 121.
24 N. Soriano, V. Migo and M. Matsumura, Chem. Phys. Lipids, 2003,
126, 133.
25 M. Mochida, Y. Katrib, J. Jayne, D. Worsnop and S. Martin, Atmos.
Chem. Phys., 2006, 6, 4851–4866.
26 D. Sparks, L. Est e´ vez and R. Hernandez, Green Chem., 2009, 11,
986.
27 A. Dobrak, A. Figoli, S. Chovau, F. Galiano, S. Simone, I. F. J.
Vankelecom, E. Drioli and B. Van der Bruggen, J. Colloid Interface
Sci., 2010, DOI: 10.1016/j.jcis.2010.02.023.
28 L. E. M. Gevers, S. Aldea, I. F. J. Vankelecom and P. A. Jacobs,
J. Membr. Sci., 2006, 281, 741.
(
1
4
2
3
C. Van Doorslaer, A. Peeters, P. Mertens, C. Vinckier, K. Binnemans
and D. De Vos, Chem. Commun., 2009, 6439.
J. A. Wojtowicz, The Kirk-Othmer Encyclopedia of Chemical-
Technology—Electronic Delivery with A Focus on Data, Wiley VCH,
2
005, 17.
4
5
K. Koike, M. Nifuku, K. Izumi, S. Nakamura, S. Fujiwara and S.
Horiguchi, J. Loss Prev. Process Ind., 2005, 18, 465.
M. Kanakidou, J. H. Seinfeld, S. N. Pandis, I. Barnes, F. J. Dentener,
M. C. Facchini, R. Van Dingenen, B. Ervens, A. Nenes, C. J.
Nielsen, E. Swietlicki, J. P. Putaud, Y. Balkanski, S. Fuzzi, J. Horth,
G. K. Moortgat, R. Winterhalter, C. E. L. Myhre, K. Tsigaridis, E.
Vignati, E. G. Stephanou and J. Wilson, Atmos. Chem. Phys., 2005, 5,
1
053.
This journal is © The Royal Society of Chemistry 2010
Green Chem., 2010, 12, 1726–1733 | 1733