Addition reactions of organometallic reagents to nitrogen trifluoride and enhanced alkyl-alkyl coupling by NF3

Article abstract of DOI:10.1016/j.jfluchem.2015.03.013

A survey of the reaction of nitrogen trifluoride (NF₃) with various organometallic reagents finds that organomagnesium (Grignard) reagents are the most useful for producing N,N-difluoroaminoalkanes. Alkyl-alkyl coupling is a persistant side reaction. Organolithiums are marginally effective. Organocopper, organozinc reagents undergo primarily alkyl-alkyl coupling catalyzed by the presence of NF₃. Organocalcium and organoaluminum reagents are unreactive.

Full text of DOI:10.1016/j.jfluchem.2015.03.013

Accepted Manuscript
Title: Addition reactions of organometallic reagents to
nitrogen trifluoride and enhanced alkyl-alkyl coupling by NF3
Author: Randolph K. Belter
PII:
S0022-1139(15)00096-2
DOI:
Reference:
http://dx.doi.org/doi:10.1016/j.jfluchem.2015.03.013
FLUOR 8538
To appear in:
FLUOR
Received date:
Revised date:
Accepted date:
2-2-2015
27-2-2015
31-3-2015
Please cite this article as: R.K. Belter, Addition reactions of organometallic reagents
to nitrogen trifluoride and enhanced alkyl-alkyl coupling by NF3., Journal of Fluorine
Chemistry (2015), http://dx.doi.org/10.1016/j.jfluchem.2015.03.013
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

*Graphical Abstract - Pictogram
Graphical Abstract
Page 1 of 14

Addition reactions of organometallic reagents to nitrogen trifluoride and enhanced alkyl-
alkyl coupling by NF3.
Randolph K. Belter*randolphbelter@gmail.com
Y-Not Chemical Consulting, 14400 Williams Road, Zachary, LA 70791, United States
Tel.: (225) 658-8792
Graphical Abstract
Highlights
R-NF2 were prepared in moderate yields by treating solutions of NF3 with R-Mg and R-
Li reagents.
R2CuX and R2Zn reagents undergo R-R coupling when reacted with solutions NF3.
R-Ca and R3Al reagents are unreactive to soltions of NF3.
Abstract
A survey of the reaction of nitrogen trifluoride (NF3) with various organometallic
reagents finds that organomagnesium (Grignard) reagents are the most useful for
producing N,N-difluoroaminoalkanes. Alkyl-alkyl coupling is a persistant side reaction.
Organolithiums are marginally effective. Organocopper, organozinc reagents undergo
primarily alkyl-alkyl coupling catalyzed by the presence of NF3. Organocalcium and
organoaluminum reagents are unreactive.
Keywords: nitrogen trifluoride; organometallic; alkyl substitution; aromatic substitution;
alkyl-alkyl coupling; difluoroalkylamines
Page 2 of 14

*Graphical Abstract - Synopsis
Abstract
A survey of the reaction of nitrogen trifluoride (NF3) with various organometallic reagents finds
that organomagnesium (Grignard) reagents are the most useful for producing N,N-
difluoroaminoalkanes. Alkyl-alkyl coupling is a persistant side reaction. Organolithiums are
marginally effective. Organocopper, organozinc reagents undergo primarily alkyl-alkyl coupling
catalyzed by the presence of NF3. Organocalcium and organoaluminum reagents are unreactive.
Page 3 of 14

*Highlights (for review)
Highlights
R-NF2 were prepared in moderate yields by treating solutions of NF3 with R-Mg and R-Li
reagents.
R2CuX and R2Zn reagents undergo R-R coupling when reacted with solutions NF3.
R-Ca and R3Al reagents are unreactive to soltions of NF3.
Page 4 of 14

*Manuscript
Corresponding author: Randolph K. Belter
(225) 658-8792 no fax
randolphbelter@gmail.com
1. Introduction
1.1 Known reactions of organometallics with nitrogen trifluoride
In an NMR study of N,N-difluoroaminoalkanes, we reported that select alkyl isomers had been
prepared by the reaction of organomagnesium (Grignard) reagents with NF3 [1]. At the time, the
difluoroaminoalkane products were isolated without concern for yield. It was the intent of this current
study to ascertain the efficiency and practicality of preparing N,N-difluoroaminoalkanes from Grignard
reagents and subsequently from organometallics in general. We were to find that organomagnesium
reagents are the only organometallic capable of producing useable quantities of difluoroaminoalkanes
and that alkyl-alkyl coupling is enhanced by the presence of NF3 and is a persistent problem with the
methodology.
The reaction of NF3 with organometallic reagents has been reported in the literature twice before.
Methylmagnesium and ethylmagnesium bromides have been reacted with NF3 to generate trimethyl- and
triethylamine [2]. Japanese Patent 11343270 and its US equivalent (referenced) disclose the preparation
of triphenylamines from reaction of complex aryllithium reagents with nitrogen trihalides. One example
with NF3 is given with a 23% yield [3]. In both cases, the NF3 is fully substituted and no
difluoroaminoalkanes are produced.
2. Results and Discussion
2.1 Reaction of NF3 with organomagnesium (Grignard) reagents
In reference 1 we had successfully prepared analytical quantities of 1-, 2-, and 3-
difluoroaminopentane and 1-, 2-, and 3-difluoroaminohexane by the reaction of the appropriate 1-, 2-, or
3-bromopentane or 1-, 2-, or 3-chlorohexane with magnesium metal followed by reaction with NF3. In
light of refs. 2 and 3, it was assumed that multisubstitution reactions would be a likely problem, so
reactions were performed at concentrations below 0.1M. Further, prepared solutions of the Grignard
reagent were slowly added to saturated solutions of NF3 at low temperature so as to maintain an excess
of NF3. This methodology suffered principally from the loss of product during the distillative removal of the
1
Page 5 of 14

gross excess of solvent. To make the process preparatively useful, we now would increase the
concentration of organomagnesium reagent to ~1.0 M and attempt to use less of the saturated NF3
solution, hopefully maintaining NF3 concentration throughout the reaction by replacing reacted NF3 via a
pressurized NF3 atmosphere. Such a procedure would be performed in a stainless steel pressure vessel
(Parr Mini-Reactor).
In our initial experiment, an ice chilled 1.3M solution of n-pentylmagnesium bromide in ether was
slowly transferred via nitrogen pressure into an equal volume of ice-chilled 150 psi NF3 saturated ether in
a SS steel pressure reactor. Product analysis by G.C. showed 22% difluoroaminopentane, 32%
pentanenitrile, and 45% n-decane. The pentanenitrile is the HF elimination product from
difluoroaminopentane, indicating an overall substitution level of 54%, very encouraging. However, the
formation of n-decane, the dimer of n-pentane, is the result of alkyl-alkyl coupling. A direct acid/water
quench of the initial Grignard reagent produced no n-decane. Clearly, the coupling was occurring post-
Grignard formation. A similar experiment with 1.0 M benzylmagnesium chloride produced copious
quantities of dibenzyl.
Our first suspicion was that iron or other metals from the stainless steel system were catalyzing
the coupling, a well known catalytic reaction [4]. We immediately converted to an all glass and
fluoropolymer system consisting of Fisher-Porter style glass pressure flasks with PTFE clad stir bars, FEP
transfer tubes and PFA valves [5]. Additionally, we began using a higher purity 99.98% magnesium
metal. The reaction was repeated, at lower concentration (0.1M) and lower temperature (-20oC), in the
glass/fluoropolymer system. Substitution products, combined, amounted to 50%. Coupling still occurred
at 28%, an improvement, but not any indication that changing from a metal reactor to a glass reactor was
stopping the coupling. We learned also, that coupling was occurring to some extent during the initial
formation of some Grignard reagents, particularly as we attempted higher concentrations. We found
suitable results generating the Grignard at lower temperatures than normal and concentrations around
0.5M.
Table 1 shows the results of reacting various alkyl, aryl and benzyl Grignard reagents with NF3 in
the glass/fluoropolymer system. All yields in the following tables are G.C. percentage unless otherwise
stated.
2
Page 6 of 14

Table 1. Reaction of organomagnesium reagents with NF3.
Reagent
1-pentyl-MgBr
Solvent
M
Temp (oC)
R-NF2
Dimer
ether
0.15
-20
50%
28%
0.15
0.5
-30
38%
29%
36%
49%
2-pentyl-MgBr
ether
-78 to 0
3-pentyl-MgBr
2-hexyl-MgCl
3-hexyl-MgCl
ether
ether
ether
0.15
0.15
0.15
-30
-22
-30
42%
24%
48%
40%
50%
50%
27%
(+ 32% c-
19%
cyclohexyl-MgCl
ether
ether
0.5
1.0
-30
-20
hexane and
22% c-hexene)
benzylmagnesium chloride
0%
100%
-ethylbenzylmagnesium
ether
ether
0.5
-78 to 0
-40
0%
0%
76%
19%
chloride
phenylmagnesium bromide
0.15
We performed several one-flask experiments where the pre-formed Grignard reagent was
subjected to a pressurized atmosphere of NF3 to no benefit.
Attempts were made to form the Grignard reagents under an atmosphere of NF3, so as to
intercept the organomagnesium reagent or the organic radical before coupling occurred. In each case,
3
Page 7 of 14

formation of the Grignard was completely inhibited. Logically, one might suppose that the NF3 is reacting
with the Mg metal to form an unreactive MgF2 coating on the Mg granules. However, at least for the case
of phenyliodide, removing the NF3 from such a system by repeated aspiration at 0oC resulted in the
formation of the Grignard reagent, Ph-MgI. Hydrolysis yielded benzene and biphenyl.
2.2 Reaction of NF3 with organolithium reagents
Assuming that our observed coupling is the combination of two alkyl radicals, we surmised that
an organometallic reagent of more ionic character than a Grignard reagent might perform the substitution
reaction in preference to the coupling. Organolithium became our next subject of study.
As organolithium reagents are apt to solvolyse ethers at temperatures above dry ice/acetone
conditions, most reactions were performed in n-pentane. Due to their limited solubility in that solvent,
they were formed with the assistance of TMEDA (and later without). n-Butyllithium and n-hexyllithium
were purchased hexane solutions. Other alkyllithiums were prepared from the organohalide and lithium
metal [6],[7] while the benzyllithium reagents were prepared from the benzylic alkane and n-BuLi [8]. The
benzyllithium solution is essentially a slurry and really not suitable for this type of reaction. 2-
lithioethylbenzene and diphenymethyllithium did not form under these conditions.
Table 2. Reaction of organolithium reagents with NF3.
Reagent
Solvent
Temp (oC)
R-NF2
Dimer
a
n-BuLi
pentane/TMEDA
0
_
Only octane obs’d
45% octane
a
_
n-BuLi
pentane
pentane
0
55% 4-octanone
n-hexyl-Li
c-hexyl-Li
-78 to 0
29%
71%
24%
7%
pentane/TMEDA -30 to RT
(+ 69% c-hexane and c-hexene)
4
Page 8 of 14

ether
0%
phenyl-Li
benzyl-Li
-78 to 0
0 to 70
31%
ether/THF
(61% benzene)
toluene
0%
Only dibenzyl obs’d
a obscured by solvent
The reactions of n-Bu-Li showed that dimer formation was still occurring for this R-Li. While both
n-Bu-NF2 and n-Bu-H were unobservable due to solvent interference in G.C., in the TMEDA free case,
substitution was evidenced by the observation of 4-octanone. Its formation via n-Bu-NF2 is described by
Scheme 1. In the case of n-BuLi with TMEDA, it is likely that proton abstraction of the TMEDA to
generate butane occurs faster than reaction with NF3. For the reaction of n-hexyl-Li, substitution is
directly observed as n-hexyl-NF2. Phenyllithium reagent generated and added to ether/NF3 or added to
THF/NF3 generated only benzene and biphenyl, hydrolysis predominating 2:1. In addition, these
reactions produced much insoluble sludge, unseen in any other reactions in this study. In attempts to
imitate Reference 3, that is the formation of trialkylamine by the slow addition of NF3 to an ether solution
of Ph-Li generated no detectable Ph3N (GC/MS). Biphenyl was formed in 13% while hydrolysis still
predominated at 58%. Again the reaction produced much insoluble sludge.
The case of benzyllithium is of particular importance because it was prepared by the
deprotonation of toluene. As such, at no time was a benzyl halide present in the reaction and, as such,
SN2 coupling of R-M + R-X is not possible (formal Wurtz coupling). Coupling of R-Li + R-Li is being
initiated by the presence of NF3 as it apparently is during the coupling of R-MgX with R-MgX. This
despite the more ionic nature of R-Li.
Scheme 1. Formation of 4-octanone
5
Page 9 of 14

2.3 Reaction of NF3 with organocopper and organocuprate reagents
Preformed copperphenylacetylide [9] was reacted as a slurry with NF3 in THF overnight at 100oC.
The only observed product is bis-diphenylacetylene. In addition, a small amount of copper plating was
observed on the wall of the flask.
(n-Hexyl)2Cu(CN)Li reacted instantaneously with NF3 in ether at -50oC (as observed by formation
of white CuFX salts) to generate dodecane in 83% isolated yield. (Benzyl)2Cu(CN)Li likewise generated
dibenzyl instantly and exclusively. Dialkylcuprates tend to be radical addition reagents (think 1,4-carbonyl
addition) and paired with NF3 makes an excellent coupling system.
Scheme 2. Reactions of copper reagents with NF3.
2.4 Reaction of NF3 with organocalcium reagents
Organocalcium reagents are very reactive but more difficult to prepare than other organometallics
[10]. Initially, we prepared organocalcium reagents from activated calcium powder by the method of
Westerhauser [11]. For the reaction of Ph-CaI with NF3, there was no reaction (only benzene observed
after work-up). For the reaction of c-hexyl-CaBr, a mixture of 39% cyclohexane, 18% cyclohexene, and
23% bicyclohexyl was isolated, indicating poor reactivity for the alkylcalcium. Both reactions were
repeated with active calcium generated from lithium biphenylide with CaI2 [12]. For the cases of PhCaI
and c-hexyl-CaI, no addition to NF3 was observed. We have no observations of coupling for the Ph-CaI
reaction as the biphenyl co-reagent masked the observation of the biphenyl coupling product. Catalytic
biphenyl (10%) was ineffective. For the reaction of c-hexyl-CaI, the products of cyclohexane,
cyclohexene, and bicyclohexyl were all again observed with no discernable c-hexyl-NF2.
6
Page 10 of 14

Ca/Hg amalgam was only partially reactive with Ph-Br to produce ~1/4 strength Ph-CaBr. The
grey suspension turned to chocolate brown upon exposure to NF3, however no reaction products (neither
substitution nor dimerization) were observed.
2.5 Reaction of NF3 with organozinc reagents
(n-hexyl)2-Zn in NF3/ether produced a white precipitate as it warmed from 0 to 20oC. Small
amounts of 1- and 2-NF2-hexane were observed, but the principle product was 46% dodecane (coupling
product). Repeating the experiment in THF/toluene [13] increased the production of difluoaminohexanes
to 22% and reduced the dodecane to 16%. However, the apparent gain was offset by the generation of
24% heptylbenzene and 25% dibenzyl by cross-coupling and coupling of the toluene respectively.
We attempted to formalize the above observed coupling of Bn2Zn by preparing Bn2Zn from BnLi
slurry and injected it into a saturated NF3/ether solution at -78oC and allowing to warm to room
temperature. The reaction generated a white precipitate but upon quenching produced only toluene.
Solvolysis was the only reaction. As with BnLi, solubility was an issue with Bn2Zn.
(c-Hexyl)2Zn in THF/toluene is mostly unreactive up to room temperature.
2.6 Reaction of NF3 with organoaluminum reagents
Organoaluminum reagents are generally found in two reagent forms, the trialkylaluminum and the
sesquihalide. The sesquihalide is formally R3/2AlX3/2 and is a mixture of RAlX2 and R2AlX. We tested both
forms for reactivity with NF3. For our tests, the sesquihalides and the trialkylaluminums were prepared
from the addition of R-Li to AlCl3 in ether [14].
Neither Ph3/2AlCl3/2 nor Ph3Al showed any reactivity in -78oC reactions that were allowed to stir to
room temperature overnight. Similarly, n-hexyl3/2AlCl3/2 and n-hexyl3Al were also unreactive under the
same conditions. Non-reactivity included no discernable formation of dimers.
7
Page 11 of 14

2.7 Mechanistic considerations
For the literature reaction of n-heptyl-MgBr + benzyl chloride (at reflux temperatures), the
principle products are heptane, heptene and bibenzyl [15]. The critical mechanistic steps are the
separation of heptyl radical from MgBr radical and the subsequent extraction of Cl radical by MgBr radical
to form MgBrCl and benzyl radical (which dimerizes). We envision a similar mechanism by which NF3
forces the separation of alkyl radical from MgX even at low temperature as shown in Scheme 3, Reaction
1. Dimerization of R. competes with R-NF2 formation, Reaction 2 vs. Reaction 3. However, Reaction 4,
the dimerization of .NF2, also competes with Reaction 3, allowing alkyl dimerization to outpace R-NF2
formation. We attempted but were unable to verify the existence of F2N-NF2 in our reaction gas space by
GC or GC/MS.
Scheme 3. Mechanistic steps.
3. Conclusion
This survey of reacting organometallic reagents with nitrogen trifluoride shows that
organomagnesium (Grignard) reagents are the reagent of choice for the generation of R-NF2. All other
organometallics were either unreactive or caused dimerization of the alkyl substituent. Even the
Grignards suffered large loss of yield due to dimer formation. In many cases, dimerization occurs to
varying degrees during generation of the organometallic and we have not differentiated that quantity from
the final observed dimer quantity. Quantities of R-NF2 and dimer listed in Table 1 and Table 2 are then
those observed for the formal conversion of R-X to R-NF2.
8
Page 12 of 14

4. Experimental
4.1 General
Alkyl halides, Li foil, CaI2, CuCN, and ZnBr2 were from Aldrich. Magnesium was Aldrich 99.98
granular. n-Butyllithium and n-hexyllithium were hexane solutions purchased from Fisher/Acros. Nitrogen
Trifluoride was from Fluoromar. THF and ether were distilled from sodium/benzophenone.
250ml and 500ml Fisher-Porter reaction flasks with a working pressure limit of 150 psi were
manufactured by Quark Glass Inc. PFA needle valves were manufactured by Whitey Inc.
Products were identified by GC analysis vs. authentic samples and GC/MS. Both instruments
used 60 M DB-1701 capillary columns.
4.2 Cautionary Notes
Nitrogen trifluoride is toxic by inhalation and should only be used in a well ventilated environment.
It is apparent to us that the N,N-difluoroamino products are toxic as well. In addition, trace by-products
may be toxic and concentrated during distillation. Compounds containing multiple NF2 groups have been
used as high energy propellant oxidizers [16] and their inadvertent concentration may pose additional
risk.
Reactions performed in pressurized glass apparatus should be performed behind an adequate
blast shield and hood sash. In addition, personal protective equipment included a heavy leather jacket,
ballistic vest, additional neck protection, leather gloves and a safety helmet with double thickness full face
shield. Hearing protection is worn at all times as the bursting of even small diameter plastic tubing can
cause serious hearing damage.
9
Page 13 of 14

5. References
[1] R.K. Belter, J. Fluorine Chem. 137 (2012) 73.
[2] O. Glemser, R. Mews, Naturwissenshaften 55 ( 1968) 229.
[3] E. Hiroyuki, U. Tsutomu, US Patent 6022997, Feb. 8, 2000.
[4] W.M. Czaplik, M. Mayer, A.J. von Wangelin, Angw. Chem. Int. Ed., 48 (2009) 607.
[5] PTFE = polytetrafluoroethylene, FEP = fluorinated ethylene propylene, PFA = perfluoroalkoxy
[6] R. Zerger, W. Rhine, G. Stucky, J. Am. Chem. Soc. 96 (1974) 6048.
[7] L.A. Walter, Org. Syn. 23 (1943) 83, Org. Syn. Coll. 3 (1955) 757.
[8] G.G Eberhardt, W.A Butte, J. Org. Chem. 29 (1964) 2928.
[9] R.D. Stephens, C.E. Castro, J.Org.Chem. 28 (1963) 3313.
[10] S.K. Ritter, Chemical & Engineering News, March 7, 2011, pg 49.
[11] R. Fischer, M. Gartner, H. Gorls, M. Westerhausen, Organometallics 25 (2006) 3496.
[12] T-C. Wu, H. Xiong, R.D. Rieke, J. Org. Chem. 55 (1990) 5045.
[13] C. Petrier, J.C. de Souza Barbosa, C. Dupuy, J-L. Luche, J. Org. Chem. 50 (1985) 5761.
[14] G. Wittig, D. Wittenberg, Ann. (1957) 606, 1.
[15] R.W. Schiessler, R.M. Speck, R.J. Dixon, J. Am. Chem. Soc. 73 (1951) 3524.
[16] examples found in:
A.J. Passante, J.R. Lovett, P.A. Albrecht, US Patent 3347872, Oct 17, 1967,
R.J. Leary, US Patent 3347873, Oct. 17, 1967,
P.A. Arbright, N.J.Cranford, US Patent 3350414, Oct. 31, 1967.
10
Page 14 of 14