Nitrogen trifluoride as an oxidative co-reagent in high temperature vapor phase hydrofluorinations

Article abstract of DOI:10.1016/j.jfluchem.2006.02.006

Nitrogen trifluoride (NF₃) has proven to be a useful additive in high temperature vapor phase hydrofluorination reactions of chlorocarbons. The activity of chromium-based catalysts is maintained by introducing a co-stream of NF₃ into the reagent chlorocarbon and HF stream. NF₃ is a desirable additive instead of O₂ as there is no water generation due to its use.

Full text of DOI:10.1016/j.jfluchem.2006.02.006

Journal of Fluorine Chemistry 127 (2006) 816–820
www.elsevier.com/locate/fluor
Short communication
Nitrogen trifluoride as an oxidative co-reagent in high
temperature vapor phase hydrofluorinations
a,
Randolph K. Belter , Mark S. Sweval , Yuichi Iikubo
b
b
*
a
Department of Chemistry and Physics, Southeastern Louisiana University, Hammond, LA 70402, United States
b
Futago LLC., W. Lafayette, IN 47906, United States
Received 12 January 2006; received in revised form 7 February 2006; accepted 14 February 2006
Available online 6 March 2006
Abstract
Nitrogen trifluoride (NF
3
) has proven to be a useful additive in high temperature vapor phase hydrofluorination reactions of chlorocarbons. The
activity of chromium-based catalysts is maintained by introducing a co-stream of NF into the reagent chlorocarbon and HF stream. NF is a
3
3
desirable additive instead of O
#
2
as there is no water generation due to its use.
2006 Elsevier B.V. All rights reserved.
Keywords: Nitrogen trifluoride; Vapor phase hydrofluorination
1
. Introduction
In an attempt to extend the life of the catalyst, Japanese
Patent 51-82206 describes the injection of about 1% of oxygen
or air continuously during the fluorination reaction of
perhalogenated compounds [1]. The catalyst used is chromium
oxide with or without other metal oxides. United States Patent
5,900,514 [2] discloses the fluorination of dichloromethane
with HF in the presence of relatively high concentrations of
oxygen (3 mol%, based on the moles of dichloromethane) at
250–450 8C, using a catalyst based on pure chromium oxide,
without the addition of another metal oxide. Excellent
selectivity to difluoromethane (R-32) is claimed, with only
minor quantities of by-products. However, Ref. [6] refutes this,
stating that by-products of the R-120 series are formed. The use
of an oxygen co-feed in the production of R-134a is claimed in
a 1980 Great Britain patent [3].
For decades chlorofluorocarbons have been useful chemicals
for refrigeration, solvent, foam manufacture and firefighting
applications. The discovery of the harmful nature of these
chemicals towards the earth’s protective ozone layer has led to
the outlawing of the manufacture and use of most of these
chemicals. The most popular non-ozone depleting replace-
ments have been the small C-1 and C-2 hydrofluorocarbons
such as R-32 (CH F ), R-125 (CF -CH F), and R-134a (CF -
2
2
3
2
3
CH F). Each is produced from the corresponding hydrochlor-
2
ocarbon or related olefin. Complete fluorine-for-chlorine
exchange, however, tends to be difficult and is accomplished
in high temperature vapor phase processes.
Numerous patents describe methods of fluorinating chlor-
ocarbons with anhydrous HF using catalysts such as chromium
oxide (Cr O ), chromium fluoride (CrF ), chromium supported
In cases where the injection of oxygen acts by oxidizing and
removing deposits of carbonaceous materials from the surface
of the catalyst, carbon dioxide and water are generated. Water
has the deleterious effect of damaging and deactivating the
catalyst through phase changes initiated by repeated addition
and loss of water from the metal halide [3–6]. Water that is
generated by the injection of oxygen is retained in the product
and recovered HF stream. It must be removed from the product
prior to any further commercial use, adding further expense to
the process. When water is in the HF stream, upon recycle
further damage to the catalyst and reactor system occurs. Each
recycle results in additional damage. Plugging of the catalyst
2
3
3
on carbon (Cr/carbon), nickel (III)/aluminum (III) fluoride
NiF /AlF , etc. It was found, however, that in order to obtain
(
3
3)
reasonably good yields and selectivity to produce, for example,
difluoromethane, the catalyst had to be replaced frequently due
to deactivation by rapid contamination with coke or due to
catalyst crystallization [1–6].
*
Corresponding author. Tel.: +1 985 549 3529; fax: +1 985 549 5126.
E-mail address: rbelter@selu.edu (R.K. Belter).
0022-1139/$ – see front matter # 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.jfluchem.2006.02.006

R.K. Belter et al. / Journal of Fluorine Chemistry 127 (2006) 816–820
817
bed occurs readily under such conditions. Water generation also
has the disadvantage of removing the protective metal fluoride
layer from the inside of the reactor and accompanying fittings.
Corrosion and erosion result and damage to the equipment can
be extensive. As such, equipment must be replaced more
frequently, adversely affecting the economics of the process.
Selectivity and reactivity are reduced and eventually the
operation must be shut down and refurbished.
exchange occurs. Samples were drawn at 15 min intervals
and analyzed by GC.
2.1. Fluorination reactions with oxygen (O2)
For comparison purposes, reactions were run on R-30, R-
23 and R-133a over chromium-based catalyst with 1–2 mol%
O as a co-feed. Catalyst beds of chromium oxide-on-carbon
1
2
Alternative oxidants have been used to replace O and its
2
were preconditioned under a stream of HF vapor diluted in
nitrogen. A period of oxygen co-feed at or above the anticipated
reaction temperature preceded the initiation of reagent feeds.
Conditions were adjusted so as to be able to maintain a constant
and acceptable level of conversion, that is to say, catalyst
activity. Conversions averaged 70%, 55% and 40% respec-
tively.
accompanying water generation problem. Chlorine has been
used as a co-feed in the manufacture of R-32 [5,6] and R-134a
[
chlorinated by-products of the R-20 series (CHCl , CHFCl ,
3]. The main impurities reported for the R-32 process are
3
2
CHF Cl) and R-110 series (CFCl -CFCl , CF Cl–CFCl ). For
2
2
2
2
2
the R-134a process, the nature and level of by-products are not
disclosed.
Fluorine (F ) has wisely not been pursued as an alternative
2
oxidant due to its high reactivity and dangerous nature.
Chlorine fluoride (ClF) or chlorine trifluoride (ClF ) likely
2.2. Fluorination reactions with nitrogen trifluoride (NF3)
3
would also be problematic, if not dangerous, co-feeds [7]. As
such, we struck upon the idea of applying the high temperature
reactivity of nitrogen trifluoride (NF ) to our advantage and
To test the efficacy of NF as a catalyst activity stabilizer,
3
reactions were run on R-30, R-123 and R-133a over chromium
oxide-on-carbon catalyst beds prepared as above. HF and the
appropriate chlorinated compound were passed over the
3
sought to use NF as an oxidative co-feed in fluorine-for-
3
chlorine exchange reactions. While NF3 may seem an
expensive choice as an O substitute, there are fluorochemical
catalyst bed and a co-feed of NF was maintained during this
3
period. As anticipated, the activity of the catalyst was
maintained at a constant level. Conversions were closely
comparable to those observed during O co-feed. Conversion to
2
production facilities around the world where fluorination
processes and inexpensive NF streams are in close proximity.
3
2
R-32 averaged 69%, conversion to R-125 averaged 58%, and
conversion to R-134a averaged 34%. Most importantly, catalyst
activity during NF co-feed was maintained at levels parallel to
2
. Results and discussion
3
that of O co-feed.
2
We have now discovered that NF can be an alternative co-
3
Under the oxidant rich conditions that we chose to maintain
for our experiments, some oxidation by-products were formed.
Interestingly, the level and nature of by-products generated by
O co-feed and NF co-feed were similar. That is, while NF is
reagent for maintaining the activity of fluorination catalysts.
Three fluorine-for-chlorine exchange reactions were used to
evaluate the efficacy of this procedure. Dichloromethane was
converted to R-32. R-123 was converted to R-125, and R-133a
was converted to R-134a (see Scheme 1). Besides conversion
rates, particular attention was given the condition of the catalyst
after recovery from the reaction.
2
3
3
acting as an oxidant, it does not act as a fluorinating agent, even
at temperatures as high as 500 8C. Rather, for both reagents,
chlorine addition products were observed at slowly increasing
levels. For the fluorination of CH Cl with oxygen co-feed, R-
2
2
Vapor phase reactions were performed in an electrically
heated tubular reactor packed with active carbon supported
chromium oxide pellets. Reagents were vaporized and passed
over the heated catalyst bed where fluorine-for-chlorine
22 (CHF Cl) and R-23 (CHF ), at the 1% level, were indicative
2 3
of oxidation (see Graph 1). For the fluorination of CH Cl with
2
2
NF , R-22 and R-23 were observed at similar levels along with
3
some R-12 (CF Cl ). For the fluorination of R-123 with oxygen
2
2
co-feed, R-113a (CF -CCl ) was observed on G.C. at 1–1.5%
3
3
(see Graph 2). For the fluorination of R-123 with NF co-feed,
3
the level of R-133a was also about 1.5%. A closer look at both
product streams with GC/MS showed lesser amounts of R-114
(CF –CFCl ), R-115 (CF –CF Cl), R-13 (CF Cl) and CF –
CHCl–CHCl–CF . For the fluorination of R-133a, oxidation is,
3 2 3 2 3 3
3
as mentioned in the introductory reference, more prevalent.
Both R-123 and R-124 were observed, with oxygen co-feed
generating about 9% of combined oxidation products and NF3
co-feed generating 10% (see Graph 3). Clearly, just as O reacts
2
with HCl to form Cl in a Deacon reaction [8,9], NF was also
2
3
oxidizing HCl to Cl . This was supported by the plainly
2
detectable odor of chlorine observed at the reactor outlet (after
scrubbing with water).
Scheme 1.

818
R.K. Belter et al. / Journal of Fluorine Chemistry 127 (2006) 816–820
Graph 1. Fluorination of R-30 to R-32.
Graph 2. Fluorination of R-123 to R-125.
Graph 3. Fluorination of R-133a to R-134a.
After the conclusion of each reaction, the reactor was purged
with nitrogen and the catalyst visually inspected. For both O2
water contact. For the corresponding NF co-feed reactions, the
3
catalyst bed remained unchanged.
and NF reactions making R-32, the catalyst beds were in good
3
condition. It is likely that with its C-1 feedstock, this reaction is
comparatively clean and longer reaction times or recycle of the
HF would be necessary to see water damage to the catalyst.
However, the reactions that use C-2 feedstocks can generate
olefinic intermediates and consequently are expected to
generate more coke and oligomers. For the O2 co-feed
reactions making both R-125 and R-134a, the outlet section
of catalyst bed contained green and white oxides, indicative of
3. Conclusion
Nitrogen trifluoride is an effective alternative to oxygen as a
co-feed for stabilizing the activity of chromium-based
fluorination catalysts. In high temperature vapor phase
reactions of C-1 and C-2 hydrochlorocarbons, catalyst activity
was maintained while reducing undesired water generation.

R.K. Belter et al. / Journal of Fluorine Chemistry 127 (2006) 816–820
819
4
. Experimental
4.5. Example 3. Preparation of pentafluoroethane (R-125)
with oxygen co-feed
4
.1. General
Anhydrous HF was flowed at 1 g/min through a HF/O2
activated catalyst bed at 400 8C. After 1/2 h, the HF flow was
increased to 1.5 g/min and a feed of CF –CHCl was started at
Anhydrous hydrogen fluoride and oxygen were Matheson-
Trigas UHP grade. Nitrogen trifluoride was Linde 99%.
Dichloromethane (R-30) was Fisher ACS/HPLC grade,
stabilized with cyclohexene. R-123 was commercial DuPont
SUVA-123. R-133a was 99% from Synquest Labs. Chromium
Oxide was Baker ACS grade. Catalysts were prepared by
treating commercial carbon support pellets with saturated
aqueous chromium oxide solution. GC/MS was performed on
an HP 5890/5971A spectrometer with a 60 M DB-1 capillary
column.
3
2
1.8 g/min. Oxygen gas was then introduced at 20 mL/min. The
outflow stream was passed through hot water and the product
composition was analyzed by gas chromatography. The
reaction conversion was 65% initially, and slowly decreased
over a 10-h reaction period to 45% (see Graph 2). Upon removal
from the reactor, the lower section of catalyst contained green
and white oxides, indicative of water contact.
Cautionary note: Anhydrous HF causes severe burns to
the skin and mucous membranes. HF should be handled with
full PPE protection. An ample supply of HF antidote gel
should be kept on hand before handling HF. A referee
recommended several references for burn treatment procedures
4.6. Example 4. Preparation of pentafluoroethane (R-125)
with nitrogen trifluoride co-feed
Anhydrous HF was flowed at 1 g/min through a HF/O2
activated catalyst bed at 400 8C. After 1/2 h, the HF flow was
increased to 1.5 g/min and a feed of CF –CHCl was started at
[
10–12].
3
2
1.8 g/min. NF gas was then introduced at 0.33 g/min. The
3
4
.2. Typical catalyst bed preparation
outflow stream was passed through hot water and the product
composition was analyzed by gas chromatography. The
reaction conversion was 65% initially, and slowly decreased
over a 12-h reaction period to 50% (see Graph 2). Upon removal
from the reactor, the catalyst bed remained unchanged.
The reactor constituted a 1 in. diameter nickel alloy tube
6 in. in length fitted with a thermocouple centered in each
half-length of the reactor. This tube was charged with catalyst
3
and purged with N at 200 8C until no water vapor was
2
detectable at the outlet with a cold mirror. Anhydrous HF was
then introduced at 1 mL/min for 1 h. The temperature was
raised to the anticipated reaction temperature for 1 h. The N2
4.7. Example 5. Preparation of tetrafluoroethane (R-134a)
with oxygen co-feed
stream was enriched to 2% O for 2 h. The oxidant flow was
2
Anhydrous HF was flowed at 1 g/min through a HF/O2
activated catalyst bed at 400 8C. After 1/2 h, the HF flow was
increased to 1.5 g/min and a feed of CF –CH Cl was started at
stopped and the temperature stabilized to the desired reaction
temperature.
3
2
1.4 g/min. Oxygen gas was then introduced at 18 mL/min. The
4.3. Example 1. Preparation of difluoromethane (R-32) on
oxygen activated catalyst with oxygen co-feed
outflow stream was passed through hot water and the product
composition was analyzed by gas chromatography. The
reaction conversion held constant at about 40% (see Graph
3). Upon removal from the reactor, the reactor outlet contained
green and white oxides, indicative of water contact.
Liquid anhydrous HF was pumped at 1 g/min through a HF/
O activated catalyst bed at 300 8C. After 1 h, the HF flow was
2
increased to 1.8 g/min and a feed of CH Cl was started at
2
2
1
.3 g/min with 20 ml/min oxygen co-feed. The outflow stream
4.8. Example 6. Preparation of tetrafluoroethane (R-134a)
with nitrogen trifluoride co-feed
was passed through hot water and the product composition was
analyzed by gas chromatography. The reaction conversion was
73% initially, and slowly decreased over an 8.5-h reaction
period to 70% (see Graph 1).
Anhydrous HF was flowed at 1 g/min through a HF/O2
activated catalyst bed at 400 8C. After 1/2 h, the HF flow was
increased to 1.5 g/min and a feed of CF –CHCl was started at
3
2
4.4. Example 2. Preparation of difluoromethane (R-32) on
oxygen activated catalyst with nitrogen trifluoride co-feed
1.4 g/min. NF gas was then introduced at 0.15 g/min. The
3
outflow stream was passed through hot water and the product
composition was analyzed by gas chromatography. The
reaction conversion was 37% initially, and slowly decreased
over a 7-h reaction period to 30% (see Graph 3). Upon removal
from the reactor, the reactor outlet remained clear.
Anhydrous HF was flowed at 1 g/min at 300 8C through a
HF/O activated catalyst bed. After 1 h, the HF flow was
increased to 1.2 g/min and a feed of CH Cl was started at
2
2
2
1
.3 g/min. NF gas was then introduced at 0.17 g/min. The
3
outflow stream was passed through hot water and the product
composition was analyzed by gas chromatography. The initial
conversion was approximately 72%, and decreased marginally
over an 8-h reaction period to 65% (see Graph 1).
Acknowledgements
This project was funded through the generous continuing
support of Futago, LLC, W. Lafayette, IN. NMR support was

820
R.K. Belter et al. / Journal of Fluorine Chemistry 127 (2006) 816–820
[
[
6] B. Requieme, S. Perdrieux, B. Cheminal, E. Lacroix, A. Lanz, US Patent
,639,115 (2003).
7] Author’s unpublished research with ClF
made possible by an instrumentation grant from The Louisiana
Board of Regents LEQSF 045UG-02.
6
3
verified propensity to detona-
tions in the vapor and liquid phase. In one small-scale reaction a 300 mL
Parr reactor was explosively breached. LaRoche Inds. (1997).
References
[
8] H. Itoh, Y. Kono, M. Ajioka, S. Takenaka, M. Kataita, US Patent 4,803,065
(
1989).
[
[
[
[
[
1] Y. Oda, H. Otouma, S. Morikawa, Japanese Patent 51-82206 (1976).
2] B. Requieme, E. Lacroix, A. Latz, US Patent 5,900,514 (1999).
3] O. Yohnosuke, S. Heikitsu, British Patent 2,030,981 (1980).
4] W. Gumprecht, US Statutory Invention Registration, H1129 (1993).
5] D. Garrait, E. Guiraud, US Patent 6,635,790 (2003).
[9] T. Kiyoura, Y. Kogure, T. nagayama, K. Kanaya, US Patent 4,822,589
(1989).
[10] J. Bertolini, J. Emergency Med. 10 (1992) 163.
[11] D. Peters, R. Miethchen, J. Fluorine Chem. 79 (1996) 161.
[12] E. Segal, Chem. Health Safety (2000) 19.