High temperature vapor phase reactions of nitrogen trifluoride with benzylic substrates

Article abstract of DOI:10.1016/j.jfluchem.2011.03.006

At temperatures around 400°C, nitrogen trifluoride (NF₃) readily reacts with benzylic substrates. Products vary with the substrate, but are all the result of difluoroamination at the benzylic position. Toluene and ethylbenzene produce benzonitrile. Cumene produces α-methylstyrene. Diphenylmethane produces benzanilide. Little or no direct fluorination or radical dimerization is observed.

Full text of DOI:10.1016/j.jfluchem.2011.03.006

Journal of Fluorine Chemistry 132 (2011) 318–322
Contents lists available at ScienceDirect
Journal of Fluorine Chemistry
journal homepage: www.elsevier.com/locate/fluor
High temperature vapor phase reactions of nitrogen trifluoride with benzylic
substrates
*
Randolph K. Belter
Y-Not Chemical Consulting, 14400 Williams Road, Zachary, LA 70791, United States
A R T I C L E I N F O
A B S T R A C T
Article history:
At temperatures around 400 8C, nitrogen trifluoride (NF3) readily reacts with benzylic substrates.
Received 11 February 2011
Received in revised form 2 March 2011
Accepted 3 March 2011
Products vary with the substrate, but are all the result of difluoroamination at the benzylic position.
Toluene and ethylbenzene produce benzonitrile. Cumene produces
produces benzanilide. Little or no direct fluorination or radical dimerization is observed.
ß 2011 Elsevier B.V. All rights reserved.
a-methylstyrene. Diphenylmethane
Available online 15 March 2011
Keywords:
Nitrogen trifluoride
Benzylic substitution
Radical abstraction
Difluoroalkylamines
b
-Elimination
Beckman rearrangement
1. Introduction
that is 99.99% or better. Any upsets in manufacturing conditions
that produce ‘‘off-spec’’ NF3 produce material that cannot be
reworked or purified. With a critical temperature of ꢀ39.2 8C and a
critical pressure of 45.28 bar, manufacturing sites cannot store any
appreciable quantity of ‘‘off-spec’’ NF3 and so are forced to vent it.
Scrubbing technologies are not up to the task and 25–100% of the
NF3 escapes trapping [3]. It is unlikely that manufacturers will
adopt more expensive technologies that to destroy the offending
NF3 waste stream. Unless an economical alternative is offered,
venting will continue as required and the atmospheric concentra-
tion of NF3 will continue to rise. Even governmental regulation is
unlikely to stop the introduction of NF3 into the atmosphere (refer
to the Kyoto HFC-23 Offset Program where manufacturers
intentionally produced waste to receive credit for its disposal) [4].
It is the overall goal of this project to develop chemistries that
will allow manufacturers to convert off-specification NF3 to
commercially valuable compounds, thus giving them economic
incentive to do so. The availability of such a process could reduce
intentional emissions and mitigate the Global Warming impact of
NF3 production.
1.1. Environmental issues of nitrogen trifluoride
Nitrogen trifluoride (NF3) is the premier etching gas for the
manufacture of silicon wafer computer chips as well as for plasma
or thermal cleaning of Chemical Vapor Deposition chambers.
World-wide production for 2008 was estimated at 4000 metric -
tons. While supposedly 96–98% of NF3 is destroyed in the process
of being used, atmospheric measurements indicate that more than
the remaining 2–4% is getting into the atmosphere. According to
research from UC, Irvine [1] and the Scripps Institute of
Oceanography [2], nitrogen trifluoride is much more prevalent
in the atmosphere than was estimated by industry. This is of great
concern as NF3 has a GWP of 17,000 and amounts to 10.5 million -
metric tons of CO2 equivalent per year. For this reason, there have
been calls to add NF3 to the list of Kyoto Protocol reportable
compounds. The studies indicated that the overall emissions are
apparently being supplemented by leakages during production.
However, such ‘‘leakages’’ would have to account for up to 16% of
production, an unlikely scenario.
The venting of ‘‘off-specification’’ NF3 by overseas manufactur-
ing plants is a more plausible explanation for high atmospheric NF3
levels. The specification for electronics grade NF3 is ‘‘four nines’’,
1.2. Precedent for reactivity
The utility of nitrogen trifluoride is due to its known
decomposition at elevated temperatures. When heated to tem-
peratures above 250 8C, NF3 exudes fluoride radicals which are
reactive with metals and metal oxides and are thus useful in
etching the silicon surfaces of computer chips. The early literature
* Tel.: +1 225 658 8792.
E-mail address: randolphbelter@gmail.com.
0022-1139/$ – see front matter ß 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.jfluchem.2011.03.006

R.K. Belter / Journal of Fluorine Chemistry 132 (2011) 318–322
319
Scheme 1. Literature evidence of fluorination by NF3.
gives examples of NF3 reacting as a fluoride source with sulfur [5],
non-metal and metal oxides [6,7] non-metal oxyhalides [8],
cyanides [9], and phosphorus–sulfur compounds [10]. The nearest
example to reaction with an organic compound is the reaction of
NF3 with COCl2. NF3 has been shown to be able to add across select
olefins to produce perfluoroamines [11].
Scheme 2. Reaction of NF3 with toluene.
Our own research has used NF3 to prevent or remove the
carbonaceous deposits that foul high temperature vapor phase
catalysts [12]. In particular, chromium fluoride catalysts used for
fluorine-for-chlorine exchange reactions in the manufacture of
hydrofluorocarbons were kept active with a co-feed of NF3.
Example reactions were the preparation of difluoromethane (HFC-
32), pentafluoroethane (HFC-125) and tetrafluoroethane (HFC-
134a) from the appropriate hydrochlorocarbons. During the course
of these reactions, it was observed that some of the product was
halogenated to a greater extent than the original starting material.
For example, appreciable quantities (10%) of pentahaloethanes
were observed during the hydrofluorination of chlorotrifluor-
oethane (HFC-133a) to tetrafluoroethane (HFC-134a) (Scheme 1).
Similar by-product formation also results from the use of the
oxidants Cl2 or O2 as a co-reagent. NF3 appears to also be acting as
an oxidizing agent. Whether the NF3 is oxidizing the CrIII catalyst to
a reactive oxidation state, is oxidizing by-product HCl to Cl2, or is
performing fluorine-for-hydrogen exchange directly is yet to be
determined. It was none-the-less the intent of this project to
harness the net fluorinating capability of NF3 to perform
fluorination reactions and produce commercially saleable hydro-
fluorocarbons.
yield the isolated benzonitrile product (Scheme 2). This all occurs
within the reactor (and not during aqueous work-up) as
benzonitrile is also isolated when the product stream is
alternatively vacuum stripped of HF. No fluorotoluenes or other
light compounds were observed on GC. A complex mixture of
heavy components made up the remaining mass balance.
2.2. Reaction of NF3 with ethylbenzene
With NF3 clearly reacting via radical substitution at the benzylic
position of toluene, we undertook the simple extension of the
reaction to include 28 and 38 benzylic compounds. We were
immediately surprised to see the results of reacting NF3 with the 28
substrate, ethylbenzene, 3. One might expect ethylbenzene to
follow a similar mechanistic pathway as that of toluene. Indeed,
difluoroamination at the benzylic position is apparently occurring,
along with the elimination of one HF molecule. However at this
point, the unstable fluoroimine intermediate apparently under-
goes a Beckmann Rearrangement with subsequent fragmentation
as the sole product of reaction is again benzonitrile, 2 (Scheme 3).
2. Results and discussion
2.3. Reaction of NF3 with cumene
2.1. Reaction of NF3 with toluene
The simplest toluene analog with a 38 benzylic position is
cumene, 4. Reaction of NF3 with cumene yielded a single heavy
product which was difficult to characterize by NMR. This heavy
product could be induced to crystallize and was identified by
X-ray crystallography as 2,3-dihydro-1,1,3-trimethyl-3-phenyl-1H-
indene, 6 [13]. This complicated molecule is, in fact, one of the three
It was assumed that at high reactor temperatures NF3 would be
generating F radicals and N2, such as it does in etching processes.
The reaction we chose to utilize this chemistry was benzylic
fluorination. Benzylic fluorination of toluene would be expected to
generate benzyl fluoride, benzyl difluoride and benzotrifluoride in
various proportions. Benzotrifluoride, 1, is a valued compound that
is considered to be an environmentally more acceptable substitute
for dichloromethane (methylene chloride) in solvent applications.
In actuality, the results of the reaction of NF3 with toluene were
quite different. In a reaction performed at 400 8C, the principle
product was benzonitrile, 2. A short retroanalysis reveals that
rather than benzylic fluorination, benzylic difluoroamination is
occurring. An initial F radical abstracts a benzylic proton and the
resulting benzylic radical is quenched by difluoroamino radical.
The resulting difluoroamine is unstable and eliminates HF twice to
known dimers of a-methylstyrene, 5. So, it appears that cumene is
reacting with NF3 in the normal manner to form a benzylic
difluoroamine, followed by an elimination of HNF2 (Scheme 4).
One might expect cumene to follow the same course of reaction
as ethylbenzene, that is, 38 substitution followed by CH3F
elimination (via Beckmann rearrangement). However, we saw
no evidence of the products of CH3F elimination such as
acetophenone, 7, or acetophenone oxime, 8 (Scheme 5). One
might then suspect that selectivity had been lost and that
difluoroamination is occurring at the 18 position with elimination
Scheme 3. Reaction of NF3 with ethylbenzene.

320
R.K. Belter / Journal of Fluorine Chemistry 132 (2011) 318–322
Scheme 4. Reaction of NF3 with cumene.
Scheme 5. Unobserved reactions of NF3 and cumene.
of HNF2 occurring in the other direction. However, in the case 18
difluoroamination, HF elimination to the nitrile 9 would certainly
prevail and ‘‘plain-old’’ styrene, 10, would result. So, cumene
appears to react as in Scheme 4 and a 3rd mechanistic end-route
has been observed for our 3rd substrate class.
2.4. Reaction of NF3 with cymene: selectivity for 38 versus 18
p-Cymene, 11, is an aromatic molecule with a 18 substituent at
one end and a 38 substituent at the other end. The structure of p-
cymene offers the opportunity to perform a competitive selectivity
test of the NF3 reaction on a single molecule. Reacting p-cymene
Scheme 6. Selectivity of NF3 reaction.
under NF3-lean conditions afforded nitrile 12 and the
a-methyl
styrene 13 (as dimer) in a 1:7 mole ratio (Scheme 6). This is a high
selectivity akin to that observed for Cl radical at 100 8C [14].
2.5. Other bifunctional substrates
We investigated the reactions of ortho-xylene, 14, with NF3 to
determine whether the rate of HF elimination from an intermedi-
ate difluoroaminobenzylic would be slow enough to allow for a
second, intramolecular, fluoroamination to occur as in Scheme 7.
In practice, o-toluonitrile and phthalonitrile (3:1) were the only
products isolated, indicative of mono- and disubstitution, respec-
tively. None of the isoindole that would result from intramolecular
substitution was observed.
Scheme 7. Unobserved intramolecular fluoroamination.
As a cross-check, p-xylene was subjected to the same NF3
reaction conditions. In this case monosubstitution was less
prevalent to disubstitution (2:3). In addition, some alkyl-alkyl
radical coupling product was observed (2-xylyl-xylene).
Scheme 8. Reaction of NF3 with diphenylmethane.

R.K. Belter / Journal of Fluorine Chemistry 132 (2011) 318–322
321
Table 1
3. Conclusion
Products and yield of NF3 reactions.
At 400 8C, NF3 reacts with benzylic substrates to generate
intermediate N,N-difluorobenzylamines. These unstable amines
decompose by differing mechanisms based on their order of
substitution. Primary difluoroamines decompose by exuding HF to
generate nitriles (i.e. toluene, xylenes). Secondary difluoroamines
decompose by rearranging and eliminating R–F (i.e. ethylbenzene,
diphenylmethane), and tertiary difluoroamines decompose by the
apparent elimination of HNF2 (i.e. cumene, cymene). See Table 1.
It was the goal of this project to find a means by which waste
NF3 could be utilized in a manner in which valuable products might
be generated. The expectation was to make fluorobenzylic
compounds such as benzotrifluoride. In reality, we have discovered
Substrate
Product(s)
Yield (%)
46
56
59
that the product profile is composed of nitriles, amides, and
a-
methylstyrene dimers, depending on substrate. Nitriles and
amides have utility as synthetic building blocks and, surprisingly,
the
a-methylstyrene dimers are valued components of dielectric
fluids. The results of this study are encouraging.
4. Experimental
66
4.1. General
Cumene, cymene, diphenylmethane, ethylbenzene and xylenes
were purchased from Aldrich. Toluene was Sunnyside distilled
from CaH2. Nitrogen trifluoride was supplied by Fluoromar, LLC.
Reactions were performed in a single 100 (2.5 cm) diameter
tubular reactor, 2400 (61 cm) long, with two electrically heated
temperature zones of equal length. An internal thermocouple array
monitored the reactor temperature at five equally spaced points.
39
63
The reactor was fed via
a vaporizer/preheater of similar
00
configuration in (1.25 cm) diameter. The reactor was charged
with CrF3-on-carbon pellets that had been previously used to
exhaustion in chlorofluorination reactions. This ‘‘spent’’ CrF3-on-
carbon was chosen for the reactor bed as it was expected to be
resistant to degradation by NF3 and devoid of any oxidizing power
(a control experiment on toluene, absent NF3, yielded unchanged
toluene with no oxidation or disproportionation observed). The
reactor bed was pre-heated to 385 8C under a flow of 50 ml/min N2.
This N2 flow was maintained throughout the experiments as a
precaution against pooling of reagents within the preheating
system in the event of flow stoppages. A backpressure of 1 atm was
maintained. The vaporizer/preheater temperature was controlled
to maintain a vapor temperature 10 8C above the boiling point of
the organic feedstock. The feedstock (i.e. toluene), NF3 and N2 were
all fed together to the vaporizer/preheater and through the reactor.
An exotherm was always observed and the NF3 feed was controlled
so as to keep the temperature of all internal thermocouples under
425 8C. Typically NF3 feeds did not exceed 50 mole%. Generally
there would be one or two thermocouples indicating the exotherm
and hence the location of the reaction. Product was collected
through a cold water condenser, washed with water to remove HF,
neutralized with saturated NaHCO3 and dried over Na2SO4.
Unreacted starting material and product fractions were separated
by fractional distillation and sometimes silica gel column
chromatography.
53
2.6. Diphenylmethane
Diphenylmethane, 15, reacted with NF3 to produce two
products. The major product was benzanilide, 17, isolated in
45% yield, as the 4:3 co-precipitate with diphenylmethane. Again,
this product is the result of difluoroamine formation, HF
elimination and Beckmann Rearrangement, but without fragmen-
tation (Scheme 8). This is the first example in which a hydrolysis
product has been isolated in appreciable amounts. The second
product was N1,N2-diphenylbenzamidine, 18, isolated in 7.5% yield
and identified by X-ray crystallography [13]. Condensation of
benzanilide, 17, with iminium fluoride, 16, followed by loss of the
benzoyl group is the likely source of this material. We isolated
small amounts of benzoyl fluoride, anhydride and benzoic acid
which support this route.
Products were identified by 1H and 13C and 19F NMR performed
on a Bruker DPX-250. MS was performed on an Agilent 6210 Time-
of-Flight spectrometer.
4.2. Cautionary notes
Nitrogen trifluoride is toxic by inhalation and should only be
used in a well ventilated environment. In addition, trace by-
products may be toxic and concentrated during distillation.

322
R.K. Belter / Journal of Fluorine Chemistry 132 (2011) 318–322
References
NF3 is an oxidant and care must be taken to ensure
operations outside the explosivity limits of the NF3/hydrocarbon
mixture. It is advisable that all metal surfaces be fluoride
passivated. In instances where the NF3 ratio exceeded 50 mole%
and/or reactor inlet temperatures raised much beyond the
boiling point of the substrate detonations were observed in the
preheater system. In an instance where the NF3 ratio was
100 mole%, an apparent NF3 flame initiated at the NF3/toluene
mixing point (stainless steel tee) as evidenced by the tee
becoming literally red-hot (700 8C) [15].
Anhydrous HF is a by-product of these reactions. Anhydrous
HF causes instantaneous 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. See reference for burn treatment
procedures [16].
[1] M.J. Prather, J. Hsu, Geophys. Res. Lett. 35 (2008) L12810.
[2] R.F. Weiss, J. Muhle, P.K. Salameh, C.M. Harth, Geophys. Res. Lett. 35 (2008) L20821.
[3] F.J. Pisacane, E. Baroody, Naval Surface Weapons Center Publication MP76-24, 1976.
[4] C&E News March 23, 2009, p. 34.
[5] O. Glemser, U. Biermann, J. Knaak, A. Haas, Chem. Ber. 98 (2) (1965) 446–450.
[6] O. Glemser, U. Biermann, Chem. Ber. 100 (4) (1967) 1184–1192.
[7] O. Glemser, U. Biermann, Chem. Ber. 100 (8) (1967) 2484–2490.
[8] O. Glemser, J. Wegener, R. Mews, Chem. Ber. 100 (8) (1967) 2474–2483.
[9] U. Biermann, O. Glemser, J. Knaak, Chem. Ber. 100 (11) (1967) 3789–3794.
[10] A. Tasaka, O. Glemser, Z. Anorg. Allg. Chem. 409 (1974) 163–175.
[11] T. Takagi, M. Tamura, M. Shibakami, H-d. Quan, A. Sekiya, J. Fluorine Chem. 101
(2000) 15–17.
[12] R.K. Belter, M.S. Sweval, Y. Iikubo, J. Fluorine Chem. 127 (2006) 816–820.
[13] F. Franczak, Louisiana State University, Dept. of Chemistry, Baton Rouge, LA 70803.
[14] J. March, Advanced Organic Chemistry, 2nd ed., Wiley, 1977,, p. 626.
[15] For details of NCl3 flame, see: N. Rubtsov, V. Kotelkin, Mendeleev Commun. 12
(2002) 241.
[16] D. Peters, R. Methchen, J. Fluorine Chem. 79 (1996) 161.