Direct fluorination of carbon monoxide in microreactors

Article abstract of DOI:10.1016/j.jfluchem.2012.06.006

Many attempts to obtain a clean stream of COF₂ have been carried out in the past by means of the direct fluorination of carbon monoxide with elemental fluorine or by electrochemical fluorination. The reaction is highly exothermic, therefore dif

Full text of DOI:10.1016/j.jfluchem.2012.06.006

Journal of Fluorine Chemistry 142 (2012) 19–23
Contents lists available at SciVerse ScienceDirect
Journal of Fluorine Chemistry
journal homepage: www.elsevier.com/locate/fluor
Direct fluorination of carbon monoxide in microreactors
c
d
d
Walter Navarrini a,b, Francesco Venturini a, , Vito Tortelli , Soubir Basak , Ketan P. Pimparkar ,
*
Andrea Adamo d, Klavs F. Jensen d
a Dipartimento di Chimica, Materiali e Ingegneria Chimica, Politecnico di Milano, 7 Via Mancinelli, I-20131 Milano, Italy
b Consorzio Interuniversitario Nazionale per la Scienza e Tecnologia dei Materiali, 9 Via G. Giusti, I-50121 Firenze, Italy
c Solvay-Solexis, R&D Centre, 20, Viale Lombardia, I-20021 Bollate (MI), Italy
d Chemical Engineering Department, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, United States
A R T I C L E I N F O
A B S T R A C T
Article history:
Many attempts to obtain a clean stream of COF2 have been carried out in the past by means of the direct
fluorination of carbon monoxide with elemental fluorine or by electrochemical fluorination. The reaction
is highly exothermic, therefore difficult to control. It can easily develop into a thermal runaway with a
poor selectivity. We have successfully circumvented these critical issues by using a stainless steel
Received 27 March 2012
Received in revised form 27 May 2012
Accepted 2 June 2012
Available online 23 June 2012
parallel channel microreactor (surface/volume ratio ꢀ 1 ꢁ 104
m
ꢂ1, residence time
t
ꢀ 0.1 s) for the
direct fluorination of carbon monoxide. Its performance in terms of operability and selectivity is
compared to that of a standard reactor assembly, namely a fluorine burner reactor coupled with a water
cooled heat exchanger. While the microreactor assembly succeeded to control the exothermic reaction,
in the same experimental conditions the standard assembly reactor underwent serious corrosion issues
that lead to nozzle meltdown lack of selectivity and consequent plant shutdowns.
ß 2012 Elsevier B.V. All rights reserved.
Keywords:
Carbonyl-difluoride
Hypofluorite
Synthesis
Microreactor
Direct fluorination
1. Introduction
Recently a flame-less approach has been patented [18]. It is
reported that the two reactants, fluorine and carbon monoxide are
High purity COF2 is required in many applications [1–10]. The
usual approach to obtaining this compound is the direct
fluorination of carbon monoxide with elemental fluorine. In
general the reaction of organic substances with fluorine, in
conventional reactors, is difficult to control due to the heat
generated by the reaction itself [11]. The fluorination of a
hydrogenated compound is a typical example where the exo-
thermicity of the whole reaction is such that any carbon–carbon
bond can be easily broken if the reaction is not properly controlled
by dilution of reagents and by adopting a low reaction temperature
and CH activation [12,13]. The improperly conducted fluorination
leads invariably to several byproducts and ultimately to CF4 and HF
[14].
Attempts to obtain a clean stream of COF2 have been carried out
mainly by trying to reduce the flame temperature via massive
dilution of the inlet streams or by switching to electrochemical
fluorination [15,16]. The introduction of inert gases implies, at a
fixed contact time, an increase of the reactor volume or, at a
constant volume, a decrease in plant productivity. As a matter of
fact, only a massive introduction of a dilution stream results in a
lower flame temperature and hence in a higher selectivity [5,9,17].
continuously bubbled in a continuous stirred-tank reactor (CSTR)
filled with a solvent [19]. This system has clearly some critical
points, i.e. solvent stability [20] and operability.
A good practice to obtain pure carbonyl-difluoride is to reduce
the flame temperature by increasing the heat exchange surface.
Due to the large amount of heat released during the reaction, the
surface/volume ratio of the reactor has to be high, thus the use of a
microreactor seems to be a suitable option. In fluorine chemistry,
there is a large interest in the development of microreactors for
chemical processing [21–25] due to the benefits that this
technological device could provide: better reaction control,
arithmetic scale-up and higher safety. Herein we report the use
of a microreactor to test the highly exothermic gas phase reaction
CO + F2 with the aim of better controlling the temperature and
increase the reaction selectivity.
2. Materials and methods
Pure fluorine (98%, from Solvay Fluor) and pure carbon
monoxide (99.8%, from Sapio) were fed in the reactor without
any further dilution. The experiments were performed by setting
the carbon monoxide flow-rate to 1 NL/h (44.6 mmol/h) and acting
on the fluorine flow-rate in order to reproduce three experimental
conditions: a lean, a stoichiometric and a rich combustion. The
composition measurements of the gas produced were taken
* Corresponding author. Tel.: +39 02 2399 3035; fax: +39 02 2399 3080.
E-mail address: francesco.venturini@chem.polimi.it (F. Venturini).
0022-1139/$ – see front matter ß 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jfluchem.2012.06.006

20
W. Navarrini et al. / Journal of Fluorine Chemistry 142 (2012) 19–23
Fig. 1. Left: exploded view of the MIT micro-reactor assembly. Right: simplified scheme of the fluorination micro-reactor plate. The dashed lines refer to structures on the back
side whereas the continuous lines indicate structures on the top side.
20 min after the imposed steps giving the system enough time to
reach a steady-state, if any.
coming from the microplant were neutralized through two
columns filled with pellets of soda lime (abatement system).
The standard reactor assembly consisted of one 1/8 in.
Swagelok tee where the two reactants were fed through two 1/
8 in. tubes. In addition 2 m of 1/8 in. tube has been installed after
the tee. The standard reactor apparatus was cooled using a stirred
bath to dissipate the heat generated by the chemical reaction. The
fluid used was Solvay Specialty Polymers Galden1 HT 110 for low
temperatures and water for tests above room temperature.
A heated safety vessel of 25 cm3 was installed after both the
reactors to provide a dead volume for transient control since a
poorly controlled plant could cyclically lead to the formation of
lean and rich mixtures. If a fluorine rich mixture would come in
contact with carbon monoxide in the same volume an explosion
would be expected. If this uncontrolled reaction happened in the IR
cell, it could damage the windows and the spectrophotometer. The
influence of the safety vessel was negligible on the conversion and
selectivity of the two reactors under comparison.
The construction of a microreactor suitable for direct fluorina-
tion reactions has several challenges. The microreactor needs to be
accurately designed and dimensioned at a microscale level. It must
also be constructed by using corrosion resistant materials able to
withstand a hostile environment, at extremely high temperature
for a long time. The microreactor used for the experiments (Fig. 1)
is based on a design of de Mas et al. [24,26] and has a volume of
65 mm3 and a surface/volume ratio of 1 ꢁ 104 mꢂ1. The original
microreactor was realized in silicon and nickel coated, but to
withstand the corrosion issues associated with pure fluorine and
carbonyl-difluoride and to increase the heat transferred via
conduction, alloys like AISI 316 and R400 were used in the present
version. The reactor plates were machined via electron discharge
while the gaskets were water-jet cut from a perfluoro-elastomer
(FFKM) sheet.
The microreactor resembled a very small heat exchanger and
was constructed by assembling many stacked parts: the enclosure
consisted of two thick plates of stainless steel that integrates the
temperature sensors, process inlets, outlets, and coolant ports. The
internal combined stainless steel reaction and cooling plates
machined with gas and coolant distribution systems were inserted
between two counter-plates and separated with FFKM gaskets as
illustrated in Fig. 1. Each reaction-cooling plate contained on the
front side 20 parallel microchannels for reaction and 20 cross flow
parallel microchannels for cooling on the back. The coolant
channels were embedded using a computer controlled milling
machine. The reaction channel dimensions were 25 mm long,
Among the signals coming from the collected IR spectra only
one ‘‘characteristic absorption peak’’ for each compound was
chosen for the on-line monitoring. The unique non-overlapping
signal of carbon monoxide lies at 2120 cmꢂ1 whereas carbonyl-
difluoride absorbs the IR radiation at 1559 cmꢂ1, 1900 cmꢂ1
2489 cmꢂ1, 2924 cmꢂ1 and 3860 cmꢂ1. The peak at 1900 cmꢂ1 is
too intense, but the other peaks at 1559 cmꢂ1, 2924 cmꢂ1
,
,
3860 cmꢂ1 are too weak and noisy to be used. Thus, the preferred
peak for COF2 is 2489 cmꢂ1. Trifluoromethyl hypofluorite (CF3OF)
has absorption bands at 1378 cmꢂ1 and 1489 cmꢂ1. Only the first
band was used.
400
m
m wide and 400
m
m deep. To ensure a uniform distribution
For every IR scan the relative ratio between peak heights
belonging to a single substance are calculated as reported in Table
1.
of the reactants in all the microchannels a neck was machined at
the beginning of each channel.
The thermal energy produced during the reaction was dispersed
from the gas phase through the microreactor plates and dissipated
into the cooling fluid (Water, Solvay Specialty Polymers Galden1
HT 110/HT 55 or CF3O–CFCl–CF2Cl, depending on the target
temperature). The amount of heat removed from the coolant was
controlled by setting the cryostat temperature. Cooling the system
was critical since the loss of temperature control of the reaction
could damage the reactor plates and seals.
Any scan that did not match the relative peak ratio of the pure
substance was discarded. Data collected from the IR spectra were
normalized against carbonyl-difluoride peaks to provide an
immediate performance comparison of the two systems.
Table 1
The analytic apparatus consists of
a Nicolet 380 FT-IR
Relative ratio of IR peak height.
spectrometer equipped with a 10 cm long PTFE cell with calcium
fluoride windows (recommended spectrum width 7400–
1111 cmꢂ1) and a GC-TCD (Dani 6500) with PFA packed column
(Kel-F oil on Hyflon MFA1 perfluorinated plastic). The output of the
TCD could be sampled and injected into a GC–MS system (Agilent
6850+5975C with PLOT-U column). After the analyses the off-gases
Substance
Peaks
Ratio
Carbonyl-difluoride
Abs (2489 cmꢂ1)/Abs (1559 cmꢂ1
Abs (2924 cmꢂ1)/Abs (1559 cmꢂ1
Abs (1489 cmꢂ1)/Abs (1378 cmꢂ1
)
)
)
1.62 ꢃ 0.04
0.502 ꢃ 0.006
0.71 ꢃ 0.08
Trifluoromethyl
hypofluorite

W. Navarrini et al. / Journal of Fluorine Chemistry 142 (2012) 19–23
21
Table 2
2.1. Safety
Operative conditions applied.
Reactants and products are toxic [27]. Pure fluorine and
Test number
Temperature
F2/CO ratio
Notes
carbonyl-difluoride are heavier than air and react quickly with
water to form the hydrogen fluoride. Burns caused by fluorine or
carbonyl-difluoride require immediate medical treatment with
calcium gluconate since the symptoms of poisoning, often become
manifest only after some hours. Both compounds have a pungent
odor and adequate warning properties, thus a small leak can be
quickly detected and subsequent actions can be taken. Localized
fluorine leaks can be detected by using a cotton tip wetted with
starch iodide solution while carbonyl-difluoride gives acid
reaction.
1
2
ꢂ18 8C
ꢂ18 8C
ꢂ18 8C
ꢂ18 8C
ꢂ35 8C
ꢂ50 8C
ꢂ62 8C
ꢂ73 8C
25 8C
0.96
1.22
0.80
1.99
1.22
1.22
1.22
1.22
1.22
1.22
CO = 1 L/h
CO = 1 L/h
3
F2 =0.96 L/h
CO = 1 L/h
CO = 1 L/h
CO = 1 L/h
CO = 1 L/h
CO = 1 L/h
CO = 1 L/h
CO = 1 L/h
4
5
6
7
8
9
10
100 8C
On the contrary, the presence of flammable, toxic and
teratogenic carbon monoxide in the environment cannot be
promptly detected since it does not have any warning properties.
Working with these gases requires good suction fume hoods
and appropriate personal protective equipment like neoprene
gloves and a full piece mask.
output coming from the FT-IR spectrometer, GC–TCD and GC–MS
system. The microreactor performances in terms of operability and
selectivity were compared with a standard reactor having a
surface/volume ratio (S/V) equal to 700 mꢂ1 by adopting a resident
time of
t
ꢀ 10 s.
The operative conditions applied while GC and IR data were
recorded are summarized in Table 2.
3. Results and discussion
In Fig. 2, we report the IR spectra recorded in preliminary tests
under lean, rich and stoichiometric conditions obtained in the
standard reactor. Under rich conditions, the adsorption band due
to the excess of carbon monoxide can be noticed along with the
signals of the main product COF2. When the reaction was carried
out using an equal reactant flows the carbon monoxide band
disappeared. When the reaction was carried out with an excess of
fluorine, trifluoromethyl hypofluorite (CF3OF) bands appeared at
1378 cmꢂ1 and 1489 cmꢂ1. GC analyses confirmed that the
amount of trifluoromethyl hypofluorite produced was approxi-
mately 10% of the COF2 stream in our experimental conditions
(room temperature F2/CO = 1.9 (v/v), standard assembly).
The presence of trifluoromethyl hypofluorite was expected
when carbonyl-difluoride and elemental fluorine are simulta-
neously present in the reactor [4,7,30,31]. When changing the
experimental conditions from the rich to the lean tests a mismatch
in peak ratio, shown in Table 1, was observed. The explanation of
this measure relies on the fact that the weaker signal was noisy,
therefore any mathematic operation on his modulus is affected by
a significant error.
The synthesis of carbonyl-difluoride (Scheme 1) is recognized
as a fast and exothermic radical gas phase reaction [28].
The equilibrium temperature for the reaction with pure fluorine
and pure carbon monoxide has been calculated to be around
3700 K at an estimated conversion of 38%. The calculated adiabatic
temperature [29] in the presence of a standard 75% nitrogen
dilution is 5015 K with a 1.5% conversion. The adiabatic tempera-
ture calculation made in the absence of nitrogen, thus with pure
fluorine, shows a value that is several orders of magnitude higher
than the one calculated in the presence of nitrogen dilution. In the
absence of proper temperature control the reaction is also prone to
thermal run-away (Scheme 2). If the local reaction temperature
exceeds 670 K many byproducts (Scheme 3) are formed.
It is also difficult to separate carbonyl-difluoride from the
byproducts formed by the reactions shown in Scheme 3 because of
their low concentration and similar boiling points. Moreover, the
similar chemical properties of these byproducts make it difficult to
separate them by adsorption or chemical neutralization methods.
To overcome these analytical disadvantages a specific online
gas analysis system was assembled. The gases produced in the
standard reactor systems or in the microreactor were analyzed by
an online analytic system allowing a continuous evaluation of the
CO + F2 → COF2
ΔH°R = - 530 kJ/mol
Scheme 1. Direct fluorination of carbon monoxide leading to carbonyl-difluoride.
CO + 2 F2 → CF4 + 0.5 O2
ΔH°R = - 820 kJ/mol
Scheme 2. Direct fluorination of carbon monoxide leading to tetrafluoromethane.
a) 2 CO + O2 →2 CO2
b) COF2 + F2 → CF3OF
c) CF3OF + COF2 → CF3-OO-CF3
d) CO2 + F2
CF2(OF)2
e) CF3OF→ CF4 + 0.5 O2
e) CF3OF + CO → CF3O-C(O)F
f) CF3O-C(O)F→ CO2 + CF4
Fig. 2. IR spectra obtained in the standard reactor. Top IR – excess of carbon
monoxide (F2/CO = 0.93, v/v) (CO band at 2120 cmꢂ1). Central IR – stoichiometric
g) 2 CO + F2 + O2 CF(O)-OO-C(O)F
mixture (F2/CO = 1, v/v) (carbonyl-difluoride bands at 1559 cmꢂ1 2489 cmꢂ1
, ,
2924 cmꢂ1 and 3860 cmꢂ1). Bottom IR – excess of fluorine (F2/CO = 1.9, v/v)
(hypofluorite bands at 1378 cmꢂ1 and 1489 cmꢂ1).
→
Scheme 3. Side reactions that might take place in the reaction environment.

22
W. Navarrini et al. / Journal of Fluorine Chemistry 142 (2012) 19–23
30%
30%
25%
20%
15%
10%
5%
25%
20%
15%
10%
5%
0%
0.75
0%
0.75
0.95
1.15
1.35
1.55
1.75
1.95
2.15
0.95
1.15
1.35
1.55
1.75
1.95
2.15
F2/CO
F2/CO
Fig. 3. IR absorbance ratios in the tests performed by varying the stoichiometry at ꢂ18 8C for the standard assembly (dashed line) and the microreactor (continuous line). Left:
AbsCF OF=Abs . Right: Abs =Abs . The vertical red line represent the ideal stoichiometric 1:1 ratio between CO and F2. (For interpretation of the references to color in
ðCOF
Þ
ðCOÞ
ðCOF Þ
2
3
2
this figure legend, the reader is referred to the web version of the article.)
20%
18%
16%
14%
12%
10%
8%
20%
18%
16%
14%
12%
10%
8%
6%
6%
4%
4%
2%
2%
0%
-90
0%
-90
-40
10
T [°C]
60
110
-40
10
T [°C]
60
110
Fig. 4. IR absorbance ratios in the tests performed by varying the temperature at constant stoichiometry (F2/CO = 1.22) for the standard reactor assembly (dashed line) and the
microreactor (continuous line). Left: AbsCF OF=Abs . Right: Abs =Abs
.
ðCOF Þ
2
ðCOF
Þ
ðCOÞ
3
2
Detailed experiments were performed at constant temperature
(ꢂ18 8C) in both the reactors assembly, standard and microreactor.
The carbon monoxide conversion (Fig. 3) is about 30% higher in the
microreactor. In both cases the amount of CF3OF produced is
linearly proportional to the amount of pure fluorine introduced in
the system. However, the amount of trifluoromethyl hypofluorite
produced in the microreactor was up to 40% lower than that in the
standard reactor test, even at high F2/CO ratio.
At stoichiometric flow rates, an unstable operation condition
was observed in the standard reactor because at this condition
(CF3OF) and (CO) were simultaneously present in the gas phase.
This thermodynamically unstable mixture may eventually burn in
the heated safety vessel, but no evidence of this has been found.
This unstable condition was less evident in the microreactor
because the amount of trifluoromethyl hypofluorite and carbon
monoxide produced was much lower compared to the trifluor-
omethyl hypofluorite and carbon monoxide produced in the
standard reactor test (Fig. 4).
without temperature control in the safety vessel, but outside the
reactor. This behavior was also confirmed by measuring the
temperature increase of the gases in the safety vessel, along with
the temperature increase of the tubes connecting the microreactor
with the safety vessel. This unstable operative condition was also
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
At constant F2/CO ratio (1.22) another unstable operating
condition appears. At temperatures lower than ꢂ50 8C (Fig. 4) the
reaction performed in the microreactor lacks conversion and
selectivity. Under those conditions the COF2 synthesis does not
ignite in the microreactor channels, probably also due to the active
termination of the radicals on the walls. A mixed stream of fluorine
and carbon monoxide then leaves the microreactor and burns
-100
-50
0
50
100
150
T [°C]
Fig. 5. Concentration of COF2 (area basis) in the tests performed by varying the
temperature at constant stoichiometry (F2/CO = 1.22) for the standard reactor
assembly (dashed line) and the microreactor (continuous line).

W. Navarrini et al. / Journal of Fluorine Chemistry 142 (2012) 19–23
23
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
CF3OF
COF2
CF4 + CO2
CO + O2
F2
100
20
-18
-50
-62
-73
100 25
-18
-35
-50
-62
-73
T [°C]
T [°C]
Fig. 6. Product distribution from GC–TCD analyses for the tests at F2/CO = 1.22 at various temperatures. Left: microreactor. Right: standard reactor.
observed in the standard reactor, but it was less noticeable due to
its large volume.
Acknowledgment
For reaction temperatures higher than ꢂ35 8C (Figs. 5 and 6) the
microreactor had 15–50% higher selectivity than the standard
reactor. GC–TCD, GC–MS and IR data of the effluent produced in
low-selectivity tests carried in both reactors revealed that along
with CF3OF many other by products were present: tetrafluor-
omethane, carbon dioxide and oxygen.
Interestingly in the low selectivity tests carried out with the
standard reactor (Fig. 6), even if a fluorine excess was fed some
unconverted carbon monoxide was still present in the output
gases. In these conditions the reaction mixture become unstable
and the overall reaction leading to COF2 was characterized by low
selectivity with clear evidence of byproducts formation as shown
in Scheme 3.
Under these conditions 1/10th or more of the carbon monoxide
was no longer converted while the whole excess of fluorine reacted
with COF2 giving CF3OF through reaction (b) of Scheme 3. At high
temperature the trifluoromethyl hypofluorite can decompose
through reaction (e) of Scheme 3 forming CF4 and oxygen [14,32],
finally CO2 was formed through the oxidation of carbon monoxide
through reaction (a) of Scheme 3. As a result, the final mixture
consists of a stream containing COF2, CF3OF, CF4, O2 and CO2.
The corresponding author is grateful to Solvay-Solexis for
financing this research.
References
[1] V. Tortelli, P. Calini, A. Zompatori, E. Antenucci, EP 1801090 A2 (2007).
[2] A. Yamamoto, H. Kume, S. Takubo, JP 4415656 B (2010).
[3] R.D. Smith, F.S. Fawcett, D.D. Coffman, Journal of the American Chemical Society
84 (1962) 4285–4288.
[4] W. Navarrini, V. Tortelli, A. Russo, S. Corti, Journal of Fluorine Chemistry 95 (1999)
27–39.
[5] A. Russo, W. Navarrini, Journal of Fluorine Chemistry 125 (2004) 189–197.
[6] W. Navarrini, S. Corti, Journal of Fluorine Chemistry 125 (2004) 73–78.
[7] W. Navarrini, F. Venturini, M. Sansotera, M. Ursini, P. Metrangolo, G. Resnati, M.
Galimberti, E. Barchiesi, P. Dardani, Journal of Fluorine Chemistry 129 (2008) 680–
685.
[8] F. Venturini, P. Metrangolo, G. Resnati, W. Navarrini, V. Tortelli, Chimica Oggi 26
(2008) 36–38.
[9] S. Rozen, C. Gal, Journal of Organic Chemistry 52 (1987) 4928–4933.
[10] W. Navarrini, A. Russo, V. Tortelli, Journal of Organic Chemistry 60 (1995) 6441–
6443.
[11] H.-C. Wei, R.J. Lagow, Chemical Communications (2000) 2139–2140.
[12] S. Rozen, C. Gal, Y. Faust, Journal of the American Chemical Society 102 (1980)
6860–6861.
[13] F. Venturini, W. Navarrini, A. Famulari, M. Sansotera, P. Dardani, V. Tortelli,
Journal of Fluorine Chemistry (2012) 6–11.
[14] G. Brauer (Ed.), 2nd ed., Handbook of Preparative Inorganic Chemistry, vol. 1,
Academic Press, 1963.
[15] R.J. Lagow, J.L. Margrave, Progress in Inorganic Chemistry 26 (1979) 161–210.
[16] W.V. Childs, US3461050 (1969).
4. Conclusions
Microreactors are an industrial technology already used in
pharmaceutical and fine chemistry. With these devices it is
possible to carefully control a highly exothermic chemical reaction
such as the synthesis of COF2 from pure CO and F2 avoiding unsafe
conditions, product decomposition and runaways. In our experi-
ence the microreactor increased the selectivity of the reaction by a
range between 15% and 50% and enhanced the carbon monoxide
conversion up to 30%. The trifluoromethyl hypofluorite, an
unstable byproduct of the COF2 synthesis produced also in the
microreactor was up to 40% lower than in the common standard
reactor. On the laboratory scale, stainless steel microreactors were
proven to be safe and reliable for direct fluorination reactions with
pure fluorine. The microreactor has the capability to control and
dissipate the large amount of heat produced during the reaction,
while at the same experimental conditions the standard reactor
undergoes serious corrosion issues that lead to nozzle meltdown,
lack of selectivity, frequent plant shutdowns, damages and poor
operability.
[17] V. Tortelli, P. Calini, A. Zompatori, E. Antenucci, EP 1801091 A2 (2006).
[18] Y. Mitsui, K. Yanase, US 7371899 B2 (2008).
[19] J.J. Kampa, J.W. Nail, R.J. Lagow, Journal of Fluorine Chemistry 102 (2000) 333–
335.
[20] H. Ohtani, Journal of Loss Prevention in the Process Industries 17 (2004) 377–
379.
[21] K. Ja¨hnisch, M. Baerns, V. Hessel, W. Ehrfeld, V. Haverkamp, H. Lo¨we, C. Wille, A.
Guber, Journal of Fluorine Chemistry 105 (2000) 117–128.
[22] R.D. Chambers, D. Holling, A.J. Rees, G. Sandford, ChemInform 34 (2003) 2002–
2003.
[23] R.D. Chambers, M.A. Fox, G. Sandford, Lab on a Chip 5 (2005) 1132–1139.
[24] N. de Mas, A. Guu¨ nther, M.A. Schmidt, K.F. Jensen, Industrial and Engineering
Chemistry Research 48 (2009) 1428–1434.
[25] G. Sandford, Journal of Fluorine Chemistry 128 (2007) 90–104.
[26] N. de Mas, A. Gu¨ nther, M.A. Schmidt, K.F. Jensen, Industrial and Engineering
Chemistry Research 42 (2003) 698–710.
[27] CDC – NIOSH Pocket Guide to Chemical Hazards, http://www.cdc.gov/niosh/npg.
[28] S. Rozen, Chemical Reviews 96 (1996) 1717–1736.
[29] Nist Chemistry Webbook, http://webbook.nist.gov.
[30] S. Rozen, O. Lerman, Journal of Organic Chemistry 45 (1980) 672–678.
[31] D. Hebel, O. Lerman, S. Rozen, Journal of Fluorine Chemistry 30 (1985) 141–146.
[32] G. Brauer, 2nd ed., Handbook of Preparative Inorganic Chemistry, vol 2, Academic
Press, 1965.