Fluoroalkane aromatization over hot sodium oxalate

Article abstract of DOI:10.1016/S0022-1139(99)00113-X

Hot sodium oxalate (465°C) aromatizes perfluorodecalin to give perfluorotetralin and perfluoronaphthalene. Decomposition of solid sodium oxalate at temperatures above 450°C involves initial formation of elemental carbon that catalyzes the subsequent decomposition of oxalate, leading to the presence of an induction period in the overall decomposition reaction. Prior incipient decomposition of sodium oxalate to carbon and sodium carbonate is necessary for this reagent to effect aromatization.

Full text of DOI:10.1016/S0022-1139(99)00113-X

Journal of Fluorine Chemistry 99 (1999) 67±72
Fluoroalkane aromatization over hot sodium oxalate
Lenore Hoyt McAlexander, Christopher M. Beck, Juan J Burdeniuc,
Robert H. Crabtree*
Department of Yale Chemistry, 225 Prospect St., New Haven, CT 06520-8107, USA
Received 4 November 1998; accepted 26 May 1999
Abstract
Hot sodium oxalate (4658C) aromatizes per¯uorodecalin to give per¯uorotetralin and per¯uoronaphthalene. Decomposition of solid
sodium oxalate at temperatures above 4508C involves initial formation of elemental carbon that catalyzes the subsequent decomposition of
oxalate, leading to the presence of an induction period in the overall decomposition reaction. Prior incipient decomposition of sodium
oxalate to carbon and sodium carbonate is necessary for this reagent to effect aromatization. # 1999 Elsevier Science S.A. All rights
reserved.
Keywords: C±F activation; Per¯uorodecalin; Ethanedioic acid; Disodium salt
1. Introduction
[11±13] by thermal gravimetric analysis (TGA) and differ-
ential scanning calorimetry (DSC). For sodium oxalate
under N2, the main product is CO, formed via Eq. (1). Some
elemental carbon was also found but the authors' conclusion
was that this is formed only in a subsequent step by dis-
proportionation of CO (Eq. (2)), promoted by the surface.
Current interest in CF-bond activation has led to the
development of a number of new reactions [1±3] for the
de¯uorination of saturated ¯uorocarbons, that operate under
milder conditions compared to the long-established indus-
trial procedure [4] using hot iron. Many involve electron
transfer from metals or metal complexes, either thermally or
by photoexcitation. For example, sodium benzophenone
reduces per¯uorodecalin to per¯uoronaphthalene [5]. Ele-
mental magnesium [6] is also effective with Cp2TiCl2 as
catalyst and Mg-anthracene is known [7] to convert C6F12 or
C6F6 to C6F5MgF. Zinc can also reduce [8] per¯uoroalkanes
to the corresponding alkenes with decamethylferrocene as
photosensitizer. Particularly relevant to the work to be
described here is the ®nding by Weigert [10] that activated
carbon can de¯uorinate per¯uoroalkanes: for example, cis-
per¯uorodecalin is converted to per¯uoro-9,10-octalin at
4508C and to per¯uorotetralin at higher temperatures. In
no case was per¯uoronaphthalene observed, however. Some
time ago, we reported [9] that hot sodium oxalate is able to
de¯uorinate cyclic per¯uoroalkanes to the corresponding
arenes. In this paper, we look at the mechanism of the
reaction and ®nd that prior incipient decomposition of the
oxalate to carbon and Na2CO3 is required for reaction.
The thermal decomposition of various oxalate salts in air
or an inert atmosphere of N2 has previously been studied
Na2C2O4 ! Na2CO3 ‡ CO
2CO ! CO2 ‡ C
(1)
(2)
We now report that C is a catalyst for sodium oxalate
decomposition, resulting in an induction period when the
decomposition is studied under constant temperature con-
ditions.
2. Results and discussion
Eq. (3) shows the aromatization of per¯uorodecalin
(PFD) by hot sodium oxalate. Apart from unchanged
PFD, we also see per¯uorotetralin (PFT) and per¯uoro-
naphthalene (PFN), but not per¯uoro-9,10-octalin.
(3)
In our early work, the temperature of the sodium oxalate
reaction bed could not be well controlled, leading to uncer-
tainty about the true temperature. We then moved to a new
*Corresponding author. Tel.: +1-203-432-3925; fax: +1-203-432-6144.
0022-1139/99/$ ± see front matter # 1999 Elsevier Science S.A. All rights reserved.
PII: S 0 0 2 2 - 1 1 3 9 ( 9 9 ) 0 0 1 1 3 - X

68
L.H. McAlexander et al. / Journal of Fluorine Chemistry 99 (1999) 67±72
via Eq. (4) and ca. 40% via Eq. (5), is not highly dependent
on conditions. The ratios of products were determined as
follows. Carbon was recovered by dissolution of the bed in
water and ®ltering off the black residue; microanalytical
data (C, 98.4%) indicated that this material was largely
carbon. Quantitative recovery was not achieved, however,
because of the very small size of the particles formed. Any
oxalate remaining in the bed after the decomposition was
determined by titration with aq. KMnO4. CO was deter-
mined by passing the exit gases from the reaction [14] over
I2O5 which leads to quantitative reduction of the pentoxide
to gaseous elemental I2. The I2 that is formed sublimes away
completely and is expelled from the system and the resulting
loss of mass from the I2O5 column is used to calculate the
amount of CO passing through it. Eq. (6) shows that 1 mol
of I2O5 is consumed by 5 mol of CO. These analytical data
were consistent with the decomposition stoichiometry of
Eqs. (4) and (5) with a 60 : 40 branching ratio.
2Na2C2O4 ˆ 2Na2CO3 ‡ C ‡ CO2
Na2C2O4 ˆ Na2CO3 ‡ CO
I2O5ꢀs† ‡ 5COꢀg† ˆ I2ꢀg† ‡ 5CO2ꢀg†
(4)
(5)
(6)
2.1.2. Initial TGA studies
Thermogravimetric analysis (TGA) data were obtained
over a range of ®xed temperatures under a ¯ow of N2, CO2,
CO or air. This ®xed temperature procedure contrasts with
prior work [11±13] in which the temperature was increased
steadily during the runs. Fig. 2 shows the loss of mass
during decomposition of Na2C2O4 at 5058C under N2.
The initial sharp drop in weight is associated with rapid
loss of the water of hydration, as also noted in prior work.
The decomposition then follows a sigmoid curve with an
initial induction time, followed by rapid decomposition.
Once past the initial induction period, the rate of the reaction
was approximately ®rst order in oxalate (Fig. 2).
Fig. 1. Apparatus for the defluorination of perfluoroalkanes over hot
sodium oxalate. The exit gases are passed through solid I2O5 (not shown)
to determine the CO evolved (see experimental).
apparatus (Fig. 1) in which the furnace temperature could
be reproducibly measured and closely controlled. The per-
¯uoroalkane aromatization reaction was easily repeated in
this apparatus, and these studies revealed that the true
reaction temperature is substantially higher than reported
in the communication. The reaction onset temperature is ca.
4508C and satisfactory yield of products can be obtained at
465±5008C. Since these temperatures are in the range
previously reported for sodium oxalate decomposition,
[13], we suspected that the decomposition itself or the
decomposition products could be the true active species
in the reaction. We have now studied the decomposition
reaction itself in more detail and ®nd that the decomposition
temperatures previously reported are systematically high,
because they did not take account of the induction period we
now ®nd. In a fast temperature scan, rapid decomposition
does not occur below 5008C, but when sodium oxalate
samples are held at 4658C for extended periods, decom-
position can occur.
To quantify these steps, we took the end of the dehydra-
tion step as the zero time, t0, for the subsequent decom-
2.1. Sodium oxalate decomposition
2.1.1. Stoichiometry
Sodium oxalate decomposes under N2 by two main
pathways (Eqs. (4) and (5)) upon heating to 450±5508C.
The branching ratio between the two pathways, experimen-
tally determined in this work from the mass change, ca. 60%
Fig. 2. A TGA plot showing the decomposition of solid oxalate under N2
at 5038C: (a) time for ramp to temperature; (b) dehydration; (c) induction
period; (d) semilogarithmic plot of decomposition (right-hand natural log
scale), showing first-order kinetics after the induction period.

L.H. McAlexander et al. / Journal of Fluorine Chemistry 99 (1999) 67±72
69
position step. We then extrapolated the initial linear part of
the rapid decomposition curve back to the weight appro-
priate for anhydrous oxalate, to give t1. Decomposition had
normally started and the induction period had ®nished by
time t1, and therefore t1 does not precisely de®ne the start of
the decomposition phase and the end of the lag phase; it is
nevertheless a well-de®ned time related to the onset of rapid
decomposition and using it allows us to compare different
runs in a quantitative fashion. The rate of the oxalate
decomposition was quanti®ed by recording the time, t2,
when the sample had undergone half the weight loss in the
decomposition part of the weight loss curve. The decom-
position did not exactly follow any simple kinetic law, such
as ®rst-order decomposition. The induction time was con-
sidered to be (t1 t0) and the reaction half-time was con-
sidered to be (t2 t1).
The half lives and induction periods for the decomposi-
tion of sodium oxalate over a range of temperatures, under
different gases, and mixed with other materials are reported
in Table 1. Heating pure sodium oxalate at 4148C produced
dehydration but no decomposition at all over a period of
several days. At 4438C the decomposition was very slow,
and had an induction period on the order of one day. At
temperatures of 4558C and higher, the induction period was
shorter and decomposition was complete in a few hours up
to two days.
constantly rising, which would have tended to mask the
existence of a lag-time and to give an apparent decomposi-
tion temperature higher than the one which would have been
found had the study been carried out at a series of ®xed
temperatures. In addition, the samples in those experiments
[11,12] were subjected to high temperatures where we ®nd
carbon formation is much less obvious, which may have
contributed to the mistaken mechanistic interpretations in
the previous studies.
2.1.3. Role of carbon in oxalate decomposition
For these reasons, we therefore considered the possibility
that elemental carbon might catalyze the decomposition
step. In an initial test of this idea, several samples were
subjected to a brief (1 h) exposure to high temperature
(5008C) to cause a small amount of the oxalate to decom-
pose to carbon; this phase was accompanied by a visible
darkening but did not lead to any substantial loss of the
initial oxalate (ca. 1.5% loss of mass by TGA over 60 min,
after loss of water). The same samples were then held at a
lower temperature (e.g., 4148C) that would normally not
permit any decomposition without the pretreatment. As
expected if carbon was important in the decomposition, a
pretreated sample held at 4148C did decompose completely
over the next 24 h. In contrast pure oxalate showed no
decomposition under these conditions. However, both pre-
treated and pure samples held at 3728C showed no decom-
position over several days.
Clearly, since the decomposition reaction produces car-
bon and carbon promotes decomposition, initial carbon
formation is expected to make the decomposition autoca-
talytic. This is consistent with the presence of an induction
period in the decomposition of pure oxalate. This picture
was con®rmed by showing that a mixture of sodium oxalate
and Darco activated carbon, ground together and heated
under N2, decomposed at 4678C without showing an induc-
tion time (Fig. 3). Control reactions, in which oxalate was
ground alone or with inert materials such as Na2CO3 or
NaCl, showed some diminution of the induction time,
presumably as a result of the mechanical effects of grinding,
but not its entire abolition, as with carbon. At 4148C a
mixture of sodium oxalate and activated carbon showed no
decomposition over several days, suggesting that the ground
mixture is less effective in activating the reaction than the
intimate mixture formed on short pyrolysis of sodium
oxalate.
The end of the induction period and onset of the rapid
decomposition near time t1 is associated with a visible
darkening of the sample in a way that suggests carbon
formation and subsequent oxalate decomposition are
related. The existence of a lag-time and the formation of
carbon were not in accord with the picture that emerges
from the prior studies, in which the initial step was taken to
be a simple conversion of oxalate to CO (Eq. (2)). In the
prior work, the temperature was not ®xed, as here, but
Table 1
Half-lives of thermal decomposition reactions of sodium oxalate
Additive
Gasa
Temperature (8C)b
t(1/2) (min)c
None
None
N2
N2
414
443
455
466
503
455
466
466
466
466
450
No reactiond
>1 daye
1700
1060
200
None
N2
None
N2
None
4%f Cx
N2
N2
1680e
None
CO2
CO2
CO2
Air
CO
130
4% Na2CO3
4% Cx
None
45
40g
2.1.4. Other effects
160
Small but distinct effects on the rate of decomposition
were found by changing the ¯ow rates of nitrogen or the
amount of oxalate. For example, decomposition was slightly
faster under increased N2-¯ow and with smaller samples of
oxalate. This suggested that inhibition of the decomposition
by the gaseous reaction products, notably CO2, could be an
issue. Faster N2-¯ow rates and taking smaller oxalate sam-
ples would both be expected to diminish the CO2 partial
None
33
a In which thermal analysis was carried out.
b Constant temperature conditions.
c Of reaction (see text).
d But dehydration of salt does occur.
e Induction period > 1 day.
f Weight %.
g No induction period.

70
L.H. McAlexander et al. / Journal of Fluorine Chemistry 99 (1999) 67±72
Fig. 3. A TGA plot showing the decomposition of mixtures of 4 wt.% of
an additive, either C or Na2CO3, ground with sodium oxalate and heated
under N2 at 4658C. Inset: expansion of the induction period region of the
plot, showing the lack of any induction period in the case of the carbon/
oxalate mixture.
Fig. 4. A TGA plot showing the decomposition of sodium oxalate under a
flow of N2, CO2, CO or dry air.
Use of either CO2 or air as the carrier gas for the reaction
inhibited the de¯uorination of PFD in these experiments
(Table 2). Under CO2, oxalate decomposition to carbon was
shown above to be slower and under air, the carbon formed
is rapidly oxidized to CO2, so both of these conditions
disfavor the presence of carbon in the reaction bed. PFD
reduction being inhibited under these conditions is consis-
tent with the proposal that carbon is an active species in the
reduction.
The prior work [10] suggests that reduction of PFD by
activated carbon alone produces per¯uoro-9,10-octalin or, at
higher temperatures, PFT but not PFN. In our work we do
not ®nd the octalin but do ®nd both PFTand PFN, and so we
considered the possibility that the presence of base in the C/
Na2CO3 experiments changes the selectivity. In our hands,
the C-reduction of PFD without base gave PFT, and under
certain conditions, small amounts of PFN. This minor
disagreement with prior data may be the result of our using
a different form of activated carbon, or simply of the
different reaction conditions we employed.
pressure. This idea was con®rmed by heating the oxalate
under a gas ¯ow of pure CO2, in which case a substantial
decrease in the decomposition rate was observed. Fig. 4
shows the substantially lower rate of a decomposition under
CO2 versus that in a standard case under nitrogen. This may
be a re¯ection of CO2-loss from an intermediate playing a
role in oxalate decomposition. The decomposition was also
slowed by heating under a ¯ow of dry air; this supports the
inference that carbon catalyzes the reaction, as the carbon
produced would quickly be oxidized to CO2 at high tem-
peratures under air. Heating sodium oxalate under a ¯ow of
CO produced no signi®cant change in the decomposition
rate as compared to N2, indicating that CO is inert with
respect to this process.
3. Perfluoroalkane reactions
3.1. Role of carbon in perfluoroalkane reduction
Prior work [10] on PFD reduction by activated carbon
indicates the formation both of graphite ¯uoride and a
mixture of light ¯uoroalkenes, the latter believed to result
from thermal decomposition of graphite ¯uoride at the
reaction temperatures. This decomposition made the mass
balance very dif®cult to obtain. In our case too, there is
Based upon the above data and the fact that elemental
carbon itself has been shown to be active in ¯uorocarbon
de¯uorination, a possible mechanism for PFD aromatization
is that the decomposition products of the oxalate reaction
are the true active species. Of the two decomposition
products, carbon and Na2CO3, only carbon is a plausible
reducing agent, but it could be assisted by the presence
of base. We could not exclude the possibility that
carbon catalyzes the reduction of PFD by sodium oxalate
itself, just as carbon catalyzes the decomposition of the
oxalate. Prior work by Weigert [10] has shown that ele-
mental carbon is an ef®cient de¯uorination reagent for
per¯uoroalkanes.
In initial experiments to test this idea, we showed that
mixtures of activated carbon and Na2CO3 are indeed active
for PFD aromatization in the complete absence of sodium
oxalate. The yields and selectivities found were very similar
to those shown in Table 2 for sodium oxalate, consistent
with decomposition products being active.
Table 2
Results of perfluorodecalin aromatization over sodium oxalate at 4658C
Gasa
Preheat
time (h)b
PFD
additionc
%Cd
%Fd
%PFNe
%PFTe
N2
1
2
16.8
16.5
17.4
15.3
2.5
41
11
CO2
Air
CO
1.2
1.2
1
1
2.1
35
30
42
14
9
2
1.5
0.085
4.7
10
a Carrier gas.
b Before PFD addition.
c Time (h) taken for addition of 2.0 ml PFD.
d Analytical data found in bed residue after reaction.
e Fraction in product mixture.

L.H. McAlexander et al. / Journal of Fluorine Chemistry 99 (1999) 67±72
71
evidence of the formation of small amounts (ca. 5%) of
¯uoroalkenes. When the exit gases from the C/Na2CO3
experiments were passed through methanol to trap any
¯uoroalkenes as methyl ethers, 19F NMR spectroscopy
showed a variety of new resonances in the range 105 to
150 ꢀ, consistent with the presence of a variety of per-
¯uorinated ethers.
Analyzer. Small amounts (4±30 mg) of sodium oxalate or
mixtures of oxalate with carbon, sodium carbonate, or
sodium chloride were heated to constant temperatures of
350±5508C until decomposition was complete. In early
experiments the solid sodium oxalate (hygroscopic) was
dried overnight at 1208C before use, but control experiments
showed that water had no effect on the decomposition. In
any case, with the inherent limits on the speed of handing
and the very small samples used, even the dried solid gained
a signi®cant amount of water during loading, so the oxalate
was thereafter used as obtained. To negate the effect of
differences in humidity on different days the time and mass
values used in the calculation were those obtained after the
loss of water from the sample.
Table 1 shows the results of heating samples of oxalate,
alone or ground with other substances, under a variety of
reaction conditions. Half-life values are derived as follows:
a tangent to the decomposition curve, before the in¯ection
point of the curve, was calculated and extrapolated back to
the initial mass of the sample. The corresponding time index
was used in calculating the half-life, de®ned as the time
corresponding to the average of the initial and ®nal sample
masses. Induction period is de®ned as the difference
between the extrapolated initial time index and the actual
initial time.
The C/Na2CO3 system shows signi®cant differences from
the activated carbon only system, however. First, only with
C/Na2CO3 did we obtain signi®cant yields of PFN, and only
small amounts of PFT. Both in our work and in prior work
with pure activated carbon as reductant, PFT was the main
product under analogous conditions. Second, in our C/
Na2CO3 system carbon appears to release some of the
¯uoride to the base as ¯uoride ion, which ends up as sodium
¯uoride. The intimacy of the mixture of Na2CO3 and
graphite obtained through the decomposition of the oxalate
salt appears to allow any ¯uorinated carbon formed to
release ¯uoride ions; a mechanically ground mixture of
Na2CO3 and activated carbon in similar proportions gave
similar results. We were able to detect F ions in the bed
after reaction by 19F NMR spectroscopy using a standard
solution of Na(CO2CF3) in D2O, indicating that the ¯uoride
abstracted from PFD does not end up only on the carbon but
also in the form of ¯uoride ion. Addition of authentic
¯uoride ion enhanced the peak con®rming the assignment
to ¯uoride ion. Graphite ¯uoride is known to react with
bases, including carbonate ion, to give carbon oxides and
¯uoride ions [14,15].
Most notably, the decomposition was slowed by use
of CO2 or air as cover gas and accelerated by grinding
the solid with activated carbon before reaction. No decom-
position of oxalate was observed at temperatures below
4208C unless the sample had ®rst been heated to a higher
temperature to initiate the decomposition. Decomposition
of sodium oxalate below 4008C was never observed, even
in samples that had previously been heated to initiate
decomposition.
4. Conclusions
We propose that carbon is the key reductant both in
sodium oxalate decomposition and in PFD reduction by
sodium oxalate or C/Na2CO3. The induction time in sodium
oxalate decomposition can be abolished by intimate incor-
poration of activated carbon. PFD reduction by sodium
oxalate does not take place until carbon begins to be formed
in thermal decomposition. The ¯uorocarbon products from
sodium oxalate resemble those from C/Na2CO3, but those
from activated carbon alone are distinctly different. We
conclude that the sodium oxalate system for reduction of
¯uoroalkanes is a modi®ed form of the prior [10] activated
carbon system.
5.2. Perfluoroalkane reactions
All per¯uoroalkane reactions used a similar procedure:
sodium oxalate, dried at 1208C, was weighed into the
apparatus shown in Fig. 1. A 100 layer of 4 mm glass beads
was added on top of the oxalate bed, and the apparatus was
¯ushed thoroughly with the cover gas (speci®ed below) at a
rate of ca. 200 ml/min. The oxalate bed was then preheated
in the furnace under N2 ¯ush for 2 h at 4658C. Per¯uor-
odecalin was placed in the dropping funnel and dropped
slowly onto the hot glass beads, so that it was vaporized and
carried through the oxalate bed. Volatiles in the exhaust
were trapped at 788C in a preweighed trap. Once the
per¯uoroalkane addition was complete, the furnace was
held at the reaction temperature 5 more minutes and then
allowed to cool to room temperature under a constant ¯ow
of gas.
5. Experimental details
Per¯uorodecalin, sodium oxalate and DarcoTM activated
carbon were used as received from Aldrich.
5.1. TGA experiments
Trapped volatiles were thawed and weighed, then imme-
diately dissolved in 1,3-bis(tri¯uoromethyl)benzene. Pro-
duct ratios were determined by GC/MS. The only signi®cant
¯uorocarbon peaks were those of PFN, PFT and PFD.
Data for the decomposition of sodium oxalate were
obtained using a Perkin-Elmer TGA-7 Thermogravimetric

72
L.H. McAlexander et al. / Journal of Fluorine Chemistry 99 (1999) 67±72
The solid bed was weighed and a small sample removed
References
for elemental analysis (Robertson Microlit). The remaining
solid was extracted with distilled water. The carbon was
recovered by ®ltration through a glass frit and dried to
constant weight. Total ¯uoride and total carbon (including
both graphitic and inorganic) in the bed were determined by
elemental analysis.
The ®ltrate was diluted volumetrically and portions
titrated with standard solutions of KMnO4 to determine
oxalate concentration and HCl to determine carbonate. In all
cases oxalate was found to have been completely decom-
posed. In samples that had been heated without addition of
per¯uoroalkane, oxalate was quantitatively converted to
carbonate. In the bed residues from de¯uorination reactions,
the sum of the ¯uoride and carbonate present after reaction
was equal to the initial oxalate.
[1] J.L. Kiplinger, T.G. Richmond, C.E. Osterberg, Chem. Rev. 94
(1994) 373.
[2] R.P. Hughes, Adv. Organomet. Chem. 31 (1990) 183.
[3] J. Burdeniuc, B. Jedlicka, R.H. Crabtree, Chem. Ber. 130 (1997)
145.
[4] B. Gething, C.R. Patrick, M. Stacey, J.C. Tatlow, Nature 183 (1959)
588.
[5] J.A. Marsella, A.G. Gilicinski, A.M. Coughlin, G. Pez, J. Org. Chem.
57 (1992) 2856.
[6] J.L. Kiplinger, T.G. Richmond, J. Am. Chem. Soc. 118 (1996)
1805.
[7] C.M. Beck, Y.-J. Park, R.H. Crabtree, JCS Chem Comm. (1998)
693.
[8] J. Burdeniuc, R.H. Crabtree, J. Am. Chem. Soc. 118 (1996)
2525.
[9] J. Burdeniuc, R.H. Crabtree, Science 271 (1996) 340.
[10] F.J. Weigert, J. Fluorine Chem. 65 (1993) 67.
[11] D. Dollimore, D.L. Griffiths, J. Thermal Anal. 2 (1970) 229.
[12] W. Balcerowiak, J. Wasilewski, C. Latocha, J. Thermal Anal. 18
(1980) 57.
Results are summarized in Table 2.
Acknowledgements
[13] Z. Gontarz, Pol. J. Chem. 65 (1991) 687.
[14] N.N. Greenwood, A. Earnshaw, Chemistry of the Elements, 1st ed.,
Pergamon, New York, 1984, p. 998.
We thank the Department of Energy for funding and Dr.
F.J. Weigert (du Pont) for helpful suggestions.
[15] Y.N. Novitov, M.E. Vol0pin, Usp. Khim. 40 (1971) 1568.